Professional Documents
Culture Documents
Vilas R. Kalamkar
Katarina Monkova Editors
Advances in
Mechanical
Engineering
Select Proceedings of ICAME 2020
Lecture Notes in Mechanical Engineering
Series Editors
Fakher Chaari, National School of Engineers, University of Sfax, Sfax, Tunisia
Mohamed Haddar, National School of Engineers of Sfax (ENIS), Sfax, Tunisia
Young W. Kwon, Department of Manufacturing Engineering and Aerospace
Engineering, Graduate School of Engineering and Applied Science, Monterey, CA,
USA
Francesco Gherardini, Dipartimento Di Ingegneria, Edificio 25, Università Di
Modena E Reggio Emilia, Modena, Modena, Italy
Vitalii Ivanov, Department of Manufacturing Engineering Machine and Tools,
Sumy State University, Sumy, Ukraine
Lecture Notes in Mechanical Engineering (LNME) publishes the latest develop-
ments in Mechanical Engineering - quickly, informally and with high quality.
Original research reported in proceedings and post-proceedings represents the core
of LNME. Volumes published in LNME embrace all aspects, subfields and new
challenges of mechanical engineering. Topics in the series include:
• Engineering Design
• Machinery and Machine Elements
• Mechanical Structures and Stress Analysis
• Automotive Engineering
• Engine Technology
• Aerospace Technology and Astronautics
• Nanotechnology and Microengineering
• Control, Robotics, Mechatronics
• MEMS
• Theoretical and Applied Mechanics
• Dynamical Systems, Control
• Fluid Mechanics
• Engineering Thermodynamics, Heat and Mass Transfer
• Manufacturing
• Precision Engineering, Instrumentation, Measurement
• Materials Engineering
• Tribology and Surface Technology
To submit a proposal or request further information, please contact the Springer
Editor of your country:
China: Dr. Mengchu Huang at mengchu.huang@springer.com
India: Priya Vyas at priya.vyas@springer.com
Rest of Asia, Australia, New Zealand: Swati Meherishi at
swati.meherishi@springer.com
All other countries: Dr. Leontina Di Cecco at Leontina.dicecco@springer.com
To submit a proposal for a monograph, please check our Springer Tracts in
Mechanical Engineering at http://www.springer.com/series/11693 or contact
Leontina.dicecco@springer.com
Indexed by SCOPUS. The books of the series are submitted for indexing to
Web of Science.
Editors
Advances in Mechanical
Engineering
Select Proceedings of ICAME 2020
123
Editors
Vilas R. Kalamkar Katarina Monkova
Department of Mechanical Engineering Faculty of Manufacturing Technologies
Visvesvaraya National Institute Technical University of Kosice
of Technology Presov, Slovakia
Nagpur, India
This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd.
The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721,
Singapore
Preface
It is our pleasure and honour to bring you these findings of research and innovation
from the International Conference on Advances in Mechanical Engineering
(ICAME 2020) held on 10 and 11 January 2020 organized by the Department of
Mechanical Engineering, Visvesvaraya National Institute of Technology, Nagpur,
India. This conference was the beginning of a year-long Diamond Jubilee year
celebrations of VNIT Nagpur’s foundation day. ICAME 2020 provided an inter-
national forum where researchers, academicians and scientists from interdisci-
plinary fields presented their synergistic solutions to frontier issues of mechanical
engineering.
We received around 200 research manuscripts from various domains like thermal
engineering, CFD, machine design, sustainability, IoT, robotics, manufacturing
engineering, biomechanics, machine learning, machine vision, optimization,
industrial engineering and many other allied domains. During 12 technical sessions
spread over two days, the conference witnessed the presentations by participants
from different NITs, IITs and universities in India as well as abroad. Out of 200 plus
received papers, only 101 manuscripts are accepted for inclusion in this proceedings.
The keynote talks, technical sessions and panel discussions of the conference were
focused on holistic contributions of mechanical engineering concerning the society
in general and industry in particular.
We are highly grateful to the authors for their contributions and all the expert
reviewers for their valuable advice. We take this opportunity to thank the members
of the organizing committees for their unwavering commitment.
We are indebted to TEQIP-III, MSME-DI Nagpur, DST-SERB New Delhi,
MOIL Nagpur, SBI VRCE Branch Nagpur and many other industries, establish-
ments and agencies in India for their generous sponsorship and support for the
conference.
v
vi Preface
We extend our heartfelt gratitude to Springer Nature for its professional assis-
tance and particularly Mr. Akash Chakraborty and Ms. Rini Christy who supported
this publication.
vii
viii Contents
xvii
Dual Quaternion-Based Kinematic
Modelling of Serial Manipulators
1 Introduction
During robot programming, the orientation and position of robot end-effector must
change as smoothly as a human hand. Quaternions, initially developed as a gener-
alization of complex numbers for three dimensions, are robust than the popularly
used Euler angles and rotation matrices for representing orientations. They are com-
pact, computationally efficient, immune to gimbal lock and mathematical singular-
ities, and also provide natural orientation interpolation [1]. However, using vectors
and quaternions for representing simultaneous translations and rotations leads to
inconsistencies [2].
A dual quaternion (DQ) uses dual numbers to unify rotations and translations
into a single state instead of defining separate vectors for them [3]. DQs have been
used for kinematic modelling and pose control of serial manipulators [4, 5]. The
available literature on modelling with quaternions and DQs shows that there are
ambiguities in representation (Hamiltonian and NASA-JPL), handedness (right and
left) and reference frames (global and body) [6].
Differential kinematics is a widely used approach for inverse kinematics (IK)
modelling. DQ and Plucker coordinates have been used to derive the geometrical
Jacobian for serial manipulators [7]. DQs and quaternion exponential maps have been
used to solve IK problems by considering joint limits [8]. Unlike the DQ differential
operator, dealing with Jacobian matrix is well discussed in the literature.
The flow of the paper is as follows: Sect. 2 introduces dual quaternions, which
is used in Sect. 3 to develop the forward and inverse kinematic models for a serial
manipulator. The approach is applied for Kuka LBR IIWA 7 R800 cobot in Sect. 4.
The results are discussed in Sect. 5, followed by conclusion in Sect. 6.
2 Mathematical Preliminaries
vector û. The quaternion t = [0, t] depicts translation t = tx , t y , tx . Then, the com-
posite transformation of rotation r followed by translation t is given by the unit DQ
p = t ⊗ r = 1 + 21 t ⊗ r̂ + 0 = r̂ + 2 t ⊗ r̂
= cos θ2 , sin θ2 û + 21 −sin θ2 (t · û), cos θ2 t + sin θ2 (t × û) (1)
For a given unit DQ p = pr + pd , quaternions r̂ and t are obtained using r̂ = pr and
t = 2 pd ⊗ pr∗ = 2 pd ⊗ r̂ ∗ . For rotational quaternion r̂ = [r0 , r], rotation parameters
θ and û are obtained as θ = 2 cos−1 (r0 ) and û = r/ sin θ2 . When a vector v 0 is
subjected to transformation p, the new vector v 1 is obtained from v1 = [1, 0] +
[0, v 1 ] = p ⊗ ([1, 0] + [0, v 0 ]) ⊗ p̄ ∗ .
q̇, JJT q̇
α=
[10], where ·, ·
is a dot product operator. This is called the Jaco-
JJT q̇, JJT q̇
bian transpose (J T ) method. To further reduce the chances of JT J losing rank, the
damped least squares method (DLS) is used [10]. Here, a damping constant δ ≈ 0.001
−1 T
in the diagonal elements modifies the expression to θ̇ = JT J − δ 2 I J q̇. The ith
∂ 0
column of Jacobian J is ( q n ) where n = number of joints.
∂θi
For given pose q = qr + qd , linear velocity v ∈ R3 and angular velocity w ∈ R3 ,
the DQ differential operator is q̇ = 21 ξ ⊗ q where ξ = [0, ω] + [0, v + t × ω] is
twist in world frame [11] and [0, t] = 2qd ⊗ qr∗ . Then, the DQ differential operator
becomes q̇ = 21 ([0, ω] + [0, v + t × ω]) ⊗ (qr + qd ).
For initial pose q 0 of a serial manipulator, let corresponding joint parameters be
θ 0 . Now, for some J (θ k ) and q̇ k , when θ̇ k is obtained from IK model, then θ k is
updated to θ k+1 = θ k + θ̇ k , and updated pose from FK model becomes X F (θ k+1 ).
However, due to various reasons such as linearization, this does not match the next
pose q k+1 = q k ⊗ q̇ k . The pose error, given by (q k ) E = X F (θ k+1 )∗ ⊗ q k+1 , is fed
back to the next differential pose q̇ k+1 .
Dual Quaternion-Based Kinematic Modelling of Serial Manipulators 5
The classical DH convention [9] is used to assign coordinate frames to the 7-dof Kuka
LBR IIWA 7 R800 cobot (or collaborative robot) and derive the robot architecture
as seen in Fig. 2.
A library for dual quaternions is developed in Python 3.6 for implementing the
algorithms. The tests are performed on a Windows 8 PC with Intel i5-4570 3.2 GHz
processor, 8 GB DDR3 RAM and GeForce 625 graphics card. The DQ outputs of
forward kinematics model for a series of joint inputs are tallied with results from
Fig. 2 Cobot under study (a) with frame assignment (b) and DH parameters [9] (c)
RoboAnalyzer software. The two IK models are applied to the trajectory shown in
Fig. 3. The red, green and blue arrows correspond, respectively, to the X , Y and Z
axes of the end-effector coordinate frame.
The DQ FK model results matched perfectly with those obtained from the homoge-
neous transform matrix (HTM) FK model and RoboAnalyzer software. The DQ FK
model required 224 additions and 312 multiplications as opposed to 234 additions
and 356 multiplications needed in HTM. It is seen that using DQs over HTMs saved
computational time, the savings being up to 50% in some cases.
Applying the J T and DLS methods on the trajectory shows that J T gives solutions
faster, but also gives more jerky motions. For J T , mean solution time ranges from
16 to 33 ms, whereas for DLS, the same ranges from 27 to 66 ms. It is also observed
that the DLS method outperforms the Jacobian transpose method when not near
singularity. The graphs of two DQ components in Fig. 4 show that fluctuations in DLS
method are less compared to those in J T method. It is seen that the trajectory loop
did not get closed in either method, which means pose error is not eliminated even
when feedback is used. This makes a case for exploring more feedback controllers.
6 Conclusion
In this work, dual quaternions (DQs) were used to develop the forward and inverse
kinematics models for a serial manipulator. The DQ differential operator and two
methods of solving inverse differential kinematics with DQs were discussed. The
approach was implemented on the Kuka LBR IIWA 7 R800 cobot, and error during
simulation of tracing a loop trajectory was studied. Though DQs seem unintuitive
Fig. 4 Comparison of deviations in trajectory traced with J T and DLS IK algorithms taking first
and sixth DQ elements as example
Dual Quaternion-Based Kinematic Modelling of Serial Manipulators 7
to humans, they lend themselves well to working with elaborate trajectories. Future
works include velocity and acceleration control, as well as DQ implementation of
faster IK algorithms such as FABRIK.
References
1. Shoemake K (1985) Animating rotation with quaternion curves. In: ACM SIGGRAPH com-
puter graphics, vol 19. ACM, pp 245–254
2. Banerjee P, Zetu D (2001) Virtual manufacturing. Wiley
3. Jia YB (2013) Dual quaternions. Technical report, Iowa State University
4. Daniilidis K (1999) Hand-eye calibration using dual quaternions. Int J Robot Res 18(3):286–
298
5. Pham HL, Perdereau V, Adorno BV, Fraisse P (2010) Position and orientation control of robot
manipulators using dual quaternion feedback. In: 2010 IEEE/RSJ international conference on
intelligent robots and systems, pp 658–663
6. Sommer H, Gilitschenski I, Bloesch M, Weiss S, Siegwart R, Nieto JI (2018) Why and how to
avoid the flipped quaternion multiplication. CoRR
7. Özgür E, Mezouar Y (2016) Kinematic modeling and control of a robot arm using unit dual
quaternions. Robot Auton Syst 77:66–73
8. Kenwright B (2013) Inverse kinematics with dual-quaternions, exponential-maps, and joint
limits. Int J Adv Intell Syst 6(1, 2):53–65
9. Siciliano B, Sciavicco L, Villani L, Oriolo G (2009) Robotics: modelling. In: Planning and
control. Springer, London
10. Aristidou A, Lasenby J (2009) Inverse kinematics: a review of existing techniques and intro-
duction of a new fast iterative solver. Technical report, University of Cambridge
11. Wang X, Han D, Yu C, Zheng Z (2012) The geometric structure of unit dual quaternion with
application in kinematic control. J Math Anal Appl 389(2):1352–1364
Performance Analysis of Corrugated
Inclined Basin Solar Distillation System
Coupled with Parabolic Trough Collector
Abstract Several designs of solar distillation system have been built over the past
century. However, the development of an economical system with high productivity
is a major challenge. Various researchers worked to improve the productivity of
solar distillation system by improving the rate of evaporation and/or the rate of
vapor condensation. In the current work, the evaporation rate is enhanced by basin
modification and using a secondary heating medium. An inclined corrugated basin
solar still is designed and fabricated and coupled with a parabolic trough collector.
Experimental study was carried at Nagpur (21.14° N, 79.0882° E) during the months
of April and May, and results indicate 13.58% increase in the thermal efficiency.
Further, CFD analysis is carried out by using RNG (k − ε) turbulence model in
ANSYS Fluent. The CFD results were found to be in good agreement with the
experimental results, thus validating the CFD model to carry out any modifications
in the future.
Keywords Inclined basin solar still · CFD analysis · Corrugated basin · Parabolic
trough collector
1 Introduction
Water is one of the basic human needs. Water treatment amounts for about 8%
of global energy consumption [1]. The use of solar still for water treatment and
distillation is one of the most popular renewable energy solutions. However, low
productivity of solar stills has been a matter of concern.
Several researches have suggested various methods to enhance the productivity
of the solar still with the use of fins, multi-basin, energy storage materials [2], wick
[3, 4] reflectors [5–7], and coating the absorber surface with different films [8]. Also,
increasing the absorber area is suggested to help improve the performance of solar
still which can be achieved by providing corrugations of V-shaped [9], wave-shaped
[10], covered absorber plate with copper chips [8], etc. Some investigators have also
used the solar collectors like flat plate collector [11, 12], evacuated tube collector
[13], tracked parabolic trough collector [14], etc., as secondary heating source.
The present work focuses on the improvement of rate of evaporation. An inclined
corrugated basin solar still coupled with a parabolic trough collector is designed and
fabricated. The performance of the system is simulated using CFD technique and
validated using experimental analysis. The experiments are performed at Nagpur
(21.14° N, 79.08° E) in the month of April and May. The details of simulation and
experimental studies are discussed in the subsequent text.
2 Working Principle
The schematic diagram of the modified system of inclined corrugated basin solar still
coupled with parabolic trough collector is shown in Fig. 1. In the modified system,
the corrugated basin receives heat from two sources, viz. incident solar radiations
and the parabolic trough collector.
The parabolic trough collector acts as the secondary heating source which heats
water, and the heat of the water is then passed through the corrugated channel in
the basin. This water is recirculated in the heating water circuit of parabolic trough
collector. The raw water to be distilled is circulated over the basin via header pipe.
The header pipe is drilled at specific interval in such a way that the raw water falls
drop by drop over the basin and thus get heated and evaporates. The evaporated
water is condensed as it encounters the top glass, and thus, the distilled condensate
is collected. The corrugated basin is modified by providing well-designed V-shaped
corrugations and orienting it in inclined position to increase the surface area of the
absorber plate.
Fig. 1 Schematic diagram of the inclined corrugated basin solar still coupled with parabolic trough
collector
Performance Analysis of Corrugated Inclined Basin Solar … 11
Fig. 2 Experimental system of conventional and corrugated inclined basin still coupled with
parabolic trough collector
3 Experimental Analysis
4 Computational Analysis
Geometry of corrugated inclined basin solar still was created in CATIA simulation
software. The geometry file was then imported in ANSYS Fluent. The rectangular
meshing was generated by AUTOMESH. Average value of the solar irradiance at
the given location and average estimated temperature values were given as input.
Glass front, and back wall temperatures were constant. Heat flux, opaque thermal,
and semitransparent conditions were implemented on; front, side, and back walls of
corrugated absorber plate and glass, respectively.
The k − ε turbulence model is the most common model used in CFD to simulate
mean flow characteristics for turbulent flow conditions. However, RNG (k − ε) model
is selected for analysis as it gives a general description of turbulence by two transport
equations. The turbulent kinetic energy Eq. (1) is same as in case of k − ε turbulence
model; however, there is an additional term in its ε Eq. (3) which is significant in
improving the accuracy for rapidly strained flows.
12 S. Joshi et al.
The temperature contours of the inclined corrugated solar still and glass cover are
shown in Fig. 3a, b, respectively. It is observed that the mean temperature of the
inclined corrugated basin is 335 K.
Fig. 3 a Temperature contour for corrugated inclined basin solar still, b temperature contour for
glass cover
Performance Analysis of Corrugated Inclined Basin Solar … 13
Fig. 4 Average distillate output of conventional and corrugated solar still at a given radiation
intensity and time of the day
It is observed that the mean temperature of the glass cover is around 302 K. Based
on the CFD analysis, the productivity of the modified solar still is calculated to be
equal to 279 ml/h.
Figure 4 shows the variation of average distillate output of conventional and corru-
gated solar still along with variation in solar radiation intensity with respect to the
time of the day.
It observed that corrugated inclined solar cell provided higher productivity than
the conventional system irrespective of the time of the day. It can be observed that the
highest output for conventional and corrugated solar still was found to be 204 ml/h
and 232 ml/h, respectively, at 1300 h.
The variation of average instantaneous experimental and theoretical efficiencies
of conventional and corrugated basin solar still with respect to time is shown in
Fig. 5. The efficiency of corrugated solar still was found to be higher, i.e., 32.9% as
compared to that of conventional solar still which was just 23.62%.
14 S. Joshi et al.
Fig. 5 Average instantaneous theoretical and experimental efficiency of conventional and inclined
corrugated solar still
6 Conclusion
The present work focuses on the productivity improvement of the solar still by
improving its rate of evaporation. An inclined corrugated basin solar still is designed.
The inclined basin is modified as a corrugated channel, and it is coupled with a
parabolic trough collector. The heat from the parabolic trough collector is trans-
ferred to the modified basin using suitable flow arrangement. Thus, the basin surface
receives heat from direct solar radiations as well as from the coupled parabolic trough
collector.
The performance of the system is simulated using CFD technique. The simulation
results show the improved performance of the modified system over conventional
one. The CFD results are validated by the experimental analysis. The modified solar
distillation system is fabricated. The experiments are performed at Nagpur (21.14° N,
79.08° E) in the month of April and May. The current work, solar still basin modifi-
cation, and secondary heating source are used to enhance the overall productivity of
solar still. Based on the experimental study, it was observed that the productivity of
the conventional solar still was found to be 204 ml/h and that by modified solar still
was 232 ml/h thus giving a clear indication of productivity improvement. Also, the
efficiency of corrugated solar still was found to be higher, i.e., 32.9% as compared to
that of conventional solar still which was just 23.62%. The CFD simulation shows
good agreement with the experimental results, thus validating the CFD model to
carry out further modification in the system.
Performance Analysis of Corrugated Inclined Basin Solar … 15
References
1. Pugsley A et al (2018) Solar distillation potential around the world (Chap. 2). In: Renewable
energy powered desalination handbook, Butterworth-Heinemann, Oxford, pp 47–90. https://
doi.org/10.1016/C2017-0-02851-3
2. Panchal H et al (2017) Various methods applied to solar still for enhancement of distillate
output. Desalination 415:76–89
3. Fathy M et al (2018) Experimental study on the effect of coupling parabolic trough collector
with double slope solar still on its performance. Sol Energy 163:54–61
4. Aybar HS (2005) An experimental study on an inclined solar water distillation system.
Desalination 180:285–289
5. Tanaka H et al (2006) Experimental study of a basin type solar still with internal and external
reflectors. Desalination 197:205–216
6. Khalifa AJN et al (2013) Effect of inclination of the external reflector of simple solar still in
winter. Desalination 264:129–133
7. Omara ZM et al (2013) Enhancing the stepped solar still performance using internal reflectors.
Desalination 314:67–72
8. Joshi S et al (2019) Productivity improvement of single using basin inclined solar still using
enamel coating and copper chips. Desal Water Treat 156:161–167
9. Shalaby SM et al (2016) An experimental investigation of corrugated absorber solar still with
wick and reflectors. Desalination 398:247–255
10. Omara ZM et al (2013) Experimental investigation of corrugated absorber solar still with wick
and reflectors. Desalination 318:25–33
11. Dwivedi VK et al (2010) Experimental validation of thermal model of a double slope active
solar still under natural convection mode. Desalination 250:49–55
12. Taghvaei H et al (2014) A thorough investigation of the effects of water depth on the
performance of active solar stills. Desalination 347:77–85
13. Singh RV et al (2013) Performance of a solar still integrated with evacuated tube collector in
natural mode. Desalination 318:25–33
14. Hassan H et al (2018) Experimental work on the effect of saline water medium on the
performance of solar still with tracked parabolic trough collector (TPTC). Renew. Energy
135:136–147
Mechanical Design of Omnidirectional
Spherical Wall Traversing Robot
Abstract Since the past few decades, many mechanisms were developed for wall-
climbing robots. Robotic systems having omnidirectional surface traversing ability
independent of its inclination require complex morphology transformation for a floor
to wall transition as well as perfect adhesion on the vertical surface. This paper depicts
the development of the mechanical design of the robot producing upthrust due to
coaxial antiparallel propeller mechanism to grip the vertical surfaces. To provide the
omnidirectional capability, 2-DoF gimbal mechanism is used. Bi-propeller coaxial
thruster situated inside the gimbal mechanism is controlled resembling its localization
on the wall and also it helps the robot to adhere to the wall. The novel design of this
robot is newly aimed and its motion is mathematically modeled.
1 Introduction
Robots having the ability to climb the vertical surfaces are categorized into wall-
climbing robots. Wall climbers have been developed across the globe for surveillance,
inspections, flaw detection, cleaning, and maintenance, etc. The proposed mechanism
has a 2-DOF bi-propeller plane which allows getting any direction of thrust on to the
wall, which helps it to move freely and remain attached to the surface.
Research work of this project is funded and supported by IvLabs VNIT, Nagpur, India.
Producing perfect adhesion with the vertical surfaces has been a considerable
quandary in wall-climbing robots. Since the past decade, mechanisms based on spiney
gripper [1], magnetic [2], elastomer adhesive [3, 4], electromagnetic/electrostatic
forces [5], grasping, wheel-driven [6, 7], and suction-vacuum [8, 9] are introduced in
the wall climber robots. But their applications are highly limited to a specific purpose.
The robots based on electromagnetic forces can climb only on ferromagnetic surfaces
[5]. Robots using a vacuum suction mechanism fail to develop sufficient adhesion
on dusty and rough walls [2–4, 10–12]. Another grasping-based bio-inspired robot
require the design of a complex mechanism and controlled gait for accurate foot
placement [10–12]. Hence, a propulsion-based gimbal mechanism is used where
required normal reaction for frictional force is provided by the thrust [13]. The
choice of two antiparallel propellers instead of one nullifies angular moment of the
total system. During motion on the wall, the required magnitude and the direction
of the thrust can be calculated. Such magnitude and direction can be adjusted by
varying angular velocity and gimbal angles vice versa. The outer spherical shape
allows horizontal to vertical surface alteration without the change in morphology.
2 Mechanical Design
The basic driving force of the proposed bot is derived from the propeller thruster. The
thruster is consists of the coaxial 1045(10 × 4.5”) ABS propellers (1CW + 1CCW
1 pair Black) and a couple of A2212 2200KV BLDC motors. On rotating, CW and
CCW propeller pair in an antiparallel direction results in the addition of individual
propeller thrust and nullifies angular moment of the system. A2212 2200KV BLDC
motors are placed adjacent to each other rather than placing coaxially, which reduces
the distance between propellers and hence further dimensions of the model by the re-
duced size of the gimbal. Coaxial propeller axes are achieved by the gear mechanism
with gear ratio 1:1.
Mechanical Design of Omnidirectional Spherical Wall … 19
2-DOF gimbal mechanism is used to control the direction of the thrust produced
by the propeller mechanism. Two couple of V0150 servo motors are placed in an
arrangement of perpendicular axes of rotation. One pair of motors is placed on the
ring and it revolves propeller mechanism by an angle θin . Another pair turns the ring
consisting of motors and propellers by an angle θout (Fig. 3).
Gimbal mechanism is gripped in a supporting frame which has a base plate at the
bottom. Weights involving battery and the controllers are placed on the base plates to
lower the center of gravity of the system. This arrangement automatically steadies the
base plate and always keeps it horizontal with respect to ground. Due to gravitational
stabilization, horizontal position can be taken as a reference which further simplifies
the mathematics (Figs. 4 and 5).
An outer shell is kept concentric with the gimbal ring with the support of ball
bearings. Bearings provide free rotation of the spherical shell over a surface in contact.
Radial perforations are provided to the outer shell for the free circulation of air
required for the thrust. But residual surface from perforation causes obstruction to
the thrust of propellers. This problem can be solved by choosing the strong material
of the shell and increasing amount of perforation.
Fig. 1 Floor to wall transition mechanism comparison of a a multi-linked track robot [12] and b
OsWalT
20 Y. Phalak et al.
Fig. 5 a Support frame and base plate, b, c, d gravitational stabilization of base plate on horizontal,
inclined, and vertical surfaces, respectively
According to the requirement analysis, the hardware of the first prototype of OsWalT
is designed to support the functional requirements of the gimbal and thruster. The RC
unit is used to manual control of the gimbal and angular speed of the thruster. The
RC unit comprises FS-TH9X 2.4 GHz 9CH Upgrade Transmitter with FS-IA10B Re-
ceiver. 1045(10 × 4.5”) ABS propellers (1CW + 1CCW-1 pair Black) and a couple of
A2212 2200KV BLDC motors are used in the bi-propeller mechanism. Four V0150
servo motors are used to control gimbal mechanism. BLDC motors are controlled
by a pair of SimonK 30 A ESCs. Orange 8000 mAh 3 S 30C/60C Lithium polymer
battery Pack (LiPo) is used as main power source to the system. Components with a
5 V requirement such as receiver and servo motors are fed by 5 V voltage regulator
IC 7805. Chassis and other framework are 3D printed using the PLA+ material (Fig.
6).
4 Results
As a first step, the floor test was performed by manually controlling the prototype
as shown in the figure. The angles and the required upthrust were adjusted remotely
by the RC module. We found that the bot could traverse require trajectory in zero
inclined surfaces. As shown in Fig. 12, the obtained results that are by changing the
thrust and angles of the servo, we can achieve the required direction and can move
the bot likewise (Fig. 7).
To perform the inclination test, the bot was placed on a wooden platform. The in-
clination angle of the platform was increased from 0◦ until the bot slips from its
stable position. The servo angles and the upthrust were adjusted manually in order
to maintain bot in position. The critical angle is found to be 45◦ . Figure 13 states the
experimental setup and time-lapse images of motion tests on different inclinations
(Fig. 8).
Mechanical Design of Omnidirectional Spherical Wall … 23
5 Conclusions
This paper presented a novel design of a wall-climbing robot. This mechanism serves
great improvement in previously proposed models due to the most simplified surface
to surface transition mechanism. The spherical shape of the robot provides symmetry
to the robot which makes it robust. Propulsion-based driving principal makes bot
independent of surface parameters and allows it to work on any terrain.
Acknowledgements The research work of this project is supported and funded by IvLabs, Robotics
lab of Visvesvaraya National Institute of Technology, Nagpur, India.
References
1 Introduction
Vibration is nothing but an oscillatory motion in any body from its mean position [1].
Mechanical vibration can be clearly observed in every structure such as a vehicle,
machine and any other solid construction when subjected to cyclic loads. Vibration
generates when unbalanced forces are applied on any equilibrant or non-equilibrant
body. Noise and vibrations are most important parameters that are analysed in any
automobile design [2]. It is an un-utilized energy of the vehicle subjected to different
road and driving conditions [3–5], that reduces comfort level in ride and also increases
fuel consumption [6]. These vibrations not only affect a vehicle component as well
as influence chassis design but also show hazardous effect on human body parts,
i.e. hand joints, nervous system, etc. [7]. Longitudinal and transverse vibration in
vehicle generates internally due to the combustion and power transmission system
[8]. However, non-uniform road surfaces are again one of the major factors that result
in production of random vibration in a vehicle [9, 10].
Energy harvesting devices are created, nowadays, that intend to collect tiny
amount of energies and convert it into useful electrical energy [11]. Such devices are
playing a major role in increasing the efficiency of a vehicle and reducing the losses
of energy in motor-operated hybrid vehicles [12, 13]. Most of the energy harvest-
ing devices are working mainly on the concepts of Faraday’s law and piezoelectric
effect [14, 15]. In this work, an attempt is made to study the harvesting of vibration
energy and for utilizing it by converting it to other usable forms of energies. For
this purpose, a fabrication method is developed using Faraday’s concept. This device
is finally tested under experimental road conditions to evaluate the performance.
Section 2 focusses on fabrication of device and selection of mounting point, Sect. 3
gives the test results on actual device and finally important conclusions are presented
in Sect. 4.
The fabrication of the device to convert vibration energy to electric energy is carried
out using Faraday’s law. It states that whenever there is a change in magnetic flux
across a current-carrying conductor, an electromotive force is produced. The device is
made to be mounted on a front wheel of a two wheel vehicle parallel to the suspension
system.
The fabrication of mechanical assembly consists of two clamps which are made up
of steel and are used to attach the device on the fork of suspension. Attaching link of
L shape made of stainless steel is welded from both ends where one end is weld to
the circumference of a clamp and another end is weld to the end face of cylinder as
shown in Fig. 1. The main function of attaching link is to fix the cylinder parallel to
the fork of suspension from the bottom. The piston having one L shape end is weld
on the circumference of the clamp and another end is kept free to move inside the
cylinder of diameter of 5 cm and length 20 cm which is again made up of stainless
steel.
The electrical assembly consists of insulated copper wire of 36 gauge which is wound
on a thick cylinder of plastic in turns of 2400 turns having inner diameter 2.5 cm, outer
diameter 3.4 cm and length 18 cm as shown in Fig. 1. The important component in
Fabrication and Performance Analysis of a Device … 27
Fig. 1 The combined assembly of a mechanical system and an electrical system of the fabricated
device
The front wheel suspension system of a two-wheeler (test device) consists of two
telescopic forks that act as dampers c1 and c2 between sprung mass m1 and un-sprung
mass m2 as shown in Fig. 2a:
28 D. H. Bonde et al.
Fig. 2 a Schematic diagram of front wheel shock absorber of a two-wheeled vehicle, b schematic
diagram of mass-spring-damper system of back wheel shock absorber of a two-wheeler
Putting the values of ẋ1 = x3 and ẋ2 = x4 we get the following equations,
1
ẋ3 = − [(c1 + c2 )(x3 − x4 )] (5)
m1
Fabrication and Performance Analysis of a Device … 29
1
ẋ4 = − [(c1 + c2 )(x4 − x3 ) + k1 (x2 − y)] (6)
m2
Equations 5 and 6 give the acceleration of the chassis and wheel, respectively. By
using these equations, theoretical values of acceleration, velocity and displacement
can be calculated.
The back wheel of a two-wheeler (test device) consists of two external spring and
two dampers attached between sprung mass and un-sprung as shown in Fig. 2b. In
Fig. 2b, m1 represents sprung mass, m2 represents the un-sprung mass of vehicle,
c1 and c2 represent the damping coefficient of the suspension, k 1 and k 2 represent
the stiffness of suspension spring while k 3 represents the stiffness of tire, x 1 and x 2
represent vertical displacement of sprung mass and un-sprung mass, respectively,
and y(t) represents road excitation which is assumed to be sinusoidal.
By applying Newton’s second law, equation of motion for Fig. 2b can be
represented by,
For sprung mass:
Putting the values of ẋ1 = x3 and ẋ2 = x4 in the above equations we get,
1
ẋ3 = − [(k1 + k2 )(x1 − x2 ) + (c1 + c2 )(x3 − x4 )] (11)
m1
1
ẋ4 = − [(c1 + c2 )(x4 − x3 ) + (k1 + k2 )(x2 − x1 ) + k3 (x2 − y)] (12)
m2
Table 1 Mechanical
Parameters Value
parameters of front wheel of
the vehicle [values are taken Sprung mass, m1 90 kg
from vehicle’s manual of Un-sprung mass, m2 14 kg
Hero Honda Achiever]
Shock absorber damping coefficient c1, c2 6000 Ns/m2
Tire stiffness k 1 105,428 N/m
the values for the back wheel. All the mechanical parameters are found and calculated
for the two-wheeled vehicle (Hero Achiever).
Above equations are solved using MATLAB for the time range of 0–10 s in
the steps of 0.1 s. Figure 3 depicts the displacement versus time plot of both front
and back wheel. The displacements are plotted providing a unit step input to the
suspension system. The graph clearly shows that the systems represented by the
given set of equations have maximum displacement in the front wheel as compared
to that of back wheel. Also, as the suspension system is designed to maximize driver’s
comfort, it is evident that the vertical displacement of back wheel will always be less
than the front wheel.
Fig. 3 Plot of displacement generated in both the wheels (x 2 ) of two-wheeler drive with respect to
time
Fabrication and Performance Analysis of a Device … 31
Considering the above result, the device is designed to be mounted on front wheel
parallel to the suspension system.
3.1 Methodology
The fabricated device is attached on the selected mounting point of vehicle. The
device output is evaluated for stationary condition. In stationary condition, pumping
of front suspension is carried out by applying external manual load at certain fixed
amplitude and frequencies. The device is also inspected by running the vehicle on
road surface at different velocities.
In stationary condition, the device is tested for varying amplitude and frequency. The
terminal voltage is recorded for different amplitude. The variation in output voltage
for different amplitude of vibrations is shown in Fig. 4a. The graph plotted, i.e.
of voltage against amplitude, shows that the induced emf increases with amplitude
almost linearly. The graph of voltage against amplitude is showing the trend as
linear law and is represented by equation mentioned below:
Fig. 4 a Representation of terminal voltage against amplitude of vibration on the mounted device,
b representation of voltage from the vibration harvester for different frequency of vibration in the
front suspension of vehicle
32 D. H. Bonde et al.
where V is the terminal voltage of device and A is the amplitude of vibration produced
in the wheel.
Figure 4b gives the idea of variation in terminal voltage for the device at different
frequency collected from the road surface at a constant amplitude of 3.5 cm. Voltage
is showing the linear trend against the frequency of vibration and it is followed by
equation given below:
Here, F is the frequency of vibration from road surface and V is the terminal
voltage across device.
Figure 5 depicts the voltage variation caused due to different speed conditions. The
data was collected for constant speed of 20, 40, 60 kmph. The graph shows that there
is increment in the induced emf when vehicle is accelerated at different constant
speeds. The graph is showing the linear trend represented by the equation:
Here, S is the speed of the vehicle which is constant and V is the terminal voltage
of the device.
Fig. 5 Representation of
voltage at different constant
speeds during motion of
vehicle
Fabrication and Performance Analysis of a Device … 33
4 Conclusion
References
1. Thomson W (2018) Theory of vibration with applications. CRC Press, Boca Raton
2. Rao MD (2003) Recent applications of viscoelastic damping for noise control in automobiles
and commercial airplanes. J Sound Vib 262:457–474
3. Ibrahim IM, Crolla DA, Barton DC (1996) Effect of frame flexibility on the ride vibration of
trucks. Comput Struct 58:709–713
4. Lee TK, Byoung SK (2008) Vibration analysis of automobile tire due to bump impact. Appl
Acoust 69:473–478
5. Wei CF, Taghavifar H (2017) A novel approach to energy harvesting from vehicle suspension
system: half-vehicle model. Energy 134:279–288
6. Abdelkareem MAA, Lin X, Ali MKA, Elagouz A, Mi J, Guo S, Liu Y, Zuo L (2018) Vibration
energy harvesting in automotive suspension system: a detailed review. Appl Energy 229:672–
699
7. Mou R, Li H, Yiqiang J, Yong Z, Yongqiao W (2015) Study of automobile suspension system
vibration characteristics based on the adaptive control method. Int J Acoust Vibr 20:101–106
8. Dawange SV, Kadlag VL (2015) A review paper on vibration analysis of DI engine. Int J Sci
Res 4:759–761
9. Ahirrao NS, Bhosle SP, Nehete DV (2018) Dynamics and vibration measurements in engines.
Proc Manuf 20:434–439
10. Burdzik R (2013) Identification of vibration transfer to car-body from road roughness by driving
car. Vibroeng Proc 1:27–30
11. Mydeen A, Kelwin I, Venkatesh M, Suthesh C (2017) Hybrid power harvester using engine
source. Int J Smart Sens Intell Syst 10:198–211
12. Wang W, Junyi C, Zhang N, Lin J, Liao WH (2017) Magnetic-spring based energy harvesting
from human motions: design, modeling and experiments. Energy Convers Manag 132:189–197
13. El-Hami M, Glynne-Jones P, White NM, Hill M, Beeby S, James E, Brown AD, Ross JN
(2001) Design and fabrication of a new vibration-based electromechanical power generator.
Sens Actuators, A 92:335–342
14. Glynne-Jones P, Tudor MJ, Beeby S, White NM (2004) An electromagnetic, vibration-powered
generator for intelligent sensor systems. Sens Actuators, A 110:344–349
15. Siddique AR, Mahmud S, Heyst BV (2015) A comprehensive review on vibration based micro
power generators using electromagnetic and piezoelectric transducer mechanisms. Energy
Convers Manag 106:728–747
16. Tan WH, Cheah JX, Lam CK, Lim EA, Chuah HG, Khor CY (2017) Vibration analysis on
compact car shock absorber. J Phys Conf Ser 908:012025
Robust Backstepping Controller
for an Omniwheeled Mobile Robot with
Uncertainties and External Disturbances
Abstract The application and implementation of portable mobile robots are increas-
ing in industrial and warehouse environments. A robot that can efficiently maneuver
in confined spaces is perfectly suited for such a task. One such variety is omniwheeled
mobile robots. In the paper, we present the design of a robust control policy for the
accurate trajectory tracking of four-wheeled mecanum mobile robot in the presence
of external uncertainties and disturbances. In the beginning, the kinematical and the
dynamical modeling of the mobile robot are performed. Afterwards, a robust control
policy is designed and implemented. The efficacy of the proposed controller was
tested on a reference trajectory. Simulation results prove that the proposed controller
tracks trajectory with greater accuracy than other standard controllers.
1 Introduction
Mobile robot platforms with rollers on the wheel’s periphery constitute a part in the
class of omnidirectional mobile robots. The rollers on the wheels leverage an edge to
these platforms as they can perform translatory and rotatory motions independently
and synchronously. These robots can steer and maneuver in very confined spaces and
hence are well suited for warehouses and factories and workshop floors. One kind of
omnidirectional mobile robot is four-wheeled mecanum mobile robot (FWMMR).
The wheels of these platforms have rollers aligned at an angle of 45◦ with the axis
of the wheel, thus helping it slide in the lateral direction.
The kinematic and dynamic modeling of the FWMMR is challenging because
of four independent motors controlling the motion. Kinematic [1] and dynamic [2]
modeling of mecanum-wheeled mobile robots have been done before [3]; however,
the majority of them were torque-based approaches. Here, the modeling is done by
considering voltage input to motors. In real world, uncertainty and disturbance will
creep in, so its effects have been added to the system. All these factors are added in
dynamic equation derived by Newton–Euler approach [4].
The system is having nonlinear dynamics, hence the linear controllers like PID,
cannot achieve accurate trajectory tracking. In the situation of bounded but unknown
uncertainties, backstepping controller, which is a nonlinear controller, can effectively
trace the trajectory. However, it has one disadvantage that it is not robust enough to
track the trajectory to a good precision in presence of higher disturbances. This
disadvantage is taken care of by adding a switching control law, which effectively
and quickly converges the system to the required trajectory. Hence, the proposed
controller is a robust backstepping (RBS) controller.
In the paper, the above methodology is proposed which is verified against tuned
PID controller. The main difference between this work and the previous works [5]
done in this domain are (1) The dynamical equation is adjusted for disturbances.
(2) The backstepping controller with and without robustness is compared to PID
controller in situations having varying uncertainty. (3) Control input voltage approach
helps include motor dynamics into consideration. (4) Controller is designed with
Lyapunov stability guaranteeing asymptotic stability.
The paper has been organized as, Sect. 2 discusses the kinematic modeling of
the FWMMR. The dynamic equation is derived in Sect. 3. The proposed controller
design is explained in Sect. 4. Simulation results and comparison of backstepping
controller with and without robustness with PID is done in Sect. 5. The paper ends
in Sect. 6 with the conclusion.
2 Kinematic Modeling
The top view of the FWMMR is shown in Fig. 1. For the system analysis, a total of
six frames are considered. Owi X wi Ywi are the frames assigned to the wheels, where
i represent the wheel number. The frame of motion of robot is Om X m Ym Z m and the
world frame is On X n Yn Z n . The angle of roller is represented by γi for the i th wheel,
which is +45◦ for first and third wheels and −45◦ for the wheels second and fourth.
The position vector is represented as Pm = [xm ym φm ]T and it shows the position of
the robot in the Om X m Ym Z m frame, where the positions of the mobile robot is given
as xm and ym and φm is the rotational orientation of the mobile robot with the X m
Robust Backstepping Controller for an Omniwheeled Mobile Robot … 37
axis. The value of velocity vector Ṗm given as [ẋm ẏm φ̇m ] in terms of the wheel’s
velocity, i.e., θ˙i and wheel’s radius R is.
⎡ ⎤
⎡ ⎤ ⎡ ⎤ θ̇1
ẋm −1 1 −1 1 ⎢θ̇2 ⎥
⎣ ẏm ⎦ = R ⎣ 1 1 1 1 ⎦⎢ ⎥ (1)
4 1 −1 −1 1
⎣θ̇3 ⎦
φ̇m d1 +d2 d1 +d2 d1 +d2 d1 +d2
θ̇4
3 Dynamic Modeling
The free-body diagram of all the forces acting on the system is shown in Fig. 2. The
driving forces due to actuator of the i th wheel in X- and Y-directions acting on the
system is given as Fxi and Fyi , respectively. The net torque on the entire platform by
all the external forces is represented by τ . The external force acting on the system is
shown by Fe at an angle of δ with Ym at a distance of b from the front wheel’s axis.
Applying the Newton’s second law of translation and rotation, i.e., F = ma , and
τ = I α in the robot frame Om X m Ym Z m and using rotation transformation matrix,
38 Z. U. Islam et al.
we have P˙n = Rmn (φ)P˙m and Fn = Rmn (φ)Fm . Upon solving and rearranging [6], we
have
where ⎡ ⎤
1 (β S(2φ) ẏ − β S(2φ) ẏ + 2β ẋ + 2C 2 (φ)(β − β ) ẋ + 4K ẋ
x n y n y n x y n 1 n
⎢ 2M
1 (−β S(2φ) ẋ + β S(2φ) ẋ + 2β ẏ + 2C 2 (φ)(β − β ) ẏ + 4K ẏ ⎥
f (ẋ) = ⎣ 2M x n y n x n x y n 1 n ⎦
1 2
⎡ 2In (−2(d1 + d2 ) K 1 − βz )φ̇ ⎤
− M1 (Sφ + Cφ) M1 (−Sφ + Cφ) − M1 (Sφ + Cφ) M1 (−Sφ + Cφ)
K2 ⎣ 1
g(x) = √ M
(−Sφ + Cφ) M1 (Sφ + Cφ) M1 (−Sφ + Cφ) M1 (Sφ + Cφ) ⎦
2 1
(d1 + d2 ) − I1n (d1 + d2 ) − I1n (d1 + d2 ) 1
(d1 + d2 )
⎡ I1n ⎤ In
2M
(2Fe S(φ)S(δ) + 2Fe C(φ)C(δ))
h(Fe , φ, δ) = ⎣ 2M 1
(−2Fe S(φ)C(δ) + 2Fe C(φ)S(δ)) ⎦ ξ(t) = [ξx ξ y ξφ ]T
1
2In
(d1 Fe C(δ) − d2 Fe S(δ) + bFe S(δ))
where C(φ) = cos(φ), S(φ) = sin(φ), C(δ) = cos(δ), and S(δ) = sin(δ) and K 1
and K 2 are motor coefficients in the driving force by motor which is given as Fdr =
K 1 R θ̇ + K 2 u [7] and ξ(t) is the uncertainty in environment.
4 Controller Design
The FWMMR system is a multiple input multiple output (MIMO) system, and here,
the purpose of the controller is to track the trajectory accurately and with least
deviation in presence of external uncertainty and disturbances by optimizing and
controlling the various inputs. In backstepping control design [8], a system is first
broken down into simpler subsystems and inputs of each subsystem is taken as virtual
Robust Backstepping Controller for an Omniwheeled Mobile Robot … 39
input. Finally, all the virtual control laws are combined by stepping back to generate
the final control law, and the stability is guaranteed by Lyapunov function. Let the
position vector be defined by x1 = [xn yn φn ]T and velocity vector by x2 = [ẋn ẏn
φ̇n ]T and control inputs as u(t) = [u 1 u 2 u 3 u 4 ]T . The two subsystems are given
as
ẋ1 = x2 (4)
here, x2 is taken as the virtual control input. The errors are defined as e1 = [x1 −
x1d ] = [xn − xnd yn − ynd φn − φdn ]T and e2 = [x2 − x2d ] = [ẋn − ẋnd ẏn − ẏnd φ̇n −
φ̇dn ]T . Here, x1d and x2d represent desirable position and velocity, respectively. Taking
the Lyapunov function as V (e1 , e2 ) = 21 [e1 ][e1 ]T + 21 [e2 ][e2 ]T and the derivative is
V̇ (e1 , e2 ) = [ė1 ][e1 ]T + [ė2 ][e2 ]T . To have, V̇ (e1 , e2 ) ≤ 0, we take [ė1 ] = −K a [e1 ]
and [ė2 ] = −K b [e2 ] where K a and K b are 3 × 3 positive definite matrix. Hence, we
have [ẋ1 ] = [ẋ1d ] − [K a ][e1 ] and [ė2 ] = [ẋ2 ] − [ẍ1d ] − [K a ]2 [e1 ]. Finally the back-
stepping controller equation is given as in Eq. (6) and its switching function is given
as in Eq. (7)
where, Sign(x) = 1, ifx ≥ 0 and Sign(x) = −1, ifx < 0. Here, ξ(t)max ≤ |K c | and
λ is 3 × 3 matrix. The robust backstepping controller is finally given as
5 Simulation Results
and the uncertainty is given as ξx (t) = 15 sin(t) ξ y (t) = 17 sin(t) and ξφ (t) =
13 sin(t) and Fe = 10 N for 4 ≤ t ≤ 6 and δ = 1rad for 4 ≤ t ≤ 6 The parameters
of the RBS controllers are obtained after over 200 iterations of parameter estimation
toolbox in MATLAB R2016a.
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
64.06 0 0 128.08 0 0 5.34 0 0
K a = ⎣ 0 70.29 0 ⎦ K b = ⎣ 0 140.04 0 ⎦ λ = ⎣ 0 17.14 0 ⎦
0 0 60.04 0 0 140.63 0 0 2.6
6 Conclusion
In this paper, a robust backstepping controller for FWMMR is developed. The robot
platform is given external disturbances and nonlinear uncertainties. Using the back-
stepping technique of integrating all the virtual inputs of the subsystem, the robust
backstepping control law is developed, and stability is verified by Lyapunov method.
Lastly, simulations were performed to check the efficacy of the proposed controller
on a reference path. The proposed controller outperforms PID in terms of trajectory
tracking, error magnitude, and practical usability.
42 Z. U. Islam et al.
References
1. Taheri H, Bing Q, Nurallah G (2015) Kinematic model of a four Mecanum wheeled mobile
robot. Int J Comput Appl (0975-8887) 113(3)
2. Williams II et al (2002) Dynamic model with slip for wheeled omni-directional robots. IEEE
Trans Robot Autom
3. Huang L, Lim YS, Li D (2004) Teoh CEL. Design and analysis of a four-wheel omnidirectional
mobile robot. In: Proceedings of second international conference on autonomous robots and
agents, December 2004. Palmerston North, New Zealand, pp 425–428
4. Mittal RK, Nagrath IJ (2015) Robotics and control, Tata Macgraw Hill Publication
5. Hamid RK (2014) Adaptive tracking with external force disturbance rejection for uncertain
robotic systems. Int J Control Autom Syst 12(1):169–176
6. Alakshendra V, Chiddarwar S (2016) Adaptive robust control of Mecanum wheeled mobile robot
with uncertainties. In: Non linear dynamics, Springer
7. Viet TD, Doan PT, Hung N, Kim HK, Kim SB (2012) Tracking control of a three-wheeled
omnidirectional mobile manipulator system with disturbance and friction. J Mech Sci Technol
26(7):2197–2211
8. Behera L, Kar I (2010) Intelligent systems and control: principles and applications, 2nd ed.
Oxford University Press
Micro-mechanical Analyses of Particle
Reinforced ex situ Bulk Metallic Glass
Matrix Composites
1 Introduction
Ex situ bulk metallic glass matrix composites (BMGCs) are broadly classified as
particle or fiber-reinforced BMGCs. Choi Yim and co-workers [1] have conducted
experiments on uniaxial compression and tension behavior of niobium (Nb) particle
reinforced Vitreloy 106 BMGCs, having particle volume fraction of 50%. The Nb
particle reinforced BMGCs show a plastic strain of up to 13% under uniaxial com-
pression and 2.21% under uniaxial tension. In the present work, finite element (FE)
analyses of a three-dimensional (3D) multi-particle model and a unit cell model of
a particle reinforced ex situ BMGC having Nb particle volume fraction of 50% is
carried out and it shows a very good agreement with the experiments of Choi Yim
et al. The unit cell modeling approach, which is validated by the experiments and
the multi-particle simulations, can further be used to study the behavior of several
types of ex situ BMGCs depending on the reinforcement shape, volume fraction and
hardening behavior.
2 Three-Dimensional Models
Fig. 1 Finite element mesh of the multi-particle simulation model having particle reinforcement
volume fraction f = 50% showing, a the matrix material with nomenclature of the node sets and
b the spherical particles
Micro-mechanical Analyses of Particle Reinforced … 45
are restricted in the Y direction. One of the corner nodes at the base is fixed in X,
Y and Z directions to prevent rigid body motion of the model. Uniaxial compres-
sion and tension loads are applied by displacing the ‘top’ nodes in the Y direction
at a quasi-static strain rate of 4 × 10−4 s−1 , corresponding to the experiments of
Choi-Yim [1].
The 3D FE simulations using unit cell method for ex situ BMGCs are developed to
study the uniaxial tension and compression response of ex situ BMGCs. A triply peri-
odic array of spherical particles is assumed to be spaced uniformly in the amorphous
matrix. Owing to this assumption, a single unit cell surrounding a reinforcement is
sufficient for capturing the response of the entire composite. The geometry of the
simulation model developed for spherical particle BMGC having reinforcement vol-
ume fraction of 50%, is shown in Fig. 2. The model represents 1/4th of the entire unit
cell (exploiting symmetry in the geometry and isotropy in the material properties)
of the composite. Node sets named ‘right’, ‘top’ and ‘front’ are created at respec-
tive sides of the unit cell as shown in Fig. 2. A set of nodes named as ‘left’, ‘base’
and ‘back’ (which are opposite to the former three node sets) are given symmetry
boundary conditions in the X, Y and Z directions, respectively. ‘Right’ and ‘front’
nodes are tied to the ‘tie’ node such that X displacement of ‘right’ and ‘tie’ nodes
are the same; similarly, Z displacement of ‘front’ and ‘tie’ nodes are the same. These
boundary conditions ensure that the unit cell boundaries remain flat and free of shear
traction throughout the analysis. This ensures compatibility between adjacent unit
cells. Uniaxial compressive and tensile loads are applied to the model by displacing
the ‘top nodes’ in the Y direction.
In both the cases of multi-particle and unit cell analyses, the material of the
particles is Nb and that for the matrix is Vitreloy 106. The matrix materials follow the
Anand-Su [2] constitutive model. The constitutive equations have been reformulated
and implemented using an implicit numerical integration scheme by Tandaiya et al.
[3]. In order to model the behavior of Nb, an empirical formula suggested in the work
of Zamiri [4] using Hollowman equation σ = K εn is used. The values of parameters
in this equation are obtained by fitting the equation to the experimental stress–strain
curve of Nb [4]. The material properties for Vitreloy 106 BMG and Nb are taken
from literature [1].
Figure 3a, b show the nominal stress versus strain curves of Nb particle reinforced
Vitreloy 106 ex situ BMGC, having particle volume fraction 50% under uniaxial
compression and tension, respectively. These figures show the simulation results
obtained from the multi-particle and the unit cell analysis. A fairly good match
is obtained between both the simulation curves and between the experiments in
literature [1].
The salient points of the stress–strain curve are marked as A, B and C. Point A
corresponds to the initial yielding of the composite. Thus, the material exhibits an
elastic behavior till point A. The slope of the stress–strain curves of multi-particle
model and the experiments matches exactly until point A. This match with unit cell
Fig. 3 Nominal stress–strain curves of Nb particle reinforced Vitreloy 106 BMGC having f = 50%
under uniaxial, a compression and b tension. A comparison between the multi-particle and unit cell
analyses is made along with the experiments of Choi-Yim et al. [1]
Micro-mechanical Analyses of Particle Reinforced … 47
model is comparable with the rest of the two curves even though an exact match is
not obtained. Further loading imparts nonlinearity to both the curves obtained from
multi-particle and the unit cell analysis up to point B. A shift in the stress–strain
curve is observed at point B. The unit cell curves deviate from the multi-particle
curves beyond elastic limit. This phenomenon is bound to occur since unit cell
considers a uniform distribution of particles. Beyond point B, the multi-particle curve
shows strain hardening effect, whereas the unit cell curve tends to flatten similar to
experiment curves. Though the simulation curves slightly under-predict the nature of
the stress–strain curves, it can be noted that a qualitative match is obtained between
the three curves. At point C, the unit cell and multi-particle simulation models show
a representative failure phenomenon.
p
Figure 4a, b show the contour plots of ln λ1 in the matrix for unit cell and multi-
particle simulation models at point A under uniaxial compression. The contours
show that the plastic deformation begins to nucleate at the diametrical center of
the sphere where the particle contacts the matrix. As the loading increases, plastic
deformation spreads toward the corners of the sphere. At point B, the deformation
spreads toward the thin ligaments in the unit cell as seen from Fig. 4c. Similar
effects are observed in the multi-particle model where there are criss-cross shear
band patterns forming within the matrix as shown in Fig. 4d. On loading beyond
B, plastic deformation begins to localize and intensify within the shear bands. At
C, the maximum principal logarithmic plastic strain reaches 10% within the shear
bands. Figure 4e, f represents this point for unit cell and multi-particle simulation
models, respectively. This coincides with an overall strain of about 20% estimated as
the failure point in experiments. Thus, under uniaxial compression, for Nb particle
reinforced Vitreloy 106, the failure point is represented by the formation of shear
bands accumulating a maximum principal logarithmic plastic strain of 10%.
Figure 5 shows the evolution of contour plots of maximum principal logarithmic
p
plastic strain ln λ1 in the BMG matrix for the unit cell and multi-particle models
under uniaxial tension. The evolution of plastic strain at points A, B and C is similar
as seen in the case of compression. Up to point A, the matrix deforms elastically and
it is seen that matrix exhibits elastic deformation as shown in Fig. 5a, b, respectively.
Beyond A, as loading increases, plastic deformation nucleates at the center of the
sphere at the particle interface. At B, plastic deformation spreads around the spherical
interface as seen in the unit cell model (Fig. 5c) and it begins to take shape of criss-
cross shear bands in the multi-particle model (Fig. 5d). With further loading, the
plastic deformation begins to localize as shear bands. The analysis is stopped at the
experimental failure strain denoted by point C in the stress–strain curve (Fig. 5b).
It is observed that the shear bands form a criss-cross pattern and plastic strain has
localized within these shear bands at this failure point. The maximum strain in the
model is about 10%, though the plastic strain within the dominant shear band is
lower than the maximum value. The failure criterion that holds true under uniaxial
compression as mentioned previously does not apply to the case of uniaxial tension.
Thus, for a qualitative study presented in the later sections, it is observed that the
strain to failure under tension is over predicted.
48 S. Gouripriya and P. Tandaiya
p
Fig. 4 Contour plots of maximum principal logarithmic plastic strain ln λ1 in the matrix at points
A, B and C marked on Fig. 3a for a, c, e unit cell, and b, d, f multi-particle model, under uniaxial
compression
4 Conclusion
The presence of particles in BMG matrix has shown to improve compressive strain
to failure in experiments. By adopting a multi-particle model and comparing with
the unit cell model and experiments, the present work elucidates the mechanics
and mechanisms of deformation and failure of particle reinforced ex situ BMGCs
at every stage of their loading process. While monolithic BMGs are known to be
brittle, particle reinforced ex situ BMGCs can exhibit enhanced ductility suggesting
that these ex situ BMGCs can be effectively deployed in load bearing structural
applications.
Micro-mechanical Analyses of Particle Reinforced … 49
p
Fig. 5 Contour plots of maximum principal logarithmic plastic strain ln λ1 in the matrix at points
A, B and C marked on Fig. 3b for a, c, e unit cell, and b, d, f multi-particle model, under uniaxial
tension
References
1. Choi-Yim H, Conner RD, Szuecs F, Johnson WL (2002) Processing, microstructure and prop-
erties of ductile metal particulate reinforced Zr57 Nb5 Al10 Cu15.4 Ni12.6 bulk metallic glass
composites. Acta Mater 50(10):2737–2745
2. Anand L, Su C (2005) A theory for amorphous viscoplastic materials undergoing finite
deformations, with application to metallic glasses. J Mech Phys Solids 53(6):1362–1396
50 S. Gouripriya and P. Tandaiya
R. Agrawal (B)
Pillai College of Engineering, Navi Mumbai 410206, India
e-mail: richasagrawal7@gmail.com
R. Uddanwadiker · P. M. Padole
Visvesvaraya National Institute of Technology, South Ambazari Rd., Ambazari, Nagpur,
Maharashtra 440010, India
1 Introduction
Fatigue leads to almost ninety percent failure of structural components. These com-
ponents are designed safe for the static load conditions but fail much below the
design stress when subjected to cyclic loading conditions. Therefore, life estimation
for fatigue becomes necessary to avoid failure during working condition. There are
two types of fatigue failure. Fatigue life cycles more than 105 are termed as high-
cycle fatigue failure and less than 105 as failure due to low-cycle fatigue (LCF).
Under LCF failure, the life cycles are less due to high stress and strain conditions.
316 LN is the most favored material [1] for structural component in liquid metal
cooled fast breeder reactor (LMFBR) due to its better mechanical properties at high
temperature and resistance to corrosive environment. LCF lives of smooth speci-
mens and notched specimens have been predicted at room temperature and at 823 K.
Notches are incorporated to study multi-axial state of stress effect present in compo-
nents due to existence of flaws, welds, and geometrical discontinuities. In the present
study, fatigue life is first determined experimentally and then estimated through finite
element and fracture mechanics approach.
2 Strain-Life Behavior
Basquin [2] stated that stress life data can be modeled using a power relationship,
which is a straight line on a log-log plot. This examination is equivalent to elastic
material behavior in the strain-life approach. Coffin [3] and Manson [4] separately
established that plastic strain-life data can be modeled using a power relationship.
The strain-life equation could be formed by adding the elastic and plastic components
and is given by Eq. 1.
σ f b c
εt = 2N f + εf 2N f (1)
E
where
εt total cyclic strain amplitude
σ f regression intercept known as fatigue strength coefficient
Nf number of cycles to failure
b regression slope known as fatigue strength exponent
E modulus of elasticity
εf regression intercept known as fatigue ductility coefficient
c regression slope known as fatigue ductility exponent
Life Estimation of Circumferentially Notch Round Bars … 53
3.1 Experimentation
Local strain-life method is applied on the assumption that the life of the component
spent during crack nucleation and growth could be estimated with the help of labo-
ratory specimens subjected to similar cyclic conditions at the site of crack initiation.
Fatigue life of a component subject to cyclic loading can be determined using this
concept once the relation between the fatigue life and localized strain is known. This
strain-life relation is represented as a curve of fatigue life versus strain and is obtained
by performing strain-controlled fatigue tests on plain and polished specimens of the
material under study. The material under stress concentration due to notches present
in it may be under plastic deformation even if the component behaves elastically dur-
ing cyclic loads, and therefore, strain-controlled axial fatigue tests are recommended
in these situations.
Low-cycle fatigue tests are conducted on plain and circumferentially notched
specimens of 316 LN SS. These tests are performed at room temperature as shown in
Fig. 1 and at 823 K and the specimens are subjected to loading conditions resulting
in different strain amplitudes. The plain specimens were fabricated as per ASTM
standard E606. The notched specimens, as shown in Fig. 2, were of same dimensions
as that of plain specimens but with a circumferential circular notch at the center of
the gauge length.
1
NODAL SOLUTION
STEP=1 JAN 24 2018
SUB =7 23:02:03
TIME=1
EPTOEQV (AVG)
DMX =.150266
SMN =.948E-04
SMX =.022947
MN
MX
Y
Z X
The strain-life parameters for the material under study obtained by conducting
LCF test on plain specimens are tabulated in Table 1. LCF life can now be estimated
by substituting these values in Eq. 1 for the corresponding value of strain. For notched
specimens, the strain value will be the maximum strain at the stress concentrated area.
The most flexible and commonly used method for estimation of the local stress
and strain magnitudes for life prediction is the finite element method (FEM). In LCF
condition, as the material is subjected to plastic strain, nonlinear analysis is performed
to study LCF behavior of the material. In the present study, rate independent kinematic
and isotropic combined hardening nonlinear plastic model is used to calculate the
stress and strain values at the notch root using ANSYS. The model was proposed by
Chaboche [5, 6] and it uses an associated flow rule with the von Mises yield criterion.
The constants incorporated in the model are derived by curve fitting experimental
plastic strain and true stress. The bilinear stress versus strain curve describes the
bilinear isotropic hardening model used in the finite element analysis. Elastic modulus
of the material defines the initial slope of the curve. The user-specified tangent
modulus (E T ) defines the slope of the line showing the stress versus total strain
beyond the user-specified yield stress when plastic strain develops in the material.
Tables 2 and 3 show the life predicted from the maximum total strain obtained by the
FEA (as shown in Fig. 3) of the notched specimens at room temperature and at 823 K,
respectively. The values show that the predicted life by finite element approach is
approximately equal to that obtained by experimentation at room temperature but
varies significantly at 823 K.
Once K I is known, the J integral is obtained using the relation between J and K I
and given by Eq. 3 [7].
K I2
J= (3)
E
Rice [8] proposed a relation to obtain the notch tip strain value at condition when
the yield strain is less than the applied strain and was differently given for large and
small scale yielding. These relations are represented by Eqs. 4 and 5.
3 J
εmax ≈ εy + large scale yielding (4)
4 σ y rt
3 σ∞ 2 a
εmax ≈ εy 1 + π small scale yielding (5)
4 σy rt
where
εy yield strain,
σy yield stress,
rt root radius,
a crack length, and
σ∞ nominal stress
The LCF life is predicted on the basis of the obtained notch root strain and the
strain-life equation and is shown in Tables 2 and 3 for the notched specimen with
radius 2.5 mm at room temperature and 823 K, respectively. Small scale yielding
specified by Eq. 5 is considered for room temperature life estimation and large scale
yielding for life estimation at 823 K.
4 Conclusion
The various life estimation techniques explained in the present paper were applied
for life prediction of circumferentially notched specimens of 316 LN SS specimens.
Average ratio of the life predicted by FEA to the experimental life and that by EPFM
and experimental life is calculated to judge the level of error in life estimation meth-
ods. The life estimated by FE analysis is within a factor of 1.23 at room temperature,
Life Estimation of Circumferentially Notch Round Bars … 57
whereas it underestimated the LCF life at high temperature. This is due to the limi-
tation of completely incorporating the material behavioral changes occurring at high
temperature in FE analysis. On the other hand, the life estimation by EPFM approach
is within a factor of 1.14 at room temperature and at 823 K. Thus, EPFM approach can
be fairly applied for life estimation of notched specimens once J integral is obtained
for the geometry under consideration. The variation in the life estimation by EPFM
may be due to the use of the stress intensity factor equation available for V notch
on round specimen, whereas the specimen used is a round specimen with circular
notch.
References
1. Roy SC, Goyal S, Sandhya R, Ray SK (2012) Low cycle fatigue life prediction of 316L(N)
stainless steel based on cyclic elasto-plastic response. Nucl Eng Des 253:219–225
2. Coffin LF, Tavernelli JF (1959) The cyclic straining and fatigue of metals. Trans Am Inst Min
Metall Pet Eng 215:794–806
3. Manson SS (1966) Thermal stress and low-cycle fatigue. McGraw-Hill, New York, Gr, pp
125–192
4. Basquin OH (1910) The exponential law of endurance tests. Am Soc Test Mater Proc 10:625–630
5. Chaboche JL (1989) Constitutive equations for cyclic plasticity and cyclic viscoplasticity. Int J
Plast 5(3):247
6. Chaboche JL (1991) On some modifications of kinematic hardening to improve the description
of ratcheting effects. Int J Plast 7(7):661
7. Dieter G (1988) Mechanical metallurgy. McGraw-Hill, London
8. Rice JR, Rosengren GF (1968) Plane strain deformation near a crack tip in a power-law hardening
material. J Mech Phys Solid 16:1–12
Placement of Heated Blocks Under
Forced Convection for Enhanced Heat
Transfer
1 Introduction
Table 1 Thermophysical
Material K, W/m K ρ, kg/m3 Cp, J/(kg K)
properties of substrates at
T ∞ = 30 °C and 1 atm FR4 0.3 1900 1370
Bakelite 1.4 1300 1465
1500, 2000, and 2500 W/m2 , three velocities of 0.6, 1, 1.4 m/s and two substrate
boards of thermal conductivities of 0.3 and 1.4 W/m K. The five representative
configurations are 110, 115, 116, 145, and 189, respectively. Heat sources in the
bottom row are assigned numbers 1–5, in the middle row 6–10, and in the top row
11–15 associated with location numbers on the substrate board from 11 to 15 in the
first lower row, 21–25 in the middle row, and 31–35 in the top row, respectively.
The substrate boards used are glass/epoxy FR4 and bakelite of thickness 5 mm.
The fifteen aluminum heat sources are arranged on these substrate boards using five
representative configurations. The photographs of FR4 and bakelite substrate boards
mounted with aluminum heat sources arranged in configuration 110 and 115 are
shown in Fig. 1. Thermophysical properties of FR4 and bakelite are given in Table 1.
3 Experimental Set Up
Steady state experiments are conducted under forced convection cooling of fifteen
rectangular heated blocks of non-identical sizes, arranged in five different configu-
rations referred to above in a vertical channel. The characteristic length (Lh ) of the
heated blocks varies from 10 to 42 mm. The experiments are conducted for a wide
range of Reynolds numbers, thermal conductivities, and heat inputs. The temperature
62 S. Durgam et al.
distribution on the substrate board and in each heat source is obtained by conducting
experiments. The vertical channel test facility fabricated in-house is shown in Fig. 2.
The total number of 90 experiments are conducted (5 configurations × 3 velocities
× 3 heat fluxes × 2 substrate conductivities). The 90 cases studied experimentally are
simulated for the same ambient temperatures that exist in each case for the purpose
of comparison and validation. The detailed procedure of experimentation is given in
[13].
4 Numerical Results
The pre-processor generated extremely fine mesh is used for the simulation. In the
present study, free tetrahedral extremely fine mesh was used for all simulations.
Complete mesh consists of 160,404 domain elements, 25,866 boundary elements,
and 1564 edge elements, and the number of degrees of freedom solved for 117,898.
The Mesh size slightly greater or less than the mesh size used gives variation of
temperature of less than 1 °C. The multi-physics used is laminar forced conjugate
heat transfer (heat transfer in solids and the fluid) to solve the governing equations
[15]. The detailed procedure of grid sensitivity study, error analysis is followed as
in reference [12]. The famous Navier–Stokes equations for mass, momentum, and
energy for three-dimensional, steady flow in the fluid domain, the solid region energy
Placement of Heated Blocks Under Forced Convection … 63
equation, and the body force in vertical direction is given in [14–16] have been solved
using simulation software. Boundary conditions are: x = 0, T = T ∞, x = L, p =
p∞. Lateral boundaries are adiabatic. The top, bottom boundaries and the edges
of substrate and the vertical channel are assumed to be adiabatic. The numerical
results of the temperature plots show that the maximum excess temperatures of
configurations 110, 115, 116, 145, and 189 are 28.4, 24.4, 27.6, 22.7, and 30.8 °C,
respectively.
The lowest value of excess temperature, in this scenario, is found in configuration
145. Figure 3 shows the substrate temperature of FR4 board is higher than that
for bakelite. The substrate temperature values are the lowest in configuration 145
compared to all other four configurations. To compare the effects of substrate thermal
conductivity on maximum excess temperature, the configuration 145 is considered
as it is found to be the optimal one. The maximum substrate temperature observed
in FR4 is about 68 °C and that in case of bakelite is about 66 °C. The plums of heat
can be observed in the direction of airflow.
Figure 4 shows the numerical and experimental results of forced convection cooling
of heated blocks on FR4 ad bakelite substrate board using flow velocity of 0.6 m/s
and heat flux value of 1500 W/m2 for different configurations.
6 Conclusions
Steady state forced convection experimental and numerical results for fifteen rectan-
gular protruding discrete heat sources showed that maximum excess temperature in
64 S. Durgam et al.
configuration number 145 is lowest in all cases considered in the present study. The
maximum excess temperature found to be lowest at higher value of inlet air velocity
considering bakelite substrate board. High heat fluxes show higher maximum excess
temperatures. From this study, it is found that position or location of heated blocks
has a significant effect in heat dissipation. The configuration number 145 was found
to be the optimal configuration. Results show that excess temperatures in case of
bakelite are lower compared with FR4. From the results, it is concluded that maxi-
mum temperature of any configuration is a function of Reynolds number, substrate
thermal conductivity, heat flux, position, and size of heated blocks.
References
10. da Silva A, Lorenzini G, Bejan A (2005) Distribution of heat sources in a vertical open channels
with natural convection. Int J Heat Mass Transf 48:1462–1469
11. Hotta T, Balaji C, Venkateshan S (2014) Optimal distribution of discrete heat sources under
mixed convection—a heuristic approach. J Heat Transf 136:104503
12. Kumar G, Rao C (2011) Interaction of surface radiation with conjugate mixed convection from a
vertical plate with multiple non-identical discrete heat sources. Chem Eng Comm 198:692–710
13. Ghorab M (2014) Forced convection analysis of discrete heated porous convergent channel.
Heat Transf Eng 36(9):829–846
14. Durgam S, Venkateshan S, Sundararajan T (2017) Experimental and numerical investigations
on optimal distribution of heat source array under natural and forced convection in a horizontal
channel. Int J Thermal Sci 115:125–138
15. Durgam S, Venkateshan S, Sundararajan T (2018) A novel concept of discrete heat source array
with dummy components cooled by forced convection in a vertical channel. Appl Therm Eng
129:979–994
16. Patankar S (1980) Numerical heat transfer and fluid flow. McGraw-Hill, New York
Analysis of Track Vibration
for Metro Rail
Abstract Buildings located near surface Metro and surface trains are subjected
to mechanical vibrations induced by their motion. High-precision instruments in
laboratories nowadays are very much sensitive to external vibrations which affect
their performance. It is thus imperative to measure amount of vibration and noise
transmitted from the railway track to ground. In this paper, frequency content and
amplitude of vibration levels are measured on the metro rail track. Track vibrations
were measured at a point on track for three different positions of metro from that point,
while it was in motion. Amplitude and frequency content were found when (a) train
was some distance away from the point of observation, (b) train was passing above the
point and (c) train had left the point by some distance. This experiment was carried out
near Khapri metro station, Nagpur, Maharashtra, India. These experimental results
are studied and discussed in this paper. Further, these results can be compared with
limits given by Federal Transit Administration (FTA) to ensure human comfort and
safety of nearby structures.
1 Introduction
nectivity, time saving and indirectly reducing traffic on roads. Therefore, metro rail
is being constructed in Nagpur, which is the third largest city in Maharashtra, India,
with population of around 2.5 million. Nagpur metro design consists of 1435 mm
gauge and maximum permissible speed of 80 km/h. Maximum acceleration and max-
imum speed of train according to design are 1 m/s2 and 90 km/h respectively. Train
composition consists of three-car train (DMC-TC-DMC). DMC and TC stand for
driving motor car and trailer car, respectively. Due to ease of access and real estates
affordable prices, construction of buildings and other facilities are often pushed in the
vicinity of subway and surface trains. Surface trains and subways induced vibrations
in the nearby structures and buildings hamper the operation of equipment which are
highly sensitive to vibration.
Vibration induced to the ground by motion of metro acts as line source of vibration
and lies in the range of 10–400 Hz [1]. Hassan [1] studied and showed that vibrations
produced by subway and metro trains propagate through the ground in different ways.
There are three different ways in which the vibration propagates, (1) surface waves,
(2) compressive wave and (3) shear wave.
Adam and Estroff [2] studied and showed that vibration transmitted to the struc-
ture through foundation and the waves propagated close to soil surface are Rayleigh
waves. Anderson [3] measured vibration in two different buildings near the vicin-
ity of railway track which were subjected to vibration at the foundation induced
by motion of railway. He found that vibration that is perceptible and can propa-
gate through ground lies in 5–50 Hz frequency domain. Sanayei et al. [4, 5] studied
vibration propagated through ground in different areas of Boston, USA, which was
induced by train and subway motion. Transmission of vibration in four-storey build-
ing through the bottom floor was also reported. Such studies help designers to design
and eliminate vibrations in areas in the building which are highly sensitive to vibra-
tion.
This study aims to measure and analyse vibration on metro rail track. These
vibration from track propagate through ground and hence act as base excitation to
nearby buildings and overhead structures. Vibrations induced by train on track are
measured and analysed. These findings provide basis for base excitation, which is
necessary in designing of vibration mitigation models and hence minimising the
induced vibration.
2 Measurement Set-Up
The measurements were carried out with the help of OROS data acquisition and
vibration analyser. It consists of DAQ, accelerometer and a software. It was carried
out on track near Khapri metro station, Nagpur, Maharashtra, India. This portion of
metro rail is on ground which makes it easy to measure track vibration.
Figure 1 shows schematic diagram of arrangement of various components used
for measurement of track vibration. Accelerometer was placed on track to mea-
sure the vertical vibration of the track when the train is moving. The accelerometer
Analysis of Track Vibration for Metro Rail 69
is in contact with the metro track with the help of strong magnet attached to its
base. Position of the accelerometer was such that its axis was perpendicular to the
track surface which helps in getting vibration in vertical direction. Figure 1 shows
the schematic of the measurement set-up and Fig. 2 shows the actual measurement
set-up. As seen in Fig. 1 accelerometer is connected to DAQ system. For frequency
analysis, Fast Fourier Transform (FFT) was used. The measurements were taken for
5 s interval when the train was in motion. Data were acquired in three different con-
ditions: when the train is away and approaching the measurement point, train, when
the train is passing over the measurement point, and when the train has crossed the
measurement point. Figure 2 shows actual measurement set-up for track vibration.
The sampling frequency was 52.1 kHz and number of data points measured were
256,000.
Results were obtained through OROS software in terms of acceleration (m/s2 ) ver-
sus time (s) graph. Waveform shows track vibration amplitude with time. As the
observed/measured data are in time domain, it is necessary to convert it to frequency
domain for ease of analysis. This conversion to frequency domain helps us to pin-
point the maximum amplitude for its respective frequency, which is done using Fast
Fourier Transform (FFT) algorithm. Time domain and frequency domain signals for
three different positions of train with respect to measurement point are discussed
below.
70 C. V. Bhore et al.
Fig. 3 Spectrum of signal when the train is away and approaching the measurement point
Figure 3 shows spectrum of signal when the train is away and approaching the mea-
surement point.
When the train is away from the point of observation maximum amplitude of
acceleration is 525.0 (mm/s2 ) at a frequency of 1600 Hz. Other significant amplitudes
corresponding to their frequencies are shown in Table 1.
Analysis of Track Vibration for Metro Rail 71
Table 1 Frequency and amplitude of vibration when train is away and approaching the measure-
ment point
Frequency (Hz) Acceleration (mm/s2 ) Frequency (Hz) Acceleration (mm/s2 )
100 299.4 1600 525.0
1050 198.3 1800 229.4
1200 183.3 2000 160.2
Fig. 4 Spectrum of signal when the train is passing over the measurement point
Table 2 Frequency and amplitude of vibration when train is passing over the measurement point
Frequency Acceleration Frequency Acceleration Frequency Acceleration
(Hz) (mm/s2 ) (Hz) (mm/s2 ) (Hz) (mm/s2 )
200 663 2000 1860 5700 499
1150 1137 4050 861 7250 480
1600 1094 4800 516 8100 2117
Figure 4 shows the spectrum of signal obtained when the train is passing over the
measurement point.
When the train is passing over the observation point for period of 5 s, maximum
amplitude is 2117 (mm/s2 ) at a frequency of 8100 Hz. Other significant amplitudes
corresponding to their frequencies are shown in Table 2.
72 C. V. Bhore et al.
Fig. 5 Spectrum of signal when the train has crossed the measurement point
Table 3 Frequency and amplitude of vibration when train has crossed the measurement point
Frequency Acceleration Frequency Acceleration Frequency Acceleration
(Hz) (mm/s2 ) (Hz) (mm/s2 ) (Hz) (mm/s2 )
1750 131.4 6800 121.6 8700 67.3
2000 181.6 7250 235.1 11,050 185.3
4050 64.5 7650 363.5 11,450 120.5
5750 122.6 7950 390.5
Figure 5 shows the spectrum of signal obtained when the train has crossed the mea-
surement point.
When the train has crossed the measurement point, for period of 5 s, maximum
amplitude is 390.5 (mm/s2 ) at a frequency of 7950 Hz. Other significant amplitudes
corresponding to their frequencies are shown in Table 3.
4 Conclusion
trial period lies in the range of 25–30 km/h. Vibration induced due to train motion
to the ground in the adjacent to its track was measured and studied. Maximum
acceleration amplitudes corresponding to its frequencies were found out. This study
helps in providing base excitation for vibration mitigation system to design safe
and disturbance free structures. Further, these results may be compared with limits
provided by Federal Transit Administration (FTA) for human comfort and safety.
References
1. Hassan OAB (2006) Train-induced ground-borne vibration and noise in buildings. Multi-
Science
2. Adam M, Von Estorff O (2005) Reduction of train-induced building vibrations by using open
and filled trenches. Comput Struct 83(1):11–24
3. Anderson DC (1994) Engineering prediction of railway vibration transmitted in buildings.
Environ Eng 7(1)
4. Sanayei M, Maurya P, Moore JA (2013) Measurement of building foundation and ground-borne
vibrations due to surface trains and subways. Eng Struct 53:102–111
5. Ngamkhanong C, Kaewunruen S (2018) The effect of ground borne vibrations from high speed
train on overhead line equipment (OHLE) structure considering soil-structure interaction. Sci
Total Environ 627:934–941
Localization of a Four-Wheeled
Omnidirectional Mobile Robot Using
Sensor Data: A Kalman Filter Approach
1 Introduction
Mecanum-wheeled mobile robots are widely used in many applications for its special
wheel structure allows for lateral movement. The rotation of the rollers around the
mecanum wheel will result in slippage of the robot, and it is difficult to model [1].
Accurate localization is very crucial for navigation of autonomous mobile robot in
any environment. The gap between two adjacent rollers causes the periodic vibra-
tion of the moving platform. Due to slippage, vibration and noise in the sensor, an
accumulation of error can exist using vision-based tracking system. The problem of
estimation is widely solved by Kalman filter method which was invented in 1960 by
Kalman [2].
Kalman filter reduces the error in least square sense from the measurement from
different sensors. Among many applications, the Kalman filter is the most essential
part of a robotic vision system. It uses the vision measurements that contain noise
and uncertainty by the visual sensors over time and produce value that is close to
the true measurement value of the robot (self-localization) [3, 4] or the target (object
detection) [4]. Object tracking is used to detect the motion of an object by accruing
input from a series of images. The vision system is used to track 3D object [5], people
[6] and some planner contour [7] by vision sensor. In the present work, a Kalman
filter is developed for a four-wheeled omnidirectional mecanum-wheeled mobile
robot. The proposed algorithm is tested for the different level of sensor noise data.
The effect of sensor noise on trajectory tracking and control command is illustrated.
In this part, referring to [8], the dynamic model of the four-wheeled mecanum robot is
introduced. The robot consists of four independently actuated mecanum wheel over
a robotic platform. Figure 1 shows a four-wheeled mecanum robot. The dynamic
model of the robot is given as (Eqs. 1–4).
Fig. 1 Four-wheeled
mecanum robot
Localization of a Four-Wheeled Omnidirectional Mobile … 77
3 Flatness-Based Controller
The dynamic model of the omnidirectional mobile robot is differentially flat, and a
set of flat output is defined [9]. Based on these flat outputs, the differential flatness
controller is defined as
⎡ ⎤ ⎡ d⎤ ⎡ ⎤⎡ ⎤
ÿ1 ÿ1 k p1 0 0 e y1
⎣ ÿ2 ⎦ = ⎣ ÿ d ⎦ + ⎣ 0 k p2 0 ⎦⎣ e y2 ⎦
2
ÿ3 ÿ3d 0 0 k p3 e y3
⎡ ⎤⎡ ⎤ ⎡ ⎤⎡ ⎤
ki1 0 0 e y1 dt kd1 0 0 ė y1
+ ⎣ 0 ki2 0 ⎦⎣ e y2 dt ⎦ + ⎣ 0 kd2 0 ⎦⎣ ė y2 ⎦ (5)
0 0 ki3 e y3 dt 0 0 kd3 ė y3
⎡ ⎤
e y1
where k p 3×3 , [ki ]3×3 , [kd ]3×3 are the controller gain parameters and ⎣ e y2 ⎦ =
e y3
⎡ d ⎤
y1 − y1
⎣ y d − y2 ⎦ is the error vector. y1 y2 y3 T = x y θ T are the flat output of the
2
y3d − y3
differentially flat system. The proposed differential flatness controller is used for
trajectory tracking of the robot.
In real systems, trajectory tracking to obtain the desired variables for the feedback
loop relies on sensors data. Sensors are prone to noise due to their internal structure
and inherent environmental conditions, and their data cannot be used directly for
measurement. The Kalman filter is a multi-input multi-output filter for optimal real-
time state estimation in systems with noisy measurements. As the filter is recursive,
it does not require all the previous sensor data to be maintained, thus simplifying
its hardware implementation. The Kalman filter can estimate current and upcoming
78 S. R. Sahoo et al.
states even when accurate models and sensor data are unavailable. In this study, we
employed the Kalman filter to timely estimate the state for feedback control by linear
difference equations of the omnidirectional mobile robot.
The vision system based on the Kinect sensor (Microsoft Corp., Redmond, WA,
USA) can determine the position of the robot from its centre of gravity using camera
data. The system equation with process noise and sensor noise can be modelled as
ẋ = Ax + Bu + w, ẏ = C x + Du + v
where w and v are the process and measurement noise following zero-mean Gaussian
distributions, respectively. Based on this model, a linear stochastic system is defined
for the implementation of the Kalman filter as
x̂t = At x̂t−1 + Bt u t + wt ,
yt = Ct xt + Dt u t + vt , (6)
where
Equation (7) provides the information of the predicted state in the next time step
provided that the current system state is available. Hence, this model can be used in
the prediction step of the Kalman filter.
The Kalman filter consists of the three processes illustrated in Fig. 2. The filter
includes a prediction or prior estimation of the state, the use of measurement data
and calculation of Kalman gain and the estimation update or posterior estimation.
Estimation is the first step of the algorithm. It provides the evolution of the system
state matrix (x̂t− ) and the process covariance matrix (Pt− ) over time if the state (x̂t−1 ) at
the previous time step is known. This is the prior estimation (x̂t−1 ) of the state. Then,
the vision system measurements are considered, and the Kalman gain calculated.
Specifically, the measurement data are collected from the sensor (yt ) and, along with
the measurement covariance matrix, the Kalman gain (kt ) is calculated. Finally, the
state is updated, and the process covariance matrix calculated. The updated value of
the system state (x̂t ) and the process covariance matrix (Pt ) are calculated based on
the Kalman gain (K t ).
The algorithm implementing the proposed Kalman filter for the omnidirectional
mobile robot is detailed as follows:
1. Algorithm Kalman Filter (x̂ t−1 , u t , Pt−1 , yt ):
2. Prediction:
3. x̂t− = A x̂ t−1 + B u t + wt : Expected state x
4. Pt− = A Pt−1 A T + Q k : Expected covariance P
5. Correction: −
6. K t = H PP−t HH+R = (H Pt− H + R)−1 Pt− H : Calculation of Kalman gain
t
7. x̂t = x̂t− + K t (yt − H x̂t− ): Correction of state x
8. Pt = (1 − K t )Pt− : Correction of covariance P
9. Return (x̂t Pt ): Return corrected x and P
10. Go to step 3
5 Simulation Results
The proposed dynamic model, control law and Kalman filter described in Sects. 2–
4 were implemented for simulation using MATLAB Simulink (Mathworks, Inc.,
Natick, MA, USA). To test the effectiveness of the proposed method, we added
various levels of noise to the feedback signals during simulations.
(a) x-axis robot position (b) y-axis robot position (c) Tracked Trajectory
Fig. 3 Estimated states during circular trajectory under low noise level
(a) x-axis robot position (b) y-axis robot position (c) Tracked Trajectory
Fig. 4 Estimated states during circular trajectory under high noise level
6 Conclusion
We use the Kalman filter to fuse position information from a vision sensor and
encoders and embed this estimation into a flatness-based controller. This control
method was implemented for trajectory tracking along prescribed trajectories on
an omnidirectional robot. We observed that the substantial amount of noise in the
measured signals coming from the Kinect sensor causes high fluctuation of the control
input. Such input may result in catastrophic damage to the actuators. However, state
estimation using the Kalman can efficiently mitigate noise. This proposed method
can be used for simultaneous localization and mapping of the robot.
References
1. Pin FG, Killough SM (1994) A new family of omnidirectional and holonomic wheeled platforms
for mobile robots. IEEE Trans Robot Autom 10:480–489
2. Kalman RE (1960) A new approach to linear filtering and prediction problems. J Basic Eng
82:35–45
82 S. R. Sahoo et al.
3. Bishop G, Welch G (2001) An introduction to the Kalman filter. Proc of SIGGRAPH, Course.
8:59
4. Correa J, Soto A (2010) Active visual perception for mobile robot localization. J Intell Rob Syst
58:339–354
5. Teslić L, Škrjanc I, Klančar G (2011) EKF-based localization of a wheeled mobile robot in
structured environments. J Intell Rob Syst 62:187–203
6. Weinstein AJ, Moore KL (2010) Pose estimation of Ackerman steering vehicles for outdoors
autonomous navigation
7. Rae A, Basir O (2009) Reducing multipath effects in vehicle localization by fusing GPS with
machine vision
8. Sahoo SR, Chiddarwar SS, Alakshendra V (2017) Intuitive dynamic modeling and flatness-based
nonlinear control of a mobile robot. Simulation 0037549717741192
9. Sahoo SR, Chiddarwar SS (2019) Flatness-based control scheme for hardware-in-the-loop
simulations of omnidirectional mobile robot. Simulation 0037549719859064
Capacitated Vehicle Routing Problem
with Interval Type-2 Fuzzy Demands
Keywords Vehicle routing problem · Interval type-2 fuzzy numbers · Clark and
Wright algorithm
1 Introduction
The very first work regarding the vehicle routing problem was done in 1959,
approximately 60 years before, by Dantzig and Ramser [3] who introduced the
vehicle routing problem, then called as truck dispatching problem as a real-world
application concerning the delivery of gasoline to gas stations. In the paper pub-
lished in 1959 by Dantzig and Ramser, the authors proposed the first mathemat-
ical programming formulation and algorithmic approach for the vehicle routing
problem. In 1964, Clark and Wright proposed [1] an effective greedy heuristic
for the approximate solution of the vehicle routing problem. If the transporta-
tion tasks are to be performed at the vertices of the network, the corresponding
problem is known as node routing problem [7], and if the tasks are to be per-
formed on street segments, then the corresponding routing problem is known as
arc routing problem [2]. The imprecision may occur in various forms, e.g. imprecise
demands of customers. This paper deals with the imprecise demands of the customers.
The capacity of the vehicle is crisply defined, but the demands of the customers are
defined in the form of an interval type-2 triangular fuzzy number. Inspired by [5, 6],
we have considered CVRP with IT2FN. The vehicle routing problem is a NP-hard
problem, and introduction of uncertainty gives a new variation of the problem. Since
VRP itself is a NP-hard problem, it is not surprising that the very first numerical
methods for solving VRPSD consisted of heuristics. These relatively simple tech-
niques based on savings were later complemented by more efficient algorithms such
as the Tabu search. This paper is structured in the following manner: in Sect. 2 of
this paper, basic preliminaries of fuzzy set theory have been reviewed. In Sect. 3, a
mathematical model of CVRP with IT2FN has been presented. Sections 4 and 5 deal
with the algorithm of the proposed model and the procedure of the method used. In
Sect. 6, an example has been presented with seven customers and a depot node where
the demands of the customers are given by interval type-2 triangular fuzzy number.
The last section comprises the concluding remarks.
Definition 1 Uncertainty in the primary membership of a type-2 fuzzy set Ø consists
of a bounded region called the footprint of uncertainty, that is, FOU = ∪x∈X Ix [4].
The footprint of uncertainty of Ø is bounded below by a lower membership func-
tion (LMF), usually denoted by μ Ã , and is bounded above by an upper membership
function (UMF), usually denoted by μ Ã [4].
Definition 2 The interval type-2 fuzzy set is called the interval type-2 fuzzy number
[4] when its UMF and LMF are type-1 fuzzy numbers.
Definition 3 The core of the interval type-2 fuzzy number Ø = [A, A] is Cor e( Ã)
˜ =
[min Cor e(A), max Cor e(A)] where φ = Cor e(A) ⊆ Cor e(A) when φ is an empty
set and μ and μ are the lower and upper membership functions of Ã, ˜
à Ã
respectively [4].
˜ is defined on the interval [a, c], and
Definition 4 Interval type-2 fuzzy number, Ã,
its lower membership function takes the value equal to h ∈ [0, 1] at b and h ∈ [0, 1]
at b, respectively, where
a ≤ a ≤ b = b ≤ c ≤ c.
Capacitated Vehicle Routing Problem with Interval … 85
3 Mathematical Model
such that αi p λ p = 1, ∀i ∈ N (2)
p∈P
λ p ≥ ϑ(N ) (3)
p∈P
λ p ∈ {0, 1} ∀ p in P (4)
4 Algorithm
1: Start.
2: Input C[n][n] and D[n]
3: i ← 1
4: for i ≤ n do
5: crdemand[i] ← defuzzy value of demand using PGMIR expression.
6: end for
7: i ← 1
8: j ← i + 1
9: for i ≤ n do
10: for j ≤ n do
11: s[i][j] ← c[0][i]+c[0][j]-c[i][j]
12: end for
13: end for
14: i ← 1
15: for i ≤ n do
16: ri ← (0 → i → 0)
17: end for
18: m[k] ← Sorted(s[i][j]). %Sorted(s[i][j]) here means sorted in descending order.
19: k ← 0
20: c ← 0 %Choose ri and r j corresponding to first entry of m[k].
21: while m = φ do
22: if qri + qr j ≤ Q and ri = r j and (i = ri [1] or j = r j [1]) then
23: qri ← qri + qr j
24: ri ← (0 → ri → r j → 0)
25: m[k] ← Sorted(s[i][j]-m[k])
26: c ← c + cri + cr j
27: end if
28: end while
29: Return c.
5 Example
To illustrate the given model, let us suppose an instance where there are seven dif-
ferent customers having their demands in the form of an interval type-2 triangular
fuzzy numbers. The network corresponding to the customers and distributor is given
in Fig. 2. The vertex node 0 represents the depot node. The adjacency matrix storing
the cost incurred in travelling any two vertices is given by C. The demand data of all
the customers is given in Table 1.
According to the algorithm, the first task we need to do is to defuzzify the
demands using parametric graded mean integration representation method; i.e. if
Ø = (A, A) = ((a, b, c), (a, b, c)) is an interval type-2 triangular fuzzy number,
then applying PGMIR on it gives
1 1
P Ø = (a + c + a + c) + (b + b)
12 3
88 V. P. Singh and K. Sharma
∞ 4 8 3 9 6 5 11
4 ∞ 9 6 10 3 4 6
8 9 ∞ 8 4 8 12 5
3 6 8 ∞ 9 8 7 3
9 10 4 9 ∞ 11 12 2
6 3 8 8 11 ∞ 8 15
5 4 12 7 12 8 ∞ 12
11 6 5 3 2 15 12 ∞
The next highest demand corresponds to the second row and seventh column. In
case, we join vertex 7 and vertex 2, the new cycle formed will be 0-4-7-2-0. But
this will happen only if all the conditions given in Step 5 are satisfied. Since crde-
mand[4]+crdemand[7]+crdemand[2]=21 < 25 which means that the capacity con-
straint is still satisfied. Here also, both 7 and 2 belong to two different cycles and
they are, respectively, the last and first customers. So, the new route is 0-4-7-2-0.
Continuing in the same way, there exist three different routes, namely 0-6-1-5-
0, 0-4-7-2-0 and 0-3-0 travelling, which never violate the constraints and fulfil the
requirement of travelling salesperson of visiting every node exactly once. The cost
of traversal of these routes comes out to be 48 units.
Capacitated Vehicle Routing Problem with Interval … 89
6 Conclusion
In this paper, we have attempted to solve capacitated vehicle routing problem when
the demands of the customers are more imprecise. In this paper, the demands of the
customers are given by interval type-2 triangular fuzzy number. This paper dealt with
defuzzifying the customers’ demand by using PGMIR and then finding the feasible
routes with the help of Clark and Wright algorithm. In this paper, it has been observed
that the cost comes out to be a crisp number.
References
1. Clarke G, Wright JW (1964) Scheduling of vehicles from a central depot to a number of delivery
points. Oper Res 12(4):568–581
2. Corberán Á, Laporte G (2013) Arc routing: problems, methods, and applications. SIAM
3. Dantzig GB, Ramser JH (1959) The truck dispatching problem. Manage Sci 6(1):80–91
4. Javanmard M, Nehi HM (2019) A solving method for fuzzy linear programming problem with
interval type-2 fuzzy numbers. Int J Fuzzy Syst 21(3):882–891
5. Singh VP, Chakraborty D (2017) Solving bi-level programming problem with fuzzy random
variable coefficients. J Intell Fuzzy Syst 32(1):521–528
6. Singh V, Chakraborty D (2015) A dynamic programming algorithm for solving bi-objective
fuzzy knapsack problem. In: Mathematics and computing. Springer, Berlin, pp 289–306
7. Toth P, Vigo D (2002) The vehicle routing problem. SIAM
Kinematic, Dynamic and Stiffness
Analysis of an Asymmetric 2PRP-PPR
Planar Parallel Manipulator
1 Introduction
The joint space velocities are mapped to the task space velocities by the Jacobian
matrix (J) of the manipulator as shown in Table 1 for the above-mentioned PPMs
(Fig. 1).
3 Performance Analysis
The numerical simulations were carried out using the Jacobian matrices to obtain the
performance index and workspace for the mentioned PPMs. Horizontal (x-direction)
and vertical (y-direction) span range is bounded to 1 m by 0.866 m (equilateral
triangle base in plane) for 3RRR and 3PPR delta shape base PPM and 1 m by 1 m
(square) for 3PPR U shape base and proposed PPM. The active prismatic actuators
(si ) displace within the interval [0, 1] meters. The rotation of the revolute joints (θi )
of 3RRR PPM limits within the interval [0°, 180°]. The shape of the mobile platform
Table 1 Jacobian of planar parallel manipulators
3RRR [5] 3PPR-base [6] 3PPR U-base [6] 2PRP-PPR [7]
⎡ ⎤ ⎡ ⎤ ⎡ ⎤ ⎡ ⎤
2h 1 0 0
D11 0 0 1 0 h cos θz 10 cos θ z
⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
a
⎥ h 3 a ⎥ −x
Jq = ⎢ J =⎢ − cos α sin α sin(α + θz ) ⎥ J =⎢ J =⎢ − tan θ z 1 ⎥
⎣ 0 D22 0 ⎦ ⎣ 2 ⎦ ⎣ 0 1 − 3 sin θz + 2 cos θz ⎦ ⎣ cos2 θz ⎦
a (s−x)
0 0 D33 cos α sin α 2 sin(α − θz ) 0 1 − h3 sin θz − a2 cos θz − tan θz 1 cos2 θz
⎡ ⎤
Q Q Q
⎢ 11 12 13 ⎥
Jx = ⎢ ⎥
⎣ Q 21 Q 22 Q 23 ⎦
Q 31 Q 32 Q 33
where where a and h are side length and height of an equilateral triangle based mobile platform respectively, α is the internal
Di j = f (x, y, θz )
T
Kinematic, Dynamic and Stiffness Analysis of an Asymmetric …
angle which is 60° in case of 3PPR PPM and x y θz represents the task space vector
Q i j = f (x, y, θz )
and J = Jq−1 Jx
93
94 D. Singh et al.
Fig. 1 Planar parallel manipulators frame diagram: (a) 3RRR, (b) 3PPR -base, (c) 3PPR U-base,
(d) proposed 2PRP-PPR
has been considered as an equilateral triangle with 160 mm side length. It is assumed
that the masses of actuators and links are negligible. For the above-mentioned PPMs,
the mass of the mobile platform is assumed to be 10 kg. Common task space vector
assigned for the workspace is: x = 10–825 mm, y = 10–800 mm, θz = 30°.
3.1 Workspace
Fig. 3 Kinematic and dynamic performance, (a) manipulability, (b) isotropy, (c) power consump-
tion, (d) energy requirement
PPMs. Isotropy also shows similar behavior. 3PPR U-base and the proposed 2PRP-
PPR PPM possess better performance indices which mean higher manipulability,
isotropy and lower resistivity as compared to other selected manipulators.
Energy and power consumption were researched, analyzed and compared reveal-
ing data on the actuator choice. The simulation was performed by Adams/View
software. The basic assumptions considered for the analysis are listed below:
1. The mobile platform is rigid.
2. The manipulator components follow Hooke’s law.
3. The manipulator was considered to be present in a gravity-less space.
The following common singular-free end-effector/task space circular trajectory
is given to the manipulator:
4 Stiffness Analysis
This paper relates the end-effector torques and forces to the corresponding angular
and linear joint displacement vectors by utilizing the joint space Jacobian stiffness
matrix in addition to matrix structural analysis (MSA) matrix. The basic assumptions
taken are listed below:
1. The mobile platform is rigid.
2. Weights of the manipulator components are negligible.
3. The joints are non-frictional and experience negligible deformation.
96 D. Singh et al.
The PPMs are divided into nine numbers of space beam elements with each node
having forces and moments in specific directions as depicted in Fig. 4.
In this section, the proposed PPM is compared with other considered manipulators.
The manipulator’s physical and geometrical parameters are listed in Table 2.
The task space variables (x, y, θz ) for the manipulators 3RRR and 3PPR base
are taken as (500, 577, 0°–45°), and for the manipulators, 3PPR U-base and 2PRP-
PPR are taken as (500, 500, 0°–45°). The horizontal span is bounded as 1000 mm,
and the orientation varies from 0° to 45° for singularity-free workspace.
To demonstrate, the stiffness analysis of the system has been performed using
MATLAB and NASTRAN (FEA) software and compared. Five sets Fx , Fy , Mθz
of end-effector force (in N) and bending moment (in N-mm) taken for the analysis are
(10, 50, 50), (20, 60, 60), (30, 70, 70), (40, 80, 80) and (50, 90, 90). The end-effector
deflection analysis has been performed in three directions on the plane, (1) x-axis
deflection (δx ), (2) y-axis deflection δ y , (3) angular deformation (δθz ).
As the simulation of end-effector deformation for multiple cases is time-
consuming, the first set of forces and moments (10 N, 50 N, 50 N-mm) has been
considered for simulation in FEA software (NASTRAN) with orientation angle θz =
30° for all the manipulators. Comparison result concludes a few points as mentioned
below:
1. The estimated analytical findings match well with those of NASTRAN software
which verified the model of theoretical stiffness.
2. In the case of 3-PPR U-base and proposed manipulator, the end-effector
deflections (δx , δ y and δθz ) are lower.
3. Stiffness of the manipulators will be less in case of higher orientation angle.
4. Stiffness in y-direction will be lower for all the manipulators as compared to other
directions. But, the stiffness in case of 3PPR U-base and proposed manipulator
will be higher in all planar direction.
This paper presents the NASTRAN model in Fig. 5 for a specific orientation angle
30° as remeshing costs a lot of computation time and is very tedious.
The comparison result only for the proposed PPM is listed in Table 3 due to page
constraint. NASTRAN results are close to analytical. This study showed that the
end-effector deflection is lower for 2PRP-PPR and 3PPR U-base PPM as compared
to others.
As deflection and stiffness are directly related, it can be concluded that 2PRP-
PPR and 3PPR U-base PPMs will possess higher stiffness as compared to other
manipulators.
Fig. 5 FEA result of PPMs, (a) 3RRR, (b) 3PPR -base PPM, (c) 3PPR U-base, (d) 2PRP-PPR
98 D. Singh et al.
Table 3 Comparative result of 2PRP-PPR end-effector for the first set of loading condition (10 N,
50 N, 50 N-mm) with 30° constant orientation angle
Deflections Analytical FEA software (NASTRAN) Difference
δx (m) 0.004173e−06 0.005733e−06 0.00155e−06
δ y (m) 0.0463e−03 0.0523e−03 6.0e−06
δθz (rad) 0.000514e−03 0.000763e−03 2.49e−07
5 Conclusions
The following are a few significant points observed from the present study:
a. The proposed 2PRP-PPR PPM possesses less singularity in the workspace due
to the partially coupled kinematic relationship.
b. 2PRP-PPR and 3PPR U-base PPMs exhibits favorable outcomes of optimal
design and performance index as it possesses lesser singularity workspace.
c. Due to structural simplicity and avoidance of interference of links, 3PPR U-
base and the proposed 2PRP-PPR PPM require minimum power and energy to
perform the operation/trajectory as compared to other PPMs.
d. The comparison showed that the stiffness of 3PPR U-base and 2PRP-PPR PPM
is higher as investigated and validated through NASTRAN.
References
1 Introduction
Today, also the rate of heat dissipation from IC chips is main concerned even simple
liquid cooling is insufficient, hence there is increased in interest of nanofluids flow.
Azizi et al. [1] studied that for concentrations of 0.05, 0.1, and 0.3 mass percentages
of CuO nanoparticle in water; the Nusselt numbers increases by 17%, 19%, and
23%, respectively, in comparison to water. Dongsheng et al. [2] used Al2 O3 -Water
nanofluid for their experiment in the case of a circular tube and found out that the
local heat transfer coefficient increases up to 47% for 1.6% concentration of Al2 O3
nanoparticles.
From the literature review, it is observed that there may be further research required
in case of some points like comparison of heat transfer enhancements using various
concentrations of different nanoparticles in base fluids. Therefore, in the present anal-
ysis, this particular aspect will be analyzed numerically. Furthermore, commercially
available CFD software ANSYS Fluent 13.0 is used for simulation purpose.
Figure 1 is the illustration of heat sink and single microchannel domain. Further
information on the model other than dimension can be found from Toh et al. [7]. In
this study, three different cases of heat sink dimensions are considered as discussed
in Table 1. Here case 0 and case 1 dimensions are considered only for validation
purpose and Case 3 dimensions are used for further numerical simulation.
The thermophysical properties of base fluid and nanoparticles are considered
from the previous study paper. Here particularly Xiang-Qi et al. [3], Khalil et al. [4],
Batchelor et al. [5], and Nilesh Purohit et al. [6] equations are considered to find out
properties of nanofluid. To find out local thermal resistance and local heat transfer
coefficient following equations are used, respectively,
Tmax(z) − Tin
R= (1)
q
q
hz = (2)
Twall − Tb
Here, the Cartesian tensor system is used for more detail information of governing
equation is found from Toh et al. [7] research article. Furthermore, numerical analysis
is performed on ANSYS FLUENT 13.0, Solver is specified at simple pressure based
on absolute velocity formation. For the simulation design solution, Standard Viscous-
Laminar model is opted with initial iteration value of 0.3 is taken as a relaxation factor.
Further second-order upwind spatial discretization is considered for momentum and
for other species. The residuals energy equation is set at 1 × 10−4 , for momentum
and the other component equations it is 1 × 10−6 and for continuity equations 1 ×
10−3 .
For validation of models, the results from the simulation are compared with previous
data from literature article [7, 8] as shown in Fig. 2a, b. The simulation results are in
good agreement with experimental data; there is little deviation seen because of the
certain assumption which is made during the study.
Fig. 2 Correlation between experimental and present study for local thermal resistance parameters
along with fluid flow: a for case-0, b case-1
CFD Analysis for Heat Transfer Enhancement of Microchannels … 103
Fig. 3 Effects of nanoparticles varying concentration in water at z = 0.9 cm plane: a on local thermal
resistance, b on local heat transfer coefficient [∅b % = percentage concentration of nanoparticles]
From Fig. 3a, it is evident that for volume fraction of 0.005; thermal resistances
is decrease up to 5.62 percentages for different nanoparticles except for SiO2 . The
decrement is sharp as even addition of a 0.5% Al2 O3 increases thermal conductivity of
base fluid by 70.17 percentage. Since heat capacity (ρcp ) of CuO is higher, decrement
in thermal resistance is more in case of CuO-Water nanofluid. For other cases of
nanoparticle, thermal resistance remains constant, increases or decreases slowly as
their volume concentration increase. Figure 3b shows that for 0.5% Al2 O3 in water
there is 47.35% of increment in heat transfer coefficient and for 4% of Al2 O3 , it is
61.53% compared to water.
