313 views

Uploaded by pdhurvey

save

- LRFD - Design - Bridge - Structures - Federal Highway Administration 9
- Automatic Generation of 2D
- +++Paper Collings UK
- Schist - Variation in Strength
- Composite Stress Analysis
- Simple Stress & Strain
- Método Empírico Para Diseño de Sostenimiento Con GSI
- Design of Shotcrete With Macro Fibres
- 10.1177_002199838201600603
- Load Combinations Due to ACI 2008
- 04502-2SA-1
- 14590_DPPS
- Aluminum Matrix Composites
- Primdeb ghfd 2
- Spalling Prediction Methods in High Stress Conditions
- Finite Element Modeling of Full Depth Precast Concrete Transverse
- ASTM D3500-90 (Tension Test).pdf
- New Microsoft Office Word Document (5).docx
- 167_1
- Tempo Spatial Characteristics
- New Microsoft Office Word Document
- Experimental Stress Analysis of Toggle Jack Using Fea Approach
- sprężyna
- 7 Design of Deep Flexural Member
- Blasting in Relation to Strength Index
- Mechanical Report
- 31486 (1)
- Cosmos 2007
- 9a01302 Strength of Materials i
- MECH13
- box
- Deshpande_uta_2502M_10951.pdf
- lec1ppt
- 9. SOM_TYS Exp._2495
- ppt ann
- 04.Bus Station Design.pdf
- arun patel paper egyptian.pdf
- EME-07
- Ahmed Amdihun
- 08-stress_analysis_of_fiber_reinforced_composite_materials.pdf
- hydro
- schedule_2018.pdf
- Schedule 2018
- Cs Tehridamproject
- IJEMS 15(6) 452-458.pdf
- Other Pages
- Index
- Teoria_Elemental_de_Estructuras_Yuan_Yu_Hsieh.pdf
- lec1.pdf
- 110CE0575-5 (1)
- Ques papers SA 2 1
- Ques papers SA 2 2
- Title Page
- 89_Fly_Ash.pdf
- SA2 Ques papers.pdf
- Ques Papers SA 2
- asokan p.pdf
- Content
- lec1.pdf
- Basics of Surveying
- catalogue-general.pdf
- All Schemes With Ministry's
- Woolybutt
- livrable_FONCTION_RSSI
- 4-Raupex Pipes 0509
- Diseño manual de funciones de una empresa
- confiabilidad.pptx
- Deutsch Catalog 2011 IE
- Case Study 2
- Chapter03 (2).doc
- Field Instrumentation.ppt
- Lista Startup
- presentacion cristalizacion
- CRC 91 BarBending Final Report ACI FDN
- M250-B17 OMM Inspection Checksheets
- HRM UNIT-1
- BTS 350
- 1008734 the EPRI Zed-Meter
- fase 1 servicio al cliente
- TUBO DE RUBENS.docx
- Efectos Del Sísmo en El Sistema Suelo-cimentación
- Demantra Installation
- tdt_caso2
- Actividad de Aprendizaje No. 1 Arquitectura de la computadora
- Boot Command
- ILOP
- doremi_datasheet_1_eng_2
- New Innovations in Cell Culture
- 19657 gruas torre.pdf
- Prosonic Flow 93PA1

You are on page 1of 457

ii

iii

iv

v

vi

vii

**International Conference on Industrial
**

Engineering Science and Applications-2014

Contents

Paper ID

02

05

06

08

10

11

13

15

16

17

19

20

21

22

23

24

25

26

28

30

31

33

34

39

42

43

Title of Papers

Progressive Failure Analysis of Laminated Composite Shells – A Review

Application of Neural Network for Identification of Cracks on Cantilever Composite

Beam

Dynamic response of a simply supported beam with traversing mass

Mathematical modeling of steady state temperature distribution due to heat loss

from weld bead of a square butt joint

Computational Analysis of Shell Fluid of Shell and Tube Heat Exchanger Allowing

the Outcome of Baffles Disposition on Fluid Flow

Thermal Analysis of Porous Fin with Internal Heat Generation

Reliability Analysis of Two Lathe Machines Arranged in a Machining System

Genetic algorithm based performance analysis of 3-phase self-excited induction

generator

Comparative Study of Machining Processes by Process Capability Indices

Heat transfer analysis in porous fin of different profiles using Vibrational iteration

method

Patient Information implantation and reclamation from compressed ECG signal by

LSB watermarking technique

Vibration Characteristics of Rotating Simply Supported Shaft in Viscous Fluid

Diagnosis of Damage in Composite Beam Structures using Artificial Neural

Network with Experimental Validation

Finite Element Analysis of Hip Prosthesis for Identification of Maximum Stressed

Zone

Multi Channel Personal Area Network(MCPAN) Formation and Routing

Vibration analysis of a cracked Timoshenko beam

Silicon on Insulator based Directional, Cross Gap and Multimode Interference

Optical Coupler design

An Analysis of Short Term Hydrothermal Scheduling using Different Algorithms

Monitoring of the lung fluid movement and estimation of lung area using Electrical

Impedance Tomography: A Simulation Study

Multi-objective design of realistic load frequency control system using particle

swarm optimization

Hybridizing DE with PSO for Constrained Engineering Design Problems

Harmonic Distortion Optimization of Generalized A-Symmetrical Series/Parallel Multilevel

Converter with Fewer Switches

**Detailed study and proposed restoration of damaged structural bracket supports for
**

three tier insulated piping system by using anchoring methodology in filter house

structure of solvent dewaxing unit

Design and Development of a Heating- Cooling Belt using Thermoelectric

Refrigeration for Medical Purposes

Power line filter design considering losses and parasitic characteristics of passive

lumped components

CFD analysis of cambered airfoil for H-rotor VAWT

Page No.

001-006

007-011

012-014

015-020

021-025

026-031

032-036

037-041

042-045

046-051

052-057

058-063

064-068

069-072

073-079

080-084

085-090

091-096

097-100

101-105

106-111

112-117

118-122

123-127

128-133

134-138

viii

45

47

49

50

55

57

59

61

62

63

66

67

68

69

70

71

72

73

74

76

82

86

87

89

92

93

94

97

98

100

101

102

**International Conference on Industrial
**

Engineering Science and Applications-2014

Modeling of active transformation of microstructure of two-phase Ti alloys during 139-144

hot working

Value Based Planning of Renewable DGs in Distribution Network Incorporating 145-150

Variable Power Load Model and Load Growth

Phase Angle Measurement using PIC Microcontroller with Higher Accuracy

151-154

Automatic Electronic Water Level Management System using PIC Microcontroller

155-158

Effect of Temperature on Photovoltaic Cell performance

159-161

Optimum Process Scheduling Using Genetic Algorithm in an Existing Machine 162-166

Layout

A review parametric performance of solar still

167-172

Differential Difference Current Conveyor (DDCC) Based Current mode Integrator 173-176

and Differentiator

Performance analysis of 2 stroke gasoline engine by using compressed air

178-181

Design of Wide Band Digital Integrator and Differentiator

182-185

Tracking Mobile Targets Through Wireless HART

186-190

Hysteresis Compensation using Modified Internal Model Control for Precise Nano 191-196

Positioning

Reversible Data Hiding using Wavelet Transform and Compounding for DICOM 197-201

Image

Study on Photovoltaic System for Isolated and Non-Isolated Source Cascaded Two 202-206

Level Inverter (CTLI)

Mass Measuring System Using Delay-and-Add Direct Sequence (DADS) Spread 207-210

Spectrum Method

Fault analysis of wind generator connected power system using differential equation 211-215

technique

Biomedical application using zigbee

216-218

Resonant Frequency of 300-600-900 Right Angle Triangular Patch Antenna with 219-222

and without Suspended Substrate

Model Free Adaptive Control in Industrial Process: An Overview

223-226

Simulation of IGBT fed Mirror Inverter based H.F. Induction Cooker

227-230

A Hybrid Intelligent Algorithm Applied in Economic Emission Load Dispatch 231-235

Problems

Performance assessment of power system by incorporating Distributed generation 236-241

and Static VAR compensator

A study of Performance Analysis on Multi-bus Power Grid Network Modeling

242-246

Reliability Assessment of Energy Monitoring Service for a Futuristic Smart City

247-252

Performance Characteristics of 2x50kwp Roof top PV Power Plant System

253-257

Multiple Distributed Generator allocation by modified novel power loss sensitivity 258-262

for loss reduction

A Novel Constraint Increasing Approach for solving Sudoku puzzle

263-267

Kinematic synthesis of six bar gear mechanism

268-272

A Simulation Based Geometrical Analysis Of MEMS Capacitive Pressure Sensors 273-277

for High Absolute Pressure Measurement

RAHSIS: A Tool for Reliability Analysis of Hardware Software Integrated Systems 278-283

Steel Fiber Reinforced Concrete in the service of Civil Engineering

284-289

Optimal Control of singular systems via Haar function

290-293

ix

a successful mode of speed control of AC motors 315-319 Using Appointment system to improve the loading process of trucks in a steel plant: 320-324 A simulation based case study Productivity Improvement Through Line Balancing using Simulation 325-329 A Genetic Algorithm Trained Artificial Neural Network Based Selective Harmonic 330-335 Elimination Technique for Cascade Multilevel Inverters Performance Analysis of Genetic Algorithm in Direction of Arrival of Wideband 336-339 Sources Over Wide SNR Range Intelligent Co-ordinated Control for Boiler Turbine Unit 340-343 Vulnerability Assessment of Reinforced Concrete Buildings having Plan Irregularity 344-348 using Pushover Analysis A comparative analysis of performance of three phase four wire DSTATCOM 349-352 topologies for power quality improvement How Ergonomics play an important role in productivity improvement of an 353-358 organization Modeling and Simulation of YNVD Transformer for Single Phase Electrified 359-362 Traction System Transient Response and Load Sharing Improvement in Islanded Microgrids 363-367 Intelligent Hybrid Fuzzy PD Control for Trajactory Tracking of Robot Manipulator 368-372 and Comparative Analysis Improvement in DC.link Voltage of Doubly Fed Induction Generator using SMES 373-377 Implementation of energy storage and FACT device with renewable power 378-381 generation system Flexible pavement cost modeling for weak subgrade stabilized with fly ash and lime 382-386 Prediction of compression index of clay using artificial neural network 387-390 On the Directivity and Multiband Characteristics of Sierpinski Fractal on Bowtie 391-396 Power Quality Improvement Using DPFC Under Fault Conditions 397-401 Benchmarking and Analysis of the User-Perceived Performance of EPICS based 402-405 ICRH DAC Power Flow Control in Smart Micro Grid using Fuzzy Controllers 406-409 Harmonics Mitigation with the help of Zsource Inverter based DVR 410-414 Assessment of Retailer’s quality in Dairy Supply Chain Using AHP Technique 415-419 Effect of Conflicting Vehicles on Service Delay Under Mixed Traffic Conditions 420-425 Defining Level of Service at Uncontrolled Median Openings through K-Medoid 426-432 Clustering Optimum Design of FRP Rib Core Bridge Deck Panel using Gradient based 433-439 Optimization Optimization design of FRP web core skew bridge deck system using Genetic 440-446 Algorithm x .103 105 107 109 110 114 115 116 117 118 119 120 123 126 127 128 129 130 131 132 133 134 136 137 140 144 146 149 150 159 160 International Conference on Industrial Engineering Science and Applications-2014 Baseline Wander and Power Line Interference Removal in ECG Signal 294-297 Prediction of relative density of clean sand: A support vector machine approach 298-301 Strengthening of Structures using Glass Fibre Reinforced Plastic 302-306 Removal of sodium do-decylsulfate from waste water using adsoption on citrus 307-310 lemettioides Design of Adiabatic Adder Structures for Low Power VLSI & DSP Applications 311-314 Variable Frequency Drives .

perhaps because of being more governing in laminated structural elements. Previously. redistribution of stresses occurs within the remaining laminae of the laminate. The laminate will be termed as damaged with degraded properties. Unlike isotropic materials. extended fatigue life and various other superior material properties.jdvu. Failures are but a multitude of independent and interacting mechanisms. 4 1 A INTRODUCTION Over the last few decades.in unaffected. But Turvey [2] focused his study where flexural load dominated limiting his research to high modulus GFRP and CFRP and cross-ply symmetric configurations. thus making first ply and progressive failure of much concern to researchers. plates. However on a macroscopic level. The constitutive relations are changed followed by reduction in stiffness.sengupta@gmail. However when the same was subjected to transverse loading. maximum strain and Tsai-Hill happened to have different results. At first-ply failure.. fibre. They inferred that when the laminate was subjected to in-plane loading all the failure criteria were capable of predicting failure. The increasing use of the composites necessitates for the precise and viable methods of analysis. mechanical. it only indicates the initiation. For a composite construction it is thus crucial to locate this change. their advantages being their high strength and stiffness to weight ratio. The stiffness of the failed lamina is not taken into account and the rest are considered to remain Tsai and Wu [1] were the first to present that an operationally simple strength criterion cannot possibly explain the actual mechanisms of failures. S E I The failure mechanisms are best understood at micro level. most initial failure analyses was concerned with the in-plane loading cases. matrix. once the first-ply failure had occurred the load 1 of 446 Copyright @ Department of Electrical Engineering. I. literature review. subjected to in-plane and/or transverse loading. the composites bear a complex response to loadings which can be analysed now by FEM. This does not necessarily mean that the whole of lamina has undergone failure. also it reflects the five failure theories working behind. The initiation and propagation of failure until final fracture of the structure assesses the life of the structure. The first ply failure occurs when stresses in the weakest lamina exceed the allowable strength of the same changing the material properties. maximum strain.Progressive Failure Analysis of Laminated Composite Shells – A Review Jayashree Sengupta Post-Graduate Student: Department of Civil Engineering Jadavpur University Kolkata.com Dipankar Chakravorty Professor: Department of Civil Engineering Jadavpur University Kolkata. India jaish. They made use of strength tensors fulfilling the invariant requirements of coordinate transformation. Upon the application of loads. Abstract—Composite materials present striking potentials to be tailored for advanced engineering applications. patch loading and hydrostatic loading varying linearly. NIT Durgapur .e. the first order shear deformation theory and a tensor polynomial failure criterion (emphasis was laid on maximum stress. the failure analysis is more intricate. aerospace and marine. This paper addresses the various literatures that have been published so far associated with the progressive failure of the composite laminated shells. composite laminates are exceedingly used in various engineering sectors like construction. A ply-by-ply progressive analysis and the damage so done is analysed by the inclusion of different failure criteria which allows for the identification of the location of the failure. and interface-interphase. Tsai-Hill. interaction terms were treated as independent components and the difference in strengths owing to positive and negative stresses were accounted for making it way too improved than the Hill criterion wherein the interactions were not independent of each other. Tsai-Wu and Hoffman’s criteria) to predict failure at elemental Gauss points. Unnoticed internal failures may lead to fatal collapse. laminated composites.ac. LITERATURE REVIEW Reddy and Pandey [3] studied the first ply failure for the laminated composite plates. The remaining laminae continue to take up load till the ultimate strength is reached. Keywords—progressive failure. uniformly distributed load. i. India prof. he however chose three cases viz. Though his research was unrestricted on grounds of loading. the laminate undergoes stresses closely related to the properties of the constituent phases. The procedure that they depicted was an iterative one. shells. A composite material undergoes transition in multiple phases. Thin walled composite panels are one of the most utilized structural elements in construction. first-ply failure.the life prediction being an important issue. II.dipankar@gmail. expressed both the deflection surface and lateral pressure deflection in Navier double series form.com dchakravorty@civil. He considered Tsai-Hill failure criterion as it is in good correlation for the GFRP laminates.

Pal and Ray [22] carried out PFA under transverse static loading in linear elastic range distinctively for both antisymmetric and symmetric angle ply laminates. with geometric non-linearity. Prior to this.5% and displacement correction convergence of 1%. et al. Cheung et al. Chang and Lessard [8] studied the damage in laminated composites containing an open hole subjected to compressive loading. The procedure 2 of 446 Copyright @ Department of Electrical Engineering. The transverse shear stresses were obtained directly from the constitutive equations. shear-out and bearing) which were judged against experimental data. They extended their research on post first ply failure emphasizing on the progressive analysis [4]. The stiffness matrix was determined based on Green’s strain and 2nd Piola Kirchoff stresses. Sahid and Chang [16] developed a model for predicting the effects of matrix crack induced accumulated damage on the in-plane response of laminated composites under tensile and shear loads. wherein the in-plane response of the laminates from initial loading to final collapse was studied. [17] presented a theoretical and experimental study of damage progression and failure modes of composite laminates under three point bending. [23] reviewed the overall computational issues and requirements for performing PFAs using STAGS (Structural Analysis of General Shells) for solving non-linear quasi-static structural response problems including special details. Echaabi et al. Hoffman. The linear loads results varied by a maximum of 35% and for non-linear loads it was 50%. Kim and Hong [11] studied macroscopic failure models evaluating the stiffness changes employing shear lag factor and fiber bundle failure. Besides. Akhras and Li [25] proposed a spline finite strip method PFA of thick composite plates based on Cho’s higher order zigzag laminate theory and Lee’s failure criterion. fiber breakage and delamination. Engblom and Ochoa [6] carried out PFA till last ply failure. Tan and Perez [12] studied progressive failure of laminated composites with holes. The model was capable of assessing type and extent of damage all throughout. Chang and Chang [7] developed the progressive damage analysis of notched laminated composites subjected to tensile loading. damage progression followed maximum stress criterion. in the Von Karman sense. stiffness decrement and damage growth followed standard laminate analysis. The Lee criterion gave the best results. of pin-loaded laminated composite plates using the penalty finite element method. was used to construct the load displacement behaviour. Ultimate ply failure loads for the stiffened panels with cross-ply and angle-ply laminations in the shell was analyzed. Gummadi and Palazotto [19] performed PFA of composite cylindrical shells with large rotations based on Langrangian approach and the due changes in the constitutive relations were discussed. They extended their work [10] to nonlinear progressive analysis using the Generalized Laminate Plate Theory (GLPT) of Reddy applying a new stiffness reduction format. These Euler stresses were the key to failure determination. Eightnoded isoparametric quadratic elements with three-noded curved beam elements were modeled and checked against various failure theories. [15] presented a PFA of composite plates by the finite strip method based on higher order shear deformation theory and 4 1 A S E I Kim et al. Tolson and Zabaras [5] studied progressive failure in laminated composite plates implementing seven degrees of freedom (three displacements. failure criteria being Hashin’s. incorporated simplified large displacement/rotation (SLR) theory and compared the results with previously published available data. considered failure modes of matrix cracking. Reddy and Reddy [9] computed linear and non-linear first-ply failure loads of composite plates for different load cases and edge conditions. The progressive failure method used a nonlinear FEA using the modified Newton-Raphson iteration scheme to work out the state of stress in a composite plate. Thereafter the stresses were employed in five failure criteria (maximum stress. The normal stress distributions were calculated at each Gauss points. the shear correction factor was not required as for the first-order shear theory. two rotations of normal about the plate midplane. Hashin’s failure criteria was performed for damage evaluation in the laminates. Prusty [24] studied unstiffened and stiffened composite panels under transverse static loadings in the linear elastic range. subjected to compressive loading predicting the extent of damage at any level of loading. [20] presented their detailed study on progressive failure and ultimate collapse of laminated composites using Hashin and Tsai-Wu’s failure criteria. this difference was much large for thin laminates subjected to transverse loading and quite small for thick laminates subjected to in-plane loadings.Lee’s failure criterion. They made use of Newton-Raphson Method with a force and moment residual convergence of 0. but the transverse stresses were assessed from the three-dimensional equilibrium equations. The in-plane stresses were calculated from the constitutive equations. Tan [13] had previously developed progressive failure model of laminated composites subjected to in-plane loading considering the environmental impacts (thermal and hygroscopic stresses). Kam and Sher [14] studied progressive failure of centrally loaded laminated composite plates. Lee. A geometrically non-linear formulation based on finite deformation theory was used. NIT Durgapur . Results showed good assessment with the experimental results. Padhi et al. [18] carried out PFA to predict the failure strengths and failure modes (tension. Spottswood and Palazotto [21] determined the response together with material failure of a thin curved composite shell resisting transverse loading. Green’s strain was transformed to Almansi strain which was further transformed into material axis system and was used to determine the Euler stresses. The Ritz method. Knight Jr. and Tsai-Wu) and were checked against whether failure had occurred or not. Hashin. Analysis was carried out on a plate subjected to uniaxial tension and four-point bending. was reduced by 20% and the process went on till the difference between any two successive failure loads was less than 1%. and two warps of the normal).

FYC Compressive strength in X and Y direction respectively FS In-Plane shear strength Normal stresses in X and Y direction . Bogetti et al. lay-up configuration. Pietropaoli [36] too worked on progressive failure of composite structures using a constitutive model implemented in ANSYS. ≥1 (2) C. A program was build and executed in MATLAB. respectively Ultimate Shear strain Generally failure criteria can be either non-interactive (independent) or interactive (polynomial). [33]. ≥ 1. A. where the non-linear equations were derived using the tangential stiffness matrix approach. Besides material nonlinearities. Standard ply discount technique was used and the onset and progression of damage was observed and the results were validated against experimental results. the effect of stress interactions is ignored. Independent Failure Criteria 1) Maximum Stress Criteria: According to maximum stress theory. Hence it is best to check for failure through both independent and non-interactive criteria.involved an incremental load analysis through modified Newton-Raphson method and standard PFA method showing good agreement with the 3D finite element solutions. ≥ 1. but in this case. 30. The stress interactions are addressed by the polynomial failure criteria. Anyfantis and Tsouvalis [35] studied post-buckling progressive and final failure response of stiffened composite panels using ANSYS. Notations FX Overall Longitudinal Strength FY Overall Transverse Strength FXT . The PFA method included intralaminar failures that stimulated material degradation of the failed layers. The intralaminar. Ganesan and Zhang [28] had conducted a detailed investigation of the progressive failure of uniform thickness laminates subjected to uniaxial compression. validation was done by comparing the results with analysis in ABAQUS. respectively Shear stress in X-Y plane Strain along X and Y direction respectively . Singh et al. fiber and matrix failure modes in compression and tension were addressed using a combined framework of Hashin’s and Tai-Wu failure. Tapered laminated plates under the action of uniaxial compression were investigated by Ganesan and Liu [27] predicting different failure loads and the associated displacements. Ahmed and Sluys [34] designed a computational model presenting a mesoscopic failure model studying matrix cracking. the lamina should be checked using the interactive failure. tensile or compressive or shear mode. Philippidis and Antoniou [38] computed a PFA model for glass/epoxy composite giving an extensive comparison between numerically calculated stress–strain response up to failure and experimental data. and is simple to apply. the numerical model simulated geometrical nonlinearities. However. delamination was presented by a shell interface model. locations and modes. III. ≥1 (1) 2) Maximum Strain Criteria: According to maximum strain criteria failure occurs if any of the following is reached. Hashin’s and Besant’s criteria were checked for different failure mechanisms. Interactive Failure Criteria Failure can be predicted by the following tensorial form. [39] presented a reduced-order FE model suitable for PFA of composite structures under dynamic aeroelastic conditions based on a Thin-Walled Beam theory predicting both onset and propagation of damage. An independent criterion gives the mode of failure. Ply discounting method was employed as the strategy for material degradation. [37] studied the nonlinear response and progressive failure of composite laminates under tri-axial loading. A mesh independent matrix cracking was modeled with discontinuous solid-like shell element (DSLS). [29. If not so. 4 1 A S E I FAILURE CRITERIA B. Progressive damage analysis methodology for stress analysis of composite laminated shells using finite strip methods based on Mindlin’s plate-bending theory were addressed by Zahari and El-Zafrany [32]. Las and Zemcík [26] proposed a PFA model of unidirectional composite panels using Puck’s fiber and inter-fiber criteria demonstrating the results with examples of tensile tests of single-ply panels. NIT Durgapur ≥1 (4) . be it longitudinal or transverse. The laminate may indicate failure using a noninteractive theory. ≥ 1. 3 of 446 Copyright @ Department of Electrical Engineering. the failure initiates if at least one of the criteria is satisfied. Sandwich composite panels under quasi-static impact were investigated by Fan et al. this analyzed and displayed the failure envelope. 31] also studied the progressive failure of uniform thickness plates subjected to different loads. The influences of the tapered configuration. FYT Tensile strength in X and Y direction respectively FXC . Shear strain Ultimate Strain along X and Y direction . Earlier. It may so happen that the independent stresses do not initiate failure but their interactions may. ≥1 (3) which when expanded in two dimensional form gives. Cárdenas et al. ≥ 1. the failure mode is disregarded. and fiber orientation were also the concern of the study. delamination and the combined effect.

1.The failure indices for different theories are as follows. IV. CONCLUSION Various literatures on laminated composite structures and are studied in this paper. however their evaluation on progressive failure is still an area of interest. Hence failure analyses of composite shells need careful attention. It is found that such studies on beams and plates have appeared in quite a number of places although similar studies on composite shells are really scarce. If > 0. Partial Ply Failure Method In this approach. B. and the maximum loads that the structures can withstand before failure occurs is essential for assessing the performance of composite laminated plates and for developing reliable and safe design. Thus this method somehow underestimates the laminate strength. This implies that if the ply undergoes matrix failure. If the ply fails due to fibre failure. Thereafter. the material properties at that point were modified with accordance to the observed failure mode. These shells are ruled. may not be the case. [40]. using a property degradation technique. FAILURE INDICES Tsai Wu = FXT else Hoffmann = FXC . the longitudinal modulus retains its value but the transverse and shear modulus are set to zero. the failure mode is taken into account. Industrial shell forms like conoids and skewed hypar need attention as well. A S E I A. TABLE I. the strength and stiffness of the failed ply is totally reduced to zero. 48] and Kumari and Chakravorty [48. Total Ply Failure Method On reaching the failure.e. 49] also studied behavioral characteristics of laminated conoidal shells. doubly curved. NIT Durgapur . 4 of 446 Copyright @ Department of Electrical Engineering. 1. A typical PFA is demonstrated in Fig. mechanical loading i. if it is a matrix controlled failure or shear failure. the stiffness of the failed ply is reduced to zero. If any failure had occurred. the stresses at each Gauss point within individual lamina were evaluated and are verified with the failure criteria to check any possible failure. However. forces and moments. Similarly if > 0. Cylindrical and spherical shells enjoyed quite an importance. Researchers like Das and Chakravorty [43-46. it is no longer able to carry load in fibre direction. Failure Tsai Indices Hill(a) (a). the initiation and growth of the damages. It is visibly evident that prediction of the failure process. Seshu and Ramamurti [41] and Qatu and Leissa [42]. Flowchart of Progressive Failure Analysis Methodology V. it is necessary to determine the first ply failure of the laminate. Based on the inputs. a load increment is performed and with the reduced stiffness the process continues till the ultimate failure load occurs and at that stage convergence of stresses cannot be achieved numerically . aesthetically appealing and easy to cast and fabricate. Hypar shell has wide applications in engineering and was studied by a number of researchers like Kielb et al.(A) Total-Ply Failure Method and (B) Partial-Ply Failure Method. if not. and the stresses were recalculated at FPF load. The material properties of the lamina. = FYT else = FYC PROGRESSIVE FAILURE ANALYSIS To carry out PFA. These shell forms are to be studied for failure also. which. a check is performed to see whether the second ply fails at FPF load. 4 1 Fig. and layer orientation are read.

95. Fan. Bogetti. October 1992. pp. 371-393. 16. pp.Zhang. pp. “Linear and non – linear failure analysis of composite laminates with transverse shear”. 1996.N. pp. N. “Progressive failure analysis of thin-walled composite structures”. “Predictions of deflection and first-ply failure load of thin laminated composite plates via the finite element approach”. T. 361–376. Lessard. Vol.. “Post Buckling Progressive Failure Analysis of Composite Laminated Stiffened Panels”. [18] S. “Progressive failure and post- [7] F.”. June 2012. Jeffrey Staniszewski. Macbain. Composite Structures.N. S. 159–176 laminates subjected to in-plane compressive and shear loadings”. Rankin. Vol.25. Levon Minnetyan. Venice.S. Composite Structures Vol. Tsai and E.C. “A General Theory of Strength for Anisotropic Materials”. “Progressive Failure Analysis of Composite Structures composite plates by the finite strip method”. Shahid and F. “Vibration of twisted cantilever plates – a comparison of theoretical results”. 5. January. Charles C. Vol. Vol. International Journal of Non-Linear Mechanics Vol.38. L. Vol. [3] J. pp.”.926-981. Hong. C. laminated plates under uni-axial compression. pp. Kielb. Bruce P Burns. Journal of Composite Materials. Pal and C.53. A Kumar and NGR Iyengar. Nos 3-4. October 1993. S.N.”. Piergiovanni Marzocca. O. K. N. “Progressive failure of laminated composites with [33] X. “Finite element analysis of progressive failure in laminated composite plates”. 2005 pp. “Predicting the nonlinear response and progressive failure of composite laminatesunder tri-axial loading. A. “Theoritical and experimental investigation of failure and damage progression of graphiteepoxy composites in flexural bending test. pp. 6. El-Zafrany. “Progressive failure of symmetrically Y. Knight Jr. 1998. 2011. 2002.R. pp. 927-940. pp. pp. [27] Rajamohan Ganesan and Dai Ying Liu. 87-102. 25.1380. pp. S E [14] T. Applied Composite Materials. pp. “Progressive failure model for the analysis of 4 1 laminated composites based on finite element approach. Composite Structures Vol. 1078-1092 [32] R. 75-80. “Progressive failure analysis of composite cylindrical shells considering large rotations. Vol. pp. Y.6(7). 29. Issue 3-4. pp. Michael Spottswood and Anthony N. Vol. A.”. pp. Li “Progressive failure analysis of thick composite [23] Norman F. Reddy. “Progressive Failure Analysis of Laminated [4] A. 41. B. 1987. [29] SB Singh.V. Computers and Structures. Sun “Damage evolution of sandwich composite laminated stiffened plates using the finite strip method”. 79-102. Italy. Issue 10. Akhras and W.79. 1980. 219-236 [36] Elisa Pietropaoli. 2002 1505-1513 [1] S. Journal of Materials. 1991.C. International Journal of Solids Structures. Hawkins. K.J.N. pp 117131 of Aeronautical Society of India. [17] J. Moy and G. Reddy “A Post First-Ply Failure Analysis of Composite Laminates” AIAA Paper 87-0898. 19. A a hole under compressive loading”. pp. 1. Li. laminates under in-plane shear: I-Positive shear. Palazotto. Ramamurti. R. Zahari and A. 44. Chang. 33. Vol. 623–637.F.3. No. Chang and L. [16] I. 2007. Computers and Structures. K. pp. International Journal of Non-Linear Mechanics. 1995. 1985. A Kumar and NGR Iyengar. J. Reddy. 24. Padhi. 1987. “Progressive failure [42] M. No.W. June 2012. Jose. T.1998 pp. Journal of Composite Materials. “Progressive Damage of Unidirectional [6] O.”. Journal Of Reinforced Plastics And Composites. No.82. 1995. [38] Theodore P Philippidis and Alexandros E Antoniou . 629-649. Vol. [26] Vladislav Las and Robert Zemcík . Composite Science and Technology. 12.5. [25] G. 2008 pp. J. “An initial flexural failure analysis of symmetrically laminated cross-ply rectangular plates”. pp. Vol.S. Gangadhara Prusty. 1989. “Progressive failure analysis of composite [12] Tan. M. Qatu and A. Hugo Elizalde. 63–70. W. 3. Procedia Engineering. and Materials Conference. Zabaras. N. “A Computational Model For Prediction Of Progressive Damage In Laminated Composites”.W. Engblom “Analysis of Failure in Composites. pp. pp. 1995.”. Computer Methods in AppIied Mechanics and Engineering.. Kim. I [15] M. “An accumulative damage model for tensile and [37] Travis A. and John Tierney. 2-43. 833–849 Unstiffened and Stiffened Composite Panels”. “Progressive failure and ultimate collapse of laminated composite plates in bending. Vol. Ochoa and J. twisted cantilever plates”. Wang and Q.E. 15. pp. Cheung. pp. Vol.S.J. Kim. Vol.pp. Journal of Reinforced Plastics And Composites. 19. 40. Journal of Composite Materials. Reddy. Journal of Reinforced Plastics and Composites Vol. Analysis”. C. pp. Christopher PR shear failures of laminated composite plates. International Journal of Mehanical Science. 65-70.11 (2–3). Palazotto . 2008. 1. 11. ECCM15 . pp. Vol. Pham and M. 58-79. 143-59. Frank A. [11] Y. 2443–2459. “Vibration studies for laminated composite analysis of composite shell. Pandey “A first ply failure analysis of composite laminates”. Sher. pp. T. Vol. “Progressive failure analysis of structure using a progressive failure analysis methodology”. [10] Y. 2012. M. 277-291 [41] P. plates using the spline finite strip method”. Wu. pp. “Vibration of twisted composite plates”. Gummadi and A. F. Issue 3-4. S. Structural Dynamics. Pandey and J. “Damage tolerance of laminated composites containing an open hole and subjected to compressive loadings: Part I.” Journal of Reinforced Plastics and Composites Vol. pp. 2001. Vol. 1971.87. 2004. S. Tolson and N.34-43 S. Vol.W. J. 227-255 [30] SB Singh. Chang. Vol. 1987. 21. Vol. “Non-Linear Progressive Failure Analysis of Laminated Composite Plates”.” Composites Science and Technology. Trochu. pp. Vol. [35] Konstantinos N. Akhras and W. 25-44 buckling response of taperedcomposite plates under uni-axial compression”. 24-28 June 2012. Vol. Leissa. Brogan. Journal of Composite Materials. 433-50. Composites Part B: Engineering Vol.6(2). Composite Structures Vol. Journal of Composite Materials. Structural Engineering and Mechanics. International Journal of Solid Structures. No.29. 2009. Vol. S. Vol. Proceedings of the AIAA/ASME/ASCE/AHS/ASC 28th Structures. “Progressive failure of symmetric laminates under in-plane shear: II-Negative shear. 1991. pp. Chang and K. “A progressive damage model for laminated composites containing stress concentrations. Vol.10. Sluys. 530–535. 1987. Vol. Applied Composite Materials. Composite Structures. Perez.N. Chang. Pages 556-577. Composite Structure. Vol. Journal [21] S.Y Kam. Science and Engineering of Composite Materials. 5. Echaabi. No. 36.633-642. No. 1365 . “Progressive Failure Analysis of Laminated Composite REFERENCES Plates by Finite Element Method”. pp-375-398. 16. 2012.K.47(5). 788-797. 33.21. Ray. Issue 5. G.H. 124 1995. Shenoi. pp. 757–72. Issue 5. pp. C. [28] R. 657-668. 1992. 28. Ouellet. 21. 1998. [31] SB Singh. X.S.46.25.37. Frank Abdi. [20] G. Vol. Leissa and J. Tsouvalis. A Kumar and NGR Iyengar. [34] A. Journal of Composite Materials. 49-61 Using a Constitutive Material Model (USERMAT) Developed and Implemented in ANSYS”. 1991. Ganesan and D. 1043-1057 [13] Tan. Issue-19-20. 451-463. Journal of Reinforced Plastics and Composites. “Progressive failure of symmetric [5] [8] [9] Composite Panels”. Composite Hoppel Christopher PR.”. “Progressive failure analysis of pinloaded laminated composites using penalty finite element method.740-755. Seshu and V. Anyfantis and Nicholas G. May 1991.15th European Conference on Composite Materials. 53–62. Journal of Composite Materials. 547–563 [40] R. 5 of 446 Copyright @ Department of Electrical Engineering. Reddy and J. 1998. Hwang and J.” Structural Engineering and Mechanics. Vol. 42. Vol. September 1998. Reddy and A. [2] G. International Journal of Numerical Methods in Engineering. 2013. [24] B. Vol. Vol. Ahmed and L. pp. Oliver Probst. Kim and C. No. “STAGS computational procedure for progressive failure analysis of laminated composite structures.” Structural Engineering and Mechanics 1997.”. Seng “ Progressive failure model for composite laminates containing openings”.11. 834-855.B. K. Vol. [19] L.. pp. Vol. Seng . AIAA Journal.N.”. Dakshina Moorthy and J. [39] Diego Cardenas. Chao and R.L. Issue 10. H. “A progressive damage FEA model for glass/epoxy shell structures”. John W Gillespie Jr. J. No. “Progressive failure analysis of composite F. NIT Durgapur .[22] P.Turvey. 30.

2011. Vol. 29. 525.542. 1793-1819. Das and D. [45] H. pp. No. Journal of Reinforced Plastics and Composites. S. Journal of Reinforced Plastics and Composites. S. Journal of Engineering Mechanics. S. Kumari and D. 4 1 A S E I 6 of 446 Copyright @ Department of Electrical Engineering. 4. Oxford University Press. 5. “On the bending characteristics of damaged composite conoidal shells – a finite element approach”.[43] H. 27. pp. Vol. 2009. Chakravorty. 2010. No. Das and D. Vol. 2010. 10. 137. Das and D. Chakravorty. “Natural frequencies and mode shapes of composite conoids with complicated boundary conditions”. 660-668. [48] H. 21. 1397-1415. Vol. 3. Vol. 45. “Design aids and selection guidelines for composite conoidal shell roofs – A finite element application’.Chakravorty. 2011. “Finite element application in analysis and design of pointsupported composite conoidal shell roofs suggesting selection guidelines”. No. No. pp. University Press. Kumari and D. Advances in Vibration Engineering. Chakravorty. 2007 pp. 8. [47] S. The Journal of Strain Analysis for Engineering Design. Daniel and Ori Ishai. No. 165 – 177. No. [49] S. “Bending of delaminated composite conoidal shells under uniformly distributed load”. 45. S. 303 – 310. [50] Mechanics of Composite Materials Mukhopadhyay. 26. Chakravorty. NIT Durgapur . Journal of Reinforced Plastics and Composites. Das and D. pp. [46] H. Vol. “Bending analysis of stiffened composite conoidal shell roofs through finite element application”. pp. 13. 17. Vol. and Structures by Madhujit [51] Engineering Mechanics of Composite Materials by Isaac M. [44] H. Das and D. “Composite full conoidal shell roofs under free vibration”. ASCE.Chakravorty. 3287-3296. Chakravorty. S. No. 2008 pp. Journal of Composite Materials.

5. NIT Durgapur . [4]. India irshadak85@gmail. Relative crack location (β1=L1/L) varies from 0.Application of Neural Network for Identification of Cracks on Cantilever Composite Beam Irshad A Khan Adik Yadao Dayal R Parhi Mechanical Engg. Which subsequently used to the design of smart system based on neural network for prediction of crack depths and locations following inverse techniques. [1]. internal. et al.5. Results from numerical analysis are compared with experimental results having good agreement to the results predicted by the neural controller.) If the stress intensity factors for a given type of crack are known. [6]. Department National Institute of Technology Rourkela. The material used in this analysis is glass-epoxy laminated composite. India drkparho@nitrkl. Damage identification on a composite cantilever beam through vibration analysis using finite element analysis software package ANSYS is established by Ramanamurthy et al. When a structure suffers from damages. Discrete Wavelet Transform based method is presented for the identification of multiple cracks in polymeric laminate composite beam by Andrzej K. Since long efforts are on their way to find a realistic solution for crack detection in beam structures in this regard many approaches have so far being taken place.125 to 0. Damage Algorithm and Damage index method used to identify and locate the damage in the composite beam. The effects of cracks on the dynamic characteristics of a cantilever composite beam are studied by Kisa M. Continuous wavelet transform is used to identification of crack in beam like structures by analysing the natural frequency and mode shape of cracked cantilever beam by Rao et al. Krawczuk M. The mode shapes were estimated experimentally using laser Doppler vibrometry. a quadratic programming model is set up the two methodologies to damage assessments. its dynamic properties can change.. NUMERICAL ANALYSIS The numerical analysis is brought out for the cracked cantilever composite beam shown in fig1. 7 of 446 Copyright @ Department of Electrical Engineering. The neural controller is developed with relative natural frequencies and relative mode shapes difference as input parameters to calculate the deviation in the vibration parameters for the cracked dynamic structure. etc. Thickness of the Beam (H) = 6mm Relative crack depth (ψ1=a1/H) varies from 0. [2] proposed two models witch gives valuable information about the location and size of defects in the beams. Mode shapes.com Mechanical Engg.0833 to 0. The numerical analysis is performed by using commercially available software package ANSYS to catch the relation between the change in natural frequencies and mode shapes for the cracked and uncracked cantilever composite beam. The effects of the crack location and depth and the fibre volume fraction and orientation of the fibre on the natural frequencies and mode shapes of the beam are explored. India adik. The material of composite beam is graphite fibre-reinforced polyamide containing multiple transverse cracks.875.com Mechanical Engg.9375.mech@gmail. INTRODUCTION I Health monitoring and the analysis of damage in the form of crack in Beam like dynamic structures are important not only for leading safe operation but also retraining system performance. Ansys. Moreover. [3]. Relative crack depth (ψ2=a2/H) varies from 0. The presence of cracks a severe threat to the performance of structures and it affects the vibration signatures (Natural frequencies and mode shapes). Two Damage identification algorithms are established for assessment of damage using modal test data which are similar in concept to the subspace rotation algorithm or best feasible modal analysis method by Hu et al. The composite beam model is used to develop a genetic fuzzy system to detect and locate the presence of matrix cracks in the structure. neural network. Crack damage leads to reduction in stiffness also with an inherent reduction in natural frequency and increase in modal damping. Department National Institute of Technology Rourkela. I. Relative crack location (β2=L2/L) varies from 0. Width of the beam (W) = 50mm.in Abstract— In the current analysis numerical and neural network methods are adopted for detection of crack in a cantilever composite beam structure containing non propagating transverse cracks. to locate the mode shape of transverse vibration at different crack depth and crack location. [5]. This method makes it possible to construct beam finite 4 1 A S E Keywords— Crack.0.833 to 0. Natural frequencies. The valuation of the crack locations is based on the estimation of natural mode shapes of crack and uncrack beams. Length of the Beam (L) = 800mm.625 to 0. A composite matrix cracking model is implemented in a thinwalled hollow circular cantilever beam using an effective stiffness approach by Pawar et al. The cracked beams of the current research have the following dimensions.ac. The output from the neural controller is relative crack depth and relative crack location. [7]. Department National Institute of Technology Rourkela. II. elements with various types of cracks (double edge.

The results of the numerical analysis for the first three mode shapes for un-cracked and cracked beam.3b.3a. having cracks orientation ψ1=0. Each node has three degrees of freedom. Poisson’s ratio = νm = 0. Modulus of rigidity of matrix = Gm = 1.277 GPa.W a1 L1 L2 a2 H L Fig.67 GPa.35. The numerical analysis is accepted by using the finite element software ANSYS in the frequency domain and obtain natural frequencies.33 gr/cm3. Modulus of rigidity of fiber = Gf = 29. The layers stacking in ANSYS shown in fig2.5 are shown in the fig3. β2=0. and mode shapes. Relative crack location from fixed end (3st mode of vibration) Fig. Fig.2 Layers Stacking in ANSYS 8 of 446 Copyright @ Department of Electrical Engineering. Relative Amplitude vs. 8 node element having three degrees of freedom at each node: translations in the nodal x. 1 Geometry Cantilever beam with multiple cracks Properties of Glass-Epoxy composite material in analysis: Young’s modulus of fiber = Ef = 72. Density of matrix = ρm = 0. Relative Amplitude vs. The natural frequencies and mode shapes are significant parameters in designing a structure under dynamic loading conditions.333 and β1=0. Young’s modulus of matrix = Em = 3.3c.4 GPa. NIT Durgapur . making a total twenty four degrees of freedom per element. Relative crack location from fixed end (1st mode of vibration) 4 1 A S E I Fig. ψ2=0. A higher order 3-D.22.6 gr/cm3.166.25. y. Relative crack location from fixed end (2st mode of vibration) Fig. Density of fiber = ρf = 2.45 GPa. Relative Amplitude vs. Poisson’s ratio = νf = 0. Numerical modal analysis based on the finite element modeling is performed for studying the dynamic response of a dynamic structure. and z directions (Specified as SOLSH190 in ANSYS) was selected and used throughout the analysis.

numerical. the second hidden layer has 36 neurons. 4. The vibrations characteristics such as natural frequencies and mode shape of the beams correspond to 1st. 2. 4. and the eighth hidden layer has 8 neurons. Relative third mode shape difference = “rtmd. 5 Multilayers feed forward back propagation neural model for damage detection IV. Function generator. Power Distribution. Data acquisition (Accelerometer). the third hidden layer has 50 neurons. The chosen numbers of layers are found empirically to facilitate training. Cracked Cantilever Composite beam 9 of 446 Copyright @ Department of Electrical Engineering. 8. EXPERIMENTAL INVESTIGATION To validate the numerical analysis result. which represents relative crack locations and relative crack depths. the sixth hidden layer has 150 neurons.III. 6. These numbers of hidden neurons are also found empirically. experimental analysis shown in table1. Relative third natural frequency = “rtnf”. The output layer has four neurons. Power amplifier. Relative first crack location = “rcl1” Relative second crack location = “rcl2” Relative first crack depth = “rcd1” Relative second crack depth = “rcd2” Fig. the fourth hidden layer has 150 neurons. 7. Relative second mode shape difference = “rsmd”. the fifth hidden layer has 300 neurons. Neural controller mechanism for crack detection The neural network used is a ten-layered perceptron. an experiment on composite beam has been performed shown in fig 6. The input layer has six neurons. 5. These results for first three modes are plotted in fig7. Corresponding numerical results for the cracked and un-cracked beam are also presented in the same graph for comparison. Vibration analyser. NIT Durgapur . 2nd and 3rd mode of vibration have been recorded by placing the accelerometer along the length of the beams and displayed on the vibration indicator. Vibration exciter. Relative second natural frequency = “rsnf”. A composite beam was clamped at a vibrating table. The term used for the outputs are as follows. Fig 5 depicts the neural network with its input and output signals.6 Schematic block diagram of experimental set-up 1. 4 1 A S E Fig. the seventh hidden layer has 50 neurons. Fig. The first hidden layer has 12 neurons. The term used for the inputs are as follows. The comparison of results between neural controller. 2nd and 3rd mode of vibration by using an exciter and a function generator.4 Neural model I A. The experimental results are in close justification with neural analysis results. NEURAL NETWORK ANALYSIS A back-propagation neural controller has been developed for detection of the relative crack locations and relative crack depth having six input parameters and two output parameters as shown in fig. Vibration indicator embedded with software (Pulse Labshop). three for first three relative natural frequencies and other three for first three average relative mode shape difference. During the experiment the cracked and un-cracked beams have been vibrated at their 1st. Relative first mode shape difference = “rfmd”. 3. Relative first natural frequency = “rfnf”.

The present method can be engaged as a health diagnostic tool for vibrating faulty structures.7c. Katunin. M Ostachowicz. pp. M. Z. pp.” International Journal of Engineering Scie and Techno. and K.” Journal of Intelli Material Sys and Struct. 3111-3126. X. The Numerical analysis results are well agreed with neural analysis results. pp. G. 2(162). Rao. Chandrasekaran. Relative crack location from fixed end (1st mode of vibration) VI. D. Relative Amplitude vs. pp. CONCLUSION The conclusions derived from the various analyses as mentioned above are depicted below. 2010. 5865-5888. Z. “Crack Identification on a beam by Vibration Measurement and Wavelet Analysis. Pawar and R.V. Ramanamurthy.X. 183(1). vol. M. Fukunaga. 38(18). Zhang. 2012.” International Journal of solids and structures.H Yao. 3. 2(5). 395 -409. 5. 41-52. Raju. “Modelling and vibration analysis of a cantilever composite beam with a transverse open crack.V. pp. K. Relative Amplitude vs.Wu. 4. vol.Srinivasa. pp. 1. vol. The investigation of vibration signatures of the cracked and un-cracked composite beam shows a variation of mode shapes and natural frequencies. “Free vibration analysis of a cantilever composite beam with multiple cracks.” International Journal of Engineering Science and Technology vol. 2. I [4] [5] [6] [7] Fig. Proposed health monitoring technique can be used for composite as well as isotropic material. 16(5). 2005. Krawczuk. They have good judgment.V.S. H. 3. The numerical analysis and neural analysis results are compared with the experimental results. Kisa. N. Relative crack location from fixed end (3rd mode of vibration) 10 of 446 Copyright @ Department of Electrical Engineering. “Identification of multiple cracks in composite beams using discrete wavelet transforms. 2010. Ganguli. REFERENCES [1] [2] [3] 4 1 A S E Fig. 2001. H. M. vol. pp. “Matrix Crack Detection in Thinwalled Composite Beam using Genetic Fuzzy System. Wang.7a. NIT Durgapur . Relative crack location from fixed end (2nd mode of vibration) A.V. and W.2005.7b. vol. Relative Amplitude vs. vol. P. E. 2004. 6989.” Scientific Problems of Machines Operation and Maintenance. Rao. Mallikarjuna.907-912. Fig.” Composites Science and Technology. 64(9). “Vibration analysis on a composite beam to identify damage and damage Severity using finite element method. Hu.” Journal of Sound and Vibration. “Damage Assessment of structures using modal test data. 1391-1402.

71 0.329 0.23 0.0036 0.162 0.73 0.164 0.620 0.412 0.49 0.373 0.26 rcd1 0.624 0.9981 0.9981 0. 2nd crack location“rcl2” 0.0025 0.21 0.47 0. 2nd crack depth “rcd2”.48 0.623 rcl2 0.0017 0.621 0.47 0.9958 0.9989 0.0069 0.009 0.0.414 0.374 0.45 0.412 0.242 0.623 0.370 0.22 0.0267 0.9987 0.372 0.9986 0.373 0.162 0.163 0.0732 0.19 0.626 0.176 0.46 0.413 0.48 0.164 0.24 0.121 0.9973 0.412 0.9937 0.0346 0.69 0.9988 0.9979 0.163 0.19 0.124 0.9975 0.243 0.0132 0.321 0.0017 0.624 0.163 0.0021 0.18 0.0082 0.0019 0.331 0. Comparison of results between Neural controller.0752 0.9985 0.23 0.0079 0.9989 0.9993 0.46 0.371 0.26 0.371 0.22 0.334 0.370 0.22 0.373 0.72 rcl2 Experimental relative 1st crack depth “rcd1” 1st crack location “rcl1” 2nd crack depth “rcd2”.731 0.0048 0.0247 0.0152 0.622 0.20 0.0046 0.9944 0.621 0.163 0.372 0.413 0.26 0. NIT Durgapur .23 0.22 0.0065 0.21 rcl1 0.9980 0.21 0.9974 0.0126 0.426 0.370 0.123 0. numerical and experimental analysis I S E A 4 1 11 of 446 Copyright @ Department of Electrical Engineering.24 0. natural natural natural mode mode mode nd nd location“rcl2”” 2 crack 2 crack location“rcl2” frequency frequency frequency shape shape shape “fnf” “snf” “tnf” difference difference difference “fmd” “smd” “tmd” rcd1 rcl1 rcd2 rcl2 rcd1 rcl1 rcd2 rcl2 Table1.0162 0.242 0.9881 0.46 0.233 0.0026 0.22 0.620 0.9989 0.0012 0.9857 0.623 0.9871 0.410 rcd2 0.9978 0.625 0.74 0.9997 0.48 0.21 0.22 0.329 0.21 0.415 Neural Controller Numerical relative Average Average Average st 1st crack depth “rcd1” Relative Relative Relative Relative Relative Relative relative1 crack depth “rcd1” 1st crack location “rcl1” 1st crack location “rcl1” first second third first second third nd 2 crack depth “rcd2”.511 0.9889 0.327 0.353 0.44 0.

the resonance is reached prior by considering the moving multi-masses system than by moving multi-forces system. Stanisic et al.Dynamic response of a simply supported beam with traversing mass Shakti. India shaktipjena@gmail. [2] developed a theory to calculate the response of a plate to a multi-masses moving structure by Fourier transformation technique. Siddiqui et al.computational technique has been developed to evaluate the dynamic response of a simply supported beam subjected to a traversing mass with different boundary conditions. [10] have investigated the response of a Timoshenko beam under moving mass and moving sprung structures. Stanisic and Hardin [3] have examined to find out the response of a simple supported beam subjected to random no of moving masses by Fourier technique in the existence of external load. Azam et al. Hence the analysis of structures subjected to traversing mass is of important problem. Sniady [13] has studied the dynamic response of a simple supported Timoshenko beam acted upon by a force traversing with constant speed. The analysis shows that for the same natural frequency of the plate. Ichikawa et al.The dynamic analysis of structures under moving loads has been an interesting subject of various researches over the last decades. The effects of speed of the moving mass on the dynamic response of the beam have been investigated and the critical influential speed (CIS) of the beam also found out. In the present paper. the dynamic deflection of a simple supported beam with moving mass has been determined and the consequences of speed on the beam’s response also found out. [11] have developed a scale-model experiment to analyze the response of a mass traversing on a Euler-Bernoulli beam by the application of theory of structural models. [15] have analyzed the necessity of Coriolis acceleration associated with the moving mass during the motion along the vibrating beam and the response of Timoshenko beam along with moving mass by introducing the coefficient of influential speed.In the present research. linear. distributed parameter systems using complex Eigen function expansions. Olsson [6] has investigated the dynamic response of a simply supported beam with a continuous traversing load at constant velocity. Parhi and Behera [7] have investigated the dynamic deflection of a cracked beam with moving mass by RungeKutta technique and the energy method. time-frequency and numerical methods.Parhi Mechanical Engineering Department National Institute of Technology Rourkela.Traversing mass. NIT Durgapur . RungeKutta Technique Introduction : A traversing mass produces larger beam deflections and stresses than that of the same mass applied statically.com Abstract. A theoretical. [9] have analyzed the dynamic characteristics of a flexible cantilever beam with moving masssprung system by using perturbation. India dayalparhi@yahoo. Lee and Renshaw [12] have investigated a new solution method to solve the moving mass structures problem for non conservative. Akin and Mofid [5] have developed an analytical-numerical method to compute the dynamic response of beams subjected to moving mass with arbitrary boundary states by variable separable technique and compared the results to finite element analysis (FEA). Keywords. Problem Formulation: 12 of 446 Copyright @ Department of Electrical Engineering. Odisha. Bilello et al. Esmailzadeh and Ghorashi [14] have studied the reaction of a Timoshenko beam subjected to a constant partly disseminated travelling mass. The governing equation of motion of the structures has been converted into a series of coupled ordinary differential equation and solved by fourth order Runge-Kutta technique. dynamic response. The analysis has been carried out with constant mass magnitude at different velocities.com Mechanical Engineering Department National Institute of Technology Rourkela. These deflections and stresses are the function of both velocity and time of the moving mass. [8] have investigated the response of a multi span Euler-Bernoulli beam carrying moving mass by using Eigen function technique and direct integration method combinedlly.Its mainly due to improvement of new high-speed vehicles and general drive to improve the active communications. I 4 1 A S E Jeffcott [1] has first developed a technique to get the inexact solution for the problem of vibration with the action of moving and variable loads. Stansic and Euler [4] have investigated an article to describe the dynamic response of the structures with moving masses using operational calculus method.Jena Dayal R.P. Odisha. Dehestani et al. CIS.

02) m. It is assumed that during the course of motion there is no separation between the moving mass and the beam . n ( x) Eigen By using the properties of Dirac Delta function and principle of orthogonality of ϕp(x) M Sp (t ) Vp x m 2 y t M [ g V 2 2 2 y x 2 2V Where δ= Dirac delta function L 2 y 2 y ] ( x ) t x t 2 f ( x) ( x )dx 0 iv EI n ( x )Tn (t ) m n ( x )Tn .35×0. m=Beam mass per unit length. the eqn (1) can be rewritten as n ( x )Tn (t ) (2) n 1 S E Tn (t ) Function of of time or amplitude function to be calculated. n= Number of modes of vibration. (5) ϕp(x) and integrating over Entire beam length 2 n // ( x)Tn (t ) 2V n / ( x)Tn.t (t ) 0 n 1 q 1 q 1 q ( )Tq. V (t) =Velocity of the moving mass with respect to time. (7) and (5) in eqn.t)=Deflection of the beam at point of consideration with respect to time. Beam size = (8×0.H. y ( x. tt n q 1 q 1 (1) 0 L . tt (t ) Vn 2V q 1 q / ( )Tq.t (t ) M [ g V L n 1 n 1 dx o n ( x)Tn. tt (t )] ( x ) n 1 ( x)S (t ) n (5) n n 1 Multiplying both sides of eqn.S of equation (1) can be written in a series form : M [ g V 2 (9) A Where (8) Substituting the values of eqn. the principal equation of motion of the beam with moving mass using Euler Bernoulli’s is given by 4 n 1 Theoretical Study: 4 y functions of the beam (without –M) with the same boundary conditions. They may be again written as in the form of I iv ( x) 4 ( x) 0 n 2 n EI m (3) 2 y 2 y 2 y 2V ] ( x ) 2 tx t 2 x ( x) S n n (4) n 1 Putting the values of eqn (2) in eqn (4). β=Distance travelled by the traversing mass from the fixed end.To get the values of Tn (t ) 0 4 n p n p Now applying equation (3) in equation (9).So the transverse of the moving mass with respect to the beam is ignored here. >0 Considering the solution of eqn in series form and by variable separation method.t n 1 ( x)T n M 4 EI n Tn (t ) mTn. n=3 Moving mass= 480 kg. Numerical Study: For the numerical analysis the dynamic analysis of the simply supported beam has been carried out and the CIS value of the corresponding beam with moving mass has been found out. tt (t ) n 1 n 1 M 2 n ( x ) V [ g V q1 q ( )Tq (t ) n n 1 2V ( )T (t ) ( ) T ( t )] ( ) q q. Nomenclature: L=Beam Length.Mild Steel.tt (t )]n ( ) q 1 Now the above equation may be reduced to : 4 1 M EI n 4Tn (t ) mTn. n. t (t ) [ g V 2 n // ( )Tn (t ) 2V n 1 n 1 p 0 Time‘t’ in Sec (Hinged-Hinged Beam) 13 of 446 Copyright @ Department of Electrical Engineering. 0 0 L (7) ( )T L Considering constant moment of inertia and neglecting the damping effect. M=Mass of the moving object. <0 . (1). f ( x) ( x ) dx f ( ) . tt (t )]n ( ) 0 q 1 Beam Type. y(x. Tn (t ) Function of time. t ) n . tt (t )] p ( ) n V p n ( x) p ( x)dx 0. I=Constant moment of inertia. t (t ) 2 [ g V q // ( )Tq (t ) q 1 (11) q ( )Tq. The R. NIT Durgapur . arranging the eqn (4) we can write: M [ g V 2 // n ( x)Tn (t ) 2V n 1 n / ( x)Tn. tt (t )] p ( x) ( x ) n 1 L S (t ) ( x) n n 1 n (6) Diflection at the midspan(mm) (10) The equation (11) must satisfy for every values of x and it is solved by 4th order Runge-Kutta Method.tt (t ) Vn 2 // / n ( x ) [ g V q ( )Tq (t ) 2V q ( )Tq .Considering a simply supported beam subjected to a moving mass ‘M’ with velocity ‘V’.t q q . EI n / ( )Tn.

R Parhi and A.vol. Stanisic. 14 of 446 Copyright @ Department of Electrical Engineering. Lafayette.115(1). It has been observed that if we are increasing the speed of the moving mass beyond the CIS value. The CIS value of the corresponding beam also calculated from the beam deflection. pp. F .799-804.199(4).67. W.75. Journal of Sound and Vibration. Dynamic response of Timoshenko beam under moving mass . 15. Journal of Sound and Vibration. M.pp. M.M Stanisic and Hardin. John E.Vol. 14. Journal of Sound and Vibration.pp.Stanisic.37-53.Akin and Massood Mofid. Thomas Telford Ltd .M. 6. 1991. 9. Vol. Journal of sound and vibration. Lou. the beam deflection will start decreasing.Y. 16.1023-1055.Vol. D. M.203(4).2004. Journal of Structural Engineering . Vol. Jeffcott.M. Ichikawa.pp.187(2). Journal of Structural Engg. K. 2008.pp.pp. S. Vibration of solids and structures (1999).A. 11. Mofid and R. Acta Mechanica.timefrequency analysis technique.Heppler .pp.229(5).pp. 2013. Bergman and Daniel Kuchma .77-87. Dehestani. Philosophical magazine series7. Vibration analysis of a continuous beam subjected to a moving mass. pp.115-123. Eftekhar Azam . pp. C. Experimental Investigation of a Small-Scale BridgeModel under a Moving Mass .66-97.C. M.pp. Scientia Iranica .H . M. 145(2). 5. Md.1889.299-307 7. Dynamic Response of a Timoshenko Beam to a Moving Force . Lawrence A.130(5).pp. Investigation of critical inﬂuential speed for moving mass problems On beams.211. 4.55(3). Baltimore. Vafai . Renshaw . Third Edition. 4 1 A S E Conclusion: The dynamic response of simply supported beam with various velocities at different boundary conditions has been investigated theoretically and computationally.J.pp3885-3895.1997. 615-628.R. Vibration analysis of a Timoshenko beam subjected to a Travelling mass. Vol.2009. Dynamics of a flexible beam carrying a moving mass using perturbation . The dynamic response of the simply supported structures of hinged-hinged & fixed-fixed type with various velocities have been examined and the deflection at the mid span of the beam also calculated. Vol.33.pp.823-827.1985. M.CBS Publishers .Behera .1996.. Vol.Journal of Applied Mechanics. Vol. Prague 17. 1929.176-185. 3.Vol.On the dynamic behavior of the structures carrying moving masses.Vol.Franklin Institute. On The Fundamental Moving Load Problem. J. C. Afghani Khoraskani . Siddiqui. Journal of Applied Mechanics. 12. 1968. Journal of Applied Mathematical Modelling. Journal of mechanical engineering science.Ingenieur-Archiv of Applied Mechanics. M. I References: 1. On the vibration of beams under the action of moving loads. 2. It has been observed that for the hinged-hinged type beam 115m/s and for the fixed-fixed type simply supported beam 215m/s is the critical influential speed of the beam. Time‘t’ in Sec(Fixed-Fixed Beam) Results and Discussion: The CIS may be analysed as the velocity of the traversing mass at which the structure produces maximum deflection with time variation. Itardin and Y.8 (48). Moﬁd and A.50–56. 13. New Delhi. 2000.24503(1-4).Olsson. M. On the response of the plate to a multi-masses moving system. Vol. The deflection of the beam mainly depends upon the velocities of the moving object. 8. 20 (1). 2000. Ind. Theory of Vibration with Application (2002) Third Edition. Golnaraghi and G.Vol.Fryba.1969. A.Thomson.493-506. Mastuda .5. NIT Durgapur . pp. E.Vol. S.T. Solution of the Moving Mass Problem Using Complex Eigen function Expansions . The consequence of critical speed and the inertia pertaining to the traversing mass has also been studied.120-132. The displacement of the traversing mass and the structure under the location of the moving object are calculated in terms of time‘t’.Numerical solution for the response of beams with moving mass.2000.Miyakawa and A. Lee and A.Deflection at midspan (mm) Vol.numerical . Pawel Sniady .K. 10. Ghorashi . M. Y..Cristiano Bilello. L. J. Dynamic deflection of a cracked beam with moving masses.On the response of beams to an atbitry number of moving masses. Esmailzadeh and M .Q.

co. 1. temperature cycle. characterization of heat affected zone (HAZ). 2nd order respectively……. is in decreasing mode towards the longitudinal direction of the welded plate. Density (kg/m3) Welding velocity (m/s) Temperature ε Perturbation parameter x Space variable β Thermal expansion coefficient (/°C) θ Non-dimensional temperature distribution term X Non-dimensional length L Length of the weld plate 4 1 0th order. 1st order. For two different materials. The analysis reveals that the non-dimensional parameter of temperature derived from analytical model has shown good agreement with the established model of moving point heat source and this entails that temperature cycle of weld bead due to heat loss by conduction. Quantification thermal aspect shows immense importance as the source of all the changes such as chemical–metallurgical phenomena in liquid solid interface of weld pool. NIT Durgapur . liquid-solid phase transformation and variation of relevant mechanical properties. All these properties are integrated to improve the weld structures in structural applications. the temperature distribution phenomena has been observed. are solely dependent of temperature gradients induced due to appearance of successive thermal cycles [1.com Department of Mechanical Engineering National Institute of Technology Karnataka. development of residual stress and distortion in weldments. S E K Particular location from fusion boundary (m) Subscripts Keywords— Perturbation technique. Professor Department of Mechanical Engineering National Institute of Technology Karnataka. 2……. Postgraduate Research Scholar Narendranath S. India Email-id: snnath88@yahoo. The thermal analysis of fusion welding covers numerous aspects such as the physics and behavior of arc with the formation of weld pool. A Asymptotic function T Thickness of base metal (m) i I Thermal conductivity t 0. stainless steel of grade AISI 316 and low carbon steel of grade AISI 1005. INTRODUCTION Fusion welding of metals is a process which comprises heating and cooling cycle. Nomenclature— y initial o Final p Peak temperature m Melting point I. Surathkal Mangalore. metallurgical properties of joints. The analysis has introduced non-dimensional variable in conjunction with variable thermal conductivity. Most prominent analytical model to predict the heat transfer characteristics is Rosenthal’s model [3] and still it is one of the basic simple model for Specific Heat (J/kg-K) Net heat input (J/m) [1] 15 of 446 Copyright @ Department of Electrical Engineering. non-dimensional parameter. moving point heat source.in Abstract—In this paper a steady state non-linear analytical thermal analysis has been carried out to determine the temperature distribution due to heat loss from weld bead formed in a butt weld joint.Mathematical modelling of steady state temperature distribution due to heat loss from weld bead of a butt joint Jaideep Dutta. The perturbation method has been incorporated to judge the nature of temperature distribution from the weld pool towards the longitudinal direction of the rectangular plate.. India Email-id: duttajd1212@gmail. 2]. Surathkal Mangalore.

[4] first proposed the heat source as Gaussian distribution on the surface of the welded plate. Elsen Van M. (e) Other modes of heat transfer (convection and radiation) have been neglected and (f) No phase change. NIT Durgapur . BASIC IDEA OF PERTURBATION METHOD (1 + εθ) Perturbation method. Based on above assumptions. and Eagar [5]. Though numerical methods are very popular for solving mathematical problems but they consist of error. MATHEMATICAL FORMULATION K(T) = K [1 + β(T − T )] (3) The boundary conditions are: T(x = 0) = T (4) T(x = L) = T (5) To identify an appropriate perturbation parameter the following nondimensional variables are used: T−T x θ= . Perturbation method can be an alternative approach for solving equations with comparatively higher accuracy. Some recent notable works done by researchers such as Hou Z. Osman Talaat and Boucheffa Abderrahmane [12] proposed an analytical model based on integral transform and finite cosine Fourier integral transform to compute 3-D temperature distribution in a solid by moving rectangular with surface cooling. Dimensional parameters were first introduced by Tsai N. the heat transfer equation on the domain (Fig. and Komanduri R. In order to implement the perturbation method the basic assumptions of this modelling are: (a) 1-D heat conduction. gives the formulation of nth order formulation. the analysis has been illustrated the temperature cycle as the distance increasing towards the rear end of the weld plate. [9] presented the analytical and numerical solution for modelling of localized moving heat sources in a semi-infinite medium and illustrated its application to laser processing materials. whereas this method is acceptable in this situations. the boundary equations (4) and (5) becomes: [2] 16 of 446 Copyright @ Department of Electrical Engineering. three dimensional temperature distribution due to a moving laser beam. Then it requires a small parameter with very small magnitude which appears as dimensionless form in the equation.X = T −T L Substituting the nondimensional variables in (2) {1 + β(T − T )θ} =0 (6) Thus the perturbation parameter can be expressed as: ε = β(T − T ) (7) Thus (6) becomes. [7] analytically derived general solutions of moving point heat source in both transient and stationary state. It is very difficult to solve higher order non-linear differential equations analytically =0 (8) Now as (8) is nonlinear. [6] assuming pseudo-Gaussian heat source. Pavelic et al. liquid weld pool. is in the range of 0 ≤ ε ≤ 1. B. [10] explored the exact solution of temperature distribution in a rectangular plated heated by a moving heat source obtained by Green’s function method. et al. the perturbation technique (asymptotic) has been adopted as it allows to evaluate the approximate solutions which can’t be determined by traditional analytical method. also known as asymptotic method is a simplification of complex mathematical problems [15]. III. Then the non-dimensionalized equation needs to be expanded in an asymptotic nature with the form [16]: ϑ = ϑ + εϑ + ε ϑ + ⋯ (1) Then the assumed equation is substituted into governing equations and equating the terms of identical powers of ε. S. Motivated by these facts the present analysis has been carried out in order to estimate the nondimensional temperature distribution from weld bead of square butt joint to the longitudinal direction by non-linear steady state analysis. The first step of implementation of this theory is to nondimensionalize the governing equation. Araya Guillermo and Gutierrez Gustavo [8] obtained analytical solution for a transient. Levin Pavel [11] developed a general solution of three dimensional quasi-steady state problem of moving point heat source on a semi-infinite solid. Winczek Jerzy [13] suggested an analytical model of computation of transient temperature field in a half infinite body caused by moving volumetric heat source with changeable direction of motion. After this appreciable work has been done by Goldak et al. 1) is [17]: K(T) =0 (2) 4 1 A S E I II. (d) No energy generation. (b) Temperature dependent thermal conductivity. By assuming variable thermal conductivity. For different temperatures starting from melting point of the metal i.illustration. Kukla-kidawa J.e. (c) Steady state heat transfer. This parameter usually denoted as ε. Parkitny Ryszard and Winczek Jerzy [14] described analytically the solution of temporary temperature field in half infinite body caused by assuming moving tilted volumetric heat source with Gaussian power density distribution with respect to depth.

d θ =0 dX Thus. NIT Durgapur .c) Applying boundary condition (9) into (1). into (1): θ (1) + εθ (1) + ε θ (1) + ⋯ = 0 (14) Again equating identical powers of ε: θ (1) = 0. Schematic diagram of square butt joint of two rectangular plates indicating the domain of heat conduction θ(x = 0) = 1 (9) θ(x = L) = 0 (10) Now using (1) asymptotic equation as mentioned. θ = (2X − X − 1) (17) Finally substituting θ . θ = 1 − X (15) (b) 1st order solution: Substituting θ as expressed in (15). θ (0) = 0 and θ (0) = 0 Proceeding same way. (11) becomes: +ε ε θ +ε + 2ε + εθ + ε[ +ε θ =0 ] + (12) Now equating the identical powers of ε. θ (0) + εθ (0) + ε θ (0) + ⋯ = 1 (13) Equating identical powers of ε: θ (0) = 1.c and applying boundary conditions (0) = 0 and (1) = 0 yields: d θ = 2 − 3X dX Thus. yields: ε : ε : = 0 +θ (12. into 12. and into (1) results: θ = (1 − X) + ε (1 − X) + ε (2X − X − 1) + ⋯ (18) [3] 17 of 446 Copyright @ Department of Electrical Engineering.a.a) +[ ] =0 (12. d θ = −1 dX Thus. 1. (8) results: {1 + ε(θ + εθ + ε θ + ⋯ )} (θ + εθ + S E I ε θ +⋯) = 0 (11) After differentiating and expanding.b) ε : +θ + 2ε +θ =0 (12. θ (1) = 0 and θ (1) = 0 The solution has been derived as follows: (a) 0th order solution: Applying boundary conditions θ (0) = 1 and θ (1) = 0 in 12. θ = (1 − X) (16) (c) 2nd order solution: Substituting θ and θ from (15) and (16) into 12. applying boundary condition (10).b and (0) = 0 and applying boundary conditions (1) = 0 gives.4 1 A Fig.

1) as derived in (18).90 Figure 2 indicates that the non-dimensional temperature distribution for the material AISI 316 decreases with the increment of non-dimensional length from weld pool towards the longitudinal direction (refer fig.00 0.30 0.05 0. Variation of temperature cycle based on perturbation method for AISI 316 workpiece Density (kg/m3) 0.00 0. The thermo-physical properties have been tabulated in table I. A AISI 316 Ni-Cr basedStainlesssteel 1. the temperature distribution peak in welded zones.20 0. 3.15 0.00 0. Non-dimensional temperature distribution along the longitudinal direction based on perturbation method for AISI 316 workpiece AISI 1005 Lowcarbon steel 1.15 0. IV. C) Temperature (D According to Masubuchi [20].30 0.45 0.15 0.05 0.30 0.45 2 4 1 1.20 Non-dimensional temperature distribution () Material No ndi m e di nsi st on rib a ut l te io m n ( pe ra tu re ( = eg.75 Non-dimensional lengthalonglongitudinal directionfromweldweldpool (X) Fig.90 0. Figure 3 also denotes the three-dimensional temperature distribution along with nondimensional length and it has depicted the fall of temperature due to heat loss after the formation of weld bead.90 Non-dimensional length fromthe weld pool towards longitudinal direction (X) Fig. NIT Durgapur .00 0. 20] Stainless steel (AISI 316) 1510 Melting point (°C) Thermal expansion coefficient (/°C) 0 Low carbon steel (AISI 1050 ) 1425 15. × × ×× + ( 1200 1000 (19) ) RESULTS AND DISCUSSION 800 600 400 10 200 TABLE I. EMPIRICAL MODEL FOR PEAK TEMPERATURE DISTRIBUTION 1600 1400 ) .7 × 10 8000 7872 Specific heat (J/kg-K) 500 481 Non-dimensional temperature distribution () S E 0.60 0.45 0.75 8 Nondime2 long nsio 4 itudi nal L 6 nal d engt irect h alo 8 ion f ng th rom e 10 weld pool (X) I 0.30 0. THERMOPHYSICAL PROPERTIES OF MATERIALS [18. Non-dimensional temperature distribution along the longitudinal direction based on perturbation method for AISI 1005 workpiece [4] 18 of 446 Copyright @ Department of Electrical Engineering.IV. specifically in the Heat Affected Zone (HAZ) and its vicinity can be determined by using (19) 1.15 0.60 0. 4.45 0.60 6 4 Fig.75 0.60 0.9 × 10 11. 2.75 0.

Peak temperature variation for AISI 1005 base metal based on empirical method Fig. but the temperature field in the neighborhood of weld pool can be obtained by specifying the heat input. C) 1200 AISI 316 (Ni-Cr based stainless steel) 6 Net heat input (Hnet) = 11. Also in both cases the temperature loss due to convection and radiation has been neglected which is very severe in liquid-solid interface region of fusion zone. Peak temperature variation for AISI 316 base metal based on empirical method In comparison with fig. the heat input with voltage V = 40volt and I = 250amps is 11. 6. 6. The other thermo-physical properties has been extracted from table I and has been considered as constant.07 × 10 J/m (time of single pass welding = 65 sec) for AISI 316 and voltage V = 40volt and I = 210amps. 7. C) Temperature (D 1400 1200 1000 800 600 400 10 AISI 1005 (Low carbon steel) 6 Net heat input (Hnet)= 9. the difference between steady state and transient phenomena has been well judged as in his paper the temperature distribution is in transient manner thus the change in thermal properties near the weld pool is notable than in rear end of the weldment along longitudinal direction whereas in present discussion the thermal properties varies linearly with linear variable thermal conductivity as mentioned in (2). Though the perturbation method is showing partial agreement (decreasing nature of nondimensional temperature) with empirical model. 6. 5. In comparison with the studies of Araya Guillermo and Gutierrez Gustavo [8].07×10 J/m S E 900 600 I 300 20 40 60 80 100 120 140 160 180 Distance fromfusion boundary (mm) Fig. 7 entails the temperature cycle for AISI 1005 material based on empirical method (19). Compared to fig.6. is 9. 1600 eg. as in (19) the empirical correlation for peak temperature distribution in Heat Affected Zone (HAZ) in welded joints. In this empirical model the heat input parameter ( = ) involves welding velocity which has a major impact for temperature cycle developed in welded joints. 6 it clearly shows the temperature decrement is more sensitive near the fusion zone (liquid-solid interface). C) 0 Fig. is more linear. [5] 19 of 446 Copyright @ Department of Electrical Engineering. which has included the velocity of moving point heat source based on Carslaw-Jeager’s moving point heat source model. 2. Whereas in fig.83×10 J/m tem N pe ond ra tu ime re ns di str iona ibu l tio n ( ) 200 8 N 2 4 longi on-dimen4 tudin al dir sional l6en 2 ectio n from gth a8long the w eld p 10 ool (X ) 6 900 Temperature (Deg. the temperature distribution in fig.83 × 10 / (time of welding in single pass = 65 sec)has been applied to determine the peak temperature from the empirical equation (19). 4 this also reveals the same nature as shown in fig. In present analysis for a 40 60 80 100 120 140 160 180 Distance fromfusion boundary (mm) Fig. In fig. 600 4 1 300 20 A Temperature (Deg. it decrease gradually.rectangular plate of dimension 180 × 50 × 10 mm. Variation of temperature cycle based on perturbation method for ASI 1005 workpiece Figure 4 and 5 also denotes the temperature distribution based on perturbation method for AISI 1005 material. NIT Durgapur . Also this investigation has brought similarity with the dimensionless temperature distribution developed by Levin Pavel [11].

5th ed. pp. 2008. [15] Jiji. 2009. and Kamtekar A. [4] Pavelic V. Inc. 4124-4131. et al. 151-157. 469-479 2013. Modeling of Casting and Welding Processes II. ASM International. Cengage Learning. 6. 250-253. pp. the temperature distribution in (18) gives θ(0) = 1and θ(1) = 0. 270-272. 2006.. the magnitude becomes negative. Welding Journal Research Supplement”. AIME New York. pp. 865-872.. The value of perturbation parameter has not varied in considerable way as the thermal expansion coefficient has been assumed as constant because it changes spontaneously with transient manner which is difficult to predict.. Fundamentals of Materials Science and Engineering / An Interactive. steady state analysis carried out by perturbation method is valid in the range of non-dimensional length X = 0 and = 1. 157-165. vol. 3rd ed. International Journal of Heat and Mass Transfer. 1997. 1969. “Temperature distribution in a rectangular plate heated by a moving heat source”. Commun Nonlinear Sci Numer Simulat. International Journal of Heat and Mass Transfer. 1497-1501. “The prediction of temperature distribution and weld pool geometry in the gas metal arc welding process”. 317. 1946. For X = 0 and X = 1. Trans. “Analytical solution of temporary temperature field in a half-infinite body caused by moving tilted volumetric heat source”. 50. [11] Levin Pavel. 2009.. “The effect of thermal properties and weld efficiency on transient temperatures during welding”... 2091-2102. 35. B. H. Mathematical modeling of weld phenomena”. 849-865. International Journal of Heat and Mass Transfer.. 49. 107-111.337. [3] Rosenthal D. [17] Kreith Frank. “The theory of moving sources of heat and its applications to metal treatments”. pp. 2001. 543-570.. 2003. [5] Tsai N. pp. “Analytical solution for the 3-D steady state conduction in a solid subjected to a moving rectangular heat source and surface cooling”. V. [19] on perturbation method for the Stefan’s problem with time dependent boundary conditions. Manglik M. The general solution of non-linear. J. From the current results and discussion the following conclusions can be drawn: [8] Araya Guillermo & Gutierrez Gustavo. “Heat and mass transfer in MHD free convection from a movable permeable vertical surface by a perturbation technique”. Also for = 0 results the temperature distribution for steady state constant thermal conductivity. Vol. “General solution for stationary/moving plane heat source problems in manufacturing and tribology”. International journal of Heat and Mass Transfer.. 48. International Journal of Heat and Mass Transfer (Pergamon). International Journal of Heat and Mass Transfer. This comprehensive model based on perturbation method suitable for non-linear steady state analysis can be useful tool for primarily measuring the surface heat loss phenomena. three dimensional temperature distribution due to moving laser beam”. 2000. I REFERENCES [13] Winczek Jerzy. 68. Vol. Also the change is very minor and it has not high impact on the temperature distribution. 14. vol. Springer-Verlag Berlin Heidelberg. W. vol. M. Vol. 7th ed.. Callister William D. pp. 2008. vol. 2. vol. A S E 4. H.. Y. CONCLUSION [7] Hou Z. [10] Kukla-Kidawa J. vol. and Eagar T. pp. Principle of Heat Transfer. Comptes Rendus Mechanique. [6] Goldak J. & Komanduri R. Heat Conduction. vol. et al. Vol. Latif M. 1. & Davies M. “A general solution of 3-D quasi steady state problem of a moving point heat source on a semi-infinite solid”. 1998.The accuracy of this model can be easily compared with the results produced by Caldwell J. pp. and Kwan Y.. 60. 1984. 2 and fig. 3. NIT Durgapur . 2952-305s.. 4872-4882. Painter M. pp. 2011. “Analytical solution to transient temperature field in a half infinite body caused by moving volumetric heat source”. pp. International Journal of Heat and Mass Transfer. [14] Parkitny Ryszard & Winczek Jerzy. John Willey and Sons. 78-79. assures that perturbation method satisfies two boundary conditions. [16] Abdelkhalek M.. vol. brazing and soldering”.. Computers and Structures (Pergamon).. “Analytical solution for transient. “Solutions for modelling moving heat sources in a semi-infinite medium and applications to laser material processing”.. [18] Jr. The Institute of Materials. 4 1 [12] Osman Talaat and Boucheffa Abderrahmane. et al. pp. Journal of Material Processing Technology. 1998. Y. vol. 5774-5781. pp. pp. A.A. 3. [6] 20 of 446 Copyright @ Department of Electrical Engineering. pp. 51. G. ASME. & Kwan Y. pp. Raj & Bohn Mark S. “On the perturbation method for the Stefan problem with time dependent boundary conditions”. ASM Handbook. [2] Little G. [9] Elsen Van M. 1679-1698. 77. “Thermal stress analysis in solids near the liquid region in the welds. 43.. pp. 68. With value higher than 1. [19] Caldwell J.. 2010. vol. 2009. Mechanics Research Communications. 46. 1094 – 1102. 53.. vol. [20] Masubuchi K. “Changes of weld pool shape by variation in the distribution of heat source in arc welding”. pp. “Experimental and computed temperatures histories in gas tungsten arc welding of thin plates. Thus it is convergent within the region as shown in fig. [1] Wahav M. S. 233-239. 2007. pp. “Residual stresses and distortion in welding. pp.

direct the shell side fluid flow along tubes. Mechanical Engineering GLA University Mathura. The results are pragmatic to be sensitive to the turbulence model selection. In this present work. have organic.Computational Analysis of Shell Fluid of Shell and Tube Heat Exchanger Allowing the Outcome of Baffles Disposition on Fluid Flow Avanish Singh Chahar Vijay Kr.tech Student. which a one pass shell is generally. attempts were made to investigate the impacts of various baffle disposition angles on fluid flow and the heat transfer characteristics of a shell-and-tube heat exchanger for three different baffles disposition angles namely 0°. Materials of baffles consists of it should have low conductivity. ANSYS 13.0. In general.com Associate Professor. heat transfer. The major roles of baffles are to Hold tube in position (prevented sagging). all of the leakage and bypass streams do not exist or are negligible [4]. NIT Durgapur . one by segmental baffles perpendicular to fluid flow and two by segmental baffles inclined to the direction of fluid flow are associated for their performance. non-metallic. The shell is a container for the shell fluid. through which one of the fluids runs. and possess high temperature. Baffle spacing is the center line distance between two adjacent baffles [7]. Although the tubes may have single or multiple passes. recirculation near the baffles. The baffles disposition depends on heat transfer coefficient and pressure drop. preventing tube vibration and sagging and to divert the flow across the bundle to obtain a higher heat transfer coefficient. Air conditioning.0. Dwivedi M.in Abstract—The modeling and replication of baffles of shell and tube heat exchanger using computational fluid dynamics (CFD). LITERATURE ASSESSMENT The tube side and shell side fluids are separated by a tube sheet [1]. E shell is the most commonly used due to its low cost and simplicity. There are two main types of heat exchangers. Typically one medium is cooled while the other is heated. the shell side outlet temperature. although shells of different shapes are used in specific applications. One common example of a heat exchanger is the radiator in a car. It is design to support the tubular bundles and direct the flow of fluid for maximum influenceiveness. Baffles Disposition. pressure drop. [8] and [9].dwivedi@gla. Mechanical Engineering GLA University Mathura. They are widely used in petroleum refineries. Turbulence Model. Indirect contact heat exchanger. [5] and [6]. A heat exchanger is a device built for efficient heat transfer from one medium to another in order to transfer and process energy. and has the highest log-mean temperature-difference (LMTD) correction factor. where both media are separated by a wall through which heat is transferred so that they never mix. refrigeration and automotive applications. as associated to the main stream. Direct contact heat exchanger. Shell and tube type heat exchanger is an indirect contact type heat exchanger as it consists of a sequence of tubes. Heat exchangers have always been an important part to the lifespan and operation of many systems. The replication results for various shell and tube heat exchangers. petrochemical plants. [2] and [3]. I. while the other fluid flows in the shell over the tubes to be heated or cooled. both in production and operation. which is amplified by both fluid velocity and the length of the heat exchanger. For this particular study E shell is considered. conventional shell and tube heat exchangers result in high shell-side pressure drop and formation of recirculation 21 of 446 Copyright @ Department of Electrical Engineering. natural gas processing. 15° and 25°. where both media between which heat is exchanged are in direct contact by each other. II. The modeling and replication is done on ANSYS 13. From the CFD replication results. The heat exchanger model used in this study is a small sized one. India avanishsinghchahar@gmail. it is cylindrical in shape by a circular cross section.ac. 4 1 A S E Keywords—Shell and Tube Heat Exchanger. chemical plants. I INTRODUCTION A baffle is the primary part of the air to water shell and tube type heat exchanger. Usually. optimal mass flow rate and the optimum baffle disposition angle for the given heat exchanger geometry are determined. Prevent the influence of vibration. in which it transfers heat from the water (hot enginecooling fluid) in the radiator to the air passing through the radiator. there is one pass on the shell side. India Vijay. Baffles are used to support the tubes for structural rigidity. Computational Fluid Dynamics. Shell-and-tube heat exchangers in various sizes are widely used in industrial operations and energy conversion systems. Baffle is provided by a cut which is expressed as the percentage of the segment height to shell inside diameter. This increases fluid velocity and the influenceive heat transfer co-efficient of the exchanger.

In this study. In this study. [degrees] θ 3 Central baffle spacing. which give better performance then single segmental baffles but they involve high manufacturing cost. a small shell-and-tube heat exchanger is modeled for CFD replications. In order to improve the thermal performance at a rational cost of the Shell and tube heat exchanger. installation cost and maintenance cost. Front outlook of arrangement of baffles of shell and tube heat exchanger by 0° 22 of 446 Copyright @ Department of Electrical Engineering. S. offering obvious advantages [11].no TABLE I. The geometry modeling of shell and tube heat exchanger is explained below. Most of the researches now a day are carried on semicircular baffles. 2. Some of the design constraints and the predetermined geometric constraints are presented in Table II. [m] L 6 Number of baffles Nb 7 Temperature. The geometric model is optimized by varying the baffle disposition angle i. In this study four baffles are placed along the shell in alternating orientations by cut facing up. NIT Durgapur . [m] Di 5 Heat exchanger length. Computational Fluid Dynamics is now an established industrial design tool. So. By modeling the geometry as accurately as possible. Front outlook of arrangement of baffles of shell and tube heat exchanger by 15° MODELING PARTICULARS In this paper we have considered small shell and tube heat exchanger in order to take observation of flow inside the shell. in order to create flow paths across the tube bundle. III.zones near the baffles. NOMENCLATURE Input Constraints Nomenclature Symbol 1 -2 Heat flux [Wm ] q 2 Baffle disposition angle.. The computational modeling involves pre-processing. 4 1 Fig. 1. Front outlook of arrangement of baffles of shell and tube heat exchanger by 25° A S E I A. Geometry Demonstrating The modeling is premeditated in ANSYS 13. a full 360° CFD model of shell and tube heat exchanger is considered. The influenceiveness and cost are two important constraints in heat exchanger design. 15° and 25°. the flow structure and the temperature distribution inside the shell are obtained.0 under the given constraints in Table II. Fig.e. baffles in the present study are provided by some disposition in order to maintain a rational pressure drop across the exchanger [10]. [m] B 4 Shell inner diameter.processing. cut facing down. [K] T 8 Velocity vector Vr Fig. 0°. The co mp l e xi t y by experimental techniques involves quantitative description of flo w phenomena using measurements dealing by one quantity at a time for a limited range of problem and operating conditions. cut facing up again etc.. solving and post. 3.

Fig. Residual plot arrangement of baffles of shell and tube heat exchanger by 25° 23 of 446 Copyright @ Department of Electrical Engineering. [m] 0. RESULT AND DISCUSSION Replication results are obtained for different mass flow rates of shell side fluid ranging from1. The choice of turbulence model is very critical in CFD replications. k-ε turbulence model are tried. 5. The inlet velocity profile is assumed to be uniform. [θ] 0°. Residual plot arrangement of baffles of shell and tube heat exchanger by 0° The shell inlet temperature is set to 300 K. B. IV. Noble and I. there is no universal criterion for selecting a turbulence model. Residual plot arrangement of baffles of shell and tube heat exchanger by 15° D. Zero gauge pressure is assigned to the outlet nozzle.5 kg/s fluid flow rate for model by 0° baffle disposition angle are validated by the Fig.5 kg/s. Mesh size selected for four baffle case: the fine mesh by approximately 283467 elements have taken.TABLE II. C. 4. The convergence benchmarks were set to 10-4 and convergence was achieved after 300 iterations. It is found that the exit temperature at the shell outlet is matching by the literature results and the deviation between the two is less than 1 %.023 3 Shell inner diameter. Boundary Conditions The working fluid of the shell side is air. The entire geometry is divided into three fluid domains Fluid Inlet.no Nomenclature Value 1 Baffle disposition angle. Turbulence Model Since the flow in this study is turbulent. INPUT CONSTRAINTS OF SHELL AND TUBE HEAT EXCHANGER Input Constraints data available in the literature G. 4 1 No slip condition is assigned to all surfaces. The simulated results for 1. NIT Durgapur .150 5 Length of baffles. turbulence influences should be taken into account using turbulence modeling.N. Mesh Assortment Mesh generation is performed in ANSYS 13.0.015 5 Number of baffles 4 B. assuming the shell is perfectly insulated. Eason. Fluid Shell and Fluid Outlet and four solid domains namely Solid Baffle1 to Solid Baffle4 for four baffles respectively. 6. 15° and 25° 2 Central baffle spacing. [m] 0. However. [m] 0. In this study. [m] 0. The residual were plotted and the plot is shown below. The standard k-ε model is a semi-empirical model based on model transport equations for the turbulence kinetic energy k and its dissipation rate ε. S E A I Fig.5 kg/s and 2. S. The surfaces of the model are meshed using square elements.030 4 Heat exchanger length. The zero heat flux boundary condition is assigned to the shell outer wall (excluding the baffle shell interfaces). Sneddon (1955).

Fig. The pressure drop is less for 25° baffle disposition associated to other two models due to smoother direction of the flow.70 2952.70 5441. for fixed tube wall and shell inlet temperatures.77 3228.51 Pascal respectively. shell side outlet temperature. It can be seen that associated to 0° baffle disposition angle. 9.32 From the CFD replication results. 15°& 25° baffle disposition angles.15 4 1 A S E TABLE IV. pressure drop values. There is no much deviation of temperature for all the three cases considered.5kg/s and 2.34 and 7440.29 2799.5kg/s Shell side outlet Temperature (K) Shell side Pressure Drop (Pascal) Heat Transfer (W) Heat Transfer Coefficient (W/m2K) 00 331.72 116684. It is seen that the temperature gradually rises from 300 K at the inlet to 340 K at the outlet of the shell side.5kg/s Baffles disposition angle (Degree) Heat Transfer (W) Heat Transfer Coefficient (W/m2K) 00 334. section displaying velocity path lines along the shell for 25° baffle disposition angle The replication results for 1.14 16300.62 210657.11 7440. 7. Baffles disposition angle (Degree) Fig.30 3248.5KG/S Mass Flow Rate = 2.02 250 337.30 7377.nearly equal to 1 m/s for all the three models at the inlet and exit surface and the velocity magnitude reduces to zero at the baffles surface.74 19120. section displaying velocity path lines along the shell for 0° baffle disposition angle From the stream line contour of Fig.71 150 337. The maximum pressure for models by 0°. It is found that for two mass flow rates 1. 15° and 25° baffle disposition are obtained. 7-9. COMPARATIVE STUDY FOR MASS FLOW RATE OF 1. 15° and 25° baffle dispositions are 19120.91 7072.5 kg/s there is no much influence on outlet 24 of 446 Copyright @ Department of Electrical Engineering. section displaying velocity path lines along the shell for 15° baffle disposition angle I Shell side outlet Temperature (K) COMPARATIVE STUDY FOR MASS FLOW RATE OF 2. provide a smoother flow by the inclined baffles controlling the fluid flow. The average temperature at the outlet surface is nearly 336 K for all the three models.62. 8.20 150 336. TABLE III.91 3761.20 339.5 kg/s mass flow rate for models by 0°.34 243720.51 279217. heat transfer and heat transfer coefficient for varying fluid flow rates are provided in Table III-IV and it is found that the shell outlet temperature declined by increasing mass flow rates as expected even the deviation is minimal.37 131191. The maximum velocity is Shell side Pressure Drop (Pascal) 250 Fig. NIT Durgapur .5KG/S Mass Flow Rate = 1. 16300.33 103033.48 3411. it is found that recirculation near the baffles prompts turbulence eddies which would result in more pressure drop for model by θ = 0° where as recirculation are lesser for model by θ = 15° and the recirculation formed for model by θ = 25° are much less in comparison to the other two models which indicates the resulting pressure drop is optimum.