Figure 4 is the illustration of the variation in local thermal resistance for varying con-
centrations of nanoparticles in EG20. It is observed that for 4% concentration of CuO
in EG20; thermal resistance is decreased by 36.52% and for the same concentration of
Al2 O3 , it is 36.13% when compared with ethylene glycol. After analyzing the above
two sections, a concrete conclusion can be made that CuO and Al2 O3 nanoparticles
are the best choices. Hence for further study, these particular two nanoparticles are
selected.
104 S. S. Bhuvad et al.
Fig. 4 Effects of
nanoparticles varying
concentration in EG20: on
local thermal resistance at
z = 0.9 cm plane
Fig. 5 Comparison between water- and EG20-based fluid for different concentration of Al2 O3 and
CuO nanoparticles, at Z = 0.9 cm plane: a on local thermal resistance, b on local heat transfer
coefficients
CFD Analysis for Heat Transfer Enhancement of Microchannels … 105
made between the local convective heat transfer coefficient, EG20 is more effective
than water-based except for lower concentration below 1% of nanoparticles, it is
almost the same for both cases. Furthermore, in both cases, convective heat transfer
coefficients increases linearly as a concentration of nanoparticles are rises.
4 Conclusions
CFD analysis is performed using ANSYS FLUENT 13.0 software and the following
observations are made from the above study:
1. Even lower concentration of oxide nanoparticles in water and EG20 base fluids;
decreases local thermal resistance and increases local convection heat transfer
coefficients effectively.
2. For ∅b = 0.5%, in case of CuO-water nanofluid, the local thermal resistance
decreases by 5.62% when compared to water, and for CuO-EG20 it is decreased
by 37.21% compared to ethylene glycol.
3. In spite of higher local convective heat transfer coefficients in EG20 compared
to water, still it is less effective for cooling of devices because of higher viscosity
for the same Reynolds number.
4. For this specific application in the case of electronics device, Water is the best
choice as base fluid compared to EG20. The use of EG20 as a base fluid in
nanofluid flow is preferable for higher working temperature condition as it has
higher boiling point.
5. From this simulation study, after comparing all nanoparticles with given two base
fluids. A concrete observation can be made that CuO-water nanofluid is more
effective for cooling of electronic devices. As with 4% concentration of CuO
in water, local convective heat transfer coefficient increases up to 61.53% other
than that it has less cost, more stability and power required by the pump is also
less.
References
Abstract This paper proposes an effective way to identify the burr using image
processing technique. The proposed method uses a simple mirror setup to capture
3D details of the burr. A set of image processing algorithm is used to calculate
the dimensions of the burr. The burr dimensions are verified using the coordinate
measuring machine. These details are then used to generate the trajectory for robot
deburring. The deburring path generated is verified using ABB RobotStudio simu-
lation software. The experimental result shows that the proposed approach provides
an efficient method for robotic deburring.
1 Introduction
Automating deburring process is highly challenging since the burr geometry is irreg-
ular and unpredictable. Though it is a critical finishing process, very few efforts are
reported in the literature. One such attempt is to extract burr parameter from the orig-
inal workpiece using parametric modeling is made by Ryuh et al. [1]. However, this
method was time-consuming. To overcome this, Lee et al. [2] developed a program-
ming technique by automatically scanning the workpiece to generate a trajectory
for robot-assisted grinding. However, for every product, the scanning has to done
separately. To overcome the above drawbacks, Jinno et al. [3] and Princely et al.
[4] developed an automatic programming method using computer vision and force
feedback. However, estimation of burr dimensions from the image data remains a
challenge. In view of this, Idaku et al. [5] proposed an advance coded structured light
2 Methodology
In order to get 3D view of the burr, a setup of four mirrors placed at 45-degree
angle is used (Fig. 1). Image acquisition is done using a CCD camera placed at
a fixed distance above the workpiece. The mirror setup transforms the 3D view
into multiple 2D projected images. These images are binarized for estimating the
height and thickness of burr by using height and thickness algorithm developed by
authors [7]. These estimated values are further used to generate the robot trajectory
necessary for burr removal. This trajectory is simulated using ABB’s RobotStudio
software for verification. On successful simulation, it is given to the robot controller
for performing a deburring operation.
Fig. 1 a Binary Image of four sections captured in mirror. b Example of Left section image with
size of 10 by 10
Burr Registration and Trajectory Planning of 3D Workpiece … 109
The proposed image acquisition setup consists of four mirrors kept at 45-degree angle.
A CCD camera is placed at a fixed height above the mirror to capture the image. This
setup enables us to capture 3D burr and convert them into 2D images. The image
consists of five sections and are named Lt (Left), Rt (Right), Up (Upper), Lo (Lower),
and C (Central). The Up, Lo, Lt and Rt section of the whole image provide back,
front, and both side views of the burr image, respectively. These sections are used to
identify and estimate the burr dimensions.
The burr dimensions, i.e., the thickness and height are estimated from the image
obtained through the CCD camera. The height data of the burr is obtained from the
binarized Up, Lo, Lt, and Rt sections of the image (Fig. 1a). It can be observed that
the presence of burr shows high intensity, i.e., 1 and 0 when the burr is absent. A
unit matrix scanning method is employed to find the height of the burr. For example,
suppose the Left section image is of size 10 by 10 as shown in Fig. 1B, each row
is scanned for the presence of “1”. If “1” is found in a location, the counter value
increases by “1” and is saved in a separate table. Then the search variable moves to
the next row and repeats the same procedure till the last row. After this, the maximum
value in the table is determined. This value gives us the height of the Left section.
The same method is used to find the height of Right section. However, in the case
of Up and Lo section, the search variable is moved column-wise as the height of the
section is directing from top to bottom and vice versa.
Trajectory is generated using thickness and height of the burr. The fundamental
inverse kinematic equations are used to calculate roll, pitch, yaw of the ABB IRB
120 robot. A data extraction algorithm is used to extract the data points for trajectory
planning. In this algorithm, first the inner edge of the burr part is detected by using
“Prewitt edge detection algorithm”, then number of passes is calculated by consider-
ing diameter of the tool. Since multiple edges are found during inner edge detection
as shown in Fig. 2a, some additional steps like closed curve fitting, smoothening and
backtracking coordinates elimination are used to assure accurate trajectory planning.
Closed curve fitting method is used to eliminate the multiple edges of the burr parts.
As the data points extracted from image are random, it was necessary to cluster data
points based on the features. Gaussian mixture model based on clustering is used
for curve fitting. Since this model is suitable only for open-loop data, the image is
110 M. R. Rahul et al.
Fig. 2 a Multiple inner edges of the burr part. b Sectioning of inner edge image to four equal
quadrants for open-loop data
sectioned into four as shown in Fig. 2b to make it open-loop data. This data is then
clustered by using K-mean clustering technique. A probabilistic Gaussian mixture
model is used to identify centers of various clusters and a curve is fitted to acquire
correct geometry as shown in Fig. 3a. After curve fitting, all quadrants are re-joined
to get a single geometry as shown in Fig. 3b. The single edge obtained by using closed
curve fitting algorithm has too much waviness, which is used directly, results in the
jerky motion of the manipulator. Hence, it is necessary to make the curve smooth.
For smoothening, an iterative discrete cosine transform algorithm is used. By min-
imizing the generalized cross-validation score, smoother curve is generated (Fig. 4a,
b). The data of inner edge has some backtracking points as seen in Fig. 4b. These
points are to be removed as they will lead to unnecessary points during data extraction
for trajectory planning. A backtracking algorithm using the directional derivative is
developed to eliminate these points. The algorithm calculates the direction vector of
Fig. 3 a Curve fitting done by using GMR model for all four quadrants. b Data obtained after curve
fitting and re-joining all quadrants
Burr Registration and Trajectory Planning of 3D Workpiece … 111
50 50 50
45 45 45
40 40 40
35 35 35
30 30 30
25 25 25
20 20 20
15 15 15
10 10 10
5 5 5
0 0 0
0 5 10 15 20 25 30 35 40 45 50 0 5 10 15 20 25 30 35 40 45 50 0 5 10 15 20 25 30 35 40 45 50
Fig. 4 a Edge before smoothening. b Edge after smoothening. c Result of backtracking coordinate
elimination algorithm
each point and saves it in a separate variable. For a particular location, if the direc-
tion vector is in reverse to earlier point, then that point is eliminated. The result of
backtracking algorithm is shown in Fig. 4c.
To verify the accuracy of the generated trajectory points, simulation using ABB
RobotStudio software is done before actual deburring. The generated coordinate
for the deburring trajectory is imported into RobotStudio software. The simulation
shows that the robot tracked the burr trajectory accurately. To access the feasibility
of the proposed method, deburring on a workpiece is conducted. The experimental
setup developed for this work consists of a CCD camera, a workpiece for deburring
on a temporary fixture and ABB IRB 120 six degrees of freedom industrial robot
with a dummy tool attached. The RAPID code generated by the ABB RobotStudio
software is transferred to the ABB IRB 120 robot controller. The trajectory generated
had jerky motion in the early stage of execution. It is corrected by applying additional
conditioning to the path like smoothening and backtracking coordinate elimination
which is discussed in the previous section. The outcome of the experiment clearly
shows that the robot tracked the burr path perfectly.
4 Results
To validate the algorithm, burr height of multiple specimens is estimated using algo-
rithm and CMM. It is observed that the average value of burr height is 4.95 mm
from the algorithm and 5.3 mm from CMM resulting into average error of 0.72 mm.
The coordinates generated by closed curve fitting had waviness and direction points.
To filter this, smoothening function and backtracking point elimination algorithm is
used. It is observed that the robot trajectory became smooth resulting in a Jerk’s free
112 M. R. Rahul et al.
motion of robotic arm (Fig. 5). It is also noticed that the total power consumption is
reduced from 35 W to 16 W after filtering which is about 50% reduction (Fig. 6).
Burr Registration and Trajectory Planning of 3D Workpiece … 113
5 Conclusion
In this paper, an attempt is made to detect the burr dimension using a simple mirror
setup. The approach provides an effective way of generating a deburring trajectory
without explicitly teaching the robot. The simulation and experiment show that this
technique can be used for deburring any workpiece. Furthermore smoothening and
backtracking point elimination algorithm helps in effective trajectory planning and
Jerk free motion of the robotic arm.
References
1. Ryuh BS, Pennock GR (2006) Robot automation systems for deburring. In: Industrial robotics:
programming, simulation and applications, p 15
2. Li CJ, Han GJS, Kyung JH, Park CH (2011) Study on teaching path reconstruction algorithm
based direct teaching and playback method. In: Eighth international conference on ubiquitous
robots and ambient intelligence, Incheon
3. Jinno M et al (1999) Teaching-less robot system for finishing workpieces of various shapes,
Kyongju, South Korea, South Korea, 17–21 Oct 1999
4. Leo S, Princely F (2014) Vision assisted robotic deburring of edge burrs in cast parts. 12th Glob
Congr Manuf Manag GCMM 97:1906
5. Idaku I, A coded structured light projection method for high-frame-rate 3D image acquisition.
In: Ventzas D (ed) Advanced image acquisition, processing techniques and applications I, 14
March 2012
6. Yohichi Nakao YW (2011) Drilling burr profile measurement method using image processing
technique. Trans Jpn Soc Mech Eng
7. Bhute RY (2017) Burr detection and trajectory planning using image processing for robotic
deburring. Dissertation, Master of Technology, VNIT, Nagpur
In-situ Microwave-Assisted Casting
of ASTM B23 Tin-Based Babbitt Alloy
Abstract Metal casting is one of the primary manufacturing processes used in indus-
tries. It is also one of the economical routes of producing components varying from
simple to complex shapes. But conventional casting methods possess some draw-
backs in terms of higher energy consumption, longer melting times, air pollution
and higher defects. Microwave-Assisted Casting is one of the latest manufacturing
technique which can overcome these drawbacks owing to its several advantages
like time-saving, clean and environmentally friendly. This paper reports on in situ
microwave-assisted die casting process of bush with 35 mm OD, 25 mm ID
and 40 mm Length, using microwave furnace at 2.45 GHz and 1.4 kW and also
conventional die casting process by using muffle furnace of 3.5 kW of ASTM B23
Tin-based Babbitt Alloy. It was found that microwave-assisted die casting process
reports time-saving of 07 min and power-saving of 1.1 kW h as compared with the
conventional die casting process. Further SEM images of microwave bush casting
showed uniform grain distribution as compared to SEM images of conventional die-
casted bush. It was found that microwave-assisted die-cast bush is having 1.4 times
higher hardness as compared with that of conventional die-cast bush.
1 Introduction
Casting is one of the most preferred manufacturing techniques used in the industries.
It is also one of the efficient techniques for producing the components. However,
there are some drawbacks in the conventional casting process like environmental
pollution, higher energy requirement, longer melting times, and a large amount of
defects. To overcome these drawbacks, new techniques are being developed and many
more in the process of development. Hence industries are now aiming at utilizing
new techniques that are clean, energy-efficient and economical [1].
2 Experimental Setup
In situ casting process of ASTM B23 Babbitt bulk alloy was done first by conventional
die casting method by using muffle furnace and then by microwave-assisted die
casting using microwave furnace. The component chosen was shaft bush used in one
the special purpose cutting machine for performing teeth cutting operation of files.
The bush size is as follows:
Outer Diameter [OD] = 35 mm, Inner Diameter [ID] = 25 mm and Length
= 40 mm (Fig. 1).
A silica crucible of 250 ml capacity, which absorbs microwaves rapidly, was
used for the melting of the charge in both cases [9]. Table 1 shows the chemical
composition of ASTM B23 Babbitt bulk alloy.
In-situ Microwave-Assisted Casting of ASTM B23 … 117
Fig. 1 Bush
Babbitt Alloy casting. Finally, the cast bush was machined by using turning operation
followed by boring operation to achieve the final size of the bush Outer Diameter
[OD] = 35 mm, Inner Diameter [ID] = 25 mm and Length = 40 mm. Figure 8
shows the finished Babbitt bush as per required dimensions.
120 S. S. Gajmal and D. N. Raut
In both cases, the time required for melting the charge was noted. Also by using energy
meter the actual power consumed during conventional and microwave melting was
noted [10]. The in-situ bush casting obtained by using muffle furnace was found
to have surface defects like porosity, blowholes, whereas the in situ bush casting
obtained by using microwave furnace was found to defects free. The results obtained
are compiled in Table 2.
In-situ Microwave-Assisted Casting of ASTM B23 … 121
3.2 Microstructure
Microstructure analysis was done by light microscopy and then by using a scanning
electron microscope [11]. Figure 9 shows the microstructure of conventional and
microwave-assisted die-casted bush done by light microscopy.
It is clear that Tin-based Babbitts contain three phases α, β and η. The basic α-
phase is a solid solution of antimony (Sb) and copper (Cu) in tin (Sn). The β-phase is
Sn–Sb and η-phase is Cu–Sn. The properties of tin-based Babbitt like hardness and
wear resistance depend upon the amount and morphological size of the hard phases
present in the alloy, especially Sn–Sb, i.e. β-phase and Cu–Sn, i.e. η-phase. When
these hard phases are well defined in the microstructure, the wear resistance of the
alloy gets improved (Fig. 10).
Further microstructure analysis was also carried out by using scanning elec-
tron microscope (Oxford Instruments, X-MaxN , 50, GeminiSEM 300) available at
122 S. S. Gajmal and D. N. Raut
Fig. 10 (continued)
Hardness test was carried out on the conventional and microwave-assisted die-casted
bush samples (polished) by using Hardness Testing Machine model-MRB-250, Sr.
No. 2001/059 (Manufactured by; Meeta Test Instruments Pvt. Ltd., Miraj, Maharash-
tra) Rockwell hardness number was measured by using 1/16 ball indentor, scale-B
with load of 100 kgf (Table 3).
It is clear from the above readings that hardness of the microwave-assisted cast
bush is almost 1.4 times higher than the conventional die-cast bush. This means there
is an improvement in the hardness of the microwave-assisted casting.
In-situ Microwave-Assisted Casting of ASTM B23 … 125
4 Conclusion
Acknowledgements The authors are grateful for the support and encouragement from the Gharda
Institute of Technology, Lavel, Taluka Khed, District Ratnagiri, Maharashtra. The authors would also
like to acknowledge Nya. Tatyasaheb Aahalye Arts, Ved. S. R. Sapre Commerce and Vid. Dadasaheb
Pitre Science College, Devrukh, District Ratnagiri, Maharashtra for providing microwave furnace
facility to carry out this research work (Ref. No.: Sr. 216/2018-19 Dated: 30-07-2018).
References
1. Mishra RR, Sharma AK (2015) A new in situ casting technique using microwave energy at
2.45 GHz. In: Proceedings of the India international science festival—young scientist meet,
Dec 2015
126 S. S. Gajmal and D. N. Raut
2. Mishra RR, Sharma AK (2016) A review of research trends in microwave processing of metal-
based materials and opportunities in microwave metal casting. Crit Rev Solid State Mater Sci
41(3):217–255
3. Agrawal D (2006) Microwave sintering, brazing and melting of metallic materials. Sohn Int
Symp Adv Process Metals Mater 4:183–192
4. Mishra RR, Sharma AK (2017) Structure-property correlation in Al–Zn–Mg alloy cast
developed through in-situ microwave casting. Mater Sci. Eng A 688:532–544
5. Kashimura K, Sato M, Hotta M, Agrawal DK, Nagata K, Hayashi M, Mitani T, Shinohara N
(2012) Iron production from Fe3 O4 and graphite by applying 915 MHz microwaves. Mater Sci
Eng A 556:977–979
6. Mishra RR, Sharma AK (2018) Experimental investigation on in-situ microwave casting of
copper. In: IOP conference series: materials science and engineering, vol 346
7. Rathee Y, Aggarwal V, Sehgal A (2016) Microwave assisted casting for fabrication of micro
components. Indian J Sci Tech 9(36), Sept 2016
8. Mishra RR, Sharma AK (2012) Effect of solidification environment on microstructure and
indentation hardness of Al–Zn–Mg alloy casts developed using microwave heating. Int J Metal
Cast 12:370
9. Chandrasekaran S, Basak T, Ramanathan S (2011) Experimental and theoretical investigation
on Microwave melting of metals. J Mater Process Tech 211(3):482–487
10. Singh S, Gupta D, Jain V (2016) Novel microwave composite casting process: theory, feasibility
and characterization. Mater Des 111(5):51–59
11. Mishra RR, Sharma AK (2016) Microwave–material interaction phenomena: heating mecha-
nisms, challenges and opportunities in material processing. Compos Part A: Appl Sci Manuf
81:78–97
Optimization of Heat Transfer Behavior
of Industrial Refrigerants Through
Different Cross-Section Microchannels
G. Chakraborty
Department of Mechanical Engineering, SDET-BGI, Kolkata, India
e-mail: gourab001@gmail.com
S. Sarkar (B)
Aeronautical Society of India, New Delhi 110092, India
e-mail: s_shubhankar@yahoo.com
A. Chanda
Mechanical Engineering Department, Jadavpur University, Kolkata 700032, India
e-mail: arunabhachanda@yahoo.com
1 Introduction
In the performance point of view, microchannel heat exchanger is way better than
conventional heat exchanger, because of its higher heat transfer performance, low
pressure drops, reduced size and above of all low coolant requirements which makes it
perfect to lower the operational cost. Removal of high heat flux still keeps the cooling
industries in Dire Straits. In the area such as Aerospace Industries, macro-electronic
devices and nanorefrigerant applications there are limited cooling techniques. For the
past few decades, it has been seen for cooling in microfluidic devices fluids pollute the
environment such a way that in recent trends fluids that have lower pollution factors
are introduced. In this numerical investigation, global warming potential have also
been considered as a prime focus for selecting suitable materials.
Tucker and Pease [1] first investigated the microchannel and introduce the concept
of electrical cooling. Zhou et al. [2] experimentally presented the Evaluation of Cor-
relations of Flow Boiling Heat Transfer of R-22 in Horizontal Channels. Shakarah
et al. [3] theoretically investigated the Boiling Two-Phase Flow in Microchannels
where they found the boiling categorized flow rules in microchannels. Duryodhan
et al. [4] experimentally simulated the Boiling flow through diverging microchan-
nel and studied the effect of heat flux and mass flux on pressure drop and heat
transfer coefficient with boiling in the diverging microchannel. Shamirzaev et al.
[5] experimentally present the flow boiling heat transfer for water and refrigerants
in microchannel heat exchangers. Ekhlas et al. [6] investigated Flow Boiling Heat
Transfer of R134a in Multi Microchannels. Lee and Mudawwar [7] had investigated
A Mechanistic Critical Heat Flux Model For Subcooled Flow Boiling based on local
Bulk Flow Conditions. Kandlikar [8] experimentally investigated A General Corre-
lation for Saturated Two-Phase Boiling Heat Transfer Inside Horizontal and Vertical
Tubes. Kew and Cronwell [9] have investigated the Correlations for the Prediction of
Boiling Heat Transfer in Small Diameter Channels. In their study, they have wanted
to set up the comparison of heat transfer coefficient between narrow existing flow
boiling heat transfer and nucleate boiling type. Hsieh and Lin [10] predicted Satura-
tion flow boiling heat transfer and pressure drop of refrigerant R-410A in a vertical
plate heat exchanger. In their prediction, they used the ozone-friendly refrigerant
R-410A which is the mixture of half percent of R-32 and half percent of R-125.
2 Theoretical Background
ρV (∇ · V ) = −∇ P + μ∇ 2 (1)
Further, in the flow field where the flow’s spatial properties are considered, The
substantial derivative applies to any flow-field variable, for instance, DT /Dt, Dp/Dt,
Du/Dt, etc., where T and p are the temperature and static pressure, respectively.
∂ T − ∂T ∂T ∂T ∂T
+ →
DT
≡ v ·∇ T ≡ +u +ν +w (2)
Dt ∂t ∂t ∂x ∂y ∂z
→
where ∂T
∂t
local derivative and −
v · ∇ T is convective derivative.
−
→ −
V ·→
n = 0 at the surface (when an inviscid flow is considered) (4)
R-410A is an HFC refrigerant, which is blended using same amount of R32 and R125
as an equal amount of mixture. It is nonflammable refrigerant and nontoxic. Pressure
generated by R-410A is 50% lower than R-22 although the operating pressure for
410A is slightly higher. R-410A has high vapor heat capacity which makes R-410A
better coolant than R-22. As per European standards for using new air conditioning,
R-22 was discontinued in 2010 because its effect on ozone layer depletion. This
numerical investigation will contribute selecting material for industrial cooling. On
the other hand, R-410A met the new standard for air conditioning application, with
no contribution to ozone layer depletion. As per the performance point of view, it
has been seen that R-410A can absorb and as well as release more heat than R-22.
Solubility of R-410A is more than R-22.
3 Model Specification
Further, this investigation shows the influence of geometry and input parameters.
It is pretty evident that input parameters have an important factor than the material
properties. In this investigation, three different fluids passed through the microchan-
nel. After the implementation of all boundary condition, it has been seen that the
heat is gradually decreasing from the surface to center in the temperature profile for
132 G. Chakraborty et al.
all microchannel. Where it has been found from inlet to outlet the temperature is
increasing with respect to position although general conditions have been applied in
the flow field. Heat flux profile abbreviation is written as H.F.P (Figs. 7, 8, 9, 10 and
11).
Heat flux profiles are very important for microchannel fluid flow. Figure shows
the temperature profile enabling the coolants to understand the most crucial parts
for cooling application. As per safety concern temperature profiles also ensure the
maximum heat profile zones to be treated priorities. In case in any anomaly standard
protocols are to be considered for dodging any kind of accidents. In this investi-
gation, it has been noted regarding inlet temperature plays a maximum role. Inlet
Optimization of Heat Transfer Behavior of Industrial … 133
temperature is to replace with coolant temperature for cooling application, thus these
two different fluids are considered for comparison in order to find the more effective
fluid. Investigation required all possible heat flux values for comparison are observed
in different temperature and the values are plotted (Fig. 12).
Investigation clearly shows that with almost the same configuration and external
condition for a microchannel coolant R-22 produces more heat flux than R-410A
for same geometries. Investigation furthermore extended to performance point of
view, simulation results are noted for heat transfer values which is a major deciding
parameter for selecting suitable coolant with respective for geometries. ANSYS
software is used for the simulation and values are tabled (Table 1).
Obtained values are pretty evident to prove the performance of R-410A over R-22.
Still, for more accurate and precised result statistical analysis has been also observed.
Optimization of Heat Transfer Behavior of Industrial … 135
100
50
R-22
R-410A
0
0 20 40 60 80
Inlet temperature in oc
Table 1 Heat flux values for different shape with respect to fluid
Shape R22 (W/cm2 ) R410a (W/cm2 ) Water (W/cm2 )
Circular 0.000825136 0.0001353165 0.001175561
Semi-circular 0.00037193 0.0002737194 0.003420678
Triangular 0.0000505 0.0004560873 0.00114894527
Rectangular 0.0009160594 0.00215477709 0.00040108186
Trapezoidal 0.0005887209 0.003740303 0.00070111689
5 Conclusion
Analysis has been carried out for different microchannels of various configurations,
in order to find optimum configuration of microchannels. It can be concluded that the
results in this study were compared with the results from literature reviews and it indi-
cated that the numerical results are in fine agreement with the reviews. Result shows
R-410A is a more suitable material for selection. Geometrically circular microchan-
136 G. Chakraborty et al.
0.004
0.0035
0.003
0.0025
R22
0.002
R410a
0.0015
Water
0.001
0.0005
0
Circular Semi-circular Triangular Rectangular Trapezoidal
nel is preferred over the other microchannels because of its heat transfer potentiality.
This work also contributes by concluding the preference of R-410A over R-22 for
selecting suitable refrigerant, keeping in mind global warming potential concerning
environmental betterment (Fig. 14).
Optimization of Heat Transfer Behavior of Industrial … 137
References
1. Tuckerman DB, Pease FW (1981) High-performance heat sinking for VLSI. IEEE Electron
Device Lett EDL-2 126–129
2. Zhou Z, Fang X, Li D (2013) Evaluation of correlations of flow boiling heat transfer of R22 in
horizontal channels. Hindawi Publishing Corporation
3. Shkarah AJ, Sulaiman MYB, Hj MRB (2013) Boiling two phase flow in microchannels: a
review. Indian J Sci Technol 6(11):5514–5521
4. Duryodhan VS, Singh SG, Agrawal A (2013) Boiling flow through diverging microchannel.
Sadhana 38(6):1067–1082
5. Shamirzaev AS, Mordovskoy AS, Kuznetsov VV (2018) An experimental investigation of
flow boiling heat transfer for water and refrigerants in microchannel heat exchangers. AIP
Publishing
6. Fayyadh EM, Mahmoud MM, Karayiannis TG (2017) Flow boiling heat transfer of R134A in
multi micro channels. Int J Heat Mass Trans 422–436. (Elsevier Publication)
7. Lee CH, Mudawwar I (1988) A mechanistic critical heat flux model for subcooled flow boiling
based on local bulk flow conditions. Elisevier Publication 14:711–728
8. Kandlikar SG (1990) A general correlation for saturated two-phase boiling heat transfer inside
horizontal & vertical tubes. ASME 112/219:219–228
9. Kew PA, Cronwell K (1997) Correlations for the prediction of boiling heat transfer in small
diameter channels. Appl Thermal Eng 17:705–715. (Elsevier Science Ltd. Publications)
10. Hsieh YY, Lin TF (2002) Saturation flow boiling heat transfer & pressure drop of refrigerant
R-410A in a vertical plate heat exchanger. Int J Heat Mass Transfer 45:1033–1044. (Elsevier
Science Ltd)
Evaluation of Two-Body Abrasive Wear
Using FIS and ANN
Abstract In the most recent three decades, many embed materials have been made
of metals, compounds, earthenware production, polymers, and so forth. Most met-
als and earthenware production are a lot stiffer than bone tissue which can bring
about mechanical bungle between the embed and the nearby bone tissue. Notwith-
standing other biocompatibility issues, metals are excessively firm while pottery is
excessively fragile and polymers are excessively adaptable and feeble to meet the
mechanical quality. Thus, composites of polymers and inorganic materials may offer
the ideal properties for embed materials. Polymers are well known because of their
low thickness, great mechanical quality, and simple formability. At the point, when
the composite is utilized as embed material, its development causes scraped spot at
the joint. Henceforth, a study on scraped spot wear of composites is fundamental
before utilizing it as embed material. In this work, two-body grating test has been
completed on HAp-HDPE and HAp-UHMWPE bio-composites to ponder the impact
of different test parameters on scraped spot wear. Fluffy derivation framework (FIS)
and Artificial neural systems (ANNs) are utilized to foresee the wear qualities of
composites. It has been seen that HAp-HDPE composite gives palatable outcomes
contrasted with HAp-UHMWPE composite as far as grating wear test. The analysis
results recommend that HAp-HDPE bio-composite has the potential for use as an
elective material for burden-bearing orthopedic applications.
M. A. Kumar (B)
Raghu Engineering College, Visakhapatnam, Andhra Pradesh, India
e-mail: amitkumarmehar@gmail.com
1 Introduction
2 Experimental Details
To do the trial work, HAp is blended with high thickness polyethylene (HDPE)
and ultra-high atomic weight polyethylene (UHMWPE) polymers with various vol-
ume rates. HAp powder is set up by blending Calcium Hydroxide (Ca(OH)2 ) pow-
der, Ortho Phosphoric (H3PO4 ) corrosive, and Ammonia(NH3) arrangement. The
technique is known as a wet concoction precipitation course [13–15].
The chosen control parameters and their qualities at various levels are recorded in
Table 1 underneath.
Evaluation of Two-Body Abrasive Wear Using FIS and ANN 141
The experimental layout using full factorial design of experiment along with S/N
ratio for HAp-HDPE and HAp-UHMWPE bio-composite are shown in Table 2.
Fig. 1 Block diagram of fuzzy inference system of Fuzzy Mamdani inference system
The experiment has been performed in the experimental layout gives the following
results.
The Analysis of Variance (ANOVA) Table for S/N ratio of HAp-HDPE composite
is shown in Table 3.
-
Data Means
% HAp Load Applied
42
40
38
Mean
1 2 3 1 2 3
No Of Cycles
42
40
38
1 2 3
Fig. 3 Main effects plot (data means) for S/N ratio of HAp-HDPE bio-composite
The probability (P) value for each factor is less than 0.05 which means that all the
factors are significant given by ANOVA Table 4. The Main Effects Plot (data means)
for S/N Ratio is also given in Fig. 3 which shows the most optimum value given by
the Main Effects Plot is 3, 3, 3.
-
Data Means
% HAp Load Applied
60
55
50
45
Mean
40
1 2 3 1 2 3
No Of Cycles
60
55
50
45
40
1 2 3
Fig. 4 Main effect plot (data means) for S/N ratio of HAp-UHMWPE bio-composite
0.05 means it is not significant. The Main Effects Plot (Data means) for S/N ratio is
shown in Fig. 4.
Here, the most optimum value given from Fig. 4 is 3, 1, 1.
In a FIS, a standard base is developed to control the yield variable. A fluffy n guideline
is a basic IF-THEN rule with a condition and an end. The assessments of the fluffy
principles and the mix of the after-effects of the individual standards are performed
utilizing fluffy set tasks and are finished by taking triangular Fuzzy sets. The tasks
on fluffy sets are not quite the same as the activities on non-fluffy sets. The fluffy set
guidelines for 27 trials which we have taken for HDPE and UHMWPE framework
for getting wanted yield.
Principle watcher for HDPE and UHMWPE are shown in Figs. 5 and 6. The
test information for HAp-HDPE and HAp-UHMWPE bio-composites through the
Fuzzy rationale. Table 2 demonstrated every one of the 27 trials with S/N proportion
in which least estimation of S/N proportion is 32.395 and the most extreme worth
is 47.33 for HAp-HDPE bio-composite and least estimation of S/N proportion is
23.999 and the greatest worth is 66.02 for HAp-UHMWPE bio-composite. Here, the
prescient worth is determined by composing the directions in MINITAB 15. At that
point, by exploratory worth and prescient worth, we determined the mean total rate
blunder by taking the normal or mean of all examinations by condition (1).
146 M. A. Kumar
By ascertaining, we can get the mean supreme rate blunder by condition (1) for
preparing is about around 0.01 and 0.085 for HAp-HDPE and HAp-UHMWPE bio-
composites through ANN model individually. So also, we can get the mean total
rate blunder for testing information. By ascertaining we can get the mean total rate
blunder by condition (1) for testing is about around 2.115 and 10.579 for HAp-HDPE
and HAp-UHMWPE bio-composites through ANN model individually.
Evaluation of Two-Body Abrasive Wear Using FIS and ANN 147
6 Conclusion
In this investigation, Fuzzy Mamdani and ANN methodologies were utilized to antici-
pate the abrasiveness in the two bodies wear trial of HAp-HDPE and HAp-UHMWPE
composites. It recommends that the Fuzzy Mamdani model is best for the foreseeing
information for the trial. The created models were assessed as far as their forecast
capacity. Fake neural systems (ANN) have potential applications in mechanized dis-
covery and finding of machine conditions. A large number of the ANNs for machine
condition checking utilized the pre-handled recurrence space highlights of the delib-
erate vibration signals. The proposed ANN and FIS models would thus be able to be
utilized successfully to foresee the two-body wear test. For HAp-HDPE proposed
level of 40% HAp, connected burden 1500 MPA and number of cycles (sliding sep-
aration) 300 are the better recommended for the structure for ideal outcome and less
wear between the bodies. For HAp-UHMWPE proposed level of 30% HAp, con-
nected burden 500 MPA and number of cycles (sliding separation) 100 are the better
recommended for the piece for ideal outcome and less wear between the bodies.
By the test performed and the outcomes recommend that the proposed compos-
ite is HAp-HDPE for the implantation as fake bone in the human body instead of
HAp-UHMWPE and further reason.
Evaluation of Two-Body Abrasive Wear Using FIS and ANN 149
References
1. Roeder RK, Converse GL, Kane RJ, Yue W (2008) Hydroxyapatite-reinforced polymer
biocomposites for synthetic bone substitutes. JOM 60(3):38–45
2. Jones MH, Scott D (eds) (1983) Industrial tribology: the practical aspects of friction, lubrication,
and wear. Elsevier Scientific Publishing Company, New York
3. Garnett J, Dieppe P (1990) The effects of serum and human albumin on calcium hydroxyapatite
crystal growth. J Biochem 266(3):863–868
4. Coombes AGA, Meikle MC (1994) Resorbable synthetic polymers as replacements for bone
graft. J Clin Mater 17(1):35–67
5. Juang HY, Hon MH (1994) Fabrication and mechanical properties of hydroxyapatite-alumina
composites. J Mater Sci Eng Part C 2(1–2):77–81
6. Sendemir A, Altintas S (1997) Production of HAp reinforced polymer composites for biomed-
ical applications. In: Proceedings of the 2nd international conference biomedical engineering
days, 5, pp. 114–117
7. Kannan S, Balamurugan A, Rajeswari S (2001) Bio-composites: a review of literature. Trends
Biomater Artif Organs 14(2):30–36
8. Chattopadhyay R (2001) Surface wear—analysis, treatment, and prevention. ASM-
International, OH, USA. ISBN 0-87170-702-0
9. Zhang Z, Friedrich K, Velten K (2002) Prediction on tribological properties of short
fibrecomposites using artificial neural networks. Elsevier Wear 252:668–675
10. Pompe W, Worch H, Epple M, Friess W, Gelinsky M, Greil P, Hempel U, Scharnweber D,
Schulte K (2003) Functionally graded materials for biomedical applications. J Mater Sci Eng
Part A 362(1–2):40–60
11. Sousa RA, Reis RL, Cunha AM, Bevis MJ (2003) Coupling of HDPE/hydroxyapatite
composites by silane-based methodologies. J Mater Sci Mater Med 14(8):475–487
12. Pramanik S, Agarwal AK, Rai KN (2005) Development of High strength Hydroxyapatite for
hard tissue replacement. J Trends Biomater Artif Organs 19(1):46–51
13. Viswanath B, Ravishankar N, Nayar S, Sinha A (2005) Synthesis, Sintering and Micro structural
characterization of nano crystalline Hydroxyapatite composites. Mater Res Soc Symp Proc
845(11–12):1–6
14. Lohfeld S, Barron V, Mc Hugh PE (2005) Bio-models of bone: a review. J Ann Biomed Eng
33(10):1295–1311
15. Jagur GJ (2006) Polymers for tissue engineering, medical devices, and regenerative medicine,
Concise general review of recent studies. J Polym Adv Technol 17(6):395–418
16. De Xie Z, Qian D Huang, Abdi F (2006) Crack growth strategy in composites under static
loading, Structures, structural dynamics, and materials conference. Am Inst Aeronaut Astronaut
5:1–8
17. Chowdhury AC, Kulkarni A Basak, Roy SK (2007) Wear characteristic and biocompatibility
of some hydroxyapatite-collagen composite acetabular cups. J Wear 262(11–12):1387–1398
18. Kothamasu R, Haung SH (2007) Adaptive Mamdani fuzzy model for condition-based
maintenance. Fuzzy Sets Syst 158:2715–2733
19. Tin-Oo MM, Gopalakrishnan V, Samsuddin AR, Al Salihi KA, Shamsuria O (2007)
Antibacterial property of locally produced hydroxyapatite. Archives Orofacial Sci 2(11):41–44
20. Neuendorf RE, Saiz E, Tomsia AP, Ritchie RO (2008) Adhesion between biodegradable
polymers and hydroxyapatite. J Acta Biomater 4(5):1288–1296
21. Huang AH, Farrell MJ, Kim M, Mauck RL (2010) Long-Term dynamic loading improves the
mechanical properties of chondrogenicmesenchymal stem cell-Laden Hydrogels. J Euro Cells
Mater 19(215):72–85
22. Eslami H, Solati-Hashjin M, Tahriri M, Bakhshi F (2010) Synthesis and characterization of
nano crystalline HAp obtained by the wet chemical technique. J Mater Sci 28(1):5–13
23. Ren Q, Baron L, Balazinski M (2011) Type-2 fuzzy modelling for acoustic emission signal in
precision manufacturing. Model Simul Eng 696947:1–12
Computational Analysis of Dual
Expander Aerospike Nozzle
Abstract To increase the performance of current space launch capability new rocket
designs are required. One concept that would undertake this need is an aerospike noz-
zle. The aerospike nozzle has proved its better performance compared to conventional
bell nozzles. The major advantage of the aerospike nozzle is its altitude compensat-
ing ability since the expansion of the jet is not bounded by a wall. In the case of
dual expander aerospike nozzle (DEAN) it uses separate expander cycles for both
fuel and oxidizer with an aerospike nozzle. The present study includes the design
of the combustion chamber and nozzle of the DEAN and its numerical analysis for
producing a thrust of 2000 KN and a specific impulse of 420 s. The given input con-
ditions are exit Mach number of 3.9, combustion chamber pressure and temperature
of 180 bar and 3818 K and use kerosene as fuel and liquid oxygen as oxidizer with
a mixture ratio of 2.65. The validation of the design is done with ANSYS FLUENT
and a C++ program is prepared for the development of geometry of the combustion
chamber and nozzle of the DEAN. The designed nozzle produces a specific impulse
of 472 s.
1 Introduction
A launch vehicle is designed to send a spacecraft designed for specific purpose into
the required orbit. Due to simplicity, manufacturability, mission effectiveness bell
nozzle rocket engine became the choice of engine and it is dominating today. The
major disadvantage of bell nozzle is that it does not work optimally at all altitudes [1].
The dual expander aerospike nozzle utilizes both fuel and oxidizer. It uses a separate
expander cycle for both fuel and oxidizer. The schematic diagram of the DEAN
is shown in Fig. 1. It uses separate turbomachinery for both fuel and oxidizer and
separate heating channels. The expander cycle is powered by using heat from the
walls of combustion chamber and nozzle. The heat obtained from the combustion
chamber is used to produce power in both expander cycle and hence in turn to power
both fuel and oxidizer pump [4].
The DEAN employs the oxidizer turbine and fuel turbine to power oxidizer pump
and fuel pump, respectively. Since both the fuel and oxidizer remain separated until
injection into the combustion chamber no interpropellant seals are required. In Fig. 1,
red dotted line represents the fuel flow and the blue dotted line represents the oxidizer
flow. The propellant first flows from the fuel tank to the pump. The pump delivers
the oxidizer to cooling channels in the combustion chamber and fuel to cooling
channels in the combustion chamber. By using the heat energy obtained from the
cooling channel it drives the turbine. From here oxidizer and fuel are injected into the
combustion chamber using injectors. After the combustion, the combusted product
expands against the aerospike nozzle to produce the desired thrust.
Computational Analysis of Dual Expander Aerospike Nozzle 153
2 Methodology
The given inputs are fuel and oxidizer, the oxidizer to fuel ratio, exit Mach num-
ber (M e ), combustion chamber pressure (Pc ), stagnation temperature of combustion
chamber (T0 ),specific impulse (I sp ), thrust (F). The kerosene and oxygen are used as
fuel and oxidizer with a mixture ratio of 2.65. The given input conditions are given
in Table 1.
F
Isp = (1)
ṁg0
With the value of characteristic velocity(C * ), we can find the value of At [3].
ṁC ∗
At = (3)
Pc
We can find the value of the combustion chamber area (Ac ) from the below
equation. Value of M ranges from 0.1 to 0.5.
γ +1
Ac 1 2 γ − 1 2 γ −1
= 1+ M (4)
At M γ +1 2
Now input the value of L * according to the fuel and oxidizer used to find the
volume of combustion chamber.
Vc
L∗ = (5)
At
Angelino’s approximate method is used for designing the contour of aerospike nozzle.
Consider Fig. 2. A sonic flow is present along throat AB. The flow is assumed to
expand by a centered wave originating at point A. For the desired contour of aerospike
nozzle, we have to consider the streamline passing through point B. A characteristic
line is a straight constant flow property line. It is inclined to the sonic line AB by an
angle A.
Where
a =μ−v
(6)
In Eq. (6), μ is the Mach angle and ν is the Prandtl–Meyer expansion angle.
1
μ(M) = sin−1 (7)
M
0.5
(γ + 1) (γ − 1) 3 0.5
v(M) = tan−1 M −1 − tan−1 M 2 − 10.5 (8)
(γ − 1) (γ + 1)
l A 1
= (9)
lt sin μ At
where A is the area normal to velocity vector and polar equation is given by Eq. (10).
In which M is the Mach number and it varies from 1 to desired exit Mach number
M e.
The input variables required for the design of expander cycle are chamber pressure
(Pc ), chamber temperature (T c ), mass flow rate of fuel and oxidizer, pump efficiency,
turbine efficiency, pressure ratio of turbine and pump. The pressure and temperature
at which fuel and oxidizer must store can be calculated using this data by using
turbine efficiency equation and pump efficiency equations.
Fig. 4 Temperature contour at sea level (Left), Mach number contour at sea level (Right)
4 Conclusion
The geometry of the Dual Expander Aerospike Nozzle is prepared for the given
conditions. The DEAN nozzle is providing superior performance compared to a bell
nozzle. A theoretical study on the various design aspects of aerospike nozzle has been
conducted for cryogenic applications with given performance parameters. Thrust
variation for different altitudes has been studied and found that specific impulse is
158 A. R. Shenoy et al.
References
V. R. Patil (B)
Department of Mechanical Engineering, AISSMS College of Engineering, Pune, India
e-mail: vrpatil@aissmscoe.com
S. S. Sane
Department of Mechanical Engineering, Pune Vidyarthi Griha’s College of Engineering and
Technology, Pune, India
S. S. Thipse
Power Train Engineering, Automotive Research Association of India, Pune, India
1 Introduction
Karanja oil from the seeds was extracted using a screw expeller. Oil was preheated
to remove water contents at about 100 °C for 10 min in the cylindrical stainless steel
tank. The Potassium methoxide is mixed with Karanja oil extract. After 8 h, glycerin
and methyl ester separates out. The suspended impurities like unconverted methanol,
unconverted catalyst were then removed by water washing method as methanol and
catalyst mixes easily with water and having more density than methyl ester. The
moisture was removed by heating the mixture at about 100–110 °C. Magnetic stirring
method was used for blending. The various properties of biodiesel blends are listed
in Table 1 which shows that density, calorific value, viscosity and cetane number of
B5, B7, B10 are closer to diesel fuel. The reduction in calorific value was observed
with the increase in proportion of the blend.
The technical specifications of the diesel engine (Fig. 1) used for experimentation
are given in Table 2. The engine speed used was 1500 RPM with a compression
ratio of 18. Eddy current dynamometer was used for load variation. The tests were
carried out first for diesel fuel and then with different blends of Karanja oil biodiesel
(B5, B7, B10, B20). The CO, HC and NOx emissions were measured with the help
of Saj make HG 540-5 channel gas analyzer. CO is measured by NDIR absorption
principle. In this method, CO gas is passed through the sample cell and air is passed
162 V. R. Patil et al.
Table 2 Specifications of
Sr. No. Descriptions Specifications
diesel engine set up
1 Make Kirloskar TV1 Engine
2 Power 3.5 kw
3 Compression ration 18
4 Speed 1500 RPM
5 No. of cylinder 1
6 No. of stroke 4
7 Cylinder diameter (Bore) 87.5 mm
8 Stroke length 110 mm
6 Orifice diameter 20 mm
7 Dynamometer arm length 185 mm
A Study on Performance and Emission Characteristics of Diesel … 163
through the reference cell. In sample gas cell nondispersive infrared rays are absorbed
by CO content. In reference, cell air is passed and there is no absorption of rays.
This variation is detected in the detector cell and with the help of photovoltaic cell
and electric signal it is measured. Zirconia sensor is used to measure oxygen in
exhaust gas which is based on solid-state electrochemical fuel cell. Its two electrodes
provide output voltage corresponding to oxygen in the exhaust gas. HC is measured
by FID method. Mass air flow sensor was used to record air flow rate. Various
engine parameters like brake thermal efficiency, brake specific fuel consumption
and emissions like CO, HC and NOx were measured and the results were plotted.
Enginesoft Software was used to capture the results.
It is important to emphasize that the lower blend fuels and the diesel fuel that was
used as reference were evaluated under the same conditions. Density, calorific value
and viscosity of B5 and B7 found to be closer to that of diesel, whereas for B10 blend
these properties are slightly higher than diesel. The increase in density as well as
viscosity of the blends (B5, B7, B10, B20…) with increase in proportion of biodiesel
was observed. It was found that the viscosity of B5, B7 and B10 is comparable with
diesel. For B20 and above, i.e. higher blends may create filter clogging problem and
poor atomisation. The calorific value decreases with an increase in blend proportion.
A cetane number of lower blends are also close to diesel.
The variation of brake thermal efficiency of lower blends of biodiesel (B5, B7, B10),
B20 and diesel against engine load is shown in Fig. 2. Brake thermal efficiency is
almost same as diesel at part load for B5 and B7 blend. For maximum load brake,
thermal efficiency of (B5, B7, B10) is high compared to diesel. Increase in brake
thermal efficiency for B5, B7, B10 was found to be 5%, 12% and 8% than diesel
as there is more oxygen chemically bonded with oxygen in biofuel, which is an
additional source of oxygen other than oxygen present in the intake air which helps
in better combustion.
Figure 3 shows the variation of Break Specific Fuel Consumption (BSFC) of all
blends and diesel against engine load. The BSFC was found to be decreasing with
164 V. R. Patil et al.
BTE in (%)
15 B5%
10 B7%
B10%
5
B20%
0
0 5 10 15
Load in (kg)
an increase in the engine load. It was noted that BSFC was lower than diesel fuel for
part load and is nearer to diesel fuel for lower blends at maximum load as biofuel
blend combustion efficiency is slightly better than diesel.
The exhaust gas emission was measured using Saj make HG-540 5-Channel Gas
Analyzer with the specifications listed in Table 3. It measures CO2 , CO, HC con-
centrations in the exhaust gas by non-diffractive infrared radiation (NDIR). O2 and
NOx measurement by electrochemical method.
3.2.1 Hydrocarbon
Figure 4 shows the variation of HC emission of lower blends of biodiesel, B20 and
diesel against engine load. Results show that the HC emission for B5, B7 and B10
A Study on Performance and Emission Characteristics of Diesel … 165
Fig. 4 Variation of 50
Hydrocarbon (HC) of lower
Karanja oil blends with 40
engine load B00%
30
HC in (%)
B5%
20
B7%
10 B10%
0 B20%
0 5 10 15
Load in (kg)
blends is low compared to diesel fuel for maximum load condition. HC emission for
B20 was also lower than diesel at maximum load. Reduction of HC for B5, B7 and
B10 was 60%, 50% and 20% as compared to diesel. Reduction of HC for B20 was
also 10% than diesel. The cetane number of ester-based fuel is higher than diesel, it
reduces the delay period and results in better combustion.
0.08
B5%
0.06
B7%
0.04
0.02 B10%
0 B20%
0 10 20
Load in (kg)
166 V. R. Patil et al.
2000
1500
Nox in (ppm)
B00%
1000 B5%
B7%
500
B10%
0 B20%
0 5 10 15
Load in (kg)
Figure 6 shows the variation of NOx emission of lower blends of biodiesel, B20 and
diesel against engine load. A higher reduction of NOx was observed for all lower
blends compared to diesel fuel at part load as well as full load conditions. These
lower NOx emissions might be due to lower temperatures in the combustion chamber.
Reduction of NOx emission for B5, B7, B10 was almost 15–17%. Trend explains
as the concentration of blend increases NOx percentage may increase. Formation
of NOx does not only depend on temperature of combustion but also on content of
nitrogen in the fuel itself. As nitrogen content in biodiesel is less, its emissions are
less. Also, inconsistencies appear among studies as engine type, engine technology
and fuel feedstock too.
4 Conclusion
An experimental investigation has been conducted in this work to find the perfor-
mance and emissions of lower blends of Karanja biodiesel (B5, B7, B10) and B20.
Results are compared with diesel. Test results showed that the brake thermal effi-
ciency of all lower blends is high compared to diesel at maximum load. Brake specific
fuel consumption of all lower blends is nearer to diesel fuel. Considerable reduction
in hydrocarbon emissions was observed. CO emission for B5, B7 and B10 blends
is low compared to diesel fuel. There was a considerable reduction in NOx for B5,
B7, B10 as compared to diesel fuel as well as a B20 blend. Thus, considering future
stringent norms of emission and need for alternative fuel it will be better to use the
lower blends of biodiesel for diesel engines.
A Study on Performance and Emission Characteristics of Diesel … 167
References
Abstract This study compared friction stir welding undertaken parameters such as
tool rotation speed, transverse speed, and tool shoulder diameter on weld joint and
compared ultimate tensile strength weld joint by underwater friction stir welding and
normal friction stir welding using vertical machining center. For statistical analysis,
the Taguchi L9 method was adopted. Ultimate tensile strength is tested as a response
which shows that the ultimate tensile strength achieved by the underwater friction
stir welding was higher than the normal friction stir welding. Regression analysis and
analysis of variance were used as a statistical analysis method. From the analysis,
it was observed that high tool rotation speed with minimum tool transverse speed
gives maximum tensile strength for underwater friction stir welding and that of
normal friction stir welding.
1 Introduction
Friction stir welding (FSW) was carried out in four stages; they are plunging stage,
the dwelling stage, the welding stage, and the tool exit or retracting stage [1]. FSW
process leads to the softening of the grains due to dissolution and coarsening of the
grains which indirectly leads to a decrease in the mechanical properties of welded
joints; for resolving these issues, underwater friction stir welding (UFSW) method
is used which improves the cooling rate and lower pick temperature [2].
In comparison with other conventional methods of welding, only FSW is the
method that consumes a considerable amount of energy. There is neither usage of
flux nor cover gas, thus making the process eco-friendly [3]. The process of FSW
is simply carried out by plunging a rotating FSW tool into the interface of two
sheets which are rigidly clamped, till the shoulder riches out to the surface of the
material being welded [4]. The movement of the tool transverses along the weld line
as well as it transverses along the joint line which causes the material movement
from advancing side to retracting side, and tool shoulder consolidates the material
at the backside of the pin leading to a solid state [5]. The surrounding environment
underwater, as well as an increase in travel speed, helps to increase the precipitation
volume fraction, and by these, it reduces the average grain and precipitation size of
the weld nugget [6].
2 Experimental Procedure
To perform the welding Al6061 was selected as work material 150 × 90 × 6 mm.
H13 tool steel was selected for this welding material for the tool. There was a 6 mm
diameter of tool pin, length of the pin was 5.5 mm, and the tool shoulder diameter
was 22, 24, and 26 mm. The friction stir welding was completed with and without
the use of water as a working medium and compared the result of them in the term
ultimate tensile strength (UTS). The experiment setup is shown in Fig. 1.
The experimental work was completed at Vertical Milling Center PX20. UTS of
the welded specimen will be tested according to ASME Section IX [7].
Effect of response characteristics ANOVA was used to find the significant vari-
ables. The main effect plots and their interaction plots were plotted using a statistical
analysis method for each parameter at all levels which were used to examine the
parametric effects on the response’s characteristics. All the results were analyzed
using “Minitab 18”—statistical software (Table 1).
The study aims to identify the effect of friction stir welding parameters on ulti-
mate tensile strength that was conducted using the Taguchi L9 method. Main
effect plots and interaction plots were plotted using ANOVA for each parameter
at all levels. The main effect plots were used to examine the parametric effects
on the responses characteristics. All the results were analyzed using “Minitab
18”—statistical software.
UTS of the welded specimen will be tested according to ASME Section IX. The
Result of the tested welded component is as shown in Table 2.
Figure 2 shows that the behavior of UTS is low rotation speed gives maximum
strength that of high rotation speed. Maximum UTS is achieved from 24 mm shoulder
diameter and shows UTS decreases as tool shoulder diameter increases. For lower
tool transverse speed, maximum UTS was achieved.
The optimum parameters were tool rotation speed 2250 rpm, tool transverse speed
10 mm/min, and tool shoulder diameter 24 mm for UWFSW. Table 3. shows that the
affecting parameters on UTS were investigated using ANOVA.
130
120
110
Mean
100
90
80
70
60
50
2250 2500 2750 22 24 26 10 15 20
At lower tool rotation speed, UTS was maximum, and at higher tool rotation speed,
UTS was minimum. The maximum UTS was achieved at lower tool transverse speed.
For tool shoulder diameter, the UTS was maximum achieved at higher tool shoulder
diameter (Fig. 3).
The optimum process parameter for the friction stir welding without the use of
water was tool rotation speed 2250 rpm, tool transverse speed 10 mm/min, and tool
shoulder diameter 26 mm. Table 4 shows that the affecting parameters on UTS were
investigated using ANOVA.
Fig. 3 Main effect plot for UTS for friction stir welding without use of water
To identify the best properties achieved by the process with or without the use of
water, comparison of both processes must be required. For better understanding or
better comparison of both processes, a graphical method was used. The bar chart was
generated for both processes. The chart is as shown in Fig. 4.
From the chart, it is cleared that the UTS achieved by the UWFSW was higher
than the UTS achieved by the FSW. The maximum UTS achieved by the use of
water was 154 N/mm2 and by the friction stir welding without the use of water was
97 N/mm2 .
10 Conclusions
This research work presents an experimental comparison between friction Stir weld-
ing and underwater friction stir welding on Al6061 alloys. The ANOVA was per-
formed for the analysis of the work, and also, regression was done; the following
conclusions were achieved:
1. UTS decreases as tool rotation speed increased, and UTS increases as tool rotation
speed decreased.
2. At lower tool transverse speed, UTS has achieved maximum, and tool transverse
speed increases, and UTS of the underwater friction stir welding decreases.
3. Tool shoulder diameter increases UTS, and in UWFSW, UTS increases up to a
certain level after that level and then decreases.
4. In FSW, at lower tool rotation speed, UTS increases, and as tool rotation speed
increases, UTS decreases.
176 H. J. Mistry et al.
5. At the lower tool transverse speed, UTS was maximum; as tool transverse speed
decreases, UTS also decreases up to a certain level; further increment of tool
transverse speed results in increment of UTS.
6. In friction stir welding, as the tool shoulder diameter increases, UTS also
decreases up to a certain level; after that level, UTS increases.
7. From various experiments performed, the maximum UTS was achieved at
2250 rpm tool rotation speed, 10 mm/min tool transverse speed, and 24 mm
tool shoulder diameter for the underwater friction stir welding, and for the FSW,
without use of water, maximum UTS was achieved at 2250 tool rotation speed,
10 mm/min tool transverse speed, and 26 mm tool shoulder diameter.
8. From the results and analysis, it is clear that the UTS achieved by underwater
friction stir welding is higher than the underwater friction stir welding.
9. According to ANOVA, most significant parameters are tool rotation speed and
tool transverse speed which affect the UTS of underwater FSW and FSW,
respectively.
References
1. Mishra RS, Ma ZY (2005) Friction stir welding and processes. Mater Sci Eng, R 50(2005):1–78.
https://doi.org/10.4236/wjet.2018.62025
2. Kumar K, Kailash SV (2008) The role of friction stir welding tool on material flow and weld
formation. Mater Sci Eng A 485:367–374. https://doi.org/10.1016/j.msea.2007.08.013
3. Wahid MA, Siddiquee AN (2018) Review on underwater friction stir welding: a variant of
friction stir welding with great potential of improving joint properties. Trans Nonferrous Met
Soc China 28:193 − 219. https://doi.org/10.1016/s1003-6326(18)64653-9
4. Clark TN (2005) An analysis of microstructure and corrosion resistance in underwater friction
stir welded 304 l stainless steel. Brigham Young University, Provo. https://doi.org/10.1007/
s11771-012-12-2
5. Lingaraju D, Salavaravu L (2016) A review on underwater friction stir welding modified with
normal friction stir welding setup. IJARSE 5(10)
6. Mofid MA, Abdollah-Zadeh A, Ghaini FM, Gür CH (2012) Submerged friction-stir welding
(SFSW) underwater and under liquid nitrogen: an improved method to join Al alloys to Mg
alloys. Miner Met Mater Soc ASM Int. https://doi.org/10.1007/s11661-012-1314-2
7. Liu HJ, Zhang HJ, Yu L (2014) Homogeneity of mechanical properties of underwater friction
stir welded 2219-T6 aluminum alloy. J Mater Eng Perform 20(8). https://doi.org/10.1007/
s11665-010-9787-x
8. Rui-Dong F, Zeng-Qiang S, Rui-Cheng S, Ying L, Hui-jie L, Lei L (2011) Improvement of
weld temperature distribution and mechanical properties of 7050 aluminum alloy butt joints
by submerged friction stir welding. Mater Design 32:4825–4831. https://doi.org/10.1016/j.
matdes.2011.06.021
Experimental Comparison Between Friction Stir Welding … 177
1 Introduction
Wear Particle Examination is one of the most important factors to monitor machine
condition. Kumar [1] discussed wear particle morphology like shape, size, color, and
texture which can be used to analyze the machine condition. Information obtained
from wear particle analysis can be used to detect abnormal wear which helps to
detect mode and mechanism of wear and wear severity. It can also help to give
insight into the condition of machine and decide the maintenance period [2]. Wear
debris is classified as rubbing, laminar, cutting, chunk (surface fatigue), sliding. There
are few techniques used for wear particle analysis like ferrography, spectroscopic
oil analysis, atomic absorption spectroscopy, _atomic emission spectroscopy, etc.
Kumar and Kumar [3] studied that data collected from wear particle analysis is
further studied under optical microscope or scanning electron microscope to check
the behavior of different wear particles. The images captured through this study acts
as an input to fractal analysis.
The images captured through this study act as an input to fractal analysis. Fractal
analysis is used to compute fractal dimension which provides characterization of
wear particle like graphic analysis, shape classification, and texture segmentation
(profile and edge detection). The generalized formula to calculate fractal dimension
is [4]:
D =1−m (1)
where
D: Fractal dimension,
m: slope of log(perimeter length) versus log(step-length),
There are different wear particle size and shape descriptor discussed by Raadnui
[5], which are mentioned in Table 1.
Amongst all these methods, most commonly used method is fractal analysis as it
gives overall wear particle morphology which can be used to detect failure mechanism
and maintenance period [6].
Ghosh et al. [7] calculated fractal dimension of wear particle using image vision
system and applied fractal mathematics to check gearbox condition. After analysis,
it is concluded that the condition of gearbox was severe and researches suggested
to change the gearbox oil. Lope and Betrouni [8] discussed the techniques used
to calculate the fractal dimension for fractal and multi-fractal surfaces and their
applications in the field of medical science. Debnath [9] discussed the brief history
of fractals, fractal geometries, and calculation of fractal dimension. He has discussed
few applications of fractal in the field of fracture mechanism and turbulence.
Kang et al. [10] applied fractal analysis for ongoing process monitoring of tool
wear with the help of fractal characteristics of machined surfaces. Researchers com-
pared the variation in fractal dimension value and surface roughness value with
change in cutting conditions. Tool wear showed an increase in fractal dimension
Wear Particle Analysis Using Fractal Techniques 181
value along with increasing surface roughness. Shah and Hirani [11] carried out
wear debris analysis on spur gearbox to study the occurrence of type of wear parti-
cle. The analysis showed the presence of pitting and fatigue wear. Increase in wear
particle and acid number showed the degradation of lubricant additives. Kirk et al.
[12] applied fractal descriptors to wear particle boundaries. They concluded that
fractal techniques can be used in machine condition monitoring to detect the change
in wear process as fractal dimension is a function of wear progression involved.
Stachowiak et al. [13] carried out oil analysis using ferrography for unused lubri-
cating oil, which showed the presence of silicon-based mineral particle, aluminum-
based mineral particle, and flat iron-rich metallic particle. Researchers applied bound-
ary fractal analysis to calculate the texture and structural fractal dimension of these
particles. So et al. [14] proposed a method named novel box counting method to
estimate the fractal dimension of arbitrary-sized image. This method helped to over-
come the problems related to conventional box counting method like problem with
large window size is number of data points are not enough.
Li et al. [15] presented a new box counting approach to improve accuracy of fractal
dimension. Researchers proposed an algorithm that can suggest minimum number of
boxes to cover the required image for all selected scale. Gonzata et al. [16] explained
visual screen box counting method to determine fractal dimension. Hong-tao and
Shi-rong [17] applied fishbone graph fractal method to depict boundaries of wear
debris.
Along with this, fractal analysis is applied to study the flow of river known as
fractal geography of rivers, to calculate area of islands, mountains, lakes known as
fractal nature of the earth, to calculate the perimeter of coastline known as fractal
nature of coastline, to construct cables and bridges, to study the complexity of maps,
to quantify, describe and diagnose cancer known as fractal medicine.
2 Fractal Analysis
Fractal geometry of wear particle is used to represent condition of wear for any
tribological system [18]. Large value of fractal dimension indicates more complex
surface [19]. Procedure to carry out the fractal analysis is mentioned in Fig. 1.
Fractal is a uneven geometric object which can be divided into subparts, each
one of which is a reduced-size replica of the whole. Fractals are self-similar but do
not depend on scale. Fractal structures are used to describe any complex shape in
the world. Multifractal is a set of tangled fractals. Self-similarity of multifractals
depends on scale. There are different algorithms used to determine fractal dimension
of any image. Few of the algorithms are listed in flowchart shown in Fig. 2.
Some of the techniques to calculate fractal dimension are discussed in this paper
as follows:
182 P. P. More and M. D. Jaybhaye
Image Processing
This method is used for connected as well as non-connected forms of images. Fractal
dimension D is calculated by covering entire three-dimensional surfaces with grid of
Wear Particle Analysis Using Fractal Techniques 183
squares. For this, sufficient amount of date is required, approximately 20 data points
with self-similarity up to second or third order of magnitude.
N (β)
q
Mq (β) = Pi (3)
j=1
This method works on the idea of self-similarity. Fractal dimension for this method
for boundary set can be defined as
1
D = lim log(Nr ) log (4)
r →0 r
n r (i, j) = l − k + 1 (5)
Nr = n r (i, j) (6)
ij
where r is scale, nr (i, j) is number of boxes covered (i, j)th block, k and l are minimum
and maximum gray level in (i, j)th block. Fractal dimension can be calculated from
the least square linear fit of log(N r ) versus log(1/r). This method has limitation
of over-counting or under-counting the box height, number of boxes, and image
intensity [15].
HP EP 90 gear oil was collected after 500 h of running gear box. Quantitative and
qualitative oil analyses were carried out on collected sample. Quantitative analysis
using DR ferrography showed the presence of large- and small-sized wear particles.
Concentration of large-sized particle (DL ) was 156.4 and concentration of small-
sized particle (Ds ) was 123.2. As the concentration of wear particles was high in
184 P. P. More and M. D. Jaybhaye
given sample, qualitative analysis was carried out using analytical ferrography to
prepare ferrogram slide to study morphology of wear particles using bichromatic
microscope. Images captured from bichromatic microscope were used to calculate
fractal dimension using Fractal Box Count Tool from ImageJ software.
Initially, the required image whose fractal dimension needs to be calculated is
selected in ImageJ tool [20]. After that the image is converted into 8-bit binary
image. Fractal Box Count Tool is selected from analyze tool bar, which is used to
select the box sizes for calculation part of fractal dimension. Change in box sizes
affects the value of fractal dimension. Following steps are used to calculate fractal
dimension of any image using ImageJ software, Fig. 3.
The image was captured at 10× magnification factor as shown in Fig. 4a. Original
image was converted to 8-bit binary image as shown in Fig. 4b. Then, box size is
selected as 2, 3, 4, 6, 8, 12, 16, 32, 64.
Table 2 shows the sizes of boxes and number of boxes occupied by wear particle
for respected number of boxes. The notation 2, 3, 4, … n shows box sizes and D
indicates the value of fractal dimension.
Graph 1 shows the values of log(box sizes) plotted against log(count), where box
size is the size of individual boxes which used to measure any object and count refers
to the number of boxes occupied by given image for respective box size. Slope of
the line in Graph 1 gives the value of fractal dimension. Fractal dimension value for
given image of normal rubbing wear particle is 1.886.
Wear Particle Analysis Using Fractal Techniques 185
Fig. 4 a Original image of normal rubbing wear particle, b 8-Bit binary image of normal rubbing
wear particle
The image was captured at 50× magnification factor as shown in Fig. 5a. Original
image was converted to 8-bit binary image as shown in Fig. 5b. Then, box size is
selected as 2, 3, 4, 6, 8, 12, 16, 32, 64. As the original image of red oxide is blur,
boundaries are not clearly separated after converting the image to 8-bit binary. Red
oxides are formed due moisture content in lubricating oil. They appear as orange–
reddish in color.
186 P. P. More and M. D. Jaybhaye
Fig. 5 a Original image of red oxide particle, b 8-bit binary image of red oxide particle
Table 3 shows the values of number of boxes corresponding to box size for red
oxide.
Graph 2 shows the log of box size versus log of count for red oxide particle.
Fractal dimension of red oxide is 1.852.
4 Conclusions
Fractal geometry is a powerful tool used to characterize and segment the wear par-
ticles in many mechanical systems. Fractal analysis is used to illustrate shape, tex-
ture/surface complexity of wear, and particle boundary. Few of the techniques used
to determine fractal dimension are discussed in this paper. Based on literature survey,
it is suggested that box counting algorithm is extensively used to calculate fractal
dimension due to its easy understanding and simplicity.
ImageJ software is used to estimate fractal dimension of these wear particle using
Fractal Box Count Tool. Fractal dimension is calculated for two different wear parti-
cles that are normal rubbing wear particle and red oxide. As these particles are neither
one-dimensional nor \two-dimensional, therefore, fractal dimension value for both
particles is between 1 and 2. By calculating fractal dimension for large data of dif-
ferent types of wear particles, they can be distinguished using the fractal dimension
ranges.
References
1. Kumar M (2013) Advancement and current status of wear debris analysis for machine condition
monitoring: a review. Ind Lubr Tribol 65:3–11
2. Stachowiak GW (1998) Numerical characterization of wear particles morphology and
angularity of particles and surfaces. Tribol Int 31:139–157
3. Kumar C Kumar M (2016) Wear debris analysis using ferrography. Int J Recent Trends Eng
Res 2(8):398–404
4. Podsiadlo P, Podsiadlo GW (2000) Scale-invariant analysis of wear particle surface morphol-
ogy: II. Fractal Dimension Wear 242:180–188
5. Raadnui S (2005) Wear particle analysis—utilization of quantitative computer image analysis:
a review. Tribol Int 38(10):871–878
6. Kirk TB, Panzera D, Anamalay RV, Xu ZL (1995) Computer image analysis of wear debris for
machine condition monitoring and fault diagnosis. Wear 181:717–722
7. Ghosh S, Sarkar B (2005) Wear characterization by fractal mathematics for quality improve-
ment of machine. J Q Maintenance Eng 11(4):318–332
8. Lopes R, Betrouni N (2009) Fractal and multifractal analysis: a review. Med Image Anal
13:634–649
9. Debnath L (2006) A brief historical introduction to fractals and fractal geometry. Int J Math
Educ Sci Technol 37:29–50
10. Kang MC, Kim JS, Kim KH (2005) Fractal dimension analysis of machined surface depending
on coated tool wear. Surf Coat Technol 193(1–3):259–265
11. Shah H, Hirani H (2014) Online condition monitoring of spur gears. Int J Condition Monit
4:1–8
12. Kirk TB, Stachowiak GW, Batchelor AW (1991) Fractal Parameters and computer image
analysis applied to wear particles isolated by ferrography. Wear 145:347–365
13. Stachowiak GW, Kirk TB, Stachowiak GB (1991) Ferrography and fractal analysis of
contamination particles in unused lubricating oils. Tribol Int 6:329–334
14. So GB, So HR, Jin GG (2017) Enhancement of the box-counting algorithm for fractal dimension
estimation. Pattern Recognit Lett 98:53–58
15. Li J, Du Q, Sun C (2009) An improved box-counting method for image fractal dimension
estimation. Pattern Recogn 42(11):2460–2469
188 P. P. More and M. D. Jaybhaye
16. Gonzato G, Mulargia F, Marzocchi W (1998) Practical application of fractal analysis: problems
and solutions. The Charlesworth Group 132:275–282
17. Hong-tao L, Shi-rong G (2009) Fishbone graph fractal description to UHMWPE wear debris
boundary. Tribol Int 42(11–12):1624–1628
18. Shirong G, Guoan C, Xiaoyun Z (2001) Fractal characterization of wear particle accumulation
in the wear process. Wear 251(1–12):1227–1233
19. Klinkenberg B (1994) A review of methods used to determine the fractal dimension of linear
features. Math Geol 26(1):23–46
20. Karperien A (2004) FracLac advanced user manual. Charles Sturt University, Australia
Strategies for Low Engine Speed Torque
Enhancement of Natural Gas Engine
Used for Commercial Vehicles:
Observations with Compression Ratio
Abstract Since many years diesel engines are powering commercial vehicles. Of
late, governments are promoting the use of natural gas (NG) as a fuel for such
vehicles to reduce pollution. Thereafter, natural gas engines have witnessed faster
development, especially for use in commercial vehicles. City bus is probably the
most common NG commercial vehicle, dedicated to ferry passengers across the city.
Some places impose a safety speed limit on such vehicles considering local traffic
conditions. Thus, a typical scenario faced by such vehicles includes low drive speeds,
high loads, frequent halts for passenger pick up and drop, signals, etc. Such vehicles
thus need high torque at low engine speeds to manage these daily occuring conditions.
Aim of this paper is to enhance torque at low-engine speed zone. Number of options
such as the use of a turbocharger, direct injection of fuel, variable valve actuation,
programmable waste-gates, etc. can help to realize higher engine outputs. The intent
here is to study the effect of compression ratio and understand the extent of change
in torque in engine low-engine speed region. Current study consists of modelling a
reference commercial vehicle engine of six cylinders. A virtual model is built and its
ability to represent actual engine performance from testbed is verified. Further, such
model undergoes iterations of change in compression ratio and different parameters
are studied for their relation with torque.
Table 1 Specifications of
Bore × Stroke (mm) 97 × 128
base engine
Number of cylinder 6 Cylinder, Inline
Firing order 1st-5th-3rd-6th-2nd-4th
Rated speed 2500 RPM
Swept volume 5.7 L
Aspiration Natural
Emissions BS-IV compliant
Fuel Natural Gas, spark-ignited
1 Introduction
Urban dwellings are home to all categories of vehicles. Based on the consumption
pattern, it has been observed that personal vehicles demand comparatively more
energy consumption and are more energy demanding than public transport. Use
of natural gas is an economical alternative which has less polluting, have ease of
control, delivers better combustion, etc. [1–3]. Compressed Natural Gas engines
for commercial vehicles are redesigned from their diesel counterparts. Although
this approach is relatively easy and quick, it imposes a number of limitations on
performance as not all parts are designed taking natural gas into consideration [4–
7]. A basic thermodynamic principle mentions that compression ratio plays a very
important part in deciding output of an engine. Compression ratio, although a function
of clearance volume, needs consideration of many parameters such as design limits,
ability to withstand high temperatures, engine knocking, etc. [8–12].
Initially, a software model of reference engine was built and compared against per-
formance data from actual engine testbed performance test. In Fig. 2, close match
is seen between in-cylinder pressure trace from actual measurement as well as a
simulated one. Other parameters of engine performance significance are highlighted
in Table 2. There was a reasonable agreement in different parameters, the model is
Strategies for Low Engine Speed Torque Enhancement of Natural … 191
NA Engine Performance
400 400
Torque Nm, Power kW
kg/hr
200 200
Torque SFC
Power kW Air_Flow
100 100
0 0
700 1200 1700 2200 2700
Engine speed RPM
Cylinder pressure
60
50 Meas
Pressure bar
40 Sim
30
20
10
0
-40 -20 0 20 40 60 80 100 120
Crank angle degrees
said to be calibrated and ready for iterating different setups such as valve timings,
compression ratios, etc.
Compression Ratio Study: It signifies the extent to which the fluid is compressed
after entering the cylinder. Study here investigates the effect of compression ratio
192 P. J. Suple et al.
Torque
420
400
Torque (Nm)
380
360
340
10.5 11.5 12.5 13.5 14.5
320 15.5 16.5 17.5 18.5
300
700 1200 1700 2200 2700
Engine speed RPM
Fig. 3 Comparision of Torque produced at various engine speed for various values of fixed
compression ratio
2.75
BMEP (bar)
8
FMEP (bar)
2.25
10.5 11.5 12.5
13.5 14.5 15.5 1.75
7 16.5 17.5 18.5
1.25
6 0.75
700 1200 1700 2200 2700
Engine Speed RPM
Fig. 4 Comparision of BMEP and FMEP produced at various engine speed for various values of
fixed compression ratio
Strategies for Low Engine Speed Torque Enhancement of Natural … 193
51
Exhaust Energy
49
Exhaust energy (%)
47
45
43
41
39
37 10.5 11.5 12.5 13.5 14.5
15.5 16.5 17.5 18.5
35
700 1200 1700 2200 2700
Engine Speed RPM
Fig. 5 Comparision of Exhaust energy (%) expelled at various engine speed for various values of
fixed compression ratio
400
34
200 32
0 30
700 1200 1700 2200 2700
Engine speed RPM
Fig. 6 Comparision of Exhaust Temeprature and Brake Efficiency (%) at various engine speed for
various values of fixed compression ratio
3 Conclusions
It has been observed that the use of higher compression ratio engine helped to enhance
the low-speed torque by about 6.5%. In all, it can be inferred that this enhancement
to low engine speed torque is due to higher compression ratio originates principally
from better utilization of fuel and enhanced thermodynamic expansion process.
Practical constraints permitted engine testing with 12.5 CR only. Engine perfor-
mance as compared to simulation can be seen as matching in Fig. 7. Hence findings
from other simulations can be taken into consideration. They can be summarized in
Table 3.
It is to be noted that, at higher compression ratio engine may knock especially at
higher engine speeds. Thus, one cannot increase compression ratio beyond a limit.
194 P. J. Suple et al.
100
Torque (Nm)
Power (kW)
300
80
200
60
100 Msr Torque Nm Sim Torque Nm 40
Msr Power kW Sim Power kW
0 20
700 1200 1700 2200 2700
Engine speed RPM
Fig. 7 Simulated and measured engine output for 12.5 compression ratio
Most common means to control knocking, i.e. retarding ignition, especially at high
speeds and loads, is not recommended as it leads to loss of efficiency, higher exhaust
temperatures, and increases emissions. Other ways to enhance torque at low engine
speeds are to be explored if higher torque is expected.
References
1. Reddy BS (1995) Transportation, energy and environment: a case study of Bangalore. Econ
Polit Wkly 30(3):161–170
2. Parker RS, Pettijohn CE (1997) The use of alternative fuels in the private trucking industry: is
there a viable target market? J Mark Theory Pract 5(4):88–93. (Fall) https://doi.org/10.1080/
10696679.1997.11501783
Strategies for Low Engine Speed Torque Enhancement of Natural … 195
3. Frost & Sullivan (2016) HD transit bus market—global analysis. In: Presentation in market
engineering, NF63-18
4. Jacob J, Abhimanyu Y, Kumar AH, Singh S (2014) Strategic analysis of compressed natural
gas (CNG) passenger cars market in Europe. In: Presentation in market engineering, MA30-18.
Frost & Sullivan
5. Gharehghani A, Koochak M, Mirsalim M, Yusaf T (2013) Experimental investigation of thermal
balance of a turbocharged SI engine operating on natural gas. Appl Thermal Eng 60:200–207.
https://doi.org/10.1016/j.applthermaleng.2013.06.029
6. Thipse S, Dsouza A, Sonawane S, Rairikar S et al (2017) Development of multi cylinder
turbocharged natural gas engine for heavy duty application. SAE Int J Engines 10(1). https://
doi.org/10.4271/2017-26-0065
7. Kalam MA, Masjuki HH (2011) An experimental investigation of high performance natural
gas engine with direct injection. Energy 36:3563–3571. https://doi.org/10.1016/j.energy.2011.
03.066
8. Semin I, Ismail A, Bakar A (2009) Investigation of torque performance effect of development
of sequential injection CNG engine. J Appl Sci 9(13):2416–2423
9. Yan B, Wang H, Zheng Z, Qin Y et al (2017) The effect of combustion chamber geometry on
in-cylinder flow and combustion process in a stoichiometric operation natural gas engine with
EGR. Appl Thermal Eng. https://doi.org/10.1016/j.applthermaleng.2017.09.067
10. Yadollahi B, Boroomand M (2013) The effect of combustion chamber geometry on injection
and mixture preparation in a CNG direct injection SI engine. Fuel 107:52–62. https://doi.org/
10.1016/j.fuel.2013.01.004
11. Wu C, Deng K, Wang Z (2015) The effect of combustion chamber shape on cylinder flow and
lean combustion process in a large bore spark-ignition CNG engine. J Energy Institute 1–8.
https://doi.org/10.1016/j.joei.2015.01.023
12. Zhao J, Ma F, Xiong X, Deng J et al (2013) Effects of compression ratio on the combustion
and emission of a hydrogen enriched natural gas engine under different excess air ratio. Energy
59:658–665. https://doi.org/10.1016/j.energy.2013.07.033
In-house Fabrication and Calibration
of Silver Thin Film Gauge
Abstract Measurement of transient surface temperature and surface heat flux got
vast attention with recent technological advancement. Thin film gauge (TFG) is
the most suitable instrumentation for such measurements. Present work aimed to
fabricate and calibrate the silver TFG. In the present work, performance parameters of
TFG were estimated through static calibration. Further TFG was subjected to constant
heat flux using in-house fabricated calibration setup in conduction mode to assess
the thermal performance and reliability of silver TFG. The silver TFG measured the
transient surface temperature, which was discretized using cubic spline technique and
heat flux was recovered by considering one-dimensional heat conduction modeling
in semi-infinite solid. Results revealed that silver thin film gauge can be considered
for heat flux measurement for short time scale (s and ms) applications.
Keywords Transient surface temperature · Surface heat flux · Silver TFG · Cubic
spline · Semi-infinite
1 Introduction
The capability to capture the transient surface temperature and heat flux is a demand-
ing task in the present engineering problems. This task is easy if the experimentation
time scale is large. However, for a short duration, capturing the transient surface tem-
perature and heat flux is very difficult. TFGs are the most worthy sensors for such
short duration measurements. TFG is the resistance temperature detector (RTD) hav-
ing a metallic thin film of sensing material (platinum/silver/nickel) on a thermally
insulating substrate such as Macor, Quartz, and Pyrex. The working principle of TFG
is the same as RTD. The resistance of TFG varies linearly with temperature. The cal-
ibration parameters of TFG such as temperature coefficient of resistance (TCR) and
sensitivity can be determined using the oil bath calibration technique [1]. When TFG
is exposed to sudden heat load, the electrical resistance of TFG varies based on the
type of heat load. This resistance response can be converted to temperature signals
with the knowledge of TCR [2, 3].
Short duration surface heat flux can be predicted by inverse analysis using the
transient surface temperature history. The first step in inverse heat prediction is the
discretization of temperature signals. Various discretization techniques are available
in the literature [4, 5]. Heat flux can be predicted analytically by considering TFG as
a thin film mounted on the semi-infinite substrate. Details of fabrication technique,
calibration and heat flux prediction are given in the following sections.
In the present work, TFG was fabricated using silver as a sensing material on Quartz
rod of 6 mm diameter and 10 mm depth. The choice of silver was made by considering
good adhesion and low cost. Quartz was selected as a substrate due to its thermal
stability over the wide range of temperature up to 1000 °C. The first step in TFG
fabrication is substrate preparation. In this process, the Quartz rod was first polished
with the commercially available metallurgical grade silicon carbide sandpaper of 400
grit size (Made-3 M). The polishing process was continued using 1000 and then 2000
grit size sandpaper until the visible polished surface was observed by naked eyes.
This substrate was washed with ethanol and further treated in an ultrasonic cleaning
machine at a high frequency of 30 kHz (Made-Telsonic). This frequency is sufficient
to remove the adhered impurities from Quartz rod. This step is necessary to ensure
the impurity-free surface. Surface impurities may reduce the adhesion of the silver
thin film to the Quartz substrate [6]. Next step is the deposition of sensing film. In
this process, silver metallo-organic paint (SPI Supplies, USA) was used for making
the sensing thin film. This paint was first mixed with thinning agent n-butyl acetate
(SPI Supplies, USA) to make the paint of required viscosity. The thermophysical
properties of materials are given in Table 1. The sophisticated brush with fine hairs
was used to deposit the silver thin film on Quartz substrate. The thickness of the film
was kept as minimum as possible (below 10 μm). If the silver film is too thin, it
may break during testing, and if it is too thick, the effective resistance of TFG will
decrease. The decrement in resistance may cause the decrement in sensitivity, and it
may affect the inverse heat flux prediction.
The final step is an electrical connection. For this purpose, the silver films were
painted on a round surface of the substrate. These connecting thin films were kept
intentionally thick compared to sensing thin film. Thickening of connecting thin film
will reduce its resistance compared to the sensing thin film. So that resistance of
connecting thin film can be neglected. This painted substrate was annealed at 350 °C
in microcontroller-based muffle furnace for 30 min to stabilize the thin film. The
In-house Fabrication and Calibration of Silver Thin Film Gauge 199
tin-coated copper wire of 0.4 mm diameter was then soldered to connecting films
with sufficient care. Tin-coated copper wire was used to ensure the better structural
stability of the solder joint. This TFG is then covered with Teflon tape to provide
thermal insulation for connecting leads on the non-sensing surface. The in-house
fabricated silver TFG by this process is shown in Fig. 1. The effective resistance of
this silver TFG was found to be 1.4590 at a temperature of 30 °C.
The performance parameters of TFG were estimated using the oil bath calibration
technique. This is also referred to as static calibration. The experimental setup fabri-
cated for static calibration is shown in Fig. 2. The oil bath calibration setup consists of
air beaker fixed inside a silicone oil beaker placed on a flat plate heater using clamps.
The silicone was selected as a heating medium for air due to its chemical and thermal
stability over the wide range of temperature. Both the silver TFG and calibrated glass
thermometer were fixed at the same height using a clamp arrangement. Mismatch in
height may affect the calibration results due to different temperatures sensed by TFG
and thermometer. TFG was first activated by 10 mA current using a constant current
source (CCS). To examine the resistance response of silver TFG, the temperature was
gradually increased from room temperature of 30 °C to 65 °C. The resistance was
200 A. Jadhav and R. K. Peetala
recorded using the data acquisition system (DAS) at an interval of 5 °C during both
heating and cooling processes as shown in Fig. 3. The good linear behavior of TFG
was observed in static calibration. However, some hysteresis was observed. This is
due to the thermal expansion and contraction of sensing thin film during the heating
and cooling processes. The coefficient of correlation (r) was calculated using the
linear regression analysis, and it was found to be better than 0.90. This reflects the
usefulness of this TFG for accurate temperature measurement. The sensitivity (S) of
silver TFG was determined using the following correlation which is essentially the
slope of the resistance-temperature response of TFG [1].
R (R − R0 )
S = = (1)
T (T − T0 )
considered for further study for greater accuracy. The average sensitivity obtained
for the present silver gauge is 0.001235 /K. To estimate the TCR, the graph was
plotted between the resistance ratio R/R0 and temperature difference T as shown
in Fig. 4. The numerical value of TCR was calculated by considering the first-order
linear fit equation. This equation was compared with correlation (2), and the average
TCR was found to be 0.0008514 K−1 .
R
= 1 + αT (2)
R0
4 Dynamic Calibration
The aim of dynamic calibration is to access the thermal performance of silver TFG
when exposed to a constant heat flux. A special experimental device was fabricated for
this purpose as shown in Fig. 5. This device consists of a Teflon insulated copper rod
(k = 391 W/mK) having a heater at one end, and another end is free to environmental
conditions. The heat input to the copper rod can be controlled with the help of a
dimmerstat. Five K types of calibrated thermocouples are attached along the axial
direction of the rod to monitor the temperature along the length. During calibration,
the axial temperature distribution was continuously monitored until the constant heat
flux was achieved in the copper rod which is given by the correlation (3).
dT
qc = −k (3)
dx
202 A. Jadhav and R. K. Peetala
Here, qc is the constant heat flux established in the copper rod (W/m2 ); k is a
thermal conductivity of copper rod (W/mK), and dT dx
is a temperature gradient along
the length of copper rod (K/m). The constant heat flux of 1.5 kW/m2 was achieved in
the copper rod. The silver TFG was first activated with a constant current of 10 mA
using a CCS, and then, its gentle contact was made to the free surface of copper rod
for 1 s. The established constant heat flux was sensed by TFG, and its resistance
varied in a parabolic pattern. This transient resistance signal was then converted to a
transient temperature signal using correlation (2) as shown in Fig. 6. This parabolic
nature of the temperature signal confirms the constant heat flux condition.
Heat flux was predicted from transient temperature signal using inverse analysis.
In this analysis, heat transfer through TFG was modeled as heat transfer in semi-
infinite solid. The following assumptions were considered in this analysis. (i) The
temperature of the sensing element and substrate surface is the same due to the
thinness of film. (ii) Heat conduction takes only in the axial direction, and lateral
heat conduction is neglected. (iii) Thermophysical properties of sensing and substrate
materials are constant within the operating temperature range. (iv) Substrate is infinite
so that for test duration of 1 s, the temperature rise at infinity is zero. The general
solution for this problem was obtained by Duhamel’s superposition integral given
by the following correlation [4–7].
t
ρs cs ks 1 d{Ts (τ )}
q= √ dτ (4)
π t −τ dτ
0
In-house Fabrication and Calibration of Silver Thin Film Gauge 203
Fig. 6 Transient
temperature response of TFG
where q is the heat flux; ρ s , cs, and k s denote density, specific heat, and thermal
conductivity of substrate material, respectively; t and τ denote experimental time
and scaled time, respectively, and T S (τ ) denotes the surface temperature of the sub-
strate. To solve the above equation numerically, first temperature signal needs to
be discretized. In the present work, the temperature signal was discretized using
the third-order cubic spline technique. Heat flux recovered by this technique was
1.42 kW/m2 . The error in heat flux recovery was 5.33%.
5 Conclusions
The silver thin film gauge was fabricated in-house successfully. TCR and sensitivity
were obtained by static calibration. However, some fluctuations in resistance–tem-
perature linear behavior and transient temperature data were observed. Further, the
thermal performance of silver TFG was assessed through dynamic calibration. Heat
flux was under-predicted by silver TFG within the range of 6%. Finally, it was con-
cluded that silver TFG is advantageous due to the low cost of fabrication and better
adhesion of thin film to the substrate compared to platinum gauge but at the cost of
accuracy in the heat prediction. This TFG can be used for small time scale (s and ms)
applications but may not be useful for ultra-short duration (μs) applications such as
shock tubes and shock tunnels.
204 A. Jadhav and R. K. Peetala
References
1. Kinnear K, Lu F (1997) Design, calibration and testing of transient thin film heat transfer gauges.
In: 20th AIAA advanced measurement and ground testing technology conference, p 2504
2. Kumar R, Sahoo N, Kulkarni V, Singh A (2011) Laser based calibration technique of thin film
gauges for short duration transient measurements. J Therm Sci Eng Appl 3(4):044504
3. Smith DE, Bubb JV, Popp O, Diller TE, Hevey SJ (1999) A comparison of radiation versus
convection calibration of thin-film heat flux gauges. ASME-Publications 364:79–84
4. Sahoo N, Peetala RK (2011) Transient surface heating rates from a nickel film sensor using
inverse analysis. Int J Heat Mass Transf 54(5–6):1297–1302
5. Niranjan S, Peetala RK (2010) Transient temperature data analysis for a supersonic flight test.
J Heat Transf 132(8):084503
6. Sarma S, Sahoo N, Unal A (2016) Thin-film gauges using carbon nanotubes as composite layers.
J Eng Mater Technol 138(4):041014
7. Jadhav A, Peetala R, Kulkarni V, Multi-walled carbon nano-tubes for performance enhancement
of thin film heat flux sensors. Heat Mass Transf
Study of Shock Wave Boundary Layer
Interaction in Hypersonic Flows Using
Various Turbulence Models
1 Introduction
Shock wave boundary layer interaction (SWBLI) and shock-shock interactions are
the prime topics of research in supersonic and hypersonic flow analysis research [1].
SWBLI is commonly observed in various parts of the vehicle such as wing-body
junctions, inlets, wedge like sections etc. These interactions enhance the heating and
pressure loads on the vehicle body and can damage it [2]. Thus it is necessary to
predict these loads accurately in order to utilize these predicted loads for designing
suitable thermal protection system for the vehicle. For understanding SWBLI, ana-
lyzing flow over complete vehicle is not required but instead canonical geometries
such as ramp, wedge, double wedge are frequently used in the studies [3]. Supersonic
and hypersonic flow over flat plate, ramp has been analyzed experimentally by few
researchers to study the wall pressure variation, condition of incipient separation and
length of flow induced separation [4–6]. The response of turbulent boundary layer to
the shock wave has been studied by Smits and Muck [7]. The shock induced separa-
tion can also be studied by numerical modelling techniques and thus capabilities and
limitations of various models must be evaluated. Knight and Degrez [8] and Knight
et al. [9] reviewed capabilities of various RANS based Navier-Stokes codes to pre-
dict pressure field, skin friction, heat transfer, shock-induced separation extent for
both laminar and turbulent shock boundary layer interactions. Hirsch [10] reviewed
the numerical work done on SWBLI considering a variety of flow solvers, grids and
turbulence models for, particularly impinging-reflecting shock case. It is observed
from the literature, reattachment of the separated boundary layer is turbulent for
stronger shocks which rises in hypersonic flow regime. Thus it is necessary to study
the behavior of shock wave and turbulent boundary layer interaction occurring in
hypersonic regime. It is also observed that the RANS based turbulence models such
as Spalart-Allmaras one equation model and k–ω SST model are suitable in case of
flows with separated boundary layer. Thus the objective of the present investigation
is the study of shock induced boundary layer separation in hypersonic flow over a
2D ramp using various turbulence models.
Turbulence models such as k–ω SST and Spalart-Allmaras models are employed for
capturing the actual flow physics of shock induced separation because these models
are found to performing well in case of flows with separated boundary layer [13].
Flow over a single ramp model studied by Marini [14] has been adopted for the
present investigation. Computational domain consists of a 0.05 m flat plate with a
15° ramp attached to it. Figure 1 illustrates computational domain and boundary
conditions. Freestream conditions are mentioned in Table 1. At inlet freestream
values of pressure, velocity and temperature are provided. Zero gradient condition is
provided at the outlet while freestream condition is employed at the farfield boundary.
The wall is considered as isothermal and no slip.
Grid is generated with proper clustering near the corner of 2D ramp for capturing
the SWBLI effects. Mesh independence study is carried out by considering grid
sizes of 180 × 90, 240 × 120, 360 × 180, 420 × 240 cells. Minimum cell size is
considered as 1 × 10−05 m with 1.12 as expansion ratio. It is found that 240 × 120
cells give mesh independent solution. The grid and clustered section of 240 × 120
mesh is shown in Figs. 2 and 3 respectively.
The high-speed flow with M = 6 reaches the compression corner and there is
a formation of an oblique shock. The shock interacts with the boundary layer and
submits it to the adverse pressure gradient and causes flow reversal which ultimately
Table 1 Freestream
M∞ 6
parameters
Ho 1.08 MJ/kg
T∞ 131.7 K
P∞ 199.4345 Pa
Re∞ (unit) 8 × 105 m−1
208 A. Kane and R. K. Peetala
Fig. 2 Mesh
Fig. 4 Mach contour and enlarged view of the boundary layer separation
Study of Shock Wave Boundary Layer Interaction in Hypersonic … 209
Fig. 5 Coefficient of
pressure variation
is considered as per the literature [15]. But the inaccuracy occurring in the present
results indicates that the value of turbulent intensity is application specific and for
the current case lower value may provide appropriate results. Thus there is need to
study the effect of turbulent intensity on the shock induced turbulent boundary layer
separation.
210 A. Kane and R. K. Peetala
20
Bubble Size (mm)
16
12
0
Experimental Spalart Allmaras K-omega SST
4 Conclusion
The investigation of shock induced separation is carried out successfully using two
turbulence models. The two models capture the SWBLI effects qualitatively but
the quantitative mismatch for size of separation is observed as compared to the
experimental results. SA model predicts the separation length to be 13 mm and k–ω
SST model predicts very small separation length of 4 mm against the experimental
value of 17.75 mm. The error reported in the prediction of the lengths of the bubble
for SA and k–ω SST models are 26 and 77%. These large errors may be because of
the higher value of turbulent intensity of 5% considered which otherwise is in the
range of 0.5–1% for high velocities. The study will be extended to investigate the
effect of turbulence intensity on SWBLI.
Study of Shock Wave Boundary Layer Interaction in Hypersonic … 211
References
1. Bertin JJ, Cummings RM (2003) Fifty years of hypersonics: where we’ve been, where we’re
going. Prog Aerosp Sci 39(6–7):511–536
2. Gaitonde DV (2015) Progress in shock wave/boundary layer interactions. Prog Aerosp Sci
72:80–99
3. Délery J, Dussauge JP (2009) Some physical aspects of shock wave/boundary layer interactions.
Shock Waves 19(6):453–468
4. Gadd GE, Holder DW, Regan JD (1954) An experimental investigation of the interac-
tion between shock waves and boundary layers. Proc Royal Soc A Maths Phys Eng Sci
226(11650):227–253
5. Greber I, Hakkinen RJ, Trilling L (1958) Laminar boundary layer oblique shock wave inter-
action on flat and curved plates. J Appl Mathe Phys (ZAMP) 9:312. https://doi.org/10.1007/
BF02424755
6. Needham DA, Stollery JL (1966) Boundary layer separation in hypersonic flow. AIAA 4th
Aerosp Sci Meeting 66:66–455
7. Smits AJ, Muck KC (1987) Experimental study of three shock wave/turbulent boundary layer
interactions. J Fluid Mech 182(1)
8. Knight D, Degrez G (1998) Shock wave boundary layer interactions in high mach number
flows a critical survey of current numerical prediction capabilities. Agard Advisory. Report, 2
9. Knight D, Yan H, Panaras AG, Zheltovodov A (2003) Advances in CFD prediction of shock
wave turbulent boundary layer interactions. Prog Aerosp Sci 39(3):121–184
10. Hirsch C (2010) Lessons Learned from the first AIAA-SWBLI Workshop CFD Simulations
of Two Test Cases. AIAA Conference pp 1–14
11. Greenshields GJ, Weller HG, Gasparini L, Reese JM (2010) Implementation of semi-discrete,
non-staggered central schemes in a colocated, polyhedral, finite volume framework, for high-
speed viscous flows. Int J Numeri Meth Fluids 63(1):1–21
12. Kurganov, Tadmor E (2000) New high-resolution central schemes for nonlinear conservation
laws and convection-diffusion equations. J Comput Phys 160(1):241–282
13. Babinsky H, Harvey JK (2011) Shock wave–boundary-layer interactions. Cambridge Univer-
sity Press
14. Marini M (1998) Effects of flow and geometry parameters on shock-wave boundary-layer
interaction phenomena. AIAA J 98(1570):319–329
15. John B, Senthilkumar P (2018) Alterations of cowl lip for the improvement of supersonic-intake
performance. J Appl Fluid Mech 11(1)
Study of Effect on Engine Performance
Using 15% HCNG Blend Versus CNG
Using a Simulation Approach
Abstract The present work deals with modeling an existing CNG engine using a 1D
simulation tool (GT-SUITE), performing baseline validation of the model, generating
simulation results of HCNG (15% hydrogen by volume), study of simulated results
to bring out various trends and studying the effects of varying fuel–air equivalence
ratio on optimum spark advance, peak in-cylinder pressure and peak in-cylinder
temperature through simulation on validated engine model. This work is intended as
a preliminary step to bring about a host of potential suggestions, which will help to
improve the current power performance of the engine while using 15 HCNG as the
fuel. The primary objective is to develop 1D engine model using the actual engine
data as input and to generate simulation results. Actual experiments will be carried
out to generate experimental results. Simulation model will be validated for accuracy
within 10%. The validated engine model is now used for the purpose of simulation
to study the effects of varying fuel–air equivalence ratio on optimum spark advance,
peak in-cylinder pressure, and power performance for selected HCNG blend.
1 Introduction
Over the preceding decade, natural gas has turned out to be a demanding alternative
fuel for the growth of sustainable transportation. Heavy commercial vehicles plying
on natural gas represent a budding technology, while natural gas heavy-duty transport
vehicles are widely popular. As a fuel source for large scale automotive applications,
natural gas provides rewards in automotive technology because of its socioeconomic
benefits in comparison with fossil-fueled engines.
K. P. Kavathekar (B)
Symbiosis International University, Pune, India
e-mail: kavathekar.edl@araiindia.com; kavathekar.kishorkumar@gmail.com
K. P. Kavathekar · S. S. Thipse · S. D. Rairikar · S. B. Sonawane · P. S. Sutar · D. Bandyopadhyay
Powertrain Engineering Lab, Automotive Research Association of India (ARAI), Pune, India
7
8
2 6
Servo motor
5
1 3
4 Signal Conditioning
1. Cat-con 2. Intake
3. Exhaust 4. Muffler
5. HCNG Injectors 6. Gas Flow Meter
7. Pressure Regulator 8. HCNG Cascade
Exhaust
3 Experimental Setup
The CNG engine test setup was used to generate test data with HCNG blends as the
fuel. It can be seen from the experimental setup that, the blend (Hydrogen and CNG)
was supplied to the engine intake manifold [2]. The storage pressure of HCNG was
200 bars. The experimental setup schematics is as shown in Fig. 2.
A four-cylinder NA CNG-fueled engine was used for the test. Full throttle per-
formance (FTP) tests were carried out for 15% HCNG fuel throughout the engine
speed range. This FTP data was then used to validate the CNG 1D model with CNG
test data.
After validation of base CNG engine, and before using the HCNG model, it was first
necessary to optimize the spark advance for the given engine specifications and each
at each selected engine speed. Brake torque was selected as the dependent variable
to find the optimum spark advance at each selected engine speed. Optimum spark
advance at any given engine speed is that spark timing which yields maximum brake
torque [3].
216 K. P. Kavathekar et al.
After completion of the baseline validation of the model with CNG fuel, the ‘fuel
object’ attribute in fuel injector object of 1D model was altered for prediction of
HCNG blend (15% by volume).
190
170
CNG (equivalence ratio of 1)
150 CNG (equivalence ratio of 0.9 i.e. Lean)
15HCNG (equivalence ratio of 1)
15HCNG (equivalence ratio of 0.9 i.e. Lean)
130
800 1200 1600 2000 2400 2800 3200
Engine speed in RPM ----->
Fig. 3 Effect of fuel–air equivalence (φ) ratio on brake torque for 15% HCNG and CNG
Study of Effect on Engine Performance Using 15% HCNG Blend … 217
45
35
CNG (equivalence ratio of 1)
CNG (equivalence ratio of 0.9 i.e. Lean)
25 15HCNG (equivalence ratio of 1)
15HCNG (equivalence ratio of 0.9 i.e. Lean)
15
1000 1500 2000 2500 3000
Engine speed in RPM ----->
210
200
1000 1500 2000 2500 3000
Engine speed in RPM ----->
the subtle changes, which go into the process of converting a stoichiometric operation
actual engine into a lean-burn engine, are not incorporated in this engine model. In
Fig. 6, the maximum in-cylinder pressure falls (by 3.63% for CNG and 5.78% for 15
HCNG at the rated speed) with change of fuel–air equivalence ratio from 1 to 0.9.
Figure 7 shows that the optimum spark advance must be advanced slightly (for
both CNG and 15 HCNG fuels) across the entire speed range when the fuel–air
equivalence ratio is changed from 1 to 0.9. This can be attributed to the fact that the
flame speed is slightly lower under lean conditions as compared to stoichiometric
condition for both the fuels thereby increasing the combustion duration [4, 5]
218 K. P. Kavathekar et al.
60
55
bar ---->
Fig. 6 Effect of varying φ on maximum in-cylinder pressure for 15 HCNG and CNG
Fig. 7 Effect of varying φ on optimum spark advance for 15 HCNG and CNG
5 Conclusions
1. The principal outcome of this study was achieved by obtaining a validated engine
model in selected 1D simulation tool (GT-SUITE) for CNG as well as selected
HCNG blend.
2. The predicted results showed that there was an encouraging improvement in
brake-specific fuel consumption (BSFC) with a small but definite improvement in
brake power and in-cylinder pressure while using 15 HCNG as fuel as compared
to CNG-fueled operation.
3. For a given fuel, the optimum spark advance needs to be slightly advanced for
both, rich and lean mixtures as compared to stoichiometric operation.
Study of Effect on Engine Performance Using 15% HCNG Blend … 219
4. The baseline engine model (CNG as fuel) was validated experimentally with the
simulation results following the same trend as the experimental results and lie
within an accuracy of 6.8% (maximum) across the entire speed range.
Acknowledgements Author would like to thank to The Automotive Research Association of India
(ARAI), Pune, and Symbiosis International University, Pune, for their support and inspiration to
work up on this exclusive area. Author also thanks Mr. Suyash Gawade for his support in simulation
work.
References
1. GT Suite user guide, flow theory, GEM-3D and Engine Performance user manuals, Version v7.4
2. Kahraman N et al (2009) Investigation of combustion characteristics and emissions in a spark-
ignition engine fuelled with natural gas-hydrogen blends. Int J Hydrogen Energy 34:1026–1034
3. Das LM et al (2005) Experimental evaluation of a hydrogen added natural gas (HANG) operated
SI engine, SAE Paper No. 2005-26-029
4. Ma F et al (2012) Effect of compression ratio and spark timing on the power performance and
combustion characteristics of an HCNG engine. Int J Hydrogen Energy 37:18486–18491
5. Zhao J et al (2013) Effects of compression ratio on the combustion and emissions of a hydrogen
enriched natural gas engine under different excess air ratio. Energy 59:658–665
Behaviour of NiTi Based Smart Actuator
for the Development of Planar Parallel
Micro-Motion Stage
Abstract Recently, the application of smart materials such as shape memory alloys
(SMAs) as actuators is gaining huge importance. SMA-based actuators are light in
weight and provide higher work/mass. SMA undergoes simple actuation process
such as Joule heating. Nitinol (NiTi) SMA can restore larger strains as compared to
others. It can serve as active prismatic joint to provide linear motion in various robotic
manipulators. The study correlates the deflection of (a) single NiTi spring and (b)
series-connected NiTi springs with input parameters (time and current) to understand
its behavioural complexity. The study revealed that the rate of NiTi spring contraction
is dependent on time and current. To predict the actuation motion, several regression
models were developed. This study defined the feasible current range for the actuation
of NiTi spring based on contraction rate and precision. The contraction rate for single
NiTi spring differs from the series connection of two NiTi springs which results a
new set of polynomial regression model. The developed mathematical models can
help control the smart actuation-based planar parallel robotic manipulators.
1 Introduction
Currently, the need of smart materials is emerging in every field due to its amazing
behaviour to change itself on application of external parameters such as temperature,
stress, electric and magnetic fields [1]. Shape memory alloys (SMAs) are smart
materials which can restore its memorized shape when heated till the austenite region
from martensite region due to change in crystal orientation from monoclinic to the
body-centred cubic structure. SMAs are being widely used in many applications
such as biomedical, commercial and aerospace industries with one of its important
application as actuators [2]. SMA actuators are very compact with higher power/mass
ratio, silent operation, low voltage activation and most importantly biocompatibility
[2, 3].
In the reference [4], the authors studied the workspace of the three-legged U-
shaped base 3PRP planar robotic manipulator experimentally by implementing niti-
nol as an actuator. The workspace was identified based on full contraction of SMA
actuator, which incurs difficulty in control of end-effector. The desired end-effector
position can be achieved by the displacement control of actuator. It can be achieved
by correlating the contraction rate of SMA with respect to current and time.
However, one of the main drawbacks of SMAs is the hysteresis which makes com-
plications in controlling the displacement because of its continuous nonlinear change
in length, particularly, when the micro-motion is in consideration. The displacement
behaviour of SMA springs under Joule heating is merely studied.
Therefore, in the present study, the correlation of the contraction of NiTi springs
with current and time as input variables was determined to understand its behavioural
complexity. Also, the series connection behaviour of the SMA spring was analysed.
2 Experimental Procedure
The selected actuation material is the readily available NiTi SMA spring of 0.75 mm
diameter with 19 helix windings. The general composition of NiTi is 50 wt% Ni and
50 wt% Ti. NiTi SMA spring has the ability to provide a maximum strain of up to
8% which is highest as compared to other SMA materials [5].
In order to provide variable DC current to the NiTi SMA spring, a variable DC power
supply has been developed and the schematic circuit diagram is shown in Fig. 1(a).
The developed variable DC power supply, used in the study, is shown in Fig. 1(b).
The necessity of the variable current is to understand the behaviour of the contraction
(deflection) of springs at various current values to predict the displacement necessary
for the manipulator. The accuracy of the DC variable power supply is 10 mA.
Behaviour of NiTi Based Smart Actuator for the Development … 223
Fig. 1 Variable DC power supply (a) circuit diagram and (b) prototype
In order to understand the behaviour of the SMA springs in series, two NiTi springs
were taken. One end of first spring was fixed at the origin (0, 0), and one end of second
spring was fixed at location B (l, 0) where l = 129 mm. The other end of the two
springs was connected together at point A (r, 0) with the help of a pin made of ABS
plastic to restrict the flow of current from one spring to other as depicted in Fig. 2.
Initially, the second spring is contracted fully by supplying electrical current and then
allowed to cool down to room temperature. Then, constant current is supplied to the
first spring, and the position of A is noted at an interval of 10 s. Similarly, different
values of constant current were supplied to the spring with the help of variable DC
power supply (as discussed in Sect. 2.2), and the position of A was determined.
In this study, an effort has been made to control the displacement of the SMA spring-
based linear actuator, so that it can effectively control the end-effector motion in
micron level. The end-effector motion is directly controlled by SMA actuator. In the
experiment, it was observed that the SMA actuator contracted by applying an electric
current from its initial elongated length (L 0 ) of 100 mm. In this study, a particular
current value was supplied to the spring, and the contraction time was noted for a
standard contraction length (L c ) of 71 mm. In the entire experiment, the current was
varied from 600 to 1220 mA in 15 steps to observe the total deflection time required
to achieve Lc . The result showed that the current range can be classified into four
different categories based on spring contraction rate as listed in Table 1.
Table 1 indicates that for higher current, the contraction rate of spring is higher
and precision is lower and vice versa. The infeasible region (category IV) indicates
the current range which cannot be utilized in the application because of incomplete
contraction over time.
The experimental data as discussed in Sect. 3.1 were used to develop a fitted nonlinear
exponential mathematical model between current (I) and time (t). The nonlinear
exponential fitted Eq. (1) has an adjusted regression value of 0.97442 which clearly
indicates the acceptability of the equation with approximately 97.442% fit. Figure 3
depicts Eq. (1) as an exponential curve of negative slope which indicates that with
decrease in current, the time required for the same contraction length (L c ) is higher.
The slope is higher initially and decreases gradually over time which proves that at
Behaviour of NiTi Based Smart Actuator for the Development … 225
higher current values, the contraction rate is higher as compared to lower current
values. Also, at lower currents (I < 700 mA), the slope becomes parallel to time
axis which clearly indicates zero contraction rate indicating the infeasible region as
discussed in Sect. 3.1. The graph can be used to interpret the time required for full
contraction (L f ) at any given current.
A full quadratic polynomial model was obtained using the experimental procedure,
which includes current versus time and deflection versus time data for distinct values
of current in Minitab 17. The second-order quadratic polynomial regression model
for the response of deflection (d) has been expressed as a function of two process
parameters—current (I) and time (t). The following full quadratic polynomial Eq. (2)
is obtained with an adjusted regression value of 0.9204 and zero probability which
is clear indication of the adequacy for the developed regression equation.
d = −30.6349 + 0.4427 × t + 0.2964 × I + 0.0002 t 2 − I 2 − 0.0008 × t × I
(2)
Equation (2) is very helpful in correlating the deflection with respect to current
and time, the only parameters necessary for the development of SMA spring-based
linear actuator. Hence, the time required for a certain deflection based on any fixed
current can easily be interpreted based on Eq. (2).
226 D. Singh et al.
The developed quadratic polynomial Eq. (2) presents the deflection trend by plotting
the deflection curve in a 2D current—time graph as shown in Fig. 4. The figure shows
the deflection trend of limiting displacement as d = 70 and 30 mm.
This typical graph depicts the amount of time necessary for a distinct given current
to obtain any desired deflection. The curves representing d = 70 mm and d = 30 mm
have negative slope which clearly reflects that to attain any desired deflection, the
time increases with decrease in current and vice versa. It can also be stated that the
slope will become zero after a certain time and the time required for full contraction
is infinite which depicts the infeasibility region as discussed in Sects. 3.1 and 3.2.
In order to obtain a three DOF planar micro-motion stage, two SMA springs are
connected with a pin as shown in Fig. 2. The developed nonlinear polynomial Eq. (2)
was applied to control the deflection of springs. Surprisingly, the deflection of the
series-connected springs was deviated from the usual as the inherent resistance of
the springs acted opposite to each other. The derived single-spring equation failed
to describe the deflection behaviour of the series-connected SMA springs. Hence,
a series of experiments were conducted again using the same current–time domain.
The deflection data of series connection at an interval of ten seconds were noted for
distinct supply of current. These data were fed to develop the second-order quadratic
polynomial model similar to single-spring connection as described in Sect. 3.3.
Behaviour of NiTi Based Smart Actuator for the Development … 227
4 Conclusion
References
1. Drossel WG, Kunze H, Bucht A, Weisheit L, Pagel K (2015) Smart—smart materials for smart
applications. Procedia CIRP 36:211–216
2. Miková L, Medvecká-Beňová S, Kelemen M, Trebuňa F, Virgala I (2015) Application of shape
memory alloy (SMA) as actuator. METABK 54(1):169–172
3. Spaggiari A, Castagnetti D, Golinelli N, Dragoni E, Scirè Mammano G (2019) Smart materials:
Properties, design and mechatronic applications. In: Proceedings of the institution of mechanical
engineers, part l: journal of materials: design and applications 233(4):734–762
4. Singh Y, Mohan S (2017) Development of a planar 3PRP planar parallel manipulator using shape
memory alloy spring based actuators. In: Proceedings of the advances in robotics (Proceeding
AIR’17), 10
5. DesRoches R, McCormick J, Delemont M (2004) Cyclical properties of superelastic shape
memory alloys. ASCE J Struct Eng 130(1):38–46
Multi-objective Optimization of Inconel
718 Using Combined Approach
of Taguchi—Grey Relational Analysis
Manav Sheth, Kunj Gajjar, Aryan Jain, Vrund Shah, Het Patel,
Rakesh Chaudhari , and Jay Vora
Abstract Nickel-based alloy such as Inconel 718 is widely used in aerospace, auto-
mobiles, gas turbines, nuclear and chemical industries. Inconel 718 is a high-strength
temperature resistance which exhibits good resistance to corrosion. In the current
paper, Taguchi’s L9 orthogonal array is implemented for wire electrical discharge
machining (WEDM) with three factors at three levels. The influence of input vari-
ables such as pulse-on time, current and pulse-off time has been investigated on the
material removal rate and surface finish. ANOVA analysis has been carried out to
check the significance of variables and their effect on output variables. For MRR,
all three input parameters are found to be significant pulse-on time and the current
is found to have an influence on SR. As Taguchi’s technique can optimize only one
objective at a time with no consideration of its effect on another output parame-
ter which may result in either lower production or pitiable quality. To satisfy such
conflicting objectives at the same time, an optimum parameter setting is required.
Grey relational analysis was used to get an optimal combination of input variables
for multiple output variables. The optimal combination of input parameters is found
to be pulse-on time 55 µs, pulse-off time 5 µs and current 2 A. Predicted values
obtained at an optimal condition using GRA have been validated by experimental
trial and show a very close relationship with negligible error.
1 Introduction
In the present study, a cylindrical bar of Inconel 718 having a diameter of mm with
density equals to 8.19 g/cm3 was utilized to machine the specimens, each of mm
thickness using a concord wire-cut EDM machine DK7732. Molybdenum wire with
0.18 mm diameter was used as electrode material. Three machining parameters, T on ,
T off and discharge current are selected on the machining process responses for MRR
and SR. Taguchi’s L9 orthogonal array is used for experimentation with three control
factors with three levels each. MRR was calculated by using difference in weight
before and after the machining of the sample divided by time taken for the machining.
The surface roughness measurement was carried out by using Mitutoyo that makes
Surftest SJ410 model. Experimentally measured values of MRR and SR for selected
nine experimental trials were shown in Table 1.
3.1 MRR
Experimentally measured values of MRR for selected nine experimental trials are
shown in Table 1. Higher value of MRR is desirable outcome for obtaining higher
productivity. For the experimental data analysis, statistical software Minitab 14 has
been used. ANOVA was implemented with 95% confidence level to determine the
influence of input parameters on each output variable and is shown in Table 2 for
MRR. Factor B (T off ) found to be most significant effect on MRR with 69.77% con-
tribution followed by factor C (Current) and factor A (T on ) with 29.27% contribution
and 0.94% contribution, respectively. The value of P should not be greater than 0.05
232 M. Sheth et al.
to keep the particular input parameter significant for the 95% confidence level [9].
All the selected input process parameters are found to be significant for MRR. The
model is found to significant for MRR as an extremely close relation was observed
between the R-squared values. Figure 1 shows main effect plot for MRR considering
the variation in levels of all input parameters. From Fig. 1, with increase current,
MRR value is found to be increased due to an increased value of discharge energy,
whereas increase in pulse-off time reduces the discharge energy which in turn also
reduces MRR. The discharge energy and spark intensity increase with increase in
pulse-on time, which will remove more materials and gives higher MRR [5].
3.2 SR
In Taguchi method, the SR is in the category of lower the better performance char-
acteristics. ANOVA for surface roughness is shown in Table 3. As seen in Table 3,
factor C has a most significant effect on the SR with 87.11% contribution followed
by factors A and B with 11.7% and 0.98% contributions, respectively. For SR, fac-
tors A and B are found to be significant while factor B is insignificant. The model is
found to significant for SR as an extreme close relation was observed between the
R-squared values. Main effect plot for SR considering the variation in level of input
parameters is shown in Fig. 1. Figure 1 shows that as T on increases, value of SR is
Multi-objective Optimization of Inconel 718 … 233
As Taguchi’s technique can optimize only one objective at a time with no consid-
eration of its effect on another output parameter which may result in either lower
production or pitiable quality. To satisfy such conflicting objectives at the same time,
an optimum parameter setting is required. By using GRA, the problem is improved
into a single objective problem. For the optimization of two quality characteristics
for Inconel 718 with WEDM process, the following steps are outlined:
Step 1 Determination of S/N ratio
Step 2 Normalization of S/N ratios for MRR and SR
Step 3 Calculation of grey relational coefficient (GRC)
Step 4 Calculation of grey relational grades (GRG)
Step 5 Calculation of response table for GRG.
234 M. Sheth et al.
Table 5 Confirmation
Optimal machining parameters
experiments
Predicted Experimental
Levels A2 B1 C1 A2 B1 C1
MRR (mm3 /s) 0.8369 0.8407
SR (µm) 5.208 5.122
By using the above grey relational method, GRC for MRR and SR are obtained
along with GRG as shown in Table 1. By using GRG values, factor response tables
have been constructed as shown in Table 4. Greater level value means more sig-
nificance and presents the optimal parameter setting to achieve multiple objectives.
Thus, GRA technique shows that for the dual objectives of MRR and SR, the optimal
parameter setting of WEDM process is A2 B1 C1 .
4 Confirmation Test
Conformation test has been conducted to validate the predicted results obtained
from the GRA. Confirmation experiment was conducted at optimal combination of
A2 B1 C1 , and values of MRR and SR are determined. Predicted and experimental
values are shown in Table 5, and good agreement can be observed between these
values which show the suitability of experimental investigation.
5 Conclusions
Inconel 718 is used for WEDM process to investigate the effect of pulse-on time,
pulse-off time and current on output responses of MRR and SR. All the input vari-
ables such as pulse-on time, pulse-off time and current are found to be significant
for MRR while pulse-on time and pulse-off time are significant for SR. Through
ANOVA, pulse-off time was found to be most significant factor for MRR with
69.77% contribution followed by current with 29.27% contribution and pulse-on
time with 0.94% contribution. Further, for SR, current was the most significant fac-
tor with 87.11% contribution followed by pulse-on time with 11.70% contribution.
The model is found to significant for MRR and SR as an extreme close relation was
Multi-objective Optimization of Inconel 718 … 235
observed between the R-squared values. The optimized process parameter settings
using GRA to maximize MRR and to minimize SR were found to be pulse-on time
55 µs, pulse-off time at 5 µs and discharge current at 2 A (A2 B1 C1 ). Validation
experiment has been conducted at final optimal parameter setting. Good agreement
between predicted and experimental values is observed which shows the suitability
of experimental investigation. Output responses of MRR as 0.8369 mm3 /s and SR
as 5.208 are obtained at optimal parameter setting of (A2 B1 C1 ).
References
1. Aggarwal V, Khangura SS, Garg R (2015) Parametric modeling and optimization for wire
electrical discharge machining of Inconel 718 using response surface methodology. Int J Adv
Manuf Technol 79(1–4):31–47
2. Yang C-B et al (2017) Single and multiobjective optimization of Inconel 718 nickel-based
superalloy in the wire electrical discharge machining. Int J Adv Manuf Technol 93(9–12):3075–
3084
3. Dabade U, Karidkar S (2016) Analysis of response variables in WEDM of Inconel 718 using
Taguchi technique. Proc CIRP 41:886–891
4. Mahapatra SS, Patnaik A (2006) Parametric optimization of wire electrical discharge machining
(WEDM) process using Taguchi method. J Braz Soc Mech Sci Eng 28(4):422–429
5. Chaudhari R et al (2019) Multi-response optimization of WEDM process parameters for machin-
ing of superelastic nitinol shape-memory alloy using a heat-transfer search algorithm. Materials
12(8):1277
6. Chaudhari R et al Pareto optimization of WEDM process parameters for machining a NiTi
shape memory alloy using a combined approach of RSM and heat transfer search algorithm.
Adv Manuf 1–17
7. Lin C, Lin J, Ko T (2002) Optimisation of the EDM process based on the orthogonal array with
fuzzy logic and grey relational analysis method. Int J Adv Manuf Technol 19(4):271–277
8. Rajyalakshmi G, Ramaiah PV (2013) Multiple process parameter optimization of wire electrical
discharge machining on Inconel 825 using Taguchi grey relational analysis. Int J Adv Manuf
Technol 69(5–8):1249–1262
9. Chaurasia A, Wankhede V, Chaudhari R (2019) Experimental investigation of high-speed turn-
ing of INCONEL 718 using PVD-coated carbide tool under wet condition. In: Innovations in
infrastructure. Springer, pp 367–374
The Effect of State Variables
on Nucleation of Earthquake Using
the Rate and State Friction
1 Introduction
Natural phenomena such as earthquakes and landslides are quite common across the
world. An early detection of such a potentially disastrous phenomenon is always
useful for saving life and loss of properties. It has been established that many crustal
N. Sinha (B)
Department of Mechanical Engineering, G. H. Raisoni Institute of Engineering & Management,
Jalgaon 425002, India
e-mail: nitishme08@gmail.com
A. K. Singh
Department of Mechanical Engineering, Visvesvaraya National Institute of Technology, Nagpur
440010, India
e-mail: aksinghb@gmail.com
A. D. Vasudeo
Department of Civil Engineering, Visvesvaraya National Institute of Technology, Nagpur 440010,
India
respectively. Motivated by the chaotic nature of the 2sRSF, Sinha and Singh [14]
have derived the governing differential equation of the 3sRSF in the dimensionless
form as follows:
⎧ dφ
⎪
⎪ = e φ
(1 − ρ ) θ̂ + (ρ − ρ ) θ̂ + (β + ρβ + ρ β − ρ)φ + ρ ψ − k
⎪
⎪ 1 1 1 2 1 2 1 3 1
⎪
⎪
dT
⎪
⎨ + keφ0
dψ
= k eφ0− eφ .
⎪
⎪
dT
⎪
⎪
dθ̂1 φ
= −e θ̂1 + β1 φ
⎪
⎪ dT
⎪
⎩ dθ̂2 = −ρeφ θ̂ + β φ
dT 2 2
(2)
where the nondimensional terms are defined as slip velocity φ, frictional shear stress
ψ, state variables θ̂1 , θ̂2 , and θ̂3 , time T, pulling velocity φ0 , spring stiffness k as
∗ ∗
ψ = τ −τ A
, φ = ln vv∗ , θ̂1 = θA1 , θ̂2 = θA2 , θ̂3 = θA3 , T = vL1t , β1 = BA1 , β2 =
B2
A
, β3 = BA3 , ρ = LL 21 , ρ1 = LL 13 , φ0 = ln vv0∗ , k = K LA1 .
Noting that the 3sRSF could be easily converted to the 2sRSF just by eliminating
the third-state variable θ̂3 . Similarly, the 3sRSF may also be reduced to the 1sRSF
again by not considering θ̂2 and θ̂3 in Eq. (2).
Although the linear stability analysis of Eq. (2) is useful for studying stability of
the spring-mass slider about steady sliding [7, 8] that analysis fails to predict sliding
behavior away from equilibrium or steady state [12]. Further, Eq. (2) is also solved
numerically with MATLAB® using ode23s solver for understanding the nonlinear
behavior as well. In the numerical simulations, the main focus is to identify the onset
of chaos in the form of irregular variation of stress amplitudes and corresponding
slip velocity [12, 19]. The results in Fig. 1 illustrate that periodic motion occurs at
k = 0.08611 and shows the period doubling at k = 0.08421 but becomes irregular
at a particular stiffness termed as chaotic stiffness kch = 0.8410. Change in stiffness
signifies the spatial variation of elastic heterogeneity of materials along the earth-
quake faults. Consequently, there is likely that variation in local stiffness of materials
may decrease kch further. Moreover, it is also evident that stress drop also increases
with decrease in stiffness in Fig. 1. Erickson et al. [20] have also suggested that
the ratio of sudden stress drop to increases in shear stress is accompanied by abrupt
change in slip velocity along the earthquake fault. It is thus concluded that the 3sRSF
gives rise to rapid nucleation of the earthquake nucleation process.
Similarly, the above numerical simulations are also repeated for the 2sRSF and
the results are presented in Fig. 2. It is observed that similar to the 3sRSF, the 2sRSF
results in chaotic motion via period-doubling bifurcation, but at reduced stiffness
kch = 0.0682. The trend of stress drop is also seen similar to the 3sRSF in Fig. 2.
240 N. Sinha et al.
Fig. 1 Stiffness dependence stress amplitudes and corresponding slip velocity using the 3sRSF for
fixed values of β1 = 1.0, β2 = 0.84, β3 = 0.38, ρ = 0.048, and ρ1 = 0.034 at initial conditions
[0, 0, 0, 0]
Table 1 presents a comparative change in the stress drop and slip velocity between
the 3sRSF and the 2sRSF. For instance, stress change ψ ∼ 0.810 at the onset of
chaos due to the 3sRSF is smaller than the 2sRSF, that is, ψ ∼ 1.703. Similarly,
Table 1 also indicates that change in slip velocity from periodic oscillation to chaotic
motion is φ ∼ 0.73 in the 3sRSF which is smaller than φ ∼ 1.1813 in the 2sRSF.
Hence, a conclusion is that the 3sRSF predicts a smaller magnitude of stress drop
and increase in slip velocity for the nucleation of earthquake compared to the 2sRSF
[5, 17, 20, 21].
A possible reason for this observation is attributed to increased number of state
variables that take into account the more complex nature of the sliding interface. The
similar conclusion is reported that the nucleation activity of an earthquake process
increases with increasing number of degree of freedom in the sliding system.
Finally, it is believed that the present study could be useful to explain chaotic
behavior of the complex sliding surfaces since the state variable is only the parameter
that increases the chaotic behavior of the RSF model.
The Effect of State Variables on Nucleation of Earthquake … 241
Fig. 2 Effect of variation on stiffness dependence stress amplitudes and corresponding slip velocity
for the 2sRSF for fixed values of β1 = 1.0, β2 = 0.84, and ρ = 0.048 at initial conditions [0, 0, 0,
0]
Table 1 Variation of stress drop and slip velocity of the 3sRSF and the 2sRSF at the onset of chaos
Stress drop Slip velocity
The 3sRSF The 2sRSF The 3sRSF The 2sRSF
Top Bottom Top Bottom
−2.269 −2.180 2.48 −1.390 2.8360 −1.6670
−1.459 −0.373 1.75 −1.017 0.6547 −0.2127
ψ ∼ 0.810 ψ ∼ 1.703 φ ∼ 0.73 φ ∼ 0.273 φ ∼ 1.1813 φ ∼ 1.4543
3 Conclusions
The present study establishes that increase in number of the state variables in the
rate and state friction law enhances the chaotic nucleation of earthquake process.
Moreover, the stiffness at which chaos begins also increases with number of state
variables in the rate and state friction law. The stress drop and corresponding slip
velocity also increase with number of state variables thus signifying the multiscale
mechanism of friction acting along the sliding surfaces.
242 N. Sinha et al.
References
Abstract The femur or thighbone is the longest and strongest bone, which bears
the maximum weight of the human body. There are various types of fracture occurs
in femur bone. The intertrochanteric and subtrochanteric fractures are very complex
to treat and generally stabilized by proximal femoral nail. The aim of the present
research work is to select the proper implant and its respective material during the
fixation of intertrochanteric and subtrochanteric fractures of femur. Such type of study
helps the orthopedic surgeons to predict the failure of implant. In the present study,
the modeling software SolidWorks 17 is used to create the 3D model of the implant.
The dimensions for 3D geometric model of implant are taken with the help of Vernier
caliper. The FEM software ANSYS 16.2 is used for simulation. The finite element
analysis is performed to study the distribution of stress and deformation under one-
legged static load boundary conditions. The deformation and stress values on the
proximal femur are compared between titanium and stainless steel implant material
in case of both subtrochanteric and intertrochanteric fractures of the femur. The bone
healing process mainly depends on the stability given to the fracture. The stability
is achieved by checking the bone contact surface area at fracture interface surface.
The frictional stress and contact pressure at fracture plane are compared by taking
both titanium and stainless steel implant material. These contact results are used to
estimate the condition of healing process at the fracture interface of intertrochanteric
and subtrochanteric fractures. The effect of frictional stress, pressure developed and
localized stress on bone healing is also discussed.
1 Introduction
The femur bone is the longest bone in the human body. It consists of an upper end,
a shaft and a lower end. The upper end of the femur consists of the head, neck,
lesser trochanter, greater trochanter, the intertrochanteric crest and intertrochanteric
line as shown in Fig. 1. The greater trochanter is a huge quadrangular prominence
positioned at the top part of the junction between the femur’s neck and shaft. The
lesser trochanter is present as a conical eminence focused medially and toward the
back from the junction between the posterior side of neck and shaft. Intertrochanteric
line marks the neck’s junction with the femur. Intertrochanteric crest is the junction
of the neck’s posterior side of with shaft of the femur bone.
The “hip” is a type of ball-and-socket joint. It allows rotation and bending of the
upper leg at the pelvis. The majority of hip fractures occur to the people who are
with age 65 or older. Most people are having hip fracture near the upper side of the
femur. There are two types of hip fractures. They are explained below.
Type I (Intertrochanteric Fracture). An intertrochanteric hip fracture is shown
in Fig. 2 which is a specific type of hip fracture. Intertrochanteric means “be-
tween the two trochanters” which are bony projections on the thighbone. Gen-
erally, intertrochanteric fracture arises between lesser and greater trochanters.
Intertrochanteric fractures are the result of such a high energy injury in younger
individuals, fall from a height or a motor vehicle accident [1].
Type II (Subtrochanteric Fracture). Subtrochanteric fractures of the femur take
place in the proximal region as shown in Fig. 3, whose anatomical definition is
controversial and difficult. Subtrochanteric hip fracture is a break in the transverse
Fig. 2 Intertrochanteric
fracture
Fig. 3 Subtrochanteric
fracture
plane between the lesser trochanter and the area around 5 cm below from the lesser
trochanter [2].
2 Methodology
The specimen is collected from the diagnosis laboratory of an orthopedic surgeon for
better measurement of screw and nail part of the implant. The 3D model of proximal
femoral nail (PFN) with a three-holed, screw plate was created with CAD software
SolidWorks. The 3D model of the superior screw, inferior screw, distal screw and
246 S. Rathor et al.
PFN nail are also prepared to utilize the geometry and measurements from the best
possible estimation by Vernier caliper. The final assembly of the implant is shown in
Fig. 4.
During the assembly in the SolidWorks, the femur, PFN implant, distal screw and
lag screw are assembled, the positioning of the implant with respect to the bone is
done, and the final assembly of bone-implant model is prepared as shown in Fig. 6.
Intertrochanteric fracture is created at an angle 45° at the intertrochanteric region
giving a 1 mm gap to create fracture as shown in Fig. 7. Subtrochanteric fracture is
also created with respect to the transverse plane of the shaft of the femur, and this
fracture is created below the lesser trochanter as shown in Fig. 8.
Fig. 7 Intertrochanteric
248 S. Rathor et al.
Fig. 8 Subtrochanteric
Different linear elastic isotropic material properties were assigned to different regions
of the femur like shaft, head, neck, trochanteric region and the implant (stainless
steel/titanium). Corresponding elastic constants are given in Table 1. Stainless steel
(SS) and titanium (TI) material properties were assigned to the implants. The fric-
tional coefficient of 0.29–0.33 is assigned to all the respective parts of the model
[2].
To generate the mesh of the PFN and the proximal femur, ten-nodded tetrahedral
elements were used. The implant and bone’s global element edge lengths were 3
and 4 mm, respectively. The no of nodes and elements generated for subtrochanteric
fracture is 87,265 and 52,713, respectively Also, for intertrochanteric fracture, the
no. of nodes and elements generated is 87,992 and 52,769, respectively. The model
is fixed at the distal end of the femur. In this study, a one-legged stance load con-
figuration is used. The load condition consists of one joint reaction force of 730 N,
and three muscle retraction force of 300, 188 and 292 N are acted [3]. All the above
forces applied to femur bone on ANSYS software are shown in Fig. 9.
Fig. 9 Loading
The von Mises stress, deformation, frictional stress and the pressure developed at the
fracture interface are the outcomes of the analysis. Figure 10a–e describes the results
obtained by stress analysis. The results include von Mises stress for subtrochanteric
fractured femur implantation with stainless steel. Also, deformation, frictional stress
and the pressure developed at the fracture interface are shown in Fig. 11a, b and c,
respectively.
The magnitude of von Mises stresses in different parts of the bone-implant assem-
bly for intertrochanteric and subtrochanteric fractures for both the implant materials
is given in Table 2. The comparison with published data [4] is also given in Table 2.
The contact status at the fracture interfaces for both implant are given in Table 3.
Fig. 11 Deformation, pressure and frictional stress in subtrochanteric fracture with SS implant
Table 2 Maximum von Mises stress induced in the parts of assembly for all the four cases
Parts of Implant material Intertrochanteric Subtrochanteric Published
assembly fracture (MPa) fracture (MPa) results (MPa)
Femur bone SS implant 61.869 63.11 –
TI implant 74.36 63.42 –
Inferior screw SS implant 109.45 126.24 –
TI implant 97.37 117.10 –
Superior screw SS implant 114.09 135.11 –
TI implant 110.37 112.16 –
Nail SS implant 127.04 165.11 152.68
(maximum in (maximum in
SS) SS)
TI implant 98.90 123.52 116.11
(maximum in (maximum in
TI) TI)
Distal screw SS implant 25.20 32.28 27.90
(minimum in (minimum in
SS) SS)
TI implant 17.67 19.31 19.04
(minimum in (minimum in
TI) TI)
Bold signifies the Maximum and Minimum Values of the stress induced in SS and TI implant
materials. Those values are compared with the published data
Finite Element Analysis of Type I and Type II Fracture … 251
Table 3 Pressure and frictional stress for all the four cases
Parameter Implant material Intertrochanteric fracture Subtrochanteric fracture
(MPa) (MPa)
Pressure SS implant 5.2157 232.69
TI implant 6.2893 195.90
Frictional stress SS implant 6.8749 208.73
TI implant 7.9341 193.55
4 Conclusion
It is concluded in the past studies that the PFN implant is the minimum invasive
implant for Proximal femoral fracture [5]. PFN is proved to be very suitable implant
for unstable proximal femoral fracture [5, 6]. It is found that the maximum von
Mises stress induced in both implant is within the range of yield stress of SS and TI,
respectively. The results obtained are compared and validated with the published data.
Stresses developed are greater for SS implant as compared to TI implant, but higher
stresses are induced in bone in case of TI implant as compared to SS implant. For
subtrochanteric fracture, maximum stress is induced in nail part in both the implant.
For intertrochanteric fracture with TI, maximum stress is induced in the superior
screw and in nail part in the case of SS implant. It is noticed that the intertrochanteric
fracture of the femur with PFN implant has high localized contact stress which does
not encourage the early healing of the fracture. Hence, PFN is more suitable for
subtrochanteric fracture as compared to intertrochanteric fracture.
Ethical Approval This article does not contain any studies with human participants performed by
any of the authors.
References
1. Liang C et al (2018) Intertrochanteric fracture: association between the coronal position of the
lag screw and stress distribution. Asian J Surg 41(3):241–249
2. Sowmianarayanan S et al (2008) Finite element analysis of a subtrochanteric fractured femur
with the dynamic hip screw, dynamic condylar screw, and proximal femur nail implants—a
comparative study. Proc Inst Mech Eng [H] 222(1):117–127
3. Fraldi M et al (2010) Topological optimization in hip prosthesis design. Biomech Model
Mechanobiol 9(4):389–402
4. Taheri NS et al (2011) Comparative study of two materials for dynamic hip screw during fall and
gait loading: titanium alloy and stainless steel. Journal of Orthopaedic Science 16(6):805–813
5. Boldin C et al (2003) The proximal femoral nail (PFN)—a minimal invasive treatment of unstable
proximal femoral fractures: a prospective study of 55 patients with a follow-up of 15 months.
Acta Orthop Scand 74(1):53–58
6. Chopra B et al (2017) Proximal femoral nail-outcome and complications: a prospective study
of 125 cases of proximal femoral fractures. Int J Res Orthop 3(5):973
Postural Evaluation of Construction
Labourers Engaged in Excavation Work
Using Newly Developed NERPA Method
and Its Validation Through REBA
and WERA Methods
1 Introduction
The first phase of any construction work is its foundation which is base of any
house. This foundation is laid down from the ground. The excavation is the digging
of soil for making pit inside the ground to lay down the foundation and erect the
column. In India, maximum construction work is being carried out manually, and the
excavation is an important aspect in initial stage. With development of modern and
new technology, construction work still demands manual intervention or physically
demanding work. In India, construction industry alone contributes 24.20% fatality
per year due to different reasons of risk factors in construction industry [1]. Many
researchers have started working in the construction area and highlighted the issues
related to physical exposure among the construction workers and labourers [2–5].
The past researches also indicate the development of assessment tools such as QEC
[6], REBA [7], OWAS [8], PLIBEL [9] and many more for measuring the physical
risk/discomfort. Many construction workers and labourers are continuously working
in unfavourable conditions without any intuition about the physical risks and suffering
from WRMSD. Therefore, this study has been carried out with the aims to (1) find
the level of pain/discomfort in different body parts of excavation labourers as per
complaints of labourers about pain/discomfort in different body parts according to
age, (2) verify this pain/discomfort by applying NERPA, WERA and REBA methods
and (3) compare the compatibility of NERPA methods’ worksheet in evaluation of
risk in construction work.
2 Literature Review
This method, NERPA, has been developed by Sanchez-Lite et al., in 2013, using a
digital human model (DMH) jointly with 3D CAD tool together with motion capture
in real time that will be used in 3D virtual environment by incorporating work process,
equipment, machineries, tools and human factors. The method has been developed
particularly for aeronautical and automobile industries which is the modified version
of RULA. The method focused on arms, neck, trunk and wrists and little focus on
legs and helps to make adequate decisions in the design and assessment [10].
Rapid entire body assessment (REBA) is a quick observational and survey method
used to assess static and dynamic movements, rapid changing or unstable body
posture which leads to the development of work-related musculoskeletal disorder
[7].
Rahman et al., in 2011, developed this pen-and-paper method to find the WRMSD in
construction for evaluation of risk and can be used in any space of workplaces without
disturbing the workers. This is quick observational method which was developed to
Postural Evaluation of Construction Labourers Engaged … 255
give screening method to working task for exposure of physical risk factor related
with WRMSD. This method comprises of six types of physical risk factors like
posture, repetition, force, vibration, contact stress and task duration including five
body parts which are neck, shoulders, wrists, lower back and legs. This method needs
no special equipment [11].
3 Methodology
The study had been carried out for ten days for daily working hours of 8–10 with
breaks. 46 male and 15 female labourers engaged in excavation work have been
observed, interviewed and recorded from 12 different construction sites whose back-
ground characteristics are given in Table 1. The height and weight had been mea-
sured using anthropometric scale and standard weighing machine. The personal data
like age, daily working hours, procedures and pain/discomfort with severity (like ‘no
pain’, ‘occasional pain’, ‘always pain’) and other related problems they faced in daily
working hours on the construction sites have been discussed and noted. The labourers
were between 20 and 59 years of age, with working experience of 1–38 years. The
mean age, weight, height, experience and BMI with standard deviation and range are
given in Table 1.
The labours’ are categorised in two groups (male and female) while an excavation
task is divided into four tasks. Table 2 shows task details, real image and body position
while doing the task. The percentage of labourers complaints for pain/discomfort in
different body parts are mentioned in Table 3.
In this study, an ergonomic risk score of postures has been evaluated by NERPA,
WERA and REBA methods using the postures of labourers while working (Table 2).
256 M. T. Gajbhiye et al.
Table 2 (continued)
Task No. and task Real images of task Body position
details
Male Female
T4 1. Shoulder extension
Throw or dump 2. Both arms above
excavated soil outside shoulder for passing
pit: Labourer 1: pan to labourer 2
3. Neck extension
4. Wrist extension
>15° (bent)
Labourer 2: 1. Trunk forward
flexion to pick pan
2. Trunk lateral
bending or twisted:
>20°
3. Arms below
shoulder while
picking pan
4. Arms above
shoulder after lifting
and throwing
5. One legs flexion
(30°–60°)
6. Other leg flexion
(>60°)
These scores are being compared with the labourers’ complaints. To validate compat-
ibility of NERPA methods’ worksheet, the NEPRA methods’ result has been com-
pared with WERA and REBA methods. From the study, it was found that the male
labourers have complained ‘always pain’ in wrists (69.57%), shoulders (93.48%),
arms (97.82%) and lower back (100%), whereas female labourers has complained
‘always pain’ in neck (60%), wrist and legs (86.67%), shoulder (93.33%), arms
and lower back (100%) (Fig. 1). The ergonomic risk score of postures for the four
different tasks had been evaluated using NERPA, WERA and REBA methods as
given in Table 4. From the scores of all the three methods, it was found that all the
tasks were under high risk and labourers were suffering from work-related muscu-
loskeletal disorders. From the score, it can be concluded that shoulders, arms, neck,
wrists and lower back were under more risk and need immediate investigations and
corrections. The scores obtained from NERPA worksheet had been compared with
the WERA and REBA scores. It was found that the risk level scores obtained from
WERA and REBA were similar to that of obtained by NERPA method. Also, the
body parts, which were under high risk, as complained by labourers and evaluated
by NERPA worksheet, WERA and REBA method were found to be same. Hence,
NERPA worksheet can be used to evaluate the risk of construction workers using
actual 3D visualisation of workstation, working postures and ergonomic design in
virtual scenario.
5 Conclusion
The complaint of the labourers and scores of the different methods show that the
labourers engaged in excavation work are in high physical risk zone and suffering
Table 4 Ergonomic risk scores of NERPA, WERA and REBA methods
NERPA WERA REBA
BP T1 T2 T3 T4 BP T1 T2 T3 T4 BP T1 T2 T3 T4
L1 L2 L1 L2 L1 L2
UA 4 4 3 4 4 Part A N 2 2 2 3 3
LA 3 3 3 2 2 S 5 4 5 5 4 T 5 5 4 2 5
W 3 2 4 4 3 W 5 5 5 5 5 L 2 2 2 1 2
WT 1 1 1 1 1 LB 6 6 6 5 6 PS-A 7 7 6 4 8
PS-A 4 4 5 5 4 N 5 5 5 5 5 F 1 1 2 2 2
M 1 1 1 1 1 L 5 6 6 5 6 TS-A 8 8 8 6 10
F 2 1 3 3 3 Part B UA 5 3 4 5 4
TS-C 7 6 9 9 8 F 5 4 6 5 6 LA 2 2 2 2 2
N 4 4 4 4 5 V 4 5 4 3 4 W 2 3 3 3 3
LB 4 5 4 2 5 C 6 6 6 5 6 PS-B 8 5 7 8 7
Postural Evaluation of Construction Labourers Engaged …
L 1 1 1 1 1 TD 5 4 6 6 6 CP 1 1 3 3 3
PS-B 7 7 7 5 8 TS-B 9 6 10 11 10
M 1 1 1 1 1 TS-C 10 10 11 10 12
F 3 1 3 3 3 A 1 1 1 1 1
TS-C 11 9 11 9 12
FS 7 7 ><7 ><7 ><7 46 45 49 44 48 FS 11 11 12 11 13
RL HR HR HR HR HR HR HR HR HR HR VHR HR VHR VHR VHR
Note BP body parts, T1 task 1, T2 task 2, T3 task 3, T4 task 4, UA upper arm/limb, LA lower arm/limb, W wrist, WT writ twist, N neck, LB lower back, L
legs, S shoulder, A activity, M muscle, F force, V vibration, C contact stress, TD task duration, PS posture score, TS total score, FS final score, RL risk
level, HR high risk, VHR very high risk
259
260 M. T. Gajbhiye et al.
Disclaimer All procedures performed in studies involving human participants were in accordance
with the ethical standards of the institutional and/or national research committee and with the 1964
Helsinki declaration and its later amendments or comparable ethical standards.
References
1. Patel DA, Jha KN (2016) An estimate of fatal accidents in Indian construction. In:
Chan PW, Neilson CJ (eds) Proceedings of the 32nd annual ARCOM conference, 5–7
September 2016, vol 1. Association of Researchers in Construction Management, Manch-
ester, pp 577–586. https://www.researchgate.net/publication/308155592AN_ESTIMATE_
OF_FATAL_ACCIDENTS_IN_INDIAN_CONSTRUCTION
2. Antwi-Afari MF, Li H, Edwards DJ, Pärn EA, Seo J, Wong AYL (2017) Biomechanical anal-
ysis of risk factors for work-related musculoskeletal disorders during repetitive lifting task in
construction workers. Autom Constr 83:41–47. https://doi.org/10.1016/j.autcon.2017.07.007
3. Chen J, Qiu J, Ahn C (2017) Construction worker’s awkward posture recognition through
supervised motion tensor decomposition. Autom Constr 77:67–81. https://doi.org/10.1016/j.
autcon.2017.01.020
4. Roman-Liu D (2014) Comparison of concepts in easy-to-use methods for MSD risk assessment.
Appl Ergonomics 45(3):420–427. https://doi.org/10.1016/j.apergo.2013.05.010
5. Valero E, Sivanathan A, Bosché F, Abdel-Wahab M (2016) Musculoskeletal disorders in con-
struction: a review and a novel system for activity tracking with body area network. Appl
Ergonomics 54:120–130. https://doi.org/10.1016/j.apergo.2015.11.020
6. David G, Woods V, Li G, Buckle P (2008) The development of the quick exposure check
(QEC) for assessing exposure to risk factors for work-related musculoskeletal disorders. Appl
Ergonomics 39(1):57–69. https://doi.org/10.1016/j.apergo.2007.03.002
7. Hignett S, McAtamney L (2000) Rapid entire body assessment (REBA). Appl Ergonomics
31(2):201–205. https://doi.org/10.1016/s0003-6870(99)00039-3
8. Karhu O, Kansi P, Kuorinka I (1977) Correcting working postures in industry: a prac-
tical method for analysis. Appl Ergonomics 8(4):199–201. https://doi.org/10.1016/0003-
6870(77)90164-8
9. Kemmlert K (1995) A method assigned for the identification of ergonomic hazards—PLIBEL.
Appl Ergonomics 26(3):199–211. https://doi.org/10.1016/0003-6870(95)00022-5
Postural Evaluation of Construction Labourers Engaged … 261
10. Sanchez-Lite A, Garcia M, Domingo R, Angel Sebastian M (2013) Novel ergonomic pos-
tural assessment method (NERPA) using product-process computer aided engineering for
ergonomic workplace design. PLoS ONE 8(8):e72703: 1–12. https://doi.org/10.1371/journal.
pone.0072703
11. Rahman MNA, Rani MRA, Rohani JM (2011) WERA—an observational tool develop to
investigate the physical risk factor associated with WMSDs. J Hum Ergol 40:19–36. https://
www.jstage.jst.go.jp/article/jhe/40/1_2/40_19/_pdf
Influence of Stress Bar Length
on the Response of a Stress Wave Force
Balance Using Finite Element Analysis
Abstract The stress wave force balance is used to measure forces in aerodynamic
facilities like shock tunnels and wind tunnels. For accurate measurement of forces, the
shape and size of the model and the balance are of importance. The dimensions of the
balance mounted inside the model (internal balance) depend on the internal diameter
of the model. However, the effect of the length of the stress bar is independent of the
model dimensions. This paper describes the variation in transient strain responses
at various locations of the stress bar with the variation of the length of the stress
bar. Impulse force of 10 N magnitude with 1 ms duration was applied at the tip of
a hemispherical model, and the strain responses were observed at various locations
along the stress bar. Impulse forces of various magnitudes were applied at the tip of
the model, and the strain responses were observed. From the simulation results, it
was observed that as the length of the stress bar increases, the strain increases. The
strain response was also found to be steady with the increase in length of the stress
bar.
Keywords Stress wave force balance · Finite element analysis · Impulse · Strain
1 Introduction
measuring the strain history with the help of strain gauges or piezo films. The stress
wave force balance was first used by Simmons and Sanderson in 1991 [1]. It was a
single component force balance and was used to measure the drag force on a conical
model. However, the stress wave force balance used by Simmons and Sanderson
was modified by researchers to measure multiple components on larger models. The
three-component force measurement using the stress wave force balance was studied
by Mee et al. [2] to measure the axial and normal forces and the pitching moment.
Tuttle et al. [3] measured the drag forces using models longer than that were used
by Simmons and Sanderson [1] at hypersonic flows. The measurement of strain in
the force balance used by Simmons and Sanderson was performed by strain gauges.
However, piezo films were also used by researchers for measurement of strain. Smith
and Mee [4] used piezo film to measure the strains to obtain the drag force in a hyper-
velocity expansion tube. The design and accuracy of the force balance have also been
verified by researchers using finite element analysis. Finite element analysis was first
performed by Daniel and Mee [5] to obtain the drag force, lift force and pitching
moment. Robinson et al. [6] performed finite element simulations on a stress wave
force balance to obtain the drag force on re-entry vehicles.
The study on force measurement using the stress wave force balance on various
models has been performed by various researchers. However, the effect of the length
of the stress bar on the strain at various locations of the stress bar has not been studied
till date. In this paper, the effect of the variation of length on the strain responses at
different locations of the stress bar has been studied. The effect of variation of load
on the strain response for the best length of the stress bar has also been obtained.
The 3D geometry of the model-balance assembly of a stress wave force balance was
modelled in CREO, and the finite element simulations were performed in ANSYS
WORKBENCH 18.0. The model material was assigned as aluminium having 60 mm
external diameter, 40 mm internal diameter and 100 mm length. Aluminium was
considered due to its low density due to which it is commonly used in aerospace
industries. The force balance material on which the strain gauges can be mounted
was assigned as brass since it is the most commonly used material used by researchers
due to its medium density [3]. The properties of aluminium and brass are shown in
Table 1. The force balance was an internal balance with diameter equal to the internal
Table 1 Properties of
Density Young’s Poisson’s
materials [7, 8]
(kg/m3 ) modulus ratio
(GPa)
Aluminium 2770 71 0.33
Brass 8450 97 0.34
Influence of Stress Bar Length on the Response of a Stress … 265
diameter of the model. The force balance was considered to be screwed to the model
at four locations. Fixed-end boundary condition at the end of the stress bar was
considered for the simulations. The mesh was generated with 34,102 number of
elements. The convergence test was performed with 34,102 and 63,260 elements,
and a very low error of 0.62% was observed. The meshed model is shown in Fig. 1.
An impulse load of 1 ms duration (Fig. 2) was applied at the tip of the model, and
the strain was observed at different locations along the length of the stress bar. Three
lengths of the stress bar from the model end to the fixed end were considered for the
simulation, i.e., 100, 150 and 200 mm.
The strain responses were observed at different locations on the stress bar for impulse
force applied on the model. For 100 mm length of the stress bar, the strain was
observed at four locations, A, B, C and D separated by a distance of 25 mm from
each other. Similarly, for 150 and 200 mm length of the stress bar, the strain was
calculated at six and eight locations of the central rod, respectively, each location at a
distance of 25 mm from the other. The location of application of force and the strain
measurement locations on a stress bar of 100 mm length are shown in Fig. 3. The
strain responses obtained at locations A, B, C and D on application of an impulse
force of 10 N at the tip of the hemispherical model are shown in Fig. 4, Fig. 5, Fig. 6
and Fig. 7, respectively. It was observed that for all the four locations, the amplitude
of strain slightly increases with increase in the length of the stress bar. The strains
obtained with a higher length of the stress bar (200 mm) were found to be steadier
than that using lower length of the stress bar (100 and 150 mm). This is because
the time required for the reflection of the stress waves from the fixed end of the
stress bar is less for lower lengths of the stress bar than at higher lengths. Thus, for
lower length, superposition of the stress waves occurs within lesser time, and thus,
the strain response was found to be more unsteady than that for longer lengths of
the stress bar. However, the strain response is similar throughout the length of the
stress bar (Figs. 4, 5, 6 and 7) as the distance from the model end only effects the
magnitude of the strain response and not the nature of the response.
268 S. Deka et al.
The optimum length of the stress bar to obtain uniform strain response was found
to be 200 mm. However, for lengths of stress bar that are higher than that, it was
observed that the stress bar was unable to support the model as the distance of the
model was too far from the fixed end. It was observed that a 250-mm-long stress bar
was unable to support the model.
The magnitude of strain along the length of the stress bar was observed for all the
three lengths of the stress bar (100, 150, 200 mm) and is shown in Fig. 8. It was
observed that the magnitude of strain decreases from the model end to the fixed
end, reaches a minimum and then slightly increases near the fixed end. This is due
to the reason that the amplitude of the stress waves passing through the stress bar
decreases from the model end towards the fixed end. However, near the fixed end,
due to superposition of the reflecting stress waves with the advancing stress wave,
the amplitude of strain slightly increases. The rate of decrease of strain with respect
to length is also low for higher lengths of the stress bar, and the strain decreases
slowly than that for lower length of the stress bar. As a result, the decrease in strain
is linear for lower lengths of the stress bar, while it is nonlinear for higher lengths of
the bar.
The stress wave force balance having a 200 mm length of the stress bar gave the
most steady and highest strain response. Thus, for further simulation, a 200 mm
stress bar was considered and loads of 10, 20, 30 and 40 N were applied having 1 ms
impulse duration. The strain response at location A with different amplitude of forces
applied at the tip of the model is shown in Fig. 9. The strain response increases with
increasing the load on the model.
4 Conclusion
The strain responses obtained from a hemispherical model using a stress wave force
balance have been analysed in this paper using finite element analysis. Three different
lengths of the stress bar were considered, and the strain responses were observed at
different locations of the stress bar. The strain responses were found to be higher and
steadier for a stress bar of length 200 mm than that using stress bar having length 100
and 150 mm. The time required for reflection of the stress wave from the fixed end of
the stress bar is higher for the 200 mm length of the stress bar is higher than that for
lower lengths of the stress bar. This leads to a steadier strain response for 200 mm
length of the stress bar. Thus, superposition of the stress waves occurs after a long
time for 200 mm length of the stress bar due to its higher length. However, a stress
bar length of 250 mm was found to be unable to support the model as the distance of
the model from the fixed end was too high. The strain was found to decrease from
the model end of the stress bar towards the fixed end, reaches a minimum and then
increases slightly near the fixed end. The stress wave passes from the model end to
the fixed end and becomes weaker as it flows through the stress bar. However, near
the fixed end, superposition of the advancing and reflected stress waves occurs due
to which the strain increases slightly near the fixed end. Thus, it was observed that
a longer stress bar gives a steadier response than a shorter stress bar; however, it
should be able to support the model and the force balance. The optimum length of
the stress bar should be used to obtain suitable response of the system.
References
1. Simmons J, Sanderson S (1991) Drag balance for hypervelocity impulse facilities. AIAA J
29:2185–2191
2. Mee DJ, Daniel WJ, Simmons JM (1996) Three-component force balance for flows of
millisecond duration. AIAA J 34(3):590–595
3. Tuttle S, Mee D, Simmons J (1995) Drag measurements at Mach 5 using a stress wave force
balance. Exp Fluids 19:336–341
4. Smith A, Mee D (1996) Dynamic strain measurement using piezoelectric polymer film. J Strain
Anal Eng Des 31:463–465
5. Daniel W, Mee D (1995) Finite element modelling of a three-component force balance for
hypersonic flows. Comput Struct 54:35–48
6. Robinson M, Schramm JM, Hannemann K (2007) An investigation into internal and external
force balance configurations for short duration wind tunnels. In: New results in numerical and
experimental fluid mechanics, vol VI. Springer, Berlin, pp 129–136
7. Nanda SR, Kulkarni V, Sahoo N (2016) Design of artificial neuro-fuzzy based methodology for
six component force balance. Proc Eng 144:528–536
8. Quadri SAP, Suman R, Mohiddin MK, Shareef MK, Malik MR, Haque WU, Shaik IM (2017) A
study of aluminium 6082 and brass 319 materials by friction stir welding process. Int J Emerg
Trends Technol 46:178–183
Relative Power Variation in Frequency
Sub-bands of the EEG Signal During
Painful Stimuli
Abstract The present study investigates trend of relative band power in the fre-
quency sub-bands of electroencephalogram (EEG) signal during painful stimuli.
Ten healthy right-handed male in the age group of 20–26 years participated in the
experiment of generating thermal pain with simultaneous EEG recording. RMS EEG
MAXIMUS acquisition and analysis software was used to record and extract rela-
tive band power while decomposing the signal in multiple frequency sub-bands, viz.
delta, theta, alpha and beta. The extracted feature during pain has source localisation
in delta and theta bands as it accounts for larger power share among all sub-bands.
The pain reporting on the Numerical Pain Rating Scale (NPRS) is subjective when
compared with the relative band power in different frequency sub-bands in varying
levels of pain over all the subjects. It was found that the extracted feature in all
frequency sub-bands has a captivating correlation with the NPRS. The EEG signal
analysis proves to be the novel tool in pain assessment for better clinical treatment
and quantification.
1 Introduction
Pain is the psychological phenomena and a warning sign for taking some necessary
measures for avoiding the problem to aggravate [1]. Pain is one of the most uncon-
vincingly addressed of all the sensations. It also develops in the form of memory
as sometimes the person feels the post-surgical obnoxious sensation after seeing the
surgical tools. Unequivocally, the painful stimuli activate specific regions in the brain
that gives the feedback while also wax and wane the experience translating it into the
proportional actions. Therefore, in order to understand the pain, it is better to peep
into the brain. At present, the most widely used clinical pain assessment tool is the
Numerical Pain Rating Scale (NPRS) or Visual Analogue Scale (VAS). However, it
has its demerits for being subjective and a non-reliable tool for pain assessment.
In the recent past, brain signals from EEG recordings for analysing variations
in the different frequency bands during sleep have shown a significant trend. The
recordings have a novel application in detecting dementia, Alzheimer or seizure.
Hence, EEG signals can also be used to analyse brain signal patterns during pain.
Kumar et al. [2] experimented on the post-anaesthetic care unit patients and validated
their pain index obtained by the EEG analysis with the respective Numerical Pain
Rating Scale (NPRS) score. The index found to be accurately estimating the level
of pain. Jian-Xiong et al. [3] profess pain as the fifth most important vital sign,
and according to them, a series of studies have found that electroencephalogram
(EEG) is a non-invasive, safe and reliable means of examination used in the field of
pain assessment. The pain can cause significant changes in EEG signals in multiple
brain regions and at multiple frequencies. EEG spectrum dominating nociception
falls primarily within four frequency bands: delta (1–4 Hz), theta (4–8 Hz), alpha
(8–13 Hz) and beta (13–30 Hz) [4].
The variations in the EEG signal are evident for asserting correct frequency range
for pain study using signals from brain. However, to use the technique for clinical
usage remains elusive as it has to be vetted and repeatedly experimented before
approving it for assessing pain. Nezam et al. [5] experimented that the effect of
mechanical stimulus (e.g. pressure or impact) on the EEGs is more likely similar to
orthopaedic pain on the hands/legs. EEG recording and validating it with person self-
reporting on NPRS depend on the feature extracted from the signal. Feature extraction
was carried out with power spectrum analysis approach to determine relative band
power [6]. Jensen et al. [7] found that the correlation of higher frontal alpha frequency
activity with the severity of pain, which was initially unexpected, is intriguing. The
results suggested that specific EEG activity patterns may associate with more pain.
Eulália Silva dos Santos et al. [8] found an increase in theta and alpha EEG power
at rest and a decrease in the amplitude of evoked potentials after sensory stimulation
and cognitive tasks for the patients in chronic pain.
The present study intends to find the relative power of each frequency sub-band of
the signal during an experimental pain and its relationship with person self-reporting
on the Numerical Pain Rating Scale.
Relative Power Variation in Frequency Sub-bands … 273
2.1 Subjects
Ten healthy male participants preferably right-handed in an age group of 20–26 years,
administered drug and narcotics free, participated in the study with their consent—
volunteers with no sign of any chronic and tonic pain in the past inducted in the
experimentation.
2.2 Paradigm
The subjects volunteered in the experimental pain generation and its assessment
using EEG recordings with simultaneous self-reporting of pain on the Numerical
Pain Rating Scale. The experiments were conducted in a lying posture with closed
eye state to avoid unwanted artefacts arising due to blinking and neck movements.
The pain condition thermally induced under cold temperature. The subjects were
asked to put their left hand in the bucket of ice and report the pain on the scale of
5–10 (Fig. 1).
2.3 Recording
The EEG recording is carried out on 10/20 electrode system with the sampling
frequency of 256 Hz in RMS EEG MAXIMUS acquisition software. An epileptic
filter of 0.5–70 Hz applied, and power frequency filter kept turned ON at 50 Hz
with an EMG filter. The experiment conducted in the sound-attenuated room. The
impedance of the sampling was kept low at 10 k. During all conditions, the EEG
experiments carried out on the 32 electrode montage system (Fig. 2) at moderate
room temperature. Initially, baseline EEG recorded for 2 min then the pain condition
Fig. 2 Pictorial
representation of 32
electrode montage system
was imposed. The subjects were asked to remove their hand at will. After withdrawal,
EEG recorded for further 2 min, and then, the experiment eventually ends.
The preprocessing stage in data processing comprises of artefact removal more likely,
infested into the signal from the power source, excessive voluntary movements and
eye blink. Data filtering remains a vital part of any data analysis. EEG data are
scaled down to the specific frequency range of study with post-processing of the
preprocessed data for feature extraction in RMS EEG MAXIMUS acquisition and
analysis software.
3 Theory
Relative power value analysis uses an EEG signal data analysis method to obtain the
correct values. The measured EEG signals were transformed into a power spectrum
using fast Fourier transforms (FFTs). The transformed EEG signals decomposed
into four frequency components (δ, θ , α, β) [9]. Fast Fourier transform is the most
critical signal processing algorithm invariably to compute Fourier transform which
returns a complex number for each frequency bin. Thus, it makes easier to extract
the amplitude and phase of the signal at a specific frequency. In spectral analysis of
Relative Power Variation in Frequency Sub-bands … 275
the EEG data, the magnitude squared of the FFT is obtained to estimate the power
spectral density expressed in watts per hertz (W/Hz).
The power spectral density is the genesis of myriad analyses. However, the com-
putation of average band power is the subject of interest which summarises the
influence of the decomposed frequency band to the overall power of the signal into
a single number. In practice, rather than reporting the absolute band power, one may
want to express the power in a frequency band as a percentage of the total power of
the signal as the relative band power. Each selected frequency range (δ, θ , α, β) is
calculated with related power values function.
Selecting range
Relative power value(%) =
Total range (1 ∼ 30Hz)
The observations from the subject’s response on the NPRS and its corresponding
relative band power at all the levels of the pain from 5 to 10 are subjective. Moreover,
there is no significant trend in the entire band, i.e. delta, theta, alpha and beta as per
the subjects’ linear reporting of their pain. However, there is source localisation of
pain in delta and theta bands as much of the band power lies in these bands.
The study aimed to configure the data points obtained from different frequency
sub-bands, viz. delta, theta, alpha and beta during pain. Although the results are
inconclusive, it speaks much about the subject of interest for studying pain. So, as to
find the trend between data points measured over different pain levels in the notable
frequency band, the correlation test was conducted between relative band power in
all frequency sub-bands and Numerical Pain Rating Scale (NPRS) score reported by
all the subjects.
The delta band of the recorded EEG signal during experimental pain follows a
moderate to strong downhill linear variation in half of the population of the test
participants with sixty per cent of the populace imitating linearly downward trend
(Table 1). In the case of the theta band, fifty per cent of the subjects follow a weak
to strong uphill trend (Table 1). The seventy per cent of the populace retains a weak
to strong uphill trend in the alpha band, while fifty per cent follows a strong uphill
trend in the beta band (Table 1).
The data analysis for determining r value is done on 10 subjects; if, however, the
number of subjects increased, then the results would be even more convincing.
The value of r is always between +1 and −1. Exactly −1, a perfect down-
hill (negative) linear relationship. Exactly +1, a perfect uphill (positive) linear
relationship.
276 S. R. Singh and A. B. Deoghare
Table 1 Correlation coefficient of different frequency sub-bands with the NPRS score
r value
Delta Theta Alpha Beta
Subject 1 0.037409 −0.1765 0.218218 −0.02619
Subject 2 0.796303 −0.88779 0.038778 0.924222
Subject 3 0.081151 0.296753 −0.39957 0.064575
Subject 4 0.186663 −0.3294 0.251651 −0.07962
Subject 5 −0.50052 0.569401 0.330341 −0.04022
Subject 6 −0.26266 −0.23435 0.736543 0.83666
Subject 7 −0.83449 0.834555 0.654654 0.654654
Subject 8 −0.85626 0.858548 0.764471 0.916515
Subject 9 −0.54674 0 0.533319 0
Subject 10 −0.68533 0.590095 0.628398 0.828079
5 Conclusions
In this study, the correlation between the relative band power in different frequency
sub-bands and the Numerical Pain Rating Scale (NPRS) score was found, and the
variation follows a linear trend in almost half of the subjects’ population. EEG signal
analysis in painful stimuli infers a lot about pain mechanism and its localisation,
in particular, frequency band as shown in the results. However, it never follows the
similar trend as the case with the patients self-reporting on the NPRS when validated
with the relative band power of the signal in different frequency sub-bands, and it
was subjective. Thus, EEG proves to be the more satisfactory and novel tool for pain
assessment as compared to the Numerical Pain Rating Scale with the scope of future
investigation for clinical use.
Disclaimer Ethical approval: The consent obtained from the participants/volunteers was verbal.
For this type of study, formal consent is not required.
References
1. Ploner M, Sorg C, Gross J (2017) Brain rhythms of pain. Trends Cogn Sci 21(2)
2. Kumar S et al (2015) Electroencephalogram based quantitative estimation of pain for balanced
anaesthesia. J Measure 59:1–390
3. An J-X et al (2017) Quantitative evaluation of pain with pain index extracted from electroen-
cephalogram. Chin Med J 130:1926–1931
4. Mohan Kumar CE, Dharani Kumar SV (2014) Wavelet-based feature extraction scheme of
electroencephalogram. Int J Innov Res Sci Eng Technol 3(1)
5. Nezam T, Boostani R, Abootalebi V, Rastegar K A novel classification strategy to distinguish five
levels of pain using the EEG signal features, IEEE. https://doi.org/10.1109/taffc.2018.2851236
Relative Power Variation in Frequency Sub-bands … 277
6. Ahirwal MK, Iondhe ND (2015) Power spectrum analysis of EEG signals for estimating visual
attention. Int J Comput Appl (0975–8887) 42(15)
7. Jensen MP et al (2013) Brain EEG activity correlates of chronic pain in persons with spinal cord
injury: clinical implications. Int Spinal Cord Soc 1362-4393/13
8. Eulália Silva dos Santos P et al (2016) Electroencephalographic patterns in chronic pain: a
systematic review of the literature. https://doi.org/10.1371/journal.pone.0149085
9. Ko K-E et al (2009) Emotion recognition using EEG signals with relative power values and
Bayesian network. Int J Control Autom Syst 7(5):865–870. https://doi.org/10.1007/s12555-
009-0521-0
A Study on the Effect of GTAW Input
Current on Surface Distortion of Thin
CRNO Electrical Steel Sheets
Abstract The thin sheets of CRNO electrical steel are widely used in motors of
electric vehicles (EV) and several domestic appliances. The thin sheets of thickness
0.5 mm are stacked and welded across the edges to form the stator core for use in the
electric motors. The present work aims to study the distortion variations observed
on the CRNO thin sheets under different range of welding current (30–110 A). The
stacks comprising of thin sheets of CRNO with total thickness of 50 mm (100 sheets
× 0.5 mm) post-welding are observed for evaluating the distortion on the surface of
stacks. The measurement of surface distortion is done using a surface profilometer for
observing the variation on the surface of stacks. The maximum and minimum surface
distortion in the range 0.66 to −0.27 mm and 0.17 to −0.08 mm, respectively, is
observed for weld samples at different welding current. The distortion of thin CRNO
sheets results in deterioration of material properties affecting the performance of
electrical systems. The present study aims to establish a relation between the surface
distortions at different locations across the weld zone with varying input welding
current.
1 Introduction
The development of highly improved and efficient electrical systems and appliances
is presently among the most critical issues along with environmental safety [1–3]. The
electrical steel and motor manufacturing industries have significantly transformed in
the areas of motor design and development to material processing and manufacturing
of laminated coatings for use in high-performance electrical systems [1, 2, 4–6]. The
CRNO electrical steel is a type of low carbon steel with the addition of alloys such as
Si and Al for improving the magnetic properties of electrical steels. They are basically
Fe–Si alloys with varying Si levels up to 6% depending on the industrial requirement
[2, 4, 5]. The CRNO electrical steel is cold rolled for varying and improving the
grain orientation in different directions for achieving the desired levels of magnetic
flux as required in rotating electrical systems such as motors. They are also the
preferred material for use in the stator cores of small- and medium-sized motors
mostly employed in hybrid electric vehicle (HEV) and electric vehicle (EV) [1, 3,
4, 7, 8]. The cold rolling operation is specially adopted for CRNO, post-annealing
for varying the grain orientation as compared to CRGO, prior to coating deposition
[5, 7].
The CRNO electrical steel of thickness 0.5 mm available in the form of sheet
rolls is cut by laser cutting and reduced to the required size by punching operation
[3, 5]. The punched sheets are further stacked and welded along the edges to form the
stator cores of electric motors [2, 5–8]. The manufacturing process commonly used
for joining the loose CRNO stacks is gas tungsten arc welding (GTAW) or tungsten
inert gas (TIG) welding. The GTAW welding process is widely used in industries
due to its easier operations and availability [3, 4, 9]. The other manufacturing pro-
cesses used for joining the loose stacks are: gluing, riveting, and interlocking [3, 8].
The manufacturing processes used for joining the stacks generate large amount of
thermal and mechanical stresses on the material surface resulting in higher losses
and affecting the performance of the electrical systems [3, 4, 6].
GTA welding generates large residual stresses and damages the material properties
of sheets [4, 9]. The CRNO sheets due to large distortion during high heat input in
welding result in lower efficiency levels, deteriorating the performance of CRNO
electrical steel sheets [3]. The present study aims to investigate the effect of GTA
welding on the surface distortion of the CRNO electrical steel sheets. The effect of
varying levels of GTA welding current on the surface distortion of thin CRNO sheets
is investigated.
2 Sample Preparation
The thin CRNO sheets (0.5 mm) of M-43 (AISI) grade is used in the study. The thin
sheets of sizes 100 mm × 40 mm × 50 mm are clamped using suitable fixtures and
welded in the direction perpendicular to the edge of sheets as shown in Fig. 1.
The sheets are clamped at a pressure sufficient to prevent any air gaps between
them. The filler material of grade SS308L as per AWS standards is used for joining the
coated samples. The chemical composition of base metal (M-43) and filler material
used is described in Tables 1 and 2.
A Study on the Effect of GTAW Input Current on Surface … 281
3 Experimentation
The stacks of CRNO post-welding are observed in the surface profilometer for
observing the distortion variation in the surface of CRNO sheets. The stack of sheets
is observed at a distance of ±2, ±1.5, ±1.0, ±0.5 mm from the weld center for
observing the distortion variations on the sheet surface due to the varying welding
current. The CRNO stacks welded at different welding current are measured as shown
in Fig. 3b.
The samples are observed under surface profilometer as shown in Fig. 3a, and the
pattern observed in the surface distortion variation is shown in Fig. 4 (for ±0.5 mm),
Fig. 5 (for ±1.0 mm), Fig. 6 (for ±1.5 mm), and Fig. 7 (for ±2.0 mm), respectively,
with reference to the weld center. Huge variations are observed in the direction away
from weld center with least distortion for ±0.5 mm and maximum for ±2 mm for
282
Fig. 3 a Experimental setup and b indication spots for measurement on the sheet surface
all sample cases (30–110 A). The intensity of variation is found to be maximum at
a distance of ±2 mm from the weld center in case of 110 A sample.
The maximum variation in range of 0.66 to −0.27 mm for ±2 mm sample, 0.46
to −0.21 mm for ±1.5 mm sample, 0.34 to −0.16 mm for ±1 mm sample, and 0.17
to −0.08 mm for ±0.5 mm sample is observed on the surface of stacked CRNO
samples welded in the range of 30–110 A.
5 Conclusion
In the present study, investigation on the effect of GTA welding current on the sur-
face distortion of the CRNO electrical steel of M-43 grade is done. The following
conclusions can be drawn from the above study:
1. The surface distortion is found to be increasing with the welding current mostly
affecting the material properties of thin sheets.
A Study on the Effect of GTAW Input Current on Surface … 287
2. The variation range in distortion on the surface of stacked CRNO sheets is found
to be increasing with input current, with least for 30 A and maximum for 110 A
weld samples.
3. The maximum variation in range of 0.66 to −0.27 mm for ±2 mm sample, 0.46
to −0.21 mm for ±1.5 mm sample, 0.34 to −0.16 mm for ±1 mm sample, and
0.17 to −0.08 mm for ±0.5 mm sample is observed.
4. The high heat input can also cause severe damage to the microstructural, mag-
netic, and mechanical properties of the thin sheets and can lead to reduction in
the performance and efficiency levels of the stator cores used in motors.
References
1. Honda A (2003), Electrical steel for motors of electric and hybrid vehicles. Kawasaki steel
technical report, vol 48, pp 33–38
2. Petrovic DS (2010) Non-oriented electrical steel sheets. Mater Technol 44(6):317–325
3. Kurosaki Y, Mogi H, Fujii H, Kubota T, Shiozaki M (2008) Importance of punching and
workability in non-oriented electrical steel sheets. J Magn Magn Mater 320(20):2474–2480
4. Lindenmo M, Coombs A, Snell D (2000) Advantages, properties and types of coatings on
non-oriented electrical steels. J Magn Magn Mater 215:79–82
5. Fischer O, Schneider J (2003) Influence of deformation process on the improvement of non-
oriented electrical steel. J Magn Magn Mater 254–255:302–306
6. Takashima M, Morito N, Honda A, Maeda C (1999) Nonoriented electrical steel sheet with low
iron loss for high-efficiency motor cores. IEEE Trans Magn, Part 2 35(1):557–561
7. Yabumoto M, Wakisaka T, Suzuki N, Kaido C, Kubota T (2003) Electrical steel sheet for traction
motors of hybrid/electric vehicles. Nippon steel technical report, vol 87, pp 57–61
8. Schoppa A, Schneider J, Wuppermann CD, Bakon T (2003) Influence of welding and sticking
of laminations on the magnetic properties of non-oriented electrical steels. J Magn Magn Mater
254(255):367–369
9. Zhang Y, Wang H, Chen K, Li S (2017) Comparison of laser and TIG welding of laminated
electrical steels. J Mater Process Technol 247:55–63. https://doi.org/10.1016/j.jmatprotec.2017.
04.010
Heat Transfer and Pressure Drop Inside
Duct with Different Surface Profiles
Abstract In this experimental study, the heat transfer rate with respect to the pressure
drop in rectangular duct is studied. The ribs with various profiles are used as an inserts
and heat transfer rate was determined. The rectangular duct of cross-section 700 ×
100 mm made up of mild steel was used. The experiments were carried for the
Reynolds number in the range 5000–22,000. Five cases were analyzed in this study.
Case 1 includes the rectangular duct with 300 mm as effective test section without
ribs. In Case 2, flat ribs were used at angle of 900 with respect to the direction of the
flow. V-shaped broken ribs inclined at an angle of 30°, 45°, and 60° with respect to
the direction of flow are used as inserts in case 3, 4, and 5, respectively. The readings
were noted at constant heat flux of test section. Initially, air was supplied with velocity
of 0.6 m/s and then it was gradually increased to 2.4 m/s. In this case, less pressure
drop is observed experimentally, as compared to transverse broken V-shaped ribs.
Keywords Heat transfer and pressure drop · Rectangular duct · V-shaped ribs
(30°, 45°, 60°)
1 Introduction
Naphon et al. [3] experimentally investigated the most important finding says with
respect to the increase in mass flow rate, the heat transfer rate also reaches higher
value. The heat transfer increase by 3.5 times when corrugated channel was used as
compared to the smooth channels, whereas pressure drop was observed 5–6 times
more in similar comparison [4].
Eiamsa-ard [5] showed the rectangular depth was varied from 0.1 to 0.3 in ratio
and tape twist ratio was fixed at 4. Due to the twisted tape, swirling action takes place
in flow direction and hence heat transfer rate increases.
S. Liu presented the study on passive heat transfer enhancement techniques [6]
that studied the effect of transverse and V-shaped broken ribs.
2 Experimental Setup
Figures 1 and 2 show the detailed view of experimental setup. The rectangular
channel is fabricated from mild steel. The acrylic glass having thermal conductivity
0.189 Wm−1 k−1 is used in the middle portion of the duct for 300 mm of its length
to observe the fluid flow.
The geometrical dimensions are shown in Fig. 2. For the calculation of heat
transfer coefficient, the concept of “Hydraulic diameter (Dh )” is used and its value
is 0.1333 m for the first 200 mm of test section the flow is allowed to enter the
developed phase [7]. After this, the air enters the heater section. Heater section is
300 mm long and 100 mm in width insulated by glass wool of 50 mm thickness [8].
After heater section for 200 mm, the flow is allowed to move freely. This is essential
to get stream line flow in the area of interest along the rectangular duct.
For uniform heat flow, plate-type heater is used which is fixed on the top portion
of the rectangular duct. The other sides of wall of the rectangular duct are well
insulated with commercial fiberglass. The eight copper–constantan thermocouples
having range of −160 to 400 °C are used for measurement of the temperature. Inlet
temperature of air is measured using one thermocouple and outlet temperature is
measured using three thermocouples. U-tube manometer is used as a flow measuring
device in the test section and also to measure the pressure drop along the length of
the duct (Fig. 3).
In Phase I, first experimentation was carried out when no ribs were present inside
the rectangular channel [9]. Transparent Plexiglas is fixed to the position to avoid
air leakage from outside into the channel. Set an input of 40 V, & 0.82 A to the
plate heater with the help of dimmer stat and waiting for the temperatures of test
section becomes steady. Start the exhaust fan and wait to reach the temperatures of
test section at the steady state.
Velocity is measured with hot wire anemometer by keeping the sensor in flow
stream. Note down the reading of temperature indicator and U-tube manometer.
Gradually increase the velocity (0.8, 1, 1.2, 1.4 m/s) by regulating air flow and note
observations for same input.
292 P. P. Shirpurkar et al.
In Phase II, solid aluminum ribs were used to enhance the heat transfer rate through
the rectangular duct. Inserts the ribs in the rib holder and ensure that ribs should
be properly placed. Insert ribs assembly inside the rectangular duct to cover by the
Plexiglas. Start the setup and follow the same procedure as above and note down the
readings.
1 0.6 20 84 86 85 84 84.75 22 23 21 22 –
2 0.8 20 83 86 85 84 84.5 22 22 22 22 –
3 1 20 82 85 84.5 83.5 83.87 22.5 22 22 22.16 –
4 1.2 20 82.5 84 84 84 83.62 22 22.5 22.5 22.33 –
5 1.4 20 82 83 83.5 83.5 83 22.5 22 23 22.5 –
6 1.6 20 81.5 82.5 83 83.5 82.62 23 22.5 22.5 22.66 0.5
7 1.8 20 81.5 81.5 82.5 83 82.12 23 22.5 23 22.83 0.5
8 2 20 81 81.5 82 82.5 81.75 23.5 23 23 23.16 0.8
9 2.2 20 81 81 81.5 82.5 81.5 24 23.5 23.5 23.66 1
10 2.4 20 80 80.5 81 82 80.87 24 24 23.5 23.83 1
293
294 P. P. Shirpurkar et al.
Table 2 Summary and calculation table for rectangular duct without ribs
S No. Velocity Mass Reynolds Heat Nusselt Pressure Friction
flow Number coefficient Number drop factor
rate (Re) (Wm2 K) (Nu) (N/m2 ) r(f )
(m)
kg/s
1 0.6 O 5155 14.6 74.1 – –
2 0.8 O 6874 19.9 101 – –
3 1 O 8591 27.5 139 – –
4 1.2 O 10,310 34.8 176 – –
5 1.4 O 12,029 44.5 225 – –
6 1.8 O 13,747 55 278 4.9 1.43
7 2 O 15,465 66.8 338 4.9 1.13
8 2 O 17,183 82.3 416 7.84 1.47
9 2.2 O 18,902 106 538 9.81 1.52
10 2.4 O 20,621 123 625 9.81 1.27
The heat transfer rate for its enhancement is studied for various cases. These various
cases include channel without ribs (smooth channel), channel with broken V-shaped
ribs and the channel with flat transverse ribs. These various cases are compared with
each other and the conclusions were drawn.
The comparison shows that the channel with broken V-Shaped ribs with angle
of inclination 60° is more efficient as that of smooth channel. There was 51.66%
of enhancement in heat transfer when V-shaped broken ribs were used. In case of
flat ribs, the heat transfer rate gets enhanced by 42.52% as compared to the smooth
channel (Table 3).
Figure 4 shows the relative change in heat transfer coefficient with respect to
the mass flow rate. The heat transfer coefficient increases monotonically with the
increase in mass flow rate and it will cause turbulence in flow and hence, the heat
transfer increases. The flow channel without ribs and with ribs of broken V shape
with angle of inclination 30°, 45°, 60° and flat ribs are shown in figure. The V-shaped
ribs with 60° angle of inclination show the maximum heat transfer rate whereas the
channel flow with no ribs shows minimum heat transfer rate. It is also observed from
the data collection that the channel with flat ribs and V-shaped ribs with 45° show a
close accordance in their data points.
Friction factor shows the nature of obstruction in flow; Fig. 5 shows that with
increase in Reynolds’s number friction factor decreases. It is evident from Fig. 6 for
some particular values of Reynolds number, the peak value of heat transfer coefficient
ratio is observed. After Reynolds number, 10,000 V-shaped ribs with 60° inclination
shows drop in heat transfer coefficient, whereas for the V-shaped ribs with angle
of inclination of 30° shows gradual increase in heat transfer coefficient ratio. From
Heat Transfer and Pressure Drop Inside … 295
Fig. 6, it is observed that maximum heat transfer coefficient ratio is obtained for
600 ribs with respect to without ribs. Minimum heat transfer coefficient ratio was
obtained for 300 as compared to 600, 450, and flat ribs. Heat transfer coefficient ratio
has a direct relationship with heat transfer rate. The heat transfer coefficient ratio
increases with increase in heat transfer rate.
296 P. P. Shirpurkar et al.
5 Conclusions
Heat transfer phenomenon was studied in rectangular duct with air flow in the channel
with heat transfer mechanism along the passage of flow. The effect of various param-
eters like mass flow rate, Reynolds’s number on the heat transfer rate is discussed
below:
Heat Transfer and Pressure Drop Inside … 297
1. The V-shaped ribs with 60° angle of inclination when used as inserts shows
maximum heat transfer rate. When compared to the channel without ribs, the 60°
inclined V-shaped ribs gives 51.66% more enhancement in heat transfer.
2. When compared to the channel with flat ribs, the 60° inclined V-shaped ribs
give 42.52% more enhancement in heat transfer. The inserts added in the flow
passage make the flow turbulent and hence, the enhancement in the heat transfer
was observed.
References
1. Uniyal M, Gupta A (2013) Experimental and numerical study on forced convection for planed
and dimpled surface in a long triangular fin bar. Int J Eng Res Technol (IJERT) 2:1–6. ISSN:
2278-0181
2. Dutta P, Hossain A (2005) Internal cooling augmentation in rectangular channel using two
inclined baffles. Int J Heat Fluid Flow 26:223–232
3. Naphon P, Wongwises S (2006) A review of flow and heat transfer characteristics in curved
tubes. Renew Sustain Energy Rev 10(3):114–121
4. Takeishiet K, Miyake Y, Motoda Y, Oda Y (2013) Convective Heat transfer and pressure loss
in rectangular ducts with inclined pin-fin on a wavy end wall. J Eng Gas Turbines Power
135:061902-1–061902-10
5. Eiamsa-ard P, Eiamsa-ard S, Thianpong C, Influences of twisted- tape with parallel rectangular-
wing on thermal performance of a heat exchanger. King Mongkut’s Institute of Technology
Ladkrabang, Bangkok 10520, Thailand
6. Tanda G (2004) Heat transfer in rectangular channels with transverse and V-shaped broken ribs.
Int J Heat Mass Transf 47:229–243
7. Kharche SS, Farkade HS (2012) Heat transfer analysis through fin array by using natural
convection. Int J Emerg Technol Adv Eng 2(4):595–598 ISSN 2250-2459
8. Pimsarn M, Promvonge P, Sriromreun P, Augmented heat transfer in rectangular duct with angled
Z-shaped ribs. King Mongkut’s Institute of Technology Ladkrabang, Bangkok 10520, Thailand,
pp 1–7
9. Ahn SW, Son KPH (2002) An investigation on friction factors and heat transfer coefficients in
a rectangular duct with surface roughness. KSME Int J 16(4):549–556
A Hybrid Process Monitoring Strategy
for Steel Making Shop
Abstract The article deals with the development of a process monitoring strategy for
a Steel Making Shop (SMS) involving an ensemble of statistical and AI techniques.
The monitoring strategy being devised was employed primarily to demonstrate the
monitoring of nonlinear processes. The monitoring strategy was based on neural
network fitting model and Hotelling T 2 control chart. Data pertaining to process and
feedstock characteristics of Steel Making Shop (SMS) was considered for checking
the efficacy of the monitoring strategy being devised. The neural network fitting
model is used for partial or full transformation of the nonlinear data into linear data.
Thereafter Hotelling T 2 chart was employed on the transformed data for monitoring
of the process. The test of nonlinearity of the data involved pairwise comparison of
any two characteristics by plotting them in a fitted line plot and computing the model
fit value which is indicative of the level of linearity. The hybrid strategy involving
the neural network model and Hotelling T 2 square chart thus devised was able to
monitor the process and detect out of control observation correctly.
Keywords Neural network fitting model · Hotelling T 2 chart · Steel making shop
(SMS)
1 Introduction
Based Statistical Monitoring Strategies (MPSMS) [3, 4]. For modern manufactur-
ing processes laden with hundreds of process, feedstock and quality characteristics,
Multivariate Projection Based Method have found widespread application owing to
their ability to deal with large number of highly correlated characteristics simultane-
ously. The Multivariate Projection Based Techniques are mainly based on Principal
Component Analysis [5, 6] and Partial Least Square Regression [7].
The MPSMS being discussed in the preceding paragraph are suitable for linear
processes and if applied to processes which deviate substantially from linearity, the
said processes may fetch erroneous result. A process may be deemed as nonlinear
if any two characteristics associated vary nonlinearly. To address the issue of non-
linearity, various techniques like Kernel Principal Component Analysis (KPCA) [8],
Bayesian inference [9] and some Artificial Intelligence based techniques like artifi-
cial neural network [10, 11] have been developed. An unsupervised neural network
[12] based technique known as auto-encoder [11] with nonlinear encoding and lin-
ear decoding can be used to transform a nonlinear data matrix into linear and then
conventional multivariate technique can be applied to get the desired results.
In this article, a hybrid process monitoring strategy is adopted to monitor the
process and feedstock characteristic of a Steel Making Shop (SMS). The hybrid pro-
cess monitoring strategy being devised consists of an artificial intelligence tool viz:
Neural Network Fitting (NNF) Model (a type of auto-encoder) used in conjunction
with Multivariate Statistical Process Control (MSPC) Technique viz: Hotelling T 2 .
The NNF Model was used for the reduction of nonlinearity in the data and then the
Hotelling T 2 chart was applied on the transformed data for monitoring of the SMS.
−1
T
x2 = X − μ X −μ (1)
where = sample covariance matrix
μ = sample mean vector.
Multivariate chi-squared control charts can be constructed by plotting x 2 versus
sample with upper control limit given by x 2 α, q where α is the level of significance
and q is the degree of freedom. When the in control covariance matrix is not known
it can be estimated by a sample from past n data the new multivariate statistic is then
given by T 2 statistics
A Hybrid Process Monitoring Strategy for Steel Making Shop 301
T
T 2 = X − μ S −1 X − μ (2)
T 2 control chart can be plotted with upper control limit given by T 2 UCL .
(n − 1) × (n + 1)q
2
TUCL = Fα, (q, n − q) (3)
n(n − q)
(n + 1) × (n + 1)q
2
TUCL = Fα, (q, n − q) (4)
n(n − q)
z = α(x) σ (W X + b) (5)
and b is a biased vector. Bias vector is an extra input which is supplied to the node to
give an output even when all other inputs are zero. The decoder part maps the feature
vector z to
x of same dimensionality.
x = β(z) wz +
σ ( b) (6)
Since linear decoding has been used, the obtained output is of linear nature and
multivariate T 2 chart can be used to establish the control limit and detect the out of
control signal.
The process is monitored in this study is a Steel Making Shop (SMS) which is used
in the production of steel billets of various grades. The SMS under consideration
consists of an Electric Arc Furnace (EAF), Ladle Refining Furnace (LRF) and a
Continuous Casting Machine (CCM). Six processes and feedstock characteristic
pertaining to the EAF was used for monitoring of the process. The list of characteristic
being monitored is listed in Table 1.
Observations pertaining to the characteristic were collected and after pretreatment
which involved the removal of outliers, 60 observations remained in the fray. The data
was then standardized to bring them on an equal scale. Thereafter linearity test was
carried out between any two pairs of characteristics. A fitted line plot was constructed
between the pair of characteristic and the model fit value (R2 ) was calculated. The
model fit value of the fitted line plot can be considered as a level of linearity. For the
given data the R2 value is found to be 0% which tells that the data is totally nonlinear
in nature.
Thereafter the data is transformed from nonlinear to linear using Neural Network
Fitting (NNF) Model and again linearity test has been performed using the R2 test.
It can be seen from the result that the data has been partially transformed into lin-
ear one since the new R2 value of transformed data is around 52%. The schematic
representation of the NNF Model is being depicted in Fig. 1.
number six has been found to be out of control which requires further diagnosis to
unearth the possible cause for the same (Fig. 2).
4 Conclusions
References
1. Montgomery DC (2004) Introduction to statistical quality control, 4th edn. Wiley, New York
2. MacGregor JF, Kourti T (1995) Statistical process control of multivariate processes. Control
Eng Pract 3:403–414
3. Akeem AA, Abubakar Y, Osebekwin A (2015) Hotelling’s T 2 decomposition approach for
five process characteristics in a multivariate statistical process control. Am J Theor Appl Stat
4(6):432–437
4. Mostajeran A, Iranpanah N, Noorossana R (2016) A new bootstrap based algorithm for
Hotelling’s T 2 multivariate control chart. J Sci Islamic Repub Iran 27(3):269–278
5. Zhou L, Junghui C, Beiping H, Zhihuan S (2018) Multi-grade principal component analysis
for fault detection with multiple production grades. Chemometr Intell Lab Syst 175(4):20–29
6. Suman S, Das A (2019) Stratified statistical monitoring strategy for a multi-product manufac-
turing facility with early detection approach. Comput Ind Eng 130:551–564
7. Wang H, Liu Q, Tu Y (2005) Interpretation of partial least-squares regression models with
VARIMAX rotation. Comput Stat Data Anal 48(1):207–219
8. Nguyen VH, Golinval JC (2010) Fault detection based on kernel principal component analysis.
Eng Struct 32:3683–3691
9. Zhiqiang G, Muguang Z, Zhihuan S (2010) Nonlinear process monitoring based on linear
subspace and bayesian inference. J Process Control 20(5):676–688
10. Hsieh WW (2001) Nonlinear principal component analysis by neural networks. Tellus A
53(5):599–615
A Hybrid Process Monitoring Strategy for Steel Making Shop 305
1 Introduction
For enhancing these properties, the hard particulate and/or fiber-type reinforcement
phases such as SiC, Al2 O3 , TiC and B4 C are introduced into the matrix material [7],
and these have extensive demands in various industries like aerospace, automobiles
and turbines [8]. Conventional machining of MMCs causes serious tool wear because
of abrasive nature of high reinforcement content, thereby shortening the life of tool
[9]. High hardness and reinforcement make it challenging to machine by conven-
tional method and have obstructed the growth of MMCs in the industrial applications
[10]. Thus, the advanced machining process has been adopted for machining of these
materials.
Simulation of the dynamic flow field of electrolyte has been done to improve the
machined flatness of electrochemical mill grinding, and it is found that electrolyte
velocity was significantly increased at edge region and decreased at the front of
middle region [11]. Experiments and simulation are carried out on machining of flat
surfaces using COMSOL software in inner jet ECG [12]. Role of electrolyte flow
in the generation of oxide layer on the workpiece by electrolytic dissolution and its
removal through grinding action of wheel and erosion action of electrolyte flow have
been studied [13]. Numerical modeling of material removal rate [MRR] in micro-
electrochemical grinding by electrolytic and grinding action has been carried out
[14]. Goswami et al. [15] studied the effect of process parameters on the machining
performance during the electrochemical grinding of alumina–aluminum composite.
But, the study of erosion caused by turbulence flow of electrolyte in inter-electrode
gap [IEG] is needed, for evaluation of shear stress and MRR during the machining
of MMCs.
Therefore, this paper presents the simulation and modeling of the micro-ECG
for analysis of turbulence flow in IEG and also shows the effect of turbulence flow
of electrolyte at various wheel speeds on shear stress generation and MRR during
the machining of MMC. In this MMC, alumina is reinforced into aluminum matrix
(Al2 O3 –68.9% and Al–31.1%) [15].
2 Model Information
Flow of electrolyte through IEG plays a significant role in the performance of elec-
trochemical grinding process [16]. A huge number of charge ions are present for
electrolytic dissolution at higher flow rate, consequential in higher material removal
rate [17]. Higher flow of electrolyte and higher rotational speed of wheel increase the
turbulence in the electrolyte flow. The turbulence in the electrolyte flow poorly affects
MRR of machined components. On other hand, shear force on interface between
wheel and workpiece reduces when flow of electrolyte decreases which badly marks
the removal of oxide layer [18]. Therefore, critical analysis of electrolyte flow rate
Analysis of Electrolyte Flow in IEG During … 309
should be in a way that MRR may increase and also avoid the turbulence at the same
time.
2.2 Assumptions
τ
Flow
IEG
Wheel Boundary
τ
IEG
Workpiece Boundary
For the analysis of turbulent flow, basically, two governing equations, i.e., mass
balance equation and momentum balance equation, are used. Since the flow of elec-
trolyte through the IEG is steady and incompressible, the mass balance is given by
continuity equation as follows:
∂U ∂V
+ =0 (1)
∂X ∂Y
A two dimensional form of time-averaged Navier–Stokes equation is used for
balancing the momentum in X- and Y- directions. The non-dimensional form of
momentum equation can be written as,
DU j ∂ P + 2K n /3 1 ∂ ∂U ∂U
=− + × 1 + vt,n + (2)
Dt ∂Xj Re ∂ X i ∂Xj ∂ Xi
As can be seen from Fig. 1, there are four boundaries in the computational domain.
The boundary conditions for mass (Eq. 1) and momentum (Eq. 2) are given as follows.
• At x = 0 (electrolyte IN), U = U i , V = 0
• At y = 0 (work piece boundary), U = 0, V = feed (m/s)
• At wheel boundary, velocity = W w × d/2
• At electrolyte OUT, x = d, P = Pa .
0.07
0.06
0.05
0.04
X[m]
0.03
0.02
0.01
0
0 2e+06 4e+06 6e+06 8e+06 1e+07 1.2e+07
Wall Shear [Pa]
Figure 3 shows the effect of wheel rotational speed on shear force. It is clear from
Fig. 3 that the shear force increases as rotational speed of grinding wheel increases.
Since turbulence flow is a function of velocity, so as rotational speed of grinding
wheel increases, the turbulence action of electrolyte in inter-electrode gap increases
and the velocity gradient in inter-electrode gap also increases. Due to this, shear
stress on the workpiece surface increases. Finally, viscous shear force increases on
workpiece boundary.
200
190
180
170
Shear Force (N)
160
150
140
130
120
110
100
2000 2200 2400 2600 2800 3000 3200
Rotational Speed of Grinding Wheel (rpm)
εC I
MRRECM = (4)
zρ F
The calculation of MRR due to mechanical action during the ECG process is
calculated by the following equation.
MRRAbrasion
⎛ ⎛ ⎞ ⎞
θ
21
⎜ ⎜ 2 ⎟ ⎟
⎜ ⎜ p × (d + a) − 2 r 2 − y 2 × t + ρgtr r − a ⎟ ⎟
⎜ ⎜ r 3 ⎟ ⎟
⎜ ⎜ ⎟ ⎟
⎜ ⎜ ⎟ ⎟
⎜ ⎜ H w ⎟ ⎟
⎜ ⎜ 1+ ⎟ ⎟
⎜
dmax ⎜ ⎟ ⎟
⎜ ⎜ Hp dg ⎟ dl ⎟
= ⎜dg × ⎜ − ⎟× ⎟
⎜ ⎜ 10 2 ⎟ dt ⎟
⎟
⎜ ⎜ 1
× (vol. fraction of abrasive) 3 ⎟
dcont ⎜ ⎜ ⎟ ⎟
⎜ ⎜ dmean ⎟ ⎟
⎜ ⎜ ⎟ ⎟
⎜ ⎜ 1 ⎟ ⎟
⎜ ⎜ (H w × V w × t × Prob. of active grains × Coeff. of friction) 2 ⎟ ⎟
⎜ ⎜ ⎟ ⎟
⎝ ⎝ ⎠ ⎠
∞
1 x2
× NTotal× √ e− 2 dx (5)
2π
p
Finally, the MRRErosion was evaluated knowing the magnitude of shearing forces
in the IEG due to electrolytic flow and given by
where
Power( p)
Shear Velocity = (7)
FSimulation
P = ρ Qg H (8)
where
Q = Vavg x AIEG .
ρ(Vavg × AIEG ) × g × H
MRRErosion = A s × ρh (9)
FSimulation
Due to relative motion between wheel and workpiece, the electrolyte enters
between wheel and workpiece, i.e., IEG splashes out as shown in Fig. 1; thus,
availability of flow power of electrolyte as well as shear force on the workpiece
reduces.
0.016
0.015
MRR erosion (mg/min)
0.014
0.013
0.012
0.011
0.01
2000 2200 2400 2600 2800 3000 3200
Rotational Speed of Grinding Wheel (rpm)
6 Conclusions
The foregoing simulation, analysis and discussions lead to the following conclusions
about micro-ECG of aluminum–alumina composite.
1. It may be observed that with increase in the flow velocity of an electrolyte at
inlet, the shear force significantly increases from 115.51 N to 187.49 N due to
increase in wheel speed from 2100 to 3000 RPM. It has also been noticed that
the maximum shear stress on the machined surface is obtained around IEG, and
moving away from IEG, minimum value of shear stress is obtained.
2. Wheel speed significantly affects the material removal rate by erosion action.
The total material removal rate increases by approximately 1% by considering
erosion action.
Analysis of Electrolyte Flow in IEG During … 315
References
Abstract The objective of this research is to analyze roll cage of single-seated off-
road all-terrain vehicle which can sustain rough terrain in all types of weather without
damage. The capability of these vehicles includes acceleration, maneuverability,
rock-crawl and hill climbing on surfaces. A roll cage is considered as a framework of
an all-terrain vehicle (ATV). It forms a three-dimensional structure around the driver.
It consists of steel members. The roll cage of vehicle plays crucial role in safety of the
driver. It contains driver, engine, brake system, steering mechanism. Roll cage must
have suitable strength in order to protect the driver in event of rollover or impact. Roll
cage safety features were first implemented in accordance with the 2019 Baja SAE
Competition Rules [1], which served as a baseline for roll cage design. Design of
roll cage is done in CATIA V5R20, and static structural analysis of roll cage is done
using ANSYS Workbench 16 for various collisions like front, rear, side, roll over
and vibration analysis. To obtain required factor of safety (FOS) in all the loading
conditions is the main objective of analysis of design.
1 Introduction
Baja SAE India is an inter-college national level design competition managed by the
Society of Automotive Engineers (SAE). Teams run by students from all over the
Nation, design and build single-seated off-road vehicle. This event in India is lead
by some of the top automobile industries.
The objective of this research is to design and analyze roll cage of all-terrain
vehicle. A roll cage is better described as a rollover protective structure. The primary
purpose of roll cage is driver’s safety, ergonomics and integration with all the other
sub-systems like engine, brake, steering and suspension [2]. In order to attain high
performance, roll cage plays a vital role. If a roll cage of higher weight is used, it
will reduce the altogether performance of vehicle. So, material selection plays an
important role in roll cage design. After the design of roll cage, it must be ensured
that it is safe for a driver in worst scenario. For that, there is a need to perform
analysis of roll cage. This paper gives a generalized idea about design parameters
and analysis performed on roll cage.
2 Design Methodology
3 Material Selection
Material selection of a roll cage has major impact on reliability, performance and
safety of vehicle [3]. Hence, the material selection process is carried out considering
the physical and chemical properties of various materials. The material selected
should have high rigidity, high bending strength and stiffness, stability and less
Static Structural Analysis of Roll Cage of an All-Terrain Vehicle 319
weight. Considering the recommended material from rulebook and the market survey,
two different options of materials are considered for roll cage. The two different
materials considered for roll cage are AISI 1018 and AISI 4130. Table 2 shows the
comparison [4] between physical and chemical properties of AISI 1018 and AISI
4130.
From above comparison between physical and chemical properties of AISI 1018
and AISI 4130, it can be stated that AISI 4130 has better yield strength, tensile
strength and carbon %. Hence, the material used for roll cage is AISI 4130 (Table 3).
Table 3 Comparison
Properties AISI 1018 AISI 4130
between properties of AISI
1018 and AISI 4130 Density 7870 kg/m3 7850 kg/m3
Tensile strength 440 MPa 560 MPa
Yield strength 317 MPa 460 MPa
Elasticity 200 GPa 200 GPa
Iron % 98.81–99.26 97.03–98.22
Carbon% 0.14–0.20 0.28–0.33
320 S. Satputaley et al.
Analysis is performed to calculate the safety offered by the roll cage to the driver in
case of any accident. The type of analysis performed on roll cage is static structural.
While meshing, the three-D element type is considered, and the element size is kept as
5 mm [5] as element size is directly proportional to percentage of error and inversely
proportional to solving time.
The force calculation at which vehicle gets impact is done by time-momentum
equation which is as follows:
M x V f − Vi
F=
T
where F = Force of impact, M = Mass, T = Time of impact.
It is a collision in which there are chances of two vehicles crashing into each other
head on during a race or event. More the time of impact the lesser the force experi-
enced by the driver, and time of impact depends on the elasticity of the material used
[5]. The time of impact is ideally considered 0.15 s.
While performing the front analysis, force is applied on front nose and front hitch
point while keeping rear suspension part fixed (Fig. 2).
With the help of impulse–momentum equation:
Velocity of vehicle = 40 km/h = 11.11 m/s, weight of ATV (including Driver) =
210 kg
f ×t =m×v
(m × v) 210 × 11.11
f = = = 15,554 N
t 0.15
From above analysis, the value of shear stress is less, i.e., 173.27 than tensile
strength of material, i.e., 560 MPa, and FOS is exceeding 1.2, i.e., 1.327 with 0.66 mm
deformation. When the factor of safety is greater than or equal to 1, then the maximum
stress is more than or equal to applied stress, so the roll cage can withstand the load.
Static Structural Analysis of Roll Cage of an All-Terrain Vehicle 321
But when the ratio of maximum stress to applied stress is equal to 1, the roll cage is
tough enough to withstand load. Hence, roll cage can sustain front impact.
322 S. Satputaley et al.
In case of a rear impact, an ATV collides to the back part of another ATV. As the
structure of an ATV is deformable in nature, hence the impact time is more than
compared to an ordinary vehicle. The impact time of ATV is 0.3 s. For analysis
purposes, the front suspension part is considered fixed, and force is applied on rear
members. Force due to virtue of velocity is considered as 40 km/h (Fig. 3).
By using impulse–momentum equation:
Velocity of vehicle = 40 km/h = 11.11 m/s, weight of ATV (including driver) =
210 kg
(m×v)
f = t
= 210×11.11
0.3
= 7777 N
From above analysis, the value of shear stress is less, i.e., 174.5 than tensile
strength of material, i.e., 560 MPa, and FOS is exceeding 1.2, i.e., 1.317 with 1.06 mm
deformation. Hence, roll cage can sustain rear impact.
In case of a side impact, an ATV hits another ATV directly perpendicular on the side
of another ATV, for analysis purposes, the impact time is considered 0.3 s, and force
corresponding to velocity is 40 km/h.
The side opposite to the side of collision is kept stationary for analysis purpose,
and force is applied on the side of rear roll hoop (RRH) and side impact member
(SIM) (Fig. 4).
By using impulse–momentum equation:
Velocity of vehicle = 40 km/h = 11.11 m/s, weight of ATV (including driver) =
210 kg
f = (m×v)
t
= 210×11.11
0.3
= 7777 N
From above analysis, the value of shear stress is less, i.e., 151.85 than tensile
strength of material, i.e., 560 MPa, and FOS is exceeding 1.2, i.e., 1.616 with 1.77 mm
deformation. Hence, roll cage can sustain side impact.
In this case, the ATV is assumed to be plummeted upon its roof from a height of
10 feet on the ground. A distance of 10 feet is significantly higher than any case
possible from event scenario. As ground is non-deformable in nature, so impact time
is assumed as 0.15 s for analysis purposes. The forces applied on roll hoop overhead
(RHO) members and lower suspension points are kept fixed (Fig. 5).
324 S. Satputaley et al.
Velocity of vehicle = 27.5 km/h = 7.73 m/s, weight of ATV (including driver) =
210 kg
(m × v) 210 × 7.73
f = = = 10, 822 N
t 0.3
Static Structural Analysis of Roll Cage of an All-Terrain Vehicle 325
From above analysis, the value of shear stress is less, i.e., 109.21 than tensile
strength of material, i.e., 560 MPa, and FOS is exceeding 1.2, i.e., 2.355 with 1.06 mm
deformation. Hence, roll cage can sustain front roll over.
326 S. Satputaley et al.
5 Conclusion
A roll cage is essentially designed for a rough terrain. This particular structural
analysis is beneficial for irregular roads especially in rural areas. The structure of the
vehicle is kept simple intentionally considering its manufacturing point of view. The
maximum shear stress values for front, rear, side, roll over impacts are 173.27 MPa,
174.5 MPa, 151.85 MPa, 109.21 MPa, respectively, which is less than tensile strength
of material, i.e., 560 MPa, and factor of safety in all the cases exceeds 1.2. Hence,
from the above analysis, it can be concluded that the roll cage has satisfying factor
of safety with minimum deformation, and it is safe for the driver in any terrain.
Static Structural Analysis of Roll Cage of an All-Terrain Vehicle 327
References
1 Introduction
Performance of vehicle not just depends upon good powertrain or fine aesthetics,
but on good levels of NVH (noise, vibration and harshness). Nowadays, vehicles are
expected to be reliable and comfortable. Comfort of the vehicle depends upon engine
noise, exhaust note and rattle noise between components, vibrations transferred to
S. S. Khune
VIT Vellore, ARAI Academy, Vellore, India
A. R. Bhende (B)
Mechanical Engineering Department, St. Vincent Pallotti College of Engineering & Technology,
Nagpur, Maharashtra 441108, India
e-mail: abhende@stvincentngp.edu.in
human body, etc. In case of motorcycles, vibrations are transferred to human body
from three points, i.e. handlebar, foot pegs and seat. Long exposure of vibrations to
rider may lead to several body disorders in ligament, bones and joints, and some-
times, it may affect the nervous system. Occurrences of such disorders are categorised
under whole-body vibration syndrome (WVBS). The disorders caused due to han-
dlebar vibrations are categorised under hand-arm vibration syndrome (HAVS) [1].
Number of international organisations have provided standards and guidelines for
measurement and evaluation of vehicle vibration such as ISO 2631 or BS 6841 [2].
Standards and guidelines for the measurement and evaluation of hand transmitted
vibrations are provided in ISO 5349-1 and ISO 5349-2 [1].
The present study is carried out to investigate the vibrations in handlebar of a
motorcycle and incorporate the modification to reduce vibrations for good ride com-
fort without affecting its handling. Handlebar is modified using tuned mass damper
to attenuate vibration amplitudes. Vibration measurements are carried out for orig-
inal (stock) and modified handlebar. It has been observed that maximum amplitude
value is reduced by substantial margins using modified handlebar.
2 Literature Review
spring and damper of TMD are replaced with actuation mass. Actuation mass is
of electromagnetic type which can be changed by merely changing coil current [6].
From the various studies carried out related to tuned mass damper system, natu-
ral frequency, amplitude and mass of the main system are the inputs for design of
TMD. TMD design is significantly dependent upon location of installation and the
production variability. If TMD is installed on a system without studying the natural
frequency of the system, then the modified design may fail due to resonance phe-
nomenon. Production variability is another crucial factor as no two components can
be manufactured exactly same with different natural frequencies.
3 Experimental Setup
4 Experimentation
Experimentation is carried out on stock handlebar for four different cases at various
RPMs. The four cases are as follows:
1. Plain handlebar (HB)
2. Bar ends attached to handlebar (HB + BE)
3. Lever, mirror and switches mounted on handlebar (HB + LSM)
4. Bar ends and other accessories installed on handlebar (HB + BE + LSM)
Design of Motorcycle Handlebar for Reduction of Vibrations … 333
Vibration signal is analysed in time domain and frequency domain plot to check
acceleration amplitude and its frequency. The acceleration and frequency readings are
recorded for each case as mentioned above. Stock handlebar is modified by inserting
designed TMD bars. The above procedure is repeated with modified handlebar.
The acceleration values are recorded and tabulated for the stock and modified han-
dlebar at various RPMs and at above mentioned four cases. Table 1 shows handlebar
(HB) reading for stock and modified handlebar. The tabulated data are graphically
represented: in Figs. 3 and 4.
For all the four cases mentioned above, vibration amplitudes are highest at 3750
engine RPM for stock handlebar. Natural frequency of stock handlebar is 392.5 rad/s
as per calculation, assuming stiffness of quad clamp as 394.384 kN/m and handlebar
bar mass as 2.56 kg. The external excitation engine frequency of a four-stroke engine
is 62.5 Hz or 392.5 rad/s. This indicates that external excitation frequency becomes
equal to natural frequency of handlebar, and system resonates with higher amplitudes
at 62.5 Hz.
TMD helps to minimise the vibration amplitude of the system but at the same time
adds weight to the system. However, to certain point, TMD becomes more effective
with increase in mass. Literature survey showed that 13.3% of handlebar mass can
be used as TMD mass which would attenuate the response and give optimum results.
Percentage attenuation achieved for above four mentioned cases is 23.5, 31.25, 65.9
and 56, respectively. The peaks have shifted in case of modified handlebar due to
shift in natural frequency.
6 Conclusion
The study presents successful implementation of tuned mass damper (TMD) for
attenuation of handlebar vibration in all the mentioned cases. Best results are obtained
in case of HB + LSM in both stock and modified handlebar. It is inferred that
location of the additional weight plays crucial role in the vibrational behaviour of the
handlebar. Addition of weight at the ends of the handlebar would result in increase of
handlebar vibrations. It is observed that weight should be uniformly distributed over
the span of the handlebar for attenuation of handlebar vibrations. From the study,
it is also observed that the effect of TMD is substantially visible at higher engine
Design of Motorcycle Handlebar for Reduction of Vibrations … 335
Conflict of Interest The authors declare that they have no conflict of interest.