2336–2345. [3] Mukherjee. Practical Thermal Design of Shell-and-Tube Heat Exchangers. as the disposition angle is amplified from 0° to 25°. ACKNOWLEDGMENT The authors take this opportunity to express our whole-hearted gratitude towards the Department Of Mechanical Engineering GLA University. Reifschneider.. WorldAcademy of Science. C... pp. [2] Schlunder. Hemisphere Publishing Corp. 4 1 A S E Hence it can be determined that shell-and-tube heat exchanger by 25° baffle disposition angle results in better performance associated to 15° and 0° disposition angles. Int. Influences of baffle disposition angle on flow and heat transfer of a heat exchanger by helical baffles. 10. E. [6] Thirumarimurugan. Influence of baffle spacing on pressure drop and local heat transfer in shell and tube heat exchangers for staggered tube arrangement. 2008. Kannadasan. [8] Li. editor. E. Malalasekera. Ramasamy. W. J. Heat Exchanger Design.. the center row of tubes are not supported. for heat exchanger by 15° baffle disposition angle and by 54.35% for heat exchanger by 25° baffle disposition associated to 0° baffle disposition heat exchanger.. Mathura that helps us for the completion of this paper. M.5 kg/s.. It is by great pleasure that we express our deep sense of gratitude to them for their valuable guidance. Satish Chand.74%.K. 1303–1311 [9] Su Thet Mon Than. for heat exchanger by 15° baffle disposition angle and by 54.74%. I It is also established that 25° is the maximum baffle angle for optimum performance of the model. Hence it can be clinched that shell and tube heat exchanger by 25° baffle disposition angle results in better enactment associated to 15° and 0° disposition angles. 1617-1628. 548-552. ninth ed.. if it is amplified beyond 2. NIT Durgapur . 601–610. Essex (England): Pearson. International Journal of Heat and Mass Transfer 49 (2006). Hence it can be pragmatic that shell and tube heat exchanger by 25° baffle disposition angle results in a rational pressure drop. [5] Uday Kapale. the mass flow rate must be below 2. pp. Khin Aung Lin.35%.. Modeling for shell-side pressure drop for liquid flow in shell-and-tube heat exchanger..e. Shell side CFD analysis of a small shell-and-tube heat exchanger. L. for heat exchanger by 25° baffle disposition angle. New York. However the shell-side pressure drop is declined by increase in baffle disposition angle i.. Energy Conversion and Management 51 (2010). [11] Versteeg. The pressure drop is declined by 14. The pressure drop is declined by 14. Tarrytown (NY): TEMA Inc. pp.Inc. H. T. [7] Sparrows. pp. 2004. Chemical Engineering and Processing 47 (2008). first ed. For the given geometry. Hence the baffle cannot be used influenceively. Heat Exchanger Design Handbook. pp. constant encouragement and patience throughout this work and we were efficiently able to complete this paper in time.. if the angle is beyond 25°. R. G. Following conclusion has made on it. Rui Li. Mi Sandar Mon. Standards of the Tubular Exchanger Manufacturers Association. M. [4] Ender Ozden. 1995. The maximum baffle disposition angle can be 25°.. 2007. We have been greatly benefited by this valuable suggestion and ideas. Ya-Fu Gao. E.temperature of the shell even though the baffle disposition is amplified from 0° to 25°. Engineering and Technology 46. Begell House. 1004 – 1014. H. [10] Yong-Gang Lei. Kottke. REFERENCES [1] CONCLUSION The shell side of a small shell-and-tube heat exchanger is modeled by appropriate aspect to resolve the flow and temperature fields. 25 of 446 Copyright @ Department of Electrical Engineering. D. American Journal of Applied Sciences 5 (2008). Gaddis. Ilker Tari. Performance Analysis of Shell and Tube Heat Exchanger Using Miscible System.V. An introduction to computational fluid dynamics: the finite volume method. 1983. V. New York. International Journal of Heat and Mass Transfer 29 (1986). Influence of inter baffle spacing on heat transfer and pressure drop in a shell-and-tube heat exchanger. pp. Heat Mass Transfer 41 (1998)... Ya-Ling He. Bureau of Energy Efficiency.5 kg/s the pressure drop increases rapidly by little variation in outlet temperature.

namely. normally the steady state condition with no heat generation is considered. Gorla [8] and Kiwan [9] investigate thermal analysis of natural convection and radiation in porous fins of rectangular shape. In the fin analysis. An important application area is in the cooling of nuclear reactors. Considering this. Kundu et al. The basic philosophy behind using porous fins is to increase the effective area through which heat is convected to ambient fluid. INTRODUCTION The enhancement of heat transfer from porous fins has become an important factor that has captured the interest of many researchers. To the best of author’s knowledge. the governing 26 of 446 Copyright @ Department of Electrical Engineering.heat porous fin and the solid fin and found that the surface temperature decreases with the increase of porosity. The internal heat generation may be a constant but in a more realistic situation it is a function of temperature. [10] investigated experimentally the impact of using porous fins on the heat transfer and flow characteristics in plate fin heat exchanger. [4] computed the enhancement of the heat transfer of porous fin of various profiles. Das and Ooi [6] have predicted the multiple combination of parameters for designing a porous fin subjected to a given temperature requirement.com Abstract—The present article investigates the temperature distribution and fin efficiency of a porous fin having temperature dependent internal heat generation. generation I. To formulate the heat transfer equation for the porous fin. This study is based on finite-length fin with insulated tip. Bhanja and Kundu [3] performed thermal analysis of a T-shaped porous fin. [5] developed an analytical solution to determine temperature distribution and fin performance of a porous pin fin used for electronic cooling. Bhanja et al. Kiwan and Nimr[1] introduced the concept of using porous fins to enhance the heat transfer from a given porous surface. a numeric scheme. The experimental work is very limited in porous fin area. their experiment was conducted under forced convection heat transfer conditions.Thermal Analysis of Porous Fin with Internal Heat Generation Aaditya Chandel & Dipankar Bhanja Department of Mechanical Engineering National Institute of Technology Silchar Silchar. they have only studied about temperature distribution of the fin. Here the heat generation is from a nuclear source and is a combination of fast neutrons and gamma rays. It is observed from literatures that a very few work is present about porous fin. The results indicate that the numerical data and analytical approach are in agreement with each other.VIM. However. Hatami and Ganji [11] investigate heat transfer study through porous fins (Si3N4 and AL) with temperature-dependent heat generation. The effects of various geometric and thermophysical parameters on the dimensionless temperature distribution and fin efficiency are studied that may help in optimum design analysis of a porous fin. They concluded that porous fins provide much higher heat transfer rate than traditional fins. the energy balance and Darcy’s model are used. Kim et al. finite difference method is adopted. However. Finally. Bassam and Hijleh [7] investigated a problem of natural convection from a horizontal cylinder with multiple equally spaced high conductivity permeable fins on its outer surface. India dipankar. They have done a comparative study between the 4 1 A S E I Keywords—porous fin. porosity. no such article is found so far which considers the effect of internal heat generation except the analysis performed by Hatami and Ganji [11]. They found that using porous fin with certain porosity may give same performance as conventional fin and save 100φ of the fin material. natural convection. Recently. An analytical technique called Variational iteration method (VIM) is proposed for the solution methodology.bhanja@gmail. The fin is exposed in natural convection condition. By using same method. in some practical applications internal heat generation is also occurred. The present paper investigates the temperature distribution as well as efficiency of a porous fin of rectangular shape with temperature dependent heat generation. To validate the analytical results. an increase in fin efficiency is observed with considering internal heat generation within the fin material. Kundu and Bhanja [2] predicted the performance and optimum design analysis of porous fin of rectangular shape using ADM. However. NIT Durgapur .

III. (2) into Eq. As the fin is porous.Schematic of porous fin with variable heat generation. 1. isotropic and saturated with a single phase fluid. NIT Durgapur (5) . Consdering these assumptions. the energy balance equation can be formulated as: dq dx +q A = ρCp gKβW γ 2 [T-Ta ] + hP (1-φ) [T-Ta ] (1) Also from Fourier‘s law of conduction: ⁄ (2) Where. A Cp Da g h k KR K L Nu P q Q qi Qi Ra t T Vw W x X β ε η φ ψ θ θt γ ρ As in Fig. The porous medium is considered as homogeneous. it allows fluid to penetrate through it. an analytical methodology followed by Variational iteration method (VIM) [12] is applied to solve the nonlinear class of governing energy equations of a porous fin attached to a vertical isothermal wall. Thus. keff is the effective thermal conductivity of the porous fin that can be obtained from: k eff φ. differential equation of the present problem is reduced to a highly non-linear equation that cannot be solved with usual analytical techniques.k s (3) Substitution of Eq. The method is proved to be powerful for solving any class of nonlinear governing equations without adopting linearization or perturbation technique. The fin is having the heat generation which is varying with temperature linearly. II. The physical properties of solid as well as fluid are considered as constant except density variation of liquid. the temperature distribution and its efficiencies are easily being evaluated for a wide range of design variables of porous fins.1. To simulate the interaction between the porous medium and fluid Darcy formulation [4] is used. a porous fin of rectangular profile has been shown. Considering the fin to be adiabatic at the tip and having the heat transfer in one dimension only. The fin looses heat to the surrounding through convection mode. which may affect the buoyancy term where Boussinesq approximation is employed. It provides an analytical solution in the form of power series where the temperature on the fin surface can be expressed explicitly as a function of position along the length of the fin.(1) leads to: d2 T Subscripts a Ambient condition b Base s Solid properties f Fluid properties eff Effective properties MATHMATICAL FORMULATION 2+ dx ρCp gKβ T-Ta keff keff tγ 2 2h 1-φ T-Ta keff t =0 (4) It is assumed that internal heat generation in the fin varies linearly with temperature as: = 1+ε (T-Ta ) 27 of 446 Copyright @ Department of Electrical Engineering. k f 1‐φ . NOMENCLATURE Cross-sectional area of the fin Specific heat of the fluid passing through fin pores Darcy number Gravity constant Heat transfer coefficient over the fin surface Thermal conductivity Thermal conductivity ratio Permeability of the porous fin Length of the Fin Nusselt number Perimeter of the Fin Actual heat transfer rate per unit width Dimensionless actual heat transfer rate per unit width Ideal heat transfer rate per unit width Dimensionless ideal heat transfer rate per unit width Rate of heat generation per unit volume Rayleigh number Thickness of the fin Temperature Average velocity of fluid passing through porous fin Width of the fin Axial length measured from the tip of the fin Dimensionless axial distance Coefficient of thermal expansion Variable heat generation parameter Efficiency of the Fin Porosity or void ratio Thickness to length ratio Dimensionless temperature Dimensionless tip temperature Kinematic viscosity Density of the Fluid 4 1 A S E I Fig. In this work.

X = ⁄ . Nis a nonlinear operator and g(x ) is an inhomogeneous term. It is clear that the temperature expressions are a function of unknown tip temperature θt .(7). X2 X4 4 (15) Fin efficiency After obtaining temperature distribution as a spatial function. Ω S 3 εg ⁄ (6) S1 θ2 S2 θ+S3 =0 (7) ⁄ 0 0 1 2S12 θt 3 3S1 S2 θt 2 S22 θt S13 θt 4 S1 S22 θt 2 S1 S32 2S12 S2 θt 3 2S12 S3 θt 2 (8b) (14) 2 2S1 S3 θt 2S1 S2 S3 θt S2 S3 X6 8 Ψ VARIATIONAL ITERATION METHOD (VIM) IV.φ=0.3 Fin temperature distribution As a starting approximation for VIM solution. a correct functional can be constructed as follows: S E I {L un (τ) +N un τ g(τ)} dτ (10) where. εg=0. Such a starting approximation automatically satisfies one of the boundary conditions given in Eq.6 θ un+1 x =un x + x λ 0 A RaDa (12) 0. Based on VIM.6 S3=0. θ0 =θt Where. η= ⁄ (18) RESULT AND DISCUSSION To illustrate the results of the present work the temperature distribution on the fin surface and fin efficiency have been displayed for a wide range of thermo geometric and thermo physical parameters for both solid and porous fins. NIT Durgapur . The subscript n denotes the nth-order approximation.4 -4 Ra=10 . KR = ks ⁄kf . 28 of 446 Copyright @ Department of Electrical Engineering. Nu=10.6 0. actual heat transfer rate per unit width is determined by applying the Fourier’s law of heat conduction at the base and can be expressed in dimensionless form as (8a) 1 S3 B.4 A. Eq. 1. 0 . which can be determined by putting boundary condition (8b) and solving by Newton-Raphson method. 4 3 ψ=0.0 without heat generation with heat generation numerical result (11) 0. ψ = ⁄ . q0 is the internal heat generation at temperature Ta. Considering all the dimensionless parameters.KR=10 0. (4).ξ ⁄ ξ λ 0 (τ) {L θn (τ) +Nθn τ g(τ)} dτ ψ + 2Nu ψ 1-φ -S3 1+εg ψKR (17) Thus. the following differential equation may be considered: Ideal heat transfer per unit width can be expressed in dimensionless form as Lu + Nu = g(x) (9) where. Eq.θ0 is assumed as constant. is a general Lagrangian multiplier that can be identified optimally via the variational theory.0 X Fig. following iteration formula can be obtained accordingly θn+1 ξ =θn ξ + Taking L= λ (τ) = τ .1. fin efficiency can be calculated as: V.6 0.8 1.2 0. When the above VIM formulation is applied to Eq.8 . S1 = RaDa⁄ψ2 Ω .(8a).e. (7) is subjected to the following boundary conditions: S1 θt 2 S2 θt Similarly the higher order terms can easily be calculated using recursive relation expressed in Eq. 2 Comparison of present work with numerical result for dimensionless temperature distribution with the variation of heat generation parameter. First two iterations are: 0.0 0. i.S2 = 2Nu (1-φ)⁄ψ2 Ω S3 = q0 L2 keff (Tb -Ta . Lagrange multiplier obtained as follows S3=0. some dimensionless parameters are introduced as follows: θ1 θ0 θ2 θ1 θ = T-Ta ⁄ T -Ta . (11).Da=10 .0. considered as arestricted variable. (4) is reduced to: d2 θ⁄dX2 (13) (16) 4 1 According to VIM. For simplifying Eq.2 0. Lis a linear operator.5.

Fig.6 0. Fin efficiency is decreased with heat generation number. Fig. 3A is drawn to show the effect of Rayleigh number and Darcy number on dimensionless temperature distribution.2 0. Tip temperature is decreased with the increase in these parameters.6 Ra=10 .2 0.5B.6 0.5 θ 0.6 A εg=0. 3 Variation of temperature distribution with (A) Ra &Da (B) (B) φ=0.6.5 ψ=0. it is clear that the fin efficiency is going to decrease with Ra and Da.0 θ Da=0.8 0. Nu=10.6.8 0. With the increase of these parameters although actual heat transfer increases but side by side ideal heat transfer rate also as defined in Eq.0.4 φ=0. 4A considers internal heat generation while Fig.0 3 0.S3=0.2 0.0 6 X (A) 0. εg=0.0 0. In both the cases it has been observed that for solid fin the dimensionless The influence of thermophysical parameters on the fin efficiency has been investigated in Fig. Nu=10.6 0.0 1.φ=0.5.8 1.2 0.8 3 ψ=0.0 0. 4 A comparison in temperature distribution between a solid fin and porous fin (A) with heat generation (B) without heat generation 0 .8 0.6 εg=0. Increasing Ra improves the convective heat transfer coefficient between the fin and the working fluid. It can be mentioned that the variation due to Da is more when Ra is increased.4 θ S3=0. On the other hand. Fig. Fig.0 1. 0.0002 0. It is also noticed that tip temperature is lowered with high porosities.1. 5A. In Fig.4 0. It is observed that fin efficiency is higher considering heat generation.6 0. Thus fin efficiency is dropped.6 0.0. εg=0.0 porous fin solid fin 0. 6 is envisioned to show the effect of heat generation on fin efficiency with the variation of Nu.4 0.(A) 0. When Da is increased more fluid can penetrate through it that reduces tip temperature of the fin.6 Ra=106.Da=10 .8 1.1.4 0.8 1.KR=10 .0 (B) 4 1 θ εg=1. 3B is plotted to show the impact of variable heat generation parameter. It is an obvious result as heat generation always maintains higher surface temperature.temperature is considerably higher than porous fin.ψ=0.6 φ=0. 3 KR=10 porous fin solidfin 1.4 0.5 0.8 0. By raising the heat generation parameter.8 -4 Ra=10 .2 0. a slight increase in fin efficiency is noticed with the increase in variable heat generation parameter.1. It is also seen that tip temperature with heat generation is higher than that of without heat generation.8 x . A comparison between a porous fin and a solid fin is also been done. 1.0 x Fig. the higher heat generation leads to higher fin temperatures as the fin must dissipate a 29 of 446 Copyright @ Department of Electrical Engineering. Fig. NIT Durgapur .0 1. 4B is showing the variation by neglecting the internal heat generation.φ=0. the variation of fin efficiency can be visualized with varying the dimensionless heat generation number as well as the internal heat generation parameter. (17).6. Nu=10.1 0.0.4 S E I 0. Nu=10.0 4 -4 0. From Fig. From the figure.1.KR=10 .KR=10 . Fig. An appreciable dependency has been observed with these parameters. With the increase of this parameter tip temperature is increased which is an obvious result.Da=10 .4 is displayed to show effect of porosity parameter on temperature distribution over the fin surface.0 0. it can be highlighted that both analytical and numerical predictions are matched convincingly with a high accuracy.Da=10-4.0 x Fig.0 4 5 0.2 0. Alternatively.S3=0. 2 is prepared for the validation of the present model by comparing with numerical values for the porous fin with and without heat generation.2 φ=0. the temperature profile reaches to the higher value.0001 Da=0.4 0.5 3 ψ=0.6 Ra=10 Ra=10 0.

Thus.0006 0. 5 Ra=10 4 Ra=10 0.6 4 3 . For the aforementioned conditions. • Fin efficiency is decreased with Ra. 3 εg=0. The following concluding remarks can be drawn from the present study: • Fin tip temperature is decreased by increasing Da and Ra. 7.KR=10 0. A comparison in fin efficiency between a solid fin and a porous fin with the variation of Nusselt number is shown in Fig.S 3 =0.8 0.6 0.3 0.0 0.0004 0.0.0 S E I with heat generation without heat generation 0. whereas the effect of variable heat generation parameter shows a reverse trend.7 εg=0.0008 0. Moreover.0 0. Variational iteration method (VIM) has been proposed for the solution of governing fin equation.KR=10 .0 3 ψ =0.6 φ=0.1.6 η -4 η 0.4 0.5 0.1. porous fin ψ=0.0.6 0.8 0.4 10 40 Nu CONCLUSION The present work is devoted to determine the temperature distribution and fin efficiency of a straight porous fin with temperature dependent internal heat generation.KR=10 0.0002 0.4 0. 1.2 4 Ra=10 .1. a better buoyancy effect is created in porous fin.6 Fig. • Solid fin maintains higher surface temperature than porous fin. namely.generation. With heat generation.S3=0.6. This method provides solution in the form of infinite power series and it has high accuracy and fast convergence.6 0.0 0.8 4 -4 Ra=10 .1 0. 3 ψ=0.8 0. 30 of 446 Copyright @ Department of Electrical Engineering. S3=0.6 εg=0. Da and Nu.1. ε g =1. Nu=10.0.4 0. larger amount of heat to the surroundings under steady state conditions.8 Nu=10. .2 Ra=10 ψ=0.0010 (A) 0. 0. It is very much interesting to be mentioned that porous fin has better fin efficiency than that of solid fin when internal heat generation is taken into consideration.4 0.KR=10 Nu=10.φ=0. Although actual heat transfer rate is enhanced with this parameter but side by side it increases the ideal heat transfer rate also which in turn reduces the fin efficiency. 5 Fin efficiency with the variation of (A) Ra & Da (B) 20 φ=0. fin efficiency is better considering heat generation within the fin material.5 (B) 0.0 35 45 50 εg=1 0.Da=10 .1 0. φ =0.Da=10 0.5 εg=0. NIT Durgapur .3 A & (B) 1.5.5 η η 0.KR=10.2 20 Fig. 6 Effect of heat generation on fin efficiency.6. 7 A comparison in fin efficiency between a solid fin and a porous fin (A) 0 with heat generation (B) without heat generation ψ=0. 3 ψ=0.3 0.2 30 40 50 Nu Fig.6 -4 0.8 solid fin 4 -4 Ra=10 .2 20 Da 25 30 (A) 1.7 solid fin porous fin 0.2 0.6 η 4 1 0.0.1. Fin efficiency is decreased with the increase of this number. temperature in the fin surface is higher than without heat 25 30 35 40 45 50 Nu VI.Da=10 .4 S3 0.Da=10 .K R =10 . an approximate analytical technique. The fin dissipates heat to the environment through natural convection.φ=0.5 0.1.

Heat Trans. “Effects of radiative losses on the heat transfer from porous Mass Transf. Bhanja. vol. 7611-7622. 55. Hasanpour. vol.T. Kundu. Lee.” Int. vol. vol. Refrigeration. “A model on the basis of analytics Energy Conver.R. correlations for porous fin in a plate fin heat exchanger. Kiwan.D. 9-16. and K.H. Heat and [5] J.” J. Das. 66. “Thermal analysis of natural convection and radiation in porous fins. and K.A. without considering heat generation solid fin has higher fin efficiency than that of porous fin. Kim. pp. 122.S. vol. D.” B.”Int. and B. J. and B. September 2011. andK. Gorla. pp. J. May 2011.Y. NIT Durgapur .. 282-288.” Energy Conver. Bakier. March 2003. 34. A S E I 31 of 446 Copyright @ Department of Electrical Engineering..vol. and D. Heat [11] M.. October 2013. Kundu. Hatami.. 74.H. S. [2] fins. vol. 2013. J. pp. 211-219. and D. Paek. pp..” Journal of Procedia Engg. 1483-1496.”Int.” ASMEJ. 125. and P. in press.” D. [1] pp. Kang. technique: some examples. [4] S. and A. A. Sci. Abu-Hijleh. J. and M. Kundu. pp. Therm.vol. B. “An analytical prediction for performance and [10] S. Kundu. Kiwan. pp. 4 1 fin used for electronic cooling.” Int. 123. [3] pp. porous fins (Si3N4 and Al) with temperature-dependent heat generation. Mandal. December 2012.” Int... “Predicting multiple combination of parameters for designing a porous fin subjected to a given temperature requirement. Non-Linear Mechnics. vol. “Thermal analysis of aconstructal T-shaped Transf. July 1999. “Thermal analysis of porous pin 699-708. Bhanja.a kind of non-linear analytical for computing maximum heat transfer in porous fins.. “Natural convection heat transfer from a cylinder with high conductivity permeable fins. Ganji. vol.W. REFERENCES [8] R. Commu. 337-352. Ooi. 34.. and Managt. D. March 2000. Heat Mass Transf. vol. pp. [6] R.K.Y. Heat Transf. 790-795. [12] J. “Heat transfer study through porous fin with radiation effects. and Managt. February 2013.“Using Porous Fins for Heat Transfer [9] Enhancement. Bassam. Al-Nimr.• Porous fin provides much better efficiency than solid fin when heat generation is taken into consideration. J. [7] A.S. Bhanja. 572-578.” Int.“Flow and heat transfer optimum design analysis of porous fins. July 2000. Bhanja. He. “Variational iteration method. pp. 46. 34. January 2011. 638-645. J. 38. pp. 1046-1055. B. October 2007. On the other hand. Refrigeration.

in time when used under stated operating conditions. II. Mathura. Some other works done in this Field are Reliability. No. Jingyi [5] pointed out the various application of reliability engineering program to the hydraulic systems. Based on the analysis suggestions are given for improvement of the design or the other parameters so as to improve the effectiveness of the system as a whole. Availability may also be interpreted as the percentage of time a component or system is operating over a specified period of time. 1 2 3 4 Machines Lathe Machine-1 Lathe Machine-2 Shape Parameter Scale Parameter L1 INTRODUCTION (Heading 1) L2 The reliability and availability of production tools can have a important consequence on the productivity and this would be a great boon for the profitability of the company. Abbreviations Used Abstract— Reliability is the most important aspects of a production unit and it is used to measure and analyses the failure data.Reliability Analysis of Two Lathe Machines Arranged in a Machining System Devashish Sharma: M. Failure data are taken from industry and these data are further used to analyse the working condition of the machine. GLA University. India Dr. B. R(t) is : R(t) = e -(t/θ )β 32 of 446 Copyright @ Department of Electrical Engineering. First of all we will analyze for Weibull distribution MLE and then Mann’s Test Characteristic will be calculated to analyze the system. Mathura. A. Keywords—Reliability. Weibull Distribution. GLA University. MTTR I.Tech Scholar Mechanical Engineering. MTBF. India A. Reliability and MTBF of the machines are estimated as per Weibull distribution. Benski and Cabau [6] emphasis to use the field data to estimate the Reliability and analyze the system performance through it. Steps of the Research Methodology A • S E The reliability analysis of a lathe machine is an important aspect for the growth of a manufacturing system and the idea behind it is generated from a paper named “Reliability Analysis of a Piston Manufacturing System” [2]. Analysis of the System The system here consists of two lathe machines arranged in a parallel system. Kurian [4] developed a simulation model to analyze the aircraft maintenance program. Block Diagram • • I LITERATURE REVIEW Abbreviation L-1 L-2 β θ Using least square method the Weibull Parameters β and θ are calculated for all the machines.This project provide me the basis for my present work. According to the Ebling [1] some of factors are given below: Reliability: Reliability is the probability that a system will perform its intended function satisfactorily in the specified environmental conditions for a stipulated period of time. Availability: Availability is the probability that a component or a system is performing its required function at a given point. 4 1 Fig. Mann’s Goodness of fit test is conducted to verify the assumption that failure data is following two parameter Weibull distributions or not. It is characterized by a hazard rate function of the form λ(t) = β/θ(τ/θ)β-1 (1) Reliability function. IV. III. NIT Durgapur (2) . Availability and Maintainability study of high precision special purpose manufacturing machines [3]. In this work a mathematical model has been developed based on the Weibull distribution to validate the failure and repair data. Sr. MACHINING SYSTEM In our present work we have used two machining systems [7] which are arranged in parallel with each other. This is the time dependent failure model in which hazard rate functions are not constant over time. Piyush Singhal: Professor Mechanical Engineering. Weibull failure distribution may be used to model both increasing and decreasing failure rates.1. MLE of the parameters are calculated. These two machines are Lathe Machine-1 and Lathe Machine-2. During this project work we have taken two lathe machines which are arranged parallel to each other and in series with other machines as they are the part of a machining system but they are not taken into consideration because our main focus is on the lathe machines. The Weibull Distribution [1] One of the most useful probability distributions in reliability is the Weibull.

…. θk)=Πf(ti│ θ1…. Mi is an approximation.1665 (12) No.lnti )/Mi] / K2 [(ln ti+1 . Zi = ln [-ln(1.The characteristic life is the interval between guaranteed life and the time at which it can be expected that 63% of the items will have failed and 37% survived.lnti )/Mi] / K2 [(ln ti+1 . For Lathe Machine – 2 Let n = 7 and r = 7 K1 = 3.25)] Let n = 6 and r = 4 L(θ1…. Maximum Likelihood Estimator [1] In common to find the MLE for any distribution with complete data. Where. θk)/ ∂ θi = 0 K1 = 2.t2.MTBF =θ . The shape parameter (β) shows infant mortality failure process for a value significantly less than one. NIT Durgapur . Zi = ln[-ln(1. Mi = Zi+1 . K2 = 1.0.5)/(n+0. of degree of freedom for Denominator = 2*k1=2*3=6 Now the value of Fcrit may be obtained from F. θk that makes the likelihood function as large as possible for given values of t1.5)/(n+0. θk) (6) The objective is to find the values of the estimators of θ1….(i-0. K1 =[ r/2] K2 =[ ( r-1)/2] Mi = Zi-1 – Zi . Weibull MLE 4 1 M = 0. Test Statistic is B. a purely random process by value fairly close to one.9748 MLE estimate of Scale Parameter θ = {1/r[Σtiβ + (n-r) tsβ]}1/β (9) Here. Since.Zi . K2 =[ ( r-1)/2] .59 (from F. of degree of freedom for Numerator = 2*k2=2*1. (7) ∂ Ln Ln(θ1….Ґ ((2/β) + 1)) – [θ (1/β) + 1))]2 (10) (4) Where.5= 3 No.28 (from F.(1/β). The hypothesis is given as: H0: the failure times come from Weibull distribution.5 I The MLE Of two parameter Weibull distribution can be calculated numerically.tn. Thus Hypothesis H0 is accepted which says that the failure times come from Weibull distribution. Scale parameter (θ) . In general the necessary conditions to find the MLEs are to obtain the first partial derivatives of the logarithm of likelihood function with respect to θ1…. M < Fcrit.distribution Fcrit = 6.25)] Where Ґ (x) gamma function is expressed by the following relation: (5) Ґ (x) = ʃy x-1 exp (-y) dy Here [ r/2] is the integer part of the number r/2. θk is found as M = K1 Σ[(ln ti+1 .(i. If M>Fcrit the hypothesis H1 is accepted otherwise H0 is accepted.lnti )/Mi] . Thus Hypothesis H0 is accepted which says that the failure times come from Weibull distribution. of degree of freedom for Numerator = 2*k2=2*3= 6 No.Ґ ((1/β ) +1)) The test statistic is: M = K1 Σ[(ln ti+1 . 33 of 446 Copyright @ Department of Electrical Engineering.distribution table). H1: the failure times did not come from the Weibull distribution. n = number of items on testing ti = failure time of ith item . K2 = 3 M = 0. ts = 1 for complete data. Since. wear out by larger values.distribution Fcrit = 4. For shape parameter the equation is [ Σ ( tiβ ln ti +(n-r) ts β ln ts )/ Σ ti β +(n-r) ts β]. For complete failure r = n. r = number of failures. M < Fcrit. the maximum of the following likelihood function with respect to the unknown parameters θ1….lnti )/Mi] (3) σ2 = Variance =θ2. of degree of freedom for Denominator = 2*k1=2*2=4 Now the value of Fcrit may be obtained from F.distribution table). (11) A S E C. θk and setting these partial derivatives equations zero.(1/ r)Σ ln ti =0 (8) No. K1 =[ r/2].

9389 520 1.120 θ A S E 352.540 1.9416 4031.9748 Characterstic 34 of 446 Copyright @ Department of Electrical Engineering.59 Failure Time (ti) β ln ti 1 140 1.385 0.680 β = 1.84 1616.120 6.3969 1292.8928 -1.177 0.120 5.12 0.680 6.120 5.197 0.1054 0.8928 -0.32 447672.80 3392.568 0.638 1.4998 154.5952 0.ln ti/ Mi 4.680 2 290 3 β θ 1.ln ti ln ti+1.680 5.468 0.671 0.1357 314.95 5077.6698 13703.8928 0.7683 639.120 θ = 666.ln ti ln ti+1.382 0.57 I 6.225 7 600 1.ln ti/ Mi 4.6359 0.41 77695.680 Mann’s Staistic Characterstic 666.622 0.1921 6 750 1.643 5 410 1.1764 0.85 228575.5952 -0.098 0.584 0.120 6.76 19923.2538 36549.56 0.2754 0.819 -0.Table 1: For Weibull Distribution MLE and Mann’s Test Statistic (For Lathe Machine .9748 Fcritic 4.145 4 320 1.8149 0.293 0.461 0.7282 0.6200 8.510 0.8928 0.0646 0.34 0.5952 -2.721 6 540 1.1936 4 570 1.92 0.77 374106.90 0.2478 0.241 0.59 0.3969 7.6114 1.45 0.89 7228.120 8269.5952 -1. NIT Durgapur .57 Mann’s Mann’s Staistic Characterstic = 0.945 300 1.1665 Staistic = 0.474 0.1665 Fcritic 6.438 0.2915 1148.0161 843.601 0.94 0.28 0.8928 -0.1) Failure Rank Weibull Parameters tiβ tiβ* ln ti 1/β Zi Mi ln ti+1.860 0.6200 67624.819 -6.28 0.120 2 170 3 β 1.2) Failure Rank Weibull Parameters tiβ tiβ* ln ti 1/β Zi Mi ln ti+1.484 1.87 0.79 Mann’s Staistic Characterstic 4 1 From the above table we can conclude that: For Lathe Machine Machine – 1 For Lathe Machine – 2 β = 1.5952 -0.19 0.28 Failure Time (ti) β ln ti 1 90 1.444 0.0918 0.120 5.43 694.63 Table 2: For Weibull Distribution MLE and Mann’s Test Statistic (For Lathe Machine .7037 594.8928 -0.5220 57360.680 6.39 3688.416 0.031 0.3456 42645.680 6.99 0.351 0.751 -0.63 θ = 352.8928 -2.191 0.751 -6.5952 0.16 270609.4027 5 680 1.811 0.680 6.

3969 0.6029 Weibull Parameters β = 1.4278 7 600 0.5781 6.5220 0.656 3 520 0.794 43.281 42.601 39.824 5.256 SUM 36.5003 37.1093 6.2732 -2.7037 -0.287 Failure Rank Straight Line Coefficient a = -24.6029 θ = 584.3456 -0.051 -2.649 2 290 0.7702 5.5079 32.52 35 of 446 Copyright @ Department of Electrical Engineering.0083 36.301 -14.1524 θ = 333.6200 0.72 Ґ1.0161 0.2297 6. 9 [1]) A So.62 = 584.147 40.89592 ( From table A.1) Failure Rank Weibull Parameters Failure Time (ti) R(t) = 1 – [(i0.155 24.Table 3: Weibull Reliability and MTBF (For Lathe Machine .4)] X= ln(t) Y = lnln (1/R(t)) X2 X*Y Weibull Parameters β = 4.1256 -9.536 1.758 4 570 0.5322 -4.4218 6.2915 0.9427 (13) 4 1 MTBF = θ Ґ(1+(1/β)) (14) MTBF = 584.3648 6.266 -0.4904 SUM 39.3)/(n+0.1242 b = 4.2538 -0.72 * 0.6874 -0.6351 5.9054 4.0588 -0.002 222.4)] X= ln(t) Y = lnln (1/R(t)) X2 X*Y 1 140 0.3665 33.5829 2.812 -3.1524 Weibull Parameters Failure Time (ti) R(t) = 1 – [(i0.3754 -6.2) 1 I 90 0.7683 -0.9328 5 680 0.2656 6.1140 5 410 0.9078 AVG SUM 6.3088 20.1934 0.3)/(n+0.2482 -3.7897 32.4846 Straight Line Coefficient a = -10.5042 4 320 0.6698 -1.7322 -1.009 AVG SUM 5. R(t) = exp(-(t/θ))β R(t) = 0.9416 -2.832 6 750 0.1357 -1.7343 5.8583 40.3428 26.3859 39.86 hrs.4998 -2.0945 6.3528 -3. NIT Durgapur .4627 2 170 0.8906 4.2123 b = 1.146 -6.8962 3 300 0.5553 229.9203 5.5000 5. S E Table 4: Weibull Reliability and MTBF (For Lathe Machine .72 Reliability( according to Weibull Distribution).110 -3. MTBF = 523.0499 6 540 0.174 32.419 -10.

Benski.44 (2). A S E I 36 of 446 Copyright @ Department of Electrical Engineering.90852 (From table A.9932 MTBF = θ Ґ(1+(1/β)) (13) MTBF = 333. SUS College of Engineering& Technology.(Tata McGraw-Hill Publishing Company Limited. [2] Kumar Amit and Lata Sneh. thesis IIT Delhi.D. MTBF = 303.RL1(t) * RL2(t) Rsys (t) = 0. Journal of Reliability and Statistical Studies.9427*0. 1988. 199-202. Journal of Scientific & Industrial Research Vol. 63. [5] Jingyi. Based on the analysis of the machining system and values obtained for these systems are used for the further improvement in the efficiency of the system. Ph. (Online):22295666 Vol. [6]. so increase in MTBF will improve the Availability of the system..9427+0. Tangri.24 = 584. Also since shape parameter is greater than one so we can say that failure rate is increasing. ISSN (Print): 0974-8024. 4.52 Ґ1. Since Availability is directly dependent on MTBF. MLE estimation and Goodness of fit test is conducted for both the machines and is shown in table 1and table 2.E. F. June 2004. R(t) = 0. Issue 2 (2011): 43-55. 4 1 References [1] Ebling C. Mohali. If the reliability of the two systems is increased then the reliability of the system is increased. [3] Sukhwinder Singh Jolly and S S Wadhwa (2004).9932 = 0. and Cabau.Reliability (according to Weibull Distribution). [4] Kurian KC. pg 187-196. "Unrelated Experimental design in Reliability Growth Programs"... An Introduction to Reliability and Maintainability Engineering. “Reliability. pp.(2000). “Reliability and availability analysis of repairable system using discrete event simulation”. Weibull Reliability and MTBF analysis is shown in table 3 & table 4. PP 512517. 23-25 January(2008) International Conference on Recent Developments in Mechanical Engineering. Organised by Deptt of Mechanical Engineering.2000). 9 [1]) So.9932-0. 1995. So Overall Reliability of the system is given by Rsys (t) = RL1(t)+ RL2(t) .72 * 0.(2011). NIT Durgapur . Zhuoru Chen and Yiqun Wang . C.009 hrs. [7] Singhal Piyush. Availability and Maintainability study of high precision special purpose manufacturing machines”. “The development and prospect of Hydraulic Reliability Engineering”.(2008)“Reliability availability and Maintainability Analysis of Machining System”. Probability in engineering and information science(USA) 1998. “ Reliability Analysis of Piston Manufacturing System”.9996 Conclusion Reliability is the most important parameter of a machining system. IEEE Transactions on reliability.

Study of steady-state analysis of self excited induction generator has importance both from the design and operational points of view and has been studied in different literature [513]. The simulated results obtained using genetic algorithm help in understanding the steady state performance of SEIG. All the machine parameters except magnetizing reactance are assumed to be constant. load absence of synchronizing equipment. equivalent circuit. Roy Dept of Electrical Engineering Dept. of Electrical Engineering National Institute of Engineering National Institute of Technology National Institute of Technology Durgapur. terminal voltage. Central grid connection is expensive or difficult to provide in remote areas but small-scale autonomous power systems may be developed for supplying the local customers thereby reducing the cost of the distribution lines and increasing the availability of electrical power. iterative method [9]. ii. reduced maintenance cost. GA. absence of dc power supply for excitation. Per unit resistances and leakage reactanes of stator and rotor are taken to be equal. The time harmonics and mmf space harmonics in the induced emf and current waveform are neglected. The equivalent circuit of SEIG has been analyzed by loop impedance method to develop circuit equation which is nonlinear in nature. In case of SEIG physical system is converted to different models and then the Loop impedance technique used by Murthy [5] and Malik [6] or Nodal admittance technique used by Quazene [7] and Chan [8] is applied to identify the operating point under saturation for the given speed. currents have been shown in this protection capability and better transient performance [2-4]. both the terminal voltage and frequency are not known and have to be calculated for a given speed. secant method [10]. load and capacitance. This can be easily solved by genetic algorithm (GA) optimization technique. India Durgapur. India Durgapur. iii. INTRODUCTION II. K. capacitance and load impedance. NIT Durgapur . SEIG becomes one of the important research areas in the field of renewable energy sector and isolated power systems due to the following although the PERFORMANCE ANALYSIS The following assumptions are made in this study i. Keywords—SEIG. The self-excited induction generator (SEIG) is a strong candidate to harness these renewable energy sources like small hydro and wind energy resources. An SEIG is normally a squirrel cage induction machine excited through an externally connected ac capacitor bank [1]. Mahato N. inherent short-circuit frequency. Ray S. The effects of capacitance and load on terminal voltage. The renewable energy resources play a major role in rural electrification and industrialization programs. N. Only one higher degree non-linear equation as the function of per unit frequency and magnetizing reactance is solved by genetic algorithm (GA) optimization technique very easily. good over-speed admittance and capacitance on magnetizing reactance. frequency. of isolated 3-phase self excited induction generator (SEIG) feeding balanced static resistive load. paper. Abstract— This paper analyzes the steady state performance reduced size. A schematic of shunt capacitor self excited induction generator has been shown in the Fig. etc.Genetic algorithm based performance analysis of 3phase self excited induction generator S. of Electrical Engineering Dept.com advantages such as brushless construction (squirrel cage rotor). The effects of load. In present paper the per phase equivalent circuit of self excited induction generator has been analyzed by loop impedance method and two higher degree non-linear equations are developed. symbolic programming method [8]. capability. Two higher degree non-linear equations as the function of per unit frequency and magnetizing reactance are generated by different conventional methods like NewtonRaphson method [5]. In an isolated power system. terminal voltage and magnetizing reactance. India raysambaran@gmail. List of Symbols R1 = Stator resistance per phase R2 = Rotor resistance per phase RL = Load resistance per phase X1 = Stator leakage reactance per phase X2 = Rotor leakage reactance per phase XL = Load reactance per phase Xm = Magnetizing reactance per phase Xc = Shunt excitation capacitive reactance per phase C = Shunt excitation capacitance per phase a = per unit generated frequency b = per unit speed YL = Load admittance per phase ZL = Load impedance per phase Eg = Generated air gap voltage per phase VL = Load voltage per phase 4 1 A S E I I.1 37 of 446 Copyright @ Department of Electrical Engineering. current and other different issues are investigated in steady state analysis.

With Eg/a. equation (1) is only valid for Zloop to be zero. Is = Eg / a − jX c RL + jX L 2 R1 a a + jX 1 + R − jX a c L a + jX L + a 2 4 1 2. Xc.2. only one equation is enough for solution with the help of GA technique very easily. speed and load the equations can be solved. 7. 1. real and imaginary part of it should be separately zero.1. RL and machine parameters known. To know the value of Xm and a. Expressions for the respective variables are given below: Fig. Active power output per phase 38 of 446 2 Copyright @ Department of Electrical Engineering. Active power input per phase Fig. calculation of the load voltage VL . Is.2. VL = aI L Z L . ZL = 6. a.a) = − ( n1 X m + n2 ) a 4 + ( n3 X m + n4 ) a3 + ( n5 X m + n6 ) a 2 − ( n7 X m + n8 ) a − n9 = 0 (4) The values of the coefficients m1 to m8 and n1 to n9 are given in appendix. Ic = Is − IL − jX = R2 + jX ║ 2 c + R 1 + jX + jX m ║ R2 + jX 2 1 2 a −b a a a −b 5.a)= − ( m1 X m + m2 ) a 3 + ( m3 X m + m4 ) a 2 + ( m5 X m + m6 ) − ( m7 X m + m8 ) = 0 (3) 8. NIT Durgapur (5) (6) (7) (8) (9) (10) (11) . From this we get the following two non-linear equations with ‘Xm’ and ‘a’ as two unknown variables on simplification. − Eg a Ir = R2 + jX 2 a −b 3. shunt capacitance. For a given value of machine parameters. IL. Ir. input as well as output active and reactive power is straightforward using the equivalent circuit of Fig. We select the first equation and solve to get the value of ‘Xm’ and ‘a’.N ( X m . Where Iloop = loop current and (2) Zloop to be zero . Xm. b. 1. Steady state equivalent circuit of self-excited induction generator with R-L load S E I Above circuit can be reduced to single loop circuit and by using loop analysis we get the expression as Iloop Zloop = 0 A (1) Since Iloop must be non-zero for energy conversion. Schematic of SEIG The per phase steady state equivalent circuit of self excited induction generator with balanced static R-L load is shown in Fig. RL + jX L a 2 R Pin = − a I r 2 a −b M ( X m . − jX c a2 IL = Is RL + jX + − jX c L 2 a a Zloop = loop impedance 4.

p. Maximum limit of Xm indicates the unsaturated value (i. 4. XL. Variation of magnetizing reactance with load admittance at u. The stronger strings will last and mate with other stronger strings to produce strong offspring. Finally one becomes best of all. III. X1=0. Step 7 : If M(Xm.u.8025 C = 1.0.u. C = Capacitance b = 1. 1. X1.f. number of generations (Gen) and tolerance (€). It is observed that ‘a’ decreases almost linearly as YL increases for any particular value of Xc.0 C = 0. S E I = −0. delta-connected squirrel cage induction machine whose per-phase equivalent circuit parameters in per unit are : R1 =0. Following are the steps involved for the study of performance of SEIG using GA optimization technique: Step 1 : Read R1. Frequency drops by 2.u and power factor (p.5 1 Load admittance (YL) in p. 230 V.u.0 p.u.6667 p. RL.42%. with b = 1.f. The magnetization curve was determined from a synchronous speed test. 0.5 Fig.u). Step 2 : Declare ranges for a and Xm Step 3 : Select population size (PS). 39 of 446 3 Copyright @ Department of Electrical Engineering.u and power factor (p. As the value of C increases.88% and 7. As the value of capacitance (C) increases the value of Xm decreases. X2.8025 2 Fig. number of chromosomes (Chm). Genetic algorithms are good at taking larger.0 p.6667 1.94 0.u.u).0 p.0563. generated frequency Vs Load ammittance 1 C = 0.22% of no-load value for corresponding values of C = 0.6667 p.p. Genetic algorithms (GA) are computerized search and optimization technique based on the mechanism of natural genetics and the Darwinian theory of evolution i.8025 and 1. shows the variation of per phase generated frequency (a) as load admittance (YL) varies at three different capacitance (C) i. 4-pole.f = 1.e C = 1. Eg/a Vs. Xm.958 1. 7. 4.f = 1.e C = 1. R2 =0.5 0 0 0. potentially huge. search spaces and navigating them looking for optimal combinations of things and solutions which we might not find in a life time [14]. The approximate equation of the curve is as follows Eg Variation of Xm 2.5 Fig. 1.00078687 X m2 − 0.8 A. Rechenberg in 1960 in his work Evolutionary strategies.0 0. b.0 0.5 4 1 A RESULTS AND DISCUSSION =0 1 Effect of excitation capacitance on generated frequency under unity power factor load (quantities are in p. M agnetiz ing reac tanc e (X m ) in p. a C = 1.0 p.u. (12) 2. R2.e Xm ≥ 1. with b = 1.0873. Step 8 : If Chm ≤ PS take the variables from the next Chromosomes and go to step 6.4. a) ≤ € go to step 10.92 0 0.Pout = I L RL 2 GENETIC ALGORITHM The idea of evolutionary computing was developed by I.u. Else go to next step. GA creates strings of binary digits and evaluates each string’s strength in terms of fitness value. Variation of generated frequency with load admittance at u.0.4212 C = 0. Step 6 : Evaluate the objective function M(Xm. 3 shows the calculated values of per phase magnetizing reactance (Xm) as load admittance (YL) varies at three different capacitance (C) i.f) = 1.5 Effect of excitation capacitance on magnetizing reactance under unity power factor load (quantities are in p. generated frequency (a) in p.0.96 C = Capacitance b = 1. respectively. Step 4 : Set Gen = 1 Step 5 : Generate initial population with two sets of chromosomes for respective variables.3.98 0. value of ‘a’ decreases very slightly for particular load admittance but the range of operation increases.8025 and 0.0562 X m + 2. NIT Durgapur .e “survival of the fittest”.6667 0.0 p.0 C = 0.6667. It is observed that Xm increases almost exponentially as YL increases for any particular value of C.0563. Else go to next step.0873 and X2=0. XC.958 ≤ X m (13) Observations are as follows: 1. 0.000004043 X m3 + 0.f) = 1.5 1 Load admittance (YL) in p. Step 11 : Stop IV. Step 10 : Print a and Xm and compute generator performance. 2.958) which makes the system unstable.8025 and 0. X m ≤ 1. In this paper the simulated results refer to a 3-phase. 0.0. Fig.2 kW. Step 9 : Generate new set of chromosomes using GA process. a) from equation (3).

C = Capacitance b = 1.u Stator current (Is) in p.4 0.0 0 0.0 p. The nature of curve shows that no-load IS is very high and its value depends on C. No-load voltage is higher for higher value of capacitance.f = 1.6 1 0.6667 1. Variation of stator current with load admittance at u.2 0.8 0.0 1. If ‘C’ is high no-load Is is also high.6667 0. stator current also increases gradually and reaches its peak value for a fixed value of C. Voltage regulation is better if we increase the value of C.8 Power output (Pout) in p. 5 the variation of stator current (IS) w.f = 1.0 p.0 p.p.5 Load voltage (VL) Vs Power output (Pout) 1.0 has been shown.0 p.u 1 Fig. 0. 1 C = 0.6 0.4 C = 1.0 p. 1.0 p. C = 0.0.t load (Pout) at different capacitance (C) with b = 1.6667 unity power factor load (quantities are in p. Effect of excitation capacitance on stator current under unity power factor load. 4.8025 Fig.0.4 1 C = 0.r. Load voltage VL Vs Load ammittance YL 1.0 1 C = 0. Variation of load voltage with load at u.u and p.7.4 1. If the capacitance increases and IS increases for fixed value of load admittance. The nature of curve shows that no-load stator current (IS) is very high and its value depends on capacitance. Variation of stator current with load at u.2 0.2 0.u and p.4 C = 1.6667 C = 0.6 0. If YL increases further stator current decreases. 8. When YL increases.0 C = 1.8025 C = 0. V. C = Capacitance b = 1.0 In the Fig.2 C = 0.0 CONCLUSION A general analysis using genetic algorithm (GA) optimization technique has been applied to predict the steady- Effect of excitation capacitance on load voltage under unity power factor load 40 of 446 4 Copyright @ Department of Electrical Engineering.p.4 C = Capacitance b =1. If the capacitance increases VL increases and range of operation also increases. 5. VL drops as the YL increases.8025 0.0 p.f = 1.f.f. Effect of excitation capacitance (C) on per phase load voltage (VL) has been shown in Fig.8 0.u. 6.u).5.f.4 Load voltage VL 1.r. 7.t load admittance (YL) at different capacitance (C) with b = 1.2 0.6 0. S E I 0.5 4 1 Stator current (Is) Vs Power output (Pout) 0. 1.u.3.f = 1. If C is high.u and power factor (p.2 Stator current Is Vs Load ammitance YL Stator current (Is) in p.f = 1. at b = 1.u).6 0 0.6.8 0 Fig. at b = 1.u 1.f) = 1. Variation of load voltage with load admittance at u. In Fig.8. 6 the variation of stator current (IS) w.f = 1.8 Power output (Pout) in p.0 p. Fig.0 0 Effect of excitation capacitance on load voltage under unity power factor load (quantities are in p.8 0.f.p. Voltage regulation is better for higher values of C.u C = 0.f) = 1.0 1.4 Effect of excitation capacitance on stator current under 0.4 0. If the value of ‘C’ is increased no-load stator current is also increased. no-load Is is high. A Load voltage (VL) in p. If the load is increased terminal voltage drops and if the load is increased beyond its maximum value both load and voltage are decreased.4 C = 1.u and power factor (p.5 1 Load admittance (YL) in p. If the capacitance increases VL increases and range of operation also increases.0 has been shown.6 0.p.8025 1 C = Capacitance b = 1. NIT Durgapur .5 1 Load admittance (YL) Effect of excitation capacitance (C) on per phase load voltage (VL) has been shown in Fig. No-load voltage is higher for higher value of capacitance.

Quazene and G.B. P. 54. 41 of 446 5 Copyright @ Department of Electrical Engineering. Malik and S. K.." IEEE Trans .. 1982. 2793–2798. Electrical Power Applications. Prabhu. 2003. PP. No.. 1990. n3 = bn1. Malik. n4 = bn2 . Issue. Vol.6. “Analysis of self excited induction generator. McPherson Jr. S.c. S. "An autonomous wind energy converter using a self-excited induction generator for heating purposes. on Energy Conversion. 129. 5. on Energy Conversion. 2009. 1994. Patton and D. Y. Syst. NIT Durgapur . J. Watson. “Steady-state analysis of self-excited induction generators”. m3 = b ( m1 − R2 X L ) .S. S.1. A. 1979. A. 10. n9 = R2 X C ( R1 + RL ) . Syst. 2. Arrillaga and T. Drives and Energy Systems for Industrial Growth. This analytical technique is very efficient requiring very little computational time. S. Milner and D.” IEEE Trans. Rajasekaran and G. C. PP. 3. VI. 3. “Analysis of series compensated self excited induction generators for autonomous power generation”. Chan. Basset and F. Potter. M. 133-139. PP. No. IET. “Steady state analysis and performance of an isolated self excited induction generator”.. PP. Inst. Vol. APPENDIX m1 = RL ( X 2 + X 1 ) + X L ( R1 + R2 ) . REFERENCES n1 = X L ( X1 + X 2 ) . [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] E. IEEE Trans. Wang and Y.687–693. 102. F. Vol.P. on P . Neural Networks. No. 1935. No. Jabri and A. “ Capacitive excitation of induction generators ”.12. Murthy. Vaishya. K. 6."Controllable d. 2. A. A S E I VII.”Proc. m4 = bX 2 ( X L R1 + RL X 1 ) . Berg. Eng. D. K.260– 265.126. Vol. Vol. PAS-103. m7 = bX C ( R1 + RL ) . Aug. IEEE Trans. Trans. J. pp.. Vol. Huang. A. Vol.“Steady state analysis of capacitor self-excited induction generators. H." Proc. 540 – 545.A. A. E. Tandon. No.1245-1248. Tandon and M. n6 = X C X 2 ( X1 + X L ) + RL ( R2 X1 + R1 X 2 ) + R1R2 X L . K. Tandon. The results show that the terminal voltage and stator current increase significantly with increase in capacitance value. m5 = X C ( R1 + R2 + RL ) . 3. n7 = b ( n5 − RL R2 ) . 1982. and G.m2 = X 1 X 2 RL + X L ( R2 X 1 + R1 X 2 ) . 2. PP. 1995. 9.” IEEE Trans. on Energy Conversion.B. IEE. No. T. The steady-state equivalent circuit can then be used to compute the performance. S. Vol. " Wind Engineering Vol. “Analysis of self-excited induction generators using iterative method”. vol. S. 1984. Mar. Power App. 1996. Amer. PP. 1. It calculates the values of saturated magnetizing reactance and the output frequency for the given capacitance . L. J. 1983. PP. Chan. I. C. n2 = X L X1 X 2 . D. Proceedings Of Internation Conference on Power Electronics. on Energy Conversion. IEEE Trans.19-23. F. I. 5. N. PAS-8. pp. Power App. IEEE Trans.. Nov. Murthy. No. S. Vijayalakshmi Pai. PP. m 6 = X C { R 2 ( X 1 + X L ) + X 2 ( R1 + R L )} + R1 R 2 R L . Inst. J. 502-507.3957-3966. Vol. and A. Fuzzy logic and Genatic Algorithms. m8 = X 2m7 . O. 612–618. 1986. "Analysis of utility protection problems associated with small wind turbine interconnections . pp. n5 = X C ( X L + X1 + X 2 ) + RL ( R1 + R2 ) . 1982 .. Vol. Elect. “ Limits on the performance of the three phase self-excited induction generators”.10. S. Haque . PP. O. Murthy. Densem. 288-296. PAS-101. “Analysis of an isolated induction generator. Curtice. speed and load. state behavior of isolated self-excited induction generators feeding static resistive load. Alolah. 4 1 n8 = b{n6 − R2 ( X1RL + R1 X L )} . to an accuracy of 1008 . PP. No. PHI. no. 445-452. T. Eng. No. power supply from wind-driven self-excited induction machines. “ Analysis of a stand-alone three-phase self-excited induction generator with unbalanced loads using a two-port network model ”.B. Electr. Watson. 350-356. 3.

NIT Durgapur . Cpk measures how close a process is running to its specification limits relative to the natural variability of the process. India e-mail. II. India. the less likely it is that any item will be outside the specification .com Associate Professor(Workshop).. roundness) has driven metal cutting industry to continuously improve quality control of metal cutting processes. USL] is the specification interval. Keywords: Machine tool capability and process capability.Cpk. Cpk. I. when shown together with Cpk.K.shukla@gmail. It is used to determine the system tendencies between tolerance limits. Mechnical Engineering Department. Cp. Indian School of Mines Dhanbad(JH). 3 3 Where [LSL.in lower Cpk.how much work is needed to bring the process to its potential.co. Cpmk for machining process has been presented. where Cp was presented by Juran (1974) and Cpk by Kane (1986). Cp and Cpk. analysis of variability relative to product requirement. μ is the process mean and σ is the process standard deviation of the incontrol process Since the indices Cp and Cpk do not take into account of the differences between the processes mean and its 42 of 446 Copyright @ Department of Electrical Engineering. gives a quick estimate of how far the actual process is from its target and thus . A possible way of working towards more accurate machining is by optimizing a process with the help of capability indices. Process Capability analysis can be applied not only to production period but also to a machine or machine tool.Mechanical Engineering Department Indian School of Mines Dhanbad (JH). Cp USL LSL 6 USL LSL Cpk min .vidhika. in short periods of time. N. 4 1 The Cpm index reflects the shifts in the process mean analogously to the way that the Cpk index reflects these shifts. Cpk is higher only when you are meeting the target consistently with minimum variation.Cpk measures how close you are to your target and how consistent you are to around your average performance. To achieve good quality products with minimum variation and to increase productivity and profitability. In this experimental study some statistical calculations have been made to describes how capability indices are calculated and comparative study between different capability indices Cp.Singh2 JRF(PHD). A process capability index is usually indicated by different indices e. Machine capability explicitly refers to the ability of a machine tool to produce very good machine parts. whereas Cp will be high. but he can be away from his target towards one of its specification limit. Cp. The two most widely used indices in industry are Cp and Cpk.nks_221@yahoo. Cpmk.g. it is necessary to all parts are machined with properties close to target values and within defined tolerances and the outcome will have high percentage of useful parts and consequently there would be very less wastage and very less re-machining of parts in order to saves time and money. The larger the index. SPC technique was used to evaluate machines’ capability (Cm=Cp) and process centering Cpk of manufacturing process to find whether the process is capable or not. S E I The machine capability study is a short-term study with the aim of discovering the machine-specific effects on the production process. A capability study is one way of visualizing the ability of a process to produce products according to defined properties of the product.Comparative Study of Machining Processes by Process Capability Indices Mrs Vidhika Tiwari1 Dr. The number of nonconforming part was determined in observed values. This gives the information about changes and tendencies of the system during the production. Abstract— Process Capability indices are indices for describing how well a product is machined compared to defined target values and tolerances. several authors have made significant contributions. Besterfield (1989) defines ‘Process Capability Analysis is a statistical technique helpful throughout product cycle for quantification of process variability. A person may be performing with minimum variation. Kane (1986) defines ‘Process capability index measures relationship between the allowable process spread (natural tolerance) and the actual allowable process spread (specification range)’. specification development activities prior to manufacturing and assistance in development and manufacturing in eliminating variability. In this study. Cpm. CAPABILITY STUDY With the origination of process capability measures since 1942 from the British Standard Institution and pioneering work initiated by Kane (1986) for their development and application. e-mail. which indicate PROCESS CAPABILITY INDICES AND PROCESS A INTRODUCTION The increase of consumer needs for quality metal cutting related products (more precise tolerances and better product surface roughness. Cpm.

02 40. Many PCI generalizations for non-normal processes suggested by Clements (1989).02 40. 18. Process Capability Index versus Defect Level [20] III. Samples were chosen randomly during the TURNING process.03 40 40. 7.02 40.04 39.01 40 40.02 40.02 40 39.02 40 40 40. 2. Kotz and Johnson [5].04 40 40 40.04 40.02 40 39. A single sampling plan was implemented by using the lot-acceptance sampling plan. The results of statistical calculations of the process with data obtained from all four machines and process capability values are listed in Table 2.01 40.01 40.04 40 40. 22.04 40. The data for the investigated characteristic was collected for 25 days in the company. Chan(1988) and Pearn(1992) considered this difference to develop indices Cpm and Cpmk as follows: USL LSL Cpm 2 2 6 T USL LSL Cpmk min .02 40. T represents the target value of quality characteristics. The table does assume that the process is in control.01 40. 25 numbers of sample parts were inspected in each lathe and four different lathe machines are considered for analysis.04 40.02 40 39. Saha and Nanda[13] Clements [22].5 133. 21. to make a clear decision about the capability of the machining process. Table 2: Where. Table 1 provides a few Cp values and their associated defect counts in terms of parts-per-million (ppm) defects.98 40 40.02 40. when μ and σ2 are unknown. The focus of the study was to investigate the process capability of the ongoing process m/c1 m/c2 m/c3 m/c4 1. Another index that is given by Boyles [10] is also worth mentioning. So total numbes of samples are 100. NIT Durgapur . 15.7 0. 11. Part name: Spiral Bevel Pinion Operation: Turning Checking method: Micrometer Make: MITUTOYO Range: 25-50mm Least Count: 0.01 40 40 40.0018 MEASUREMENTS A PPM can be estimated based on empirical data (i. A number of new approaches to process capability analysis have been attempted by Carr [11] and Flaig [17].02 40.02 40 40. The work pieces were machined under dry conditions. Cp PPM 0.02 40. 39. Tolerance: + 0.01 mm Target: 40. The process parameters μ and σ2 are estimated from the sample mean variance for and S2. 14. 10.02 39. and Sommerville and Montgomery [16].01 40.98 43 of 446 Copyright @ Department of Electrical Engineering.04 40.02 40. Pearn and Kotz [19]. 8.02 40. 40 40 40 40 4. is normally distributed. 12.98 40 40 40.02 40 40. 16. and that the specifications are two-sided. ANALYSIS AND APPLICATION OF PCI The work piece Spiral Bevel Pinion was machined using lathe machine in a production organization.04 40.02 40 40. The indices suggested so far whose assessment is meaningful regardless of whether the studied process is discrete or continuous are those suggested by Yeh and Bhattacharya [14]. This index indicates the ratio between the number of pieces exceeding the specification limits and a million produced units.02 40.0 2.98 40.614 1.02 40 40. Maiti.e.02 40 40.98 40 40.target value.96 39.01 40.98 40.01 40 40. 2 2 3 2 T 3 2 T so as to decide upon the suitability of the machine to hold particular tolerance. 20.01 39. 17. In this study.00 mm dia Another process capacity index is Parts per Million (PPM).0 0.02 40. 40.05 mm S E I Table 1. 4 1 Sample Nos.02 3. 9.01 40 40. 19.01 39. the number of exceeding elements over one million) or with the cumulative distributions function of the theoretical distribution.700 1.96 40.98 40 40.34 2.3 96 1.02 40.02 40. the process is centered on the target value. .01 40 40. Pearn and Chen [18] and Mukherjee [12]. 6.05 40. and Perakis and Xekalaki [9].02 40.02 40. Borges and Ho [21]. 5.02 40. 13.

601536 0.01 40 39.33) Fig 3.99 40 40.For a barely capable process this process will produce greater than 64 PPM but less than 2700 non-conforming PPM.379343 1.018723 0.080276 1. Fig 2 RESULT AND DISCUSSION For machine 2 Cp is 1.885438 1.054093 (1 < CP < 1.158648 0. 25.08 0.023274 0.02 40.e.677223 0.000542 0.22274 0.120897 0.946847 0.14129 0.054093 0. NIT Durgapur .102316 1.9832 40.00025 0.33) Fig 4.830693 0.704454 0.716115 and 0.985521 0. This process will produce conforming products as long as it remains in statistical control.00014 0. a significant portion of product will start falling outside one of the specification limits. This should translate into higher profits.02 40. effort must be immediately made to improve the process (i. 4 1 A S E I Fig 3 Fig 1 For machine 4 Cp is 1. A fairly capable process this process should produce less than 64 nonconforming PPM. In an incapable process this process will produce more than 2700 non-conforming PPM.568868 IV.01 40 39.008 40.121805 0. reduce variation) to the point where it is capable of producing consistently within specifications.72 0.885438 0. 39.379343 (CP > 1. 24. This process must be closely monitored.006 SD σ 0.23.716115 1. For machine 1 and machine 3 Cp is 0. It should be noted that if the process mean moves to the left or the right.102316 0.02 Average 40.06 0. It is impossible for the current process to meet specifications even when it is in statistical control.121805 respectively (CP < 1) Fig 1&2. This process has a spread just about equal to specification width.0128 39.158648 0.33 40. 44 of 446 Copyright @ Department of Electrical Engineering.867258 CPL Cpk Cpm Cpmk 0.01 40.60003 1. The process owner can claim that the customer should experience least difficulty and greater reliability with this product.01208 0.601536 0.136831 0.015811 Var σ2 Range R 0.04 Cp CPU 0.98 40.

[11] [12] [13] 4 1 A S E I [14] [15] [16] [17] [18] [19] [20] [21] ACKNOWLEDGMENT The authors remain ever grateful to Shri Sandeep Mittal Manager in HAL Bangalore for his guidance and suggestions. Quality Progress. Statist. Statist. No. 447458(2001). Mukherjee. Z. 279-300. D. Engng. Chapman and Hall. Theory Methods. Quality and Reliability Engineering International. New York 1974. Saha and Nanda. Qual. J..E.. March 2005.643-648. South Korea. 41–52.B. Montgomery DC. Quality Technology & Quantitative Management Vol. 176–187. A fraction defective based capability index. S.and Cpk <1 the machine tool process was not capable it was necessary to stop the process.Simulation and Computation. W. Communications in Statistics . K. Sommerville. Application of Clements' method for calculating second. (1998). Kotz.. Qual. and Chen. Technol.L. 613-643(1994). CONCLUSION Statistical Quality Control has great deal of improvements in companies for the mass production. S. (1996). L. 27(2). A. Yeh. “Bootstrap Confidence Limits for Short Run Capability Indices”. Pearn.M.” On Generalizing Process Capability Indices”. Juran... pp. 17. Ekvall.. Quality Control Handbook. 152(1991). 18 (1986). 1–19.. McGraw-Hill. pp.L. S. 243-249(1995). Rodriguez Recent developments in process capability analysis J. Introduction to Statistical Quality Control. 19 (1990). NIT Durgapur . Flaig. Maiti. Manufacturing planning. Bhosale. 305–316. Vol. 24(8). A.. S. [22] 45 of 446 Copyright @ Department of Electrical Engineering. 2005. Quality Engineering. 3. W.Simulation and Computation. When examining results of Table 2 for the machine 2 both machine tool and process capability were accomplished. It is used to determine the system tendencies between tolerance limits. B. Process Capability analysis gives the information about changes and tendencies of the system during the production. S. London.. pp. machine tool and cutting parameters which are three important components of machinability were reevaluated. process parameters were required to be redetermined.” process capability index for discrete processes”.These obtained values showed that. L. 2002. 364–413 Balamurali. Chou. J. 22. 2010. A robust process capability index. A new approach to process capability analysis. Using the manufacturing data obtained from each line respectively and the machine tool process capability was determined.REFERENCES [1] [2] [3] [4] [5] [6] [7] [8] [9] Fig 4 [10] V. Technol. reexamine the process parameters and rearrange the process in order to reassure machine tool and process stability. 34. In: Juran. Quality Progress. Pearn.N.N. and A. 11. New York: John Wiley & Sons Inc.. Technol. A. S. September 2005. N. Chan. 11–18. In this study. pp. third ed.K. No. The work piece material. P.” Proceedings of the SMTAI Conference. It was determined that the parts were manufactured out of tolerance. Owen On the distributions of the estimated process capability indices Comm. S. Kane Process capability indices J. D. 7. Quality Engineering. and Bhattacharya.J. Clements. In machine 4 while the machine capability was accomplished since Cp>1. 7. Different Results: Methodologies for Analyzing a Stencil Printing Process Using Process Capability Index Analyses. 3. L. statistical calculations for a product manufactured using mass production in a medium sized company were carried out. W. Xiong. R. Theory Methods. Seoul. 4549– 4560. Kotz. 24 (1992). and Ho. Proceedings of the International Conference on Statistical Methods and Statistical Computation. pp. It was detected that the machine tool and process capability for the whole process was inadequate and the mass production was unstable Since in the machines 1 and 3. Cp < 1 . D.C. 9 (1996). London (1993). 1992–2000. A new Process capability index: parts per million. J Qual.L.M. R. 565-589. Santos. 139145(1994-95). and Kotz. S. A. Y. and which made the submission of this paper possible in the present form. V. L. 389-39(1995). J.E. S. (Ed. D. Montgomery Process capability indices and non-normal distributions Qual. Quality and Reliability Engineering International.and third-generation process capability indices for non-normal Pearsonian populations. Process capability calculations for non-normal distributions. Estimating process capability indices for non-normal Pearsonian populations. 9. Quality Engineering. N. Borges. D. Chicago. 2003 Michael perakis and Evdokia xekalakia. Johnson Process Capability Indices Chapman and Hall. “One Process. 95-100(1989). W. pp. Aravamudhan. pp. 18 (1989). Journal of Statistical Computation and SimulationVol. Process Capability Indices—A review. it was determined that the process capability was not achieved because Cpk was less than 1. 23. 75.). 205-211. Zhang On the asymptotic distribution of some process capability indices Comm. Carr. Johnson. Process capability with asymmetric tolerances..15. M. Communications in Statistics . L. Process capability indices and associated inference problems. 175–187 Boyles. J. IL.

India Department of Mechanical Engineering Dr. the system becomes bulky and uneconomical. To satisfy industrial demands. However. Roy Engineering College Durgapur. heat transfer in porous medium has gained considerable attention of many researchers in recent years. porous fin of any geometry transfers high heat rate in comparison to solid fin. convection.com Abstract—In this work. This is justified by the fact that porous media play a vital role in different thermal systems.e. The results indicated that the temperature distribution is strongly depending on the Darcy and Rayleigh numbers. the energy balance and Darcy’s model are used. profile. The concept of using fins made of porous materials was firstly introduced by Kiwan and Al-Nimr [1] and they introduced Darcy model for analyzing the porous fins for the first time. Consequently. 46 of 446 Copyright @ Department of Electrical Engineering. By using same method. To formulate the heat transfer equation for the porous fin. Under a constant heat transfer coefficient and a given temperature difference between primary surface and ambient condition. [2] performed heat transfer study through porous fins with temperature dependent heat generation. Kiwan [6] performed thermal analysis of natural convection porous fins. finite-length fin with insulated tip and a finite-length fin with tip exposed to a known convection coefficient and it was found that the effect of different design and operating parameters such as: Ra number. Da number. Kundu et al. radiative as well as combined convective-radiative straight fins with various profiles considering all the non-linearities. Adomian decomposition method (ADM) was used as solution method . this fact depends upon the magnitude of porosity and flow parameter. the effects of temperature dependent heat transfer coefficient on thermal analysis of porous fin is investigated. B. [3] worked on the performance and optimum design analysis of porous fins of various profiles operating in convection environment. Collocation Method (CM) and Least Square Method (LS) for predicting the temperature distribution. air-cooled craft engines. Bhanja et al. Nevertheless. There is a continuous effort by the designers to determine the optimum fin profile and shape which will maximize the rate of heat transfer for a specified fin volume or minimize the fin volume for a given heat transfer rate. Bhanja and Kundu [4] performed thermal analysis of a T-shaped porous fin. The fins are exposed in natural convection condition. It is illustrated that the magnitude of temperature is increased with decreasing the Darcy number and consequently Sh parameter. The thermal performance of porous fins was then studied for three types of fins: long fin. Darcy model. I. Also they observed that the temperature distribution in the porous fins is highly dependent upon the fin profiles.bhanja@gmail. Fins have proved to be an inexpensive and effective means of enhancing heat transfer between a hot surface and its surrounding fluid. convex and triangular types are chosen. It is observed that the exponent associated with heat transfer coefficient has direct effect on the efficiency of the fins and rectangular profile surpasses all the other profiles in study for the best efficiency. NIT Durgapur . light weight. the rate of heat transfer depends mostly on the surface area of the fin. 4 1 A Keywords—porous surface. The effects of various geometric and thermophysical parameters on the dimensionless temperature distribution and fin efficiency for all the profiles are studied in a comparative way that may help to choose best possible geometry. increasing surface area will result in increasing overall dimension of the equipment and thereby increasing material cost and size of the equipment. An analytical technique called Variational iteration method (VIM) is proposed for the solution methodology. Fins find extensive application in airconditioning. VIM. India dipankar. He used energy balance and Darcy’s model to formulate the heat transfer equation. This study is based on finite-length fin with insulated tip. rectangular. A large amount of work has been done in the field of solid extended surfaces regarding the evaluation of thermal performance and efficiency of convective. C. [5] developed an analytical solution to determine temperature distribution and fin performance of a porous pin fin used for electronic cooling.They made a few remarkable observations i. The use of porous fins is an excellent passive method for providing high heat transfer rates for electronic components in a small. The basic philosophy behind using porous fins is to increase the effective surface area through which heat is convected to the ambient fluid. Also using Al as a porous fin’s substrate leads to higher value of temperature rather than Si3N4 element. Hatami et al. Recently. low maintenance and energy free package. Three different profiles namely.Heat transfer analysis in porous fin of different profiles using Variational iteration method Subhasis Chakravarthy. refrigeration. Dipankar Bhanja Pabitra Kumar Mandal Department of Mechanical Engineering National Institute of Technology Silchar Silchar. cooling of computer processor etc. They chose Si3N4 and AL as fin materials and used Differential Transformation Method (DTM). INTRODUCTION (Heading 1) S E I Heat transfer analysis has been the quintessential part of study in all scientific fields and extended surface (fins) heat transfer is the field that is grabbing attention for the past few decades.

The method is proved to be powerful for solving any class of nonlinear governing equations without adopting linearization or perturbation technique. the temperature distribution and its efficiencies are easily being determined for a wide range of design variables of porous fins. A Dimensionless tip temperature Cp Specific heat of the fluid passing through fin pores Da Darcy number g Gravity constant h Heat transfer coefficient over the fin surface hb Heat transfer coefficient at sink temperature k Thermal conductivity KR Thermal conductivity ratio K Permeability of the porous fin L Length of the Fin m Exponent of heat coefficient parameter n Variable fin profile index Nu Nusselt number P Perimeter of the Fin q Actual heat transfer rate per unit width Q Dimensionless actual heat transfer rate per unit width qi Ideal heat transfer rate per unit width Qi Dimensionless ideal heat transfer rate per unit width Ra Rayleigh number tb Thickness at fin base T Temperature Vw Average velocity of fluid passing through porous fin W Width of the fin x Axial length measured from the tip of the fin X Dimensionless axial distance β Coefficient of thermal expansion η Efficiency of the Fin φ Porosity or void ratio ψ Thickness to length ratio θ Dimensionless temperature γ Kinematic viscosity ρ Density of the Fluid 4 1 A S E I The present work studies analytically the heat transfer phenomenon through porous fins for three different profiles. Kiwan [7] studied the effect of radiation heat transfer on the natural convection heat transfer from a porous fin attached to a vertical isothermal surface. NOMENCLATURE Subscripts a Ambient condition b Base s Solid properties 47 of 446 Copyright @ Department of Electrical Engineering. Das and Ooi [9] have predicted the multiple combination of parameters for designing a porous fin subjected to a given temperature requirement. Kr. NIT Durgapur . and convection parameter in conventional fin were discussed suggesting that by increasing “S” the heat transfer increases but in some cases it has exemption. no such article is found on porous fin of various geometries other than rectangular shape except the work performed by Kundu et al. It was found that as the surface temperature parameters increase the radiation effect becomes important and as Ra increases the radiation effects become less important. Thus. convex and exponential were considered in that analysis. It provides an analytical solution in the form of power series where the temperature on the fin surface can be expressed explicitly as a function of position along the length of the fin. variation of heat transfer coefficient with temperature has not been considered in that work. and the surface-ambient radiation parameter. He adapted Rosseland approximation for the radiation heat exchange and the Darcy model for simulating the solid–fluid interactions in the porous medium. rectangular. and length to thickness ratio on the temperature distribution along the fin is grouped into one newly defined parameter called Sh. [3]. The effect of varying Rayleigh number. Ra . They concluded that porous fins provide much higher heat transfer rate than traditional fins. triangular and convex with temperature dependent heat transfer coefficient with polynomial variation. Their study is based on finitelength fin with insulated tip. the surface temperature parameter. to the best of author’s knowledge. II. rectangular. Bassam and Hijleh [10] investigated a problem of natural convection from a horizontal cylinder with multiple equally spaced high conductivity permeable fins on its outer surface. whereas further increase in these parameters adds no improvement to the fin performance. were presented. The effect of the variation of Sh on the porous fin thermal performance was established. it has been observed that a considerable amount of work is present about porous fin. He concluded that increasing the fin length and effective thermal conductivity enhances the heat transfer from the fin up to certain limit. From the brief literature survey mentioned above. the radiation–conduction parameter. The effect on porosity parameters “S” and convection parameter in porous fin “n”. R2. Variational iteration method (VIM) [11] is used as an analytical method for solving the non-linear governing differential equation. However. Saedodin and Olank [8] investigated the temperature distribution in porous fins in natural convection condition and compared it with temperature distribution in conventional fins.thermal conductivity ratio. θb. Three different profiles namely. Rd. However.

In this solutions for some weell-known non--linear problem t problems aare initially appproximated wiith possible method. Heaat is traansferred from m the fin surfa face through conduction c andd conveection. and FOR RMULATION OF THE PROB BLEM The problem prresented here is the one-dim mensional heaat transffer in longituddinal porous fiins of rectangu ular. S E I Introd ducing the folloowing non-dim mensional param meters: 0 =0 0 (6) (7a) 1 (7b) VARIATIONAL ITERA ATION METH HOD VIM [11] is an anallytical method which gives appproximate ms. and g(x) is an inhomogeneouus term. The poro ous medium is considered as homogeneouss. the unknown ns. The governing differential equation off longittudinal thin fin ns is as follows: ∆ 1 φ Consiidering all thee dimensionlesss parameters. The ambieent temperaturee a to bee Ta iss taken as connstant and thee fin tip is assumed insulaated. 1. L is a linnear operator. andd . triangularr and convex c profile with the lengthh L as sketched d in Fig. The conv vective heat tran nsfer coefficiennt is consideredd to be function of tem mperature of th he form: Ω φ θ φ θ ω θ2 ⁄ V. ω . Then a corrrection functioonal is construucted by a general Lagrange L multipplier. Eq.e δ δu u this form mulation . Lan ngrangian multiiplier λ (t) is obbtained as folloows: λ 48 of 446 Copyright @ Department of Electrical Engineering.ω . 2 . Ω . By using restricted d variation i. which caan be identifiedd optimally via the variational v theoory. thee following differentiial equation maay be considereed: 4 1 A (4)) l semi-fin thicknesses fo or rectangular. ψ. and N is a non-linear operator. IV (3)) . NIT Durgapur (13) .f Fluid propertties eff Effective prooperties . . which may m affect thee buoyaancy term wheere Boussinesq q approximation is employedd. triangular andd The local conveex fins are giveen by ω θm+1 1 (2)) W Where. Ω . . Porous fins of different d profiles (aa) Rectangular (b) Triangular nvex (c) Con λ t @ @ 0 0 (11) (12) By soolving (10) witth respect to thhe boundary coonditions in (12). The physicaal propeerties of solid as well as fluiid are considerred as constannt excep pt density varriation of liquuid. To siimulate the intteraction betw ween the porouus medium andd fluid Darcy formulaation [7] is used. Basedd on VIM. (6) is subjected to thhe following booundary condittions: wheree. Accordingg to VIM. ψ θ . and respectively. As the fin is porous. the subscript s n denotes thhe nth order appproximation and a u is conssidered as a 0. Nu . (2) is reduced to: t (1)) ∆ 1 (8) Wherre.. it allows fluuid to penetratee throu ugh it.1. isotroopic and saturaated with a single phase fluidd. a differentiial equation forr obtaining the Lagrangian muultipler λ(t) is λ" t 0 (10) With corresponding conditions. λ t 1 Fig. (5) Eq. a correctionnal formula cann be constructeed as: λ t ξ ξ g ξ dξ (9) wheree λ is a generaal Lagrangian multiplier whhich can be identified d optimally viia variational theory . hb is the heaat transfer coeffficient at the convection c sinkk tempeerature. Da III. .

1. (7a).0 0.4 (16) X 2 ω ω A 4 0.8 1. It is clear from the figure that rectangular fin profile provides highest tip temperature whereas the variation of dimensionless temperature is considerably high for the triangular profile. it is solved by Gauss-Seidal iterative method by satisfying desired convergence criteria.5. φ =0.7 As a starting approximation for VIM solution. 2. m = 0 with the corresponding results for temperature-dependent convective heat transfer coefficient of exponent m = 0.0 0. -4 Ra=10 . To validate the present work.3 φ =0.5 (15) ω A convex rectangular triangular Num erical work 0.8 (19) θ Thus. first two iterations are: θ -4 3 ω θ X X X ω θ 4 Ra=10 . Comparison of dimensionless temperature distribution among various profiles predicted by analytical work and numerical work. and consequently the convective portion of surface heat loss is reduced in spite of the reduced porosity. 2 is drawn for the validation of the present analytical model. NIT Durgapur .2 0. X Ψ 0.4 Fig.1. ω A X 2 X 0. θ is assumed as constant.6 using Taylor series central difference scheme. Nu=10.9 θ θ 1.2 X ω A X 8 ω ω A 8 X 0.6 A 2 ω A ω A 4 ω A X 2 ω A ω ω A X X A AX A 0. Such a starting approximation automatically satisfies one of the boundary conditions given in Eq. Da=10 .2 0.6 ψ =0. Finally.3 0. . The fin surface temperature decreases with the increase of porosity parameter. (2).0 X Fig. a high porosity value reduces the effective thermal conductivity of the fin due to the removal of the solid material. results mainly temperature distribution and fin efficiency are generated with a wide range of thermo-physical and thermo-geometric parameters.m =0.e. Figure compares the temperature distributions in fins with a constant heat transfer coefficient hb. φ =0.25 0. Fig.0 (17) Similarly the higher order terms can easily be calculated using recursive relation expressed in Eq.5. convective heat transfer increases due to more fluid passing through pores.If the above VIM formulation is applied to (6) the following iteration formula is obtained X θ t X dt V. 0. (14). actual heat transfer rate per unit width is determined by applying the Fourier’s law of heat conduction at the base and can be expressed in dimensionless form as Ra Da 0.. Ideal heat transfer per unit width can be expressed in dimensionless form as + 0.0.0 Fig. This figure also compares the temperature distribution among various profiles.K R =10 3 m =0. 3 illustrates the effect of the exponent m on the temperature distribution in rectangular fin with variation of porosity parameter.4 RESULTS AND DISCUSSIONS Based on the above analysis. Effect of power factor m on temperature distribution for rectangular porous fin with the variation of porosity parameter 49 of 446 Copyright @ Department of Electrical Engineering. 3. At the same time. Actually.0 0. ψ 2 Nu ψ 1-φ (18) 4 1 1.4 0.K R =10 . It is clear that the temperature expressions are a function of unknown tip temperature A which can be determined by putting boundary condition (7b) and solving by Newton-Raphson method. Applying VIM in Eq. SOLUTION WITH VIM ψ =0. Combining these two effects result in the decrease in lower tip temperature.0 4 VI. fin efficiency can be calculated as : η= ⁄ 1. 0.8 A S E I After obtaining temperature distribution as a spatial function. a difference equation is formulated by discretization of governing equation of the present problem φ =0.6 0. The latter situation results in higher fin temperature because the average convective heat transfer over the surface of the fin is less than hb.25 (20) 0.9 m =0.8 (14) θ θ A n n A 0. Nu=10. i.0. Da=10 .25.

1 0.0. Similarly. ψ=0.2 ψ=0.0. High porosity presents a lower temperature distribution with the subsequent values of m = 0 . 4.0. On the other hand.φ=0.8 0.2 0. Fig.1. Variation of fin efficiency of rectangular porous fin with the variation of Ra.0 4 -4 3 ψ=0.3 4 0. 4 illustrates the effect of the exponent m on the temperature distribution in triangular fin with variation of porosity parameter.25 0. Nu and m. as the value of Darcy number increases the slope of the η-Da curve decreases.2 0.25 VII. 5.4 0.2 m=0.25 and porosity high.KR=10 0. η S E I 1. Increasing Ra improves the convective heat transfer coefficient between the fin and the working fluid and thus actual heat transfer rate is enhanced.1 Ra=10 5 m=0. 0.1.0004 0.04 0. 6.5.1.0 4 The variation of fin efficiency of rectangular porous fin with the variation of Ra.0010 X Da Fig.0002 0.6 0. Fig. Although with m = 0. Da=10 . triangular and convex with temperature dependent heat transfer coefficient using 50 of 446 Copyright @ Department of Electrical Engineering.5.0006 0.8 0.3 φ=0. 0.3 η θ φ=0. The present study investigates the temperature distribution and fin efficiency of porous fins for three different profiles.5. Da=10 . The figure clearly depicts that at constant heat transfer coefficient and high porosity the tip temperature of the fin is on the lower side. φ=0. fin efficiency is decreased with the increase of Nusselt number.25 and vice versa for convex profile as in triangular and rectangular profile. Fin efficiency comparison among various profiles with the variation of Darcy number. Ra=10 .6 m=0.4 0. m=0. φ=0.7 25 30 35 40 45 50 Nu Fig. (19). Although actual heat transfer rate is enhanced with this parameter but side by side it increases the ideal heat transfer rate also as defined in Eq.4 φ=0.4 0.0 20 3 θ -4 Da=10 .8 1. which is more pronounced for triangular profile. 5 depicts the effect of the exponent m for convex fin with variation of porosity parameter.KR=10 . Effect of power factor of heat transfer coefficient on temperature distribution for convex porous fin with the variation of porosity parameter Heat transfer through fins of different profiles has been under study for many years to determine the optimum parameters and design of a fin for maximum heat transfer.1 1.2. the convex or triangular fins may reach to the lowest value of fin efficiency. 7.6 0. NIT Durgapur . Nu=10. rectangular. However. One can conclude from the comparison given in the figure for various profiles that the efficiency.9 m=0.0008 0. This figure shows that for the fixed parameters which have been specified in the figure.0 0.5 1. 4 1 A -4 Ra=10 . the temperature is less as compared to m = 0 and low porosity parameter.3 φ=0. The fin temperature is higher for the latter stage although less profound than the rectangular profile. depending on the values of parameters. Fig.2 0.6 0.0. φ=0.6 0.4 0. Effect of power factor of heat transfer coefficient on temperature distribution for triangular porous fin with the variation of porosity parameter. Fig. which in turn reduces the fin efficiency. Nu=10.4 0. 0. is always the highest for rectangular fins. But. Nu=10. 6 illustrates the effect of Darcy number on the fin efficiency for different profiles. CONCLUSION 0.5 m=0.0 0. Porous fins have been found to be quite effective and economical in this regard in terms of commercial and economic perspective. for fixed values of parameters.KR=10 0.0 0.8 4 rectangular triangular convex Ra=10 .9 m=0.5. As can be seen from the figure higher values of m and Ra the fin efficiency is higher. Nu and m is depicted in Fig. 7.2 0.02 Ra=10 0.0 0. 0. 3 KR=10 . 3 ψ=0.0 X Fig.0 0.

Ooi. [7] S. Bhanja. • Both the tip temperature and the fin efficiency are high for the profile of rectangular shape as compared to other geometries. Therm.” Int. 74.S.. J. in press. 123.” Energy Conver. Saedodin. vol. Kiwan. March 2003. Mandal. February 2013. October 2007. July 1999.T. [6] S. vol.” Energy Conver. Hasanpour. vol. “Heat transfer study through porous fins (Si3N4 and Al) with temperature-dependent heat generation. and M. Bassam. “Predicting multiple combination of parameters for designing a porous fin subjected to a given temperature requirement. vol. 2011.. Refrigeration. and B. 55. Bhanja. pp. pp. 1046-1055. 2007.D. 67. 1483-1496. Kundu. [11] J.” Transport in porous media. American Sci. • A higher value of Rayleigh number presents a higher efficiency with high value of exponent m. December 2012. Abu-Hijleh. [10] A.. 9-16. pp. “A model on the basis of analytics for computing maximum heat transfer in porous fins. vol.. Kiwan. 476-481. and D. B.. “Effects of radiative losses on the heat transfer from porous fins. Kundu. 34. “Temperature distribution in porous fins in natural convection condition. Heat and Mass Transf. 2013. Hatami. September 2011. 4 1 [4] A S E I 51 of 446 Copyright @ Department of Electrical Engineering. and Managt. Heat Transf. “Thermal analysis of porous pin fin used for electronic cooling. Olank.” Int. Bhanja. “Thermal analysis of a constructal T-shaped porous fin with radiation effects. Following points have been obtained from the study: • Fin surface temperature decreases with the increase of porosity parameter for all the profiles.. • The temperature distribution as well as fin efficiency is highly dependent on exponent m of the heat transfer coefficient. and P. and K.” ASME J.. Kundu. 282-288.” Int. M. A. 17–29.” J. Lee. 699-708. D. “Using Porous Fins for Heat Transfer Enhancement. vol. “Natural convection heat transfer from a cylinder with high conductivity permeable fins. vol. [8] S. [9] R. vol. J. 34. vol. “Variational iteration method. 790-795. Non-Linear Mechnics.” Int. 66. pp. July 2000. and K. 211-219. pp. [5] D. Al-Nimr.A. 7(6).a kind of non-linear analytical technique: some examples.K.H. pp.. B. Ganji.1007/s112 4 2-006-0010-3. pp. 7611-7622. “Thermal analysis of natural convection porous fins. pp. Das. Procedia Engg. He. pp.” J. NIT Durgapur . Kiwan. and K. and M. Heat Trans.Variational iteration method (VIM). 46. and Managt. J. Sci. 125. DOI 10. October 2013.” J. J. vol. REFERENCES [1] [2] [3] S. pp. D.

sekhar@gmail. storage and transmission.com.B (India). In [9]. INTRODUCTION Telemedicine is the use of telecommunication and information technologies in order to provide clinical health care at a distance. S. W.org) and gives a highly compressed result that can be stored using far less digital space without distorting important ECG characteristics which are essential for proper medical diagnosis and at the same time. bandwidth for the transmission can be reduced in telemedicine applications. ASCII character. A typical ECG monitoring device generates massive volumes of digital data.com Abstract— In the application of telemedicine. nishantmishra21@hotmail. E-mail: basudev_h@yahoo.Mishra S.B (India).Bose. ECG signals need to be combined with patient confidential information when sent. To deal with the huge amount of electrocardiogram (ECG) data for analysis.Patient Information implantation and reclamation from compressed ECG signal by LSB watermarking technique Basudev Halder.com In this paper ECG signals are watermarked with patient information using LSB watermark technique and compressed the huge amount of ECG data using ASCII character encoding in order to confirm patient or ECG linkage integrity and reduced the bandwidth in telemedicine applications.743368. To avoid this confusion. Throughout these techniques the embedded data will be ‘invisible’ to maintain the quality of the host data [2]. SD. who can receive care from doctors or specialists far away without the patient having to travel to visit them. N. As a result. The “Electrocardiogram” (ECG) as shown in fig 1 is an invaluable tool used for the diagnosis and treatment of various cardiac and other related diseases. ECG compression techniques have primarily focused on lossless methods. Department of Electronics. Neotia Institute of Technology. and embedded inside medical images like CT and MRI. Other researchers [8] implemented a wavelet based watermarking technique for ECG signal. and drug administration oversight. Recent developments in mobile collaboration technology can allow healthcare professionals in multiple locations to share information and discuss patient issues as if they were in the same place. 24-pgs(s) . Although there is a large volume of works on watermarking and steganography. (Affiliated to WBUT). Management & Science. It can be beneficial to patients living in isolated communities and remote regions. Keywords— telemedicine. W. authors proposed to insert an encrypted version of the electronic patient record (EPR) in the LSB (Least Significant Bit) of the gray scale levels of a medical image. Kolkata 700 092. The main goal of any compression technique is to achieve maximum data volume reduction while preserving the significant features and also detecting and eliminating redundancies in a given data set. Watermarking technique can play a crucial role in ECG by combining the confidential information with the ECG signal. Due to the diagnostic uses of medical data and since a small detail may be very important. In [7].physionet.S. (Affiliated to University of Calcutta). the ECG signal is used as a secret data. ECG signal is transmitted to the Doctor end without any patient details. an embedded information can be completely retrieved. NIT Durgapur 1 . an efficient ECG compression technique is needed to reduce the amount of data as much as possible while pre-serving the clinical significant signal for cardiac diagnosis [1]. confusion is arisen between signal and patient’s identity. potentially significantly reducing the overall cost of medical care. watermarked. siddhartha. 52 of 446 Copyright @ Department of Electrical Engineering. I. This huge amount of bandwidth for the transmission of the ECG signal can be avoided if the signal is compressed after embedding the patient’s personal information within the ECG signal. CR Watermarking is a process by which a discrete data stream called a watermark is hidden within a host signal by imposing imperceptible changes on the signal. An effective data compression scheme for ECG signal is required in many practical applications such as ECG data storage. Remote patient monitoring through mobile technology can reduce the need for outpatient visits and enable remote prescription verification 4 1 A S E I PRD.Mitra Department of Information Technology. not many researchers have addressed issues related to ECG data. Telemedicine can also facilitate medical education by allowing workers to observe experts in their fields and share best practices more easily. Huge amount of bandwidth is required for the transmission of the ECG signal for telemedicine purposes. Email: s_mitrasarkar@rediffmail.com. ambulatory recording systems and ECG data transmission over telephone line or digital telecommunication network for telemedicine. Since the ECG signal and the patient details integrated into one. The whole module has been applied to various ECG data of all the 12 leads taken from PTB diagnostic database (PTB-DB) of physioNet (www. The volume of ECG data produced by monitoring systems can be quite large over a long period of time. Netaji Nagar Day College.

The remainder is kept in the variable ‘rem’. The basic idea of this process is to watermark the patient Id.org as a ‘txt’file is copied into notepad and used as ECG data file.e. increased or decreased by 1 or will remain unchanged. All these steps are explained sequentially in the rest of the sections. Hence concentration is given only to compress the “Voltage” values discarding the ‘Time’ value. To implement our algorithm. The length of the patient Id is calculated firstly. the possible change in ECG voltage values will only be by 1 viz.2] c[1.7] c[1. Say for example N is 4000. e.10a and the following steps are involved: For getting smaller numbers and consequently better compression a difference array is constructed which contains the difference of every ECG sample and it’s preceding except the first ‘‘Voltage’’ value which is kept unchanged. only the ‘Voltage’ values are taken which we intend to compress. This binary data is embedded in the voltage values of the first ECG lead using LSB embedding technique. wavelet transform (WT) [5] etc. For this we create a 1D array from normalized 2D array of each lead of the ECG data. In the reconstruction module of the proposed scheme. the content ecg[i. Now the sign bit is printed in the output file in its corresponding ASCII character. It is observed that the proposed scheme achieves much better result as compared to in [1]. Fig1: Normal ECG signal ECG data compression algorithms have been mainly classified into three major categories: 1) Direct time-domain techniques. a binary (1) is taken as sign bit for that corresponding ‘Voltage’ value. age and sex into the ECG signal by using the LSB watermarking technique [9]. we divide the number by 255 and keep the quotient at this (1st Column) position. NIT Durgapur 2 .g. We propose a novel approach of doing both of the required functions together. Decimal equivalent of this binary string will be used as the sign-bit of these corresponding eight ‘Voltage’ values. and then the 2D array becomes 500 x 8 viz. e. From this file. we focus on how to implant the patient personal information into the ECG signal by LSB watermarking technique [9] and how to compressed embedded ECG signal using ASCII character encoding [1]. amplitude-zone-time epoch coding (AZTEC) [3]. fast Fourier transform (FFT). 500 rows and 8 columns. As we are using LSB technique. the first ‘voltage’ value is essential otherwise data retrievals will not be possible. the “Time” axis can be easily generated during data reconstruction. After getting the sign-bit... The proposed algorithm allows not only the embedding but also the efficient retrieval of the embedded data. After embedding the patient’s personal information. a binary zero (0) and for every negative difference. discrete sine transform (DST).5] 14 0 Example 2 c[1. Embedding patient’s personal Information process: The second phase of the proposed scheme is to embed the patient’s personal information such as patient ID. Let as an instance Example 1.j]) of each lead of the ECG data of length (N/8) x 8. but individually.3] 0 2 5 c[1.the 12 leads. we make a 2D array (ecg[i. ECG data selection and Normalization: The ECG signal which is taken from PTB diagnostic ECG database from www.METHODOLOGY Example 1: A. Since the sampling frequency of the input ECG data is 1000 sample/sec. The whole compression scheme is such that the compressed file contains only ASCII characters. II. where i=1 and j=1 to 8. we will embed patient’s personal information by watermarking techniques. using the reverse logic of compression which is proposed in [1]. patient’s personal information is reconstructed. A sample array after normalization is shown below: c[1. embedded data i.the first value of every row is becoming high (possibly greater than 255) as it remain unchanged during creation of difference array. and then converted to its corresponding binary data form. At the time of reconstruction. There are extensive work on ECG data compression and hiding patient information in ECG signals. original ECG signal is brought back and using the reverse logic of ‘LSB watermark technique’. To bring this number in the range of 0-255. 3) Parameter extraction methods [6] are mainly based on linear prediction and long-term prediction methods. voltages are recorded up-to three decimal points.1] c[1. The logic of ECG data compression with embedding patient personal information is written in the MATLAB 7.j] array is as below.g. we compress the ECG data by characters encoding.4] c[1. discrete cosines transformation (DCT). Data bits will be scattered throughout the 53 of 446 Copyright @ Department of Electrical Engineering. all negative numbers in the difference are made positive by multiplying -1. coordinate reduction time encoding system (CORTES) and Fan algorithm. Now. During creation of the difference array and amplification (multiplying by 1000) .6] c[1.8] 4 7 15 B. In such an array. turning point (TP) [4]. Suppose we have ECG data with N voltage values in each of 4 1 A S E I The proposed scheme is divided into following five main sections: (A) ECG data selection and Normalization (B) Embedding patient’s personal Information (C) Data Compression (D) Data reconstruction (E) Patient personal information retrieval. In this paper. Now each and every number in the difference array is multiplied by 1000 because in standard ECG database and also in PTB-DB. For every positive number. the sign of the every element of difference array is checked. age and sex etc in to the ECG signal. 2) Transformational approaches.physionet.

84+2*7). First. 184 368 552 736 920 1104 1288 1472 46 20 3 26 110 5 21 49 A 00101110 00010100 00000011 00011010 01101110 00000101 00010101 00110001 B 1 1 1 0 0 1 1 0 C 00101111 00010101 00000011 00011010 01101110 00000101 00010101 00110000 D A: Actual data.e. i. Here. This algorithm involves the following steps: • Embedding the length of the patient Id in first lead: Let the length of the patient Id or the number of characters is 12. All these data’s are converted to its corresponding ASCII value. 24th and 32nd position in first lead. Similarly. These bits will be embedded with original ECG signal in the interval of 8.Similarly the number of character of patient sex is one.. For this we scatter the Here 50 is added as an offset because a situation might arise where the sign byte for a lead is zero. Three types of grouping are considered here as in [1] namely (a) forward grouping.2nd column of actual input ECG file is length of patient Id plus two. in 8th. (7*184).e. So.etc positions. again embedded array is converted to 2D array.2nd column of actual input ECG file. used in PTB data base for each ECG file.e number of character is one . The interval for every bit will be calculated by using following equation: Interval = sign bit +50 + (8 – mod (sign bit. Suppose the 54 of 446 Copyright @ Department of Electrical Engineering. patient age and sex are used as data for embedding in each ECG signal. First lead is already reserved for the length of embedding patient Id.txt’. ECG data compression: To compress the data. NIT Durgapur 3 . D= embedded data: Example 3 Let us calculate now the interval at which these bits will be embedded in the respective lead will depend upon the first sign-bit of that respective lead. Three variables have been taken ( f : for forward. B: binary form of actual data. (8*184). suppose the sign-bit of a lead is 128 then Interval is 184 (ie. embedded integers are grouped maintaining some essential logical criteria. the length of the embedded data (i.24 and 32 in lead 1 are 15.remaining 11 leads.e TL = 11.e. Example 4 shown below. the length of the patient Id is essential otherwise data retrievals will not be possible. The number of bits to be embedded in0 remaining 11 leads will be calculated by ‘Bitload1’ and ‘Bitload2’. (2*184). D:Actual with embedded data Example 4 After the Text bits have been embedded in the leads. The ASCII value of each character is converted to its corresponding binary form with seven bits.2nd column of actual input ECG file. A= actual data B= binary value of actual data C= actual with embedded data. (b) reverse grouping. So.(TL-1)) …. initialized with zero (0)) for denoting the positions of these groupings. Suppose the four values of position 8. the total number of bits that has to be embedded is 12*7 =84 bits.Corresponding ASCII character of age is ‘A’. These two characters (i. [128+50+(8. the patient Id. 16th.e. C.e age and sex) again converted to its corresponding binary form with seven bits each. (4*184). r : for reverse and n : for no grouping position respectively. Bitload2= mod(TB.mod(128. Suppose the file name is ‘s0015lre.8))]) . then after embedding ‘1100’ the result will be shown below in example 3 .. 8th 15 00001111 00001110 14 16th 10 00001010 00001010 10 24th 11 00001011 00001011 11 32nd 4 00000100 00000101 5 A B C D embedded data bits in all the 12leads of ECG data because embedding all bits in a particular lead might increase the PRD of that lead. i. (3*184). It is either ‘M’ or ‘F’.e TB=98 and number of leads for embedding is 11 i. • Calculation of the number of bits and interval for embedding data: Let us describe now how calculate the number of bits of a particular lead for embedding data. So the total number of character for embedding in the first lead i. Suppose the total number of bits to be embedded is 98 i. Suppose the length of the patient Id is 12 then total number of character is 14 which are to be embedded in the first lead i. suppose the patient age is 65 and its corresponding ASCII value is ‘A’ and patient sex is either ‘F’ or ‘M’. • Embedded data generation: 4 1 A S E I Let us describe now how the embedded data is generated. At the time of reconstruction.. and used as patient ID which is of 12 characters. Here patient Id and sex both are in text format and patient age is in numeric value. So number of bits to be embedded in lead 2 to lead 11 is 9 and number of bits to be embedded in lead number 12 is 8.. (6*184). the total number of bits that has to be embedded in ECG file is 98(ie.The equations are given below: • Bitload1=floor (TB/(TL-1)) ……. The length of the characters is to be embedded in the first lead i.12) is converted to its corresponding binary value i.e. 4 respectively. If the first sign-bit of a lead is 128. By making this offset we can be sure that the interval is always 50 and above. (1) (2) Where TB is the total data bits and TL is the number of leads for embedding. 552nd….e. 8))… (3) Calculation of the length of patient age and sex: Suppose the patient age is 65. 11. 368th. (5*184). ‘1100’. (c) no-grouping..16. C: Embedded data. the bits will be embedded at 184th. 10.Hence the corresponding value of Bitload1 and Bitload2 will be 9 and 8 respectively. then the positions at which 8 bits will be embedded are (1*184).

If so. i. the length of embedded data is ‘1100’.2]. Equivalent ASCII values of those characters are saved in an array and then reconstruction algorithm is applied on those to restore original eight ‘Voltage’ values. 4 1 A 20 2 11 3 Actual data 00010100 0 00000010 00001011 00000011 0 1 1 binary of actual data Example 6 reconstructed data From the above example 6.e. ECG Data reconstruction: The fourth phase of the proposed technique is the ECG data reconstruction . 2. we need to multiply the value at first value of the array with 255 and add the value of ‘rem’ to it.8th and 7th are the grouping position. so. Therefore the total number of bits which are to be embedded in the first lead of ECG file will be (84+2*7) =98. In the next step. r. Suppose the four values of position 8. mentioned below.16. E. 3 respectively. 24th and 32th position in first lead.j+1]array position and checked whether the value is less than 255. we have embedded the size of data in the LSB of four values in 8th. Using the reverse logic of grouping technique. For example in the second pair. the original eight voltage values are generated and stored in an array . n. Therefore the number of bits embedded will be 2*7=14. The value of f=7. Data position holds the position where forward. • Calculation of the length of Patient’s Identification: As during embedding.its corresponding decimal value is 12.24 and 32 in lead 1 are 20. initially ‘i’ is set to ‘one’ and incremented by a factor of two. Therefore the number of bits embedded will be NC*7. Following ASCII values are the grouped / no grouped integers. If forward grouping is not possible then it is to be checked that whether any ‘(c[i. We will calculate the length of the embedded data (patient’s ID) from the LSB identification of these values. the sign-bit will be converted into its corresponding 8-bit binary equivalent. it is necessary to decompress the data for retrieving the original signal and patient’s information at the receiving end.5] and c[1. n=8 and data position=9. 16th. which for the above example will be 12*7=84. then this reversed grouped value is printed in character form. i.j]’ is less than 255 or not. etc.e. 11. If both forward and reverse groupings are not possible then. Each set of grouped or not-grouped ‘Voltage’ values along with other necessary information (sign byte.e. for pairs c[1. n=8 and sign bit=196 and rem=35. For all the other pairs we take the value “1” (which is the first position of the pair) for denoting the grouping position. c[1.After compression with embedded data. In the example 5. • Calculation of the number of bits and interval for embedding data: 55 of 446 Copyright @ Department of Electrical Engineering.) will be printed in the output file in ASCII character form maintaining the following format. So the total number of character for embedding in the first lead i. the value of f=7. After all the pairs are examined. grouping positions.2nd column of actual input ECG file is length of patient Id plus two. There will be at most 14 characters in a set. each number in c[i.j] array position is multiplied by 100 and added with the number in c[i. Patient personal information retrieval: The fifth phase of the proposed technique is to retrieve the patient’s identification from the reconstructed ECG Data. • Calculation of the length of Patient’s age and sex: The number of character of patient age is one and number of character of sex is one which means that number of character is two.e. reverse and no grouping position ends. For forward grouping.6] we would take value “6” and so on. This array contains the reconstructed ECG voltage values. From the compressed data file at a time one set of ASCII character is taken. a modified version of the algorithm proposed in [1] is used to decompress the ECG data in lossless manner. value at those positions are left unchanged and both integers are printed separately in character form.1] and c[1. we sum the values and store it in the respective variable.4]. we take the value “1” (which is the first position of the pair) for denoting the grouping position. r=4. All these are necessary to get back the original eight voltage values. They will be interpreted providing the size of data embedded as : Here. f. S E I Example 5 The above compression algorithm is executed iteratively until all the ECG data contained in the input file are become compressed.j + 1] *100) + c[i. From the Example 5 we know that. The first ASCII value among those contains the ‘Sign byte’ of those eight voltage values. data position is 9.sign bit=196 and rem=35. which means number of characters (NC) is twelve characters. Suppose the length of the patient Id is 12 then total number of character is 14 which are to be embedded in the first lead i. For this purpose. For all the other pairs we take the second position for denoting the grouping position. the 9th. Data position and ‘rem’ respectively. To obtain the actual value at the first position.2nd column of actual input ECG file. we would take value “4”. c[1. for all three grouping procedure. r=4. then this grouped integer value is printed in character form. In the binary string if any bit is ‘1’ then the corresponding positional element of array will be multiplied by (-1). NIT Durgapur 4 . This algorithm is the reverse logic of embedding patient’s identification.first pair of c[ ] array.3] and c[1. D.

e TL = 11.The character thus obtained is the data which was embedded. (4) Bitload2= mod (TB. every number is divided by 1000.004349 7.00455.295404 3623.987 s0465_re. PRD: percent root-mean Square difference • The first parameter that expresses the effectiveness of a data compression technique is the Compression Ratio (CR).txt 50 M 12Lead 0.txt 56 M 12Lead 0. QS: Quality Score. The actual interval in this case will be next multiple of 8 from 178.004635 7. To get the original.982 The following measures are computed to evaluate the performances of the proposed algorithm.txt AGE 58 s0464_re.419356 1927. the closer the reconstructed signal is to the original ECG data. The new values are stored in another array. CR = n PRD = 100 X QS = CR PRD PRD.002265 7.txt 55 M 12Lead 0.txt 41 F 12Lead 0.429676 3133.893 s0480_re. each number is added with the previous value.003365 7.007101 7.txt 44 F 12Lead 0.331693 1608. Table 1: Recovering the Embedded Data bits and Text reconstruction: FILE NAME 4 1 s0499_re.txt 34 M 12Lead 0.187249 1736.046 s0463_re.003913 7. we will convert the value at the intervals obtained into binary format and will store its LSB for construction of the characters.03 s0028lre.txt 100 M 12Lead 0. 56 of 446 Copyright @ Department of Electrical Engineering.It is defined as below. the better of the performance of the compression algorithm.txt 75 F 12Lead 0. • Another numerical measure Quality Score (QS) was proposed in [15] to quantify the overall performance of compression algorithm. The interval for every bit will be calculated as: • ( y i− yi )2 ∑ AVERAGE SEX LEAD PRD % CR QS M 12Lead 0. (8*184). (Every decimal value has corresponding character is ASCII character set). The lower the PRD.238283 4006.004234 7.241806 1791. and QS obtained for Myocardial Infarction and Normal patients are given in Table 1.902 0.351 s0462_re. 3 is normal (N) and Fig. 184.047 s0027lre.453826 2649. i. average CR of about 7. (7*184).187249 1729. (3*184). (2*184).txt 67 F 12Lead 0. 8))… (6) In the previous step. Now. we can obtain the character which was embedded. This measure evaluates the distortion between the original and the reconstructed signal.003604 7. (5*184).004557 7.txt 51 F 12Lead 0.003799 7.454503 2470.333 S0026lre. NIT Durgapur 5 .004884 7.e TB=98 and number of leads for embedding is11 i.478 s0474_re.424361 3824.6 S0025lre. The higher the value of CR.txt 30 M 12Lead 0. we will calculate the value of Bitload1 and Bitload2 by the equation: Bitload1=floor (TB/ (TL-1)) ………….txt 45 F 12Lead 0.401489 2544.203689 1180. we can see that using the proposed algorithm we can get average PRD of about 0.88. we add 50 to it making it 178. Following the reverse procedure of what we applied during embedding.Hence the corresponding value of Bitload1 and Bitload2 will be 9 and 8 respectively. i =1 ∑ Now the interval at which these bits will be embedded in the respective lead will depend upon the first sign bit of that respective lead.88 A S E I III.33 and average QS of about 1608. From the above tables.00274 7.004687 7. n i =1 Interval = sign bit +50 + (8 – mod (sign bit.005413 7.313354 6407. For the intervals at which the bits are embedded in a lead with sign byte 128 are (1*184). (5) Original ECG Filesize Compressed ECG Filesize • Percentage Mean Square Difference (PRD) is a measure of error loss. where X will be the number of characters that were embedded. By iterating the procedure of LSB obtaining method mentioned above for each lead as per Bitload1 and Bitload2 values. If the first sign bit of the lead is 128.35102 2963.192171 1379.txt 25 F 12Lead 0.By reserve procedure as followed during the embedding. This array is identical to the original ECG voltage values.290111 2201. Fig.txt 24 F 12Lead 0.135 s0475_re.907 s0015lre. (TL-1)) ………….757 s0467_re. (6*184).528023 2279.358899 1294.005393 7. we will obtain a 2D array of X*7. PRD calculation is as follows: Suppose. (4*184).055 s0031lre.233 s0466_re. Following figures (Figs. 2–3) shows the original and reconstructed ECG signals of 2’s sample.txt 77 f 12Lead 0. A high score of QS represents a good compression performance. The binary value will again be converted into decimal for application of further procedure.txt 55 M 12Lead 0. Converting each row with seven binary values to corresponding decimal value. 2 is Myocardial Infarction (MI) ECG.003433 7.00627 7. So number of bits to be embedded in lead 2 to lead 11 is 9 and number of bits to be embedded in lead number 12 is 8. for a particular lead we calculated the interval at which the text bits have been embedded.707 s0029lre.007313 7.e. MEASURES OF PERFORMANCE y i2 represents original signal samples and Where represents reconstructed signal samples. CR: compression ratio.004631 7. the total number of bits to be embedded is 98 i. CR.596 S0022lre.txt 77 F 12Lead 0.

-0. A. Engin. 13:604–22. 14.” Biomed.6 12.V: CONCLUSION Original ECG SIGNAL Amplitude -0. IEEE international conference on engineering in medicine and biology society.W. and S. Digital Signal Process 2008.VB.M. M. LLEDCCE [1] Proposed Method CR 5.Hilton “Wavelet and wavelet packet compression of Electrocardiograms” IEEE Trans Biomed Eng 1997.33) as compared to other methods reported in Table 2. S.4 -0. IEEE Trans Biomed Eng 1993.‘A novel compression algorithm for electrocardiogram signals based on the linear ediction of the wavelet coefficients’.S. Lucchese. Nolle.1 2. In Conference proceedings: Annual International Conference of the IEEE Engineering in Medicine and Biology Society.9 1.61 28 2.Tompkins. 1978.13. vol. MS.Fabrizia. vol. Table 2 Algorithm Name AZTEC [3] Hilton [5] Johan [10] SPIHT[12] Perceptual Masks[13] Al-Shrouf [14] Fira and Goras [15] TP [4] USZZQ and Huffman coding of DSM [16] Wang and Meng [11] S.0045).‘Wavelet compression of ECG Signals by the set partitioning in hierarchical trees (SPIHT) algorithm’.0030402 1500 2000 2500 3000 . and G.Cox.MM. Computers and Electrical Engineering 37 (2011) 486–497. J. C. AZTEC. Ahmadian.5 -1 [3] -1. G. A. October 1998.619 .CR =7. Mukhopadhyay et al.14. F. D. Journalof Medical Systems. The performance of the proposed algorithm was compared with some other proposed ECG signal compression methods. and E.CR =7.4. M. p.47(7):849–56. Volume 33. R.S. Ghavami. Dilmaghani. pp 241-259 J.8 -1 0 500 1000 PRD =0.August 2009.6534 2500 3000 .Zahhad.6 12 0.Montreal. Engin. Mitra. M.Leonardo.QS =6757.J.8 -1 0 500 1000 1500 2000 2500 Time Reconstructed ECG 3000 3500 4000 Amplitude -0. [14] [15] [16] 57 of 446 Copyright @ Department of Electrical Engineering.55(4):1319–26. ‘A 2-D ECG compression algorithm based on wavelet transform and vector quantization’.Q.A. H. 2005. China. 1. M. Karimi-Fard.Vlachos .D. pp.txt (MI) Original ECG SIGNAL Amplitude -0. the denominator part in the expression of PRD becomes high and shows a very low PRD (0.Nave. 18:179–88. Que. p.553 [4] 3500 4000 Time --> [5] Fig 3: File name: S0499_re.Yu.C. A. 4 1 A S E I PRD S.Cohen. Z. Due to this low PRD. 29(6):589–594. 167–8.DY. S. Anand and U.Wang.0045 7.4 -0. Cidam. Oliver. Kozat . an excellent QS value often gets raised.”Embedding and Retrieving Private Metadata in Electrocardiograms”.Mueller.5. O. CR and PRD of the proposed method is comparable with the methods [3.Biomed Signal Process Control 2006. pp.3 0. Hong Kong.33 [6] [7] [8] [9] [10] [11] [12] [13] It is observed that the proposed method achieves high CR (7. Proposed algorithm shows better performance for both normal and abnormal ECG signals. Eng.6 -0.R.Kim. “Arrhythmia detection software for an ambulatory ECG monitor.Manikandan.74 2 11.T.S.V.Fira. vol. M.K. IEEE Trans Biomed Eng 2008. M. if there is any shift of base line.L. BME-15. CMD.Dandapat. NIT Durgapur 6 . It is interesting to note that in the original ECG signal. 40:877–85.0652 3500 4000 Time --> Fig 2: File name: S0025lre. Inst.P.6 -0.Goras L. Niranjan. M.”ECG compressions using discrete symmetric wavelet transform”.Meng. Fozzard.Johan.1:261–70. Mitra.5 11.M.Biomed. 703-706.Shrouf.16.” IEEE rans. 2006. C. Volume 1. ‘A novel blind watermarking of ECG signals on medical images using EZW algorithm’.2 -0. 128–129.WA.Nguyen.Herle. C. Mukhopadhyay.K.11. In Proceedings IEEE/EMBS Conference. CLAIB 2007.Rodrigo. Digital Signal Process 2003.06 1.Pearlman.15.0011326 1500 2000 .. ‘An ECG signals compression method and its validation using NNs’.”A preprocessing program for real time rhythm analysis. 229–31.A.5 -1 References: -1.2 We proposed a new approach for implantation and reclamation of patient information from compressed ECG signals by using LSB watermarking technique. We proved statistically that the implanting of patient information does not affect the significant features of the ECG signals and it also does not induce any changes in the diagnosis.The results of this research will likely provide an improvement on existing techniques.M. IEEE Trans Biomed Eng 2000.44(5):394– 402.‘Near-lossless compression of ECG signals using perceptual masks in the DCT domain’. ‘Wavelet Transformation Based Watermarking Technique forHuman Electrocardiogram (ECG)’.18 0. Sci. “Watermarking Medical Images with Patient Information”.Lu. 81–85. X. PRD and QS of various lossless and near lossless methods with the proposed method is given in Table 2.5 [1] 0 500 1000 1500 2000 2500 Time Reconstructed ECG 3000 3500 4000 [2] Amplitude -0. pp. Canada.12.S.Ahmed.73 10 8 8 8 3. W.A. A.H. In: IFMBE Proceedings 18.1] as listed in Table 2.10.. specially in [1] and it is observed that the proposed algorithm gives a high compression ratio with low distortion and an excellent Quality Score than in [1].24 5.18 1.5 -2 -2.023 7. 1968.‘Wavelet threshold based ECG Compression using USZZQ and Huffman coding of DSM’.”ECG compression using long-term prediction”.L. “A lossless ECG data compression technique using ASCII Character encoding”.6 3.5 0 500 1000 PRD =0.5 -2 -2. Nambakhsh.QS =2506.txt (N) IV: PERFORMANCE COMPARISON Comparison of CR.

Keyword: Rotating shaft. because of its importance in design. Walston et al. [4] in the paper discussed the behavior of a rotating shaft in viscous fluid. D. I. influnce coefficient method. respectively Ω = Angular velocity of whirling.mech@gmail.Fy=Fluid forces on shaft in x and y direction. but no clear-cut distinction was made between virtual mass effect and damping effect on the rotating shaft. The fluid force affected the natural frequency and amplitude of the simply supported rotor. investigators have concentrated on the critical speed of rotors in vibrational analysis. Idea [10] also treated the same problem taking an infinitely extending water region instead of a cylinder. v = Tangential flow velocity v = Coefficient of viscosity V1= Poisson’s ratio ƍ = Fluid density ƍ1 = Mass density of the shaft ψ1. Papadopoulos and Dimarogonas [3] have discussed the performance of a rotating cracked shaft. Wauer [1] have presented a review on the dynamics of rotor with cracks. under various conditions. He established that the presence of a crack influences vibration amplitudes at the critical speed. but virtual mass effect (added mass) and external damping effect were not discussed separately. Sol [9] has proposed an analytical model and compares his results with the experimental rig. ωxx. The effect of fluid forces on the rotor is analyzed with the help of Navier-Stokes equations. due to varying the coefficient of viscosity and gap ratio there is change in the amplitudes of vibration. amplitude. Rourkela Rourkela. It is observed that when the rotor rotates in a viscous fluid. The influence coefficient method is used to find the fundamental natural frequency of the simply supported shaft damping effect on rotating shaft due to viscous fluid are analyzed with the help of Navier Stoke’s equation. They used mathematical and the theoretical consequences for eigen frequency exposed to different fluid parameters. G = Shear modulus I = Section moment of inertia of the shaft L = Total length of the shaft Kn (x) = Modified Bessel function of second kind of order n L = Length of span Ms = Mass of the shaft per unit length m = Fluid mass displaced by the shaft per unit length p = Pressure R1 = Radius of the shaft R2 = Radius of the cylinder t = Time u = Radial flow velocity. India adik. [12] have examined the dynamic behavior of a swinging shaft which is 4 1 A NOMENCLATURE:A1= Shaft cross-sectional area a1= Crack depth D =Diameter of the shaft δ =Whirling radius of the shaft E = Modulus of elasticity of shaft material ɛ = Eccentricity Fx. S E I For decades. ψ2= Stream functions ω= Rotating speed INTRODUCTION 58 of 446 Copyright @ Department of Electrical Engineering.com ω0= Natural angular frequency of the shaft.R. the analysis of natural frequency and amplitudes becomes difficult. viscous medium. NIT Durgapur . Here we have varying the coefficient of viscosity of fluid and gap ratio. The surveys carried out the vibration analysis problem of a rotor in viscous fluid. He took the De-Laval rotor for the investigation and explored some prospects for early crack detection. Brenner [11] has analyzed theoretically the fluid forces acting on a circular rod rotating in a circular cylinder for high and low Reynolds numbers Walston et al. Gasch [5] have presented the vibration analysis of a rotating shaft with a transverse crack. Fritz [8] have studied a circular rod spinning in a cylinder. [2] have examined the oscillations of a rigid cylinder in a tubular duct occupied with a viscous incompressible fluid. Kadyrov et al. ωyy = Critical speeds in x and y directions. taking account of the influences of turbulence and Taylor vortex for a very small rod cylinder gap. respectively. Ostachowicz and Krawczuk [6] have studied the effect of transverse cracks on coupled torsional and bending vibrations of a rotor Branner [7] have examined the fluid forces performing and acting on a rotating cylinder for low and high Reynolds number. damping effect. Abstract— In this article we have presented the vibration analysis of a rotating simply supported shaft in viscous medium.Vibration Charcterstics of Rotating Simply Supported Shaft in Viscous Fluid Adik Yadao1 Irshad A Khan2 Dr. parhi3 Department of Mechanical Engineering National institute of technology. When a shaft rotates in viscous medium.

Kito [13] have studied the influence of fluid forces on a eccentrically rotating circular bar in a viscous medium. In this investigation. t) the above equation can be written as. The radial and tangential components of flow velocity at point A in figure are. Damping effect due to viscous fluid is determined with the help of Navier Stoke’s equation. u A R1 sin a sin(t ) (4a) vA R1 cos a cos(t ) (4b) Where a is the angle between the lines ÓA and OA. viscosity and mass established by using the statistical regression analysis. Achenbach and Qu [14] have discussed the resonant vibrations of a submerged beam of circular crosssection using mathematical methods. sin a / R1 sin t and cos a 1 for R1 We can write the above expressions for r=R1 as follows II. . Natural frequency of the rotating shaft used for finding the critical speed of the system is determined using the influence coefficients. u 1 p 2u 1 u u 1 2u 2 v t r r 2 r r r 2 r 2 2 r 2 (1b) A S E Where u and v denote flow velocities in radial and tangential directions. NIT Durgapur (8) . The influences of the fluid on the critical speed and amplitude of the rotor were determined experimentally. Considering the relationships. the boundary conditions for r = R2 are Equation [2] can be divided into two parts i. I We obtain. respectively.immersed in a viscous fluid. Where v 1 p 2v 1 v v 1 2v 2 u t r r 2 r r r 2 r 2 2 r 2 4 1 vr R1 cos t R1 Re e jt R1 (1a) (5a) ur R 0 1 vr R1 Re e j t R1 (6) When the shaft is immersed in an infinitely extending fluid region. (2) ur vr 0 (7) When the shaft is immersed in a fixed circular cylindrical fluid region with radius R2 . and p means a pressure. 2 0 1 2 0 v t ur R2 vr R2 0 59 of 446 Copyright @ Department of Electrical Engineering.e.stroke equation for fluid velocity is expressed in the polar coordinate system r-θ as follows. The solution of equation [2] can be given by 1 2 Where 1 and 2 are the solution of equation [3] In this article. stiffness elements. a systematic analysis for the vibrational behavior of a rotating simply supported shaft in viscous medium is presented. denotes a real part of . he supposed the flow velocity distribution to be linear over a gap between the bar and cylinder.For a special case equation is reduced to. They compared the exact results for the beam deflection with the approximate results that are available in the literature. the boundary condition for r are. and a correspondence equation for amplitude as a function of velocity. and boundary conditions of the simply supported shaft. θ. 1 4 2 0 t Where 2 2 1 1 2 2 r r r r 2 (5b) j = 1 and Re . THEORETICAL ANALYSIS ur R1 sin t Re j e jt The Navier. Rewriting the above equation with the help of a stream function ᴪ (r.

Thus the no stationary components of flow velocities induced by the whirling motion of a shaft are given as follows: S E I R1 2 R1 A B C I1 kr 1 r r e j t ud j R r D 1 K1 kr r R1 2 R1 A B C I1 kr kR1I 0 kr r r jt vd e r R1 D K1 kr kR1K0 kr r (12a) Fy 0 rr (15a) 1 sin r cos R1d im 2 A B CI1 DK1 eit Where (15b) kR1m R12 Only the real parts of equation are meaningful.(10b) is Bessel’s equation . 2 d 2 F2 1 dF2 dr 2 r dr (13) Normal stress (9a) 2 r . ANALYSIS OF FLUID FORCES Substituting the flow velocities given by equation (12) to equation (1). NIT Durgapur . the no stationary component of pressure p can be written as Where H A B CI1 DK1 and Re H Im H denotes the real and imaginary part of H . 2 cos sin R d m A B CI DK e Fx rr r 1 0 2 AR 2 F1 r 1 Br r i t 1 (11a) 4 1 . The coordinates of the center of the shaft are x cos t and sin t 60 of 446 Copyright @ Department of Electrical Engineering. . no stationary components of solution 1 and 2 can be expressed as 1 r .C and D are arbitrary constants .B.the general solution of these equations are easily derived as. t F2 r e j t p A 2 2 R1 Br ei (t ) r (10b) and 1 ui r (14) Fluid forces acting on the surfaces (i. F2 r R1 CI1 Kr DK Kr (11b) A 2 Where A. and I1(Kr) and K1(Kr) modified Bessel functions of the 1st and 2nd kinds . .(10a) is a Euler’s equation and Eq. t F1 r e j t the p (9b) r due to flow rr ui r p 2 d F1 1 dF1 1 2 F1 0 2 dr r dr r rr (10a ) i r r r r 1 2 k 2 F2 0 r and tangential stress can be obtained as.e.Under these conditions. So Fx and Fy after simplification can be express as Fx m 2 Re H cos t Im H sin t (16a) Fy m 2 Re H sin t Im H cos t (16b) (12b) A..respectively. r R ) per unit length of the shaft in the x and y direction are obtained Since eq.

2… n=1&2 61 of 446 Copyright @ Department of Electrical Engineering. and Av 2 (18b) ms =mass of the shaft per unit length. ms 2 L*4 4 1* 2* sin 2* 2 4 4 (17b ) mass to the inertia force of the shaft and ms 22 m*2* Im H 2 2 C0 n mn*n* Im H fn K vn * * 1 mn Re H 1 mn Re H The steady. 2 * * 2 * 2 1 1 L sin v 2 Av1 v 1 m1* Re H The equation of motion for the shaft having uniformly distributed mass and stiffness are.state solution of equation [21] is easily obtained as v 1 2 v* cos 1*1 1 Yv 1 2 v* cos v* 2 2 Where. ms m Re( H ) I 2 x x 4 x m Im H EI ms 2 cos t (19a) 2 4 t t z 2 * * 2 * 1 2 L cos2 v 2 v 1 m1* Re H 4 1 (20b) A S E m =fluid mass displaced by the shaft per unit length. B. 2 x cos t 4 x EI F t 2 z 4 EI = bending stiffness of the shaft. NIT Durgapur . (20a ) Yv 2 C02Yv Kv 2Yv Av 2 sin 2* 2 . a simply supported rotating shaft immersed in the fluid region is considered. ms m Re( H ) 2 y y 4 y m Im H EI ms 2 sin t (19b) t 2 t z 4 * vn K Avn vn n* 2 2 C0 nn* 2 Equation [19] can be reduced to 1 m* Re H f1 22 m*2* Im H 2 2 2 L*4 4 1* 2* sin 2* 2 4 4 vn tan 1 C0 nn* K * 2 n vn (20a) For =1.Fx m Re H d 2x dx m Im H 2 dt dt 1 m* Re H (17a ) f1 d2y dx Fy m Re H 2 m Im H dt dt In equation [17] m Re H denotes the virtual or added m Im H X v 1 C02 X v Kv1 X v Av1 cos 1*1 . ANALYSIS OF SIMPLIY SUPPORTED SHAFT MOTION Here. we obtain denotes the viscous damping coefficient. (18a ) 2 x sin t 4 y EI Fy t 2 z 4 Where (20b) Applying the Fourier transform to both sides of equation (20).

due to eccentricity III.156. there is a decrease in critical speed as well as decrease in amplitude of vibration and increases the amplitude ratio due to increase the gap ratio.01m.01m. ‘q1’=((R2-R1)/R1) 6) Radius of the viscous container A I S E = 3.0m) Gap Ratio (q1=3) Shows the effect of changing the damping coefficient of the viscosity of the fluid and the gap ratio parameter of the container on the frequency and amplitude of the simply supported shaft. as L* .0 = R2 The result obtained from numerical analysis using the expression with the program which plotted in figure1. respectively .0814 stokes 4) Equivalent mass of fluid displaced / Corresponding mass of the shaft ‘M*’=0.Frequency ratio vs. from these figures we observed that as the viscosity of the external fluid increases. 4 1 Figure 2. In this figure the graph are plotted between dimensionless amplitude ratio and frequency ratio. L=1. NIT Durgapur . 2 and 3.0844 and 2. the whirling motion for fundamental bending mode in x and ydirection can be written as respectively. Mild steel simply supported shaft rotating in viscous fluid (finite region) 1) Length of the shaft ‘L’ =1.01m.8/0.0m 2) Radius of the shaft ‘R1’=0.L=1. Dimensionless Amplitude ratio Mild steel shaft (R1=0. NUMERICAL RESULTS AND DISCUSSION In the present investigation. 5) Gap ratio.487/0. 1 . the specimen being has got the following specification. Frequency ratio vs. 2 12* sin 2* 2 12 2 (22b) Figure 1. Frequency ratio vs. Gap Ratio (q1=5) 62 of 446 Copyright @ Department of Electrical Engineering. Dimensionless Amplitude ratio Mild steel shaft (R1=0.461/0. From the above equations. Gap Ratio (q1=3) and are the dimensionless deflection in x and ydirection.7/0.0m). Dimensionless Amplitude ratio Mild steel shaft (R1=0.0/5.Similarly for Yv it can be done. 1 11* cos 1* 1 11 2 (22a ) L* .0m). L=1.01m 3) Damping coefficient of viscous fluid ‘v’=2. Figure 3.

“Dynamic stability of rotating shafts in viscous fluids. and A.Fritz. Sol. Gasch.212 1002. “Vbration detection of a transverse crack in a rotating machine shaft.398 V=0. [1] 1 1053 V. Vol.Kadyrov.” Applied Mechanics Review 43 (1).Idea. Materiaux. S.Clark. “On the flow in an annulus surrounding a whirling cylinder.804 0.” Mecanique.V. F.494 4597. J. 4 1 A 3.1980. Brennen. The effect of damping coefficient of viscosity and gap ratio are taken into consideration while concluding the result. 923–930. the investigation carried out presently can be used for vibration analysis of high speed rotating shaft in viscous fluid analysis or rotating shaft in high speed centrifugal condition monitoring at dynamic rotor system. and M. 13–17 . 81–93.61 0.IV.” Journal of Sound and Vibration 117.Wauer and S. 1986. 278-283. Journal of Applied Mechanics. and L.Ames and L.J. “On the dynamics of cracked rotors: A literature survey. W. Krawczuk.G.432 3 890. Walston. 191–201. 173–191. Sorokin.342 569. 1987. C. 292299.489 63 of 446 Copyright @ Department of Electrical Engineering.066 0.926 0. 22.738 -0. 1976.A. Amplitude ratio Error REFERENCES [2] 1 Experimental analysis The vibration characteristics of simply supported rotating shaft in viscous fluid. Container 6.G. 1958. D.95 0. J. Variac 1053. No. “A potential technique in the theory of interaction between a structure and a viscous.462 890.213 3 1002. 1964. 1996.412 2 29110.G. 4: Schematic diagram of experimental set-up 1.M. W H.185-198. C. F. Kito. NIT Durgapur . Power supply [6] 4.8 Experimental analysis Error [7] [8] [9] [10] [11] 1 605. 313332.” Journal of Sound and Vibration (126)0. 121 (1956-9).663.312 CONCLUSION J. 405–417. Electricite 371-372 . Vol. 1970.D.Walston.654 534.F. W. 2001. 5 1 Theoretical analysis Theoretical analysis V=2.096 605.Wauer.” Archive of Applied Mechanics 62. The speed is controlled by a variac which is connected to the shaft motor from the end of the simply supported shaft the amplitude of the vibration was measured with the help of vibration pick-up and vibration indicator for simply supported shaft rotating in different viscous fluid and different . 1993. 404-409. R. R.742 2 11837. 3 1 1 2 4 3 1 6 [3] 1 [4] Fig.018 0. Rotating shaft I TABLE I.” Journal of Sound and Vibration (120)5.” ASME Journal of Applied Mechanics. 292–299. W. . Ostachowicz.H. part 1 – Theory.” Archive of Applied Mechanics 71. and J. “Dynamic stability of rotating shafts in viscous fluids. "Resonant vibrations of a submerged.806 0. Dimarogonas. “The effects of an annular fluid on the vibrations of a long rotor.951 11837. 1956 Achenbach.487 1 568. ANALYSIS RESULT OF SIMPLY SUPPORTED ROTATING SHAFT IN VISCOUD FLUID FOR DIFFERENT DAMPING COEFFICIENT OF VISCOSITY. 1990. W. “Coupled torsional and bending vibrations of a rotor with an open crack. Motor [5] 2. The effect of increasing the gap ratio increases the amplitude of vibration. compressible fluid. Vibration pick –up S E 5. “A survey of the dynamic behaviour of a simple rotating shaft with a transverse crack. J. Transactions of the Japan Society of Mechanical Engineers (in Japanese).” Journal of Basic Engineering 92 (4). “Coupled longitudinal and bending vibrations of a rotating shaft with an open crack. C.367 2 4597. Transactions of the Japan Society of Mechanical Engineers (in Japanese). Ames. Papadopoulos. 141 (1958-5). 1964. S.461 29110. Qu.276 [12] [13] V=0.” ASME. No. Amplitude ratio EXPERIMENTAL ANALYSIS A test are conducted on simply supported shaft which is rotating in viscous medium for determining the amplitude of vibration by varying damping coefficient of viscosity of fluid . Due to increase in coefficient of viscosity the amplitude of vibration decrease and also critical speed shifts towards left.0844 1 553.463 0. 24. Clark.” Journal of Fluid Mechanics 75 (1).

of Mechanical Engineering Faculty of Engineering Siksha‘O’Anusandhan University Bhubaneswar. Chinchalkar [1] has extracted the first three natural frequencies of the cracked beam to identify the crack using a finite element by considering the different boundary conditions and crack depth.akram@gmail. A compliance matrix is formulated by Tada et al. Over the time with the evolution of different methodologies for damage detection established the fact that. such as pattern recognition or data classification. Loutridis et al. Agarwalla. The most interesting feature of this artificial neural network (ANN) is the novel structure of the information processing system. vibration signatures. which are hard to solve using analytical or numerical methods. India amiyadash@yahoo. Odisha. The parallel computing capability and the ability to perform under changing environment make the neural network a potential tool to address applications. It is composed of a large number of highly interconnected processing elements (neurons) working in parallel to solve specific applications. the predicted results from the ANN model have been compared with the experimental results and are found to be in close agreement. ANN model. Saavedra et al. Ostachowicz et al. Krawczuk et al. In the present work. India sahabaz. [7] have proposed a finite spectral element method & wave propagation analysis to determine the modal parameters of a cracked Timoshenko beam.co. Kim et al. During 4 1 A S E Keywords— damage. The current feed forward ANN model has been trained with adequate number of data (vibration parameters) obtained from finite element analysis and experimental analysis. [2] have interpreted the dynamic behavior of the cracked structure theoretically and experimentally by a new technique based on instantaneous frequency and empirical mode decomposition. glass fiber reinforced composite. of Mechanical Engineering Faculty of Engineering Siksha‘O’Anusandhan University Bhubaneswar. The proposed neural network model for damage identification has been trained with requisite amount of data from different damaged scenarios of the glass fibre reinforced composite cantilever beam. The outputs are relative damage positions and relative damage severities. a glass fibre reinforced composite cantilever beam has been considered to develop the neural network models for structural damage diagnosis. through a learning process. experimental analysis the training. I INTRODUCTION Neural networks. [10] have come up with a computer based damage diagnosis system for concrete structures using Fuzzy set theory. with their remarkable ability to derive meaning from complicated or imprecise data. finite element analysis. The change in dynamic characteristics of the vibrating composite structures lay down the main platform for the damage diagnosis mechanism. the effects of the crack locations and sizes on the vibration behavior. the models have been fed with six input parameters i. Asst. He has modeled the problem using finite element and the component mode synthesis methods. NIT Durgapur . first three relative natural frequencies and first three mode shape differences. position and severity. Professor Akram Sahabaz. Dash. Odisha. Kisa [9] has investigated to trace the cracks and nature of cracks in a composite structure made of graphite fiber reinforced polyamide cantilever. the smooth functioning of all structural members is only possible when the development of damage of any form is prohibited. Finally. [4] on an aluminium sheet having micro cracks generated by compression loading and the deformation is tracked using the acoustic emission technique. of Mechanical Engineering Faculty of Engineering Siksha‘O’Anusandhan University Bhubaneswar. He has studied the effects of location and depth of crack and the volume fraction and orientation of fiber on the natural frequencies and mode shapes of the beam.Diagnosis of Damage in Composite Beam Structures using Artificial Neural Network with Experimental Validation Deepak K. India deepak_cet08@yahoo. In the current research. Professor Amiya K. [6] have proposed a method assuming an open and closed crack with triangular disk finite elements. He has analyzed the forced vibrations of the beam. Odisha. A modal analysis is conducted by Ravi et al. Various experiments have been performed on the cantilever beam with different damage characteristics.in Dept. UG Student Dept. [3] in damaged structure for determining the crack location and crack depth. They have applied the enhanced technique to diagnose the damage using the 64 of 446 Copyright @ Department of Electrical Engineering. [8] have developed a finite element stiffness matrix for the vibration analysis of the multibeam structure with different boundary conditions.com Abstract— Damage diagnosis techniques have been widely used in various engineering applications.e. Owolabi et al.e. I.com Dept. can be used to recognize patterns and detect trends that are too complex to be noticed by either humans or other computer techniques. which subsequently validate the results obtained from ANN technique. the deviations of the vibration signatures derived from the intact and damaged beams have been utilized to model the feed forward artificial neural network (ANN) with back propagation technique for prediction damage characteristics i. [5] have investigated the position and severity of crack for Al fixed-fixed and simply supported beams by analyzing the modal responses.

the flexibility matrix Cintact of the intact beam element can be expressed as. 1. which they have implemented on sandwich composite for the purpose of damage detection. the computational analysis software has been used to determine the vibration responses i. the natural frequencies and mode shapes of the damaged and undamaged composite cantilever beam.2) is for the intact beam. Shekhar et al. reliable by engaging the available information such as stator current. which are subsequently used for damage diagnosis. Saravanan et al. [11] have implemented the fault diagnosis in bevel gear box with the help of an artificial neural network. Where. [14] have coupled the neural network method with a combination of vibration and thermal damage detection signatures to develop a damage defection tool. The forces acting on the beam element for finite element analysis are shown in Fig. As far as damage detection in composite structural members are concerned. The relationship between the displacement and the forces can be expressed as. the application of artificial Intelligence (AI) techniques have not been explored much by the researchers.e. D= (1) Øj (2) Øj The displacement vector in equation (1. Under this system.damage attributes as building blocks to model the fuzzy inference system. Uj Cintact Where. The vibration responses are extracted from the experimental test structures for the modeling of neural network. EXPERIMENTAL ANALYSIS The experimental analyses have been carried out to measure the natural frequencies and mode shapes of the damaged and 65 of 446 Copyright @ Department of Electrical Engineering. [15] have developed a multi layer perceptron neural network based classifier for damage detection in induction motors which is inexpensive. The DWT technique has been amalgamated with the selective feature of energy spectrum for the development of the purposed fault detection algorithm. 1 Damaged cantilever beam element subjected to axial and bending forces. Ghate et al. Finite element analysis (FEA) has been performed on the cantilever beam with different damage characteristics to evaluate the modal parameters. [16] has determined the dynamic responses utilizing a model based on finite element analysis. The FEA tool follows the methodology for solution of the stiffness matrix Kc as mentioned in equation (4) and derive the modal response parameters of the intact and damaged beam members. u j – ui θj – θ i II. (1). From the literatures cited above. This gives the basis for the direction of the current research to apply AI technique i. scientists have made their effort to formulate different techniques for diagnosis of damaged structures. Fig. [12] have proposed a damage diagnosis technique for internal combustion engine using discrete wavelet transform (DWT) and neural network.e. The total flexibility matrix Ctot of the damaged beam element can now be obtained by. III. Covl = R11 -R12 -R21 R22 Cintact = 0 Le/EA 0 Le/EI 4 1 A Ctot = -R12 Cintact + Covl = Le/EA+ R11 (3) Le/EI+ R22 -R21 Through the equilibrium conditions. Agosto et al. wave let and proximal support vector machine. FINITE ELEMENT MODELING In the current section. Wu et al. the stiffness matrix Kc of a damaged beam element can be obtained as [16] (4) Kc=DCtot-1DT Where D is the transformation matrix and expressed as. The results from the FEA for the first three modes of the damaged beam with different damage configurations are compared with that of the experimental analysis and are presented in Table 1. it is observed that different diagnostic tools have been used by the researchers for damage detection in different types of engineering applications and structures. [13] have come up with a technique for online structure health monitoring of blades with axial flow utilizing neural network. uj – ui θj – θ i = Covl Uj Where overall flexibility matrix Covl can be expressed as. For damage diagnosis in structural members made from composite material. S E I = -1 0 1 0 0 -1 0 1 In the current analysis. artificial neural network to develop the methodology for estimation of damage parameters present in vibrating cantilever structures. R11: Deflection in direction 1 due to load in direction 1 R12= R21: Deflection in direction 1 due to load in direction 2 R22: Deflection in direction 2 due to load in direction 2. NIT Durgapur . Oberholster et al. The pictorial view of the cantilever beam with damage severity a1 subjected to axial and bending load has been presented in fig.

one output layer and eight hidden layers. (vii)Power amplifier. nine numbers of data set have been presented in table 1 corresponding to nine different damage configurations. This technique is an approximate steepest gradient algorithm in which the performance of the network is based on mean square error. diagnosing the Cardiovascular System and etc. (viii)Power supply. The modal parameters in terms of natural frequencies and mode shapes for the first three modes of vibration have been recorded by suitably positioning the accelerometer at the damaged sites across the length of the composite beam members. The relative natural frequencies.625).undamaged glass fibre reinforced composite cantilever beam with different damage characteristics. (iii)Composite cantilever beam platform. The input layer contains six neurons. I IV. SMD and TMD. 0. 1. 0. FMD. sales forecasting. The parallel computing capability and the ability to perform under changing Fig. The outputs from the neural model are as follows. (iv)Vibration Analyzer. The back propagation technique can be used to train the multilayer networks. Experiments have been performed on the damaged beam structures with different damage positions and damage severities to validate the results obtained from finite element analysis and the proposed ANN model. Relative damage position = “RDP” Relative damage severity = “RDS” Activation function.25. The inputs to the neural network model are FNF. 1 from the several experimental observations are for the three relative damage positions (0. NIT Durgapur . chosen in this work as the hyperbolic tangent function which is expressed in equation (5). environment make the neural network a potential tool to address applications. SNF. The error in each hidden layer just before the output layer in a direction opposite to the way activities propagate through the network have to be computed and fed to the network by back propagation algorithm to minimize the error in the actual output and desired output by adjusting the parameters of the network. IV A. modeling. 2. The multilayer neural system would calculate the change in error due to increase or decrease in the weights. Neural controller mechanism for damaged detection A back propagation neural model [17] has been proposed for identification of damage (i. relative damage severities) of structural beam members (Fig.The neural model has been designed with six input parameters and two output parameters. For the above said damage configurations the first three relative natural frequencies and relative mode shape differences derived from the FEA and experimental analysis have been presented and compared in the table no.75) and three relative damage severities (0. ex − e−x = f (x ) e x + e −x ARTIFICIAL NEURAL NETWORK MODELING Neural network has been successfully implemented in many industrial applications such as industrial process control. relative damage positions. electronic noses.375. which are hard to solve using analytical or numerical methods. The schematic diagram of the experimental set up has been shown in Fig. (ii)Delta Tron accelerometer.50 and 0. 3 Neural network of the proposed model The back propagation neural network has been made with one input layer. In order to train the neural network.e. 4 1 A S E Fig. The algorithm first computes each error weight by computing the rate of the error changes with the change in synaptic weights.3). The number of neurons in each hidden 66 of 446 Copyright @ Department of Electrical Engineering. where as the output layer contains two neurons. 2 Schematic diagram of experimental setup Out of several observations (relative natural frequencies and relative mode shape differences) obtained from the experimental analysis for measuring the modal response for different damage configurations of the composite cantilever beam. (vi)Function Generator.500 and 0. The required instrumentation for measuring the vibration parameters of the cantilever beam consists of eight numbers of elements and they are (i)Vibration exciter. the weights for each input to the neural system should be so adjusted that the error between the actual output and desired output is minimum. (v)Vibration Monitor. & relative mode shape differences presented in the table no. TNF.6.

(3). The experimental results are found to be in close agreement with the finite element results. 1 to 4 to compute the vibration parameters (relative natural frequency and relative mode shape difference) from the damaged model subjected to axial and bending load as shown in fig.e. FEA and ANN model have been presented and compared in table no. Fig. different materials can also be used for carrying out the diagnosis of beam structures subjected to various loading conditions and end conditions with the above mentioned technique. 3(a) to 3(c) for experimental and FEA method. The experimental section describes the complete instrumentation used to derive the relative first three natural frequencies and relative first three mode shape differences of the damaged composite cantilever beam with different damage scenarios. the ANN model can be used as a robust tool for diagnosing the structural integrity of engineering structures of different materials in various working scenarios. The back propagation trained feed forward ANN model with five numbers of layers with the complete architecture has been shown in fig. experimental analysis and ANN model. 1. The deviations in the mode shapes are quite significant at the vicinity of the damaged site present on the composite cantilever beam which acts as the basis not only to identify the presence of damage but also. 2nd. The results from both FEA and experimental analysis have been compared and shown in table no.e. 4 1 A S E I V. The finite element analysis has been carried out for computing the vibration parameters of the damaged and intact composite cantilever beams at the first three modes of vibration. The hidden layers i. However. (2). The neurons associated with the input layer of the network represent the first three relative natural frequencies and first three average relative mode shape difference. CONCLUSIONS The conclusions drawn by analyzing the results from the above mentioned sections are depicted below. (1). first three relative natural frequencies and first three mode shape differences and the outputs are relative damage positions and relative damage severities. The major deviations in the mode shapes have been observed at the third mode of vibration. The developed feed forward neural network model trained with back propagation technique has predicted the damage characteristics present in the composite cantilever beam effectively and the results are found to be close to the experimental results. Table 1 Comparison of damage characteristics of composite cantilever beam obtained from ANN. RESULTS AND DISCUSSIONS VI. 11 neurons and 9 neurons respectively. 1 and a close agreement has been found between the results. the models are fed with six input parameters i. From the analysis of results from table (1). Hence. During the training. The relative damage position.layers are different in order to give the neural network a diamond shape and for better convergence of results (Fig.. The table (1) presents a comparison of results derived from the FEA. 3). The experimental setup for vibration analysis has been shown in fig. experimental and the ANN model used for damage diagnosis in the composite cantilever beam model. FEA and Experimental method The FE model describes the required procedures from equation no. 3rd and 4th layer of the network comprises 9 neurons. the position and severity of the damage. it is evident that the neural network model gives results in close proximity to actual as compared to the finite element analysis. 67 of 446 Copyright @ Department of Electrical Engineering. 3 illustrates the neural network with its input and output signals for the composite cantilever beam. This section of the paper list down the discussion in regard to the results obtained from the FEA. The outputs from the experimental. NIT Durgapur . The proposed neural network model for damage identification has been trained with 600 patterns of data for composite cantilever beam featuring various states of the beam members. The variations of relative amplitudes of vibration have been illustrated with respect to the change in relative distance from the fixed end of the beam for first three modes of vibration in Fig. relative damage severities are represented by the two neurons of the output layer of the neural network. The number of neurons in each hidden layer has been decided using the empirical relation.

pp. “Forced vibration behaviour and crack detection of cracked beams using instantaneous frequency”. S. 247. Wu and J. 4. M.1451-1459.A. Hadjileontiadis. 3 (c) illustrate the variations of relative amplitude with the relative distance from fixed end for 1st. Paris and G. 13. “Free vibration analysis of a cantilever composite beam with multiple cracks”. 157. pp. Computers and Structures. 14.S. 2010. NIT Durgapur .C. Vol. Del Research Corp. Owolabi. 3 (a). Shafiq and A. 2006. E. pp 1139-1153. 3. 3 (a) A S E I Fig. “Faulted gear identification of a rotating machinery based on wavelet transform and artificial neural network”. Kisa. 2. Vol.S.3468–3481. “On-line fan blade damage detection using neural networks”. Saravanan . 20. pp. second and third mode shape differences Length of the composite beam element A Young’s modulus of the composite cantilever beam in transverse direction Area of cross section of the composite cantilever area 4 1 Fig. pp.78–93. 8. Vol. 2001. 3 (b). Swamidas and R. 38 (5). 344–360. pp. 85.M. Journal of sound and vibration. 36. and M. P. Shekhar and B. Pennsylvania. “Fuzzy set based crack diagnosis system for reinforced concrete structures”. 8862–8875. 10. TNF FMD. J. “The stress analysis of cracks hand book”. F.Cuitino.J . A. H. SMD.I. C. 2010. N. Serrano.1-22. Vol. and proximal support vector machine (PSVM)”. Vol. 64 (9). 2004. “Crack detection in beams using changes in frequencies and amplitudes of frequency response functions”. 16. Haykin. Vol. pp. Tada. Vol. FNF.J. “The dynamic analysis of a cracked Timoshenko beam by spectral element method”. 2009. NDT&E International. Composites of science and Technology. Ramachandran. Ostachowicz.Chan. “Neural Networks.Agosto. Oberholster and P. “Crack detection and Vibration behavior of Cracked beams”. M.SNF. Vol. 1097–1102. second and third natural frequencies Relative first. H. 2006. A. Dudul. Vol. S. pp. 411–419. Liew. 2nd and 3rd mode of vibration 68 of 446 Copyright @ Department of Electrical Engineering. Computers and Structures. Krawczuk. Expert Systems with Applications. S. N. 245250. S. Expert Systems with Applications. P.1828–1844. 17. 2001. D. 12. 3 (c) Fig. “A study of the effect of micro crack on the vibration mode shape”. V. “Vibration analysis of a cracked beam”. B. 22. W. W. M. J. pp. Cecchini. Prabhu. 2003. Seshadri. D. M. pp. 37. 3 (b) Fig. Vol. Vol. 2008. 2005. Composites: Part B.J.N. 15. pp. Ghate. Heyns.D. Applied Soft Computing. 79 (16). 10. Krawczuk. 1990. M. pp.J. “Fault diagnosis of spur bevel gear box using artificial neural network (ANN).264 (5). 375-381. Chinchalkar. Vol. 217–225. 7. Journal of Sound and Vibration. “Neural network based nondestructive evaluation of sandwich composites”. 11. pp. 417-429.K. Kim and G. Loutridis. Vol. Journal of Sound and Vibration.REFERENCES 1.S. A.V.S. “Determination of crack location in beams using Nomenclature natural frequencies”. 1992. Saavedra and L. “Crack detection and vibration characteristics of cracked shafts”. Palacz. pp 1391-1402. Engineering Structures. 36. TMD Le E Relative first. 2000.R. 2007. Vol. 6. M. Hellertown. S. Douka and L. 1973. 5. V. 265. Hong. Computers & Structures.N. 39. 9. Kumar Siddabattuni and K. Ravi and K. G. “Optimal MLP neural network classifier for fault detection of three phase induction motor”. 2003. Kim. Y. Ostachowicz. Irwin. pp. Journal of Sound and Vibration. PearsonEducation. A comprehensive foundation”. Vol. Mechanical Systems and Signal Processing.

When the femoral head section of the Weight bearing surface Fig. [2] have investigated a theoretical model to predict the 69 of 446 Copyright @ Department of Electrical Engineering. the mechanical properties of the prosthesis play a vital role for their optimum performance at the place of fixation. finite element analysis (FEA) has been carried out on a stainless steel hip prosthesis to find out the stress induced on the prosthesis and at the interface of the prosthesis and the femur bone. Dash. NIT Durgapur . Agarwalla. when the normal biological femoral head of the femur bone of a human being does not function properly and produces severe pain at the hip joint (Fig. Acetabulum (Socket) 4 1 A INTRODUCTION Femur (Thigh Bone) S E The hip prosthesis is generally implanted in humans. 1 View of the ball and socket hip joint Pelvis I Hip Prosthesis Femur Fig. 2 (b) Pictorial view of the implanted hip prosthesis at the hip joint The number of patients undergoing pelvic surgery for hip prostheses is growing due to increasing life expectancy. Harish Ch. In the current research. Siksha‘O’Anusandhan University Bhubaneswar. India femur bone ceases to function. deteriorating the movement of the hip joint of the patient. Stress. India *amiyadash@yahoo. The prosthesis is a mechanical element which is made from either stainless steel or titanium. of Mechanical Engineering Faculty of Engineering. Besides biocompatibility. of Ortopaedics. However. Deepak K.Finite Element Analysis of Hip Prosthesis for Identification of Maximum Stressed Zone *Amiya K. Stainless steels. 2 (a). They have considered a threedimensional physiological human body and carried out the analysis to measure wear at the hip joint following the contact stress variation from the finite element analysis. clinical researches have shown that the function of hip prosthesis deteriorates after long term use due to loosening of fixation. Uddin et al. [1] have introduced a wear prediction model with the aid of the finite element analysis. failure I. hip prosthesis. Bhubaneswar. Rabiteja Patra. Odisha. 2 (b). The view of the implanted prosthesis at the hip joint has been shown Fig. This establish the fact that. 2 (a) Pictorial view of the hip prosthesis Fig. The hip prosthesis used in the current work made of stainless steel which. Institute of Medical Science & SUM Hospital Siksha‘O’Anusandhan University. Baleani et al. Odisha. Abstract— Hip prosthesis is an important tool for reducing pain and restoring function of the hip and femur of patients suffering from musculoskeletal disorders such as arthritis. In the present work. cobalt–chromium alloys are the common metals used for applications in the human body in the form of prosthesis and implants. From the analysis of the result. titanium. The generation of wear and wear debris and loosening of the stem of prosthesis in total hip replacement surgery posses a great challenge for the successful long term performance. Metals are mostly used as biomaterials to produce implants and prostheses for fixation of bone joint disorder and fractured bone since last few decades. then implanting the prosthesis on the femur bone gives the required mobility to the patient. Das Jagannath Sahoo Dept. mainly comprises of four parts such as stem. finite element analysis (FEA) of the prosthesis has been carried out by simulating the actual conditions and the effect of stress on the performance of the prosthesis has been evaluated. the failure of the hip prosthesis starts at the maximum stressed zone resulting in loosening of the prosthesis and thereby.com Dept.1) when subjected to load. it is observed that the maximum stress is generated at the contact region of the femur bone and the prosthesis. femoral head and acetabular cup which has been shown in Fig. neck. Pelvis (Hip Bone) Femoral Head (Ball) Joint Capsule Keywords— finite element analysis.

They have stated that. They have compared their finding with the previous published literature and expressed that. cobalt chrome and ceramic. [10] have investigated the effect of three-dimensional prosthesis shape on the response and failure probability of a cemented hip prosthesis. [6] have studied and discussed the advantages of ultra high molecular weight polyethylene (UHMWPE) prosthesis over the poly tetra fluoro ethylene (PTFE) and stainless steel bearing combination prosthesis.al. In the present analysis. In their analysis they have used four different hip prostheses with stem of different geometries. It comprises of rounded head of the femur joining the pelvis in a ball and socket arrangement. Chao et al. wear debris from the head and acetubular cup and loosening of the prosthesis at the neck of the femur bone are found to be the prominent cause of failure. The analysis of the prosthesis at the hip joint using finite element analysis package can throw some lights in finding out the reasons for failure due to induced stress.[5] from Computerized Tomography (CT) scan data. [15] have done static & dynamics finite element analysis of stems of hip prosthesis to calculate the safety factors for fatigue life. finite element analysis has been carried out on the hip prosthesis fixed on a femur bone at the hip joint to find out the maximum stressed zone.al. [3] have investigated on the fatigue failure of implant fixation by numerical approaches. In this research they have compared the Von Mises stress profile of the biological femur with that of resurfaced femurs using different prosthesis material like titanium. Cabrer et al. In their analysis they have dealt with two different loading cases. Shinge et. allowing the detection of possible implant failures in a very short time. cortical and cancellous bone and bone cement elastic properties to evaluate the failure criteria at the bone cement and the prosthesis–bone cement interface. A S E I 4 1 II. Dowson et al. They have studied the mechanical behavior of the model and analyzed the influence of stepped structure on the stress distribution and concluded that the step structure reduces the stress in femur. which subsequently makes the loosening of the prosthesis resulting in failure of the device.al.884 nodes. Finite element simulations for stress profile have been done using the static analysis module of the simulation software. Abdulla et al.maximum stress induced in the stem by experimental test setup. walking and stair climbing. NIT Durgapur . Hung et al. FEA is a numerical tool that enables the investigation of the parameters related to the failure of the hip prosthesis which are not possible experimentally and theoretically. They have used parameters such as joint and muscle loading. During the design and development of the hip prosthesis the stress conditions are considered under different types of loading as those have got a greater effect on the mechanical behavior and life of the prosthesis. [11] have studied the wear rate of retrieved metal on metal hip prosthesis using co-ordinate measuring machine and a computer program. Pizzoferrato et al. FINITE ELEMENT ANALYSIS (FEA) HIP PROSTHESIS The hip joint is the largest load bearing surface in the human body. Heijink et. it has been found that researchers have made substantial effort to identify possible reasons for failure of the hip prosthesis. They have concluded that proper stress & strain distribution along the femur enhances the bone growth and keeps normal functioning of the femur. Finite element analysis (FEA) has been widely used in the field of bio-mechanics to analyze the mechanical behavior of hip prosthesis under different types of complex loading. Lord et al. They have demonstrated that the wear production caused by the alumina-alumina combination is very less as compared to metal-polyethylene coupling and also mentioned that alumina-alumina combination is thermodynamically stable and chemically inert unlike the metals and polymers. In the current work. They have analyzed the geometric characteristics. and a micro structural study of the stem material. A computer algorithm based on finite element analysis and continuum damage mechanics has been proposed to quantify the fatigue damage rate of cement mantle under physiological conditions. [9] have developed and described a new method single-point imaging (SPI) technique which. a solid CAD finite element model of the hip prosthesis for 30 year old male person has been developed using the three dimensional modeling software. From the published work. Measurements of the prosthesis have been done and are 70 of 446 Copyright @ Department of Electrical Engineering. hip prosthesis is subjected to a wide range of time dependent loads during normal activity once it has been implanted. [7] have conducted study under constant load conditions in a prosthesis model which resembles to a hip prosthesis. fracture surface. It’s applications include the design and development of hip prosthesis and fixation devices.[14] have compared the force analysis of the Somchesi Modular Prosthesis (SMP) and the prosthesis developed by authors by changing the steel ball of SMP with ultra high molecular weight polythene (UHMWPE). Stress has been computed using beam theory and finite element analysis (FEA). Strain measurements have been done to validate the theoretical and finite element calculation. Fatigue tests have been performed to authenticate the theoretical results about the fatigue stem performance. [12] have investigated the reasons of fracture of cement less hip prosthesis model manufactured with Ti6A14Valloy. the wear rates measured by them are of higher order as compared to others. Nicolella et al.[13] has evaluated the effect of stress on the prosthesis design for the proximal femur. [8] have analyzed the biological affect of alumina wear particles found in tissues around hip joint prostheses. enable the direct visualization of the polymeric elements of the implants. A FE model of hip joint prosthesis with stepped stem has been generated by Tan et. Their results indicate that the most of the time the fracture has been initiated at the junction radius of the neck and the collared device due to a fatigue mechanism. Barbour et al. it is evident that wear. By analyzing the literatures mentioned above. The meshing has been done using triangular elements with 40. Hence beam theory can be used to estimate the maximum stress. Kayabasi et. [4] have performed finite element analysis of the implanted proximal femur to examine the stress profile in cemented hip prosthesis with different design. They have stated that the (UHMWPE) prosthesis produce less wear debris as compared to the PTFE prosthesis.al. Analysis has been done using analysis software & it has been observed that displacement & equivalent stress on UHMWPE ball are more compared to the stainless steel ball. Thus this tool has become an ideal tool for prediction of failure of hip prosthesis.

iii) A point load of magnitude 400N has been acting at the centre of the head of the prosthesis at an angle of 45° to the horizontal axis in the downward direction iv)An uniformly distributed load of magnitude 600 N has been acting at the bottom surface of the femur bone in the upward direction. Head Neck Stem Femur Bone Fig. neoprene and bone tissues whose mechanical properties have been shown in Table 1. head and acetubular cup are presented in Table 1. NIT Durgapur . Young’s modulus and poisson’s ratio are also presented in Table 1. 3(b) Enlarged view of the Maximum stressed zone III. The prosthesis model under consideration has been made of three distinct materials such as stainless steel. The outcome from the analysis has revealed that.e. 4. 3(a) Meshed model of hip prosthesis with the femur bone 4 1 A Table 2 Variation of Maximum Stress across the length of the stem and neck of the prosthesis with System Parameters S E I Fig.Table 1 Material properties used for analysis of Hip Prosthesis presented in Table 1. 4 Stress distribution of the prosthesis with the femur bone 71 of 446 Copyright @ Department of Electrical Engineering. The magnitudes of stress developed at different section of the prosthesis have been presented in Table 2. Boundary conditions for the analysis have been laid down and are i) the displacement in the z-direction has been made zero ii) the bottom surface of the femur bone has considered as a fixed surface. By analyzing the Fig. the maximum stress has been induced at the interface of the femur bone and the prosthesis which has been presented in Fig. The stress induced at different section of the stem and neck of the prosthesis has been mentioned in Table 2. RESULTS AND DISCUSSION The results derived from section 2 of the manuscript have been briefly discussed and presented in this section. neck. The assembled model of the prosthesis with the femur has been analyzed using the finite element simulation package to measure the stress induced across different cross sections of the prosthesis following the boundary conditions as cited in section 2. The materials used in the different parts of the prosthesis model and the bone tissues along with their mechanical properties i. The geometrical dimensions of the different parts of the hip prosthesis such as stem.

2011. Vol. 28(4).5 mm. Van der Toorn.5 mm. Roberto Muccini. Morpho metric and micro analytical analyses of alumina wear particles hip prostheses. Predicting the wear of hard-on-hard hip joint prostheses. 203-215. it is observed that maximum stress along the stem has been developed at the interface of the femur bone and hip prosthesis and is of the order of 21.S. [8]A. Prosthesis designs and stress profiles after hip resurfacing. 22 (10). Advances in engineering software. T. Mohd S. Vol. Mulik.033 MPa. The outcome of the research can be helpful to orthopedic surgeons to take measures while implanting the prosthesis at the hip joint. On the neck the maximum stress magnitude is 25. Pizzoferrato. The view of the hip joint with the artificial prosthesis has been shown in Fig. Mauro Ansaloni. Jian-Horng Chen. Failure analysis of a Ti6Al4V cementless HIP prosthesis. the magnitude of stress induced and their distributions are significantly non-uniform. “Finite element modeling and analysis of a new cemented hip prosthesis”. 72 of 446 Copyright @ Department of Electrical Engineering. 583-590. pp. 3 (a). Engineering Failure Analysis. A. 39 (7). Zhang X. March 17-19. we have identified the maximum stressed zone (section AA of Fig. 4. Wear. Joyce.192-200. Thus. 2011. This joint is known to be maximum load bearing surface in a human body.e. ME Zobitz.K. pp. Vol. Vol.865–871. 14 (8). 301 (1-2). Ramos-Cabrer. Journal of Biomechanics. pp. 3 (b). A. MRI of hip prostheses using single-point methods: in vitro studies towards the artifact-free imaging of individuals with metal implants. Vıctor Lopez. Nargol. 1985. Vol. Eaching Kumar S Mahadik. Since the different parts of the prosthesis i. Normally the patient undergone pelvic surgeries to implant the prosthesis at the hip joint often complain about improper functioning of the joint. 250-257. Hence it can be concluded that due to the persistent maximum stress occurring at the interface of femur and stem of the prosthesis. Hejink.363 at a neck length of 2. Barton. The enlarged view of the stress analysis at the interface of the femur bone and the prosthesis has been presented in Fig. it is observed that maximum stress of the order of 21. The influence of contact stress on the wear of UHMWPE for total replacement hip prostheses. wear debris from the hip prosthesis and loosening of prosthesis are the major reasons for the failure of the artificial hip joint. Fig. Tardan Giha “ Finite element analysis of cemented HiP Arthoplasty: influence of stem tapers”. Hong Kong. A. The meshed model of the prosthesis with the femur bone used for FEA has been shown in Fig. pp. Fig. III. Pramod V. 477-483.S.J. Aitavade. Volumetric wear assessment of failed metal-on-metal hip resurfacing prostheses.C. 1265–1278. pp. [7]P. Computer Methods and Programs in Biomedicine. M. proceedings of the international Multi conference of engineers and computer scientists. 2 (a) presents the photographic view of the metal on metal hip prosthesis with it’s different parts that has been used in the current research.C. Vol. Zhang. International Journal of Advanced Biotechnology and research. [14]Abhijit. Vol. stem. 278-285. Nicolella.1097–1103. 104 (3).. neck. pp. R Nuyts. the research output is validating the clinical diagnosis. S. Even the present outcome can be utilized by prosthesis design and development organization to remodel their design to prevent the failure of the prosthesis. Endurance verification of custom-made hip prostheses. Barbour. From clinical diagnosis in most of the cases. 2006. [12]Jesu´s Chao. 16 (3). [2]Massimiliano Baleani. Luo J. [5] Tan R. Marco Viceconti. 2013. it is found that the failures have been due to the loosening of the prosthesis from the femur bone. Wear. Biomaterials. this may be the reason for loosening of the device from the femur.REFERENCES data from Table 2. pp. Finite element analysis of modified Hip Prosthesis. 76 (2). A. 326-332. 14 (5). Vol. Wallbridge. D. 1995. Alias. The complete stress analysis of the assembled prosthesis and femur model has been presented in Fig. Vol. Wear. Hamid Katoozian. 1993. NIT Durgapur . Mohd N. KN An. Shinge. J. Computer simulation on fatigue behaviour of cemented hip prostheses: a physiological model. Journal of orthopaedic surgery. Vol. at the interface of femur and stem of the prosthesis. Stea. Davy.C. Fan H. International Journal of Fatigue. 2(b). Vol. [6] D. a finite element analysis”. Er Zincanli.F. 272 (1). 3(b)) i.Wear.V. Mhod Asri. [3]Jui-Pin Hung. Vol. 2011.103-113. 2007. The effect of three-dimensional shape optimization on the probabilistic response of a cemented femoral hip prosthesis. Uddin. Dowson and N. 37(7). G. D. [13]A. 2004. J. Hsiu-Lu Chiang. Vol. [9]P. head and cup are with distinct geometrical dimensions. Gualtierit and S. pp. Toni. 2000. pp. From the FEA. Abdullah. K.79– 87. IV.181-183 (part 1). Dwight T. [11]J. Wu F. Lord. CONCLUSIONS The conclusions drawn by analyzing the above sections are described below. Sudanese.732-736. [15]Kayabasi Ogut. Anasane.P. Chen J. [10]Daniel P. Vol.M. James Shih-Shyn Wu. L. 2 (2). Squarzoni. Ben H. R.J.e. pp. Fisher. van Duynhoven. pp. pp. Magnetic Resonance Imaging. pp. 1 depicts the different parts of the human hip joint which resembles to a ball and socket joint. 2004. 22 (8). Wear. 4 1 A S E I [1]M. Langton. [4] Abdul H. Thacker. IMECS 2010.033 MPa has been developed on the stem at a length of 147. 2008.M. Nicolay. Ek nath N. 2006. Laboratory wear tests and clinical observations of the penetration of femoral heads into acetabular cups in total replacement hip joints: I: Charnley prostheses with PTFE acetabular cups. 822–830. BF Morrey. Eondep S. Nigrisol. so as to avoid failure of the prosthesis. “Three dimensional finite element analysis of bone stress distribution around the hip joint prosthesis with stepped stem” Sheng Wu Yi Xue Gong Cheng Xue Za Zhi. From the FEA. pp.

. 17]. sector-III. After the great success in PAN this technique has also been tried to apply in dynamic Network too.Multi Channel Personal Area Network(MCPAN) Formation and Routing Debabrato Giri1 Uttam Kumar Roy2 Dept. Kolkata-700098. WPAN. being referred to as the 'branching factor' of the hierarchical tree. of Information Technology Jadavpur University. Compact Routing etc.request’ primitive . In that A. As a result of this feature WPAN allows small. Low memory devices. India debabrato. tree routing is simplest as it does not require any Routing table.. I. In tree network topology a central 'root' node (the top level of the hierarchy) is connected to one or more other nodes that are one level lower in the hierarchy (i. Compared to the other kinds of routing. and a reasonable battery life. NIT Durgapur . simple mathematical equations are used for address assignment and routing. the number. Plot-8. India u_roy@it.giri@tcs. We have addressed the network depth problem in our paper “Address Borrowing in Wireless Personal Area Network” [8].The devices forming this networks are Low rate. We have addressed this problem in our paper “Address reorganization in wireless personal area network”. while each of the second level nodes that are connected to the top level central 'root' node will also have one or more other nodes that are one level lower in the hierarchy (i. Index Terms—PAN.com2 Dept. Kolkata-700098. power-efficient. of Information Technology Jadavpur University. reliable data transfer. also with a point-to-point link. sector-III. Distance vector Routing [1. NWK layer will attempt to discover networks operating within the device’s POS by performing an active scan over the channels 73 of 446 Copyright @ Department of Electrical Engineering. The main advantage of tree address allocation is that it does not require any routing table to forward a message. f. short-range operation. Mesh. Salt Lake Campus Block-LB. the second level) with a point-to-point link between each of the second level nodes and the top level central 'root' node. Tree and Mesh are the two common topologies for this network. II. WPAN require relatively less or no infrastructure compared to wireless local area networks (WLANs).jusl. Now in some other network configuration the maximum breadth of the network may be attained but the maximum depth of the network may not be attained (because of the asymmetric nature of the physical area) at that part and hence address lies unused. Salt Lake Campus Block-LB.Multi Channel. each node in the network having a specific fixed number. Low powered. The hierarchy of the tree is symmetrical[2].in Abstract—802. Networkdiscovery When a ZigBee co-coordinator is powered up and set up for forming a network the 1st thing that it used to carry out is check for discovery of any network within its transmission range using ‘NLME-NETWORK-DISCOVERY.com1 mail_debabrato@yahoo. INTRODUCTION Wireless Personal Area Networks (WPANs) are used to communicate information over relatively short distances. 11. ZigBee Alliance [18] has provided the specification for Network layer and Physical layer (PHY). The main problem in applying this tree routing in this dynamic network is all the devices are connected to each other using a single link with parent child relation ship and if the link in broken then that part will be disconnected from the rest of the network and transmission will be disrupted. Plot-8.e.15.ac.e.Tree. of nodes connected to it at the next lower level in the hierarchy. the top level central 'root' node being the only node that has no other node above it in the hierarchy. f. Tree and Mesh are the two topologies used in this kind of network. Here In this paper we have provided a unified Multi channel routing scheme which can be easily applied to tree network so that the network can be used in dynamic application space and overcome the link disruption problem by Multi channeling but without adding any extra overhead of having a routing table.4-2003 is the standard for Wireless personal area network(WPAN). 4 1 A S E I Address OVERVIEW OF TREE ROUTING Routing. In tree network no routing table is required for routing. the third level) connected to it. Reorganizing. inexpensive solutions to be implemented for a wide range of devices.

request’ primitive. The others device are router capable device which can have children either end device or router capable device. The scan is performed by means of the MLMESCAN.confirm primitive with the Status parameter set to STARTUP_FAILURE. IfRm 1 Lm d 1 1 C m Rm C m . nwkMaxRouters (Rm). For a given values for the maximum number of children a parent may have. Network formation After completing the Network search the next step is forming the network.0x0e Description The 27 least Significant bits indicate Which channels are to be scanned A value used to calculate the length of time to spend scanning each channel: Table 1 : The result of the scan is reported by the . NIT Durgapur . ScanChannels. Network Join Once the network is set up the PAN coordinator opens the network for other device to join by the ‘NLME-PERMITJOINING. nwkMaxDepth (Lm). Device willing to join the network first checks for existence of any network within its radio range using a energy scan by the primitive ‘NLME-ED-SCAN. request primitive to the MAC sub-layer with the ScanType parameter set to indicate an energy detection scan and then issues the primitive again with the ScanType parameter set to indicate an active scan. The network descriptor information fields contain the information about all the PAN network operating in the current network space. It has the following 5 parameters.request Parameters attribute nwkExtendedPANId is equal to 0x0000000000000000. After getting this request from the next higher layer the ZigBee Network layer If the device is to be initialized as a ZigBee coordinator. then Cskip(d). Otherwise (1) 1 Rm Network addresses should be assigned to router-capable child devices using the value of Cskip(d) as an offset. Once a suitable channel and PAN identifier are found.request primitive to the MAC sub-layer to set the MAC PIB attribute macShortAddress. on receipt of the MLME-SCAN. D. the maximum depth in the network. If no suitable channel or PAN identifier can be found. RejoinNetwork.request primitive. SecurityEnable 4 1 S E I A The 64 PANID is used for specifying the network to which it wants to join to. To do this. If the NIB ExtendedPANId.request’.Rm C skip (d ) . This sub block is calculated based on the following parameters. is: 1 C m . Distrubuted address assignment Each router capable device in the network is handed over a sub block of address by its parent.request’. ScanDuration. B. The NWK layer will pick a PAN identifier that does not conflict with that of any network known to be operating on the chosen channel. C.request primitive to request that the device start a new ZigBee network with itself as the coordinator and subsequently make changes to its superframe configuration. d. After the completion of the active scan. the NLME issues the MLMESCAN. CapabilityInformation. the NLME requests that the MAC sub-layer first perform an energy detection scan and then an active scan on the specified set of channels. NLME-NETWORK-DISCOVERY.specified in the ScanChannels argument for the period specified in the ScanDuration parameter. Zigbee uses the NLME-NETWORKFORMATION. the NLME issues the NLMENETWORK-FORMATION. nwkMaxChildren (Cm). This information is useful for setting up a new network. Name ScanChannels Type Bitmap Valid range 32-bit field ScanDuration Integer 0x00 .(Lm d 1). and the maximum number of routers a parent may have as children. End device has no routing capability and can have no child node. the NLME will choose 0x0000 as the 16-bit short MAC address and inform the MAC sub-layer. essentially the size of the address sub-block being distributed [18] by each parent at that depth to its router-capable child devices for a given network depth. This allows the other node to join the network.confirm primitive from the MAC sub-layer.confirm’ primitive. the NLME selects a suitable channel. Once the suitable network is found the device tries to join that network by using the primitive ‘NLME-JOIN. A parent assigns an address that is 1 greater than its own to its first router-capable 74 of 446 Copyright @ Department of Electrical Engineering. This information includes the 64 bit PAN Identifiers of the PANs and the logical channels used by them. Devices that join the network may be of two kind end device or router capable device. To do this. the NLME issues the MLME-SET.NLME-NETWORKDISCOVERY. this attribute will be initialized with the value of the MAC constant macExtendedAddress.

So if we try to use this network in a dynamic network where each devices are movable and still needs to continue data transmission then there will be disruption due to breaking of the link between its parent or child when the device is crossing the transmission range . A device only communicates with its Parent and Childs. Rm=2 Simplicity-It’s a very simple network no complex path is formed. 14 Figure 2: Address Allocation F. 4 1 Table 2 0 A 16 1 2 9 17 S E I 24 Transmission range 3 4 6 7 10 8 13 H.Now if node 9 and node 13 started moving in the indicated path then after some time they will not fall outside each others radio transmission range and thus the link between node 9 and node 13 will be broken and thus they transmission halted until node 13 rejoin the network using any other node as parent.(R th 1) 1 a child end device then the address N of the next hop device is given by: N=D. In Fig2 node 13 is connected with node 9 and 14 only where node 9 is the parent and 14 is the child.17. NIT Durgapur . A maximum of Rm of such addresses shall be assigned. Even though node 17 and node 10 are in the radio transmission range of node 13 they are not connected because of the limitation of Zigbee address distribution and tree routing algorithm. (6) Otherwise. Limitations of ZigBee tree routing Now as we can see although there are 27 channels available (in best case where there is no network in the given transmission range) available for communication only a single channel is chosen and used for forming the Network and subsequent data transfer. Tree/Hierarchical Routing If the destination is a descendant of the device.10 and 11 only even though node 2. Depth in the Network 0 1 2 3 4 Offset Value [Cskip(d)] 15 7 3 1 0 Scalability – The tree network can grow more than that of mesh networks.node12. Lm=4. Say for in Fig2 if node9 used to be connected with node2.C skip ( d ) (5) If it is ARth Aparent Cskip (d ). Now each device is connected to any other device (its parent or child) with only a single link (Channel). R th Router address at depth D is: If it is determined that the destination is a descendant of the receiving device. Rm n E. Advantage Network formation Let Cm=2. then it is checked that if the destination is a child end device using the equation D A Rm .Also the portion of the network will be unreachable. the device shall route the frame to its parent. If the destination is not a descendant.6 are in the radio transmission range . In our proposed algorithm we will try to overcome this problem. then a destination device with address D is a descendant: A D A C skip (d 1) (4) III. An AParent C skip ( D ).hild device. For a ZigBee router with address A at depth d. Subsequently assigned addresses to router-capable child devices are separated from each other by Cskip(d). (3) D ( A 1) N A 1 C skip (d ) C skip (d ) (7) 1 n Cm Rm And A parent is address of parent G. Similarly node 9 is only connected with node 1. if the following logical expression is true.node 6 and node 13 be connected to node 17 and node 10 as they are also in the radio transmission range then even though the node started in 75 of 446 Copyright @ Department of Electrical Engineering. the device shall route the frame to the appropriate child. PROPOSED MULTICHANNEL ROUTING In this scheme a node will be allowed to have multiple parents and multiple children so that if the link between a parent to its child get disconnected even then it will be able to stay connected with the network via the other channels. The next hop address is given by: (2) Network addresses to the end devices are assigned using the following equation [18].

request’. Now if a new node is powered and it also wants to start a new network with the same PAN ID then it have to chose a separate channel for transmission. It also chooses a suitable channel to be used for communication. As per our proposed scheme a PAN will be identified with combination of two parameters. Then it can start another PAN with the same PAN ID but with a different channel say channel 2. After that it will start accepting the node joining the network using the primitive ‘NLME-PERMIT-JOINING. SecurityEnable So that multiple PAN can have same PAN Identifier but the combination of PAN ID and Channel identifier would be unique. 0 4 1 2 A S E I B.The final network with its address distribution is depicted in Figure 3. NIT Durgapur . There are 27 channels available for communication but only a single channel is used by a particular PAN. Then only it accepts the request. PAN Identifier Channel Identifier ExtendedPANId.request’. ScanChannels.9). This allows the other node to join the network. The final network along with its address distribution is depicted in Figure 4. In this multi channel pan we are proposing a scheme which will allow a node to connect with multiple neighbors so that it can stay connected inspite of its mobility. 16 1 3 4 9 6 7 10 8 17 24 13 14 Figure 3: PAN 1 Now say node 2 of figure 3 wants to start the network with the same set of nodes It will scan the underlying channels and select a separate channel(2) and use the same PAN ID (ABC11234x). All the nodes that joins the network uses this channel for sending /receiving message. This primitive needs to be modified to add one extra parameter to specify the channel along with the PAN ID. On receiving a network join request it will first check if the PAN ID and channel ID is matching with its PAN ID and selected channel. MCPAN Network Formation: The Network discovery part will remain as it is and there will not be any change to form a network the PAN coordinator will perform the network discovery and then it will go for forming the network. In this paper we have explained the proposed algorithm with 3 concurrent PAN. 76 of 446 Copyright @ Department of Electrical Engineering. That is if as a result of Network discovery a node finds that there is a network with PAN ID ABC11234x and its operation on channel 1 and there are no other PAN with this ID. PAN coordinator chooses a 64 bit PAN Identifier by which the underlying network is identified.2. CapabilityInformation.request’ primitive.request’. A. Device willing to join the network for the 1 st time first checks for existence of any network within its radio range using a energy scan by the primitive ‘NLME-EDSCAN. traversing in the indicated path there would have been paths for transferring data within the network. MCPAN Network Join Once the network is set up the PAN coordinator opens the network for other device to join by the ‘NLME-PERMITJOINING. Now say node 0 first powered up and finds no network is present it does an energy scan and finds that all the channels are free. In our given operating space say there are 3 devices which are capable of setting up a network of its own marked in rose color (0. Once the suitable network is found the device tries to join that network by using the primitive ‘NLME-JOIN. The next step it will do is choose a particular channel (1)and PAN ID (ABC11234x) to setup its own network .The only change is even if there is a network within its radio range even then it can chose the same PAN Identifier but it must chose a separate channel for communication. This helps us in setting up multiple PAN existing within same network area. Channel ID RejoinNetwork. ScanDuration.

It will have the following 3 fields. NIT Durgapur .14).9 16 1 0 17 16 18 24 19 PAN ID Channel ID Network Address Primary Network The table maintained at node 0 of Figure 3 will look like the following .24.10.14) in normal PAN but in Multichannel PAN node 13 is now connected with 4 node(9.request’. Each address is associated with a particular channel. To support this each router capable node needs to maintain a special table called Multichannel Map.13. If we look at Figure 2 which shows a normal PAN we can find that node 9 is connected with only 3 node 1. The resulted network is shown in Figure 6 Now say node 9 of figure 3 wants to start the network with the same set of nodes It will scan the underlying channels and select a separate channel(3) and use the same PAN ID (ABC11234x). Now because of this additional neighbor even if node 9 and 13 started moving in the indicated path marked in blue and come out of each other transmission range even then the will stay 77 of 446 Copyright @ Department of Electrical Engineering. After that it will start accepting the node joining the network using the primitive ‘NLME-PERMIT-JOINING. A 10 11 9 1 3 4 0 2 5 6 7 16 17 S E I 24 8 16 1 2 3 4 0 9 6 7 17 10 8 24 13 14 Figure 6: Multichannel network with all the connectivity Figure 5 : PAN 3 Please note that in this scheme a node is having multiple network address.17. 4 1 After the completion of network formation it may form a network depicted in Figure 5.At max this table can have 27 entry 1 for each channel 24 3 2 PAN ID Channel ID Primary Network Network address ABC11234x 1 Y 0 ABC11234x 2 N 9 ABC11234x 3 N 10 6 25 7 Table 3 : Multichannel Map Figure 4 : PAN 2 Now if we see the whole set of connectivity we can find that the no neighbor a node can have has been increased considerably to show that we can merge the 3 PAN .and 13 but with our proposed modification in Multichannel PAN node 9 now is connected with 5 node(1.2.10. Similarly Node 13 was connected with 2 node(9.17).