Ethical Approval All procedures performed in studies involving human participants were in accor-
dance with the ethical standards of the institutional and/or national research committee and with
the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
Ethical Approval This article does not contain any studies with human participants or animals
performed by any of the authors.
References
1. Fasana A, Giorcelli E (2010) A vibration absorber for motorcycle handles. Meccanica 45:79–88
2. ISO 2631-1, Second Edition, 1997, Mechanical vibrations and shock—evaluation of human
exposure to whole body vibrations
3. Sivakumar A, Tiwari A, Ragavendra G, Kolhe V (2009) Elastomer blend for vibration isolators
to meet vehicle key on—key off vibrations and durability. SAE Technical Paper
4. Tathawadekar P, Liu K-J, Rajan S, Johnson P, Application of tuned mass damper to address
discrete excitation away from primary resonance frequency of a structure
5. Barbetti M, Carvalho L, Calcada M (2017) Vibration absorber application, case study: mid-size
truck steering wheel vibration. SAE Technical Paper
6. Lee P, Rahbar A (2005) Active tuned absorber for displacement-on-demand vehicles. SAE
Technical Paper
Development of a Diagnostive Tool
for Prediction of Severity of Coronary
Artery Disease
Abstract The objective of the present work is to assess the percentage rise in pres-
sure in stenosed coronary artery using computational fluid dynamics (CFD) and
develop a diagnostive tool to predict the extent of stenosis severity. Correlation for
blood pressure with cross-sectional area stenosis (AS) is included in this study. A
healthy artery model and a total of 15 coronary stenosis in three different stenosed
models are analyzed. The hemodynamic parameters computed through CFD are used
to determine blood pressure in stenosed models during systole as well as diastole. It
is observed that stenosis is critical for % AS > 80%. CFD results are then mapped
by standard curve fitting techniques to develop a mathematical model. A good and
significant correlation between blood pressure and % area stenosis is found. The
developed mathematical model is further used to develop an inexpensive and handy
diagnostive tool for preliminary diagnosis of severity of coronary artery disease by
clinicians.
1 Introduction
Coronary artery disease (CAD) is one of the leading causes of high blood pressure and
heart attack in human. Heart attack or in clinical terms myocardial infarction occurs
when blood flow to a part of heart decreases or stops. Physiological importance of
stenosis severity cannot be ascertained by just depending on the modern imaging
techniques such as angiography, magnetic resonance angiography (MRA) and intra-
vascular ultrasound (IVUS) especially for intermediate area stenosis (AS—it defines
blockage in artery based on reduction in lumen cross-sectional area). These modern
techniques are used by radiologists to find the exact location and severity of CAD.
However, these diagnostive methods are expensive and are also inaccurate for lumen
2 Methods
In order to develop a diagnostive tool which correlates the percentage area stenosis
(AS) in coronary artery with increase in blood pressure, sixteen 3-D models of
coronary artery are generated. In many previous literatures [3, 4], it is mentioned
that the geometry of a perfect coronary artery is nearly symmetrical and cylindrical.
In the current study, the three-dimensional arterial models are generated with this
assumption. Based on clinical data, the healthy coronary artery is modeled as of
16 mm in length and 4 mm in diameter.
In clinical language, blockage in artery is specified in terms of % DS. However, the
hemodynamic parameters are the functions of % AS. For the same % DS, the % AS
varies depending on the shape and orientation of plaque deposition (refer Table 1).
The hemodynamic parameters behave differently for same % DS as % AS can be
different. Henceforth in this work, % area stenosis is considered to determine severity
of stenosis. Three cases of stenosis model: (1) asymmetric plaque deposition (ASD),
(2) one-sided plaque deposition (OSD) and (3) symmetric plaque deposition (SD)
as shown in Fig. 1 are considered. For each stenosed model, five cases of diameter
stenosis (DS), 25, 50, 66.67, 75 and 83.3% are investigated. In all the three cases of
stenosed models, the shape of plaque is assumed to be circular. Although the three
stenosed models are generated with same extent of DS, these differ with each other in
Table 1 % AS of stenosed
% DS Stenosed artery model
artery models corresponding
to their % DS ASD OSD SD
25 14.3 19.51 43.75
50 39 49.97 75
66.67 58.4 71.8 89.44
75 68.4 80.44 93.75
83 78.78 88.72 97.11
Development of a Diagnostive Tool for Prediction … 339
Fig. 1 3-D stenosed artery models of 66.67% DS for (1) asymmetric plaque deposition (ASD), (2)
one-sided plaque deposition (OSD) and (3) symmetric plaque deposition (SD)
terms of AS. Therefore, these arterial stenosed models result in 15 different models
which vary from 14.3 to 97.22% in terms of area stenosis (AS) as shown in Table 1.
All the 16 arterial models (healthy artery and 15 stenosed arteries) are designed
in SolidWorks 2013. Then, these models are imported to pre-processing program of
ANSYSv14.5 for generation of mesh. Grid independency test is performed to avoid
any error in computation and to minimize the cost of computation.
No-slip boundary condition is executed for velocities of the solid arterial walls. The
solid walls of the coronary artery models are assumed to be rigid to simplify the
analysis as it does not have significant effect on the output results [5, 6]. At the
outlet, a constant 100 mm of Hg gauge pressure which is mean of normal systolic
(120 mm Hg) and diastolic (80 mm Hg) blood pressure is imposed. Physiologically,
flow of blood inside vessels (artery and veins) is pulsatile in nature. Therefore, at
the inlet, a uniform time-varying velocity profile is forced to incorporate pulsatile
behavior of blood flow in the simulation. The sinusoidal wave pattern of flow [6, 7]
is assumed as a substitute to physiological pulsation during systole. During diastolic
phase, there is cut-off of supply from the heart. Therefore, during this phase, the
blood flow is assumed to have a fixed velocity of 0.1 m/s. The imposed velocity
profile delivers an inflow velocity of 0.5 m/s during peak systole at t = 0.1 s with
a fast heartbeat of 120 per minute. The complete cycle is of 0.5 s in which systolic
phase is up to 0.218 s, and rest represents diastolic phase. Simulation of two cycles
is necessary to achieve a cycle independent solution in the present analysis. Hence,
to develop the diagnostic tool, results of the second periodic cycle are only used.
To analyze the flow conditions in the coronary artery, blood is considered as a
homogeneous and incompressible fluid with specific mass of 1060 kg/m3 as in [8].
The laminar flow has been assumed in the present work since the average Reynolds
number ranges from 150 to 800 [9, 10]. Blood is a shear thinning fluid, and hence, in
340 P. Jhunjhunwala et al.
this work, blood is modeled as a non-Newtonian fluid which follows Carreau model
[11],
μ − μ∞ n−1
= 1 + (λγ̇ )2 2 (1)
μ0 − μ∞
where μ, μ∞ and μ0 are the effective viscosity, dynamic viscosity at infinite shear
rate and dynamic viscosity at zero shear rate, respectively. Characteristic viscoelastic
time of the fluid, shear rate and power law index are represented by λ, γ̇ and n
respectively. For blood, n = 0.3568, λ = 3.313 s, μ0 = 0.056 kg/ms and μ∞ =
0.0035 kg/ms.
The simulation results are computed with time step size of 0.01 s for pressure distri-
butions in all the 16 artery models for the complete cardiac cycle. In spite of having
same % DS, large variations in rise in maximum pressure computed with CFD is
observed as the % AS of these models are different. In general, large drops in pressure
are observed after stenosis, and pressure has maximum value at the inlet.
The simulation results obtained through CFD are then further used to compute
rise in systolic and diastolic BP due to increase in stenosis severity. To compute
the rise in systolic BP, the maximum pressure obtained at peak systole (t = 0.6 s)
for healthy and stenosed models is considered. Then, from the values of pressure
obtained for stenosed model, the corresponding % pressure rise to healthy artery
case is obtained. Now, if 120 mm Hg (normal systolic BP) is considered as standard
baseline for healthy coronary artery, the rise in blood pressure because of increase in
stenosis severity can be obtained simply by adding % increase in pressure to baseline
pressure of 120 mm Hg.
In the similar fashion, to calculate the rise in diastolic BP, maximum pressure
obtained during end of cycle (t = 1 s) is used. For diastolic blood pressure, the stan-
dard baseline is 80 mm Hg. The obtained and calculated values from CFD analysis
for increasing severity of stenosis and corresponding % pressure rise for only healthy
and SD stenosed artery models are listed in Table 2 as an illustration.
The computed CFD results for % rise in blood pressure are then plotted using
standard curve fitting techniques of MATLAB to develop a mathematical equation
which relates % rise in blood pressure with % AS. Two different five-degree poly-
nomial equations, y = ax 5 + bx 4 + cx 3 + dx 2 + ex + f, appreciably map the data
for % pressure rise at peak systole and diastole with R-square values of 0.93 and
0.9, respectively. As R-square value equals to 1 represents no error, therefore, it can
be said that the selected mathematical model can be utilized with confidence for
the formulation of diagnostic tool. The constants of selected five-degree polynomial
equations for the calculation of rise in pressure (%) at the systolic and diastolic phases
are given in Table 3.
Development of a Diagnostive Tool for Prediction … 341
Table 2 Maximum pressure and corresponding % pressure rise values obtained for SD stenosed
artery models
Area stenosis Systolic Diastolic
(%) Maximum Pressure rise Maximum Pressure rise
pressure (mm (%) pressure (mm (%)
Hg) Hg)
0 102 0 100 0
43.75 103 0.95 100 0.08
75 116 12 101 0.81
89.44 187 45 105 5
93.75 345 70 114 12
97.11 1280 92 164 39
Using these equations and taking into account the errors, a code is developed in
the MS-Excel for the formulation of diagnostic tool. The diagnostic tool is developed
in the form of a “blockage calculator” as shown in Fig. 2.
To use this tool, a user has to just open the excel file in which the tool has
been developed and click the button “Click to Calculate.” The developed blockage
calculator will pop-out after then. As shown in Fig. 2, to get the prediction of probable
% AS as output through this tool, it requires four inputs, namely normal systolic BP,
normal diastolic BP, current systolic BP and current diastolic BP. In the developed
tool, the normal systolic and diastolic BP represents the blood pressure of a person
when he/she is healthy (e.g., 120/80 mm Hg), while current systolic and diastolic
BP represents the elevated blood pressure or the blood pressure at which a person
is showing discomfort (e.g., 160/85 mm Hg). It can be observed here that this tool
can predict the probable % AS just by measuring the blood pressure of a person, and
hence, it is patient specific, inexpensive, non-invasive, easy to use, handle and less
time-consuming. If normal blood pressure of a person is unavailable in case history,
then 120/80 mm Hg can be inserted as a default value.
342 P. Jhunjhunwala et al.
4 Conclusion
This study proposes using a diagnostive tool “blockage calculator” to estimate the
extent of stenosis severity. This quantified tool is based on the CFD study of blood
flow through three different stenosed models of same percentage of lumen diameter
stenosis with varying cross-sectional lumen area stenosis. This tool is easy to use
and can help clinicians in preliminary diagnosis for patients of high blood pressure.
References
1. Pijls NH, Fearon WF, Tonino PA, Siebert U, Ikeno F, Bornschein B, van’t Veer M, Klauss
V, Manoharan G, Engstrøm T (2010) Fractional flow reserve versus angiography for guiding
percutaneous coronary intervention in patients with multivessel coronary artery disease: 2-year
follow-up of the FAME (fractional flow reserve versus angiography for multivessel evaluation)
study. J Am Coll Cardiol 56(3):177–184
2. Tonino PA, De Bruyne B, Pijls NH, Siebert U, Ikeno F, vant Veer M, Klauss V, Manoharan
G, Engstrøm T, Oldroyd KG (2009) Fractional flow reserve versus angiography for guiding
percutaneous coronary intervention. N Engl J Med 360:213–224
3. Jhunjhunwala P, Padole P, Thombre S (2015) CFD analysis of pulsatile flow and non-Newtonian
behavior of blood in arteries. MCB Mol Cell Biomech 12(1):37–47
4. Pericevic I, Lally C, Toner D, Kelly DJ (2009) The influence of plaque composition on under-
lying arterial wall stress during stent expansion: the case for lesion-specific stents. Med Eng
Phys 31(4):428–433. https://doi.org/10.1016/j.medengphy.2008.11.005
Development of a Diagnostive Tool for Prediction … 343
Smit V. Motghare
1 Introduction
The earliest successful limb amputation ages back to 484 B.C., when Hegesistratus
had to cut his feet to escape from prison. He created a wooden foot for himself to
compensate the loss. In 1696, first notable lower-extremity transtibial prosthesis with
an unlocked knee, a thigh corset and external hinges was introduced by Verduyn, a
Dutch surgeon. Potts in 1816 introduced first transfemoral prosthesis having articu-
lating knee and ankle, the Anglesey leg. Transfemoral artificial limb with polycentric
knee, suction socket and multi-articulated foot was introduced by Parmelee in 1863.
S. V. Motghare (B)
Department of Mechanical Engineering, Government College of Engineering, Nagpur, India
e-mail: smit.motghare@gmail.com
measurement of developed shear stress and total deformation for stainless steel and
carbon fibre material were performed.
Fig. 1 Model of lower-limb prosthesis. a Socket and femoral structures. b Tibial structure and foot
348 S. V. Motghare
On the basis of analysis and results, it was observed that the structure made of stain-
less steel experiences maximum principle stress 0.8539 MPa at tibio-femoral joint,
Design, Modelling and Optimization of Artificial … 349
this keeps the weight in permissible limits to maintain stability as well as ease to lift
by the amputee limb. During gait, the foot acts as a shock absorber and supports the
structure by balancing it. The principle components may be covered with cosmetic
cover of flexible polymeric material to impart shape of a normal leg.
The purpose of prosthetic knee is to restore normal flexion–extension movement,
that is alignment controlled, i.e. knee axis is positioned perpendicular to the weight
line, heel strikes to mid stance in such way that an inherently stable knee extension
moment is provided. The free movement occurs only in one direction that too is
restricted by two springs on the posterior side. The energy required for bending the
knee would be very less as the springs does not need to be compressed but only
a transverse movement should occur. The client’s ability to voluntarily control the
knee during stance would depend on the hip musculature, length of residual limb
and stability of the knee unit (prosthesis). A flexible urethane foam cover similar to
the anatomical limb may be applied over the proposed design.
4 Conclusion
A lower limb was simulated and analysed using finite element analyser (ANSYS
R16.0). Total deformation, maximum principle stress and mass of the proposed model
were obtained for stainless steel and carbon fibre materials. On analysing results for
material optimization, carbon fibre was selected on the basis of percentage reduction
in weight of the principle components of artificial limb. Future work will involve
reproducing the FE model for experimental validation. Also, design a limb that will
show frictionless movement along rotational axes of the tibio-femoral joint to support
sitting (on floor) posture and gait.
References
1. Sanders GT, May BJ, Hurd R, Milani J (1986) Lower limb amputations: a guide to rehabilitation.
F. A. Davis, Philadelphia
2. Thomas CL (1993) Taber’s cyclopedic medical dictionary, 17th edn. F. A. Davis, Philadelphia
3. Barbara Silver-Thorn M, Standard handbook of biomedical engineering and design. Marquette
University, Milwaukee, Wisconsin
4. Shurr DG, Cook TM (1990) Prosthetics and orthotics. Appleton & Lange, East Norwalk
5. Wilson AB (1989) Limb prosthetics, 6th edn. Demos Publications, New York
6. May BJ (1996) Amputations and prosthetics: a case study approach. F. A. Davis, Philadelphia
7. Kaul V (2014) J Biomech Eng 136/011002-1. https://doi.org/10.1115/1.4025692
8. Seelen HAM, Anemaat S, Janssen HMH, Deckers JHM (2003) Effects of prosthesis alignment
on pressure distribution at the stump/socket interface in transtibial amputees during unsupported
stance and gait. Clin Rehab 17:787–796
9. Krouskop TA, Cosmetic covers for lower-limb prostheses. Bioengineering
10. Donahue TL, Hull ML, Rashid MM, Jacobs CR (2002) A finite element model of the human
knee joint for the study of tibio-femoral contact. ASME J Biomech Eng 124(3):273–280
Design, Modelling and Optimization of Artificial … 351
Abstract In the present study, an attempt has been made to slice tungsten carbide
rod with an abrasive water jet machining technique. The objective of this work is to
evaluate and characterize the surface roughness of the machined surface. Experiments
are carried with varying traverse speeds keeping other parameters constant. The
surface is examined in detail using a 3D Optical Profilometer to characterize the
roughness parameters which include centre line average deviation (Ra ), root mean
square deviation (Rq ), total height (Rt ), maximum peak height (Rp ) and maximum
valley depth (Rv ). It is observed that except Rv , all other parameters increase with an
increase in traverse speed.
1 Introduction
wear resistance and (hot) hardness. This is the reason why WC has gained popularity
as a cutting tool material for die and mould making. Hard to cut materials can be easily
machined by these carbides, and these carbides are difficult to machine by conven-
tional manufacturing methods. Since the thermal expansion and thermal conductivity
are low with a high melting point, brittleness is the major source of machining dif-
ficulties which generate high thermal stresses. The micro-cracks enlarge and lead
to macro-crack and fragmentation [2]. The available literature reveals that most of
the work is accomplished in machining different metallic, non-metallic and com-
posite materials only, but a limited literature was found on machining of carbides
by AWJM technique. Ahsan Ali Khan et al. carried machining of carbide grade P25
using AWJM with an objective to study the effect of jet pressure, abrasive flow rate
and traverse speed on the smoothness of the surface produced [3]. It was reported
that better surface is obtained at higher pressure, lower traverse speed and abrasive
flow rate and concluded that at the jet entrance, good surface is obtained than at the
bottom of the cut. This is because of the type of cutting wear depth and deformation
wear depth [4, 5].
The present work is an attempt to cut WC rod with a high-pressure AWJM which
has the advantages of minimum heat generation and no metallurgical changes in
the work material, and also, the surface is almost free from residual stresses. Here,
investigations are carried on the sliced surface of WC produced by AWJM with
different traverse speeds indicating the interaction time between the abrasive jet
and work material. Machined surface characterization is carried using a 3D Optical
Profilometer.
2 Experimentation
2.1 Material
Table 1 Mechanical
Property Value
properties of WC
Density 14.9 g/cm3
Hardness 1580 HV30/91.8 HRa
Young’s modulus 630 MPa
Thermal conductivity 80 W/mK
Coefficient of thermal expansion 5.0 × 10−6 K−1
Fracture toughness 6.9 N/mm2 m0.5
All the experiments were performed on a 3-axis Abrasive Water Jet Machining Cen-
ter (Model No. OMAX 2652) which has a cutting pressure of up to 40 KSI. The
maximum possible traverse speed is 4000 mm/min with an accuracy and repeatabil-
ity of ±0.025 mm, respectively. Once the material selection is done and considering
the machine parameters for experiments, fixtures are fabricated for holding the work
piece. While conducting the experiments, some of the machine parameters are kept
unchanged and are shown in Table 2. The experiments were performed at three dif-
ferent quality levels: 1, 2 and 3, respectively, which indicate that as traverse speed
decreases, the quality level increases. This makes the difference in quality with the
traverse speed of the jet. The traverse speed conditions for different quality levels
are given in Table 3. The machined sample is shown in Fig. 1.
2.3 Measurements
The sliced samples are measured for surface roughness using a 3D Optical Profilome-
ter. The 2D top surface, 3D view of the sample and the surface profile extracted from
the Optical Profilometer for the quality level 1 machined surface at one location are
shown in Fig. 2a–c, respectively. Table 3 gives the analytical results obtained after
measurements.
Each sample is measured at three different locations, and the average of these
values is considered for analysis, as given in Table 4. Table 5 gives the average
values of the machined samples at different traverse speeds (Table 6).
The graphs indicated in Fig. 3a–e show a similar increasing trend of the roughness
parameters with increasing traverse speed. The basic reason is that when the traverse
speed increases, the abrasive particles get lesser time to cut the material, and their
interaction with the material is less, leading to lesser cutting action. So it can be
An Experimental Study on Surface Roughness … 357
concluded that to achieve lesser surface roughness on the machined surface of WC,
the traverse speed should be as low as possible.
4 Conclusions
Acknowledgements The authors thank Head, 4i Laboratory, IIT Kanpur, for permitting us to
conduct the experiments and for the support provided in measuring the surface roughness using 3D
Optical Profilometer available in Manufacturing Science Laboratory, IIT Kanpur.
References
1. Bonny K, De Baets P, Vleugels J, Van der Biest O, Lauwers B, Liu W (2009) EDM machinability
and dry sliding friction of WC-Co cemented carbides. Int J Manuf Res 4(4):375–394
2. Mahamat ATZ, Rani AMA, Husain Patthi (2011) Machining of cemented tungsten carbide using
EDM. J Appl Sci 11(10):1784–1790
3. Khan AA, Bin Awang ME, Bin Annuar AA (2005) Surface roughness of carbides produced by
abrasive water jet machining. J Appl Sci 5(10):1757–1761
4. Hashish M (1991) Wear modes in abrasive water jet machining. Trans ASME PED 54:141
(Tribological Aspects in Manufacturing, 141, ASME)
5. Momber AW, Kovacevic R (1998) Principles of abrasive water jet machining, 1st edn. Springer,
Berlin
Energy Absorption Characteristics
of Single and Double-Walled Square
Tubes Subjected to Axial Crushing
Abstract Experimental and numerical study of single and double-wall square tube
under axial compression was carried out at a displacement rate of 100 mm/min. Two
configurations of double-wall tube, viz., parallel and diamond were explored. During
axial compression of double-wall tubes, crumpling takes place either at same end
or at opposite ends for the two tubes. Deformation at the same end absorbs 3–5%
more energy than opposite end crushing. For the same type of crumpling, diamond
arrangement absorbs 5–8% more energy than parallel configuration. Two different
lengths of unequal tubes for parallel and diamond arrangement were examined. It
was observed that unequal length tubes improve crashworthiness characteristics.
1 Introduction
Fig. 1 Cross section of specimen a single wall tube 1, b single wall tube 2, c parallel arrangement,
d diamond arrangement
double-wall tubes with different configurations. For better safety, the structure should
absorb as much as energy with minimum peak force possible. Hence, unequal length
tubes are examined.
2 Geometrical Details
In this study, commercially available extruded square aluminum columns are used.
Thin-walled square tubes of two different cross sections, one with 25.4 mm ×
25.4 mm and another with 50.8 mm × 50.8 mm (outer dimensions) having a length
of 150 mm and 1 mm thickness, are used for this study (Ref. Fig. 1a, b). Prior to
this, single wall tubes (SWT) of above two different cross sections are investigated.
Two tubes kept one inside other is called double-wall tube (DWT). In this study,
two arrangements of aluminum tubes which are parallel (P) and diamond (D) are
explored as shown in Fig. 1c, d. In the case of DWT, two tubes can have equal or
unequal lengths. DWT with an equal length of 150 mm with parallel and diamond
arrangement is examined. For DWT with unequal length (Un), outer tube length is
kept constant at 150 mm while two variations of inner tube length, viz., 147 and
153 mm are considered for parallel and diamond arrangement. The cross section,
length, arrangement and abbreviations used in this study for simplicity are shown in
Table 1.
Belytschko-Tsay shell element with five integration points is used. Mesh size of 2 mm
was used for square columns, as it gave reasonable convergence of the solution.
Piecewise linear plasticity material model for an elasto-plastic material was used
for numerical simulation of the specimens. In this material model, the true stress–true
plastic strain data obtained from tensile tests were used as input. The rigid block is
fixed at the bottom and velocity is given to the upper plate in the downward direction
for a specific time. For upper plate, all other degrees of freedom except z-translation
degree of freedom are constrained. Automatic nodes to surface contact algorithm
were used to model contact between the square column and upper plate, square
column and lower block. Automatic single surface contact algorithm is used for each
square column of the specimen to depict self-contact between shell elements. For
double-wall tube specimens, the automatic surface to surface contact algorithm is
used to capture the interaction between two tubes. The static and dynamic friction
coefficients used were 0.2 and 0.3.
To change the mode of buckling from extensional to symmetric mode (as observed
in experiments) an imperfection of the order of 0.1 mm was added to the tube using
the Perturbation node command. The crushing characteristic of the tube depends on
the way it is manufactured, viz., whether is formed out of metal sheet or extruded
[5]. Extruded square columns are used in this study. It was observed that corners are
thicker compared to side thickness. Measurements were taken and corner modeling
was done. For the rows of two elements at the corners, the thickness is given as 1.54
times thickness in side. Hourglass control was employed to control hourglass mode
deformation.
364 S. S. Toshniwal and R. V. Prakash
Fig. 2 a Arrangement for the compressive test, b true stress versus strain curve from tensile test
4 Experimental Investigation
To carry out a numerical simulation on the crushing of specimen, data about properties
of the material from which tube walls are made are required. For this purpose, tensile
specimen is cut from the side of the square tube, parallel to the direction of extrusion.
Tensile test specimens are prepared according to ASTM E-08 standard. Tensile tests
are done at a displacement rate of 5 mm/min. An axial extensometer is used for strain
measurement. Stress is calculated based on the load value measured with the help of
load cell. The engineering stress-strain curve is obtained from the test as shown in
Fig. 2a. The material has yield strength (σ y ) of 111 MPa, Young’s modulus (E) of
68.59 GPa, Poisson ratio (μ) of 0.3, density (ρal ) of 2700 kg/m3 .
as the area under force versus displacement curve. Crush force efficiency for each
specimen is determined by taking the ratio of mean force of crushing to Peak force of
crushing.
Numerical and experimental force versus displacement curves for crushing of SWT1
and SWT2 are shown in Fig. 3. During crushing, plastic hinge is formed first which
results in peak load. After hinge formation, bending of tube follows which completes
the plastic fold. A symmetric progressive collapse mode is noticed in the tubes.
In SWT1, formation of two full plastic folds and an incomplete fold is observed
during crushing. In SWT2, one complete fold and next plastic hinge is formed.
This corresponds to the second peak force in Fig. 3b. Experimental and numerical
deformation of SWT1 is shown in Fig. 4a, b. Deformation in SWT2 is shown in
Fig. 4c, d. The basic fold length in SWT2 is more than that of SWT1. Due to smaller
fold length, more folds and more plastic hinges are formed in SWT1; hence, it can
be said that SWT1 has more specific energy absorption capacity.
Figures 4 e-i show various types of deformations in double-wall tubes. In the
case of DWT, two types of interactions are observed between the tubes. In one type,
two tubes start their progressive collapse at two different ends such that there is no
contact between plastic folds of two tubes but a collapsed part of one tube will contact
non-collapsed part of another tube as shown in Fig. 4f, i. This is referred as opposite
end crushing. Another interaction is when two tubes collapse at same end. In same
end collapse case, plastic folds of two tubes contact each other and penetrate one
into the another as shown in Fig. 4g, h. It is termed as the same end crushing.
In case of DWT-P, same end of crushing specimen absorbs 5.5% more energy
compared to opposite end of crushing. In case of same end crushing, due to inter-
penetration between the folds, there is more interaction and subsequently, resistance
offered for crushing increases. Hence, small increase in energy absorption is noticed.
Fig. 3 Experimental and numerical force versus displacement for SWT1 and SWT2
366 S. S. Toshniwal and R. V. Prakash
Fig. 5 a Comparison between force versus displacement diagram of SWT1, SWT2, and DWT-D,
summation of force of SWT1 and SWT2, b Comparison between force versus displacement diagram
of SWT2, Un-147-D, and Un-153-P
368 S. S. Toshniwal and R. V. Prakash
with inner tube length 153 mm and outer tube length 150 mm is also studied. It will
have peak force that is between the values of Un-147 and DWT.
In this study for all unequal length tubes, the same end crushing mode of interac-
tion was observed only in Un-147-P. For all other configuration opposite end, crush-
ing is observed. As seen in DWT, in this case too, the same end crushing absorbs
6% more energy compared to opposite end crushing. In Un-153 case, the inner tube
collapsed first, while in Un-147 outer tube buckled first.
In case of Un-147-D specimen, for the same peak force as SWT2, energy
absorbed is 91.6% higher. When compared to DWT diamond configuration, peak
force decreased by 37.6% and energy absorbed by 12.4% as shown in Fig. 5b. Com-
parison between Un-153-P and DWT-P suggests a decrease in peak force by 18%
while energy absorption is decreased by only 3%. Un-153 tubes displayed increased
peak force and corresponding energy absorption All unequal length tubes have better
crush force efficiency (CFE) than DWT (Table 2).
The numerical model showed less energy absorption compared to experimental
results. Peak force values obtained by numerical method correspond well with exper-
imental results. The deformation pattern by numerical and experimental methods was
found to be similar. The numerical force versus displacement curves followed the
trend of experiments.
6 Conclusion
It is found that the energy absorption of the double-wall tube is greater than the sum
of energy absorption of two tubes it is made of. This is due to interaction effect
between tubes. The minimum increase was 12% for parallel and maximum increase
up to 25.8% for diamond. Two types of interactions are observed in double-wall
tubes. Same end crushing absorbed about 3–6% more energy than opposite end
crushing. Due to formation of plastic hinge for same type of crushing interaction,
diamond arrangement absorbs 5–8% more energy than parallel arrangement. In tubes
with unequal length, for a small loss in energy absorption, a large decrease in peak
force is observed. Unequal tubes have better crush force efficiency than DWT. When
compared to DWT, decrease in peak force is around 37% for Un-147 and 18%
for Un-153 tubes. Hence, with proper design unequal length tubes can optimize
crashworthiness characteristics.
References
1. Abramowicz W, Jones N (1984) Dynamic crushing of axial square tube. Int J Impact Eng
2(2):179–208
2. Prakash RV, Ghogare RS (2014) Energy absorption characteristics of rib reinforced thin-wall
extruded structure subjected to axial crush. Int J Veh Saf 7(3/4):345
Energy Absorption Characteristics of Single and Double … 369
1 Introduction
operators and operating conditions. Work study improves upon the existing process
or method and helps in standardization and simplification resulting in appropriate
manpower planning, production planning, etc. [1, 2]. Further, lean manufacturing
techniques are evolved for shop floor optimization. Lean manufacturing is a method-
ology that focuses on minimizing waste within manufacturing systems while simul-
taneously maximizing productivity [4, 5]. The benefits of lean include reduced lead
time, reduced operating costs, and improved product quality, to name just a few [6,
7]. Together, all these standard plant optimization techniques help to improve the
productivity and reduce the wastages of all kind.
In this work, the dispatch section of a medium-scale packaging industry is studied
so as to identify the bottleneck operations. This work is carried out at a printing and
packaging industry located at MIDC, Hingna region of Nagpur, India, with an annual
turnover of 5000 million Indian rupees. During the initial stage, layout of dispatch
section is studied, various packaging processes are observed, and process flowchart
is prepared. Time study is performed on packaging processes (film wrap stretching
process). During time study, bottleneck operation and non-valuable operation and
movements are identified. Finally, using line balancing technique and classifying
operations into stations, cycle time is evaluated. Section 2 presents layout of dispatch
section and process flowchart, Sect. 3 gives the time study, and Sect. 4 presents
important conclusions.
In this section, the layout of dispatch section of the packaging industry is discussed.
The section consists of stretching, packing, weighing, and storage as main operations.
Figure 1 gives the layout of dispatch section. It can be observed that machines are
arranged in a linear pattern, and there is ample free space on the floor. Dispatch is a
crucial section in that industry and hence is selected for time study.
Tables 1 and 2 give the sequence of operations carried out in packaging of a roll
and of a box. It is observed that there is a considerable delay introduced in weighing
process due to slip generation. This slip mentions all the parameters of the roll or
box and is computer generated.
A considerable difference between roll packaging and box packaging operations is
that in box packaging operation, additional supporting operations such as preparation
of boxes is required.
Time study is carried out by classifying operations into workstations and observing
operation as well as station time. Three workstations were classified as given in
Field Data Analysis Using Work Measurement … 373
time required at a particular workstation based on the above sample. Based on line
balancing concept, the cycle time for roll packaging operation t cr is fixed at 20.16 s
and cycle time for box packaging operation t cb is fixed at 41.29 s.
3.1 Analysis
This section focuses on the analysis of total production output if a certain percentage
of total time is used for box packaging and remaining time is used for rim packaging
operation, based on effective working time in a day of 14 h, i.e., 840 min:
Box packaging: Considering cycle time of 41.29 s and a pallet consisting of 35
boxes, total time required to prepare a pallet is 24.09 min. Considering additional
pallet handling time of 2.67 min as observed, total time required in preparing and
handling a pallet for box packaging is 26.76 min. Number of boxes that can be packed
in a working day, considering complete time is utilized for same is 1098.65 boxes.
Field Data Analysis Using Work Measurement … 375
Fig. 2 a Production output from box packaging and roll packaging against percentage of time,
b total production output against percentage of time (zero on x-axis signifies that hundred percent
time is dedicated to box packaging and hundred percent signifies that hundred percent of time is
dedicated to roll packaging)
Roll packaging: Operation cycle time for roll packaging is 20.16 s. Considering
same assumptions as above and if total time is dedicated to roll packaging, a total of
2034.18 rolls are packed per day.
Figure 2a gives total output for box packaging and roll packaging if a certain
percent (incremental from 0 to 100%) of working time is utilized for the operation.
It can be clearly observed that the output of roll packaging is always higher than box
packaging due to low operation cycle time.
Figure 2b gives the total production output if total time available is distributed
between box packaging and roll packaging. Here, zero percent on time scale signifies
that 0% of time is dedicated to roll packaging and 100% of time is utilized for box
packaging and vice versa. It is observed that with increase in total percentage of time
for roll packaging, total production output increases. This analysis is useful for floor
supervisor to fix that during a working day, what proportions of box and rolls can be
packed so as to maximize the productivity, simultaneously maintaining the demand
constraints.
4 Conclusions
This study provides a basis for identifying and analyzing the packaging operations
and fixing appropriate cycle time. Operations are classified into workstations and
operating cycle time is fixed using line balancing approach. It is clearly established
that box packaging requires higher cycle time than roll packaging. At the same time,
the study for distributing total available time for box packaging and roll packaging
is also conducted in the context of a particular packaging industry. This study helps
to establish the appropriate production blend so as to maximize the total output for
376 C. M. Salkar et al.
an industry involved in handling a product mix. However, this study can further be
repeated for higher sample size and different packaging industries.
References
1 Introduction
with the application of ultrasonic vibrations given to the Si (100) workpiece [9]. The
comparison of surface roughness presented in the present work will highlight the
process capabilities of DDMAF on Si (100) integrated with chemical assistance and
ultrasonic approach.
The DDMAF technique was integrated with chemical processes like chemical etching
of Si (100) wafer and the preferential application of chemical oxidizers in the polish-
ing region to enhance the polishing characteristics of Si (100) wafer. The experimental
setup of DDMAF and its combination with ultrasonic vibration is shown in Fig. 1a,
b, respectively. The DDMAF experimental setup is comprised of primary magnetic
disk and secondary magnetic disk, having the dimensions ∅ 25 mm × ∅ 3 mm. In
each magnetic disk, four blind holes existed to form a magnetic pole. The silicon
workpiece with diameter 100 mm, thickness 525 ± 25 µm, cut and lapped, p-type
boron and resistivity of 0.001–1000 -cm was inserted in between the rotating mag-
netic disks aligned in the vertical direction, placed on the tray, meant for executing
the polishing process. The allied processes which are used with the DDMAF are as
follows:
Fig. 1 a DDMAF experimental setup [7, 8]. b DDMAF experimental setup integrated with
ultrasonic vibration unit [9]
Comparative Study of Nanofinishing of Si (100) … 379
The use of chemical oxidizers with DDMAF has further enhanced the surface rough-
ness value of Si (100) wafer to 6.9 nm on atomic force microscopy scale at experimen-
tal process parameters namely finishing speed of 175 rpm, working-gap of 5.40 mm,
and flow rate of slurry in the polishing region being 10 ml/min. Throughout the exper-
imental procedure, the pH of the slurry was kept at 12.29. The chemical oxidizers
in proportionate amount, namely sodium hydroxide (15% w/w), Marshall’s acid salt
(1% w/w), Caro’s acid salt (1% w/w), and hydrogen peroxide (0.05% w./w.) are used
in slurry of alumina (20% w/w) in 100 ml distilled water. The addition of Marshall’s
acid salt in the alumina slurry at pH 12.29 provided elevated oxidizing power, and the
addition of Caro’s acid salt generated a hydrophobic over the surface of abrasive that
leads to the improvement of surface quality of Si (100); also, the mixing of hydrogen
peroxide with slurry created a passivation layer on the wafer surface to protect itself
from the attack of hydroxide ions. Thus, the preferential use of chemical oxidizers
with DDMAF improves the surface characteristics of polished Si (100) [8].
380 K. Pandey et al.
The surface characteristics of finished Si (100) were carried out on atomic force
microscopy scale. The height of the peaks in raw Si (100) originally existed at
5.43 µm. The AFM image of raw Si (100) is presented in Fig. 2a. The peak height in
raw Si (100) has been considerably reduced to 266, 96.8, and 59.9 nm by chemical
etching process, implementation of chemical oxidizers, and integrated approach of
Fig. 2 AFM images of a raw Si (100) [7–9], b polished Si (100) in chemically etched in 60% KOH
at optimum parameters by DDMAF [7]
Comparative Study of Nanofinishing of Si (100) … 381
Fig. 3 AFM images of a polished Si (100) using chemical oxidizers with DDMAF at optimum
parameters [8], b polished Si (100) using chemo-ultrasonic approach with DDMAF at optimum
parameters [9]
382 K. Pandey et al.
Fig. 4 a Raw silicon wafer, b chemically etched and finished silicon wafer at optimal parameters,
c finished silicon wafer using chemical oxidizers with magnetic abrasive particles at optimum pro-
cess parameters, d polished silicon wafer using ultrasonic vibrations along with chemical oxidizers
supplied to magnetic abrasive particles at optimum process parameter
4 Conclusions
i. Chemically etching the Si (100) wafer in KOH solution with isopropyl alcohol
and finishing by DDMAF process at optimal experimental levels enabled to
obtain the surface roughness of 18.20 nm on atomic force microscopy scale.
ii. The improvement in polishing quality of Si (100) was noticed with the use of
oxidizers, which were permitted to flow in the finishing region at the rate of flow
of 10 ml/min. The surface roughness obtained in this case was 6.9 nm.
iii. The process efficacy was further improved with the combination of chemical
oxidizers and ultrasonic vibration provided to the Si (100) work material. The
optimum experimental parameters in this case registered a surface roughness of
1.5 nm.
Comparative Study of Nanofinishing of Si (100) … 383
Fig. 5 Comparison of surface roughness on atomic force microscopy scale of polished Si (100)
using different processes integrated with DDMAF
References
1. Sreejith PS, Udupa G, Noor YBM, Ngoi BKA (2001) Recent advances in machining of silicon
wafers for semiconductor applications. Int J Adv Manuf Technol 17(3):157–162. https://doi.
org/10.1007/s001700170185
2. Jain VK, Kumar P, Behera PK, Jayswal SC (2001) Effect of working gap and circumferen-
tial speed on the performance of magnetic abrasive finishing process. Wear 250–251:384–390.
https://doi.org/10.1016/S0043-1648(01)00642-1
3. Kala P, Pandey PM (2015) Comparison of finishing characteristics of two paramagnetic materials
using double disc magnetic abrasive finishing. J Manuf Process 17:63–70. https://doi.org/10.
1016/j.jmapro.2014.07.007
4. Sihag N, Kala P, Pandey PM (2015) Chemo assisted magnetic abrasive finishing: experimental
investigations. Procedia CIRP 26:539–543. https://doi.org/10.1016/j.procir.2014.07.067
5. Yamaguchi H, Shinmura T (1999) Study of the surface modification resulting from an
internal magnetic abrasive finishing process. Wear 225–229:246–255. https://doi.org/10.1016/
S00431648(99)00013-7
6. Mulik RS, Pandey PM (2011) Magnetic abrasive finishing of hardened AISI 52100 steel. Int J
Adv Manuf Technol 55:501–515. https://doi.org/10.1007/s00170-010-3102-8
7. Pandey K, Pandey U, Pandey PM (2018). Statistical modeling and surface texture study of
polished silicon wafer Si (100) using chemically assisted double disk magnetic abrasive finishing.
Silicon 11(3):1461–1479. https://doi.org/10.1007/s12633-018-9961-6
384 K. Pandey et al.
8. Pandey K, Pandey PM (2018) Use of chemical oxidizers with alumina slurry in double disk
magnetic abrasive finishing for improving surface finish of Si (100). J Manuf Process 32:138–
150. https://doi.org/10.1016/j.jmapro.2018.02.007
9. Pandey, K., & Pandey, P. M. (2019) An integrated application of chemo-ultrasonic approach for
improving surface finish of Si (100) using double disk magnetic abrasive finishing. Int J Adv
Manuf Technol 103:3871–3886. https://doi.org/10.1007/s00170-01903829-5
Optimization of Thickness of Hollow
Punch–Die for Proposed Solar-Assisted
Leaf Plate and Cup Making Machine
Abstract Leaf plates are traditionally made by hand in Indian villages. These are
commonly used for serving food at family, religious and social functions. The labo-
rious craft can now be converted into machine operation to make these containers
in elegant shape and sizes and to make the plates much stronger than the normal
one by the conventional method of making plates. The punch–die assembly of the
traditional machine consists of an electric heater where the effective heating temper-
ature is maintained from 110 to 120 °C. Alternatively, this temperature can also be
achieved easily by using solar concentrator to heat the thermic oil which in turn can
be used to heat the punch–die assembly. The hot oil will be circulated through hollow
space of punch and die so that the green leaf will be heated from both the sides and
occupied the required shape of cup or plate in minimum time while moulding in die.
The thickness of such punch–die assembly is optimized analytically and also by FE
software ANSYSTM .
1 Introduction
of days (approx. 300–330 days). So, the solar concentrator can be used to heat the
thermic oil. Thermic oil is a suitable fluid for passive heat storage and for supply of
heat to the work station at punch–die assembly. Thus, the use of electric heater can be
eliminated from traditional machine, and the cost of electrical energy consumption
will be reduced to zero. So the above machine can be regarded as a green machine
and a boon for rural people to earn their livelihood. The first important aspect is to
optimize the design of hollow punch–die assembly. The hot oil will be circulated
through hollow space of punch and die so that the green leaf will be heated from
both the sides and occupied the required shape of cup or plate in less time while
moulding in the die.
2 Literature Review
Leaf plate and cup manufacturing is a well-known process in India. Use of banana
leaves is a common practice in India to serve breakfast and lunch/dinner during
family, social and religious functions. Hence, manufacturing of cups and plates of
different size and shapes has emerged in many small-scale industries. The conven-
tional type of machines is high-grade electrical energy based, i.e. electric energy
(heaters) is used to heat the punch of die–punch assembly.
All the machines available or manufactured in India are electric energy based.
Some of them are bench model, stand model, multi-set type model-mass production
type, foot press type and hand lever press type.
In all of the above machines, either punch is lowered or die is moved upwards. In
the majority of machines, punch is heated by means of electric heater. The punch–
die assembly used in conventional machines are solid, rugged type, and hence, the
design is easy and simple. Literature survey [1, 2] indicates that no researcher has
developed solar-assisted type (thermic oil based) leaf plate/cup making machine, and
no research paper is cited in the area of solar energy utilization for leaf plate/cup
production.
3 Present Design
In the present design, both punch and die are heated simultaneously by means of
thermic oil heated at solar concentrator. A preliminary leaf cup/plate making machine
[2] was developed on conventional (electric energy). In this machine, the die is heated
and punch is pressed. The researchers had got an idea of manufacturing the leaf
plate along with optimum temperatures needed at punch and die. The temperature of
around 120 °C for the duration of one minute is sufficient to produce the leaf plate/cup.
It is found that the punch should also be provided with heating arrangement so that
both sides of plate are uniformly heated. The thermal energy needed for producing
the leaf plate would be easily collected by solar insolation using effective solar
Optimization of Thickness of Hollow Punch … 387
concentrator where thermic oil is heated. The heated thermic oil is circulated in
hollow punch and die to produce the leaf plate. The solar-assisted leaf plate making
machine (SALPMM) consists of punch, die, thermic oil tank, solar concentrator and
valves.
The main operating parts in the machine SALPMM is hollow punch–die system.
The thickness of punch–die system should be proper to get the requisite shape of
leaf plate, and it should also be able to withstand the various forces acting during
the cup/plate manufacturing operation. Line diagram of punch and die is shown in
Fig. 1.
In the present research, the optimization of thickness of punch and die is carried
out by analytically and by finite element analysis to get a true picture of the effect of
forces acting thereupon.
60
Finite element model is shown in Fig. 3. The pressure load calculations for conical
and bottom surfaces of the punch are as follows:
Fig. 3 FE model
Pc Pa Pb
Pressure Pa
Optimization of Thickness of Hollow Punch … 389
References
1. Kalita P, Dixit US, Mahanta P, Saha UK (2008) A novel energy efficient machine for plate
manufacturing from Areca Palm Leaf Sheath. J Sci Ind Res 67:807–811
2. Potdukhe PA, Nilawar AS (2011) Design, fabrication & performance studies on leaf plate making
machine. In: National conference on advances in mechanical engineering, Nov 2011
Development of Briquette Cum Pellet
Making Machine
Abstract Fossil fuels are on the edge of extinction as a very less quantity of them is
available now. They also have large ash content, and pollution takes place to a greater
extent during combustion. So, there is a need to replace these fuels with something
that has the same calorific value and causes less pollution. Biomass briquettes and
pellets can be an alternative to coal. These briquettes and pellets are formed by
compression of biomass that can be used as fuel. This reduces pollution and pro-
vides alternate low-cost fuel. Various machines have been developed for producing
briquettes and pellets, but the cost of these machines is very high. After analyzing
various machines, a machine that can make both briquettes and pellets through a
single setup was not available. So, decision was made to produce a low-cost machine
that can produce both briquettes as well as pellets through a single manually driven
setup. The extrusion process has been used for producing briquettes and pellets,
wherein the forward extrusion process was used for producing the briquettes and the
backward extrusion process was used for producing pellets. Analysis, fabrication,
and experimentation were carried out on fabricated briquette cum pellet machine.
The objective of the experimentation was to find out the density of the briquette and
pellet obtained using sawdust and aspen chips as raw material and a binder (bentonite
or starch) combination. This machine can reduce farm waste into a useful product
by also providing a source of livelihood.
1 Introduction
In today’s world, humans are mostly reliant on fossil fuels like coal as a primary
energy source. These sources have large ash content, and pollution takes place to
a greater extent during combustion. So, there is a need to replace these fuels with
something that has the same calorific value and causes less pollution. As India is an
agricultural country, it has a huge prospective in converting biomass into a renew-
able form of energy. But most of these loose biomass is directly burnt, which causes
extensive air pollution. This can be reduced by using briquetting of biomass. Biomass
briquette or pellets are obtained by pressing loose biomass into definite shape and
size. The biomass densification leads to increased net calorific value per unit volume,
reduced transportation cost, improved bulk density, and moreover handling charac-
teristics [1]. Briquettes generally have a diameter range of 40–90 mm whereas pellets
have a diameter range of 5–10 mm [2]. Various types of raw materials like sugarcane
residue, sawdust, groundnut shells, maize and cotton stalks, rice husks, paper waste,
wood chips, forest residues, etc., can be used for the briquetting process [3, 4]. The
ash content, calorific value, moisture content, size of particle, and easy availability [5]
are the key factors for the selection of raw materials to get quality briquettes/pellets.
Starch, clay, molasses, bentonite, etc., can be used as binders [6]. Materials required
for making briquettes and pellets are already available in abundance in rural areas,
and this was one of the best solutions to the problem. Briquettes and pellets are not
only low-cost fuel, but it can be a source of income, if the briquettes and pellets are
sold in the market, thus creating wealth from waste.
The machine is designed to produce briquettes and pellets using a single setup man-
ually by the process of extrusion, wherein the forward extrusion process will be used
for getting the briquettes and backward extrusion process will be used for getting pel-
lets. Various parts of the machine include Scotch yoke mechanism, piston, cylinder,
and hopper. While designing machine, various mechanisms (like slider crank and
toggle mechanism) were considered. The Scotch yoke mechanism best suits require-
ment for making machine, as this mechanism is easy to operate and linear motion of
connecting rod and parts is easy to replace [7]. Due to the advantages of Scotch yoke
mechanism, it was finally selected. Various parts like Scotch yoke, piston, cylin-
der, and hopper were designed according to the requirement [8–10]. Piston design
includes six tapered holes for the production of pellets through reverse extrusion.
After the finalization of design, fabrication work for the machine was done. Various
parts were fabricated and assembled to produce the machine. The CREO model of
the machine dimensions is shown in Fig. 1.
Development of Briquette Cum Pellet … 393
The working of this briquette cum pellet making machine is simple due to the
Scotch yoke mechanism. The rotary motion of the Scotch is obtained with the move-
ment of the handle. This rotary motion from Scotch is transmitted to yoke with the
help of the movement of a pin attached to it. Reciprocating motion from yoke is
transmitted to the piston through the connecting rod. When the raw material is added
to the cylinder through the hopper, this reciprocating motion compresses the mate-
rial. As piston consists of the drilled holes, it allows the formation of pellets through
the reverse extrusion process. Briquettes are formed through the forward extrusion
process.
3 Experimentation
The photographs of different parts of the machine are shown in Fig. 2, and the devel-
oped machine is shown in Fig. 3. Experimentation was done to find out the density
of the briquette and pellet obtained by this machine, by using various raw material
and binder combination [11, 12]. Sawdust was used as raw material as sawdust is
easily available, low-cost, and finely ground material. Another raw material used
was aspen, which is also easily available. Binders used for experimentation purpose
are bentonite and starch along with a certain mixture of water. Clay was used just
for the experimentation purpose. In rural areas, rice husk along with cow dung as the
binder can be used to produce briquettes and pellets (Fig. 4).
During compression stroke, the material inside cylinder is compressed to 2:1
compression ratio. The excess material inside the cylinder after compression comes
394 Y. M. Sonkhaskar et al.
through the holes drilled on the piston. Holes on the piston are tapered so that the
material comes out and is also compressed producing compressed pellets.
Table 1 Density of
Raw material Binder Density (kg/m3 )
Briquettes
Sawdust Bentonite 956
Aspen chips Bentonite 798
Sawdust Starch 936
396 Y. M. Sonkhaskar et al.
5 Conclusion
Analysis of raw materials, the tentative design of components and mechanisms was
done. The extrusion process has been used for producing briquettes and pellets,
wherein the forward extrusion process was used for producing the briquettes and
the backward extrusion process was used for producing pellets. Experimentation
includes finding out the density of the briquette and pellet obtained by this machine,
by using various raw material and binder combinations. Briquettes and pellets were
obtained of diameter 70 mm and 8 mm, respectively. The findings of experimentation
state that sawdust along with bentonite as a binder has more density for briquette as
well as pellet, among the various raw material and binder combinations used. This
machine can help reduce farm waste into a useful product by also providing a source
of livelihood.
References
1. Sengar SH, Mohod AG, Khandetod YP, Patil SS, Chendake AD (2012) Performance of Bri-
quetting machine for briquette fuel. Int J Energy Eng 2(1):28–34. https://doi.org/10.5923/j.ijee.
20120201.05
2. Sonkhaskar YM, Deshpande VS, Modak JP (2013) Design and fabrication of dual powered
mini pelletizer. Appl Mech Mater 465–466:242–247
3. Saptoadi H (2008) The best bio-briquette dimension and its particle size. AJEE 9(3 and 4):161–
175
4. Shukla S, Vyas S (2015) Study of biomass briquettes, factors affecting its performance and
technologies based on briquettes. IOSR J Environ Sci Toxicol Food Technol (IOSR-JESTFT)
9(11):37–44
5. Oladeji JT (2015) Theoretical aspects of biomass briquetting. J Energy Technol Policy 5(3)
6. Križan P, Matúš M, Bábics J, Šooš L, Beniak J (2019) Relationship between raw material
composition and pellets physical properties. In: TSME-international conference on mechanical
engineering. https://doi.org/10.1088/1757-899X/501/1/012004
7. Shyamalee D, Amarasinghe ADUS, Senanayaka NS (2015) Evaluation of different binding
materials in forming biomass briquettes with sawdust. Int J Sci Res 5(3)
8. Mani S, Sokhansanj S, Bi X, Turhollow A (2006) Economics of producing fuel pellets from
biomass. Am Soc Agric Biol Eng 22(3):421–426, ISSN: 0883-8542
9. Bhattacharya SC, Augustus Leon M, Mizanur Rahman M (2000) A study on improved biomass
briquetting. In: International conference on biomass-based fuels and cooking systems, Pune,
India (2000)
10. Sonkhaskar YM, Saluja HS, Srivastava R, Parikh R, Singhai U (2018) Low cost manual bri-
quette making machine using Scotch Yoke mechanism. Int J Innov Res Sci Eng Technol
7(7)
Development of Briquette Cum Pellet … 397
11. Mushiri T, Mugodo P, Mbohwa C (2017) Design of a sawdust pelleting machine. In: Interna-
tional conference on industrial engineering and operations management, Rabat, Morocco, pp
1763–1776 (2017)
12. Sonkhaskar YM, Deshpande VS, Modak JP (2015) Mathematical modeling of pellet making
process using human powered flywheel motor. Ind Eng J 8(11):6–10. ISSN: 0970-2555
Towards the Development of Low-Cost
Vacuum Setup for Customized Implant
Manufacturing
Abstract This study introduces a new vacuum casting setup used to cast customized
implants. Setup is equipped with the induction furnace, vacuum pump, top vacuum
chamber, thermocouple probe and bottom mold chamber along with Pirani gauge
and vacuumized sealed doors and valves mounted. Experiments were conducted
with aluminum alloy to investigate the accuracy of the setup. Implant fabrication is
carried out with the aid of machining process which is difficult to machine, if the
shape is intricate. Casting is one of the best routes to cast the complex geometry with
the application of rapid prototyping technology. The proposed setup in this paper
for the casting of aluminum under vacuum was designed and developed in-house,
equipment opted and setup design after thermodynamic calculation for a vacuum
chamber and minimizes required vacuum level for melting of Al. Melting is by
induction furnace, metal is directly poured with the help of bottom pouring crucible
mechanism mold prepared by zircon sand, and the pattern used for mold making
is made on rapid prototype machine having acrylonitrile butadiene styrene (ABS)
material. A promising new technique for direct melting and pouring of titanium alloy
for medical implants was emerged from this study. This research demonstrated the
feasibility of direct melting and casting of the aluminum alloy by vacuum technology.
S. Randiwe (B)
Shri Ramdeobaba College of Engineering and Management, Nagpur 440013, India
e-mail: randiwesm@rknec.edu
D. Bhiogade
MSPGCL Co., Ltd., Nagpur, India
A. M. Kuthe
CAD-CAM Centre, VNIT, Nagpur, India
1 Introduction
Today, Ti6 Al4 V titanium is the most widely used material in the biomedical field
due to its high mechanical properties, excellent corrosion resistance and excellent
biocompatibility. However, titanium is characterized by high thermal reactivity, low
thermal conductivity and reduced machinability [1]. It is the most suitable alloy used
in biomedical implants in everyday life because of its lightweight as compared to
that of stainless steel SS316L (CF3M).
The cost of titanium and its alloy is high compared with existing materials such
as aluminum alloys and steels. The cost includes the raw materials as well as the pro-
cessing cost that requires a controlled environment due to the reactivity of titanium.
Currently, research is being aimed to reduce the cost of primary titanium [2].
Prior to actual experimental setup, the various studies were carried out to inves-
tigate the properties of zircon sand mold which is used to pour the molten metal so
that the process becomes very cost effective. The zircon sand is selected as a primary
molding material because it is considered to be common foundry sand of the most
refractory [3]. Zircon has a fusion point of 2700 K with good flowability, relatively
high conductivity and better dimensional stability than graphite. In addition, zircon
sand molds have been used for many years in ferrous and nonferrous foundries with
good results for heavy walled premium castings.
Malone et al. suggested that molten titanium reacts violently with zircon, but later
studies by Lang et al. [4] investigated that zircon rammed molds using sodium silicate
as binder withstood the attack of molten titanium. But casting of thickness of 12 mm
or more produced in the above molds suffered from pinholes and poor details.
Koch et al. [5] benefited using sodium silicate in lower quantities and reported
very good results in castings up to 100 mm cube size. He also tried a number of mold
washes and found that zirconia wash with water or isopropyl alcohol base is very
effective for inhibiting metal-mold reactions. Bentonite has been tried as a binder
since it produces excellent grain and good dry strength in rammed zircon molds. This
combination indicated good results in castings of 150 mm cube size.
In the recent study at DMRL, zircon sand was used with CO2 molding method
for casting titanium. Here, sodium silicate is used as binder and hardened by CO2
gas which releases colloidal silica as given in the simplified reaction.
Among various melting methods, vacuum induction melting (VIM) and vacuum
arc remelting (VAR) are widely used for commercial production of NiTi alloys. In
VIM, the electrodynamic forces result in the first stirring of the melt and thereby
ensuring greater chemical and microstructural homogeneity in the alloy. The main
disadvantage of the VIM is carbon contamination from the crucible. VAR uses the
consumable electrodes to melt the metal and is carried out under water-cooled copper
crucible.
Many researchers have focused on machining titanium alloy and achieved the
optimal results. The present experimental study focused on the melting of titanium
alloy in a vacuum environment with low-cost vacuum casting setup.
Towards the Development of Low-Cost Vacuum … 401
Much effort has been expended in the modification of induction melting furnace
for use with titanium. This system uses many advantages, including single-stage
melting, uniformity of temperature and composition within the melt and the ability
to melt scrap, ingot or sponge. Many researchers have focused on the melting of the
titanium alloy in vacuum environment only, and there are many research gaps in the
titanium alloy melting. The main aim of the research is to develop a cost-effective
and miniature vacuum setup to melt titanium alloy to make the intricate shape such
as customized prosthesis for patients in India.
2 Experimental Method
Ti6 Al4 V alloy is most commonly used in aircraft and biomedical industries. It covers
50% of the whole production of titanium alloys. It is also the most explored and
tested titanium alloy with balanced properties such as low density, ductility, good
corrosion and oxidation resistance. Standard composition of the ASTM F136 Grade
5 titanium is given in Table 1, and the chemical analysis of the titanium alloy (ASTM
F136) along with composition is given in Table 2. Figure 1 shows SEM image of
microstructure of work material Ti6 Al4 V.
Fig. 1 Microstructure of Ti6 Al4 V (light area represents α phase rich in aluminum, and the dark
area represents β phase rich in vanadium)
2.2 Experimentation
The alloy to undergo vacuum remelting is formed into cylinder typically by vacuum
induction melting. This cylinder, refined to as an electrode, is then put into large
cylindrical enclosed crucible and brought to a metallurgical vacuum (0.001 mm Hg
or 0.1–13.3 Pa). At the bottom of the crucible, there is a small amount of alloy to
be remelted which the top electrode is brought close prior to starting the melt. The
modification done in our system was to pull the molten metal from the bottom of the
crucible to fill the mold cavity which melted under induction heating.
Towards the Development of Low-Cost Vacuum … 403
The selection of the refractory and insulating bricks was made based on the ser-
vice requirements of the furnace vacuum chamber. These service requirements were
identified as follows:
1. Temperature: A maximum furnace temperature of 2000 °C (3632 °F) was put
into consideration. Since titanium melts at 1668 °C (3034 °F), this temperature
would allow for the superheating of the charge to be more easily accomplished
and also makes the furnace more durable.
2. Size: The furnace volume should be enough to melt a maximum 5 kg of titanium
necessary for the casting of small titanium components. The refractory used to
be the high alumina R/M or LCC, Al2 O3 % = 90–92, Fe2 O3 % = 0.5 (max), ST
= 1100 °C, AT = 1750 °C, Grading = 0–6 mm.
3. Top chamber with door assembly: The material used for the construction of the
vacuum chamber was stainless steel. All the mechanical design aspect was taken
into account for the development of the top assembly to avoid failure of the
chamber under vacuum. The top chamber is mainly used for the charging of
the material and stirring of the metal, and eyepiece was provided to observe the
condition of the molten metal. The chamber was lined entirely with the refractory
to avoid the heat loss.
4. Heating chamber: The central part of the chamber consists of the bottom pour-
ing crucible surrounded by the induction coil. Two electrical connections were
directly attached to the middle section where the induction coil was wounded to
the crucible.
5. Bottom part of the chamber: Bottom portion of the chamber was designed to
serve the purpose of casting. The molds were placed below the bottom pouring
crucible for directly pouring the molten metal.
6. Bottom pouring crucible: The specially designed graphite crucibles with zirco-
nium coating were used for the pouring of the molten Ti. This technique helps
to avoid complicated tilt-pouring system.
404 S. Randiwe et al.
The material was charged into a crucible at room temperature; the furnace was then
closed, oxygen was evacuated in 10−3 Torr. As titanium is very reactive material,
it is essential to degas the chamber. Melting took place in a vacuum. The melting
temperature reading was noted using immersion-type thermocouple. For Ti6 Al4 V, the
casting was poured at temperature set in between 1665 and 1680 °C. The rectangular
Towards the Development of Low-Cost Vacuum … 405
bar was cast as shown in Fig. 3 having 100 mm length, 20 mm depth and 20 mm
width. After shake out, it observed good surface finish with no significant defect.
3.2 Simulation
The casting simulation of rectangular bar 100 × 20 × 20 mm3 was simulated in Auto-
CAST software, properties assigned to cast Ti were fed into the database, which was
obtained from thermophysical properties calculations, which are density 4430 kg/m3 ,
liquid density 4230 kg/m3 , thermal conductivity 6.70 W/m K, specific heat 526 J/kg K
and latent heat 91.8 kJ/kg [7]. Also, for zircon powder mold, properties are derived
from the experimental results and software database, which are density 1770 kg/m3 ,
thermal conductivity 0.98 W/m K, specific heat 735 J/kg K and compressive strength
20 kPa. Moreover, a constant value of 1000 W/m2 K was used for the interface heat
transfer coefficient. Total filling time is 2.21 s. The average pouring rate was 0.98 kg/s.
4 Conclusion
This article discusses about the aluminum casting for which the low-cost miniature
setup is developed with a focus on mechanical properties of aluminum alloy. The
experimental trials were carried out to as-cast implant (hip joint) with the mechanical
properties of tensile strength 350 Mpa, yield strength 270 Mpa and Young’s modulus
of 110 Mpa. Figure 4 shows the hip joint.
The vacuum casting equipment constructed was able to replicate the advantages
of inclusion-free casting and improved in castings properties produced via vacuum
406 S. Randiwe et al.
casting process. Earlier, the vacuum casting equipment was constructed with costing
of US$18,000.
The model can be easily scaled up for large-sized casting by increasing furnace
capacity and the size of the molding chamber.
References
1. Ezugwu EO, Wang ZM (1997) Titanium alloy and their machinability—a review. J Mater Process
68(3):262–274
2. Suzuki Y (1998) Fabrication of shape memory alloys. In: Otsuka K, Wayman CM (eds) Shape
memory materials. Cambridge University Press, Cambridge, pp 133–148
3. Randiwe SM, Bhiogade DS, Kuthe AM (2017) Investigation & critical analysis of titanium
casting in zircon sand mold for customized implant manufacturing. In: Proceedings of 5th
International Conference on Powder metallurgy and Advanced Materials (RoPM-AM-2017),
Technical University, Cluj-Napoca, Romania
4. Lang RM et al (1954) Development of processes for making Ti casting and other ingots. No.
DRDTB-I-12046, Battelle Memorial Institute, Columbus
5. Koch RK, Hoffman JL et al (1977) Casting titanium in zircon sand moulds. IS Bureau of Mines,
Report RI-8203, p 44
6. Aremo B, Adeoye MO (2010) A low-cost vacuum equipment for aluminium alloys. Russ J
Non-ferrous Metals 51(2)
7. Bhiogade DS, Randiwe SM, Kuthe AM (2017) Study of Hot tearing in stainless steel CF3M
during casting using simulation and experimental method. Int J Metal Cast. https://doi.org/10.
1007/s40962-017-0170-7
Simulation Study on Effect of Variable
Curvature on the Modal Properties
of Curved Cantilever Beams
Abstract Curved beams are widely used in many engineering fields due to the high-
strength capacity compared to their straight forms. It is essential that curved beams
be analyzed for their modal properties as structures should have the natural frequency
away from that which occurs in their working condition to avoid resonance. In the last
few decades, many researchers had carried out investigation on several parameters
of the curved beam. The present research work proposes a generalized mathematical
equation for predicting mode shapes of curved cantilever beams, which take into
account the change in length and curvature of the beam. Modeling of curved beam
subjected to cantilever boundary condition is done by using ANSYS software. Twelve
cases of cantilever beams, each having different curvature and length, were created.
The mode shapes of the curved beams were obtained using the normalization method
for each mode shape. The effect of the variation of the curvature on the mode shapes
and natural frequency was analyzed for the first six transverse modes. It is found
that the frequency of vibration and amplitude of vibration increase as the radius of
curvature of the beam increases. Curve fitting equations for these mode shapes were
obtained using MATLAB software. The generalized equation obtained for the mode
shapes of the curved cantilever beams generates the mode shapes which are in good
agreement with those obtained from ANSYS.
1 Introduction
Some of the structures such as arches, arch bridges and long-span roof structure use
curved beam elements due to their enhanced strength. The modal analysis of these
curved beams plays an important role in analyzing the modal properties which can
be used to understand the vibration behavior of the structure. Various researchers
have studied the modal properties of curved beams.
Lee et al. [1] developed the governing equation for the variable curvature arch.
Numerical method is used to calculate the natural frequency and mode shapes of
the parabolic, sinusoidal, and elliptic arches. Guierrez et al. [2] obtained the low-
est frequency coefficients of symmetrical and unsymmetrical arch-type structures,
using geometric parameters coordinate functions and the Ritz method. Kang et al.
[3] developed the governing equations of motion for in-plane and out-of-plane vibra-
tion of circular arches considering shear deformation using differential quadrature
method. Oh et al. [4] derived the governing differential equations for free in-plane
vibrations of non-circular arches, including the effects of rotatory inertia and shear
deformation. Raveendranath et al. [5] investigated the performance of a curved beam
finite element with coupled polynomial distributions for normal displacement and
tangential displacement for in-plane flexural vibration of arches. Yang et al. [6] devel-
oped the governing differential equations for the free in-plane vibration of curved
beams using finite element method.
From the literature review, it is found that curvature plays an important role in the
modal properties of curved beams. Hence, it is necessary that the effect of curvature
on the modal properties be evaluated.
The present research work proposes a generalized mathematical equation for
predicting mode shapes of curved cantilever beams, which take into account the
change in length and curvature of the beam. Modeling of curved beam subjected
to cantilever boundary condition is done using ANSYS software. Twelve cases of
cantilever beams, each having different curvature and length, are considered.
2 Methodology
The methodology for carrying out the research work is shown in Fig. 1.
Simulation study was carried out for the various cases of the curved beams, viz.
Case I: Constant radius and varying length.
Cases II, III, IV: Constant length and varying radius.
These cases are shown in Table 1.
Mode 1
Mode 2
Mode 3
Mode 4
Mode 5
Mode 6
Further, modal analysis of beam of constant radius 478 mm and length 750 and
1000 mm was performed and fitness functions of the mode shapes of each were
obtained.
Equation to obtain the general curve is obtained using geometrical proof consid-
ering the arc T1CT2 of radius R and chord length L as shown in Fig. 4 and to obtain
the equation considering any vertical distance Ox at distance x.
Ox = OE − DO
In ODT1,
DO = R 2 − (L/2)2
In OE E,
OE = R2 − x 2
Ox = R2 − x 2 − R 2 − (L/2)2 (1)
The curve represented using Eq. (1) starts from the negative half of chord length;
therefore in order to initiate the curve from zero, shifting operation was performed
on the equation after which Eq. (1) becomes
412 A. Siddiqui et al.
Fig. 5 Comparison of original curve and curve generated using Eq. (2)
Ox = R 2 − [x − (L/2)]2 − R 2 − (L/2)2 (2)
Figure 5 shows the comparison of actual curve plotted from ANSYS data points
and curve plotted using Eq. (2) which is seen to be in good agreement with each
other. Similar procedure is followed for the case two of constant length and varying
radius, and fitness functions for the mode shapes were obtained and observed to be
similar. Thus, the polynomial equations were generalized for the mode shapes for
any length and radius expressed as:
m
y= R2 − x 2 − R 2 − (L/2)2 + F Ci X i (3)
i=0
where
R Radius of curvature
L Chord length
C Coefficient of equation
x
X=
L
R θ
F = × 1 − cos
5 2
m = mode_no + 3
ANSYS for the beam of radius 478 mm and length 500 mm. Similarly, mode shape
plot for the lengths 750 and 1000 mm can be tested.
4 Conclusion
The generalized equation obtained for the mode shapes of the curved cantilever
beams generates the mode shapes which are in good agreement with those obtained
from ANSYS.
Hence, generalized equation obtained using the current methodology can be used
for generation the mode shapes of a curved cantilever beam.
414 A. Siddiqui et al.
References
1. Lee BK, Wilson JF (1989) Free vibrations of arches with variable curvature. J Sound Vib
136(l):75–89
2. Gutierrez RH, Laura PAA, Rossi RE, Bertero R, Villaggi A (1989) In-plane vibrations of non-
circular arcs of non-uniform cross-section. J Sound Vib 129(2):181–200
3. Kang K, Bert CE, Striz AG (1995) Vibration analysis of shear deformable circular arches by the
differential quadrature method. J Sound Vib 353–360
4. Oh SJ, Lee BK, Lee IW (1999) Natural frequencies of non-circular arches with rotatory inertia
and shear deformation. J Sound Vib 219(1):23–33
5. Raveendranath P, Singh G, Pradhan B (2000) Free vibration of arches using a curved beam
element based on a coupled polynomial displacement field. Comput Struct 78:583–590
6. Yang F, Sedaghati R, Esmailzadeh E (2008) Free in-plane vibration of general curved beams
using finite element method. J Sound Vib 318:850–867
Variation in the Properties of Spot
Weldments of Cold Rolled Mild Steel
Welded with Filler Metal by Annealing
Treatment
Abstract Filler metal addition has been verified as an effective way to refine the
mechanical behaviour of cold rolled mild steel in resistance spot welding. Negligible
quantity of filler metal if added to the spot weld is found to improve mechanical prop-
erties of spot weldments, if no variation in the composition of base metal and filler
metal is allowed. Looking at practical applications, the sensitivity of the resistance
spot welding process with filler metal to variation in annealing treatment was exper-
imentally investigated. Filler metal quantity was from 30 to 70 mg and annealing
heat treatment was kept between 30 and 70%. The material of the filler metal is the
same as that of base metal and was added at the centre of overlap in lap joint. The
experimentation was carried out by spot welding of specimen varying filler metal
from 30 to 70 mg maintaining annealing cycle constant at 30%. The experimenta-
tion was then repeated for different annealing cycles. For 30–70% of anneal cycle,
30 mg quantity of added filler metal yielded maximum breaking point and plasticity
ensuring optimum load-bearing and energy absorption capability under the button
pullout mode.
Keywords Spot welding · Filler metal · Failure mode · Anneal · Cold rolled mild
steel
1 Introduction
For selecting materials for cars, car designers today seek materials with best stiffness,
mass reduction, safety performance, mass saving, formability, weldability, corrosion
resistance, fatigue resistance, cost, and environmental factors. Spot welding is the
S. T. Ambadkar (B)
Department of Mechanical Engineering, Government College of Engineering, Chandrapur,
Maharashtra, India
e-mail: sushipme@gmail.com
D. V. Bhope
Department of Mechanical Engineering, Rajiv Gandhi College of Engineering, Research and
Technology, Chandrapur, Maharashtra, India
2 Experimental Procedure
The material, i.e. cold rolled mild steel sheets used in this study is specifically for auto-
motive body applications and selected sheet had 1.25 mm thickness. The chemical
composition of this material is shown in Table 1.
Tensile testing of welded specimen was carried out to obtain tensile stress–strain
curves of the CR MS sheets. A Resistance spot welder, K J Thermoweld make with
specifications as mentioned in Table 2 was used to spot weld the specimen (Fig. 1).
Variation in the Properties of Spot Weldments … 417
Fig. 1 Resistance
spot welding machine
418 S. T. Ambadkar and D. V. Bhope
The cycle in the spot welder can be controlled with the help of parameters as
mentioned in Table 3. The machine was microcontroller based and energy levels
were programmable.
The dimensions of the spot weld specimen were 70 mm by 25 mm. The thickness
of CRMS sheet as mentioned was 1.25 mm thick. The electrode material was made
C15000 (copper alloy) and it had a face diameter of 5 mm. The coolant, i.e. water
flow-rate in the electrode was at 4 l/min (Fig. 2).
The spot welding schedules and parameters for experimentation are shown in
Table 3. The purpose of experimentation was to determine the effect of annealing on
the addition of filler metal in relation to mechanical behaviour. The composition of
filler metal was the same as that of base metal and it was added at the centre of overlap.
The extent of filler metal added in experimentation was from 30 to 70 mg for each
annealing energy level from 30 to 70%. The schedule followed is mentioned in Table 3
and was maintained unchanged with addition of filler metal and varying annealing
energy level. Specimens were tested in tensile tests and load–displacement curve
was plotted to obtain breaking capacity. The failure mode was also observed and was
interpreted by visual inspection. Some of the spot weld specimens were sectioned
through the cross-section perpendicular to the length direction and mounted for
macrostructure and microstructure observations following standard metallographic
procedures (Fig. 3).
Investigations are carried out by adding a small quantity of filler metal from 30 to
70 mg in spot weldments. The annealing energy level is varied from 30 to 70%.
Changes in is strength, plasticity is then determined using load–displacement curve
(Figs. 4 and 5).
Interfacial fracture and nugget pullout are two distinct failure modes observed
during tensile testing (Fig. 6). Table 4 shows, for 30–70% of annealing energy levels,
filler metal with 30 mg quantity yield maximum breaking strength and plasticity.
Based on this, it can be concluded that filler metal with 30 mg quantity produces
Fig. 4 Variation of displacement at max load (mm) and average breaking strength (kN) on Y-axes
with filler metal quantity (mg) on X-axes at 30–70% annealing energy level
420 S. T. Ambadkar and D. V. Bhope
Fig. 5 Specimen No. 135, weld metal and HAZ (filler metal 30 Mg, anneal 60%) weld metal
showing Widmanstatten ferrite along with normal ferrite (whitish) and pearlite (blackish), HAZ
showing tempered martensite (black) and ferrite
Fig. 6 Failure modes in spot weld specimens: button pull out and interfacial
optimum tensile performance, at all annealing levels. Most desirable button pull out
mode of failure is found in the 30 and 40 mg of filler metal with gradual transformation
to interfacial mode. This eventually is true at all annealing levels. Load carrying
capacity and energy absorption capability were found to be more for weldments
welded with 30 mg filler metal failing under the button pullout mode.
Maximum nugget diameter of 4.71 mm is obtained at 30% annealing correspond-
ing to addition of 30 mg filler metal but its further addition from 40 to 70 mg leads to
a gradual reduction in nugget diameter. At all annealing energy levels, 30 mg filler
metal yielded maximum nugget diameter with uniform reduction. Not much varia-
tion is visible in nugget diameter. For all filler metal and annealing combinations,
the properties seem to be governed by recrystallization annealing at 30% annealing
and full annealing at 40–70% annealing level.
In the present study, filler metal from 30 to 70 mg is added with annealing variation
from 30 to 70%. One way of improving the performance of resistance spot welds
is to change the microstructure by selecting a suitable post-weld heat treatment.
A localized post-weld heat treatment can be accomplished by inducing in process
annealing [6]. Annealing relieves internal stresses, refines structure and induce duc-
tility. The joint region for 30% annealing consisted of three separate zones: (i) fusion
Variation in the Properties of Spot Weldments … 421
Table 4 (continued)
Specimen Filler Breaking Annealing Type of Displacement Nugget
No. material load (kN) variation failure at max load diameter
by (%) (mm) (mm)
weight
(mg)
176 40 10.44 Button pull 3.8 4.59
out
177 50 10.3 Button and 3.76 4.01
interfacial
178 60 9.68 interfacial 3.72 3.8
179 70 9.52 interfacial 3.7 3.91
area also called weld nugget; adjacent to it is (ii) heat-affected zone and at the end
(iii) base metal. The material in the heat-affected zone must have faced maximum
temperature and must have cooled by cooling rate varying inversely with respect to
its distance from the fusion zone. The heat-affected zone phases near the fusion line
consist of martensite, ferrite. The treatment seems to be low-temperature tempering
which is relieving internal stresses thus improving toughness. The strength obtained
is found to be maximum for 30 mg filler metal with gradual reduction with respect
to filler metal. The fusion zone boundary phases of specimens produced tempered
martensite profile along with a significant fraction of ferrite. Tempering of marten-
site in the specimen is synonymous with tempering of steels as mentioned by Chuko
and Gould [6]. The distant microstructure from fusion zone consisted of ferrite and
pearlite phase. With an increase in filler metal and annealing temperature, a gradual
drop in strength and plasticity is observed. Any martensite formed after the welding
might have retransformed to austenite during high-temperature annealing with the
formation of new grains (recrystallization). This can be attributed to increased carbon
percentage due to the addition of filler metal locally, reducing critical transformation
range. Fast cooling rates were observed after re-austenitization and, again, predomi-
nantly martensite was reformed which resulted in reduction in plasticity and strength
as confirmed by Baltazar Hernandez et al. [7].
Interfacial fracture and nugget pullout are observed failure modes as shown in Fig. 6.
At 30–70% of annealing, addition of 30 mg of filler metal lead to maximum breaking
strength and elongation. Hence, it is concluded that the optimum tensile performance,
based on the peak load and energy absorbed at failure, is achieved with 30 mg filler
metal. Pull out mode of failure is found in the 30 and 40 mg of filler metal with smooth
and gradual transformation to interfacial mode. Pullout failure in the specimen can
Variation in the Properties of Spot Weldments … 423
The microstructure and the mechanical behaviour of spot-welded cold rolled mild
steel was successfully modified by adding filler metal and utilizing annealing heat
treatment cycle in resistance spot welding. The main results in this study are listed
as follows:
1. From experimentation, it is found that maximum breaking strength is obtained
with 30 mg filler metal for all annealing levels from 30 to 70%. Gradual drop in
breaking strength is visible with a further addition of filler metal. It is concluded
424 S. T. Ambadkar and D. V. Bhope
References
1. Ambadkar ST, Bhope DV (2018) Resistance spot welding of cold rolled mild steel with filler
metal. In: Advanced manufacturing and material science. Springer, Cham, pp 63–73
2. Rathbun RW, Matlock DK, Speer JG (2003) Fatigue behavior of spot welded high-strength sheet
steels. Welding J 82(8):207-s–218-s
3. Matlock DK, Krauss G, Ziaebrahimi F (1984) Strain hardening of dual phase steels: an evaluation
of the importance of processing history. In: Krauss G (ed) Deformation, processing, and structure.
ASM International, Materials Park, pp 47–87
4. Smith S, den Uijl N (2006) Resistance spot welding of advanced high strength steels for the
automotive industry. In: The 4th international seminar on advances in resistance welding. Wels,
Austria, pp 30–60
5. Khan MI, Kuntz ML, Biro E, Zhou Y (2008) Microstructure and mechanical properties of
resistance spot welded AHSS. Mater Trans JIM 49(7):1629–1637
6. Chuko WL, Gould JE (2002) Development of appropriate resistance spot welding practice for
transformation-hardened steels. Welding J 81(1):1-s–7-s
7. Baltazar Hernandez VH, Okita Y, Zhou Y (2012) Second pulse current in resistance spot welded
TRIP steel—effects on the microstructure and mechanical behavior. Welding J 91(10):278-s–
285-s
8. Kou S (2003) Welding metallurgy, 2nd edn. Wiley-Interscience, London, p 232
Comparison of Metro Track Vibration
with Federal Transit Administration
Limits
Abstract Structures located in the vicinity of at grade (surface) and elevated metro
rail are subjected to vibration induced by metro movements. Instruments with a
very high level of precision such as MRI, X-Ray and microscopes are sensitive to
external vibrations. Externally induced vibrations may affect overall performance of
such instruments. Also, considering the case of human comfort, noise produced by
such vibrations above audible limit causes discomfort in adjacent buildings, residents
and industries. It is thus important to measure quality and quantity of vibrations
transferred from metro track to ground. These measurements can be compared with
FTA limits and respective measures may be taken. In this paper, frequency and
amplitude of vibration are measured on the metro track and compared with the limits
provided in Federal Transit Administration (FTA) for human comfort. While the
metro train was in moving condition, track vibration was measured at a fixed point.
Measurements were taken for three different positions of metro train with respect to
the measurement point, such as when metro train is (a) approaching, (b) above and (c)
crossed the measurement point. amplitude and frequency content was also found for
all the three conditions. These experimental results are studied and discussed in this
paper, and they will provide the basis for excitation to be considered for analysing
vibrations of nearby structures or instruments.
1 Introduction
2 Measurement Set-Up
The measurements were carried out with the help of OROS data acquisition and
vibration analyser. It was carried out on track near Airport South metro station, Nag-
pur city, Maharashtra, India. This portion of metro rail is on ground which makes
it easy to measure track vibration. Figure 1 shows schematic diagram of measure-
ment set-up showing details of various components in it. Accelerometer position on
the track base is such that it will measure vibration amplitude in vertical direction.
Strong magnetic base at accelerometer end maintains firm contact with metro rail
track. Figure 1 shows the schematic of the measurement set-up and Fig. 2 shows
the photo of actual measurement set-up. As shown in Fig. 1 accelerometer is con-
nected to DAQ system. This signals through DAQ are communicated to the computer
through computer bus, which can be interpreted by computer software. For frequency
analysis, fast Fourier transform (FFT) was used.
The measurements were taken for 6 s interval when the train was in motion. Data
were acquired in three different conditions when metro train is (a) approaching, (b)
above and (c) crossed the measurement point. Figure 2 shows actual measurement
set-up of track vibration showing accelerometer, data acquisition system, battery and
laptop.
Table 1 Root mean square (RMS) values of amplitude for track vibrations for different conditions
Case Condition of measurement RMS (mm/s2 )
(w.r.t measurement point)
a When the train is approaching 2296.2
b When the train is above 3789.9
c When the train has crossed 756.1
Fig. 3 Spectrum of signal when the train is approaching the measurement point
3 Results
Measured vibration results were obtained through OROS software in terms of accel-
eration (m/s2 ) versus time (s) graph. Table 1 shows root mean square values of accel-
eration amplitude (m/s2 ) of track vibration when the train is at different positions
from the measurement point. Fast Fourier transform (FFT) is used to convert time
domain signal to frequency spectrum. With the help of this frequency spectrum, one
can pinpoint maximum amplitudes for its respective frequency. Frequency spectrum
of all the conditions are found.
Figure 3 shows spectrum of measured time domain signal when train is some
distance away and approaching the measurement point.
It was observed that when metro train is some distance away and approaching
the measurement point, maximum amplitude of acceleration occurred at 1600 Hz
with magnitude of 525 (mm/s2 ). Similar graphs were plotted for remaining two case
as shown in Fig. 3, and maximum acceleration amplitudes corresponding to their
frequencies were found. They are shown in table.
4 FTA Criteria
where for comparison of vibration speed, speedref is used. If the measured speed is
lower, then we must increase the dB level for comparison with speedref dB level.
Reference speed is defined as the most common speed in one set of measurements.
Structures and humans are having some sort of response to ground-borne vibrations,
some of which vibrations sources are illustrated in FTA [4]. Many different applica-
tions are having their specific range of vibration, but in this case, we are interested
in vibration level ranging from 50 to 100 vdB. For human comfort as well as many
instruments which are vibration-sensitive, the threshold specified is 65 vdB. In case
of highly sensitive equipment, such as electron microscope and lithography equip-
ment, below 48 vdB and 54 vdB, respectively, are considered. But when we consider
extremely sensitive equipments, vibration of 42 vdB is considered as safe.
Using Eq. (1), peak accelerations from Table 2 are converted to velocity. These
velocities are further converted to velocity levels in decibels using Eqs. (2) and (3).
Peak vibration level found at different train position with respect to measurement
point shown in Table 3. The values obtained as shown in Table 3 are around 10–15
vdB less than the limits provided in FTA. Limits provided in the FTA are given for
objects which are more than 15 ft away from the track. As the vibration levels directly
measured on the track are under limit, it can be said that the vibration at distance
more than 50 ft which is the minimum distance for having a structure near railway
track, will also be under limit.
430 C. V. Bhore et al.
Table 2 Maximum acceleration amplitude and corresponding frequencies for all three cases
Case Condition of Frequency (Hz) Acceleration (mm/s2 )
measurement (w.r.t
measurement point)
a When the train is 525 1600
approaching
b When the train is 2117 8100
above
c When the train has 390.5 7950
crossed
Table 3 Peak vibration velocity levels (db) and corresponding frequencies at different positions of
train
Case Condition of Velocity level (vdB) Frequency (Hz)
measurement (w.r.t
measurement point)
a When the train is 56.23 1600
approaching
b When the train is 58.40 8100
above
c When the train has 43.88 7950
crossed
5 Conclusion
Vibration induced due to train motion to the ground adjacent to its track was measured
and studied. Maximum value amplitude of acceleration was found out for three
cases of metro train position with respect to measurement point. As the metro rail in
Nagpur is yet to start in its full capacity, therefore, its speed in the trial period lies in
the range of 25–30 km/h. According to FTA guidelines, the velocity amplitudes of
vibration corresponding to their measured acceleration amplitudes at their respective
frequencies were calculated. On comparing with the FTA limits, it shows that they
are in limit and are around 10–15 vdB less than which was suggested. Therefore, it
can be said that there will be no disturbance to vibration-sensitive equipment as well
as to human comfort in nearby buildings.
References
1. Adam M, Von Estorff O (2005) Reduction of train-induced building vibrations by using open
and filled trenches. Comput Struct 83(1):11–24
2. Anderson DC (1994) Engineering prediction of railway vibration transmitted in buildings. Env-
iron Eng 7(1)
Comparison of Metro Track Vibration with Federal … 431
3. Xia H, Chen J, Wei P, Xia C, De Roeck Guido, Degrande Geert (2009) Experimental investigation
of railway train-induced vibrations of surrounding ground and a nearby multi-story building.
Earthq Eng Eng Vib 8(1):137–148
4. Hanson CE, Towers DA, Meister LD (2006) Transit noise and vibration impact assessment.
Technical report
Effect of Moisture Content and Fiber
Orientation on the Mechanical Behavior
of GFRP Composites
Abstract The glass fiber reinforced polymer (GFRP) has provided better strength
to weight ratio for various structural applications. Due to their non-corrosive nature,
they are particularly suited for corrosive environments where materials like steel can
cause problems. The aim of this research is to determine the combined effect of mois-
ture and fiber orientation on the static strength of GFRP composites. Accordingly,
twenty-four layered woven glass fiber/epoxy laminated composite was fabricated
using the autoclaving technique. Moisture absorption and mechanical tests were car-
ried out with two samples with different fiber orientation, i.e., 0°/90° and +45°/−45°.
The specimens accelerated moisture aged with 70 °C tap water immersion, and the
percentage change in moisture content was calculated up to saturation point. Then,
the tensile test was carried out using the Instron universal testing machine with a
constant crosshead speed of 1 mm/min. Higher tensile strength was observed for
0°/90° laminate with fiber allied parallel to the loading direction. Tensile strength
was decreasing post-moisture absorption as compared to virgin laminates. After the
mechanical tests, fracture analysis of tested specimens was carried out using scan-
ning electron microscopy (SEM) which revealed a reasonable post aging fiber–matrix
bonding degradation.
1 Introduction
Glass fiber reinforced polymer (CFRP) composite structures are widely employed as
aerospace and automobile components to reduce structural weight [1–3]. Composite
structures have gained importance in many weight-sensitive applications due to their
superior properties over metals. However, the use of GFRP composite laminate is
subjected to its physical and mechanical performance in the desired environmental
condition [4, 5]. The application sector involving high-temperature humid environ-
ment requires static and dynamic analysis of composite in that particular condition.
The main advantages of using fiberglass-reinforced thermosetting resin to produce
marine structures include lightweight structures, lower fabrication and raw material
cost, ease of fixing additional devices and ease of use [6]. Even though composites
have interesting mechanical properties, they are also characterized by disadvantages
such as their sensitivity to temperature and environment, which may influence their
mechanical properties and reduce the stiffness of the structure [7, 8]. This paper
examines the effect of long-term high-temperature water aging and fiber orientation
on static behavior of GFRP composite laminates.
Wf − Wi
%Moisture content = × 100% (1)
Wi
where W f = final weight of the composite after a particular time interval and W i =
initial weight of the composite both in grams. The tensile tests of both virgin and
moisture aged specimen were carried out using the Instron universal testing machine
(Make: INSTRON, Model: 8802, Capacity-250KN). The tests were carried out as per
ASTM D3039 standard with a constant crosshead speed of 1 mm/min. The tensile
tests of moisture aged specimens were carried out as soon as it reached a moisture
saturation point. Finally, the fracture morphology and mechanism were analyzed
using scanning electron micrographs (Make: JEOL, Model: 6380). for both virgin
and moisture aged specimens. Due to the non-conducting nature of the composite,
the sample was coated with palladium using a coater.
Effect of Moisture Content and Fiber Orientation … 435
Fig. 1 Samples before (a) and after (b) the burn-off test
The burning method (ASTM D2548-68) has been considered as a simple and effective
technique to determine the volume fraction of cured resin composite materials. A
furnace was preheated to 600 °C, and the specimen (20 mm × 10 mm × 4.5 mm)
was inserted into it placing it in a crucible. Prior to that, the mass of the specimen
and the crucible together was determined. The crucible was placed till all the epoxy
was degraded due to high temperature leaving behind only reinforcement. Then, the
crucible was removed from the furnace and placed carefully on a gram scale after it
reaches room temperature. A detailed image of sample before and after the burn-off
test is shown in Fig. 1a, b, respectively. The volume fraction of fibers was observed to
be 0.55, and weight fraction was 0.6. The mechanical properties of GFRP composite
laminate are related to the mechanical properties and fractional volume of each phase,
matrix and reinforcement. Amount of fiber in a fiber-reinforced composite directly
corresponds with the mechanical properties of the composite [9]. Adding too little
fiber reinforcement in the composite will actually deteriorate the properties of the
material.
Series 1–6 in Fig. 2 represent six static strength specimen which had minor vari-
ation in weight before aging. It can be noted that all the six samples showed a linear
slope increase for initial 4–5 weeks which was due to capillary action along the fibers.
Then, there was a minor weight gain or loss in the moisture absorption curve. The
initial rate of moisture absorption was 1%, and it was maximum after two months.
Time to reach the saturation point was about two months. This rate of moisture uptake
was dependent on the temperature, relative humidity, exposure time and mechani-
cal load. This absorbed moisture adversely affects the performance of laminate in a
number of ways like (1) dimensional changes (swelling); (2) reduction in the glass
transition temperature and (3) variation in mechanical and physical properties [5].
The effect of moisture is to cause hydrolytic breakdown of the fiber–matrix interface
resulting in a loss in the efficiency of load transfer between the matrix and the fiber
reinforcement [8].
The tensile strength of aged samples was observed to be decreased by about 50% as
shown in Fig. 3. The aged samples failed at half the load value as compared to virgin
samples as revealed from true stress vs true strain curves given in Fig. 4a–d. This
shows that there was a decrease in strength due to moisture absorption, irrespective
of the fiber orientation, i.e., 0°/90° and +45°/−45°. Even the percentage degradation
was very similar to each other. Fibers–matrix bonding degradation was the primary
concerning issue. The loss of tensile strength was dependent on exposure time, tem-
perature and degree of humidity [4]. On initial exposure to a high-temperature water
environment, the rate of fiber degradation is relatively rapid.
Effect of Moisture Content and Fiber Orientation … 437
Fig. 4 True stress versus true strain curve for a virgin 0°/90°, b Aged 0°/90°, c virgin +45°/−45°,
d aged +45°/−45° laminates
438 A. Behera et al.
Fig. 5 Scanning electron micrographs of a virgin 0°/90°, b virgin +45°/−45°, c aged 0°/90°,
d Aged +45°/−45° laminates
The fracture surfaces of both virgin and aged tensile test specimens were analyzed
using a scanning electron microscope. Figure 5a illustrates individual fibers separated
showing clearly the 0° and 90° orientation of fibers [9]. Similar layered wise tearing
of fibers was observed in virgin +45/−45 laminate as shown in Fig. 5b. This shows
that the fiber failure was in the direction of their orientation, i.e., 45° direction. The
fiber pull-out in brittle mode and fiber breakage was observed [10]. The moisture aged
specimen showed major delamination and weak interface caused due to the moisture
uptake as shown in Fig. 5c [6]. Matrix erosion was also observed in saturated moisture
aged specimen as shown in Fig. 5d.
4 Conclusions
1. The burn-off test analysis showed the fiber content of the sample to be 55%
2. There was a significant 50% decrease in tensile strength of aged 0°/90° laminate
as compared to virgin, whereas the difference was not that huge in +45°/−45°
specimen.
3. The thickness of the sample exposed to moisture was increased due to absorbed
moisture by epoxy.
4. The mode of failure was brittle matrix dominated with fiber pull-out, matrix
erosion and delamination in both virgin and aged specimens.
References
1. Menail Y, El Mahi A, Assarar M, Redjel B, Kondrats A (2009) The effects of water aging on the
mechanical properties of glass-fiber and Kevlar-fiber epoxy composite materials. Mechanika
76:28–32
2. Behera A, Dehury J, Thaware MM (2019) A comparative study on laminated and randomly
oriented Luffa-Kevlar reinforced hybrid composites. J Nat Fibers 16:237–244. https://doi.org/
10.1080/15440478.2017.1414653
3. Behera A, Thawre MM, Ballal A (2018) Effect of matrix crack generation on Fatigue life of
CFRP multidirectional laminates. Mater Today Proc 5:20078–20084. https://doi.org/10.1016/
j.matpr.2018.06.374
4. Poodts E, Minak G, Zucchelli A (2013) Impact of sea-water on the quasi static and fatigue
flexural properties of GFRP. Compos Struct 97:222–230. https://doi.org/10.1016/j.compstruct.
2012.10.021
5. Fang Y, Wang K, Hui D, Xu F, Liu W, Yang S, Wang L (2017) Monitoring of seawater immersion
degradation in glass fibre reinforced polymer composites using quantum dots. Compos Part B
Eng 112:93–102. https://doi.org/10.1016/j.compositesb.2016.12.043
6. Wang Z, Zhao XL, Xian G, Wu G, Singh Raman RK, Al-Saadi S (2017) Durability study on
interlaminar shear behaviour of basalt-, glass- and carbon-fibre reinforced polymer (B/G/CFRP)
bars in seawater sea sand concrete environment. Constr Build Mater 156:985–1004. https://
doi.org/10.1016/j.conbuildmat.2017.09.045
7. Malpot A, Touchard F, Bergamo S (2016) Influence of moisture on the fatigue behaviour of a
woven thermoplastic composite used for automotive application. Mater Des 98:12–19. https://
doi.org/10.1016/j.matdes.2016.02.123
8. Katunin A, Gnatowski A, Kajzer W (2015) Evolution of static and dynamic properties of GFRP
laminates during ageing in deionized and seawater. Adv Compos Lett 24:47–52
9. Behera A, Dupare P, Thawre MM, Ballal AR (2019) Effect of fatigue loading on stiffness
degradation, energy dissipation, and matrix cracking damage of CFRP [±45]3S composite
laminate. Fatigue Fract Eng Mater Struct 42:2302–2314. https://doi.org/10.1111/ffe.13065
10. Behera A, Thawre MM, Ballal A (2019) Failure analysis of CFRP multidirectional laminates
using the probabilistic weibull distribution model under static loading. Fibers Polym 20:2390–
2399
Experimental Investigation
and Simulation of Modified Evaporative
Cooling System
Abstract This paper presents design and fabrication of modified evaporative cooler
for producing cooled air without an increase in humidity. An experimental investi-
gation has been carried out in Ahmedabad, India, and subsequently, a mathematical
model is developed for the analysis of heat and mass transfer involved during the
process. The model is validated with the measured experimental data, and further, the
validated model has been used to analyze the performance of the system for Ahmed-
abad weather conditions throughout the year. The result shows that the minimum and
maximum reduced ambient temperature obtained using a modified evaporative cool-
ing (MEC) system is 6.52–34.31 °C and the maximum temperature drop is 17.85 °C.
This system is made of kraft paper and aluminum sheet and it can give a better result
than the direct and indirect evaporative cooler. The air can be cooled here lower than
the wet-bulb temperature without an increase in humidity, and subsequently, it is an
economical solution that can provide better comfort.
Nomenclature
Subscript
a Average
db Dry bulb
dp Dew point
in Inlet
l Wall
out Outlet
p Primary air
s Supply air
w Working air
wb Wet bulb
w_d Dry working air
wf Water film
wv Water vapour
Greek
1 Effectiveness
ρ Density (kg/m3 )
1 Introduction
Technological development and increased population growth along with the improve-
ment in living standards resulted in enlarging demands comfort in life. Therefore,
the demand for air conditioning systems is continuously increasing which leads to
large electricity consumption and also affects the environment both directly (i.e., use
of harmful refrigerant) and indirectly (i.e., electric consumption). However, there is
a large population who cannot afford air conditioning; hence, they have to depend
on the evaporative cooler, as it is economical. The conventional evaporative cooling
system can give satisfactory cooling but the uncontrolled humidity is the major disad-
vantage of the system, which creates discomfort. One of the ways to solve the above
Experimental Investigation and Simulation of Modified … 443
problem is to use an indirect evaporative system, which can control the humidity,
but it reduces the cooling and effectiveness compared to the direct evaporative sys-
tem. In order to overcome these restrictions, it needs to be redesigned and modified
for better comfort. There are recent developments of evaporative coolers, which are
summarized below.
Zhao et al. [1] presented a theoretical study on an indirect evaporative cooling
system with counter-flow arrangement. The wet-bulb effectiveness (WBE) and dew-
point effectiveness (DPE) were found up to 1.3 and 0.9, respectively, at operating
conditions of 28 °C dry-bulb temperature (DBT) and 20 °C wet-bulb temperature
(WBT). Riangvilaikul et al. [2] have presented an experimental investigation on
counter-flow dew-point evaporative cooler for dry, moderate, and humid climates.
The range of WBE and DPE obtained was 92–114% and 58–84%, respectively.
Subsequently, the theoretical model has also been presented, and the range of WBE
and DPE achieved was 106–109% and 65–86%, respectively [3]. Later, Jradi et al.
[4] presented on the cross-flow system, where WBE and DPE were found 112% and
78%, respectively, with 30 °C DBT and 50% relative humidity (RH) and 0.33 working
to intake airflow ratio. Heidarinejad et al. [5] have done study on indirect evaporative
cooler with a single and double stage, using a finite difference method. The results
show 50% improvement in WBE with two-stage system. Duan et al. [6] developed and
tested regenerative evaporative cooler with counter-flow arrangement for different
operating conditions and found 31% and 40% improvement in WBE and energy
efficiency ratio (EER), respectively. Khalid et al. [7] presented a study of counter-flow
indirect evaporative cooler for hot, moderate, and humid climates. It was reported
that WBE and DPE vary from 104–120 to 68–87% at inlet air temperature 25–45 °C.
Furthermore, Liu et al. [8] concluded from the parametric study for two-dimensional
numerical model that with increase in air ratio leads to higher temperature drop and
cooling capacity. Further, Wang et al. [9] presented a numerical study on dew-point
counter-flow indirect evaporative cooler. The optimum geometrical and operating
conditions were reported.
It has been observed from the literature that, a very limited study has been reported
in Indian climatic conditions. In present work, a counter-flow MEC system has been
designed, fabricated, and experimentally tested with local climatic conditions. A ther-
modynamic model has been developed and simulated using MATLAB® and validated
with experimental data. This validated model has been used to analyze the year-round
performance of the system using hourly ambient air condition (DBT and RH) data
(Taken from the IITRAM weather station, installed at IITRAM, Ahmedabad, India)
for Ahmedabad city.
2 Experimental Work
The MEC system is designed and fabricated in the laboratory. The schematic diagram
of MEC is shown in Fig. 1. The aluminum-coated kraft paper is used as a sheet
material for making heat and mass exchanger unit (HMU). Thickness of aluminum
444 M. Lata and D. K. Gupta
sheet is 0.2 mm while kraft paper’s thickness is 0.27 mm. Dry and wet channel is
of 1000 mm * 550 mm * 1000 mm (length * Width * Height). The dry and wet
passage gap is 5 mm with perforation diameter of 5 mm. The total number of 40 dry
and 41 wet passages is used in HMU. There are three streams of air, i.e., air flowing
in dry passages (product air), wet passages (working air), and the final output air
ready to supply (supply air) as shown in Fig. 2, which pass between the dry and wet
passage (aluminum-coated kraft paper) alternatively. A fraction of product air, after
the dry passage, is allowed to pass through the wet passage from the top end of each
sheet, and it is controlled by the perforation made in each sheet. Only sensible heat
transfer takes place from product air to working air in the dry passage, and it divides
into two streams at the top end of the dry passage, i.e., one stream as working air
to wet passage and another as supply air. The working air gets pre-cooled before
entering the wet passage and it flows in reverse direction of the product air, which
enhances the effectiveness due to a higher temperature gradient between dry and wet
passage. In the wet passage, both latent and sensible heat transfers take place. The
working air comes in contact with wetted kraft paper, and evaporation of water takes
place by absorbing the heat from it, which cools the wet surface, and working air,
subsequently, allows sensible heat transfer from product air to working air. Finally,
humidified air is discharged to the atmosphere from bottom of the wet passage.
Exhaust Wet
Working air
air Passage
Water film
3 Mathematical Model
A mathematical model has been developed to study the process of heat and mass
transfer, involved in the HMU. The heat and mass transfer equation is solved using the
finite difference equation and simulated with the help of MATLAB® . The following
assumptions have been made for the modeling: (a) negligible heat loss between
system and surrounding, (b) interface temperature between working air and water
film is water temperature, (c) the uniform properties and velocity of the fluid in a
control volume, (d) uniform distribution of water in the wet channel, and (e) negligible
thermal resistance.
Energy balance equation for dry passage
ṁ p Cp tp_ in − tp_ out = h p dAs tp_ a − tl_ a (1)
ṁ wf_ in − ṁ wf_ out = ṁ w Ww_ in − Ww_ out (3)
ṁ w Hw_ out − Hw_ in = h w dAs twfa − twa + Hwv h m dAs ρwf_ a − ρw_ a (4)
Equations 1–5 have been solved simultaneously using the iterative method, and
the cooling performance of MEC is evaluated by
tdb,in − tdb,out
εwb = (6)
tdb,in − twb,in
tdb,in − tdb,out
εdp = (7)
tdb,in − tdp,in
4 Model Validation
The mathematical model has been validated with the real-time data taken for the
month of June 2018 (i.e., peak summer conditions) from 11.00 a.m. to 10.00 p.m.
446 M. Lata and D. K. Gupta
0.8
Effectiveness
20 0.017
Fig. 3 Comparison of experimental and simulation data by varying ambient air conditions
(IST) at Ahmedabad, Gujarat, India. It is observed that the ambient air temperature
(AAT) and humidity ratio (HR) during the experimentation period varies from 34.4
to 39.5 °C, and 0.0206 and 0.0145 kg/kg of dry air, respectively. The minimum
OAT is 34.4 °C with 0.0204 kg/kg (HR), whereas the maximum AAT is 39.5 °C
with 0.0151 kg/kg (HR). Figure 3 depicts the comparison of experimental and sim-
ulated results, which includes the variation of reduced ambient temperature (RAT),
temperature drop (TD), WBE, and DPE with different AAT and ambient HR. The
operating range of AAT and ambient HR varies between 36.8 and 39.5 °C and 0.0173–
0.0151 kg/kg of dry air, respectively. A very good agreement between the simulated
and experimental results has been found.
The validated mathematical model of the MEC system is used to analyze the year-
round performance of the system for Ahmedabad city using hourly ambient air con-
dition (DBT and RH) data. Figure 4 shows the variation in bin hours with respect to
the ambient air conditions throughout the year. Ambient air condition is represented
40
35
30
Bin hours
25
20
15
10
5
0
0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500
Ambient air condition
45
temperature ( °C)
Reduced ambient
35 RAT TD
25
15
-5 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400
Ambient air condition
on X-axis, and Y-axis represents bin hours. Annually, DBT and RH vary between
8–44 °C and 23–100%, respectively.
Figure 5 shows the variation of reduced ambient air temperature and temperature
drop, using the MEC system at different ambient air conditions throughout the year
as shown in Fig. 4. It has been observed that minimum and maximum reduced
ambient temperature obtained by using the MEC system is 6.52 °C at the ambient
condition of 8 °C DBT and 78% RH, and 34.31 °C, at the ambient condition of 42 °C
DBT and 54% RH. However, maximum temperature drop obtained is 17.85 °C at
40 °C and 24% RH. It has also been observed that the performance of the system is
better with the dry and hot climate. However, the performance is always better than
the conventional system, as it reduces temperature even below the WBT, without
increasing humidity. The system ensures that the maximum output temperature of
air never goes beyond the 34.3 °C with the respective humidity throughout the year
in Ahmedabad’s climatic conditions.
6 Conclusions
Experimental investigation and thermodynamic study have been carried out for the
MEC system having a counter-flow configuration of HMU. The year-round perfor-
mance of the MEC system has been predicted, for Ahmedabad weather conditions. It
is concluded that the maximum temperature drop obtained using the MEC system is
found 17.85 °C at 40 °C and 24% RH. The proposed system can provide better com-
fort than the conventional configurations, throughout the year. This system can also
be used to supply secondary air for any other cooling system of HVAC application.
448 M. Lata and D. K. Gupta
References
1. Zhao X, Li JM, Riffat SB (2008) Numerical study of a novel counter-flow heat and mass
exchanger for dew point evaporative cooling. Appl Therm Eng 28:1942–1951
2. Riangvilaikul B, Kumar S (2010) An experimental study of a novel dew point evaporative cooling
system. Energy Build 42:637–644
3. Riangvilaikul B, Kumar S (2010) Numerical study of a novel dew point evaporative cooling
system. Energy Build 42:2241–2250
4. Jradi M, Riffat S (2014) Experimental and numerical investigation of a dew-point cooling system
for thermal comfort in buildings. Appl Energy 132:524–535
5. Heidarinejad G, Moshari S (2015) Novel modeling of an indirect evaporative cooling system
with cross-flow configuration. Energy Build 92:351–362
6. Duan Z, Zhan C, Zhao X, Dong X (2016) Experimental study of a counter-flow regenerative
evaporative cooler. Build Environ 104:47–58
7. Khalid O, Butt Z, Tanveer W, Rao HI (2017) Design and experimental analysis of counter-flow
heat and mass exchanger incorporating (M-cycle) for evaporative cooling. Heat Mass Transfer
53:1391–1403
8. Liu Y, Li JM, Yang X, Zhao X (2018) Two-dimensional numerical study of a heat and mass
exchanger for a dew-point evaporative cooler. Energy 168:975–988
9. Wang L, Zhan C, Zhang J, Zhao X (2019) Optimization of the counter-flow heat and mass
exchanger for M-Cycle indirect evaporative cooling assisted with entropy analysis. Energy
171:1206–1216
Effects of Different Vegetable Oils
and Additives in Gearbox Operation
and its Condition Monitoring
Keywords HP EP 90 · Coconut oil · Castor oil · Garlic oil · Rapeseed oil · Worm
gear box · Oils · Additives
1 Introduction
For the proper operation and compliance of worm gear, thermal stability plays an
important role and it primarily depends on lubricating oil and additives. Temperature
has direct effect on the lubrication oil viscosity, and however, the oil viscosity as well
significantly affects the processes in the sliding zone, and hence the temperatures.
The optimum lubrication for the chosen lubricating oil can be achieved only within
the suitable range of the operating temperature. Given a much higher proportion of
sliding relative to rolling movement, worm gears have far more friction between
the teeth sides than in the case of cylindrical and cone gears, which results in a
significantly lower efficiency [1]. It is known that the efficiency of a worm pair
depends on lubricant, sliding speed, load, and temperature [2].
Ruggieroa et al. [3] analyzed hydro-treated vegetable oil from raw rapeseed oil
and raw Jatropha carcass oil. They have observed that the selected oils showed good
physicochemical properties and can favorably be used as lubricant feedstock.
Thottackkad1 et al. [4] in their research work have explored the possibility of
use of vegetable base oils as automobile lubricants. They have concluded that as a
lubricant, friction-reduction properties of coconut oil are enhanced by the addition
of CuO nanoparticles to a moderate concentration.
Mannekote and Kailas [5] in their research work have evaluated coconut and palm
oil as lubricants in four-stroke engine. They have observed that antiwear properties
of fresh vegetable base oils and engine oil were comparable at ambient conditions.
Li et al. [6] evaluated natural garlic oil (NGO) as an extreme pressure additive
for lubricating oils. They have demonstrated that NGO could provide superior load-
carrying ability in the selected base fluids than the conventional extreme pressure
additive SIB.
Jayadas et al. [7] experimentally evaluated the effect of an AW/EP additive on the
tribological performance of coconut oil. They have observed that the addition of the
AW/EP additive has shown substantial reduction in wear with coconut oil as 2T oil.
Wu et al. [8] evaluated tribological properties of castor oil. Results observed
showed that castor oil as the additive could efficiently reduce the friction and wear.
Shanhua et al. [9] claimed that a small amount of ionic liquid can postpone the
peak coefficient of friction and decrease the wear behavior of castor oil.
Penga et al. [10] given detailed information on the dependent and independent
roles of vibration and wear debris analyses in predicting and diagnosing machine
faults.
Mineral oils are petroleum by-products and have very poor biodegradability, and
they are major source of environmental pollution. Oils of natural ester (vegetable
oils) have capability to act as a base for the environmentally friendly lubricating oils.
They have good lubricity, biodegradation capability, viscosity versus temperature
characteristics, low evaporation capacity, etc. Vegetable oils have high viscosity
index (VI). Table 1 [11] provides details regarding rating of vegetable oil [rapeseed
oil]properties compared with mineral oils ratings: A stands for excellent, B stands for
Effects of Different Vegetable Oils and Additives … 451
very good, C stands for good, D stands for moderate, and E stands for poor. The main
important property of vegetable oil which is completely different from mineral oils is
the carbons may or may not be saturated with hydrogen. Because of the presence of
triglyceride units, vegetable oils are more polar compared with the mineral oils also
triglycerides have a higher affinity to metal surfaces. The structure of the triglycerides
of vegetable oils provided good properties of a lubricant. The important properties
of comparison of the coconut oil and rapeseed oil are mentioned in Table 2 [11].
Biolubricants can act as potential substitute lubricants in different applications.
One of the most effective properties of the biolubricant is their fast biodegradability
and good VI. An increased number of studies have shown that the biolubricants have
very good potential to be used as alternative lubricant. There are many researches
going on in this area worldwide but still there is need of a systematic research to
evaluate and validate the tribological behavior of different biolubricants for different
applications and under different working conditions. As mentioned in Table 1, the
biolubricants have advantages over the mineral oils in comparison with the important
oil properties. Biolubricants have few disadvantages which are as well mentioned in
Table 1.
Various researches have worked on the possibility of using vegetable oils as lubri-
cants using various methods of evaluation. Table 3 demonstrates the work done with
a comparison done with the standard SAE reference lubricant [12].
452 A. Dube and M. D. Jaybhaye
Researchers have also shown that natural oils can also be used as an additive (like
rapeseed oil, palm oil, etc.) as an antiwear additive and extreme pressure additives.
Antiwear additives protect the sliding machine parts from wear and extreme pressure
additives and create a thin film which prevents welding and seizure of contacting
surfaces during working life of the sliding components. Based on the literature survey
and availability in the market, natural garlic oil and rapeseed oil are used as an additive
for the selected base oils that is coconut oil and castor oil. Literature survey suggests
that natural garlic oil can act as a good lubrication oil additive. It has pale yellow
color and has refractive index as 1.566, specific gravity (20 °C) as 1.081, acid value
(mgKOH/g) as 1.6, and peroxide value (meq/kg) 2.03.
Base oils and additives both are natural oils and are completely soluble without
any solid particles remains in the blends. Table 4 shows the blends which are selected
for this study. The behavior of the selected blends is compared with the standards
HP EP 90 gear oil. Physical properties of selected natural oils are listed in Table 5
[13].
A worm gear test rig system is designed for carrying out this research work. Figure 1
shows the components of the designed test rig. The designed test rig apparatus con-
tains a model gear box with half HP motor attached to it. A system is designed for
applying loads at the output shaft of the gear box using hanger bearing. A square
base is designed for this test rig. Rubber bushings are provided at the four legs of the
rig to avoid any vibrations due to uneven resting of the test rig.
In the designed test rig for applying load on the output shaft of the gear box
assembly, a load cell and hanger bearing arrangement is used. The load cell and the
hanger bearing are attached with each other using a threaded rod. A bolt is present at
the top end of the load cell attachment plate. On tightening, the bold load gets applied
Effects of Different Vegetable Oils and Additives … 453
on the output shaft of the gear box. The applied load can be measured using an output
display unit attached with the load cell assembly. The output display is a commonly
used weighing scale which displayed applied load in KG unit. The selected load cell
is calibrated using a KG weight block. On keeping the KG blocks on this system, the
output shows exactly same quantity in the display unit scale. Digital output devices
are used for viewing the motor RMP value, gear box oil temperature, and load applied
on the out rod in KG.
For measuring accurate temperature of entire system, a smart vibration sensor
QM42VT2 with bi-axial accelerometer and temperature measurement capability was
selected. This is a plug and play type vibration sensor and does not need a separate
data logger. All the electronics including filters and ATD converter are packed in a
single unit that can be mounted on any parts for which vibrations and temperature
are to be measured using a special mounting bracket and RS485 to USB connector.
454 A. Dube and M. D. Jaybhaye
5 Method of Experimentation
For the current experimentation, gear box temperature and gear box oil temperature
are selected as parameters. The test rig is operated for 70 min and applied load is
increased by 10 kg after each 10 min run. Readings were taken after every 10 min.
Two different readings were taken to check if there is any sudden change in the
readings and then average of 2 readings is taken as final reading. A test plan was
developed for taking different readings at the selected locations.
Readings were taken for each selected five oil blends mentioned in Table 4. For
recording gear box temperature responses, the Banner vibration–temperature sensors
are used, and for recording gear oil temperature, RTD sensor with a digital display is
used. The output load applied on rod readings was taken using a load cell mechanism
with weighing scale display.
6 Results
Graphs are generated for the temperature changes for the selected locations with
respect to the varying loading condition for each selected oil blends. The readings
are compared with the standard HP EP 90 commercial gear oil.
Figure 2 shows gear oil temperature vs load applied. It has been observed that the
temperature changes are less for coconut, castor additive, and coconut additive and
higher for castor oil as compared with the standard HP EP 90 oil. Coconut and its
additive perform way better than all.
Figure 3 shows gear box temperature vs load applied. It has been observed that
the temperature changes are less for coconut and coconut additive and higher for
Effects of Different Vegetable Oils and Additives … 455
castor oil and almost same for castor additive as compared with the standard HP EP
90 oil. Coconut and its additive perform way better than all.
7 Conclusions
For the designed test rig and for the selected oil blends and for the selected loading
conditions, it can be concluded that coconut additive blend has very good tempera-
ture absorbing properties as compared with other selected oil blends. Performance
of coconut additive blend is better than that of the coconut oil blend without any
additive. [For maximum load condition coconut additive shows 14% less temper-
ature while coconut oil shows 9% less temperature compared to the standard HP
EP 90 oil]. Castor oil has somewhat less temperature observing properties as com-
pared with other selected oil blends. Castor oil with additive performs better than
that of the castor oil without additive. Gear box housing and gear oil temperature are
almost same throughout the test. The selected oil blends have potential to be used as
lubricating oils with proper additives added in them.
456 A. Dube and M. D. Jaybhaye
References
Abstract Among the recent challenges in India, one is the extensive consumption
of non-renewable energy sources due to rising demands of fuels and environmental
issues. India is an agriculture-based country with the second-highest population in
the world. Converting the waste material into the energy and use of renewable sources
is the best way to serve the energy needs of the population. The biogas production
in India is around 2.07 billion cubic meters per year which are targeted to increase
up to 30 billion cubic meters per year by using various advanced technologies. To
extract more methane percentage, various plants are approaching toward pressure
swing adsorption (PSA) technology. PSA technology increases the percentage of
methane by reducing carbon dioxide and other gases’ percentage. In this paper, we
have studied and analyzed various parameters of a bio-mechanization plant which
is based on segregated organic municipal waste for the production of bio-CNG and
organic fertilizer and is working on the vapor pressure swing (VPSA) technology.
The effect of season, temperature, waste obtained, etc. on bio-CNG production has
been researched.
1 Introduction
The most important factor for India to reach the status of developed countries is
the renewable source of energy. The annual LPG consumption in India is 101.4 kg
and 119.3 kg for rural and urban households, respectively [1]. Biogas is a renewable
energy source that is generated by the action of the methanogenic bacteria on the
municipal solid waste (MSW), industrial wastes, hotel waste, animal and agricultural
wastes. Biogas as a renewable energy source contributes 14% of the total renewable
energy produced in India [2]. It has been estimated that India has biogas production
potential of 40,734 Mm3 /year from different organic wastes [3]. More than 127,486
tons per day of MSW is been generated [4]. National Biogas and Manure Management
Program (NBMMP) organization of India is taking efforts to install biogas plants in
rural and semi-urban areas in various states of India [5].
Biogas is a mixture of various gases such as methane, carbon dioxide, hydrogen
sulfide, nitrogen, and water vapors. The percentage of methane (CH4 ) is in a range
of 50–70% and that of carbon dioxide (CO2 ) is 30–50%. Small traces of nitrogen
(N2 ) and oxygen (O2 ) are present at concentration of 0–3% and water vapors (H2 O)
at concentration of 5–10% [6]. The purification of biogas can be done by the various
technologies. The percentage extraction of methane depends on the technology being
used. The technologies used are cryogenic separation of gases, physical and chemical
adsorption, membrane-based technology, water scrubbing, chemical scrubbing, and
other biological methods of separation [7, 8]. Water scrubbing is the oldest technol-
ogy which is mostly used in India in old biogas plants. The percentage of methane
obtained from water scrubbing is about 58% which is less and needs to be improved
[9]. The biological method of methane production is an upgrading technology and
research is being carried out to improve the bio-mechanization efficiency [6]. Various
methods have been tested for the reduction of carbon dioxide in biogas by the use
of alkali metal compounds. Biogas plants are looking forward to some modern tech-
niques which improve methane percentage up to maximum level by reducing carbon
dioxide for bio-CNG. Pressure swing adsorption (PSA) is a promising technology
which improves the methane percentage up to 85% by reducing carbon dioxide into
methane [10]. The purification of biogas depends on the various parameters. For
example, in summer and winter, the amount of methane produced is different [11].
Also, the temperature and type of waste affect the rate of purification in different
seasons [12].
In VPSA, biogas is compressed at high pressure and then fed to the column.
The column consists of adsorbents like activated carbons, zeolite molecular sieves,
etc. which have high porosity. Methane-rich gas is obtained at top of the column
and CO2 is removed. Hydrogen sulfide adsorption is irreversible so it is separated
first. As the adsorbent gets saturated with CO2 after several passages, it has to be
regenerated. It can be easily done by reducing the pressure up to vacuum pressure.
This method yields highly pure form of bio-CNG, and hence, its working was studied
in bio-mechanization plant based on VPSA.
The bio-mechanization plant of Indore has been researched by our team. The name
of the plant is “Design, Construction, Supply, Installation, and Commissioning of 15
TPD Bio Mechanisation Plant-Based on Segregated Organic Municipal Waste for
Production of Bio-CNG And Fertilizer” and the work is executed by Mailhem Ikos
Environment Private Limited. The plant is situated at Kabit Khedi, Indore. Various
readings were taken to plot the graphs. The recordings obtained were analyzed to
come across some crucial conclusions.
Study and Analysis of Various Parameters … 459
2 Methodology
As briefed earlier, the potential of food waste for biofuels was realized and research
was carried out. Various technologies were studied and it was found out that PSA
and water scrubbing technologies are effective in India. The most recent plant with
advanced technology setup in 2018 by the Government of India is in Indore, Madhya
Pradesh. It purifies biogas with the use of vapor pressure swing adsorption (VPSA)
technology. The plant supplies bio-CNG to be used by the local buses by collecting
local hotel waste. We visited the plant on June 8–9, 2019, to know more about
VPSA technology. The data collected included waste intake and bio-CNG filled for
three months (from February 2019 to April 2019) (Fig. 1).
The data collected was conditioned for parameter selection. A quantitative
approach was used and analysis was done after parameter selection and correla-
tion. This research paper is basic, exploratory, and inductive. The data is primary,
quantitative, and descriptive. It is fixed, field research with longitudinal studies and
non-probability sampling. Based on the waste-related data, the quantity of useful
waste was taken into consideration. The record of CNG filled in the vehicles was
collected. The data is organized in a pivot table in Excel, in order to extract dataset.
We plotted graph for waste rejected and net waste for a particular period. Also, to
establish the effect of season on the biogas purified and temperature variation, graph
was plotted on the corresponding parameters.
The flow of materials and plant description:
Daily waste is collected from local hotels by municipal garbage trucks and brought
to the plant. The waste is weighed on the weighing machine, which is capable of
measuring tons of weight. Then the waste is dumped in an area allocated for dumping
near to the feeding conveyor, on which the waste is manually segregated by the
workers. Segregation is required as the waste comprises both biodegradable and
non-biodegradable types of waste. From the feeding conveyor, after the separation,
the waste comes on to shredding table where a 5Hp motor shreds the biodegradable
waste into finely chopped particles of the waste. Recycled water from the digester is
being used in the shredder for proper mixing of the biodegradable wastes. From the
shredder, the slurry of bio-digester goes into a pre-digester where it is stored for a
while before inserting it into the primary digester. The larger sized particles which
are not chopped into shredder are again brought back onto the shredder table for
shredding. Organic waste compost is obtained from the solid–liquid separator which
is useful as a fertilizer in farms. From the pre-digester, the digest is pumped inside
a slurry tank, also called primary digester and secondary digester, with the help of
two 5HP motors.
The digest is injected from the top in the slurry tank. The anaerobic bacteria
digest the waste and produce slurry mixed up of some raw biogas. An agitator helps
to reduce the formation of foam/scum. Slurry form of waste is injected into the
Mailhem Ikos MUASB digester to produce the purified form of methane gas after
the pre-digester. The digester has a temperature and pH level sensor to maintain the
temperature and pH level of the digester within prescribed limits. The recycle and
sludge lines are present to reuse the highly viscous digestate present in the digester.
From the digester, the lines go to the storage of the raw biogas. Moisture trappers
on the lines help to reduce the moisture content in the raw biogas. The balloon has
a capacity of 375 m3 each to store the raw biogas into it. From the balloon, the raw
biogas is let into the VPSA purification system wherein biogas is purified to bio-CNG
(Fig. 2).
Along the way, a mass flow meter is used to measure the amount of raw biogas and
a flare unit is inserted to flare the gas into the atmosphere if the purification system
is not in working condition. A blower helps to increase the mass flow rate of the raw
biogas and increase its capacity of purification. The raw biogas is purified inside the
purification system (VPSA technique) and is regularly monitored by gas chromatog-
raphy. After the purification, the purified gas is compressed by a compressor so as
to fill in the tank for storage purpose. Further, the gas is dispensed through a CNG
dispenser and filled in government buses.
The observed readings of CNG produced, net waste, waste rejected and atmospheric
temperature is taken and analyzed to plot the effect of temperature variation on
CNG production, effect of used MSW variation on CNG production, and variation
of the MSW and waste rejected. Figure 3 shows the variation of the MSW and waste
rejected. The MSW generated in the first half of March summed to the value of
approximately 160,000 kg. It shows that more the waste generated, more care is to
be taken to reject the useless waste to produce good quality CNG.
Figure 4 shows the variation of CNG produced from the MSW in the interval
of seven days. It can be observed that the CNG production is directly proportional
to the amount of solid waste fed. The month of March shows the highest waste
generation which leads to increased production of CNG which is approximately
equal to 1800 kg. The readings taken were used for the ratio of CNG produced (kg)
per kg waste and are plotted in graphical form against the atmospheric temperature.
Figure 5 shows the effect of temperature variation on the ratio of the amount of
purified CNG produced to used MSW. It can be concluded that there is a seasonal
effect on the purified CNG produced via V-PSA, which increases with an increase
in the temperature from winter to summer season.
4 Conclusion
With the change in season, the quality of biogas is altered. As temperature increases,
quality of biogas increases, thereby quality of bio-CNG. The quality of biogas affects
the VPSA purification process. It can be concluded that seasonal effect occurs on the
purification system and bio-CNG produced per kg of waste increases as temperature
increases from winter to summer.
Acknowledgements Special thanks to Mailhem Ikos Environment Pvt. Ltd. which is a global waste
management conglomerate offering customized solutions in solid waste treatment and management.
The plant is a joint venture between two leading waste management firms Mailhem Engineers Pvt.
Ltd., India, and Lhotellier Ikos Group, France. All the guidance and mentoring during the tenure
of project was provided by research and development department.
References
1. Kadam R, Panwar NL (2017) Recent advancement in biogas enrichment and its applications.
Elsevier
2. Overview of Biogas in India. https://www.globalmethane.org/documents/16.Ag2.1.Dhussa.
pdf. Accessed 04.09.19
3. Rupnar AK, Jain S, Panwar NL (2018) Biogas in India: potential and integration into present
energy systems. Int J Curr Microbiol Appl Sci 07(07):2175–2186
4. CPCB India (2012) Central pollution control board. Government of India
5. Ministry of New and Renewable Energy [National Biogas and Manure Management Pro-
gramme (NBMMP)]. https://mnre.gov.in/biogas. Accessed 25.07.19
6. Angelidakia I, Treua L, Tsapekosa P, Luoc G, Campanarob S, Wenzeld H, Kougiasa PG (2018)
Biogas upgrading and utilization: current status and perspectives. Elsevier
7. Prussi M, Padella M, Conton M, Postma ED, Lonza L (2019) Review of technologies for
Biomethane production and assessment of Eu transport share in 2030. Elsevier
8. Ong MD, Williams RB, Kaffka SR (2014) DRAFT comparative assessment of technology
options for biogas clean-up. In: Public interest energy research (PIER) program draft interim
project report, Oct 2014
9. Olugasa TT, Oyesile OA (2015) Design and construction of a water scrubber for the upgrading
of biogas. J Fundam Renew Energy Appl 05(05):1–6
Study and Analysis of Various Parameters … 463
10. Magomnang AASM, Maglinao AL, Capareda SC, Villanueva EP (2018) Evaluating the system
performance of a pressure swing adsorption (PSA) unit by removing the carbon dioxide from
biogas. Indian J Sci Technol 11(17):1–17
11. Shukla PV, Bhalerao TS, Ingle ST (2010) Effect of seasonal variation on biogas production
from different food waste. J Pure Appl Microbiol Res Gate 4(1):333–337
12. Ramaraj R, Unpaprom Y (2016) Effect of temperature on the performance of biogas production
from duckweed. Chem Res J 1(1):58–66
Robust Sliding Mode Controller (RSMC)
for an Omniwheeled Mobile Robot with
Uncertainties and External Perturbations
Abstract Mobile robots at present are used extensively in the industrial and com-
mercial sectors. There is a need for a robot that can easily maneuver in the sparse
space at the warehouse, stores, etc. Omnidirectional robots have an upper hand over
conventional mobile robots as they need not change their orientation while maneu-
vering. In this paper, we are putting forth the design of robust controller for omni-
wheeled mobile robot (OWMR). In the beginning, kinematic and dynamic modeling
of OWMR was done; then, PID and sliding mode controller (SMC) were imple-
mented on the OWMR. The proposed controller tracks the trajectory with greater
accuracy as compared to other standard controllers.
1 Introduction
45◦ with the side of the base (Fig. 1). The adjacent wheels are kept perpendicular to
each other, while opposite wheels are made parallel to each other.
Kinematic and dynamic modeling of OWMR was done by taking voltage as an
input to motors, whereas most of the previous researches were focused on the torque-
based approach. Kinematical modeling [2, 3] was done considering different frames,
and transformation matrices were used to obtain the kinematic equations in the world
frame. Dynamic modeling was done by using the Newton–Euler method taking into
account all external perturbations and uncertainties.
The dynamic model of the robot is nonlinear. As the external disturbances and
uncertainties are bounded but unknown, the standard linear controllers like PID
cannot track the trajectory accurately. In this case, sliding mode controller [4] is very
effective and gives much better results in trajectory tracking. It consists of nominal
as well as the discontinuous controller. The nominal part of the controller is used to
achieve the trajectory, while discontinuous part of the controller makes the OWMR
robust [5].
The paper has been divided into six sections. We have dealt with the kinematic
model of the OWMR in Sect. 2. In Sect. 3, we have discussed the dynamic model of
OWMR. Section 4 focuses on the controller design for OWMR. Behavior of OWMR
while using standard controllers and SMC has been shown in Sect. 5 with simulation
results. The paper will end with the conclusion in Sect. 6.
2 Kinematic Modeling
Figure 1 shows the top view of OWMR. The system is studied from six different
frames. Owi X wi Ywi represents the frame assigned to four wheels of the robot,
where i is referred to the ith no. of wheel. Robot moving frame attached with robot
platform is represented by Or X r Yr , and the world frame is represented by Oq
X q Yq . Pr = [xr yr φr ]T is the position vector of the robot in Or , where xr and
yr are used to represent the position of the robot and φr represents the rotational
orientation of robot [6]. Pq = [xq yq φq ]T is the position vector of the robot in Oq .
Ṗr = [ẋr ẏr φ̇r ]T represents the velocity vector in Or . The velocity vector Pr in
terms of [θ̇1 θ̇2 θ̇3 θ̇4 ]T , where θ̇i represents the velocity of the i th wheel [2], is
given by:
⎡ ⎤
⎡ ⎤ ⎡ ⎤ θ̇1
ẋr 0 1 0 −1 ⎢ ⎥
⎣ ẏr ⎦ = R ⎣ 1 0 −1 0 ⎦ ⎢θ̇2 ⎥ (1)
2 −1 −1 −1 −1 ⎣θ̇3 ⎦
φ̇r 2a 2a 2a 2a
θ̇4
Robust Sliding Mode Controller (RSMC) for an Omniwheeled Mobile Robot … 467
3 Dynamic Modeling
The free-body diagram of all the forces acting on the robot is shown in the Fig. 2. Fi
represents the driving force due to the actuator on the ith wheel of the robot [7]. It
is given as Fi = αu i − β R θ̇i . α and β are coefficients of motors and are calculated
using datasheet of motor and u i is the controller input. τ is the net torque due to all
the forces acting on the robot. Fex is the unknown external force acting on the robot
at an angle ψ with Yr at height b from top corner of the robot. Applying Newton’s
second law of translational and rotational motion, i.e., F = Ma and τ = I α in Or
q q
and using rotational transformation matrix [6, 8] Ṗq = Rr (φ)Ṗr and Fq = Rr (φ)Fr
and rearranging , we get
where
⎡ ⎤ ⎡ 1 ⎤
K 1 ẋq + K 2 ẏq − M Sφ − M1 Cφ M1 Sφ 1
M
Cφ
⎢ ⎥
f (ẋ) = ⎣ K 3 ẋq + K 4 ẏq ⎦ , g(φ) = α ⎣ M1 Cφ − M1 Sφ − M1 Cφ 1
M
Sφ ⎦ ,
K 5 φ̇q a a a a
I I I I
⎡ ⎤
1
F S(ψ − φ)
M ex
⎢ ⎥
h(Fe x, φ, ψ) = ⎣ 1
F C(ψ − φ)
M ex ⎦ , ξ(t) = [ξx ξ y ξφ ]
T
1
F (bC(ψ)
I ex
− (b − a)S(ψ))
468 M. Saad et al.
⎡ ⎤
Cφ Sφ 0
Rrq (φ) = ⎣−Sφ Cφ 0⎦
0 0 1
where C(φ) = cos(φ), S(φ) = sin(φ), C(ψ) = cos(ψ) and S(ψ) = sin(ψ)
K1 = 1
M
((2β − Bx )C 2 (φ) − (2β + B y )S 2 (φ))
K2 = K3 = 1
M
(4β − Bx + B y )CφSφ
K4 = 1
M
((2β − Bx )S 2 (φ) − (2β + B y )C 2 (φ))
K 5 = 1I (4βa 2 − Bz )
and Bx , B y and Bz are linear coefficients of friction in x-, y- and z-directions, respec-
tively, and ξ(t) is the uncertainty in the environment.
4 Controller Design
The main objective for controller design is to track the desired trajectory in the
presence of bounded but unknown external perturbations and uncertainties. OWMR
is a system of the type multiple inputs multiple outputs (MIMO). In this case, robust
nonlinear controller is appropriate rather than the standard linear controller.
Robust Sliding Mode Controller (RSMC) for an Omniwheeled Mobile Robot … 469
Sliding Mode Controller: Let Pd (t) = [xqd yqd φqd ]T is the vector of desired
trajectory. P̂ is the error vector given by P̂= Pq (t) − Pd (t). Let s be the sliding
surface defined by s = ( dtd + λ)n−1 P̂. The value of n = 2, as we are considering
till second derivative of P [3], i.e. P̈
ˆ + λ P̂(t)
s = Ṗ(t) (3)
Equation of the best estimated continuous or nominal controller (û nom ) is derived
by equating the first derivative of Eq. (3) with zero and is given by
ˆ + λ Ṗ(t)
ṡ = P̈(t) ˆ =0 (4)
ˆ
û nom = g −1 (φ)[ẍqd − f (ẋ) − h(Fex , φ, ψ) − ξ(t) − λ Ṗ(t)] (5)
ˆ − ksgn(s)] (7)
u = g −1 (φ)[ẍqd − f (ẋ) − h(Fex , φ, ψ) − ξ(t) − λ Ṗ(t)
5 Simulation Results
To verify the efficacy and accuracy of the proposed controller, trajectory tracking
was performed. The physical dimension and parameter values for the robot were
taken as mass of OWMR is M = 5 kg, moment of inertia Iq = 0.0806 kgm2 , friction
force on the wheels Bx = B y = Bz = 0.2 Ns/m, motor coefficients α = 0.087 N/V and
β = 10 kg/s,length of the robot a = 0.22 m, radius of wheel R = 0.05 m. The results
of error as deviation
from the trajectory are quantified in the form of integral square
error (ISE =
2 dt) , integral average error (IAE = |
|dt), integral time average
Error (ITAE = t|
|dt), where
is error magnitude and t is time. (Table 1). The
smaller values of ISE, IAE and ITAE for RSMC indicate that the controller has very
470 M. Saad et al.
small response time to external perturbations and disturbances. The equation of the
trajectory ∀t ∈ (0s, 190s)
3 sin(t/30) 9 sin(t/30) cos(t/30)
xqd = yqd = φqd = 0 (8)
1 + cos2 (t/30) 1 + cos2 (t/30)
and the bounded but unknown uncertainties are given as ξx (t), ξ y (t), ξz (t) ∈ (−1, 1),
Fex ∈ (−3, 3), ψ ∈ ( −π , π ) and b ∈ (0, a). The parameters of the SMC controller is
2 2
obtained
⎡ after ⎤ repetitive estimation,
⎡ ⎤ and the one giving most robust behaviors is
500 300
λ = ⎣0 5 0 ⎦ η =⎣0 4 0⎦
005 003
The initial position and orientation of the OWMR are [xn yn φn ] = [0 0 0]T .
Trajectory tracking of robot with SMC and PID is shown in Fig. 3. It was observed
that the trajectory tracking when SMC is applied on the robot, is far more accurate
than that of the trajectory tracking by PID. Figures 4 and 5 show the error in trajectory
tracking of robot in x- and y-directions with respect to time. The error in x- and y-
direction using SMC is almost zero, while error in PID is more prominent [9].
4
smc
pid
desired
3
2
1
metres
0
-1
-2
-3
-4
-4 -3 -2 -1 0 1 2 3 4
metres
error(metres)
0.1
0.05
0
-0.05
-0.1
-0.15
-0.2
-0.25
0 20 40 60 80 100 120 140 160 180 200
time(s)
In this paper, a robust sliding mode controller for OWMR has been proposed. Kine-
matic equations are developed. Dynamic equations are derived in the presence of
unknown external perturbations and uncertainties. The second-order sliding mode
control law is developed considering external disturbances and uncertainties. The
proposed controller efficacy has been tested for a reference trajectory. The simula-
tion result proves that the proposed controller excels in comparison with the PID
controller. Addition of static and dynamic boundary layer in SMC can be consid-
ered for improving the performance of SMC by trading off with accuracy. Adaptive
control law can also be developed for unbounded uncertainties.
References
1. Oliveira HP, Sousa AJ, Paulo Moreira A, Costa PJ (2008) Precise modeling of a four wheeled
omni-directional robot. Proc Robot
2. Muir PF, Neuman CP (1986) Kinematic modeling of wheeled mobile robot, Carnegie Mellon
University
3. Muir PF, Neuman CP (1990) Kinematic modeling for feedback control of an omnidirectional
wheeled mobile robot. In: Cox IJ, Wilfong GT (eds) Autonomous robots vehicles. Springer,
New York, NY
4. Edwards C, Spurgeon SK (1998) Sliding mode control theory and application, CRC Press Taylor
and Francis Group
5. Slotine J-JE, Li W (1991) Applied non linear control, Prentice Hall
6. Mittal RK, Nagrath IJ (2015) Robotics and control, Tata Mcgraw Hill Publication
472 M. Saad et al.
7. Alakshendra V, Chiddarwar S (2016) Adaptive robust control of Mecanum wheeled mobile robot
with uncertainties. Non Linear Dyn Springer
8. Muir PF, Neuman CP Kinematic modeling for feedback control of an omnidirectional wheeled
mobile robot. In: 1987 IEEE international conference on robotics and automation (ICRA)
9. Control Tutorials for MATLAB and Simulink by University of Michigan, Carnegie Mellon
University, University of Detroit Mercy
The CFD Analysis of Convection Heat
Transfer with Magnetic Field in the 2D
Domain Using OpenFOAM
Abstract The present numerical study focused on the enhancement and regulation
in the heat transfer with bifurcation in the flow along with improvement in the aver-
age temperature (T avgE ) of enclosure with the application of magnetic field. The
magnetic field-based in-house solver is developed using open-source CFD toolkit
OpenFOAM. The electric potential formulation with Boussinesq approximation is
employed in the present solver to analyze the buoyancy-driven natural convection
flow with the magnetic field. The buoyancy force is constant in the fluid by main-
taining the Rayleigh number of 106 . The impact of the force of magnetic field on
the fluid stream and heat transfer rate is reported. The magnetic field is applied in
terms of Hartmann number of Ha = 0, 10, 25, 50, 75, and 100. It is noticed that
the intensity adjusts the strength and orientation of Lorentz force in the domain and
alters the flow pattern as well as regulates the heat transfer. The detail discussion
on the impact of the magnetic field on the isotherms, streamlines, and the run-time
average Nusselt number is reported.
1 Introduction
The natural convective heat and fluid stream inside the closed cavity is one of the
important areas of the industrial application such as cooling of electronic devices,
heat dissipation in lubrication system, heat exchangers in nuclear industries, solar
energy receiver, solar desalination method, melting and solidification of molten metal
[1–4]. The fluid motion is brought in the cavity by the density variation as the conse-
quence of the temperature difference in the domain. The MHD is one of the attentive
applications in the case of natural convection flow, as it assists in regulating and sup-
pressing the flow features and heat transfer performance in the cavity even at same
Ra [5]. Thus, the application of additional magnetic force on the system retards the
buoyancy force which reduces the heat transfer [6]. The alignment of the magnetic
field on the system also plays a significant role in the convectional roll formation,
which further brings more unsteadiness in the fluid [7, 8].
In the case of unrestricted convection flow with the exposure of the magnetic field,
the above literature shows that that the presence of Lorentz force refrained the heat
transfer and the flow of fluid in the domain. However, the present study reports the
favorable role of the existence of Lorentz force for boosting the heat transfer and
regulating the fluid flow in the cavity at low strength of magnetic field. In this numer-
ical analysis, in-house developed code on the open-source CFD toolkit OpenFOAM
is used for MHD-based buoyancy driven. The detail discussion of relevant quanti-
ties such as run-time average Nusselt number, local Nusselt number, streamlines,
isotherms, the average temperature in the enclosure (T avgE ), Lorentz force strength
and its direction in the domain is reported.
The present solver couples the Navier–Stokes equation with Poisson’s electric poten-
tial equation and energy equation to obtain the velocity distribution, temperature vari-
ation, and Lorentz force in the domain. The Lorentz force and Boussinesq approxi-
mation are added in the Navier–Stokes equation as a source term to show its influence
on the flow physiognomies. The fluid is supposed to be incompressible, viscous, and
electrically conducting in nature. The complete set of the partial differential equation
is taken from reference [6].
The non-dimensional parameters incorporated in the solver that govern the fluid
flow under the influence of the magnetic field are:
σ
Ha = B(2L) Ra = Gr · Pr
μ
gβ(T − Tref )(2L)3 ν
Gr = Pr =
ν2 α
where Ha is the Hartmann number, which measures the power of the magnetic field
imposed on the system. Ra is the Rayleigh number, Gr is the Grashof number, and
Pr is the Prandtl number. The characteristic length (2L) is considered as a bottom
length of the cavity. Direct numerical simulation (DNS) is applied in OpenFOAM
tool to solve the complete sets of Eqs. (1–6) [6]. Collocated finite volume method is
used to solve all terms in the equations. The PIMPLE algorithm is set in the solver to
solve the entire sets of the equation with two outer correctors and pressure correctors
which assist in refining convergence of velocity and pressure field at every interval.
The CFD Analysis of Convection Heat Transfer … 475
The Rhie and Chow interpolation scheme [9] is used to couple the pressure and
velocity. The first-order accurate implicit Euler scheme is supposed to discretize the
time derivate terms. The second-order central difference method is implemented to
discretize the convection and diffusion terms of the governing equations.
The gradually constricted cavity from the top and the flat bottom surface is used
as a computational domain. The characteristic length is taken as the base length of
the surface (2L), and the opposite vertical adiabatic surface is of length (L). The
geometry details, coordinates system, and mesh distribution in the present study are
shown in Fig. 1. All computations are performed at a constant Rayleigh number of
Ra = 106 and at Pr = 0.71 [10]. The grid independence study is performed on the
three different sizes, Grid 1 = 100 × 200, Grid 2 = 125 × 250, and Grid 3 = 150 ×
300, at Ra = 106 and Ha = 50. The average temperature of the enclosure (T avgE ) is
calculated for all three grids as shown in Fig. 2b. The T avgE variation for all grids
overlaps with each other; therefore, Grid 2 is fine enough to capture the flow physics
near the boundary layers and in the central core region. Hence, Grid 2 is adopted for
the entire simulation in the present study.
The validation test case is considered for two-dimensional MHD-based natural con-
vection flow in the rectangular domain and obtained data is compared with reference
[6, 11]. The geometrical size, aspect ratio, and boundary conditions are maintained
as mentioned in the given reference [6, 11]. The streamwise normalized velocity
(U x /U x(max) ) obtained in the present case is compared with the result mentioned in
the reference [6, 11] as shown in Fig. 2a, and the comparison trend shows a good
→ → →
agreement with published data [6, 11]. The Lorentz force ( F = j × B ) is the con-
sequence of the interaction among the working fluids with the magnetic field. Due
Fig. 2 a Validation of the result obtained from present solver with reference [6, 11], b the grid
independence test
to the heated bottom surface, fluid movement triggered and formation of convection
rolls in the domain are observed.
Now, the fluid interacts with the magnetic field (B) to produce electric potential
and subsequently electric current. In the two-dimensional computational domain,
fluid flows in x-direction (U x ) and y-direction (U y ). The imposed magnetic field
→ →
is in x (Bx ). Hence, the electric potential (∂φ ∂z ≈ U × B ) and electric current
→
( j ) are developed in the z-direction [6–8]. However, the present analysis is two-
z
dimensional; the distribution of an electric current and an electric potential variation
in the cavity is not available in this study. The orientation of Lorentz force is induced
in the reverse direction of the flow of the fluids, which affects the flow pattern and
hence heat transfer. Figure 3 shows the streamlines variations in the enclosure for the
different intensity of the magnetic field applied in the direction parallel (Bx ) to the
heated surface. The streamlines for non-MHD free convection flow (Ha = 0) in the
enclosure has two convection rolls and two primary vortices at the top and bottom
parts of the cavity as shown in Fig. 3a. The sharp constriction of the top surface of
the enclosure is responsible for the flow separation in the top and lowest curves of
the cavity. Hence, the fluid gets cooled in the mid-span of the top surface, and after
that, it starts gravitating leaving eddies behind at top corners of cavity. Similarly, at
the bottom corners of enclosure, eddies are generated for non-MHD flow (Ha = 0) as
shown in Fig. 3a. The vortices at the bottom corners of the cavity are first eliminated
with the application of magnetic field at Ha = 10. Although, the strength of Lorentz
force is not sufficient to cop with the primary vortices at the top corners as shown
in Fig. 3b–d. As the intensity of the magnetic field (Ha = 50) increased, secondary
vortices near the top surfaces start emerging; this is due to the dissimilar orientation
of the Lorentz force with respect to the heated surface. The dissimilar variation of
Lorentz force suppresses the fluid toward the bottom surface and pulls the fluids
toward the top corners. The increase in the intensity of the magnetic field (Ha = 75
and 100), two convectional rolls are formed and occupy the entire domain without
any traces of smaller vortices as shown in Fig. 3e, f.
Figure 4 shows the isotherms variation in the enclosure. It is observed from Fig. 4a
for non-MHD flow; the plume is developed from the center of the cavity and bifur-
cated into two equal halves from the top converged point. Consequently, for Ha =
0, the interaction area for the cooling is lesser as the fluid becomes denser and starts
gravitating leaving the cold surface without occupying the entire upward slope. How-
ever, the plume formation is shifted near the adiabatic walls as soon as the magnetic
field is imposed on the system (Fig. 4), which is observed from the streamline plots
(Fig. 3). The shift in the position of plume generation increases the heat dissipation
area. When the hot fluid started following the cold surface and gravitates along with
the downward slope of the wall, it increases the contact area for cooling as shown in
Fig. 4b–f. The furthermore increase in the magnetic strength (Ha = 50 − 100), the
fluid is occupying the entire cavity into two equal halves without any traces of eddies
as observed in the streamlines plots (Fig. 3e, f). Hence, the average heat dissipation
area of fluid gets increased. However, the flow velocity of the fluid in the domain
478 R. J. Singh and T. B. Gohil
is highly suppressed by the presence of the magnetic field; hence, the rate of heat
absorption and dissipation gets retarded at the higher Ha.
Figure 5 illustrates the run-time average Nusselt number (N avg ) variation over
the hot surface. It has been observed that the Nuavg upsurge at Ha = 10 compared
to Ha = 0 (Fig. 5a); this is because, at the small intensity of Ha, the Lorentz force
assists the fluid to dissipate heat in the entire area of cavity as the buoyancy force is
dominant over viscous force and Lorentz force. The flow of fluid becomes unsteady
at Ha = 25 (Fig. 5b); this is happening because the Lorentz force starts dominating
over buoyancy force and it tries to pull the fluid according to its nature of flow and
direction. The unsteady behavior at Ha = 25 keeps the Nuavg same as to Ha = 0.
Further, with the increase in the Bx magnetic field, Nuavg begins to drop down as the
strength of the induced Lorentz force in the Bx magnetic field increases with the rise
in the Ha and has direct opposing nature to buoyancy force which suppress the flow
motion and heat transfer as shown in the Fig. 5c. For these cases, all reported results
are time-averaged.
5 Conclusions
References
1. Chiang H, Kleinstreuer C (1991) Analysis of passive cooling in a vertical finite channel using a
falling liquid-film and buoyancy-induced gas vapour flow. Int J Heat Mass Transfer 34:2389–
2394
2. Payvar P (1991) Laminar heat transfer in the oil groove of a wet clutch. Int J Heat Mass Transfer
34:1791–1798
3. Dayem AMA (2006) Experimental and numerical performance of a multi-effect condensation–
evaporation solar water distillation system. Energy 31:2710–2727
4. Das D, Roy M, Basak T (2017) Studies on natural convection within enclosures of various
(non-square) shapes—a review. Int J Heat Mass Transfer 106:356–406
5. Gajbhiye NL, Eswaran V (2015) Numerical simulation of MHD flow and heat transfer in a
rectangular and smoothly constricted enclosure. Int J Heat Mass Transfer 83:441–449
6. Singh RJ, Gohil TB (2019) The numerical analysis on the development of Lorentz force
and its directional effect on the suppression of buoyancy-driven flow and heat transfer using
OpenFOAM. Comput Fluids 179:476–489
7. Singh RJ, Gohil TB (2019) Influence of the presence of the lorentz force and its direction on the
suppression of secondary flow in two different orifices: a numerical study using OpenFOAM.
J Appl Mech 12(3):751–762
8. Singh RJ, Gohil TB (2019) The numerical analysis on the variation of electric potential, electric
current and Lorentz force with its influence on buoyancy-driven conjugate heat transfer and
fluid flow using OpenFOAM. Fusion Eng Des 148:111300
9. Ferziger JH, Peric M (2002) Computational methods for fluid dynamics, 3rd edn. Springer,
Berlin, pp 247–251
10. Chatterjee D, Mondal B, Halder P (2014) Mixed convective transport in a vertical lid-driven
cavity including a heat conducting rotating circular cylinder. Numer Heat Transfer Part A
65:48–65
11. Garandet JP, Alboussiere T, Moreau R (1992) Buoyancy driven convection in a rectangular
enclosure with a transverse magnetic field. Int J Heat Mass Transfer 35:741–748
Design of a Remote Racking Module
for Racking Operation
Abstract Circuit breaker is a device used to disrupt the flow of current when an
abnormal condition occurs in switchgear. Withdrawal-type circuit breakers have the
provision for displacing the circuit breaker in and out of the switchgear compartment
while inspection and maintenance operations. This paper discusses a new mechanism
for displacing the circuit breaker during racking operation by stationing the operator
at a remote location. The first part deals with the design of lead screw and variation of
various thread parameters on the stress distribution and performance. The later part of
the paper deals with the development and analysis of the novel racking mechanism.
1 Introduction
Locations such as industries, offshore oil platform, refineries, and hospitals employ
multiple circuit breakers, typically located within a circuit breaker cabinet. The instal-
lation and removal of circuit breaker involves the engagement and disengagement of
contacts of the contact breaker with a power bus within the circuit breaker cabinet.
Conventionally, this installation and removal is performed manually by a technician.
One of the main issues faced by the operator working in the electrical industry is
the sudden explosion of high energy arc due to proper methods of engaging of the
breaker, handling of tools, etc. During the service and inspection period, these circuit
breakers are often dislodged from the electrical system. The manual racking action
necessitates that the operator stands directly in front of the breaker. As the breaker
is methodically extracted from the bus, several potential problems may occur.
This creates a vulnerable surrounding to the operator engaging with the breaker.
As a safety precaution, the electrician often wears equipment to protect oneself from
fatal injuries while working with the electrical breaker. This heavy equipment often
reduces the dexterity and hinders the movement of the operator. The personnel are
advised to keep a minimum distance with working with the breaker. The most easy
and feasible solution is to keep the worker out of the flash boundary around the
breaker by maintaining a safe distance. These devices pose advantages over manual
operation like protects operator, creates a safe workplace around the breaker, reduces
errors due to manual racking, and preserves integrity of interlocks, and self-stopping
system helps to stop abnormal conditions. This apparatus allows the insertion and
withdrawal of electric components while the operator is outside the safe boundary.
They are designed specifically to remove the personnel from proximity to the circuit
breaker in operation. The device provides a portable, motorized device that can be
hand-carried to the work location and affixed to the circuit breaker without the need
for any modification to the breaker or its enclosure.
Hawkins et al. [1] designed a remote racking module along with a structural
member for mounting. Bower et al. [2] and Vaill et al. [3] developed a mechanism
that interfaces with the interlock system in switchgear as well as assembly for fixture
thereby facilitating displacing of circuit breaker. Varanasi et al. [4] designed a model
for a system consisting of a lead screw taking into the inertia as well as damping
effects.
Withdrawal-type circuit breakers for medium voltage application usually are placed
over a withdrawal part. The circuit breaker moves with the help of a lead screw
inside the withdrawal part. The vital element in transmission of power from the
remote racking device to the breaker for displacement is the lead screw. This session
discusses the lead screw design and the effect of thread parameter on the stress and
deformation [5]. As per the various forces acting on the lead screw, the lead screw
is designed as shown in Table 1.
The various thread parameters like thread pitch, nominal diameter, etc., play an
important role in power transmission. The following section discusses the influence
of thread pitch on the stress distribution on the screw and the frequency of the system.
The study is done by considering square thread profile.
Fig. 1 Effect of thread pitch of square thread profile on stress distribution and deformation
Fig. 2 Stress and displacement plot of square profile lead screw of pitch 3.4 mm at position of nut
102 mm from front end
The stress distribution is evaluated for the lead screw for different thread profiles
such as square, acme, and trapezoidal with pitch ranging from 2.5 to 5.5 mm. The
influence of pitch variation on stress values at various positions of the nut during
its travel on the square profile lead screw is shown in Fig. 1. The stress plot and
deformation plot for the square profile lead screw of pitch 3.5 mm are shown in
Fig. 2. It is observed that stress in the lead screw decreases as the position of nut
moves toward the center. Later, there is a slight increase in the stress value as the
nut further travels to the other end. Another observation is that the variation of stress
values is less for all pitch when the nut is at the center of the lead screw.
The effect of thread pitch on the system’s frequency is studied by considering the
square thread profile of pitch varying from 2.5 to 6.5 mm. It is observed that the
484 A. S. Syriac and M. R. Rahul
a b
Fig. 3 Dynamics analysis of square profile lead screw of pitch 3.5 mm a Mode 5 and b Mode 6
system’s frequency is not affected much by the change in thread pitch. The trend
seen in all the different pitch considered is that there is a small drop in frequency as
we increase pitch. Further, the value of frequency is seen to increase as we move to
pitch 6.5 mm. The mode shape of the lead screw is shown in Fig. 3.
A new system has been developed to be used for rack-in and rack-out of the cir-
cuit breaker. The breaker placed inside a compartment should allow the racking-in
and racking-out movements behind a closed door. This new racking truck brings
many improvements to existing functions and integrates new features linked with
the position of the circuit breaker. The main components of the mechanism are con-
fined inside a box. This novel racking module brings many improvements to features
linked to the present racking solutions. The basic system consists of a mechanism
for mounting, a device to actuate racking and controller.
The mounting mechanism is designed to fasten the racking device onto the panel
of the breaker. This mechanism is developed in a way that no modifications are to
be made on the front panel of the breaker to accept the device. The structure of the
mounting mechanism is robust and shaped for easy handling by the user as shown in
Fig. 4. The device can be engaged and dislodged by magnetic means. The provision
for angular rotation of the device helps for precise mounting of the device in front
of the breaker door.
The main component of the system is the racking device which helps to actuate
the racking screw of the withdrawal part of the circuit breaker for maneuvering of
the circuit breaker forward and backward. The entire components are housed inside a
box. The box is attached to the mounting mechanism for the operation. The rotation
of the racking screw is done using the selected motor with the required torque. The
exact torque for rotation is calculated to displace the breaker between the connected
and disconnected position. The torque value selected must help in maintaining the
integrity of the interlocks in the withdrawal device as well as overcome the weight
of the breaker along with friction. Various safety devices are present which help to
sense the overtorque condition.
To keep the operator out of the arc flash boundary, the device is designed to
control from outside the flash boundary by the means of a remote console module.
The controller contains buttons to configure the racking device in the racking-in and
raking-out operation. In case of any abnormal condition senses by the system, emer-
gency button is present to disable the racking operation immediately. The proposed
mechanism is shown in Fig. 5.
To test the performance of the device for various stress induced in the system,
vertical deformation due to the weight of the device, analysis of the device is carried
out in ANSYS Workbench. The boundary conditions for the study are shown in the
figure. The static analysis boundary conidtion is shown in Fig. 6a. Modal analysis
of the remote racking device is carried out to study the various types of vibration
induced in the system which operates as shown in Fig. 6b.
4 Conclusion
This paper discusses a novel racking mechanism that allows the medium voltage
circuit breaker to be racked in or out by using a remote racking device that can
be operated from a remote location. This proposed method of racking stations the
operator outside the arc flash boundary thereby protecting the operator from fatal
486 A. S. Syriac and M. R. Rahul
Fig. 6 a Racking mechanism device with boundary condition and b modal analysis of the device
depicting Mode 1 (bending about Z-axis)
injuries. The mechanism consists of a motor to rotate the racking screw for displacing
the breaker, unit for gripping the remote racking device as well to operate the device
from a remote location.
References
1. Hawkins et al (2011) Circuit breaker racking apparatus, systems, and methods of using same.
US patent 8, 553, 394, B2, 17 Mar 2011
2. Bower et al (2012) Remotely—operated switchgear racking device and a mounting method for
the same. US patent 2012/0199450 A1, 8 Feb 2012
3. Vaill et al (2005) Racking device and power module therefor. US patent 7,019,230 B1, 3 Aug
2005
4. Varanasi KV, Nayfeh SA (2004) The dynamics of lead-screw drives: low-order modeling and
experiments. J Dyn Syst Meas Control 126:388–396
5. Dragoni E (1994) Effect of thread pitch and frictional coefficient on the stress concentration in
metric nut-bolt connections. J Offshore Mech Arct Eng 11:21–27
A Coupled Heat Transfer and Artificial
Neural Network Based Model for
Accelerated Direct Cooling of Steel Plate
Abstract This article proposes a novel approach for controlling the accelerated cool-
ing of hot steel plates in the existing plate mill at AM/NS India, Hazira. Difference
in top and bottom surface temperature or uneven temperature distribution at the exit
of accelerated cooling equipment leads to the property variation and shape deforma-
tion. Therefore, the model aims to predict top and bottom surface temperature of the
plate for better online operational control. The control algorithm is a fundamental
heat transfer calculation coupled with statistical artificial neural networks (ANN).
The model prediction shows an excellent agreement with the plant measured data
and almost 90% of the plates are within error range of ±20 ◦ C.
Keywords Accelerated cooling · Mist spray cooling · Plate mill · Heat transfer
calculation · Artificial neural network
1 Introduction
The schematic diagram of ADCO machine at AM/NS India Hazira facility is shown
in Fig. 1. It contains 6 modules of mist spray cooling arrangement (module 1–6).
Each module has 4 top and 4 bottom air-assisted slit-type sprayers. Total length of
each module is 4 m in which the sprayers are placed at equidistance of 1 m. Width
of the sprayers is 5.1 m. Maximum water flow rate through a module is limited to
1000 m3 /h which is controlled by electro-pneumatic valves. The operating water
pressure is 1.5 bar for module 1–6. Air pressure and flow in each module are set
constant as 780 mm of water column and 55,000 ∗ 2 m3 /h, respectively. Apart from
that one additional module (module 0) with high-pressure spray nozzles is installed
for intense cooling of the plate. The maximum water flow rate through module 0 is
the same as module 1–6. However, operating pressure is set as 7 bar for air-water
mist spray.
Three scan pyrometers are installed for plate surface temperature measurements
before and after ADCO machines. The first pyrometer (P1) measures top surface
temperature of the plate at ADCO entry. At ADCO exit, the second (P2) and the third
(P3) pyrometers are positioned to measure the top and bottom surface temperature of
the plate, respectively. The plate temperature measurement inside the cooling system
is very difficult due to radiation effect of high mass flux density mist spray and steam
generation over the plate.
Earlier modeling approach was to combine heat transfer calculation with the phase
transformation model of steel [3, 4, 8]. However, for accurate prediction of phase
fraction, the detailed characterization of each steel grade is required at the rolling
temperature. Hence, instead of pure fundamental model, an hybrid model of the
basic heat transfer calculation and artificial neural networks is proposed in the present
study. The first principle heat transfer calculation without incorporating metallurgical
phase transformation model forms a base of the model prediction, while ANNs help in
compensating the error in basic heat transfer calculation and online implementation.
For fast prediction, heat conduction in length and width direction is assumed much
smaller than the thickness direction. Therefore, unsteady one-dimensional heat dif-
fusion equation with the moving boundary condition is used to predict the top and
bottom surface temperature of the plate.
∂T ∂ ∂T
ρC p = k + Q̇ (1)
∂t ∂z ∂z
where T is the temperature (K), ρ is the density (kg/m3 ), C p is the specific heat (J/kg
K) and k is the thermal conductivity of the plate (W/m K). Q̇ is the heat source due
phase transformation (kJ/kg). It is assumed that density of the plate depends only on
the chemistry, and it remains constant. The variation of thermal conductivity with
respect to temperature is illustrated in Eq. (2) [2].
The specific heat of the plate varies significantly with the change in plate chem-
istry, temperature and phase transformation. Since the phase transformation model
is not incorporated in the heat transfer calculation (neglecting the source term Q̇),
transformed phase fraction (α in Eq. (3)) is assumed to follow sigmoidal curve from
Ar3 to Ar1 (critical temperatures) (cf. Fig. 2a). C p values of ferrite, pearlite and bai-
nite are assumed to be same. Figure 2b also shows the variation of specific heat for
low-carbon steel in working temperature range [8].
490 S. Dave et al.
1200 1400
Austenite M0 Top
1 0.75
Transformed Phase Fraction (α)
Ferrite, Pearlite, Bainite M0 Bottom 162*G
1200 M1 to M6 Top
Cp Value (J/kg K)
800
0.5 800 120*G0.7
600 250*G0.3
400
600 140*G0.42
200
0
Ar Ar 400 0
1 3 300 400 500 600 700 800 900 0 2 4 6 8 10 12 14 16
Fig. 2 Heat transfer calculation parameters a assumption for transformed phase fraction (for C p
value), b specific heat variation for low-carbon steel [8], c heat flux profile for different modules
% Plates
800 10
750
700 5
650
x is compensated by ANN x 0
600
<-60
-60--55
-55--50
-50--45
-45--40
-40--35
-35--30
-30--25
-25--20
-20--15
-15--10
-10--5
-5-0
0-5
5-10
10-15
15-20
20-25
25-30
30-35
35-40
40-45
45-50
50-55
55-60
>60
550
0 1 2 3 4 5 6 7 8
Time (sec) Error in Temperature Prediction (°C)
(a) (b)
Fig. 3 Results of heat transfer calculation a transient calculation for R O T configuration [8], b
error in temperature prediction
First, the developed model has been validated with the results of Sun et al. [8]
for their R O T configuration (cf. Fig. 3a). In present ADC O setup, the model is
dedicated to predict top finish cooling temperature (TFCT) and bottom finish cooling
temperature (BFCT) at the ADC O exit. Total 5000 plates data which include broad
range of thicknesses (10–60 mm) and more than 60 grades of the steel plate (0.05–
0.24 %C), are used to train the model. Results of unsteady heat transfer calculation
are summarized in Fig. 3b.
Temperature prediction error shown in Fig. 3b is closely related with the variation
in plate chemistry. To compensate this error due to the phase transformation source
term, ANN-based model (ANN1) is coupled with the present heat transfer calculation.
Multilayer perceptron (MLP) networks with three layers (input-output and one hid-
den layer) are trained to develop the coupled model by applying the second-order
Broyden–Fletcher–Goldfarb–Shanno (BFGS) algorithm. Different combinations of
nonlinear activation functions (like logistic, tangent hyperbolic, exponential, sine,
etc.) and node weights are tuned during the learning process in order to improve the
output prediction. The input parameters used for training and testing of the network
(ANN1) are the plate chemistry (%C, %Mn, %Mo, %Nb), plate speed, length/width
ratio and plate entry temperature. The prediction statistics of the coupled model are
detailed in Fig. 4.
A comparison of predicted top and bottom surface temperature with the pyrom-
eters reading at ADC O exit is illustrated in Fig. 4a. The prediction error in TFCT
and BFCT is less than ±20 ◦ C for 91.56% and 95.42% of the plates, respectively (cf.
Fig. 4b). It shows that after coupling ANN with the basic heat transfer calculation,
the results are in excellent agreement with the plant measurements.
492 S. Dave et al.
30
650 20
% Plates
15
600
10
5
550
0
<-50
-50--45
-45--40
-40--35
-35--30
-30--25
-25--20
-20--15
-15--10
-10--5
-5-0
0-5
5-10
10-15
15-20
20-25
25-30
30-35
35-40
40-45
45-50
>50
500
500 550 600 650 700
Measured Cooling Temperature (°C) Error in Temperature Prediction (°C)
Two more networks were trained for the online implementation of the model. ADC O
entry temperature is predicted with the help of ANN2 as the model has to run before
actual cooling starts for its dynamic performance. Apart from that there is inbuilt
dynamic water control system to modify water flow rate for head, tail and body of the
plate, and it is very difficult to use instantaneous flow for the heat transfer calculation.
Therefore, ANN3 is introduced to predict averaged water flow rate per module using
head and tail set points which are fixed by the operator. Input parameters for ANN2
and ANN3 are as follow: (1) ANN2: plate chemistry (%C, %Mn, %Mo, %Nb, %Ti),
target thickness and finish rolling temperature (2) ANN3: plate speed, head flow rate,
tail flow rate and module index.
20
Deviation (°C)
-20
-40
-60
520 550 580 610 640 670 700
Measured Pyrometer Reading (°C)
A Coupled Heat Transfer and Artificial Neural Network … 493
After successful installation, the model has been tested for 300 plates for online
operation. Figure 5 shows the deviation in TFCT and BFCT prediction from the
measured pyrometer readings which is inline with the model claim.
5 Conclusion
A hybrid model comprising fundamental heat transfer calculation and statistical arti-
ficial neural network has been developed and tested for the top and bottom surface
temperature prediction of hot steel plate after accelerated cooling. It was found that
combining the first principle heat transfer model with ANN model for estimating
error due to variation in chemical composition of the steel has given better results.
Prediction statistics of the coupled model indicates an excellent agreement with
the plant measured data. The TFCT and BFCT prediction error fall within band of
±20 ◦ C for almost 91% and 95% of the plates respectively. The model online predic-
tion after plant scale implementation is inline with the model claim which indicates
the applicability of this type of hybrid and computationally efficient model in the
steel industries.
References
1. Al-Ahmadi H, Yao S (2008) Spray cooling of high temperature metals using high mass flux
industrial nozzles. Exp Heat Transf 21(1):38–54
2. Darken L, Gurry R (1953) Physical chemistry of metals, vol 582. McGraw-Hill book company,
Inc., New York, Toronto, London
3. Edalatpour S, Saboonchi A, Hassanpour S (2011) Effect of phase transformation latent heat on
prediction accuracy of strip laminar cooling. J Mater Process Technol 211(11):1776–1782
4. Han HN, Lee JK, Kim HJ, Jin YS (2002) A model for deformation, temperature and phase
transformation behavior of steels on run-out table in hot strip mill. J Mater Process Technol
128(1–3):216–225
5. Hnizdil M, Chabicovsky M, Raudensky M, Lee TW (2016) Heat transfer during spray cooling
of flat surfaces with water at large reynolds numbers. J Flow Control Measur Vis 4(03):104
6. Lee P, Raudensky M, Horsky J (2013) Development of accelerated cooling for new plate mill.
Ironmaking Steelmaking 40(8):598–604
7. Liang G, Mudawar I (2017) Review of spray cooling-part 2: high temperature boiling regimes
and quenching applications. Int J Heat Mass Transf 115:1206–1222
8. Sun C, Han H, Lee J, Jin Y, Hwang S (2002) A finite element model for the prediction of thermal
and metallurgical behavior of strip on run-out-table in hot rolling. ISIJ Int 42(4):392–400
Effect of Air Distribution on Cooling
of Photovoltaic Panel and Its
Performance
1 Introduction
Solar energy is one of the most abundantly available resources on earth which can be
used in various applications such as power generation systems, heating and cooling
system, and photovoltaic cell systems. A solar cell device is used to convert irradiance
into electricity through the photovoltaic effect. The solar cells convert only 15–20%
of incoming irradiance into useful electrical energy and rest of it is converted into
heat which increases the surface temperature of the PV panel [1]. For 1 °C rise in
temperature of PV cells above the working limit reduces conversion efficiency by
0.4–0.65% [2]. Thus, it becomes necessary to keep the temperature of the PV panel
close to the standard operating temperature. Several cooling techniques have been
investigated to reduce the temperature of the PV panel such as water cooling, phase
change material, heat pipe cooling, and refrigerant cooling, etc. [2–5]. The non-
uniform temperature on the surface of the PV panel produces thermal stresses (hot
spots). Thus, uniform cooling of the PV panel is required to reduce thermal stresses
which lead to an increase in the life of PV panel [6]. Tonui and Tripanagnostopoulos
et al. [7] studied the performance of PV/T system using thin flat metal sheet suspended
at the back wall of an air channel. The experimental data and result of theoretical
model are in good agreement. The suggested modification improves electrical and
thermal efficiencies. This system is tested with and without glass cover in natural
and forced convection [8]. One of the important conclusions of above study is that
setting glass cover on PV panel increases thermal efficiency but decreases electrical
efficiency. Popovici et al. [9] presented a numerical model to reduce the temperature
of PV module at different configuration of heat sinks by changing angles between fins
and base plate. Improving efficiency of PV panel has been proposed using aluminum
plate heat sink under natural convection and forced convection [10, 11]. Nabil et al.
[12] presented a novel geothermal air cooling system coupled with a photovoltaic
system for improving the efficiency of the PV panel.
In the current study, the effect of a newly designed diffuser with a photovoltaic
panel on the uniformity of air distribution was investigated at a different flow rate.
Also, the electrical efficiency and power output of a PV panel with a varied flow rate
were compared with the reference module.
Fig. 1 Computational
domain of diffuser
Effect of Air Distribution on Cooling of Photovoltaic Panel … 497
Table 1 Dimensions of
Parameters Dimensions
computational domain
Panel dimension 0.655 × 0.455 m
Height of diffuser 0.2275 m
Radii of sidewalls 0.275 m
Diffuser outlet 0.455 m
Number of air deflectors 3, 4
to solve the given numerical fluid flow problem in ANSYS Fluent 16. A pressure-
based solver and ambient temperature of 27 °C (300 K) chosen for a case to obtain
simulation results.
For uniform distribution of air, a diffuser with three and four deflector plates with
curved sidewalls was validated [13]. A diffuser with three deflector plates and curved
sidewalls shows a more uniform distribution of air. Figure 3a, b shows the effect of
three and four deflector plates on the distribution of air, respectively.
Figure 4 shows the magnitude of velocity at the exit of a diffuser with three
deflector plates.
Fig. 3 a Effect of three deflector plates on the distribution of air, b effect of four deflector plates
on the distribution of air
498 S. S. Bhakre and P. D. Sawarkar
Fig. 4 Comparison of
velocity magnitude results at
the exit of the diffuser (three
deflectors)
3 Experimental Setup
The diffuser with three deflector plates with curved sidewalls was fabricated for uni-
form distribution of air as per the dimension given in Table 1. The experiments
were performed to investigate the effect of diffuser on the performance of the
PV panel. All experiments were performed at VNIT Nagpur, India (21.1458° N,
79.0882° E) in April 2019 under sunny weather. A schematic and photograph of
experimental setup were shown in Figures 5 and 6, respectively.
Fig. 5 Schematic of
experimental setup
Effect of Air Distribution on Cooling of Photovoltaic Panel … 499
Fig. 6 Photograph of
experimental setup
The maximum power and electrical efficiency of the photovoltaic (PV) panels are
calculated using Eqs. (1) and (2), respectively.
where
I solar radiation (W/m2 ),
Pmax maximum power (W),
V max maximum voltage (Volt),
I max maximum current (Ampere)
Ac C/s area of PV panel m2
500 S. S. Bhakre and P. D. Sawarkar
The temperature of the PV panel was measured using k-type thermocouple for
every 30 min. Irradiation was measured using solar meter. The rheostat was adjusted
to obtain maximum voltage and maximum current. The current and voltage are
measured using a multimeter.
Fig. 10 Variation of
electrical efficiency of PV
panel versus time of the day
5 Conclusions
For uniform distribution of air, different geometries of the diffuser are simulated.
The result of simulation shows that among all the geometries, diffuser with curved
sidewalls and three deflector plates is the best geometry to be considered for uniform
distribution of air velocity. The performance of the PV panel with diffuser increases
with an increase in air velocity. The average electrical efficiency of PV panel is
increased by 15.33% and power output is increased by 1.87 W at a flow rate of
2.5 m/sec.
502 S. S. Bhakre and P. D. Sawarkar
References
1. Rahman MM (2015) Effects of various parameters on PV-module power and efficiency. Energy
Convers Manag 103:348–358. https://doi.org/10.1016/j.enconman.2015.06.067
2. Stropnik R, Stritih U (2016) Increasing the efficiency of PV panel with the use of PCM. Renew
Energy 97:671–679. https://doi.org/10.1016/j.renene.2016.06.011
3. Moharram KA, Abd-Elhady MS (2013) Enhancing the performance of photovoltaic panels by
water cooling. Ain Shams Eng J 4:869–877. https://doi.org/10.1016/j.asej.2013.03.005
4. Anderson WG (2008) Heat pipe cooling of concentrating photovoltaic cells. IEEE 7(8):978–
985. https://doi.org/10.1109/PVSC.2008.4922577
5. Sajjad U, Amer M (2019) Cost effective cooling of photovoltaic modules to improve efficiency.
Case Stud Therm Eng 14:100420. https://doi.org/10.1016/j.csite.2019.100420
6. Baig H, Heasman KC, Mallick TK (2012) Non-uniform illumination in concentrating solar
cells. Renew Sustain Energy Rev 16(8):890–909. https://doi.org/10.1016/j.rser.2012.06.020
7. Tonui JK, Tripanagnostopoulos Y (2008) Performance improvement of PV/T solar collectors
with natural air flow operation. Sol Energy 82:1–12. https://doi.org/10.1016/j.solener.2007.
06.004
8. Shahsavar A, Ameri M (2010) Experimental investigation and modeling of a direct-coupled
PV/T air collector. Sol Energy 84:1938–1958. https://doi.org/10.1016/j.solener.2010.07.010
9. Popovic G, Sebastian VH (2016) Efficiency improvement of photovoltaic panels by using air
cooled heat sinks. Energy Procedia 85:425–432. https://doi.org/10.1016/j.egypro.2015.12.223
10. Mays A (2017) Improving photovoltaic panel using finned plate of aluminum. Energy Procedia
119:812–817. https://doi.org/10.1016/j.egypro.2017.07.103
11. Soliman MN (2019) An experimental study of the performance of the solar cell with heat sink
cooling system. Energy Procedia 162:127–135. https://doi.org/10.1016/j.egypro.2019.04.014
12. Elminshawy N (2019) Performance of PV panel coupled with geothermal air cooling system
subjected to hot climatic. Appl Therm Eng 148:1–9. https://doi.org/10.1016/j.applthermaleng.
2018.11.027
13. Amanlou Y, Hashjin TT (2018) Air cooling low concentrated photovoltaic/thermal (LCPV/T)
solar collector to approach uniform temperature distribution on the PV plate. Appl Therm Eng
141:413–421. https://doi.org/10.1016/j.applthermaleng.2018.05.070
Numerical Investigations of Photovoltaic
Phase Change Materials System
with Different Inclination Angles
Abstract Excess unused solar radiation falling on the PV system rises its temper-
ature and drops the electrical conversion efficiency. Phase change materials can be
effectively used for the PV thermal management systems due to its high latent heat.
In the current research work, effect of inclination angle on the thermal performance
of PV PCM system is investigated. The inclination angle is increased from 15° to 90°
with constant PCM thickness of 30 mm at 1000 W/m2 of incoming solar radiations.
It is observed that the PCMs can be effectively used for the PV thermal manage-
ment systems. The increase in inclination angle of PV PCM system reduces the time
required for melting PCM and increases the PV surface temperature.
1 Introduction
The natural resources for fulfilling ever increasing demand for heat and electricity
across the globe are on the verge of extinction. Solar energy has been considered
as one of the prominent alternatives for natural resources due to its abundant avail-
ability, free of cost and probability to efficiently replace the use of natural fuels for
the production of heat or electricity. Solar energy from the sun can be effectively
converted into electricity with photovoltaic (PV) solar cells. PV solar cell is a very
mature technology now; however, it has a serious issue of decreasing its efficiency
with the increase in its surface temperature. Solar PV panels generally convert 16–
20% of the incoming solar radiation into electricity and most of the excess unused
solar radiation causes to increases the temperature of PV cells; which affects its per-
formance [1]. The solar PV cell efficiency is measured at the standard test conditions
of 25 °C, AM1.5 and the incoming solar radiations of 1000 W/m2 . The increase of
1 °C in the surface temperature of PV solar cells above 25 °C causes to decrease the
efficiency of PV solar cells around 0.45–0.6% and thus, PV thermal management
systems are considered to be one of major challenge across the globe especially, in
the country like India with very hot climate in most of the time around the year.
During the last 2–3 decades various active and passive thermal management sys-
tems have been investigated by various researchers; though, the best possible thermal
regulation systems are still not available. The PCMs may be organic or inorganic and
have a common tendency to absorb heat during charging and releasing the absorbed
heat during the discharging period [2]. Phase change materials (PCMs) have been
used as thermal energy storage systems in the past but, due to its property of high
latent heat and maintaining a constant temperature during phase change; it can be
efficiently combined with the PV systems [3]. Choubineh et al. [4] performed exper-
imental investigations of PV systems with salt hydrates as a PCM for the applica-
tions in the building storage systems. They found a 3.7–4.3 °C reduction in the PV
panel surface temperature. Karthick et al. [5] experimentally investigated the effect
of inorganic Glauber salt as a PCM on the performance of building-integrated PV
PCM system. The results showed a 12% reduction in PV cell temperature compared
to the conventional PV system. The CO2 mitigation over the lifetime of the devel-
oped PV PCM system was observed to be 1.74 tCO2 . Su et al. [6] performed on-site
experimental analysis of concentrated PV systems. The performance of the PV water
cooling and PV PCM system was compared. They concluded that PCM cooling was
a more effective technique for the performance improvement of CPV-T systems.
Ma et al. [7] performed mathematical modelling and sensitivity analysis of PV sys-
tem integrated with PCM. They developed a 1D resistance model and observed that
with every 100 W/m2 increase in solar radiation causes a 5 °C increase in the peak
PV temperature, however, they suggested that the PCM melting temperature should
be slightly higher than the surrounding ambient temperature. Lu et al. [8] investi-
gated the performance of building-integrated concentrated PV systems with the fins
and PCMs. A 12% improvement in PV efficiency was observed with the developed
fin-based PV PCM system.
It is observed from the previous literature that the PCM can be effectively utilized
for the thermal regulation of PV systems. Huge research work has already been
carried out to analyze the performance of PV system with PCM; however, it is
noticed that the numerical analysis of PV systems integrated with PCMs need to be
performed more effectively as most of the previous literature is based on vertical
building-integrated PV system and very few studies are available with inclined PV
systems. It is well known that the PV systems are mounted with different inclination
angles depending on the geographical location on the earth.
In the present work, numerical investigations are carried out with a PV PCM
system at different inclination angles. The numerical model is validated first with
Numerical Investigations of Photovoltaic Phase Change Materials … 505
the past experimental results available in the literature. The effect of temperature on
the specific heat and thermal conductivity is considered during the analysis. Natural
convection plays a major role during inclined PV PCM system; thus, the effect of
natural convection is also considered during numerical analysis using Boussinesq
approximations. The inclination angle of the PV PCM system is varied from 15° to
90° and the effect of inclination angle on the different performance parameters are
analyzed.
2 Methodology
In the current research work, a polycrystalline PV panel is considered for the numer-
ical analysis integrated with the PCMs. The PV panel consists of five layers and at
the back of the PV panel, the PCM container is attached with the aluminum metal
plate as a top side as shown in Fig. 1. For the numerical analysis few assumptions
were considered as heat loss is neglected from the bottom and the sidewall, solar flux
is distributed uniformly and the PV panel contact resistance is neglected.
The melting process of PCM inside the PCM container is modelled using enthalpy-
porosity technique explained in Voller and Prakash [9]. The governing equations used
during the numerical analysis are given in Table 1.
The developed numerical model is validated first with the experimental results
obtained by Kamkari et al. [11] with different input temperatures of 60 and 70 °C
applied to the vertical PCM tank as shown in Fig. 2. The grid independence study
and time-independent study has also been carried to analyze the effect of changing
time step size and grid size on the numerical results. Three different time step size
with different number of nodes is investigated and finally, 18,620 number of nodes is
selected for the analysis with 0.1 s of time step size. The temperature variations of the
PV surface, PCM melt fraction inside PCM container and velocity of PCM during
Table 1 Governing
Boussinesq approximations ρ = ρl /(β(T − Tl ) + 1) (1)
equations used in the →
numerical analysis [10] Continuity equation ∂ρ
∂t+ ∇. ρ −u = 0 (2)
− →−
Momentum equation ∂
∂t ρ→u + ∇. ρ − u→ u =
−
→
μ∇ 2 −
→u −∇ p +ρ − →g + S (3)
Energy equation ∂
−→
∂t (ρ H ) + ∇. ρ u H =
∇.(k∇T ) (4)
−
→ −
→ (1−γ )2 −
→
The source term S in the S =A mush γ 3 +ε u (5)
momentum equation
Where, β thermal expansion coefficient, Amush mushy zone
constant, H enthalpy
Fig. 2 Validation of
numerical model with 1.0
experimental [11]
0.8
700C
Liquid fraction
0.6
600C
0.4
Numerical simulation
Experimental results
0.2
0.0
phase transition is investigated at different tilt angle of the system with 1000 W/m2
of incoming heat flux. ANSYS Fluent 16 is used for the numerical analysis. SIM-
PLEC scheme is used for the pressure velocity coupling and PRESTO is used for
the pressure spatial discretization. Energy, velocity and continuity residuals were
selected to be 10–8, 10–4 and 10–4, respectively. Piecewise linear changes are used
to accommodate change in specific heat as suggested by Biwole et al. [12]. RT25 is
selected for the analysis with the melting point of 26.6 °C. The ambient temperature
is selected to be 20 °C.
Numerical Investigations of Photovoltaic Phase Change Materials … 507
360
350
150
0
350 15
Temperature of PV surface(0C)
450 450
340
600 600
330
900 900
330
320
320
310
310
300
300
290
0 50 100 150 200 250 300 0 50 100 150 200 250 300
Time (min) Time (min)
(a) (b)
1.0 150 0.0005 150
Melt fraction inside PCM container
300 300
450 450
0.8 0.0004
Velocity magnitude(m/s)
600 600
900 900
0.6 0.0003
0.4 0.0002
0.2 0.0001
0.0 0.0000
0 50 100 150 200 250 300 0 50 100 150 200 250 300
Time (min) Time (min)
(c) (d)
Fig. 3 a Temperature of PCM, b temperature of PV surface, c melt fraction, d velocity magnitude
508 T. Sathe et al.
at a lower inclination angle, the heat transfer is quite low, the time required to melt
the PCM is also more and thus it takes more time to convert into liquid Fig. 3c. The
velocity magnitude of PCM inside the PCM container is more in 90° system and it
decreases with the decrease in the inclination angle as shown in Fig. 3d. Figure 4
shows the contours of temperature variation with respect to time at 45° of inclination
angle. The temperature of the PV PCM systems is observed to be more on the top-
most corner of the system compared to the bottom side during the analysis and it is
not distributed equally. Thus, it is suggested that some passive systems like extended
surfaces can be attached on the top part of the PV PCM systems to extract excess heat
from the outside. The extended surfaces can also be placed on the heated surfaces to
make sure even distribution of temperature throughout the heated surface.
4 Conclusions
Inclination angle plays a major role in the design of PV PCM system. In the current
research work, effect of inclination angle on the thermal performance of PV PCM
systems is investigated. It is observed that as the inclination angle increases the time
required to melt the PCM decreases and PV surface temperature increases. Thus, it
can be concluded that optimum PCM thickness needs to be selected for the required
inclination angle of PV panel during the design of PV PCM system.
References
1. Browne MC, Norton B, Mccormack SJ (2015) Phase change materials for photovoltaic ther-
mal management. Renew Sustain Energy Rev 47:762–782. https://doi.org/10.1016/j.rser.2015.
03.050
2. Barnes FS, Levine JG (2011) Large energy storage systems handbook. CRC Press
3. Nazir H, Batool M, Bolivar Osorio FJ, Isaza-Ruiz M, Xu X, Vignarooban K et al (2019) Recent
developments in phase change materials for energy storage applications: a review. Int J Heat
Mass Transf 129:491–523. https://doi.org/10.1016/j.ijheatmasstransfer.2018.09.126
4. Choubineh N, Jannesari H, Kasaeian A (2019) Experimental study of the effect of using phase
change materials on the performance of an air-cooled photovoltaic system. Renew Sustain
Energy Rev 101:103–111. https://doi.org/10.1016/j.rser.2018.11.001
Numerical Investigations of Photovoltaic Phase Change Materials … 509
Abstract Nowadays, breast cancer is the most commonly occurring cancer among
women. Early diagnosis of such disease improves the survival rate. Mammography
is considered as the gold standard for the diagnosis of breast disease. But a frequent
examination of the breast through mammography increases the chance of cancer
occurrence due to radiation effect. However, taking advantage of non-radiating nature
of thermography, it can be used for regular screening of breast region temperature
distribution noninvasively. The presence of an inflammatory region in a breast ther-
mogram is a signal of abnormality. In a thermogram, the inflammation is represented
with higher gray level pixels. Observing this, in this study we aim to classify sick
and healthy breast thermograms by comparing the edge features of the images. For
the elimination of unwanted edges, the anisotropic diffusion is used as the filtering
process of thermograms before edge detection. The outcome of the Artificial Neural
Network-based classification using edge features as input generates 93.3% accuracy,
with an area under the curve value 0.92, sensitivity 100%, and specificity 85.7%.
1 Introduction
Breast cancer is the most commonly occurring cancer among women. In India, the
occurrence rate of breast cancer in a female is 25–32% among all types of cancers
[1]. Also among all cancer, worldwide it is the second most common type of cancer
[2]. 97% of survival rate can be achieved through early detection and treatment of
S. Bardhan
Department of Computer Science and Engineering, Tripura University (A Central University),
Suryamaninagar, Agartala, Tripura 799022, India
e-mail: shawli.cse@gmail.com
S. Roga (B)
Department of Mechanical Engineering, Visvesvaraya National Institute of Technology, South
Ambazari Rd., Ambazari, Nagpur, Maharashtra 440010, India
e-mail: rogasukanta@gmail.com
breast cancer [3]. Different imaging procedures are present for breast abnormality
diagnoses such as mammography, ultrasound imaging, magnetic resonance imaging,
etc. But those techniques contain the drawback of radiation effect, invasive nature of
examination, patient discomfort, etc. Infrared breast thermography is a noninvasive,
radiation-free method of breast screening that can detect abnormality in the early
stage of breast disease. The thermography technique is based on receiving the infrared
radiation emitted from the breast skin surface related to internal inflammation and
represents it in image format. In general, the inflamed regions are represented with
higher gray level pixel intensities in the thermogram. Analysis of the thermogram can
help in early stage abnormality detection. Sathish et al. [4] performed local energy
features of wavelet sub-bands based contralateral asymmetry analysis and achieved
91% accuracy. The left and right breast thermogram asymmetry is measured for
abnormality detection by Kirubha et al. [5]. Garduño-Ramón et al. [6] also performed
image processing methods for contralateral breast asymmetry analysis and achieved
0.8684 and 0.8943 rate of sensitivity and specificity, respectively, for the detection
of breast abnormality.
The abovementioned recent studies show that contralateral asymmetry analysis is
the most commonly used method for breast abnormality detection. But the method
fails if the temperature of both the breast increases due to the presence of abnormality.
Focusing on the limitation of existing methods, here in this study, we perform the
classification of sick and healthy breast thermograms by extracting holistic edge
features of the image.
2 Methodology
The method targets to classify the thermograms for abnormality detection. One group
consists of healthy thermograms and another is of sick thermograms of breast. The
methodology is dependent on edge features and performed through seven different
steps as shown in Fig. 1. The rest of the section detailed steps of the methodology.
The experiment has been performed over the publically DMR dataset [7]. The dataset
contains thermograms of sick breast and healthy breast. We collected 51 thermograms
and among those 23 belong to sick category and rest are from healthy groups. The
first image of Fig. 2 demonstrates an example of the collected sick thermogram.
2.2 Preprocessing
The experiment concentrates on the pixel of thermogram breast area. But the
thermograms contain unwanted regions like arm, stomach, logo, temperature bar,
etc.
Information from those unwanted regions may generate errors in analysis and
classification. Therefore, in the preprocessing step, the breast region is manually
cropped. In the next step of preprocessing, the blue channel from the cropped RGB
thermogram has been selected as this channel contains maximum information about
inflammation spreading. The example related to the outcome of preprocessing is
shown in Fig. 2.
2.3 Filtering
Fig. 3 Outcome of filtering using anisotropic diffusion and canny edge detection
Therefore, the noise of images is removed without affecting edge details and as a
result, unwanted edges are removed. To control the diffusion mechanism, the method
requires to give the iteration number, conduction coefficient, and speed of diffusion.
Here we use 50 iterations with conduction coefficient value 30 and speed of diffusion
0.25. The output will preserve the information of high contrast edges for abnormal-
ity analysis. The outcome of filtering is shown in the middle image of Fig. 3 below
where input is the blue channel thermogram.
In this step, the region base high contrast edges are detected. For that, the canny edge
detection mechanism has been applied over the outcome of filtering mechanism [10].
Though canny uses the Gaussian filtering for smoothing base noise removal, it does
not diminish the edges as the high contrast edges are preserved through anisotropic
diffusion-based filtering. After smoothing, the gradient magnitude is computed for
finding out the edges. In the third step of canny, non-maximal suppression performed
to eliminate pixels that do not represent the edge of a region. In the final step,
hysteresis thresholding is applied depending on the image gradient for edge detection.
The last image of Fig. 3 shows the outcome of canny edge detection over the outcome
of anisotropic diffusion.
2.6 Classification
For classification, Artificial Neural Network (ANN) is used in this study with two-
layer feed-forward channel. The sigmoidal activation function is applied to generate
nonlinearity in the network. Here we used 40 hidden neurons for optimal training
of the network through trial and error mechanism. 30% of total images are sent for
testing purposes of the system where each represented with 11 features. The output
layer divides the input images into two classes: one is healthy group and another
represents sick thermograms. The Support Vector Machine (SVM) classifier is also
used for the same classification for the comparative study of classification.
2.7 Validation
The validation is performed for the outcome of the system measurement. The classi-
fier outcome (sick or healthy) of each thermogram used for testing is matched with
the actual status of the breast thermograms.
3 Result
4 Conclusion
Fig. 4 a ROC curve of ANN classifier with AUC: 0.92; b confusion matrix of ANN classifier;
c ROC curve of SVM classifier with AUC: 0.82
from sick and healthy thermogram images. The anisotropic diffusion base filtering
is applied over the blue channel of color thermogram for noise cancellation. The
outcome of the analysis shows that use of the edge-based extracted features in the
ANN classifier provides maximum accuracy of 93.3%, with sensitivity 100% and
specificity 85.7%.
Acknowledgements Authors would also like to thank Dr. M. K. Bhowmik, Assistant Professor,
Department of Computer Science and Engineering, Tripura University, Suryamaninagar 799022,
Tripura, India for his support during knowledge development in the area of thermal image
processing, image feature extraction, and classification.
Edge Feature Based Classification of Breast Thermogram … 517
References
1 Introduction
Mechanisms used in space applications are mission critical and need to be evalu-
ated on various dimensions like modal, quasi-static loading and frequency response
analyses during configuration studies. Analytical methodology using finite element
analysis (FEA) is used to estimate characteristics of the system that are used as input
for the improvement of system configuration via iterative studies.
Mechanism for dual-axis gimballing of the antenna is a key spacecraft appendage
that enables the communication between earth stations and satellite. It provides two-
axis steering of the antenna to ensure hemispherical coverage and ground station
Satellite-level requirements for positioning an antenna on-orbit are based upon the
communication system, envisaged between the ground stations and satellite, which
depends upon various parameters that can be broadly classified as below:
a. Mission parameters like orbital height, inclination, field of view, pointing accu-
racy, data transmission (real time or non-real time), ground passes, steering
rate.
b. Environmental constraints like extreme temperatures, vibration loads during
launch, radiation impacts on-orbit.
c. RF specifications in the form of frequency band, spot beam focus, RF power.
Based on the satellite-level requirements, mechanical system-level requirements
are derived. These requirements can be classified as:
a. Drive system parameters, to provide torque for gimbal along two axes, based
upon global parameters like steering rate, vibration loads, pointing accuracy [1].
b. Environment support system considerations in the form of stiffness character-
istics, stress distribution, moments along gimbal axes that depend on global
parameters like launch vibrations, on-orbit thermal gradients, geometry of
antenna.
c. RF transmission system considerations like waveguide or cable management unit
for contactless data transmission during steering depending on global parameters
like orbital height, inclination, frequency band, RF power [2].
From the above, it is evident that multiple variants of mechanism configuration
are required to meet the mechanical requirements corresponding to varied satellite
demands. However, the realization of different mechanism configurations is not war-
ranted from a practical point of view. Thus, an optimized solution in the form of a
mechanism configuration that caters to the varied satellite requirements is essential.
In this regard, the development of a modular mechanism configuration, where with
Analytical Approach to Develop a Robust Mechanism … 521
modification of specific modules, the mechanism configuration can cater to the var-
ied gimbal requirements, is worth considering. Advantages of modular configuration
include:
a. Benefit of catering to multiple requirements with minimum changes.
b. Ease of testing, as it can be carried out module wise.
c. Optimization of time, as modules can be produced in batches that can be used as
per satellite requirements.
Development of such robust mechanism configuration demands analytical
approach toward iterative parametric studies in order to identify as well as minimize
the specific modules that determine variants of mechanism configuration.
The mechanism for antenna gimbal is configured with several sub-systems like motor,
gear, RF transmission system and hold-down system to fulfill the mechanical require-
ments. The prime objective in the development of this mechanism configuration is
to fulfill the varied antenna gimballing requirements across satellites with minimum
modification of the mechanism configuration. Toward this, the mechanism is divided
into several modules where the majority of these modules are the same across the
variants of mechanism configuration. For instance, identical high torque capability
drives are used for both the gimbal axes in all the variants. Also, the mechanism
configuration has been made keeping in view the requirements from different multi-
disciplinary systems involved in satellite development. For instance, mechanism can
suit X-band as well as Ka-band rotary joints that are determined by the communica-
tion systems. Thus, robustness in the configuration of mechanism for antenna gimbal
has been enabled by adoption of modular approach and multidisciplinary perspective
during the mechanism development process.
Analytical approach has been adopted to analyze the mechanism configuration
module wise, using iterative parametric studies. This enabled the development of
variants of mechanism configuration to fulfill the varied satellite demands, with
minimum changes in the configuration. For instance, the gimballing mechanism for
1-m-diameter antenna and 0.35-m-diameter antenna is identical except the hold-
down system, as shown in Fig. 1. This characteristic of robustness in mechanism
configuration enables quicker design iterations and effective parametric analyses in
order to address larger and diverse antenna gimbal requirements.
522 V. S. P. RaviChand et al.
Common
modules –
Drive modules
Antenna
Variable
modules –
Hold-down
unit with
satellite inter-
face structure
Patch antenna gimbal mechanism Parabolic dish antenna gimbal mechanism
These studies are carried out sequentially in an iterative manner during the entire
mechanism development process, i.e., from mechanism conceptualization phase till
on-orbit performance review stage, so as to minimize modifications across different
variants of gimbal mechanism configuration while meeting the respective satellite
mission requirements.
a. Determination of geometric shape for gimballing
Kinematic simulation is carried out to obtain the configuration of mechanism in the
form of geometric shape for steering along the two gimbal axes. It is also used to
carry out the interference studies during steering.
b. Evaluation of dynamic loads during gimballing
Analytical Approach to Develop a Robust Mechanism … 523
After ensuring that all the components are seamlessly rotating without any interfer-
ence, dynamic simulations are carried out to estimate loads on the joints by simulating
drive speeds, acceleration and inertia parameters.
c. Estimation of stiffness characteristics of mechanism
Satellite-based systems are primarily driven by stiffness characteristics. Hence, the
configuration of mechanism needs to be evaluated with respect to stiffness by carrying
out normal mode analysis with requisite boundary conditions. This FEA provides
details on dynamic characteristics of the mechanism with respect to mode shapes,
natural frequency values and participating mass details. Based on the FEA results,
the improvements with respect to mechanism configuration in the form of providing
stiffeners, etc. can be carried out iteratively.
d. Evaluation of strength parameters of system
Spacecraft mechanisms are characterized by intricate geometry so as to withstand
various eccentric loads like extreme thermal gradients and large launch accelerations.
In this regard, quasi-static FEA, for inertia loading of 30 or 20 g based on launch
vehicle specifications, is carried out to evaluate stress distribution and joint forces
in the system. Accordingly, the improvements in configuration, by way of providing
fillets, can be carried out to reduce stress concentrations and high joint loads.
e. Evaluation of amplitude responses of system during vibration
Integrity of the mechanism system under harmonic vibrations across a range of
frequencies, determined based on mission specifications, needs to be ensured so as to
avoid resonance. This can be ensured by evaluating amplitudes within the mechanism
using frequency response FEA and accordingly reconfiguring the system to have beta
values within environmental specification limits.
f. Estimation of responses at satellite level
Mechanism system’s fundamental frequency should be kept away from the satellite
frequencies so as to avoid resonance phenomena. This can be achieved by evaluating
amplitudes at the spacecraft level using coupled load analysis. This coupled load
FEA also provides the input loads for the mechanism system.
g. On-orbit performance analysis
The functioning of mechanisms on-orbit is contingent upon the body rates experi-
enced by the satellite. To evaluate on this account, analysis from multi-body dynamics
perspective is carried out, and the corresponding data is correlated with the actual
on-orbit satellite data received, by carrying out fast Fourier transform (FFT) on body
rate plots, to evaluate the overall system performance.
The analytical approach has been applied for devising a mechanism for dual-
axis gimbal of parabolic dish antenna from the available mechanism for dual-axis
gimbal of a patch antenna, as shown in Fig. 2. The variations with respect to satellite
requirements for these configurations are that the mass and size of parabolic dish
antenna are about twofold more than that of the patch antenna. Also, the stowing
524 V. S. P. RaviChand et al.
Hold-down
structural
interface
angles are different for both cases. Accordingly, there needs to be variations on
mechanical requirements to accommodate stiffness characteristics.
Based on the above-mentioned analytical approach, configuration of mechanism
for gimbal of parabolic dish antenna has been devised from that of the mechanism for
gimbal of the patch antenna, as shown in Fig. 2, by appropriate modification. It is to
be noted that both the mechanism configurations have the same modules except for
hold-down unit and structural interface of mechanism with satellite. Using analytical
studies, the hold-down system module related to patch antenna gimbal mechanism
configuration has been modified to meet stiffness requirements of parabolic dish
antenna, while the structural interface base has been modified to meet stowing angle
requirements. It is important to note that both these mechanisms are planned to be
used on agile low mass satellites of nearly the same structural frequency, thereby
the fundamental frequencies of both these variants need to be almost the same in
spite of differences in mass or geometry of respective antennas. For the given case
study, adoption of the systematic methodology enabled devise of both the variants of
mechanism configuration, shown in Fig. 2, as per requisite specification of around
70 Hz fundamental frequency with minimal modification of mechanism configura-
tion. Thus, the adoption of the systematic analytical approach during detailed para-
metric studies enables the development of a robust configuration of modular antenna
gimbal mechanism that can cater to the varied satellite demands.
4 Conclusion
This paper brings out a systematic methodology using analytical approach for devis-
ing a robust configuration of mechanism for dual-axis gimbal of the antenna to cater
to the varied satellite requirements. This is achieved by adopting a modular approach
with multidisciplinary perspective during mechanism development process based on
Analytical Approach to Develop a Robust Mechanism … 525
References
Abstract With the introduction of tunnel boring machines (TBMs), the rate of devel-
opment and construction of tunnels has increased across the world. Breakdowns,
repairs and replacements that occur in a TBM have to be addressed within the tun-
nel. Of the several components within a TBM, the cutter-picks within the cutter-head
undergo maximum wear and tear, thereby requiring periodic replacements. In order
to optimize the consumption of cutter-picks, it is important to analyse the forces
involved in the cutter and the properties that govern rock–cutter interactions. One
such property is the abrasivity, which can be determined by CERCHAR abrasiv-
ity index (CAI) tests. Research suggests that the value of CAI is largely depen-
dent on the mineralogical, physical and mechanical properties of rocks. Therefore,
this study aims to perform experimental analysis of the various properties and their
corresponding effects on granite and to determine the respective changes in CAI.
1 Introduction
Mechanized tunnelling has improved the rates of excavation and exploration along the
last few decades. This was achieved through the introduction of various rock-cutting
machines like tunnel boring machine (TBM), road-headers, etc. Economics of the
tunnelling operation is largely dependent on the machine availability. One of the most
important components of a tunnelling machine is the cutting mechanism, which in
the case of a TBM is the cutter-head that comprises several cutting discs, also known
as cutter-picks arranged in a specific and concentric pattern, as illustrated in Fig. 1.
Since the cutter-picks undergo continuous wear and tear due to rock interaction, their
condition determines the efficiency of the cutting operation. Frequent replacements of
the worn-out picks have to be carried out within the tunnelling site which increases the
Fig. 1 Types of tunnel boring machines (TBMs) and their respective cutter-head
total project cost and duration. Studies suggest that the consumption of cutter-discs
and its rate is mainly governed by the strata (geological) and machine parameters.
While the machine parameters are within our control, the type of geological formation
and its characteristics should be studied in order to design and deploy the proper type
of TBM. One such characteristic is the abrasivity which accounts for gradual loss of
the cutting tool due to pick–rock interaction. The physico-mechanical properties of
the rock surface being cut and the stresses involved in the pick–rock interaction also
play a major role in the total wear [1].
Of the several tests developed for analysing the abrasivity of rocks, the CERCHAR
abrasivity index (CAI) is one of the most widely used tests due to its simplicity. The
test was developed in 1973 by Laboratoire du Centre d’Etudes et Recherchers des
Charbonages de France and involves the formation of a scratch on the surface of
a rock sample with the help of a steel stylus. The original design of the equipment
involved making a 10-mm scratch on a stationary rock specimen within one second
by means of a pin that carried a static load of 70 N [2]. However, a modification
to the design is proposed by West [3], wherein a stationary pin with a 70 N static
load creates a scratch on a rock sample that moves at a velocity of 1 mm/s. The test
is conducted for 10 s. Thereafter, the wear to the steel stylus is examined in order
to calculate the CERCHAR abrasivity index (CAI) [4]. The CAI of the rock can
be used to estimate the cutter-pick consumption with the help of predefined charts
and equations [5]. The magnitude of wear depends entirely on the rock type, and
therefore, in this study, various physico-mechanical properties of a granitic rock
have been analysed with respect to its CAI.
In this study, granitic rock samples were used for testing the dependence of physico-
mechanical properties on its abrasivity. The samples were obtained from the core
specimens sent to the Department of Mining Engineering, Visvesvaraya National
Institute of Technology (VNIT), by Central Institute of Mining and Fuel Research
Impact of Rock Abrasivity on TBM Cutter-Discs During Tunnelling … 529
(CIMFR). The core samples were cut to appropriate lengths with the help of a core
cutter. Care was taken to ensure that the length and attributes of the specimen were
in accordance with the standards and methods as specified by the American Society
for Testing and Materials (ASTM) and International Society for Rock Mechanics
(ISRM), respectively [6–9]
Cylindrical specimens of the sample were used for testing the compressive (σ C ) and
tensile (σ T ) strengths. Samples with a length-to-diameter ratio of 2:1 were tested in a
compressive testing machine, and the axial strain was recorded with the help of strain
gauge which was installed on the sample before the test. The load and strain data was
recorded with the help of DataTaker 80G data acquisition system. The tensile strength
was analysed using the indirect Brazilian testing method. The cylindrical samples for
the tensile test had a length-to-diameter ratio of 0.75:1. The elastic modulus (E) of
the samples was measured by analysing the stress–strain plots. It was ensured that the
linear section of the plot was selected for the modulus measurement. Additionally,
it was ensured that the corresponding stress window of the linear section amounted
to at least 20% of the peak stress. This helped in obtaining representative data of the
elastic modulus.
Furthermore, the porosity of the samples was tested using the buoyancy and
saturation techniques. The limitations of this technique are that it can only be applied
to rocks that are non-swelling in nature, and that the technique only measures the
effective porosity. In this technique, the samples were submerged in water for 24 h
before measuring the saturated mass (M SAT ), submerge-saturated mass (M SUB ) and
the oven-dried mass (M S ). The porosity (Φ) was calculated using Eq. (1).
MSAT − M S
φ= × 100 (1)
MSAT − MSUB
The CERCHAR abrasivity set-up used for this study is illustrated in Fig. 2. The set-
up, as developed by West [3], is a modification of the original design, wherein the
surface of the rock sample moves across a stationary pin [1]. The test was conducted
in accordance with the recommendations prescribed by the ASTM standard D7625-
10 and the ISRM suggested method [4, 8]. The steel stylus with a Rockwell hardness
of 54–56 and a 90° conical tip was used in this study. A static load of 70 N was applied
to make a scratch across a predefined path. The rock sample was moved across the
stationary pin at a velocity of 1 mm/s. In order to attain statistical coherence, five
530 N. N. Sirdesai et al.
scratches were made across each sample using re-sharpened styluses, as suggested
by Käsling and Thuro [2]. The damage to the tip was observed and measured using
an optical microscope with a 30× magnification. The change in diameter at the tip
(DW ) was recorded to an accuracy of 0.01 mm to calculate the CAI as illustrated in
Eq. (2).
CAI = 10 × DW (2)
The results of the physico-mechanical and abrasivity analysis have been enlisted
in Table 1. The correlation between the physico-mechanical properties and CAI
has been plotted in Fig. 3. As seen in this figure, there exists a linear correlation
between the strength and elastic properties, and porosity with CAI. The equations
and correlation coefficients are illustrated in Eqs. (3–6).
Fig. 3 Correlation between a compressive strength, b tensile strength, c Young’s modulus and
d porosity with respect to CAI
The equation states that except for porosity, the CAI increases with an increase
in strength and elasticity. Similar correlations have been suggested by several
researchers. For instance, Tripathy et al. [10] studied the correlation between abra-
sivity and physico-mechanical properties of several Himalayan rock samples using
statistical and soft-computing tools. The results of their analyses suggest that an
increase in the mechanical properties of the rock increases the abrasivity of the rock.
This can be attributed to the increase in the resistance provided by the rock to the
formation of a scratch. Additionally, the rocks formed by relatively tougher minerals
offer high resistivity to deformation. Research conducted by Deliormanlı [11] and
Hamzaban et al. [12] have also suggested similar findings.
Further, Alber [1] performed a rigorous experimental analysis to correlate the
effect of in situ stresses with abrasivity. The results suggest that an increase in the
porosity of a rock sample tends to decrease the sample’s resistance to wear and tear.
This makes it easier to create a fracture (scratch) within such rock types. Therefore,
porosity has a negative correlation with CAI.
532 N. N. Sirdesai et al.
Several studies have been conducted to classify the abrasivity of rocks based on
its CAI value. The studies suggest that the CAI values can be used for predicting
the wear and tear of tool bits and pick consumption in mechanized excavators such
as road-headers and TBMs [5, 13, 14]. For instance, Thuro and Plinninger [14]
proposed a graphical representation to predict the rate of consumption of a 45-mm
button drill bit as a function of CAI, as represented in Fig. 4(a). The study suggests
that rocks with high abrasivity (≥3.0) have an adverse effect on the condition of drill
bit. Further, the plot suggests that rocks belonging to the metamorphic and igneous
variety tend to be highly abrasive when compared to sedimentary rock types. The
rock samples used in this study have an average CAI of 2.92, thereby suggesting that
if a 45-mm button drill bit were to be used for drilling into such rocks, the tool bit
would have to be replaced after approximately ~300 m of drilling. Such correlations
help in developing techno-economic models for any drilling operation.
Further, Girmscheid [15] suggested a graphical correlation between the UCS
and CAI to predict the wear and tear of the cutter bits of a road-header. Figure 4b
illustrating the Girmscheid graphical correlation provides an insight on the cutter-bit
lifecycle for a road-header and provides an estimation on the volume of loss upon
the excavation of 1 m3 of fresh solid rock. Based on this correlation, if a road-header
were to be used to excavate a tunnel in a granitic rock formation similar to the one
used in this study, the magnitude of wear would be 0.3 picks per 1 m3 of excavation.
This suggests that the pick needs to be replaced after excavating 3.3 m3 of rock.
In order to estimate the loss of cutter-disc while excavating a tunnel using TBM,
Gehring [16] suggested a 17-inch cutter-disc should be replaced if the loss of weight
is about 3500 gm. In order to simplify the disc replacement protocol, Gehring [16]
correlated the specific ring weight loss VS (mg/m-rolling distance) as function of the
CAI, as illustrated in Eq. (7).
Fig. 4 Estimation of a 45-mm button drill bit lifetime and b cutter-pick wear
Impact of Rock Abrasivity on TBM Cutter-Discs During Tunnelling … 533
Based on this equation, the loss in weight for a 17-inch cutter-disc fitted onto a
TBM which is excavating through a granitic rock formation similar to that analysed
in this study would be 5.85 mg/m. This suggests that the cutter-discs should be
replaced after roughly ~600 km of travel (rolling distance). It should be noted that
the rolling distance travelled by the cutter-discs of a TBM depends entirely on their
position and distance from the centreline. A disc at the peripheral section would cover
larger rolling distance to that near the centreline for a unit length of excavation. This
suggests that the peripheral cutter-discs would endure higher wear and tear than that
of central discs. Similar to the example illustrated by Alber [1], if TBM with a 2.5-m
cutter-head were to be used for constructing a tunnel within a granitic rock and if the
advance per revolution is 1 mm, then the peripheral cutter-discs would have to be
replaced after roughly ~40 m of advance. Such correlations help in estimating and
planning the maintenance schedule for the equipment.
5 Conclusion
References
1. Alber M (2008) Stress dependency of the Cerchar abrasivity index (CAI) and its effects on
wear of selected rock cutting tools. Tunn Undergr Space Technol 23(4):351–359
2. Käsling H, Thuro K (2010) Determining rock abrasivity in the laboratory. Rock mechanics in
civil and environmental engineering—Proc EUROCK:425–428
3. West G (1989) Rock abrasiveness testing for tunnelling. Intl J Rock Mech Mining Sci
Geomechanic Abs 26(2)
4. Alber M, Yaralı O, Dahl F, Bruland A, Käsling H, Michalakopoulos TN, Cardu M, Hagan P,
Aydın H, Özarslan A (2013) ISRM suggested method for determining the abrasivity of rock
by the CERCHAR abrasivity test. In: The ISRM suggested methods for rock characterization,
testing and monitoring: 2007–2014. Springer, pp 101–106
534 N. N. Sirdesai et al.
5. Plinninger RJ, Kasling H, Thuro K (2004) Wear prediction in hardrock excavation using the
CERCHAR abrasiveness index (CAI). In: Schubert (ed.) EUROCK 2004 & 53rd Geomechanics
Colloquium, VGE, pp 1–6
6. ASTM (2008) D3967-08. Standard test method for splitting tensile strength of intact rock core
specimen. ASTM International, West Conshohocken, PA. www.astm.org, https://doi.org/10.