16. Reactive Routing a. 2=>1 =>0 =>16 =>17 16 2 10 It Will 1st try to send the message using the its Primary network that is PAN 1 (Channel 1). Source scheduled b. In this case node 6’s address is 2 and Node 13’s address is 17. C. 6=> 2 =>1 => 9 =>13. Source scheduled: . 1. Proactive Routing 9 Reactive Routing overview: In this mode a node will use its primary network first to send its message if it finds that the message is not delivered then it tries to send the massage using the 2nd Network and if that is also failed then it tries with the other network and so on until all the network is exhausted or the message has been delivered. Figure 3) in this Node 6’s address is 16 and Node 13’s address is 6. 0 17 There are two version of this routing.10 in this case. 16 1 78 of 446 Copyright @ Department of Electrical Engineering. And it will follow the following route. a.It will try to send the message using the path 16 => 0=> 1=>2 =>6. Now if the link between node 9 and 13 is broken because of the movement or any other reason then Node 6 will eventually receive a failure send message.connected via other node 24. Routing overview in MCPAN: There are two ways with which the routing could be done one is reactive mode and the other is proactive mode. and its Multichannel Map is as follows. In this paper we will only discuss about the source scheduled routing only. In this example I have used only 3 channels if we increase the no of channel then it will have more neighbor and hence better connectivity and mobility. NIT Durgapur 24 . Channel ID Primary Network Network address ABC11234x 1 Y 6 ABC11234x 2 N ABC11234x 3 N 6 4 1 25 7 A S E I PAN ID 24 19 After receiving the failure notification it will send the message using PAN 3 Figure 5.But as the link between node 2 and 6 is broken then this message transfer will also fail. On receiving this it will send the message using Pan 2(Channel 2. The reactive mode again is divided into two sub categories source scheduled and Failure point scheduled. It will use the following path to send the message.In this the source node takes the responsibility of the fail transmission and tries to send the message a separate channel.2. Suppose Node 6 of Figure 6 wants to send a message to node Node 13. Failure point scheduled 2. Lets discuss the algorithm using an example.1. 11 9 1 0 16 0 2 3 16 1 9 17 4 6 7 10 8 2 6 17 24 4 3 5 7 8 13 14 24 3 2 16 18 1.

IEEE Signal Processing Magazine. [11] [12] [13] [14] [15] [16] [17] REFERENCES [18] A. T. Das.Uttam Kumar Roy. Srivastava. C. LA. July 2003. “Adaptive clustering protocol for sensor networks”. Unpublished IEEE Stndard 802.D Amis. that is demand-driven (e. Energy aware operation of wireless devices is the dominant theme in [14].15. These routing protocols have to deal with the challenges of wireless networks. high error-rates and often energy and memory constraints. and A. Los Angeles. E. B.4. Proceedings of the 2 nd IEEE Workshop on Mobile Computing Systems and Applications. R. E.4”. “A Review of Current Routing Protocols for Ad-Hoc Mobile Wireless Networks”. 4 1 A S E I CONCLUSION C. ISPAN' 99. Authors provide one of the first studies of the MAC sub layer while the recent paper [15] is a comprehensive performance evaluation of 802. E. low-power PAN is relatively new. V. UCLA/CSD-TR 02-0013.Patiala. IEEE Personal Communications Magazine.[3] The advantage of this mode is that it doesn’t put any extra load on the network and sends a 2nd message using separate networks only in the case of failure. However. July 2004. H. IEEE INFOCOM’2000. S. IEEE Aerospace Conf. Park and M. pp.4: A Developing Standard for Low-Rate Wireless Personal Area Networks”. Paul Gorday. Volume: 19. Draft Standard: ”Low rate Wireless Personal Area Networks. Hu. C.4-2003: Wireless Medium Access Contron(MAC) Physical Layer (PHY)Specification. M. Routing in wireless networks has been a fascinating topic of research for long. June 2002. 1994. "A direct combination of the Prim and Dijkstra constructions for improved performancedriven global routing. W. P. C.15.”An Energy-Efficient MAC Protocol for Wireless Sensor Networks”. In Proc. Dept. February 1999. April 1999. [1] 79 of 446 Copyright @ Department of Electrical Engineering. 2005. Bob Heile and Venkat Bahl. “Max-Min dcluster formation in wireless ad-hoc networks”. ZigBee Alliance (ZigBee Document 02130r7) Draft Version 0. “Home Networking with IEEE 802. Jianliang Zheng and Myung J.g. Feb. New Orleans.g. Tech. Jose A. page no 1074-1079 Ed Callaway." University of California. Lee. So if the node is part of 27 channel network then the delay can be increased up to 27X.IEEE Communications Magazine August 2002. 1-8. of IEEE International Advanced Computing Conference. Ganeshan. Alpert. Kahng. Rep. J.4/ZigBee although its applications have been discussed in [14]. In Proc. Belding-Royer. Bandara and A.15. For example if for normal transmission it takes X amount of time then for each retransmission will increase the time by X amount at an average. Elizabeth Royer and C-K Toh. In [7] authors explored the complex behavior of a large number of small lowpower sensor nodes. Bhagwat. Lance Hester. [2] A. RFC 3561. Tel Aviv. Basagni . Computer Science. NY. Marco Naeve. D. [4] IV.15. 2002. 15. Thai H P Vuong and Dung T Huynh. B. March 2000.A comprehensive survey of these protocols has been done in [9]. H. Prakash. Proceedings of ACM SIGCOMM. R. B. Durresi and V.90: Network Specification. Perkins. J. Australia. [5] [6] RELATED WORK [7] [8] Wireless networks have rapidly gained popularity since their introduction in 1970s. “Address Borrowing in Wireless Personal Area Network”.March 6-7). DSDV [5].15. low-rate. S. “Highly Dynamic Destination-Sequenced Distance-Vector Routing (DSDV) for Mobile Computers”. C. [9] [10] In this paper we have provided a unified Multichannel network formation and routing scheme which can be used to use WPAN in Business networks where mobility is involved such as wireless sensor network in Mine by allowing the network to be created using multiple channel with marginal overhead in . Perkins and P. “Ad hoc On-Demand Distance Vector (AODV) Routing”. Debabrato Giri. For example a chatting application might not meet the user experience if the delay is prolonged. N. Paruchuri. 1999. 2003”. E. Huang. WRP [Shree_Murthy]) or source-initiated.Address Borrowing has been discussed in [8]. an investigation into low-cost. AODV [4.India. 6. “Complex Behavior at Scale: An Experimental Study of Low-Power Wireless Sensor Networks”. 2004. 1992 C. “EnergyAware Wireless Microsensor Networks”. Perkins and Elizabeth Royer. D. [16]. CSD-920051. “Ad-hoc On-Demand Distance Vector Routing”. But the main disadvantage is every time a resend is done the delay is increased as well. “Challenges in cluster tree formation with top-down approach for wireless sensor networks”. Gutierrez.15.310315. NIT Durgapur . These algorithms are either table-driven (e. June 23-25. (IACC’09. Schurgers. Heidemann. “A Comprehensive Performance Study of IEEE 802.There is relatively scant literature on 802. March 2002. Estrin. “Distributed clustering for ad-hoc networks”. Jayasumana. pp.Proc. IEEE P802. DSR. New York. Issue: 2. [9].4/D18. and S. 3]). March 2005. Krishnamachari. If the networks demand a continuous data transfer with short delay in those scenario it will not be suitable. Proceedings of the 21st International Conference of the IEEE Computer and Communications Societies (INFOCOM 2002). Ye and D. UCLA Computer Science. namely low-bandwidth. J.

amplitudes of forced vibration. longer life. It has been observed that most of the structural members fail due to the presence of cracks. [13] which is based on combination of wave-let based elements and genetic algorithm. They have observed that the presence of crack in the beam..in Abstract The structures and machineries designed in the present day are based on maximum strength. [10].Behera Department of Mechanical Engineering National Institute of Technology Rourkela.in T. applied cyclic loads etc. [14] have investigated experimentally the effects of cracks and damages on the structures . analysis of these changes is essential. INDIA. minimum weight and low cost etc. INTRODUCTION For the last several years. Dynamic stiffness method [7].ac. Beams are one of the most commonly used elements in structures and machines. mechanical vibrations. The lowest four natural frequencies of the cracked structure have developed by Shin et al. extensive research work has been commenced to investigate the faults in structures. Timoshenko beam. Euler-Bernoulli’s beam model where only translation mass & bending stiffness have been considered and Timoshenko beam model where both the rotary inertia and transverse shear deformation have been considered. direct and inverse methods on free vibration of simply supported beams with a crack was derived by Hai-Ping Lin [8]. Frequency contour method has been used to identify the crack size and location properly. A method for shaft crack detection have proposed by Xiang et al. Crack localization and sizing in a beam from the free and forced response measurements method is indicated by Karthikeyan et al. The presence of crack in a structure causes a local variation in the stiffness which affects the mechanical behaviour of the entire structure to a considerable extent. rkbehera@nitrkl. In this paper. [12] by using finite element method. Due to the existence of such cracks the frequencies of natural vibration. [9].ac. Ali et al. finite element method. NIT Durgapur . It has been observed that the local flexibility matrix is mainly appropriate for the analysis of a cracked beam if one employs an analytical method by solving the differential equations piece wisely [2]. 4] and mechanical impedance method [5].Al-Qaisia et al. which allow development of a high level of stresses in them which further leads to the development of crack in their elements. Keywords: Cracked beam. INDIA alokbiswal82@gmail. The information from the analysis enables one to determine the degree of sustainability of the structural element and the whole structure.R.com. and areas of dynamic stability change. I. [11] have studied the free vibration analysis of a cantilever beam. Department of Mechanical Engineering National Institute of Technology Rourkela. Many engineering structures may have structural defects such as cracks due to long-term service. Direct and perturbative solutions for the natural frequencies for bending vibrations of cracked Timoshenko beams are provided by Loyaa [6].K. Generally for the observation proposes the beam is modeled by two types. affects the natural frequency. Papadopoulos and Dimarogonas [1] revealed the introduction of local flexibility due to presence of transverse crack in a structural member whose dimension depends on the no of degrees of freedom considered. In their work Timoshenko and Euler beam formulations have been used to estimate the influence of crack size and location on natural frequencies of cracked beam. the presence of transverse and open crack in the Timoshenko beam has been considered and natural frequencies and mode shapes of the cracked Timoshenko beam have been studied by using finite element method (FEM) and MATLAB programme. Owolabi et al. They have obtained the approximate crack location by using Armon's Rank-ordering method that uses the above four natural frequencies. The 4 1 A S E I effect of crack on the dynamic behaviour of the structural elements has been the subject of several investigations for the last few decades.Biswal R.Roy Department of Mechanical Engineering National Institute of Technology Rourkela. [15] have utilized 80 of 446 Copyright @ Department of Electrical Engineering. The magnitude for the change of natural frequency depends upon the change of number. It is also appropriate to use a semi-analytical method by using the modified Fourier series [3. So it is very much essential to know property of structures and response of such structures in various cases. depth and location of the crack.Also the change of dynamic property effects on stiffness and dynamic behaviour. In order to identify the magnitude and location of the crack.Vibration analysis of a cracked Timoshenko beam A. and fatigue cracks are the main cause of beams failure.Identification of cracking in beam structures using Timoshenko and Euler beam formulation has been studied by Swamidas et al. INDIA tarapada@nitrkl.

The equation of motion can be derived by using Hamilton’s principle as the total strain energy being equal to the sum of change in kinetic energy and the work done due to the external forces and is given by t2 U T We dt 0 . as well as experimental approach to the crack detection in cantilever beams by vibration analysis. The overall additional flexibility matrix parameters are computed by 128-point (1D) or 128 ×128(2D) Gauss quadrature and then further best fitted using the least-squares method. Where N w & N are the shape functions for displacement and rotation. [17] have developed an analytical. y and z directions of the beam is given by w u x.A regular simply supported Timoshenko beam element The regular beam element is shown in fig-1 . which will lead to a generalized eigenvalue problem and thus the natural frequencies can be obtained. Where the axial displacement along the x-axis is u. [16] have introduced in their research different damage scenarios by reducing the local thickness of the selected elements at different locations along finite element model (FEM) for quantification and localization of damage in beam-like structures. An investigation of a beam crack identification method by using the standard finite element formulation has been produced by Demosthenous et al.The longitudinal axis of the regular beam element lies along the X-axis. can indeed give more accurate results than those obtained from using the local flexibility matrix. taking the shear (ϕ) into consideration 12EI k y 1 Fig-1. t wx. t z x. The shear is β(x) which has no contribution during finding out the axial displacement at a point at a distance z from the centre line. ρI N θ 0 N θ 0 T Le The stiffness matrix of the regular beam element is the sum of the bending stiffness and the shear stiffness and is written in matrix form as 2. Substituting the mode shapes N w & N we get the stiffness matrix of the regular beam element. It has been observed that the consideration of ‘overall additional flexibility matrix’. NIT Durgapur . z. Nahvi et al. The mass matrix of the regular beam element is the sum of the translational mass and the rotational mass and is given in matrix form as N ρA 0 N w M b w dx .a1. a2 a3 and a4 are the unknown coefficients and are found out by using boundary conditions at the beam ends.The AG L2 . When ϕ is 81 of 446 Copyright @ Department of Electrical Engineering. the lateral displacement along the y axis is v which is equal to zero and the time dependent transverse displacement of the centroidal axis (along z axis) is w. The analysis reveals a relationship between crack depth and modal natural frequency. [18] have calculated natural frequencies for a beam with open edge crack using theoretical and finite element analysis. t1 Here U is the strain energy. modal natural frequency with crack location has been studied by Sutar [20]. displacement relation in the x. y. The bestfitted formulas agree very well with the numerical integration results [22]. Let w be the approximated by a cubic polynomial and is given as A S E II. Sahin et al. I MATHEMATICAL FORMULATION 4 1 w a1 a2 x a3 x 2 a4 x 3 . t z x x vx. due to the presence of the crack. After getting the stiffness matrix of a cracked beam element standard FEM procedure can be followed. The modeling of shaft was done by finite element method and coupled to an experimentally identified foundation model. The time dependent rotation of the cross-section about y axis is θ. [19]. y. t 0 . Behzad et al. Only single crack at different depth and at different location are evaluated. we get the governing equation of motion (Timoshenko beam equation) of a general shaped beam modelled with Timoshenko beam theories.1 Finite element modeling of regular beam element N θ Le EI x K b N 0 N θ w 0 x N θ 0 . Where x is the distance of the finite element node from the fixed end of the beam.the reduction of Eigen frequencies and sensitivity analysis to localize a crack in a non-rotating shaft coupled to an elastic foundation. The finite element analysis of a cracked cantilever beam and the relation between the modal natural frequencies with crack depth. Zheng [21] has described an overall flexibility matrix instead of local flexibility matrix in order to find out the total flexibility matrix and the stiffness matrix of the cracked beam. mass density and length [23].Sustituting the values of strain energy and kinetic energy. k ky . z. x dx N kGA N w θ x T . y. z. T is the kinetic energy and W e is the external work done which is equal to zero and t is the time. modulus of elasticity. t . wx. The element has constant moment of inertia.

835 2. and α 1 D 2 4 x 2 D 2a .The beam length is divided into forty no of elements.3-5. y j yi θi V j θ θ Covl M .053 250.3 α/D =0.2 α/D =0.2.716 249. The various geometric and physical parameters of the beam are L=1m.2 2183.Natural frequencies for uncracked Timoshenko beam.2 Flexibility matrix for intact beam element: The flexibility matrix Cintact of the intact Timoshenko beam element can be written as y j yi θi V j θ θ Cint act M . ‘a’ denotes the crack depth.378 1944. Natural Frequencies of Cracked Timoshenko beam.703 1013.Under the FEM co-ordinate and notation system.2 1013. Xc/L=0.3Total flexibility matrix of the cracked Timoshenko beam element: The total flexibility matrix of the cracked Timoshenko beam element is obtained as 2. The corresponding displacements are denoted as ‘y’ and ‘θ’. The crack is at a distance of ‘Xc’ from the left end of the beam.4Stiffness matrix of a cracked Timoshenko beam element: Through the equilibrium conditions.275 ω2 0. A S E I Table-1. 0 1 0 1 Where b Da a 2 . 28] 1 K c LCtotal LT .2 simply supported beam with crack subjected to shear force and bending moment 1 Le L 1 0 A simply supported beam of circular cross-section having diameter ‘D’ with a single transverse crack with constant depth ‘a’ is shown in Fig.1Overall additional flexibility matrix under conventional FEM co-ordinate system: A cracked beam element subjected to shearing force ‘V’ and bending moment ‘M’ at each node is considered. NIT Durgapur .2.315 1552. the relationship between the displacement and the forces can be expressed as A simply supported Timoshenko beam with a transverse crack is considered for numerical analysis. ω1 253.396 1012. i j j RESULT AND DISCUSSION 4 1 h D 2 4x 2 Table-2.952 1008. ‘Lc’ denotes the distance between the right hand side end node and the crack location.875 ω3 0. The first three mode shapes of uncracked and cracked Timoshenko beam are shown in Fig.It is observed that there is a sudden jump in the mode shape due to presence of crack in the beam.703 ω3 2280.553 1010.258 82 of 446 Copyright @ Department of Electrical Engineering. 2 2. III. The natural frequencies for the uncracked and cracked Timoshenko beam are shown in Table 1-5. Fig. 2.3. The beam element has length ‘Le’.835 2280.425 253.371 248. Where L is the transformation matrix. ρ=7800Kg/m3& µ=0.381 1700.5 ω1 0. the stiffness matrix ‘Kc’ of a cracked beam element can be obtained as follows [27. D=2cm. i j j ω (rad/sec) Present [FEM] Theoretical [29].20. E=206Gpa.2.2. The beam is divided into number of equal elements having length ‘Le’. 2.2 Finite element modeling of a cracked beam element Ctotal Covl Cint act .2.4 α/D =0. ω (rad/sec) Xc/L α/D=0.neglected. the mass matrix and the stiffness matrix reduce to the mass and stiffness matrix of an Euler-Bernoulli beam. cross-sectional area is A and flexural rigidity EI.426 ω2 1013.2 252.

REFERENCES C. Table-5.5. Natural frequencies for bending vibrations of Timoshenko cracked beams. R.6 2162. A.H.2. ω (rad/sec) Xc/L α/D=0.619 2369.50. Crack identification in beam structures using mechanical impedance.α/D=0. S.609 927. Zheng.Tiwari.711 1024.8 250.6 251. Xc/L=0. Engineering Structures 26 (2004) 427–436.85 7 α/D =0.405 2350. Natural frequencies of a non-uniform beam with multiple cracks via modified Fourier series.418 1031.006 CONCLUSION In this paper an overall additional flexibility matrix is considered for evaluating the stiffness matrix as well as natural frequencies of a cracked Timoshenko beam.389 250. ω (rad/sec) Xc/L α/D=0. S.4 α/D =0.2.Trochidis.A. P.Viola.P.0831 977.2 α/D =0.615 982.2 α/D =0.3 α/D =0. Natural Frequencies of Cracked Timoshenko beam.544 247.028 1037. The magnitude for the change of natural frequencies depends upon the change of relative crack depth and location for the crack. E. ω (rad/sec) ω1 Xc/L 0. Journal of Sound and Vibration 242 (4) (2001) 701–717.997 ω2 0.60. D. M.3 α/D =0. J.Dimarogonas.I.5 ω1 0. E.A. E.Crack localization and sizing in a beam based on the free and forced response measurements.067 2248.4137 907. Journal of Sound and Vibration 117 (1) (1987) 81–93.5 ω3 0.Loya.130 242. H. Shifrin.11 989. Free vibration analysis of axially loaded cracked Timoshenko beam structures using the dynamic stiffness method. Ruotolo.3 First mode shape for uncracked and cracked beam.6 1018.2. XC/L=0.8 2321.Y.5 Table-4.132 2377.321 968.5 α/D=0.Fernandez-Saez. Journal of Sound and Vibration 222 (1999) 409–423.2 α/D =0.51 2268.100 251.8 1001.089 ω3 0. Lin. Journal of Sound and Vibration 246 (2) (2001) 297–317.α/D=0. Natural frequency changes of a cracked Timoshenko beam by modified Fourier series.5.453 1966. R.5 Third mode shape for uncracked and cracked beam.A.4 252.3 α/D =0.032 ω3 0. Karthikeyan.960 245. 4 1 A S E I Fig.069 6 2284.221 249.α/D=0. Journal of Sound and Vibration 256 (2) (2002) 287–297.4 α/D =0.Table-3. Bamnios.875 Fig.404 1900.C.4 Second mode shape for uncracked and cracked beam.80.Papadopoulos. Xc/L=0. Natural frequencies of a beam with an arbitrary number of cracks. XC/L=0. Fan. From the results obtained it has been observed that the presence of crack in the beam decreases the natural frequencies.Coupled longitudinal and bending vibrations of a rotating shaft with an open crack.XC/L=0.5 ω1 0. D.Ricci.240 ω2 0. S. Natural Frequencies of Cracked Timoshenko beam.088 246. Fan. NIT Durgapur . IV. Aliabadi. [1] [2] [3] [4] [5] [6] [7] [8] [9] Fig. Journal of Sound and Vibration. Direct and inverse methods on free vibration analysis of simply supported beams with a crack.772 252. Mechanical Systems and Signal Processing 21 (2007)1362–1385. L.Talukdar.Rubio. Zheng.823 2058. Natural Frequencies of Cracked Timoshenko beam. Douka. It is observed that with increase in relative crack depth the natural frequency decreases and there is an abrupt change in mode shape due to presence of crack in the beam.5 ω2 0. J.Y.8555 2281.5. Xc/L=0. 83 of 446 Copyright @ Department of Electrical Engineering. Y.D. M.C. Journal of Sound and Vibration 304 (2007) 124–153.784 953. 290 (2006) 640-653.810 245.

L. A new approach for vibration analysis of a cracked beam.A. Kosmataka. 4 1 A S E I 84 of 446 Copyright @ Department of Electrical Engineering.[10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] A.C.Catania. Mechanical vibrations.Zhong. M. L.M.A. Triply coupled bending–torsion vibration of Timoshenko and Euler–Bernoulli shaft beams with arbitrarily oriented open crack.S. Y. NIT Durgapur .D. 1975.R.Irwin. A.Kessissoglou.He. Davis.T.Nisreen. Journal of Quality in Maintenance Engineering 9(2003)176 – 201.Paris. X.Owolabi. M. Crack detection in beams using experimental modal data and finite element model. Proceedings of the 5th WSEAS International Conference on Circuits (2006) 209-214. G. G.Behzad.Nahvi. R.Bejarano. Free vibration Analysis and Dynamic Behaviour for Beams with cracks. H.Zheng. Z. Vibration of cracked shafts in bending. Formulation of cracked beam element for structural analysis. A. A.Shin.Swamidas.Sutar. L.N. Y. Computers and Struct.Ali. The Stress Analysis of Cracks Handbook. International Journal of Advanced Engineering Research and Studies E-ISSN2249–8974. R.Seshadri.Demosthenous.J.K. J.S.Sahin. P.Rao. Journal of Sound and Vibration 273 (2004) 457–475.Maiti. Crack localization in non-rotating shafts coupled to elastic foundation using sensitivity analysis techniques.J. Z. G. 2002.J. Free vibration analysis of a cracked beam by finite element method. Journal of Sound and Vibration 324(3-5) (2008) 1067-1085.Y. New York. Investigation of a beam crack identification method using the standard finite element formulation. Quantification and localization of damage in beam-like structures by using artificial neural networks with experimental validation. A.Chen.M. International journal of solids and structures 45(2008)47824795. ASME Press.Z. Crack detection in beams using changes in frequencies and amplitudes of frequency response functions.Xiang. A. Crack Identification of a Rotating Shaft with Integrated Wireless Sensors. Rabinowitz.Naik.A. C.S. C. G. M. International Journal of Science Engineering and Technology.Meghdari. X.Papadopoulos. Friedman. S. M. D.S. B. fourth edition. P.Y.Viola. M. Tada. Journal of Sound and Vibration 265(2003)1-22. Computers and Structures 77(2000)327342.Faizal. International Journal on Smart Sensing and Intelligent Systems 2(4) (2009).Jabbari. Identification of cracking in beam structures using Timoshenko and Euler beam formulations. J.Jia. A. A study on crack detection using Eigen frequency Test data. Crack detection in a shaft by combination of wavelet based elements and genetic algorithm. 2004. 2000.J.Al-Qaisia. 2009. E. Dimarogonas. S.K. H.J. Engineering Structures 25(14)(2003)1785-1802. International Journal of Mechanical Sciences 47(2005) 482-490. Journal of Sound and Vibration 91 (4) (1983) 583–593.Swamidas.Jee.Aswan. S.Federici. Finite element analysis of a cracked cantilever beam.Shenoi. P. 47( 3) (1993) 473–481. International Journal of Engineering 18(4) (2005) 319. F.Nobile. A.Methods of Numerical Integration. An improved two-node Timoshenko beam finite element. N.Yang.

India. When discussing about coupler and splitters it is customary to refer to them in terms of number of input and output device [2]. Cross Gap and Multimode Interference Optical Coupler design Vishal Sorathiya Amit kumar Electronics and communication department.com Electronics and communication department. switches. (b) Picture of cross section [8] The design of the three types of coupler namely Cross Gap Coupler (CGC). Abstract— A new and innovative interconnection technology in opto-electronics based chip fabrication is presented. and nonlinear optic devices)[1].48 and 1. INTRODUCTION P lanar light wave circuits are optical integrated circuits (ICs) or optical circuit boards which usually perform their functions in the optical domain. vishal. There for it could not be bending to small curvature. the waveguide components have to be reduced considerably [3].21 μm. This structure makes it possible to bend the wave guide with micrometer or nanometer curvature.Planar light wave circuits are analogous counterpart of electronic ICs in optical domain. A S E I To make these planar light wave circuits of smaller size. gives maximum coupled optical power and reduce the back hand coupled power losses. 2: Fused fiber coupler [2] 85 of 446 Copyright @ Department of Electrical Engineering. [9]. Optical coupler design. Multimode interference coupler. optical passive components. with very small cross-section. So to design these optical devices we use high index contrast waveguides. SI PHOTONIC WIRE WAVEGUIDE 4 1 Fig. III. the focus of this paper is on to develop such optical coupler. They are usually of nanometer size and can act as splitter. For the requirement of sharp bends in waveguide it is necessary to have high index contrast of reflection (high ∆). Rajkot. Marwadi Education Foundation Group of Institute. Silicon photonics is receiving much interest because it enables the use of well-developed Si processing technologies as well as Si substrates that are cheaper than the compound semiconductor (GaAs or InP) substrates to fabricate a broad range of optical devices (light emitters.kumar@gmail. I. Hence with the advent of photonic integration nowadays strongly guiding ridge waveguides are used as building blocks for optical devices such as small length couplers (with coupling length of few micrometers). [10]. Directional coupler. Fig. Cross gap coupler. chaurasia.46 respectively and the width and height of the Si core was 0. optical switches. 1 the schematic diagram of waveguide structure with Si wire core is in nanometer size. After giving overview of different types of optical coupler design with planer light wave circuit. NIT Durgapur . photo-detectors . silica and various polymers.45 μm and 0. As shown in Fig. Directional Coupler and Multi Mode Interference Coupler (MMI) were based on Si photonic wire waveguide with refractive index of the core and cladding material were 3. Marwadi Education Foundation Group of Institute. popularly known as Si photonic wire waveguides. India. power tap etc. This technology gives the advantages over existing technology of electrical interconnection in terms of greater speed. 1: Si photonic wire waveguide (a) Layered structure.sorathiya9@gmail. efficiently with less space required and with greater speed. II. [3]. Rajkot. Index Terms—Silicon on insulator. FUNDAMENTALS. Planar light wave circuits uses optical waveguides to route photons in the the similar way as the metal traces are used to route electrons in electronic ICs and circuit boards . TYPES AND APPLICATIONS Optical coupler of N×N coupler (with N≥2) can be used as power splitter. couplers etc. The index contrast between the Si core & silica cladding is of 40% order.com However in Silicon on Insulator based waveguide at several hundred micrometer range of banding waveguide causes the maximum bending loss. They are optical circuits patterned on Si wafer and are useful because of their small size and their potential for integration of multiple devices. For the design of coupler by using planar waveguide like lithiumniobate(LiNbO3 ).Silicon on Insulator based Directional.

as shown in Fig. A S E I Fig. 6 is to regard it as a combination of two planar waveguides. coupling region size and the difference in radii of the two fiber in coupling region [6]. We start by calculating the effective index in the y direction. and obtain the effective index in this direction. This process is known as frustrated total internal reflection. Many functions realize using MMI such as power splitter [12]. Pb=Pincos2(zπ/2Lc) (3) Pc=Pinsin2(zπ/2Lc) (4) Fig. So. IV. is very simple [1]. optical switching etc. The main aim to design directional coupler is to give power from any one of port and get maximum coupled output at two ports which is in opposite side and make minimum power output at isolated back port. for example. NIT Durgapur . .6: waveguide cross section and index distribution. and so this method. waveguide width and refractive index between waveguides in coupling region. The transfer of the light is proportional to the propagation distance (z).4 with 2×2 type and that have mainly four pot design. easy and large fabrication tolerances. 86 of 446 Copyright @ Department of Electrical Engineering. (2) Multimode interference coupler (MMI) Multimode interface coupler has attracted much interest in the past decade. low loss. 7(a). In CGC waveguides are kept with few wavelength distances from interface to make efficient corner mirror. For identical core fiber we have equation is P1=P0 -P2 =P0 [1-sin 2 (kz)]=P0 cos2 (kz) (5) ne and n0 is effective indexes of even and odd super nodes. owing to it compact structure. 3: Normalized coupled powers as a function of the coupler draw length [2] The process of total internal reflection is not always satisfied because of boundary surface cannot be make 100% reflection and create attenuation with loss of power. 7(a) and 7(b). one horizontal and one vertical.The optical power of coupler is given as P2 =P0 sin2 (kz) The coupling length which is as small as possible is given as [4] (1) Lc=π/βe-βo= λ0/2(ne-n0) Where z is axial distance and k is coupling coefficient . (1) Directional coupler (DC) Directional coupler is as shown in below Fig. The approach to finding the propagation constants for the waveguide shown as an example in Fig. taking the effective index of the first as the core refractive index for the second. and considering a TE mode. We then successively solve the planar waveguide eigenvalue equations first in one direction and then the other.4: Schematic core pattern of a directional coupler [5] 4 1 EFFECTIVE INDEX AND MODE POLARISATION The effective index is used to find approximate solutions for the propagation constants of two dimensional waveguides. Fig. shown respectively in Fig. This is done without dealing directly with the electric fields within the waveguide. 6. For the process in which waveguide is design in such way that the internal high refractive index material is surrounded by low refractive index material with minimum distance and when a optical signal pass through the waveguide than it is differ from ordinary condition to pass energy.Here P2 is depends on axial length.5: Schematic pattern of directional coupler in PLC and its crosssection [5] The DC considered as consist of two parallel symmetric wave guides [7]. All the power is expressed in db. The basic parameter id to require for designing 2×2 coupler is gap between waveguide. the previously defined strip waveguide. The Bar and Cross output of DC is given as followed equations. In this the TE and TM mode of polarization operation is possible. (2) (3) Cross Gap Coupler (CGC) Fig. when considering. with the index distribution shown in Fig. known as the effective index method.

87 of 446 Copyright @ Department of Electrical Engineering. with respect to this imaginary vertical waveguide. The general solution to the wave equation for TE modes in planar waveguides [1].46 10 μm 5 μm 0. has to be solved in order to obtain the propagation angle . we must use the TM eigenvalue equation because. equation ℎ cos − = tan + tan TABLE 1: SPECIFICATIONS USED FOR MODELING THE DIRECTIONAL COUPLER Power of the input light Polarization and wavelength of the incident light Width and height of the core Refractive index of the core Refractive index of the cladding Radius of curvature for sshaped waveguide Coupling length Gap between two linear waveguide (6) =2 With the propagation constant in free space and h the height of the core layer. The DC was having silicon insulator material with 0. ( ) or the intensity distribution. 9: Planar view of H y component of light in directional coupler The corresponds to a propagating sinusoidal type field.10: Refractive index distribution (2) MMI Coupler: MMI coupler having smaller size compare to DC and design is robust less parameters like pitch (Pt). while the represents fields propagating in the y direction through the claddings. | ( )| for the different modes. we obtain that: = sin (7) Then the analogous equation for TM modes has to be solved for the other decomposition in order to find the effective index of the mode.55 μm 0. power combiner and switches [8-9] is directional coupler.21 μm 3.Then. from the definition of the effective index of the mode. 7: (a) vertical (b) horizontal calculation of effective index For the first decomposition of the waveguide.20 μm Opti FDTD-8 (optiwave corporation) 4 1 A Fig. then when solving the threelayer planar waveguide in the y direction we use the TE eigenvalue equation.45 μm. For TE modes. The polarisation has to be chosen carefully.48 1. (1) Directional Coupler: The most fundamental component in optical device for add-drop multiplexer. However. NIT Durgapur . the field is polarised in the TM direction.45μm thick single crystal Si surface fabrication. If we are considering an electric field polarized in The x direction (TE polarisation). which is of the form: ∇ With =1 = DESIGN AND ANALYSIS Simulation Software 1mw(0 dBm) TM MODE. coupling length and width [11].0. So after obtaining this solution. it is possible to plot the field distribution. in order to have an idea of how the field propagates through the waveguide [1]. 8: Design of Directional Coupler S E I (8) the velocity of wave is following: = (9) Fig. The surface was first patterned by using electronic beam lithography to wave guide core pattern of DCs [14].V. That means that the field penetrates the cladding with a decay constant . so part of the field is propagating through the cladding. Fig. Fig. when we subsequently solve the vertical three-layer planar waveguide. 1.

Fig.206 -4.37 -3. [15].97 1.959 -4.741 -3.424 -3.767 -4.973 -3.173 -6.458 -5.949 -3. [16].39 -3. In this design the wavelength 1.064 -3.483 -3.6744 -4. Fig.735μm and 0.58 -3. All the structures are design in planar light wave circuit (PLCs).29 -3.993 -4.165 -3.796 -4.743 -3.281 1.55 -3.183 Cross Output (dBm) CGC 60° Bar Cross Output Output (dBm) (dBm) -3.882μm respectively.53 -3.2829 -3.215 -3.445 -3. 16: Planar view of H y of light in CGC (θ=30°) A RESULT AND COMPARISON S E I TABLE 2: DIFFERENT TYPES OF LOSSES Types of losses and ratio parameter Coupling Ratio Excess Loss Equation of loss = = 10 log Insertion Loss Return Loss and Reflectance × 100% + + = 10 log = 10 log TABLE 3: CROSS AND BAR OUTPUT FOR CROSS GAP COUPLER WITH DIFFERENT ANGLE Coupler 1.55 -3.5902 -3.959 -3.853 -4.734 -3.985 -3.262 -3.899 1. for particular wavelength.5μm) Fig. 15: Structure of Cross Gap Coupler (θ=30°) Fig.32 -5.55μm.0315 -4. power is divided equally at both cross and bar wave guide port. 11: Structure of the MMI Coupler [7] Here we design for two widths 1.328 -3.632 -6.5μm) (3) CGC Coupler: 4 1 The cross gap coupler is structure of two waveguide with crossing each other with some angle 30.26 -3. This gap is function as half mirror or corner mirror [6].884 -4.323 1.927 -4.342 -4.56 -3.492 1. NIT Durgapur .397 -4.8 with pitch distance is 0.mirror.431 -4. 14: Structure of Cross Gap Coupler Fig.52 -3.54 -3.5595 -7.027 -4.3382 -6.88 -5.136 -4.236 1. The gap is replaced with silica material which acts as corner VI.6 -4. 50 and 60.065 -3.5 and 1.656 -4.43 88 of 446 Copyright @ Department of Electrical Engineering. 12: Structure of the MMI Coupler (width 1.423 -6.794 -4.179 1. 40. Fig.846 Wavelength (in µm) CGC 40° Bar Output (dBm) -3.334 -3.575 -4.073 CGC 50° Bar Cross Output Output (dBm) (dBm) -3.5 CGC 30° Bar Cross output Output ( dBm) (in dBm) -3. 13: Planar view of H y of light in MMI (width 1.742 -4.921 -3.1816 -4.

52 1.982 0.8 1.7972 2.95 -3.139 0.29 S.664 -3.6682 1.6 -4.066 1.96 -4.3 43.678 -3.5μm Wavelength (in µm) Bar output ( dBm) Cross Output (in dBm) Bar Output (dBm) Cross Output (dBm) Bar Output (dBm) Cross Output (dBm) 1. NIT Durgapur .8 µm(dB) 1.35 0.45 1.413 -4.083 1.171 0.542 -4.415 -3.5 1.0797 1.58 52.125 1.83 46.89 48.19 1.54 48.07 1.53 47.79 47.67 52.07 52.538 0.768 -3.345 -3.634 -3.75 1.15 -2.6 1.73 31.54 1.08 -4.TABLE 4: CROSS AND BAR OUTPUT FOR MULTIMODE INTERFACE AND DIRECTIONAL COUPLER Coupler Directional MMI 1.073 0.12 50.4 1.R.32 1. for 40˚ S.7 50 50 50 51.5 -2.896 0.45 -3.47 55.56 -3.56 51.8 TABLE 5: SPLITTING RATIO FOR DIRECTIONAL.53 48.97 S.464 -4.0764 1.94 -4.5 52.8 µm width 58.53 -2.4 61.55 50 49.26 53.52 -2.44 55 52.58 1.8μm MMI 1.785 -4.R.68 0.657 1.669 0.85 1.994 0. of MMI 1.0697 0.9 45.02435 0. MULTIMODE INTERFACE AND CROSS GAP COUPLER Wavelength(μm) S.41 TABLE 6: EXCESS LOSS FOR DIRECTIONAL.71 45.5 µm(dB) MMI 1.19 46. MULTIMODE INTERFACE AND CROSS GAP COUPLER Wavelength (μm) CGC 60 ˚(dB) CGC 50 ˚(dB) CGC 40˚(dB) CGC 30˚(dB) DC(dB) MMI 1.08 -3.98 1.67 57.72 46.R.72 34.12 0.89 36.948 -4.528 0.58 -3.914 0.211 -4.91 36.56 1.6 40.134 -3.35 1.7608 1.116 0.066 0.0316 0.45 1.09 -4.76 0.908 -3.89 -3.8457 89 of 446 Copyright @ Department of Electrical Engineering.0782 1.9848 0.1036 0.08 -3.234 -3.94 45.02 -3.81 -3.04 55.389 -3.98 50 42.024 1.R.86 45 44. of MMI 1.8 1.6996 1.9269 1.32 48.R.86 -3.9 57.9 1.53 1.08 -3.198 1.546 0.55 -3.613 0.245 -2.5958 1.34 1.063 1.5 41 4 1 A S E I 34.of DC 46.55 1.76 45.6 53. for 30˚ S.95 -4.56 -4. for 60˚ S. for 50˚ S.52 47.76 46.06293 0.904 0.R.5 µm width 57.902 0.7 54.01 45.999 0.67 1.12 -2.775 -2.1 1.888 -3.48 -4.14 42.1255 0.R .288 -4.7956 1.73 1.69 51.752 0.115 1.1013 0.71 1.635 -4.09 1.54 -2.732 1.28 46.998 -4.127 1.124 0.

Jalali. vol. Marom. L. Exp. Condens. 938–947. “Advances in silicon-on-insulator optoelectronics. and J. T.” J. Hirohito Yamada. vol. E82-B. MULTIMODE INTERFACE AND CROSS GAP COUPLER Excess Loss(in dB) CGC of 30˚ CGC of 40˚ CGC of 50 ˚ CGC of 60 ˚ DC MMI of 1. “Switched directional couplers with alternating Δβ. 3. Osgood. B. T.Satomi Ishida. G. Tata McGraw Hill. no. U. Forber and E. Lightw. Tao Chu. Nov. Gerd Keiser. “Small-area bends and beam splitters for low-index-contrast waveguides.3 45 44. 6. J. Nordin. “Silicon photonics: an introduction. R. 2003. E. For 30 degree design has large excess loss and poor wavelength dependency. no. Rendina. vol. Technol... S. no. 14. 4. T. Reed. Kogelnik and R. November/December 2006. vol. 8–13.. Satomi Ishida.” John Wiley&Sons. Tao Chu .171 0. 1976. Hirohito Yamada.5µm width MMI Coupler of 1. With 40 degree design has least excess loss. M. 1986. Ofusa. Kitamura. U.on-Insulator. 2003. Y. 911–919. 11. Urino. Quantum Electron.” IEICE Trans. “Optical Fiber Communication” fourth edition. S. Ahmad. 6. “An optical add-drop multiplexer with a grating-loaded directional coupler in silica waveguides. November 2007. Li. “Ultracompact Corner-Mirrors and TBranches in Silicon.. Coldren.100 0. 12.8457 [10] [11] [12] [13] TABLE 8: VARIATION IN SPLITTING RATIO FOR DIRECTIONAL. C. Norberg.. A.8 µm width Minimum 1.. Member. 1771–1776. In 60 degree design has less variation in splitting ratio and best wavelength dependency but in design of 50 degree with moderate excess loss and splitting ratio for designing PLCs with silicon on insulator substrate. I. P. 3. Matter. QE-12. 4 1 A S E I REFERENCES [1] [2] [3] [4] [5] [6] [7] G.595 0. “Silicon micro photonics. Pavesi. 2000.125 0. Yoshimoto. T. [8] COMPARISONS OF THE SIMULATION RESULTS [9] TABLE 7: VARIATION IN EXCESS LOSS FOR DIRECTIONAL.083 1. Saito. vol. Keita Kawabe. R. pp.” IEEE J. QE-22. G. Feb. Sci. Pages 267–271. J.” J. Multimode interface coupler is very easy in design but it has large excess loss and variation in splitting ratio is large.896 1. Rickman. 396–401. 2009. pp. and R. Hawaii. “Silicon-on-insulator optical rib waveguide loss and mode characteristics. 15.” Opt. IWA1.01 52.” OSA/IPNRA/NLO/SL 2009. Makoto Fujimaki.60 40. NO. pp. Yoon. 1994. 1. “Will silicon be the photonic material of the third millennium?. pp. “Improving the self-imaging in multimode interference (MMI) couplers. 7. and F. “Symmetric directional coupler switches. 2002 “Tutorials and Background FDTD” Optiwave Inc.999 1. “InGaAsP/InP based Flattened Ring Resonators with Etched Beam Splitters. Phys.” IEEE J.014 0. L. vol. Surf. and M.76 0. MULTIMODE INTERFACE AND CROSS GAP COUPLER Splitting Ratio (in %) CGC of 30o CGC of 40˚ CGC of 50˚ CGC of 60˚ DC MMI Coupler of 1. Parker. Jiang. and Yasuhiko Arakawa.Reed and A. pp. V.” Optical Directional Coupler Based on Si-Wire Waveguides. Masahiro Suzuki. 12. Koichi Awazu. NIT Durgapur .797 0. 10. Jun. L.Junji Tominaga “Control of the properties of directional couplers using proton irradiation.5 µm width MMI of 1. VOL.73 46. pp. 282-290.8µm width Minimum 31. and F. G. R. vol. 90 of 446 Copyright @ Department of Electrical Engineering.73 57. Sel. Yoshimichi Ohki.P.94 Maximum 45. Honolulu.. 1248–1251. T. no. Issue 2. pp. “Si Photonic Wire Waveguide Devices. Vol. F. July 12-17.Ltd..80 53.” Volume 264. Cross Gap Coupler has different result variation with change in angle. Quantum Electron.198 Maximum 2. pp. Shimoda.14 42.” IEEE Journal Of Selected Topics In Quantum Electronics. 1998. 17. R1169– R1196. XLI. S. Pizzuto. M.85 55.no. Member. RUI YIN.. Krishnamachari.. Coppinger. G.” Optical Application. 65-67. No. Oct. H. and L. Technol. 2011. Yegna narayanan.” Appl.–Dec. no.” IEEE photonics technology letters.Khights. R. J. L.86 43.116 0. no. 159–160. Kimerling. and Yasuhiko Arakawa. T. march 2005. Schmidt. pp. 2006. English. Communication. 1999. 3. Jul.” IEEE Photon. N. vol. Hanada. H.2009. Rao. Member. S. Guzzon.VII.528 1. Namavar.97 58.44 61. 6.T. Camarda. vol.29 [14] [15] CONCLUSION [16] [17] As conclusion Directional coupler having very less transmission losses but it is having large splitting ratio. vol.” IEEE J. Jan. Espinola. Lett. Topics Quantum Electron.

com2 Abstract— The paper presents Particle swarm optimization (PSO). esi Cost curve coefficients of ith thermal unit. In the present work. In this paper. So this algorithm have drawn a wide attention in the world. B. csi . Hydrothermal scheduling problem is referred to find out the discharge of hydro plants and power generation of thermal plants to minimize the total fuel cost at a schedule horizon while satisfying various constraints. evolutionary programming technique [2] and genetic algorithm [3-5] have been applied for determine optimum fuel cost from hydrothermal power plant. From the results it is clear that PSO approach gives the best global optimum solution than the PSOCFA. d si . Particle swarm optimization with constriction factor approach (PSOCFA). In this paper particle swarm optimization (PSO). Particle swarm optimization with constriction factor and inertia weight approach (PSOCFIWA) which are applied to determine the optimal hourly schedule of power generation in a hydrothermal power system. Durgapur2 .Shibpur3 Howrah. PSOCFA. There are some limitations that are obtained in [12]. Output power of jth hydro unit at time m. Phj max Lower and upper generation limits for jth hydro unit. the varying hourly reservoir inflows. Sumit Banerjee2 Chandan Kumar Chanda3 1. the physical limitations on the reservoir storage and turbine flow rate and loading limits of both thermal and hydro plants. applied PSO technique to solve shortterm hydrothermal scheduling [15].2 Electrical Engineering Department3 Bengal Engineering and Science University. the effects of valve point loading in the fuel cost function of the thermal plants are also considered. The main constraints time coupling effect of the hydro sub problem. 91 of 446 Copyright @ Department of Electrical Engineering. Particle swarm optimization(PSO). we consider the cost characteristics of individual thermal units instead of considering an equivalent thermal unit. India Electrical Engineering Department .com1. Water discharge rate of jth reservoir at time m. Particle swarm optimization with inertia weight approach (PSOIWA).An Analysis of Short Term Hydrothermal Scheduling using Different Algorithms Koustav Dasgupta1. PSOIWA. 6-13]. sumit_9999@rediffmail. Santiniketan Institute of Polytechnic. This method have been provide very faster performance and very efficient than other conventional method. NIT Durgapur . valve point effect. Roy Engineering College. The result of proposed method is very innovative and encouraging compared with other methods. al. PSO has small number of free tunable parameters to arrive desire goal. PSO has been applied in many field of power system to determine the optimum value of the parameter of power system. min max Lower and upper generation limits for ith thermal unit. The developed algorithms are illustrated for a test system consisting of four hydro plants and three thermal plants. understand from any other technique. PSOCFIWA based hydrothermal scheduling is developed through the test system [14]. This optimization technique has a faster convergence rate to global solution. Particle swarm optimization with inertia weight approach(PSOIWA). Yu et. The objective of the hydrothermal scheduling problem is to find out the discharge of hydro plants and power generation of thermal plants to minimize the total fuel cost at a schedule horizon while satisfying various constraints. Psi . It has highest potential of finding more nearly global solutions to hydrothermal coordination problems [12]. Bolpur1. Psi PDm Phjm Phj min Qhjm Load demand at time m. asi . the time varying system long demand. Different performances of stochastic techniques have been studied in the literature [1. Particle swarm optimization with constriction factor approach (PSOCFA). where the water flow in an earlier time intervals affects the discharge capability at a later period of time. particle swarm optimization with constriction factor and inertia weight approach (PSOCFIWA) method are proposed for short-term optimal scheduling of generation in a hydrothermal system which involves the allocation of generation among the multi-reservoirs cascaded hydro plants and thermal plants with valve point loading effects so as to minimize the fuel cost of equivalent thermal plant while satisfying the various constraints on the hydraulic and power system network. C. . PSO. I. bsi . Literature survey shows that particle swarm optimization technique is very simple optimization technique and easier to 4 1 A Keywords— Cascaded reservoirs. Dr. India Koustav2009@gmail. Recently simulated annealing technique [1]. PSOIWA and PSOCFIWA techniques. NOMENCLATURE Psim Output power of ith thermal unit at time m. Hydrothermal scheduling. the cascade nature of the hydraulic network. INTRODUCTION S E I Hydrothermal scheduling problem plays an important role in power system operation today.

pbest id is the best position of the d th dimension of the i th particle. m ∈ M M NS 2 + d ×sine × Pmin−P f = ∑ ∑ asi +bsiPsim+csiPsim si si si sim m=1 i =1 (1) B. C6 j Power generation coefficients of There are some operational constrains of hydrothermal units which consist of some water balance equation as well as reservoir storage and release target. Ns Nh ∑ Psim + ∑ Phjm − PDm =0 m∈M (2) i =1 j =1 The hydropower generation is a function of water discharge rate and reservoir head P =C V 2 +C Q2 +C V Q +C V hjm 1 j hjm 2 j hjm 3 j hjm hjm 4 j hjm (3) (8) PARTICLE SWARM OPTIMIZATION (PSO) 4 1 A . C5 j . called pbest. Each particle keeps track of its previous best position. C4 j . vidk . m ∈ M (iii) Hydraulic network constraints (5) And discharge rate of jth reservoir. These constraints include: (a) Physical limitations on reservoir storage volumes V hjmin ≤ V hjm ≤ V hjmax j ∈ N h . If promising new solution is discovered by a particle then all other particles will move closer to it. m ∈ M (6) I hjm jth hydro unit.Q (4) Psimin ≤ Psim ≤ Psimax j ∈ Nh . Based on PSO concept. Nh m. m ∈ M volume of jth reservoir. 1]. Minimum and maximum water max hj . Constraints (i) Power balance constraints Total power generated by hydel and thermal units must balance the predicted power demand. r1 and r 2 are random numbers uniformly distributed within [0. Here the objective function and associated constraints of the hydrothermal scheduling problem are formulated.min hj Q (ii) Generation limits Storage volume of jth reservoir Vhjm Phjmin ≤ Phjm ≤ Phjmax at time m. j∈ Nh. and corresponding fitness in its memory. Target is to generate maximum power as possible by hydro generating unit. Time index and scheduling period. Spillage of jth reservoir at time m. is a population based optimization algorithm. ( PROBLEM FORMULATION A. Number of thermal generating units. The population is called ‘swarm’. In terms of real power output.Vhjmax j ∈ Nh . C1 j . at each time interval over the scheduling horizon. So aim of the hydrothermal scheduling is to minimize the fuel cost of thermal units. Velocity updating equation: (k +1) (k) v = w× v + C × rand( ) × pbestid −xk + C × rand( ) × gbestid −xk id 2 id id id 1 (9) Position updating equation: ( k + 1) k + v ( k + 1) x id = x id id where xidk . Incremental cost of hydrogenating unit is zero. NIT Durgapur . Minimum and maximum storage Vhjmin . C 2 are the cognitive and the social parameters. 92 of 446 Copyright @ Department of Electrical Engineering. Inflow rate of jth reservoir at time m Ruj Number of upstream units directly S hjm above jth hydro plant. T1 j Water transport delay from reservoir Q hjmin ≤ Q hjm ≤ Q hjmax (b) The continuity equation for the hydro reservoir network Ns l to j. M and discharge rates. Each potential solution is called particle which is given a random velocity and is flown through the solution space (similar to the search process for food of a bird swarm) searching for the optimal position. total cost can be expressed as the following: j ∈ Nh . mathematical equations for the searching process are: A S E I Q +C 5 j hjm 6j (7) Number of hydro generating units. vidk +1 are the velocity of the d th dimension of the i th particle at the k th and the (k + 1 )th iteration and C1 . The best value of pbest is called gbest. which is the best position discovered by the swarm. gbest d is the group best position of the d th dimension and w is the inertia weight factor. C2 j . +C m∈M Ruj V =V +I −Q −S + ∑ Qhl(m−t ) + Shl(m−t ) hj(m+1) hjm hjm hjm hjm lj lj l =1 III. first developed by Eberhart and Kennedy (1995). II. Objective function Fuel cost function of each thermal generating unit is expressed as a sum of quadratic and sinusoidal function because considering valve point effect [14]. xidk +1 are the position (10) of d th dimension (variable) of the i th particle at k th and (k + 1)th iteration. These bounds are determined by the physical reservoir and plant limitations as well as the versatile requirements of the hydro system. C3 j .

The water discharge rate of the jth hydro plant Qhjd in the dependent interval d is then calculated by V..…. PSOCFIWA algorithms.Psi. . VII. PARTICLE SWARM OPTIMIZATION WITH INERTIA WEIGHT FACTOR APPROACH(PSOIWA) For particle swarm optimization with inertia weight factor approach (PSOIWA). k max = maximum no of iteration cycle. PSOIWA. Constriction factor (CFa) varies 0. PARTICLE SWARM OPTIMIZATION WITH CONSTRICTION FACTOR AND INERTIA WEIGHT FACTOR APPROACH (PSOCFIWA) For particle swarm optimization with constriction factor and inertia weight factor approach (PSOCFIWA). inertia weight ( w (k + 1)th wk +1 = k +1 Q (13) Velocity updating equation: ( k +1) (k ) v = w k +1 × v + C × rand id id 1 + C × rand 2 k ( ) × gbest id − xid k ( ) × pbest id − xid S E I VI. hydrothermal scheduling problem have been solved by applying PSO.w=w max − w max − w min × iter itermax Psi = [Psi1.. . the velocity of equation no. The total optimum fuel cost obtained from PSO... Nh]T be a trial matrix designating the kth individual of a population to be evolved. Qhjm. (9) is manipulated in accordance with equation no. PSOCFA. 93 of 446 Copyright @ Department of Electrical Engineering. The generation limits... PSOCFA.6 to 0. hourly storage volume of hydro reservoirs. PSOCFIWA techniques are shown in Table IV. In inertia weight factor approach (IWA). Psi2. Qhj2. IV.. the velocity of equation no. reservoir inflows. The scheduling period has been maintained to 24 h with 1 h time interval. Assuming the spillage to be zero in (8). hourly plant discharge are obtained using PSO technique is shown in Tables I to III. (14).V + ∑ I + ∑ ∑ Q − ∑ I hjm hjm hl m − tl hjd hjo hjM m =1 m =1 l =1 m =1 N N s s P =P − ∑ P − ∑ P sdgm Dm sim hjm i =1 j =1 (18) i≠d VIII. Psim. The elements Psim is the the power output of the ith thermal unit at time m. respectively.4. SIMULATION RESULTS PSOCFIWA has been applied to a test system [11] which consists of a multi-chain cascade of four hydro units and three thermal units. (9) is manipulated in accordance with the following equation w k + 1 × v ( k ) + C × rand ( )× pbest k 1 id − x id id ( k + 1) v = CFa × id k + C 2 × rand ( )× gbest id − x id (15) Here also.73.. j ∈ N h (16) To met accurately the boundaries on initial and final reservoir storage in (8).PsiM ] Qhj = [Qhj1. Hydrothermal generation (MW) scheduling. . The optimum fuel cost obtained from PSO. DEVELOPMENT OF THE PROPOSED ALGORITHM In this section. NIT Durgapur ... Qh. wmin =0.. (9) is manipulated in accordance with equation no. QhjM]. the velocity of equation no..60 to 0. PSOCFIWA are shown in Figure 1-4. The range of the elements Psim and Qhjm is to be satisfied the thermal generating capacity and the water discharge rate constraints in (5) and (7). (17) The dependent water discharge rate must satisfy the constraints in (7). . . The hydro sub-system configuration and network matrix including the water time delays has been considered. PSOCFA. . cost coefficients of thermal units are given in [11]. The load demand. reservoir limits are given in [11]. So we can determine the reservoir volume on different interval. Let pk= [Ps1. (12). M M Ruj M V −V = ∑ Qhjm − ∑ ∑ Q − ∑ I hjm m −t hj 0 hjM hl m =1 m =1 l =1 lj m =1 w × v ( k ) + C × rand ( )× pbest k 1 id − x id (12) id ( k + 1) v = CFa × id k + C 2 × rand ( )× gbest id − x id Constriction factor (CFa) varies from 0.. The program is written in Matlab language and executed on a CORE i3 personal computer. the thermal generation Psdgm of the dependent thermal generating unit dg can then be calculated using the following equation: 4 1 A (14) Where wmax =1. The elements Qhjm is the discharge rate of the jth hydro plant at time m. The hydraulic continuity constraints are (11) Where iter is the maximum number of iterations and max iter is the current number of iterations. PSOIWA. hydro unit power generation coefficients. water discharge rate to meet exactly the power balance constraints in (2). PSOIWA.73 and the inertia weight factor approach (IWA) follows equation (13).. Ps2.PsNs. m≠d ) at cycle is given by wmax − (wmax − wmin ) × (k + 1) k max R M M uj M =V .. Qh1…. PARTICLE SWARM OPTIMIZATION WITH CONSTRICTION FACTOR APPROACH (PSOCFA) For particle swarm optimization with constriction factor approach (PSOCFA).

0000 140.0000 189.9941 9.9901 186.0000 40.0632 56.0000 157.8486 27.4867 6.8718 82.3440 20.6932 10.6197 116.5817 21 102.3218 92.9643 60.6212 3 222.0898 25.0000 40.0000 163.5618 253.9525 30.3198 159.0000 40.Table I: HYDROTHERMAL GENERATION (MW) SCHEDULING USING PSO Hour Ph1 Ph2 Ph3 Ph4 Ps1 Ps2 Ps3 1 207.2714 20.7016 115.1556 110.4555 17.0000 40.8659 64.1664 99.2327 6.1901 20.9185 Qh2 10.1852 7.0035 123.8605 2 85.1578 96.7001 100.2936 17.9531 205.9820 20.1147 233.7871 137.7142 4 105.6615 166.2122 211.3007 5.3041 20.0000 40.8853 73.0000 285.0600 12 401.6525 14.0148 73.0977 70.0791 12.1768 6.0166 7.0000 40.0768 18 100.0000 40.8404 23 289.0000 170.8120 85.4470 27.7977 15 333.8165 8 137.9087 5.0826 167.3591 20.0000 40.7270 12.0000 154.0000 40.8958 26.0000 368.6145 172.0560 22 368.6449 152.8797 42.9825 15 119.1556 19 130.5187 61.1860 192.0000 40.4598 256.0000 40.7597 7.8841 2 294.0000 40.8939 50.0000 22 96.0000 40.2198 24 238.0006 11.7740 20.7489 483.4045 17 146.5015 20.1391 174.4946 126.6422 90.1590 23 118.0000 172.5824 100.8511 13 119.0000 205.2326 74.3598 9.5522 8.0254 141.3390 163.4782 141.9506 215.9612 106.2784 193.6891 23.0000 70.5659 16 412.4627 65.4973 14 128.0000 75.4632 23.5214 161.4531 55.9251 11 93.4906 95.0000 277.8183 6.1788 3.5014 12.2868 65.1587 7.8369 4 186.0599 70.0000 1 119.8595 64.6584 75.0000 40.5231 11.5203 20.5736 85.4086 15.4310 112.8924 5.0000 137.5314 89.0000 40.3159 75.3818 209.9328 141.0000 40.6140 177.7732 107.6956 16 86.9480 84.8120 145.0688 63.3550 166.2049 38.0601 100.3913 114.5191 8 258.7390 24 120.0899 82.2060 5 203.2453 9 10 450.4037 84.5429 218.6581 121.1832 59.0000 113.5046 89.2169 82.2117 Qh3 16.1540 19.7163 194.0000 40.7157 132.5909 20.2859 13 365.3369 132.1143 251.9400 93.6301 258.9220 20.0000 Table III: HOURLY PLANT DISCHARGE(X 104 M3) OBTAINED USING PSO HOUR 1 2 3 4 5 6 7 8 9 10 11 12 Qh1 5.0000 237.6241 20.8161 23.9573 92.0000 170.0371 6 313.8723 248.9277 20.0000 425.2811 18 492.0000 120.0000 208.2035 7 420.3356 20.9469 17 371.5532 5.1407 14 500.8214 98.2759 20.2928 12 87.4887 100.3307 10 137.0000 40.0000 80.4242 8.0000 145.6525 20.8454 11.1915 20 99.7894 115.0000 196.0000 40.0000 107.8706 118.0000 205.7181 93. NIT Durgapur Qh4 12.0000 40.6628 58.6053 .9435 14.4351 80.0449 20.9477 6 122.3304 207.0000 40.6698 0 156.9727 21 405.6456 33.3372 250.0000 241.6607 86.0000 20.4952 20.4481 3 93.4902 46.9538 0 124.9271 134.1766 21.0000 40.4499 14.7372 S E I Table II: HOURLY STORAGE VOLUME OF HYDRO RESERVOIRS(X 104 M3) USING PSO HOUR Vol1 Vol2 Vol3 Vol4 0 100.6771 20.5656 20.9591 20.9055 167.7713 28.0000 140.9304 6.2798 88.0000 224.1747 75.0199 76.6114 5.2936 7.6326 6.2615 193.8607 78.5989 163.4425 113.0268 7 138.8468 53.0000 389.8904 20.7979 8.7205 53.8161 94 of 446 Copyright @ Department of Electrical Engineering.7944 27.2139 30.0000 227.6778 120.0786 20 500.4812 128.0000 172.7710 42.2472 20.5020 146.3555 9.0199 93.5950 215.4889 9 135.0106 19 344.7261 107.5532 6.8819 112.0254 88.8325 24.4728 4 1 A 153.4947 181.0000 252.0136 76.9970 91.3499 52.7170 20.0000 140.1428 74.0000 340.0308 5 144.0000 40.8900 11 283.5446 12.0000 90.4986 62.9754 12.9086 78.0000 86.0000 40.

3 3. 3 Convergence Behavior of the PSOCFA Method 4 x 10 3.14 3.28 3.1 0 10 20 30 40 50 60 Iteration --> 70 80 90 100 Fig.0301 11. Constraints on the operational limits of the thermal and hydro units on the reservoir volume limits are also considered. The effect of valve point loading is also considered.0779 18.8363 14.7747 9.7002 10.0932 13.7184 10.1918 24.2584 11.6471 7.1240 10.4891 7.9 2.06 3.1890 7.05 3.2478 16.3 Fuel Cost --> Fuel Cost --> 4 1 3.85 0 10 20 30 40 S E I 50 60 Iteration --> 70 80 90 100 3.6207 13.892 PSOIWA 30.0244 12.0068 13.0795 9.4103 9. PSOIWA. NIT Durgapur .08 3.22 3.4381 9. Four hydro and three thermal plants are considered to solve hydrothermal scheduling problem. PSO.95 2.12 3.2529 9.1556 8.26 3.2 3.12.5247 15.1409 9.4 Fuel Cost --> 13 14 15 16 17 18 19 20 21 22 23 24 3.1487 9. 4 Convergence Behavior of the PSOCFIWA Method Fig. From the results it is clear that PSO approach gives the best global optimum solution than the PSOIWA.4876 17.32 3.2 3.1 3.9540 12.601 3.6781 21.402 PSOCFA 31.4540 15.15 Method Cost ($) PSO 28.16 0 10 20 30 40 50 60 Iteration --> 70 80 90 100 Fig. The generator load power balance equation and total water discharge equation have been subsumed into system model. PSOCFIWA has been successfully introduced to solve the hydrothermal scheduling problem.18 Fuel Cost --> 3.8809 24.3471 11. The numerical results show that the proposed approach PSO is better than other techniques in terms of having better solution quality and good convergence characteristics.18 3. 1Convergence Behavior of the PSO Method 4 x 10 3. PSOCFA.500 PSOCFIWA 31.22 3.1 3.9619 14. CONCLUSIONS In this paper.4518 10.5816 11.9518 4 x 10 3.5 3.8330 10.55 3. PSOCFA and PSOCFIWA techniques.2162 8.16 3.7838 22.0788 16.24 A 3 3.34 4 x 10 3.2256 10.8808 14.35 3.1129 11.6191 6.7481 14.45 3.0102 18.4542 12.6849 11. 2 Convergence Behavior of the PSOIWA Method 95 of 446 Copyright @ Department of Electrical Engineering.25 3.04 0 10 20 30 40 50 60 Iteration --> 70 80 90 100 IX.1560 21.15 3.2 2.2 Table IV: TOTAL OPTIMUM FUEL COST 3. Fig.

69 pp. C. Ontae. NIT Durgapur . [5] J. [12] Chandrasekar Samudi. Power syst. 2008. P. 1998.T. 2000. Ho. 4 1 A S E I 96 of 446 Copyright @ Department of Electrical Engineering. IEE Proc. 15(November (5)) pp.-Y.S. Short-term hydrothermal scheduling part 1: simulated annealing approach. 1994. The short-term hydrothermal coordination via genetic algorithms.K. Ramirez. Electr..M. 501-518.Yu. J. H. Energy convers.-C. X. H. 2004. Gil. pp. 1256–1264.K.M.O. M.R. Power Syst.E. S. C. Chang. Gen. R. H. P. M. P. Appled soft computing 8 pp. A genetic algorithm modeling framework and solution technique for short term optimal hydrothermal scheduling. An Interactive fuzzy satisfying method based on evolutionary programming technique for multiobjective short-term hydrothermal scheduling. Res. [13] Wang. IEEE Trans. [11] K. Vol. Huang. 189-196. 1998. 13(2) pp. 1996. 2003. [10] N. Y. IEEE Trans.=Y. N. Sinha. 978-1-4244-1904-3/08/ 2008 IEEE. IEEE Trans.. 2006.R. 8. Electr. IEEE Trans. Power Syst.-H Chen. cascaded reservoirs and uncertain data. J. Wong. [15] B. Chattopadhyay. 497–501. 975-981.Das. Transm. EPSR 52 (November (2)) pp. “Genetic added scheduling of hydraulically coupled plants in hydro-thermal coordination”.Sreeni. [6] Y. “A genetic algorithm modeling framework and solution technique for short-term optimal hydrothermal scheduling”.”. Basu. 277–285. R.area hydrothermal power systems with tie-line constraints”. 1902-1908. Power Components Syst. Shahidehpour. [8] P. 3. [9] P. IEEE Trans. 18 (4) pp. Transm. Scheduling shortterm hydrothermal generation using evolutionary Programming techniques. Power syst. T. “Power generation Scheduling for multi. Piyush C. Ojha. “A diploid genetic approach to short-term scheduling of hydrothermal systems”.Wang. 1993. Wu.Chakraborty.K. Wang. 2007. 2003. Distribution 143 (4) pp.C.L. Irving. Bustos. 1–19. 1268-1274. [7] S.W. Yang. No.Basu. 1999. “Shortterm hydrothermal scheduling through evolutionary technique”. [3] S. Rudnick. 1333-1340. Manage. Wong.48 pp. “Short-term hydro-thermal scheduling using Particle swarm optimization method”. Yang. Yuan. Chattopadhyay. [4] E. Orero. 371–376. IEEE Trans. GauthamP. [14] M. PWRS. Irving. Mandal. Gen. “Fast evolutionary technique for short-term hydrothermal scheduling”.O. 1996.Chakrabarti. Chakrabarti. Short-term hydrothermal generation scheduling model using a genetic algorithm. Power syst.P. 11(may(2)) pp. IEE Proc. [2] P. J. “Particle Swarm Optimization technique based short-term hydrothermal Scheduling”. 214-220. M. D.REFERENCE [1] K.-G.K. IEEE Trans. Distribution 141 (5) pp. 18(1) pp. Orero. 34 pp. PWRS 13 (2). Power Syst. C. 1392-1399. Hota.

97 of 446 Copyright @ Department of Electrical Engineering.e. we first created a 2D finite element model of the human thorax and divided the lung into three coronal layers. water in the lung) and thus. the electrodes are equidistantly attached to the patient’s skin around the thorax. More objectively. 9].. In order to achieve maximum benefit. we examined which current-injection and voltage measurement electrode pairs are best suited for a proper separation between lung and heart activity. and head imaging [6]. I. During a measurement.g. Finite Element Method (FEM). Thus. Acute Respiratory Distress Syndrome (ARDS). analysis is made to find the electrode pairs capable of separating the lung and heart activity when a particular amount of constant current is injected through them are also carried out. EIT-derived numeric parameters would help the physician to evaluate the state of the lung. Finally. a finite element method (FEM)model of a human thorax in 3 dimensional platform of FEM Multiphysics software is created and is tested with new ventilation indices regarding their ability to quantitatively describe structural changes in the lung due to the gravitationally dependent lung collapse. adjacent drive/adjacent receive). In this procedure. variations of the conductivity 4 1 A S E I Keywords— Electrical Impedance Tomography (EIT). Kolkata: 700 107. METHODS A. new ventilator settings. The attending physician should be able to immediately recognize ventilation changes in the lung resulting from.. we altered the fluid contents of the dorsal layers close to the back in order to simulate lung collapse of increasing severity . Kolkata: 700 032. Inspite of low spatial resolution (about one tenth of the thorax diameter). as EIT seems also to be capable of distinguishing between heart and lung related changes of the conductivity in the thorax [10]. EIT-derived numeric parameters could be able to evaluate the objective more precisely in comparison with the visual observations of the EIT images (subtle changes are simply hard to detect). Heritage Institute of Technology. Additionally.e. lung collapse) and pulmonary edema (i. we looked at different methods to quantify the given ventilation changes and compared them with the actual modifications made in the thorax model. reconstructed images with this data visualize the change in conductivity of the specific area of interest[5]. EIT is well suited in providing an instantaneous feedback to the medical staff [8. ARDS is a life-threatening state of the lung characterized by atelectasis (i.. adjacent electrodes are fed with a small alternating currents and the potentials at the remaining electrodes are measured (i. India E-mail: madhurima02@gmail.e. Then. further investigations is to be carried out to process the fast sequence of EIT images (up to 50 images/s) and also the way out for the extraction and quantification of the underlying physiological information.Monitoring of the lung fluid movement and estimation of lung area using Electrical Impedance Tomography: A Simulation Study Deborshi Chakraborty Madhurima Chattopadhyay Dept. Then. EIT is use to monitor a variety of clinical problems such as respiratory failure.. II. ARDS patients have to be mechanically ventilated with higher oxygen concentrations and changing pressure levels resulting in dynamic shifts of body fluids and lung aeration. Additionally. NIT Durgapur . gastric emptying. e. Therefore. Modeling An adequate geometric model of the human thorax is designed with realistic values of the electrical properties of biological tissue for the proper assessment of lung collapse. here we have performed a Finite Element Method based simulation study for monitoring the condition of lungs and heart of ARDS patients. of Instrumentation Science Jadavpur University. EIT-based trend can analysis for different kinds of diseases with the help of numeric parameters. we calculated the average current density for the different tissues and compared them based on the position of the current injection. For visualizing those shifts. After reconstruction. Therefore. India E-mail: dcphy123@gmail. of Applied Electronics and Instrumentation Engg. lungs activity. INTRODUCTION The estimation of the spatial conductivity distribution in a thorax cross section is carried out by thoracic electrical impedance tomography (EIT) as it is a non-invasive technique[1-4]. cardiac volume changes. but due to its non-invasiveness and high temporal resolution it seems to be particularly useful for the bedside monitoring of patients suffering from the acute respiratory distress syndrome (ARDS) [7]. In recent years. From this EIT based reconstructed images an approximate determination of the lung area is obtained.Then the thorax model from the 2D plate were transformed to the 3D plane for the analysis. heart activity. distribution in the thorax.com Abstract— Patients suffering from the acute respiratory distress syndrome (ARDS) requires thoracic electrical impedance tomography (EIT) for the monitoring their conditions ranging from dynamic shifting of body fluids to lung aeration right at the bedside. Hence.com Dept.

Due to the weight of its upper parts. a correct geometric dimension of the model (shown if Fig. 17]. body fluids etc. the geometric changes during a breath cycle are omitted.35182 10094. During the mesh generation. 1. we divided the lung into three coronal layers and modeled different amounts of lung collapse by subsequently replacing the bottommost layers with collapsed lung tissue [Fig.To solve this equation. c) Forward Problem A relation can be obtained between the voltage measurements made on the boundary (∂Ω) and the domain conductivity can be found [12. equation (1) can be solved numerically using FEM to calculate the nodal potentials of the domain for the known conductivity (σ). -I is for sink electrode and 0 otherwise. Visceral movements (i. 1000) were assumed to be a mixture of deflated lung tissue and well-conducting body fluid[14]. of 98 of 446 Copyright @ Department of Electrical Engineering. cf. is given by. conductivity and relative permittivity of the lung were varied between deflated and inflated lung as given in [12].Thus.50 Heart* 0. If K and σ are known. In order to simulate patient breathing.200 KHZ [12]. a relationship can be established between the electrical conductivity (σ) and spatial potential (Φ). to obtain qualitative character. Thus. It is known as the “forward problem”. geometric changes) during a breath cycle are not considered as it has been reported to have a rather low influence for state-differential reconstruction algorithms [13]. 1(a)) from real CT images of a human male. The currents of 10 mA @200 kHz (as frequencies between 20 and 200 kHz are widely used for thoracic single-frequency EIT) are injected across the electrode pair. This straightforward simplification should be sufficient. PROPERTIES OF THE MODELED TISSUES Conductivity (S/m) Relative permittivity Lung* (deflated) 0. ribs.19543 16982. ELECTRICAL BETWEEN 20 .42 Muscle* 0. the following boundary conditions are to be known a) Dirichlet Boundary Condition: (2) 4 1 i Where. +I is for source electrode.00 Body Fluid 1. 1(a)] without any breathing activity.02064 264.50000 98.40 Lung *(inflated) 0. Mathematical Model For a known conductivity. As before.45 S/m. NIT Durgapur . i =1…N. the nodal potentials within the phantom area are computed using the forward solver problem of EIT [15. The value of the conductivity and permittivity of the blood. The electrical properties of the collapsed areas (0. TABLE I.e. 1 provides the details overview of the electrode placement and numbering of adjacent electrode pairs. bones (spinal column.19 Fat* 0. necessary simplifications and conditions are made were similar to the ones described in [11].02424 172. 13] as.70080 5197. and sternum). variation of its electrical properties between “normal” heart and blood-filled heart (a mixture of blood and heart properties: 0.10265 4272.70 (Asterisk-Marked (*) properties were used for the simulation of the reference state) (b) Fig. 0 (1) The potential field inside of the conductor having no internal electrical sources or sinks. must satisfy the equation (1). ∂Ф is the boundary and η is the outward normal vector to the electrode surface. are the measured potentials on the electrodes. called the Governing Equation [17] of EIT. for a known value of a constant current. The description of this problem is discussed below. Ф is the vector of voltage measurements and K is the transformation matrix relating Ф to σ. In total. to calculate the nodal potential [17]. b) Neumann Boundary Condition: A S E I Tissue Blood (a) I I 0 (3) Here.00 Bone* (cortial) 0. In this paper forward problem solver algorithm is used. Table I. 11 000) were considered. For the simulation of the heart activity.26197 8531.. five different types of tissue were being modeled with electrical properties taken from the literature: muscle. heart.56 0. (a) Electrode placement and (b) numbering of electrode pairs B. the lung tends to collapse in the lower parts first (commonly termed as gravitationally dependent collapse). and lung. The resulting mesh consisted of 454847 first-order triangular elements are created with FEM based Multiphysics software. Electro-dynamics of EIT is governed by a nonlinear partial differential equation. Therefore. as mainly the missing ventilation of the collapsed lung tissue will influence the reconstructed images of conductivity change. Fig. Patients suffering from acute respiratory failure are usually treated in supine position.6 S/m. (4) K Where σ is a vector of conductivity values. fat.

6 0. Determination of Heart Activity The amount of current through the different tissues is obviously influenced by the position of the current-injection electrode pair. 3. D. 2). 2. 2 4 6 8 10 12 14 16 18 Position of the current injection electrode pair Fig. Especially. 3 it is seen that depending on the position of the current-injection electrode pair. Field Simulation The FEM Multiphysics software is used to simulate the models. The reconstructed images represent the conductivity changes based on either the breathing or the heartbeat activity. Electrode pair #5 will lead to a greater amount of current flowing through the (right) lung. J III. Plot of current densities of heart and lungs J*heart and J*lung vs. Simulations are conducted similar to the measuring procedure of the EIT system. #8. Hence. Deflated lung Inflated lung 1.2. #15. 4 shows that the collapse of different layers of the lungs starting from the lowermost part.2 * lung 1. electrode pairs #1. The voltage distribution plots as of the lungs in Fig. The qualitative results for the best electrode pairs to measure the heart activity were additionally backed up by looking at the voltage changes at the boundary. #9.8 0. In order to quantify dorsoventral as well as lateral shifts of the ventilation distributions resulting from different amounts of lung collapse are studied [18]. SNR was maximized (here. 0. For each physiological state of the lung. 3. C.4 0 From the Fig. the lung activity is regarded as “noise”). when using the identified optimal current-injection electrodes as receiver electrodes (e. 4 it is seen that the different amount of collapsed lung area which is determined by forward solver algorithm.0 * A. From these figures approximate 99 of 446 Copyright @ Department of Electrical Engineering. NIT Durgapur . 4 1 RESULTS AND DISCUSSIONS 1. current-injection pair: #16.g. the amount of current through heart and lung varies.the designed FEM model are obtained from Luepschen et al. For the detection of the heart activity.6 heart S E I A Fig.Current density distribution and potential isolines plots from the FEM based simulation study for different current-injection positions. Image Reconstruction and Processing State-differential EIT images were reconstructed from the simulated 208 voltages using the forward algorithm [17] and the boundary conditions as discussed in section B. 2. 1. Determination Of Lung Collapse And Lung Area From the Fig. [20]. receive pair: #8).. whereas injection through electrode pair #16 maximizes the current flow in the heart it can easily be visualize from the plot of the ratio of the current density of heart to that of the lungs as shown in Fig. and #16 provide the best chances for the observation. voltages distributions is determined for each state of the lung (shown in Fig. the position of the current injection electrode pair 3.4 /J The corresponding current density images of the FEM simulations are provided in Fig. 16 simulations are to be carried out to account for all current-injection positions while determining the voltages between the remaining 13 electrode pairs.

H.-Sci Mens. 2003. R.cnr. 100 of 446 Copyright @ Department of Electrical Engineering. pp. 43. vol. “Dielectric properties of body tissue. vol. Sarwan Kumar. Wiley India. G. Anderson. Ying Li. Luepschen. No.” Am. 1403–6. Volume 37. Vol. T. IEEE Transactions on Medical Imaging.. vol. H. Care Med. 2007. Node Back Projection Algorithm”. Meas.Med. vol. Epub ahead of print. and William R. “Imbalances in regional lung ventilation—A validation study on electrical impedance tomography. Leibecke.6-7 March. Zou & Z. Vol. 6. “Impedance imaging of lung ventilation: Do we need to account for chest expansion?. #9. Shuai Zhang. studies on the electromagnetic field distribution in the human thorax are important to understand because it provides an estimated idea relating to the varying current paths. Proceedings of the 20th Annual International conference of IEEE EMBS. 2007.Biomed. REFERENCES [1] [13] [14] [15] [16] [17] [18] [19] [20] Mohd Tahir Erwati. Meier.“Medical Instrumentation Application and Design”.. British Columbia.Nagaraju. Leibecke. pp. T. vol. [3] [4] [5] [6] [7] [8] [9] [10] Fig . 28. Lionheart.”Impedence based image reconstruction of the field distribution inside the closed phantom using finite element method”.. Meier. 31.R. J. Govt. IEE Prac. J. 24. John G.ifac. and S. Jianjun Zhang. July 2010. 97–108. Vol. “Electrical impedance tomography (EIT): A review.L. A.(SCOReD 2009). J.” IEEE Trans. of India. 2007. 169. Karsten. 2. Proceeding of 2009 IEEE International Advance Computing Conference (IACC 2009). 2004. Issue 4. 1996 Alistair Boyle. M. H.” Am. Y. Eng.C.[2] determination of the lung area can also be estimated. Canada. Tushar kanti Bera. “Shape Deformation in Two-Dimensional Electrical Impedance Tomography”.. pp. Karsten. Adler. pp. which can provide an overview condition of the lungs to the physician. no. and #16 provide the best SNR for the detection of the heart activity. December 2012. T. India. Leonhardt. Institute for Applied Physics. “Protective ventilation using electrical impedance tomography. Crit. Webster . 2nd Edition. R. Malaysia. J.” Jun. Respir. The influence of the lung activity is minimized while considering these pairs of electrodes. “A New Algorithm to Reconstruct EIT Images. France August 23-26. Shuo Yang and Weili Yan”Three Dimensional Electrical Impedance Tomography in Thorax Complete Model” 30th Annual International IEEE EMBS Conference Vancouver. 2007 [Online]. E. and S Leonhardt. Smith et al. of 2009 IEEE Students Conference on research and development.. and Y. J. 4 1 A [12] During EIT measurements. #15. P. Magn. & Nagi Farrukh.. S247–S260. Lippincott Williams & Wilkins. 27. Guo. Leidholdt Jr. “Modeling of Fluid Shifts in Human Thorax for Electrical Impedance Tomography”. Guizhi Xu & Quanming Zhao.. Eng. 2. a real time implementation of the system as well as completve three dimensional analysis of the human thorax with EIT [19] is necessary so that the results obtained from the simulation based study can be correlated with the practical obtained experimental results. No. UPM Serdang. no. In considering the structural changes due to the breathing and the heart beat would give more precise understanding of the occurring reconstruction artifacts.it/tissprop A. 6. J. H.” Intensive Care Med. Victorino et al. Mohammed and F.. Luepschen. 29. Huanli Wu. Mar. D.Tozer. Reconstructed differential EIT images for different amounts of collapsed lung with forward solver algorithm IV. CONCLUSIONS [11] S E I Finally in a future.”Physiol. 2003.Available: http://niremf. IEEE Transactions On Magnetics. 16-18 Nov ’09.757-764... pp. Jafari “A 3D hybrid BE–FE solution to the forward problem of electrical impedance tomography”. 414–420.”IEEE Trans. Lyon. vol. 491–9. This knowledge is very much vital to understand the pathological changes obtained from the EITbased observations.. pp. Henning Luepschen et al. Hongli Yu. #8. A.. Boone. 2013. “A FEM-Based Forward Solver for Studying the Forward Problem of Electrical Impedance Tomography (EIT) with A Practical Biological Phantom”. No 7. Grossherr. “Analytic solution of the forward problem for induced current electrical impedance tomography systems”. Brown. ISBN-10: 0683301187. M. Italian National Research Council. O.Care Med. 2009. Berthiaume. Guardo. Andy Adler.. (IJCNS) International journal of computer and network security. A. June 2008. 791–800. Technol. 12. Guizhi Xu. John M. “Applications of electrical impedance tomography for imaging in biomedical and material technology”.Gehring. No. “The Essential Physics of Medical Imaging”.. pp. Respir. B. August 20-24. TechnoL. Sneh Anand & Amit Sengupta. “Electrical impedance tomography to measure pulmonary perfusion: Is the reproducibility high enough for clinical practice?.. Vol. Grossherr. Patiala.” J. Proc. November 1995. T. 44. Bushberg. third edition. H. T.4. Engineering Analysis with Boundary Elements. J. Ghaderi Daneshmand. The electrode pairs #1. Cite Internationale. July 25. 2007. M.IEEE . “Detailed 2-D and 3-D finite element modeling of the human body for the evaluation of defibrillation fields.. for providing financial assistantship as INSPIRE Fellowship. Weili Yan. 2008978-1-4244-1815-2/08. ACKNOWLEDGMENT The author Deborshi Chakraborty like to acknowledges Department of Science and Technology.I. 1993. Crit.Freeston. 2. B. “A review of electrical impedance techniques for breast cancer detection” Elsevier Medical Engineering and Physics 25 / (2003). Seibert. NIT Durgapur . “Assessment of regional lung recruitment and derecruitment during a PEEP trial based on electrical impedance tomography. J. K. 142. Uler.

The supplementary control action is used to minimize the frequency deviation. A multi area adaptive LFC based on SelfTuning Regulator (STR) for Automatic Generator Control Simulator (AGCS) is investigated in [6]. PI and PID controller parameters tuning based on Ziegler-Nichols method for distributed generation system (DGS) has been proposed in [2]. advance control engineering techniques are used to design LFC system of large scale complex and uncertain power system. have been introduced to design the LFC system. not-defined parameters and model. In recent past. if persist after primary control. [9. LFC is important to keep the system frequency and the inter-area tieline power as close as possible the schedule values. A Proportional Integral Derivative (PID) controller is used for the design and analysis of the proposed model. with two or more independently controlled areas. The control of tie-line power and frequency is commonly known as load frequency control (LFC). 10]. A systematic way of P. governor dead band. Yao Zhang et al. A unique objective function is also formulated considering the transient specifications. Bevrani et al. In an interconnected system. The Particle swarm optimization (PSO) algorithm is developed to obtain suitable control parameters to achieve the optimum performance. consumers in order to efficiently deliver reliable. the use of renewable sources results in frequent variations of frequency in the system as the power supplied is not constant. of Electrical Engineering National Institute of Technology Durgapur. LFC is accomplished by two different control actions: primary control and supplementary control. The frequency of a system is dependent on active power balance. The first one is based on H∞ control design using LMI technique and the second controller is a PI type. and is tuned by GALMI technique to mimic the same robust performance of the first one. PID controller. a detail structured singular value method and eigen value method is proposed for local area and tie-line robustness analysis. The governor will try to minimize the frequency and tie line power deviation to zero by manipulating input to the turbine. LFC using controller based Dynamic Demand Control (DDC) can be used to maintain the frequency at nominal value. India pain. Takaki-Sugeno fuzzy model of the power system has been constructed for designing the fuzzy model based LFC. In [5]. 4 1 A S E I In general. The controllers designed so far using traditional and advanced control techniques for LFC system are model based controllers that performed better on the specific model and are not usable for large power systems like power systems with non-linearities. Khodabakhshian and Edrisi [4] used a new tuning algorithm which is based on maximum peak resonance specification supported by Nichols chart to tune the PID controller. of Electrical Engineering National Institute of Technology Durgapur. model uncertainty and large disturbances. India Abstract— This paper presents a practical model for load frequency control (LFC) of two-area interconnected power system. Keywords—load frequency control. Regulation of frequency can be done by control of load as well as the governor at the generating stations. INTRODUCTION For large scale interconnected power system. Alireza et al. to zero through integral control action [1]. When load changes. particle swarm optimization I. NIT Durgapur . the primary speed control perform the initial readjustment of the frequency and tie-power by the action of governor itself. In [8]. smart grid predicts and intelligently responds to actions of all suppliers. The proposed model is a non-linear power system and is framed with thermal generating units including security constraint governor dead band imposed by the governor. the generation within each area has to be controlled so as to maintain scheduled power interchange. addressed a new decentralized fuzzy-logic based LFC scheme for minimization of frequency deviation and tie-line power changes in the presence of high penetration wind turbines [12].com Dept. the design of intelligent controller which are more adaptive than linear and robust/optimal controller.Multiobjective Design of Realistic Load Frequency Control using PSO Santigopal Pain Parimal Acharjee Dept. economic electricity services. This will reduce the burden on generating station alone to regulate the frequency. two robust decentralized control design methodologies have been proposed. In [11]. The closed loop performance is examined in the presence of constraint scenario and the results and characteristics show that the developed algorithm gives better performances. Recent growth and development of power system industry and increased power demand has necessitated intelligent methodologies for practical control of the power system. A robust control design that utilizes loopshaping ideas for LFC regulation is proposed in [3] considering large uncertainty. two area power system. [7] proposed a new robust optimal MISO-PID controller for LFC.santigopal@gmail. An adaptive fuzzy gain scheduling scheme for conventional PI and optimal controllers has been proposed in 101 of 446 Copyright @ Department of Electrical Engineering. In modern days. Therefore. developed an active disturbance rejection control (ADRC) based robust decentralized LFC solution considering wide range of parameter variations. In smart grids.

2 and the system parameters are given in appendix A.1 where the blocks are: non-reheat steam turbine = 1/ (Tt s 1) . turbine and generator. As the load varies. LFC Model For understanding the control action of LFC. such as the energy from steam or water.8 (0. The classical/conventional approaches provide very poor performance for large network under practical constraints like governor dead band. Tg and Tt are the governor and turbine time constant. PID gains of Sugeno fuzzy logic based automatic generation control of multi-area thermal generating plants are optimized using classical particle swarm optimization.2 / )] / ( sTg 1) 102 of 446 Copyright @ Department of Electrical Engineering. The turbine unit is used to transform the natural energy. In multi area power system. The PID parameters of LFC system must be designed in such a way that ensures safe. Model of single area power system In the above configuration.1. The Particle Swarm Optimization (PSO) algorithm. A S E I This paper presents a realistic approach to optimize the proportional integral derivative gains of load frequency control system. load and machine = K p / (Tp s 1) . The short comings of the conventional approaches are the motivating factor of applying soft computing techniques for the designing of LFC system. The transfer function model of two area non-reheat thermal power system is depicted in fig. NIT Durgapur (1) .1. Governors are used to sense the frequency bias 4 1 B.[13]. caused by load change and cancel it by varying the input of the turbine. The transient performance shows the significant superiority of the proposed design approach. it is obvious that PSO is a power full tool which can be effectively used to design the LFC system. 17]. For quality electric power services. each area connected to others via tie line which is the basis for power exchange between them. The governor dead band non-linearity tends to produce a continuous sinusoidal oscillation of natural period of about T0 = 2 s. load frequency control is an important factor. The speed governor helps to match active power generation with the demand by controlling the throttle valves which monitor the steam input to the turbine. robust in nature and can give multi solutions. has been proposed in [14]. generation rate constraint. load frequency stabilization by coordinated control of thyristor controlled phase shifters (TCPS) and super conducting magnetic energy storage (SMES) have been investigated using craziness-based particle swarm optimization (CRPSO). The ACE of each area is the linear combination of the frequency and tie line error. A governor dead band is defined as the total magnitude of a sustained speed change where there is no change in valve position of the turbine. time delay etc. Genetic algorithm (GA) based parameter optimization of PID sliding mode LFC used in automatic generating control (AGC) of multi-area power systems with non-linear elements. u is the load reference and ∆Pd is the load change. Basic block of power generating unit consist of the combination of governor. System Under Study As an example of multi area power system. For the two area case the effect of governor dead band non-linearity is taken into account to make this approach a bit realistic one. The generator unit of the power systems converts the mechanical power received from the turbine into electrical power. to provide load sharing between generators proportionately and to maintain the tie line power exchange at schedule value. reliable and uninterrupted power supply. There will be area control error (ACE) for each area and this area will try to reduce its own ACE to zero. But for LFC. Under critical situation and security constraints. Soft computing techniques are not model specific. In [19]. ACE = Frequency error + Tie line error. The linear model of the LFC system is shown in fig.e. soft computing techniques can give highly satisfactory results. the primary objectives of the LFC are to keep the system frequency at nominal value. that will be met by the increase in generation in all the areas associated with a change in the tie line power and a reduction in frequency. the focus is on the rotor speed output (frequency of the power systems) of the generator instead of the energy transformation. The LFC analysis for single and multi area power system using PSO has been reported in [16. governor = 1/ (Tg s 1) . To make the analysis realistic one. hybrid particle swarm optimization. A two area thermal-thermal power system is considered for the design purpose. we have considered two area interconnected power system for LFC analysis. The transfer function of governor with non-linearity [17] can be expressed as: G ( s) [0. The particle swarm optimization algorithm is developed to optimize the PID gains and a novel objective function is designed to calculate the optimal controller gains more accurately in least time. the governor dead band is considered in this model which make the system non-linear. the speed/frequency of the generator changes. But the normal operating state of the power system is that the demand of each area will be satisfied at a normal frequency and each area will absorb its own load changes. For an interconnected system. in its different form have assumed much importance in recent years for optimization of complex control problems and it is being widely used in LFC design. From the above discussions. When there is change in power in area one. Kp=1/D and Tp =2H/fD where D is the ratio of load changes percentage to the frequency changes percentage and H is the inertia coefficient of generator. consider the non-reheat thermal power system shown in Fig. Fig. SYSTEM MODEL A. into mechanical power that is supplied to the generator. II. hybrid genetic algorithm simulated annealing and is reported in [18]. i. The decentralized LFC is formulated as multi-objective optimization problem in [15] and GA has used to tune the PI controller parameters of multiarea power system.