1520/d3967-08
7. ASTM (2014) D7012-14E: standard test methods for compressive strength and elastic moduli
of intact rock core specimens under varying states of stress and temperatures. Annual Book of
ASTM Standards. ASTM International, West Conshohocken, PA
8. ASTM (2010) D7625-10, Standard test method for laboratory determination of abrasiveness
of rock using the CERCHAR method, ASTM International, West Conshohocken, PA. www.
astm.org, https://doi.org/10.1520/d7625-10
9. ISRM (1981) Rock characterization testing and monitoring—ISRM suggested methods.
Pergamon Press for the Commission of Testing Methods
10. Tripathy A, Singh T, Kundu J (2015) Prediction of abrasiveness index of some Indian rocks
using soft computing methods. Measurement 68:302–309
11. Deliormanlı AH (2012) Cerchar abrasivity index (CAI) and its relation to strength and abrasion
test methods for marble stones. Constr Build Mater 30:16–21
12. Hamzaban M, Memarian H, Rostami J (2018) Determination of scratching energy index for
Cerchar abrasion test. J Mining Environ 9(1):73–89
13. Plinninger RJ, Spaun G, Thuro K (2002) Predicting tool wear in drill and blast. Tunnels &
Tunneling International Magazine:1–5
14. Thuro K, Plinninger R (2003) Hard rock tunnel boring, cutting, drilling and blasting: rock
parameters for excavatability. In: Proceedings of the 10th ISRM Congress. International Society
for Rock Mechanics and Rock Engineering
15. Girmscheid G (2000) Construction management and construction techniques in tunneling.
Berlin Ernst and Sohn
16. Gehring K (1995) Leistungs-und Verschleißprognosen im maschinellen Tunnelbau. Felsbau
13(6):439–448
Tool Condition Prediction Using Acoustic
Signal Processing and Learning-Based
Methods
Abstract In the work presented here, signal processing techniques are used for con-
dition monitoring of boring tool using acoustic signals. To accomplish this work, 108
experimental runs are performed to diagnose the tool life of the boring operation, by
using two different sensor data. The acoustic signals are captured from the tool which
is performing boring operation on mild steel materials. The acoustic signature being
the main signal for study, vibration signals have also been measured for comparison.
Four sets of boring experiments were performed. After the data had been collected
for all runs of boring operation, signal processing methods are applied for doing the
analysis of captured signal. The acoustic signals are captured from the boring tool
and further analyzed by using learning-based methods for determining the condition
of boring tool.
1 Introduction
The problem of tool failure like tool wear and tool breakage is becoming more and
more important in the manufacturing industry. In metal cutting operation, such as
boring, the cutting tool becomes dull after a certain working time and requires chang-
ing. The consequences of tool wear while machining may change the dimensions,
tolerances, and surface finish of the part or product. So, measures for avoiding this
must be taken in early stage before the tool breakage or it effects on the quality of
the product in order to reduce cost and time [1]. Use of signals such as acoustic
emissions and vibrations that are influenced by the boring operation can be used
effectively for monitoring the condition of boring tool. The existing approaches may
not be capable of predicting the tool life and monitoring the tool health effectively
[2]. This necessitated the need for an in-process tool condition prediction approach,
which may be used while the machining process is going on and which is based upon
the signals that are believed to be influenced by the machining process [4]. So, an
approach to determine the condition of boring tool using acoustic signals is presented
here.
2 Methodology
Four sets of boring experiments were performed. After the data had been collected
for all runs of boring operation, signal processing methods are applied for doing
the analysis of captured signal. Figure 1 shows the graph between tool life (mm2 )
and tool wear (gm). The tool wear is measured as a loss of weight during each
boring operation run. The tool life is measured in terms of the circumference of the
workpiece machined. The circumference is given as DL where the length of the
work piece is fixed which is taken as 125 mm. The diameter is getting changed after
machining so the tool life is taken in terms of the circumference of the workpiece
getting bored. It is observed from the graph that as the tool wear is increasing, there is
a spontaneous decrease in the life of tool which is true for all four sets of experiments.
The experiments are conducted to collect the data of acoustic and vibration signals,
and the corresponding output of weight loss of tool and roughness of the surface
machined is recorded [3]. It was found that the tool is worn out at experiment no. 39.
This is found from the fact that there was a reduction of 0.05 gm of the total weight
of the cutting tool. Figure 2 represents the tool wear and the tool worn out. It can
be observed from the graph that tool worn out took place at 39th experimental run.
As the same insert was used to carry out all the experimental runs, the insert started
wearing out after carrying out experimental runs. As a consequence of which the
remaining useful life of the cutting tool started reducing. Figure 2 shows that when
the tool wear reduces, the remaining useful life of the tool increases gradually. The
red mark in the graph shows that there is a failure of tool at that point.
The acoustic signal [6] data captured during boring operation is processed using an
appropriate ANN signal processing technique. Vibration signals are too acquired and
processed for comparison in the later part of this research [5]. The acoustic signals
captured through microphone are analyzed [3]. The raw data undergoes de-noising,
amplification, and conditioning for effective analysis of the signal.
The wavelet families that are selected and optimal wavelet de-noising parameters are
estimated based on SNR criterion. These wavelet families represent the comparison
of acoustic signal de-noising performance based on different levels as shown in Fig. 3.
538 P. S. Deole and P. M. Khandekar
The preprocessed signal acoustic signals contain a large volume of information that
needs to be broken into meaningful information. The information so extracted is
the feature of the signal and is generally processed in the time domain to obtain its
statistical characteristics, in the frequency domain to analyze the power of signal
distribution over a range of frequencies, and in the time–frequency domain to obtain
spectrum of non-stationary signals. The time domain features which contain impor-
tant information are extracted from the signal. As the time domain features reflect
only changes in signal with respective time, there is a need to extract features in the
frequency domain also. Every de-noised acoustic signal is further processed using
wavelet packet transform to extract the time–frequency domain features. The signal
was decomposed into 6 levels, and 64 nodal points were determined using WPT tool-
box of MATLAB. Coefficients of these nodal points were further estimated. For every
array of coefficients, features estimated are logarithmic energy, skewness, kurtosis,
threshold crossing rate, pulse width, and effective value.
52 features are extracted from every de-noised acoustic signal in time, frequency,
and time–frequency domains. It is desirable that the features should be optimized
according to a criterion such as Pearson’s correlation coefficient (PCC), R-square
statistical methods, and Pearson’s chi-squared test. The features for which PCC is
greater than 0.9, R squared statistics is greater than 0.9 and Pearson chi squared test
for 95% significance level are selected. The features of effective value are extracted
Tool Condition Prediction Using Acoustic … 539
for the said three methods. The features from Pearson’s correlation coefficient are
selected for further analysis, as the Pearson’s correlation coefficient is the best method
for feature selection. Hence, 12 features for acoustic emissions are selected.
For developing a model for tool life estimation and estimating remaining useful life
of the tool, the artificial neural network is used [7, 8]. There are a wide variety of
networks depending on the nature of information processing carried out at individual
nodes, the topology of the links, and the algorithm for adaptation of link weights.
Multilayered perceptron (MLP) neural network and radial basis function (RBF) neu-
ral networks are used in the present work for developing the model for tool life and
remaining useful life estimation [7]. There are 17 total inputs to neural network, i.e.,
cutting speed, feed rate, depth of cut, hardness of material, tool slenderness ratio, and
12 features of acoustic emission. The expected output of neural network is surface
roughness, tool wear, and remaining useful life of tool. Table 1 shows the parame-
ters used for neural network for both multilayer perceptron (MLP) and radial basis
function (RBF).
By doing the error comparison between MLP and RBF for different error perfor-
mances, it is found that R-square prediction is higher, i.e., 0.9861 in RBF. So, RBF
is best suited for the analysis of the acoustic signals.
Fig. 4 Measured tool wear versus predicted using MLP and RBF
6 Experimental Validation
For validation of developed neural network model, a fresh set of experiments were
conducted and performance parameters were measured.
Figure 4 gives the performance of MLP and RBF for measured tool wear and the
one predicted with MLP and RBF. The validation of the proposed model is done by
performing a new set of experiments with some fixed set of process parameters to
observe the prediction of tool wear using the two ANN approaches.
The average error in the prediction of tool wear is 0.0334 (gm) for MLP and 0.021
gm in RBF, whereas the percent average error in the prediction of tool wear is 1.58
in MLP and 0.0168 in RBF. So, the RBF ANN architecture is preferred over MLP.
It is found that there is a similar trend in the change of surface roughness (microns)
measured for validation experimental runs for RBF neural network. It is observed
from the validation experiments clearly that MLP is not a good performing neural
network for surface roughness. The features proved good enough to predict the tool
condition and remaining useful life of the tool with radial basis function neural
network being more accurate learning-based method for determining the condition
of boring tool.
Tool Condition Prediction Using Acoustic … 541
References
1. Schneider G (2010) Cutting tool applications chapter 10: boring operations and machines. Boring
increases the inside diameter of a hole, and achieves three things: sizing, straightness, and
concentricity. American Machinist. 27 July
2. Singh S (2011) Handbook of mechanical engineering, 2nd edn. S. Chand publication, pp 807–885
3. Tana CK, Irvinga P, Mba D (2007) A comparative experimental study on the diagnostic and
prognostic capabilities of acoustics emission, vibration and spectrometric oil analysis for spur
gears. Mech Syst Sig Proc 21:208–233
4. Dongre PR, Chiddarwar SS (2013) Tool condition monitoring in various machining operations
and use of acouastic signature analysis. Int J Mech Eng Robot (IJMER) 1(1). ISSN (Print):
2321–5747
5. Orhan S, Osman A, Camus-cu N, Aslan E (2007) Tool wear evaluation by vibration analysis
during end milling of AISI D3 cold work tool steel with 35 HRC hardness. NDT and E Int
40:121–126
6. Mathew MT, Pai PS, Rocha LA (2008) An effective sensor for tool wear monitoring in face
milling: acoustic emission. Sadhana 33(3):227–233
7. Xu C, Fan X, Luo W (2009) Research of tool wear based on radical basic function network. In:
Proceedings of the international multi conference of engineers and computer scientists, vol 1,
March 18–20 2009. Hong Kong, pp 1–6
8. Zhang C, Yao X, Zhang J, Jin H (2016) Tool condition monitoring and remaining useful life
prognostic based on a wireless sensor in dry milling operations. Sensors 16(6–795):1–20
Finite Element Simulation of Ballistic
Response of Metallic Sandwich
Structures with Aluminium Foam Core
1 Introduction
The performance of sandwich structure under impact loading has been studied widely
[1–6] and characterization of their behaviour with varying conditions of ballistic
impact, viz. configurations of individual components, boundary conditions, thick-
ness, shape and size of the targets, and nose shape, incidence angles, size, impact
velocity of projectiles are reported. The output parameters such as residual velocity,
ballistic limit, failure modes and deformations observed locally and globally and
damage initiation in sandwich structure; work and nature of work amount absorbed
in the components are also reported in the literature. The parameters specific to the
sandwich structure such as stress versus strain characteristics, load versus indenter
depth and shock enhancement under varied rates of loading have also been presented
in the literature. These cellular structures have been studied using all three techniques
experimental, numerical and analytical/empirical methods.
Zhao et al. [1] experimentally investigated the performance of sandwich pan-
els against hemispherical projectile by carrying out inverse perforation tests where
instead of free-flying projectile, the sandwich is made to impact on a stationary per-
forator. Hou et al. [2] experimentally determined the performance of sandwich panels
with Al5005-H34 alloy face sheets and CYMAT foam cores against projectiles with
hemispherical, conical and flat-shaped projectile. Effect of core density, projectile
nose shapes, impact velocities of the projectiles and thickness of face sheets and
core were studied. Hanssen et al. [3] studied the effect idealized bird strike on dou-
ble sandwich panels, and LS-DYNA code was employed to carry out impact on
sandwich panels. A model based on continuum damages theory was employed for
the panel sheets, and phenomenological Deshpande–Fleck model was employed for
the foam core. Mohan et al. [4] suggested that the load vs. penetration depth of the
indenter revealed that foam material provides resistance by cell wall crushing under
the impactor nose and tearing of cell wall near the edge of the impactor. Increasing
the thickness of foam and incorporating face plate increased energy absorption of
the sandwich panels. Elnasri et al. [5] calculated the top piercing force required for
top face sheet penetration by RPPL shock model and found to be good agreement
with experimental result. Su et al. [6] studied ballistic response of curved sandwich
shells under normal/oblique impact including different curvatures, projectile nose
shapes and angle of impact of the projectile. Graded foam core and thicker face
sheets were proven to increase the ballistic resistance of the sandwich structure. For
aluminium foams and sandwich structures, it is evident that complete finite element
analysis considering complex behaviour of sandwich structure under impact condi-
tions requires a detailed constitutive model for both face plates and core as based on
the review of the literature. Hence, this study attempted to combine the two detailed
constitutive models: Johnson–Cook material model for face plates and Deshpande–
Fleck foam material model for core. The focus of this study is to describe the ballistic
performance against projectiles under different conditions of impact.
3.75 mm
25.50 mm
(c)
(a)
(d)
(b)
Fig. 1 Finite element model for parametric study: a assembly of the parts, b hemispherical
projectile, c enlarged view of primary impact zone, d number of elements over thickness
elements close to 1. The sides of elements were kept as 0.50 mm in the primary
impact zone and across the thickness of 1 mm; 6 elements were taken as shown in
Fig. 1d for face plates. The core was modelled with the same element size, and 21
elements were taken across the thickness. The total elements in the primary impact
zone for each layer were 246 for face plate while 5376 for the core. Material model
used for the face sheet and foam has been shown in Tables 1 and 2.
The sandwich target of face plate thickness of 0.60 mm and core thickness of 25 mm
was impacted with hemispherical projectile, and residual velocities upon impact at
various impact velocities are given Table 3 and Fig. 3. The numerical values are
presented and compared with the experimental values. It is observed that higher
impact velocities, the numerical model is in good agreement with the experimental
results while the model numerically over predicts the residual velocities at lower
impact when compared to the experimental values (Fig. 2 and Table 4).
546
(a) (b)
(c) (d)
Fig. 2 Residual velocity variation with impact velocity and perforation energy. a and b indicate
experimental and numerical values obtained for bare plates and core for residual versus impact
velocity and perforation energy respectively. c and d indicate experimental and numerical val-
ues obtained for sandwich structure for residual versus impact velocity and perforation energy,
respectively [2]
The ballistic limit for sandwich structure and bare individual parts was obtained by
both experimental and numerical methods and is given in Table 4. As indicated by
Fig. 3, the simulation gives slightly lower value for ballistic limit when compared
with the experimental results. Excellent agreement is observed for bare plate and
foam core, whereas for full sandwich structure, the numerical simulation gives 4.3%
lower ballistic limit than calculated by the experimental values.
548 N. Khaire et al.
Table 3 Experimental and numerical results for sandwich structures individual parts
Target Impact Residual velocity (m/s) Perforation energy (J)
configuration velocity (m/s) Experimental Numerical Experimental Numerical
[2] [2]
Sandwich 186.34 143.34 142.73 49.28 50.45
181.82 139.65 140.07 47.87 47.24
174.22 124.29 128.86 54.99 50.72
157.73 113.30 113.36 44.29 42.33
106.31 46.84 50.07 36.47 35.21
98.88 24.08 27.90 37.48 31.63
77.76 0 0 24.79 27.90
Face plate 105.93 102.23 102.39 3.20 3.06
only
Core only 103.45 65.85 68.08 26.42 25.18
Diameter = 100 mm, Hemispherical projectile (Length = 25.50 mm, Diameter = 7.5 mm)
4 Conclusion
A finite element model was prepared to carry out the validation and understand
the behaviour of sandwich structures, and numerical results were compared against
experimental results available from literature and following were the major findings
of the study:
• It is observed that higher impact velocities, the numerical model is in good agree-
ment with the experimental results, and a difference of 0.05% was observed
while the model numerically over predicts by the maximum difference of 6.9%
between the residual velocities at lower impact velocities when compared to the
experimental values.
• The prepared numerical model was able to successfully capture the physical
phenomenon, and failure modes were in good agreement with the experimental
observations.
The ballistic limit obtained numerically was compared against the experimental
values. A good agreement was observed for bare plates, and bare foams, while there
was a difference 4.50 m/s between experimental and numerical ballistic limit which
accounts for 4.3% error.
Acknowledgements This research was fully supported by SERB/DST under project number
DST/SERB ECR/2016/001440 for providing resources for numerical simulations and experimen-
tation. We thank our colleagues from Visvesvaraya National Institute of Technology, Nagpur, who
provided insight and expertise that greatly assisted the research.
References
1. Zhao H, Elnasri I, Girard Y (2007) Perforation of aluminium foam core sandwich panels under
impact loading—An experimental study. Int J Impact Eng 34(7):1246–1257
2. Hou W, Zhu F, Lu G, Fang DN (2010) Ballistic impact experiments of metallic sandwich panels
with aluminium foam core. Int J Impact Eng 37(10):1045–1055
3. Hanssen AG, Girard Y, Berstad T, Langseth M (2006) A numerical model for bird strike of
aluminium foam-based sandwich panels. Int J Impact Eng 32(7):1127–1144
4. Mohan K, Yip TH, Idapalapati S, Chen Z (2011) Impact response of aluminum foam core
sandwich structures. Mater Sci Eng 529:94–101
5. Elnasri I, Zhao H (2016) Impact perforation of sandwich panels with aluminum foam core: a
numerical and analytical study. Int J Impact Eng 96:50–60
6. Su B, Zhou Z, Zhang J, Wang Z, Shu X, Li Z (2015) A numerical study on the impact behavior
of foam-cored cylindrical sandwich shells subjected to normal/oblique impact. Lat Am J Solids
Struct 12:2045–60
7. Bendarma A, Jankowiak T, Łodygowski T (2014) Experimental and numerical analysis of the
aluminum alloy AW5005 behavior subjected to tension and perforation under dynamic loading.
Theor Appl Mech. 55(4):1–15
Crushing Behavior of Thick Circular
High Strength Aluminum Tube Against
Quasi-static Axial Loading
Abstract The present study investigates the crushing performance of circular tube,
made of high strength aluminum alloy AA-7005 and AA-7075 by numerically and
experimentally. The tubular structures were exposed to quasi-static axial loading
with variation in length (51, 68 and 85 mm) while diameter (34 mm) and thickness
(3.55 mm) kept constant. The uni-axial tension tests for both the material were
carried out to explore the behavior of stress–strain which was used as input for
numerical simulations. The quasi-static compression test was conducted on Instron
compressive testing machine while for simulation finite element code LS-Dyna was
used. The crashworthiness parameters such as initial maximum peak load, crash
force efficiency, energy absorption capacity, and specific energy were found from
the obtained deformation behavior of structures. It is found that the tubular structures
made of AA-7005 show higher crash force efficiency whereas the structures made
of AA-7075 absorb significantly higher energy during the collapse.
1 Introduction
The vehicles are now being widely used for transport purposes. Increasing number
of such vehicles has led in a rise in number of crashes, reflecting a greater safety
requirement. This caused the designer to explore the energy absorbing component
that mitigates the effect of crash or collision by absorbing the impact energy. The
energy absorbers are designed in such way that it undergoes progressive collapse in
order to absorb maximum amount of impact energy. Concertina mode of deformation
reflect a good characteristic for a thin wall tube structures compare with the diamond
and mixed mode. Starting phase of research suggested some theoretical model for
(a) (b)
both concertina mode [1] applicable for a small crushing length and diamond mode
[2] relevant for a large deformation. Earlier lots of studies [3–5] have been done on
conventional circular thin walled tubular structure. Other than circular, researchers
[6–10] have explored different cross-sections such as triangles, squares, rectangles
and polygons. In actual crash incident, the structures collapse under different loading
condition like axial, lateral or bending. In such crash event the impact is dynamic
during which large distortions occur in a short time.
The study of thick tubular structures is limited. In this paper, a thick circular
tubular structure made of aluminum alloy AA-7005 and AA-7075 is introduced. The
crashworthiness performance has studied in terms of deformation modes and load-
displacement curves against quasi-static loading by experimentally and numerically.
2 Experimental Details
The AA-7075 and AA-7005 tubes were used for experiment and numerical test. To
obtain material properties, a tensile test was performed on the samples that were cut
from the tube as per ASTM-E8 standard. Figure 1 shows the true stress–strain curve.
The specimens of the required dimension were cut from both grade of high strength
aluminum tubes, see Fig. 2. The prepared specimen having three height variations
(51, 68 and 85 mm) while diameter and thickness kept constant as 34 mm and
3.55 mm respectively.
Crushing Behavior of Thick Circular High Strength … 553
Table 1 shows the specification of all specimens. The complete set-up for com-
pression testing machine with a load capacity of 5000 kN is shown in Fig. 3a. The
lower part of machine is fixed while upper part moving downward with a constant
rate of quasi-static loading of 6 mm/min.
3 Numerical Modeling
The finite element code LS-Dyna [11] is used to investigate the crushing performance
of thick tubular structures. The complete part of the numerical set-up shows in Fig. 3b.
The modeled tubular structure compressed between two rigid plates. The bottom plate
was fixed while upper plate moving axially downwards at a rate of 6 mm/min. The
circular tube was taken as a solid element of EQ-2, having an element size of 1 mm.
The piecewise linear plasticity material model was assigned to the tube, while the
upper and lower plates were allocated as rigid properties. The contact between the
tube and plate taken as automatic surface to surface while for tube, automatic single
surface was assigned in order to minimize the effect of interpenetration. The friction
554 V. Patel et al.
(a) (b)
coefficients 0.3 and 0.2 considered for the static and dynamic friction, respectively,
between two contact surfaces.
The load-displacement curve describes the structural behavior during the collapse.
The all crashworthiness parameters [Peak Force (PF), Energy absorption (E a ),
and Crash Force Efficiency (CFE)] defining the feasibility of the tubular structure
are calculated on the basis of load-displacement curve. Figure 4 shows the load-
displacement curve of tubular structure made of AA-7005 and AA-7075 for all six
specimens. It can be observed that PF values for AA-7075 specimens are higher than
that of AA-7005 specimens. The higher thickness of specimens resulted in the for-
mation of only one-fold while compressing. The PF indicates the resistance offered
by structure at initial state of collapse. After attained maximum value it decreases
as the plastic hinge formed. The obtained graph from both experimental and numer-
ical observation was approximately close that indicate the correctness of numerical
models.
When compressing the structure, it absorbs energy (E a ) that is calculated by incor-
porating region under the load-displacement curve. The obtained value of experimen-
tal and numerical are compared in Table 2. The percentage error for peak load varies
0.4–14% while for energy it is in the range of 0.4–16%. The specific energy absorp-
tion (SEA) is defined as the amount of energy per unit mass absorbed by specimens
Crushing Behavior of Thick Circular High Strength … 555
(a) (b)
AA-05-01 AA-75-01
AA-05-02 AA-75-02
AA-05-03 AA-75-03
and it should be higher. The fracture failure was observed in AA-75-03 specimens
that reduce the structures’ ability to absorb impact energy.
The CFE, defined as the ratio of mean force to the peak force, value close to unity
is desirable. It can be observed that CFE value for AA-7005 specimens is higher than
that of AA-7075 specimens.
556 V. Patel et al.
Fig. 5 Compressed specimen: a AA-7005, b AA-7075 and c compared deformation mode obtained
through experimental and simulation
For axial compression concertinas are the best deformation mode for higher energy
absorption. But it is only observed in very thin circular columns. With a change
in geometry and thickness, deformation mode changes to diamond folds. Figure 5
shows the deformation behavior of tested samples. The double folding happens for
shorter height while as an increase in height it deforms from the middle that led to
buckling. Fracture failure was observed in AA-7075 specimens.
5 Conclusion
• Increase in peak crushing force was observed for increase in length of speci-
mens. For AA-7005 specimen, the peak load increased by 19.09%, when length
increased from 51 to 85 mm whereas for AA-7075 it increases by 14.71%.
• The crush force efficiency of AA-7005 specimens was consistently better than
the AA-7075 specimens.
• The total amount of energy absorbed by AA-7075 specimens was 16.81% higher
than AA-7005 specimens of the same dimensions.
• AA-7005 specimens had a complete plastic collapse, while AA-7075 specimens
had a plastic collapse, resulting in an undesirable fracture failure.
• Although AA-7075 has a higher strength than AA-7005, the tubular structure
made of AA-7005 performs better than the structure made of AA-7075.
References
1. Alexander JM (1960) An approximate analysis of the collapse of thin cylindrical shells under
axial loading. Q J Mech Appl Math XIII:10
2. Pugsley SA, Macaulay M (1960) The large-scale crumpling of thin cylindrical columns. Q J
Mech Appl Math 8:1
3. Andrews KRF, England GL, Ghani E (1983) Classification of the axial collapse of cylindrical
tubes under quasi-static loading. Int J Mech Sci 25:687–696
4. Jones N, Abramowicz W (1985) Static and dynamic axial crushing of circular and square
tubes. In: Proceedings of symposium on metal forming and impact mechanics, Pergamon
Press, Oxford, p 225
5. Karagiozova D, Jones N Dynamic effects on buckling and energy absorption of cylindrical
shells under axial impact. Thin Wall Struct
6. Reddy et al (2015) Multi-cornered thin-walled sheet metal members for enhanced crashwor-
thiness and occupant protection. Thin Wall Struct 94:56–66
7. Abbasi et al (2015) Multiobjective crashworthiness optimization of multi-cornered thin-walled
sheet metal members. Thin Wall Struct 89:31–41
8. Reyes A, Langseth M, Hopperstad OS (2003) Square aluminum tubes subjected to oblique
loading. Int J Impact Eng 28(10):1077–1106
9. Rossi A, Fawaz Z, Behdinan K (2005) Numerical simulation of the axial collapse of thin-walled
polygonal section tubes. Thin Wall Struct 43(10):1646–1661
10. Zhang X, Huh H (2010) Crushing analysis of polygonal columns and angle elements. Int J
Impact Eng 37(4):441–451
11. Hallquist J (2006) L.-D.K.U.s. Manual, Livermore Software Technology Corporation, Google
Scholar, 2007
Estimation of Burr Dimensions Using
Image Processing for Robotic Deburring
Abstract This paper discusses an effective method to identify the burr location for
2D and 3D workpiece. The proposed method uses an image processing technique
to estimate the dimension of the burr from an image. A set of image processing
algorithms are developed for estimating the location as well as the dimension of the
burr. The burr dimensions are verified using coordinate measuring machine. The burr
data generated are used for planning the deburring trajectory. ABB IRB 120 robot is
used to validate the result experimentally. The result clearly shows the effectiveness
of this approach.
1 Introduction
R. Y. Bhute (B)
Dassault Systèmes Solutions Lab, Pune 411057, India
e-mail: rohinibhute6@gmail.com
M. R. Rahul
Visvesvaraya National Institute of Technology, South Ambazari Road, Ambazari, Nagpur,
Maharashtra 440010, India
dimensions from the image data remains a challenge. In view of this, Idaku [5]
developed a method to acquire three-dimensional image from an advanced structure
light projection at a high frame rate. However, this method requires a sophisticate
camera setup which makes it impractical to implement in an industrial scenario. The
major drawback of these methods is: (1) The burr location and its dimensions cannot
be estimated. (2) The deburring trajectory planning has to be done manually.
Therefore, these methods are not effective in finding the burr location and its
dimensions. In this paper, identification of burr location and burr dimension estima-
tion for 2D and 3D workpiece is discussed. The outline of the paper is as follows:
Sect. 2 discusses the methodology for burr dimension estimation. Section 3 discusses
the trajectory planning for robotic deburring, its simulation, and experimental val-
idation. The result of the proposed method is discussed in Sect. 4. Finally, Sect. 5
discussed the conclusion drawn from the work.
2 Methodology
The proposed method consists of two phases: In phase one, burr thickness estimation
for 2D workpiece is done and phase two estimates the burr thickness and height for
a 3D workpiece. The images are captured using a CCD camera placed at a fixed
height above the workpiece. In the first phase, a subtraction algorithm and thickness
algorithm are developed to extract the burr details. In the second phase, the image
from a mirror setup is employed to estimate the thickness and height of the burr.
The estimated burr data points are used to generate the deburring trajectory. ABB
RobotStudio software is used for simulating the deburring trajectory. The simulated
deburring path is verified on ABB IRB 120 robot with a dummy tool. The workflow
diagram for the proposed method is shown in Fig. 1.
The thickness of burr for 2D workpiece is estimated by using two algorithms, namely
the subtraction algorithm and thickness calculation algorithm. The subtraction algo-
rithm identifies the burr present in the workpiece, whereas the thickness calculation
algorithm estimates the burr thickness in pixel values. The subtraction algorithm
uses two images, a masterpiece image and a machined piece image as shown in
Fig. 2. The images captured using a CDD camera is cropped, filtered, and resized to
make them suitable for subtraction. The masterpiece image is subtracted from the
machined piece image, which will give data of the burr part. These data are used
further for calculating burr thickness.
Once the burr is identified, the burr thickness is estimated using thickness algo-
rithm. The algorithm uses a novel unit matrix search method to estimate the burr
thickens in terms of pixel values. The pixel values are calibrated to get the actual
Estimation of Burr Dimensions Using Image … 561
value of the burr in mm. The burr image is converted to binary image such that ‘1’
represents the presence of burr and ‘0’ represents the standard part or no burr. It is
observed that the thickness of the burr is equivalent to a maximum size of the order
of unit matrix found in the binary image of the burr part. The results generated by
this method are validated by measuring the thickness of the same burr on CMM.
562 R. Y. Bhute and M. R. Rahul
A mirror setup as shown in Fig. 3a is used to estimate the burr thickness and height of
the workpiece. The setup consists of four mirrors placed at 45°, which help in taking
side images of the burr specimen. Thus, an original three-dimensional workpiece
can be transformed into a two-dimensional image. The image captured using CDD
camera is sectioned into five parts namely Lt (Left), Rt (Right), Up (Upper), Lo
(Lower), and C (Central) as shown in Fig. 3b. Up, Lo, Lt, and Rt part of the image
help in estimating the burr height and part C in the image helps in estimating the
thickness of the burr using subtraction and thickness algorithm.
The main objective of this phase is to select the data points of the burr part for robotic
deburring. Once the thickness of the burr is estimated for 2D workpiece, a systematic
approach for data point collection is carried out for trajectory planning. To determine
the number of passes (n) for robotic deburring, thickness of the burr is divided by
the radii of the tool. From the reference edge, all the X-coordinate and Y-coordinate
for the ‘n’ passes are stored and an optimized trajectory is generated for deburring
process.
The thickness and height data are used to plan the trajectory for 3D workpiece.
Prewitt edge detection algorithm is used to detect the inner edge of the workpiece.
Then, the number of passes for the deburring is determined by considering the diam-
eter of the tool. However, multiple edges are detected for 3D workpiece. To find the
optimal deburring trajectory, various correcting techniques like backtracking coor-
dinate elimination, cure smoothening, and curve fitting are carried out. This ensures
the accuracy of the planned trajectory.
The data points generated for deburring for both 2D and 3D workpiece are with
respect to the camera coordinate system. These data points are transferred to robot
coordinate system using a transformation matrix. The transferred data points are
given to ABB RobotStudio software for simulation. The simulation shows that the
generated trajectory is tracked accurately by the robot. The verified data points are
converted into RAPID code, which can be directly given to ABB IRB 120 robot
controller.
To test the feasibility of the proposed method, experimental validation using
dummy tool is carried out on ABB IRB 120 robot. The experimental setup consists of
a CCD camera, MATLAB for data processing, RobotStudio software for simulation
and RAPID code generation, and ABB IRB 120 robot. The workpiece was fixed on
a dummy fixture, and the deburring process was carried out. The experiment shows
that the robot tracked the deburring trajectory accurately.
To estimate the burr thickness and location, various existing methods like pattern
matching, canvas overlapping, and edge detection techniques were used. The pattern
matching method matches the same intensity points of the two images but fails
to match the correct points, whereas canvas overlapping overlaps the two image
and identifies the unmatched part in the image, and however, this method failed as
it replaced the old canvas with a completely new canvas. The third method, edge
detection technique, identified the complete edges and provided good information of
burr but was insufficient to calculate the burr thickness. The subtraction and thickness
algorithm identified the burr as well as estimated the burr thickness from the image
564 R. Y. Bhute and M. R. Rahul
(Fig. 4). The burr dimensions estimated were verified using coordinate measuring
machine, and the average error of 0.8 mm was observed in the dimension.
The simulation result using ABB RobotStudio clearly shows that the robot tracked
the deburring path accurately (Fig. 5).
5 Conclusion
The proposed method discusses an effective technique to identify the burr location
and estimate its dimension for 2D and 3D workpiece. The subtraction algorithm,
thickness algorithm, and height algorithm can effectively estimate the burr dimen-
sions. An effective trajectory planning for the deburring process can be done using
the data points generated by the algorithms proposed in this paper. The simulation
and experimental results clearly show the effectiveness of the proposed method.
Estimation of Burr Dimensions Using Image … 565
References
1. Ryuh B-S, Pennock GR (2006) Robot automation systems for deburring. In: Industrial robotics:
programming, simulation and applications, p 15
2. Li CJ, Park CH, Kyung JH, Chung GJ (2011) Study on teaching path reconstruction algorithm
based direct teaching and playback method. In: Eighth international conference on ubiquitous
robots and ambient intelligence (Incheon 2011)
3. Jinno M et al (1999) Teaching-less robot system for finishing workpieces of various shapes,
Kyongju, South Korea, 17–21 Oct 1999
4. Princely FL, Selvaraj T (2014) Vision assisted robotic deburring of edge burrs in cast parts. In:
12th Global Congress on Manufacturing and Management (GCMM)
5. Idaku I (2012) A coded structured light projection method for high-frame-rate 3D image acqui-
sition. In: Ventzas D (ed) Advanced image acquisition, processing techniques and applications,
14 Mar 2012
Study of Structural and Mechanical
Behaviour of Severe Plastically Deformed
Al–Mg(AA 5052) Alloy Processed
by Constrained Groove Pressing
Technique
Jaya Prasad Vanam, Vinay Anurag Potnuri, and Sree Vidya Sravya Nallam
Abstract This paper discusses the effects of Constrained Groove Pressing (CGP)
on Aluminum–Magnesium(AA 5052) alloy specimens at room temperature. CGP
is one of the severe plastic deformations technique (SPD) by which Ultra Fine-
Grained (UFG)/Plane metallic materials can be processed. A comprehensive study
is made on the structural and mechanical properties of the Aluminum specimen before
and after constrained grooves pressing. The entire process is simulated in AFDEX
CAE Software. Further, simulated results of differently oriented workpieces (90°
and 180°) which are processed by the CGP technique are compared. It is found that
most of the properties are superior to Aluminum samples such as Yield Strength,
Ultimate Tensile Strength, Hardness, Strain rate, etc. are found to be better for the
CGP processed specimen. The results are discussed with respective graphs.
1 Introduction
2 Experimentation Procedure
In the present study, AA 5052 alloy sheet with sample size 85 × 85 × 3 mm3 was
used. Further sample was deformed in between corrugated and flat dies alternatively
as per Fig. 1 (Table 1).
In the first pass, the die is placed in between the corrugated die Fig. 2a with
the slightest offset and deformed in between the dies. Then the deformed sheet is
removed and kept in between the flat pair of dies and now it will be flattened as per
Fig. 2b. After first flattening, now the sample is rotated by 90°/180° about the axis
which is normal to plane of the sample specimen. Here by rotating the specimen, the
un-deformed areas are in between the dies for better stress distribution [7].
Further then flattened work piece is again deformed in between the corrugated
dies as per Fig. 2c. Followed by flattening the deformed sample after pass 3 in next
consecutive pass Fig. 2d.
Theoretical equations for calculating shear and effective strain accumulated after
each pressing induces an engineering strain [1]
Table 1 Chemical
Material Composition (%)
composition of AA 5052
alloy Mg 2.2–2.8
Cr 0.15–0.35
Fe 0.0–0.40
Si 0.0–0.25
Others 0.0–0.15
Cu 0.0–0.10
Zn 0.0–0.10
Mn 0.0–0.10
Other (each) 0.0–0.05
Al Remainder
Fig. 2 a First pressing of corrugated specimen, b Second pressing of flat specimen, c Third pressing
of flat specimen with 90° rotation, d Fourth pressing of specimen with flat dies
where H, T and θ are groove height, width and groove angle, respectively. Shear
strain is given as
εx y = γx y /2 (2)
570 J. P. Vanam et al.
Since CGP is assumed as a pure shear deformed under plane strain condition,
correspondingly, the effective strain [4]
4(εx y)2 tan θ
εeff = = √ (3)
3 3
2 tan θ
εtotal = n √ (4)
3
Our corrugated dies were designed and manufactured with groove angle θ is 45°
and with equal groove width (T ) and height (H) which theoretically√concludes the
engineering shear strain γ xy = 1, and the effective strain εeff = 1/ 3 ≈ 0.58 1st
pass). At the end of first flattening (Pass 2), the sheet has a uniform strain magnitude
of 1.16 [8].
3.1 Simulation
Dies are modelled in CATIA V5 R20 and Simulation (FEA) is carried out in AFDEX
(Advisor for metal forming process Design Expert) software which is for the first time
incorporated for present work simulation which is a forming tool which widely used
in industries. AFDEX reduces the complexity of explicit analysis of ANSYS and
continuous simulation of DEFORM 3D, Entire Process was simulated seamlessly
with minimum effort.
Mechanical properties when work piece is oriented 90° and 180° at the end of the
fourth pass.
Here in this case after the first flattening, i.e., after the second pass, the work piece was
rotated by 90° [9] to deform the un-deformed areas of specimen which are strained
in particular direction in the earlier pass. After the four passes, different mechanical
Study of Structural and Mechanical Behaviour of Severe … 571
properties are varied in the following trend. These are the respective simulation results
after all four consecutive stages don in AFDEX software. Mechanical properties
varied in the following fashion. Effective strain rate at the end of four stages is 4.49
from Fig. 3a. In the study conducted earlier by Satheesh Kumar and Raghu [10]
strain rate can be up to 5.8.
Hardness variation in the above software is given by Brinell hardness load in
the simulated studies variation in terms of kgf/mm2 can be converted to N/mm2 by
multiplying above result with 9.8. Total hardness variation after four passes increased
by 1.42 times from Fig. 3b and according to Kia-Huai Yang and Ze-Chang hardness
may increase up to 1.51 [6] times the initial hardness value.
Strength varied up to 194 mpa as showed in Fig. 3c. Here the load local min and
max value represents respective yield strength and Ultimate tensile strength at that
particular pass. Both Yield and Ultimate strength are showed improvement at the
end of every pass which is discussed in the coming graphs.
In this case after the first flattening, i.e. second pass the specimen will be rotated
180° [6] normal to plane of the specimen to deform the un-deformed areas in the
specimen which are present in the earlier passes. Variation of mechanical properties,
in this case, are as follows: effective strain rate after pass 4 is 4.6 in this case as per
Fig. 4a.
Hardness value varied in terms of Brinell hardness load in AFDEX and after pass
4 hardness is increased by 1.41 times. The variation of each pass is plotted as a graph
and will be discussed with the help of graphs.
572 J. P. Vanam et al.
Yield strength and ultimate tensile strength are improved significantly after four
passes and they are shown in Fig. 4c and variation is discussed in the coming graphs.
700
600
Yield Strength(Mpa)
500
Brinell Hardness Load (N/sq.mm)
400
Effective strain Rate
300
Ultimate Tensile Strength(Mpa)
200
100
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
CGP Stages
(b) AA 5052 Property Variation under CGP with Work piece orientation 180o
1600
Mechanical Properties Variation
1400
1200
Yield Strength (Mpa)
1000
Brinell Hardness load(N/sq.mm)
800
Effective Strain rate
600
Ultimate Tensile Strength(Mpa)
400
200
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
CGP Stages
Fig. 5 a Mechanical properties variation for 90° orientation of work piece, b mechanical properties
variation for 180° orientation of work piece
Yield strength variation is nominal and increasing after every pass until pass 4 in
the 180° oriented CGP process but the intensity of yield strength is 150% more in case
of 180° orientation. In Fig. 5a, b, yield strength at the end of pass 4 of 90° is almost
identical to yield strength at the end of pass 2 of 180° case of specimen. Ultimate
tensile strength variation is linear up to pass 3 then variation is more dynamic from
pass 3 to pass 4 in 90° specimen orientations. In the 180° Specimen orientation is the
ultimate tensile strength variation is linear up to pass 1 but after that improvement in
UTS is very dynamic from pass 2 to pass 4 which can be observed from Fig. 5a, b.
So after studying the graphical variation of above two simulated cases, the inten-
sity of property improvement is significant in the case of 180° than 90° even at almost
similar strain rates, From the simulated results, in the case of 90° specimen orien-
tation, improvement in properties is limited because in the pass 3 s corrugation, i.e.
574 J. P. Vanam et al.
4 Conclusion
In this current work, AA 5052 alloy sheets were subjected to an SPD method called
CGP and cumulative strains imposed after four passes on the sheet is 4.6. Also,
the effect of CGP on Mechanical Properties variation like Yield Strength, Ultimate
Tensile Strength (UTS), and Hardness was investigated with respect to our simulation
and experimental results. Then the main findings of our study can be presented as
follows:
1. Ultimate Tensile Strength of AA 5052 alloy is increased by increasing number
of passes in the CGP process. Maximum rate is during the initial stage (first
pressing) and further rate of increase is paced down with increasing deformation
in case of both 90° and 180° orientation of work piece.
2. The improvement of Yield and UTS of sample specimen is more dominant is
180° orientation of work piece which has been rotated to deform un-deformed
zone of sample specimen rather than the 90° oriented work piece because in
this case, the deformation is perpendicular to its earlier case, so it is limiting the
property improvement like in case of 180° oriented work piece.
3. The average Brinell hardness load increased by a factor of approximately 1.42
after the second flattening or pass 4.
4. Strain rate for 90° is varied with 5% when compared to 180° so strain rate is
almost similar for both the specimen orientation.
References
1. Wang ZS, Guan YJ, Wang GC, Zhong CK (2015) Influences of die structure on constrained
groove pressing of commercially pure Ni sheets. J Mater Process Technol 215(1):205–218
2. Valiev RZ, Alexandrov IV (2000) Nanostructured materials produced by severe plastic
deformation. Logos, Moscow
3. Dobatkin SV (2000) Severe plastic deformation of steels: structure, properties and techniques.
In: Lowe TC, Valiev R (eds) Investigation and applications of severe plastic deformation,
Kluwer, Netherlands, vol 3, p 13
4. Shin DH, Park JJ, Kim YS, Park KT (2002) Constrained groove pressing and its application to
grain refinement of aluminium. Mater Sci Eng 328A:98–103
5. Huang JY, Zhu YT, Jiang H, Lowe TC (2001) Microstructures and dislocation configurations in
nanostructured Cu processed by repetitive corrugation and straightening. Acta Mater 49:1497–
1505
Study of Structural and Mechanical Behaviour of Severe … 575
6. Yang K, Zou Z, Zeng J, Chen W (2018) Microstructures evolution and mechanical properties
of 5052 aluminium alloy processed by constrained groove pressing. Int J Comput Mater Sci
Surf Eng 7(3):4
7. Krishnaiah A, Chakkingal U, Venugopal P (2005) Applicability of the groove pressing tech-
nique for grain refinement in commercial purity copper. Mater Sci Eng 410–411(A):337–340
8. Nagaraju KN, Sunil AR, Sachin K, Sujay H, Siddesha H, Anand Kumar S (2018) Influence of
constrained groove pressing passes and annealing characteristics on the mechanical properties
of 6061 aluminum alloy
9. Soon Fong K, Jen Tan M, Lan Ng F, Danno A, Wah Chua B (2017) Microstructure stability
of a fine-grained AZ31 magnesium alloy processed by constrained groove pressing during
isothermal annealing. J Manuf Sci Eng 139(8):081007
10. Satheesh Kumar SS, Raghu T (2011) Tensile behaviour and strain hardening characteristics of
constrained groove pressed nickel sheets. Mater Des 32:4650–4657
Shear Rate Dependent Frictional
Behavior of the Granular Layer
Abstract The present experimental study investigates shear rate dependent fric-
tional properties of the granular layer between two hard surfaces. Slide-free-slide
(SFS) experiments were performed on the layer in direct shear mode. It is observed
that static stress increases with both normal stress as well as shear velocity. The
Mohr–Coulomb (MC) failure criterion is used for determining adhesive stress as
well as coefficient of friction of the sliding interface. Both components of friction
increase with shear velocity. Their scaling laws, in terms of shear velocity, reveal
that the Coulombic friction is more pronounced over the adhesive friction and these
results are also justified.
1 Introduction
A granular material is a system of discrete particles and the physical and mechanical
properties of such material depend on distribution and voids among the particles [1].
Granular materials find applications in sugar, salt, food grains, and pharmaceutical
industries and the examples include sand, soil, rock, gouges, etc. [2]. Motivated by
the interesting frictional properties of the granular solids, we perform the experiments
for understanding the frictional properties of the granular layer in low shear velocity
conditions. The aim of the present work is to understand the shear rate dependent
frictional characteristics of the granular solid sandwiched between two hard surfaces.
Krantz has performed sliding tests on the mixture of sand and cement, sand and
clay and dry quartz sand and reported that the coefficient of friction and cohesion
increase with density of the materials [3]. Schellart has also studied frictional behav-
ior of sugar grains, glass particle, and quartz and established that shape such as
rounding and sphericity of the granular particles, instead of their size, determine
their frictional properties [4]. Duttine et al. investigated the frictional properties of
the glass and natural sand particles having varying sizes and their observations show
the rate-independent sliding behavior [5]. Recently, van den Ende et al. have exper-
imentally determined the cohesion and frictional properties of an earthquake fault
during the interseismic period [6]. Nasuno et al. have explored experimentally the
stick-slip behavior of glass particles under the condition of low sliding velocity [7].
Geminard et al. have reported that sliding velocity has a negligible effect on steady
friction of the saturated granular layer consists of the glass beads [8]. Lieou et al.
proposed a mechanistic model for the gouge materials subject to acoustic vibrations
and shear deformation [9]. Morgan used the particle dynamics method for simulating
the frictional behavior of the gouge layer for validating the rate and state friction laws
[10]. Dorostkar et al. have studied the effect of liquid contain on frictional properties
of wet granular layer [11]. Frye and Marone have found in their experiments that
coefficient of friction is independent of relative humidity, but at high humidity, there
is a transition from velocity weakening to velocity strengthening as shear velocity
increases [12]. Guo and Julia have experimentally studied the effect of size and
shape of the particle of granular material on frictional properties of granular mate-
rial [13]. Marone has reviewed the frictional behavior of gouge materials in view of
the rate and state friction laws [14]. Further, if friction increases as sliding velocity
increases, the phenomenon is known as velocity strengthening (VS) process. While
friction decreases as sliding velocity increases, this is known as velocity weakening
(VW) process and stick-slip instability occurs in the VW regime [14]. In the present
study, friction experiments were performed in VS regime to understand the steady
frictional properties of thin granular layer.
According to Mohr–Coulomb (MC) friction criterion, shear strength arises due
to the cohesion and Coulomb friction [14, 15]. In other words, frictional stress τ =
c + σn μ, where σn and μ are effective normal stress and coefficient of friction,
respectively. Further, c is generally characterized as cohesive stress [16]. However,
in the case of sliding interface forming dissimilar solids for instance soil and rock
surfaces, the MC failure criterion takes the form as τ = a +σn μ, where a is adhesion
and μ is corresponding interfacial coefficient of friction [15]. The significance of
this simple model is that role of both Coulombic, as well as non-Coulombic friction
(adhesive) could be determined between the granular material and rock surface in a
single test. The objective of the present study is to understand the effect of normal
stress and shear rate on frictional behavior of granular layer.
2 Experimental Set Up
Figure 1 shows the schematic sketch of the experimental set up used in the present
study. The experimental set up is based on principle of direct shear sliding [15]. The
lower surface is fixed while the upper block acts as the slider. Test sample consists
of 10 g of black cotton soil having grain size of 425 micron sheared between two
Shear Rate Dependent Frictional Behavior of the Granular Layer 579
rough granite blocks. In order to maintain the thickness of the layer, a fixed amount
of granular material was spread over a fixed area (30.0 cm2 ) of the substrate. The
normal load was varied in the experiment in range of 10–50 N. A load cell (50 N)
connected with data acquisition system (DAS) was used for measuring the friction
force at the sampling rate of 10. Although the hold time was fixed for 60 s, the
external shear velocity V0 was varied from 0.2 to 1.0 mm s−1 in direct shear test
using slide-free-slide (SFS) test [15].
Figure 2 presents the relation between static frictional stress τs versus normal stress
σn . The linear relationship between τs versus σn is linear, thus validating the MC
criterion. At the same time, the slope of the straight signifies coefficient of friction,
which intercepts adhesive stress. It is also observed from the plots in Fig. 2 that τs
increases with v0 at a fixed σ n .
Figure 3a presents the scaling law between adhesive stress as and shear velocity
vo for static friction as as ~ v00.012 and Fig. 3b, on the other hand, shows the scaling
Fig. 3 shows the scaling law between adhesive stress as versus v0 as as ~ v00.012 in (a) as well as
static coefficient of friction μs versus v0 as μs ~ v00.1244 in (b)
law between static coefficient of friction μs and shear velocity v0 for as μs ~ v00.124 .
It is seen that as increases with increase in v0 . Experiment results also show the
exponent of as is lower in magnitude than corresponding μs . In other words, frictional
properties of the wet granular layer still arises mainly due to the Coulomb stress (hard
solid friction) than the adhesion process.
The rate-dependent behaviors of the granular solids could be explained using the
Arrhenius theory of rate reaction. This theory predicts the variation of shear strength
as logarithm of shear velocity [14]. Future studies include hold time-dependent adhe-
sion and frictional properties also validating the rate and state friction laws on granular
materials at low normal stress [6, 15].
4 Conclusion
The present experiments show that the magnitude of adhesion and friction coefficient
increases with shear velocity. Static stress follows the increasing trend with normal
stress. The exponent of adhesive stress with sliding velocity is larger in magnitude
than corresponding to the coefficient of friction. The reason for this observation may
be attributed to less surface energy of the granular layer.
References
1. Campbell CS (1990) Rapid granular flows. Annu Rev Fluid Mech 22(1):57–90
2. Wornyoh EY, Jasti VK, Higgs CF (2007) A review of dry particulate lubrication, powder and
granular materials. J Tribol 129(2):438–449
3. Krantz RW (1991) Measurements of friction coefficients and cohesion for faulting and fault
reactivation in laboratory models using sand and sand mixtures. Tectonophysics 188(1-2):203–
207
4. Schellart WP (2000) Shear test results for cohesion and friction coefficients for different gran-
ular materials, scaling implications for their usage in analogue modelling. Tectonophysics
324(1–2):1–16
5. Duttine A, Tatsuoka F, Kongkitkul W, Hirakawa D (2008) Viscous behavior of unbound granular
materials in direct shear. Soils Found 48(3):297–318
6. Van den Ende MPA, Niemeijer AR (2019) An investigation into the role of time-dependent
cohesion in interseismic fault restrengthening. Sci Rep 9(1):9894
7. Nasuno S, Kudrolli A, Bak A, Gollub JP (1998) Time-resolved studies of stick-slip friction in
sheared granular layers. Phys Rev E 58(2):2161
8. Géminard JC, Losert W, Gollub JP (1999) Frictional mechanics of wet granular material. Phys
Rev E 59(5):5881
9. Lieou CK, Elbanna AE, Carlson JM (2016) Dynamic friction in sheared fault gouge:
implications of acoustic vibration on triggering and slow slip. J Geophys Res Solid Earth
121(3):1483–1496
10. Morgan JK (2004) Particle dynamics simulations of rate-and state-dependent frictional sliding
of granular fault gouge. In: Computational earthquake science part I, pp 1877–1891
11. Dorostkar O, Guyer RA, Johnson PA, Marone C, Carmeliet J (2018) Cohesion induced sta-
bilization in stick slip dynamics of weakly wet, sheared granular fault gouge. J Geophys Res
Solid Earth 123(3):2115–2126
12. Frye KM, Marone C (2002) Effect of humidity on granular friction at room temperature. J
Geophys Res Solid Earth 107(B11):ETG-11
13. Guo Y, Julia KM (2004) Influence of normal stress and grain shape on granular friction, Results
of discrete element simulations. J Geophys Res Solid Earth 109:B12
14. Marone C (1998) Laboratory-derived friction laws and their application to seismic faulting.
Annu Rev Earth Planet Sci 26(1):643–696
15. Koupouli NJ, Belem T, Rivard P, Effenguet H (2016) Direct shear tests on cemented paste
backfill–rock wall and cemented paste backfill–backfill interfaces. J Rock Mech Geotech Eng
8(4):472–479
16. Debasis D, Kumar VA (2016) Fundamentals and applications of rock mechanics. PHI Learning
Pvt. Ltd.
Mathematical Overview on
Omnidirectional Spherical Wall
Traversing Robot
Abstract Since the past few decades, many mechanisms are being developed for
wall climbing robots. Robotic systems having omnidirectional surface traversing
ability independent of its inclination require complex morphology transformation
for a floor to wall transition as well as perfect adhesion on the vertical surface.
This paper depicts the development of the mathematical model for omnidirectional
maneuverability of the robot. The novel design of this robot is newly aimed, and its
motion is mathematically modeled.
1 Introduction
Robots having ability to climb the vertical surfaces are categorized into wall climbing
robots. Wall climbers have been developed across the globe for surveillance, inspec-
tions, flaw detection, cleaning, maintenance, etc. The proposed mechanism has a 2
DOF bi-propeller plane which allows getting any direction of thrust on to the wall
which helps it to move freely and remain attached to the surface.
As the working principle of the proposed Omni-directional Spherical Wall
Traversing Robot is based on propeller-based thrust, hence its performance is inde-
pendent of surface parameters, thus outspreading the competence of wall climbers.
The proposed design is resulting in robustness and simplified mobility of the robot.
2 Historical Overview
Various designs of wall climbing robots have been developed in the past few years.
These robots are primarily built keeping in mind that they can be used for rescue,
wall inspection, and firefighting [1]. Some of them were also used in nuclear plants,
construction industries, military areas, etc. [2]. The challenging part of these robots
is the adhesion mechanism which sticks the bot on the surface of the wall. One
technique is to use vacuum suction to produce the adhesive force. However, this
method fails miserably if the surface is uneven and rough or if there are obstacles
on the path. This is because, in order to have maximum contact force the surface
must be smooth, any loose contact will make the bot to fall down [3]. The other idea
is to use magnetic/electromagnetic wheels on the bot, but then, this will work only
on iron, steel, and other ferromagnetic structures. Another mechanism is to use dry
elastomer adhesives on the legs of the bot [4]. However, this technology is currently
not mature enough to be used for climbing robots. Lastly, there is one mechanism in
which the thrust generated by the propellers is used to drive the robot on the surface
of the wall. Comparing with other mechanisms, the bot moves fastest in this one. It
does not depend on the properties of the wall and can work on uneven surfaces also.
Including a pair of coaxial propellers mounted on a gimbal structure increases the
degrees of freedom of the bot and helps in the steering of the bot on the wall [5].
3 Mathematical Overview
The coaxial bi-propeller mechanism consists 1045 (10 × 4.5”) ABS Propellers
(1CW + 1CCW-1 pair Black) and a couple of A2212 2200KV BLDC Motors. Thrust-
angular velocity characteristics are given by the Eq. 1 (Fig. 1)
Perforations are made onto the spherical shell to pass the air required for producing
thrust. As propulsion thrust is directly proportional to the area of the propeller plane,
perforated shell hinders some factor of the thrust; the thrust hindrance function ζ is
introduced to include this problem into account. Due to non-uniform thrust hindering
factors into the space of the spherical robot, the thrust hindrance function will depend
upon space angular space coordinates, i.e., ζ (α, β, γ ). Therefore, the thrust vector
is given by
F = f · ζ (α, β, γ ) · [cos(α) cos(β) cos(γ )]T (2)
Assuming equilibrium position of the sphere during the motion heading in a direction
making angle φ with the positive x-axis, the frictional force fr depends upon wall and
sphere surface material characteristics. μws is the kinetic friction coefficient between
two touching surfaces of wall and robot [6]. On resolving into three-axis components,
the resulting thrust is given by the equation.
⎡ ⎤ ⎡ ⎤
cos(α) fr · cos(φ)
F = f · ζ (α, β, γ ) · ⎣cos(β)⎦ = ⎣fr · sin(φ) + mg ⎦ (3)
cos(γ ) − μfrws
Transformation matrix for rotation of plane with angle θ about axis with unit
vector [l m n]T is given by,
⎡ ⎤
l 2 · (1 − cθ ) ml · (1 − cθ ) nl · (1 − cθ )
⎣lm · (1 − cθ ) m 2 · (1 − cθ ) nm · (1 − cθ )⎦
ln · (1 − cθ ) mn · (1 − cθ ) n 2 · (1 − cθ )
⎡ ⎤
cθ −n · sθ m · sθ
+ ⎣ n · sθ cθ −l · sθ ⎦
−m · sθ l · sθ cθ
The 2DOF gimbal mechanism is used to control the direction of the thrust. This
mechanism is made of two rotating motors with perpendicular axes of rotation. Let,
angle of rotation of inner motor be θin about axis [1 0 0]T , and an outer motor is a
θout about axis [0 0 1]T ; hence, transformation matrices of both individual rotations
are given by: ⎡ ⎤
1 0 0
T rin = ⎣0 cos(θin ) −sin(θin )⎦ (4)
0 sin(θin ) cos(θin )
⎡ ⎤
cos(θout ) −sin(θout ) 0
T rout = ⎣ sin(θout ) cos(θout ) 0⎦ (5)
0 0 1
Hence, thrust vector with direction controlled by gimbal angles is given as:
⎡ ⎤
cos(θin ) × sin(θout )
F = f · ζ (α, β, γ ) · ⎣cos(θin ) × cos(θout )⎦
sin(θin )
Mathematical Overview on Omnidirectional Spherical Wall … 587
⎡ ⎤
fr · cos(φ)
= ⎣fr · sin(φ) + mg ⎦ (6)
− μfrws
Let the robot is initially located at point (x1 , y1 ) in the x-y plane which is supposed
to reach point (x2 , y2 ). Therefore, a linear velocity of the bot is taken proportional to
the instantaneous Euclidean distance between its position and the destination. As it
is an accelerated motion (with acceleration ‘a’) pseudo force (ma) is taken opposite
to the direction of motion. Kinetic frictional force fr is opposing the motion. Let
∴ vx y = η
· sec(φ) × [sin(φ) cos(φ)]T (7)
∴ a = η2
· sec(φ) × [sin(φ) cos(φ)]T (8)
⎡ ⎤
fr · cos(φ) − mη2
Arduino Nano V3.0 CH340 Chip is used as CPU of the system. Figure 3 shows
the overview of the closed-loop control system of the bot. The angular velocity
of the thruster is measured using a Hall effect sensor (BMES A3144 Hall Effect
Magnetic Sensor Module) and given as a feedback PID controller used [3] with self
tuning ability given in [7]. Similarly, two PID controllers are used to control gimbal
angles. Tilt angles are calculated using potentiometers attached to the motor shafts
and given as a feedback to the PID controller. Pairs of servomotors in gimbal are
operated synchronously with opposite angels.
5 Future Work
Acknowledgements The research work of this project is supported and funded by IvLabs, Robotics
lab of Visvesvaraya National Institute of Technology, Nagpur, India.
References
1. Nishi A (1991) A wall climbing robot using propulsive force of propeller. In: 5th international
conference on Advanced Robotics. Robots in Unstructured Environments, vol 1. Pisa, Italy,
320–325. https://doi.org/10.1109/ICAR.1991.240633
2. Briones L, Bustamante P, Serna MA (1994). Wall-climbing robot for inspection in nuclear
power plants. Proceedings IEEE international conference on Robotics and Automation, vol 2,
1409–1414. https://doi.org/10.1109/ROBOT.1994.351292
3. Liu G, Liu Y, Wang X, Wu X, Mei T (2016) Design and experiment of a bioinspired wall
climbing robot using spiny grippers. In: 2016 IEEE international conference on Mechatronics
and Automation. Harbin, 665–670
4. Murphy MP, Sitti M (2007) Waalbot: an agile small-scale wall-climbing robot utilizing dry
elastomer adhesives. IEEE/ASME Trans Mechatron 12(3)
5. Tanaka Y, Nozaki K, Ioi K (2017) Motion control of a wall climbing robot with coaxial propeller
thruster. In: 2017 2nd IEEE international conference on Intelligent Transportation Engineering
(ICITE), 2017, 360–364
6. Schmidt D, Hillenbrand C, Berns K (2011) Omnidirectional locomotion and traction control
of the wheel-driven, wall-climbing robot. Cromsci Robot 29:991–1003
7. Babu VM, Das K, Kumar S (2017) Designing of self tuning PID controller for AR drone
quadrotor. In: 2017 18th international conference on Advanced Robotics (ICAR), Hong Kong,
167–172, 2017. https://doi.org/10.1109/ICAR.2017.8023513.
8. Lee G, Kim H, Seo K, Kim J, Kim HS (2015) Robotics and autonomous systems multitrack: a
multi-linked track robot with suction adhesion for climbing and transition, vol 72. https://doi.
org/10.1016/j.robot.2015.05.011
9. Ma B, Liu R, Zhang R, Tian L, Nashunbuhe (2007) Design of wall climbing robots with
transition capability. In: IEEE international conference on Robotics and Biomimetics, 15–18
December
Finite Element Analysis of Ballistic
Impact on Monolithic and Multi-layered
Target Plate with and Without Air Gap
Abstract In the present work, Finite Element Analysis is applied to analyze the
ballistic impact on 1100-H14 aluminum and weldox 460 E steel multi-layered plate
using FEM package ABAQUS/CAE explicit. A conical projectile is projected on
the plate with different velocities. The 1100-H14 aluminum plate is model using
Johnson–Cook material modeling and the Bao-Wierzbicki failure model is used for
fracture. The Cut-off on negative triaxiality has been incorporated in the present
work. The values of material parameters such as elastic and plastic are taken from
the literature and the projectile is assumed to be rigid. Fracture pattern of the plate,
residual velocity, and velocity drop on different thicknesses of plates are calculated.
The number of petals formed in the plate after the fracture has been reported and
the maximum deformation experienced by the plates are studied. It is found that
the fracture pattern by the numerical analysis on the plate is almost similar to the
experimental result as reported in the literature. And also the ballistic performance
of multi-layered metal plates with and without spaced, subjected to impact by blunt
projectile is investigated by numerical simulation. Further, the effect of the air gap
on ballistic resistance is investigated. Ballistic limit velocities of layered plates are
decreased with multi-layered target. The residual velocity, ballistic limit velocity
and perforation time are determined. The result also showed that there is a consistent
increase in ballistic resistance of target as the number of layered is increased.
1 Introduction
over a relatively longer timeframe. A large number of works have been carried out
with impact modeling, impact damage assessment and the assessment of post-impact
residual properties. Borvik et al. [1] demonstrated the impact of nose shape (blunt,
hemispherical, conical) projectile on the impact of weldox 460E steel plate of thick-
ness 12 mm. They found that ballistic cut-off speed for hemispherical and conical
were found near one another yet for blunt shot is lower. Kurtaran et al. [2] simu-
lated the ballistic effect on plate of 2 mm thickness by projectiles with hemispherical
rounded nose of 7.62 mm distance across moving with speeds in the range of 500,
1000, 1500 m/s. They saw that the influence of thermal softening effect must con-
sider for reproduction of better outcomes. Gupta et al. [3] considered the effect of
shot nosed shape and aluminum plate of thickness in the scope of 0.5–3 mm both
experimentally and numerically. It was seen that ogival projectile were the best pen-
etrators for exceptionally thin plate thickness and blunt projectile for 3 mm thick
plate thickness. Teng et al. [4] have shown the ballistic performance of doubled lay-
ered steel defensive plates against projectile effect utilizing finite element analysis by
conical-shaped projectiles. Iqbal et al. [5] analyzed the effect of projectile nose shape.
Aluminum plates were influenced by steel projectiles having double nose shape viz.
conico–blunt, blunt–blunt, blunt–conico. It was seen that failure mode was exposed
to nose shape, and was particularly not exactly the same as that saw with single nose
projectile. Kumar et al. [6] examined simulation using continuum damage mechan-
ics (CDM) model on ballistic effect of steel target impacted by rigid projectiles. The
continuum damage model was made by them realized in ABAQUS/Explicit through
a client-specified material model subroutine (VUMAT). The attain of temperature
and strain rate were incorporated. Zhang et al. [7] thought about the ballistic impact
of monolithic and multi-layered steel plates affected by blunt rigid projectile. They
examined the impact of air gap between layers, the number, order and thickness
of layers on the ballistic resistance. Corran et al. [8] done a no. of tests on the
multi-layered in contact plate and found that layered plates in contact were superior
to relative monolithic plate. Gupta et al. [9] drove a movement of perforation trail
of single- and three-layered metallic objectives with practically identical thickness
against level, ogival, and hemi-circular nosed steel projectile. The results exhibited
that when the amount of layers were exhausted, the speed drop was viewed as higher
than one record of single plate. Ogival nosed shots were viewed as the most prof-
itable penetrator of layered objective. Hemi-circular nosed shot requires the most
outrageous vitality for perforating the target.
2 Methodology
Cut-Off Value of Triaxiality—It is reported that fracture did not start for the
triaxiality esteem not exactly (−1/3). In this manner, in the present work, damage
increment is considered zero for triaxiality esteems not exactly (−1/3).
Criteria for Element Deletion—A basic critical damage criteria for element
delete, at whatever point the fracture strain in an element arrives a critical value, the
element is deleted. The criteria is executed over the entire space of workpiece.
The simulation of blast impact is performed to study the failure of aluminum plate
by steel projectile subjected to normal impact on different target thicknesses. The
impact velocity and coefficient of friction are kept 100 m/s and 0.05, respectively.
Single nosed cylindrical projectile was used by the researcher on the aluminum plate
of 0.82 mm thickness. When the bullet strikes on the plate this type of deformation
generated. The number of petals are developed exactly the same in both experi-
mentally and numerically as shown in Fig. 4. The fracture pattern is matched with
experimental result of Iqbal et al. [7].
The residual velocity of the projectile is found to be 87 m/s. The fracture pattern
of 1.82 mm thick target is shown in Fig. 5. It is found that the number of petals is
Fig. 4 Shearing failure is occurred on 0.82 mm thick target impacted a experimental [7] and
b numerical results
596 R. Kumar et al.
seven, when the conical projectile strikes on the surface of the plate. A small crack is
developed on the plate and equivalent stress is very high around the projectile. The
red color on the target plate shows the high-stress zone. After some step time, the
projectile perforates the plate and on the other hand caused failure through ductile
hole enlargement and petal formation. The number of petals is more with respect to
first plate thickness is because the strength of the second plate is more.
Multi-layered Without Air Gap—The objective plate of steel with each layer
of 3 mm has been taken. Here two layers are jointed with considering friction (μ =
0.1) between them. What’s more, the diagram is plotted in between Von-mises stress
and equivalent plastic strain. And ballistic resistance is investigated (Figs. 6 and 7).
Multi-Layered Target with Air Gap of 1.5 mm—The target plate of steel with
each layer of 3 mm with air gap of 1.5 mm has been taken. Two layers are joined
together with air gap of 1.5 while considering friction (μ = 0.1) between them. The
diagram is plotted in between von-mises stress and equivalent plastic strain, and also
ballistic resistance is investigated (as shown in Fig. 8).
Fig. 10 Graph impact velocity versus residual velocity, i.e., (ballistic resistance)
4 Conclusions
The number of petals formed in the plate of thickness 0.82 and 1.82 are four and
seven, respectively. The deformation in the thick plate is more because more petals
are formed. The residual velocity of the bullet is large in the case of thin plate.
The projectile velocity decreased continuously after penetration of target plate. It is
additionally seen that the equivalent stress, equivalent plastic strain, triaxiality and
strain rate are most elevated in the region of contact between the projectile and target.
Projectile caused failure mode of frontal plate is bulging. Whereas on the back plate
is bulging and dishing which include bending and extending. For impact velocity
500 m/s, the residual velocity in without and with air gap is found 352.84 m/s and
357.26 m/s, respectively. The velocity drop in case of multi-layered target without
air gap is more. So The ballistic resistance is more in the case of multi-layered target
without air gap. Air gap reduces the velocity drop that is why it is having less ballistic
resistance.
References
1. Borvik T, Langseth M, Hopperstad OS, Malo KA (2002) Perforation of 12 mm thick steel plates
by 20 mm diameter projectiles with flat, hemispherical and conical noses part I: experimental
study. Int J Impact Eng 27:19–35
2. Kurtaran H, Buyuk M, Eskandarian A (2003) Ballistic impact simulation of GT model vehicle
door using finite element method. Theoret Appl Fract Mech 40(2):113–121
3. Gupta NK, Iqbal MA, Sekhon GS (2007) Effect of projectile nose shape, impact velocity and
target thickness on deformation behavior of aluminum plates. Int J Solids Struct 44(10):3411–
3439
4. Teng X, Wierzbicki T, Huang M (2008) Ballistic resistance of double-layered armor plates. Int
J Impact Eng 35(8):870–884
5. Iqbal MA, Khan SH, Ansari R, Gupta NK (2013) Experimental and numerical studies of
double-nosed projectile impact on aluminum plates. Int J Impact Eng 54:232–245
6. Kumar M, Deep U, Dixt PM (2017) Simulation and analysis of ballistic impact using continuum
damage mechanics (CDM) model. Procedia Eng 173:190–197
Finite Element Analysis of Ballistic Impact on Monolithic … 599
7. Wei Z, Yunfei D, Sheng CZ, Gang W Experimental investigation on the ballistic performance
of monolithic and layered metal plates subjected to impact by blunt rigid projectiles. Int J
Impact Eng
8. Corran RSJ, Shabolt RJ, Ruiz C (1983) Impact loading of plates—an experimental investiga-
tion. Int J Impact Eng 1:3–22
9. Gupta NK, Iqbal MA, Sekhon GS (2006) Experimental and numerical studies on the behavior
of thin aluminium plates subjected to impact by blunt and hemi spherical nosed projectile. Int
J Impact Eng 32:1921–1944
10. Bao Y, Wierzbicki T (2004) On fracture locus in the equivalent strain and stress triaxiality
space. Int J Mech Sci 46–1:81–98
Additive Manufacturing Process
Selection Using MCDM
Abstract Additive manufacturing (AM) has huge benefits over traditional manu-
facturing, viz. cost saving, lesser product development time and lead time. AM eas-
ily produces complex geometry. However, selecting the right AM process/machine
compatible for part as per customers’ specification and manage manufacturability
and functionality is a critical issue. This study uses a multi-criteria decision-making
(MCDM) methodology for deciding the most suitable AM process that is presented.
For this, 17 criteria under five group criteria are used.
1 Introduction
Additive manufacturing (AM) has been introduced in the 1980s for producing pro-
totypes and models [1]. It was invented to manufacture a three-dimensional object
through computer-aided design (CAD). Using data of 3D model, AM joins mate-
rials layer by layer to produce objects [1]. AM has been defined as, “the use of a
computer-aided design (CAD)-based automated additive process to construct parts
that are used directly as finished products or components” [2]. AM produces complex
customized parts at rapid pace and significantly provides the benefits like low inven-
tory turnover, minimum time to market, low wastage of material and higher flexibility
[1]. Selecting a right process/machine for producing such a complex part through
AM always remains a tedious task. To address this issue, the focus of the current
study is to develop a framework for evaluating most suitable AM process/machine
for producing parts.
2 Literature Review
The problem of determining the most suitable AM process has received a huge atten-
tion and addressed by researchers in different ways using different methodologies
[3]. Some of them includes question table and comparison chart [4], rule-based expert
system [5], modified TOPSIS [6], fuzzy synthetic evaluation with experts system [7],
graph theoretical approach [8], AHP [9], fuzzy logic and artificial neural network
[10]. Despite having huge literature on AM process determination, as far as authors
knowledge, there is a lack of literature which has discussed and used the fuzzy rep-
resentation for objective criteria. In this article, a rating approach of AHP is used
to evaluate subjective criteria while the fuzzy representation is used for objective
criteria [11] to solve the problem under study.
3 Research Methodology
3.1 AHP
Level 4
VH H M L VL
Fig. 1 Hierarchy of criteria for AHP (*: local weights of criteria, **: global weight of criteria)
604 V. Dohale et al.
measured using AHP shown in Table 3. The objective criteria are evaluated using
fuzzy approach, explained in Sect. 3.2. Finally, the overall score obtained by each
alternative is calculated as:
I
J
Sa = Wi wi j Pi ja (1)
i=1 j=1
The performance of the alternatives is evaluated using AHP. For doing this, the
values for subjective and objective (qualitative and quantitative) criteria are taken
from the literature [8, 9, 15, 16]. However, due to space constraint, a sample data
and calculations of subjective and objective criteria are provided in Tables 4 and 5.
The alternatives considered in this study are:
Machine 1 (M1) Machine 2 (M2) Machine 3 (M3) Machine 4 (M4) Machine 5 (M5)
Table 4 Sample data and performance evaluation of alternatives—objective criteria
Sub-criteria (objective) Local weight (wi ) Global weight (GWi ) Sample data Performance evaluation (fuzzy
method)
M1 M2 M3 M4 M5 M1 M2 M3 M4 M5
Dimensional accuracy 0.57 0.251 120 150 125 185 95 1 0 0.49 0 1
(µmm)
Additive Manufacturing Process Selection Using MCDM
In this approach, a graph plotted in which Y-axis represents the capabilities with
values ranging from 0 to 1, and the X-axis represents evaluation criterion with its
value. A tolerance of 10% is considered for getting the result [11, 17, 18]. In this
article, the fuzzy representation helps to measure the level of compatibility between
the customers’ requirement for a specific part. Here, in this case, component A with
the following specification is given. Further, based on the values, the graphs of fuzzy
representation for all criteria are plotted. A sample graph of dimension accuracy
criteria for all alternatives is shown in Fig. 2.
As per the AHP results shown in Table 6, the machine 3, has obtained the highest
overall performance value, i.e., 0.268, whereas machine 1 received the least value,
i.e., 0.168.
5 Conclusion
The study helps to identify the most compatible AM machine for a given part. Five
different machines are considered in this study. In this study, a robust methodology
is developed to accommodate dynamics in the criteria. At first, the relative weights
of both subjective and objective criteria are calculated through AHP. Then, further
the performance rating of alternatives is determined using AHP to find compatible
AM machine regardless of part criteria. Further, the part specifications, i.e., objective
criteria, are matched with the machine’s criteria value using fuzzy representation for
providing the compatibility score to specific machine concerning particular criteria
to address the purpose of study.
References
1. Khorram Niaki M, Nonino F (2017) Additive manufacturing management: a review and future
research agenda. Int J Prod Res 55(5):1419–1439
2. Hopkinson N, Hague R, Dickens P (2006) Rapid manufacturing: an industrial revolution for
the digital age. Wiley, Chichester
3. Kek V, Vinodh S, Brajesh P, Muralidharan R (2016) Rapid prototyping process selection using
multi criteria decision making considering environmental criteria and its decision support
system. Rapid Prototyp J 22(2):225–250
4. Brown R, Stier KW (2001) Selecting rapid prototyping systems. J Ind Technol 18(1):1–8
5. Masood SH, Soo A (2002) A rule based expert system for rapid prototyping system selection.
Robot Comput Integr Manuf 18(3–4):267–274
6. Byun HS, Lee KH (2005) A decision support system for the selection of a rapid prototyping
process using the modified TOPSIS method. Int J Adv Manuf Technol 26(11–12):1338–1347
7. Lan H, Ding Y, Hong J (2005) Decision support system for rapid prototyping process selec-
tion through integration of fuzzy synthetic evaluation and an expert system. Int J Prod Res
43(1):169–194
8. Rao RV, Padmanabhan KK (2007) Rapid prototyping process selection using graph theory and
matrix approach. J Mater Process Technol 194(1–3):81–88
9. Borille A, Gomes J, Meyer R, Grote K (2010) Applying decision methods to select rapid
prototyping technologies. Rapid Prototyp J 16(1):50–62
10. Munguía J, Lloveras J, Llorens S, Laoui T (2010) Development of an AI-based rapid
manufacturing advice system. Int J Prod Res 48(8):2261–2278
11. Akarte MM, Ravi B (2007) Casting product—process—producer compatibility evaluation and
improvement. Int J Prod Res 45(21):4917–4936
12. Saaty TL (2008) Decision making with the analytic hierarchy process. Int J Serv Sci 1(1):83–98
13. Saaty TL (1980) The analytic hierarchy process. Mc-Graw Hill, London
14. Tam MCY, Tummala VMR (2001) An application of the AHP in vendor selection of a
telecommunications system. Omega 29(2):171–182
15. Mançanares CG, Zancul EDS, Cavalcante J, Miguel PAC (2015) Additive manufacturing
process selection based on parts’ selection criteria. Int J Adv Manuf Technol 80:1007–1014
16. Peko I, BajiĆ D, Veža I (2015) Selection of additive manufacturing process using the AHP
method. In: International conference: mechanical technologies and structural materials, pp
119–129
Additive Manufacturing Process Selection Using MCDM 609
17. Giachetti RE (1998) A decision support system for material and manufacturing process
selection. J Intell Manuf 9:265–276
18. Lovatt AM, Bassetti D, Shercliff HR, Bréchet Y, Lovattt AM, Bassetti D, Sherclifft HR, Brechet
V (2000) Process and alloy selection for aluminium casting. Int J Cast Met Res 12(4):211–225
Evaluation and Improvement
of Makespan Time of Flexible Job Shop
Problem Using Various Dispatching
Rules—A Case Study
1 Introduction
operations, the capacity of production and the effect of dispatching rule on makespan
time was carried out.
The small-scale unit follows flexible job shop production and it is involved in
the manufacturing of more than 200 components. Out of these 200 components,
for few components, there are regular orders, and such components that contribute
to the 80% of production are identified using Pareto analysis. The data for these
components, i.e. various processes it undergoes, its sequences and the processing
time of each operation is obtained. The information about the number of machines
and other facilities of flexible job shop was also obtained with their production
capacity. This complete information is the input to the scheduling activity and for
obtaining schedule with different dispatching rules with minimum makespan time.
An open-source LEKIN software was used to obtain the makespan time and the Gantt
charts based on the input data of the production unit. This analysis was focused on
the machines of the press shop with 12 machines. The literature was reviewed to
identify the methodology followed by other researchers to reduce makespan time
for the different production systems. The literature on the scheduling of flexible job
shop was focused for study with the application of different dispatching rules. Shen
et al. [1] proposed a Tabu search algorithm for solving flexible job shop scheduling
problems. Moslehi et al. [2] proposed a hybrid approach of using a combination
of particle swarm and local search algorithm for multi-objective flexible job shop
scheduling problem. Vaessen et al. [3] presented deterministic and randomised local
search method to solve job shop scheduling problem while Phanden et al. [4] proposed
a hybrid method of simulation and genetic algorithm to minimize makespan time of
job shop schedule. Kundake et al. [5] considered dynamic search with situations such
as a breakdown, transportation time, change over times using a genetic algorithm to
reduce makespan time of a job shop. It is found that researchers have used a various
method such as fuzzy method [6, 7], ant colony algorithm [8], beam search method
[9], hybrid fruit fly optimization algorithm [10], quantum behaved particle swarm
optimization [11] for minimize makespan time.
The aim of this work was to reduce makespan time and develop a stable load plan. The
objective was to evaluate the makespan time for existing resources using different
dispatching rules for scheduling and identify bottleneck operations. Based on this
analysis, the requirements of resources are proposed by applying a heuristic approach
along with the dispatching policy so that the delivery schedules can be met without
any delay. The methodology followed in this work is shown in Fig. 1.
Evaluation and Improvement of Makespan Time … 613
Fig. 1 Methodology
Problem Identification
Literature review
System Analysis
Scheduling
Conclusion
The data were collected from the industry for the items ordered to produce and Pareto
analysis was carried out to identify 20% of the components which contributes to
80% of the total demand in a flexible job shop. It was found that only 51 components
contribute to 80% of production. Figure 2 shows the Pareto chart.
The existing production capability of the power press was evaluated to study its
ability to produce the number of components. For this, the number of strokes per
shift of individual machine is evaluated for the power presses which is shown in
Fig. 3. Using this information and the information of the number of strokes required
to complete the operation on individual component was evaluated. Based on this,
the requirement of the number of days required to produce one-month demand using
these machines was obtained and it is shown in Fig. 4.
The makespan time of the production is depended on the individual process time,
Number of components to be produced, the sequence of operation, the machines to
be used, etc. This information for 51 components was collected and it is given as
614 M. Bihari and P. V. Kane
5000
No. of strokes
4000
3000
2000
1000
0
Shearing 300TPP 200T Power 150TPP 100TPP 50TPP1 30TPP1 20TPP1
Press
Machines
45
30TPP1 18
30
Machines
100TPP 30
7
200TPP 22
6
SHEARING 26
0 10 20 30 40 50
No. of days
input to open-source LEKIN scheduling software. Based on this input data, and for
the various dispatching rules such as FCFS, LPT and SPT, the makespan time for
existing resources is evaluated. Different dispatching rules were applied to do the
scheduling. It is observed that by LPT, 78% of schedule deliveries can be met. With
SPT, 63% delivery schedule can be met while with FCFS, 66.13% schedule can be
met. Thus, it is found that the LPT would be a better option compared to others for
the existing resources. The analysis was carried out for the two shifts and with one
shift of 10 h, leading to the due date by 7200 min and, in all cases, the makespan time,
i.e. time required to complete the manufacturing of components found to be is more
than that due date. In this work, 84% of the total demand is taken into consideration
and for producing this amount of demand more than 7200 min is required. The
objective to adhere to schedules with the existing resource was not met by using
different dispatching rules for existing resources. Hence, there is a need to have an
additional machine and human resources to adhere to the scheduled delivery dates
of all the components. For this purpose, based on the Gantt charts obtained from
LEKIN Software, bottleneck operation for the existing resources were identified.
The Gantt chart for LPT dispatching rule is shown in Fig. 5. How the addition of
the machines would affect the makespan time for different dispatching rules was
evaluated by simulating these conditions in LEKIN software.
The various such modification is made and the resulting makespan time is summa-
rized in Table 1. After making individual modifications and combined modification
by heuristic approached two sets of modifications found to be effective to reduce
the makespan time and the results obtained are summarized in Table 1. Although
the makespan time has been reduced by 27% by the introduction of 20TPP, 100TPP
and 200TPP machines (i.e. set 2), it has led to an increase in demand for resources
like dies, which can financially overburden the industry. So, to cope up with this
problem a shearing machine is added with 20TPP and 100TPP machines (i.e. set 1)
and different dispatching rules are applied to study the effect on makespan time. It
is observed that the makespan time improves by 26% and the shearing machine will
not require any dies and hence the investment.
Table 1 Comparison between the makespan time for a proposed set of machines and existing
system
Dispatching Makespan Makespan Makespan % Reduction % Reduction
rule time time time in makespan in makespan
(minutes) (minutes) (minutes) time (for set time (for set
with existing with set 1 with set 2 1) 2)
resources modifications modifications
FCFS 9149 7146 7766 22 16
LPT 7691 5756 5661 26 27
SPT 9567 6646 8935 31 7
5 Conclusion
In this work, the efforts are made to evaluate and minimize the makespan time of a
flexible job ship using different dispatching rules to address the issue of late deliv-
eries faced by the small-scale unit. The information and the data of the components
manufactured, the machines required for its processing, the operation cycle time
of each process for each component, etc. were collected and the schedules were
obtained using different dispatching rules in LEKIN software. The necessary addi-
tion of resources and the scheduling policy that can be followed to adhere to the
delivery schedules are suggested. It is observed that the LPT dispatching policy with
the addition of a shearing machine and two power presses of 20TPP and 100TPP
machines lead to the reduction of makespan time by 26%.
References
1. Shen L, Dauzère-Pérès S, Neufeld JS (2018) Solving the flexible job-shop scheduling problem
with sequence-dependent setup times. Eur J Oper Res 265:503–516
2. Moslehi G, Mahnam M (2011) A Pareto approach to multi-objective flexible job-shop
scheduling problem using particle swarm optimization and local search. Int J Prod Econ
129:14–22
3. Vaessens RJM, Aarts EH, Lenstra JK (1996) Job shop scheduling by local search. Informs J
Comput 8:302–317
4. Phanden RK, Jain A (2015) Assessment of makespan performance for flexible process plans
in job shop scheduling. IFAC-PapersOnLine 48:1948–1953
5. Kundakcı N, Kulak O (2016) Hybrid genetic algorithms for minimizing makespan in dynamic
job shop scheduling problem. Comput Ind Eng 96:31–51
6. Palacios JJ, González MA, Vela CR, González-Rodríguez I, Puente J (2015) Genetic tabu
search for the fuzzy flexible job shop problem. Comput Oper Res 54:74–89
7. Palacios JJ, González-Rodríguez I, Vela CR, Puente J (2015) Coevolutionary makespan opti-
misation through different ranking methods for the fuzzy flexible job shop. Fuzzy Sets Syst
278:81–97
8. Chaouch I, Driss OB, Ghedira K (2017) A modified ant colony optimization algorithm for the
distributed job shop scheduling problem. Procedia Comput Sci 112:296–305
9. Birgin EG, Ferreira JE, Ronconi DP (2015) List scheduling and beam search methods for the
flexible job shop scheduling problem with sequencing flexibility. Eur J Oper Res 247:421–440
Evaluation and Improvement of Makespan Time … 617
10. Liu Q, Zhan M, Chekem FO, Shao X, Ying B, Sutherland JW (2017) A hybrid fruit fly algorithm
for solving flexible job-shop scheduling to reduce manufacturing carbon footprint. J Clean Prod
168:668–678
11. Singh MR, Mahapatra SS (2016) A quantum behaved particle swarm optimization for flexible
job shop scheduling. Comput Ind Eng 93:36–44
The Impact of Building Orientation
on Microhardness and Surface
Roughness of Direct Metal Laser
Sintered Inconel Alloy
1 Introduction
Inconel 718 (IN718) superalloy is often used for various applications, such as jet
engines, aircraft turbines, and steam turbine power plants in view of its fine creep
resistance, good weldability, high yield resistance, and good fatigue strength at higher
temperatures [3–5]. IN718 alloys are suitable for high-temperature applications it
maintains its mechanical properties up to 650 °C [6]. Therefore, IN718 attracts
researchers’ more and more attention.
From the last two decades, it has attracted much interest of the researchers to
understand the connection between built orientation and mechanical properties (such
as surface roughness, microhardness, density, and strength) of DMLS fabricated
IN718 components [7, 8]. Different researchers have optimized the different process
parameters for AM fabricated samples to minimize the porosity and to produce
crack-free parts for different alloys such as Fe-based alloys [9], Al-based alloys [10],
Ti-based alloys [11] and Ni-based alloys [12].
Tarun et al. [13] studied the effect of three build orientation (0°, 45°, 90°) on
surface roughness and microhardness and found minimum surface finish for inclined
orientation (45°) than horizontally (0°) and vertically (90°) build orientations while
higher microhardness was found for horizontal orientation for 18Ni300 Maraging
steel. Mohsen et al. [14] studied the effect of the different process parameters such
as scanning speed, laser power, hatch distance and hatch pattern on the surface for
AlSi10Mg_200C using DMLS process. Meneghetti et al. [15] studied the effect
of two building orientations (0, 90) and did heat treatment on 18Ni300 and found
maximum roughness for 0° build orientation for heat-treated parts. Rashid et al.
[16] analyses that the parts printed using the single scanning strategy exhibit lower
hardness value than the part printed with double scanning strategy for 17-4PH steel.
Hrbackova et al. [17] performed heat treatment for maraging steel printed samples
and found that the heat-treated samples have higher hardness than the as-built (non-
heat-treated) samples and it has also found that non-heat treated AlSi10Mg samples
have higher hardness than heat-treated samples. Fatemi et al. [18] studied the different
treatment and surface finish process on Ti6Al4V samples printed by laser powder
bed fusion process and found that higher surface roughness for As-built annealed
samples. Kamariah et al. [19] studied the effect of heat treatment of different build
oriented (0°, 45°, 90°) selective laser melted 316L stainless-steel and found that
higher hardness for 45° oriented heat-treated samples. Liverani et al. [20] found
the influence of different process parameters for microhardness of selective laser
melted 316L stainless samples. Haghshenas et al. [21] compared that sample printed
by additive manufacturing process have higher hardness than sample built by the
conventional manufacturing process. Cyr et al. [22] investigated the effect of different
heat treatment on microhardness of MS1-H13 samples and found that samples which
are heat-treated for age hardening have higher microhardness than any other heat-
treated samples. Lu et al. [23] reported the effect of the different heat-affected zone
on microhardness and found that the higher microhardness value is at top surface
layer which is two times more than the base metal layer. Uhlmann et al. [24] studied
the different finishing process for SLM shape specimen to improve the surface finish.
Mantrala et al. [25] studied the different heat treatment process and investigated the
effect of it on microhardness and the results revealed that the higher hardness can be
The Impact of Building Orientation on Microhardness … 621
achieved by suitable heat treatment process after fabrication. Gong et al. [26] studied
the nanoindentation for top and bottom layer of SLM printed Inconel 718 sample.
Wang et al. [27] studied the effect of annealing on microhardness of selective laser
melted Alsi7Mg alloy and found that after annealing the value of microhardness
decreased by nearly 41%. This previous research supports us for further research.
Some studies have investigated the effect of heat treatment, build orientation,
and changing the process parameters on microhardness and surface roughness for
additive manufactured samples of different materials such as Al alloy, Ni and Ti based
alloys, and Fe-based alloys. Till date, no literature has been found that investigated
microhardness and surface roughness for Ni-based alloys considering three-building
orientations i.e. (0°, 45°, 90°) of DMLS printed Inconel 718 alloy.
Therefore, the objective of this study is to examine the impact of three building
orientation on microhardness and surface roughness of DMLS printed Inconel 718
samples are presented.
In this work, Inconel 718 spherical shaped powder particle size 15–45 µm was taken.
The chemical composition of metal powder of IN718 is listed below in Table 1 which
made available by EOS.
Samples has been fabricated using the setup EOSINT M280 direct metal laser
sintering additive manufacturing system which is available at Central Tool Room and
Training Centre (CTTC), Bhubaneswar, as shown in Fig. 1a, b. The available system
has a capacity of Ytterbium (Yb) fiber laser of laser power (≤400 W), wavelength
(1060–1100 nm) and scan speed (≤7000 mm/s). This system has a build platform of
(250 mm × 250 mm × 325 mm) before printing the samples it is required to preheat
the platform up to 40 °C at an oxygen level of 1.6%). The Inconel 718 samples have
been fabricated using EOS suggested standard process parameters that is listed in
Table 2.
Samples have been built in three different building orientation i.e. (0°, 45°, 90°)
of size 100 mm × 10 mm × 5 mm using the above DMLS system. Three samples
of the same size for each building orientation were fabricated. Microhardness and
surface roughness were performed for all the samples (Fig. 2).
To measure the variation in surface properties such as microhardness and surface
roughness four different points (a, b, c and d) marked along the building layer at an
equal distance.
A radial digital Vickers microhardness tester RMHT201 is used to measure the
microhardness. To make an indent a load of 200 g was applied to the samples top
layer for a dwell period of 15 s. Indentations were taken at four selected locations at
an equal distance for each sample and then taken an average of all samples for same
indentation location.
A linear motion digital surface roughness tester Mitutoyo SJ410 is used to measure
the surface roughness. The cut off length set for the tester is 0.25 mm and its traveling
622
speed set to 0.1 mm/s. The measurement was taken at four selected locations at an
equal distance for each sample and then taken an average of all samples for the same
location.
direction for each sample then taken its average of all. As shown in Fig. 3a–c a little
variation is found in the microhardness along with the building layer for all three
building orientations. The microhardness is found decreasing along the building
direction for all three building orientations.
The reason to decrees in microhardness along the building direction is due to
decrees in solidification time between the layers which increases the coarser grains
along the building direction. A decrease in solidification time occurs due to remelting
of the previously deposited layer which is caused by changes in a specific area of heat
flow [13]. The result of the average microhardness value for all building orientation
is given below in Table 3. Maximum microhardness value is calculated for 90°
orientation and minimum microhardness value is calculated for 0° orientation.
Fig. 3 Variation in microhardness along building direction for a 0° orientation, b 45° orientation
and c 90° orientation
Table 3 Average
Building orientation Microhardness (HV)
microhardness for three
building orientations 0° 327.95 ± 31.27
45° 337.61 ± 34.07
90° 343.34 ± 25.12
The Impact of Building Orientation on Microhardness … 625
Fig. 4 Variation in surface roughness along building direction for a 0° orientation, b 45° orientation
and c 90° orientation
626 A. K. Maurya and A. Kumar
for 90° building orientation. The average surface roughness Ra (µm) is summarized
in Table 4 for all three-building orientation.
4 Conclusions
This work has analyzed the variation of microhardness and surface roughness for
three building orientations, of DMLS fabricated Inconel 718 samples. Variation in
microhardness and surface roughness has been found for each building orientation
along the building direction. From the result, it is clear that the building orientation
has a significant effect on microhardness and surface roughness. Study shows that the
vertically fabricated samples have maximum microhardness and minimum surface
roughness than the horizontally fabricated samples. These results will be helpful to
the manufacturer to choose the suitable building orientation for better surface finish
and microhardness. Further heat treatment and microstructure studies also can be
performed to examine the changes occurs in microhardness and surface roughness.
Acknowledgements Thanks to staff of CTTC, Bhubaneswar for providing the DMLS facilities.
The support of Dr. A. K. Dubey, Dr. A. Narayan and Dr. S. K. Saini MNNIT Allahabad for conduct-
ing microhardness and surface roughness tests is gratefully acknowledged and financial support
provided by MHRD, GOI.
References
1. Frazier William E (2014) Metal additive manufacturing: a review. J Mat Eng Perf 23:1917–
1928. https://doi.org/10.1007/s11665-014-0958-z
2. Yap CY, Chua CK, Dong ZL, Liu ZH, Zhang DQ, Loh LE, Sing SL (2015) Review of selective
laser melting: materials and applications. Appl Phys Rev 2(4):41101. https://doi.org/10.1063/
1.4935926
3. Sharma AD, Sharma AK, Thakur N (2014) Crystallographic, microstructure and mechanical
characteristics of dynamically processed IN718 superalloy. J Alloy Comp 597:175–180
4. Wang ZM, Guan K, Gao M, Li XY, Chen XF, Zeng XY (2012) The microstructure
and mechanical properties of deposited-IN718 by selective laser melting. J Alloy Comp
513:518–523
5. Chen Y, Lu FG, Zhang K, Nie PL, Hosseini SRE, Feng K (2016) Dendrite microstructure and
hot cracking of laser additive manufactured Inconel 718 under improved base cooling. J Alloy
Comp 670:312–321
The Impact of Building Orientation on Microhardness … 627
Keywords Alloy 718 · Reciprocating wear test · COF · Wear rate · Glaze layer
1 Introduction
2 Experimental Methodology
The chemical composition (wt%) of the material alloy 718 used in the present study
is indicated in Table 1. Alloy 718 employed in the present study undergone heat-
treatment steps such as solutionising and DA reported in [9] and which had the elastic
modulus = 208 GPa, Poisson’s ratio = 0.29. The alloy 718 samples were polished
using SiC emery sheets of different grades 400, 600, 800 and 1000 and subsequently
polished with diamond paste of 0.25 μm. The polished sample was etched using
Kalling’s reagent. Microstructural observation was under using a scanning electron
microscope. Figure 1 shows a typical alloy 718 microstructure consisting of FCC
nickel matrix after DA heat treatment. The hardness value of the alloy 718 samples
is found to be 42 HRC.
The high-temperature reciprocating wear tests were performed on the sample
dimension of 20 × 20 × 6 mm3 . The comparison wear tests were carried out on
samples at ambient temperature. The wear test samples prepared from the alloy 718
material subjected to DA heat-treatment condition. The wear test parameters are
tabulated in Table 2. The wear loss was measured using weight-loss method using an
analytical balance of 0.0001 g accuracy. Counter body material used in the present
study was WC-Co of ball diameter 6 mm had young’s modulus = 450 GPa and
Poisson’s ratio = 0.22 [16]. Both Hertzian contact pressure and contact depth were
calculated using Eqs. (1)–(3) as proposed by Hertz theory [17] for WC-Co ball on
alloy 718 plate configuration.
6N E ∗2
P=
3
(1)
π 3r 2
where
1 1 − γ12 1 − γ22
= + (2)
E∗ E1 E2
3 3Nr
a= (3)
4E ∗
Figures 2 and 3 show the variation of COF with the wear test duration for ambient
and high temperature at a sliding distance of 100 m for three different loading con-
ditions. In has been well established that, metals and alloys subjected to tribological
632
Figure 4 shows the influence of temperature on wear resistance of the alloy 718 sam-
ples. The wear volume is increasing with increasing the normal load in both ambient
and high-temperature condition, which could be attributed to the increase in contact
pressure and contact depth (Table 3). It could be seen that the alloy 718 samples
tested at high temperatures had a poor wear resistance compared to samples tested
at ambient temperature. It is presumed that products of tribo-chemical reaction such
as harder wear debris resulted in developing a three-body abrasion wear mechanism
among tribo-pair. It is has been well established that, various factors such as phys-
ical, chemical, thermal and mechanical properties of the tribo-pair synergistically
aggravate the for tribo-chemical reaction leading to generation of reaction products
with hard wear debris [19].
Among various properties of tribo-pair, the mechanical and thermal are exhibiting
the significant contributions for accelerating the tribo-chemical reactions. However,
Investigation on Elevated Temperature Tribological … 635
Fig. 4 Influence of
temperature on the wear loss
of alloy 718 samples
4 Conclusions
The testing temperature had a significant influence on the COF and wear resistance
of alloy 718. The lower COF values of the alloy 718 samples subjected to high
wear testing temperature conditions were attributed to the extent and presence of
glaze layer formation, which is insignificant in the case of samples subjected to
ambient temperature. The higher wear rate values of the alloy 718 samples subjected
to high wear testing temperature conditions were attributed to the due to occurrence
tribo-chemical reactions at the contact zone of tribo-pair.
636 S. Anand Kumar et al.
References
1. Herman H, Sampath S, Stern KH (eds) (1996) Thermal spray coatings, metallurgical and
ceramic protective coatings. Chapman & Hall; London, UK, pp 261–289
2. Erdemir and Fenske GR (1988) Wear resistance of metals and alloys. ASM International,
Chicago, p 89
3. Sequera A, Fu CH, Guo YB et al (2014) Surface integrity of Inconel 718 by ball burnishing. J
Mat Eng Perf 23:3347–3353
4. Nabah BAA, Hassan WT, Ryan D et al (2010) The effect of hardness on eddy current residual
stress profiling in shot-peened nickel alloys. J Nondestruct Eval 29:143–153
5. Amanov A, Pyun YS, Qingyuan W et al (2014) Fine grain structure as palliatives for fretting
wear of Inconel 718 alloy at various temperatures. 8th international symposium on superalloy
718 and derivatives. Wiley, Pennsylvania, pp 555–562
6. Kumar S, Sudhakar Rao G, Chattopadhyay K et al (2014) Effect of surface nanostructure on
tensile behaviour of superalloy IN718. Mater Des 62:76–82
7. Chen T, John H, Xu J et al (2013) Influence of surface modifications on pitting corrosion
behaviour of nickel-base alloy 718. Part 1: effect of machine hammer peening. Corr Sci 77:230–
245
8. Kamal S, Jayaganthan R, Prakash S (2010) Mechanical and microstructural characteristics of
detonation gun sprayed NiCrAlY þ 0.4 wt% CeO2 coatings on superalloys. Mater Chem Phys
122:262–268
9. Pichon CL, Cormier J, Dubois JB et al (2013) Plasma assisted nitriding of Ni-based superalloys
with various microstructures. Surf Coat Tech 235:318–325
10. Anand Kumar S, Ganesh Sundara Raman S, Sankara Narayanan TSN et al (2012) Fretting wear
behaviour of surface mechanical attrition treated alloy 718. Surf Coat Tech 206:4425–4432
11. Anand Kumar S, Sundar R, Ganesh Sundara Raman S, Gnanamoorthy R, Kaul R, Ranganathan
K, Bindra KS (2017) Effects of laser peening and counterbody material on fretting wear
behaviour of alloy 718. J Eng Tribol 231(10):1276–1288
12. American Society for metals ASM handbook volume 4: heat treatment. ASM International,
USA (1991)
13. Wang C, Li R (2004) Effect of double aging treatment on structure in Inconel 718 alloy. J Mater
Sci 39:2593–2595
14. Maru MM, Tanaka DK (2006) Influence of loading, contamination and additive on the wear of
a metallic pair under rotating and reciprocating lubricated sliding. J Braz Soc Mech Sci Eng
28:278–285
15. Ward R (1970) A comparison of reciprocating and continuous sliding wear. Wear 15(6):423–
434
16. Metals Handbook Committee (1961) Metals handbook. Metals Park, OH7 American Society
for Metals, p 1201
17. Johnson KL (2003) Contact mechanics. Cambridge University Press, UK
18. Waterhouse RB (1981) Fretting at high temperatures. Trib Inter 14:203–207
19. Ismail R, Yaacob II (2005) The formation of aluminides in intermetallic nickel aluminide-based
nanocomposites. J All Comp 392:214–219
Digital Twin for Shell and Tube Heat
Exchanger in Industry 4.0
Abstract Digitalization of the heat transfer in shell and tube type heat exchanger is
what sets the foundation of the future of thermal industry. In traditional computational
model design, a simulation model is validated with the experimental results. A digital
twin is a virtual representation of the real system or processes and imbibes the
validation of the model. The only difference is that the exchange of information is
carried out in real time and is more reliable. A digital twin is not a static representation
of the real space but rather a dynamic phenomenon connecting the two spaces. In
the present analysis, a comparative finite element analysis of the shell and tube heat
exchanger has been performed with an intention to enhance the effectiveness of the
heat exchanger process. An approach toward development of the next generation of
heat exchangers is discussed in this work.
Keywords Digital twins · Shell and tube heat exchanger · Comsol multiphysics ·
Turbulence model
1 Introduction
Shell and tube heat exchanger (STHE) is the most essential and vital component of
the thermal industry. It is used to recover heat between two flowing fluids. Because
of the simplicity involved in its manufacturing technique and its adaptability to the
wide range of operating conditions, it is largely preferred [1, 2]. Energy consumption
plays a critical role in the economic development of a country. It has been reported
previously that two-third of the energy utilized in a diesel engine is wasted [3]. The
work done in this paper focuses on the proper utilization of sustainable sources and
methodologies for energy conservation using STHE. In this type of heat exchanger,
several tubes are packed inside a shell. One fluid (usually oil/water) flows through
the tubes, while the other fluid (usually air) flows outside the tube in the shell. Xiao
et al. [4] reported that the heat exchange quality in double tube pass STHE is two
times greater than that in single tube pass STHE. Such a heat exchanger finds its
importance in oil refinery and various chemical industries [5]. Depending upon the
materials used for its construction, it can be used to cool or heat the water in swimming
pool or charging air [6]. The most significant advantage of using such a kind of heat
exchanger is that it is elementary to service, especially in a model which involves
floating tube bundle.
With the advent of the next industrial revolution (Industry 4.0), digital twins find
their importance in the factories of future [7]. A digital twin can be defined as a
linkage between the real and virtual space. This linkage can be obtained by using a
simulation software. In the current work, a simulation model of STHE is developed
using COMSOL Multiphysics 5.3. This is the first step toward establishing a digital
twin of the heat exchanger. COMSOL API utilizes Java; the Java model file from
the software can further be used to communicate with an external system, by using
dynamic link library files (dll-files). Benefiting from the Java ecosystem, it is also
possible to implement the virtual space as a Web service (such as a Java-based Web
service running inside Tomcat) that could present, for example, a representation state
transfer application programming interface (REST API) for communicating with the
real space. This would be a central part of the digital twin environment [8]. However,
the current work is limited only to the development of the simulation model. Its
integration via simulation apps leading to the establishment of the actual digital twin
will be discussed in future works.
2 Model Definition
Hot fluid (water) enters from the tube inlet with a velocity of 0.1 m/s and temperature
of 80 ◦ C. The material used for the heat exchanger is structural steel. Hot fluid and cold
fluid have different inlet temperatures when it enters the domain, but after circulating
inside the domain, the fluids are brought closer to the equilibrium temperature. Baffles
are provided in a crisscross manner, which enhances the rate of heat transfer and
also minimizes the vibrations caused by the motion of the fluid [9]. Finite element
analysis of STHE has been performed by coupling two physics. Heat transfer module
is coupled with the k- turbulent model.
Cold fluid (air) enters from the shell side with the velocity of 1 m/s [10]. At tube
inlet and at shell inlet position, the turbulent length is considered as 0.07 R, where
R is the channel radius. Due to symmetry, one-half of the geometry is simulated.
Influence of various parameters on the effectiveness of heat exchanger and overall
heat transfer coefficient are evaluated and discussed. Figure 1 shows the geometry
of the STHE. The nominal diameter of tubes is set to 3/4". Other specifications of
Digital Twin for Shell and Tube Heat Exchanger in Industry 4.0 639
the heat exchanger are: number of tubes = 84, shell diameter = 400 mm (ID), shell
material = stainless steel, tube material = stainless steel and copper, tube length =
1.5 m, total length of the heat exchanger = 1.85 m.
2.1 Meshing
A physics controlled mesh is utilized for the current study. Free tetrahedral meshing
has been performed within the cavity, and near the wall, boundary layer meshing is
used. Figure 2 shows the mesh elements.