Ki is the integral gain and Kd is the derivative gain. Inertia weight is another important factor for swarm optimization problems. NIT Durgapur . …n and n is the size of the swarm.e. the PSO system might not converge at all. It is randomly selected within a certain range. proportional part reduces the error responses to disturbances. Transfer function model of two area interconnected power system with governor dead band III. The parameters of the PID controller have been designed using developed particle swarm optimization algorithm.4]. The control input of the power system u1 and u2 are the output of the controllers and these are obtained as u1 K p1 ACE1 K i1 ACE1dt K d 1d ( ACE1 ) / dt (7) u 2 K p 2 ACE2 K i 2 ACE2 dt K d 2 d ( ACE2 ) / dt (8) In LFC system. in order to convergence to optimal solution. The error input to the controllers are the respective area control errors (ACE) given by (6) ACEi Bi fi Ptie where i = 1. two different unique objective functions are formulated. upto the kth generation Fig. 3. Inertia weight (w) must not be constant for better results. CONTROL STRATEGY The control configuration for multi area power system is depicted in fig. S E I vik 1 w * vik c1 * rand() * ( pbestik xik ) c 2 * rand () * ( gbestik xik ) (3&4) w wmax 4 1 ( wmax wmin ) * present _ iteration max imum _ iteration (5) A IV. In this technique. Bi = frequency bias factor = (1 / Ri ) Di Control input to the power system is obtained by use of PID controller together with the area control errors ACE1 and ACE2. vik 1 : updated velocity of ith particle w PID CONTROLLER c1 & c2 : constriction factors In this study PID controller is used as supplementary control for LFC. k represent the no. The transfer function of PID controller is GPID ( s) Y (s ) / E ( s) K p K i / s K d s rand () : random number between 0 and 1. The inertia weight w is determined according to the following equation: (2) where Y(s) and E(s) are the controller output and tracking error signals in s-domain respectively. 2. Another best fitness value is also tracked in the iterations obtained so far. In every iteration all the particles will be updated by following the best previous position (pbest) and best particle among all the particles ( gbest) in the swarm. According to fitness value the best solution is determined in the current iteration and also the best fitness value is stored. The desired result should have minimum settling time. xik : current position of ith particle at kth generation vik : current velocity of ith particle at kth generation pbestik : pbest of ith particle for kth generation gbestik : gbest of ith particle considering the whole generation i. Random selection of w provides successful tracking for a dynamic optimization problem. Kp is the proportional gain. no overshoot and zero steady state error. robust performance and simplicity. The objective functions are derived considering steady state and transient response specifications and proper 103 of 446 Copyright @ Department of Electrical Engineering. For the optimization process. applicability. of iteration.2. The each particle updates its velocity and positions with following equations: xik 1 xik vik 1 : inertia weight for ith particle At the end of the iteration. the integral part minimizes the steady-state error and the derivative term improves the transient response and stability of the system.where i = 1. the best position of the swarm will be the solution of the problem. V. PSO does not provide always the optimum solution of the problem but the obtained solution will be optimal one. Usually c1 equals to c2 and ranges from [0. The best solution is known as pbest. In PID controller. a group of random particles (solutions) are generated. This best fitness value is a global best and its corresponding particle (solution) is called gbest. To get the optimum performance from the considered system. The PID controllers are widely used in industry because of its clear functionality. if c1 & c2 are not selected properly according to the problem. easy implementation. the gains of the PID controller must be tuned in such a way that the close loop system produces desired result. PARTICLE SWARM OPTIMIZATION Particle swarm optimization is a population based stochastic optimization technique which is introduced by Kennedy and Eberhart in 1995. This computational technique is developed inspired by social behavior of bird flocking or fish schooling.

0027 Tiepower - - - 10.0068 Area1 0.5 and wmax = 0.5 0.54 1 0. It is clearly seen from the results that the proposed PSO based PID controller with the derived cost function improved the control performance by minimizing frequency and tie line power deviation to zero.2 pu MW in area-1. M p = Maximum overshoot and SSE = Steady state error. To meet the design specifications. but a family of solution is obtained which may be useful for design and analysis.5350 0. NIT Durgapur 14 16 18 . The proposed method yields true optimal gains. The simulation is realized in case of step load change. following objective functions are used.078 Area2 0. ∆PL = 0. 4-6.0003 Area1 0.4.4785 1.001 0.5 0.8 0.0924 0. Best value of constriction factors c1 and c2 are taken as c1=c2=1. J1 (e1 M p1) 10 04 ( SSE1) (9) J 2 (e 2 M p 2) 1004 ( SSE 2) (10) Integral Square Error (ISE) = ( ACEi ) 2 dt (11) where e1 and e2 = square integral of ACE1 and ACE2. simultaneous optimization of multiple objectives is carried out.05 -0. The observations show that the proposed control strategy with developed objective function produces good dynamic performances of the considered power system. 3. The results are compared with Zeigler – Nichols tuning PID control method.selection of weighting factors.07 Tiepower - - - 14. For the multi area LFC system. Frequency deviation of area. Wrong selection of weighting factors leads to incompatible numerical value of each term of objective functions which gives erroneous result.1 104 of 446 Copyright @ Department of Electrical Engineering.4806 7.9 software run on a PC of dual core processor with 2 GHz speed and RAM of 2 GB. In multi objective optimization. -0. 0.0924 0.5 0 Tiepower - - - 8. of iterations for optimization are 15. occurring at t = 1 sec and the frequency change in area-1.45.25 12 0.6240 0. The transient response characteristics are depicted in Figs.54 1 0.2 0 4 1 A Proposed PSO-PID S E I Fig.0117 Area2 0. Table 1: Controller parameters and transient specifications of two area power system by PSO-PID method ZieglerNichols PSO-PID [with standard cost function ISE] Kp Ki Kd ts Mp Area1 0. the solution is not a single point.15 -0.6240 0.1 0.1 -0. Specially. RESULT AND DISCUSSION The simulation is carried out by MATLAB 7. area-2 and tie-line power change is observed. In this paper.25 12 0.7496 1. the population size is chosen as 40 and the maximum no.95 and wmin = 0.4806 9 0.05 0 D e lt af 1 (H z ) VI.2 -0.1831 5. Unlike single objective optimization.2 0 Area2 1. minimum settling time and zero overshoot of the transient responses compared to others and establishes its superiority over the others.0342 7.25 0 2 4 6 8 10 12 Time(s) Fig. Simulink model of the LFC system The tuned parameters of the control system and the transient response specifications are shown in table 1. the proposed objective function extract better solution compared to Zeigler – Nichols tuning method and standard cost function [17]. multi objective optimization using PSO algorithm is used to tune the PID control parameters.

issue 9. pp. no. Power system stability and control. “A robust decentralized load frequency controller for interconnected power systems”. pp. Control and Power Engineering . Lili Dong.Ghoshal. A.02 [1] 0 -0.2 [6] 0. Communications. and F. vol.02 [2] -0.0707 pu. G. "Maiden application of ZieglerNichols method to AGC of distributed generation system".04 Delta f2(Hz) -0. T12 = T21 = 0.2. “Robust load frequency control for uncertain nonlinear power systems: A fuzzy logic approach”. Sakharuk.N. vol. It has formulated to optimize a composite set of objective functions. Rerkpreedapong. International Journal of Electrical Power & Energy Systems.145-150. "On robust control analysis and design for load frequency regulation". 2012.425 pu MW/ Hz. Tt1 = Tt2 = 0. Daneshmand. pp. 996-1005. December 2004.14 [5] -0. no. Tp1 = Tp2 = 20s. PSO based PID controller strategy using multi objective optimization has been proposed for load frequency control. 16.1818-1821. December 2006. 3520-3537. 1111-1124. March 2012. “Load frequency stabilization by coordinated control of thyristor controlled phase shifters and superconducting magnetic energy storage for three types of interconnected two-area power systems”. and T.2773-2778. ISA Transactions.2s.14. 72. Power Systems Conference and Exposition. pp. pp. Energy Procedia. Feliachi.1." 2002 IEEE Region 10 Technical Conference on Computers.1-4. P. March 2009. 10691080.08 -0. vol.and Hong Zhou.R. pp. A two area power system with governor dead band has been considered to demonstrate the proposed methodology. and K. 410-419.541-547. 2010 Annual IEEE India Conference (INDICON). “Robust analysis of decentralized load frequency control for multi-area power systems”. Zhu Hengjun. Ranjit Roy. May 2003.449-455.1. Talaq. Hassan Bevrani. vol.R.1 [4] -0. 43. Fatemeh Daneshfar. vol. Mishra. S. Ghoshal. vol.05 -0. 6.2. Frequency deviation of area. The objective functions are uniquely formulated by considering the transient specifications and appropriate selection of weighting factors.A.13. vol.1-7. pp .01 Delta Ptie [8] -0. Yao Zhang. H. Mallesham. Oct. vol. 2010.06 [3] -0.12 -0. Electric Power Systems Research.02 -0. Gautam. and Ehsan Hamedrahmat. Li Pingkang. pp. 32. May 2012. "Fuzzy logic based load frequency control concerning high penetration of wind turbines". pp. 176. 257-263. [12] [13] [14] [15] [16] [17] APPENDIX A [18] The typical values of the system parameters are given bellow.18 0 2 4 6 8 10 12 14 16 18 Time(s) Fig. "Improved particle swarm optimization based load frequency control in a single area power system". McGraw-Hill. Mohammad Hossein Ramezani. and Young Hoon Joo. December 2012. 4 1 A [11] Fig. pp. pp. S. S. 14. Jativa Ibarra. December 2010. Yao Zhang. Control Engineering Practice.2075-2080. and J.01 [7] 0 -0. Praghnesh Bhatt. vol. International Journal of Electrical Power & Energy Systems. 57-65. Hasanovic. Wen Tan. Alireza Yazdizadeh. Jin Bae Park. 2002. and Zhiqiang Gao. Kp1 = Kp2 = 120 Hz/pu MW. and M. issue 3. Ho Jae Lee. "Multi-area adaptive LFC developed for a comprehensive AGC simulator.S.35.. Feb 1999. “Performance verification of PID controller in an interconnected power system using particle swarm optimization”. issue 1. and A. vol.6. pp. Khodabakhshian. Dec. A.4 Hz/pu MW. NIT Durgapur . Stanković. vol. Vadirajacharya. Edrisi. IEEE Transactions on Power Systems . and A.A. "Genetic algorithm optimization for AGC of multi-area power systems. Ashok Mohan Jadhav.5.REFERENCES 0. issue 3.16 -0. IEEE Systems Journal . Shoults. 105 of 446 Copyright @ Department of Electrical Engineering. New York. vol. May 1998. G. “Decentralized load frequency control using a new robust optimal MISO PID controller”. June 2009. this methodology can be applied to other control system problems quite effectively. A. pp. 42. CONCLUSION In this paper. no. Information Sciences. International Journal of Electrical Power & Energy Systems. pp.Tadmor. IEEE Transactions on Power Systems . vol.K. issue 1. issue 1. Due to its simple structure and superiority.173-180. Al-Basri.2. and P. Bevrani. February 2012. vol. PSCE '09. pp.M.04 -0. [19] Tg1 = Tg2 = 0. and Li Yuyun.06 [10] 0 2 4 6 8 10 12 14 16 18 Time(s) S E I Simulation results emphasis that the designed PSO based PID controller gives very good transient and steady state performance for frequency and tie line power deviation compared to Zeigler-Nichols tuned PID controller and PSO based PID controller with standard objective function ISE. November 2012." IEEE Transactions on Power Systems . 203-212.P. issue 23. and Zhiqiang Gao. no. "Load frequency control for multiple-area power systems.855-861. R1 = R2 = 2. Tie line power deviation VII." American Control Conference. “Multiobjective design of load frequency control using genetic algorithms”. B1 = B2 = 0. and N.03 [9] -0. Lili Dong.8. “A new robust PID load frequency controller”. May 1993. R.3s. pp.3. September 2008. "Adaptive fuzzy gain scheduling for load frequency control". pp. D. no. pp. ACC '09. Goyal. A. Jha. issue 10. International Journal of Electrical Power & Energy Systems. 1994.P. "Robust load frequency control using genetic algorithms and linear matrix inequalities". Kundur.18. vol. IEEE Transactions on Power Systems. 51. IEEE/PES. “Optimizations of PID gains by particle swarm optimizations in fuzzy based automatic generation control”. J.

military. The algorithm begins with a high exploration rate with minimum effort and then gradually switches to higher exploitation rate till the promising areas of the search space are identified. Nwankwor et al. The investigated mutation combination for DEs in DPD algorithm is shown to enhance the local search ability of the basic Differential Evolution (DE) and to get a better trade-off between convergence rate and robustness. Differential evolution (DE) [1] and Particle Swarm 4 1 A S E I Optimization (PSO) [2] is an efficient variant of EAs for solving engineering optimization problems. this paper puts forward a new algorithm hybridizing differential evolution (DE) with particle swarm optimization (PSO). etc. Non-redundant search. The experiment results show the algorithm located the global optima in all tested problems. Silchar Assam-788010. 106 of 446 Copyright @ Department of Electrical Engineering. Section 5 contains the results and discussions. In some cases. The combination of PSO and DE seems to be very effective as suggested in the report by Xin et al. India rparouha@gmail.Hybridizing DE with PSO for Constrained Engineering Design Problems Dr. Also the investigation of mutation combination for proposed algorithm has the advantage of higher accuracy and efficiency for constrained optimization problems. and even found a better solution than those previously reported in the literature. according to their increasing order of function value. INTRODUCTION Constrained optimization problems frequently appear in engineering optimization. It can be concluded that the hybridization of DE and PSO came out as a giant optimizer for the optimization problems. The rest of the paper is organized as follows. These methods do not require the derivatives of the objective function and constraints. Section 2 introduces the different components of the proposed hybrid system. Section 3 represents the proposed algorithm. Initially the suitable mutation operators for both DEs used in DPD are investigated. it outperforms other methods in terms of both solution accuracy and computational cost. Mid Group and Superior Group. In view of the above problems. many evolutionary algorithms have been proposed for solving constrained engineering optimization problems. two strategies namely Elitism (to retain the best obtained values so far) and Non-redundant search (to improve the solution quality) have been employed in DPD cycle. Then under the guidance of the bracket operator penalty (exterior penalty). The experimental setting is reported in section 4. The novel hybrid algorithm thus proposed is named as DE-PSO-DE for solving constrained engineering optimization problems.com Abstract—Hybridizing of the optimization algorithms provides a scope to improve the search ability of the resulting method. DE is incorporated to update the previous best positions of PSO particles to force them to jump out of local attractor in order to prevent stagnation of population. [3] used a hybrid PSO called PSO-DE to solve constrained numerical and engineering optimization problems. Recently. I. A novel hybrid DE-PSO-DE (DPD) algorithm is proposed for solving the constrained engineering design optimization problems. NIT Durgapur . the algorithm quickly finds better feasible solution. The performance of DPD is tested on four engineering design problems. Because of the individual shortcomings of each of DE and PSO. Initially all individual in the population are divided into three groups – Inferior Group. the solution leads to a premature convergence or getting stack in some local optima. The Conclusion and future works is drawn at the last section 6. Further. researchers started working on the hybridization techniques between DE and PSO in order to maintain a proper balance between exploration and exploitation in the search space. Especially. In recent years. In this study. The said approach moves around the enhancement of capabilities of PSO and DE in various aspects. However. [6]. there is no single EA algorithm which is able to solve a wide range of constrained optimization problems (COPs) consistently. The PSO-DE integrates PSO with differential evolution (DE) to obtain a good performance. In PSO-DE. DPD combines DE with PSO on the basis of an optimal information sharing mechanism firstly.com Department of Mathematics National Institute of Technology. [5] proposed a hybrid version of PSO and DE for the optimal well placement problem. Hence. The proposed algorithm integrates differential evolution (DE) with particle swarm optimization (PSO) on the basis of an optimal information sharing mechanism. DPD is based on tri-break-up concept of population. Keywords— Differential Evolution. India kedar. Khamsawang et al. Heuristic methods are quite suitable and powerful for obtaining the solution of engineering optimization problems. from the literature. It is usually difficult or even impossible for the traditional methods to solve them because they could not provide the derivative information of the objective function and constraints. Silchar Assam-788010. Elitism. network and management science. economic. the process of hybridization is being emphasized with a tri-breakup of the population. Kedar Nath Das Raghav Prasad Parouha Department of Mathematics National Institute of Technology. Liu et al. Particle Swarm Optimization. [4] proposed an improved hybrid algorithm based on conventional particle swarm optimization and differential evolution (called PSO-DE) for solving an economic dispatch (ED) problem with the generator constraints.iitr@gmail.

r4 . 1] and jrand ∈ (1.. to discourage premature convergence. which is witnessed by the rapid evolution of diverse hybridization techniques in the past decades. D) is the randomly chosen index. “DE/rand-to-best/2/bin” vi xr1 F ( xbest xr2 ) F ( xr3 xr4 ) F ( xr5 xr6 ) “DE/current-to-best/2/bin” vi xi F ( xbest xi ) F ( xr2 xr3 ) F ( xr4 xr5 ) vi xi F ( xbest xi ) F ( xr2 xr3 ) viii. vi is the mutant vector.i.. In this paper. r1 and r2 are random numbers. G . i . uniformly distributed in the interval [0. if ( rand j CR ) or ( j jrand ) X j . i . In the EAs community exterior Penalty function method is preferred to the interior Penalty function approach.g (t) . r3 . Crossover and Selection. G . “DE/rand/2/bin” vi xr1 F ( xr2 xr3 ) F ( xr4 xr5 ) “DE/best/1/bin” vi xbest F ( xr2 xr3 ) “DE/best/2/bin” vi xbest F ( xr2 xr3 ) F ( xr4 xr5 ) “DE/rand-to-best/1/bin” vi xr1 F ( xr2 xr3 ) iii. Infinite barrier penalty. Particle Swarm Optimization (PSO) PSO is a robust stochastic optimization technique based on the movement and intelligence of swarms. exterior penalty functions. are briefly described in this section as follows. During the evolution.G1 V j . r2 . like parabolic penalty. 2. r5 .2. r6 1.. which penalize feasible solutions.. Mutation.. 2. Exterior Penalty function method starts with infeasible or feasible solution. whilst maintaining an emphasis on optimality. Selection:. 1]. . Penalty functions were originally proposed by Courant in the 1940s and later expanded by Carroll and Fiacco & McCormick in the 1960s.G1 Ui . [8]). the approaches for solving constrained optimization problems can be roughly classified into three categories in general. Bracket operator penalty. Each individual in the population is called particle that involves with a position x and velocity v. PSO also starts with an initial population with the population size... The constrained optimization problem transformed into unconstrained problem is of the form: F(x(t) . fitness function of each individual in the population find by using the bracket operator penalty (exterior penalty).G ) X i. 3. vi xr1 F ( xbest xr2 ) F ( xr3 xr4 ) vi. which penalize infeasible solutions. especially for continuous constrained problems. CR is the crossover constant takes values in the range [0. each particle flies to its previous best position pbest and the global best position gbest found so far. f stands for the objective function and Ω stands for penalty term which is given by: = R g (x) where 2 0. a particle‟s velocity and position are updated as follows v v c1r1 ( pBest x) c2 r2 ( gBest x) xi N xiN v iN where ω is called the “inertia weight”. and it was pointed out that the penalty function is the most widely used technique to handle constraints due to its simple principle and easy implementation.It is a central/core operator in DE. COMPONENTS OF HYBRIDIZATION Hybridization has turned out to be an effective and efficient approach to design high-performance optimizers. if 0 “DE/rand/1/bin” ii. Michalewicz [7] summarized several constraint-handling techniques. 1]. if ( rand j CR ) or ( j jrand ) where j = 1. Use of penalty functions has been commonly reported in literature for use in constrained optimization. B. Differential Evolution DE works with three major steps such as Initialization. There are many mutation strategies in DE: A. NIT Durgapur .. Most of the researchers use methods based on penalty functions to handle constraints in evolutionary algorithms (EAs).II. 1. xi is referred to the target vector. repair and tournament selection methods. 107 of 446 Copyright @ Department of Electrical Engineering.The second and main operator of DE is crossover operation defined as: A S E I i. G+1 . i.G1 ) f ( X i . The components participated in framing the hybrid system proposed in this paper. if f (Ui . The main advantage of the use of penalty functions is their simplicity.. G1 . their main shortcoming is that penalty function methods require the fine-tuning of the penalty function parameters.R(t) )= f(x(t) )+ (R(t) . c1 and c2 are positive constants. which is the main drawback of this method. If it starts with infeasible solution then penalty function forces the algorithm to move towards the feasible solution and if it starts with feasible solution the penalty function becomes zero so it does not affect the value of the objective function. Although the use of penalty functions is the most common technique for constraint-handling. This penalty is mostly used in handling inequality constraints (Deb.…. Mutation: . r1 .D. called “acceleration coefficients”. Log penalty. Hence. U j . if 0 = . and Inverse penalty.. iv. x best represents the best individual in the current generation 4 1 i r1 r2 r3 r4 r5 r6 . Np . Constrained Handling Technique An efficient and adequate constraint-handling technique is a key element in the design of optimization algorithm. randj ∈ [0. v. There are various forms of penalties reported in the literature.1 is the mutation coefficient. however. Until now. otherwise C. respectively. They are penalty.The last operator of DE is Selection which is defined as : xi. “DE/current-to-best/1/bin” vii. Crossover: . The reason is that interior function approach requires an initial feasible solution. and interior penalty functions.h(t) ) Where. Two basic types of penalty functions exist. where a random number F0.

The motivation behind the hybridization of DE and is PSO to take advantage for providing better solution. go to step-ii.The purpose of incorporating these two giant techniques in a single one is to provide a better and more robust algorithm for solving constrained optimization problems. The local mutation model is applied in the superior group. DE has the ability to balance exploration and exploitation. Steps of proposed DPD: i. iv. To combat this we can use a feature known as elitism. It diversifies the candidate solutions in a better way 4 1 A S E I and probably also helps in avoiding some shortcomings of DE. to B) to overcome the shortcomings of DE. Due to the robust behavior of mutation operators. when we are solving some complex constrained optimization problems where the ratio of the feasible region and search region is very small [11]. The GDE algorithm performs two mutation operations based on different groupings to effectively search for the optimal solution. PSO and DE on Group A. It is also observed that the behavior of PSO is to wildly seek the potential solution.In [10] author developed a dynamic groupbased differential evolution (GDE) algorithm for global optimization problems. proposed in [10].D. sometimes.e. The idea of this kind of hybridization gets more importance. B and C respectively. However. It initiates with a random population. Population breakup (into 3 parts namely A. a tribreakup-population based mechanism is proposed in this paper and applied in constrained optimization problem (COPs). Employment of DE. DE works better at two different situations where the local search or the global search is essential. to search for potential solutions. ix. Elitism If crossover or mutation performed in an EA algorithm then good candidates may be lost in offspring that are weaker than the parents. ii. III. Hence many a time the diversity in the population need to be maintained. B and C as described above). to A and C). Often the EA will re-discover these lost improvements in a subsequent generation but there is no guarantee. due to the inherent shortcomings of DE. However the key steps are as follows. it is allowed to being employed in both the inferior and superior groups (i. Elitism is therefore is a mechanism to retain the overall best individuals. Of course the population size is kept fixed to a multiple of 3 to favor the tri-breakup mechanism. Addition of new individuals selected randomly instead of these redundant ones. Else go to step x. In GDE algorithm. If the stopping criterion is satisfied. vi. This algorithm works as follows1. As reported in [10]. initially all individuals in the population are grouped into a superior group and an inferior group based on their fitness. Apply NRS. At the end of iteration both the populations (obtained before and after the iterations) are combined and the best half is considered for next generation. viii. Find the fitness function of each individual in the population by using the bracket operator penalty. v. in the hybrid system. 2. The GDE algorithm provides a generalized evolution process based on two mutation operations to enhance search capability. The detailed mechanism (flow diagram) of the proposed DPD is also outlined in Fig 1. Therefore introduction of another mechanism becomes essential in the group based hybridization methods. B and C namely inferior group (first 1/3rd of the population). x. PROPOSED METHOD Inspiration: . respectively. Report the Optimal Solution obtain and Stop. stacking in some local minimal or choosing the path to premature convergence is unavoidable. Hybridization process hopefully helps to get rid of falling in premature convergence and getting trapped in local optima. mid group (middle 1/3rd of the population) and superior group (last 1/3rd of the population). non-redundant search improves the search ability to find the optimal solution. Deletion of individuals with the same chromosome in the current population. both swarm intelligence and evolutionary process. Therefore the synergy of DE-PSO-DE (DPD) is the hybrid method proposed in this paper. The strings are then sorted according to the increasing order of their function value. Non-redundant search (NRS) [9] is used as a local search. which is composed of individuals with lower fitness values. Sorting population according to fitness function value. Observation/Motivation of DPD: . At the same time Particle Swarm Optimization (PSO) is used in the mid-group (i. NIT Durgapur . Apply „Elitism‟ with the population of pre and post generation.e. This kind of hybridization between DE and PSO could provide us a better solution for global optimization problems. over search space. Idea of the hybridization:. Then find the fitness function of each individual in the population by using the bracket operator penalty. 108 of 446 Copyright @ Department of Electrical Engineering. Now the population is being allowed to break in to three different groups A.Differential evolution (DE) and Particle swarm optimization (PSO) scheme is the contemporary. Consequently. Random initialization of population. simultaneously. The global mutation model is applied to the inferior group. which has both exploitation and exploration abilities. According to the local and global searching behavior of DE (as observed above). NRS Algorithm In proposed algorithm. Keeping in view the above observations and inspired by the concept of GDE. to search for better solutions near the current best position. iii. Combination of the resultant sub-populations vii. The two groups perform different mutation operations. E.

.. 5).... 7).. 8 different strategies of most efficient mutation operators have been reconsidered for analysis. to measure the winner combination.e. The list of those strategies is reported in section 3. researchers used many mutation strategies in the DE algorithm. 100 independent runs with 240000 function evaluations are fixed to start the simulation with same parameter setting according to 5.2) include with different number of decision variables and a range of types (linear inequalities. 2.e. (3. (3. (a-c) for 30. this combination i. in the present study. (3. N p i Sr = Number of successful runs of ith problem Tr i = Total number of runs of ith problem Moi = Mean optimal objective function value obtained by an algorithm of ith problem Fig. the role of mutation operator is as a kernel operator that has the ability in the search of potential solutions. 50. if Sr i > 0 i = Af 3 0. Some experiments are carried out so as to identify the top four combinations are (3. (3. 2(c). Hence. 3) combination are better than other three shown in Fig.. 5).. Constrained Engineering Design Benchmark Problems and Parameter Settings A set of four well-known constrained engineering design problems (reported in [14]) are carried out to evaluate the performance of DPD: (1) E01: Welded Beam design problem 109 of 446 Copyright @ Department of Electrical Engineering. On the basis of review of the past literature. a Performance and its extended form is used and is defined as N below: 1 i i i k a +k a +k a p Performance = subject to 1 Np 1 2 1 2 3 1 and where where 3 3 k 1 k2 k3 Sr i Tr i Moi . 3). nonlinear equalities.2.. Selection of Mutation Operators for DEs in DPD In DE Algorithm. Performance evaluation for top 4 combination (for 100 run) Aoi = Minimum of mean optimal objective function value obtained by all the algorithms of ith problem Mf i = Average number of function evaluations of successful runs required by an algorithm in obtaining the solution of ith problem B. Hence. Performance evaluation for top 4 combination (for 30 run) 4 1 A S E i=1 k +k +k 2 Out of top 4 mutation combinations {(3. Therefore. if Sr i 0 i for i = 1.. Hence.. While dealing with DPD. and nonlinear inequalities) and number of constraints. 50. PEXPERIMENTAL SETTING Af i = Minimum of average number of function evaluations of successful runs required by all algorithms in obtaining the solution of ith problem N p = Total number of problems A. For this 30.. Finally out of top four combinations check the performance which is defined in [12].. Top 4 combinations (i. (3. 2(b). if Sr i > 0 i i = Ao 2 0. 1). 100 run respectively. Performance evaluation for top 4 combination (for 50 run) I Mf i . 3) is considered for further study. 1)} the performance of (3. (3..IV. a total of 8X8 = 64 combinations with all possible permutation of mutation operators. 3).. (3. 7). all forms of DPD) are tested in 4 constrained engineering problems (reported in 4.. all combinations of mutation strategies are considered as: DE (with 8 different strategies of mutation) + PSO + DE (with 8 different strategies of mutation) = DPD Therefore.2. are generated by considering one case at a time and keeping PSO fixed. For DE. 2(a). if Sr i 0 i 1 = Fig. NIT Durgapur . selecting a mutation operator is indeed a difficult task.2. DE is being used two times in a generation. Fig.

471719 9. the robustness. NIT Durgapur .0 and implemented on CORE i3. The results provided by compared approaches were directly taken from SC-ABC [14].198353 PRESSURE Best Mean STD Best Mean STD Best Mean STD Best Mean STD SC-ABC 3.00000 7.1E-4 2996.9.800000 3.808660 0. multiple of 3.012665 0. As the problems studied in this work are minimization ones.350215 5. the statistical results provided by DPD were compared with some algorithms used in literature. the penalty function just adds a very high value (R=1e10) to the objective function. TABLE I.94E−18 5886.012667 0. However. DPD is able to consistently find the best value with a very small standard deviation which indicates that the proposed DPD has a remarkable ability to solve constrained problems.198353 5946. DPD uses an information exchange mechanism that helps to avoid the 110 of 446 Copyright @ Department of Electrical Engineering. the acceleration constants C1 and C2 are both set to 2.538862 0. and the crossover weight CRA=CRC=0.0000 1. Also.350272E-15 2993.2514 6059.0131 4.348165 2996. on the comparison above.700000 17. and the efficiency of search.640750 6059. 60 independent runs are performed and statistical results which give information about the robustness of the algorithm.098187 176.TABLE III. F A=0. which integrates PSO with DE.012665 STATISTICAL RESULTS FOR ENGINEERING DESIGN PROBLEMS BY DIFFERENT METHOTDS Problems E01 SC-ABC 0.0127 2.01302E-08 0. it kept fixed to 42 for all problems to favor the population-tri-breakup system.8 GHz machine with 2GB of RAM. (2) E02: Pressure Vessel design problem (3) E03: Tension/Compression Spring design problem Methods x1(d) x2(D) x3(P) f(x) (4) E04: Speed reducer design problem These problems have been previously solved using a variety of other techniques. TABLE IV.343364 5.899497 180.399960 5886.700000 17 7.3482 0. DPD shows a very competitive performance with other algorithms in terms of the quality. As a conclusion.205836 1.695255 BEST SOLUTION OBTAINED VESSELDEASIGN PROBLEM Methods x1(Ts) x2(Th) x3(R) x4(L) f(x) SC-ABC 0. This penalty value was empirically chosen to be considerably bigger than the objective function values generated by the tested problems.012665 4.0000 1. all problems deliver best value with low standard deviations.0.0498 12. the exploration and exploitation abilities of DPD algorithm are balanced. These approaches are: Sic-PSO [13] and SC-ABC [14]. V.253253 9.72458 2. Moreover.500000 0.1725 0.758872 Sic-PSO Methods SC-ABC DPD 1.0574 0.758872 2993.051515 0. Out of four engineering design problems DPD delivered best result compared to other algorithms in all manners.300000 7.348165 2996.2097 0. C.4E-7 2996. 2.800000 3. The population size for DPD is set to a value.399721 41.361108 11.714335 6092. Methods x1 (h) x2(l) x3(t) x4(b) f(x) TABLE II. In Table 2.042758 0. Only for E03 it performs marginally better but mean and std are very low compared to others.348165 DPD 3.352529 11.e.012667 BEST SOLUTION OBTAINED FOR SPEED REDUCER DEASIGN PROBLEM Methods x1 x2 x3 x4 x5 x6 x7 f(x) S E BEST SOLUTION OBTAINED FOR WELDED BEAM DEASIGN PROBLEM BEST SOLUTION OBTAINED FOR TENSION/ COMPRESSION SPRING DEASIGN PROBLEM In order to show the effectiveness and superiority of DPD. This illustrated that the results generated by DPD is robust.812500 0. TABLE V.205724 3.8. and standard deviations over 60 independent runs.5 and for group 3 i. it is compared with other classical algorithms designed for constrained engineering design problems.72665 1.051871 0.286683 2996.768058 FOR DPD 0. It includes the best. Values in “bold face” in tables represent the best obtained in the listed comparisons.e.726625 DPD 0. which is useful to show the validity and effectiveness of the proposed algorithm.00000E+00 4 1 A E04 DPD 0. it shows that DPD can generate best solutions for most of the problems for all 60 runs. Values in “bold face” in tables represent the best obtained in the listed comparisons.036644 0. For PSO.3482 0.012665 0. FC=0. For each problem.036860 0.768058 6060.205563 3. mean. By using global information obtained from DE and PSO. Stimulation Results and Discussion for five engineering design problems The best results and statistical results of the DPD on the 4 engineering design problems are summarized in Table I-IV and Table V respectively.300000 7.205729 1.285351 2993. However for DE in DPD. CONCLUSION AND FUTURE WORKS This paper proposes a novel algorithm named DPD.7298.500000 0. the inertial weights (w) are set to 0. E02 E03 I SC-ABC 0.790134 3. the mutation factor in group 1 i. The DPD are coded in C-Free Standard 4.695255 2.695255 1. Environment.0069 0. Almost all constrained engineering problems values obtained by DPD are better. Maximum numbers of function evaluation DPD are set as 240000.758872 0.29E-4 6059.7413 2.437500 42.

Deep and K. pp. Y. T. K. MIT Press. Vol.” Journal of Global Optimization. Coello. 4 1 A S E I 111 of 446 Copyright @ Department of Electrical Engineering. Chang and C.” Applied Intelligence. Brajevic. 1995. “Hybridizing differential evolution and particle swarm optimization to design powerful optimizers: a review and tax-onomy. 5. and K. DPD algorithm is used for only solving single-objective constrained optimization problems. X. S. pp. pp. it is simple.” Informatica.” 2nd International conference on autonomous robots and agents. 2004. B. Therefore. T. 2010.N. 135–155. Runarsson. 319-326.” Applied Soft Computing. 1. Subotic. Zhang.” In Proc. Khamsawang. no. vol. Ishikawa. Xin. Liang. “Differential evolution – a simple and efficient heuristic for global optimization over continuous spaces. In this paper. pp. J. Peng. pp. F. no. Conf. In general. pp. 11. 1995.” In Proceedings of the IEEE International Conference on Computer and Automation Engineering. Z. H. pp. Das. 42. M. Eberhart. 1997. Kennedy and R. P. it is very effective for solving constrained engineering optimization problems. pp. M. An investigated mutation strategy for DEs in DPD. Nagar and D. vol. 2010.” IEEE Transactions on Systems Man and Cybernetics Part C: Application sand Reviews. K. Lin. Liao. J. 39. 5. no.” Computational Geosciences. Deb. vol. no. Nwankwor. 341-359. 629–640. Price. C. different fashion of hybridization of DE and PSO algorithms may be investigated. “Particle swarm optimization. no.[5] premature convergence. NIT Durgapur . 744–767. 32. Liu. The two strategies Elitism and NRS illustrate the attractiveness of the proposed method. 41. Jiriwibhakorn. 1995. 4. 17. Wannakarn. Deb. and able to find the optimum solution in almost every run. Michalewicz. it is applied to solve four engineering design problems. vol. 1942-1948. 244–248. Also. H. Chen. 2. 135–139. Reid. 4. [6] [7] [8] [9] [10] [11] REFERENCES [1] [2] [3] [4] R. I. 4. vol. J. DPD may be used for solving multi-objective-real-world problems. H.135-143. 2. “Hybrid PSO-DE for solving the economic dispatch problem with generator constraints. A. Cai. 2013. M. J. New Delhi. “Performance of the improved artificial bee colony algorithm on standard engineering constrained problems. Neural Networks. and Y. 5. “An extended hybrid genetic algorithm for exploring a large search space. 1–20. no. and S. pp 4156. 10. Suganthan. 2. J. Further.” International Journal of Mathematics and Computers in Simulation. Intelligent Defence Support Systems. “Solving Engineering Optimization Problems with the Simple Constrained Particle Swarm Optimizer. P. [12] [13] [14] E. Tuba and M. Fang and Z. Cagnina and S. Esquive. L. no. E.” In Proceedings of the 4th Annual Conference on Evolutionary Programming. tri-beak-up technology for the population really makes the DE-PSO-DE faster and robust. “Dynamic groupbased differential evolution using a self-adaptive strategy for global optimization problems. J. “Performance improvement of real coded genetic algorithm with Quadratic Approximation based hybridization. enhance the local search ability and advance the convergence rate. “Hybridizing particle swarm optimization with differential evolution for constrained numerical and engineering optimization. H. pp. 2008. S. 2006. vol. Z. converges fast.” Journal of Applied Mechanics. vol. “Problem definitions and evaluation criteria for the cec 2006 special session on constrained real-parameter optimization. Clerc. In order to demonstrate the effectiveness of the proposed method. Cambridge. Wang.” Int. “Optimization for Engineering Design: Algorithms and Examples. Mezura-Montes. Storn and K.” Prentice-Hall of India. 2012. C. “A survey of constraint handling techniques in evolutionary computation methods. pp. Han. C. Simulation results and comparisons show that the proposed DPD has higher accuracy and better robustness. 2011. 319-334. IEEE Int. robust. vol. As a future research. 2. pp. Zhang and M. vol. 2009. “Hybrid differential evolution and particle swarm optimization for optimal well placement. 2013. no.

2 M. multilevel inverters are extensively applied for high voltage and high-power applications in many more industries has transform to greater extent democratic and high performance superior to that of formal full bridge 2-level & 3-level inverters. INTRODUCTION Life style changes in. NIT Durgapur .Tech. cleanness and reliability should be very high. there is a need to independent on fossil fuels as well as reduce the effect of greenhouse gases and necessitates the increase of electricity generated by renewable energy systems. Cascaded H-Bridge (CHB) based multilevel inverters have been speculating as an alternate in high voltage & medium voltage converter trade.com 1 Abstract— Renewable Energy Systems (RES) modeling ride must communicate variance of socio-economic sensitivities and ecological in organize to efficient budget related decisions and guide policy. 1 gallasatya. Application of power electronics in that. Andhra Pradesh. 7-level. 2Department of Electrical & Electronics Engineering 1 University of Hyderabad. To secure the microgrid performance.1 Structure of Symmetrical Series/Parallel 5-Level Inverter Renewable energy sources (RESs) have versed an expeditious growth in last decades due to the huge technological improvements. economic insights will promote this needful transition in way of sustainable energy systems. 2Vishnu Institute of Technology. A competent asymmetrical based series/parallel multilevel technique is presented with low switches compared to CHB & ACHB. which comprise photovoltaic. Hyderabad. India. which are environment friendly and sustainable nature. 2 klganesh201@gmail. SA Sc vbu S1 s ibus Vdc SB Sc Fig. has been increased the efficiency at similar time and progressively reduced their costs [1]. 2Vdc S2 Generalized Asymmetrical Series/Parallel Inverter. Practically. which induce the possibility of eliminating the load side output filter using Matlab/Simulink Platform. fuel cell.Tech. Vdc S2 SA Sc vbus S1 Vdc S3 ibus Sc SB Fig.1 represents the structure of symmetrical series/parallel 5 level multilevel inverter using 6 active semiconductors utilized in cogeneration systems. most of the distributed generation system (DGs) consists of power grid and further it should be fabricate to microgrid with local loads. Total Harmonic Distortion (THD). Bhimavaram.eee@gmail. Assistant Professor School of Management Studies.Harmonic Distortion Optimization of Generalized A-Symmetrical Series/Parallel Multilevel Converter with Fewer Switches G. and preparedness. influential source is fuel cells (FCs) is rising as a favorable complementary power due to these advantages such as high efficiency. a 5-level.2 Structure of A-Symmetrical Series/Parallel 7-Level Inverter Today. I. A Simulation study & comparative analysis has been carried out to evaluate high performance of the proposed asymmetrical series/parallel configuration to accost a very low total harmonic distortion (THD) of output voltage. photovoltaic (PV) energy performs quite fascinate for generation of electric power because of low maintenance.com. The main theme of this work is to originate a foremost decision support creature for distributed hybrid renewable energy system nothing as co-generation systems. the asymmetrical Cascaded H-Bridge (ACHB) inverter. Moreover. it can be settled majorly by utilizing to photovoltaic and wind energy generation. such as fuel cells. Hence. Lakshmi Ganesh2 1 M. Renewable Energy Sources (RES). pollution free. 11-level and 21 level converters can be succeeding with a low number of active semiconductors. 4 1 S E A I Keywords— Asymmetrical Cascaded H-Bridge (ACHB) Inverter. This paper interpolate a prenominal regular methodology for reliability and modeling assessment of miniature co-generating systems. K. behaves the high voltage levels by expend the dc supplies with unequal voltages. technical. PGDBM. some mismatch power flow in between the load and source (FC) compulsory achieved by using energy storage 112 of 446 Copyright @ Department of Electrical Engineering. DG is normally necessity to work in grid connected and stand-alone connected systems. Many more research things are going on co-generation of renewable energy sources (RESs). Hence. and battery to integrate grid by exploiting power electronics technology. The startup period is very long and dynamic response also very slow. Other. inadequate points of fuel cells. Fig. Their divergent. Satyanarayana1. Societal. noiseless and more flexibility and should be constituted by another interchange energy sources to ascertain a reliable & constant power supply. PV arrays and wind energy generation. primarily regarding consumption and production it will ultimately be forced on many things by economic and ecological pressures.

are key role in these systems [10]–[14]. adjustable speed drives. neutral point clamped inverters (NPC) and cascade multilevel inverters [3] [4]. there is a need of that generated power essentially renovate to ac power and is integrate to the grid by using grid-connected (VSI or CSI) inverters. semiconductor drive control. output filters also eliminated. IEEE-1547 and EN-61000-3-2. NIT Durgapur . Hence. converters and inverters. high voltage capability. The virtually standard advantages of this proposed multilevel inverter resemble with formal voltage source inverter (VSI) are more concern in high efficient operation. hence reducing the overall cost.2 represents the structure of A-symmetrical series/parallel 7 level multilevel inverter using 7 active semiconductors. Although this type of approaches is well convenient for excellent multilevel inverter. Nevertheless. a theoretical analysis is used to conceive an asymmetrical based series/parallel conversion strategy compatible with several multilevel configurations are implemented and compared [9]. Integrating such type of energy sources innovate a FC/PV/Battery co-generation based hybrid generation system. static Var compensators and renewable energy sources. the CHB multilevel inverter has some disadvantages like large number of active switches. A charge pump attains a larger voltage at output side than the input side voltage with respect to proposed switching pattern. the co-generation system have high potential to provide high quality energy. the performance evaluation of power electronic converters are critical. power systems control.e. better electromagnetic interference. high power converters have been linearly developing in many fields such as power quality. harmonic currents and notch free.3 Structure of A-symmetrical Series/Parallel 21-Level Inverter When the various switches are connected in series and parallel configuration. such as IEEE-929. By create so. The grid connected DG systems need an inverter to convert dc to ac. presenting good performance and very low THD values. controlling the startup transitions as well as dynamic characteristics. diverse international standards. Nevertheless. for increases the number of levels. However. Asymmetrical CHB (ACHB) inverter employ dc supplies with several voltages [5]-[8]. and co-generation. communicate the requirement on inverter output must be maintained quality. with include of the flying capacitor inverters . a charge pump is presented [15]. attaining the dc power from various energy systems. equipment testing. the proposed asymmetrical series/parallel 21 level multilevel inverter with 113 of 446 Copyright @ Department of Electrical Engineering. high reliable and high efficient power. and it can get maximum number of levels if the dc supply voltages are scaled out with different values and 4Vdc - vbus S4 + - S6 PV Source S5 (Vdc) SD SB Fig. the THD is prominently reduced consequently. a transformer or an inductor in the step up converter makes the very large system. In spite of that. output voltage and current with outstanding quality can be acquired with small circuit and control complexity also. comparisons with respect to single-sourced system. i. when the input voltage is lower than that of output voltage. their prolongation to a increased number of levels are not a easy thing. lower switching losses. In co-generation application. There are many different topologies. During this proposed concept. as well as isolated dc supplies in balanced condition. Respective workings have partly declare and propagate these prospection to alleviate the and control scientific design of such topologies favored in huge power applications like industrial applications. with addition of one active component we get 7-level output voltage. PROPOSED ASYMMETRICAL SERIES/PARALLEL MULTILEVEL TOPOLOGY The high range power electronics. by getting reduced filter size with the support of increased harmonic characteristic of the output voltage. Boost converters or transformers are extensively used in these systems. and 3) simpler operation & structure 4) low filter range 5) commercial advantageous. Many of the co-generation systems are interfaced in between the RES and grid.element. because they must have large and bulky magnetic cores to endure the high power. 4 1 S E A + S7 I S3 FC (3Vdc) 2Vdc S1 SA Sc S8 + S9 ibus - TO GRID Va S2 + As Fig. A Series/Parallel inverter is similar to a charge pump in this topology with the less number of semiconductor switches we get high voltage levels based on asymmetrical configuration. and uninterruptible power supplies (UPS). charge pumps have many switching devices for that system should be more complicated [16]-[17]. Power electronic conversion applications to medium voltage. such topologies has introduced in last four decades. and no importance of transformer. II. This paper recommend to a modern solution for eliminating the main drawbacks and to keep the advantages of asymmetrical series/parallel inverter are as follows 1) automatic voltage balance among H-bridges 2) very low THD at all operation ranges. and decorative the peak power capacity [2]. the charge pump outputs the addition of several voltage of the input side values are to be get at high output side value with low harmonic content.. Most of the conditions batteries are normally preferred as storage system for enhancing output power. transformers. total harmonics distortion (THD) of the output voltage and current. As an enactment against the problem. Matlab/Simulation results obtained for an asymmetrical series/parallel inverter fed grid connected system approve the eminent dynamic performance of the proposed method. high power quality.

f. S7. further switches are inactive. SD. further switches are inactive. S7. SA. Mode 2: (Vdc Level) This level can be acquired by five switching actions S1. SB are active. SA. SB are active. further switches are inactive. associated the load positive voltage(Vbus). S6. Mode 8: (7Vdc Level) This level can be acquired by five switching actions S4. further switches are inactive. The proposed inverter doesn’t present any inductors which make the system to be very large. further switches are inactive. the voltage across the load terminal is 6Vdc. h. 5(A)–(K). associated the load positive voltage (Vbus). S6. Fig. Mode 7: (6Vdc Level) This level can be acquired by five switching actions S2. Mode 4: (3Vdc Level) This level can be acquired by five switching actions S4. S6. S5. associated the load positive voltage(Vbus). the voltage across the load terminal is 4Vdc. S3. SA. S8.4 shows the 21-level output voltage of asymmetrical series/parallel multilevel inverter with hybrid sources and with the need of different operational switching modes The required twenty levels of output voltage with low THD value supported to grid/load as follows. the voltage across the load terminal is 3Vdc. S8. k. SB switches & for negative half cycle SC. 10Vin 9Vin 8Vin 7Vin 6Vin 5Vin 4Vin 3Vin 2Vin vin b.3. 114 of 446 Copyright @ Department of Electrical Engineering. SA. SB are active. Mode 9: (8Vdc Level) This level can be acquired by five switching actions S4. Mode 3: (2Vdc Level) This level can be acquired by five switching actions S5. SB are active. SA. S8. c. SB are active. associated the load positive voltage (Vbus). S5. the voltage across the load terminal is Vdc. and the voltage across the load terminal is zero. SD) and mainly eleven switching states (positive and zero levels). associated the load positive voltage (Vbus). Mode 10: (9Vdc Level) This level can be acquired by five switching actions S2. further switches are inactive. SB are active. Mode 5: (4Vdc Level) This level can be acquired by five switching actions S4. the output harmonics of the proposed inverter are reduced by the multilevel output [18]. further switches are inactive. SB are active. S7. S5. the voltage across the load terminal is 8Vdc. As Fig. the voltage across the load terminal is 9Vdc. the voltage across the load terminal is 7Vdc. load terminals should be short circuit.5 Operating Modes of +ve Half Cycle & -ve Half Cycle a. SB. the voltage across the load terminal is 2Vdc. SA. S8. SA. SB are active. S7. S3. 4 1 g. SB are active. SA. S7. Fig. SB are active. S5. associated the load positive voltage(Vbus). associated the load positive voltage (Vbus). SA. SC. Mode 11: (10Vdc Level) This level can be acquired by five switching actions S2. further switches are inactive. associated the load positive voltage (Vbus). Mode 6: (5Vdc Level) This level can be acquired by five switching actions S2. as shown in Fig. Vbus Time d. 5 shows the path for current flow and active at this mode and for getting positive half cycle turn on SA. S5. associated the load positive voltage (Vbus).different sources such as pv source. the voltage across the load terminal is 10Vdc. S7. fuel cell and battery energy storage system as shown in Fig. Modes of Operation: S7 S8 SA 3Vdc S2 2Vdc Sc I S9 S3 S1 S E 4Vdc ibus vbus S4 S6 Vdc S5 SD SB Fig. SA. SD are active. further switches are inactive. further switches are inactive. further switches are inactive. i. NIT Durgapur . associated the load positive voltage (Vbus).4 Output Voltage of Proposed 21-Level Asymmetrical Series/Parallel Multilevel Inverter The proposed asymmetrical multilevel inverter’s operation majorly can be divided into two mode of operation such as positive mode & negative mode by turn on the switches (SA. the voltage across the load terminal is 5Vdc A A. j. -Vin -2Vin -3Vin -4Vin -5Vin -6Vin -7Vin -8Vin -9Vin -10Vin e. Mode 1:(0 Level) This level can be acquired by two switching actions such as SB.

7 shows the 5-Level output voltage of proposed symmetrical series/parallel multilevel inverter topology. here get 16. Fig.11 shows the grid voltage of proposed A-symmetrical 7-Level series/parallel multilevel inverter topology.9 FFT Analysis of Output Voltage of Proposed 5-Level Symmetrical Series/Parallel Multilevel Inverter Topology.10 shows the 7-Level output voltage of proposed Asymmetrical series/parallel multilevel inverter topology with the addition of single switch and different voltage sources.12 shows the FFT analysis of output voltage of proposed 7-Level A-symmetrical series/parallel multilevel inverter topology.9 shows the FFT analysis of output voltage of proposed 5-Level symmetrical series/parallel multilevel inverter topology. Fig. 11 Grid Voltage Fig. Fig.6 shows the Matlab/Simulink model of proposed symmetrical series/parallel multilevel inverter topology using Matlab/Simulink software package. 4 1 A S E Fig.12 FFT Analysis of Output Voltage of Proposed 7-Level A-Symmetrical Series/Parallel Multilevel Inverter Topology. 10 7-Level Output Voltage Fig. PERFORMANCE E VALUATION OF MATLAB/SIMULINK MODELING & RESULTS The proposed system as shown below has been modeled & simulated using Matlab/Simulink software package to achieve the newly proposed asymmetrical series/parallel multilevel inverter topologies interfaced to grid system. I Fig.8 shows the grid voltage of proposed symmetrical 5Level series/parallel multilevel inverter topology. The circuit consists. Fig.82%.6 Matlab/Simulink Model of Proposed Symmetrical Series/Parallel Multilevel Inverter Topology Fig. Fig. NIT Durgapur .48%. here get 20. Fig. 7 5-Level Output Voltage Fig. 115 of 446 Copyright @ Department of Electrical Engineering. group of power electronic switches rendered by different sources such as co-generation systems and form series/parallel configuration. Fig. 8 Grid Voltage Fig.III.

18 FFT Analysis of Output Voltage of Proposed 21-Level A-Symmetrical Series/Parallel Multilevel Inverter Topology.A 20 Flying Capacitor 40 N.15 FFT Analysis of Output Voltage of Proposed 11-Level A-Symmetrical Series/Parallel Multilevel Inverter Topology.Fig. Switching Devices Clamping Diodes Balancing Capacitor DC Bus Diode Clamped 40 380 N. 4 1 Table I Comparison of Various Multilevel Topologies with Respect to Proposed 21-Level Asymmetrical Series/Parallel Topology A Item S E Fig. in proposed concept no need of any clamping diodes. clamping diodes. 14 Grid Voltage Fig. As Table I shows the number of switches. mainly reduction of switching devices compared to asymmetrical cascaded HBridge topology and preferred equal number of DC sources.17 shows the grid voltage of proposed A-symmetrical series/parallel multilevel inverter topology. Fig.16 shows the 21-Level output voltage of proposed Asymmetrical series/parallel multilevel inverter topology with the addition of six switches and different voltage sources. Fig.A N. here we get 11.62%. Fig.86%.13 shows the 11-Level output voltage of proposed Asymmetrical series/parallel multilevel inverter topology with the addition of three switches and different voltage sources.A 190 20 Cascaded H-Bridge (Symmetrical) 40 N. here get 13. balancing capacitors to achieve high quality of output voltage with low harmonic distortions. 17 Grid Voltage Fig. I Fig. dc bus required for various multilevel configuration in comparison of proposed 21 level series/parallel multilevel configuration.15 shows the FFT analysis of output voltage of proposed 11-Level A-symmetrical series/parallel multilevel inverter topology.18 shows the FFT analysis of output voltage of proposed 21-Level A-symmetrical series/parallel multilevel inverter topology. 13 11-Level Output Voltage Fig.A 4 13 N. Fig.A N. NIT Durgapur .A 10 Cascaded H-Bridge (Asymmetrical) 16 N.A N.14 shows the grid voltage of proposed A-symmetrical 11-Level series/parallel multilevel inverter topology. 16 21-Level Output Voltage Fig. for 31 Level output voltage need 116 of 446 Copyright @ Department of Electrical Engineering.A 4 Configuration Proposed Series/Parallel Fig. capacitor.

hybrid electric. ―Switching characterization of cascaded multilevel-inverter-controlled systems. Zhang. G. G. Lu. 1607–1617. I CONCLUSION The perception of high power multilevel converters has been more acceptable because of low switching losses. vol. Reg.. Mag.. 7 pp... Jan. D. Sustainable Alternative Energy (SAE).. Valencia.‖ in Proc.. 4. Beuning. USA. enhance electromagnetic compatibility. S. Papers. pp. and economically it is more perfect for industrial applications. 2010.‖ in Proc. pp. Prasad. 56. McGrath and D. pp. 49.. K. and C. I. [5] Y. Conf. Rodriguez. Familiant. A. Jul. S. 21. I. no.. vol. Fig. Dec. Symp. Leeb. [12] X.additional three switches and another DC sources with the help of proposed series/parallel asymmetrical multilevel inverter. Anil Kumar. no. 19 Performance of Proposed Multilevel Inverter Based on Switch Count with Respect to Various Formal Multilevel Topologies. Electron. no. J. Electron.‖ IEEE Trans. P. MN. 3.48% 16. vol. and A. [8] J. 2643–2650.V. [14] B. 19 shows the 2D line diagram of performance comparison of proposed multilevel inverter with respect to various formal multilevel topologies based on switch count. NIT Durgapur . pp. [9] K.. T. and M. no. 2004 [13] S. Prats. Sádaba. "Comparison of different topologies of cascaded H-Bridge multilevel inverter. IEEE Int.‖ IEEE Trans. Franquelo. 2013 International Conference on . ―A cascade boost switched capacitor. Electron. Qiu. ―A single phase multilevel inverter using switched series/parallel DC voltage sources. 2013 International Conference on . vol. Minneapolis..‖ IEEE Trans.‖ IEEE Trans. Vamsee Kiran. Ding. 2010. R. [17] V. Koizumi. "Realization of cascaded H-bridge 5-Level multilevel inverter as Dynamic Voltage Restorer. Circuits Syst. 2008. Starzyk.‖ IEEE Trans. G. Aug. The most important attractive multilevel converter topologies are cascaded H-bridge multilevel configuration. E. Kou.‖ in Proc. Theory Appl. no. Mar. G. [18] A. vol. May 2006. IEEE PES/IAS Conf. and P." Computer Communication and Informatics (ICCCI). 21. Ioinovici. vol.82% 13. no. 858–867. Corzine. A. 117 of 446 Copyright @ Department of Electrical Engineering. 3. no. Durrwachter. ―A switched-capacitor inverter using series/ parallel conversion. no. vol. pp.62% 11. Berkovich. [10] B. In this paper. Holmes. ―Power electronics intensive solutions for advanced electric. [6] S. Oct. Agelidis.1. 2005 [11] J.. Table II FFT Analysis of Output Voltage with Proposed System THD (%) Output Voltage (Vo) 5-Level Topology 7-Level Topology 11-Level Topology 21-Level Topology 20.‖ IEEE Ind.‖ IEEE Trans. pp. 2010. Sotkiewicz. pp. and higher voltage capability. [15] R. 1506–1515. Hawkins. 2.Ranjith Kumar.N. ―A multilevel inverter topology for inductively coupled power transfer. 2002. Porter. pp. 2001. Gokdere.V Prasad. pp.converter-two level inverter with an optimized multilevel output waveform. Satyanarayana.1. 3. Circuits Syst. Schuyler. 2007. and M. Joshi." Computer Communication and Informatics (ICCCI). pp. ―On attaining the multiple solutions of selective harmonic elimination PWM three-level waveforms through function minimization. 3. ―Integrated magnetics for interleaved DC–DC boost converter for fuel cell powered vehicles.. here we get better voltage quality with respect to increased number of levels and THD also drastically reduces. Chandrasekaran and L. Axelrod. IEEE Power Electron. no. 567–577. Khaligh. Y. Electron. 2763–2770.. Jun. Goyena and O.Satyanarayana. Emadi. 12 pp [3] K. A. A. 57. 55. 28–30. Power Electron. pp. Balouktsis. 19. attempt to lower THD and high quality of voltage waveform. Hinago and H. 4-6 Jan. S. G.. ―Sizing and analysis of big scale and isolated electric systems based on renewable sources with energy storage. 3188–3191. Mar. 25–29. I. I.. Power Electron. K. Spec. ―The age of multilevel converters arrives. 4. H. J. Zhang. Ind. Lowell. REFERENCES Fig. [4] L. ―Study on a modified multilevel cascade inverter with hybrid modulation. 356–361. Ind. vol.. and F. no. pp. 2009. 8. 12. Technol. Leon. [2] J. and A. B. 2001.6.6. Electron. no.‖ IEEE Trans. 2013. J. and A. and Y. 48. ―A unique fault-tolerant design for flying capacitor multilevel inverter. Smith. 379–383.Ranjith Kumar. 979–987. Rodriguez and S. no. vol. mostly applied to interfacing in-between RES to grid connected system. B. Circuits Syst. ―A dc–dc charge pump design based on voltage doublers. Portillo. Corzine. 4. 2008. May/Jun. vol. Mar. 2008 [16] J. Jun. L. Ela. Kouro. 52. 2006. I. high power quality. This configuration is very appealing due to its simple structure and modularity of control. A. Koizumi. Cossar.. and K. 1047–1058. Power Electron. vol. IEEE Power and Energy Society General Meeting. as low frequency preferred to high and medium power range applications. K. Veh. Ind. 2004. Zou. [7] Y. A. Jul. ―A unique ultra capacitor direct integration scheme in multilevel motor drives for large vehicle propulsion. Y. U. G. Nov. vol. and fuel cell vehicular power systems. C. 2013. pp. Ghosh. Hinago and H. pp.‖ IEEE Trans. [1] S. no. X. 2. Sep. ―Multicarrier PWM strategies for multilevel inverters. pp. switching devices contrive for high-voltage as well 4 1 A S E In Table II show the FFT analysis of output voltage for proposed asymmetrical series/parallel multilevel inverters. proposed converter requires less number of switches for getting same levels. a generalised asymmetrical multilevel converter with series/parallel configuration which degrades the switching devices by controlling the switching condition with respect to dc voltage sources in comparison of formal multilevel converter topologies for corresponding number of voltage levels at output side and also reduces the THD with increased number of levels. S. ―Impact of variable renewable energy on US electricity markets. 4-6 Jan. A. Ferdowsi. Spain. Ind. The performance of the proposed multilevel converter topology was examined in detail. 6. Jul.‖ IEEE Trans. Aug. 28–39. vol.S. Gupta. G. M. 55. Lasher. Y.. Fundam. IEEE Power Electron. pp. Drive Syst..N. W.‖ IEEE Trans. Williamson.‖ in Proc. Y. 996–1004. Kirby.86% IV. 350–359. G.‖ in Proc. K.

Subsequently. guiding. The structural bracket supports for the three tier insulated piping system at an elevation ranging from 5. mechanical stresses and vibrations.0 to 8. or any other operating characteristics that affect the forces at play — whether inside the pipeline or in the environment surrounding the system [1]. as necessary.in Abstract—Pipe supports are structures that suitably support pipes and ancillaries between process areas of industrial plants. Furthermore. or anchoring) [1]. the importance of the pipe supports is frequently overlooked. which through proper inspection and maintenance could have been avoided. Movement results from changes in temperature. Overall. critical to every piping system. and ancillaries between process areas of industrial plants. load. Haldia Refinery. a detailed study of the existing piping system along with modified pipe support system (both structural bracket and reinforced concrete framed structure options were analyzed). recommendations were provided for both short term and long term restoration measures. or movement. Engineering Services Department. vibrations or displacements could result in time-consuming and costly retrofitting or repair in addition to any labour and materials that might have been wasted. India. is that they accommodate pipe displacement. as per site condition was carried out using computer modeling. Making sure a system can meet functional and operational requirements efficiently requires that designers thoroughly account for each component's design from the conditions during installation (cold) to those during system operation (hot). as necessary. leading to higher levels of stress. shock loading). NIT Durgapur . • • • • INTRODUCTION Every component of a piping system has a role to play. chemical anchoring technology was recommended for fixation of the new pipe supports to the existing Filter house framed structure. sahua@indianoil. along with use of high build epoxy primer and paint for reducing corrosion levels. 4 1 A S E I Keywords—pipe support. bracket support. Based on the findings of the analysis. restoration.in Manager (Civil). Engineering Services Department. India. to prevent sagging. Pipe support assemblies can be classified according to both their level of engagement (active or passive) and function (load-bearing. They are.Detailed study and proposed restoration of damaged structural bracket supports for three tier insulated piping system by using anchoring methodology in filter house structure of solvent dewaxing unit Moni Sankar Hazra Amarjit Sahu Civil Engineer. mechanical stresses and vibrations. chemical anchoring. The reasons for damage may be largely attributed to high degrees of corrosion and improper structural arrangement of the existing bracket supports. examinations and suitable modifications can enhance plant safety. Haldia. deflections and support reactions was found to have undergone significant reduction for both the new bracket and portal frame support cases. or lost. to prevent sagging. damaged. Failure to consider the stresses. The proposed modifications would thereby enhance plant safety. Critical piping and pipe support failures have occurred. these two general problem areas are often the result of four broad categories acting independently or in various combinations [2]. Two areas of consequence which can evolve from improper support performance are the development of unpredictable stress and elevation changes in the connected equipment [2]. Indian Oil Corporation Limited. Haldia Refinery. Indian Oil Corporation Limited. generated during operation without adding excessive stress to the overall system. Thorough inspection of the piping and the overhang bracket support system revealed that the supports at most places have been dislodged from the connection with the main reinforced concrete framed structure. Periodic piping system walk downs.0m along northern side of Filter house structure in Solvent dewaxing unit of Haldia Refinery has been in severely corroded and deteriorated condition. The recommended new pipe support system will result in a considerable reduction of stresses. Pipe supports are structures that suitably support pipes Design (product-specific deficiencies or an oversight in their application) Installation (supports improperly positioned or calibrated) Fabrication (poor workmanship/quality) Service (hostile environments. I. One requirement of pipe supports. reliability 118 of 446 Copyright @ Department of Electrical Engineering. Haldia. particularly with respect to how they influence the safe and continued operation of plant equipment [2]. reliability and operating life. hazrams@indianoil. the member stresses. Whether they represent active or passive designs. Due to execution constraint.

coastal saline in nature and close proximity with river often introduces salinity in the fresh water. Existing column layout of the filter house structure in grid pattern. replacement and shutdown time can also be avoided. the metal band joints were severely damaged at most locations. II. Haldia is also one of the biggest ports in the Eastern region and a focal point for industrial development in West Bengal. For the two larger dia. assess the present condition of the structural piping supports. vulnerability to corrosion related failures is significantly high in the region. filter house section including vacuum compressors and exchanger. The high costs associated with repair. but no proper foundation has been provided to these temporary supports. The support condition of piping along northern side of Filter house in SDU was found in severely corroded and damaged condition. although an external temporary arrangement has been provided to take the load of the piping. for better structural stability. • The structural members used for supporting the two larger diameter pipes in the upper tiers have totally corroded away at many locations and also the supports were wrongly provided viz. The load bearing structural beams and angles were found in severely perforated condition. pumps and column section.0m along north side of the filter house section structure. pipes in 2nd and 3rd tier. Piping system walkdown inspection and observations A site visit for walkdown inspection was carried out to Fig. The study carried out by Engineering Services department (civil section) also includes recommendations and proposed corrective measures for repair and restoration of the damaged supports. Also. WALKDOWN INSPECTION AND OBSERVATIONS The important observations during detailed walkdown inspection of the pipe supports and structures along north side of Filter house section of the unit has been mentioned below. The bracket support for pipes should be preferably provided from the main structural column. foundation provided to the temporary support appeared improper. Haldia is located at 22. Hence. high unsupported length). Haldi & Rupnarayan in East Midnapore district. by extending a bracket from a member meant for only supporting a B. A. The solvent de-waxing Unit (SDU) of Haldia Refinery. At a few places.06°E longitude and has an average elevation of 8 meters above mean sea level (msl) [3]. the bracing should be provided from the end of the overhang portion of the bracket. 1). The present study was carried out to assess the condition of the structural bracket supports for the three tier piping system at an elevation ranging from 5. Average annual rainfall over Midnapore district during 2007– 2011 is 1680 mm and average rainfall during southwest monsoon season (July – October) is 1288 mm [4]. It is located at a distance of 125 km South-West of Kolkata and 50 km from the Bay of Bengal at the confluence of three rivers Hooghly. 1.e. the insulation blocks have fallen of the pipe. Also.0m to 8. At places. The existing column layout of the filter house structure in a grid pattern has been shown below for reference (Fig. • The insulation for the pipes was found severely damaged in many locations. 4 1 S E I A • The temporary support provided to the pipelines was by means of using a slender member (i. Haldia Refinery and its township area fall under deltaic and residual soil areas. NIT Durgapur . Study Area Haldia is one of the most rapidly growing towns in West Bengal and is on the deltaic tidal range of the Ganga basin.03°N latitude & 88. which are in severely corroded and deteriorated condition. the metal band joints were found dislodged from the pipes.and operating life. The support beams in most cases was found dislodged from the connection with main column of the structure. The aluminum foil cladding was also in damaged condition. Haldia Refinery also falls in the most severe zone in atmospheric corrosivity map of India. In a few cases. The unit has been divided in three separate sections – Ammonia section including chillers. • Bracket support for pipes in Grid-26A has been provided from a non-load bearing structural member. 119 of 446 Copyright @ Department of Electrical Engineering. which is structurally not recommended. Indian Oil Corporation Limited was commissioned in the year 1976 along with other lube units for production of lube oils. The soil of the region is brackish.

Density of insulation material = 260 kg/m3. A.small walkover or extending a ‘L’ shaped bracket from a member meant for only supporting grating & platform load. • For other two tiers. Member Sections The member sections for new overhang structural bracket support system (for an immediate corrective measure as shown in Fig.0 t/m2.e. Fig. • Safe Bearing Capacity (SBC) of soil at 2. NIT Durgapur . Typical view of new bracket support system model in STAAD. single insulated pipe of diameter higher than 300mm has been considered. unsupported length (i. • Lowest tier has been considered to be full of pipes with diameter lower than 300mm. • Basic maximum wind speed = 55 m/s. • Beam (2nd tier) - ISMB 250. The support system was then re-modeled considering a new overhang structural bracket system (for an immediate corrective measure) as well as considering a reinforced cement concrete (RCC) portal frame support system (for a long term sustainable measure). • Beam (bottom tier) - ISMB 300.0m. span of pipe between two consecutive support) for these pipes was found to be very high and even the minimal supports that were present were heavily corroded and were not providing any support to the pipe system. 2. • All structural members joint shall be welded joints with continuous fillet weld.0m. Severe damage in the fire proofing was observed in the structural beams and bracings. the support condition of piping in SDU was found in severely damaged condition.0m. Some of the beams are getting corroded due to the paint peel off because of inadequate surface cleaning during painting.5:3 mix. Design Basis The important design basis considerations are as follows. mainly because of dislodgement of the pipe brackets.0m depth = 5. • Struts (between tier) - ISMC 150. Overall.8m. maximum width over which pipe is spread = 2. • Maximum overhang length for bracket support = 2. Re-assessment and immediate repairs or replacements are to be carried out to increase service life of these structures.Pro 120 of 446 Copyright @ Department of Electrical Engineering. The cold insulation of piping connected to filters was found damaged in most piping circuits causing ice formation and leading to severe corrosion of the pipe and supports due to water dripping. • For portal frame. However. • B. • Maximum unsupported length of pipes for grid 22A – grid 26A = 8. • Concrete for RCC work (nominal mix) = 1:1. rd A S E I STRUCTURAL MODELING AND ANALYSIS The existing filter house frame structure along with the overhang structural bracket supports along the north side was modeled using STAAD. • Bracings - ISA 150x150x15. • • At 1st floor level of the Filter House. III. The series of pipes in the lower tier was found heavily sagged between Column 19A and 21A. • Pipe loads has been considered as water filled pipes for the respective diameter and with 75mm insulation thickness.5m. 2) are as follows. • Depth of water table below natural ground level (NGL) = 0. two parallel pipes were found insulated together. Such arrangements are structurally unstable and not recommended. 4 1 • Beam (3 tier) - ISMB 250. towards northern side.Pro v8i (Select series 4) and the adequacy of the existing support system was checked. • Maximum unsupported length of pipes for all other grids = 6. The severe corrosion and perforation of load bearing beams and channels were observed at most of the locations. The approach to these pipes was highly congested and appeared to be one of the main reasons for poor maintenance.

50% in case of new bracket support. Therefore. The proposed modifications would thereby enhance plant safety. 4 1 • The maximum nodal displacement in a member showed a significant reduction by about 73 – 92% for the new supports.69 27.TABLE I.764 0.0m. maximum support moment at the foundation level was reduced by about 9 – 20% for the new supports. However. Overall. Fixing of the prefabricated structure by use of chemical anchoring technology was thereby envisaged as a workable solution. it was observed that for the bottom tier of pipes. Analysis and Results The results obtained after analysis of the support models has been summarized below (in Table I) for the 3 cases – existing support system. deflections and support reactions was found to have undergone significant reduction for both the new bracket and RCC portal frame support cases. nodal displacement mm 17. stress ratio in a steel member - 0. • The maximum stress ratio for a steel member was reduced by about 16. from that in the existing case. it is observed that. RECOMMENDATIONS FOR RESTORATION A.87 25. Typical elevation for new RCC portal frame support system C. during walkdown inspection and structural analysis.656 0. B2. nodal moment at connection interface From the tabulated results. B3) - 400mmx300mm. • Main column (square) - 400mmx400mm.328 Max. nodal reaction at connection interface Max. NIT Durgapur . 3.77 130.94 147. Therefore. IV.35 280. which also consists of other running process units.14 Max.59 Max.48 4. support moment kNm 161. The member sections for new RCC portal frame support system (for a long term measure as shown in Fig.55 1.99 283.26 kNm 15. the length of overhang portion was 2. • Beams (B1. in running plant condition. Site constraints and plant safety considerations The Solvent de-waxing unit is located in Lube Oil block. SUMMARY OF ANALYSIS RESULTS Pipe Support Cases Parameter Unit Existing support New bracket support New RCC support kN 32. the member stresses. and with no provision for future piping expansion. Also. support reaction kN 280.77 Max. Fig. • The maximum nodal moment at the connection interface of the structural members with the RCC column of filter house structure was reduced in the range of 22 – 40% for the new supports.34 12. 3) are as follows. A S E I • The maximum support reaction at the foundation level remained almost similar for all three cases of supports.59 9. welding works during execution of work at site would not be permitted for ensuring plant safety aspects. the long term and sustainable 121 of 446 Copyright @ Department of Electrical Engineering. new bracket support system and new RCC portal frame support system. • The maximum nodal reaction at the connection interface of the structural members with the RCC column of filter house structure was reduced in the range of 15 – 23% for the new supports.18 Member max. reliability and operating life.

the following has been recommended for restoration of the damaged pipe supports. 3. 19A.S. prevent forced shutdowns. B. Mulvaney. • • • • • Structural bracket supports for the three tier piping system along northern side of filter house for grid– 18. NIT Durgapur . All new structural supports shall be painted with epoxy based high build corrosion resistant primer and paint. Care shall be taken so as to ensure that during drilling of the holes for anchor bolt. It is worthwhile to mention that. M.com/technical/articles/consideringmovement. After completion of the fixation of the new supports as per proper procedure. 4th. Embedded depth of the chemically anchored reinforcement into the substrate shall be 150mm or 10 times the diameter of the reinforcement. Ohio. improper weight balance. Proceedings of International Conf. Removal of damaged support and installation of new support shall be executed in a staggered manner.removed. USA: Technical Articles.0m could be avoided by the portal frame arrangement. and shall extend the life of piping and other associated equipments. • ACKNOWLEDGMENT We humbly acknowledge the contributions of Inspection department. 6th Grid and the same shall be replaced with new support. 4 1 A S E I For a long term and sustainable corrective measure. Web address: http://www. no damage is caused to the main reinforcement of the column. Puducherry. 1996. “Considering all movement in pipe support design”.T. December 2010. i. Periodic pipe support system inspection and survey along with carrying out minor modification works shall help to increase plant safety and reliability. after proper surface preparation. 21A. Piping Tech. January 2010. the temporary structure shall be The insulation for the pipe system shall be thoroughly inspected and replaced. The chemical anchor bolts shall be fixed strictly following the application procedure mentioned in the manufacturers’ data sheet. piping system support problems are mainly attributed to improper design of supports not allowing sufficient flexibility for operational displacements. Recommendations for restoration works Based on the structural analysis results for the various support systems. Green Campus Summit – 2013. Once. p. the pipes shall be first supported using temporary structure. “Comprehensive environmental pollution abatement action plans for the industrial clusters in West Bengal – action plan for Haldia”.e. 5th Grid and the same shall be replaced with new support. 24A and 26A (a total of 7 grids) shall be replaced with new bracket support as an immediate corrective measure. and the failure to consider all loading conditions. Kolkata. measure of a RCC portal frame structure was envisaged without affecting the existing utilities. 2014.htm. Indian Oil Corporation Limited in carrying out the walkdown inspection of the damaged piping system supports. Sahu. Texas. India. During restoration works. April 2013.pipingtech. The anchor plate shall be fixed to the existing main RCC columns using chemical anchor bolts (M20). the new supports shall be placed in position. Geehan. damaged support shall be removed from 2nd. Subsequently. National Board Classic series. and Products. whichever is higher. Haldia Refinery. of grids is recommended. 3rd. the damaged supports has been dismantled and removed. “Pipe support performance as it applies to power plant safety and reliability”. the bolt position shall be suitably adjusted. REFERENCES [1] [2] [3] [4] A. 122 of 446 Copyright @ Department of Electrical Engineering. 22A. West Bengal Pollution Control Board (WBPCB). The National Board of Boiler and Pressure Vessel Inspectors. Hazra and A. • The sagged pipes shall be restored to position after the new supports has been placed. CONCLUSIONS The restoration of the pipe support system as per the recommendations shall enable safe and reliable operating condition for the concerned unit. save repair and replacement costs. India: Technical Publication. The concrete surface shall be completely dry and cleaned thoroughly before fixing of the chemical anchor bolts. poor material selection and workmanship. G. V. If the anchor bolt position and existing reinforcement location are found to be fouling with each other. 70. USA: Technical Articles. 20A. Also. The anchor plate used for fixing of the new bracket support to existing column shall be minimum 16mm thick. but providing an adequate support to the existing piping system along with provision for extension in all three tiers. accessed on February 08. p. damaged support shall be removed from 1st. an extended overhang length of 2. portal frame structure for supporting the three tier piping system along northern side of filter house for aforesaid 7 nos. The connection of the new beams to the existing RCC columns shall be executed by using chemical anchoring of reinforcement. “Study on effectiveness of recharge structure form of rain water harvesting in high groundwater table area and coastal saline environment of Haldia refinery and township”.

Cancer patients undergoing chemotherapy are susceptible to low white blood cell count (neutropenia) [3]. which might open opportunities for solid-state cooling. India goswami.com M. it has gained application in a wide range of advanced medical devices in biomedical engineering. infection [4]. who subsequently won the Nobel Prize in Chemistry in 1993 for his work and now. such as electrical hazards from fluid leakage [11]. Peltier effect. water blankets possess certain drawbacks. S E I In order to avoid these problems and minimise their effects.Cooling Belt using Thermoelectric Refrigeration for Medical Purposes Dr. Thermoelectric Refrigeration has emerged as a promising breakthrough in the field of refrigeration and with the development of various thermoelectric materials. interchangeable direction of operation. Moreover. this alternative refrigeration system has recently attracted considerable attention of a large number of scientists all over the world due to their distinct advantages such as lesser maintenance due to absence of moving parts. which in turn is the main contributing factor towards various environmental threats such as stratospheric ozone layer depletion and the greenhouse effect due to the presence of CFCs and HCFCs in the troposphere. PCR has become a ubiquitous and indispensible method used in medical and biological laboratories for DNA amplification [2]. Tech Scholar. Treatment utilising water blankets. independent of position and temperature controllability to within ± 0. operation in severe environments. ice bags called ice packs are commonly used to alleviate the pain 123 of 446 Copyright @ Department of Electrical Engineering. Also in the field of professional sports therapies. thermoelectric module. These side effects can lead to temperature sensitivity. reliable and quiet in operation as there is no vibration. leaving the patient feeling cold [6]. there has been tremendous increase in atmospheric CO2 percentage over the past few decades. Thus. and patient neuroprotection following cardiac arrest [9. use of thermoelectric blankets with proper electrical insulation and temperature controlling mechanism can help in avoiding these problems. cooling of electronic devices etc. development of a wearable thermoelectric heater/cooler can prove to be a boon for athletes who require constant and immediate therapy for low-grade tissue injuries in order to allow them to return to their activities. Rahul Dev Misra Joly Goswami Department of Mechanical Engineering National Institute of Technology Silchar Silchar. Nonetheless. NIT Durgapur . Department of Mechanical Engineering National Institute of Technology Silchar Silchar. light weight. cooling. Therapeutic cooling and heating of tissue injuries is also an effective treatment and has been known to reduce the healing time [12]. these kinds of refrigeration systems cannot be utilised for remote applications due to the limitation of use of grid power. Keywords—Seebeck effect. In such cases. With advancement in the field of thermoelectrics. compact and rugged in construction. bismuthtelluride.1°C [1]. or wraps dates back to the 1950s. Abstract—Thermoelectric Refrigeration is an environment friendly solid state technology for small scale localized cooling or heating in various applications such as refrigeration of biological samples/specimens during tissue preparation and storage and other medical applications. when tested for heating and cooling applications.joly@gmail. This kind of temperature controllable jackets or vests can be very helpful for military personnel and soldiers during extreme climatic situations. heating.com other such harmful fluids. I. swelling etc. Hypothermia blankets are pads that patients rest on during surgery to keep their body at a certain temperature. 8]. and low red blood cell count (anemia) [5]. 10].. researches have been carried out to develop new eco-friendly alternatives. During various injuries such as burning and swelling. vaccine storage and other medical and pharmaceutical services. no CFCs or 4 1 A INTRODUCTION Conventional refrigeration systems use ChloroFluoro Carbons (CFCs) and Hydro ChloroFluoro Carbons (HCFCs) as heat carrier fluids which release carbon dioxide (CO2) and other toxic gases to the atmosphere causing global warming. Modern polymerase chain reaction (PCR) thermal cycles for rapid heating and cooling of DNA using thermoelectric devices was developed in 1983 by Kary Mullis. long operational life. CPU cooling. India rdmisra@gmail. This paper presents the designing and development of a belt using thermoelectric module.Design and Development of a Heating. therapeutic products such as temperature controllable blankets or couches using thermoelectrics may potentially help patients cope with these side effects. burns. thermoelectric refrigeration. The belt. With the rapidly increasing demand for refrigeration in the fields of surgery. electronics and optoelectronics. sprains. Therapeutic hypothermia has been particularly effective for treatment of neonatal encephalopathy (brain disorders) [7. showed desirable results and hence can be expected to be useful from medical point of view for treatment of joints and muscle pain. vests. infrared detectors etc.

(4) The proportionality constant of (4) is called Peltier coefficient ab. E. The emf output is given by where q/dx is the Thomson heat transfer. The proportionality constant of (1) is denoted by ab = E / T where ab. discovered that the passage of a current through a junction of two dissimilar conductors caused a temperature change. ab should be high. A (2) (9) qj = I2R (10) where I and R are the current (amperes) and electrical resistances (ohm) respectively.ab = a – b and in some cases. is the coefficient of a pair of different metals (say A and B) such that ab = a ‒ b S E I SI unit of ab is volt per unit kelvin. French watchmaker. Thomson effect. Thomson effect It is a reversible thermoelectric phenomenon discovered by Thomson in 1857. five effects are observed: Seebeck effect. there is heat transfer from hot end to the cold end and is related as qcond = U (Th ‒ Tl) (11) where qcond is the rate of heat transfer in watts per unit square meter. When a current passes through a single conductor having a temperature gradient. Conduction effect If the ends of any element are maintained at different temperatures. This effect is known as Joulean effect [13]. known as the Seebeck co-efficient or thermoelectric power [13]. But high current will render high heat generation due to Joulean effect [13]. (6) SI unit of ab is volt. NIT Durgapur (12) . such that q = ab I D. He found that the heating or cooling rate (q) is directly proportional to the current (I) qI 4 1 From (9). Joulean effect When an electrical current flows through a conductor. If there is only one conductor of crosssectional area A (square meters). In this project. Joulean effect and Conduction effect. Jean Charles Athanase Peltier. there is a dissipation of electrical energy (qj). a belt is designed such that it can provide both heating and cooling effect. II. hot water bags and such other heating mechanisms are employed for pain relief. the overall conductance is given by (5) U = kA / L 124 of 446 Copyright @ Department of Electrical Engineering. THERMOELECTRIC REFRIGERATION EFFECTS C. it is evident that to get high value of cooling or heating. which is given by q / dx = I (dT / dx) Whenever direct current passes through a pair of thermocouples with junctions maintained at different temperatures. found that an electric current would flow continuously due to generation of electromotive force (emf) in a closed circuit made up of two dissimilar metals provided that the junctions of the metals were maintained at two different temperatures. being Thomson co-efficient having unit volt per unit kelvin. (7) A. there is heating at one junction and cooling at the other depending upon material combination [13]. Peltier effect. it is found E T (1) q = ab IT where E and T are emf output (volt) and temperature difference (kelvin) of the two junctions respectively. According to Joule this is related as (3) B. This generation of emf is known as Seebeck effect. Tl are the high and low temperatures in kelvin respectively. Zemansky [14] using first and second laws of thermodynamics obtained the relation between Seeback and Peltier coefficient as ab = abT (8) Using (8) into (5). i. if direct current is passed through a pair of dissimilar metals. U being the overall conductance in watts per unit square meter per kelvin and Th. the desired heating and cooling effect can be achieved. conductivity k (watts per cubic meter per kelvin) and length L (meters). even to stop bleeding from cuts etc. Thomas Seebeck. which is the co-efficient for two different metals (say A and B). heat transfer takes place [13]. This is called Peltier effect and it is the main concept behind working of a thermoelectric refrigerator. Seebeck effect The first important discovery relating to thermoelectricity occurred in 1821.e. Again. Peltier effect In 1834. when a German scientist. A thermoelectric module is used as the basic component of the belt such that by reversing the polarity of the applied voltage. I is the flow of current through the conductor in amperes and dT/dx is the temperature gradient along the conductor in kelvin per meter. otherwise large current would be required. in case of muscular and joint pain.

In addition to Bismuth Telluride. electrical resistivity () and thermal conductivity () over the operational temperature range of cold and hot side named as figure of merit (Z) and expressed as Z = 2 / () Higher the value of figure of merit. A large number of thermocouples are connected electrically in series to form a module which is the main component of a thermoelectric cooler or heat pump. Both the p-type and n-type materials have different free electron densities at the same temperature. In short. Single stage system A simple thermoelectric refrigeration system consists of a single-stage semiconductor thermoelectric device. His work explained that thermoelectric cooling materials needed to have high Seebeck coefficients. Goldsmid and Douglas demonstrated that cooling from ordinary ambient temperatures down to below 0°C was possible. the performance of the modules depends on not only the physical properties of the thermocouples. As current flows up and down through the module it attempts to establish a new equilibrium within the materials. Altenkirch give the basic theory of thermoelectrics. A S E Conventional thermoelectric devices have various specifications for various applications. but also the configuration of the thermoelectric modules. the rates of heat rejection from the device to the high temperature heat sink and input from the low temperature cooled space to the device. Electrically conductive materials. which is composed of a large number of p-type and n-type semiconductor elements connected electrically in series and thermally in parallel. The ceramics also serve as insulation between the modules’ internal electrical elements and a heat sink that must be in contact with the hot side as well as an object against the cold side surface. Bismuth Telluride began to be the primary material used in the thermoelectric cooling and it is the most commonly employed material for commercially available thermoelectric coolers.5 when operating at temperature difference of 20°C. the hot side temperature of one stack is the cold side temperature for the adjacent stack. Thermoelectric systems are mainly classified as single stage and multistage systems. The current treats the p-type material as a hot junction needing to be cooled and the n-type as a cold junction needing to be heated. reversing the polarity will switch the hot and cold sides. It is typically composed of two ceramic substrates that serve as a foundation and electrical insulation for p-type and n-type semiconductor pellets that are connected electrically in series and thermally in parallel between the ceramics. Since the material is actually at the same temperature. P-type dice are composed of materials having a deficiency of electrons while n-type has an excess of electrons. Shortly after the development of practical semiconductors in 1950’s. and low thermal conductivity to reduce heat transfer from junctions to junctions. from its surroundings except at the junction–reservoir contacts. However. Solder is most commonly used at the connection joints to enhance the electrical connections and hold the module together [15]. such as the allocation of the junction temperature difference in the module. the dimensions vary from 3 mm square by 4 mm thick to 60 mm square by 5 mm thick. The maximum temperature difference between the hot and cold side can reach 70°C. For cooling. of a present thermoelectric refrigerator is typically less than 0. the result is that the hot side becomes hotter while the cold side becomes colder. The pyramid shape is used because the stages closer to the heat sink must remove the heat dissipated by the stages closer to the heat source in addition to the heating load of the heat source or cooling target. Multistage thermoelectric modules are often used for extending operating temperature range of the thermoelectric refrigeration. NIT Durgapur . Silicon Germanium (Si-Ge). good electrical conductivity to minimize Joule heating. where Qh and Qc are. both electrically and thermally. there are other thermoelectric materials including Lead Telluride (Pb-Te). COP is given by COP = Qc / (Qh ‒ Qc) (13) 4 1 A. The lowest practically achievable temperature is about ‒100°C. multi stage systems are developed where one group of thermoelectric modules are stacked on top of the other. theory of semiconductor thermo-elements was developed and in 1954. A typical multi-stage thermoelectric cooler is a pyramid stack of singlestage coolers. respectively. I The coefficient of performance (COP) used to define the cooling efficiency is given by the heating or cooling effect achieved per unit energy supplied or work input. the maximum heat pumping rate from 1 to 125 W. The direction of the current will determine if a particular die will cool down or heat up. have the highest figure of merit and most suitable in refrigeration. He also observed that if the current was reversed the ice could be melted. maintain the electrical connections inside the module. which is connected thermally in series. The cooling effect of a semiconductor material depends on a combined effect of the material’s Seebeck co-efficient (). In 1909 and 1911. In 1949.III. usually copper pads attached to the ceramics. Multistage system As the COP of single-stage thermoelectric refrigerator decreases with the increase of temperature ratio of the heat sink to the cooled space [16]. and BismuthAntimony (Bi-Sb) alloys that may be used in specific situations. Materials based on bismuth-telluride mixed crystal and modification of Bi2Te3. It is like a cascading system where. The devices contain from 3 to 127 thermocouples. demonstrated that water could be frozen when placed on a bismuth-antimony junction by passage of an electric current through the junction. better are the thermoelectric properties of the material. The COP 125 of 446 Copyright @ Department of Electrical Engineering. the length and number of thermocouples [17]. The thermoelectric device is insulated. (14) B. SYSTEM DESCRIPTION Emil Lenz in 1838.

This thermoelectric belt is designed to provide wet hot water massage in case of muscle or joint pain and cold massage for burns etc.” Cancer. “Magnetic polarization of metals and Minerals. “Chemotherapy-induced anemia in adults: incidence and treatment. S. The thermoelectric belt shown in Fig. Mr. 2010. when tested using soaked sponge. C. Dale.” New Age International Publishers. 1 has been fabricated as a prototype. this belt can be expected to be quite helpful for medical purposes as it is very handy and easy to operate and maintain. the test is carried out without using the sponge and a fan is used to continuously remove the heat from the hot side of the module. pp. Manohar Prasad. 1821. “Managing cancer side effects to improve quality of life: a cancer Psychoeducation program. Edwards et al.8mm.J. F. Second Edition. Martin. R. V. Mr. Rubber is used to provide flexibility and to hold the module in position with the strap. D. “Clinically induced hypothermia: why chill your patient?. When tested using a fan and without sponge.5 V [10] Current = 0. A. J. Agarwal. 1999. 2285–2289. 34(7). “Treatment of comatose survivors of out-of-hospital cardiac arrest with induced hypothermia. Time of use = 8-12 minutes For heating: Voltage = 5. 2002. 2006. UG students (Mechanical Engineering. 2004. 26(1). 557–563. c363. easy to use. Kowal and J. A. T. 100(2). “Neurological outcomes at 18 months of age after moderate hypothermia for perinatal hypoxic ischaemic encephalopathy: synthesis and meta-analysis of trial data. DESIGN AND DEVELOPMENT OF THE THERMOELECTRIC VI.. Oddo. S.a review. Stirling. “Refrigeration and Air Conditioning. M. D. Feihl. 37–44. 246(8). As a temperature of 12-14°C is obtained experimentally. The module used is a single stage bismuthtelluride module which is powered with a regulated power supply and is proposed to be powered using mobile charger. 1989. 39–45. M. Makic. “Possible benefits of hyperthermia to chemotherapy. 91(19). J.. New Jersey: Humana Press. M. aluminum sheet and thermoelectric module. H.80 A I [5] [6] [7] [8] Temperature = 40-50°C [9] For cooling: Voltage = 1. 23(1). D. So. “From evidence to clinical practice: effective implementation of therapeutic hypothermia to improve patient outcome after cardiac arrest. Eisner. simple in construction and reliable in operation. 2003. Lyman. ACKNOWLEDGMENT The authors are pleased to thank Mr. T. [12] [13] [14] 126 of 446 Copyright @ Department of Electrical Engineering. Schaller. Dutta and Mr. 53–54. 1865–1873. J.50 A Temperature = 12-14°C [11] The test is carried out using a soaked sponge for heating effect and the temperature obtained is within the range of 4050°C. and G. REFERENCES [1] [2] [3] Fig. Ribordy. M. 228–237. The design consists of sponge. S.” Journal of the National Cancer Institute. The model has been designed and modified considering various constraints and resource availability.” AACN Advanced Critical Care. Groopman and L.” vol. Pradeep Bansal.” British Journal of Sports Medicine. M. bleeding etc.” Cancer Research. 1200. Thus. “PCR Protocols.” Critical Care Medicine. L. Bernard et al.. 125–132. Omar Abdelaziz. “Assessment of brain tissue injury after moderate hypothermia in neonates with hypoxic–ischaemic encephalopathy: A nested substudy of a randomised controlled trial. the belt can be used for burns. The belt. Altman. 17(2). a heating effect of 40-50°C was obtained. 2010. Kakoti. Crawford. 226. BELT CONCLUSION The present study develops a design method of a simple belt using a thermoelectric module. Borah. a cooling effect of 12-14°C was attained. The cost of the thermoelectric module is Rs. “Advances in household appliances . for pain relief. A. 39. 3–6. Aluminium sheets are used to properly conduct the heat and the velcro holds the sponge in position. NIT Durgapur . Itri. NIT Silchar). 1616–1634. C.” Cancer Nursing.” BMJ. E. 1. 2003. F. 4 1 A [4] V. pp. and M. 2003. S. Holden and M. In case of cooling. Bertino. and C.” The Lancet Neurology. M. Edward Vineyard. 9(1). The main advantages of this belt are portability. Seebeck. As a result. McLean. A. straps.7 V Current = 0. bleeding etc. swelling. sprains etc. Golant. velcro. 2006.” Abhandlungen der Deutschen Akademie der Wissenschaften zu Berlin.” New England Journal of Medicine. for their help and support in the fabrication and experimentation. it can be easily used as a first aid measure during sudden cramps. Thermoelectric heating-cooling belt EXPERIMENTATION AND TEST RESULTS The model is powered from a regulated power supply and the experimentation details are as follows: S E Module size = 40mm×40mm×3. Further experimentation is going on for better results and modification of the belt. Bartlett.” Applied Thermal Engineering 31 (2011) 3748-3760. B. D. 265:1822-23. M. D. 340(9). “The use of cold and superficial heat in the treatment of soft tissue injuries. Liaudet.IV. D. D. Rutherford et al. it can be used for pain relief in case of joints and muscle pain as well as for burns. 1979. “Chemotherapy-induced Neutropenia.

Wang. Hongjie Wang.” Cryogenics 47 (2007) 89–93. Mr.” Applied Energy 73 (2002) 285–298.April 2013. [16] Jincan Chen. Kangshan Xie. Hua Zhao. “Comparison of the optimal performance of single and two-stage thermoelectric refrigeration systems. Yinghui Zhou. [15] Prof. Miss. “Analysis of optimum configuration of two-stage thermoelectric modules.” International Journal of Engineering Trends and Technology (IJETT) – Volume 4. “Fabrication of solar operated heating and cooling system using thermo-electric module. NIT Durgapur . Issue 4. Priti G. Jin T. Mukesh P. Mangtani. Gandhewar. Vivek R.[17] Jianlin Yu. Bhadake. 4 1 A S E I 127 of 446 Copyright @ Department of Electrical Engineering.

The difference in the direction of propagation of these coupling currents along the phase and neutral line.e. Gitam University. all the power control techniques deliberately deliberatel distort sinusoidal wave form of power frequency and generate unwanted interfering signals. • High power handling capacity capacity. Widely used Switched mode power supplies (SMPS) for powering today’s electronics loads is a most common example. AP.Srinivasa Rao2 1. Obviously.[1] The currents flowing through the phase and neutral conductors are depicted as ( )and and ( ). Fig. Rushikonda Visakhapatnam-530 530 045. 1: Relationship between circuital and modal voltages and currents.e. the filter design process is carried out. In order to meet the limit set for conducted EMI. Sri. composite filter. power supplies require a filter at its input.This This paper attempts to describe a design technique for Power Line filter using classical design technique. 1. by considering the effect of non non-ideal behaviour of passive lumped components.com Abstract-A new methodology to design power line filter. • Two modes of propagation i. The propagation of interferences through phase. INTRODUCTION S E I Switching and modulation techniques are used for efficient use of electrical power. CE can be controlled by using filters and increasing line impedance. Unfortunately. where the parasitic effects of passive lumped elements are not taken into account. The main differences between aan EMI filter and a conventional filter are as given below: • Unknown noise source and load impedance. India 2. The coupling mechanisms of these interferences are different. Gandhi Nagar.com Keywords.power line filter. Conducted EMI generated by one equipment gets coupled to another equipment mostly through power line cables. Gitam University. The filter’s circuits were first simulated using Advanced design software (ADS) based on S-parameter parameter to verify the basic filter response. differential mode. AP. A proposed model of the power line filter design.Power Line filter Design considering losses and parasitic characteristics of Passive lumped components Shashwatee Paul1.S. Insertion los. Characterization of CM and DM currents 1. Rushik ikonda Visakhapatnam-530 530 045. they all are often cited as one of the main source of conducted emission (CE). Gandhi Nagar. based on the parasitic effect of the passive lumped elements is presented. Usually the power supplies which are using controlling techniques cannot comply with the strict electromagnetic compatibility (EMC) regulations. Shashwatee. The CM and DM currents separately can be determined from the conducted emission measurement data: 128 of 446 Copyright @ Department of Electrical Engineering. Therefore design of EMI filters demand decoupling of these two modes[1]. and DM are to be considered. Conducted EMI has two components: The common mode (CM) interference and differential mode (DM) interference. common mode. India 1. NIT Durgapur . 4 1 A A.CM i. using the conventional technique of composite low pass filter has been presented. leads to differential mode ( ) current and common mode( ) current.paul@gmail. parasitic elements. neutral and groun ground conductor is shown in “Fig. These currents can be decomposed into two auxiliary currents. It is based on a model where common mode and differential mode interference rference is separately considered into the filter design. which are referred to as the common mode current ( ) and differential mode current ( ) [2 [2]. In this paper.1” [3]. srinu16ssr@gmail. The simulation result is compared with the cases.

2” [4] emission Fig2: Equivalent circuit of the conducted emissions measurement. 4 1 B. 3” 3”.Therefore according to the definition of Insertion loss. one for phase line and the other for neutral line. (1) (2) Input impedance of measuring instruments is mostly 50Ω.It consists of two line impedance stabilization networks (LISNs).From the data of conducted emission test. 4EMI Receiver data C. The measurement data of the test est Receiver is shown in the ““Fig. 4”.The measurement is carried with Line impedance stabilising network (LISN) and the the conducted emission test is viewed in the EMI Test Receiver.Therefore. Determination of Insertion Loss Basic definition: The Insertion loss (IL) as a function of frequency is the most fundamental characteristic of a filter and is defined by [5] IL (dB) = 20 (5) Where. Measurement is carried out in frequency domain the graphical display of the conducted emission is shown in the EMI Receiver. A S E I Fig. = the output voltage of the signal source without the filter connected to the circuit Fig 3: Measurement set up at SAMEER Kolkata = the output voltage of the signal source at the output terminal of the filter with the filter inserted in the circuit. it can be viewed that the peak level of emission exists at 80 dB V. Conducted emission measurement An example of a measurement setup for conducted emission tests according to CISPR 11/22is carried out at the centre of SAMEER Kolkata as shown in “Fig.the CM and DM Voltage can be defined as: (3) (4) The equivalent circuit for the conducted measurement setup is shown in the “Fig. Fig. the filter must be designed esigned in such a manner. NIT Durgapur . 129 of 446 Copyright @ Department of Electrical Engineering.Therefore 40dB. But according to the standard specification. The EUT is supplied by the power source through an artificial mains network (AMN). that the attenuation of the power linee filter must be more than 40dB 40dB. the limitation of the emission is40dB.

The significance of constant-k filter is such that the product of (series impedance) and (shunt impedance) is a constant real term. but for HF signals. we present a design methodology to realize low pass filter using a unique combination of constant-k. This type of LPF is called composite filter[5].15 to 30 Mhz range.2. 6”[5] 4 1 Constant-k Section: The value of Capacitance and Inductance is determined as A (6) For low-frequency. Therefore. the sharpness of the cutoff is increased. the m-derived filter section is a modification of the constant-k section. Hence the power delivered by the source is mostly absorbed by the load. The model of equivalent circuit of capacitor is as shown in the fig (5). And the prototype section (constant-k) provides the necessary attenuation at frequencies remote from cut off. In this paper. the parasitic capacitance become more dominant. It is designed separately for the differential mode filter in the frequency range of 150 kHz to 1 MHz and the common mode filter in the frequency range of 1 MHz to 30 MHz. such as noise currents in the 0. Role of parasitic effects of passive elements B. The lead inductance of the capacitor can be determined from its self resonant frequency (SRF). we present a design methodology to realize low pass filter using a unique combination of constant-k. So. With this. The impedance of the capacitor is given by[5]: S E I Inductor: The equivalent circuit of an inductor and the non ideal behaviour the inductor can be represented as shown in “Fig. Filter Design Process In this paper. capacitors are connected in parallel with the noise source. But at the same time it has some demerits. m-derived section. (7) Fig5: Equivalent circuit of capacitor including its non ideal effect. FILTER DESIGN PROCEDURE and after that. . m-derived section. it is low impedance. The impedance of the inductor is given by: A. a capacitor appears as high impedance.

The model of the equivalent circuit of Capacitor consists of a series combination of series resistor ( ). bisected. the frequency response of the inductor shows that the inductor is dominant unto a certain frequency ' & 130 of 446 Copyright @ Department of Electrical Engineering.e. NIT Durgapur (13) . The lumped elements of the terminating half sections at the input and output port of the low pass filter can be determined as Fig6:Equivalent circuit of an inductor including its non ideal effect. ! (8) M-derived Section: The cut-off frequency ( " ) and the infinite attenuation occurred at frequency ("# ) is defined as "# √& " √ (9) The independent element values of m-derived LPF is obtained as ' & (10) ! ' () '' & *& (11) (12) Matching section: This Circuit can be further modified using the Matching section i. Unlike capacitor. In the conducted emission range (150 kHz–30 MHz). Capacitor: The mathematical models that yield considerable insight into the non ideal behaviour of components have been used. the behaviour of these elements is far from the ideal. lead inductor ( ) and capacitor as shown in the “Fig.5”.+ section of the Mderived section that are implemented as a terminating Half Section.

47µH 127. 131 of 446 Copyright @ Department of Electrical Engineering.77µH 19. It is based on the propagation of CM current from source to load. This non ideal behaviour of the filter characteristic is due to the effect of parasitic components of the lumped elements (capacitor & inductor).The The design value of the components is shown in “Table 1”. The design is carried out at a cut off frequency ): 333 kHz and maximum attenuation pole ( 30 MHz as in ““Fig.The The direction of propagation of DM current is considered while designing the circuit. E. NIT Durgapur .33khz Fig9:Simulation result for DM & CM composite filter. thee series inductors of DM section are connected in series with the circuit of CM filter. The non ideal response of the individual passive components was simulated as shown in ““Fig. 11”. 4 1 3. 10” and “Fig. 8”. “ 9”.7”. It shows a ripple formation in the frequency response of the graph. Table 1 Design components of Filter circuit Mode L C DM 318. Filter Performance ideal behaviour of Inductor Fig 10: Simulation result for non-ideal The differential mode & common mode circuits are simulated in ADS based on S-parameter as shown in “Fig. While the CM filter shows aan attenuation of 125 dB at the frequency of 30 MHz. as shown in “Fig. ROLE OF PARASITIC EFFECTS Fig7:Circuit of lumped element DM composite filter. Common mode Filter S E I In this circuit. The non ideal response of the filter can be noticed from the frequency characteristic graph of the Common mode filter at higher frequency. A D.1nF 333. Fig8: Circuit of lumped element CM composite filter.(14) (15) (15 C. The simulation result of the DM filter has an attenuation of 84dB at the 1 MHz frequency.39 nF 50kHz CM 47. Differential mode Filter The composite low pass filter is designed for the differential mode filter with the cut off frequency : 50 khz and infinite attenuation pole :1Mhz.

13”. NIT Durgapur . by changing the value of a power supply filter element nt that affects that component [7]. Modified Common mode filter design In the similar manner. 13” Fig12:Illustration of the dominant component mponent of current over a particular frequency range of the conducted emission test. This is due to the result of parasitic effect of the passive element (capacitor or inductor) [6].The he level of a particular (dominant) component can be reduced. namely the self-resonant frequency (SRF). A. MODIFIED FILTER CIRCUIT. Common mode section of the filter is also modified by considering the parasitic components of the passive lumped elements as shown in ““Fig. Each of the passive elementscan can have good high frequency characteristics due to the pres presence of the ever present parasitic. the dominantcomponent at that frequency must be reduced. 4 1 A S E I (16) 100pF Fig13:ADS schematic of DM filter B. the original circuit of the filter is modified by adding the parasitic components in the basic equivalent circuit of the passive lumped elements(inductor and capacitor). Hence in order to reducethe total conducted emission at a particular frequency. Fig13: ADS schematic of DM filter 132 of 446 Copyright @ Department of Electrical Engineering. the total current of interference is the resultant of the dominant component nent current as shown in “Fig. Table 2 Parasitic components of modified filter circuit Components 5. one ne component maydominate over the other. given by The concentration is mainly on their rolee in suppression of conducted emissions and on their non ideal behaviour. any suppression component resonates at some frequency. As a result of their presence.The ADS schematic circuit of the differential mode filter ddesigned as shown in “Fig. it is clear that over a certain range of frequency.6 nH Fig11:Simulation result of Capacitor From the simulation result of the passive elements. 12”. Modified differential mode filter design In order to obtain the desired frequency response and the Insertion loss of the differential mode filter (DM).4. Over a certain range of frequency. capacitor).

IEEE Trans Electromagne lectromagnetic compatibility. J. 2001. IEEE Trans. [3] [4] [5] [6] [7] V. design of power line filter using constant-k constant and mderived composite low pass filter gives a better frequency response. it is clear that the desired attenuation is achieved within the range of conducted emission. New York. “Power line filter Design Des for Conducted EMI using Time-domain domain measurements”.The The design of power line filter with composite LPF is simple and hence it avoids any complicacy. SIMULATION RESULT OF MODIFIED CIRCUIT.Prasad kodali.3. “ 14 and 15”.“A Modal Model of Common-Mode Mode Chokes for Cond Conducted Interference Prediction”. Inc. The he institute of Electrical and Electronics Engineers. Espoo. CONCLUSION From the simulation result of the filter circuit it is noticed that.5. Technology 2009.M. Engineeringg Electromagnetic Compatibility. Finland.Wiley & Sons. Now after tuning the circuit with the modified filter section. Joan Ramon Regue. Konstantin Kostov. D. Konstantin antin Kostov and Jorma Kyyra. Miquel Ribo. Nordic Workshop on Power and Industrial Electronics. 1996. [2] The final result of the power line filter is shown in “Fig. Fig14:ADS schematic of DM filter 4 1 A Fig15:Simulation result of the modified CM M filter. circuit REFERENCE [1] Mohit Kumar and Vivek Agarwal.48. February 2006 133 of 446 Copyright @ Department of Electrical Engineering. Design and Characterization of Single SinglePhase Power line Filters. August 2010.Wiley & Sons. ““Insertion loss and Network parameters in the Analysis of Power line Filters”. Electromagnetic compatibility. no. Electromagnetic compatibility. Inc. Vol. 2006.52.1. complicacy From the simulation result of the modified filter circuit. Vol. Inc. NO.Paul. NIT Durgapur . Clayton R. Microwave Engineering. Albert Miquel Sanchez. June 2008.J. Pozer. Further the he performance of the filter response are enhanced when hen the parasitic effect of the lumped elements are considered in the modified circuit. the frequency response is equivalent to the practical filter response. S E I 6. Helsinki University of Technology. Pablo Rodriguez-Cepeda and Francisco Javier Pajares.

This paper presents CFD (computational fluid dynamics) based 2D. Additionally. The second principle also referred to as aerodynamic or airfoil lift principle.co Abstract— Wind energy as a clean.com Department of Mechanical Engineering National Institute of Technology. NIT Durgapur . straight blades are used. S826 and S831 at various wind speeds and angle of attacks. oil and natural gas will gradually lower the ambient level of CO2. Assam. I. S825. The unspoken hypothesis is that having these technologies replaces coal. It is concluded from the results that NREL wind turbine profile S818 is the most efficient blade profile and suitable for working at low wind speeds. Maximum lift coefficients of the blade profiles are obtained at stall angle of 150. India agnibis@yahoo. because they are not powered by fossil fuels. CO2 reduction will be less than presently anticipated by their proponents. Economic Development). Department of Mechanical Engineering National Institute of Technology. VAWT are not always self-starting. the cost of wind-generated electricity has dropped by more than 80% since the early 1980s. As an alternative derived from the standard configuration. install and maintain. which is a major drawback. hydroelectricity production can be diminished substantially and thus become intermittent because of drought. However. and thus alleviate or even eliminate possible climate change. CFD. Keywords—2D. in April 2010 [1]. in which no stalling is documented. such as geothermal and hydroelectricity. directly connected to the rotor shaft by struts. angle of attack. particularly for conditions corresponding to low wind speeds and/or urban areas. The Hrotor is also able to accept wind from any direction. wind power is cost competitive with fossil fuels such as coal. wind speed. because they are reliable and not intermittent. Agnimitra Biswas M. state-of-the-art utility-scale wind turbine can generate electricity for 4 cents per kilowatt-hour. There are two different physical principles to extract power from wind. The lift coefficient is increased in the angle of attack range of 00 to 150 for all the proposed velocities. highly reliant on this source of electricity. and is becoming more so. Unlike solar power. H-Darrieus rotor [3] 134 of 446 Copyright @ Department of Electrical Engineering. clearly reduce green house gases. leading to extremely low costs.CFD analysis of cambered airfoil for H-rotor VAWT Shikha Bhuyan Dr. Straight Darrieus wind turbines (called also in what follows H-rotors. As other energy sources become more expensive.1) have even more attractive characteristics. is at present predominantly applied for wind energy conversion. which remains extremely expensive. steady state simulation for 4 NREL cambered profiles S818. This occurred in Venezuela. the present technical design relying exclusively on horizontal axis turbines is not yet adequate to develop reliable wind energy converters. The airfoil drag method is based on the wind drag force incident on a wind-blown surface. which is nearly competitive with fossil fuels (AWEA. 1. a number of studies suggest because of the intermittent nature of some renewable technologies. Assam. Wind Energy and 4 1 A INTRODUCTION The desire to reduce dependence on fossil fuels and adversities in climate change is resulting in numerous policy incentives for renewable energy. Silchar Silchar. wind will become even more cost competitive. Debate has raged for decades on the role of carbon dioxide in climate change. In order to economically gain the maximum energy from the wind turbine. Wind energy is found to be growing industrial sector fuelled by economic environmental and political motivations and is now a key sector for employment and investment. it is very easy to build. However. A modern. Research work is ongoing in both academic and industrial centres to improve the efficiency and reliability of the wind turbine devices that will service the growing energy demand [2]. Fig. is taken into consideration by many developed and developing countries as a promising means to provide electrical energy. Indeed.Tech Scholar. environment friendly and cost effective renewable energy resource. It is often assumed that renewable energy technologies. and wind technology continues to improve. blade profile. which is based on flow deviation inside the rotor. S E I Certain renewable. the performance of the blade profile must be evaluated. but suffer from a low efficiency compared to horizontal axis turbines. Vertical Axis Wind Turbines (VAWT) like the Darrieus turbine appear to be particularly promising for such conditions. Fig. Instead of curved rotor blades. will reduce CO2 contribution to overall energy use. Although considerable progress in wind energy has already been achieved. Silchar Silchar. India shikhabhuyan@gmail.

5.11. S831. to carry forward the present work. analyzed different wind-turbine-blade profiles using finite-volume method. So. They found that at low AOA the lift coefficient can be increased and drag can often be decreased by using a cambered airfoil [9]. The tangential force FT has the instantaneous responsibility of the torque and the power outputs from the Darrieus turbine [3]. using a Computational Fluid Dynamics (CFD) method based on the finite-volume approach. automatically lead to the best possible performance. III. leading to the best possible performance of a Darrieus turbine. 3 [10. 2. 4 [13]. 2. Fig. maximum thickness and thickness distribution of the profile and the trailing edge angle. The aerodynamic simulations were performed on two-dimensional S-series wind-turbine-blade profiles. 4 1 The geometric parameters that affect the aerodynamic performance of an airfoil include: the leading edge radius. Drag is also generated perpendicular to the lift and its presence is highly undesirable. 11. the present work concentrates on 4 cambered airfoil profiles by NREL namely S818.2% more than symmetric one[8]. Its purpose is to induce suction on the upper surface of the blade to generate lift. An alternative geometry might be much better. some attempts have already been documented to improve the performance of the Darrieus turbine. numerical simulation techniques. PURPOSE OF PRESENT WORK The airfoil is one of the fundamental parts of a rotor blade design. These lift and drag forces can then be resolved to get the tangential force FT and the axial force FN as shown in Fig.4. The main effort of the wind turbine industry in the field of aerodynamics related to the development of blades which offer better performance. A S E I There is nevertheless no proof that NACA profiles.0 A numerical analysis is performed on the airfoil using Ansys ICEM-CFD and Fluent software at angle of attacks ranging from 0 to 45 at velocities from 4 to 12 m/s at equal interval of 2 and 15 m/s with a turbulence intensity of 10%. A hexahedral mesh is generated around the airfoil with a total of 500. increased reliability and faster control of larger wind turbines [4]. developed by the National Renewable Energy Laboratory (NREL). Mohamed A. as shown in Fig. Cl and the lift to drag ratio (L/D ratio) for the airfoil must be maximized [5. S826 shown in Fig. 12]. 3. Cp and the torque generated.12].000 nodes as shown in Fig. In order to maximize the power coefficient .The implementation of an innovative aerodynamic control technique in wind turbines is a point under extensive investigation since the conventional wind turbine blade technology is reaching its limits. in particular for such very specific applications. 0 0 Fig. NIT Durgapur . 6. Airfoil nomenclature [13]. the main ideas and proposed improvements are summarized below. NREL wind turbine blade profiles [10. As a consequence. If the airfoil is set at an angle of attack α in a fluid flow and according to the standard airfoil theory. are used to investigate the response of airfoils against the blowing wind. Beyond pure aerodynamic studies of the airfoil. studied the influence of airfoil asymmetry on VAWT performance and reported higher performance for a cambered profile ( NACA 0012) with Cp of 0. it will generate a lift force FL normal to the free stream and a drag force FD in the direction of the free stream. 7]. Forces and velocities distribution on Darrieus rotor airfoil [3].447. S825. based on Computational Fluid Dynamics (CFDs) software technologies. II. Sayed et al. NUMERICAL ANALYSIS WITH ANSYS 14. Raciti Castelli et al. 4. the lift coefficient. mean camber line. 135 of 446 Copyright @ Department of Electrical Engineering. Fig.

The mesh (as shown in figure 5) is based on structured Ogrid and C-grid topology which is then subsequently converted to a high quality unstructured grid to optimize spatial discretization and meet turbulence model requirements [14]. The simulations are performed for the NREL meshed blade profiles. there is a dramatic decrease in Cl indicating that 150 is the stall angle which is also true for other 3 profile Similarly. Fig. As thickness is of S818 is 24%. For the convergence criteria. Cd increases with increase in α. From 140 onwards there is a rapid increase in Cd values. Initialization and setting of post processing parameters. it can be concluded that S818 blade profile is the optimum one amongst the selected profiles.lift and drag distribution of S818 airfoil against angle of attack ranging from 0 to 45 at wind velocities of 4. 7 shows the result of torque coefficient(Ct) for S818 NREL profile which is plotted against the angle of attack for the velocity 4 m/s. The solver settings used were as follows: Steady State Density Based solver with absolute velocity formulation.10. imitates the fuel of wind energy therefore. 5. Boundary condition setting on fluid domain as inlet. 6 shows the computational 0 0 Fig. Mesh density near the leading and trailing edges was increased to capture peak suction. The lift and drag coefficient of S818 NREL profiles at different wind speeds. IV. (c) wind speed of 8 m/s (d) wind speed of 10 m/s. The basic objective of the simulation is to find the optimum airfoil profile for H-Darrieus rotor that produces the maximum power from the wind turbine blades based on the maximum lift to drag coefficients leading to maximum torque coefficient. Fig. A lower angle of attack is therefore suitable for the airfoil to operate at to safeguard it from stalling. comparatively highest. 0 0 Fig.6. the values of Cl also increase. a better understanding of wind characteristics may lead to a better visibility of the use of wind energy [15-17]. A transitional SST turbulence model was chosen. Fig. the drag coefficient. Fig. Transitional Shear Stress Transport (SST) model 4equations. NIT Durgapur . give the highest value of Ct at all considered velocities. as a meteorological variable. Therefore. The boundary conditions are kept with air as the flow medium having a density that corresponds to a temperature of 250C. It is seen that with increase in velocities. wall and pressure outlet respectively. conducted for the whole range of the α from 0 to 45 because it is the normal operating range of the wind turbine designs. (b) wind speed of 6 m/s. 7. Positive values Ct are found from 40 angle of attack flaunting its self starting capability at these value of angle of attack.15 m/s.(e) wind speed of 12 m/s and (f) wind speed of 15 m/s. farfield and outlet (velocity inlet. possible transition and flow separation. 136 of 446 Copyright @ Department of Electrical Engineering.12. stagnation point. The walls around the 2D airfoil are assigned no slip condition. 8 and 9 show pressure and velocity contours at wind speed of 10 m/s and angle of attack of 120. RESULTS AND DISCUSSIONS Wind speed.8. The transitional model was done with Gamma Theta Model.) method for I Least Squares Cell Based method for gradient. 6. the residual type of RMS and residual target value was set to 10-6. 4 1 A S E First order upwind discretization Momentum and turbulence equations. It increases linearly with α incurring flow being smooth and attached over most of the upper surface of the airfoil. The simulations is carried out at different airflow velocities. The final mesh. Torque coefficients for the proposed airfoils at velocity 4 m/s. (a) Wind speed of 4 m/s. Past 150 angle of attack. indicating that the flow has separated from the upper surface well before the trailing edge. Control monitor of the iterative process to check convergence.

It is concluded from the results that angle of attack has a dominant effect on determining the optimum profile while the wind speed does not affect the optimum profile. 0 S E The velocity and pressure contour results in Figs. obtained from a wind tunnel experiment [18]. the working hour of the wind turbine would increase [19]. it is concluded from the results that NREL wind turbine profile S818 is the most efficient blade profile and suitable for working at low wind speeds. the pressure drops to a quite low value. The flow then decelerates. 137 of 446 Copyright @ Department of Electrical Engineering. the airfoil tends to rotate and the pressure of the lower surface is always higher than the pressure on the upper surface. Fig. Fig. so that CP quickly becomes zero and then reasonably goes to a large negative value. A VALIDATION Fig. Pressure contours for S818 profile at 12 angle of attack and 10 m/s wind speed. An airfoil with high coefficient of torque has higher efficiency than the one with lower coefficient of torque. such as velocities. As a result. I V. 0 4 1 Fig. pressure. and the magnitude of CP drops. NIT Durgapur . Velocity contours for S818 profile at 12 angle of attack and 10 m/s wind speed. On the trailing edge. 8 and 9 illustrate that. 11. the separation point moves forward on the airfoil as well as the separation of flow occurs earlier at higher angles of attack. 10. Experimental result [18]. The profile having maximum thickness should be used at low positive operating angle of attack in order to extract the maximum power from the wind. Thus. Computational results.11 shows the result of the present computational study of the S809 airfoil at the same operating conditions. the flow on the upper surface decelerates and merges with the flow from the lower surface. pressure coefficients etc. Moreover. 10 shows the pressure coefficient distribution for S809 airfoil at zero angle of attack and 105 Reynolds number. As the strength of the adverse pressure increases. Fig. as the flow accelerates over the airfoil. 9. the pressure rises. pressure coefficient becomes more negative. VI. The negative torque zone has been minimized by optimum cambering of airfoil profile at acceptable angle of attacks thus the positive torque that provides higher power can be obtained. 8. The comparison shows that the pressure coefficient distribution of the S809 profile obtained in the present study is in a good agreement with the experimental results at 105 Reynolds number. Since the VAWT is able to self-start at a lower wind speed. CONCLUSION Performing CFD calculations provide knowledge about the flow in all its details. As the velocity of air increases. the same operating conditions are taken for the present work .Fig.

USA. Amherst.” Journal of Aircraft 1997. Swansea University. 2011. “The S825 and S826 airfoils. “DMLHFLC (Dual mode linguistic hedge fuzzy logic controller) for an isolated wind-diesel hybrid power system with BES (battery energy storage) unit. Somers.” Progress in Energy and Combustion Science 2004. “CFD calculations of S809 aerodynamic characteristics. Selig MS.35:188-201. John Wiley and Sons Ltd. Wolfe WP. Venezuela„s Chavez extends electricity decree for 60 days. S817..” a subcontract report by National Renewable Energy Laboratory. “The S816. Mohamed A. “Simulations technique for the design of a vertical axis wind turbine device with experimental validation.B. LSS-S01/11. M. Mayer C. Renewable Energy. M. Guglielmo JJ.” Energy 2010.”South Korea. W. August 2005. A. Mohammad Asid Zullah. UK. Engineering and Technology 61 2012.. Singh. Mohamed. NREL/SR-500-36339.” a subcontract report by National Renewable Energy Laboratory. Hampsey M. Kandil. Rogers AL.D Thesis. “Design and performance of a double-pitch wind turbine with non-twisted blades. Marco Raciti Castelli. December 2004. Wood DH. “An assessment of wind energy potential as a power generation source in the capital of Iran.H.M. 2010.” Ph. “Aerodynamic analysis of different wind-turbine-blade profiles using finite-volume method. 34. 501-43. . S. Energy 47 (2012) 522-530. of Mechanical Power Engineering Dept. Messina M. Hsiao. Fazlizan .C. and S832 airfoils. Velusami S. NREL/SR-500-36344. Abbaszadeh R. Ansari MM. YoungHo Lee.” a subcontract report by National Renewable Energy Laboratory. F. and S818 airfoils.35:3827-37.” College of Engineering. “The starting behaviour of a small horizontal-axis wind turbine. Hamdy A. Lanzafame R. Somers. “Design of a low Reynolds number airfoil for small horizontal axis wind turbines.” 1997 [AIAA-97-0973].” Energy 2010. Bechly ME. “The design.M. Poh . “Design optimization of Savonius and wells turbine. simulation and testing of an urban vertical axis wind turbine with the omni-direction-guide-vane. 1994 – 1995.” University of Massachusetts. Hew . Germany. Sahin AD. “Performance investigation of H-rotor Darrieus turbine with new airfoil shapes. D. 34:1413-20.T.C. October 1991 – July 1992. 2013. Applied Energy. Univ.” Renewable Energy Lab. K. Keyhani A. Mohamed MH.P. S.. Pan . Mattaria.com. Ronit K. Giulia Simioni and Ernesto Benini. Rolland et al. “Wind energy explained. Ochs SS. NIT Durgapur . Khanali M. 2002. Manwell JF.” 2009. NREL/SR-500-36333.M.30. ISBN: 0-470-84612-7. “High-lift low Reynolds number airfoil design.” Renewable Energy 2001. Swansea SA2 8PP.” World Academy of Science. “Numerical Analysis of the Influence of Airfoil Asymmetry on VAWT Performance. Bloomberg. Rafiuddin Ahmed.REFERENCES [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] Cancel D. D. D. January 2005. “The S830. 4 1 A S E I 138 of 446 Copyright @ Department of Electrical Engineering. Somers. November 2001 – November 2002. “Progress and recent trends in wind energy. W. Sayed. of Magdeburg. Chong .22:4117. April 8. Varnamkhasti MG. Tehran. S831.” Malaysia. McGowan JG. Helwan University. Ahmed Shaltot. Faculty of Engineering.” Energy Conversion and Management 64 (2012) 541–550.

I. Indeed. This regime is experimentally the set of conditions. I. or refining of an unprepared microstructure during manufacturing of the product. and at what point during the deformation it ceases to be so. for the majority of alloys for maintenance of the superplastic condition it is necessary that the temperature (T) is in the order 0. Ti alloys require special techniques to manufacture the parts with complex shapes out of them.Yakushina Institute of Technical Education and Research Siksha ‘o’ Anusandhan University Bhubaneswar. Phenomenological scalar model with internal variable based on statistical description of microstructure is proposed and used for simulation of the high temperature deformation of Ti6Al-4V alloy. There is even an approach of interactive change of loading conditions to maintain the maximum rate sensitivity during the deformation process [6].com Advanced Forming Research Centre. Low strain-rates typical for SPD make the forming process time taking and inefficient. const In practice.g. superplasticity. due to the approximate nature of these constraints.Modelling of active transformation of microstructure of two-phase Ti alloys during hot working O. but currently it is often being done deliberately e. where optimum superplastic conditions are violated. However. The conditions required for such behavior of materials. Many of these technologies are based on superplastic and near-to-superplastic deforming. University of Strathclyde Renfrew. Going beyond the limits of the optimum SP happens occasionally. In these processes the transformation of microstructure of the material can be very significant and lead to changing the mechanical properties of the material along with deformation. are classified as "near-tosuperplastic deformation". with lowering the temperature to reduce power consumption ('low-temperature SP").yakushina@strath. Such processes. T . but the material deforms almost homogeneously without visible flow localizations or fractures till relatively high strains (compared to a conventional hightemperature deformation). microstructure transformation. Because of this proper description of the correlation between mechanical loading.melting temperature on absolute scale). its initial average grain size). B. moreover. with increasing strain rate to speed up the process at some stage (the so-called "high strain-rate SPD"). Inchinnan. simulation. NIT Durgapur . d 0 ) The change in microstructure of the material in this case is not significant.Pradhan E. taking into account only the state of the material before deformation (e. A INTRODUCTION I log m log T . and its transformation to a first approximation can be neglected. the material behaves like a viscous liquid and the flow stress can be considered to be a function of strain rate. For this reason. the design of technological processes using the phenomenon of superplasticity is generally done in such a way that the entire process of deformation is carried out in the "optimal superplasticity" regime. S E Here and are the dimensionless stress and strain rate respectively (hence it only refers to uniaxial loading). especially for large dimensions of the workpiece and to prepare the required original ultrafine microstructure uniform enough.4Tm (where Tm. Odisha. Thus even materials that are brittle in normal conditions can be deformed to hundreds and sometimes few thousand percent frequently without occurrence of localization of deformations which accelerates the process of failure.ac.K. it is economically inefficient.uk Abstract — Being very strong and brittle. changes in microstructure and mechanical behavior of material is required. in which parameter of strain-rate sensitivity [3-5] reaches the maximum. strain-rate lies in the range of 10-5 to 10-1 s-1 and microstructure of the material is fine-grained (the average size of grains ≤ 10μm) and globular. biplab93@gmail. UK evgenia. it is quite difficult and costly to implement the process of isothermal deformation for a long time. It is established [1. maintaining the optimal conditions of superplastic deformation (SPD) in industrial processes is sometimes difficult or even impossible and.2] that almost all polycrystalline materials and even ceramics are capable of showing under certain conditions a special kind of mechanical behavior called "superplasticity" (SP).g. are usually experimental and thus by nature empirical. Due to these reasons the cases of terminating the optimal regime are coming to the stage more and more frequently. Bylya. For example. it is practically impossible to say exactly when the deformation of the material becomes superplastic. 139 of 446 Copyright @ Department of Electrical Engineering. The problems of obtaining the parameters of the model and sensitivity of the model to the accuracy of this process are discussed.in.ac. temperature and initial grain size: 4 1 f (. India olgabylya@soauniversity.B. During optimum superplasticity. Keywords— Constitutive modelling.

has physical sense of the average grain size and is defined as a probabilistic mean of distribution: Fig. but also non-perfect ones like bimodal. 2. The choice of this parameter is difficult. For the ability of numerical simulation (especially FEM) of such deforming processes a special model is proposed. specifies the relative volume (Vi) occupied by the grains which belong to a size range di ( di 2 d di 2 ). each grain is identified only by one size . The distribution function p(d). It is assumed that the representative volume contains grains of various sizes. Here and hence all parameters of the model are dimensionless. stress. will have phenomenological character and will remain within continuum mechanics. For medium size grains deformation growth will be a dominating mechanism (depending on the stress and individual size of grain). if grains are small and globular. This model might approximately describe behavior of other Titanium Aluminates. a dimensionless combination Fig. For example. an additional characteristic variable will be assigned to a representative volume. A. d i Ds t RT where t=time. unlike a number of physical models [2]. 4. The designed model. strain and temperature. 10]. Description of the Model. II. This fact was proved by many experiments showing the dependence of mechanical behavior of the material on the history of deformation [2. The purpose of this model is to consider active transformation of the microstructure during deformation and its effect on the mechanical properties of the material.nature. This scheme is shown in the Fig. illustrating main mechanisms of microstructure transformations.5. Instantaneous change of this parameter will depend not only on current strain rate.2. T=850C.the characteristic size d. To take into account active transformation of a microstructure during deformation. The SEM snaps of Ti-6Al-4V alloy with initial Widmanstätten microstructure after deformation of =0. Here each physical law is applied only to such types of grains for which it is obtained [1]. This variable () is to be in charge of a current state of the microstructure. This approach will provide to the change of the average grain size some functional I i As it was mentioned above in the same point of material few physical mechanisms can operate simultaneously applied to different types of grains. So to begin with. 10 4 s 1 . For each grade of grains special micro-model should be used. Using the function of distribution provides us the ability to take all of them into account. NIT Durgapur . 140 of 446 Copyright @ Department of Electrical Engineering. the basic mechanism of their growth will be slow diffusion growth (governed by temperature). The main difference between these processes from the optimal SP is an active transformation of microstructure occurring during deformation which leads to a functional nature of constitutive models describing these processes. The slow temperature-dependent growth of small grains is described by the equation: 4 1 A S E CONSTITUTIVE MODELLING p( d i ) d i d i Q k exp . 1. the modification of the sizes of the grains of each type is calculated with different mathematical law. micro-level mechanisms working in these grains will be also different. To take it into account in the model. the direct kinetic equation for the average grain size will have very few physical sense. but also on the current state of the microstructure. but the mechanisms behind most Ti alloys are too different to be predicted by it. the grains which are bigger than some critical value (depending on the value of the strain energy) start to refine (by dislocation creep and recrystallization). The scheme of microstructure transformation used in the model. as microstructure transformations happen at different scale levels and are not trivial [11]. When accumulated energy of deformation increases. lamellar or Widmanstätten. The limitation d i Ds selects fine grains smaller than parameter Ds. we use the most obvious and easily experimentally measurable parameter . 9. If microstructure simultaneously contains fine grains as well as big ones and they have different nature. Mathematically above mentioned approach will be described by using statistical methods by introducing the function of distribution of the grain sizes.the average grain size. The parameter characterising a microstructure. If we are going to involve in our consideration not only uniform fine-grained equiaxial microstructure.

C1. That means that critical size Dcr is not a constant Fig. C3 – parameters.constants. for which: d i C1 4 C2 5 Q exp . or as some value depending on Dcr. 4. To complete the phenomenological model the relation between macroscopic parameters of loading such as stress and strain rate are required. Here the model of Perevesentsev is used [12]. d i Dcr t Grains grow only until they reach some critical size Dcr. The flow chat of the algorithm used for simulation of microstructure transformation. The size of the medium grains changes in the process of deformation because the Dcr decreases and the original sizes of the grains which undergo refinement also become smaller. energy of activation Q. As soon as the increments of the grain sizes are calculated. 4 1 Refinement of large grains can happen due to two basic physical mechanisms: rearrangement of dislocations into dislocation walls and recrystallization. then it grows and also divides a grain into few parts. It can be taken in the form of two serial 141 of 446 Copyright @ Department of Electrical Engineering. The example of the results of simulation value. The elementary mathematical model of refinement used here is based on the assumption that all grains after exceeding critical size break into few small and medium grains in proportion χ/(1-χ). D0 is fixed size of new recrystallized grains. as soon they exceed it refining process starts to these grains. A S E I Vref p( di ). t 1 C3 4 d i RT Ds d i Dcr where is dimensionless stress. grain is getting divided into two or several grains by them. and even relatively small grains can get recrystallized at large deformations. universal gas constant R. C2. that even long enough lamellar grains remain unbroken if deformation is small (e. In the first case when dislocations form a ‘wall’. but can be also considered as phenomenological and found from experiments. Mechanism of deformation growth dominating for the D d i Dcr medium size grains ( s ) was studied by researchers and a number of models are proposed. NIT Durgapur . Experimental results show. It was also observed that at high stresses refinement is more active than at low ones. Parameters D0 and Dm are mean sizes of small and medium grains gained in result of refinement correspondingly. 3. So Dm can be assumed to be equal to either total average grain size from the previous time step. is always almost the same. in the centre of the specimen under torsion). d i d i Dcr p( D0 ) : p( D0 ) Vref p( Dm ) : p( Dm ) (1 ) Vref Fig.g. where Vref is the total relative volume of “large” grains with the sizes greater than Dcr. In the second case nucleus of new grain arises in some part of the border of large grain. Q RT involves temperature Т (absolute scale). which in initial model [12] represent combinations of physical constants of a material. and material constant k. Here it is assumed to depend on accumulated strain energy: Dcr D * exp( q d ) where D* and q> 0 . distribution function is modified accordingly: p(d i ) t t d j d j ( d i ) p( d j ) .

To find required parameters. λ. and in our case several experiments with different strain rates and several experiments with different temperatures are required. The example of it is shown in the Fig. 1050 and 1100C). It is assumed that they are constants for a given material. To study the problem of finding m and Q and robustness of this process.basic elements: viscoelastic and "superplastic". The second problem is related with the need to extrapolate experimental data from few different experiments. A. 1000. Obtained results and their deviation from the original ones are also listed in Table I. Simulation Numerical simulation of the uniaxial (tensile) loading with the constant strain rate and under the constant temperature was done in MATLAB. 800. As a viscoelastic element the generalized body of Maxwell [2] modified for the temperature can be taken and superplastic one can be taken in standard form but involving our internal parameter η: e v sp Q exp E RT e v sp A 1m k Q exp RT where Е.410 0. const If experimental data is taken from few different tests. Such mechanical behavior is typical for Ti alloys deformed in nearto-superplastic regimes. THE DEPENDANCE OF STRAIN RATE SENSITIVITY PARAMETER TO THE ORDER OF POLYNOMIAL INTERPOLATION AND ACCURACY OF INPUT DATA Degree of Interpolation Quadratic Cubic 4th Degree 5th Degree 6th Degree Fig. So for all further calculations 4th order polynomial approximations were used.26 0. However for practical use of this model in real technological simulations the question of experimental obtaining parameters of the model must be solved.415 v1-v3 (SenSitivity) 0. These experiments were conducted for the Ti alloy Ti6Al4V with fine globular microstructure for 7 constant strain rates (0. 4. B. 3 and the results of simulation in the Fig.0003.385 0. log m log T .35 Value with randomerror in log(sigma) (v2) 0. The flow chart of the used algorithm is shown in the Fig. 900. reality both of them strongly depend on the strain rate and temperature.29 0. These results show that increasing the order of polynomial order from 4 to 6 increases the sensitivity to inaccuracies almost twice.18 v1-v2 (SenSitivity) 0. 142 of 446 Copyright @ Department of Electrical Engineering.001.28 0. const log R Q ( 1 / T ) m . Approximation of the experimental data with splines of different orders for differentiation required for obtaining strain rate sensitivity. 0. But higher order polynomial approximations are more sensitive to the inaccuracy of input data and results of calculations are thus less robust.35 0. 0.035 0. the amount of data is very limited and contains certain inaccuracies. randomized error of 10% was introduced to initial experimental data (to the value of stress). which may imply that highest-possible order (6th) would lead to the most accurate results. Original Value (v1) 0.273 0.211 0.36 0. 4 1 A FINDING THE PARAMETERS OF THE MODEL S E Two parameters of the model: m – strain rate sensitivity and Q – activation energy. There are two main problems with obtaining these parameters from experimental data.065 To check this. 10 and 100s-1) and 8 temperatures (750.01 0.01 0. 850. but in I TABLE I.30 0.17 Value with random error in log(sigma) (v3) 0.16 0. 950. the experimental data published in [16] were used. m and k are the constants of model. the approximation of available experimental data with polynomial splines of different order was made.027 0. It can be seen from the results of simulation that obtained stress-strain curve has hardening in the first part and softening with gradually decreased slope at the second one. obtained with interpolation of different order are given in the Table I. 5. NIT Durgapur . 1. The values of strain rate sensitivity m. 5.1. This range of conditions covers both superplastic as well as neighboring hot visco-plastic regimes of deforming.050 0. It can be seen that from 4th order the values do not change much.25 0.039 0.07 0. are the most important for the accuracy of calculations because the results of simulation are very sensitive to them. this depicts the actual accuracy limit of majority of high temperature mechanical tests.2 0. III.

20 466.79 373.12 0. Being based on statistical approach this model can deal with several micro-mechanisms at a time.08 471. /T 750 800 850 900 950 1000 1050 1100 0.23 0.18 330.02 305. However for other titanium alloys. For equation (8) and the total model to give correct results.67 349. having fundamentally different microstructure.95 223.16 0.25 0.16 0. The reason for higher inaccuracies pertaining to polynomials of greater degree is purely mathematical.55 266. analysis of proper Proposed model involves grain growth as well as refinement of the microstructure that gives ability to describe both hardening and softening of the material.12 0.08 0.14 100 0.51 493.26 0.27 416.41 0.17 0.88 670. the greater number of stationary points it will contain.31 0.03 0.21 0. In FEM simulation of real technological processes temperature and strain rate can vary from one point of workpiece to another.34 0.83 508.85 583.20 923. This feature is very important for two-phase alloys. 143 of 446 Copyright @ Department of Electrical Engineering.30 0.06 0.48 0.22 0.31 0.06 0. 7.20 0.23 0.52 279. e.67 522.01 0. The procedure of obtaining parameters of the model from experimental data and robustness of results is discussed and optimal method is chosen.97 275. intermelallic titanium aluminates. and the fields of these values are show in the Figures 6 and 7.76 697. Proposed model can be suitable for the family of + Ti alloys with similar type of microstructure and similar mechanisms of its transformation.05 0. Using these experimental data [16] and approximations which were found to be optimal.24 0.03 0.1 1 10 563.01 VALUES OF ACTIVATION ENERGY Q(IN KJ) /T 0.45 727.15 0.19 0.38 0. choice of physical models and possibilities of obtaining the parameters describing micro-mechanisms has to be investigated systematically. It was shown that these parameters significantly depend on these conditions and it must be taken into account in simulation.47 524. In this paper due to limited space these questions are not discussed.32 0.20 Contour plot for values of Q (activation energy) 0.77 366. The importance of consideration of microstructure in simulation of processes of hot forming including ones utilizing the superplasticity was analyzed and constitutive model able to take into account active transformation of microstructure was proposed.02 0.24 0.90 394.65 493. Nevertheless. VALUES OF STRAIN RATE SENSITIVITY M 0.32 0.23 0. TABLE II. especially for titanium alloys.15 1074.28 0.35 448.37 0. The transformation of the microstructure at every step of the simulation gives the ability to trace functional (history dependent) character of the process as well as the effect of initial microstructure on material properties. The higher the degree of a function.79 513. Obtained results clearly show that both parameters significantly depend on temperature and strain rate.0003 0.00 1143.69 583.47 486.g.67 CONCLUSION 4 1 A S E I IV. the choice of the microstructure parameters as well as kinetic models of their behavior must be different.06 0. parameters m and Q must be all the time verified and adjusted according to local conditions.27 0. Obtained results are given in the Tables II and III.14 0.16 1007.35 0.31 459.93 566.26 Other parameters of the model are also important but less critical because the model is much less sensitive to them than to the above described ones.59 408. which can have very different morphology of microstructure.40 0.04 0.11 0. 6. The values of strain rate sensitivity and activation energy were calculated for different temperatures and strain rates.33 732.1 1 10 0.26 801.000 3 0.05 647. Contour plot for values of m (strain rate sensitivity) Fig.03 0.27 0. TABLE III.18 0.83 723.01 750 800 850 900 950 1000 1050 1100 367.001 0.22 0.43 854.19 0. NIT Durgapur .48 1369.Fig.10 0.001 0.09 0.27 0.72 501.78 495.04 0.18 0. This can be done with the help of the fields shown in Figures 6 and 7. the values of strain rate sensitivity m and activation energy Q were calculated for different temperatures and strain rates.36 421.27 0.26 329.81 486.57 460.50 589.

A.. Maximum m superplasticity deformation for Ti-6Al-4V titanium alloy//J. A. Perevezentsev V. A. Mater. 278p. Eng.. Frazier W.192-193. 184-194.L (Eds. Ridley N. Stability criterion to optimise the process parametres and some applications to design industrial [11] [12] [13] [14] [15] [16] processes//Proceedings of the International Conference on Recent Advances in Material Processing Technology (RAMPT ‘ 05) / National Engineering College.W. P. A. Muravlev. V. Medeiros S. 44. Dynamic рекристаллизация and instability of a plastic current of superplastic alloys.14-22. and Enikeev F.U...301. Kovilpatti. 2001. A. The influence of simple and complex loading on structure changes in two-phase titanium alloy. K.A. M. P. Prasad Y. 4 1 A S E I 144 of 446 Copyright @ Department of Electrical Engineering. R. K. (2000).. V. V. 463. Tech.A. Bylya O. P. Y. A325. Warrendale. Raj A Model for the evolution of grain size distribution during superplastic deformation//Acta Metall. 1990. Medeiros.:Металлургия. pp. 3.. 2005. Proc.//PMM. [10] REFERENCES [1] [2] [3] [4] [5] [6] [7] [8] Kajbyshev O.. Larin S. N. 224-230. A. Scripta Materialia. V. 2007. 2007 V. and R. O.. 280p. Materials Science and Engineering: A.V. 2007. Sc. W. p. Eylon D.K. N. Sc.A. Eng. Srinivasan R.I. Bylja. & Prasad.. Vasin R...V. N. 6.. Lee W.U.. 1998. Ufa: Publishing house "Gilem". Fu M. Enikeev F. [9] ACKNOWLEDGMENT The work carried out is partly supported by Russian Fund of Basic Research (RFBR 11-08-00961). 1997. pp.. Sc. I.463.1-2. A.A. Enikeev F. Kim Y. Advances in the Science and Technology of Titanium Alloy Processing. S. Seshacharyulu.W. V. 430p. 447-456.112-125. 951-958. P. G. 1986 V. Bhattacharya S. Bate P. 1975.. Chaudhury P. P. 36-40. 34.U... TMS. Sivaprasad P.). 93-100. 284(1). 2001. 253-254. K. PA.R. Yang H.introduction in superplasticity mechanics. N.. Frazier. Effect of strain rate path on cavitation in superplastic aluminium alloy//Mater.. and Padmanabhan K. Microstructural mechanisms during hot working of commercial grade Ti-6Al-4V with lamellar starting structure//Mater.. GaoChao Wang. 9. On the measurement of strain-rate sensitivity for deformation mechanism in conventional and ultra-fine grain alloys//Mater. Superplastic flow: phenomenology and mechanics. Thirukkonda M.. 555-560 Seshacharyulu T. 29. T.S. M. 2002.K. Mechanical behaviour of titanium alloy Ti-6Al-4V with unprepared microstructure under jumpwise variations of the strain rate in the superplastic state//Mechanics of Solids. Venugopal S. G. Ermachenko.41-56. in: I... Vasin. Semiatin S. 1993 V. Guillard S. Berlin: Springer Verlag. Hot working of commercial Ti–6Al–4V with an equiaxed α–β microstructure: materials modeling considerations. 2009 V. NIT Durgapur .S.Simulation with the model using calculated parameters approximately matched the experimental results.. R. On the superplastic flow under non-uniform stress-strain state//Mater. B. P. S. V. Sc. V. India. P. Vasin R.. Vasin R.C. N. Mukheriee Superplastic behaviour in a two-phase TiAl alloy//Scripta Metallurgica et Materialia. ZhangB. Bania P. (1997). Ghosh A. A.. Ghosh A. Padmanabhan K.G. P. Eng. Eng..1403-1408. C. A.. P. & Karavaeva.A. Plasticity and superplasticity of metals..K. 36(8). Chistjakov.

7]. Proper locations and sizes of renewable DGs in the distribution network can cause maximum utilization with lowest capital cost. The output power generation of variable speed wind turbine is given by 145 of 446 Copyright @ Department of Electrical Engineering. loss cost and reliability cost. In the last decade many papers have appeared focusing the different planning techniques for location and size selection of DG [2-7]. The proposed planning method tries to evaluate appropriate locations and sizes of wind and solar power generation units in distribution network aiming maximization of benefit to cost ratio. A. PERFORMANCE MODELING OF RENEWABLE DG Power generation of wind turbine and PV array depend on resource such as wind speed. Performance modeling of wind and solar based DGs are portrayed in section II to measure the actual power generation of PV array and wind turbine in particular geographical location. load growth was not considered in their study. In [2]. load growth. In section IV.in Dept. Chanda Dept. solar radiation and ambient temperature. the authors have proposed a technique to locate dispatchable distributed generation optimally considering load growth. II. benefit to cost ratio. As wind and solar energies are promising renewable power generation sources and free from green house gas. This paper presents a novel approach of renewable DG planning considering variable power load model and load growth of a distribution network. In [8]. A very few number of papers have studied the effect of load model to accommodate DG optimally in distribution network [6. Reinforcement of DG largely depends on the type and demand of customer loads. India partha_kayal@yahoo. Planning of these renewable DGs are ardently essential to extract maximum potential in economic way. The objective function of the optimization problem is also analyzed in detail. optimization problem of DG planning is formulated. I. C. The proposed method is tested on a 28-bus Indian distribution network using discrete particle swarm optimization technique and its effectiveness is demonstrated. The paper has described load growth on five year basis on viewpoint of constructional cost. However. of Electrical Engineering Future Institute of Engineering and Management Kolkata. As resource data are depend on the geographical location.co. The paper is concerned with value based worth assessment (cash outflow and inflow) for DG planning satisfying voltage and power limit constraints. Wind Power Generation Wind power generation unit converts kinetic energy of wind into electricity. Some papers have investigated the DG planning problem with optimization approach. Sedighi et al [5] have proposed particle swarm optimization (PSO) based maximum size determination of DG to reduce power losses. Abri et al [3] have presented a mixed integer non-linear programming technique for sitting and sizing 4 1 A S E Keywords—distributed generation. Shibpur Howrah. K. INTRODUCTION I Worldwide the rapid depletion of fossil fuel resources has given momentum to the power system research with application of alternative energy sources. The cost function has been modified with a segmentation based power loss reduction cost calculation technique. Section III illustrates the variable power load model and effect of load growth on customer load demand. This paper highlights on allocation of multiple PV array and wind turbine units considering both load type and load growth. The rest of the paper is organized as follows. A value based DG planning method is described by Teng et al [4]. The authors have used genetic algorithm to optimize the cost function. NIT Durgapur .Value Based Planning of Renewable DGs in Distribution Network Incorporating Variable Power Load Model and Load Growth Partha Kayal Chetan Kumar. the authors have proposed loss sensitivity method to accommodate DG in distribution network. discrete particle swarm optimization of DG in distribution network. power generation model in terms of resource parameters have significant impact in proper planning of renewable DGs for a region. Finally some relevant conclusions are drawn in section VI. India Abstract—Presently one of the major challenges for the distribution companies is to supply cost effective and quality power to the customers utilizing distributed generation (DG). they can be alternative to the fossil fuels [1]. Photovoltaic (PV) array and/or wind power generation units are normally small scale power generation and can be directly connected with the distribution buses. They have suggested that better result can be obtained with optimization technique in comparison to analytical approaches. of Electrical Engineering Bengal Engineering and Science University. Simulation of the proposed method is carried out on a typical Indian distribution network and obtained results are discussed in section V. The paper lack of illustrating the customer load type and annual load growth scenario.

**constant power load model may misguide the DG selection for
**

distribution system.

⎧ 0 vw < vcin or

vw > vcout

⎪

v −v

⎪

Pw = ⎨ Prated w cin vcin ≤ vw ≤ vN

(1)

vN − vcin

⎪

⎪⎩

Prated vN ≤ vw ≤ vcout

vcin , vcout , vN are cut-in speed, cut-out speed and nominal

speed of wind turbine respectively; vw is the average wind

speed at hub height of wind turbine; Prated is the rated output

power of turbine.

**According to type of customer, loads are categorized into
**

residential, commercial and industrial load. Voltage

dependencies of different class loads are different. The variable

power load model can be formulated as follows.

⎛ TN ⎞

G ⎜ ⎟

1 + c2 * ln N ⎝ Ta ⎠

Ga

VNoc

c3

**introduced to show non-linear relationship between solar
**

irradiance, photo-current and cell temperature [9].

III.

VARIABLE POWER LOAD MODEL AND LOAD GROWTH

**According to IEEE report [10] loads are dependent on
**

voltage magnitude of buses. As a result, consideration of only

P0,i , Q0,i and V0,i are nominal

x y

PD' ,i = (1 + 100

) * PD ,i

'

x y

QD ,i = (1 + 100 ) * QD ,i

TABLE I.

4

1

yth

(8)

VALUES OF EXPONENTS USED FOR DIFFERENT TYPES OF LOAD

A

S

E

I

(7)

**and reactive load for x% of annual load growth at bus-i at
**

year becomes

(4)

⎛G ⎞

I sc = I Nsc ⎜ a ⎟

(5)

⎝ GN ⎠

GN and Ga are the nominal and actual solar irradiance on

module; TN and Ta are nominal and actual PV module

temperature respectively; VNoc and I Nsc are nominal the open

circuit voltage and short circuit current of PV module; C1 ,

C2 and C3 are the three different constant which are

β

**Load growth is a continual phenomenon of distribution
**

network which is unavoidable. The modified demand of active

Pmd = FF *Voc * I sc

(3)

Voc , I sc and FF are the open circuit voltage, short circuit

Voc =

⎞

⎟⎟

⎠

**active load, reactive load and voltage magnitude at bus-i;
**

α and β are active and reactive power exponents. Active and

reactive power exponents are different for different types of

load and shown in TABLE I.

**Where Pmd is the maximum electrical power generated by
**

PV module which is formulated as

c1

⎛V

QD ,i = Q0,i * ⎜ i

⎜V

⎝ 0,i

(6)

voltage magnitude at bus-i;

(2)

**current and fill factor (FF) of PV module. FF is a design
**

parameter and depends on V-I characteristics of the module.

Voc , I sc are the function of solar irradiance and PV module

temperature; and these are obtained as follows

⎞

⎟⎟

⎠

PD ,i , QD ,i and Vi are active load, reactive load and

**B. Solar Power Generation
**

Solar radiation and ambient temperature are the main

governors for power generation of PV module. PV module

cannot generate bulk amount of electrical power. So, large

numbers of PV modules are connected in series and parallel to

design PV array. Series and parallel connection of PV modules

boost up voltage and current to tailor PV array output. For a

PV array consist of N S × N P PV modules, maximum output

power can be calculated as

Ppv = NS N P Pmd

α

⎛V

PD ,i = P0,i * ⎜ i

⎜V

⎝ 0,i

Load type

α

β

Residential

0.92

4.04

Commercial

1.51

3.40

Industrial

0.18

6.0

IV.

PLANNING PROBLEM FORMULATION

**With growing load demand in the distribution network, it
**

provides potential scope for research in terms of planning of

DG in distribution network to meet the demand with the

present infrastructure.

A. Power Loss Reduction Cost

Significant cost of electrical energy is lost in network

power loss. Proper location and size of DG in distribution

network can be helpful to reduce network power loss and save

energy cost as well. In this section segmentation based power

loss reduction cost (PLRC) calculation method is presented.

Network power loss is a function of customer load demand.

As load growth modifies the power demand, network power

loss would become different in future years. So, it would be

more convenient to calculate the PLRC by segmented the total

planning period into a number of single years. After computing

the PLRC for each year, they are summed up to generate total

PLRC for the planning horizon.

146 of 446

Copyright @ Department of Electrical Engineering, NIT Durgapur

**Power loss reduction due to allocation of renewable DGs
**

for

**includes cost saving due to reduction of power
**

consumption from main substation and power loss reduction.

Total benefit that can be achieved from operation of

distribution network with renewable DGs is given by

BenefitrenwDG

th

i year can be calculated as

i

i

i

ΔPlossrenwDG

= ( PlosswithoutDG

− PlosswithDG

) *8760 (9)

**Cumulative Present Value (CPV) converts the future cost
**

into present value. Interest rate, inflation rate, and planning

period have taken into account to calculate CPV. It can be

determined as follows

PV =

IF )

(1+ 100

N

**Benefit renwDG = {( ∑ ∑ j∈type PrenwDG ,ij * ni * li )}* C hr * 8760 * CPVt m + PLRC
**

i=2

(10)

(16)

From (16), it is seen that more the value of BCRrenwDG , utility

will be more benefitted.

d )

(1+ 100

C. Optimization Problem

Maximization of BCR is considered as the objective of this

study. As BCR of wind and solar based DG planning in

distribution network is function of locations and sizes of DGs,

the objective function can be formulated as

**Where PV is the present value of cost; IF and d are the
**

percentage of inflation rate and discount rate per year

respectively; CPV for total number of year t would be

CPVt i =

( PV − PV t +1 )

(1− PV )

(11)

OF = max { BCR ( PrewDG ,i , PD' ,i , QD' ,i )}

ith year can be determined as

CPV for

(17)

CPV i = CPVt i − CPVt i −1

**The optimization algorithm should satisfy some equality
**

and inequality constraints as depicted below.

be

**1. Power balance equation of total system is always
**

maintained and represented as

(12)

Power loss reduction cost for total planning period m would

m

i

PLRC = ∑ (ΔPlossrenwDG

⋅ CPV i ) * Chr

i =1

Chr is the cost of electricity per kilowatt-hour.

I

N

N

i=2

Cost renwDG = ∑ ∑ j∈type ICij * ni * li + ( ∑ ∑ j∈type OMCij * PrenwDG ,ij * ni * li ) * CPVt m

Where

ICij

i =2

(15)

and OMCij are investment cost; and operating and

**maintenance cost of type-j renewable DG at bus-i respectively.
**

ni is the number of DG unit that can be connected at bus-i. li is

location variable at bus-i. It is to be noted that value li equals to

0 or 1 indicate DG is connected or not connected at bus-i.

PrenwDG ,ij is the power generated by type-j DG at bus-i. N is the

number of buses in the network and it is considered that

substation is connected at bus-1. As operation and maintenance

cost of renewable DGs are future cost, it is required to include

CPV.

4

1

i=2

(18)

(19)

**It is considered that DGs are operated at unity power factor.
**

PGSS and QGSS are active and reactive power fed by substation

respectively.

A

S

E

**Cost renwDG is composed of investment, operating and
**

maintenance cost. Investment cost includes cost of the DG unit,

DG site preparation, electric system interface, SCADA system,

protection system etc. Operating and maintenance cost includes

labour cost for service, tax and preventive maintenance cost.

Total cost of renewable DGs due to install, operation and

maintenance is given by

i=2

i =2

N

(14)

BenefitrenwDG

CostrenwDG

N

QGSS − ∑ QD' ,i − QL = 0

**B. Benefit to Cost Ratio
**

Benefit to cost ratio (BCR) indicates the economical benefit

that can be realized with respect to cost for implementation of

the project. BCR for renewable DG planning is represented as

BCRrenwDG =

N

PGSS + ∑ P renwDG ,i *ni * li − ∑ PD' ,i − Ploss = 0

(13)

2. Generation limit constraint at bus-i:

PrenwDG ,i * ni ,min ≤ PrenwDG ,i * ni ≤ PrenwDG ,i * ni ,max

(20)

ni ,min and ni ,max are minimum and maximum value of size

**variable respectively at bus-i.
**

3. Bus voltage tolerance constraint at bus-i:

0.95 p.u ≤ Vi ≤ 1.05 p.u

Vi is the magnitude of bus voltage at bus-i.

(21)

**4. Line power flow constraint for line connecting between
**

bus-i and bus-j:

Sij ≤ Sijmax

S

(21)

max

ij

**is the maximum power flow through the line between
**

bus-i and bus-j.

5. DG penetration constraint in the network

N

∑P

renwDG ,i

i =2

N

⋅ ni ⋅ li ≤ ∑ PD' ,i

i =2

V.

(22)

SIMULATION RESULTS AND DISCUSSION

**The proposed planning method is tested on a 28-bus rural
**

distribution network situated in the district of South 24

147 of 446

Copyright @ Department of Electrical Engineering, NIT Durgapur

**Parganas in the state of West Bengal, India. The structure of
**

the distribution network is radial [7] in nature with bus-1 as

substation as shown in Fig. 1. It has total load demand of

714+j622 kVA at base configuration. Some of buses namely

bus-7, 8, 9, 10, 22, 23, 24, 25, 26 in the network suffer with

low voltage problem and minimum bus voltage of the network

is obtained as 0.927 p.u at bus-26. So, voltage magnitudes of

low voltage buses can be enhanced by selecting proper location

and size of renewable DGs. Discrete PSO [11] is used to solve

the optimization problem in MATLAB environment. Variables

of PSO are rounded off by round function in MATLAB to

construct DPSO. Number of particle and maximum number of

iteration were set 100 and 200 respectively.

**Optimal location and size of wind turbines evaluated by
**

DPSO for maximum value of BCR are tabulated in TABLE III.

The obtained results are compared with constant power load

model. Constant power load model uses the base value of load

and voltage effect is not reflected on load demand. The overall

load growth for the region is measured 2% per year. Location

and size of wind turbines with consideration of constant power

load model in planning is also simulated and obtained results

are displayed in TABLE III. BCR value is obtained 0.753 in

case of variable power model and 0.641 for constant power

load model. Experimental results show that location and sizes

of wind turbines are also different in two cases. The constant

power load model can not reflect the customer load demand

properly. As a result optimal location and sizes of wind

turbines differ from case of variable power load model.

TABLE III.

LOCATION AND SIZE OF WIND TURBINES OBTAINED THROUGH

DG PLANNING WITH VARIABLE AND CONSTANT POWER LOAD

Variable power

load model

Constant power

load model

Fig. 1. Single line diagram of 28-bus Indian distribution network with bus-1

as substation

A

S

E

**the complexity. For total planning period of ten years IF, d and
**

Chr were taken 5%, 3% and Rs 5.8/kW-hr respectively. Mean

load growth per annum in case of industrial, commercial and

residential type loads has been calculated 1.3%, 2.1% and 3.8%

respectively.

I

**A. Placement and sizing of wind turbine
**

In South 24 parganas, annual average wind speed is 3.52

m/s at 30m height from the level of ground. Power generation

of wind turbine (specification given in TABLE II) is calculated

as 44 kW.

TABLE II.

Value

Rated output power, Prated

500 kW

Cut-in-speed, vcin

3 m/s

Nominal wind speed, vN

9 m/s

Cut-out speed,

vcout

25 m/s

Investment cost

80000 Rs/kW

Operating and maintenance cost

1600 Rs/kW-year

1

Size (kW)

44

bus-11

5

220

bus-18

5

220

bus-25

4

176

bus-27

1

44

bus-11

3

132

bus-18

3

132

bus-19

3

132

bus-25

3

132

bus-27

5

220

1

without DG

with wind turbine before load growth

with wind turbine after nine years of load growth

0.99

0.98

0.97

0.96

0.95

0.94

0.93

0.92

SPECIFICATION OF WIND TURBINE

Attribute

Number of unit

bus-5

4

1

Voltage magnitude

**Newton-Raphson power flow solution was used in the main
**

function of PSO to calculate network power loss, voltage

magnitude of buses and line power flow for each particle.

ni ,min and ni ,max were considered 1 and 6 respectively to defy

Location

0

5

10

15

Bus number

20

25

30

**Fig. 2. Comparative study of voltage profile of buses in the distribution
**

network with wind turbines at starting and ending of project

**Voltage magnitudes of the buses for allocation of wind
**

turbines in the network are illustrated in Fig. 2. The figure

clarifies about improvement of bus voltage magnitudes with

planned deployment of wind turbines in the network.

148 of 446

Copyright @ Department of Electrical Engineering, NIT Durgapur

**load. Planning of PV arrays with constant power load is also
**

examined and solution generated by DPSO is tabulated in

TABLE V. BCR value is obtained 0.389 in this case.

**B. Placement and sizing of PV array
**

With annual average solar irradiation of 192.917

and annual average ambient temperature of 26

power of PV array is determined as 16 kW.

W /m

C , output

1

SPECIFICATION OF PV MODULE

parameter

Value

Maximum output, Pmax

100 W

Nominal solar intensity on module, GN

1000 W / m

VNoc

Nominal short circuit current at I Nsc

6.5 A

Fill Factor, FF

0.985

Nominal open circuit voltage at

Nominal cell operating temperature,

0.98

2

21 V

TN

0.93

1000 Rs/kW-year

0.92

0

DISTRIBUTION NETWORK FOR VARIABLE AND CONSTANT POWER LOAD

Location

Number of unit

Size (kW)

bus-8

4

64

bus-9

6

96

bus-10

4

64

bus-11

6

96

bus-20

3

48

bus-21

4

64

bus-25

5

80

bus-26

4

64

bus-28

1

bus-4

1

bus-6

1

bus-8

bus-9

6

A

16

16

96

96

96

2

32

bus-14

3

48

bus-15

3

48

bus-16

3

48

bus-22

6

96

bus-24

3

48

bus-25

5

80

15

Bus number

20

25

30

4

1

16

6

bus-12

10

**Improvement of voltage magnitudes at distribution buses
**

by allocating PV arrays at appropriate locations in the

distribution network is shown in Fig. 3. Simulation results

show that load growths at distribution load points are

responsible for lower down the voltage magnitudes at buses

after the certain planning period. Thus, load growth has

significant impact on DG planning.

S

E

I

6

5

**Fig. 3. Comparative study of voltage profile of buses in the distribution
**

network with PV arrays at starting and ending of project

LOCATION AND SIZE OF PV ARRAYS OBTAINED IN

bus-7

0.95

25 C

Operating and maintenance cost

Constant power

load model

0.96

0.94

200000 Rs/kW

Variable power

load model

0.97

o

Investment cost

TABLE V.

without DG

With PV arrays before load growth

With PV arrays after nine years of load growth

0.99

Voltage magnitude (p.u)

TABLE IV.

o

2

VI.

CONCLUSION

**In this paper, a new renewable DG planning methodology
**

is proposed considering load growth. Load types are

categorized on the basis of voltage dependency and included in

the study. The method evaluates the maximum value of benefit

to cost ratio using segmentation based calculation of power

loss reduction cost. The methodology generates potential

solution for selection of proper locations and sizes multiple

wind turbine and PV array units in distribution network. The

obtained results can support the load growth keeping voltage

magnitude of buses and line loading limit within desired value.

Comparative study shows that consideration of constant power

load model is no further a good choice for DG planning.

Modeling of wind and solar based DGs using location specific

resource parameters helps in quantifying accurate measurement

of power generation in planning process. All these outcomes

reveal that the lowest cost planning is possible by proposed

comprehensive planning technique.

REFERENCES

[1]

**50 × 20 PV modules were used to design PV array. The
**

specification of PV module is given in TABLE IV. The

proposed planning technique identifies the appropriate location

and size of PV arrays in distribution network for maximum

value of BCR which is shown in TABLE V. DPSO results the

value of BCR 0.457 in case of planning with variable power

[2]

[3]

**Z. W. Geem, “Size optimization for a hybrid photovoltaic-wind energy
**

system”, Int. J. of Electric Power and Energy Systems, vol. 42, pp. 448451, 2012.

N. Acharya, P. Mahat, N. Mithulananthan, “An analytical approach for

DG allocation in primary distribution network”, Int. J. of Electric Power

and Energy Systems, vol. 28, pp. 669-678, 2006.

R. S. A. Abri, E. F. El-Saadany, Y. M. Atwa, “Optimal placement and

sizing method to improve the voltage stability margin in a distribution

system using distributed generation”, IEEE Trans. Power System, vol.

28, pp. 326-334, 2013.

149 of 446

Copyright @ Department of Electrical Engineering, NIT Durgapur

[4]

**J. H. Teng, Y. H. Liu, C. Y. Chen, C. F. Chen, “Value based distributed
**

generator placements for service quality improvements”, Int. J. of

Electric Power and Energy Systems, vol. 29, pp. 268-274, 2007.

[5] M. Sedighi, A. Igderi, A. Parastar, “Sitting and Sizing of Distributed

Generation in Distribution Network to Improve of Several Parameters

by PSO algorithm”, IEEE 9th International Power and Energy

Conference , Singapore, pp. 1083-1087, October, 2010.

[6] D. Singh, D. Singh, K. S. Verma, “Multiple optimization for DG

planning with load models”, IEEE Trans. Power System, vol. 24, pp.

427-436, 2009.

[7] R. K. Singh, S. K. Goswami, “Multi-objective optimization of

distributed generation planning using impact indices and trade-off

technique”, Electric Power Comp. and Systems, vol. 39, pp. 1175-1190,

2011.

[8] I. Zairi, G. Ledwitch, A. Ghosh, “Optimal distribution network

reinforcement considering load growth, line loss and reliability”, IEEE

Trans. Power System, vol. 28, no. 2, pp. 587-597, 2013.

[9] W. Zhou, H. Yang, Z. Fang, “A novel model for photovoltaic array

performance prediction”, Applied Energy, vol. 84, pp. 1187-1198, 2007.

[10] IEEE Task Force on Load Representation for Dynamic Performance,

Load representation for dynamic performance analysis, IEEE Trans.

Power System, vol. 8, no. 3, pp. 472-482, 1993.

[11] M. Clerc, “Discrete particle swarm optimization, illustrated by the

travelling salesman problem”, In New optimization techniques in

engineering, pp. 219-239. Springer Berlin Heidelberg, 2004.

4

1

A

S

E

I

150 of 446

Copyright @ Department of Electrical Engineering, NIT Durgapur

**Phase Angle Measurement using PIC
**

Microcontroller with Higher Accuracy

Khairul Alam2

Research and Development Wing

Advance Integrated Tech. Lab

Kolkata, West Bengal, India

system.khairul@gmail.com

Tanmoy Chakraborty1*

Department of Computer Science and Engineering

Saroj Mohan Institute of Technology

Guptipara, Hooghly, West Bengal, India

tanmoy.chakraborty@yahoo.com

Satadal Mal3

Department of Electrical Engineering

Kalyani Government Engineering College

Kalyani, Nadia, West Bengal, India

satadal14@gmail.com

Utpal Biswas4

Department of Computer Science and Engineering

University of Kalyani

Kalyani, Nadia, West Bengal, India

utpal01in@yahoo.com

**Abstract — The importance of measuring phase angle
**

between two sinusoidal waves of same frequency is useful for

control system, robotics, communication, medical instruments,

power system etc. The power is single/three phase system cannot

be measured without measuring the phase angle between voltage

and current using a microcontroller, a boon of very large scale

integrated technology. This is a digital method of measuring

phase angle and displaying lagging and leading state between the

waves on a liquid crystal display. Development of programmes

related to this phase angle measurement in high level language,

C, converting into hex/binary level and integrated into the flash

ROM of microcontroller has been tested in the laboratory.

4

1

**Fig. 1(a) Phase angle between V1 and V2; V2 is lagging V1 by an angle Ø
**

Fig. 1(b) Phase angle between V1 and V2; V2 is leading V1 by an angle Ø

A

S

E

I

**Keywords— Zero Crossing Detector; Phase Angle; PIC
**

Microcontroller; Flash ROM; Liquid Crystal Display.

INTRODUCTION

A sinusoidal instantaneous voltage wave (V1) can be

expressed as V1=Vmax sin ωt

(i)

ω=2πf, f=50Hz

Where Vmax = maximum/ peak voltage in volts.

ω= angular frequency in rad/sec.

Another sinusoidal voltage (V2) can be expressed as

V2=Vmax (sin ωt+ Ø)

(ii)

Where Vmax = maximum/ peak voltage in volts.

ω= angular frequency in rad/sec.

The V2 is lagging, shown in fig. 1(a) and that is leading shown

in fig. 1(b).

**To measure the phase angle digitally between V1 and V2, each
**

wave is converted into square wave through Zero Crossing

Detector, and then its level is restricted to 5V and passed

through an XOR gate. The output of XOR gate will generate

the ON period proportional to the phase angle between waves.

The ON period is measured by the PIC microcontroller. The

state of lagging and leading is detected and displayed on LCD

using the software developed.

RELATED WORK

In the paper [1] very simple method is used in phase

measurement error compensation technique for automation.

The paper [2] deals with measurements of phase angle for fault

tolerant systems using software methodology. Abdulrahaman

151 of 446

Copyright @ Department of Electrical Engineering, NIT Durgapur

**K. Al-Ali et.al. [3] worked on microcontroller based phase angle
**

measurement and correcting phase angle using voltage

controlled capacitor. In paper [4] Weiguo Que et.al. measured

phase angle between voltage and current for polymer

insulators. The paper [5] has developed a simple algorithm for

16 bit microcontroller, its performance simulation and

presentation of practical results. The paper [6] measured phase

angle in electric power systems using the discrete Fourier

Transformation algorithm.

An efficient electronic system based on a PIC microcontroller

has been developed to measure the phase angle between two

sinusoidal waves. The microcontroller through its hardware

and software programmes determines the phase angle in

degree and also shows lagging and leading with respect to the

reference. The system is also low cost and user friendly. The

display system of lagging and leading is an additional

attribute.

**Fig.3 (a), Fig.3 (b) and Fig.3(c) show the case of lagging signals and its phase
**

difference.

PROPOSED WORK

The frequency sample is collected by a transformer

connected to power supply of 50 Hz. and allowed to pass

through a Zero Crossing Detector (ZCD), converted into a

square wave signal say V1. The square wave is supplied to a

RC Phase Shift circuit for simulating Leading and lagging the

signal V2. Depending on the RC network value, the time

duration will change accordingly. This change is actually the

phase shift of the V1 signal which is representing as V2. Fig.2

(a) shows the reference signal V1 and V2 being the leading

signal.

Fig.2 (a): shows the reference signal V1 and V2 being the leading signal.

Fig.2 (b): with respect to V1 after converting it into square wave form of 5

volt.

Fig.2(c): shows the XOR output being the phase difference between the two

signals.

**The XOR output is fed to bit 4 of port A (PA.4) to distinguish
**

the lagging/ leading of V2 and to measure the period for which

it is lagging or leading. If the output at time t=0 is logical zero,

the signal V2 is leading else lagging. This will help to display

lagging or leading condition of the signal.Fig:4 shows the data

flow diagram of the system designed.

4

1

Fig. 4: Data Flow Diagram

A

S

E

I

152 of 446

Copyright @ Department of Electrical Engineering, NIT Durgapur

**Fig. 6: The setup of Phase Angle measuring system showing the leading phase
**

angle

**A counter initialized is started at the instant XOR output is
**

logic one, incremented by 400 ns pulse and stopped when

XOR output is logic zero. The count value incremented

multiplied by 400 ns presents the ON time period of XOR

output signal. This time period calibrated in terms of phase

angle will provide phase difference between the two signals.

For a 50 Hz signal, time taken for 90 degree phase

difference is 5 milli-seconds. With this reference, the time

period measured is calibrated in degrees. The

microcontroller will calculate this angle or phase difference

and display both degree and leading/lagging condition on

16X2 Liquid Crystal Display (LCD).

The

overall

system overview of the instrument designed is shown in

Fig.5.

Fig.5: System Overview

**Fig. 7: The setup of Phase Angle measuring system showing the lagging phase
**

angle

4

1

A

S

E

I

**In order to display, the decimal value on LCD, the difference
**

count value in hex is converted into decimal value, and then

corresponding to each decimal digit is converted into ascii and

send it to Liquid Crystal Display (LCD).Fig. 5 shows the block

diagram of the controller used for the purpose along with the

personal computer connected through universal synchronous

asynchronous receiver transmitter (USART) and serial

communication port for displaying the result in computer

terminal as well.

RESULTS

Table.1: Results of Phase Angle measurement and Error percentage

R in

KΩ

C

in µF

Degree

Measured

Time

in Sec.

Simulated

Time

Delay

t=1.1RC(ms)

Simulated

Time

Delay

(ms)

Degree

Measured

% Error

(Degree)

1

0.1

0.00011

0.11

1.98

0.109

1.962

-0.02

4.7

0.1

0.000517

0.517

9.306

0.512

9.216

-0.081

10

0.1

0.0011

1.1

19.8

1.109

19.962

0.1458

14

0.1

0.00154

1.54

27.72

1.525

27.45

-0.243

22

0.1

0.00242

2,42

43.56

2.41

43.38

-0.162

30

0.1

0.0033

3.3

59.4

3.32

59.76

0.324

33

0.1

0.00363

3.63

65.34

3.6

64.8

-0.486

42

0.1

0.00462

4.62

83.16

4.6

82.8

-0.324

43

0.1

0.00473

4.73

85.14

4.7

84.6

-0.486

153 of 446

Copyright @ Department of Electrical Engineering, NIT Durgapur

**extending the infrastructural facilities without which it was
**

difficult to implement the system.

**In order to verify the accuracy level of the instrument,
**

measuring phase angle, the simulated phase angle is measured

using our developed instrument and calculated the difference

between the measured value and calculated value used for

simulation. The result is given in the Table 1. And the

percentage error versus phase angle is shown in Fig: 8.

References

Fig.8: Percentage error versus phase angle graph

**[1] M.F. WAGDY, M.S.P. LUCAS “A PHASE-MEASUREMENT ERROR
**

COMPENSATION TECHNIQUE SUITABLE FOR AUTOMATION” IEEE TRANSACTIONS

ON INSTRUMENTATION AND MEASUREMENT, IM-35 VOL.1, MARCH 1986.

[2] W.V. SUBBARAO, V.ORDAX “DIGITAL MEASUREMENT OF PHASE

ANGLES FOR FAULT TOLERANT SYSTEM” SOUTHEASTCON '93,

PROCEEDINGS, IEEE, ISBN 0-7803-1257-0, 4-7 APR 1993.

[3] A.K. AI-AIi, M.T. Abuelma’atti, I.Hussain “MICROCONTROLLERBASED

PHASE-ANGLE

MEASUREMENT

AND

CORRECTION

TECHNIQUE” Industrial Automation and Control: Emerging Technologies,

1995., International IEEE/IAS Conference, 22-27 May 1995.

[4] W.Que, E.P.Casale, S.A.Sebo “Voltage-current phase angle

measurements during aging tests of polymer insulators” 2002 Annual Report

IEEE Conference on Electrical Insulation and Dielectric Phenomena.

[5] I. Adam, A. Mohamed, H. Sanusi “Simple Phase Angle Measurement of

Two Periodic Signals”, Proceedings of 2009 IEEE Students Conference on

Research and Development (SCOReD 2009), 16-18 Nov. 2009, UPM Serdang,

Malaysia.

[6] T. Xia, Y. Liu “Single-Phase Phase Angle Measurements in Electric

Power Systems”, IEEE Transactions Power Systems, Vol.25, No.2, MAY

2010.

4

1

**[7] Hewlett Packard, Inc., "Gain-phase meter 3575A," Operating and
**

Service Manual, June 1976.

**It shows that the maximum percentage of error from the graph
**

is -0.8%. The accuracy level belongs to Class 0.5. In the

programmer, the 400 ns interval sample pulse is used. The 100

ns interval sample pulse can easily be used. This will

definitely improve the accuracy class with this same hardware

system.

CONCLUSION

A

S

E

I

**The contribution to this field of measurement is the
**

implementation of the concept available using the latest

development of programmable devices (PIC microcontroller).

The developed microcontroller based advance system has been

implemented on a printed circuit board with a microcontroller,

LCD display system and some other interfacing components.

Its cost is about Rs.2000.00 only, which shows very cost

effective and efficient displaying system. This system will be

very useful in power system when the phase-angle needs to be

measured and monitored.

[8] Vicente Ordax Jr., "Prediction of component or system failure with multiphase power simulation" Florida International University MS thesis

dissertation, August, 1990.

[9] B. Djokic, and E. So,‟„Phase Measment of Distorted Periodic Signals

Based on Nonsynchronous Digital Filtering”, IEEE Tram. Inrmmmfotion and

Meanrremenls, Vol. 50, No. 4, pp.864-867, August, 2001.

[10] Al-Ali, A.K. Abuelma’atti, M.T.Hussain, I. “Microcontroller based phase

angle measurement and correction technique”, International IEEE/IAS

Conference on Industrial Automation and Control: Emerging Technologies,

1995, pp. 569-571, May 1995.

[11] R. Michelliti, “Phase Angle Measurement Between Two Sinusoidal

Signals”, IEEE Transactions on Instrument and Measurement, Vol.30, No. 1,

pp.40-42, February 1991.

[12] P. Bonanomi, “Phase angle measurements with synchronized clocks—

Principle and applications,” IEEE Trans. Power App. Syst.,

vol. PAS-100, no. 12, pp. 5036–5043, Dec. 1981.

[13] J. Chen, “Accurate frequency estimation with phasor angles,” M.Sc.

thesis, Bradley Dept. Elect. Comput. Eng., Virginia Polytechnic Inst. State

Univ., Blacksburg, VA, 1994.

[14] A. G. Phadke, “Synchronized phasor measurements in power systems,”

IEEE Comput. Appl. Power, vol. 6, no. 2, pp. 10–15, Apr. 1993.

ACKNOWLEDGMENT

The authors are really thankful to the Department of Electrical

Engineering, Kalyani Government Engineering College,

Kalyani, Nadia and the Department of Computer Science and

Engineering, University of Kalyani, Kalyani, Nadia for

**[15] Z. Zhong, C. Xu, B. J. Billian, L. Zhang, S. S. Tsai, R. W. Conners, V.A.
**

Centeno, A. G. Phadke, and Y. Liu, “Power system Frequency Monitoring

Network (FNET) implementation,” IEEE Trans. Power Syst., vol. 20, no. 4,

pp. 1914–1921, Nov. 2005.

154 of 446

Copyright @ Department of Electrical Engineering, NIT Durgapur

I. safety analysis of drinking water. Nadia. however. NIT Durgapur . India tanmoy. Lab Kolkata. MOSFET. Otherwise. Depending upon the water level of both the tanks. Automation.com Abstract — Automatic Electronic Water Level Management System (AEWLMS) of a multistoried residential building is very useful in urban area. digital latch etc. West Bengal. the pump/motor will run for the excess period during which the water will be wasted due to overflow and the energy will be consumed simultaneously. It is realized using PIC microcontroller based electronic design and tested in laboratory. it is very difficult for a 4 1 person to be very active and sincere for all the time during discharge and loading of upper roof top tank. the motor will automatically started for pumping water. PROPOSED WORK It has been assumed that there is an underground reservoir in the ground floor and a tank at the top of a multistoried building. In this paper the proposed system is designed by comparator. Another one is on microcontroller based drip irrigation system. Operational Amplifier.com Tanmoy Chakraborty1* Department of Computer Science and Engineering Saroj Mohan Institute of Technology Guptipara. the motor will be in ON or OFF state to maintain the water level of roof top tank minimum level to maximum level through constant monitoring. Water will be automatically store in the reservoir as per the service provided by the local municipal authority. amplifiers. One pump is coupled with a motor. India system. West Bengal. The water will be consumed by the residents of that building.Automatic Electronic Water Level Management System using PIC Microcontroller Khairul Alam2 Research and Development Wing Advance Integrated Tech. This paper deals with two reservoirs. Energy Efficient. II. this design will be very effective for the society as a whole. On the other hand. Our 155 of 446 Copyright @ Department of Electrical Engineering. INTRODUCTION This sort of water level management appears to be very simple but very useful. paper [1] deals with water level sensing and controlling for agriculture. A person if assigned for the purpose has to be very sincere in operating the pump/motor switching on or off. one is in the roof top and another one is underground. West Bengal.com Utpal Biswas4 Department of Computer Science and Engineering University of Kalyani Kalyani. subject to the minimum maintenance of water level in the underground reservoir. These two tanks are connected by a pump operated by a single phase/three phase induction motor. Water Level Management. When that water level reaches to a minimum level. Nadia. This will save energy and eliminate human personnel making it cost effective and low priced as well. West Bengal. receives water from the local municipal corporation and another tank placed on the roof top of the same multistoried building. industry and domestic consumption. A S E I Keywords— PIC Microcontroller. India satadal14@gmail. The exact related work is almost not available. to save energy and to eliminate the pump operator. Water will be pumped to the roof top tank and up to the limiting level. Hooghly.com Satadal Mal3 Department of Electrical Engineering Kalyani Government Engineering College Kalyani. India utpal01in@yahoo. waste water handling system etc. The proposed system basically consists of one underground reservoir. Then the motor will be automatically in stop condition.chakraborty@yahoo.khairul@gmail. So. This design will save energy due to mal operation of water level management manually. Digital Latch.

The condition 2 is applicable when condition 1 is satisfied. 2 System Overview Fig. when the water level is in the intermediate level. In fig. one placed at the upper roof top and another is underground one. else logic 1. The motor is connected to lift water from lower tank to upper tank. the need of previous state to be preserved in the EEPROM contained in the microcontroller. float A generates logic 1. When the water level of the upper tank reaches in the minimum level. The motor will be stopped. A single phase induction motor is coupled with the pump. Fig. When the water level of the lower tank reaches in the minimum level. 1 Actual System Flow In order to realize the control circuit using microcontroller. when float C=1 and float B=0. shows the system overview of AEWLMS. NIT Durgapur .2. float C generates logic 0 and when the water level of that tank is higher than the minimum level. float B fitted to the bottom level of the upper tank and float C fitted to the minimum level to be maintained of the lower tank. either float C=0 or float A=1. the float C will generate logic 1. 3 Data Flow Diagram 4 1 A S E I The fig. Fig. The pump is connected in between two reservoirs. When the water level of upper tank reaches the top level. else logic 0. float B generates logic 0.1 shows the two tanks. subject to the condition. the motor will be started to pump water to the upper tank and the motor will be automatically stopped when the water level will reach up to the desired level. The object is to develop a microcontroller based high performance dedicated system to take care the pumping of water following the above conditions. There are three mechanical floats. 156 of 446 Copyright @ Department of Electrical Engineering. float A fitted to the top level of the upper tank. The motor and the pump will be started when the water level in the underground reservoir will maintain certain minimum level.object is to pump water to the roof top tank under the following conditions. The water level in the roof top tank when falls below certain level.

IEEE. it is started when the level of water falls below the minimum level. I. [8] Vicente Ordax Jr. Abuelma’atti. AI-AIi. float C and motor Y. Inc. There is a chance of damage to the motor. H.P. A. University of Kalyani. Step 3: Then start the motor and store the present state of float A. [7] Hewlett Packard. International IEEE/IAS Conference. but that of our present system costs Rs. 4 Snapshots of 16X2 LCD Display S E I References [1] M. [2] W. 1995. Adam. MAY 2010. Kalyani." Operating and Service Manual.. IM-35 VOL. 2009. PROCEEDINGS.25. CONCLUSION Step 2: If float C=1 and float B=0. float B. Vol. The total algorithm is also shown in the Fig. State 8: Again go to Step 2. float B and float C. float B. June 1976. "Prediction of component or system failure with multiphase power simulation" Florida International University MS thesis dissertation. 4-7 APR 1993. 22-27 May 1995. August. [3] A.K. Proceedings of 2009 IEEE Students Conference on Research and Development (SCOReD 2009). 300 approx. V. Hence it is cost effective and efficient.P. it will move to the next step 3. Liu “Single-Phase Phase Angle Measurements in Electric Power Systems”. UPM Serdang. Again store the state of float A.F. Step 5: Float B will be logic 1 after receiving some water into the upper tank. If float C=0 ACKNOWLEDGMENT or float A=1. Nadia for extending the infrastructural facilities without which it was motor Y will be stored in the EEPROM. The microcontroller based automated system will take care of water level at the ground level reservoir and that of roof top tank. difficult to implement the system.ORDAX “DIGITAL MEASUREMENT OF PHASE ANGLES FOR FAULT TOLERANT SYSTEM” SOUTHEASTCON '93.T. float C and motor (Y=1). WAGDY. SUBBARAO. Step 4: Input the present state of float A. [5] I.Step 1: The state of float A. 1400 approx. 157 of 446 Copyright @ Department of Electrical Engineering. meaning the motor will be Kalyani. E.1.A. float B and float C is read by the microcontroller through the input port and stored in the EEPROM during the runtime. then the motor will be stopped.. "Gain-phase meter 3575A. NIT Durgapur . float C and Engineering. Y. M.Hussain ―MICROCONTROLLERBASED PHASE-ANGLE MEASUREMENT AND CORRECTION TECHNIQUE” Industrial Automation and Control: Emerging Technologies. III.S. IEEE Transactions Power Systems. M. float B and float C to EEPROM. If float B=1. pump water from reservoir to the roof top tank relieving the residents to deploy person for running the pump as and when necessary.Que. Kalyani Government Engineering College. Step 6: Check the state of float A and float C now. float B. ISBN 0-7803-1257-0. store float A. i.. [6] T. Xia. This kind of water level controller is available at Rs. Y=0 else The authors are really thankful to the Department of Electrical continuously monitor the state.Sebo “Voltage-current phase angle measurements during aging tests of polymer insulators” 2002 Annual Report IEEE Conference on Electrical Insulation and Dielectric Phenomena.2. This possibility is eliminated with the installation of this automated system and rather protecting the motor from damage. Nadia and the Department of Computer Science and stopped. 3 Flow chart. State 7: Reset the output Y=0. No. MARCH 1986. S. else Step 2.Casale. the person deployed could not estimate the minimum level of water of the ground level reservoir. Mohamed. Malaysia. 16-18 Nov. 1990. 4 1 A Fig. Engineering. Again the present status of float A. [4] W.V. LUCAS “A PHASE-MEASUREMENT ERROR COMPENSATION TECHNIQUE SUITABLE FOR AUTOMATION” IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT.e. Moreover. Sanusi “Simple Phase Angle Measurement of Two Periodic Signals”.