The flow inside STHE is turbulent; hence, it is critical to choose the right turbulence
model so that flow can easily be defined, and physics could be captured properly. For
the current work, k- turbulence model for both the fluids (air and water) is used.
k represents the turbulent kinetic energy and is representative of the dissipation
of kinetic energy per unit mass. Reynold stresses are identified using Boussinesq
relationship [11]. The constitutive equations for turbulent viscosity μT , k and at
the boundary conditions are discussed below (Eqs. 1–17).
640 H. Singh et al.
u.n = 0 (1)
u = −Uo n (2)
u = −Uo n (7)
3/2
3 k
k= (Ur e f IT )2 , = (Cμ )3/4 . (8)
2 LT
Digital Twin for Shell and Tube Heat Exchanger in Industry 4.0 641
u.n = 0 (11)
∇k.n = 0, ∇.n = 0 (12)
u u/d .n = 0 (13)
∇ku/d .n = 0 (14)
2
Cμ ku/d
du .n = + (15)
kv δw,u/d νu/d
u uu
u tang . = − (u u/d .n)n (16)
d d
Heat transfer in fluids
3 Results
Two cases are considered for the current study (a) when shell and tube both are made
up of structural steel and (b) when shell is made up of stainless steel and tube is
made up of copper material. The findings are shown in Tables 1 and 2, respectively.
It can be seen that the value of heat transfer coefficient is increased for the second
case. The effectiveness of the heat exchanger varies from 0.68 to 0.76 for the first
case and 0.73–0.83 for the second case. On increasing the inlet temperature of water
and decreasing the inlet temperature of air, a higher value of effectiveness can be
achieved. Figures 3 and 4 show the temperature profile and streamline velocity profile
of the STHE. The streamlines show variation of temperatures at inlet and outlet of
the shell and tube heat exchangers. It can be seen that there is an existence of the
recirculation zones at the inlet of water.
642 H. Singh et al.
Table 1 Simulation results for SS shell and SS tube versus SS shell and Cu shell
u_water u_air Tin _water Tin _air Tout _water Tout _air h (w/m2 K)
(m/s) (m/s) (◦ C) (◦ C)
(SS + (SS + (SS + (SS + SS + SS SS + Cu
SS) (◦ C) Cu) (◦ C) SS) (◦ C) Cu) (◦ C)
0.001 1 80 5 66.94 65.93 56.24 59.65 4.5 4.81
0.0505 1.5 75 7 74.60 74.56 59.24 63.60 7.59 8.29
0.1 2 70 9 69.77 69.75 53.62 58.27 9.66 10.78
Table 2 Simulation results for SS shell and SS tube versus SS shell and Cu shell
u_water u_air Tin _water Tin _air ∇p_water (Pa) ∇p_air (Pa) u_max (m/s) Effectiveness
(m/s) (m/s) (◦ C) (◦ C)
SS + SS + SS + SS + SS + SS + SS + SS +
SS Cu SS Cu SS Cu SS Cu
0.001 1 80 5 0.02 0.03 11.90 12.40 1.64 2.0 0.68 0.73
0.0505 1.5 75 7 11.27 11.63 25.50 26.65 2.42 3.0 0.76 0.83
0.1 2 70 9 38.30 39.07 42.90 45.02 3.11 3.9 0.73 0.81
4 Conclusions
References
1. Fesanghary M, Damangir E, Soleimani I (2009) Design optimization of shell and tube heat
exchangers using global sensitivity analysis and harmony search algorithm. Appl Therm Eng
29(5–6):1026–1031
2. Hadidi A, Hadidi M, Nazari A (2013) A new design approach for shell-and-tube heat exchangers
using imperialist competitive algorithm (ICA) from economic point of view. Energy Convers
Manag 67:66–74
3. Pandiyarajan V, Chinna Pandian M, Malan E, Velraj R, Seeniraj RV (2011) Experimental inves-
tigation on heat recovery from diesel engine exhaust using finned shell and tube heat exchanger
and thermal storage system. Appl Energy 88(1):77–87
4. Xiao J, Wang J, Jian G, Wang S, Wen J (2016) Performance investigation on shell-and-tube heat
exchangers with different baffles based on fluid-structure interaction, pp 379–384
5. Pignotti A, Shah RK (1992) Effectiveness-number of transfer units relationships for heat
exchanger complex flow arrangements. Int J Heat Mass Transf 35(5):1275–1291
6. Hesselgreaves JE (2000) Rationalisation of second law analysis of heat exchangers. Int J Heat
Mass Transf 43(22):4189–4204
7. Saxena P, Papanikolaou M, Pagone E, Salonitis K, Jolly MR (2020) Digital manufacturing for
Foundries 4.0. In: Tomsett A (ed) Light Metals 2020. The Minerals, Metals & Materials Series.
Springer, Cham
8. Lyu P (2017) Which porous media and subsurface flow interface should I use? COMSOL Blog
9. Bejan A (1980) Second law analysis in heat transfer. Energy 5(8–9):720–732
10. Ogiso K (2003) Duality of heat exchanger performance in balanced counter-flow systems. J
Heat Transf 125(3):530–532
11. Ahsan M (2015) Prediction of gasoline yield in a fluid catalytic cracking (FCC) riser using k-
epsilon turbulence and 4-lump kinetic models: a computational fluid dynamics (CFD) approach.
J King Saud Univ - Eng Sci 27(2):130–136
Model-Based Synchronized Control
of a Robotic Dual-Arm Manipulator
1 Introduction
A robotic system, which consists of more than one robot, has more capacity than
a system consisting of only a single robot for assembly tasks and carrying heavy
material. Not many researchers have contributed their time studying the coordinated
control of multiple robot arms actively as compared to the study of the control of a
single robotic arm. When a system comprising of multiple robots grasps one object,
A. Katpatal
Robotics Engineering, Worcester Polytechnic Institute, Worcester, MA, USA
A. Parwekar (B) · A. K. Jha
Department of Mechanical Engineering, Shri Ramdeobaba College of Engineering and
Management, Nagpur 440013, India
e-mail: parwekaraa@rknec.edu
it forms a closed chain mechanism which is coupled and nonlinear. The application
of control algorithm [1] for a system of multiple robots, performing assembly tasks
utilizes motion and internal force control along with a force distribution technique.
In the literature [2–4], control schemes for coordinated system of multiple robotic
arms, which require hybrid position and force control, have been presented. The
control algorithms are given in [6–8] convert the trajectory of payload into individual
robot arm end-effector trajectory by considering the payload motion with respect to
the independent robot arm movements. Researchers have been recently using the
robust, adaptive, passive, and adaptive variable structure control methods, which are
formulated to deal with uncertainties and disturbance, for the control of single-arm
manipulators and can also be utilized for system of multiple cooperating robotic arms
[9–14]. Not taking into account the disturbance and uncertainty effects, may cause
problems in the control capabilities for most of the proposed control algorithms.
Joint-1 acts as the shoulder and it rotates about the z1 axis. This joint rotates in the
x 1 y1 plane and has been assigned a rotational angle as θ 1 . Joint 2, which is the elbow,
rotates about the z 2 axis, which is perpendicular to z 1 axis. θ2 has been assigned as
the rotational angle for joint 2. Joint 3 facilitates the tool pitching motion denoted
as θ 3 and joint 4 facilitates tool roll motion along z4 axis which is perpendicular to
joint 3. Joint 5 is at a vertical offset with joint 4 and provides an angular motion of θ 5
which is also identical to gripper rotation. The gripper sliding motion while picking
is not considered as a degree of freedom of the manipulator (Fig. 1).
Table 1 contains the DH parameters for the above manipulator. Here θi denotes
the angle of rotation about z-axis, ai denotes the length of the link, d1 denotes the
link offset along z-axis and αi is a link twist about x-axis.
The overall transformation matrix which relates the end-effector coordinate frame
to the base frame is given as:
Fig. 1 General
configuration of the 5 DOF
robotic arm
Model-Based Synchronized Control of a Robotic … 647
0
5T = 01 T 12 T 23 T 34 T 54 T
⎡ ⎤
c1 c234 c5 + s1 s5 −c1 s234 s5 + s1 c5 c1 c234 c1 (d5 c234 + a2 c23 + a2 c2 )
⎢ s1 c234 c5 + c1 s5 s1 s234 s5 + c1 c5 s1 s234 s1 (d5 c234 + a4 c23 + a2 c2 ) ⎥
=⎢
⎣
⎥
−c234 c5 c234 s5 −s234 d1 − a2 s2 − a3 s23 − d5 s234 ⎦
0 0 0 1
Here, c1 , s1 , c23 , and c234 are, respectively, cos θ1 , sin θ1 , cos(θ2 + θ3 ), and
cos(θ2 + θ3 + θ4 ). This product matrix is 4 × 4 matrix with the three elements of
last column represent position of end-effector (denoted as px , p y , and pz ), while
the remaining nine elements indicate the end-effector orientation vectors. In other
words, the end-effector is represented with its position and orientation as follows:
⎡ ⎤
nx ox ax px
⎢ py ⎥
⎢ ny oy ay ⎥
5T = ⎣
0
nz oz az pz ⎦
0 0 0 1
Here,
s12 = s1 c2 + c1 s2
c12 = c1 c2 − s1 s2
c234 = c2 (c2 c4 − s3 s4 ) − s2 (s3 c4 + c3 s4 )
s234 = s2 (c3 c4 − s3 s4 ) − c2 (s3 c4 + c3 s4 )
Next, for the considered manipulator arm (first 3 degrees of freedom), the dynamic
equations are formulated.
Inertial matrix derived from Euler–Lagrangian method is given below:
⎡ ⎤
m 11 0 0
M(θ ) = ⎣ 0 m 22 m 23 ⎦
0 m 32 m 33
where
Here, m1 , m2 , m3 , and m4 are, respectively, masses of first, second, third link and
end effector, and load masses. The terms in Coriolis and centrifugal vector is given
as:
g1 = 0
g2 = (m 4 + m 3 + 0.5m 2 )ga2 cos θ2 + (m 4 + 0.25m 3 )ga3 sin(θ2 + θ3 )
g3 = (m 4 + 0.5m 3 )ga3 sin(θ2 + θ3 )
Substituting these in the basic dynamic equation, we are able to calculate the
torque at the individual joints using the following equation:
τ = M(θ )θ̈d + C θ, θ̇ + G(θ )
After the complete analysis of the kinematics and the dynamics involved, it is impor-
tant to study the control algorithm used for tracing the desired trajectory. For this
purpose, a nonlinear control algorithm like the Computed Torque Control Method
is used to make the controller desirable in real-time conditions. CTC was one of the
very first models for control of nonlinear systems. Because of the nonlinear nature,
CTC is a desirable controller. As the presence of feedback (function of joint posi-
tion and velocity) loop helps in making the robot dynamics linear. It cancels all the
650 A. Katpatal et al.
nonlinearities and applies exactly the torque needed to overcome the inertia of the
actuators. The final control law is given by:
τ = M(θ )θ̈d + C θ̇ + G + M(θ ) −kv ė − k p e
τff τfb
The displacement and velocity at each joint were taken as input and used in computer
programs in MATLAB to calculate the error in the CTC control technique.
The trajectories chosen for the three joints are as follows:
π t 1 t
θ1 = − sin 2π
5 T 2π T
π t 1 t
θ2 = − sin 2π
8 T 2π T
π t 1 t
θ3 = − sin 2π
12 T 2π T
Model-Based Synchronized Control of a Robotic … 651
0≤t ≤T
where T is the time for which the trajectory is executed and whose upper limit is
10 s. Initially T = 0. The equation for joint friction is given as:
2 2
−q̇i −q̇i
τfriction = Fci sgn(q̇i ) 1 − exp 2
+ Fsi sgn(q̇i )exp + Fvi q̇i
vsi vsi2
where F ci , F si , and F vi are the coefficients for Coulomb, static, and viscous friction,
respectively.
Figures 3, 4, and 5 indicate the corresponding joint torques.
Here, i = 1, 2; The Cartesian coordinates of the ith robot is denoted by xid . β ∈ R 3x3
and Λ ∈ R 3x3 are two diagonal positive-definite matrices. ci = diag{ ci1 , ci2 , ci3 } ∈
R 3x3 ∩ C 2 , is an invertible coordination matrix.
From the above equations, time derivative of u i (t) can be given as:
The vector u i (t) considers only the reference signals and states.
θi denotes the dynamic parameter vector of the ith robot; the regressor matrix Yi
can be defined as,
Yi (xi , ẋi , u i , u̇ i )θi = Hi (xi )ci−1 (t)[u̇ i (t) − ċi (t)ẋi (t)]
+ C(xi , ẋi )ci−1 (t)u i (t) + G i (xi )
The resultant force matrix for a system comprising of two manipulators holding
a payload can be written as:
Model-Based Synchronized Control of a Robotic … 653
F1 γ I3 I3 FO
=
F2 (1 − γ )I3 −I3 FI
where Fi ∈ R 3 , i = 1, 2 denotes the contact force for the tip of the ith robot.
I3 ∈ R 3x3 is an identity matrix; γ ∈ [0, 1] is the load distribution constant between
the two arms; F1 ∈ R 3 represents the internal force, and FO ∈ R 3 is the net force on
the payload. The inverse for the above equation can be given as:
FO I3 I3 F1
=
FI (1 − γ )I3 −γ I3 F2
6 Conclusion
The equations for Kinematics and Dynamics for a Dual-Arm Manipulator System
using the DH parameters were obtained. The control technique for tracking the joint
torques of a given serial manipulator has been shown. The results of the simula-
tions for the CTC technique show the effectiveness of the solution. The mentioned
control technique can be applied to the actual model of the manipulator with some
modifications. Moreover, the gain selection process for the control system can be
made automatic using a tuning approach. The stated synchronization algorithm for
the purpose of controlling both the manipulators will be implemented in the future.
The ultimate aim with the application of the synchronization algorithm is to have
multiple arms working in a coordinated manner.
References
1. Liu S, Wu L, Lu Z (2007) Impact dynamics and control of a flexible dual-arm space robot
capturing an object. Appl Math Comput 185(2):15
2. Cannon RH, Schmitz E (2014) Initial Experiments on the end-point control of a flexible one-link
robot. Int J Robot Res 3(3)
3. Azadi M, Eghtesad M, Gharesifard B (2005) Inverse dynamics control of two 5 DOF coop-
erating robot manipulators. In: International design engineering technical conferences and
computers and information in engineering conference. California, USA, 24–28 Sept 2005
4. Khatib O Object manipulation in a multi-effector robot system. In: Robotics research: the forth
international symposium, pp 137–144
5. Hosek M (2006) An actuation mechanism for a dual-arm robotic manipulator. In: International
mechanical engineering congress and exposition. Chicago, Illinois, USA, 5–10 Nov 2006
6. Krishnasamy J, Hosek M (2007) Design optimization of an actuation mechanism for a dual-
arm robotic manipulator. In: International mechanical engineering congress and exposition.
Seattle, Washington, USA, 11–15 Nov 2007
7. Ghaemi Osgouie K, Meghdari A, Sohrabpour S (2007) Optimal configuration of A 4-DOF dual-
arm cam-lock manipulator. In: 2007 International design engineering technical conferences and
computers and information in engineering conference
654 A. Katpatal et al.
1 Introduction
Transportation is the backbone for economic and social development of any coun-
try. According to the 2011 census, 31.2% of India’s population is living in urban
areas. This 31.2% urban population is responsible for approximately 63% of India’s
gross domestic product (GDP) [1]. With rapid growth in urbanization, a strong need
of improving the existing public transportation is felt throughout the country. The
present paper deals with understanding travelling criteria considered by intra-city
commuters in Nagpur city. The criteria are further prioritized using AHP.
Nagpur is the third-largest city and second capital of Maharashtra state in India.
It is the second-largest city by population in central India as per as 2001 and 2011
censuses. According to [2], the GDP of Nagpur would increase from 8.41 to 48.36%
by 2035 making it fifth fastest growing city in India. Upcoming metro railway project
has given new dimension to the intra-city transportation.
It is estimated that Nagpur Municipal Corporation (NMC) has incurred a loss of
about 941 million rupees for operating PTS for Nagpur city for year 2018–19 [3]. To
minimize these losses, NMC runs buses on longer route catering all rural areas and
villages which surround Nagpur city. With addition of rural population, city buses
have become overcrowded and the passengers from the downtown hardly find any
space to board the bus. Hence, a need is felt to change the focus from infrastructure-
based model to customer-centric model so as to accommodate more and more urban
passengers.
2 Literature Review
3 Methodology
evaluation of these parameters could be done using 9-point Likert’s scale as given in
Table 1.
In order to conduct the survey in an effective way, survey card was designed to indicate
the relative importance, expected level and observed level of each of the 14 travelling
parameters. A commuter will perceive different importance for each of the travelling
parameters. Hence, a rank (from 1 to 14) based on the relative importance can be
provided to each of the 14 parameters. The expected level and observed level were
rated using 9-point Likert’s scale as described in Table 1. The survey was conducted
658 P. Lanjewar and D. A. Jolhe
among 50 potential commuters in the age group of 20–60 years. For the purpose
of further analysis, the survey participants were classified into three segments as
follow: segment 1: bus commuters (BC), segment 2: two-wheeler riders (age group
20–40 years) (TWR-20/40), segment 3: two-wheeler riders (age group 40–60 years)
(TWR-40/60). Table 2 shows a survey card filled in by a commuter.
TTU 0.1 0.1 0.2 0.2 0.3 0.5 2.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
AE 0.1 0.1 0.2 0.2 0.3 0.3 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
P 0.1 0.1 0.1 0.1 0.3 0.5 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
TP 0.1 0.2 0.2 0.3 0.3 0.3 2.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
BA 0.1 0.2 0.2 0.3 0.3 0.5 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
OS 0.3 0.3 0.3 1.0 0.5 0.5 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
659
660 P. Lanjewar and D. A. Jolhe
Table 4 Relative weights of all travelling attributes for the three categories of commuters
S. No. Travelling parameter Commuter category
BC TWR-20/40 TWR-40/60
1 Travel time 0.16 0.10 0.09
2 Travel cost 0.22 0.09 0.03
3 Safety and security 0.14 0.03 0.21
4 Reliability 0.10 0.05 0.05
5 Mobility 0.09 0.24 0.03
6 Accessibility 0.06 0.10 0.03
7 Comfort/discomfort 0.03 0.18 0.03
8 Driving stress 0.03 0.03 0.14
9 Travel time utilization 0.03 0.03 0.08
10 Additional employment 0.03 0.03 0.03
11 Air and sound pollution 0.03 0.03 0.05
12 Traffic problem 0.03 0.03 0.11
13 Bus stops ambience 0.03 0.03 0.09
14 Overall satisfaction 0.04 0.03 0.03
indicated in Table 4. During the study, a consistency ratio of 0.06 is obtained, which
is less than 0.10, indicating the survey undertaken was reliable and consistent.
The pair-wise comparison carried out in the previous section provides some useful
directions to the PTS under study. The three segments of commuters have different
need sets with a small overlap. The priorities of the travelling parameters based on
the relative weights, for each segment of commuters, are given in Table 5.
It can be observed that the bus commuter (BC) segment, i.e., those who are using
bus service of the PTS, are expecting lower travel cost, lower travel time, safety and
security during travel and reliability of the service. This is very natural since most of
the commuters are from long distance rural areas. A big portion of these commuters
is college students which make reliability of service more important for them.
The two-wheeler riders (age 20–40 years) (TWR-20/40) are more inclined to
mobility (i.e., fast movement between origin and destination), comfort during jour-
ney, accessibility of transport vehicle and lower travel time. These are young age
commuters (large portion of which are employed youth) who prefer to use personal
vehicles for their conveyance.
The third segment, i.e., two-wheeler riders (TWR-40/60), expects more safety
and security, and lesser driving stress; and wants to avoid traffic problems and to
reach their destination on time. It should be noted that travel cost does not bother
to this segment. Obviously, this segment is likely to pay more to ensure safety and
minimum driving stress. Hence, a customized bus service with premium fare can
be provided that will boost revenue collection stream in NMC, and it will help to
overcome its financial losses.
6 Conclusion
The study provided systematic approach to priorities the travelling parameters using
pair-wise comparisons. A total of 14 travelling parameters were identified through
discussion with the PTS authorities. The survey was conducted among the poten-
tial users of PTS to understand their importance, expected level of satisfaction and
observed level of satisfaction about the 14 parameters. The pair-wise comparison
helped to find relative weights of the travelling parameters for each of the three
commuter segments. It was observed that the three segments have different need
sets, which could be fulfilled by providing customized services. The study helped
to identify a new commuter segment, i.e., TWR-40/60, which could be attracted
by providing premium customized bus service. It should be noted that to improve
the accuracy of findings, more sample size is required. Further, more segments of
potential commuters need to be defined to widen the scope of customized transport
services.
Acknowledgements Note Before conducting the survey, oral consent was obtained from each of
the survey participants.
662 P. Lanjewar and D. A. Jolhe
References
1 Introduction
Magnesium and its alloys are light-weight materials compared with steel and alu-
minum alloys. High strength to weight ratio, low density, and good damping capacity
of magnesium alloys make them as promising candidates for applications in trans-
portation and automobile industries [1, 2]. However, the wear resistance of magne-
sium alloys is very poor which limits their applications where structures subject to
relative motion. Several techniques have been demonstrated to enhance the mechani-
cal and tribological properties of magnesium alloys [3, 4]. Tribological properties can
also be improved by altering the microstructure of materials through grain refinement.
2 Experimental Details
As cast AZ91 Mg alloy with a chemical composition of 8.67% Al, 0.85% Zn, 0.002%
Fe, 0.03% Mn, and remaining being Mg by wt% was obtained from Exclusive Mag-
nesium, Hyderabad, India, and plates of size 100 mm × 50 mm × 5 mm were
machined from the billet. Figure 1a shows the image of FSP tool used in the present
study. Tool was fabricated from a round bar of H13 tool steel and was hardened to
52 HRC [8]. Figure 1b shows the stirring of material during FSP and Fig. 1c shows
the FSPed plate.
Samples were extracted from the processed plate and base material (BM) for
conducting microstructure, hardness, and wear studies. For microstructure study,
samples were polished using different grades of emery papers. Polishing was further
continued through disk polishing machine using diamond paste (1–3 μm). Samples
were cleaned with ethanol and then, chemically etched using picric acid reagent to
observe the microstructure of the FSPed sample. Hardness of samples was measured
across the processed zone using Vickers microhardness tester (Omnitech, India)
by applying 100 g load for 10 s. To explore the tribology characteristics of BM
and FSPed samples, wear tests were conducted on both samples by using a linear
reciprocating tribometer (Ducom TR-281, India). AISI52100 steel balls of 6 mm
diameter and 62 HRC were used as counter body material. Tests were conducted for
a sliding distance of 500 m, under the normal load of 10 N and sliding velocity of
0.06 m/sec at room temperature. Scanning electron microscope (SEM) and energy
dispersive X-Ray (EDS) analysis were carryout on the worn surface of the tested
samples obtained after the tests to understand the governing wear mechanisms for
the samples.
Effect of Friction Stir Processing on the Sliding Wear … 665
Fig. 1 a Photograph of FSP tool used for FSP, b photograph obtained during FSP and c FSPed
AZ91 Mg alloy plate
Figure 2a, b shows the optical micrograph images of BM AZ91 Mg and FSPed
AZ91 Mg alloy. Microstructure of BM reveals the presence of solid solution grains
(α), intermetallic phase (β) Mg17 Al12, and eutectic region (α + β). Microstructure of
FSPed AZ91 sample reveals a decrease in grain size compared to BM. Grain size was
reduced from 166.5 μm of BM to 4.08 μm of FSPed AZ91. Intermetallic phase (β)
Mg17 Al12 was observed as reduced after FSP. Due to frictional heating, intermetallic
phases started to dissolve in the solid solution grains.
Figure 3a shows the hardness distribution across the stir zone of FSPed AZ91
sample. Hardness was observed to be increased in the stir zone. Figure 3b shows
the average microhardness of BM and FSPed sample. FSPed sample shows higher
hardness (107.4 HV) compared with the hardness of BM (95.5 HV). As reported
in the earlier studies on FSP, intense plastic deformation in the material, this in
turn refines the microstructure of FSPed material. According to Hall–Petch relation,
decrease in grain size leads to the increment in the hardness of substrate. However,
dissolution of hard brittle intermetallic phase in the solid solution phase decreases
the hardness of substrate [9, 10]. Enhancement in hardness reveals that the grain size
reduction plays more dominant role than the dissolution of hard brittle intermetallic
phase in solid solution during FSP.
Figure 4a shows the friction curves obtained for BM and FSPed AZ91 samples
during the wear tests. Average coefficients of friction were observed as the same
for both the samples. However, more fluctuations were observed in friction curve
obtained for FSPed AZ91 compared with BM. Figure 4b shows the mass loss obtained
for both samples after a sliding distance of 500 m. FSPed AZ91 sample shows lower
mass loss compared with BM sample. Mass loss results confirm the enhancement
in wear resistance of AZ91 Mg alloy after the FSP. Figure 5a, b shows the SEM
micrograph of worn surface morphology obtained for BM and FSPed samples. Deep
parallel grooves were observed on the wear track of BM and FSPed samples. Hard
Fig. 3 a Hardness distribution across the stir zone of FSP AZ91 Mg and b average hardness of
BM AZ91 Mg and FSPed AZ91
Effect of Friction Stir Processing on the Sliding Wear … 667
Fig. 4 a Coefficient of friction and b mass loss graph for BM AZ91 Mg and FSPed AZ91 Mg
Fig. 5 SEM micrographs of worn surface of wear track obtained after sliding distance of 500 m
for a BM AZ91 Mg and b FSPed AZ91 Mg
surface of the counter body removes the material from the surface of the soft matrix.
This is the typical abrasive wear mechanism usually observed in the wear testing
of magnesium alloy [11]. Some craters can also be observed on the wear track of
BM sample. Due to repeated sliding action of counter surface, subsurface cracks
formed. Series of subsurface cracks are joined together and lead to the formation of
craters on the wear track. This is a typical feature of de-lamination wear as reported
in previous studies [12]. However, due to higher hardness of FSPed sample, more
delay in the formation of cracks occur underneath the surface compared with BM
sample and thus shows more wear resistance than BM sample. Figure 6a, b shows
the morphology of debris obtained from BM and FSPed samples. Due to the abrasive
action of counter body, material was removed from the test sample in the form of
flake and sheet-like structure. EDS analysis of wear debris confirms the presence of
oxide layer formed during the wear tests in both samples. Formation of oxide layer
in FSPed sample is more compared with BM. Oxide layers act as a protective layer
for soft matrix surface by avoiding the contact between hard counter surface and soft
matrix surface under the certain limit of loads and sliding velocities [11].
668 H. Patle et al.
Fig. 6 SEM micrographs of wear debris obtained after the wear tests of a BM AZ91 Mg and
b FSPed AZ91 Mg
4 Conclusions
In the present study, effect of FSP on microstructure, hardness, and wear resistance of
AZ91 Mg alloy was studied. Microstructural observation showed the significant grain
refinement after the FSP. Intermetallic phase (Mg17 Al12 ) was observed as reduced
after FSP. Hardness of FSPed AZ91 Mg (107.4 HV) was observed as increased
compared with BM AZ91 Mg (95.5 HV). It was also observed that the effect of grain
refinement was found to be more effective than the dissolution of the intermetallic
phase on increasing the hardness. Wear resistance of FSPed AZ91 Mg was observed
as slightly increased after the FSP. This is because of increased hardness of sample
after FSP. Abrasive and oxidative wear were the common activewear mechanisms
observed in both the samples. However, increased hardness of FSPed AZ91 Mg
delays the formation of microcracks underneath the surface and thus increases the
wear resistance for FSPed sample compared with BM sample.
Effect of Friction Stir Processing on the Sliding Wear … 669
References
1 Introduction
A laminated composite comes under the class of thin-walled structures, which are
regularly found in several engineering fields, especially in the field of aerospace
industry. The demand for low weight and high strength aerospace structures lead to
the development of fiber-reinforced composite structures. While in operation, these
aerospace structures are subjected to in-plane loads. A laminated panel under the
influence of in-plane loads is susceptible to buckle at mere small stress levels, and
K. S. S. Chandra (B)
BMS College of Engineering, Bengaluru 560019, India
e-mail: siddachandra@gmail.com
K. V. Rao
Department of Mechanical Engineering, BMS College of Engineering, Bengaluru 560019, India
T. Rajanna
Department of Civil Engineering, BMS College of Engineering, Bengaluru 560019, India
© Springer Nature Singapore Pte Ltd. 2021 671
V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering,
Lecture Notes in Mechanical Engineering,
https://doi.org/10.1007/978-981-15-3639-7_80
672 K.S.S. Chandra et al.
hence is the topic of concern for researchers for over several years. The buckling
behavior of laminated panels responding to uniform edge loads has remained the
problem of extended investigations [1, 2]. Nevertheless, in several realistic stability
problems, the enforced edge load is non-uniform in nature [3, 4]. In such instances,
the stress distribution is likewise non-uniform. The solution for such problems is
rendered quite complicated. The difficult in solving the buckling problem is further
enhanced by inducing cutout within the panel. The panel with cutout is commonly
found in aerospace and ship structures in order to facilitate inspection for fuel/water
lines and sometimes to simply lessen the overall weight of structures. Introduction
of cutout within a panel reduces overall strength resulting premature failure due to
buckling. Hence, a sound consciousness of buckling characteristics is necessary to
design the laminated panel with cutout during the act of non-uniform edge loads.
Ghannadpour et al. [5] analyzed buckling of circular cutout in cross-ply laminate
under uniformly distributed load. Jain and Kumar [6] interpreted the effect of com-
pressive load with circular cutout on post-buckling responses. Baba et al. [7] studied
centered circular cutout in E/glass epoxy composite plate under uniaxial compres-
sion. Recently, Rajanna et al. [8] investigated buckling characteristics of centrally
placed square and circular cutouts under compressive and tensile edge loads.
It is evident from previous studies that the effect of different sizes of cutouts
and their position along with various types of non-uniform edge loads is rarely
reported and no comprehensive work is identified in the literature. Hence, in the
present investigation, a four-layered antisymmetric angle-ply and cross-ply panel
is considered to study the consequence of cutout and non-uniform edge loads on
buckling response by using commercially available FE package ABAQUS [9].
The square laminated panel considered for the study has dimension ‘a’ in x-direction
and ‘b’ in y-direction with constant thickness ‘t’ and has a centered circular cutout
Fig. 1 a Geometry of a laminated panel including a circular cutout, b illustration of linearly varying
loads for different values of α
Effect of Varying in-Plane Loads and Cutout Size on Buckling … 673
of diameter ‘d’ as in Fig. 1a. It has four laminas; each lamina has different ply
orientation ‘θ ’.
The governing equilibrium equation for a plate under in-plane load on edge written
[m]{q̈} + [κ] − pcr κg {q} = {0} (1)
For buckling analysis {q̈} becomes zero, so Eq. (1) reduce to static
[κ]{q} − pcr κg {q} = {0} (2)
The Eigenvalue in Eq. (2) gives the buckling load for modes. The buckling load
having least value is known as the critical or fundamental load.
Fig. 2 a A perforated square plate under varying load when α = 0.5, b selected meshing pattern
and simply supported boundary condition with loading edges
674 K.S.S. Chandra et al.
Table 1 Comparision of buckling loads of cross-ply laminate with circular cutouts for uniformly
distributed uniaxial load under simply supported condition relevant to previous literatures [5, 8]
with the properties mentioned below
d/b Ratio (Present) Rajanna et al. (2016) Ghannadpour et al. (2006)
0.1 12.82 12.83 12.80
0.5 6.41 6.40 6.40
E 11 = 130.0 GPa, E 22 = 10.0 GPa, G12 = G13 = 5.0 GPa, G23 = 3.4 GPa, υ 12 = 0.35
Pcr b2 /E 22 t 3 (4)
where Pcr is the absolute critical load and E 22 is laminate stiffness in the lateral
direction.
3 Comparative Studies
In order to assess the accuracy and effectiveness engaged in the buckling analysis
of present investigation, the evaluation of present work along with previous studies
is essential. The results obtained for a laminated square panel with the cutout is
validated with those of Ghannadpour et al. [5] and Rajanna et al. [8]. The present
results obtained using S8R5 element in ABAQUS are in good agreement with the
above said literature (Table 1).
The effect of size of the cutout and non-uniform edge loads on the buckling behavior
of cross-ply and angle-ply laminated cutout panels are shown in Figs. 3 and 4 respec-
tively. It is evident from Fig. 3 that the buckling parameter (γ cr ) decreases with an
increase in cutout regardless of loading types (α). Higher values of γ cr were observed
in case of panels with angle-ply orientations, when compared to those with cross-ply
orientations irrespective of loading conditions (α). When panels of angle-ply orien-
tations are under the influence of pure bending case (α = 2.0), exhibit increasing
tendency in buckling load parameter (γ cr ) for d/b > 0.5 as seen from Fig. 4. This may
Table 2 Effect of loading parameters, thickness, and boundary conditions on (± 45o )2 panel
d/b = 0.6 b/t = 100 b/t = 20 b/t = 10
Load forms SSSS CCCC SSSS CCCC SSSS CCCC
α = 0.0 26.61 42.62 9.84 9.91 2.49 2.49
α = 0.5 34.54 53.68 12.64 12.64 3.18 3.18
α = 1.0 46.98 71.96 17.28 17.36 4.37 4.37
α = 1.5 71.77 108.9 26.91 27.05 6.89 6.89
α = 2.0 146.9 220.23 46.4 46.6 11.87 11.88
be due to the redirection of stresses from the cutout region to the stiffer region, which
is usually observed in the panels with higher cutout size. This phenomenon of redi-
rection of stresses is predominating in the panels with angle-ply laminates-especially
for α = 2.0.
The effect of circular cutout offset ratio (e/b) from center of panel towards positive
x-axis and y-axis is studied and the outcomes are shown in Figs. 5 and 6, respectively
by considering the cutout size, d/b = 0.6 and layup scheme, θ = (0/90)2 .
It is evident from Fig. 5, for any given loading conditions, the buckling resistance
of the panel remains almost constant for any value of cutout offset ratio (e/b) except
for the loading parameters (α), α = 1.5 and α = 2.0, wherein the result of γ cr
decreases with increasing cutout offset ratio. This behavior is completely reversed
Effect of Varying in-Plane Loads and Cutout Size on Buckling … 677
when the cutout offset moves towards positive y-axis, whereupon the results of γ cr
increases along with cutout offset, which can be observed in Fig. 6.
5 Concluding Remarks
The effect of geometry of panel with cutout and loading conditions on buckling
behavior of laminated panel with cutouts is studied in the present work. Based on
the results obtained, the following conclusions are drawn:
• The buckling intensity in cutout panel decreases when cutout size is increased.
This is true for all loading conditions except α = 2.0 for angle-ply layup which
678 K.S.S. Chandra et al.
exhibits increasing tendency in buckling load parameter (γ cr ) for d/b > 0.5 due to
the redirection of stresses from cutout region to stiffer region.
• The buckling resistance of the angle-ply layup scheme is significantly higher as
compared to that of cross-ply laminate scheme irrespective of cutout size and load-
ing conditions because the stresses in angle-ply cutout panel are more distributed
throughout the panel than in cross-ply cut out panel.
• For all the provided cutout ratios (d/b), the buckling resistance of cutout panel
increases with an increase in load parameter (α) and is found to be highest when
panel subjected to pure bending (i.e., α = 2.0) and is least for α = 0.0.
• With regard to buckling, the heavy panels with SSSS and CCCC boundary
conditions are least affected, while it is significantly considerable for thin panels.
• Reduction in buckling strength is observed when cutout is moved towards loading
edges and it is found to increase when the cutout is moved towards unloaded edges.
References
1. Timoshenko SP, Gere JM (1961) Theory of elastic stability. McGraw-Hill, New York
2. Whitney JM (1987) Structural analysis of laminated anisotropic plates. Technomic Publishing,
Lancaster
3. Zhong H, Gu C (2007) Buckling of symmetrical cross-ply composite rectangular plates under a
linearly varying in-plane load. Compos Struct 80:42–48
4. Hwang I, Lee JS (2006) Buckling of orthotropic plates under various in-plane loads. Struct Eng
10(5):349–356
5. Ghannadpour SAM, Najafi A, Mohammadi B (2006) Buckling of Cross-ply laminate composite
plates due to circular/elliptical cutouts. Compos Struct 27:3–6
6. JainP Kumar A (2004) Post buckling response of square laminates with a central/elliptical cutout.
Compos Struct 75:179
7. Baba BO, Baltaci A (2007) Buckling characteristics of symmetrically and anti-symmetrically
laminated composite plates with central cutout. Appl Compos Mater 14:265–276
8. Rajanna T, Banerjee S, Desai YM, Prabhakara DL (2016) Vibration and buckling analyses of
laminated panels with and without cutouts under compressive and tensile edge loads. Steel and
Compos Struct 21(1):37–55
9. Dassault Systems Simulia Corp. (2009) Abaqus analysis user’s manual, Version 6.4, Providence,
RI, USA
Effect of Machining Parameters
on Surface Roughness and Tool Flank
Wear in Turning of Haynes 25 Alloy
Abstract This work presents an analysis of the effect of cutting parameters on sur-
face roughness and tool flank wear in turning of Haynes 25 superalloy using coated
carbide tool under minimum quantity lubrication. Response surface method is used
for experimentation and analysis. Experiments were designed using the central com-
posite design of response surface methodology. Prediction models are developed
for each response using response surface methodology. Analysis of Variance is car-
ried out to find the significant parameters and percentage contribution of process
parameters on response. Main effect plots are used to analyze the effects. It is found
that the effects of cutting parameters are different for the two responses and further
optimization is necessary to get the best machining performance.
1 Introduction
2 Experimentation
For experiment design, three input parameters (cutting speed, feed, and depth of
cut) were selected and two responses (surface roughness average and tool wear)
were measured. Experiments were designed with the help of the Central Composite
Design technique of Response Surface Method. According to this design, five levels
of inputs were used for experimentation. The experiment sequence was obtained
from Minitab software and 20 experiments were required to be carried out according
to experiment design.
The experiments were performed on CNC lathe machine; MTAB MAXTURN+.
The workpieces used in this experiment were bars of Haynes 25 cobalt-based super-
alloy having diameter = 32 mm and total length = 100 mm. The cutting length was
20 mm for each experiment. Widia make of PVD coated Tungsten carbide inserts with
Zirconium coating (TNMG160406 UF) mounted on a tool holder MTJNL 2525M16
were used for machining. MQL system of Dropco make was used to provide a mist
of fluid during experiments.
Surface roughness (Ra ) was measured by Mitutoyo make Surftest SJ-410 rough-
ness tester. Flank wear of the tools was measured by means of USB Microscope
ISM PM600SB from INSIZE Ltd. Table 1 shows the responses measured for the 20
experiments.
3 Analysis of Responses
The responses (Ra and VBmax ) as recorded in Table 1 are used to develop models using
Response Surface Methodology and the developed models are analyzed statistically
by ANOVA technique to determine the significance of various parameters. Using the
data of Table 1, regression coefficients obtained for the model of surface roughness
are shown in Table 2.
From Table 2, the prediction model for surface roughness is developed which
contains speed, feed rate, and depth of cut as independent variables and it is shown
in Eq. 1.
For this model, R2 = 96.84% which is closer to 100%, so, regression fit and
estimates are pretty good and the model is able to predict new observations. R2
(adj) value is 94.00% which is also very good and model can predict the roughness
correctly. ANOVA results of the above model are shown in Table 3. From this Table,
it can be seen that for 95% confidence level, the interaction terms V * f and V * d
are not significant and can be removed.
Effect of Machining Parameters on Surface Roughness … 681
Here, R2 = 89.03% and R2 (adj) = 79.15% which indicates that adjusted model
is poor than the original model. ANOVA results of the above model are shown in
Table 5. From this Table, it can be seen that for 95% confidence level, the square
terms and interaction terms are not significant.
Figures 1 and 2 show main effect plots for surface roughness and tool wear. The main
effects plot of surface roughness (Fig. 1) shows pronounced improvement in the sur-
face roughness value with increasing cutting speed and feed rate, but the improvement
is only up to a certain threshold value of 45 m/min speed and 0.075 mm/rev feed.
Beyond that value, the surface roughness degrades. With the increase of cutting
speed, temperature increases at cutting region, so the material becomes soft and it
is easy to remove material, so surface roughness is decreasing. However, excessive
increase in cutting speed leads to early tool wear, because of which surface roughness
degrades at higher cutting speed. Increased feed rate results in increased feed force
that affects negatively on surface roughness. Depth of cut has a linear relationship
with the surface roughness; it increases as the depth of cut increases as more volume
of material is cut.
The main effects plot of Fig. 2 shows a variation of tool wear. Tool wear has
increased with an increase of cutting speed because increase of cutting speed results
in an increase in volume of material being cut, along with increase in friction and
temperature. All these effects cause increase in tool wear. Tool wear has decreased
with increase in feed rate, because of the lower availability of contact time between
tool and workpiece. Tool wear has increased with increase of depth of cut, because
the depth of penetration increased. The main effect plot clearly shows that minimum
tool wear occurs at lower cutting speed, higher feed rate, and lower depth of cut.
Depth of cut is a more significant parameter for the tool wear followed by feed rate
and cutting speed.
5 Conclusion
References
Y. V. Deshpande (B)
Department of Industrial Engineering, Shri Ramdeobaba College of Engineering and
Management, Nagpur 440013, India
e-mail: deshpandeyv@rknec.edu
A. B. Andhare · P. M. Padole
Department of Mechanical Engineering, Visvesvaraya National Institute of Technology, South
Ambazari Rd., Ambazari, Nagpur, Maharashtra 440010, India
1 Introduction
2 Experimentation
Inconel 718 work parts (22 × 150 mm) were selected for machining. Kennametal
make, uncoated tungsten carbide inserts (TNMG 160408) with dual chip breaker
geometry, was selected for experiments. Each experiment is performed using a new
tool edge at a cutting length of 60 mm. Insert type, cryogenically treatment cycle,
MQL and LN2 flow rate are selected based on the literature [1, 9, 10].
The machining conditions used in turning of Inconel 718 are: (1) dry machining
with cryogenically treated inserts (DRY-CT), (2) minimum quantity lubrication with
Performance Appraisal of Cryogenically Treated Tool … 689
cryogenically treated inserts (MQL-CT) and (3) MQL and liquid nitrogen (cryo-
gen) with cryogenically treated inserts (MQL-CRYO-CT). For condition 2, BioCut
3600 coolant is used through the MQL spraying system (make: DROPCO) with the
pressure of four bars and flow rate of 80 mlph. Similarly, for condition 3, liquid
nitrogen is passed through a vacuum insulated pipe at the rate of 1 lph beside MQL.
The deep cryogenic treatment (−196 °C) cycle can be helpful for better wear resis-
tance of tungsten carbide tools [1, 2]. Therefore, a similar deep cryogenic treatment
cycle was preferred in this research. Uncoated carbide tools are treated using deep
cryogenic cycle at a rate of 0.5 °C [1].
Turning operation was executed on CNC lathe machine as shown in Fig. 1. The
piezoelectric dynamometer is used to measure the cutting forces. The accelerometer
is placed to measure tool vibrations. National Instrument’s DAQ 9178 is used to
acquire the output from dynamometer and accelerometer. Surface roughness of work
parts was measured with portable surface roughness tester. The tool flank wear at
nose radius was measured with the help of USB microscope. The cutting temperature
was measured using a digital thermometer (Mextech IR 1800 with thermocouple).
For designing the experiments, five levels are coded for three input factors like cut-
ting speed (v), feed rate (f ) and depth of cut (d) using central composite design (CCD)
of response surface methodology (RSM) in MINITAB 17 software. The ranges of
v, f and d are selected as 9.5–110.45 m/min, 0.06–0.23 mm/rev and 0.32–0.17 mm,
respectively. The experimental strategy is applied, and the values of responses such
as cutting temperature (T ), cutting force (F c ), tool vibration (As ), surface rough-
ness (Ra ) and tool flank wear (VBmax ) were measured for the three conditions used,
i.e., DRY-CT, MQL-CT and LN2 -MQL-CT. The input and measured values of pro-
cess parameters are considered for further processing but not presented here by
considering the page limit of this paper.
Fig. 1 Experimental setup (left) and use of cooling conditions in turning of Inconel 718 (right)
690 Y. V. Deshpande et al.
The values of input and response parameters are used for evaluating their relation-
ship using RSM. The empirical models display an association between response
parameters (T, F c , As , Ra and VBmax) and cutting parameters (v, f, d) in the turning
of Inconel 718. The capability of developed models was confirmed using analysis of
variance (ANOVA) and correlation coefficient (R2 ). The developed models for esti-
mated responses have been compared with experimental data using regression plots.
Based on R2 values (greater than 90%), the RSM models are found to be capable for
estimating responses for all machining conditions. It is concluded from the entire
assessment of regression plots that the models are established a strong relationship
between estimated and experimental responses.
The models are designed for assessing the performance in the turning of Inconel 718
for three machining conditions. Figures 2 and 3 show the plots of the RSM models
for surface roughness and tool wear, respectively. The cutting parameters (v, f and
d) against the responses such as surface roughness and tool wear are presented in the
context of three machining conditions like DRY-CT, MQL-CT and LN2 -MQL-CT.
Figure 2 shows that there is a significant enhancement in the surface roughness by
increasing cutting speed. However, for increasing feed and depth of cut, the value
the surface finish degrades. It is also observed that the reduction in roughness values
is observed at high speed and moderate feed and depth of cut for three machining
conditions shown in Fig. 2a–c. The high value of roughness is noticed for DRY-CT
condition, whereas the low value of Ra is noted in LN2 -MQL-CT condition. The
value of roughness is highest for the DRY-CT condition at low cutting speed and
3 2 2
2.5
1.5 1.5
2
Ra (μm)
1.5 1 1
1
0.5 0.5
0.5
0 0 0
0 50 100 150 0 0.1 0.2 0.3 0 0.5 1 1.5
v (m/min) f(mm/rev) d(mm)
(a) (b) (c)
1 1.2 1
0.8 1 0.8
0.8
VBmax (mm)
0.6 0.6
0.6
0.4 0.4
0.4 DRYc
Fig. 3 Performance plot of cutting parameters against tool flank wear (VBmax )
lowest at high cutting speed. The same results are reported by Devillez et al. [11] and
Deshpande et al. [12] in turning of Inconel 718. They concluded that at high cutting
speed, more heat is built up at tool–work interface which softens the workpiece
result in low cutting force. The increase in feed and depth of cut results in breaking
of cutting tool which severely increases the surface roughness and tool wear.
Tool flank wear at the nose is selected as one of the response parameters. It is
the most significant wear in the different forms of tool wear, and it considerably
increases the cutting forces and vibrations which affect the quality of machined part.
Tool wear variations against cutting parameters (v, f, d) are shown in Fig. 3. The
behavior of wear is almost the same for all the machining conditions. The cutting
speed is more significant for wear compared to f and d. The wear is very high in the
DRY-CT condition, as an extreme amount of heat is produced and material erosion
takes place. The material removal rate gradually decreases, high for the DRY-CT
condition and low in case of LN2 -MQL-CT condition as shown in Fig. 3a–c.
In LN2 -MQL-CT condition, the cutting force, tool vibration and cutting temper-
ature are observed to be less as compared to DRY-CT and MQL-CT conditions.
MQL condition helps to remove heat from the cutting edge through cool air and
the lubrication at the tool–chip interface. The dual effect of MQL and cryogenically
treated insert results in a reduction in surface roughness and tool wear. The spraying
of cryogen (LN2 ) at the tool–chip interface reduces the cutting temperature beside
the use of cryogenically treated insert. This is accountable for the increased wear
resistance of tool which further helps to maintain the tool sharpness even at high
temperature. Here also, the combined effect of LN2 and MQL with cryogenically
treated insert results in minimum cutting force, tool vibration and cutting tempera-
ture. These parameters therefore result in a reduced value of surface roughness and
tool wear.
692 Y. V. Deshpande et al.
4 Conclusion
References
1. Deshpande YV, Andhare AB, Padole PM (2018) Experimental results on the performance of
cryogenic treatment of tool and minimum quantity lubrication for machinability improvement
in the turning of Inconel 718. J Braz Soc Mech Sci Eng 40(1):6
2. Deshpande YV, Andhare AB, Padole PM (2018) How cryogenic techniques help in machining
of nickel alloys? A review. Mach Sci Technol 1–42. https://doi.org/10.1080/10910344.2017.
1382512
3. Tazehkandi AH, Shabgard M, Pilehvarian F (2015) Application of liquid nitrogen and spray
mode of biodegradable vegetable cutting fluid with compressed air in order to reduce cutting
fluid consumption in turning Inconel 740. J Clean Prod 108:90–103
4. Pusavec F, Deshpande A, Yang S, M’Saoubi R, Kopac J, Dillon OW, Jawahir I (2015) Sus-
tainable machining of high temperature nickel alloy–Inconel 718: part 2–chip breakability and
optimization. J Clean Prod 87:941–952
5. Kaynak Y, Karaca H, Jawahir I (2015) Cutting speed dependent microstructure and trans-
formation behavior of NiTi alloy in dry and cryogenic machining. J Mater Eng Perform
24(1):452–460
6. Homami RM, Tehrani AF, Mirzadeh H, Movahedi B, Azimifar F (2014) Optimization of turning
process using artificial intelligence technology. Int J Adv Manuf Technol 70(5–8):1205–1217
7. Thakur D, Ramamoorthy B, Vijayaraghavan L (2009) Optimization of high speed turning
parameters of superalloy Inconel 718 material using Taguchi technique. Indian J Eng Mater
Sci 16:44–50
8. Jawahir I, Attia H, Biermann D, Duflou J, Klocke F, Meyer D, Newman S, Pusavec F, Putz M,
Rech J (2016) Cryogenic manufacturing processes. CIRP Ann Manuf Technol 65(2):713–736
9. Pusavec F, Deshpande A, Yang S, M’Saoubi R, Kopac J, Dillon OW Jr, Jawahir IS (2014)
Sustainable machining of high temperature nickel alloy—Inconel 718: part 1—predictive
performance models. J Clean Prod 81:255–269. https://doi.org/10.1016/j.jclepro.2014.06.040
10. Thakur DG, Ramamoorthy B, Vijayaraghavan L (2009) Machinability investigation of Inconel
718 in high-speed turning. Int J Adv Manuf Technol 45(5):421. https://doi.org/10.1007/s00170-
009-1987-x
Performance Appraisal of Cryogenically Treated Tool … 693
11. Devillez A, Le Coz G, Dominiak S, Dudzinski D (2011) Dry machining of Inconel 718,
workpiece surface integrity. J Mater Process Technol 211(10):1590–1598
12. Deshpande Y, Andhare A, Sahu NK (2017) Estimation of surface roughness using cutting
parameters, force, sound, and vibration in turning of Inconel 718. J Braz Soc Mech Sci Eng1–10.
https://doi.org/10.1007/s40430-017-0819-4
Finite Element Simulation for Turning
of Haynes 25 Super Alloy
1 Introduction
Machining of superalloys is a challenging task due to problems like rapid tool wear.
Doing experiments for analysis of metal machining is costly and needs lot of time.
An alternative and faster way to analyze machining process of such materials is to
use computational technique like finite element method.
The present work analyzes the effects of speed, feed rate and depth of cut on cutting
and thrust forces in turning of Haynes 25 alloy using PVD coated tungsten carbide
tool. A three-dimensional machining model for simulation of cutting forces is used in
DEFORM 3D. Johnson–Cook material model is used in the finite element model to
study the deformation behavior of Haynes 25 material during the machining process
and the Coulomb friction (constant friction) model is considered to incorporate the
friction characteristics. Later, simulation results are compared with experimental
data.
While using FE method, some assumptions are made for simulation. These assump-
tions help to define the problem and to apply the boundary and loading conditions.
The assumptions are as below.
• Feed is much smaller than the width of cut and both are assumed to be constant.
Thus, the plain strain conditions are present at the cutting region.
• The vector of cutting velocity is perpendicular to the cutting edge.
• The workpiece is a solid of homogeneous, polycrystalline, isotropic, and
incompressible material.
• The workpiece temperature is at 20 °C at the start of cut.
• Boundary conditions for temperature are adiabatic.
• The lathe is perfectly rigid and its vibrations have no effect on machining.
In the machining process, work materials undergo high plastic strains (1–5) at very
high strain rates (103 –106 s−1 ), and high temperatures (20–1500 °C). All these effects
of strain, strain rate, and temperature on flow stress are included in the material con-
stitutive laws. These models are obtained from test results by applying the regres-
sion technique. Johnson–Cook material model is popularly used for the analysis
of machining processes. It is found to satisfactorily represent the effect of strain,
strain rate, and temperature on flow stress for many materials. The values of model
Finite Element Simulation for Turning of Haynes 25 Super Alloy 697
constants are obtained from data of Split Hopkinson Pressure Bar (SHPB) test or
Ballistic Impact Tests. This model is suitable for modeling of turning process for any
material as the process involves high strain, strain rate, strain hardening, and non-
linear material properties. The equation for Johnson–Cook model is given as Eq. 1.
The constants of Johnson–Cook model for Haynes 25 are shown in Table 1.
ε̄ T − Tr m
σ̄ = A + Bεn ∗ 1 + c ln ∗ 1− (1)
ε0 Tm − Tr
where σ̄ is Flow stress, ε is Equivalent plastic strain, A is constant for yield stress, ε is
rate of strain, B is coefficient of strain hardening, ε0 is reference value of strain rate,
n is exponent for strain rate dependence, T r is reference temperature, m is coefficient
for temperature dependence, and T m is temperature for melting of work material.
Fig. 1 a Shape of a finite element with six nodes and three quadrature points, b initial shape, c after
re-meshing
is high. This mesh adaptation happens at regular intervals; new corner nodes are
added at the mid-sides of elements undergoing refinement, completely redefining the
element connectivity. Residual distortions are overcome by the Laplacian smoothing
algorithm. For computational ease, a mesh coarsening algorithm is also used to keep
the number of nodes and elements within permissible limits. This algorithm is applied
in the inactive areas of the mesh.
Tool geometry is imported to DEFORM™ 3D. The size ratio for tool insert is
considered as 4. Similarly, the size ratio for workpiece is 7. Tetrahedral element type
for a workpiece with a relative mesh size of 50,000 and tool insert with a relative
mesh size of 40,000, respectively, is considered. Here, workpiece is assumed to be
plastic material and tool insert is assumed to be of a rigid material. For simulation,
workpiece of 32 mm diameter and 20 mm length is assumed. The tool is positioned
with respect to the workpiece so that proper contact between them is generated. The
simulation steps are 500.
Material properties used for simulation are: Poisson’s ratio = 0.296, emissivity
= 0.85, density = 9.13 g/cm3 , melting range = 1330–1410 °C, Young’s modulus =
225 GPa, thermal conductivity = 10.5 W/m °C, and coefficient of thermal expansion
12.8 µm/m °C. Variation of Young’s modulus, thermal conductivity, and coefficient
of thermal expansion of Haynes 25 alloy with temperature was obtained from refer-
ence [4]. Flow stress versus strain at different temperatures is obtained from graphs
reported in Ref. [5].
3 Experimentation
Experiments were performed as per the design of experiments prepared using Cen-
tral Composite Design of Response Surface Methodology. Three input parameters
(cutting speed, feed, and depth of cut) and five levels were used for the design of
experiments. Twenty experimental runs were performed as per the experiment plan.
The experiments were performed on CNC lathe machine; MTAB MAXTURN+,
equipped with a Siemens 828D controller. The workpieces were bars of Haynes 25
alloy having diameter of 32 mm and total length of 100 mm. The cutting length
was 20 mm for each experiment. Two components of machining forces; (a) Cutting
Force (F c ) and (b) Thrust force (F t ) were recorded during experimentation by using
a Kistler dynamometer type 9257B. MQL system of Dropco make was used during
experiments.
4 Results
Simulations are performed for all the runs of experiments using DEFORM™ 3D
software. Figure 3 shows plots of simulation result for cutting and thrust force against
time, for cutting speed = 60 m/min, feed rate = 0.05 mm/rev, and depth of cut =
0.5 mm. Figure 4 shows plots of actual cutting and thrust force with time. Analysis
of force data from experiments and simulation for every machining run has been
done in a similar manner as stated above. Table 2 gives cutting and thrust forces data
700 A. B. Andhare et al.
Fig. 3 Simulation results for cutting force (left) and thrust force (right) versus time (V = 60 m/min,
f = 0.05 mm/rev, d = 0.5 mm)
Fig. 4 Experimental values of cutting force (left) and thrust force (right) (V = 60 m/min, f =
0.05 mm/rev, d = 0.5 mm)
for simulation and experiments and relative error between the actual and simulated
values. It is seen that maximum relative error between simulation and experiment for
cutting force is 15.94% and for thrust force it is 13.75%. Thus, it can be said that the
simulation results are acceptable and can be considered as reliable. The error may be
further reduced by introducing sticking and sliding friction model in the simulation.
5 Conclusion
References
Abstract In this paper, minimization of surface roughness, tool wear, cutting force,
and thrust force is carried out by optimizing machining parameters in turning of
Haynes 25 superalloy. Coated carbide tool was used for machining, and response
surface desirability method is used for optimization. Central composite design of
response surface methodology was used for design of experiments. Prediction models
were developed for each response using response surface methodology. Analysis of
variance was carried out to confirm the correctness of individual models. Later,
optimization was performed. The optimal parameters are validated by performing
additional experiments. It is found that predicted values and experimental results are
closely matching.
1 Introduction
2 Literature Review
Different methods of parameter estimation and optimization are available in the liter-
ature and are used for optimizing machining processes by various authors. Sarıkaya
and Güllü [1, 2] have performed multi-objective optimization of minimum quantity
lubrication parameters for Haynes 25 alloy using Taguchi gray relational analysis.
Deshpande et al. [6, 7] have used various responses to predict surface roughness in
turning of nickel-based superalloy Inconel 718. Sahu and Andhare [8] have used
response surface method for predicting surface roughness in machining of titanium-
based superalloy, Ti6Al4V. They have also performed multi-objective optimization
using desirability approach in machining of the titanium alloy [9, 10]. Neşeli et al.
[11] have used response surface method for optimizing tool geometry in machining
of Haynes alloy. It is found from the literature that RSM is a popular method for
response analysis and optimization as it gives interaction effects and a simple desir-
ability function can be used to perform multi-objective optimization. This paper has
used response surface desirability method for optimization.
3 Experimentation
Cutting speed, feed, and depth of cut are considered as the input parameters for
turning. Surface roughness, tool wear, and cutting forces are the responses recorded.
Five levels of input parameters were selected for experimentation. Experiments were
designed using central composite design method of response surface technique. 20
experiments were carried out according to experimental design as given in Table 1.
Turning was carried out on CNC lathe machine. The workpieces used had diameter
of 32 mm and total length of 100 mm. For each experiment, 20 mm length was used for
turning. PVD coated tungsten carbide inserts with zirconium coating (TNMG160406
UF, Widia-make) were used for machining. MQL system of Dropco-make was used
for providing cutting fluid.
Surface roughness was checked by Mitutoyo-make Surftest SJ-410 rough-
ness tester. Tool flank wear was measured by INSIZE-make USB microscope
ISM PM600SB. Cutting force and thrust force were measured by Kistler-make
dynamometer type 9257B. Table 1 shows the responses measured for the 20
experiments.
Optimization of Machining Parameters for Turning of Haynes … 705
The responses (Ra , V Bmax , F x, and F t ) as recorded in Table 1 are used to develop
models using response surface methodology with the help of Minitab 19 software,
and the developed models are analyzed statistically by ANOVA technique to deter-
mine the significance of various parameters. Due to lack of space, only models are
presented here without ANOVA results. Equations (1)–(4) show the models for sur-
face roughness (Ra ), tool flank wear (V Bmax ), cutting force (F C ), and thrust force
(F t ), respectively. R2 values of these models were 96.84%, 89.03%, 97.82%, and
95.39%, respectively. As all the values are more than or close to 90%, these models
are acceptable.
Main effect plots were also obtained for the responses considered. However, due
to space constraints, these are not shown here. Their findings are discussed below.
The surface roughness decreased with increase of cutting speed and feed rate.
However, the improvement was only up to 45 m/min speed and 0.075 mm/rev feed.
Beyond that value, the surface roughness degraded. With increase of cutting speed,
temperature increases in cutting region, and material becomes soft and easy to cut, so
surface roughness is decreasing. However, excessive increase in cutting speed leads
to rapid tool wear, because of which surface roughness deteriorates at higher cutting
speed. Increased feed rate results in increased feed force that affects negatively on
surface roughness. Depth of cut has linear relationship with the surface roughness;
it increases as depth of cut increases.
Tool wear increased with increase of cutting speed, because cutting speed leads
to an increase in chip volume, friction, and temperature. Tool wear decreased with
increase of feed rate, because of lower contact time at high feed rate between cutting
tool and workpiece. It increased with increase of depth of cut, because the depth of
penetration increased. Minimum tool wear occurred at lower cutting speed, higher
feed rate, and lower depth of cut. Depth of cut was more significant parameter for
the tool wear followed by feed rate and cutting speed.
The cutting and thrust forces decreased with increase of cutting speed because of
thermal softening of material. Both the forces increased with increase of feed rate
and depth of cut. With increase of feed rate, the time to remove the material from
workpiece decreases causing forces to increase. With increase of depth of cut the
volume of material removal is increased, so forces are increased. Minimum cutting
force occurred at higher cutting speed, lower feed rate, and lower depth of cut. Depth
of cut was more significant parameter for the cutting force followed by feed rate.
However, feed rate was more significant parameter for the thrust force followed by
depth of cut.
Optimization of Machining Parameters for Turning of Haynes … 707
Optimization of the machining parameters (cutting speed, feed rate, and depth of cut)
for minimization of surface roughness (Ra ), tool wear (VBmax ), cutting force (F c ),
and thrust force (F t ) is performed. Default values in Minitab 19 software are used
for optimization of the machining parameters using desirability function in RSM.
Minitab 19 software shows the global desirability solution. At the beginning, the
algorithm picks up various points for search of optimal variables. Local and global
searches are performed. Different local solutions are obtained, and one of them is the
best solution. It gives the best relation between input variables for obtaining the best
possible outputs. There are various benefits of using RSM optimization over con-
ventional experimental or optimization methods. In conventional methods, at a time,
only one variable can be optimized. Also, RSM is more economical because fewer
experiments are required to get main and interaction effects. Therefore, desirability
approach of response surface method has been used for optimization in this work.
The response function values are converted into a dimensionless desirability index
(d) that lies between 0 and 1. If d = 0, the response is unacceptable, and if d = 1,
the response is acceptable and desirable. Geometric mean of individual desirability
values (di) is used to combine them. The desirability function is obtained as Eq. (5)
n 1ri
D= diri (5)
i=1
6 Conclusion
In turning of Haynes 25 cobalt-based superalloy using coated carbide tool, the opti-
mum process parameters were determined to minimize surface roughness, tool wear,
and cutting, and thrust forces using desirability function in response surface method-
ology. The optimum values are V = 41.17 m/min, f = 0.0593 mm/rev, and doc =
0.329 mm with minimum values of responses as: Ra = 0.1517 µm, F c = 121.9397 N,
F t = 92.7130 N, and V Bmax = 0.0340 mm. These response values were later con-
firmed by performing validation experiments. The maximum error was limited to
7.63, 8.51, 15.21, and 5.07% in surface roughness, tool wear, cutting, and thrust
force in that order. Thus, response surface desirability approach can be confidently
used for machining parameter optimization.
Table 2 Validation of results (V in m/min, f in mm/rev, and d in mm)
V f d Experiment values RSM predictions % Error
Ra (µm) V Bmax F c (N) F t (N) Ra (µm) V Bmax F c (N) F t (N) Ra (µm) V Bmax F c (N) F t (N)
(mm) (mm) (mm)
41.2 0.06 0.3 0.157 0.047 144.3 97.19 0.148 0.043 122.4 92.26 5.73 8.51 15.21 5.07
55 0.07 0.6 0.459 0.042 274.9 126.2 0.424 0.039 237.7 120.6 7.63 7.14 13.54 4.45
65 0.09 0.9 0.975 0.054 425.2 153.6 0.936 0.05 374.2 146.6 4.09 7.41 11.99 4.56
Optimization of Machining Parameters for Turning of Haynes …
709
710 A. B. Andhare et al.
References
Abstract This paper describes the detailed design, analysis and an overall develop-
ment of the hinge mechanism used for the deployment of radial rib antenna (RRA).
RRA is a deployable mesh-type reflector used for strategic applications under higher
frequency bands. The challenges which are faced in the optimization of the mech-
anism elements are also presented in this paper. Stiffness of the hinge brackets is
very much crucial in determining the overall stowed frequency of the antenna during
launch as the hinges are bonded to CFRP parabolic shaped ribs which are slender
in nature, thus making the hinges compliant. Since all the hinges are connected in
series, the overall frequency will be still lower than the component frequencies. The
paper also provides the suitable configuration changes done for the hinge brackets
and its overall impact on the stiffness of the antenna in the stowed configuration.
Overall configuration of the antenna is also projected in the paper.
1 Introduction
In recent years of spatial antenna technology, mesh deployable antennas are becom-
ing popular because of its inherent ability of folding and unfolding to get an increased
deployed aperture in space [1]. Since the payload fairing size of the launch vehicle is
limited, the accommodation of large shaped antennas becomes a challenge. Typical
mesh-type deployable antennas consist of gold-plated molybdenum wires of suffi-
ciently small diameters which are interlinked with each other with a definite pattern
of knitting. A variety of techniques can be used to make the mesh deploy and attain
its required parabolic shape such as ribs, trusses and hoops. In modern days, space
scientists and engineers concentrate mainly on the deployable truss-type and the
rib-type deployment techniques. The areas of application of these large unfurlable
2 System Description
The RRA hinge mechanism is the deploying member in the entire antenna assembly.
Fig. 1 shows the RRA hinge mechanism. RRA hinge mechanism consists of mainly
the following components:
• Fork end bracket and eye end bracket
• Deep groove ball bearings
• Concentric and eccentric bushes
• Bearing preloader and its associated clamping components.
Bearing Preloader
Concentric bush
Eye end bracket
Eccentric bush
Hinge Pin
Alignment screw
Fork end bracket
The material selection of the hinge components has been done based on strength-
to-mass optimization, availability, coefficient of thermal expansion and machinability
of the materials. The mechanism also consists of concentric bushes and eccentric
bushes to take care of any azimuthal misalignment present in the assembly and has
also the provision for countering elevation misalignment through its internal design.
This will be explained further in the subsequent sections.
RRA consists of parabolic shaped ribs made out of CFRP bonded to the eye end
bracket of the hinge mechanism. Ti6Al4V alloy is chosen due to its higher strength-
to-mass ratio and lower coefficient of thermal expansion (CTE). Bearings are made
out of SS 440C (hardened and tempered) and have been associated with Ti alloy
components because under extreme temperature conditions, since the CTE of bearing
alloy is higher than Ti alloy components, there will be still clearances present to make
the bearings roll without any interference. Finally, the machinability of the materials
is also taken care by annealing for stress relieving during the fabrication of the
components.
3.2 Design
RRA consist of CFRP ribs stowed against a plate called as the top plate and is held
in the stowed configuration by hold down and release mechanism (HDRM). Once it
is deployed, the ribs attain final configuration based on the semi-aperture size. The
hinge point has been found out by equalization of the distances from hinge point
to stowed point and from the hinge point to the deployed point. Let (xs , ys ) and (xd ,
yd ) be the stowed and deployed point coordinates, respectively. Let (x and y) be the
location of the hinge point.
(x − xs ) + (y − ys
2
)2 = (x − xd )2 + (y − yd )2 (1)
It can be seen from Eq. (1) that it is a single equation with two variables, namely
x and y. Thus, based on the physical constraints and using Eq. (1), the hinge point
has been evaluated. The RRA mechanism is a single degree of freedom mechanism
having a single drive cable which has to be routed through all the hinges and has to
be terminated in a particular bracket [2]. Thus, the design of the hinge has been done
in such a way that there are pulleys and retainer pins through which the drive cable
has been routed. Figure 2 shows the configuration of the pulleys and retainer pins.
714 R. Ghatak et al.
Hinge pulley
Location of drive cable
Retainer pin
Latch Spring
CFRP Rib
RRA Hinge
From Fig. (1), it can be seen that in the stowed configuration, the drive cable line of
action does not pass through the line of center of the hinge (hinge axis). As the drive
actuator starts, tension is induced in the drive cable. From the above configuration,
there should be an angle of 7.5° between any two consecutive ribs. Thus, the radially
inward component of the tension force acts on every hinge which creates the moment
for the deployment of the hinge. The design is made such that the force does not pass
through the hinge axis in stowed configuration. Thus, the deployment starts with the
magnitude of the moment as:
Mt = 2 T sin θ L (2)
where
Mt = Moment about hinge axis, Nm
T = Drive cable tension, N
θ = Half of the angle between consecutive hinges, degrees
L = Distance between the hinge axis and the line of action of the force.
As the deployment progresses, L increases reducing the value of T to keep the
moment constant. Finally near to the end of the deployment, the hinge latch [3] gets
arrested in latch spring as shown in Fig. 3.
A Compact Hinge Mechanism for Radial Rib Antenna 715
RRA hinges taken the out of plane launch loads generated by the launch vehicle for
the full antenna. Thus, deep groove ball bearings have been used to take up the loads.
The moment resistance capacity of the bearings has been enhanced by separating
the bearings with suitable spacers made out of SS to make it compatible with the SS
material of the bearing races. Bearing preloader has been designed to close the axial
gap and to apply a particular preload to the bearing outer race. Proper tolerances have
been incorporated in the drawings of all the hinge components keeping in view of
the alignment requirements for the RRA. The main challenge is that RRA operates
on X-band frequency spectrum having very stringent RMS requirements. In order
to achieve that particular RMS and to amalgamate the deploying member with the
central solid reflector in fully deployed configuration, very stringent tolerances have
been given in the fabrication drawings of the hinge brackets. The stiffness of the hinge
bracket is very much important from the point of view of stowed natural frequency
and deployed natural frequency of the antenna. Since there are 45 numbers of hinge
assemblies and all are connected in series, it is important to design the fork end
bracket of the hinge which is connected to the baseplate of the antenna with higher
stiffness.
716 R. Ghatak et al.
As described earlier, the fork end bracket is directly connected to the RRA baseplate
with clamping pins. Therefore, its stiffness is important in meeting the desired natural
frequencies of the antenna. Figure 4 shows three different configurations of fork end
bracket. The FE model of the fork end bracket is shown in Fig. 5. All six DOFs have
been constrained at the hinge clamping locations as shown in Fig. 5. Tet-10 element
has been used during FE model generation. 11,473 elements have been used in FE
analysis. Solid (3D) meshing has been performed with tetrahedral elements with a
global edge length of 5 mm. 20,392 nodes have been generated after meshing the
component. TetMesh has been used as the mesher. Maximum aspect ratio Lh has
been taken as 0.1 with minimum edge length of 0.2 times the global edge length.
1N
Land Area
Mid MPC
A Compact Hinge Mechanism for Radial Rib Antenna 717
Table 1 Mechanical
Mechanical properties Magnitude
properties of Ti6Al4V
considered during FE analysis Modulus of elasticity (E), GPa 108.0
Poisson’s ratio (µ) 0.279
Density (kg/m3 ) 4400.0
Table 2 Comparison of
Sl. No. Configuration Stiffness (N/m) Natural
stiffness and natural
frequency (Hz)
frequencies for different
bracket configurations 1. (a) 1.248 × 106 4161.684
2. (b) 1.219 × 106 4241.521
3. (c) 1.295 × 106 3837.942
Linear static analysis of the brackets shown in Fig. 4 has been carried out in
MSC PATRAN® . Table 1 describes the mechanical properties which have been used
during FE analysis. The results are presented in Table 2. Ti6Al4V has been used as
the material for fork end bracket.
To check the integrity of the FE model, free-free normal mode analysis showed
six rigid body modes. Furthermore, a force of 1 N has been applied on the mid-MPC
of the bracket (a) as shown in Figs. 4 and 5. The corresponding support reactions are
0.519 N and 0.481 N, and the algebraic sum of these forces gives the input force. It can
be observed from Table 2 that configuration (b) has the highest frequency and is the
most desired configuration. But from the practical point of view, this configuration
does not contain any land area (Fig. 4) for the mounting of thermal control elements
like heaters and sensors for maintaining the temperature of the bearings near to
ambient temperature on orbit. Thus, configuration (a) has been selected for the fork
end bracket as this has frequency closer to configuration (b) and has the land area
for mounting the thermal control elements.
5 Conclusion
In this paper, the design and configuration of compact hinge mechanism were per-
formed. It can be concluded that the RRA hinge mechanism is one of the most
precise mechanisms on RRA because it has to perform tasks like the alignment of
the deployed reflector, amalgamation of the solid and deployable parts of the antenna,
sustain launch loads during launch and finally give sufficient stiffness to achieve the
required stowed natural frequency of the antenna. It can also be concluded that the
hinge mechanism plays a vital role in maintaining the RRA a single DOF system.
718 R. Ghatak et al.
References
1. Roederer AG, Rahmat-samii Y (1989) Unfurlable satellite antennas: a review. Springer Link
44(9–10):475–488
2. Srivatsava AK, Murali NS, Keshavamurthy KA (2016) Design and Realisation of a novel 3 m
hybrid rib type Unfurling reflectors. In: 10th national symposium and exhibition on aerospace
and related mechanisms (ARMS 2016). Thiruvananthapuram, Kerala, India
3. NASA (1983) Large space antenna systems technology 1982. NASA Langley, NASA SP,
7046(07)
Suntracker on Rocker-Bogie Mechanism
Abstract This paper introduces an all-terrain robot, i.e., robot with the capability
to traverse all kinds of terrain. For this purpose, the rocker-bogie mechanism was
chosen for the robot design and differential gear mechanism is used to stabilize the
platform. Another focus of ours in this paper is to increase the efficiency of the solar
energy absorbed by the solar panels on the robot to make the robot self-powering. For
this, a dual-axis sun-tracking mechanism is used. All parts of the robot were designed
using CAD modeling software. The robot can be controlled via a Bluetooth joystick
app of an Android phone making it extremely easy to change its speed and direction.
1 Introduction
An all-terrain robot is different from other mobile robots since it is specially de-
signed to traverse through all kinds of uneven surfaces like slopes, mounds and even
stairs. To complete the above tasks, the robot has to be robust enough to sustain all
kinds of jerks. To fulfill these kinds of requirements, the key to building such an
optimized robot is its design. A very famous example of such type of robots is the
Mars Exploration Rover Mission robots Spirit and Opportunity, which were devel-
oped by NASA to conduct research-based activities on the surface of Mars. Thus, the
application of the rocker-bogie mechanism has several advantages over other robot
mechanisms. One of the most important advantages being its simplicity of design and
robustness to traverse uneven terrain. Any suspension system limits the tilt stability
Murarka, Wadichar, Tanawde, Rehpade, Agrawal: These authors claim equal contribution.
due to the height of the center of gravity, but the rocker-bogie mechanism can climb
obstacles that are up to twice the size of the wheel diameter. As given in reference
[1], to achieve dynamic stability, the weight of the robot is equally distributed to all
six wheels. Another main feature of the project is the solar panels that are always
pointed toward the sun. Since in most places conventional sources of energy are lim-
ited, the sun provides a reliable non-conventional solution to this problem. Thus, a
suntracker becomes a very important asset as it can trap more energy than stationary
solar panels as per research [2, 3]. The suntracker uses the application of a dual-axis
rotation mechanism with the help of two servomotors for the movement of the so-
lar panels. The solar panels are connected in parallel with each other to prioritize a
greater amount of current than voltage. The information of light intensity is sensed
by light-dependent resistors (LDRs) and is processed to move the solar panels (Figs.
1 and 2).
2 Development
In Fig 4.
AD = AH 2 + D H 2 = 150 (2)
AB = BC = C D = x
Fig. 3 Dimension
calculations- 1
Fig. 4 Dimension
calculations- 2
722 S. Murarka et al.
In Fig 3.
BC
I n D EC, ∠EC D = 120◦ , C E = = 37.5
2
C E 2 + C D2 − D E 2
cos(120◦ ) = ⇒ D E = 99 (5)
2 × CE × CD
In Fig. 3 ∠C DG = 120◦
EC ED
= ⇒ ∠E DC = 19.14◦ (6)
sin(∠E DC) sin(∠EC D)
In Fig. 4
143
∠AD H = θ = sin−1 = 72.43◦ (8)
150
∠E DG = α − θ = 66.7◦ (9)
DG 2 + D E 2 − E G 2
cos(∠E DG) = ⇒ E G = 234.2 (10)
2 × DG × D E
Relation of diameter of wheel (D), rpm of motor (n) and speed (V)
n× D×π
V = = 141.37 mm/s (12)
60
Suntracker on Rocker-Bogie Mechanism 723
A major problem spotted during the simulation phase was that the platform was
extremely unstable since there was no support as seen in [4]. In order to stabilize
the platform efficiently, we have used the three-gear differential mechanism. In this
mechanism, three identical bevel gears are arranged such that gear A and gear C are
perpendicular to gear B as shown in the figure. Gear A and gear C are attached to two
rockers. When one rocker tends to rotate, the gear at that end transfers the rotation to
the gear B and eventually to the third gear such that gear A and gear C tend to rotate
in the opposite direction, thus stabilizing the platform about the axles. We can see
similar concepts in [1, 5, 6] (Fig. 6).
Fig. 8 Suntracker
The robot is controlled by the joystick Bluetooth commander app [7]. The image of
the joystick app is shown in Fig. 7. The joystick is free to move anywhere inside
the circle, thus giving the user more freedom while robot controlling. When the app
gets connected to the Bluetooth module HC−05 present in the robot, it is ready to
send signals. The app sends x and y coordinates of a joystick as mentioned by Kaz
in [7]. When y > 0, the speed of the left wheel is given by y + x and the speed of
the right wheel is given by y − x. However, if y < 0, it’s vice versa. These values are
then mapped according to the range of micro-controllers, so at extreme positions,
maximum power is obtained. Thus, robot can move in any direction and at any speed
(Fig. 8).
2.5 Suntracker
As the robot is continuously moving through various terrains and time, fixed solar
panels cannot absorb enough energy all the time. In the sun-tracking mechanism,
light-dependent resistors (LDRs) are used to detect the direction of the maximum
intensity of solar rays. Two servomotors are placed perpendicular to each other such
Suntracker on Rocker-Bogie Mechanism 725
that panels can move in a specific region. A special arrangement is made such that
if solar panels are not perpendicular to rays, two LDRs will be exposed to more
light than the other two, and then, the servomotors will rotate in such a way that the
intensity of sunlight falling on each LDR is same.
3 Result
After the completion of our robot, it was tested to determine the extent of its capa-
bilities on various terrains and lighting conditions and the following features were
determined:
1. Better stability of the platform.
2. Ability to climb a slope of gradient up to 45◦ in rough terrain.
3. Ability to traverse extremely uneven surfaces.
4. Ability to climb the steps of the max height of 15 cm.
5. Accurate control of speed and angle of rotation using Bluetooth joystick.
This graph shows a comparison between power output by stationary solar panels
against that of the dual-axis suntracker. The angle is the angle between the sun
and a horizontal line parallel to ground in a plane containing the sun and robot.
Maximum power is obtained when the sun is near about 90◦ angle since solar rays
are perpendicular to panels in both situations. While in other situations dual-axis
suntracker tries to keep panels perpendicular to solar rays. Hence, more power is
generated with dual-axis suntracker as compared to stationary panels (Fig. 9).
Fig. 9 Efficiency of
stationary solar panel versus
dual-axis solar panel
726 S. Murarka et al.
⇒ = 38.96%.
In the future, our objective on the further development of the robot deals in making
the robot robust since the hardware of the robot is not strong enough to handle high-
force impacts. By mounting a camera on the robot to receive live visuals of the
environment and reducing its size, it can be further used as a reconnaissance robot.
In certain fields to survey where human life is in danger, the robot can be used as a
surveillance robot. Overall since it is a very capable all-terrain robot, it can work as
a base for various purposes.
Acknowledgements The authors would like to thank IvLabs—Robotics club of VNIT to give us
an opportunity to work on Suntracker on Mobile Robot. This work is motivated from Deep Learning
Based Stair Detection and Statistical Image Filtering for Autonomous Stair Climbing [8].
References
1. Barlas F (2004) Design of a mars rover suspension mechanism. Izmir Institute of Technology
Izmir, Turkey
2. All Earth Solar Power Blog. Comparing the ROI of a solar tracker versus fixed mount solar. Ac-
cessed from https://www.allearthrenewables.com/blog/dealers/pv-tracker-vs-pv-fixed-mount-
system
3. Dhanabal R, Bharathi V, Ranjitha R, Ponni A (2013) Comparison of efficiencies of solar tracker
systems with static panel single-axis tracking system and dual-axis tracking system with a fixed
amount. IJSEIT 2(2). ISSN:2319–5967 (VIT University, India)
4. Bhaskaran B, Dhayanidhi N, Dhamotharan S (2018) Design and fabrication of Rocker-Bogie
mechanism geosurvey rover. IJ SDR 3(8)
5. Harrington B, Voorhees C (2004) The challenges of designing the Rocker-Bogie suspension for
the mars exploration rover. ESTIMATES, AMS papers, NASA
6. Lindemann R, Voorhees C (2005) Mars exploration rover mobility assembly design, test and per-
formance. Jet Propulsion Laboratory, National Aeronautics and Space Administration, Pasadena,
CA
7. Kas (2013) Arduino bluetooth joystick. Accessed from the Arduino forum https://forum.arduino.
cc/index.php?topic=173246.0
8. Patil U, Gujrathi A, Kulkarni A, Jain A, Malke L, Tekade R, Paigwar K, Chaturvedi P (2019)
Deep learning based stair detection and statistical image filtering for autonomous stair climbing.
In: IEEE international conference, robotic computing (IRC). ISBN: 978-1-5386-9245-5
Market Basket Analysis: Case Study
of a Supermarket
Abstract The relationships among various items in a group can be deciphered using
a data mining technique such as market basket analysis (MBA). It plays a significant
role in the analytical systems in supermarkets to determine the arrangement of goods,
design of sales promotion and discounts for different customer segments to improve
customer satisfaction and thereby increase the sales. This case study involves the use
of data gathered from a supermarket as a database. Measures such as support, con-
fidence and lift are used to measure the association between each product. Based on
these values, association rules are generated. This information can give supermarket
managers an edge over their consumer counterpart to strategically promote products
and improve sales. These results also provide valuable insights for cross-selling,
up-selling and new product integration tasks.
1 Introduction
Reduction in the price of a product not only increases its own demand but also
increases demand for its complementary product. In simple words, if two products
are complementary to each other such as coffee and sugar, their demands tend to be
positively associated. Many such product pairs purchased together by consumers are
commonly known. However, considering the fact that a typical supermarket contains
hundreds of items flung by thousands of customers buying a large number of products,
it becomes difficult to realize the associated product pairs that are not obvious. One of
the best-known examples in the data mining industry is that beers and diapers tend to
be purchased together in American supermarkets [1]. Market basket analysis (MBA)
aims to discover such less obvious types of product associations with minimal human
interaction.
The present study deals with an application of MBA to understand the purchasing
pattern at a supermarket. The paper consists of five sections. Section 2 briefly reviews
basic concepts in MBA. The case study dealing with application of MBA for a
supermarket is discussed in Sect. 3. Section 4 presents the discussion on the result,
while Sect. 5 provides the conclusion.
2 Literature Review
2.1 Support
2.2 Confidence
Confidence measures how much the consequent item is dependent on the antecedent
item, i.e., conditional probability of the consequent item given the antecedent item.
2.3 Lift
The lift tells us how much better a rule is at predicting the result than just assuming
the result in the first place. Greater lift values indicate stronger associations.
3 Case Study
A case study was carried out to find out associations between various items for
the transactions taken place in a popular supermarket in Mumbai for the month of
February 2019. The transaction data was collected and was set up in a spreadsheet
by classifying all the items in 51 classes. There exist multiple brands for each of the
products. The data was collected bill-wise; i.e., each bill indicates one transaction.
The data was stored in a binary form such that the items brought in a transaction
are assigned ‘1’ otherwise ‘0’ irrespective of the type, brand and the quantity of
the product. For the period of study, data about 576 transactions was collected. The
minimum number of items purchased in a transaction was 2, while the maximum
number of items was found to be 32. The transaction history was segregated into 12
time zones, each of 1-hour duration from 10 am to 10 pm. The maximum number of
transactions and the maximum number of items purchased were found to be during
evening time zones, especially from 6 pm to 8 pm.
3.1 Methodology
In order to establish association rules, the measures, viz. support, confidence and lift,
were calculated using the transaction history collected. Considering the vast number
of products, demographics, season and past experiences, the minimum measures for
support, confidence and lift were set up as 15%, 55% and 240%, respectively. The
number of transactions in which both the items, A and B (i.e., A ∩ B), are purchased
together can be obtained by multiplying the two-itemset columns and getting their
sum. SUMPRODUCT function of Microsoft® Excel® may be used for this purpose.
Similarly, for all other associations between itemsets, support was computed. The
following 22 associations satisfied the condition of minimum support of 15%—
Biscuits-Noodles, Biscuits-Other Snacks, Biscuits-Milk, Biscuits-Sugar, Coffee-
Eggs, Coffee-Milk, Coffee-Sugar, Coffee-Tea, Eggs-Milk, Eggs-Other Snacks, Eggs-
Sugar, Milk-Other Snacks, Milk-Others, Noodles-Milk, Noodles-Other Snacks, Tea-
Coffee, Tea-Sugar, Tea-Other Snacks, Sugar-Milk, Sugar-Tea, Sugar-Other Snacks
and Sugar-Coffee.
Market Basket Analysis: Case Study of a Supermarket 731
Similarly, for all other associations between itemsets, confidence was com-
puted. The 42 associations that satisfy the constraint of minimum confidence,
i.e., 55%, are—Cake-Biscuits, Chips-Biscuits, Groundnuts-Biscuits, Hygiene and
Grooming Products-Biscuits, Jams and Spreads-Biscuits, Ketchup and Sauce-
Biscuits, Noodles-Biscuits, Crockery-Bread, Pooja Needs-Coffee, Tea-Coffee, Milk-
Other Snacks, Milk-Others, Noodles-Milk, Noodles-Other Snacks, Tea-Coffee,
Tea-Sugar, Tea-Other Snacks, Sugar-Milk, Sugar-Tea, Sugar-Other Snacks, Sugar-
Coffee, Milk-Eggs, Tea-Milk, Sugar-Milk, Toor Dal-Rice, Coffee-Milk, Eggs-
Milk, Health Drinks-Milk, Chawli-Other Snacks, Bathing Soap-Others, Cosmetics-
Others, Coffee-Others, Detergent-Other Snacks, Poha-Rice, Moong-Rice, Matki-
Salt, Ice Cream-Soft Drinks, Eggs-Milk, Eggs-Sugar, Eggs-Coffee, Eggs-Milk and
Detergent-Washing Soap.
Similarly, for all other associations between itemsets, lift was computed. The 5
associations that satisfy the constraint of minimum lift, i.e., 240%, are—Chawli-
Chocolates, Deodorants-Crockery, Housing and Cleaning Products-Deodorants,
Moong-Rajma and Moong-Toor Dal.
The three measures, viz. support, confidence and lift, play an important role in decid-
ing various promotional strategies in a supermarket. The object pairs with highest
and lowest values of these three measures are shown in Table 1. (The percentage
values of support, confidence and lift are shown in bracket.)
732
Table 1 Object pairs showing highest and lowest values of the three measures
S. no Highest values Lowest values
Support Confidence Lift Support Confidence Lift
1 Milk-Eggs Eggs-Milk (72%) Moong-Toor Dal (264%) Dry fruits—Dates Eggs-Toor Dal (9%) Eggs-Toor Dal
(24%) (2%) (47%)
2 Sugar-Milk Cake-Biscuits (71%) Housing and Cleaning Stationery Noodles-Oats (9%) Eggs-Vegetable and
(23%) Products-Deodorants Products—Matki (2%) Fruits (53%)
(252%)
3 Tea-Sugar Milk-Eggs (62%) Chawli-Chocolates Stationery Eggs-Stationery Oats-Noodles
(19%) (243%) Products—Groundnuts Products (10%) (58%)
(3%)
4 Sugar-Eggs Rajma-Others (61%) Moong-Rajma (242%) Dry fruits—Poha (3%) Soft Matki-Eggs (62%)
(19%) Drinks-Stationery
Products (11%)
5 Tea-Milk (18%) Hygiene and Deodorants-Crockery Ketchup and Sugar-Stationery Poha-Eggs (62%)
Grooming (242%) Sauces-Poha (3%) Products (11%)
Products-Biscuits
(61%)
A. R. Pillai and D. A. Jolhe
Market Basket Analysis: Case Study of a Supermarket 733
It can be observed that (Milk, Egg) has shown maximum support and confi-
dence. Similar observations can be found for (Sugar, Milk), (Cake, Biscuits), (Tea,
Sugar), (Sugar, Eggs), (Tea, Milk), etc. This highlights the need of placing these
items together. Other items that need to be placed together are (Moong, Toor Dal)
and (Moong, Rajma) since they have highest lift values. This association rule mining
has also revealed some interesting facts. The item combinations for which support,
confidence and lift have minimum values may be placed in different sections. The
items (Eggs, Toor Dal), (Stationery Products, Matki), (Eggs, Vegetable and Fruits),
(Noodles, Oats), (Stationery Products, Groundnuts), (Eggs, Stationery Products),
(Dry fruits, Poha), (Soft Drinks, Stationery Products), (Matki, Eggs), (Sugar, Sta-
tionery Products), (Poha, Eggs), etc., did not reveal any correlation, and hence they
need not be together.
Items with higher lift are to be placed in the same rack or nearby racks. Moong
and Toor Dal both belonging to the food grain category are easy to arrange and kept
in nearby racks. Items with higher confidence are to be placed in adjacent racks in the
near vicinity (distances typically covered in 10–15 s by walking). Items with higher
support are to be placed nearby enough that the image of the earlier product is in
the minds of the customer (distances typically covered in 15–20 s by walking). Eggs
and Milk are items enjoying high confidence and support and kept nearby enough to
remind the customer that he/she needs to buy it.
This is very much necessary as a supermarket usually contains a plethora of
products and customers often tend to forget their purpose of shopping and end up
buying other unnecessary products. These associations are indicative of the buying
behavior of the consumer. In addition, these associations could be used for planning
inventories, workforce according to the buying pattern. Besides, such information
enables retailers to engage in private branding of products.
5 Conclusion
References
1. Blattberg RC, Kim BD, Neslin SA (2008) Market basket analysis. In: Database marketing.
international series in quantitative marketing, vol 18, Springer, New York
2. Musungwini S, Zhou TG, Gumbo R, Mzikamwi T (2014) The relationship between (4 ps) &
market basket analysis. A Case Study of Grocery Retail Shops in Gweru Zimbabwe. Int J Sci
Technol Res 3(10):258–264
3. Kawale NM, Dahima S (2018) Market basket analysis using apriori algorithm in R language.
Int J Trend Sci Res Develop 2(4):2628–2633
4. Silvers F (2012) Data warehouse designs- achieving ROI with market basket analysis and time
variance. CRC Press, Taylor & Francis Group, Boca Raton
5. Boztug Y, Reutterer T (2008) A combined approach for segment-specific market basket analysis.
Euro J Oper Res 187(1):294–312
6. Griva A, Bardaki C, Pramatari K, Papakiriakopoulos D (2018) Retail business analytics:
customer visit segmentation using market basket data. Expert Syst Appl 100:1–16
7. Kaur M, Kang S (2016) Market basket analysis: identify the changing trends of market data
using association rule mining. Procedia Comput Sci 85:78–85
Experimental Investigation of Effect
of Nanoparticle Concentration
on Thermo-physical Properties
of Nanofluids
1 Introduction
The researchers from across globe identified some parameters such as concentration,
particle size and shape, which have great influence on heat transfer characteristics of
nanofluids. Out of these parameters, concentration has significant effect on thermo-
physical properties of nanofluid. Bhuiyan et al. demonstrated the role of nanoparticle
concentration on surface tension of nanofluids. Also, this work correlates the surface
tension and thermo-physical properties of nanofluids for heat transfer. The result
of this investigation shows that surface tension has direct dependence on nanopar-
ticle concentration [1]. Khaleduzzaman et al. focused on the correlation between
heat transfer and thermo-physical properties of nanofluids through surface tension
[2]. Gaganpreet et al. analyzed the role of aggregation in enhancement of thermal
conductivity. The present work pointed out that anomalous increase in the ther-
mal conductivity of nanofluid is caused due to clustering, which is mainly an out-
come of nanoparticle concentration [3]. Albadr et al. studied the forced convective
heat transfer performance of heat exchanger for different concentrations of Al2 O3
nanofluid. The results of this study conclude that heat transfer coefficient increases
with nanoparticle concentration in base fluid but increase in nanoparticle concen-
tration results in increase of friction factor through viscosity [4]. In light of above
discussion, it is planned to prepare five different nanofluids, namely Al2 O3 , TiO2 ,
MgO, CuO and ZrO2 by two-step method. This study found enhancement in thermal
conductivity and other parameters as a function of concentration of nanoparticles.
Also, pumping power characteristics of all these nanofluids are studied for optimized
samples.
2 Experimentation
The nanofluids prepared using CuO, MgO, TiO2 , ZrO2 and Al2 O3 nanoparticles
were directly procured from Sigma-Aldrich (India) of analytical grade. The two-step
method was employed to prepare nanofluids. In this method, metal oxide nanoparti-
cles with concentration range (0.5–2.5 wt%) were forcefully dispersed in base fluid,
i.e., water by using probe sonication technique. The base fluid water used in this
work has electrical resistivity of the order of 18.2 M cm.
X-ray diffraction was carried out in nanoparticles for two different purposes, first
to check the structural purity and second to determine the particle size. XRD data
of samples was recorded on Rigaku, Miniflex II (Japan) X-ray diffractometer in
the 2θ range 20–70° with step height 0.02°. Additionally, morphology of dispersed
nanoparticles was analyzed using scanning electron microscopy (SEM) technique.
The SEM images of nanoparticles were captured using JEOL JSM-7500F scanning
electron microscope. Thermal conductivity of nanofluids was estimated by using the
relation
k = (q/4π K ) (1)
where
K is a slope between T and time,
k is a thermal conductivity (W/m K),
q is the power generated by heating wire per unit length (W/m).
Experimental Investigation of Effect of Nanoparticle … 737
mA
Home built setup of hot wire method (Fig. 1) was used to measure thermal con-
ductivity of nanofluids. The values of viscosity and thermal conductivity were deter-
mined thrice, but no significant deviation was observed. Furthermore, the stability
of nanofluids was checked using settling velocity and Brownian velocity which was
computed using hydrodynamic radius of dispersed nanoparticle in base fluid. The
hydrodynamic radius data of dispersed nanoparticles was estimated by using dynamic
light scattering technique (NanoZS, Malvern).
Figure 2a–e shows the XRD pattern of Al2 O3 , TiO2 , MgO, CuO and ZrO2 , respec-
tively. The structural purity of all these nanoparticles reflects from their diffraction
peak positions and relative intensities. For more exact confirmation, all XRD pat-
terns are compared with standard data cards that are provided by Joint Committee on
Powder Diffraction Standards (JCPDS). Figure 2a shows the XRD pattern of Al2 O3 ,
which is in concurrence with JCPDS card number: 10-173. This indicates that Al2 O3
is in alpha phase. Figure 2b indicates the XRD pattern of TiO2 , and it is in accor-
dance with JCPDS card number: 76-1940, which confirms TiO2 is in rutile phase.
Figure 2c depicts the XRD pattern of MgO, which exactly matches with JCPDS card
number: 78-0430; this JCPDS card is indexed to the face-centered cubic system of
MgO. Figure 2d represents the XRD pattern of CuO, which is in accordance with
JCPDS card number: 80-1917. This card shows the monoclinic structure of CuO.
Figure 2e shows the XRD pattern of ZrO2 , and it is in agreement with JCPDS card
number: 79-1771, which is indexed to tetragonal phase of ZrO2 .
The average particle size of metal oxide nanoparticles was calculated by using
the Scherrer formula (Eq. 2) [5],
Kλ
D= (2)
β cosθ
738 P. Maheshwary et al.
CuO
Intensity (A.U.)
(c)
MgO
(b)
TiO2
(a)
Al2O3
20 30 40 50 60 70
2 Theta Position (Deg.)
Fig. 3 SEM images of a Al2 O3 , b TiO2 , c MgO, d CuO and e ZrO2 nanoparticles
nanoparticles
1.2
1.1
Al2O3
1.0
TiO2
MgO
0.9 CuO
ZrO2
0.8
XRD analysis and visualized in SEM study. The presented data of UV–Vis is col-
lected for 2 times, by an interval of 5 days for 1 month, which reflects good stability
of nanofluid. Special feature of present work is that stability is achieved without any
surfactant.
Addition of any foreign matter in base fluid results in the increase of dissipation
energy of base fluid. This is obviously resulting in increase of viscosity [6]. Generally,
the viscosity for base fluid containing foreign particles is expressed by Einstein
equation within concentration limit (2.5 wt%) (Eq. 3) [6],
where μnf is viscosity of base fluid containing foreign particles and μbf is viscosity
of base fluid that is water in our case.
Figure 5 shows the variation of viscosity with concentration of nanoparticles in
base fluid. It is observed that viscosity of base fluid linearly increases with concentra-
tion of nanoparticles. The highest value of viscosity observed was for Al2 O3 –H2 O
nanofluid. The minute observation of viscosity plot shows irregular increment in
viscosity of nanofluids. This irregular increment in viscosity is attributed to the
degree of hydrodynamic interaction between dispersed nanoparticles and base fluid
molecules. The aggregation of nanoparticles in nanofluids is also responsible for
irregular increment in viscosity of nanofluids [7].
Figure 6a–e depicts the variation of thermal conductivity with concentration of
nanoparticles for Al2 O3 , TiO2 , MgO, CuO and ZrO2 nanofluid systems, respectively.
From plot, it is observed that thermal conductivity increases linearly with temper-
ature. The increase in thermal conductivity with nanoparticle concentration shows
irregular increment. The highest enhancement in thermal conductivity is observed
for Al2 O3 –H2 O nanofluids.
1.2
1.0
0.8
0.6
0.1 0.2 0.3 0.4 0.5
Concentration (Wt.%)
Experimental Investigation of Effect of Nanoparticle … 741
0.8 0.8
0.4 0.6
310 320 330 340 350 310 320 330 340 350
310 320 330 340 350 310 320 330 340 350
Temperature (K) Temperature (K)
Thermal Conductivity (W/m.K)
Base Fluid
1.5
0.5 Wt.% ZrO2 (e)
0.5 Wt.% ZrO2
0.5 Wt.% ZrO2
0.5 Wt.% ZrO2
1.2 0.5 Wt.% ZrO2
0.9
0.6
Fig. 6 Effect of concentration on thermal conductivity of a Al2 O3 –H2 O, b TiO2 –H2 O, c MgO–
H2 O, d CuO–H2 O and e ZrO2 –H2 O nanofluids
4 Conclusions
Comparative experimentation with five different metal oxide (Al2 O3 , TiO2 , MgO,
CuO and ZrO2 ) nanoparticles–water-based nanofluids in the context of thermo-
physical properties shows good dependence on concentration parameter. Within the
range of validity of existing correlations (0.01–2.5 wt%), thermal conductivity and
viscosity of all nanofluids show linear dependence on concentration of nanoparticles.
742 P. Maheshwary et al.
Similarly, data for viscosity shows significant increase with increase in nanoparti-
cle concentration. The data of thermal conductivity shows that Al2 O3 –H2 O-based
nanofluid shows higher thermal conductivity among all the nanofluids. The highest
thermal conductivity in case of Al2 O3 shows enhancement of 159% over base fluid,
and lowest is in case of CuO.
References
1. Bhuiyan MHU, Saidur R, Amalina MA, Mostafizur RM, Islam A (2015) Effect of nanoparticles
concentration and their sizes on surface tension of nanofluids. Procedia Eng 105:431–437
2. Khaleduzzaman SS, Mahbubul IM, Shahrul IM, Saidur R (2013) Effect of particle concentration,
temperature and surfactant on surface tension of nanofluids. Int Commun Heat Mass Transf
49:110–114
3. Gaganpreet, Gaganpreet S (2012) Effect of aggregation on thermal conductivity and viscosity
of nanofluids. Appl Nanosci 2:325–331
4. Albadr J, Tayal S, Alasadi M (2013) Heat transfer through heat exchanger using Al2 O3 nanofluid
at different concentrations. Case Stud Therm Eng 1:38–44
5. Hernández FJR, Rubio MFA, Navarro JFV, Rosales FJG (2006) Intrinsic viscosity of SiO2 ,
Al2 O3 and TiO2 aqueous suspensions. J Colloid Interface Sci 298:967–972. https://doi.org/10.
1016/j.jcis.2006.01.009
6. Beck MP, Yuan Y, Warrier P, Teja AS (2009) The effect of particle size on the thermal conductiv-
ity of alumina nanofluids. J Nanopart Res 11:1129–1136. https://doi.org/10.1007/s11051-008-
9500-2
7. Maheshwary PB, Handa CC, Nemade KR (2017) A comprehensive study of effect of concen-
tration, particle size and particle shape on thermal conductivity of titania/water based nanofluid.
Elsevier Appl Therm Eng 119:79–88. https://doi.org/10.1016/j.applthermaleng.2017.03.054
A Framework for Robot Programming
via Imitation
1 Introduction
A. Jha (B)
G H Raisoni College of Engineering, Nagpur, India
e-mail: abhishekduttjha@gmail.com
S. S. Chiddarwar
Visvesvaraya National Institute of Technology, Nagpur, India
S. G. Sakharwade
Rungta College Engineering and Technology, Bhilai, India
© Springer Nature Singapore Pte Ltd. 2021 743
V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering,
Lecture Notes in Mechanical Engineering,
https://doi.org/10.1007/978-981-15-3639-7_89
744 A. Jha et al.
the action, behaviour or trajectory demonstrated to it [2]. For industrial robots, the
trajectory level imitation is a simple and straightforward way to teach the requisite
task. The demonstrated trajectory is adopted by the robot either in joint or task space
for execution.
The present work describes a multistep framework for the programming of indus-
trial robots through direct imitation of the human arm motion. In the first step,
an executable trajectory is modelled for the robot through direct imitation of the
demonstrated action. The imitation principle utilized for this purpose is based on the
kinematic model-based transformation strategy as discussed by Jha and Chiddarwar
[3]. Here, multi-sensor data fusion is used for accurate motion tracking. The second
step of the framework is generalization of the trajectories, where multiple trajectories
of the same task are further processed to retrieve an executable one for the robot.
For this purpose, a probabilistic model-based generalization framework [4] is used.
The execution of the framework is shown in terms of spray painting operation for a
six-degrees-of-freedom industrial robot.
The rest of the paper is organized as follows. Section 2 describes the overview
of the developed framework. The experimental procedure is described in Sect. 3
followed by experimental observations in Sect. 4.
The robot programming architecture described in this work consists of two phases,
namely trajectory generation phase and the generalization phase. The trajectory gen-
eration phase utilizes demonstrations of the requisite trajectory by a human demon-
strator for end-effector trajectory modelling. The trajectories thus modelled are fur-
ther processed in the generalization phase to retrieve an executable trajectory for the
robot. The various aspects of the approach are described in the following sections.
Imitation is one of the simplest modes to teach specific task to a robot. In this work,
the trajectory generation phase utilizes imitation of the human demonstrated action
for end-effector trajectory modelling for the robot. For this purpose, the kinematic
model-based imitation framework as discussed in [3] is modified and used for the end-
effector trajectory modelling. This model could represent the motion of the human
arm in the form of hand trajectory. The hand trajectory such generated becomes the
reference trajectory for the robot to imitate.
The trajectory generation framework utilizes multi-sensor fusion technique to
systematically track the demonstrated motion. A Kinect sensor along with inertial
sensor is used to capture the position and orientation data of the human hand during
A Framework for Robot Programming via Imitation 745
Analogous
I.K Model of Pose Mapping into Hand Pose in Arm
End-Effector
Robot Robot Workspace Reference Frame
Configuration
motion tracking. The sensor fusion provides complete specification of hand during
task demonstration. The steps involved in this process are shown in Fig. 1.
At the time of task demonstration, the Kinect-based motion capture system pro-
vides 3D position (x-, y-, z-coordinates) of the hand. The inertial sensor mounted
on the wrist of the demonstrator provides orientation (roll, pitch and yaw angles)
of the hand. Thus, the dimensionality of the data captured at each instant during
arm motion tracking is 6. The hand poses such captured are used for describing the
hand trajectory in the world reference frame. The hand trajectory can be defined as
a sequence of hand poses obtained from the two sensors and can be represented by
where Hi (ti ) represents the six-dimensional hand pose at time t i defined in the arm ref-
erence frame. For mapping purpose, each state of the trajectory Hi (ti ) thus obtained
is represented by a time-dependent homogeneous transformation matrix as
where
n H oH aH pH
TH i (ti ) = (4)
0 0 0 1
746 A. Jha et al.
The hand trajectory represented in Eq. (4) is mapped into the robot workspace for
imitation. The key intent behind the kinematic modelling of the robot is to optimally
reproduce the mapped trajectory within the kinematic limitations of the robot. The
mapping from human arm to the robot workspace is achieved by using a homogeneous
transformation matrix, defined between the arm and the robot reference frames. This
mapping function transforms the hand trajectory as a Cartesian space target trajectory
for the robot. For each component of this target trajectory, the incremental inverse
kinematic model of the robot produces a joint space configuration of the robot.
This is achieved by updating the initial joint space configuration of the robot in an
incremental manner. The end-effector trajectory obtained in such a manner can be
represented as a sequence of states in the robot workspace given by
The main objective of the generalization phase is to process the recorded set of
trajectories and retrieve a best-approximating trajectory for execution by the robot.
Generally, the approaches which are solely dependent on the human motion suffer
from its inconsistencies [2]. The variability present in human arm motion causes
multiple demonstrations of the same task to characteristically differ from each other.
Therefore, the requisite form of skill is extracted through the generalization process.