2005.30. Power App. 6. 569-571.. 1995. M.‟„Phase Measment of Distorted Periodic Signals Based on Nonsynchronous Digital Filtering”.K. G. A. Chen. no. Vol. August.Sc. Phadke. Apr. B. vol. 4. 1993.vol. L. no. 5036–5043. 2. Liu. Zhang. thesis. W. International IEEE/IAS Conference on Industrial Automation and Control: Emerging Technologies. 12. Eng. [11] R.. 1. no. So. “Microcontroller based phase angle measurement and correction technique”. 1914–1921.864-867. Bonanomi. Syst. and Y. 10–15. Inrmmmfotion and Meanrremenls.. pp. J. Power. A. 50. [13] J. May 1995. [12] P. pp.” M. Centeno. V. 1981. C. Xu. pp. VA. No.” IEEE Comput. Power Syst. [10] Al-Ali. Michelliti. NIT Durgapur . [9] B.T. S. 4.40-42.A. Elect. pp.” IEEE Trans. pp.” IEEE Trans. 2001. R. Bradley Dept. No. Djokic. IEEE Tram. S. Phadke. Virginia Polytechnic Inst. Appl. “Power system Frequency Monitoring Network (FNET) implementation. Conners. Nov. I. Comput. Dec. “Synchronized phasor measurements in power systems. PAS-100. February 1991. and E. 4 1 A S E I 158 of 446 Copyright @ Department of Electrical Engineering. Blacksburg. “Accurate frequency estimation with phasor angles. 20.Hussain.. “Phase angle measurements with synchronized clocks— Principle and applications. Billian. [15] Z. G. [14] A. Vol. Tsai. vol. State Univ. Zhong. “Phase Angle Measurement Between Two Sinusoidal Signals”. 1994. Abuelma’atti. IEEE Transactions on Instrument and Measurement. pp.

4 1 INTRODUCTION Photovoltaic cells are the devices which directly converts light energy into electric energy. India anildau@yahoo. It is located within the latitude of 29.2. to 3:00 P.1 and variation of solar intensity over a day is shown in fig. photovoltaic cells. A solar cell is basically is a p-n semiconductor junction. I The variation of solar intensity at different day time and in different season is shown in fig. Kurukshetra. power output. monsoon (July-Sept) and winter (Dec. low maintenance cost. Solar cell operates under varying temperature changes. the variation of solar intensity and temperature is recorded in NIT Kurukshetra. at an interval of one hour.c can be given as: Ƞc = Pmpp / Pin = Impp x Vmpp / I (t) x Ac Fig. India. a direct current is generated PVs offer several advantages such as high reliability.M. The average solar radiation intensity (W/m2) is recorded for the seasons of summer (March-June).com Department of Electrical Engineering National Institute of Technology Kurukshetra.2 Variation of solar intensity over a day 159 of 446 Copyright @ Department of Electrical Engineering.89°C in India. PV MODULE EFFICIENCY AS A FUNCTION OF OPERATING TEMPERATURE The maximum conversion efficiency ƞmax.97°C and longitude 76. Keywords— solar intensity.co. which is free abundant in most parts of the world has proven to be an economical source of energy in many applications. The effect of solar intensity and temperature on the efficiency of the cell is investigated. India ashishmlochan@gmail. A Fig. When exposed to light. there has been an enormous increase in the understandings of the operational principle of photovoltaic devices. NIT Durgapur . It uses materials which absorb photons of light and release electron charges. Recently. the change in temperature will affect the power. no environmental pollution and absence of noise. output from the cells. I.Feb) from 9:00 A. This site is perfect for receiving maximum solar radiations and there was no shading of any structure or any object in the path of solar rays falling on the pyranometer from dawn to dusk. II.1 Solar intensity in various seasons S E The data were recorded in National Institute of Technology. temperature.M. In this paper. which led to a rapid increase in the power conversion efficiencies of such devices. PV module efficiency.in Abstract—Solar energy.Effect of Temperature on Photovoltaic Cell performance Ashish Lochan Anil Kumar Dahiya School of Renewable Energy & Efficiency National Institute of Technology Kurukshetra.

which is partially converted into useful electrical output and the rest is III. βref . and at solar radiation of 1000 W/m2. corresponding to solar intensity I (t) and Ac is area of solar cell. 3 I-V characteristic of a PV module Authors usually consider Tref = 25°C. temperature βref = 1 / (T0 – Tref) In which T0 is the (high) temperature at which the PV module’s electrical efficiency drops to zero. having values of about 0.βref x {Ta –Tref + (TNOCT -20) x (I(t) / I(t)NOCT} + γ log10 I(t)] x 100 Fig.4. γ. dissipated into the surroundings. will be discussed in this section. in particular. the variation of solar intensity is recorded over the year at different day time. Tc =Ta + [(TNOCT – 20) / 0. depends not only on the PV material but on Tref as well. Impp and Vmpp are the current and voltage for maximum power. There is a point on the curve named as ‘Maximum Power Point’ (MPP).8 kW/m2] x I(t) From above equations: Ƞc = ƞref [1. A common simplifying assumption is made that the cell-ambient temperature drop increases linearly with irradiance. for crystalline silicon modules. we can see that efficiency decrease with temperature increase. NIT Durgapur . The variation of PV cell efficiency as a function of temperature is investigated and shown graphically.0045 K. wind speed. and the solar radiation coefficient . The correlations expressing the PV cell temperature (Tc) as a function of weather variables such as the ambient temperature (Ta). It is given by the ratio: Fig.The basically effect leads to a relation in the form: From fig. The effect of temperature on the performance of photovoltaic module is investigated. Convection is the main mechanism for heat dissipation in terrestrial. etc. This work clearly indicates a decrease in the efficiency of the PV module with increase in 160 of 446 Copyright @ Department of Electrical Engineering. solar radiation I (t). ambient humidity and so on.12.0045 K and 0. respectively. average ƞref = 12% and average βref =0.8 kW/m2 and wind speed is 1 m/s. 4 1 A S E Ƞc = Ƞref [1. Tref .βref (Tc –Tref) + γ log10 I(t)] In which Ƞref is the module’s electrical efficiency at the reference temperature. The temperature coefficient. CONCLUSION In this paper. The actual value of the temperature coefficient. and radiation is the second non negligible mechanism of heat dissipation. This information is contained in the Nominal Operating Cell Temperature (NOCT). The coefficient depends on module installation. TNOCT usually values around 45 °C. though a single value is used to characterize a module type. The effect of temperature on the electrical efficiency of a PV cell/module can be obtained by using the fundamental equations. For variations in ambient temperature and irradiance the cell temperature (in oC) can be estimated quite accurately with the linear approximation. flat plate applications. irradiance is 0. The steady-state power balance determines cell temperature: the input is the absorbed luminous power. I The quantities of ƞref and βref are normally given by PV manufacturer. For crystalline silicon solar cells this temperature is 270°C. are mainly material properties.The characteristics curve of a PV module is shown in fig. which is defined as the cell temperature is measured under open-circuit when the ambient temperature is 20◦C.4 Photovoltaic cell efficiency vs. However.3. they can be obtained from flash tests in which module’s electrical output is measured at two different temperatures for a given solar radiation flux.

September 2012. Merten. I. Lorenzo Ciani. [3] Abdul Jaleel J. 2334–2340. Ramzan M . Yu. Omega A R. [6] Marcantonio Catelani. Renewable Energy 36 (9) (2011). For a particular PV module. vol. 79. NIT Durgapur . we can find out the value of ambient temperature for a desired efficiency of PV module.8. Zini. [5] Parashchuk D. Russian Journal of General Chemistry. Renewable and Sustainable Energy Reviews 15 (2011) 1625–1636. ISESCO journal of science and technology. solar energy 80 (2006) 170-176. Electrical performances optimization of Photovoltaic Modules with FMECA approach. Thermal photovoltaic solar hybrid system for efficient solar energy conversion. S. 2543–2555. Measurement 46 (2013) 3898–3909. Vol. Iniyanb. REFERENCES [1] Hanif M.temperature. IJAREEIE. Ranko Goic. [7] G. vol-1. pp. 4 1 A S E I 161 of 446 Copyright @ Department of Electrical Engineering. A review of solar photovoltaic technologies. Simulation on maximum power point tracking of the photovoltaic module using Labview. No. 11. Bulat L. 2009. issue 3. [4] Bhubaneswari Parida. Loredana Cristaldi. 14 Nov 2012 (9-12). Mangeant. Studying power output of PV solar panels at different temperatures. Reliability of largescale gridconnected photovoltaic systems. J. C. Modern photoelectric and Photochemical Methods of Solar Power Conversion. and Kokorin A. [2] Vorobiev Yu.

m5 S E I In cellular manufacturing. West Bengal. machine layout. 1). It thereby improves productivity of a manufacturing facility. in job shop manufacturing. Layout affects the cost. On the other hand.com 2 samyajit_talukdar@rediffmail. and arrangement of cells. Group technology (GT) or cellular manufacturing is usually applied to manufacturing systems that has fairly large variety with fairly big product volume. Three major types of arrangements in the GT cell are the single-row. Hassan and Albin [5] did an extensive study on the type of data required in the machine layout problem. Dr. namely the number of machines. A detailed description of this layout is given by Kouvelis et al. NIT Durgapur . m1 m1 m2 m3 m4 m5 m3 m4 (c)semi-circular layout m2 m3 m8 m4 m7 m6 m5 (d) loop layout Fig.741235. M. m4 m3 (a) single-row layout Keywords—genetic algorithm. or GT cell layout. West Bengal. The single-row layout is used when the parts assigned to the group follow the same machine sequence. The multi-row. Each component needs to undergo four to nine operations using the machines. In the present work. multi-row. In a case where product volumes are high and varieties are low. Each of these steps is a combinatorial problem and many heuristics have been proposed over the years to solve them. GA.com any other machine. Machines can be arranged in a single row. machine 4 1 layout.K. standard Genetic Algorithm (GA) is employed to find out a machine schedule to achieve minimum handling time and cost in a typical manufacturing industry. Cellular layout suites this condition. Reetabrato Bhattacharyya2 Dr. and hence. minimization. In the loop layout (Fig.me@gmail. Santanu Das3 Department of Mechanical Engineering Kalyani Government Engineering College Kalyani . The multi-row layout is usually linear and the movements of parts can be between any machines in different rows as well as in the same row. or loop layout depending on the sequence of operations. linear or U-shaped. multi-rows or loop layout. The machine layout data was considered in a hierarchy depending on the designed layout. The GA gives an optimum result within the domain considered. the cell or machine layout problem is basically concerned with finding the best arrangement of machines in each cell. 1 Types of layout in the GT cell The development of the GT layout basically involves 3 steps [4] such as formation of part family. 162 of 446 Copyright @ Department of Electrical Engineering. There are different ways to plan machine layout in an effective way. [6].com Abstract— Choosing an appropriate machine layout plays an important role to minimize material handling time and related cost. Pakhira Department of Computer Science and Engineering Kalyani Government Engineering College Kalyani . Another type of product layout is the fixed product layout in which production equipment moves toward the product. that is known as flow shop process.Optimum Process Scheduling Using Genetic Algorithm in an Existing Machine Layout Samyajit Talukdar1. permits a part to move from any machine to m2 m5 m6 (b) multi-row layout m1 m2 A INTRODUCTION Plant layout problem deals with finding out the best arrangement of physical facilities to provide an efficient operation. available space. as shown in Fig. layout is normally based on type of product. handling time. 1. GT centre layout is based on functional arrangement of machines. However. when productmix changes frequently. Eight different types of components are considered to produce that need ten different machines. overall productivity and efficiency of the plant [1][2]. India e-mail 1reetabrato@outlook. The single row layout may assume different configuration such as semi-circular. India e-mail malay_pakhira@yahoo. and it needs job shop layout [3]. which is suitable for FMS. time and throughput of material handling. m1 robot I. there is large variety of products with low production volume.com 3sdas.741235. Layout of machines depends on the type of manufacturing industry. The type of the machine layout are affected by a number of factors. machines are arranged in an oval path with unidirectional movement.

Ideally. Standard genetic operators are selection. [16]. mutation. A S E I Kouvelis et al. as shown in the row matrix. Banerjee and Zhou formulated [13] the facilities design optimisation problem for a single-loop layout using genetic algorithms. Some common techniques used had been assembly charts giving the graphical representation of flow of parts. multi-product process charts. evaluating fitness values. volume moved. etc. [15]. such as maneuvering an overhead crane [17]. The algorithm was developed is for the cell systems layout. and selecting handling equipment need be considered simultaneously. string diagrams. Koopmans and Beckman developed [8] a quadratic assignment problem (QAP) model to solve for the multi-row layout with an aim to minimize the material handling cost. Single-row layout. Genetic algorithm is a heuristic search algorithm based on evolutionary ideas of natural selection through the survival of the fittest following the principles introduced by Charles Darwin [14]. These machine tools are to employ to manufacture eight different components. The actual layout of machines within cells was not considered in this work. In the present work. rate with which the part travels and the cost of the move [1]. frequency of move. especially the GA. It works very well on continuous and discrete combinatorial problems. transportation programming. associated usually with the flow-line problems. operation process charts. Quantitative approaches were also tried for optimizing the location of equipment with the movement of material [1]. distance. transshipment programming. The genetic algorithm (GA) is used in this work as detailed by Goldberg [14] and later used by Ghosh et al. detailed analysis needs be done on the routing of parts. Since the cost associated with the layout is proportional to the distance travelled by the parts. fitness scaling. inversion. Simulated annealing and genetic algorithm were also employed to settle machine layout. Ghosh et al. and providing a penalty function are inherent for the working of a typical genetic algorithm [14]. and others. the algorithms developed had the objective of minimizing the distance travelled by the parts. controlling the joint position of a humanoid robot [19]. The type of material handling equipment determines the travel time. Determining an objective function. assignment problem. etc. although the GA is susceptible to have local optima. The QAP was also tried successfully by Garey and Johnson [9]. the resultant layout will actually minimize the cost of travelling the distance. etc. combinatorial nature of these poises difficulty to adopt this consideration [5][7]. preparing an optimum time table of a city tram service [18]. NIT Durgapur . A genetic algorithm (GA) is an adaptive search technique which imitates the process of biological evolution [14]. and hence. developing machine layout. To choose a machine layout. Hassan described [7] the differences between machine layout and facility layout or block layout.1. similarity of sequences and the material handling system used [7]. Sequence of machining processes required are given by the sequence matrix. THE GENETIC ALGORITHM III. II. THE GA APPLIED FOR SCHEDULING MACHINING PROCESSES The GA is applied in this work to solve a machine layout problem. Among the techniques used to solve the layout problems are linear programming. the material handling equipment used for the multi-row layout was AGVs or human operators to handle fork lifts or other equipment. Suresh et al. did not consider the layout of machines within a cell. Each component requires different machining operations to perform on it using suitable machine tools for its finishing. B in which the ith row corresponds to the machine tools needed to 163 of 446 Copyright @ Department of Electrical Engineering. It can be stated that different methods.Hence. The GA was also successfully applied in many fields for solving optimization problems. and integer programming. Number of batches of a part (frequency) to produce on each of ten machines may be different. and multi-row layout appropriate to job-shop problems were considered in these works. V. the GA is used to find out an optimum schedule of machining of eight components to apply to the existing layout of ten machine tools such that handling time and related cost would be minimum corresponding to the present problem. but. A method using genetic algorithms was developed to solve the machine layout problem by Chipperfield et al. and affects throughput and flexibility of the FMS. Genetic algorithm (GA) was introduced by John Holland in 1970s. It is computationally less expensive and suits well to solve a host of combinatorial problems. First element of this matrix signifies that the first machine tool has the task to machine 4 batches of components while the second element related to second machine tool has only 1 batch of components to machine. minimized [11] total backtracking distance for material handling in a row layout of an automated manufacturing system. [16]. used [12] GA to find part family as well as the layout between cells with linear single or double row layout. etc. where the objective was to minimize the cost of interaction between various departments. 4 1 In a genetic algorithm (GA). crossover. or may sometimes be the same. some genetic operators are used to reach the optimal solution. was successfully used to solve machine layout problem. A multi-row layout of ten machine tools existing in the manufacturing shop is shown in Fig. used [10] a genetic approach to solve the facility layout problem. Unlike the single-row layout. Gupta et al. and from/to charts. V = [4 1 2 2 3 5 4 1 6 8].

NIT Durgapur . Part 5: Machining 5 should be before Machining 4. The penalty cost function used is slightly modified over that given by Abe et al._ of _ generation (2) where. 1 2 10 10 10 10 0. Eight different parts can undergo from a single to 10 different machining processes. The penalty cost function used in this work is given by 4 1 A S E I PC = K .The objective function [4] is given as follows: carry out machining of ith component in sequence to initialize. To each pool. B = ⎡ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎣⎢ 1 2 5 6 1 2 3 4 5 7 0 2 4 7 8 9 0 1 2 5 6 8 1 5 4 8 7 9 0 2 0 1 3 1 4 4 3 7 5 4 9 8 8 9 N N ∑∑ Obj = min fij c ij d ij i =1 j =1 ⎤ ⎥ ⎥ ⎥ ⎥ 9 ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎦⎥ (1) where.8 0. The technological constraints have been kept in mind. the objective is formulated as minimizing the total traveling distance of parts. Mutation and Elitism. before Machining 9. Part 3: Machining 7 should be before Machining 8. For example: chromosome {5 2 1 3 4} would represent a layout shown in Fig.8 4 10 10 0.8 5 location 1 3 location 4 2 location 2 1 location 3 4 location 5 Fig. as well as the possible locations of machines. and hence. Part 4: Machining 2 should be before Machining 8. where the location of the gene represents the location of the machines [4].8 5 100 10 0. TECHNOLOGICAL CONSTRAINT ------------------------------------------------------------------------Part 1: Machining 1 should be before Machining 6. If the tested chromosome is found invalid. Part 2: Machining 2 should be before Machining 4.8 7 100 10 0. The optimal machining sequences for each part have been found out using GA.2 Assuming the cost cij remain constant. and then it is extended to 100 to find out any benefit achievable by it. the 164 of 446 Copyright @ Department of Electrical Engineering. Based on the optimal sequences the parts can be classified in different group (according to manufacturing attributes) and the optimal cell formation can be obtained for a machine layout problem. Table 1 shows the technological constraints for the manufacture of each component.8 8 100 10 0. Each chromosome consists of genes representing machines required for their machining.05 0. The initial population is generated by permutation of each chromosome. The criterion most often used in the layout problem is to minimize the travel cost of the parts. Each generated chromosome corresponding to each part is checked for validity with the imposed technological constraints. 3 10 10 0. There are 10 machine tools which can be used for the manufacturing process.2 0. with the cost related to the distance traveled. are found out first. GA is applied using Roulette Wheel Selection. PC = Penalty cost K = Number of stations or machines N = Number of constraints The chromosome representation is based on the permutation of integer which represents the machine numbers. before Machining 7. the objective would be reduced to minimizing the total distance travelled for the parts.n.3 generation no. Elitism is used here to preserve the best chromosome of each pool.8 0. [20].1 0. a heavy penalty cost is charged to prevent its further selection into new pool. Part 8: Machining 8 should be before Machining 9.1 0. VALUES OF THE PARAMETERS USED TABLE 1. The values of the genetic parameters used are given in Table 2. 2.03 0.03 0. Consequently. ------------------------------------------------------------------------- Conditions Generations Population Size Crossover Probability Mutation Probability In the present problem. 2. eight parts have been considered for manufacturing. Machine layout for chromosome {5 2 1 3 4} The distances between locations.8 6 100 10 0.05 0. Part 7: Machining 4 should be before Machining 8. before Machining 5. Partially Matched Crossover (PMX) [12]. Maximum number of generations are allowed up to 50 first. Part 6: Machining 7 should be before Machining 9. fij = frequency/volume of movement cij = cost to move one unit load per one distance unit between two machines dij = distance between machine i and j TABLE 2.

variation of total travel length with number of generations is shown in Fig. Fig. Fig. there are 10!. Each part undergoes between 4 to 9 operations. The amount of backtracking for this example has been reduced by 198 units or by 53% from the initial random layout generated. ---------------------------------------------------Part[0]:1256 Fitness:20. these sequences may be chosen for adopting in the present case. and results obtained after this are presented. Hence.000000 165 of 446 Copyright @ Department of Electrical Engineering.050000 ---------------------------------------------------.000000 **************************************** * CONSOLIDATED RESULT * **************************************** Chromosome [1]:2165 Fitness [0]:12. NIT Durgapur .000000 Part[1]:123457 Fitness:30. Table 1 and Table 2. which indicates the number of times a part has to move from one location to another. 3 corresponding to mutation probability of 0. which are being processed in different machines.. of generations : 100 Cross over probability : 0. After determining the fitness values for the initialized random sequence. the possible combination for n locations is n!. 150 100 50 0 1 350 300 250 200 150 100 50 0 1 9 17 25 33 41 49 57 65 73 81 89 97 No.000000 Part[6]:23458 Fitness:37. each part undergoes from 4-9 different operations..1 Sample results of the genetic algorithm corresponding to initialized random sequence and the last generation is given in the following. objective is to find the best position of machines that would minimize the total distance travelled by parts. cross over probability of 0. In this.800000 Mutation probability : 0..05 To give an indication of the complexity of the problem.000000 Part[7]:013489 Fitness:20. of generations Fig. Initializing . 5 shows an example of how the movement of one part is simplified as the layout evolves from generation 1 to generation 20 corresponding to 10 machines and 8 parts.1. 100 generations of the GA are allowed. With a mutation probability of 0.1.000000 Part[5]:01479 Fitness:176.. Total travel length (Fitness value) 400 350 300 SIMULATION RESULTS AND DISCUSSIONS 250 200 The data used in this work are shown in sequence matrix V. The GA-based method is able to find the solution in a matter of a few minutes compared to using enumerative methods.000000 Part[3]:0125689 Fitness:28. In this example.05. and the total travel distance is reduced from 372 to 174. 6 of Table 2 Population size : 10 Total No. IV.6288 x 106.000000 Part[2]:024789 Fitness:27. 4. 4.8 and that of mutation of 0. 5.05 are used. 8 Sample Result for Condition No. of generations 85 92 99 Initial Length Travel length Reduced Length 400 350 300 250 200 150 100 50 0 372 372 372 372 372 372 372 372 175 184 182 174 176 174 176 174 1 2 3 6 7 8 4 5 Conditions 4 1 Fig. 3 shows the evolution of the machining sequences at different generations corresponding to mutation probability of 0. The reduction in the traveling distance reduces the cost of traveling as well as time of traveling from one machine to another. Another important data required for the layout problem is the frequency of movement between machines. Hence. Variation of total travel length with number of generations with mutation probability of 0.fitness values which show remarkably less value.000000 Part[4]:154879 Fitness:34. possible combinations. Chromosomes related to each of the component types indicate the machine sequence and these are evaluated in terms of 36 43 50 57 64 71 78 No. Movement of each part from generation 1 to generation 20 A S E I Length of travel (Fitness value) 400 15 22 29 Fig. 3. These results are comparable with that of Fig. Variation of total travel length with number of generations with mutation probability of 0. or 3.000000 Initial Length of Travel: 372.

1995. Vol. Vol 34(2). 2006. “Managing data requirement for the machine layout problem”. Int. 53-76.M. and S. [11] [12] CONCLUSIONS [13] In the present work. Journal of Computer Applications in Technology. Int. Operation Management. Journal of Production Research. 1979. Wiley.000000 [7] [8] [9] [10] Reduced Length of Travel: 174. Vol 27. K. 2003. Operations Research. pp.V.R. Molecki. Econometrica. IEEJ Trans. 99-104. India. Koopmans and M. J. 447-482. Dilworth. “A genetic algorithmbased approach to cell composition and layout design problems”. Zhou. 1994. Int. pp. “Machine layout problem in modern manufacturing facilities”. NIT Durgapur . of the Conf. “Optimum scheduling of a robotassisted assembly task using genetic algorithm”. Ghosh.000000 Percentage of Reduction: 53. Kim. M.000000 Chromosome [4]:2105689 Fitness [3]:24. 28-37. Suresh. Banerjee. Ph. Proc. pp. M. 1994. V. “Assignment problems and location of economic activities”. Hassan. pp. 337-343. Kouvelis.D. 1957.226% =========== END OF REPORT ========== V. 1994. pp. Das. Gupta. and G. G. IIE Transactions. M. Y. pp. Journal of Production Economics. of the Workshop on Intelligent Information Systems.S. and Y.000000 Chromosome [7]:24385 Fitness [6]:24. Beckman. Addison-Wesley. “Timetabling of city tram service using a genetic algorithm”. P. pp.126. Goldberg. 1996. Samanta. Optimization and Machine Learning. 1996. Freeman and Co. Ustron. D. G. Int. The GA gives a minimum distance of part travel from one machine to another machine satisfying all the restrictions and technological requirements that would help in achieving good productivity. 2559-2584. Journal of Production Research. M. 1995. Vol. P.C. Int. 1995. 38. Gupta. 1999. Fonseca. Australia. Chiang.000000 Chromosome [8]:013849 Fitness [7]:14. Computers and Intractability: A Guide to the Theory of NP-Completeness. Yu. Vol 33(1). on Robotics and Automation.000000 Chromosome [5]:517849 Fitness [4]:19. pp. [14] [15] [16] REFERENCES [1] [2] [3] [4] [5] [6] J. on Advances and Recent Trends in Manufacturing. for Use with MATLAB: User Guide”. and N. “Genetic Algorithms TOOLBOX.M. P. and C. New York. N. Hassan. Research Report No.000000 Chromosome [6]:01794 Fitness [5]:56. Sundaram. Genetic Algorithms in Search. International Journal of Production Research. Vol. M. 533-550. Johnson. 2003. 25(1). Wigeth. Fleming. H.3102.000000 Chromosome [3]:027849 Fitness [2]:17. Kwasnicka. S. and C. and M. pp. Vol. 1992. W. T. Deb. Brisbrane.M. Michigan. “Improved joint control using GA of a humanoid robot (GuRoo)”. “Unidirectional loop network [17] layout problem in automated manufacturing systems”.E. Optimization of Cellular Manufacturing Systems Using Genetic Algorithms.173-188. 1042-1053. 34113423. McGraw Hill. Genetic and Evolutionary Computation– GECCO-2004. Gupta. 4 1 S E I A [18] [19] [20] 166 of 446 Copyright @ Department of Electrical Engineering. Albin. A. “Optimal algorithms for row layout problems in automated manufacturing systems”.183-203. Vol. pp. Hassan. Apple. Genetic Algorithm is used to solve a machine layout problem associated with a manufacturing cell to find out optimal sequence of operations for each part to manufacture.W. of the 2003 Australian Conf.B. Morad. 512. 7(1-2). and M. Yamada. Chipperfield. and M. and D.K.C. “Facilities layout design optimization with single loop material flow path configuration”. A. “In search of optimal operating principles of an overhead crane maneuvering using multi-objective evolutionary algorithm”. 32(11). and J. and S. Kumar. Chiang. Sahu. Kee. 2004. Journal of Production Research.M. Vol.Chromosome [2]:127435 Fitness [1]:8. P. Kouvelis. EIS. 33(12). Polheim. Vol. T. Matsui. pp. K. Plant Layout and Material Handling. 2nd Edition. 1997. 1989. S. “Layout design in group technology manufacturing”. Abe. 1977. Proc. 40(3). Poland.D.D. Vinod. “A genetic algorithm for facility layout”. 1995.D. Roberts. D. H. Lecture Notes in Computer Science. Department of Automatic Control and Systems Engineering. A Design Problem of Assembly Line Systems using Genetic Algorithm under the BTO Environment. Part 2-3. Proc. and B. Kalyani. Thesis submitted to University of Sheffield.. Vol. Garey.

Hyderabad and My Dear Parents. whether domestic or industrial. but active systems are more expensive than passive systems. This makes such systems generally more reliable. Active systems use electric pumps.com Department of Mechanical Engineering JNTU H College of Engineering Hyderabad Kukatpally. gets heated up causing the vaporization of water.com Ravi Gugulothu Department of Mechanical Engineering JNTU H College of Engineering Hyderabad Kukatpally. A detailed review on previous experimental investigations done by other researchers in the field of solar distillation using active and passive techniques is made in this paper.gugulothu@gmail.P. Hyderabad. Solar still is a useful device that can be used for distillation of brackish water. Water is a nature’s gift and it plays a key role in the development of an economy and in turn for the welfare of a nation. A. Solar energy is available in abundant. Moreover. A.P. thermal energy flow occurs by radiation. Keywords— Solar. Desalination of brackish or saline water using solar energy is a good method to get fresh water. I. majority of the health issues are owing to the non availability of clean drinking water. Hyderabad. Hyderabad. conduction and natural convection. India e-mail: mohanbukya@gmail. like coal. fresh water is getting depleted and polluted. The incident solar radiation on solar still is transmitted through an inclined glass cover to the water which is available in the solar still. gas and oil as well as to mitigate the increasing concentration of gaseous CO2 in the atmosphere. So many experimental investigations were conducted on solar distillation techniques in many countries like India. The water absorbs the solar radiation. Clean water is evaporated from the brackish water and condensed on the glass cover. Solar thermal systems: The purpose of using any type of solar thermal collector is to convert the solar radiation into heat to be used in a specific application. 167 of 446 Copyright @ Department of Electrical Engineering.Vijaya Kumar Reddy. K. Iran and Iraq. Department of Mechanical Engineering. The availability of drinking water is reduced day by day due to increase in the population and industrial usages. India e-mail: nagasaradaso@gmail. the saline or impure water is purified by using this technique. A.P. In such technologies. Solar still is a simple device which can convert available waste or brackish water into potable water using solar energy. in active solar system. 4 1 A S E I which requires electricity for its operation. JNTUH College of Engineering. The evaporated water particles condense on the glass cover and the condensed distilled water exits through the outlet channels. there is a need to explore alternative energy sources. INTRODUCTION Water is the only available source of drinking and it is the gift of god on this beautiful earth. In passive systems. Energy storage materials.P. passive solar system is preferred because it is cheaper compared to active solar still. The work done by previous researchers in obtaining distilled water using solar energy in different methodologies is listed below: Among active and passive systems. A. Jordan. easier to maintain and possibly longer lasting than active systems. Two major groups of solar distillation devices are: active solar still and passive solar still. Guntur. free of cost and hence solar distillation is the best solution for obtaining clean drinking water. Turkey etc.com Abstract— To combat the present and future energy supply challenges which are associated with shortage of the non renewable natural resources.A REVIEW ON PARAMETRIC PERFORMANCE OF SOLAR STILL Naga Sarada Somanchi Hima Bindu Banoth Department of Mechanical Engineering JNTU H College of Engineering Hyderabad Kukatpally. The basic principle of solar water distillation is the sun’s energy which heats water to the point of evaporation. Potable water is a basic necessity for human being. Today. we should depend on oceanic and brackish water resources for fulfilling our demand of water. valves and controllers to circulate water or other heat transfer fluids through the collectors. NIT Durgapur . distillation. They are active and passive thermal systems. extra thermal energy is obtained with the addition of pump and collector. India e-mail: nagasaradaso@gmail. active and passive techniques. Passive systems have the advantage that electricity outage and electric pump breakdown are not issues. We have two different types of thermal systems. solar energy is the only source of the energy while in active systems. Finally. Due to increase in population and human activities. In future. Non availability of drinking water is one of the major problem faced by both the under developed and developing countries all over the world.com Mohan Bukya Department of Electrical and Electronics Engineering Koneru Lakshmaiah University. desalination. India e-mail: ravi. Active systems are more efficient. This process removes impurities such as salts and heavy metals as well as destroys microbiological organisms. which is drained out for use. Passive solar technologies are means of using sunlight for useful energy without use of active mechanical systems. there is a need of pump Sponsors: Dr.

60 kg 3.B et al Joseph.15 m 16 Hikmet S Aybar 2006 (Turkey) Solar still Inclined at 300 13 Solar still 200 150 2003 (Egypt) 12 350 optimum Still glass cover A 30 mm water depth Efficiency (Distillate) 3. NIT Durgapur .G. JNTUH College of Engineering.No Authors Year of Publication Distillation Process Augumented by 50 ppm 100 ppm 50 ppm Black 172.1 m.05 kg 4.N 2005 (India) Active solar distillation Different water depths are 0. 250.J et al 2005 (India) 2005 (India) Floating tilted wick solar still 14 Nijmeh. 450 and 550 Collector Inclination Red 1 Anil K Rajvanshi 1981 (USA) Solar still Adding Dyes 2 Bilal A Akash et al 1998 (Jordan) Single basin solar still 3 Bilal A Akash et al 2000 (Jordan) Solar still 4 Sanjeev Kumar et al 2000 (India) Active solar still 5 Nafey A.5 ppm 50 ppm Green 100 ppm Black rubber mat Black ink Black dye Cover tilt angles are: 150.4 kg/m2/day Solar still 2005 (Turkey) Three different variants Janarthanan. Hyderabad and My Dear Parents.S and Yogi Goswami.5 l/2m2 /day) 17% 26% 29% Evaporative heat transfer coefficient is more at 0.S et al 11 Hikmet S Aybar et al 2002 (India) Solar still with black rubber Solar still with gravel Basin type solar still Solar still with floating aluminum perforated black plate S E 2003 (USA) I Water desalination system Thickness: 10 mm 20% Size: 20-30 mm 19% Glass inclination angles 200. 350.P Al-Kharabsheh.5-5. 0.15 m water depth 3.60 kg 5. 450 450 is optimum 3 cm 15% 4 1 6 cm 40% Normal still Perforated and black Aluminium sheet Aluminium sheet folded into “V” wave 1600 kg/m2 Flat basin solar still 3-4 kg/m2/day Low grade solar heat 6. K.31 kg 5. 0.S.19 kg 38% 45% 60% 1700 kg/m2 43% 2400 kg/m2 Sponsors: Dr.Vijaya Kumar Reddy.S et al 2001 (Egypt) 6 Irene De Paul 2002 (Argentina) 7 Safwat Nafey et al 2002 (Egypt) 8 9 Valsaraj.D 10 Fath H.5 kg/m2/day Pyramid shaped Single sloped Bare plate Black cloth wick Black fleece wick Open cycle Closed cycle Single stage solar desalination K2Cr2O7 KMnO4 Violet dye 30% 33% 1290 ml/day 1705 ml/day 2995 ml/day More efficiency in closed cycle 15%-26% (8.E.05 m. 168 of 446 Copyright @ Department of Electrical Engineering. Department of Mechanical Engineering.55 kg 5.S et al 2005 (Jordan) Single basin solar still using 15 Rajesh Tripathi and Tiwari.

G.05% 11.04m.N 2006 (India) Solar desalination Solar stills for different depths of water are 0. 300.8 l/m2/day) 150 optimum 0.04 m 0.B et al 24 Velmurugan.98 l/8h) 35% (3. K.693 l/day) 18% (5.4 l/m2/day) 60% (5. Department of Mechanical Engineering.V et al 2008 (India) Solar still with cover inclination angle Single slope solar still Double slope solar still I Stepped solar still using Rahul Dev and Tiwari.8 l/m2/day 10% to 34% 1.12m.M.K and Verma.9 l/m2/day 1. NIT Durgapur .G.37 l/m2) Sponsors: Dr.04m optimum 67% 70% 80% 0. 450 Vibration Inclination angles of glass 150. 0.Tiwari and Tiwari.0 cm 20 Phadatare.75 l/m2/day 1.01 l/8h) 60. JNTUH College of Engineering. 0. 300.92% 14.V et al 2009 (India) Solar still 28 Velmurugan.8 l/m2/day 1.05 m.10 m Solar still with different water depths 0.83 l/m2 53.756 l/day) 19% (5.1 l/m2/day 1.12 m 21 Anil Kr. Hyderabad and My Dear Parents.11% (1.N 2008 (India) 22 Hitesh N Panchal 2008 (India) 23 Selva Kumar.3% (1.15 m June 2003 August 2003 516 l/day 355 kg/day Active still Active still produces more distilled water Passive still 0.S.02 m 0.N 2009 (India) Passive solar still with 27 Velmurugan.85 l/m2/day 150 is optimum 20% 35% 24.883 l/day) 19% (5. 450 Water depths 0.47% 30. 0.16m Using Pebbles Using Baffles Using Fin and Sponges Solar still alone Fin Fin + pebble 23% (6.27 l/m2) 65% (1.5 cm 19 Muafag Suleiman K Tarawneh 2007 (Jordan) 2. 169 of 446 Copyright @ Department of Electrical Engineering.1 m.0 cm Solar still with varying water depths 3. 0.39% (1.G.06 m 0.0 cm 4.08m.Vijaya Kumar Reddy.62 l/8h) 76% 96% (1. 0.K 2007 (India) 0.V et al 2009 (India) Stepped solar still with 26 Sponges Fin Backed helical wires Solar still with Khaled M S Eldalil Without charcoal With charcoal With boosting mirror Boosting mirror and charcoal Stepped solar still alone Sponges + Fin 2009 (Egypt) 25 4 1 Using sprinkler (Latitude angle less than 150) A “V” type solar still S E 2008 (India) 150.660 l/day) 2.08 m 0.17 Imed Houcine et al 2006 (Tunisia) 18 Rajesh Tripathi and Tiwari.

33. Department of Mechanical Engineering. Black ink and Sponge are some absorbing materials 4 cm of water depth 350C of sea 6 cm of water water depth 8 cm of water depth Rubber mat gives more yield 3930 ml/day 4940 ml/day 6670 ml/day Sponsors: Dr. 170 of 446 Copyright @ Department of Electrical Engineering.65 l/m2) Adding isolating dry sand to evaporator 8.R and Walke P.Fin + sponge Fin + pebble + sponge 29 Bechki.Z.Vijaya Kumar Reddy.M et al 2011 (Egypt) A Quantity of water 50 l S E I 4 1 Quantity of water 30 l Conventional Finned Corrugated Conventional Finned Corrugated 33 34 35 Rajendra Prasad et al Teltumbade T. JNTUH College of Engineering. NIT Durgapur .03 l/m2day North side South side cover produces more distillation Solar still glass cover South side 30 Hitesh N Panchal and Shah P.J.4% . Hyderabad and My Dear Parents.5% (4550 ml/m2/day) 41% (3950 ml/m2/day) 36% 49% Conventional Water depth 50 mm Finned Corrugated 32 Omara. K.T et al 2011 (Iraq) Charcoal cloth is absorbing material 4 mm.V Sabah A AbdulWahab et al 2011 (India) Solar still 2011 (India) Single basin solar still 2012 (Oman) Inverted solar still with Refrigeration cycle Without gel With graphite filled silica gel Rubber mat.7% .28% Outdoor Test 53.K 2011 (India) Solar still glass cover thickness 31 Mahdi. 8 mm and 12 mm 4 mm optimum Indoor Test 37.7% 34% (3400 ml/m2/day) 41% (4100 ml/m2/day) 45% (4400 ml/m2/day) 34% (3200 ml/m2/day) 41% (3850 ml/m2/day) 40% (3750 ml/m2/day) 35% (3250 ml/m2/day) 47.4 l/m2) 98% (1.D et al 2010 (Algeria) 68% (1.

sponges and pebbles. JNT University for his suggestion of reading numerous research publications in the area of solar distillation and to write review papers.40% (11963 ml) 25. India. He did B.Rajvanshi. “Effect of various dyes on solar distillation”. Hyderabad and My Dear Parents. Hyderabad. 156. and Optimal Control. University College of Engineering.38% (13173 ml) 1965 ml/day 8% 49. West Bengal. [3] Anil Kr. She has 26 journal publications. depth of water in basin.Vijaya Kumar Reddy.K). His areas of interests are Design of Mechanical Elements. Her areas of interest are: Heat Transfer and Thermodynamics.G. Sita Rama Raju. Hence. it is observed that. Sudheer Premkumar. She has 10 years of teaching experience and 5 years of industrial experience. “Analysis of an innovative water desalination system using low grade solar heat”. Solar Energy Vol. NIT Durgapur . 27.48% tubes Kukatpally.D. Andhra Pradesh. gravel. Hima Bindu Banoth is a B. Dornakal Mondal. India. varying glass cover inclination angles.S and Yogi Goswami. pp: 51-65. Warangal District on 4th August 1985. Desalination. Andhra Pradesh. 10080 ml/day 9500 ml/day 10104 ml 18. 1981. Madras. Department of Mechanical Engineering. I am thankful to Professor A. Solar energy technologies and its usage are very important and useful for the developing and under developed countries to sustain their energy needs. Vaddeswaram. His interested areas are Control Systems. Andhra Pradesh. JNTUH College of Engineering. JNTUH College of Engineering. there is a scope to conduct experimental investigations in this area. Thermal Analysis and Heat & Mass Transfer. He is the member of RINA (U. sponges and nanomaterials are limited. Vijaya Kumar Reddy and Professor B. JNT University and my parents for their encouragement in preparing this review paper on solar energy. REFERENCES [1] Al-Kharabsheh.7% 3910 ml/day Black gravel with evacuated 59.48% (12679 ml) 30. The length of solar still. Nonlinear Control. But the work using combination of sun tracking system (single axis or double axis) coupled with PCM. absorber plate area and solar radiation are the major parameters which affect the performance of the still. Design of Offshore Structures. 2003.N. He completed his B. India. 171 of 446 Copyright @ Department of Electrical Engineering.Tech in JNTUH College of Engineering Hyderabad and M. inlet water temperature. The solar stills are friendly to nature and eco-system. BIOGRAPHIES Naga Sarada Somanchi is a Professor in the Department of Mechanical Engineering. From the literature review. Kukatpally. Hyderabad. K.Tech from Indian Institute of Technology. dyes. He born in Champlathanda. previous researchers conducted experimental investigations on obtaining distilled water with the help of solar energy in the presence of different water depths. Mohan bukya is a Assistant Professor in the Department of Electrical and Electronics Engineering at Koneru Lakmaiah University. Guntur. [2] Anil K. India. Ravi Gugulothu is a lecturer in the Department of Mechanical Engineering at JNTUH College of Engineering Hyderabad. 323-332. “Effect of Cover Inclination and Water Sponsors: Dr.V. Andhra Pradesh. He has 3 years of teaching experience.Tech in Electrical & Electronics Engineering. inclination angle of glass.Tech in Control Systems Engineering at IIT Kharagpur.Tiwari and Tiwari.Tech student in the Department of Mechanical Engineering at JNTUH College of Engineering. Kakatiya University and M. dyes and energy storage materials like black rubber mat. free surface area of water.4 cm of water depth 300C of sea 6 cm of water water depth 8 cm of water depth Without dye 36 Mitesh I Patel et al 2013 (India) With Red dye Single sloped active solar still Water depth of 10 cm With Blue dye With Black dye Still alone Black gravel 37 Sampathkumar K et al (India) Solar still with RESULTS Energy and water are the basic necessity for all of us to lead a normal life in this wonderful world. water glass temperature difference. Local climatic condition and application are to be considered while selection of solar still. I am thankful to Professor K. 4 1 A S E I Evacuated tubes 6400 ml/day ACKNOWLEDGMENT Foremost. The use of solar energy in distillation process is one of the best applications of renewable energy. He has three years of teaching experience.

2005. “An experimental study on solar distillation in a single slope basin by surface heating the water mass”.N.M. 2013. Sep.A.S. [34] Valsaraj. [18] Mitesh I Patel. [27] Rajesh Tripathi and Tiwari. [11] Hitesh N Panchal and Dr.Mohsen and Waleed Nayfeh. 2005.B. 2009. Thermal Science: Vol.P. I [15] Joseph. [31] Sanjeev Kumar. 2000. [20] Nafey. “Effect of Various Parameters on Productivity of Single Slope and Double Slope Solar Still”. 187200.N and Singh.B.12.Shah. Energy Conversion and Management 41. [14] Janarthanan.1-4. “Characteristic equation of a passive solar still”. “Performance of finned and corrugated absorbers solar stills under Egyptian conditions”.T and Srithar.S. Volume 1. 25. 1179-1186. 2011. Sanjay Kumar and Jayaprakash. pp. 233-238. Thirteenth International Water Technology Conference. 1401-1408. “Effect of Water Depth on the Performance Evaluation of Solar Still”. Mousa S. Department of Mechanical Engineering. 159. Venkata Rajeev.E and Sharif. Mayilsamy K.G. 2006. pp:1781-1787.R and Renganarayanan. [23] Phadatare. 285289. Naveen Kumar. [17] Mahdi. Odeh. Desalination. Desalination.A. A [12] Imed Houcine.P. Vol. Jordan Journal of Mechanical and Industrial Engineering. 153-163. 2002. Padma Pujitha. Vol. Desalination.S. 2006. Nos.B. Desalination. K.Z. “Experimental study of the basin type solar still under local climate conditions”.A. International Journal of Engineering Science and Technology (IJEST). [19] Muafag Suleiman K Tarawneh. 69-79.J. 2009. “New model of a basin type solar still”.H. 307310.K and Hassabou. [8] Hikmet S Aybar. International Communications in Heat and Mass Transfer. No.A.V. “Effect of partial intermittent shading on the performance of a simple basin solar still in south Algeria”. Abdelkader. Meena P.A and Mabrouk. Desalination 229. Shanmugam and Senthilkumar P. [30] Sampathkumar K. Niranjan Prabhu. “Effect of Varying Glass cover thickness on Performance of Solar still: in a Winter Climate Conditions”. Abdelmotalip.T. 291-305.R. 4 1 [28] Sabah A. 77-82. Fuat Egelioglu and Atikol. Noorul Haq. NIT Durgapur . Journal of Solar Energy Engineering. 14. [6] Bilal A.S. Desalination 217. Mousa S. Boughali. 196. Procedia Engineering.Eldalil. Renewable Energy. May. 4. Vol. 32.D.A.B. 424-434. “Performance analysis of a V type solar still using a charcoal absorber and a boosting mirror”.N. [16] Khaled M. Mofreh H. 2002.M and Sunil Inkia. Desalination 173. 2012. “Mathematical modeling of an inclined solar water distillation system”. “Experimental Evaluation of A Single basin Solar Still Using Different Absorbing Materials: An Overview”. 4. Smith. 3.U. 2008. 2008. 2007. [22] Omara.M. “Energy efficient Solar Water Still”. Senthil Kumaran.K. “Solar still productivity enhancement”. Oct-Dec 2011. 2001. [4] [5] Bechki.Performance Analysis”. Amanallah Guizani and Mohammed Maalej.Akash. 937-946. 34. IWTC 13. Sponsors: Dr.K. August.Hamed and Kabeel.V. “Study of the Performance of the Inverted Solar Still Integrated with a Refrigeration Cycle”. April 2011. 172 of 446 Copyright @ Department of Electrical Engineering. “Experimental and theoritical study of a single basin solar still in Jordan”. 2010. 2009. 180.130/ 024502.N.R and Walke P. “Evaporative heat loss and heat transfer for open and closed cycle systems of a floating tilted wick solar still”. Blanco-Galvez. Desalination.M.212-223. “Enhancement of solar still productivity using floating perforated black plate”.A.Mohsen. International Journal of Engineering Research and Applications (IJERA).2007. No. “Studies on a single stage solar desalination system for domestic applications”.A. 1.H.Depth on Performance of a Solar still for Indian climatic sonditions”.S and Akash. [35] Velmurugan. “Thermal modeling of passive and active solar stills for different depths of water by using the concept of solar fraction”. Bilal A. 2002. [36] Velmurugan. [24] Rahul Dev and Tiwari. Bouchekima. “Performance analysis in stepped solar still for effluent desalination”. [33] Teltumbade T. “Productivity Enhancement of Stepped Solar Still. 245. Part-1”. Desalination 180. Desalination. 2003.V and Srithar. S E [13] Irene De Paul.A. pp. [29] Safwat Nafey.G and Vikky. Journal of Solar Energy Engineering. 80. 127. Fahmy. International Journal of ChemTech Research (IJCRGG).M. “An experimental study on an inclined solar water distillation system”. “An experimental wick-type solar still system: Design and construction”. 260. Desalination. Energy Conversion and Management 43. 2011. Chandrasekaran.S. [25] Rajendra Prasad. 2005.Akash. pp: 456-460. “Effect of dye on distillation of a single slope active solar still coupled with evacuated glass tube solar collector”.K.S and Mahcene. “Influence of water depth on internal heat and mass transfer in a plastic solar still”.J. 883-890. Desalination.B. Mahmoud Ben Amara. NSC. El-Samanoudy. Number 1. 217-230.B. Vol. 173. 246-265.S. 2005. Vol. 124.O. “Pilot plant testing of a new solar desalination process by a multiple effect humidification technique”. Desalination 277. 33. 267275. 281-287.J.G. XXXII National Systems Conference.E.A.Energy Conversion and Management 42. Renewable Energy. 65-69.M. [9] Hikmet S Aybar. Vol. JNTUH College of Engineering. Hyderabad and My Dear Parents. 2006. [32] Selva Kumar. [7] Fath H. [26] Rajesh Tripathi and Tiwari.K. 956-967. “New Concept for Improving Solar Still Performance by Using Vibratory Harmonic Effect Experimental Prediction. 1998.A and Mabrouk. 2011.Vijaya Kumar Reddy. Babay. 1.K .G.3. 2000. No. 2008.S.4. December 17-19. 79-88. Omar Osta and Yaser Elayan.P. “Annual performance of an active solar distillation system”.Abdul-Wahab and Yousuf Y Al-Hatmi. Energy. [10] Hitesh N Panchal.E.S. 2008. pp: 23-29. Abdelmotalip. 565-572. Vol.G. Issue 3. “Thermal economic analysis and comparison between pyramid shaped and single sloped solar still configurations”. [21] Nijmeh. Desalination 267.S.K. 2005.J. 63-70. Bouguettaia. 105-124. 607-612.N. “An Experimental Study on Single Basin Solar Still Augmented with Evacuated Tubes”. “Experimental Evaluation of a single basin solar still using different absorbing materials”.K and Verma. pp. No. 190. International Journal of Renewable Energy Research (IJRER). Solar Energy. “Effect of water depth on internal heat and mass transfer for active solar distillation”. Saravanan.V.J and Kumar.3. Tiwari. Abdelkader.H.

com performance over previously reported circuits [1-5]. The circuit performance is critically studied and verified through PSPICE simulation. Many active RC integrators and differentiators have been reported in the technical literature [1-4]. The use of grounded passive elements in the circuits makes them suitable for using in integrated circuits (IC). The circuit also enjoys the attractive features.K. There is a high impedance current output terminal Z. India sajalkpaul@rediffmail. However all of these circuits are in voltage-mode operation or use an excessive number of active/passive elements. low bias voltage and low sensitivity Fig. Recently in [5] a differential voltage current conveyor (DVCC) based dual input current mode integrator and differentiator is presented employing three grounded passive components. The performance of the circuits is demonstrated using PSPICE. I. waveform generation. such as. I Y1 0 0 0 0 0 VY1 I Y2 0 0 0 0 0 0 0 0 VY2 I Y3 0 = 0 VX 1 IZ 0 0 0 0 IX IZ 2 0 3 0 VY3 The port relationship may also be represented as IY1 VX IZ IY 2 V 1 Y1 IY 3 0 V 2 Y2 (1) V 3 Y3 IX 173 of 446 Copyright @ Department of Electrical Engineering. The proposed circuits are realized without any matching constraints. The proposed CM integrator and differentiator circuits are realized using only two passive components and without the requirement of component matching.com Department of Electronics Engineering Indian School of Mines Dhanbad. Current mode active RC integrator and differentiator circuits serve as very useful basic building blocks for various signal processing applications. low component count. It can be characterized by the following matrix relations between various voltage and current variables [6]. 4 1 Introduction Tunable active RC integrator and differentiator networks carry a lot of importance in microelectronic circuits and systems. In this paper.India das_rupam@rediffmail. Differentiator. Also the circuits have high output impedance.Differential Difference Current Conveyor (DDCC) Based Current Mode Integrator & Differentiator Rupam Das Sajal. Jharkhand. and calibration circuits. single resistor tunability and low sensitivity performance. 1. Integrator. Keywords— Differential Difference Current Conveyor (DDCC). of low bias voltage. process controller. which enables easy cascading in current mode. Terminal X is a low impedance current input terminal. active filters. which is attractive in IC realization. a simple circuit for realizing integrator and differentiator in current mode is proposed using a differential difference current conveyor. and Y 3 are voltage input terminals with high input impedance.Ccurrent Mode Circuits. CIRCUIT DESCRIPTION The symbolic representation of a differential difference current conveyor (DDCC) with only Z+ output is shown in Fig. Only two grounded passive components are used. NIT Durgapur (2) (3) . only two passive components.1: Symbol of DDCC The Y 1 . the circuit also enjoys attraction features of low component count. along with.Paul Department of Electronics and Communication Engineering Asansol Engineering College Asansol. Y 2 . Abstract— This paper presents a new circuit realization for current mode integrator and differentiator using single differential difference current conveyor (DDCC) as active device. S E A I Many active RC integrators and differentiators implemented with conventional operational amplifiers (OPAMPs) or current feedback amplifiers (CFAs) have been reported in the literature [6-9]. The passive sensitivities of the circuits are low.

β 1 . Fig.4 M11-M12 45 Identify applicable sponsor/s here. III. Fig. are two admittances of the circuit of Fig.5 0.5 . β 2 and β 3 represent the voltage tracking errors from Y 1 .5μm MIETEC process with supply voltage of +/-2. sRC I out I in 1 sRC To verify the theoretical analysis of the proposed circuits. The internal circuit diagram of DDCC used in simulation is shown in Fig. 2. having SIMULATION RESULTS S E I (5) where (RC) is the time constant of the integrator. F x (9) The sensitivities of integrator time constant and differentiator time constant are evaluated with respect to passive elements are summarized as S R =1. Reanalysis of the differentiator circuit assuming non-ideal DDCC yields.2: A generalized circuit topology for realizing various analogue functions Routine analysis of the circuit yields the following voltage transfer function I out I in 1 / Z1 1/ Z 2 (4) I out I in 1 . The generalized topology for realizing various analogue functions. the configurations have been simulated using SPICE circuit simulation program. Y 2 .5 0.5 0. 4 1 A Where y 1 =1/Z 1 and y 2 =1/Z 2 . S C =1 It reveals that the time constant of the proposed integrator and differentiator circuits have low sensitivity to passive components. delete this text box (sponsors). II. Y 3 to X terminal. NIT Durgapur L ( m) 0.where α represents the current tracking error from X to Z terminal. 1 (6) (b) Differentiator: If Z 1 =1/sC and Z 2 = R then a differentiator is realized. (a) Integrator: If Z 1 = R and Z 2 = 1/sC then an integrator is realized. Reanalysis of the integrator circuit assuming non-ideal DDCC yields. The DDCC is simulated using PSPICE level 3 parameters in 0. SENSITIVITY ANALYSIS The relative sensitivity of a function F with respect to a parameter x is defined as S xF x F . The dimensions of the MOS transistors are given in Table I. Table-I: Dimensions of the MOS transistors TRANSISTOR W ( m) M1-M4 0. having I out I in sRC (7) where (RC) is the time constant of the differentiator.3 [10]. 2. based on a differential difference current conveyor (DDCC) with only Z+ output is shown in Fig. CMOS process model parameters given in Table II. If no sponsors. Ideally β 1 = β 2 = β 3 = α = I out I in sRC 1 (8) 1.3 Internal structure of DDCC.7V . 174 of 446 Copyright @ Department of Electrical Engineering.5 0.8 M5-M6 10 M7-M8 4 M9-M10 14.5V and biasing voltage of -1.

[3] [3] J. K.68E-10 +MJSW=. [2] [2] J.120 GAMMA=0.81 +NFS=0. Vol. 331. Lee and S.03E-6 ETA=0 +VMAX=113E3 NSUB=2.4(a) and Fig.76 KAPPA=2 AF=1 +WD=.14E-6 CJ=85E-5 MJ=0. Vol. The realization is free from matching constraints.81 LD=0.38E-10 +CGBO=3. The input and output waveforms are shown in Fig.) waveform.37.TableII: 0.15E-6 RS=417 RSH=2.2E11 .1E-6 RS=886 RSH=1.631 CGSO=1.429 CJSW=4.0E-8 TPG=1 VTO=. NIT Durgapur . The proposed circuits are insensitive to passive component changes and suitable for IC implementation.MODEL PT PMOS LEVEL=3 +UO=100 TOX=1E-8 TPG=1 VTO=-.4(b) respectively. R. Nandi.8E-6 +XJ=. No. “Dual-input singletunable integrators and differentiators using current feedback amplifier”. Int. with the designed passive component values as: C = 80nF and R = 40K.MODEL NT NMOS LEVEL=3 +UO=460. Venkateswaran. It verifies that the proposed differentiator circuit also performs well. 2003. L.42 +NFS=1. Electronic letter. . I.333.45E-10 KF=3.905 +THETA=0. The proposed circuits are expected to find applications in instrumentation systems. L. Vol.07E-28 DELTA=0.04E-6 ETA=0 +VMAX=130E3 NSUB=1.761 PHI=0.357 CJSW=5.5 TOX=1.129 GAMMA=0.08E-29 DELTA=0. CONCLUSION In this paper new current-mode integrator and differentiator circuits using single DDCC and two grounded passive components are presented.109-115. REFERENCES P.38E-10 +CGBO=3. Electronics letters. pp.67E-10 +MJSW=.38E-10 CGDO=1.11E-6 CJ=76. 4 1 A Fig:5(b) Shows output square wave of the Differentiator S E I Fig. Liu.1 AF=1 +WD=.905 +THETA=0. “Dual-input RC integrator and differentiator with tunable time constant using current feedback amplifiers”. It verifies that the proposed integrator circuit functioning properly. It is evident that there is good quality of conversion from rectangular to triangular (appx.4E-5 MJ=0. a triangular current waveform I in with 10A (PP) was applied.62 JS=1. S.69 KAPPA=0.5(b) respectively.71E17 PB=. J. Electronics. “Integrator and differentiator with time constant multiplication using current feedback amplifier”.2.302 CGSO=1.73 LD=0.45E-10 KF=1.38E-6 +XJ=0.4(a) Shows input Square wave of the integrator IV. Sanyal and R. Lee and S.5(a) and Fig.08E17 PB=.58 JS=.52E11 Fig:5(a) Shows input triangular wave of the Differentiator To test the performance of integrator circuit a rectangular current waveform I in with 1mA (PP) was used with the design values of R = 50 and C = 50 F.5 LEVEL-3 m MIETEC CMOS Process Model Parameters Similarly for differentiator circuit. The input and output is shown in Fig. signal processing circuits and analog filters. pp. I. K. Nagaria. 2001.90.911 PHI=0.38E-10 CGDO=1. Liu.35. pp. [1] Fig:4(b) Shows output waveform of the integrator 175 of 446 Copyright @ Department of Electrical Engineering.19101911. 1999.

Vol. IEEE MELECON 2004. U. 1992. pp. 1990.[4] [5] [6] [7] [8] [9] [10] J. 1988). “Dual-input current mode integrator and differentiator using single DVCC and grounded passive elements”. pp. Minaei. Communication and Computers. 709-714.V.” IEEE Transactions on Instrumentation and Measurement. pp. Recent Rsearches in Circuits. and R. 39. 39. “A high-quality dualinput differentiator.K.” IEEE Transactions on Instrumentation and Measurement. IEEE Transactions on Instrumentation andMeasurement. NIT Durgapur .C. 4 1 A S E I 176 of 446 Copyright @ Department of Electrical Engineering. Sarker. 41.Naveen Kumar. Vol. 2004. S. Franco. Vol. Design with Operational Amplifiers and Analog integrated circuits (New York: McGraw Hill.C.5.123-126. Tsao. and R. Nandi. Nandi. S. “Increased timeconstant dual-input integrators.”Differential Difference Current conveyor based cascadable voltage mode first order all pass filters”. S. Tsao. and H. S. 726-729. W. 1990.Murlikrishana. Sanyal.”. 41.Y.K. Sanyal. pp. Sarker.2011. pp. Lee. No. Y. 1992.lal . Systems. Lee and H. 672-673. IEEE Transactions on Instrumentation and Measurement.709-714. P. J. “True RC integrators based on current conveyors with tunable time constants using active control and modified loop technique”. Vol.K.128-131. pp. U.W. “True RC integrators based on current conveyors with tunable time constants using active control and modified loop technique.S.R.

and then proceeded with the creative invention of mono-energy compressed air engine. Chau and Wong [1] emphasized the issue of energy expenditure with respect to environmental protection. The reading shown by pressure sensor which is placed at the exhaust of the engine is 270 KPa at 1600 rpm. Based on the fact that air can be used as a fuel to power the vehicle. In our project we uses compressed air to run the modified compressed air engine and simultaneously investigates various dynamic parameters of the engine. This is because of the lift presented in both inlet and outlet valve. I.08 Nm and 1. Compressed air was also used in some of the vehicles for increasing the initial torque using turbo chargers [5. One of the best solution of above problems is air powered car. Air as a fuel has got lots of advantage like pollution free. The cylinder pressure was monitored using an pressure sensor.com Department of Mechanical Engineering GLA University Mathura rathoresamurai@gmail. 100 cc. which is fixed in place of spark plug. hybrid electric vehicles and 4 1 A S E I Keywords – Compressed. abundantly available and free of cost. The measured outlet air temperature is 255 K at 400 rpm and 280 KPa at 1600 rpm. NIT Durgapur . Engine. Nelson [2] find out the performance and requirements of hybrid electronic vehicles (HEV). 177 of 446 Copyright @ Department of Electrical Engineering. single cylinder. Compressed air can be stored in the air tank for use anytime. Torque hydrogen vehicles. The output torque and output power obtained is 12. Therefore lots of R&D activities focus on zero pollution vehicles and green engines including other alternates for transportation like electric vehicles. and fuel cell electronic vehicles can be used as healthy and environmental friendly replacement of conventional vehicles [3]. Guy Negre [7] designed a bi-energy engine incorporating gasoline and compressed air. diesel. hybrid electronic vehicles (HEV). These fuels are depleting day by day at a very fast rate and also exhaust gasses produced by these fuels are very harmful to the atmosphere and to the mankind. fuel cell vehicles. In 1991. The measured air pressure at exhaust of the engine is above atmospheric pressure. All these alternatives have definitely reduced the emission of exhaust gasses and hence pollution also but complete eradication is still rigorously pursued. Power. INTRODUCTION Today in every corner of the earth most commonly used power generating system for transportation sector is internal combustion engine which mostly uses conventional fuels like petrol. if the storage is sufficient then there is no need to run the air compressor continuously. 6]. The electronic vehicles (EV). The engine was modified and made to run on two strokes using a modified cam shaft. The experimental kit consist of an four stroke. batteries used in EVs and HEVs takes lots of time to get fully recharged[4]. the fuel cell have got a problem of recycling.Performance Analysis of 2-Stroke Gasoline Engine by Using Compressed Air Rahul Nath Tripathi Pushpendra Kumar Singh Rathore Department of Mechanical Engineering GLA University Mathura rntrahulraj@gmail. spark ignition engine.1 Kw at 400 rpm and at 1180 rpm respectively. and compressed natural gas. Other problems associated with these types of vehicles are. An experimental kit has been developed in order to find out certain dynamic parameters of compressed air engine.com Abstract – This paper shows that compressed air can be used as alternate fuel for internal combustion engine and gives an powerful option to mankind to replace the conventional fuels.The pressure-volume diagram shows that cylinder pressure increases and decreases gradually. this study uses piston type engine for feasibility study for vehicle applications.

All the data is then gathered to a computer for further study.0 X 51. Further on this compressed air is feed into the engine cylinder via inlet valve using pipe. Due to this there is a small lift at inlet valve and at exhaust valve. Four processes namely: intake. the detail specifications are given in table 1. for monitoring the air pressure two pressure transducers are placed at inlet and outlet of the engine. Z121) is fixed in order to monitor the cylinder pressure. To accomplish this goal the company fitted camshaft was replaced by a modified cam shaft as shown in fig.9 KW of power. In order to run the engine using compressed air all these thermodynamic processes must be completed in one revolution of crankshaft. Similarly. EXPERIMENTAL PROCEDURE A single cylinder. expansion . 1. This expansion of gas lasts till the piston reaches the BDC. NIT Durgapur . The experimental set up consist of an air compressor (ELGI. The pressurized air strikes the piston and process of expansion starts. Specification of IC Engine Company Hero Honda. The new camshaft was first designed then analyzed using pro-e and then it was fabricated according to the calculated dimensions. For measuring the output power a dynamometer having electromagnetic brakes of range 30 Nm is used. and it remains open for the next 210° of crank rotation. In order to use it in our experimental kit it has been disassembled from the motorcycle. which is adjusted using starter motor. HV 21 100) which produces 14. When the engine starts the position of piston must be at the TDC. 1 Modified camshaft of compressed air engine 178 of 446 Copyright @ Department of Electrical Engineering. The air will move out from exhaust valve.e. For monitoring the inlet and outlet temperature digital thermometer are placed at the inlet and outlet of the engine respectively. India Engine Type 110. Before the piston reaches the TDC the exhaust valve closes and compression will start. and then the inlet valve closes.0 Cm3 (4-Stroke) Horse power 7. expansion. The inlet pressure is equal to the air storage pressure. Table 1. A tank having storage capacity of 600 L is connected with the compressor in order to store the compressed air. The movement of piston from TDC to BDC will complete two processes.04 Nm @ 4500rpm (Gasoline) Bore X Stroke 50.5 mm 4 1 A S E I The thermodynamic cycle of the compressed air engine starts with inlet valve opens and pressurized air from the compressor enters the cylinder. which in the case of conventional IC engine get finished at 720° movement of piston. The exhaust valve is started to open when the crank is at 75° after TDC. The spark plug of the engine is removed and in its place a pressure sensor (AVL. the inlet valve will open when the piston is at TDC i. The movement of the piston from BDC to TDC will complete the next two processes.The conventional IC engine was used in motorcycle for transportation in India. 4-stroke engine of Hero Honda make was modified and converted to 2-stroke in order to run on compressed air. valve timing diagram changes significantly.8 HP @ 7500 rpm (Gasoline) Torque 8. With the modified cam profile the Fig. at 0° and remains open at 20° after TDC. After 20° to TDC the closure of the inlet valve is started and it is completely closed when the crank is 45° after TDC. The exhaust valve starts to close when the crank is 75° before TDC. will be finished at 360° movement of the piston. discharge and compression takes place in order to complete one thermodynamic cycle of the engine. From BDC the piston starts moving towards TDC and exhaust valve opens. compression and discharge. Thus in the case compressed air engine all the four processes namely : intake. The crankshaft completes two revolutions during these processes.II.

Outlet pressure sensor 11. 7 shows the monitored outlet pressure and fig. The graph shows an incremental movement in cylinder pressure with the increase in inlet air pressure and it remain constant after a certain increase in value. Engine 9. Fig. 6 shows calculated efficiency of the engine at varying supplied air pressure and rotation speed. The highest torque output 12. The highest power output 1. Results and Discussion A S E I Torque (Nm) Various performance parameters of the modified engine were measured at a air pressure ranging from 400 KPa to 800KPa and at rotation speed varying from 400 to 1600 rpm. In order to find out the variation of temperature and pressure in the modified engine thermometer and pressure sensors were installed at inlet and outlet. Starter motor 4 1 III. Fig 3 and 4 shows the variation of torque and power with rpm. 3 Engine torque acquired at different rotation speed and supplied air pressures.1 KW was obtained at 1180 rpm and at 800 KPa of pressure. Power supply 14. The output torque and output power of the modified engine were calculated on the experimental kit at varying supplied air pressures. 14 400 KPA 12 500 KPA 10 600 KPA 700 KPA 8 800 KPA 6 4 2 0 400 800 1200 1600 Rotation Speed (RPM) Fig. 179 of 446 Copyright @ Department of Electrical Engineering. Inlet thermometer 5. The highest efficiency of 15% is achieved at lower rotation speed at around 400 rpm and then at higher rpm the efficiency gets reduced. Fig. Optical tachometer 6. Torque Sensor 12. Buffer Tank 3. Flow Meter 7. Compressor 2.08 Nm was obtained at 400 rpm and at 800 KPa. 5 shows PV diagram for cylinder pressure at 1100 rpm. Fig. Inlet Pressure Sensor 4. Electromagnetic brakes 13.Fig. Outlet thermometer 10. Data acquisition unit 1. NIT Durgapur . 2 Experimental Setup 8.

Cylinder Pressure (KPA) Outlet Pressure (KPA) 400 400KPA 800 1200 1600 Rotation Speed (RPM) Fig. 16 400 KPA 14 500 KPA 12 600 KPA 10 700 KPA 8 800 KPA 6 1. 6 Calculated efficiency of the engine at various rotation speed and supplied air pressure 0. 180 of 446 Copyright @ Department of Electrical Engineering. NIT Durgapur . 7 Measured value of air pressure at engine outlet at varied rotation speed and supplied air pressures.2 300 0 800 1200 1600 Rotation Speed (RPM) S E I 800 400 KPA 700 500KPA 600 600KPA 300 200 150 600KPA 700KPA 800KPA 100 50 400 800KPA 400 500KPA 0 700KPA 500 4 1 250 A Fig. The inlet temperature lies between 309 to 313K and temperature at the outlet decreases and lies below the room temperature. 9 shows monitored temperature at the engine inlet and outlet respectively at various rotation speed.2 400KPA 1 500KPA 2 600KPA 0 0.8 0. 4 Power produced by the engine at varied rotation speed and supplied air pressures.Power (KW) Efficiency (%) 8 and fig.4 0. As the inlet pressure increases from 400 KPa to 800 KPa The outlet pressure also shows a significant increment of 270 KPa. 200 100 0 25 50 75 100 Volume (Cm3) Fig.6 4 700KPA 400 8OOKPA Rotation Speed (RPM) 800 1200 1600 Fig. Further if we increases the inlet pressure the pressure at the outlet will also increases. 5 Measured cylinder pressure at 1100 rpm and varied supplied air pressures.

R. “Electric and hybrid vehicle technology”. Onder. Guzzella. C. Ochsner. 226. Available online . Voser. Donitz. C. Proc. C. Onder. Inst. This outlet pressure can be utilized as feed back to the storage tank in order to reduce the work done by the compressor and hence increasing the efficiency of the system. The value of outlet pressure rises upto 270KPa at inlet pressure of 800KPa and it will continue to rise if we uses higher inlet air pressure. Guzzella. hybrid electric-vehicles and fuel cell electric vehicles: state of the art and perspectives:. piston pin. In-cylinder boosting of turbocharged spark-ignited engines. therefore there was no need of combustion and hence there is no production of harmful exhaust gasses. In-cylinder boosting of turbocharged spark-ignited engines. Mech. NIT Durgapur . Part D J. Eng. 2012.43:1953-68. Proc. I. 312 700 KPA 311 800 KPA 310 309 308 307 400 800 1200 1600 Rotation Speed (RPM) Fig. The outlet cylinder temperature is 257 K at 800 KPa and at 400 rpm which will continue to fall if the air pressure is increased furthermore. “Overview of power management in hybrid electric-vehciles” Energ Convers Manage 2002.26(4):9-26. and crank pin. Part D J. The outlet temperature falls to 258K at 800 KPa at 400 rpm. Eng. T. Conclusions 314 400 KPA 313 500 KPA 600 KPA The data gathered during the experiment shows that the compressed ignition engine if made to run using compressed air as a fuel by doing some modification on the cam shaft then it will produce 1. J. T. C. Voser. L. http://www.1997. 226. 8 Measured air inlet temperature at varied rotation speeds and supplied air pressures. Robinson. The cylinder pressure will continue to rise with the increase in value of inlet air pressure. Eng.V Mierlo.91:2-26 G.F. Autom. Since the power inside the engine is produced by the expansion of the compressed air. which will lead to severe lubrication problem to certain components like piston rings. Part 1: Model-based design of the charge valve. C. L. Autom. The low temperature is obtained at low rotation speed but as the speed is increases above 1600 rpm the temperature is also increases. References A [1] [2] 285 S E I 400 KPA Outlet Temperature (K) 280 500 KPA 275 270 265 [3] 600 KPA [4] 700 KPA [5] 800 KPA 260 [6] 255 250 [7] 245 4 1 K. 9 Measured air outlet temperature at varied rotation speeds and supplied air pressures. 181 of 446 Copyright @ Department of Electrical Engineering. Part 2: Control and experimental verification. “Electric vehicles. Donitz. Chau. Ott. J Power Source 2000. Vasile.Inlet Temperature (K) IV. Ann Chim Sci Mat 2001.com/ (accessed on 27 August 2013) 240 400 800 1200 1600 Fig. Y. Nelson. Inst.theaircar. Wong.08 Nm of torque using air pressure ranging from 400 KPa to 800KPa. Maggetto. “Power requirements for batteries in hybrid electric vehicles”.1 KW of power and 12. UK:International press. G. Eng. C. 1408–1418. The efficiency of the engine is high at lower rpm and stays low at higher rpm. Mech.T. Motor Development International (MDI) web page.S. 1564–1574. Furthermore if the inlet pressure increases the outlet temperature is expected to drop. This is because at higher rpm air gets less time to expand inside the cylinder. 2012.

Keywords— Differentiator. Padre Conceicao college of Engineering.(z1+1/z1). FIR. The Tick’s integrator on which [4] is based. obtains an Ni (z) ≈ 1. Di () At high frequencies.com. N i () 1 . K. H i () A S E I -π < ω ≤ π . A digital integrator finds applications in diverse areas such as Radar [1] and Sonar [2] signal processing and vibration control [3]. (4) respectively. such that. response. A. H i () (2) At low frequencies the numerator (denominator) function Ni (ω) (Di (ω)) are synthesized such that Ni (ω) ≈ 1. (7a) and its magnitude response is Di (ω) = 2sin(ω). subjected to degree of approximation suggested by (3) and (4). II. its frequency response will be. Similarly. the acceleration can be computed from the position measurements using second order differentiator [1]. Abstract—— In this paper. instrumentation. (7b) The magnitude of Di (jω) approximates ‘ω’ over low frequencies. Frequency Integrator. The magnitude response of an ideal integrator is. (9) then. Hdiff(ω) =|ω|.The differentiators have applications in signal processing systems such as. over a range of frequencies thus satisfying (5). by modifying its gain. Integrator and differentiators thus obtained have magnitude response close to the ideal ones over a wide range of frequency. the operator z-1is defined as a binary fraction (1/zq. Di (jω) = 2sin(ω)exp[j(ω+π/2)].Design of Wide Band Digital Integrator and Differentiator Jayalaxmi Devate. q an integer) of the sampling period ‘T’. (3) Di(ω) ≈ ω. biomedical and communication systems. For example. Pai E&C Department. Bengaluru. S. (1a) (1b) DESIGN N i () . In [4] Di (ω) is synthesized by an FIR filter function of second order. we propose the design of wideband digital integrator and differentiator where the unit delay. India. (10) This modification actually increases the range of frequencies over which (4) is satisfied. Differentiator is obtained by inverting the transfer function of the integrator.If the denominator function (6) is modified as Di (z) = z (z-1). z2=1/z1 and p = . in radars.S. M. (8) Where. Magnitude Response. Di (jω)=2sin(ω/2) exp[j(3ω/2+π/2)]. -π < ω ≤ π . control systems. NIT Durgapur . Dr. India. 4 1 Di (z)=z2-1. the ratio between synthesized numerator and denominator function approximates an integrator response. Relative Error. namely. Integrators Let the magnitude response of the integrator be. Ni (ω) [4] is synthesized by an FIR filter function of second order of the form Ni (z) = z2+pz+1= (z-z1)(z-z2). Goa.R. The transfer function for integrator synthesized according to the above design achieves improved bandwidth over the full-band. INTRODUCTION Hint(ω) =|1/ω|. And that of differentiator is. I. D i () (5) (6) Its frequency response is. Verna. namely. ETC Department.Y. It can be seen that if 182 of 446 Copyright @ Department of Electrical Engineering. Ramaiha Institute of Technology. E-mail:jayaa_2002@yahoo. Kulkarni Dr.

where z’=z1/r with a frequency rfs=2qfs. we encounter a constraint which stipulates that the sampling clock period ‘T’ is also the unit delay. by making Ni1(ω) equal to unity. Fig. Relative error in dB (18) Alternatively. (16) (17) RE is as low as -149. Let the magnitude response be. (13) and the corresponding frequency response.3 and Fig. i1 Relative error is less than 1% over the entire range except at ω=0.1. H i1 () D i1 () 32 sin( / 32) Fig2. H i1 () N i1 () D i1 () . and (15) Ni1(jω)= 2cos(ω/32) exp [j(3ω/32)] Therefore.2 shows the plot of RE in dB. H i1 ( j ) N i1 ( j ) Di1 ( j ) cos( / 32) 32e j / 2 sin( / 32) . it is easy to obtain both the clocks from single clock generator using a binary divider chain. The error in proposed approximation is estimated by a Relative error (RE) defined as (12) The denominator Di1 (ω) is synthesized by an FIR filter function with r=24.4 show the new magnitude and error response respectively given by (20). 183 of 446 Copyright @ Department of Electrical Engineering.32% at ω=π.we are able to synthesize in (2) a denominator of the form Di (ω) = ksin(ω/r).which is close to ideal. (11) where. In the literature [4-6]. Fig. 1 Magnitude response of proposed integrator. NIT Durgapur . Maximum value of RE is 0. Di1 (z’) =32 z’ (z’-1). one exclusively for sampling with frequency fs and period ‘T’ and the other to determine the system delay (z’)-1. 4 1 A (14) The numerator Ni1 (ω) is synthesized by an FIR filter function (19) S E I Ni1 (z’) = z’ (z’+1). k is a gain constant and ‘r’ is preferably a large binary integer r=2q. q an integer. Di 2 ( ) 32 sin( / 32) (20) Fig. then it can be shown to yield a wideband integrator/differentiator.7dB at very low frequencies. N i 2 ( ) 1 . If we can split the clocks into two. RE | 1 ωH (ω) | . an improved integrator response is obtained as shown below. The new magnitude response is H i 2 ( ) shown in Fig. Di1(j ω) =32(2sin(ω/32) exp [j(3ω/32+π/2)]. If ‘r’ is a binary power. and its magnitude response becomes N i1 () cos( / 32) . unit delay (z’)-1 as where z’=z 1/16 .

The magnitude response of (23) is shown in Fig. and.005. Relative error in dB (22) COMPARISON WITH EXISTING INTEGRATORS AND DIFFERENTIATORS The proposed integrators and differentiators have small relative error over a wide range of frequencies compared to [4-6]. cos( / 32) The relative error is again less than 1% over the entire range. Thus wideband differentiators obtained by inverting (18) and (20) are. III. H d1 () Fig6. Differentiators A wideband differentiator is obtained by using the reciprocal of (2) as its transfer function i. S E I B. Fig. 5 Magnitude response of proposed differentiator.6 below shows the plot of RE in dB. 1 H d (z' ) . 4 1 A Fig4. Fig. Di 2 () N i 2 () 32 sin( / 32) . RE is as low as -155.16% at ω=π. Relative error in dB Note that the Maximum value of RE is 0. Di1 () N i1 () 32 sin( / 32) . 3 Magnitude response of proposed integrator. NIT Durgapur . Fig. (20) H i (z' ) where Hi(z’) is the transfer function of the integrator. H d 2 () respectively.8dB at low frequencies. 1 (23) In the clocks (or frequency frequency CONCLUSIONS proposed integrator/differentiator. The absolute error at ω=0 is zero and is maximum at ω=π which is of the order 0.. IV. Thus the response given by (20) is better than that of (18). The high clock can be termed the system clock and the low clock or the sampling clock is obtained by a ÷2N 184 of 446 Copyright @ Department of Electrical Engineering.5.e. there are two rather a single clock and a divider chain).

Suggested approximations give the required response through simple computations and have error less than that in [4-6].4.. M. Sept. 2006.345-353. it is shown that a wideband integrators/differentiators are obtained by using a transfer function of order one (or a single delay).4." IEEE Trans.pp. 936-40.Systems Signal Process. New York. As a result. Any function satisfying (3) and (4) may be used to get the required response. No. M. ”Novel Digital Integrator and Differentiator” Electron. AI-Alaoui.R.4. II: Express briefs. 29.” Mech.9. Lett. Hodges. NIT Durgapur . pp. Mc Graw-Hill. July 1992. February 1993. Principles of underwater sound. 28( 15). 53.I Skolnik. pp. By splitting the clock. Vol.J Uriek. 376-1378.J Elliott.counter following high frequency clock.. A.vol. Circuits Syst. and S. AI-Alaoui. Vol. The proposed scheme can be implemented by a simple hardware or software. 1983. pp. no stability problems are encountered with the single pole placed on the unit circle in z’-plane. T. 4 1 A S E I 185 of 446 Copyright @ Department of Electrical Engineering. New York 1980. A. 1376-1377. Introduction to Radar Systems. P. Nam Quoc Ngo "A New Approach for the Design of Wideband Digital Integrator and Differentiator.vol. Nelson. Mc Graw-Hill. REFERENCES [1] [2] [3] [4] [5] [6] M.. no. Lett.1990.”Novel Approach to Designing of Digital Differentiators” Electron.A.No. “The design of a precision digital integrator for use in an active vibration control system. J.

WirelessHART. In this condition small negligence may cause heavy human loss and economical loss. Maharashtra Asst Prof.)JNEC Aurangabad.C. II. XDM2510H With WirelessHART we can design such system.e. assets. ME(E. field devices and mobile devices i. Localization refers to the process of obtaining location information on a node with respect to a set of known reference positions. Ukarande Shaikh F. Field devices are installed in factory and location of field devices is stored in network manager when there is gas leakage and worker carrying handheld device moving in leakage area then network manager will determine location of worker by RSSI based localization method and alert will be given to worker and his life can be saved. This system can give real time data for gas leakage areas also it will track location of worker moving in that particular area and alert him from hidden danger. In design of this system we will have two types devices mainly stationary devices i.JNEC Aurangabad. Localization. so that his life can be saved by alerting him about hidden danger[6]. Now when worker moves in such areas where chemical or gas leakage has occurred then at that time signal strength of workers handheld device will be calculated and neighbour table of network manager will be automatically refreshed and when signal strength of field device and handheld device are matched at that time workers location is determined and alert will be given to worker by activating buzzer.e. 4 1 A S E Key words: Wireless sensor networks. It is more important to know location of field engineer or worker. I I. Location of field devices is already stored in network manager database. Network manager will maintain neighbour table. LOCALIZATION Localization in wireless sensor network has grabbed attention of many researchers recently. With advancement in wireless sensor networks there is huge necessity to develop such a safety system which would be reliable and will offer guaranteed performance in such harsh environment. infrared [6] are some techniques used for localization but GPS fails in indoor environment because GPS and infrared signals can not go through wall and other obstacles and power consumption is also more in such 186 of 446 Copyright @ Department of Electrical Engineering.Tracking Mobile Targets Through WirelessHART Vaibhav V. INTRODUCTION In large refineries and petrochemical industries there is possibility of hazardous gas or chemical leakage. and portable instruments can increase management efficiency. The real-time knowledge of the location of personnel. .com Abstract— WirelessHART has emerged as strong alternative to other WSN. I. Maharashtra ukarandevaibhav@gmail. worker carrying handheld device. NIT Durgapur .In this paper we are going to present design for tracking mobile targets through WirelessHART in process industries. GPS.com faisalshaikh29@yahoo.

and Application layers.e. As of version 7. TDOA(Time difference of arrival) and AOA(Angle of arrival) out of which TDOA and AOA can not be used be because they suffers from line of sight problem and requires special hardware. WirelessHART Architecture. Network.. WirelessHART is first open wireless communication protocol. WirelessHART could suffer with a interference from several other sources.4-2006 DSSS physical layer with a data rate of 250 Kbps and operating frequency of 2. WirelessHART uses Time Division Multiple Access (TDMA) technology to communicate between network devices.4-based wireless mesh network.4-2006 4 1 A S E I Figure 1.15.devices.483 GHz.4 radio platform operating in the license-free 2. it is built on an IEEE 802. Most automation networks in process industry are based on traditional 420mA analog wiring. So we will make use of RSSI method while designing our application For use of RSSI we do not require any special hardware and it can be implemented via software also. HART also provides IEEE 802. Figure 1 shows It is based on IEEE 802. and cost effective communication. WIRELESSHART BASIC FEATURES OF WIRELESSHART WirelessHART [2] is an international standard defined by IEC 62591-1 as of march 2010. Transport.From experiments it is found that with increase in distance error in RSSi measurement increases[7]Another method for localization is to divide area in to number of regions and then locate workers location if he is in that particular region. WirelessHART [2] is a wireless networking standard based on HART that adds wireless flexibility to an existing HART network. All devices are time synchronized and communicate in pre-scheduled time window which enables collision-free. “hops” across the 16 channels between 11 187 of 446 Copyright @ Department of Electrical Engineering.4 GHz ISM band.4GHz band.15. standard but specifies its own Data-link.15. There are various distance indication methods are available[4]such as RSSI(Received signal strength indication).. The HART[1] (Highway Addressable Remote Transducer) Protocol is the global standard for sending and receiving digital information across analog wires between smart devices and control or monitoring system. To tackle this problem WirelessHART uses FHSS technique i. wired HART technology plays a critical role because the digital information is simultaneously communicated with the 4-20mA signal. Like other technologies that use the 2. power-efficient. Predefined time slots also enable the Network Manager to create the optimum network for any application without user interference.15. NIT Durgapur . WirelessHART uses OSI model and is based on the PHY layer specified in the IEEE 802. III.4-2.

it is critical to keep the clocks of all devices in the network synchronised. It provides mechanisms for devices joining and leaving the network.to 26 defined by the IEEE 802. When the device receives a packet containing a graph id. There is only one active network manager for whole network..network becomes easy When interference or other obstacles interrupt a communication path. A graph route consists of several different paths from source to destination. Field Device: These are devices that are connected to the process or to the plant equipment. Network Figure 2 shows the different Network device types that comprise a WirelessHART network.Gateway Manager: Enables communication between host applications and WFDs in the WirelessHART network. As WHART uses TDMA to avoid collisions by assigning timeslots .The Data Link Layer (DLL) provides mechanisms for time synchronisation and transmission between neighbours. WirelessHART Components. Router is a special kind of WFD that does not interfere with the process. Since all devices can be treated equally in terms of networking capability. and expansion of a WirelessHART . Each device maintains a graph table with a list of graph ids. information in the graph table is used to decide the next hop. Channels which are affected by consistent interferences could be put in the black list and can be disabled. Security: WirelessHART gives hop-to-hop security measures through encryption and message authentication on the Network and Data-link layers. Graph routing: Graph is collection of paths created by network manager. this benefits the complete network. All WirelessHART devices have routing capability. installation. Network manager is responsible for collecting and maintaining diagnostics and overall health of the network and reporting these diagnostics to the host application.15. So in this way messages are routed in network manager. and optimization of the WirelessHART network. NIT Durgapur . WirelessHART 188 of 446 Copyright @ Department of Electrical Engineering. A S E I 4 1 Figure 2. Automatic clear-channel assessment (CCA) before each transmission and channel "blacklisting" can also be used to avoid specific areas of interference and minimize interference to others with help of blacklisting feature which is not available in ZigBee. WirelessHART uses two different kind of mechanisms for message routing. Source routing: It is not concerned with process related messages it work only during network diagnostics as a supplement to graph routing. Network Manager: It is responsible for the overall management. scheduling. Handheld Device: Supports direct access to adjacent WirelessHART field devices. the network immediately and automatically reroutes transmission to other path And because WirelessHART uses an path optimized.4 radio standard. redundant mesh topology.Graph routing and source routing. as there are no reduced function devices (RFD) like in ZigBee[3].

PIC18F24K22 IC. Signal strengths from field device as well as mobile device will be matched by network manager using RSSI and based upon computation he will discard untrustworthy pairs determine location of handheld device whose difference with field device is minimum. [5].So when there is chemical leak or gas leakage occur and worker carrying handheld device moves near to leakage area at that time neighbour table will get refreshed.MQ-135 CO2 gas sensor. Join keys are provisioned using a hand-held device. Our Solution is as follows: Firstly signal strength of field devices is stored in database of network manager. PROPOSED IMPLEMENTATION Hardware Platform: We are using RFM’s XDM2510H WirelessHART RF Transceiver Module[6]. METHODOLGY In WirelessHART even though various commands are specified already but no command has been specified for location determination. IV. Network keys are used by data link layer to authenticate messages on a onehop basis. So this is the challenging part when we think about designing localization application.Utilizes AES-128 block ciphers with symmetric keys for the message authentication and encryption. Figure 3: Discarding Untrustworthy Pairs 4 1 A S E I V. Join keys are unique to each device.. session and join keys.Since neighbor health reports are sent periodically (it can be set by the network manger) to the network manager. User can describe its own commands at application layer. Network manager maintains neighbour table which gets updated or refreshed after certain time and which can be set by network manager Locating a worker with mobile device is done by the network manager using trilateration method.Session keys are allocated for device-to-device communications. CLOC K CIRCU IT MCU POWE R ON RESET WHAR T MODE M INTER FFACE WHAR TMOD EM SENS OR INTER FACE CIRCU ITT CO2 SENS OR PROGRAM MING PORT Block Diagram implementation. POWE R SUPPL Y for 189 of 446 Copyright @ Department of Electrical Engineering.XG2510HE gateway manager[6] . WirelessHART defines Network. NIT Durgapur proposed .

and deterministic. Shambhu Upadhyaya.RFM’S XDM2510H module[6] provides WirelessHART compatibility. Mohammad Iftekhar Husain. NIT Durgapur . The XG2510HE acts as both as a gateway and network manager for up to 250 XDM2510H modules (motes). but such a implementation is very much costly. IEEE SECON. StefanSvensson. The XDM2510H is tailored for use in battery.com/products/wirelesshart [7] Is RSSI a Reliable Parameter in Sensor Localization Algorithms – An Experimental Study by Ambili Thottam Parameswaran. Xiaoyan Li. Fredrik Hekland “A Comparison of WirelessHART and ZigBee for Industrial Applications” IEEE Paper 2008. In combination. energy savaging and loop powered devices for applications that demand robust wireless network reliability. J et al (2008)). The XG2510HE is designed to manage a network of RFM’s XDM2510H radio modules that use Dust Networks’ low power radio technology. excellent communications reliability and long battery life in a wide range of industrial automation applications. Deji Chen. [2] Song. The Limits of Localization Using Signal Strength: A Comparative Study. “HART7 Specification”. Richard Martin. [3] Tomas Lennvall. reliable wireless mesh network. October 2004. When worker carrying this handheld device move in leakage area his location will be tracked as explained in methodology. low latency.Aloysius K.HART Communication Foundation. September 2007.Wei Dong.4 GHz frequency operation and also supports proprietary protocols. [4] Eiman Elnahrawy. 190 of 446 Copyright @ Department of Electrical Engineering. Mok. [6] www.rfm. The XDM2510H employs dynamic network optimization and intelligent routing to achieve carrierclass data reliability.Mark Nixon “A Locationdetermination Application in WirelessHART” 2009 15th IEEE International Conference on Embedded and Real-Time Computing Systems and Applications. and deterministic power management. So Same thing we can implement using low cost MRF24J40 transceiver module which supports 2. creating a self-configuring. 4 1 A S E I Discussion: This is first kind of implementation on WirelessHART with this specialized hardware. these products provide carrier class data reliability. [5] Xiuming Zhu. low latency. Song Han. And worker will carry handheld device which comprised of XD2510H transceiver module and PIC controller will be interfaced with each other. References: [1] HCF . MQ-135 sensor used will be a stationary device. Jianping Song. WirelessHART: Applying Wireless Technology in RealTime Industrial Process Control.

in II. This paper presents a new control scheme for the hysteresis compensation and precise positioning of a piezo electrically actuated manipulator. Maxwell model [3].x T . Nevertheless PZT introduces some non linearity due to the hysteresis property occurring at voltagedriven strategy. the hysteresis is compensated by feedforward control resorting to an inverse of the hysteresis Uma Datta E&I Group CSIR-CMERI Durgapur.arpita@gmail. high stiffness. INTRODUCTION Nano technology is the science of understanding matter and control of matters at dimension of 100 nm or less. Besides. Section II gives a theoretical background of the Dahl model to appropriate the hysteresis effect. and integrated circuit chip assembly line. & Science University. Prandtl–Ishlinskii model [5]. An important aspect of research in nanotechnology involves precision control and manipulation of devices and materials at a nano scale.com Saikat Kr Shome E&I Group CSIR-CMERI. The rest of the paper is organized as follows. Based on the Inverse Dahl model and Internal model control is demonstrated in section IV.res.Shaping the system input can also serve as a promising tool against parametric plant uncertainties [13]. In the present research. and x represent the equivalent mass. Section VI concludes the paper. India saikatkshome@cmeri. Duhem mode [4]. damping coefficient.e nanopositioning. hysteresis. It is assumed that the accurate system dynamics are reflected. system non-linearity is developed through a second order Dahl model which includes a lesser number of parameters. They are Preisach model [1. The implementation of this control strategy involving the second order Dahl hysteresis model has not been reported to best of our knowledge. K. nanopositioning.Hysteresis Compensation using Modified Internal Model Control for Precise Nano Positioning Anirban Bhakta School of Mechatronics Bengal Engg.stcet@gmail.11]. i. The hardware model of the Feedforward control using TMS320C6713 for piezo actuators in the context of nano positioning is done in [14]. The performance of sliding mode control using extended Kalman filter is studied in [12]. The observations from Matlab has been studied in section V. model control(imc). Durgapur. uncertainties. In 191 of 446 Copyright @ Department of Electrical Engineering. Dahl model [6] and Bouc– Wen model [7]. D. The efficiency of dither as an effective strategy in the paradigm of nanopositioning has been reported in [8. Micro and nano manipulators are fundamental components for a variety of devices applied in micro and nanotechnologies.x K .u Fh (1) where the parameters M. NIT Durgapur . SYSTEM MODELLING AND IDENTIFICATION A. Results envisaged that this control strategy improves the positioning accuracy of the system. which attenuates the accuracy of the manipulator. Therefore an inverse Dahl model is developed for the Internal model control technique. stiffness.2]. and x-axis displacement of the XY micromanipulator. There are a number of hysteresis models available in the literatures.in Abstract—One of the crucial aspects of research in nano technology is precision control and manipulation of devices at the nano scale i.res. such as scanning probe microscope (SPM). biological cell operator. Dynamic modeling of the system Considering the mechanical part as linear and of second order. Brief description of the conventional close loop control is given in Section III. Arpita Mukherjee E&I Group CSIR-CMERI Durgapur.com Keywords—Micro-nano manipulator. fast response and employ a flexure-based mechanism of piezoelectric actuators (PZTs) to deliver a submicron or sub nanometer resolution positioning. Shibpur ani. This scheme follows an adapted inverse Dahl model based Internal Model Control algorithm. large blocking force. The objective of the research is to find a robust control law which maintains system response and error signal within prescribed tolerance despite the effect of uncertainty on the system along with the primary advantage of using Internal model control which is to get an efficient set point tracking. the plant parameter uncertainty and the robustness against the process and measurement noise are also dealt with in the paper.. dynamic model of whole system with nonlinear hysteresis can be established as follows: M . India ee.x D.Typically. Nano positioners are precision mechatronic systems designed to move objects over a small range with a resolution down to a fraction of an atomic diameter. model[6]. the finer is the positioning performance of the piezo driven micro manipulator. Moreover this proposed strategy is able to handle the possible model uncertainty errors and errors due to measurement noise.e. and the nonlinearity arising from the PZT actuator. One of the popular ways to overcome the nonlinearity is to model the hysteresis first and then develop a feedforward compensation scheme. 4 1 A S E I Therefore hysteresis must be eliminated in order to achieve high precision applications. They are capable of producing nanometer position resolution. respectively. India umadutta@cmeri. The dynamic model of the actuator is established with the addition of Dahl hysteresis model. internal I.