In this work, a generalization architecture proposed by Calinon et al. [4] is modified to
suit the considered application. This generalization framework consists of Gaussian
mixture model (GMM) to encode the multiple end-effector trajectories of the same
task, and retrieval of the suitable trajectory is done using Gaussian mixture regression
(GMR). The retrieved trajectory is given input to the robot controller for execution.
The encoding of the trajectories through GMM is achieved by grouping the spa-
tial and temporal components of the trajectories. The data set of trajectories can
be modelled by a mixture of K components, defined by a probability distribution
function
K
p(ε j ) = p(k) p(ε j |k) (5)
k=1
where p(k) is the prior and p(ε j |k) is the conditional probability density function.
These parameters can be defined as
p(k) = π K (6)
A Framework for Robot Programming via Imitation 747
1 −0.5 (ε j −μk )T −1
K (ε j −μ K )
p(ε j |k) = N (ε j ; μk , K ) = e (7)
(2π ) D K
Therefore, GMM parameters for Eq. (5) can be described by prior probability
π k , mean vector μk and covariance matrix k, respectively. The trajectories encoded
through GMM are used by the GMR model to retrieve a generalized trajectory.
The regression process estimates the conditional expectation of spatial component
for the given temporal component at each instant. Thus, by evaluating conditional
expectation at each time step, the generalized form of the trajectory can be retrieved.
3 Experimentation
4 Experimental Observations
375 Target
Actual Target
300 Actual
Z (mm)
225 375
Z (mm)
150 300
225
75 150
0 75 -400
0 0
375 -250
75
300 -100
150
350 225 50
225 200 Y (mm)
X (mm) 300 50 X (mm) 150 200
-100 75
375 -250 350
-400 Y (mm) 0
Case 1: Rectangular trajectory Case 2: Spiral trajectory
workspace (target trajectory) and the trajectory obtained from the inverse kinematic
model of the robot (actual trajectory) for the two cases.
It can be seen from Fig. 2, that for both the cases, the actual trajectory is overlap-
ping with the target trajectory. The overlapping nature of the two trajectories reflects
the effectiveness of the mapping strategy and accuracy of the trajectory generated for
robot through imitation. Despite various differences in the kinematic structure and
motion spaces, the proposed framework is capable of generating geometrically sim-
ilar trajectories for the robot by direct imitation of the human arm motion. However,
the inconsistencies associated with the human motion may result in the non-smooth
segments in the trajectory during demonstration. Non-smooth segments of the tra-
jectory segments may lead to jerky motion of the robot at the time of execution.
Such errors can be eliminated in the generalization phase. Figure 3 shows the gener-
alization process of the encoded trajectory of the first case, while Fig. 4 depicts the
generalized trajectories obtained for the two cases.
It can be seen from Figs. 3 and 4 that for all the sampled points, the regression
process results in a smooth distribution of the encoded variables. The main advantage
of the adapted generalization approach is that it provides a smooth approximation of
the encoded data set which does not require further post-processing. Therefore, the
trajectories retrieved from the regression process can directly be used for execution
by the robot.
It can be observed from Fig. 4, that the generated trajectories are free from the
inherent inconsistencies present in the data set. In the present case, the generaliza-
tion process reduces the inconsistencies, and variations present in the multiple end-
effector trajectories of the same task and retrieve a best-approximating trajectory for
execution by the robot.
A Framework for Robot Programming via Imitation 749
375
300
Z (mm)
225
375
Z (mm)
150 300
225 -400
75 150
0 75 -250
0
375 -100
300 -400 375
-250 300 50
225 -100 Y (mm)
225 200
X (mm) 150 50 150
75 200 Y (mm) X (mm) 75 350
0 350
Case 1 Case 2
References
1. Pan Z, Polden J, Larkin N, Van Duin S, Norrish J (2012) Recent progress on programming
methods for industrial robots. Robot Comput Integr Manuf 28(2):87–94
2. Billard A, Calinon S, Dillmann R, Schaal S (2008) Robot programming by demonstration. In:
Springer handbook of robotics. Springer, Berlin, pp 1371–1394
3. Jha A, Chiddarwar S Robot programming by demonstration using teleoperation through
imitation. Ind Robot 44(2):142–154
4. Calinon S, Guenter F, Billard A (2007) On learning, representing, and generalizing a task in a
humanoid robot. IEEE Trans Syst Man Cybern B Cybern 37(2):286–298
Optimizing EDM Parameters
for Machining Cu102 and Finding
Regression Equation of MRR
and Surface Finish
Abstract This experimental study is conducted for finding the optimum parameters
in Electro Discharge Machining of Cu102 workpiece with the ETP copper electrode.
The varying significant parameters are discharge current, pulse on-time, and pulse
off-time and other parameters such as gap voltage, duty cycle, polarity, flushing
pressure, etc. are kept constant for the proposed machining. The obtained output
data has been used for statistical analysis where surface regression equations for
MRR and Surface Finish are acquired with the help of ANOVA. The best surface
finish obtained is 0.5 µm and pulse on-time is the significant parameter for it while
current intensity is a significant parameter for MRR.
1 Introduction
has also been employed for solving complicated inter relationships between process
parameters and multiple performance measures of the EDM process by Lin et al.
[4]. Response surface modeling (RSM) is also utilized for determining the relations
between the various EDM process parameters with the various machining criteria by
Habib [5], Singaram Lakshmana et al. [6], Shashikant et al. [7].
Studies have been done by mixing powders in the dielectric medium for improving
performance measures. Jeswani [8] used graphite powder in the kerosene resulting in
a considerable increase in MRR and TWR. Wong et al. [9] compared various powders
(graphite, Al, Si, Sic, etc.) of different mesh sizes mixed with dielectric and their
effects on spark gap along with the surface finish. The most important characteristics
of powders affecting the EDM performance have been identified as the particle
size, the particle concentration, the particle density, the electrical resistivity, and the
thermal conductivity by Tzeng and Lee [10].
Normally Most of the EDM work has been carried on steel materials, EN series,
Ti–6AL–4V, S45C, SiC, B4C, WC–Co, Al2 O3 + Ti, Inconel 718, and other hard to
machine materials. But unfortunately, little work has been found for machining the
soft material such as pure copper. Rebelo et al. [11] presented the experimental study
on copper-beryllium alloy for analysis of rough and micro finishing. They compared
the results of MRR and surface quality with their past studies of machining steel,
Rebelo et al. [12]. The machinability of EDM on a copper alloy grade HR750 with
a copper tungsten electrode in terms of the material removal rate and electrode wear
ratio was evaluated by Muttamara Apiwat et al. [13].
In this study, the influence of EDM parameters on machining the pure copper
material Cu102 or ETP copper Cu110 is presented. Although a number of data is
available for rough EDM regime, very little is available about optimal parameter
settings for finishing operation. In this experiment the oxygen-free copper grade
Cu102 workpiece material is selected as the component to be manufactured is difficult
to machine or finish by conventional methods due to its shape and size.
The material used for machining is oxygen-free copper Cu102 (99.95 Cu + 0.05
O2 ) having melting point of 1085 °C and thermal conductivity of 388 W/m K. The
machine used for experimentation was Electronica 500 × 300 ZNC Die sinking
EDM with industrial-grade EDM fluid and 10 mm diameter electrode of ETP copper.
Experiments were carried out in two stages of Roughing and Finishing whose varying
parameters are shown in Table 1 while parameters fixed are shown in Table 2. The
pulse on-time and pulse off-time are set according to the technology chart of the EDM
machine in order to make machining stable. It is set so that the duty cycle should be
nearly constant for the whole stage. Full factorial DOE is used for performing the
experiments so as to not miss any significant output.
The EDM machine was set at positive polarity and other non-significant param-
eters were also fixed shown in Table 2. The depth of machining was fixed to 1 and
Optimizing EDM Parameters for Machining Cu102 … 753
2 mm (in some cases) during roughing operation while it was 0.5 mm for the finishing
operation with their respective combinations of current and pulse on-time made from
the Table 1. Machining time for each combination of operation was noted down and
the material removal rate is calculated by using the machining depth, cross-section
area of electrode and machining time. The surface roughness of each operation is
then measured with the help of Mitutoyo SV514 surface roughness tester.
The highest value of MRR is obtained at pulse on-time of 58 µs for discharge current
of 30 A while it is obtained at 28 µs for both 15 and 5 A. MRR value obtained for
finishing operation is highest at pulse on-time of 11 µs for all current intensity, i.e.,
4, 2 and 1 A. The decrease in MRR after a certain pulse on-time is seen unlike steels
or other materials as shown in Fig. 1. This decrease is because of the high thermal
conductivity of copper resulting in quick resolidification of molten material in the
MRR (mm3/min)
8.000 0.600
6.000
0.400
4.000
0.200
2.000
0.000 0.000
0 100 200 300 0 50 100 150
Pulse On-time Ton (ђsec) Pulse On-time Ton (ђsec)
(a) (b)
Fig. 1 MRR versus T on plot for roughing (a) and finishing (b) operations
754 A. Motwani et al.
4 2
2 1
0 0
0 100 200 300 0 50 100 150
Pulse On-time Ton (ђsec) Pulse On-time Ton (ђsec)
(a) (b)
Fig. 2 Surface finish versus T on plot for roughing (a) and finishing (b) operations
deepest part of the crater. These results show close similarity to those achieved in
past experimentation by Rebelo et al. [13].
A similar plot for surface roughness measured by the parameter Ra as a function
of pulse on-time is shown in Fig. 2. The best surface finish obtained during rough
machining is Ra value of 1.35 µm for 5 A whereas during finishing operation the
Ra value obtained was 0.5 µm for all the three discharge currents of 1, 2 and 4 A at
the same pulse on-time of 5 µs. The roughness value increases with the increase in
pulse on-time which is also similar to steels and the results obtained by Rebelo et al.
[13].
These experimental data are used further for statistical analysis in Minitab 18
software where various plots such as surface plot, contour plot, Pareto chart, scatter
diagram can be obtained. Response surface regression is also performed which is
described in the below section where with the help of ANOVA analysis, regression
equations for Surface Finish and MRR are obtained.
See Table 3.
Surface Finish (Ra) = 0.437 + 0.290I (A) + 0.02701Ton (µs) − 0.0453I (A) ∗ I (A)
− 0.000037Ton (µs) ∗ Ton (µs) − 0.00118I (A) ∗ Ton (µs)
Optimizing EDM Parameters for Machining Cu102 … 755
See Table 4.
MRR mm3 /min = −1.79 + 0.171I (A) − 2.55 Ton (µs) + 4.33Toff (µs)
− 0.00004I (A) ∗ I (A) + 0.0141Ton (µs) ∗ Ton (µs)
− 0.0403Toff (µs) ∗ Toff (µs) − 0.0411I (A) ∗ Ton (µs)
+ 0.0701I (A) ∗ Toff (µs)
5 Conclusions
References
1. Lee HT, Yur JP (2000) Characteristic analysis of EDMed surfaces using the Taguchi approach.
Mater Manuf Process 15(6):781–806
2. Jose Marafona CW (2000) A new method of optimising material removal rate using EDM with
copper–tungsten electrodes. Int J Mach Tools Manuf 40:153–164
3. Lina L, Wang KS, Yan BH, Tarng YS (2000) Optimization of the electrical discharge machining
process based on the Taguchi method with fuzzy logics. J Mater Process Technol 102:48–55
4. Lin CL, Lin JL, Ko TC (2002) Optimisation of the EDM process based on the orthogonal array
with fuzzy logic and grey relational analysis method. Int J Adv Manuf Technol 19:271–277
5. Habib S (2009) Study of the parameters in electrical discharge machining through response
surface methodology approach. Appl Math Model 33:4397–4407
6. Singaram Lakshmanan MK (2013) Optimization of EDM parameters using response surface
methodology for EN31 tool steel machining. Int J Eng Sci Innov Technol 2(5):64–71
7. Shashikant AKR, Kuma K (2014) Optimization of machine process parameters on material
removal rate in EDM for EN19 material using RSM. IOSR J Mech Civil Eng 24–28
8. Jeswani ML (1981) Effect of the addition of graphite powder to kerosene used as the dielectric
fluid in electrical discharge machining. Wear 70:133–139
Optimizing EDM Parameters for Machining Cu102 … 757
9. Wong YS, Iqbal Rahuman LCL, Tee WM (1998) Near-mirror-finish phenomenon in EDM
using powder-mixed. J Mater Process Technol 79:30–40
10. Tzeng YF, Lee CY (2001) Effects of powder characteristics on electrodischarge machining
efficiency. Int J Adv Manuf Technol 17:586–592
11. Rebelo JC, Morao Dias A, Mesquita R, Vassalo P, Santos M (2000) An experimental study on
electro-discharge machining and polishing of high strength copper-beryllium alloys. J Mater
Process Technol 103:389–397
12. Rebelo JC, Morao Dias A, Kremer D, Lebrun JL (1998) Influence of EDM pulse energy on
the surface integrity of martensitic steels. J Mater Process Technol 84:90–96
13. Muttamara Apiwat TN (2013) an experimental study on electrical discharge machining of
copper alloy grade HR750. Adv Mater Res 816–817:75–79
Multidisciplinary Solution Avenues
in Mechanical Engineering
Pradnya Gharpure
1 Introduction
The discipline of Mechanical Engineering has been unfolding new pathways in recent
times covering several fields of science and technology. It has been playing a signifi-
cant role as a core catalytic component in overcoming varied sets of challenges within
each field. This has led to recognition of the modern-day ‘Mechanical Engineering’
Multidisciplinary Solution Avenues in Mechanical Engineering 761
2.1 Biomaterials
Labs-on-chip have emerged as cost-effective and easy to use devices that can integrate
many laboratory operations scaling them down to a chip-as tiny as a few millimetres
in dimension. As these systems handle extremely small volumes (a few picoliters)
of laboratory reagents, micro-fluidic principles play a vital role in their design and
functioning. The devices are compact and low cost due to the small amount of material
required for fabrication and the low volumes of chemicals used. They offer quick
response and reduce the risk of chemical hazard, which make them a popular choice
for point of care diagnostics [4]. In near future, they are likely to influence the public
health sector positively, considering the growing need for economically viable and
user-oriented healthcare devices.
2.3 MEMS/NEMS
which on integration of electrical and mechanical phenomena, are able to sense elec-
trical, mechanical, chemical, optical and even biological impulses to convert them
as input for further processing. The field is seeing a shift from the micro-scale down
into the nano and pico domain in recent times. Traditional manufacturing has been
complemented by techniques like micro-machining, lithography and etching which
are the basics of fabrication of these scaled-down devices. They can be better at
precision and control over macro-mechanical systems and also are cost effective [5].
As these devices scale down to micro/nanosizes, their applications range over areas
like nanotechnology, assembly of particles and sensors that monitor macro as well
as micro- and nano-level phenomena.
Environmental conditions play a huge role in our well-being, and enviro-sensors keep
a check on key environmental parameters from time to time, thus helping us adapt
to the surroundings in a better fashion. Common examples of these include pressure
and temperature transducers, humidity and air quality sensors. Enviro-sensors based
on advanced chemical and mechanical principles are likely to detect toxic gases and
unfavourable environmental parameters such as radiation levels. The results from
these shall be available directly on smartphone, when complemented by Internet of
things (IoT).
Soil quality monitors aid farmers to adopt efficient methods of farming. Mechanobiol-
ogy principles that study the interaction of cells with their surrounding environment
are the next generation of devices that shall monitor and enhance the growth and
development of plants and trees, ensuring an appropriate supply of nutrients to them
and shielding them from the unwanted ones. Each small step towards a greener planet
matters the most, and these innovations possess the potential to crucially impact many
environmental aspects.
Figure 2 depicts the association of the principles of Mechanical Engineering with
the various disciplines where they are leveraged.
Green energy has become the need of the hour, and many technologies have made
their mark in the energy sector by harnessing energy in all possible natural forms.
Wind turbines, hydropower, tidal energy and biomass are a few to name amongst
Multidisciplinary Solution Avenues in Mechanical Engineering 763
In spite of the non-conventional energy boom, natural fuels still drive the energy
sector to a large extent, especially in India. This brings in the necessity of fuel
catalysts that shall contribute towards a healthier environment by improving fuel
efficiency during combustion and minimizing incomplete combustions. Nanocata-
lysts are seen to influence this sector, and soon, conventional engines will require
lesser heat for combustion and will produce minimal to no toxic exhaust, owing to
these fuel additives.
764 Pradnya Gharpure
Study and analysis of biological systems is now a field of prime importance with
the increasing risks of biological hazards in all walks of human life. This makes it
critical to mimic biological systems for experimentation without causing any harm
to the actual entity. Biomimetic systems is a sector of research that attempts to
collaborate biological, chemical and mechanical principles with bio-phenomena to
replicate biological systems artificially in laboratories leading to fast and low-risk
study and trials on them. These prototypes can further be implemented on the actual
system for a specific cause or function.
In contrast to the era a few decades back, Mechanical Engineering is now accompa-
nied by automation and a tremendous scale shift. Machines are acquiring intelligence
and are becoming capable of decision making. As the size of mechanical functions
reduces with their varied application, nanomachines are looked upon as potential
devices as they can work in a precise and detailed manner, leading to higher accuracy.
Cost-reduction of the involved equipment due to low material and energy require-
ments are the other benefits that come as add-ons. Nanomachines and nanobots are
likely to influence the diagnostics and healthcare sector by means of sensing, robotic
surgical procedures and drug delivery [7]. Nanorobotics is also radical in reducing
the industry risk factors by performing perilous functions with ability to work and
transduce in places humans find difficult to reach. The likely applications include
testing of gases in mines, cases in the infrastructure sector-sensing or correcting struc-
tural flaws and failures well before occurrence of disaster by initiation of needful
action. In many modern-day applications, micro- and nanoscale devices have been
found to provide enhanced functional and economic output than their macro-scale
counterparts that reigned the industry for decades.
Traditionally, Mechanical Engineering had been associated solely with the design
of man-made systems and machines, which would carry out the laborious industrial
tasks with higher accuracy and lesser efforts, purely based on their design parameters
and arrangement of the mechanisms involved. This had revolutionized the means of
production by the ease of use and efficiency of these machines.
Multidisciplinary Solution Avenues in Mechanical Engineering 765
With advances in the discipline, symbiosis has been achieved between inanimate
machinery and biological systems, as Mechanical Engineering principles with a
flavour of biology have been applied to analyse them (Fig. 3). It can be seen that the
principles of fluid mechanics are being applied to biological fluid flows, making it
simpler to study biological transport phenomena and design devices that use these
concepts for a highly specific application. Solid mechanics has found way into the
analysis of biological structures like human bones/muscles and is aiding teams of
healthcare professionals with mechanical engineers as key members, to devise novel
solutions to biomechanical issues. The energy sciences, thermodynamics being the
major one, are likely to contribute to research in bioenergetics and related phenomena.
Computational fluid dynamics can be adopted with its different models to simulate
biological processes and draw meaningful conclusions.
The second notable synergy is seen between mechanical systems and artificial
intelligence (Fig. 3). Computers, with the help of operator delivered instructions, are
granting decision making powers to machines, and machines are becoming integrated
units with mechanical power and intelligence together. This aims at further reducing
human intervention in many situations and applications where it was a necessity
before the dawn of the automation era. Machine Learning is now the new buzzword,
that uses an algorithm-based approach to enable machines to learn, using a well
defined set of observations, and this is set to transform the way the industry functions,
even post automation.
766 Pradnya Gharpure
4 Sectoral Linkages
Various sectors influencing day to day life are linked with technologies mentioned in
the sections above. This portrays how the discipline of Mechanical Engineering has
changed form over the years and yet has remained an integral part of a wide range
of human activity segments. The growth is widespread and diverse and connects to
almost every task in the industry and each research domain. An overall perspective
towards these shows how far-reaching the essentials of the discipline are, and how
they flow into all sectors of human life.
Table displays these linkages.
The sectors mentioned above are the driving parameters for human life and well-
being. Viewing them in a holistic manner, rather than as independent drivers, reveals
the fact that all of them are interdependent and together produce an environment that
is cohesive for development, leading to improved quality of life. Figure 4 depicts
this graphically.
The linkages indicated in Table can be extended across numerous domains depend-
ing on their relevance and perception and are not limited to the ones specified above.
It can thus be inferred that Mechanical Engineering shall continue to expand and
enhance various sectors of industry, research and everyday life.
5 Conclusion
The impacts of Mechanical Engineering observed centuries ago, rhyme well with
those in the modern era if a holistic approach is adopted towards the newer technolo-
gies emerging from the discipline. Thus, it can aptly be concluded that Mechanical
Engineering, till date, is the driving force for change and revolution and continues
to steer the engineering areas that influence human life, just as it did centuries back.
References
1 Introduction
Emulating the social behaviors of insects to tackle real-life problems has become an
important topic of research. Researchers are curious about this new way of achieving
a form of artificial intelligence, “swarm intelligence,” [2] which is the emergent
collective intelligence of simple agents. It offers an alternative way of designing
systems, in which autonomy, emergence, and distributed functioning replace control,
pre-programming, and centralization [3].
The same principle of imitating swarm behaviors when extended in the field of
robotics can cooperatively achieve a goal. Designing, building, and utilizing several
simpler robots are easier as they are simple, more flexible, more dependable, and
fault-tolerant because failure of one or more robots will not affect goal achievement
(although efficiency may be affected). Swarms on various robotic platforms like
drones [6], ground vehicles [4], etc. have proven their potential of solving some
real-life problems.
However, it is fair to state that very few applications of swarm intelligence have
been developed and again, very few of them are deployed outside of the highly
sophisticated lab environments [1]. We identified two reasons for this relative lack
one being the fact that swarm systems are difficult to program, as the solutions are
not predefined but emergent [8], and tackled by the interactions among individuals.
Another main reason is the lack of robust and easy-to-deploy swarm communica-
tion stack. To overcome these challenges, this paper proposes one possible solution
to develop an integrated library involving an easy-to-deploy communication stack
and essential building-block functions for executing swarm behaviors. The library
is likely to become a one-point solution for swarm robotics as it is versatile when
it comes to robotic platforms. The library is open source and also expects continu-
ous growth from the contributions of the community. After a keen study of swarm
behaviors, we identified recurring basic common functions in most of the behaviors
which are explained further in the paper.
In this article, the second section elaborates the tools and methods we incorporated
during our work and also reasoning behind choosing specific tools. The third section
gives the precise explanation about the structure of library. In the fourth section,
we have demonstrated the use of the library with some examples by targeting some
problems and tackling them with the help of library functions. The final section
concludes the paper along with the limitations and future contribution to the library
and reveals the direction of future work.
2.1 Tools
purpose, standard Wi-Fi might prove to be too rapid. Another major factor that should
be considered is the power consumption and cost. ZigBee-based networks generally
consume 25% of the power consumed by Wi-Fi networks. Also, Wi-Fi-controlled
devices tend to be more expensive than those operating under ZigBee.
Bluetooth is a wireless technology standard created and used for short-range
wireless exchange of data. One of the major constraints of Bluetooth is the range.
Networking range of ZigBee is almost four times that of Bluetooth. Also, creating a
mesh network in Wi-Fi and Bluetooth networks is a tedious job.
And hence ZigBee proves to be the most appropriate standard for creating a
communication network among multiple communication protocols. It is relatively
easy to configure ZigBee. In fact, this protocol was designed as “assemble and forget”,
in the sense that once configured, there is no need to reconfigure it later. Also, there
is no need of explicitly addressing the source address, i.e., the source address is
transmitted with the data packet by default. Different network topologies like tree,
star, and mesh can be implemented on ZigBee.
Here, the star topology is implemented on ATmega328p which could be extended
to any micro-controller in the ATmega family since whole code is written at reg-
ister level. So a few changes in the nomenclature of the registers would enable the
user to port the code to the required platform. But ATmega poses some limitations
like limited number of Interrupts, UART [9] modules and Timers. So for interfacing
encoders, generating PWM signals, and communicating with ZigBee, the processing
and time complexity was a major issue. In this situation, ARM-based Tiva-C series
TM4C123GXL Launchpad proved to be very helpful. It already has separate hard-
ware for encoders, thereby helps the user to save processing time. Another advantage
of these boards is that they function on 3.3 V logic level, thus reducing the power
consumption [10].
The communication stack can be used for different platforms like aerial vehicles,
mobile robots, or humanoids. In this paper, we have implemented this over a swarm
of mobile robots. We have transferred odometric data of the robot, which is used to
carry out some tasks. This part is covered in the Behaviors and Algorithms section
of this document.
2.2 Methods
A single coordinator initiates the whole network by setting the channel and PAN
ID. In the established network, the routers act as a messengers for communication
between other devices that are not in the range to convey information on their own.
Finally, there are end devices. As the name suggests, an end device cannot allow other
devices to join the network. An end device can power itself down intermittently, thus
saving energy by entering a non-responsive sleep mode. Micro-controller on each bot
receives data from its own sensors, which is transmitted over the ZigBee network to
all the devices with the same PAN ID or to the device with the mentioned destination
address of same PAN ID.
772 H. Zade et al.
3 Structure of Library
The purpose of the library is to provide more user-friendly functions for coding on
any platform. To implement swarm communication, we wrote some functions for
basic tasks like sending and receiving data, creating and joining network, etc. It also
includes some of the odometric functions which are required for swarm of wheeled
robots. Our library can be classified as follows.
Most swarm algorithms share some common functions to implement some of the
services. Software-support section consists of most commonly used functions, cre-
ated by using already existing functions as parameters. Further, library can be built
for more of the famous algorithms, as necessity demands.
• These functions are model specific. For example, we created library functions
specifically for incremental encoders in mobile robots and ZigBee as a communi-
cation protocol.
• For all the sensors, whichever are required, their different sensor-specific libraries
can be included so that sensor interfacing is done by using the simplified and basic
functions for that particular sensor.
Using this library, we have developed a ZigBee stack. These libraries include func-
tions for directly sending frames over a network by just giving data as an input
parameter. It also has another parameter which governs nature of communication as
broadcast or peer-to-peer communication. Similarly, our library includes functions to
extract data from the received frame. An interrupt function is included in the library
which will return data, source addresses, etc. in the form of a structure.
4.1 Behaviors
5 Conclusion
Implementation of algorithms for swarm behavior requires tools and libraries with a
proper platform having sufficient support modules. Also, observing several existing
algorithms showed the need for basic functions for their implementation. To over-
come this, the paper presents an open-source integrated library, tools, and information
about proper support modules.
It gives a brief explanation about the advantages of Tiva boards and ZigBee com-
pared to other tools available. The library consists of basic functions for developing
and testing algorithms. Some applications for mobile robots are shown in the Behav-
iors and Algorithms section. In a very similar manner, this library can be utilized for
swarming various robotic platforms. The standalone communication stack can also
be used for IoT applications. The library also expects continuous growth from the
contributions of the community and in the coming future, we see it as a one-point
solution for all swarm-related applications.
References
1. Bayindir L, Sahin E (2007) A review of studies in swarm robotics. Turk J Electr Eng Comput
Sci 15:115–147
2. Bonabeau E, Dorigo M, Theraulaz G (1999) From natural to artificial swarm intelligence.
Oxford University Press Inc, New York, NY
3. Brambilla M, Ferrante E, Birattari M, Dorigo M (2013) Swarm robotics: a review from the
swarm engineering perspective. Swarm Intell 7:1–41. https://doi.org/10.1007/s11721-012-
0075-2
4. Bürkle A, Segor F, Kollmann M (2011) Towards autonomous micro UAV swarms. J Intell
Robot Syst 61(1):339–353. https://doi.org/10.1007/s10846-010-9492-x
OLSAC: Open-Source Library for Swarm Algorithms and Communication 775
Abstract Devising a best possible design is often considered as the main important
stage in the development of a new device for making curved holes effortlessly. This
study deals with the development of a crank–connecting rod attachment for the
circular motion of the electrode into the metal body to generate a curved hole by
electric discharge machining (EDM). This attachment helps to convert the periodical
linear motion of EDM ram to the electrode (curved) circular motion into the metal
object. To make the curved hole in the metal body, the electrode also should be
in a curved shape. In electric discharge machining, both tool and workpiece are in
interaction during sparking; therefore, while producing the curve hole, the curved
electrode also gets melted and is not useful for the next production. The fabrication
of the curved electrode is also a difficult task. Hence, the attachable electrode inserts
(copper) have developed to protect the curved electrode during the EDM process.
So that, the combined setup (electrode element + insert) could move in a circular
motion during EDM machining to generate the circular holes. Hence, the copper
insert only gets melted during ED machining, and it is changeable. The insulating
bearings (nylon bushes) were also developed for a rotating shaft to avoid the short
circuit between the electrode linkages (electrically negative) and supporting base
(electrically positive). By using this device, the EDM experiments were conducted
with four rough setting parameters consists of pulse-on time (T on ), pulse current
(IP), sensitivity of ram (SEN), lifting time of ram (LFT). Finally, the curved hole
was generated successfully in a hard metal block using this device.
1 Introduction
Material selection is a very important phase while developing the attachment for
the EDM process to generate a curved hole. This attachment consists of a number
of mechanical components which would be submerged in the dielectric fluid. How-
ever, the electric power supply must give to the electrode material. In this, mainly
Development of Crank–Connecting Rod Attachment … 779
two linkages are present; the first linkage consists of yoke-connecting rod—crank
wheel—crankshaft—electrode assembly, and other linkage is the support base. The
complete first linkage acts as an electrode (−), and the second linkage is connected
with the positive (+) terminal. These two linkages were isolated with nylon bushes
to avoid the short in the electric circuit. In the first linkage, some components like
connecting rod and crank wheel made with acrylic plastic, to avoid the electric shock
to the human operators at any stage of working.
These two parts could be designed based on the slider-crank mechanism and its
theory and space available in the EDM machine. It is primarily used to change recip-
rocating motion into rotary motion or vice versa. Here, we used some formulae to
quantify the dimensions of crank and connecting rod according to the space con-
straints and required crank rotation which is less than 360°. Here, the maximum
crank rotation to make circular curve hole is taken as 180°, i.e. = 180°. So that
extreme positions of crank-slider mechanism are obtained along the center lines of
the crank and work head. If the eccentricity is zero between connecting rod centerline
and crank center, the slider-crank mechanism is termed an in-line slider-crank and
the stroke is twofold the crank radius. If the eccentricity is not zero, it is typically
called an offset slider-crank mechanism. Here, eccentricity is measured between
crank wheel centerline and work head centerline. The below formulae would give
the link length dimensions for the unit stroke of the ram.
1 1 − λ2 sin Φ
C = ,
2 1 + λ2 + 1 − λ2 cos Φ
1 1 1 1
a32 = = Φ Φ
2 1 + λ2 + 1 − λ2 cos Φ 4 cos2 2
+ λ2 sin2 2
1 λ2 1 λ 2
a22 = a32 λ2 = = Φ Φ
2 1 + λ + 1 − λ2 cos Φ
2 4 cos2 2
+ λ2 sin2 2
in Fig. 1b, c. Remaining all the components are designed based on these two links.
These schematic views are shown below.
By using all link length dimensions, the schematics of electrode setup and the
whole device assembly were modeled as shown in Fig. 2a, b.
Fig. 1 Schematics of all required components a yoke, b connecting rod (acrylic plastic), c crank
wheel (acrylic plastic), d shaft, e nylon locker, f nylon bushes, g base, h electrode element, i clip
for clamping the curved element, j electrode holder with slot
Fig. 3 a Actual crank–connecting device and b Installation of the crank–connecting device with
EDM machine
Dimensions of the components with an accuracy place a prime role while assembling
of each component. The accuracy of the dimensions helps the smooth running of the
device during the operation. The crank–connecting rod dimensions were obtained by
the existed formulae which are briefly described in the materials and methods section.
Based on dimensions obtained from numerical solution, the links were manufactured
with mild steel. Figure 3a, b shows the complete assembly of crank–connecting rod
attachment with appropriate installation in available EDM room.
The curved shape of the electrode would produce a provision to move in a circular
path. But, during EDM, both electrode and workpiece get machined while moving
in a circular path. This situation leads to the tool wear, which causes the curved
electrode that gets shorten continuously during EDM. Therefore, each time, new
curved electrode needs to be generated using wire EDM to produce the next curve
hole. This situation makes the generation of curve hole costlier and time-taking
process.
782 D. Makireddi et al.
Fig. 4 a Curved electrode assembly with copper inserts. b Actual operation of the curved electrode,
power cable, radius (R) measurement, and shaft
The curved hole was successfully achieved with crank–connecting rod mechanism in
electric discharge machining as shown in Fig. 5a, and the cross section of the curved
hole is also illustrated in Fig. 5b.
4 Conclusions
The crank–connecting rod attachment for making curved holes has been developed
and analyzed. The working of this device was concluded as follows.
Development of Crank–Connecting Rod Attachment … 783
Fig. 5 a Curved hole produced by using crank–connecting rod device. b Sectional curved hole
after EDM
This attachment drives the curved electrode into the workpiece in a curved trajec-
tory during electric discharge machining successfully. Also, this device converted the
periodical reciprocating motion of EDM ram into periodical rotary motion of curved
electrode smoothly. The idea of copper inserts could be the main profit during curve
hole production.
References
1. Ishida T, Takeuchi Y (2002) L-shaped curved hole creation by means of electrical discharge
machining and an electrode curved motion generator. Int J Adv Manuf Technol 19:260–265.
https://doi.org/10.1007/s001700200032
2. Fukui M, Kinoshita N (1989) Developing a “mole” electric discharge digging machining. CIRP
Ann Manuf Technol 38:203–206. https://doi.org/10.1016/S0007-8506(07)62685-7
3. Bayramoglu M, Duffill AW (1995) Manufacturing linear and circular contours using CNC
EDM and frame type tools. Int J Mach Tools Manuf 35:1125–1136. https://doi.org/10.1016/
0890-6955(95)90407-D
4. Goyena R (2019) Curved hole machining by self-movable mechanism with EDM function. J
Chem Inf Model 53:1689–1699. https://doi.org/10.1017/CBO9781107415324.004
5. Westerdal M, Rights A, Copyright I (2002) Development of electrode feed mechanism for
electrical discharge curved hole machining using slider crank chains 68:103–119
6. Miyake Y, Ishida T, Takeuchi Y (2005) Electrical discharge machining of constant curvature
curved hole by means of electrode motion control device. J Japan Soc Precis Eng Contrib Pap
71:1388–1392. https://doi.org/10.2493/jspe.71.1388
7. Uchiyama M, Shibazaki T (2004) Development of an electromachining method for machining
curved holes. J Mater Process Technol 149:453–459. https://doi.org/10.1016/j.jmatprotec.2004.
02.027
8. Ishida T, Kogure S, Miyake Y, Takeuchi Y (2004) Creation of long curved hole by means of
electrical discharge machining using an in-pipe movable mechanism. J Mater Process Technol
149:157–164. https://doi.org/10.1016/j.jmatprotec.2003.11.043
Effect of Crack Angle on Stress Shielding
in Bone and Orthopedic Fixing Plate
Implant: Design and Simulation
R. R. Lam
Department of Mechanical Engineering, Vignan’s Foundation for Science Technology &
Research, Vadlamudi 522213, India
V. V. Kondaiah · B. Ratna Sunil (B)
Department of Mechanical Engineering, Bapatla Engineering College, Bapatla 522101, India
e-mail: bratnasunil@gmail.com
Y. Naidubabu
Department of Mechanical Engineering, Rajiv Gandhi University of Knowledge Technologies,
Nuzvid 521202, India
R. Dumpala
Department of Mechanical Engineering, Visvesvaraya National Institute of Technology, Nagpur,
Maharashtra 440010, India
1 Introduction
Laminate composites are a class of composite materials in which laminas are arranged
in a sequence to develop a bulk structure [1, 2]. Recently, composites are becom-
ing a promising choice of researchers in the biomedical field along with metals,
ceramics and polymers for bone implant applications. Particularly, for load-bearing
bone implant applications, composites offer the best solutions [3]. The materials
(matrix and fiber) used to develop composites should be biocompatible. Among
many other metallic systems, titanium and its alloys have become promising can-
didates to develop orthopedic implants [4]. However, titanium and its alloys raise
stress shielding issue due to higher mechanical properties compared with fractured
bone. Schatzker et al. [5] and Perren et al. [6] demonstrated that the compression
stresses developed in the bone effects the bone remodeling during the process of
healing. Similarly, Kristina and Gholamreza [7] using finite element analysis (FEA)
clearly explained how the compression stress and strain energy density result the
stress shielding effect in cortical bone and trabecular bone when implanted with
two titanium alloys of having two different young’s modulus (105 and 40 GPa). On
the other hand, polymethyl methacrylate (PMMA) is a biopolymer widely used for
several medical applications. But PMMA is not suitable for load-bearing implant
applications which is a limitation [8]. Therefore, developing PMMA-Ti composites
to combine the properties of both the materials can offer several advantages over their
counterparts. In our previous work, the decreased stress shielding effect by develop-
ing PMMA-Ti composites was demonstrated by CAD modeling [9]. It is true that the
stresses within the fractured bone resulted due to the applied load influence the bone
cell activities and healing rate at the fractured regions [10, 11]. The crack orientation
with respect to the screw fixing direction also affects the stress distribution within the
bone. Hence, in the present study, PMMA-Ti composite was considered as material
for implants which can reduce the stress shielding effect, and titanium was selected
as material for screws. CAD model was developed to simulate and demonstrate the
effect of the angle of the cracks with respect to the longitudinal axis of the bone and
with the positions of the screws on the stress distribution within the bone and the
implant.
In the present study, the middle region of the human femur bone (cortical bone)
was selected where the bone can be assumed as a cylinder. For implant material,
commercial pure titanium (CP-Ti, grade 1) and polymethyl methacrylate (PMMA)
were selected to design the composite. For fixing screws, CP-Ti was selected.
CP-Ti is well known for its promising properties suitable for biomaterial applica-
tions. PMMA is an FDA, USA approved polymer that is widely accepted and used in
Effect of Crack Angle on Stress Shielding in Bone … 787
the medical applications [3, 11]. While designing the composite, the optimized com-
position was obtained from theoretical calculations such that the composite exhibits
mechanical properties close to the human femoral bone. Lamellar type composite
was developed by considering PMMA as the matrix and titanium as the dispers-
ing fiber for each lamina. Several laminas were arranged in different orientations to
develop a laminate of required thickness (2 mm). While optimizing the composition,
in order to simplify the analysis, a few assumptions were made and the elastic con-
stants was calculated by using MATLAB® (Math Works, USA) software as explained
elsewhere [9]. The objective behind optimizing the composition of the composite is
to reduce the difference in the stresses which produce in the implant and the fractured
bone. From the preliminary calculations, the fractions of titanium were considered
between 0.25 and 0.5 to obtain both modulus of elasticity and shear modulus of the
designed composite close to that of natural bone values. The thickness of individual
lamina was considered as 0.25 mm and the diameter of titanium fiber was considered
as 0.2 mm. A total of 8 laminas were arranged in different orientations (0°, 30°, 45°,
60° and 90°) to get a composite plate of thickness 2 mm. Different volume fractions
of fiber (V f ) were considered (from 0.2 to 0.5) to calculate the mechanical properties
of laminated composites.
Human femur bone and the region that was selected for the analysis were designed
as shown in computer-aided drafted (CAD) model in Fig. 1. Unigraphics NX8.0
software was used to develop CAD models. The bone screws, implant plate and the
fractured bone assembly by using four screws were modeled. It was assumed that the
crack in the bone is inclined with the axis of the bone at three different angles (20°, 40°
and 50°) as shown in Fig. 2a–c. In the CAD model, 120 and 28.6 mm dimensions were
selected for the bone as length and diameter, respectively. Implant was designed with
100 × 16 × 2 mm3 dimensions. Fixing screw has a conical head of 10 mm diameter
and 8 mm shank diameter for a length of 18 mm including 3 mm V-shaped end. A
total of four holes of 8 mm diameter were designed by placing two holes on both
sides of the crack as shown in Fig. 1e keeping a distance of 20 mm between the
holes. Analysis was carried out by adopting finite element analysis (FEA) with the
help of ANSYS Workbench® (ANSYS Inc. USA). The stress distribution in bone and
implant was studied by applying static load conditions. The CAD model was imported
into ANSYS for FE analysis. The meshing was optimized from coarse mesh to fine
Fig. 1 3D modeling shows different parts and assembly: a femur bone, b cross section showing
selected region of femur bone with hole drilled for screw fixation, c screws, d implant plate with
four holes and e assembly of the implant plate, screws with the fractured bone
788 R. R. Lam et al.
mesh. The contacts were defined between the mating surfaces by considering the
relative movement and the requirement. Before providing the boundary conditions,
material properties were given as input data. A compressive load of 50 kg (490 N)
was applied at both ends of the CAD model by considering a human self-weight and
a reaction forces applied on the bottom end due to the self-weight.
V f ) [90 60 90 0]s
PMMA-Ti (0.3 V f ) 15,100 19,020 6430 0.2 0.3 0.4 5877 2470 2310
[90 60 −45 0]s
PMMA-Ti (0.4 V f ) 11,130 21,040 7160 0.1 0.4 0.45 5681 3060 2710
[60 30 90 45]s
789
790 R. R. Lam et al.
Table 2 Maximum, minimum and the difference between maximum and minimum stress
Angle Combination Max von Mises Min von Range (max–min
(V f _Orientation) stress (Pa) Mises stress stress) (Pa)
(Pa)
20 0.25_[90 60 90 0]s 2.620e6 4749.5 2,615,450
0.3_[90 60 −45 0]s 3.085e6 3472.2 3,082,327
0.4_[60 30 90 45]s 3.024e6 4924.3 3,019,275
40 0.25_[90 60 90 0]s 2.930e6 4659.9 2,925,341
0.3_[90 60 −45 0]s 2.516e6 6257.8 2,510,042
0.4_[60 30 90 45]s 2.422e6 5584 2,416,716
50 0.25_[90 60 90 0]s 3.481e6 3548.2 3,477,851
0.3_[90 60 −45 0]s 3.656e6 4658.6 3,651,441
0.4_[60 30 90 45]s 3.948e6 3589.8 3,944,410
in the bone fixing assembly as listed in Table 2. Among all combinations, if the
crack angle is 20°, [90 60 90 0 0 90 60 90] orientation by using laminas of
0.25 V f yielded higher uniformity in the stress distribution. If the crack angle is
40°, [60 30 90 45 45 90 30 60] orientation with 0.4 V f resulted in better stress
distribution. Interestingly, when the crack angle is increased to 50°, again 0.25 V f with
[90 60 90 0 0 90 60 90] orientation resulted in relatively better stress distribution.
These observations explain that the same composite may not be suitable for all kinds
of bone fractures which are having different inclinations with the bone longitudinal
axis. Depending on the crack inclination, using laminated composite with specific
orientation certainly helps to increase the homogeneity in stress distribution and
directly contributes toward better bone remodeling.
In all the cases, the lowest stress was observed in the screws. As typically shown
in Fig. 3d–f in the case of bone with 20° crack inclination with three kinds of implants
(0.25, 0.3 and 0.4 V f ), the interface between the screws and the bone is experienced
higher stresses. This kind of behavior was also noticed for 40° and 50° crack. The
screw–bone interface is the actual region undergoes drilling while fixing the screws
and desired to have higher stresses to promote higher healing rate after implantation.
In order to investigate the effect of the positioning of screws on stress distribution
while fixing the implant, 20° crack inclination was considered and three different
positions 11, 22 and 33 were adopted. Based on the results, the laminated composite
with [90 60 90 0 0 90 60 90] orientation of 0.25 V f was selected. From the overall
observations, it is understood that the use of PMMA-Ti laminated composites helps
to obtain better stress distribution in the bone to accelerate the bone remodeling in
the process of healing. The crack inclination also influences the stress distribution
which can be fairly addressed by adopting appropriate laminated composite with
optimum orientation. This flexibility opens the opportunities to design and develop
case-specific and patient-specific implants to improve the healing rate. This kind of
understanding helps to develop the next level of implants for orthopedic applications
which are case specific and patient specific.
Effect of Crack Angle on Stress Shielding in Bone … 791
Fig. 3 Von Mises plots of the model with 20° crack subjected to loading when three different
implant materials are used: a composite with 0.25 V f with [90 60 90 0]s orientation, b composite
with 0.3 V f with [90 60 −45 0]s orientation, c composite with 0.4 V f with [60 30 90 45]s
orientation, d cross section corresponding to 0.25 V f , e cross section corresponding 0.3 V f and
f cross section corresponding to 0.4 V f
4 Conclusions
In the present work, 3D model of bone fixing including fractured bone, implant
plate and screws were designed by CAD and simulation was done by using ANSYS
Workbench® . PMMA-Ti laminated composites with different lamina orientation
were designed by using MATLAB software and optimum laminated composite was
obtained in view of achieving mechanical properties close to that of natural human
bone. Stress distribution was varied with respect to the crack angle in the bone. As the
crack angle is changed from 20 to 40 and 50, composites of different volume fractions
with different orientations were observed as optimum. The results demonstrate the
role of crack angle and screw positions on stress distribution in the fractured bone.
Understanding such factors helps to develop specific implants which are optimum
for individual cases.
792 R. R. Lam et al.
References
1. Chung DDL (2010) Composite materials science and applications, 2nd edn. Springer, New
York
2. Nikhilesh C, Krishna KC (2010) Metal matrix composites, 2nd edn. Springer, New York
3. Park J, Lakes RS (2007) Biomaterials an introduction. Springer, New York
4. Niinomi M, Nakai M (2011) Titanium-based biomaterials for preventing stress shielding
between implant devices and bone. Int J Biomater 1–10
5. Schatzker J, Horne JG, Sumnersmith G (1975) Reaction of cortical bone to compression by
screw threads. Clin Orthop Relat Res 111:263
6. Perren SM, Huggler A, Russenberger M (1969) The reaction of cortical bone to compression.
Acta Orthop Scand 125:19–29
7. Kristina H, Gholamreza R (2013) A discussion on plating factors that affect stress shielding
using finite element analysis. Comput Biol Med 43:1748–1757
8. Jaeblon T (2010) Polymethylmethacrylate. J Am Acad Orthop Surg 18(5):297–305
9. Naidubabu Y, Mohana Rao G, Rajasekhar K, Ratna Sunil B (2017) Design and simulation of
polymethyl methacrylate-titanium composite bone fixing plates using finite element analysis:
optimizing the composition to minimize the stress shielding effect. Proc IMechE Part C: J
Mech Eng Sci 231(23):4402–4412
10. Chen JH, Chao L, Lidan Y et al (2010) Boning up on Wolff’s Law: mechanical regulation of
the cells that make and maintain bone. J Biomech 43:108–118
11. Wolff J, Paul M, Ronald F (1986) The law of bone remodelling. Springer, New York
IoT-Based Ambiance Monitoring System
Abstract This paper introduces a flexible ambiance monitoring system that records
the real-time values of its surrounding’s environmental factors and then plots their
line graphs. Its architecture include humidity, temperature, air quality and sound
sensor connected to a single ESP8266 NodeMCU micro-controller. These sensors
monitor the corresponding environmental aspects and return these readings to the
NodeMCU. Since the accuracy of these sensors is quite respectable, this system could
be utilized in many sectors like hospitals, military, mining, etc. Due to its simplicity
and economical price, it can be potentially extended by adding more sensors and
hence can be interfaced to meet specific requirements.
1 Introduction
“The most profound technologies are those that disappear. They weave themselves
into the fabric of everyday life until they are indistinguishable from it” was Mark
Weiser’s central statement in his seminal paper in Scientific American in 1991 [1], the
concept of the Internet of things is one such thing. The IoT can be expressed in many
ways, and it encircles many aspects of life that include its implementation in different
sectors like hospitals, military, metropolitan cities, industries, etc. One of the best
and the largest implementation of IoT can be seen in the case of mobile phones which
can be redefined as the eyes and the ears of applications that help in interconnecting
various other applications. As fascinating as it might sound IoT has helped companies
to become more flexible, minimize errors, helped in their developmental process
and incorporate complex and efficient functioning of organizational systems. All
around we see various IoT platforms that have the capability to interconnect millions
of devices simultaneously. Presently, IoT covers a wide range of departments and
instances that vary from a huge connection of multiple platforms and that include the
deployment of previously installed technologies with online cloud database systems
to a small single constraint device in real time. The collaboration of all these devices
is a cumbersome task.
The paper has been divided into five sections. The main device and app are dealt
with in Sects. 2 and 3 discusses the former works related to the topic. Section 4
focuses on the advantages and limitations the system can overcome. The paper ends
with the conclusion and future work in Sect. 5.
The general proximity monitoring system that gives the general data of the closed
premise. This project aimed to create a system to monitor and visualize the ambiance
of its surrounding environment. The collected data would be then transmitted to its
online database such that anyone with its mobile application can access this data,
anywhere in the world in real time.
Arduino IDE by using the pre-existing libraries. The components of this device are
as follows.
NodeMCU is an open-source firmware and development kit that helps you to design
or assemble IoT products. It comprises firmware which runs on the ESP8266 Wi-Fi
SoC developed by Espressif Systems, and hardware which is based on the ESP-12
module. The Lua scripting language is used by the firmware.
MQ135 Gas Sensor is highly sensitive to ammonia, sulfide, benzene, smoke and
other harmful gases. This sensor comprises a micro Al2 O3 ceramic tube, tin dioxide
(SnO2 ) sensitive layer, measuring electrode and heater that is fixed into a crust made
with plastic and stainless steel net. The heater provides necessary work conditions
for the work of sensitive components [3].
DHT11 Temperature and Humidity Sensor includes a calibrated digital signal
output with the capability to measure temperature and humidity. It can be broadly
categorized into three main components, a thermistor with a negative temperature
coefficient, resistor humidity sensor (its conductivity is dependent on moisture con-
tent)and an 8-bit micro-controller which gives a digital output by converting the
received analog signals from both the components [4].
LM386 Sound Sensor all the common sound sensors work by detecting differences
in air pressure and transforming them into electrical signals. Amplifiers and micro-
phones are usually integrated into these sound sensors to increase the strength of the
incoming signal.
This circuit is a digitally generated circuit that can be used to study the pin configu-
ration of the project. The NodeMCU micro-controller is powered using a USB cable.
The sensors are connected to the 5 V power supply. Each of the sensors is connected
to a GPIO pin on the NodeMCU (Fig. 1).
The android application made for this project displays and plots real-time graphs for
air quality, temperature and humidity and reports if the sound intensity exceeds any
preset permissible limit. This has been made possible because the app is connected
with the NodeMCU micro-controller via Google’s real-time Firebase Database.
Hence, the real-time values can be displayed and line graphs can be plotted on
the phone. This app has features like graph readings and details, for the analysis
of temperature, humidity and sound. The app is a multi-screen mobile application
that is made with Android Studio using Java programming and XML for its layout
(Fig. 2).
The environmental factors are first measured by the sensors and sending them
to the ESP-8266 micro-controller. If the micro-controller is connected to a Wi-Fi
connection, then the readings are transmitted to the online database Firebase. This
data can be accessed by any android device with the app and an Internet connection.
3 Related Work
In [5], a similar IoT-based system is implemented for personal use that monitors its
surrounding conditions. While considering the safety of some premise this system
consisting of a gas leakage sensor, temperature sensor for detecting the gas leakage
IoT-Based Ambiance Monitoring System 797
and fire in the house. Also, this system includes a door on/off sensor that can detect
the presence of any unauthorized person entering the home. The sensor sensed data
is displayed on the Web page. The control button is used for controlling devices
through the Internet. Light, fan and small motor are controlled by these buttons. By
hitting the on and off button, all devices can be controlled. If any problem is detected
in the system, then it can report that problem, by message, to the service center [5].
In [7], the system consists of a micro-controller as the main processing unit and
all the sensors and devices are connected to the micro-controller. The sensors can be
operated by the micro-controller to retrieve their data, then it processes that data and
uploads it to the Internet through the Wi-Fi module connected to it [7].
4 Advantages
The system in [5] is more home-based, it has an LPG gas sensor, while our project is
made customizable by replacing the gas sensor with other MQ gas sensors according
to need. The system is made compact which increases its scope in the industrial
sector for getting surrounding aspects of places like ducts, food storage rooms, etc.
The paper introduces a self-built, easy to use app, so the user does not need to go to
the cloud to get data and has easy access to data in real time from anywhere in the
world. In [5–7], the systems either used two micro-controllers or an external Wi-Fi
module, our device operates on NodeMCU which has an internal Wi-Fi chip thus
make the device more compact, cost-efficient and less power consuming.
This system monitors the environmental changes in its surroundings. This infor-
mation is accessible through the cloud to anyone with the mobile application. The
DHT11 sensor monitors the humidity and temperature, the MQ135 sensor records
the air quality and pollution level in its surroundings. The ESP8266 module then
uploads this data to the Firebase cloud. The app displays this data then it analyzes it
to plot a real-time graph. It also notifies the user if the noise in the proximity of the
device exceeds the permissible limit [6].
This system can be modified to develop a control automation system where the
values observed by the sensors can be used to activate some process and control the
devices for heating or cooling, ventilating space, via the mobile application [2].
The mobile application can be modified to display trends over days, weeks or
months. The mobile application can be guided to send a notification to some members
over a rise in some readings or damage of sensor and hence making the system
smarter. This system can be a crucial step in the development and implementation
of IoT-based products and can serve as a building block of several innovations in the
field.
798 H. S. Parihar et al.
References
Abstract Using hand gestures for controlling laptop features is a creative and inno-
vative strategy of communication between users and devices. As compared to existing
methods, the technique of hand gesture has the benefit of being easier to use. By using
this technique, the conventional method of using mouse, keyboards, and controllers
will have to change since by utilizing the proposed solution, one can interact with the
computer with hand gestures. In this technique, ultrasonic sensor is used to classify
the hand movement in real time. Ultrasonic sensors (HC-SR04) measure the distance
of the user’s hand by using sound waves to calculate the time difference between
reflected waves. The principle thought of our methodology is to accelerate the inter-
action with the computer, by using universally used equipment like a laptop, Arduino
UNO board, and low-cost sensors like the HC-SR04. By the use of PyAutoGUI and
PySerial, the device was able to communicate and interpret the input data by the
Arduino. Hence by acting along these lines, any user can easily interact with the
computer using simple, easy to learn hand gestures.
1 Introduction
Since the dark ages’ humans have been using gestures to communicate. This tech-
nique of communication, though evolved from using gestures into written and spoken
languages, even today, the significance of gestures in our everyday communication is
immeasurable. Not only that but sign languages such as the American Sign Language
Rishabh Runwal, Shivraj Dhonde, Jatin Pardhi, Suraj Kumar, Aadesh Varude: contributed equally
to this work.
Related Works
The IEEE paper titled ‘A system for controlling personal computers by hand gestures
using a wireless sensor device’ is proposed by Kaoru et al. [1]. There is a great deal of
home machines and individual PCs around us. In this study, as an interface centering
on the convenience, we build up a framework to control individual PC by Arduino-
based hand gesture control of PC application by applying the regular conduct of
human. Research issues incorporate the meaning of the relationship between the PC
tasks and signals, the acknowledgment of hand signals, the change of the mistake of
the signals, and how to understand the framework. The IEEE paper titled ‘Radar and
ultrasound hybrid system for human computer interaction’ is proposed by Nguyen
and Li [2]. Touchless hand gesture is one of the rising innovations for human PC
communication. In this paper, we explore the achievability of a cross-breed frame-
work utilizing frequency-modulated continuous wave (FMCW) radar and ultrasound
sensors with one transmitter and one collector to recognize hand development for
controlling a PC. This sort of information likewise raises issues that are not significant
with conventional info. On the client’s side, these issues are to learn, to recall, and
to precisely execute signals. Sziladi et al. have researched in the filed of how mouse
cursor control can be done based on hand gestures, this can be read in their paper
titled “The analysis of hand gesture based cursor position control during solve an IT
related task,” in which they detail how they realize this mouse control movement by
the leap motion device [3]. Rajkanna et al. have also worked on controlling a mobile
robot via hand getures through a microcontroller as demonstrated in their 2014 paper
titled “Hand gesture based mobile robot control using PIC microcontroller” [4]. Anju
Latha et al. have similarly worked on ultrasonic sensors using ToF measurements to
build a distance meter device [5].
Hand Gesture Control of Computer Features 801
2 Working
Gesture controlling is based on how one is able to determine hand position from the
ultrasonic sensor. For preparing the crude information, a microcontroller is essential;
for that one uses the Arduino UNO board. By means of USB connection, the micro-
controller moves the processed and thus determined distance value which is given by
the sensor. The information which is sent by the sensor is handled by a software in PC
where the required computations are performed and the data is coordinated with the
predefined conditions (gesture resolution). In this model, two ultrasonic sensors are
used to recognize hand position and are associated with the Arduino board. As we are
aware, ultrasonic sensors constantly emanate sound waves, which in our application
then get reflected from the user’s hand. The microcontroller analyses the reflected
back sound waves and determines the position of the hand.
3 Implementation
There are no ends to the applications of our project, but for the sake of testing and
implementation, we covered the following spheres of use by any user.
Media Player
One of the most basic and widespread uses of computers today is to view videographic
content. Realizing this, it was decided to firstly gain control to the most fundamentally
used functions of media players. Looking at the uses of media player, we deduced
the following actions to be essential and, therefore, brought them under control of
our apparatus and hence the user’s hand gestures.
Gestures:
1st Hand gesture: Allows us to ‘Play/Pause’ media player by placing two hands
in front of the ultrasonic sensor at a particular far distance.
2nd Hand gesture: Allows us to ‘Rewind’ the video by hand motion away set in
front of the left sensor at a particular distance.
3rd Hand gesture: Allows us to ‘Forward’ the video by hand motion toward the
left sensor at a particular far distance (Fig. 1).
4th Hand gesture: Allows us to ‘Increase Volume’ of the video by hand motion
toward the right sensor.
5th Hand gesture: Allows us to ‘Decrease Volume’ of the video by hand motion
away from the right sensor.
Game
After covering the basic operation, we decided to cover the most prevalent use of
computers, in which hundreds of thousands of users spent an immense amount of
time on gaming. In particular online games (Subway Surfers), the script of the python
802 R. Runwal et al.
Fig. 1 Fast-forwarding
Media
matches and enables the background python code to run and activate the keypress
which is not as easily possible in other such games.
Gestures:
1st Hand gesture: Allows us to move right by swiping hand on the right sensor.
2nd Hand gesture: Allows us to move left by swiping hand on the left sensor.
3rd Hand gesture: Allows us to roll on the track by moving both the hands toward
both the sensors.
4th Hand gesture: Allows us to jump on the track by moving hand away from both
the sensors (Fig. 2).
Chrome
The third and the final use which was decided upon was the most complex one with
almost twice the amount of gestures. ‘Controlling a Web browser,’ doing this without
making the gestures unlearnable for the user proved to be a feat. Web browsers
today are able to access and edit incredible amounts of data (Video, Document,
Image, and Communication). The following gestures allow user to carry out the
major fundamental tasks possible through a Web browser (Fig. 3).
Gestures:
1st hand gesture: It allows us to change the tab by moving our hand toward the
left sensor.
2nd hand gesture: It allows us to open the drop-down menu by hand motion away
from left sensor.
3rd hand gesture: It allows us to scroll up by hand motion toward right sensor.
4th hand gesture: It allows us to scroll down by hand motion away from the right
sensor.
5th hand gesture: It allows us to enter by motion of both hands in front of both
the sensors at once.
6th hand gesture: It allows us to forward the page by swiping hand in front of the
right sensor.
7th hand gesture: It allows us to rewind the page by swiping hand in front of the
left sensor.
Schematic Diagram
4 Links
Code: https://github.com/108pbsuraj/laptop-control-using-hand-gesture.git.
Video: https://www.youtube.com/watch?v=KIszs4WeCiQ.
5 Further Prospects
Disclaimer Ethical approval: This article does not contain any studies with human participants or
animals performed by any of the authors.
References
1. Kaoru Y, Lie J, Cheng Z (2014) System for controlling personal computers by hand gestures
using a wireless sensor device. IEEE
2. Nguyen MQ, Li C (2010) Radar and ultrasound hybrid system for human computer interaction.
IEEE
3. Sziladi G, Ujbanyi T, Katona J, Kovari A (2017) The analysis of hand gesture based cursor
position control during solve an IT related task. 2017 8th IEEE International Conference on
Cognitive Infocommunications (CogInfoCom), Debrecen, pp. 000413–000418
4. Rajkanna U, Mathankumar M, Gunasekaran K (2014) Hand gesture based mobile robot control
using PIC microcontroller. 2014 International Conference on Circuits, Power and Computing
Technologies [ICCPCT-2014], Nagercoil, pp. 1687–1691
5. Anju Latha N, Rama Murthy B, Bharat Kumar K (2016) distance sensing with ultrasonic sensor
and Arduino. Department of Instrumentation, Sri Krishnadevaraya University, Anantapur, A.P.,
India
Industry 4.0 Applications in Agriculture:
Cyber-Physical Agricultural Systems
(CPASs)
Abstract The increase in the demand of agricultural food products that includes
processed food, meat, dairy and fish will put the current food production and supply
systems under pressure. The high demand for food resources needs high agricul-
tural productivity. Industry 4.0 technologies such as cyber-physical systems (CPSs),
Internet of Services (IoS), Internet of things (IoT), cloud computing (CC) and big
data promise huge potential in the field of agricultural supply chains through digi-
talization. This article proposes a framework for cyber-physical agricultural systems
(CPASs) which is an intelligent integration of CPS, IoT, CC and big data with agri-
cultural systems. CPAS can be applied in leveraging agricultural supply chains and
boosting productivity.
1 Introduction
There is a tremendous scarcity for food and agricultural resources in the current
times [1, 2]. It is estimated that one in seven people do not have access to enough
protein and energy from their diet and even more suffering from some form of
R. Sharma (B)
Department of Operations and Supply Chain Management, National Institute of Industrial
Engineering (NITIE), Mumbai, India
e-mail: rohit.sharma.2017@nitie.ac.in
S. Parhi
Department of Industrial Engineering and Manufacturing Systems, National Institute of Industrial
Engineering (NITIE), Mumbai, India
e-mail: shreyanshu.parhi.2017@nitie.ac.in
A. Shishodia
Department of Engineering Technology and Project Management, National Institute of Industrial
Engineering (NITIE), Mumbai, India
e-mail: anjali.shishodia.2014@nitie.ac.in
micronutrient malnourishment [3]. The world now faces, altogether, a new set of
intersecting challenges such as the increasing competition for water, land and energy
resources leading to food insecurity [4]. Also, food producers all over the world are
experiencing greater competition for land, water, and energy and the need to curb
many adverse effects of food production on the environment [5, 6].
The previous industrial developments did not have a prolonged effect on agricul-
tural productivity. However, the Industry 4.0 technologies provide an opportunity to
the agricultural sector to increase its productivity substantially with the use of CPS,
IoT, cloud computing and big data. The fourth industrial revolution can be termed as a
digital revolution in manufacturing which has emerged as result of improved network-
ing and computerization in all aspects of manufacturing and production [7]. Industry
4.0 brings a new paradigm to manufacturing where the latest technologies collaborate
for delivering maximum output with minimum resource utilization [8]. The amalga-
mation of digital and physical worlds is brought through CPS which would herald
the way for future developments in industry, thereby allowing the improvement of
efficiency and productivity in various sectors [9]. However, integrating CPS with
agriculture has not received much attention. There are a few studies that have made
use of CPS platforms in some way or the other such as pest control in farms, smart
farming [10–12], applications of big data and IoT in agriculture [13] and robotics in
farming [14]. In the following sections, the authors propose a conceptual framework
of CPAS based upon the underlying principles of Industry 4.0. This article focuses on
the concepts of cyber-smart farming systems based on CPAS which forms a possible
solution for answering food security problems.
the historical field data, the decisions pertaining to sowing of the crops are taken
for enhancing productivity. The CCU consists of computers where data analytics
is performed using software as a service (SaaS) [19, 20] which helps in improved
decision-making capabilities. Another aspect to focus here is the creation of visibility
in the agricultural supply chain by having access to data from all the phases.
The farm vehicles are equipped with automated sensors which uptake GIS data
and perform instructions as instructed by the CCU. There is a provision for smart
warehousing which utilizes the data for post-harvest storage and paves the way for big
data-enabled value-added warehousing [21]. This helps the farmers in decisions on
how much to store and how much to sell their produce. The CCU is directly connected
with the cloud and performs decision-making based on business intelligence and
domain knowledge. It serves as a decision support system (DSS) to the farmers/users.
The decisions taken at the CCU can be routed to the farmers via Wi-Fi-enabled smart-
phones/computers which serve as a user interface. The user interface can control the
entire activities going in the field from their smartphones/computers. CCU and the
cloud interface act as a service platform for the farmers/users. An important aspect
of the cyber-smart farming systems is that many organizations/stakeholders can par-
ticipate and benefit from the improved decision-making capabilities from the DSS.
CCU also performs the task of cybersecurity by preventing data leakage and espi-
onage. This concept of the cyber-smart farming systems is called as the digitization
in agricultural systems. This entire idea of the cyber-smart farming systems uses the
concept of the cyber-physical agricultural systems. The next section deals with the
basic architecture of the CPAS.
810 R. Sharma et al.
Industry 4.0 deals with the digitization of industries with the help of technologies
like IoT, big data analytics, cloud computing, cyber-physical systems (CPSs) and
blockchain technologies [22]. Xu and Duan [23] proposed nine enablers for real-
izing Industry 4.0. Based on these enablers, a cyber-physical production system
(CPPS) is realized [24]. While there have been applications of IoT [25], big data
[26], drones [27], smart farming [28] in agriculture, very few studies have touched
upon the CPAS phenomena. Figure 2 highlights a five-tier architecture of the CPAS
which is an evident technology for the cyber-smart farming systems. The physical
layer is embedded with IoT infrastructure which captures the real-time data from
physical agricultural assets like farming equipment, machineries, soil, etc. The IoT
infrastructure consists of devices like sensors, radio frequency identification devices
(RFIDs), geographic information systems (GISs) and big GIS analytics [29]. The
function of IoT infrastructure is to capture raw data from field and transfer it to upper
layers. Real-time data is captured to track the behavior of the cyber-smart farming
systems. This data is stored in the cloud from where it is retrieved on a real-time
basis. The data captured is moved to the cyber-layer where a digital twin of the entire
facilities available at the farm is present. The concept of the digital twin enables the
people at CCU to study the behavior of the systems virtually and observe the ana-
lytical outcomes by predicting, modeling and analyzing by algorithms, etc., on the
cyber platform. This entire integrated technology is referred to as the cyber-physical
agricultural systems as shown in Fig. 2.
The platform tier consists of computer units where data analytics is done. The data
collected is converted into information and is disseminated as knowledge by using
analytics tools. The platform tier meets the enterprise tier where central computer is
located, and most of the decision making is being done using business intelligence
by the experts. The enterprise tier merges with cloud tier which is the centralized
data storage and retrieval platform. The application layer is used to integrate appli-
cations such as blockchain technology with the cloud and would help in providing
transparency, provenance solutions for the stakeholders.
Unlike smart farming, the CPAS focuses on autonomous decision making. The
analytics (prescriptive and predictive) algorithms guide CPAS to take appropriate
and accurate decisions out of the best available alternatives and instruct the physical
assets in the farm. The CPAS simulates/models the actual farm scenarios and pre-
dicts the outcomes to the users based on the soil condition, weather data, irrigation
requirement, seeding (predicts the crop to be sown), harvesting, farm management,
etc. Blockchain as an application will enhance the traceability and provenance in the
agricultural supply chains with the help of a decentralized database [30]. Other fea-
tures such as data anonymity and privacy will enhance cybersecurity. CPAS differs
from cyber-physical production systems (CPPSs) in the way that a CPAS is versatile
in design specifications in comparison with a CPPS. It must be robust as it operates
in a natural environment unlike the CPPS, which is controlled in a human-centered
environment. This makes the CPAS more responsive toward forecasting of weather-
related events. High investment is required for setting up a CPPS as it requires setting
up of sophisticated machinery.
4 Conclusion
As Industry 4.0 is going to create a waterfall effect in the manufacturing sector, the
other sectors should try to embrace Industry 4.0 as its benefits are many. Agricul-
ture sector can reap the potential benefits by deploying CPAS-enabled cyber-smart
farming systems. An important implication of this article is that the governments
from various countries can step forward and incorporate Industry 4.0 solutions in
their agricultural policies. The future scope lies in actualizing the illustrated CPAS-
based cyber-smart farming framework in real-time agricultural systems. This will not
only help in creating visibility across the agricultural supply chains but also leverage
them for an increased performance, thus presenting a possible solution in feeding
the coming generations.
References
1. Scanlon BR, Ruddell BL, Reed PM, Hook RI, Zheng C, Tidwell VC, Siebert S (2017) The
food-energy-water nexus: transforming science for society. Water Resour Res 53(5):3550–3556
2. Mancosu N, Snyder R, Kyriakakis G, Spano D (2015) Water scarcity and future challenges for
food production. Water 7(3):975–992
3. Food and Agriculture Organization of the United Nations (FAO) (2015) State of food insecurity
in the world 2015. FAO, Rome
812 R. Sharma et al.
4. Evans A (2009) The feeding of the nine billion: global food security for the 21st century.
Chatham historical society incorporated
5. Tilman D, Fargione J, Wolff B, D’antonio C, Dobson A, Howarth R, Schindler D, Schlesinger
WH, Simberloff D, Swackhamer D (2001) Forecasting agriculturally driven global environ-
mental change. Science 292(5515):281–284
6. Millennium Ecosystem Assessment (2005) Ecosystem and human well-being: biodiversity
synthesis. World Resources Institute, Washington, DC
7. Schröder C (2016) The challenges of industry 4.0 for small and medium-sized enterprises.
2016 Available at https://www.researchgate.net/profile/Christian_Schroeder10/publication/
305789672_The_Challenges_of_Industry_40_for_Small_and_Medium-sizedEnterprises/
links/57a1dfe808ae5f8b258a5e69.pdf
8. Kamble SS, Gunasekaran A, Sharma R (2018) Analysis of the driving and dependence power
of barriers to adopt industry 4.0 in Indian manufacturing industry. Comput Ind 101:107–119
9. Pereira AC, Romero F (2017) A review of the meanings and the implications of the Industry
4.0 concept. Procedia Manuf 13:1206–1214. https://doi.org/10.1016/j.promfg.2017.09.032
10. Wolfert S, Ge L, Verdouw C, Bogaardt MJ (2017) Big data in smart farming–a review. Agric
Syst 153:69–80
11. Doyle M (2018) https://www.ptc.com/en/product-lifecycle-report/how-the-internet-of-things-
helps-grow-our-food. Accessed on 4 Sept 2018
12. Ward J (2018) https://www.forbes.com/sites/sap/2017/07/17/how-the-internet-of-things-is-
fueling-the-f-35-of-the-farm-fields/#51e453028739. Accessed on 3 Sept 2018
13. Chen M, Mao S, Liu Y (2014) Big data: a survey. Mobile Netw Appl 19:171–209
14. Faulkner A, Cebul K (2018) Agriculture gets smart: the rise of data and robotics, cleantech
agriculture report. Cleantech group. https://www.cleantech.com/wp-content/uploads/2014/07/
Agriculture-Gets-Smart-Report.pdf. Accessed on 4 Sept 2018
15. Marnewick C, Marnewick AL (2019) The demands of industry 4.0 on project teams. IEEE
Trans Eng Manage
16. Karadayi-Usta S (2019) An interpretive structural analysis for industry 4.0 adoption challenges.
IEEE Trans Eng Manage
17. Ahuett-Garza H, Kurfess T (2018) A brief discussion on the trends of habilitating technologies
for industry 4.0 and smart manufacturing. Manuf Lett 15:60–63
18. Lee J, Bagheri B, Kao HA (2015) A cyber-physical systems architecture for industry 4.0-based
manufacturing systems. Manuf Lett 3:18–23
19. Dutt A, Jain H, Kumar S (2018) Providing software as a service: a design decision (s) model.
Inf Syst e-Bus Manage 16(2):327–356
20. Dempsey D, Kelliher F (2018) Business-to-business client relationships in the cloud computing
software as a service realm. In: Industry trends in cloud computing. Palgrave Macmillan, Cham,
pp 83–109
21. Sharma R, Parhi S (2017) A review on use of big data in warehousing to enhance accessibility of
food. In: communication and electronics systems (ICCES), 2017 2nd international conference
on IEEE, pp 663–667
22. Xu LD, Duan L (2018) Big data for cyber physical systems in industry 4.0: a survey. Enterp
Inf Syst: 1–22
23. Rüßmann M, Lorenz M, Gerbert P, Waldner M, Justus J, Engel P, Harnisch M (2015) Industry
4.0: the future of productivity and growth in manufacturing industries, vol 9. Boston Consulting
Group
24. Bocciarelli P, D’Ambrogio A, Giglio A, Paglia E (2017) A BPMN extension for modeling
cyber-physical-production-systems in the context of industry 4.0. In: IEEE 14th international
conference on networking, sensing and control (ICNSC), pp 599–604, May 2017
25. Luthra S, Mangla SK, Garg D, Kumar A (2018) Internet of things (IoT) in agriculture
supply chain management: a developing country perspective. In: Emerging markets from a
multidisciplinary perspective. Springer, Cham, pp 209–220
26. Sonka S, Ifamr I (2014) Big data and the ag sector: more than lots of numbers. Int Food
Agribusiness Manage Rev 17(1):1–20
Industry 4.0 Applications in Agriculture: Cyber-Physical Agricultural … 813
27. Tripicchio P, Satler M, Dabisias G, Ruffaldi E, Avizzano CA (2015) Towards smart farming and
sustainable agriculture with drones. In: Intelligent environments (IE), international conference
on IEEE, pp 140–143
28. Jayaraman PP, Yavari A, Georgakopoulos D, Morshed A, Zaslavsky A (2016) Internet of things
platform for smart farming: experiences and lessons learnt. Sensors 16(11):1884
29. Sharma R, Kamble SS, Gunasekaran A (2018) Big GIS analytics framework for agriculture
supply chains: a literature review identifying the current trends and future perspectives. Comput
Electron Agric 155:103–120
30. Kamble SS, Gunasekaran A, Sharma R (2019) Modeling the blockchain enabled traceability
in agriculture supply chain. Int J Inf Manage. https://doi.org/10.1016/j.ijinfomgt.2019.05.023
Person Following Mobile Robot Using
Multiplexed Detection and Tracking
Abstract Helper robots are widely used at various places like airports and railway
stations. This paper presents a pipeline to multiplex the tracking and detection of a
person in dynamic environments using a stereo camera in real time. Recent devel-
opments in object detection using ConvNets have led to robust person detection.
These deep convolutional neural networks generally fail to run with high frame rates
on devices with less computing power. Trackers are also used to retain the identity
of the target person as well as impose fewer constraints on hardware. A concept of
multiplexed detection and tracking is introduced which makes the pipeline faster by
many folds. TurtleBot-2 is used for prototyping the robot and tuning of the motion
controller. Robot operating system (ROS) is used to set up communication between
various nodes of the pipeline. The results found were comparable to recent state-of-
the-art person followers and can be readily used in day to day life usages.
1 Introduction
With immense progress in the field of science and technology, robots have almost
become an inevitable part of the human race. We have relied on robots for all kinds of
help ranging from industrial manufacturing to household work. This paper proposes
a special helper robot that can follow a person without losing its track in real time.
This is especially helpful at various places like airports, shopping malls, and railway
stations as a baggage carrier. The user is just required to keep baggage on the robot
which will maintain a constant distance from the person. If the person moves forward,
the mobile robot will also move forward and if the person moves backward, the robot
will also move backward. Its usage can further be extended at restaurants for helping
the waiter to carry more quantity of food at a single go. These robots can be thus
considered as societal aiding.
There were many challenges associated with the project. The first challenge is
to track the back of a person. Since it is a follower robot, it will mostly track the
backside of a person. Generally, front side is easier to track (e.g., face) since facial
features are more distinguishable compared to the backside. So, we require a highly
accurate detector, which provides feedback to the tracker after regular intervals to
compensate for the accumulated error. This also helps in solving the challenge of
losing a person in a frame. The second challenge was of partial/complete occlusion.
Using only trackers was found to be very inefficient since it loses the target person
and starts tracking another person in a case where a random person crosses between
the camera and the targeted person. Apart from this, while following a person, a robot
can come across obstacles which can be solved using distance sensors like TSOP
and ultrasonic sensor.
The architecture described in this paper was tested on a system running on Ubuntu
16.04 with Intel Core i7-8750H @ 2.20GHz CPU. Intel RealSense D415 camera
was used which uses Active IR stereo technology for depth estimation. Camera has
resolution up to 1920 × 1080 and frame rate up to 60 fps.
Section 2 presents related work in this field and compares it with our work.
Section 3 presents a systematic overview of the abstracted parts of the pipeline along
with the Cyber-physical Architecture of the robot. Section 4 presents the results of
our work. Section 4 concludes our paper and presents the future scope and limitations
of the robot.
2 Related Works
Although people detection and tracking with a single camera is a well-explored topic,
it has been observed that detecting humans from a single image while maintaining a
low false detection rate is a very difficult problem. There are various deep learning
detection algorithm for robust person detection like Faster R-CNN [7], SSD [4], etc.
These highly sophisticated algorithms fail to run in real time on CPU. One of the
work includes tracking using HOG features which works at very high frame rates
[1] but the drawback is that it produces many false results. This can also be achieved
using face detection and tracking [3] but the major downside of this approach is that
it requires a person to face the robot always which defies the robot’s purpose. Leader
detection and tracking have been demonstrated using stereo and feature tracking [2],
mean-shift color histograms [10], and machine learning [5]. Other researchers have
used laser rangefinder (LRF) [9] to detect and track leg movements. Authors of [8]
have used three depth templates (Left, Front, Right) where the target is 2 m away
from the camera. For tracking a person, extended Kalman filter is used.
Person Following Mobile Robot Using Multiplexed … 817
Depth
Detector
Image
RGB
Image
Cartesian Error
TurtleBot
Tracker Coordinate Estimator
Movement
Estimator and Controller
3 Blocks of Pipeline
The complete pipeline for the person following robot can be broken down into the
following four parts (Fig. 1).
3.1 Tracking
Before the robot starts to track a person, the user needs to manually select the tar-
geted person’s bounding box. This locked target is then tracked by the tracking
algorithm. There are various implementations of trackers pre-coded in the OpenCV
library. Tracking is used to determine the position of a person in successive frames.
After testing of diverse kind of trackers (Table 1), Kernelized correlation filter (KCF)
tracker was found to be best suited for our case. This tracker provided high tracking
accuracy and a frame rate of around 60 fps (max frame rate of camera).
Camera
TurtleBot
3.2 Detection
Yolov3 [6] is used for detection of all the person in the given RGB frame. RealSense
stereo camera returns images with RGB and depth channels (RGBD). Depth channel
contains distance information for every pixel encoded in 12-bit integer value. A list
of coordinates of bounding boxes of persons is obtained on passing an RGB image
through the model. Distance of all persons in this list are obtained from the depth
image. Persons who do not fall within 1–4 m range are removed from the list. This
approach creates a search region in z direction of the robot’s frame of reference
(Fig. 2). The robot searches other parts only when the tracked person is lost.
The major advantage of tracking is that it preserves identity. The output of the detec-
tion model is a list of bounding boxes with no clue of identity. To overcome this
problem, we use a tracking algorithm.
However, tracking algorithms alone are prone to accumulation of errors over time.
To overcome this problem, the detection model is used to correct this error after every
hundred frames which act as feedback for the tracker. Deep learning-based detection
is more robust to occlusion but also require high computing power. The computation
speed of tracking algorithms also depends on the area to track. Reducing the area
increases computation speed but it also decreases its accuracy and the tracker is
lost easily. We fixed the tracker size to a 100 × 100 pixel window as it gave the
best empirical results and also performed at 60 fps (without the detector running).
The depth frame contains information of distance in the form of a grayscale image
where each pixel value is mapped to the pixel’s distance from the camera. Center
of bounding box obtained from Yolov3 can be considered to estimate the person’s
Person Following Mobile Robot Using Multiplexed … 819
distance from the robot. However, the depth contains a lot of noise and hence just
selecting a single pixel to estimate depth can lead to a highly noisy estimate of depth.
To overcome this problem, we considered a region of size 50 × 50 pixel in the center
of the tracking window (100 × 100 pixel). Gaussian filtering is performed on this
small patch to suppress unnecessary spikes/noise. A 2-D average is taken over the
region to obtain an averaged scalar depth information of the person. This depth is
considered in d̂ direction in Fig. 2 of the person in robot’s frame of reference. The x-
coordinate can be obtained directly as the x-coordinate of the center of the bounding
box.
TurtleBot (Fig. 3) uses a differential drive mechanism for locomotion. Setpoint for
d̂ direction is set to be 1.5 m. This signifies that the robot will always try to maintain
1.5 m from the person. This value can be adjusted according to person’s convenience.
The algorithm is designed such that if the person moves away from the center of the
frame (1280/2 = 640 pt) the robot will rotate about its axis such that the person
always remain at the center of the frame. Error in the current coordinate and set
point is obtained in the direction x̂ and d̂, respectively. This error is passed through
the proportional, integral, and derivative (PID) controller to continuously obtain
modulated control velocities. The obtained values of velocities are passed to the
TurtleBot using the ROS framework.
820 K. Agrawal and R. Lal
3.5 Algorithm
In this paper, we have presented a deep learning and computer vision-based pipeline
for person detection and tracking for person following in real time at very high
frame rates (Fig. 4). We have performed the experiments of the person following
robot under various conditions and found it as a useful service robot. Future work
will include making the algorithm robust against full occlusions and failed detection.
In the current version, we are tracking the person random of the person but for better
tracking, we can use SIFT/SURF feature tracking. The obstacle avoidance feature can
also be integrated using various sensors. Since we prototyped the model on TurtleBot,
efforts can be made to make it all-terrain ground vehicle which will diversify its usage.
5 Disclaimer
standards. Informed consent was obtained from all individual participants included in
the study. Additional informed consent was obtained from all individual participants
for whom identifying information is included in this article.
References
1. Brookshire JD (2011) Person following using histograms of oriented gradients. US Patent App.
12/848,677, 3 Feb 2011
2. Chen Z, Birchfield ST (2007) Person following with a mobile robot using binocular feature-
based tracking. In: 2007 IEEE/RSJ international conference on intelligent robots and systems.
IEEE, pp 815–820
3. Kansal S, Makhal A, Chakraborty P, Nandi GC (2012) Tracking of a target person using
face recognition by surveillance robot. In: 2012 international conference on communica-
tion, information computing technology (ICCICT), Oct 2012. https://doi.org/10.1109/ICCICT.
2012.6398104
4. Liu W, Anguelov D, Erhan D, Szegedy C, Reed S, Fu CY, Berg AC (2016) Ssd: Single shot
multibox detector. In: European conference on computer vision. Springer, Berlin, pp 21–37
822 K. Agrawal and R. Lal
1 Introduction
Electroless method of coating is an autocatalytic process where metal ions are reduced
on the activated substrate after immersing it in the reducing agent comprising elec-
troless bath, and the process is continued till the bath solution gets depleted [1].
The wear properties of the Ni–P coatings at elevated temperatures are topic of inter-
est for several years. It is well established that wear resistance of Ni–P coatings
is improved by incorporating particulate matter into Ni–P matrix [2, 3]. Silver as
a solid lubricant has excellent chemical stability, oxidation resistance and thermal
conductivity, thereby making it applicable for high-temperature applications [4, 5].
The nanocrystalline Ni–P/Ag composite coating prepared using electroless deposi-
tion was first reported by Ma et al. [6]. Alirezaei et al. reported the amount of silver
particles in Ni–P matrix, and heat treatment had shown significant effect on hardness
and mechanical properties of Ni–P/Ag coating [7]. Further, the Ni–P/Ag coatings
showed good tribological performance at elevated temperatures due to the formation
of self-lubricating thin layer containing silver during sliding wear test [8]. The study
on Ni–P/Ag coatings using electroless deposition is restricted mostly to deposition
mechanism, structural and mechanical properties. The studies on high-temperature
sliding wear characteristics of Ni–P/Ag composite coatings are limited.
In the present study, the silver particles incorporated Ni–P coatings were deposited
on steel substrate using electroless deposition technique. The prepared coatings were
heat-treated at 400 °C, and the influence of silver particles and heat treatment on
elevated temperature sliding wear characteristics of Ni–P/Ag coatings were analyzed.
2 Experimental Details
Electroless bath for coating was prepared from a commercial solution of reagent
quality. The operating conditions for the deposition of Ni–P and Ni–P/Ag coatings
are specified in Table 1. The mild steel samples (AISI 1018 grade) were finished using
SiC papers of grade 150, 320, 600 and 800 sequentially on a disc polishing grinder.
The finished sample is subjected to degreasing with 10% NaOH solution. Subse-
quently, the sample is subjected to ultrasonication in ethanol solution for 5 min. The
activation of sample is done using 40% HCl for 2 min. Silver particles of 2–3.5 µm
size procured from Sigma-Aldrich (99.9% pure) were used for the experiment. The
particles were subjected to ultrasonication with suitable amount of cationic surfac-
tant (CTAB) solution before adding into electroless bath. After deposition, some of
the Ni–P and Ni–P/Ag samples were heat-treated in a tubular furnace at 400 °C for
1 h.
The hardness of the coatings was measured using Mitutoyo microhardness tester
at different locations on the sample at an applied load of 100 g. The wear character-
istics of the coatings were evaluated using linear reciprocating tribometer (Ducom
TR281). The tests were conducted in room temperature and 300 °C under dry sliding
conditions against the AISI E52100 steel ball (Ø 6 mm). A load of 10 N was applied
Table 1 Operating
Operating condition Ni–P Ni–P–Ag
conditions for deposition of
Ni–P and Ni–P–Ag coatings Nickel source 5–6 g/L 5–6 g/L
Sodium hypophosphite 24–36 g/L 24–36 g/L
Silver particles – 0.2 g/L
Surfactant (CTAB) – 150 ppm
pH 5.5–6 5.5–6
Temperature 87 ± 2 °C 87 ± 2°C
Deposition time 2 h 30 min 2 h 30 min
Sliding Wear Characteristics of Silver Particles … 825
As shown in Fig. 1, the hardness of Ni–P/Ag was 10% less than the Ni–P coating,
which is attributed to the low hardness of soft silver particles. Several authors reported
a decreased value of hardness with the incorporation of soft particles in the Ni–P
matrix [7, 9]. The higher hardness of Ni–P/Ag coatings under heat-treated conditions
is due to Ni–P coating crystallization and Ni3 P phase precipitation [10].
Figure 2 presents the variation in friction coefficient with respect to time of
as-deposited Ni–P and Ni–P/Ag coatings at room and 300 °C sliding conditions.
Under room temperature sliding conditions, a better trend in friction coefficient was
observed for Ni–P coatings compared to Ni–P/Ag coating. In as-deposited condi-
tions, both the coatings have similar amorphous structure. However, slightly lower
friction in AD Ni–P/Ag coating is due to the solid lubrication effect of silver parti-
cles in the coating. The wear rate observed in AD Ni–P/Ag coating is significantly
high compared to Ni–P coating attributed to the weaker matrix structure obtained
with the incorporation of silver particles. As shown in Fig. 2b, AD Ni–P/Ag coating
showed much higher friction coefficient than the AD Ni–P coating at 300 °C. Coating
delamination and failure occurs in as-deposited Ni–P/Ag coating at high temperature
attributed to weaker structure of the coating. The lower friction coefficient observed
in Ni–P coatings when tested at high temperature is due to slight crystallization
because of in situ heat treatment during 300 °C wear test. This transformation leads
Fig. 2 Frictional characteristics of as-deposited a Ni–P and Ni–P/Ag coating at room temperature,
b Ni–P and Ni–P/Ag coating at 300 °C
Fig. 3 Frictional characteristics of heat-treated a Ni–P and Ni–P/Ag at room temperature and
b Ni–P and Ni–P/Ag at 300 °C
to increased hardness and strength in coatings which results in reduced wear rate and
decreased friction coefficient.
Figure 3 shows the variation in friction coefficient of heat-treated (HT) coatings
at room temperature and 300 °C. Under room temperature sliding conditions, HT
Ni–P/Ag coating displays much better frictional characteristics than the HT Ni–P
coating. Further, it is also observed (from Fig. 4b) that the weight loss of HT Ni–P–
Ag coating is 25% less than that of HT Ni–P coating under similar conditions. It is
hypothesized that solid composite particles with low shear strength act as lubricants
reservoir in coatings. Slow sequential flow to the interface during wear test leads
to a lower friction coefficient. The mechanism explained is valid only when the
surrounding matrix, i.e., the Ni–P matrix provides enough structural support to the
coating. In AD coating, the structural support provided by the Ni–P matrix is observed
to be weak and resulted higher wear rates. Under 300 °C sliding conditions, the HT
Ni–P coating (as shown in Fig. 3b) showed a lower friction coefficient compared to
that of Ni–P/Ag HT coating. It is also observed weight loss of Ni–P/Ag HT is more
when compared to that of the Ni–P HT coating.
Sliding Wear Characteristics of Silver Particles … 827
Fig. 4 Bar diagram showing the weight loss of a as-deposited, and b heat-treated, Ni–P and
Ni–P–Ag coatings at room temperature and 300 °C
Figure 5 shows the wear track morphology of as-deposited and heat-treated Ni–
P/Ag coatings at room and 300 °C. At room temperature sliding conditions, the
more wear track width is observed for AD coating (~1.22 mm) compared to that
of HT coating (~244 µm). Under cyclic loading, due to its weak structure, the AD
coating fails to withstand the stresses resulting in delamination. The wear in as-
deposited coatings is governed by abrasive and adhesive wear mechanisms. The
Fig. 5 Wear track morphology of a as-deposited, b heat-treated, Ni–P/Ag at room temperature and
c as-deposited, d heat-treated, Ni–P/Ag at 300 °C
828 B. Ramakrishnan et al.
wear track formed in heat-treated coatings is much more resilient than as-deposited
coatings. Heat treatment of the coatings leads to the formation of crystalline Ni
and precipitation of Ni3 P. The silver particles are entrapped in the crystalline phase
and do not contribute to the crystallization. The crystalline phase formed in the
coating provides the structural support to the coating, thereby increasing load bearing
capacity. The silver entrapped in this crystalline phase acts as a solid lubricant during
sliding resulting in low friction coefficient and low wear rate. At 300 °C sliding
conditions, more wear track width (~1.24 mm) is observed for the AD Ni–P/Ag
coating. The reduced strength of the coating at high temperature contributes to severe
plastic deformation resulting in higher wear rate. The mixed mechanism of adhesive
and fatigue wear is observed in AD and HT Ni–P/Ag coatings at 300 °C.
4 Conclusions
Silver particles are successfully co-deposited into Ni–P matrix by the electroless
deposition technique. The silver particles in the coating affected the friction and
wear characteristics significantly. At room temperature sliding conditions, the heat-
treated Ni–P/Ag coatings showed greater wear resistance attributed to the synergetic
effect of self-lubrication provided by silver particles and higher hardness due to heat
treatment. At 300 °C sliding conditions, irrespective of the coating condition, the
variation in friction coefficient and wear resistance is negligible. It is concluded that
the addition of silver has lubricating effect only if the coating matrix can sustain
the test load. Otherwise, the friction coefficient and wear rate deteriorate due to
delamination of the coatings. It is proposed that the quality and mechanical integrity
of Ni–P/Ag coatings can be improved by optimizing bath parameters and surfactant
quantity.
Acknowledgements The authors are thankful to SERB, Department of Science and Technology,
New Delhi, for funding to carry out this research work under the grant: ECR/2016/000654.
References
1. Sarret M, Müller C, Amell A (2006) Electroless NiP micro-and nano-composite coatings. Surf
Coat Technol 201:389–395
2. Agarwala RC, Agarwala V (2003) Electroless alloy/composite coatings: a review. Sadhana
28:475–493
3. Sahoo P, Das SK (2011) Tribology of electroless nickel coatings—a review. Mater Des 32:1760–
1775
4. Torres H, Rodríguez M, Prakash RB (2018)Tribological behaviour of self-lubricating materials
at high temperatures. Int Mater Rev 63:309–340
5. Alirezaei S, Monirvaghefi S, Saatchi A, Ürgen M, Motallebzadeh A (2013) High temperature
friction and wear behaviour of Ni–P–Ag–Al2 O3 hybrid nanocomposite coating. Trans IMF
91:207–213
Sliding Wear Characteristics of Silver Particles … 829
Abstract The present work investigates the performance of a single cylinder, air
cooled, direct injection (DI) compression ignition engine, using blends of biodiesel
from non-edible feedstock-Citrullus colocynths L. methyl ester (CCME) and Diesel.
The engine was tested with diesel (B0), 10% Biodiesel (B10), 15% biodiesel (B15)
and 20% biodiesel (B20). The biodiesel is produced using transesterification process.
The investigations have been studied for two compression ratio (CR) values of 14 and
16 with different blends of biodiesel at different load conditions. CO2 emissions were
found to increase with increase in blending and CR, whereas increase in load showed a
decrease in exhaust CO2 values. HC emissions decrease substantially with increasing
blend ratio of biodiesel, compression ratio and load on engine. NOx emissions were
increased with all blends of biodiesel, and increase in compression ratio and load
also evidenced a subsequent increment in NOx emissions in exhaust.
1 Introduction
M. Zeeshan
School of Mechanical Engineering, Lovely Professional University, Phagwara, Punjab, India
S. K. Sharma (B)
Department of Mechanical Engineering, Amity University Chhattisgarh, Raipur, Chhattisgarh,
India
e-mail: sksharma@rpr.amity.edu
noise and particle size concentration/distribution. They concluded that the transient
operation showed the decrease in PM, HC and CO and an increase in NOx emissions
when biodiesel % was increased in the blend. Highly unsaturated feedstocks result
in higher NOx emissions; CO and HC, which generally decreased with the increase
in biodiesel, show reverse trends in cold starting conditions due to limited efficiency
of the catalytic convertor.
2 Experimental Methodology
The production of biodiesel includes the conversion of oils and fats esters or biodiesel.
It is also known as fatty acid methyl esters (FAME), and the process is named as
transesterification. About 45.36 kg of oil or fat when reacted with 5.54 kg of a
short chain alcohol (usually Methanol) in the presence of a catalyst (usually Sodium
hydroxide [NAOH] or Potassium hydroxide [KOH]) formed 45.36 kg and 4.54 kg
of biodiesel and glycerine respectively. Figure 1 provides the experimental set-up of
the variable compression engine [1].
The process variables were the preheating temperature of the oil was maintained
at 60 °C. The transesterification reaction was carried out at temperature close to the
boiling point of methanol, i.e. 60–70 °C. The molar ratio of 6:1 (alcohol to oil) and
temperature ranging from 60 to 80 °C yield maximum ester. Sodium alkoxide as
catalyst in the range of 0.5–1% by weight gives 94–99% conversion of vegetable oil
into esters [9].
Figure 3a, b shows the variation of CO2 emission at different CR. It is evident from
the results that CO2 emissions with biodiesel blends are increased at all loads due to
the presence of oxygen in the cylinder helps to increase conversion of CO to CO2 ,
and hence, CO2 emissions are found to be the highest with higher blend ratio, i.e.
B20. It is also noticed that CO2 emissions are increased nominally with increase in
CR from 14 to 16 which can be attributed to the fact that increasing the compression
ratio improves the combustion of fuel, and hence, more CO2 is emitted as a result of
complete combustion.
4 Conclusion
References
1. Rashid U, Anwar F, Moser BR, Knothe G (2008) Moringa oleifera oil: a possible source of
biodiesel. Biores Technol 99:8175–8179
Investigations on Engine Emission Using Biodiesel … 837
2. Santori G, Nicola GD, Moglie M, Polonara F (2012) A review analysing the industrial biodiesel
production practice starting from vegetable oil refining. Appl Energy 92:109–132
3. Ng JH, Ng HK, Gan S (2010) Advances in biodiesel fuel for application in compression ignition
engines. Clean Techn Environ Policy 12:459–493
4. Qiu F, Li Y, Yang D, Li X, Sun P (2011) Biodiesel production from mixed soybean oil and
rapeseed oil. Appl Energy 88:2050–2055
5. Lapuerta M, Armas O, Fernandez JR (2008) Effect of biodiesel fuels on diesel engine emissions.
Prog Energy Combust Sci 34:198–223
6. Vasudeva M, Sharma S, Mohapatra SK, Kundu K (2013) Optimizing the compression ratio of
compression ignition engine fuelled with esters of crude rice bran oil. Int J Theor Appl Res
Mech Eng 2:131–134
7. Roy MM, Wang W, Bujold J (2013) Biodiesel production and comparison of emissions of a DI
diesel engine fueled by biodiesel–diesel and canola oil–diesel blends at high idling operations.
Appl Energy 106:198–208
8. Giakoumis EG, Rakopoulos CD, Dimaratos AM, Rakopoulos DC (2012) Exhaust emissions
of diesel engines operating under transient conditions with biodiesel fuel blends. Prog Energy
Combust Sci 38:691–715
9. Planning Commission of India (2003) Report of the committee on development of bio-fuel, New
Delhi, India
Investigation of Thermal Desalination
System Using Heat Recovery
Abstract The work presented in this paper is an attempt to investigate the per-
formance of thermal desalination system using heat exchanger. Helical coil heat
exchanger has been used as a condenser to condense vapors passing over the coil
where cooling water in the coil absorbs latent heat of condensation as well as sen-
sible heat of vapors. Experiments were conducted for different initial temperature
of water. Results have shown that application of a separate condenser for thermal
desalination system augments rate of distillate output. Condenser also performs the
function of heat recovery. In present work, heat recovery has augmented distillate
output by 50%. Use of warm water for desalination improved vapor generation and
its temperature. This has improved heat recovery of the system.
1 Introduction
Studies have revealed that freshwater available on earth for human consumption is
only about 1%. Developments in social life have boosted use of freshwater. Freshwa-
ter is also required for industrial applications. Desalination process is worldwide used
for conversion of saline water into potable water. Most of desalination techniques use
electricity. Solar still is a low-priced device for thermal desalination which can be
used without electricity. Researchers have investigated solar stills incorporated with
different energy storage materials to use solar stored energy during night time [1].
R. K. Petkar (B)
Symbiosis International (Deemed University), Pune, India
e-mail: rajanpetkar@gmail.com
Department of Mechanical Engineering, Zeal College of Engineering and Research, Pune, India
C. R. Sonawane
Department of Mechanical Engineering, Symbiosis Institute of Technology, Pune, India
H. N. Panchal
Department of Mechanical Engineering, Government Engineering College, Patan, India
In conventional solar still, use of glass cover for condensation wastes latent heat of
vapor condensation to the atmosphere. Panchal et al. [2] have performed experiments
on double basin solar still in order to use latent heat of condensation in lower basin.
They used vacuum tubes in order to boost performance of solar still. It was observed
that calcium stones as energy storage materials yield more distillate output.
Kumar et al. [3] connected inclined solar still with pyramid type solar still. Baffles
were used in inclined solar still to increase surface area for solar radiation capture.
Inclined solar still yields distillate as well as provides hot water from its outlet to the
pyramid type solar still in order to augment its performance. These integrated stills
give more distillate output per unit area as compared to that of individual solar stills.
Effect of environmental conditions, physical parameters, solar reflectors and different
output enhancing arrangements has been studied on the performance of solar still [4].
Environmental conditions include climatic condition, wind speed, etc., and physical
parameters include glass cover inclination angle, depth of water in the still, glass cover
temperature, etc. Li et al. [5] used compound parabolic concentrator and evacuated
tube absorber coupled with bar type heat pipe to provide heat energy to seawater
desalination system. They used one desalination unit connected in series with six
units for heat recovery without power consumption. Rashidi et al. [6] studied effect
of partitioning a conventional solar still by experimental and numerical method. It has
been observed that partitioned solar still allows more vortexes in the still which leads
to augment heat and mass transfer. This modification increased hourly as well as daily
distillate output of partitioned solar still as compared to that of conventional solar
still. The results reveal a clear relation between ambient temperature and distillate
output. Ma et al. [7] investigated adsorption desalination system using shell and tube
type heat exchanger for simultaneous evaporation on shell side and condensation
on tube side. They observed 74.7% efficiency of the heat recovery system. Kaushal
et al. [8] used multieffect diffusion with vertically arranged plates for heat recovery.
A conventional solar still involves evaporation and condensation in the same space
and thus, prone to more heat loss. Present work has been studied for heat recovery
system that can be coupled with a solar still.
2 Experimental Study
Nozzle Cooling
water outlet
Cooling
Tco Water Tank
Pressure
cooker
__ __ __ __ _ __ __ __ __ _
__ __ __ __ __ __ __ __
__ __ __ __ _ __ __ __ __ _
Heat
source Tank-C Tank-D
pressure cooker has been used for vapor generation. Cooling water tank was provided
to supply water at room temperature for condensation of vapors. Cooling water was
fed using pump, passes through the coil side of the condenser and collected in cooling
water outlet tank (Tank-C). The condensed water in shell side of condenser flows
from bottom of condenser to the distilled water collection tank (Tank-D).
Key characteristic of the present work is application of a separate condenser. It
provides advantage of following points as compared to a conventional solar still: (1)
condensation area is increased irrespective of the size of vapor generating component
of the system, thus increased condensation rate; (2) use of cooling water reduces
condensation resistance as in case of conventional solar still; (3) all condensed water
flows to the distilled water collection tank, thus freshwater falling back to the still in
case of conventional solar still is evaded; (4) latent heat of condensation of vapors
increases sensible heat of cooling water, and thus, it can be reused for vapor generation
with augmentation in distilled water output.
All experiments were conducted for fixed quantity of water in pressure cooker as
500 ml, before initiating experimentation. Saline water was provided to the cooker
as well as to the cooling water tank to be used as cooling water in the condenser.
Cooling water flow rate was kept constant as 400 ml/min.
Initially, sample experiments were conducted to check stable results of the system.
Later experiments were recorded for two cases followed consecutively for same time
duration. In case-1, water at room temperature was filled in the cooker, and then, it
was heated for 10 min. Vapors generated in the cooker were allowed by means of
842 R. K. Petkar et al.
nozzle to pass through the hose pipe carrying vapors to shell side of the condenser.
As cooling water was passed through coil of the condenser, vapors condense over the
coil, accumulate at bottom of the condenser shell and flow to distilled water collection
tank (Tank-D). Cooling water was collected in Tank-C with increased sensible heat.
In case-2, water collected in Tank-C was filled in the cooker, and then, it was
heated for 10 min. All other operations were same as in case-1. For both cases, the
following data were recorded: (1) vapor temperature before entering the condenser
(T v ); (2) cooling water temperature at inlet of condenser (T ci ); (3) cooling water
temperature at outlet of condenser (T co ); (4) amount of distilled water collected in
Tank-D.
Experimental performance of the present setup was studied for two cases. Case-1 was
performed by filling water at room temperature (36 °C) in pressure cooker, while case-
2 was performed by filling warm water from Tank-C (43.4 °C) in the cooker. Figure 2
shows comparison of vapor temperature, cooling water inlet temperature and cooling
water outlet temperature for the two cases. Though cooling water inlet temperature
in case-2 was slightly lower, corresponding cooling water outlet temperature was
higher. This is due to the warm water used in case-2 generated vapors at higher
temperature.
From Fig. 3, it is observed that distillate output as well as cooling water tem-
perature rise are higher in case-2. It reveals that use of warm water for desalination
has improved vapor generation. It is also clear that heat transfer in case-2 is higher.
Thus, sensible heat of cooling water at outlet of condenser is higher, and it can be
further utilized for desalination to get enhanced results. This confirms that condenser
conditions 80
60
40
20
0
Case-1 Case-2
Experimental condi ons
Vapor temperature (Tv)
Cooling water inlet temperature (Tci)
Cooling water outlet temperature (Tco)
Investigation of Thermal Desalination System … 843
has not only improved distillate output but also performed heat recovery for further
increment in the output.
4 Conclusions
The experimental setup has been developed to study thermal desalination process
with the help of a separate condenser using helical coil heat exchanger. The per-
formance of desalination system is analyzed by considering different experimental
conditions. Following conclusions have been recovered from the present work.
• Use of separate condenser allowed all condensate to be collected in distilled water
collection tank (Tank-D), and thus, loss of condensed water due to re-evaporation
is eliminated as in case of conventional solar still.
• Condenser allows its use for heat recovery as well. Warm water from condenser
coil outlet (Tank-C) was used for further desalination and augmented distillate
output by 50%.
• Vapor generation and its temperature are influenced by initial temperature of
water used for desalination (water used in cooker). Warm water for desalination
has improved heat transfer in the condenser and thus heat recovery.
References
1. Panchal HN (2016) Use of thermal energy storage materials for enhancement in distillate output
of solar still: a review. Renew Sustain Energy Rev 61:86–96
2. Panchal HN, Shah PK (2014) Enhancement of distillate output of double basin solar still with
vacuum tubes. Front Energy 8(1):101–109
3. Kumar PN, Manokar AM, Madhu B, Kabeel AE, Arunkumar T, Sathyamurthy R, Panchal HN
(2017) Experimental investigation on the effect of water mass in triangular pyramid solar still
integrated to inclined solar still. Groundwater Sustain Develop
844 R. K. Petkar et al.
4. Awasthi A, Kumari K, Panchal H, Sathyamurthy R (2018) Passive solar still: recent advance-
ments in design and related performance. Environ Technol Rev 7(1):235–261
5. Li S, Liu Z, Shao Z, Xiao H, Xi N (2018) Performance study on a passive solar seawater
desalination system using multi-effect heat recovery. Appl Energy 213:343–352
6. Rashidi S, Esfahani JA, Rahbar N (2017) Partitioning of solar still for performance recovery:
experimental and numerical investigations with cost analysis. Sol Energy 153:41–50
7. Ma H, Zhang J, Liu C, Lin X, Sun Y (2018) Experimental investigation on an adsorption desali-
nation system with heat and mass recovery between adsorber and desorber beds. Desalination
446:42–50
8. Kaushal AK, Mittal MK, Gangacharyulu D (2017) An experimental study of floating wick
basin type vertical multiple effect diffusion solar still with waste heat recovery. Desalination
414:35–45