Once the trajectory is assigned.u h M M Fig 1Entire system overview (2) Where. .8485 x 10 - b0 0 - c/m III. the left hand side equation can be obtained as a linear second-order system in the form x 2. Traditionally.b1. b0 . The envisaged Feed forward control has been implemented as Fig. The objective of feedforward control is to force the output platform of the micromanipulator to track a given position trajectory. Gdc denotes the D.F sgn(dx).1773 x 10 6 - b1 6 1. the determination of the input voltage which is applied to the PZT is the target of the controller design. u denotes the input voltage. CONVENTIONAL CONTROLLER A. which provides the hysteresis force Fh based on the output velocities x * is shown in Fig.9874 - a2 1. dividing by M. T is the piezo-electric coefficient.5973 x10 4 N. n the natural frequency of oscillation and ξ is the damping ratio of the second order system.a1. 4 1 Table:1 The Dahl model parameters adopted for the micro nano manipulator K Values Unit 0.a1 1 sgn(dx) b0 system Parameters q ( x) q 1 q2 ( x) D (4) Fh C . C b1 sgn(dx). and Fh indicates the hysteretic effect of the system in terms of force.q (5) S E I Where the hysteresis model parameters can be derived based on the following formulas a1 2 peakn 1 ln( ) peakn T 2 a1 a2 T 2 b1 Gdc . peakn represents the nth overshoot with respect to the steady state values.a2 2 A M dx dq dx Aq Bu dt dt dt 2 b0 s0 T denotes the period of the damped oscillation. Hysteresis Modeling: It has been shown that the hysteretic force term Fh can be predicted more accurately by the second-order than the firstorder Dahl model for the bearing friction in disk drive actuators [15]. 2. Feedforward Control: The input to the plant can be generated from the required trajectory obtained through the inverse transfer function of the plant model.addition.6065 x 10 0.x n2 T F . a2 . Specifically.1 which provides the output displacement ( x * ) for a given reference trajectory.1828 Kg 3 2. NIT Durgapur .0468 a1 121.s/m N/m T 2. The second-order Dahl model with state space representation d 2F dF du sgn(dx).C gain and S0 denotes the initial slope respectively.n . B.u (3) dt 2 dt dt Fig 2 Dahl hysteresis model The second order Dahl model. 3. C. Fig 3 Feedforward controller 192 of 446 Copyright @ Department of Electrical Engineering. B . Model Implementation Based on the dynamic plant parameters and the Dahl model specifications the plant can be designed as a feedback system. The entire system model is shown in Fig. the system model of can be identified by finding the model parameters through the experiments. The time domain state representation of the equation is 0 A a2 0 1 .

an additional feedback control is adopted to compensate for the model imperfection and other disturbances of the system. Y(s) is the output and N(s) is the measurement noise. if Gm ( s ) G ( s ) and D(s) and N(s) are not zero. the feedback signal B(s) is B( s) G(s)U (s) D(s) N (s) Gm ( s)U ( s) t U FB (t ) K p [e(t ) 1 de(t ) e( )d Td ] Ti 0 dt (7) 4 1 B( s) (G(s) Gm (s))U (s) D( s) N ( s) (8) Now. Conventional Internal Model Control voltage can be calculated by the inverse Dahl model developed from the system model 1 (6) U FF Mxd Dxd Kxd Fh T Where.e Gm ( s ) G ( s ) and the disturbance D(s) and the noise N(s) are both zero.(5) and xd is the desired trajectory. This is the condition when there is no uncertainty. U(s) is the control. Fig 7 Block diagram of an IMC system Consider the system shown in Fig 7 G(s) is the actual plant model and Gm (s ) is the nominal plant model. The concept is illustrated in Fig. Feedforward and Feedback controller Feedforward control is an open loop system. if the nominal and actual plant model is exact i. we are having Y (s) G( s)U (s) D(s) B(s) Y (s) N (s) Gm (s)U (s) U (s) C (s)(R(s) B(s)) Now eliminating U(s) and B(s) from the above equations Y (s) Fig 5 PID controller (1 C ( s )Gm ( s )) D( s ) G ( s )C ( s ) R( s ) 1 C ( s )(G ( s ) Gm ( s )) 1 C ( s )(G ( s ) Gm ( s )) G ( s )C ( s ) N ( s ) (9) 1 C ( s )( G ( s ) Gm ( s )) The sensitivity function S(s) that relates Y(s) and D(s) when R(s)=N(s)=0 is Y 1 C ( s )G ( s ) (s) S (s) D 1 C ( s )(G ( s ) Gm ( s )) And the complementary sensitivity function. 7. 5 and Inverse Dahl model is developed in Fig. From the fig 7. NIT Durgapur (10) . D(s) is the disturbance input. Fig 6 Inverse Dahl model 193 of 446 Copyright @ Department of Electrical Engineering. then B(s) A Fig 4 Feedforward+ feedback control S E I expresses the uncertainty of the process. 6. C(s) is called the IMC controller and is to be designed so that y(t) is kept as close as possible to r(t) at all the times. then B(s) is also zero and the control system is effectively open loop. B. 7. The PID controller is demonstrated in the Fig. Fh is the hysteresis force from Eqn.For a given position trajectory xd the expected input IV. R(s) is the desired value. INTERNAL MODEL CONTROLLER A. The feedback control input can be written as in Eqn. where a PID feedback controller is employed due to its robustness and ease of implementation properties. However. U FF is the output voltage from controller. Therefore. From the Fig. So from the controllability point of view it is not desired for a system.4.

The schematic is shown is the Fig. 13 C ( s) Now C (s) C ( s) / 2 1 C ( s)Gm ( s) substituting C(s) by (13) C (s) in the Y (s) ( R(s) D(s) N (s)) / 2 Eqn 9 (14) As. N(s)=D(s) 0 Y (s) R(s) N (s) from Eqn. the equation becomes Y (s) T (s) R(s) S (s) D(s) (12) If T(s)=1 there is perfect set point tracking. Fig 11 Feedforward+ feedback controller In this research. Modified IMC gives the best performance for set point tracking as reflected in Fig. there will be a perfect setpoint tracking as the external noise is sufficiently removed. Modified Internal Model Structure A. if S(s)=0 there is perfect disturbance rejection. Fig 9 Rectified sinusoidal signal B. This modification has been done to remove the noise more accurately. From the perfect set point tracking point of view. the PID controller parameters have been taken as K p 0. this model is found to be more efficient in the presence of process and measurement noise. Ti 0. 9. Td 0. Indirectly which denotes the decrease in non-linearity.009 s.T (s) 1 S (s) C ( s )G ( s ) 1 C ( s )( G ( s ) Gm ( s )) (11) Thus when N(s)=0. This will occur if Gm(s)=G(s) and C(s)=1/G(s). D( s) N ( s) . 4 1 Fig10 Feedforward controller S E I A V. Comparison of the hysteresis loop In this approach the IMC controller C(s) has been replaced by a negative feedback closed loop combination of C(s) and Gm(s). 8. it is observed that the area under the hysteresis loop is decreasing. The IMC controller As we go through the hysteresis plot in the Fig. So if the input trajectory is chosen double of the required reference. Y (s) R(s) / 2 (15) It is observed from the Eqn. shown in Fig. RESULT AND DISCUSSIONS The effectiveness of the proposed control strategies are verified using a rectified sinusoidal signal of amplitude 100um and frequency 1Hz. We have considered Gm(s)=G(s) and C(s)=1/Gm(s). Further we verify the efficiency of the proposed control technique by varying the plant parameters by 2%. Again this will occur if Gm(s)=G(s) and C(s)=1/Gm(s).044 s Fig 12 Internal model controller Fig 13 Modified IMC Fig 8 Modified IMC structure 194 of 446 Copyright @ Department of Electrical Engineering. NIT Durgapur . 9. 10-13. The peak to peak tracking errors and R.M.S tracking errors are tabulated for the comparative study. 15 that the output trajectory has become half of the input.12 .

a Set point tracking of modified IMC Fig 17. 16.S error 6x10−7 Feedforward+ Feedback control 15x10−7 5.M.a Fig 16.a-b respectively. The dynamic plant parameters like equivalent mass (M).A.a Set point tracking of Feedforward control Fig 14.b The tracking error of Fig 16.M. 195 of 446 Copyright @ Department of Electrical Engineering.9 x108 as compared to the other control approaches. 17. IMC and modified IMC in the Fig 14. NIT Durgapur . Feedforward+ feedback controller.S error (m) Control strategies Peak to peak error 20x10−7 Feedforward control R. Fig 14.9x10−8 B.a-b.a-b. Here the blue line showing the given trajectory and the green lines are showing the output trajectory obtained from the plant.8x10−7 10x10−8 1. 15. The difference between actual path and the obtained path and the tracking errors are highlighted for the Feedforward controller.b The tracking error of Fig 17.a Fig 17.a Set point tracking of Feedforward+ Feedback control S E I Fig 15. damping coefficient (D) and stiffness (K) are varied bidirectional.a Set point tracking of IMC Fig 16.85x10−7 Conventional Internal model control Modified internal model control 12x10−7 5.a-b. Performance in the presence of parameter uncertainty The efficiency of the proposed strategy has also been examined under the plant parameter variation. The corresponding error values are tabulated in Table 2. The outcomes in terms of error from the actual trajectory are recorded in the Table 3.b The tracking error of Fig 15.a 4 1 A Fig 15. Comparison of Set point tracking Here we have studied the deviations from the actual trajectory assigned to the plant.b The tracking error of Fig 14.a Table 2: Peak to peak error and R. It is observed that Modified IMC provides a comparatively reduced error of 1.

18. no.. [17] Q.Y.” Mechatronics.pp. Eng... I. S Pradhan... U Datta. W. S K Shome.. 2007 “Robust Multiple Frequency Trajectory Tracking Control of Piezoelectrically driven Micro-nano [7] Ha. July 2010. VOL. Lim. pp 4234-4239. (Accepted).”Real time compensation of hysteresis in a Piezoelectric-stack actuator tracking a Stochastic Reference. the basic control schemes such as conventional Feedforward control representing the open loop system and PID control as a closed looped version have been compared with the robust Internal Model Control approach. Robotics and Embedded System (IEEE CARE).Xu.Li. N.. 2003. Control syst. actuators. CSIR-CMERI for carrying out this research. pp. Michellod. control syst.” IEEE Trans. [18] SantoshDevasia.. REFERENCES [1] Ge. [20] Basem M. and Jouaneh. W. 2013. [5] Bashash.. Y. H. NIT Durgapur . measurement and control. The results demonstrate the effectiveness of the proposed controller under the presence of external disturbance and plant parameter uncertainty.. [9] S K Shome. Automation and Mechatronics. 5. 669-684. “Adaptive Sliding Mode Control With Perturbation Estimation and PID Sliding Surface for Motion Tracking of a PiezoDriven Micromanipulator” IEEE Transactions on Control Systems Technology. 196 of 446 Copyright @ Department of Electrical Engineering. 3712–3716 [16] Cahyadi.. 2013. But in real life environment as the plant is more prone to noise injection.J..M. H. and Jalili. and Celanovic.M. M. Eng. ASME.2008. [8] S K Shome. A mukherjee..Y. S K Shome and S Pradhan. 2013. S K Shome.79x10−6 Conventional Internal model control 4. Evangelos Eleftheriou. Energy & Communication.S error under parameter variation ( 2%)(m) 1. VI. Y-S. Jywe. 15(5) pp. CONCLUSION AND FUTURE SCOPE In this experiment.1998 “Intelligent control of peizoelectric actuators” proceeding to the international conference on descision and control.”Discrete Modeling and Sliding Mode Control of Piezoelectric Actuators”. Due to a simple model.-Y.G. [11] S Pradhan. [22] Choi. "Performance Evaluation of Dither Distributions of Piezo Electric Actuators for Nano-Positioning" in proceedings of IEEE International Conference on Control. 2006 “A comparison of fitnes function for the identification of a piezoelectric hysteresis actuators based on the real-coded generic algorithm”.L. vol. Proceedings of the 2010 IEEE International Conference on Automation and Logistics August 16-20 2010. 1995 “Modeling Hysteresis in Piezoceramic Actuators. we observe that modified IMC performs more efficiently under the presence of process and measurement noise whereas conventional IMC design gives a satisfactory performance under plant parameter uncertainty. G. and Yamamoto Y. S. [6] [13] S K Shome. The authors acknowledge the continuous support of the Director.-L..2006 “modeling a micro manipulator system with flexure hinge” proceeding to the IEEE International Conference on Robotics. Computing and Control. India. A Mukherjee. M Prakash and M K Patel. Ali “Nanopositioning Fuzzy Control for Piezoelectric Actuators”.. CIEC 14.positioning System.17(3). Automation. N. 2013. the proposed methodology is attractive from practical implementation point of view. 4.-C.technol. Fung.. and Shu. and Gillet. J. D. Mullhaupt. and Messner. [10] M Prakash..Kolkata (accepted) [15] Helmick.”Proceeding of the American Control Conference. pp 1-5. 2002. U Datta. Robotics and Embedded System (IEEE CARE).. S. U Datta. [12] QingsongXu and Yangmin Li. A.R-F.. 867-878. Badr and Wahied. [21] Shen.pp 69-79. A Mukherjee. “Input Shapers Against System Parametric Uncertainty” proceedings of IEEE International Conference on Machine Intelligence Research and Advancement. Austria.. the proposed modified form of IMC is more suitable for industry purpose. [14] A Bhakta. September 2007... and Choi. S Sen. Reza Moheimani “A Survey of Control Issues in Nanopositioning” IEEE Transactions on Control Systems Technology. control structure and robustness to noise interference. IEEE Computer Society. pp. A Mukherjee.pp 71-78. “Dither Based Precise Position Control of Piezo Actuated Micro-Nano Manipulator” proceedings of 39th Annual Conference of IEEE Industrial Electronics Society. Technol. Table 3: R.821x10−6 Feedforward+ Feedback control 1. Chiang. NO. Automation. and Hsien. M Prakash.”Sliding Mode Control of a PiezoDriven Micropositioning System Using Extended Kalman Filter”. Y and su. 17(3). [4] Stepanenko. This work can further be extended to design two degree of freedom IMC along with the proposed modification in the IMC controller.” Proceedings of the 42th IEEE Conference on Decision and Control.IEEE trans. 2010. C-Y. A.” Precis. 2008. Y.M.. [3] Goldfarb. India. 2013. D.Xu.211-221 [2] Weibel.pp.Li. P.Kung. India. Hong Kong and Macau 4 1 A S E I ACKNOWLEDGMENT Q. “Dahl Model-Based Hysteresis Compensation and Precise Positioning Control of an XY Parallel Micromanipulator With Piezoelectric Actuation. [19] Yu Cao.17x10−7 From the above studies.. “Dither Control For Dahl Model Based Hysteresis Compensation” IEEE International Conference on Signal Processing. “Higher Order Modeling of Hysteresis in Disk Drive Actuators..” journal of dynamic system..S-C.sens.. “Dahl model based Feedforward control for Precise Positioning of Nano Actuators using TMS320C6713” proceeding to IEEE International Conference on Control. Y. F.1997 “Modeling piezoelectric stack actuators for control of micromanipulation”. G.12(5).”Precision Tracking Control of Piezoelcetric-Actuated System.32(2). 15.-K.” Precis.056x10−7 Modified internal model control 7. “Tracking Position Control of Piezoelectric Actuators for Periodic Reference Inputs.S error under parameter variation Control strategies Feedforward control R. "A Comparison of Dithers for Hysteresis Alleviation in Second Order Dahl Model Based Piezoelectric Actuator" in proceedings of IEEE International Conference on Control. P.. Instrumentation.

India. statistical invisibility. or distortion-free. Maity D.com Department of Information Technology. storage and impression of medical images in standard formats. reversible data hiding is desired to extract the embedded data as well as to recover the original host signal. Digital watermarking technique. India.mtech@gmail. Kolkata. The essential requirements of digital watermarking scheme are imperceptibility. MCKV Institute of Engineering. low embedding cost and high embedding rate. India. On the other hand. Howrah 711204. Chakraborty Department of Electronics & Communication Engineering. W. In other words. India. integer wavelet transform. Companding Rapid growth in digital techniques and wide use of Internet technology have put a serious threat to the owner of the digital medias to place the works in their web pages or in other public places. santipmaity@it. Shibpur. Liluah. The irreversibility is not admissible to some sensitive applications. P.. also often referred to as lossless. the original cover media usually cannot be restored completely. Bengal Institute of Technology. Medical images are normally collected in hospitals or research centers due to the nature of the specialized equipments. is being used widely as a solution to the kind of problems by imperceptible embedding of an auxiliary message in the multimedia signals. A reversible data hiding scheme based on integer transform is presented in [5]. Moreover.ac. drdpc@gmail. Data embedding directly in pixel values’ (spatial domain) of the cover image offers advantage of low computation cost but robustness performance particularly against lossy compression operations is poor. There are standard means like DICOM [7] (Digital Imaging and Communications in Medicine) that enable the management. The goal is achieved by embedding watermark based on companding in integer wavelet transform (IWT) domain. The superiority of the proposed scheme is verified by simulation results and compared with selected other methods.. W. Bengal Engineering and Science University. DICOM image. invertible. Shibpur. Howrah 711 103.in Ex Professor. Bengal Engineering and Science University. Moreover. The companding technique is introduced into the embedding process to increase data hiding capacity.com Department of Information Technology. W. such as legal and medical imaging.B. unlike previously proposed reversible watermarking techniques. The problem of 197 of 446 Copyright @ Department of Electrical Engineering. data hiding has been a very active research subject in the last a few years. a reversible watermarking technique is proposed that offers high embedding capacity and fidelity of watermarked image. Xuan et al. based on Quantization Index Modulation (QIM) is proposed for healthcare information management systems. higher resiliency is achieved in 4 1 A S E I I. For these applications. Thabit and Khoo[6] proposed a novel robust reversible watermarking scheme based on Slantlet transform that transforms small blocks of the original image and hiding the watermark bits by modifying the mean values of the carrier subbands. In this paper. which facilitates the selection of compression and expansion functions. poulami. This is because of the fact that anyone having access to those forums can copy that information in unlimited numbers and without any loss. Department of Electronics & Communication Engineering. Even though the distortion is often small and imperceptible to human visual systems (HVS). NIT Durgapur . reversible data hiding is very important and useful in telemedicine as it extracts the embedded data as well as recovered the original host signal. amitphadikar@rediffmail.B. hence. they are irreversible data hiding. simple linear function is used in companding to make the scheme easy to implement.com Abstract—In recent years.becs. In [3] a reversible watermarking scheme.B. Data embedding is carried out by expanding the differences between one pixel and each of its three neighboring pixels.Reversible Data Hiding using Wavelet Transform and Companding for DICOM Image Poulami Jana Amit Phadikar Santi P. But it is difficult to meet all these requirements simultaneously with: the highest degree of accuracy. INTRODUCTION transform domain watermarking scheme at the cost of higher computation and complexity of the algorithms. distort the cover media in order to insert the additional information. and robustness against common signal processing operations as well as deliberate attacks. over the last decade. this novel approach does not embed the location map in the host image that ultimately helps to achieve high fidelity of the watermarked image. The concept of reversible watermarking firstly appeared in the patent owned by Eastman Kodak [1].. [2] embedd data into the high frequency wavelet coefficients by companding technique. Most multimedia data hiding techniques modify and. The scheme takes advantage of the Laplacian-like distribution of integer wavelet coefficients. Keywords—reversible watermarking. Tian [4] expanded the value differences between two neighboring pixels to embed a bit without causing overflows or underflows. Reversible.

The symbol (5) represents bit by bit x-or operation. to get the final watermarked HH subband. Moreover. as this one offers the best compromise between robustness and invisibility. The superiority of the proposed scheme is verified by simulation results and compared with selected other methods. The function abs(. Step 7: Change of dynamic range: The coefficients of HH bands are divided by two. Moreover. This operation will prevent overflow/underflow after data embedding. Data Encoding Step 1: Image transformation: Original image (I) is decomposed into four subbands i. low-high (LH). 4] embed location map as side information for restoration of host image. Step 5: Generation of random binary matrix based on user defined key: A binary pseudorandom matrix (R) is generated based on the secret key (K).e. A S E I II. To solve the above problems. This is analogous to compression as the above division operation changes the dynamic range of the original signal to a narrower range. In this section. A. Then the sign matrix(S) is multiplied with HHw. the watermark can be detected efficiently from those subbands without waiting for the others. the choice of embedding the watermark information into the HH subbands was motivated by (6) Step 8: Watermark bit embedding: Then the watermark bits are embedded into the HH subband according to the following rule: (7) The symbol HHw represents the watermarked coefficients. due to the reduced number of inserted bits. Step 4: Generation of sign matrix: A sign matrix (S) is generated from HH subband according to the following rule. One may use higher level of decomposition for batter results. NIT Durgapur . low-low (LL). The objective of the encoding scheme is to hide a watermark into the cover image. 1 and 2. a novel reversible watermarking scheme combining an integer wavelet transform with companding technique is proposed. On the other hand. Step 3: Generation of odd-even matrix: The odd-even matrix (P) is generated from HH subband according to the following rule. in case of scalable decoding if only the high-energy subbands are sent to the decoder. A reversible watermarking scheme based on integer wavelet transform and companding is presented in Section II.) returns the absolute value of a signal. high-high (HH) using IWT. PROPOSED ALGORITHM 4 1 Step 6: Generation of watermark bit: A content dependent watermark (W) is generated based on P & R. The rest of the paper is organized as follows. high-low (HL). The block diagrams of the proposed watermark encoding and decoding process are described in Figs. Moreover. ) (1) (2) It is to be pointed out that the size of P and HH are same.overflow/underflow is avoided by histogram modification process. according to the following rule. as this component contains both the characteristics of horizontal and vertical edge components. Step 2: Coefficients selection criteria for watermark embedding: The scheme selects HH subbands. unlike previously proposed reversible watermarking techniques. which are indicators of the strength of the authentication and originality mechanisms of solution. the decoding scheme extracts the embedded watermark and also recovered the original host signal. The multi-resolution nature of wavelets reflects the anisotropic properties of human visual system (HVS) more precisely and also helps to design watermarking scheme that offers high fidelity of the watermarked image. That ultimately reduces the effective embedding capacity of a scheme. respectively. Most of the current solutions found in the literature compromise between the quality of the reproduced medical image and between the numbers of bits inserted. Lastly. 198 of 446 Copyright @ Department of Electrical Engineering. The symbol HHn represents the high-high subband after change of dynamic range. experimental tests. The experimental results are shown in Section III and finally we conclude the paper in Section IV. the schemes proposed in [2. Because small changes in wavelet coefficients distort the image very less. the strength of the current solutions is highly vulnerable and it can be broken with simple security algorithms or brute force attacks. (3) (4) It is to be pointed out that the size of S and HH are same. this novel approach does not embed the location map in the host image that ultimately helps to achieve high fidelity of the watermarked image. we present the encoding and the decoding schemes of the proposed method. Moreover.

yi)= HHw(xi.yi)+1 if P′(xi.88 54.Step 9: Inverse image transformation: Then inverse IWT is applied and watermarked image (Iw) is formed. it is cleared that the watermark is embedded in the host image without decreasing the fidelity of the image. low-high (LH). 3(a) shows one of the test images. Fig. respectively. TABLE 1. Figs. where as relative entropy distance (Kullback Leibler distance (KLD) [9] is used as measure of security ( ) of hidden data.38 74. Then inverse IWT is applied to get back the original image (I). The present study uses peak–signal-to-noise-ratio (PSNR) and meanstructure-similarity-index-measure (MSSIM) [8] as distortion measure for the watermarked image. All experimentations are conducted in Pentium IV. III. 3(b) shows the decomposed image using integer wavelet.000 40.000 62. MSSIM and values and the restored images. The high PSNR (dB) & MSSIM values of the watermarked images and low security values indicate better imperceptibility and security of the hidden data. B. KLD 4×10-7 27×10-7 69×10-7 83×10-7 199 of 446 Copyright @ Department of Electrical Engineering. 3(c)(d). MSSIM AND KLD VALUES FOR DIFFERENT PAYLOAD SIZE. 2. Data encoding.yi) otherwise PSNR (dB) 80. VARIATION OF PSNR. Payload (in bits) 2. Step 3: Generation of random binary matrix based on user defined key: A binary pseudorandom matrix (R) is generated based on the secret key (K) that was used at the time of watermarked embedding.yi)–1 if P′(xi. Step 2: Watermark bit extraction: The watermark bits (W′) are extracted by examining the least significant bit (LSB) of HH subbands.32 MSSIM 1 1 0. Data decoding. 4 1 S E I A Step 6: Restoration of HH subband of watermarked image: The watermarked HH subband is restored to its original form according to the following rule. Step 4: Extraction of odd-even matrix pattern: Then odd-even matrix pattern (P′) is generated based on the extracted watermark W′. 2. the image is restored completely by the proposed scheme.yi)=HHw(xi. Fig. PERFORMANCE EVALUATION Fig.yi)=odd & W′(xi.yi) =1 HHR(xi. Data Decoding Step 1: Image transformation: Decompose the watermarked image (Iw) into 4 subbands i. high-low (HL). Moreover.9980 The symbol HHR represents the restored HH subband.80 GHz processor.500 HHR(xi. with 512 MB RAM using MATLAB 7. NIT Durgapur . 1. (8) The performance of the proposed scheme is evaluated over various medical images having varied image characteristics.e. Step 5: Generation of sign matrix: A sign matrix (S) is generated from HH subband according to Eq.yi)=HHw(xi. (3) & (4).yi)=even & W′(xi. 3(c)-(d) show the watermarked images along with the PSNR. respectively. low-low (LL).9992 0. high-high (HH).yi) = 0 HHR(xi. The size of all test medical images is (512×512).500 10.36 62. From Figs. using IWT. Fig.

(c): Watermarked image. The capacity vs. Each point on the curves is obtained as the average value of 100 independent experiments. That ensures that the scheme is secured. As expected with the increase in payload size the value of KLD is also increased. 4. M. the value of MSSIM is till approaching to one (1). M: 1) Fig. M: 0. MSSIM and security values of the watermarked image. 3. 5. 4 shows that as expected the value of PSNR is decreased with the increase of payload size. Variation of MSSIM for different watermark size (bit). K: 83×10-7) (b) (P: ∞. NIT Durgapur . 6 shows the variation of KLD for different payload size. Fig. 3(a). till the value of KLD is very low and approaching to zero (0). However. MSSIM and KLD values.(a) (P: 54. 5 shows that the structure of two images (original image and watermarked image) are identical till the size of embedded watermark is 1024 bits. and Xuan et al. Figs. S E I Fig.32. Fig. (c) 4 1 (d) Fig. (a): Original image.4-6 show the average variation of PSNR. It is observed from the results of Fig. ((P. 200 of 446 Copyright @ Department of Electrical Engineering. That implies that the data embedding mechanism does not alter the structure of watermarked image so much. Fig. Variation of PSNR (dB) for different watermark size (bits). 7. Table 1 shows the variation of PSNR. 6. [2] is shown in Fig. It is also seen that with the increase of payload size the MSSIM value is decreased. Variation of KLD for different watermark size. distortion comparisons between our method. (b): Decomposed image using integer wavelet. MSSIM and KLD values for the image Fig. K) above each image A represents the PSNR (in dB). However. (d): Reconstructed image. Fig.998. 7 that the proposed method offers better gain in term of PSNR (dB) over other methods.

pp. Rabbani. Tian. ―A novel reversible watermarking based on an integer transform‖. J. a reversible watermarking technique is proposed based on integer wavelet transform and companding. 4. 2001. 2005. of International Workshop on Digital Watermarking. [6] R. 2004. 1997.‖ IEEE Transactions on Image Processing. ―Understanding and using DICOM. Stoffel.3 0. vol. the data interchange standard for biomedical imaging‖.‖ Proc.‖ IEEE Transaction on Circuits and Systemsfor Video Technology. Horii. [2] Proposed 75 70 PSNR (dB) 65 60 55 50 45 40 0. C.[7] W. 3. The proposed scheme is simple. ―Image quality assessment: from error measurement to structural similarity. 2007. [5] S.‖ US patent: 6278791. 88. H. S. 1–8. A S E I REFERENCES [1] C. Maity. and M. T. vol. 2012. and R. The experimental results show that the proposed scheme offers high embedding capacity with no compromise in watermarked image fidelity and security. 890–896. H. ―Reversible data hiding using integer wavelet transform and companding technique. Wang. 13. P. of IEEE International Conference on Image Processing.25 Payload Size (bpp) 0. Bovik. pp. W. Journal of the American Medical Informatics Association.‖ Computational and Mathematical Methods in Medicine. 4 1 IV. R Sheikh.35 Fig. distortion comparison. Bidgood. [2] G. Honsinger. Khoo. A. Zhen. Yang. Prior. 80 Xuan et al. pp. Y.1 0. Hsin. “Robust image watermarking using multiresolution analysis. 3304. pp. pp. vol.2 0. Shi. Sung. Capacity vs. vol. Chen. V. Ko. and is a possible solution for hospital data management. ―Reversible data embedding using a difference expansion. [3] L. ―Robust reversible watermarking scheme using slantlet transform matrix. [8] Z. S. 2004. Das. Y. 115-124. P. Z. ―Lossless recovery of an original image containing embedded data. Jones. W. K. P. P Simoncelli. Weng. and D. 7. pp. Kundu. of the IEEE INDICON.179. 2014. Xuan. 74-86. Shieh. vol. pp. in Proc. 201 of 446 Copyright @ Department of Electrical Engineering. Thabit.‖ Journal of Systems and Software. [4] J. Nandy. Y. cost effective and easy to implement. C. M. 174 . 2003. C. and T. E.15 0. Y. 13. Pan. ―Nested quantization index modulation for reversible watermarking and its application to healthcare information management systems. NIT Durgapur . J. Y. C. and J. and E. CONCLUSSION In this paper. R.‖ in Proc. B. 241 244. F. Ni. Q. Syckle. [9] S. Zhao. no. D. Lecture Notes in Computer Science. 1-14.

PV systems become much popular. The model considers irradiance and temperature as variable parameters and obtains the I/V characteristic for that particular cell under the above conditions. and have poor spectral performance in lower switching frequencywhich is sometimes necessary for high power and medium power applications to reduce the switching losses. The system. 2. low dv/dt and low EMI problems [3]. a feature that enables the inverter to continue operating at lower power levels after cell failure [7]. Durgapur -713209. is connected with a distribution grid of 400V (ac-line to line). one control technique is proposed to keep the dc link voltages constant for the isolated sources and non-isolated source by balancing the active power flow. the conventional three phase invertersproduce three output voltage levels with higher total harmonic distortion (THD). Durgapur -713209. Multilevel voltage source inverters (VSI) topologies have three major classes. the cascaded multilevel inverter has the potential to be the most reliable and achieve the best fault tolerance owing to its modularity. and [6] with isolated dc A S E I 4 1 The model of a photovoltaic cell that can be implemented in any simulation environment is proposed in [9]. [5]. multilevel converter. dc link voltage balance. and inverter output is connected to allow frequency three phase transformer open winding primary side. reduced current and voltage harmonics on the ac side. high voltage capability. INTRODUCTION In the present day the power demand is increasing repeatedly and it can be highly depended on the use of conventional or non-conventional energy source. a grid-connected photovoltaic system with CTLI is connected to the primary of an open-end. tapassahanit@gmail. Keywords—Photovoltaic system. however.com Department of Electrical Engineering. such as. 1. Photovoltaic system with isolated source cascaded two level 202 of 446 Copyright @ Department of Electrical Engineering. The regular development of photovoltaic system makes it cost effective and highly efficient. NIT Durgapur. which is then fed to the conventionalcascaded two-level inverter (CTLI). The framework of PV system with isolated source (CTLI) topology model is depicted in Fig. [11] and [12]. NIT Durgapur . Various control strategies are available in literature to control the output voltage of the power converters [10]. and switching frequency can be decreased hence reduction in switching losses. The power converter. many grid-connected photovoltaic systems use a three phase inverter to perform this connection [2]. In this work. Thus. However. for grid connected photovoltaic (PV) system. 1 and non-isolated source (CTLI) topology model is depicted in Fig. The CTIL is one of the widespread topology in multilevel inverter. The grid connected photovoltaic system use a photovoltaic cell to generate electricity. I. among the three topologies.isolated and non-isolated sourceinverter In this paper. Fig. NIT Durgapur. The photovoltaic (PV) system is verified through simulation using MATLAB-Simulink environment.com source and non-isolated dc source.com and deybiswasjayati@gmail. both topology isolated source and non-isolated source CTLI system are operated to supply active power with available solar irradiance and reactive power in absence of it. which make these kinds of power converters very interesting for the power industry. Generally. Photovoltaic systems are smart renewable energy sources tomeet the energy demand. three-phase isolated source and non-isolated source cascaded two level inverter (CTLI) topology. The PV systems are connected to the cascaded two-level inverter (CTLI). In recentyears. Power converters play a vital role in the overall efficiency of a Photovoltaic system [1]. In the present study. thereby spectral performance can be improved and switching frequency can be decreased with low THD in output voltage. with space vectorpulse width modulation (SVPWM) control topology. [8]. The PV source is considered to produce a dc voltage as low as 48V under rated operating conditions.The grid connected photovoltaic system is very popular today. Abstract—This paper presents a.Study on Photovoltaic System for Isolated and NonIsolated Source Cascaded Two Level Inverter (CTLI) Nayan Kumar Tapas Kumar Saha and Jayati Dey Department of Electrical Engineering.Using multilevel inverters the number of levels in the output voltage can be increased. India. There are other salient features. supplies both controlled active and reactive power for conventional CTLI topology. In this paper. show in [4]. single diode model of solar cell is used following the equations developed in [13]. India nayansays@gmail. low frequency three-phase transformer.

Photovoltaic system with non-isolated source cascaded two level Fig. dual two-level inverter topology gains popularity. For solar cells used in power industry.1. Non-isolated Source CTLI The two PV source are connected to single capacitor of inverter as shown in Fig. The equivalent circuit determined from the equation is used for simulation model.CTLI topology perform power conversion in multilevel voltage steps to obtain improved power quality. = photogene rated current (linear with irradiance) (A). shown in Fig. GRID CONNECTED CASCADED TWO LEVEL ISOLATED AND NON-ISOLATED INVERTER MODEL OF THE PHOTOVOLTAIC ARRAY In recent years. The control scheme successfully controls (i) reactive power (ii) dc link single capacitor voltage. On the other hand.For this topology. Fig. 3.38 x 10-23 I/K = electron charge. 4 1 A B. should be very large.2. [14]. the series resistance Rsshould be very small and the parallel resistance. and the number of sets Fig. PV cell connected inverter-1 and inverter-2 dc link voltage. The control scheme adopted successfully controls (i) reactive power (ii) average dc-link capacitor voltage. 1. Thehv side of the transformer is connected across the grid. A. in the order of battery voltage. The change of available solar power is done by changing the solar radiation level. The control design presented here is made simpler since it is only used to control the AC currents. and same. are Vdc1=Vdc2.connected in parallel to form one array has been considered to provide maximum power at 48V at the normal Indian solar irradiance. 1. Rp is much greater than the forward resistance of a diode.The number of cells connected in series to form one set. b and c windings are as follows: (2) (3) (4) Where are the pole voltages of first inverter and are the pole voltages of second inverter. Rp. Isolated source CTLI topology A three phase transformer of suitable rating is connected in between the two inverters. 2. In the isolated source CTLI. = saturation current due to diffusion mechanism (A) T = ambient temperature. S E C. respectively. 3. = Boltzmann constant.Cells connected in parallel increase the current and cells connected in series increase the output voltages. 4. IsolatedSource CTLI The two PV source are connected to each inverter capacitors as shown in Fig. lower switching losses. dc-link capacitor voltages of both the inverters are low. better electromagnetic compatibility and higher voltage capability.6. non-isolated source CTLI topology dc-link voltages of two converters are low. the voltage across a. II. considering these advantages. NIT Durgapur . 4 and Fig. Kelvin. This work 203 of 446 Copyright @ Department of Electrical Engineering. Development of Photovoltaic Array Photovoltaic (PV) array is a device that converts available solar irradiance into electricity by the photovoltaic effect. III. Photovoltic cellequivalent circuit The topology consists of two inverters and a three phase transformer. The parameter values of the model have been selected following the data given in [13]. A mathematical description of the current/voltage terminal characteristic for PV cells has provided by the equation as an exponential equation (1) I (1) Where.6 x l0-19 C To design high quality solar cell. The both (isolated and non-isolated source) topology is constructed by using conventional two level inverters as shown in the Fig. In the isolated source CTLI topology. being the references usually given in a d-q reference frame.

The d-q axis voltages in synchronous rotating reference frame are expressed as follows: (10) (11) Where reference is average dc link capacitor voltage of inverter.5 shows the transformer RL equivalent circuit. The reference current now can be written as: For isolated source CTLI Fig. The first and second converters are receiving phase shifted modulating signals. L is leakage inductance of transformer primary and secondary windings. NIT Durgapur . In this paper. S E I (9) Where. The control scheme is shown in Fig. In this paper. . R is resistance of transformer primary and secondary windings.isolated source CTLI. and voltage controller is selected as . is considered as positive to provide active power from the converter to the grid. 6. respectively. The objective of the system controller is to achieve voltage regulation of the inverter dc-link capacitors.Due to this conversion. Thus the reduction in solar irradiance will result in reduction of in the controller.presents the modeling and design of the cascaded two-level inverter in grid connected photovoltaic system. Equivalentcircuitfor‘a’phase (12) For non-isolated source CTLI are considered to be equal to and the isequal to .The 204 of 446 Copyright @ Department of Electrical Engineering. The cascaded two level inverter (isolated and nonisolated source) topology based photovoltaic system are as shown in Fig. where is source voltage. The three phase transformer lv-side open end winding are connected to CTLI output.5. and is considered as negative to supply lagging reactive power from the converter to the grid. dc-link capacitor voltage is considered to be . The Fig. both active and reactive currents are decoupled and can be controlledindependently. The bandwidths for the current controllers are taken as ten times of voltage controller.2. Non-Isolated source CTLI topology The reference value of d-axis current is generated from the dc-link capacitor voltage controller. in absence of solar energy. The d-q to abc conversion has been achieved with the standard vector control practice. IV. the dc link capacitor voltages are maintained by keeping the zero orlittle bit negative to supply the losses. is the voltage across primary of transformer which is controlled by first and second inverter. respectively. (6) and (7) a state space model is developed for the above system as follows. and hv-side transformer directly connected to grid. The is the actual average dc link voltage of inverters for non-isolated source inverter. The controllers are tested to operate satisfactorily in the simulation environment. From equation (5). the reference dc voltage is kept at the required average level of . The gate signals are generated after comparing them with triangular carrier wave of the switching frequency. Here for phase ‘a’ is primary voltage. so that dcomponent of source voltage. DEVELOPMENT OF SYSTEM CONTROLLERS The photovoltaic system is connected to both the inverters with equal ratings and two equal groups of panels supplying them for isolated source and non. . is aligned with synchronous rotating reference frame.For three phase a. 5. A (8) Equation (8) is transformed intosynchronous rotating reference frame. b and c (5) (6) (7) Fig. is d-axis voltage component of ac source. The q-component of the source voltage. The current controllers for and are set up identically. qcomponent of the source voltage and are d–q axes components of cascaded inverter output voltage and current. On the other hand. and are the actual dc-link voltage of the first inverter and second inverter for isolated source inverter. The system variablesin d-q frame are expressed as follows: 4 1 (13) The dc link voltage controller ensures that the available power from solar photovoltaic system be supplied to the grid. for ‘a’ phase. 1 and Fig. In this study.

This ensures maximum utilization of the available solar power. 15. As a result. 4 1 Fig. shown in Fig. S E A Fig.The dc average voltage changes asshown in Fig. The variations of the currents are shown in Fig. The switching frequency of converter is taken as 1.10.9. Fig. In an isolated source dualinverter. 14 and Fig. The is designed to be of constant value since it is expressed from the constant grid voltage. the current increases to 5 Amp from 3.isolated source dual-inverter dc-link voltage undergoes a higher change in transient as compared to isolated source dual-inverter. Proposed inverter controller V.5 Amp from the initial value of 5 Amp.isolated source dual-inverter. 12. The change in the quadrature axis current during the entire operation is negligible.5 sec.10 and Fig.10. It is observed that non. 11.5 Amp for both the cases. The dc voltage changes for the non-isolated source dual-inverteris as shown in Fig. 9. I Fig. The controller is same as isolated source inverter as shown in Fig. 12. remains at so as to ensure no reactive power exchange with the PV system. The nature o and variations are. Direct axis and quadrature axis current of the non-isolated source cascaded inverter in response to the reduction of solar irradiance.7.5 sec.the values of for both the non-isolated source dual-inverterand isolated source dual-inverter are sameas shown in Fig. The variation in dc link average voltage of isolated source dual-inverter is only by 3V in transient which is shown in Fig. Average DC link voltage of the isolated source cascaded inverter in response to the reduction of solar irradiance. It settles down to 3.8 shows the output voltage of the isolated source dual-inverter. when the solar irradiation reduces. The is kept at 0 by the control action. Isolated source inverter output voltage The output voltage of the non-isolated source dual-inverter is shown in Fig. The active power supplied is .16. 13. as a result of which the current reduces as shown in Fig. given that . Non-isolated source inverter output voltage At time 0. 10. Fig. 205 of 446 Copyright @ Department of Electrical Engineering. Direct axis and quadrature axis current of the isolated source cascaded inverter in response to the reduction of solar irradiance.The non. The system parameters are taken as [17]. 30% reduction in active power supply reduces the dc. The . This is the fundamental nature of the output voltage. The Here the voltage is retrieved to its initialvalue within 0. 8. COMPARISON OF CONVERTER PERFORMANCE The entire grid connected photovoltaic system based onthe cascaded two level isolated and non isolated inverter has been simulated in MATLAB -SIMULINK environment. the solar irradiance decreases at time 0. Fig.link voltage by 7V. It is clearly seen that the 30% reduction inactive power supply reduces the dc voltage only by 3 V. Isolated source CTLI and Non-Isolated source CTLI The Fig. It is seen that the is reduced an amount of 30% in a step. following its reference value . 11. respectively. From this structure it is possible to verify the multilevel operation of the power converter with the proposed controller. respectively.secondary is star connected for the transformer. 7.2 kHz. The active power (P) is directly dependent on the magnitude of .05 sec only. A. The solar irradiance is now increased at 1 sec. NIT Durgapur .

Surendra Babu and B.The power converter system with the proposed controller has been designed and verified in MATLABSIMULINK environment. “A New Multilevel Conversion Structure for Grid-Connected PV Applications. D. A. Kumar.“A new approach to enhance power quality for medium voltage AC drives” IEEE Trans. Saha. Fig.performance characteristic in transient condition compared to that obtained in non-isolated topology.Hammond. Vannier. on Computational Intelligence and Communication Systems.May 2009. Vol. 5. 692 . Direct axis and quadrature axis current of the isolated source cascaded inverter in response to the increment of solar irradiance [3] [4] [5] Fig.. 4. E. These results show satisfactory performance both in steady state and transient conditions for both isolated source and non-isolated source inverters. Antunes. Ind."Cascaded two-level inverter-based multilevel static VAr compensator using 12-sided polygonal voltage space vector modulation" IET Power Electron. 2012. "Simplified Reactive Power Control for Single-Phase Grid-Connected Photovoltaic Inverters." IEEE Transactions on Industrial Electronics. March 1999 J.pp. F. pp 468. August 2013. and increase in the active power supply with change in the availability of solar power. “Cascaded Two Level Inverter Based Grid Connected Photovoltaic System: Modeling and Control”. V. Power Appl.F.Electron. Arzande. [6] I VI. November 2009." IEEE Transactions on Power. Vol.” IEEE 36th Power Electronics Specialists Conference. 13. J. 2005. pp. Manning. Marcelo Gradella Villalva.Prajapat. pp. A. 22. pp.. 1 . 1198 .1208 .2. Jan 2008." IEEE Transactions on industrial electronics. Vol. The proposed scheme successfully controls the dual level inverter. 4416-4426..-Electr. ”Analysis and proposal of capacitor voltage control for an asymmetrical cascaded inverter. Silva. and J. and C.. Ind. 425 . K. 15.” Elsevier Solar Energy.434 MARCH 2007. Abhishek Katariya. and J. No." Int.PESC '05. 24. C. Dey. Iss. 1. Rahim. G. Conf. No. pp. on Ind. A. Iss. 16. 61. and D. “Narrow pulse elimination PWM for multilevel digital audio power amplifiers using two cascadedH-bridges as a nine-level converter. 60. Aug.” IEEE Trans.697 7-9 Oct.473 [12] [13] [14] 206 of 446 Copyright @ Department of Electrical Engineering. Silva.” IEEE Trans. Ashok Kumar and Shuchi Shukla.Vol. “Individual Voltage Balancing Strategy for PWM Cascaded H-Bridge Converter-Based STATCOM.” IEEE Trans. and A. Yaow-Ming Chen and Yung-Ruei Chang. pp. 5. 55. “Development of a photovoltaic array model for use in power-electronics simulation studies.pp. 86. Victor M. L. 49. D. C. 2002. The controller maintains the dc-link capacitor voltage under both the isolated source and nonisolated source dual inverter topology. pp. Pimentel.G. and J. pp. Average DC link voltage of the non isolated source cascaded inverter in response to the increment of solar irradiance References [1] [2] Fig. no.for non-isolatedsource dual-inverterthe dc link voltage increases by 7V in transient as shown in Fig. Barrena. Vol. Pomilio. M. Average DC link voltage of the non-isolated source cascaded inverter in response to the reduction of solar irradiance. Pires.Yu-Hui Lin. Chia-Hsi Chang. IEEE ICIT. Ostojic. Fernão Pires. 56.IEEE. P.” lEE Proc.24th July 2012 Nasrudin Abd. Escalante. Casadei. NIT Durgapur .V. No. 2. N. The isolated CTLI has good 4 1 A S E Fig. No. Fernando A.This transient dc-link voltage variation in non-isolated source dual-inverter inverter is higher. Proc. Ind.Jan/Feb 1997. F. Fernandes.Electron. N. “Dual-inverter for grid-connected photovoltaic system: Modeling and sliding mode control. Vol. P. Electronics. "Simulation and Testing of Photovoltaic with Grid Connected System. Iss. 8. 809–815.W. 17. Rossi."Transistor-Clamped H-Bridge Based Cascaded Multilevel Inverter With New Method of Capacitor Voltage Balancing. K. and Ernesto Ruppert Filho "Comprehensive Approach to Modeling and Simulation of Photovoltaic Arrays. and C. 202 208 . 33. No. 809-815. Martins. 2014. T.5. 14. 1500–1509. 2106-2115. 8. Gow. J. 11. 21-29. J. Direct axis and quadrature axis current of the non isolated source cascaded inverter in response to the increment of solar irradiance. Jonas Rafael Gazoli. 2011. Hao. Vol. IEEE Transactions on Power Electronics. A. 146. May 2014 V. pp. 16-16 June 2005.N.. “Flying capacitor multilevel inverters and DTC motor drive applications. vol. Electronics. Vol.K. [7] [8] [9] [10] [11] CONCLUSION The grid connected photovoltaic system isproposed with space vector control scheme for dual-level inverter. vol. Grandi. S. Average DC link voltage of the isolated source cascaded inverter in response to the increment of solar irradiance On the otherhand. No. Vol.Mohamad Fathi Mohamad Elias and Wooi Ping Hew.

Durgapur. Kolkata. Application of capacitive sensor without microcontroller (6) cannot meet our requirement. Bankura. Subrata Chattopadhyay Dhananjoy De Electrical Engineering. I. products are weighed using electronic weighing scale that weighs a package fast and accurately. thermally unstable circuits and the gain programmable by software (through PWM's output of the microcontroller) (3) strain-gage based sensor (i. technology.mahato@ee. this means high accuracy and efficiency of weighing are also constantly high on the agenda. India sankar.Mass Measuring System Using Delay-and-Add Direct Sequence (DADS) Spread Spectrum Method Karunamoy Chatterjee Sankar Narayan Mahato Electronics and Communication Engineering. food. etc.g. In the mail sorting and grading machinery. When microcontroller was first interfaced with sensor (5) then 207 of 446 Copyright @ Department of Electrical Engineering. 4 1 In the practice. A load cell is an uncontrollable weighing device. That is. DADS.. load cells) particularly sensitive to vibration. India subrata0507@sify.com Electrical Engineering. National Institute of Technology. NIT Durgapur . a novel algorithm and architecture is proposed for the weighing system. ingredients for production. I The weighing of articles is an essential part of modern Life. based on the least-mean square (LMS) algorithm (2) is useful for long duration measurement . the micro-controllers using general algorithms cannot meet higher accuracy. However. chemistry. There is a constant need for knowing the exact weight of many items.com Abstract— In the area of mass production. the usual internal division is 1137500 accuracy at 10 pieces /sec. In the area of mass production. the products are weighed using load cell based weighing system.in. India karuna_ds@rediffmail. National Institute of Technical Teachers Training and Research. India dhananjoyde2@gmail. Continuation of this trend brings benefits for both the customer and the producer. The main constraint in achieving higher accuracy and in increasing throughput rate of the passing products is a superimposed noise on a useful signal from the weighing system. INTRODUCTION A S E Keywords—accuracy. manufacturing efficiency is increased and hence profitability while package quality and quantity are assured to the customer's satisfaction.nitdgp. the government bodies are trying internationally to maintain the same constant pace. It is high precision for useful signal recovery and relative simplicity of construction. But it is not suitable for long duration measurement and particularly for industry environment. Mean while electronic capacitive transducer (4) was designed. A moving average filter algorithm (1) is particularly efficient when processing for very short-duration. noise. The common sources are: (1) constant values (2) the noises: electromagnetic pick-up.. mass measurement. For that a strong filtering circuit should be implemented with high precision and fast response time.ac. In production. In this paper. the implementation of pipelined in low-power 6 taps adaptive filter. The type and the number of products that require weigh control are increasing. Bankura Unnayani Institute of Engineering. Bankura Unnayani Institute of Engineering. But the technique for measuring (3) low-frequency and low-level capacitance variations is opened the door of implementation of weighing system using capacitive sensor.com Electronics and Communication Engineering. Bankura. power harmonic. Improvement in filtering increases the speed of weighing and enhances the measurement accuracy.e. e. whereas moving average filter response time was less than LMS filter. The algorithm has been verified and implemented in the weight measurement system successfully. Although this filter has more response time. pharmacology. Consequently.

resulting in a signal sk= {sk. But PLC based intelligent Signal Processing method for high speed weighing system (8) can tolerate impact of environmental changes. DESIGN METHOD In fig. 3. In particular. although response time is improved. The two main aims for improvement are increasing the speed of weighing and achieving good measurement accuracy. Any sampling rate selected must therefore be greater than 80 Hz. The micro-controller (8031) transfers the detected digital signal to the communication unit and the display unit according to the procedure set on the memory. Improvement in SPM that provides any one or both aims brings significant benefit to the overall static weighing system. In this paper.at approximately 2. The two main aims for improvement are increasing the speed of weighing and achieving good measurement accuracy. and precision.. the noise is amplified and an error is introduced in the system. II. with L<<M. and therefore. S E I Owing to the physical characteristics. there is always a RC low pass filter in the output of the circuit. 2. stability. it produces a weak current. In fig. 1. the amplitude spectra reveal the significant noise components of the weighing system.+1} with data rate 1/M. Then make a comparative study with other filtering algorithm used for signal processing module. And finally the two signals are added together. In this paper our objective is to implement new algorithm for filtering method in signal processing module for weighing system with even capacitive sensor with least average variation and fast response time with higher accuracy and precision. in which a pseudo-random bit sequence (PRBS) generator generates a bipolar spreading code x={xi}.2 Hz. as shown in fig. 2. the proposed novel DADS filter algorithm used in weighting system can meet the requirements of speed. A/D converter and the two stage digital filter into digital signal. In the lower branch of the modulator. coming from the sensor . and the second and third peaks appear to be integer multiples of this noise. With this novel method. 1. when the load cell detects the external force. The common sources are amplified before transferred into the A/D converter. They can be implemented in real time or as post processed applications. The weak current is passed through AD7730 that has the PGA (Programmable Gain Amplifier).e. A signal processing module (SPM) acquires the electrical signal from the weighing device and estimates a value of weight for the passing product as its output. the schematic block diagram of DADS transmitter is shown. the focus is on using digital filtering techniques to remove measurement noises from the extremely low frequency noise of the static weighing system and make a comparative study with other filter. The first peak is centered 4 1 Fig. the same spreading code is delayed by L. Hence we can assume that there are no significant harmonic components beyond 40 Hz. Again its average variation is bounded for higher load. it is known that linear phase filters are the best with respect to transient response time. A Fig. a programmable filter can also be designed for weighing system. Digital filters are ideal for the treatment of low frequency noise. Noise spectra of Mass Indicator system. but its effect on low frequency noise is few. a new spread spectrum technique (i. From the filter theory. The communication unit consists of a transmitter and a receiver and it uses the Delay-and-Add Direct Sequence (DADS) spread spectrum modulation technique. 208 of 446 Copyright @ Department of Electrical Engineering. DADS) in the measuring system is proposed. Signal processing module (SPM) and Mass measurement. In the upper branch. This kind of filters can restrain high frequency noise. this spreading code is multiplied with the digital data dk ϵ{-1. Improvement in SPM that provides any one or both aims brings significant benefit to the overall static weighing system.i} with. In this paper we discusses the development of a load cell based the static weighing system. In order to reject the high-frequency random oscillatory component in the circuit. NIT Durgapur .

sk.2 = sk. separated by 2.+2}.i has only two possible values ϵ {4dk. a value that equals zero. The signal sk will be up-converted by the RF front-end and send out. First of all.1 = -sk. as there is no correlation between xi. 2 samples to the left and to the right. (3) Fig.i} for dk= +1 and a delay L=2. Fig.g. will always have opposite signs.i} has two important characteristics.11. separated by L. The block diagram of DADS demodulator is shown in fig. xi-L and xi-2L for any value of i. After the integrate and dump the signal sk will be. This is a direct consequence of the second characteristic of sk. e.g. 6. 4.i is a three level signal ϵ{+2. 209 of 446 Copyright @ Department of Electrical Engineering. Block diagram of DADS modulator. The base band signal sk={sk.0}. In this case. Block diagram of DADS demodulator. Fig. sk. 4 shows the first 15 samples of a modulated signal {sk... 5 shows the same for L = 2 and dk = -1 . 0.i. resulting in a signal {r’k. Every 2 non-zero samples sk. the received signal r k ={rk. will have the same sign if the data bit dk = +1 and will have opposite signs if dk = -1.4 or sk. 3. sk.9 = sk. every 2 non-zero samples. as described above. The second and the third term tend to zero. 6.i} will be delayed by the same delay L as in transmitter and multiplied with its non-delayed version.i}.(1) Fig. And also this property has eliminated the bit synchronization technique. Fig. The mathematical expression for r’k. Structure of modulated signal sk for dk=+1 with delay L. The second characteristic is of essential importance for the receiver. = + (2) It can be verified that r’k. Fig. This important property will be used for the demodulation.i will be given by. The first term in equation (3) is the useful signal. 5.3. Structure of modulated signal sk for dk = -1 with delay L= 2. NIT Durgapur . or a value with the same sign. A S E I 4 1 In the DADS demodulator. Every non-zero sample will have. e.

Wen Cai . NIT Durgapur . Meas.1 Pages616 . RESULT AND DISCUSSION In this paper. REFERENCES TABLE 1. E. we present the comparison between the proposed digital filters and the other filters which can be implemented in the mass measuring system.. “On some efficient algorithmsfor movingaverage filtering. Marioli. E: Sci. E. D. and resolution are satisfactory within the tolerable limit of industrial mass measurement.” IEEE Circuits and Systems. Rizkalla and P. L. without the need for code synchronization. without losing the benefits of a spread spectrum modulation. accuracy of the system will be increased and the system will become very fast. 21. It is found that using DADS filter. Dibo Hou . Huimei He. 1992. 2002. 15563758.” IEEE Circuits and Systems.4K 10 -2-+2 93. This results in a very simple receiver structure. Hence. [8] I 210 of 446 Copyright @ Department of Electrical Engineering. 2002.9K 7 -1-+1 172K 0. Beck. Dukel. Pallis-Areny.304 [4] [5] S E CONCLUSIONS In the mass measurement systems conventional filtering method employed have limitation in improving the accuracy and in throughput rate. M. LMS adaptive filter algorithm. V. Salama.1K 5 -3-+1. vol. G. Bridge Transducer ADC Data Sheet.287 101 TAPS FIR WITH HAMMING DADS FILTER DIVISION K=1000 -5-+5 37. 1771-1775.8-1 [2] [3] 64 TAPS MOVING AVERAGE FILTER LMS FILTER RESPONSE TIME (Seconds) AVERAGE VARIATION 135. Instrum.619 9-12 Aug. Gasulla and R. the average variation and response time are reduced drastically. Meas. Baxler..563-487. 3. Instum. Lyandres and S. Phys. linearity. It will enable measurement accuracy to be 1/172000. Guagxin Zhang. “Electronic transducers for industrial measurement of low value capacitance. an alternative technique 4 1 A [6] [7] IV.” IEEE Trans.” J. 2004. The result shows that new DADS filter can be employed in a practical high accuracy mass measuring system. K.” New York: IEEE Press. “Implementation of pipelined LMS adaptive filter for low-power VLSI applications. is embedded in the transmitted signal. Comparison of digital filters. Such as Moving average filter algorithm. Huang.2002 Vol: 2. 1997. S. Taroni. R. “An Intelligent Signal Processing method for high speed Weighing System”. the signal is again delayed by L and multiplied with the non-delayed signal for dispreading. S. 1991. Stott. From the experimental study. MWSCAS. [1] ZERO CALIBRATION FULL CALIBRATION UNFILTERED 148.. 4. Sardini and A. 1988. 242-280. “Capacitive Sensors Design and Application.” Instrum. 2012. “Measurement of small capacitance variations. has been explored to find a solution. B. L. At the receiver. a copy of the spreading code. In this case. Pages: 11-533-536 Aug. III. 24-Bit Sigma-Delta. Green and M. M. A.5 83.580-524. Reverter. delayed by L. 40 (2). Pingjie Huang. it is evident that the repeatability. F. Conventional FIR filter algorithm and Intelligent signal processing which can be used in the development of microcontroller based mass measurement system. Proceedings of the 35th Midwest Symposium on.” Analog Devices Tnc. In DADS. Briskin. 426-428. “A low-cost microcontroller interface for low-value capacitive sensors. Zhe Liu . “AD7730/AD7730L: CMOS. International Journal of Food Engineering. M.5K -1-+4 75.

This formula was taken from” Non conventional energy sources by GD ROY” λ (1) Where the inputs for turbine are shown below Pw = mechanical power extracted from the airflow [W]. renewable energy sources like wind energy generation got importance Mechanical power output of wind turbine was calculated based on the pitch angle.INDIA durga206@gmail. The simulation study is carried out for different types of faults on transmission line connected with wind farm. 211 of 446 Copyright @ Department of Electrical Engineering. more energy is required to satisfy the basic needs and to attain improved standards of human life.INDIA mahesh.Cp = performance coefficient or power coefficient. first stage consists of knowing the impedance values in healthy condition and in stage 2 the fault impedance values are compared with healthy values.INDIA vijayvihari5@gmail.yek@gmail. Section 1 in this paper shows different ways to integrate wind generator to the feeder and observe the short circuit current wave form for different types of integration. Section 4 represents the R-X graphs for different transmission line faults Section 5 represents the conclusions that are based on short circuit fault level of different types of wind generator connecting system and classification of faults. identify and classifying the faults was done. B.λ = tip speed ratio Vt/Vw. classification of faults was done based on the level of fault impedance values. Differential Equation Technique. wind velocity and λ (which described below) and given input to the induction generator to distribute electrical power through feeder In this paper section 1 represents the different ways to connect wind generation to the feeder.GITAM UNIVERSITY VISAKHAPATNAM. and Ar = area swept by the rotor [m2] Fig 1. section 2 represents the fault analysis of type 1 connected system and R-X diagrams for different 4 1 A S E Keywords: squirrel cage induction generator. The simulation was carried out in MATLAB/SIMULINK environment. NIT Durgapur .DURGA PRASAD ASSISTANT PROFESSOR:EEE Dept. INTRODUCTION power systems are complex systems that evolve over years in response to economic growth and continuously increasing power demand. By using DET.MAHESH BABU PG STUDENT(PS&A):EEE Dept. Double feed induction generator. which will be used in setting the fault current value in the relay for protecting the grid connected to the wind generator. With growing populations and the industrialization of the developing world. D= Air density [kg/m3].GITAM UNIVERSITY VISAKHAPATNAM.VIJAY VIHARI PG STUDENT(PS&A):EEE Dept.com Modeling of wind turbine The power output of a wind turbine is given by the formula (pw) as shown below.Fault Analysis of Wind Generator Connected Power System Using Differential Equation Technique Y. Section 3 shows the study of faults and impedance diagrams for different faults using Differential equation algorithm technique. Fault analysis in section 3 was done in two stages. The fault analysis is carried out on squirrel cage induction connected Type 1 system by using DET. Abstract--This paper presents an approach to analyze the fault severity in a different types of wind farm connected to the grid that occur on transmission line and identify the instant of occurrence. by using Differential equation technique (DET). In order to make energy economically available with reduced carbon emission.com B. I I.com symmetrical faults that occurred on transmission line using Differential equation algorithm technique.GITAM UNIVERSITY VISAKHAPATNAM. (the ratio between the blade tip speed VT and the wind speed upstream the rotor Vw [m/s]) Ѳ = blade pitch angle [deg]. turbine model.

6 0.7 0.5 Voltage Voltage 2000 0 -0.5 -1.7 0.2 0.Voltage form of type 1 system at healthy condition 0.The connection of wind turbine generator to the feeder 8000 4 1 6000 4000 0 A -2000 -4000 -6000 -8000 Fig.4 0.10.03 0.4 0.01 0.2 0.4 0.5 Time (sec) 0.2 0.2 0.9.5 Time 0.8 0.5 1 -1 voltage 0.9 1 Fig.8.2.5 Fig 4. Type2 wind turbine generator connected to grid: In this model wound rotor induction generator is used to add resistance to the rotor terminals as shown in below figure8 Fig.5 -1 4 0 0.02 0.5 -0. NIT Durgapur 0. x 10 1 Current 0.6 0.9 1 short circuit current -1.4 -2 0 0.3 0.1 0.2 0 x 10 1.3 0.6 0.7 0.5 Time (sec) 0.8 Current 4 2 1 0.06 0.4 0. 0 0.9 1 Fig.3 0.5 x 10 0 -0. Voltage wave form for L-G fault at phase ‘a’. Thee phase short circuit current for Type 2 system 212 of 446 Copyright @ Department of Electrical Engineering.9 0.7 0. -0. 2 x 10 0 -2 0 0.2 -0.1 Fig.3 0. Here Asynchronous machine is used as generator because in wind power system the torque of the turbine is not constant.04 -0.05 Time 0. Type 1-Squirrel cage induction generator: Fig2 shows the connection of wind turbine couple with induction generator and then connected to the feeder.5 Fig 5.08 0.3 0.5 I 0.11.9 1 . Type 2 WTG connected to grid.9 1 Fig.7 0. DIFFERENT WIND FARMS Different ways to connect wind turbine generator to the system is taken from reference[1] 3 2 current (Amps) 1 A.1 0. Voltage wave form at healthy condition.6 -0.1 0.4 0.6.II.8 Fig.5 Time 0.6 0.8 0.07 0.6 0.4 0.5 0 1 2 3 Time (sec) 0 4 5 6 4 x 10 Fig.09 0.1 0.5 -1 -1.2 0.6 S E 0. 1 4 1.5 Time 0.3 0.8 0. Current wave form with L-G fault at phase ‘a’: ig.8 0.1 0.5 Time (sec) 0. 6 short circuit current L-G fault is applied on the transmission line and the variation in the current and voltage wave form are shown 2 x 10 0 -2 -4 0 0.8 0 0.8 0.4 0.6 0.3.7 0.1 0. 0 -1 -2 -3 -4 -5 0 0.Current wave form for L-G fault at phase ‘a’: isc.Current wave form of type 1 system at healthy condition.2 0.7 Three phase schort circuit current B.5 4 1.

1 0.45 0.35 0.22.3 0.25 Time 0.16.9 Fig.Voltage wave form when L-G fault occurred at load end on phase a 5 4 3 2 1 Current 200 voltage 100 -100 0 -1 -200 0 0.4 0.5 0.02 0.C.Type3 wind turbine generator connected to the grid: The double feed induction generator connection to the feeder through ac\dc\ac converter as shown in fig.05 0.04 0.2 0.c.7 0.09 0.15.6 0.1 -2 -3 -4 -5 Fig.3 0.14. NIT Durgapur 1 .5 1 voltage 0.2 0.2 0. Current wave form at healthy condition.08 0. by using this type of system reduction in transient currents and transient voltages are reduced compared to squirrel cage induction generator connected Type1 system.08 0.35 0.25 Time 0.04 0.Type4 wind turbine generator connected to grid: In this model wind turbine generator is connected as shown below figure in which ac to dc and dc to ac converter is connected to get continuous output.01 0.5 Fig.45 0.5 Time (sec) 0.06 0.3 0.5 voltage 150 0 100 50 voltage -0.06 0.03 0. Type 4 WTG connected to feeder 4 1.07 0. 0.03 0.3 0.Thee phase short circuit current for Type 3 1 213 of 446 Copyright @ Department of Electrical Engineering.06 0.5 Time (sec) 0.1 Fig. By using this connection the fault current is reduced than previous models as shown in Fig18.03 0. 0 0. 4 2 0 x 10 1.04 0. Type 3 WTG connected to feeder Fig.8 0.6 0.3 0. voltage wave form at healthy condition 150 A 100 500 50 Current 400 0 300 200 -50 -150 0 0.5 Fig .19.4 0.5 1000 0 -1000 -2000 4 short circuit current 1 x 10 0 0.4 Fig .13.4 0.01 0.09 0.5 x 10 1 200 0.05 Time 0.15 0.04 0.5 -1 0 -50 -1.7 0.20.06 S E 0. Fig. short circuit current -0.2 0.12.1 0.05 Time 0.1 0.05 Time 0.18.07 0.Ccurrent wave form when L-G fault occurred at load end on phase a.25 Time 0.05 Fig. this was achieved because of ac\dc\ac converter connection as shown in fig 12.02 0.01 0.05 Time 0.05 0.9 Fig.02 0. Current wave form when L-G fault occurred phase a.5 -1 0.45 0.1 0.08 0.07 0.2 0.07 0.02 0.35 0. D.08 0.21.4 0.03 0.1 -100 4 1 -150 -200 0 0.1 Fig.15 0.15 0.5 0 -0. Thee phase short circuit current for Type 4 0.5 0 2000 -1 -1.17.1 0.5 .Voltage form at healthy condition 0.8 0 0.01 0. I 150 100 Current 50 0 -50 -100 -150 0 -100 -200 -300 -400 -500 0 0.09 0.09 0.5 -2 0 0. Voltage form when L-G fault occurred at phase a.

Let i0. when the voltage returns after a while. This paper mainly concentrates on faults like L-G. Faults on wind farm connected power system cause flexible drive trains to exhibit torsional oscillation and these oscillations causes power fluctuation. if wind farm integrated with synchronous generators system due to fault on wind turbine synchronous generators exhibit speed oscillations. of 446 Copyright @ Department of Electrical Engineering. External faults which occur on transmission line are mainly symmetrical faults and unsymmetrical faults. L-L-L faults. the following definitions are made Severity of fault on transmission line with different wind generator Type system is tabulated below CA=i0 + i1 DB=i 2 . The analysis of faults was done and differentiation was done based on the level of impedance deviation from the healthy Impedance values . so care should be taken on wind turbine system to detect fault quickly and to set fault current value to the relay. NIT Durgapur .On comparing the impedance values with previous cycle we can find out whether the fault is occurred or not. they are mainly classified as internal faults and external faults. i1. C. v2 be the voltage samples For calculating (R. FAULTS By using ODE formulas we can find out the impedance values at every cycles. Type 2 and Type 3 CB=i1+i2A=v0+v1 4 1 A S E B. In wind generation internal faults mainly occur at rotor terminals. Based on the magnitude of the impedance the type of fault is classified A.FAULT ANALYSIS: There are different types of faults which cur in the system.ORDINARY DIFFERENTIAL EQUATION TECHNIQUE : In this technique three voltage samples and current samples 214 are required with two equal intervals of time. based on this Algorithm the instant of fault Occurrence and type of fault was observed and shown in Results D. Result Analysis Category Transient period Fault current :isc Type1 Greater than 10 sec High Type2 Less than one cycle Type3 Less than one cycle Less than Type 1 Less than Type 1and Type 2 Type4 Greater than 10 sec Less than Type 1.Flow chart for differential equation technique Table 1.III. both of these are done in this paper to protect the grid. W=2*П*F.Effects of faults on wind turbine integrated grid: I In squirrel cage induction generator which by being connected to the grid during grid faults may require large currents to energize their induction generator. X). L-L-G. Fig. v1. i2 be the sample currents and v0. Y axis=X in ohms. These faults are classified and R-X diagrams are shown by using Differential equation technique X=W*DT*(VA*CB-VB*CA)/(2*(DA*CBDB*CA) Z=R+JX. Flow chart for differential equation Technique The effect of fault on wind turbine interconnected power system may cause to blackout due to tripping signal of the relays’.i1 If the penetration of energy is more than 20% of system energy then there is an increase in the voltage and frequency. IV. By comparing the three phase impedances in every cycle.23.Fluctuations in the system and this leads to power quality problems R= (DA*DB-DB*VA)/ (DA*CB-DB*CA) X=WL. The effect is minimum at low penetration levels but could be challenging at high penetration level [6]. ANALYSIS OF RESULTS In this section R-X diagrams for different faults applied Type1 induction turbine generator connected to the feeder are shown below Where X axis=R in ohms. gear box and winding. This can lead the possible violation of grid codes and security standards [5].

James Lyons.24. J.A.” Simulation of the impact of wind power on the transient fault behavior of the Nordic power system”. Roy “Non conventional energy sources”. Clemens Jauch. B. [1] Fig.R. N. signal processing and Renewable energy sources B. Poul Sørensen.R-X diagram for L-L-G fault at phase a.L Munda. Nicholas Miller. and Abraham Ellis. Ochieng1. “Short Circuit Current Contribution for Different Wind Turbine Generator Types”.W. Visakhapatnam. A. GITAM University. Jun Li. [8] [9] [10] GD. Nader Samaan. India. Fig. J. Eduard Muljadi. India. Shichikha” Mathematical Modeling of Wind Turbine in a Wind Energy Conversion System:” Jawahar lal Bhukya” Modeling and Analysis of Double fed Induction Generator for Variable Speed Wind Turbine”. Hansen.The Trajectory in Fig. He is currently doing his Post Graduate course in Power Systems and Automation at Electrical and Electronics Engineering department of GITAM Institute of Technology. F.Final report” .” Integrating Large Wind Farms into Weak Power Grids with Long Transmission Lines”.25. Helen Markou. Visakhapatnam. C. VII.24 is outside the circle which we can say it is under healthy condition. VI.27.” Challenges of Grid Integration of Wind Power on Power System Grid Integrity: A Review”. M. Visakhapatnam. Results show the effectiveness of the Algorithm for different fault conditions and that faults can be identified and classified easily by using Differential equation technique. V. GITAM University. His research interest includes Power System Protection. if the trajectory in the R-x diagram is inside the circle then the system is said to under faulty condition as shown in Fig. Onyango. India. Xiaoming Yuan. 4 1 A Fig. “Effect of Variable Speed Wind Turbine Generator on Stability of a Weak Grid”. 215 of 446 Copyright @ Department of Electrical Engineering. India. In section 3 fault analyses is carried out on Type 1 system. Florin Iov” Grid fault and design-basis for wind turbines . R-X diagram for L-L-L-G fault at phases a. NIT Durgapur . R. b.Manyonge. M. CONCLUSION Table 1 shows that fault severity and transient period is more in Type 1 system. Cutululis.R-X diagram for L-G fault at phase b and ground [7] S E I Fig.Durga Prasad received his Master’s degree in 2008 in Power Systems and Automation from AU Visakhapatnam. [2] [3] [4] [5] [6] REFERENCES Eduard Muljadi. phase b and ground. He is currently doing his Post Graduate course in Power Systems and Automation at Electrical and Electronics Engineering department of GITAM Institute of Technology. T. Jimoh. Anca D. The simulations were performed for different kinds of faults on transmission line.Mahesh babu was born in the year 1991 and received his Bachelor’s degree in 2012 from Bapatla engineering college. GITAM University. Badri ram “power system protection and switch gear”.26.D and currently working as an Assistant Professor in Electrical and Electronics Engineering Department at GITAM Institute of Technology. Richard Piwko. Renchang Dai. Subbaiah Pasupulati. Ayodele. P.27. Visakhapatnam. Nicolaos A. Ian Norheim. A. Carsten Rasmussen. Juan SanchezGasca. Brian Parsons. c and ground. J. India.25.X diagram at healthy condition. Poul Sørensen. Vahan Gevorgian. Butterfield.vijaya vihari was born in the year 1991 and received his Bachelor’s degree in 2012 from AU.R. where as the Type 4 system is less effected from faults and having less transient period.T Agee. India.26. BIOGRAPHIES Y. He pursuing Ph.

The laboratory KARD system for stress test A. If needed ECG can be connected for heart rate measuring. 6]. The embedded measuring and managing systems represent an important integral part for development of such systems. B) The TELEKARD is a telemetric portable exercise testing system that allow to monitor cardio respiratory function in laboratory as well as in field (sport medicine. People who suffer from specific diseases. and the peripheral circulation [1. Keywords—Telekard System. 3]. the heart. The equipment is composed of a unit (carried by object) which transmits the measured data to a receiver in area time. which are placed directly on their bodies sensing the necessary patient’s data. JNEC.This methodology was a basis for successive applied research projects that were developed in cooperation with hospitals. KULKARNI VISHWASHRI Department of Electronics. It is also used in medical rehabilitations. the pulmonary circulation. Aurangabad. India Email id:Chandrakironji@gmail. Zigbee technology II. the block diagram of the TELEKARD system is drawn [1]. The applicability of this idea has a great potential. India Email id:vishwashri@rediffmail.An appreciation to the normal response profiles of the gas transport systems that support cell respiration is essential to recognize the abnormal response patterns that characterize the many disease states that affect them. notebook). This system is based on AVR Atmel16 microcontroller. 2] was designed to use different types of old gas analyzers (Spirolyt or Infralyt). The heart rate meter SPORTTESTER is used for heart rate measuring and information is transmitted.BIOMEDICAL APPLICATION USING ZIGBEE Ms. The gas-meter with digital serial output was used for ventilation measuring.com Department of Electronics. In application of measuring and control systems in medical systems in medical diagnostics we always face the problems of signal processing.2. MGM’s. Simultaneously. 4 1 A S E I I. This paper considers noninvasive sensors for monitoring of some physiologic parameters. pulmonary ventilation and breathing frequency are non-invasively measured in patients breadth. KARD system The KARD system is used for laboratory testing which is based on microcontroller. the CO2 produced by muscles must be removed to avoid severe tissue acidosis with its adverse effects on cellular function. This involves the lungs. there are increasing number of people whose health requires frequent health check up or examination. The increased metabolic rate during exercise requires an appropriate increase in O2 the muscles. the subsequences of which can occur unexpectedly and result in critical health situation.1.com ABSTRACT In society. Their ability to respond adequately to this stress is a measure of their physical competence (or “health”). rehabilitation etc). HARDWARE There are two systems developed for data measuring: A) The KARD is a system for exercise testing which is used in laboratory. standard parameters are calculated by means of computer [4. From these data. In addition this paper reflects exercise testing which is used for sportspersons to increase the consumption of oxygen and increase the endurance capacity of individual. Remote monitoring.Heart rate. INTRODUCTION Physical exercise testing requires the interaction of physiologic mechanism which enable the cardiovascular and respiratory systems to support the increased energy demands of contracting muscles. The solution for this is continuous remote monitoring of patients. This is a “wireless” device (telemetric system). WAWHAL SONIYA Ms. PROF. JNEC. Both systems are consequently stressed during exercise. 2. Aurangabad. The 216 of 446 Copyright @ Department of Electrical Engineering. In Fig. but new type of gas analyzers can also be used. archiving and visualization at supervising place using computing equipment (desktop PC. data transport and their processing. 5. MGM’s. The KARD system [1. NIT Durgapur . Under remote monitoring the sensing of important data related to person’s health wherever they are can be transmitted into technological centre where they will be evaluated and understood. many of other Fig.

**receiver is connected to personal computer (PC) which shows
**

the information on the The following devices are used for

B. TELEKARD system

**heart beat. The digital output can be connected to
**

microcontroller directly to measure the Beats per minute

(BPM) rate. It works on principle of light modulation by blood

flow through finger at each pulse.

SENSOR CO2: MQ-6 is used as CO2 sensor for measuring the

content of Carbon dioxide in human breath.

SENSOR O2:OOM102-1 is used as oxygen sensor.

Its measuring range is 0-100%.

**a) %O2 Expiratory- Oxygen sensor, range 0 to 100%.
**

b) %CO2 Expiratory-Infrared carbon dioxide sensor, 0-10%.

c) Heart pulse-Heart rate sensor or heart rate meter

SPORTMETER is used.

Additionally, other parameters can be measured like,

d) Flow-turbine flow meter, 26mm, VE range 10-250 1/min.

e) Pressure Sensor-differential dual ports integrated silicon

pressure sensor on chip signal conditioned, temperature

compensated and calibrated MPX5010 is used.

f) Temperature sensor-LM 35 can be used as temperature

sensor for measurement of temperature.

2.

Microcontroller

**In this paper we are using Tmega16U4/ATmega32U4
**

microcontroller. This is a high performance, low power AVR

8-bit microcontroller. It has non volatile program and data

memories. It has 131 instruction and these mostly require

single clock for execution so the time required for execution

of data is far less than other microcontrollers.

3.

Zigbee RF module

a)

Serial Communications

**The Xbee/Xbee-PRO OEM RF modules interface to
**

host device through logic-level asynchronous serial port.

Through its serial port, the module can communicate with

any logic and voltage compatible UART, or through a

level translator to any serial device (For example through

RS-232 or USB interface board) [9].

4

1

Fig.2. TELEKARD System

**The TELEKARD system is used. The wireless data is
**

used. The signals a) - f) are processed in AVR Atmel16

microcontroller is used and transmitted to a 2.4GHz RF

module [6,7],where they are elaborated and presented to the

user using PC.

A

S

E

I

**Devices that have UART interface can directly
**

connect to pins of RF-module as shown below.

Fig.4. System Data-Flow in UART-interfaced Environment

c)

Serial Data

**Data enters the module UART through the DI
**

pin(pin-3) as an asynchronous serial signal. The signal

should be idle high when no data is being transmitted.

Each data byte consist of a start bit(low), 8-data

bits(least significant bit) and stop bit(high). The following

figure illustrates the serial bit pattern of data passing

through the module.

Fig.3.Block Diagram of Telekard System

1.

b) UART Data Flow

Heart Rate Sensor

**Heart beat sensor is designed to give digital output of
**

heart beat when a finger is placed on it. When the heart beat

detector is working, the beat LED flashes in unison with each

217 of 446

Copyright @ Department of Electrical Engineering, NIT Durgapur

Fig.5.UART data packets as transmitted through RF-module

**The module UART performs tasks, such as timing
**

and parity checking, that are needed for data communication.

Serial communication depends on two UART’s to be

configured with compatible settings (baud rates, parity, start

bit, stop bit, and data bits) [9, 10].

III.

TABLE I. RESULTS

**LIST OF MEASURED AND CALCULATED
**

DATA

Name

30

29

Heart Rate

70

73

V.

**The system’s non-invasive nature and ease of use makes it
**

ideally suited for clinical situations where monitoring of

cardiovascular and ventilatory systems responses under

conditions of precisely controlled stress is desirable. Exercise

testing with appropriate gas exchange measurements can also

serve to grade the adequacy of cardiorespiratory function and

benefit from the convenience of this simple, inexpensive test.

From the measured data, many of other parameters are

calculated by means of computer.

The tested results with reference parameters are as under

shown in table 1.

4

1

REFERENCES

**[1] M. Stork, ” The System for Spiroergometric Data Measuring and
**

Evaluation, in Proceedings of measurement 2001”,ISBN 80-967402-5-3,

EAN 9788096740253,2001, pp.305-308.

[2]

M. Stork, ”Hardware and software for Spiroergometric

Examination”, Medicina sportive Bohemica & Slovaca, ISSN 1210-548,

Vol.10, No.2, 2001.

[3] Microcontroller Atmel16 data sheet http: // www. engineersgarage.

com/sites/default/files/Atmega16.pdf

[4] R.S.Khandpur, (2003)”Handbook of Biomedical Instrumentation”,

Tata McGraw Hill Education.

[5] M.Stork ,”Hardware and software for cardiopulmonary exercise

testing,in proceedings of SIMBIOSIS 2001”,VI international

conference, ISBN 80-214-0893-6.Poland,2001,pp.149-154.

[6] M.Stork, The laboratory and telemetry systems for cardiopulmonary

exercise testing”, in proceedings of IFMBE,2nd European and biological

engineering Conference, EMBEC’02, Vienna,Austria,ISBN-3-90135162-0, Part1, December 2002, pp.518-519.

[7] Heart rate sensor Available: http://www.sunrom.com/sensors/heartbeat-sensor/

[8] Xgbee PRO-RF-Module http://www.digi.com/products/wirelesswired-embedded-solutions/zigbee-rf-modules/point-multipointrfmodules/xbee-series1-module#overview

[9] XBee Pro 802.15.4 OEM RF module datasheet Available:

http://www.picaxe.com/docs/xbe001.pdf

A

S

E

I

IV. RESULT

CONCLUSION

**The embedded system for some physiologic data
**

during physical activities was presented. A properly

administered exercise test allows the objective and

quantitative assessment of the patient’s performance, reserves

and limits that are necessary to make a correct differential

diagnosis and institute appropriate therapy. The programs and

systems for automatic stress testing have been developed in

cooperation with doctors for more than 15 years and now are

used in approx.20 hospitals in Czech Republic.

Unit

VE

Ventilation

1/min, BTPS

RF

Respiratory frequency

1/min, STPD

Vt

Tidal Volume

1, BTPS

VO2

Oxygen uptake

1/min, STPD

VCO2

Production of carbondioxide

1/min, STPD

FeO2

mixed expired O2

%, dry

FeCO2

mixed expired CO2

%, dry

HR

Heart rate frequency

bpm

RQ

Respiratory quotient

--VE/VO2 Ventilatory Equivalent for O2

--VE/CO2 Ventilatory Equivalent for CO2

--VO2

Oxygen pulse

ml/bpm

VO2/Kg

VO2 per Kg

ml/min*Kg

VO2 peak Maximum Value of VO2

ml/min, STPD

VEmax

Maximum value VE

1/min, BTPS

HRmax

Maximum value HR

ml/bpm

Measured

Parameters

Temperature

**During the exercise testing the KARD or TELEKARD
**

acquires the following main signals:

Heart frequency, Flow, %O2 Expiratory, %CO2 Expiratory.

From these signals, the standard parameters are calculated by

the program are the following:

Symbol

Standard

Parameters

Name

218 of 446

Copyright @ Department of Electrical Engineering, NIT Durgapur

**Resonant Frequency of 300-600-900 Right Angle
**

Triangular Patch Antenna with and without

Suspended Substrate

M. Biswas

Department of Electronics & Tele-Communication

Engineering, Jadavpur University

Kolkata, India

e-mail: mbiswas@ieee.org

M. Dam, B. Chakraborty

Dept. of Electronics & Communication Engg.

Hooghly Engineering and Technology College

Hooghly, India

e-mail: mihirdam@yahoo.co.in

**Abstract— A new, very simple expression for the effective side
**

length is formulated to compute the resonance frequency of a 300600-900 right angle triangular patch antenna with and without

suspended substrate. An effective permittivity expression has

been used for the proposed antenna structure. The present model

is found to be valid for both low and high permittivity substrate.

A standard commercial simulator (HFSS) has been employed to

validate the present model. Good correlation (average % error

1.357) is observed between theoretical and simulated values

compared to other model available in literature.

ρ

I.

0

30

0

a

εr1

h1

I

4

1

g

**Keywords— resonance frequency, CAD, 300-600-900 right angle
**

triangular patch, suspended substrate.

εr2 = 1

S

E

I

h

h2

II

A

INTRODUCTION

**The triangular microstrip patch is a well-known microwave
**

network element due to its compactness compared to other

patch shape. It finds extensive application in the design of

many useful microwave circuit components like circulator,

resonators and filters [1]. Narrow band structure of triangular

microstrip patch antenna supports its application in array

configuration with reduced coupling [2].

Among all triangular geometry only equilateral shape has

been widely investigated by several researchers [1-8]. Among

other geometry of triangular patch that have attracted much

attention lately is the scalene triangular patch antenna. This is

due to its small size and less complexity in its design

compared to other triangular shapes (isosceles and

equilateral), circular as well as rectangular patch antennas as

described by James [8]. The 300-600-900 right angle triangular

patch antenna (scalene category) is most attractive patch shape

because it having least area among all the standerd triangular

patch operating in same frequency as shown by Olaimat [9].

The 300-600-900 right angle triangular patch antennaa (RTPA)

is least investigated till now. Only Olaimat [9] has predicted

theoretically the resonant frequency of the 300-600-900 RTPA

for dominant and higher order mode. The effect of fringing

field is not included in the theoretical model proposed by

Olaimat [9]. So, large discrepancy found between the

simulated and theoretical values. Full wave analysis [7] and

60

**Fig. 1. (a). A schematic diagram of probe-fed
**

triangular patch antenna

300-600-900 right angle

**other commercial software [10] are not efficient due to their
**

complexity and large time requirement. So, they are not

effective in of direct synthesis of antenna, whereas CAD

model containing closed form expression is highly suitable for

synthesis of patch antenna.

We have addressed these issues and proposed a very simple

and accurate CAD model based on cavity model analysis to

compute the resonant frequency of 300-600-900 RTPA with

and without suspended substrate. A new, very simple

expression for the effective side length is formulated for better

accuracy. The proposed model is found to be valid for both

low and permittivity substrate. A standard commercial

simulator (HFSS) has been employed to validate the present

model. Good correlation (average % error 1.357) is observed

between theoretical and simulated values compared to other

model available in literature. For designing the MIC on

semiconductor materials with εr≈ 10, this model will be

widely applicable.

219 of 446

Copyright @ Department of Electrical Engineering, NIT Durgapur

TABLE I

THEORETICAL AND SIMULATED VALUES OF RESONANT

FREQUENCIES OF A 300-600-900 RIGHT ANGLE TRIANGULAR PATCH

ANTENNA ON SINGLE SUBSTRATE OPERATED IN DIFFERENT

MODES

h

p =

a

ε

+ 0.3

r , eff

ε

−

r , eff 0.258

**Resonant frequency(GHz)
**

Mode

a

(mm)

εr = εr1

60

h = h1

(mm)

2.4

0.8265

Simulated

[10]

50

4.5

0.7

Present

Olaimat [9]

1.810

1.825

1.863

TM11

3.105

3.161

3.227

TM20

3.540

3.650

3.727

TM21

4.765

4.829

4.930

TM30

5.325

5.475

5.590

1.979

4.116

Avg. %Error

(3)

The q is obtained from [13] by taking equivalence relation

between rectangular and triangular geometry considering

equal area as

Computed

TM10

TM10

1.620

1.611

1.633

TM11

2.809

2.790

2.828

TM20

3.233

3.222

3.266

TM21

4.341

4.262

4.320

TM30

4.847

4.833

4.899

Avg. %Error

0.736

0.811

Total Avg. %Error

1.357

2.464

ε

r , eff − 1

a

+

0.1925 h + 0.5092 + 0.0947

2

ε

r , eff

−

0

.

1925

ε

+1

a

r , eff

1.675 +1.1547 ln 0.1667 + 0.94

2πε

h

r , eff

WL = (1 / 2 3 ) a 2

(4)

Now we equate the zeroth order resonant frequencies of

rectangular patch and triangular patch without fringing to get

f r ,10 =

c

c

=

2 L ε r a 3ε r

(5)

3

Hence, L =

a

2

From relation () and () we get

1

(6)

W= a

3

The expression for computing the εr,eff can be proposed as [3]

ε r ,eff =

4

1

ε re + 1 ε re − 1

2

+

2

A

12 h

1+

3 a

2

−1 / 2

(7)

where, εre is the relative permittivity of the medium below the

II.

patch can be computed as

THEORY

**Based on cavity model we have considered side walls of
**

the patch are perfectly magnetic. According to this assumption

the resonant frequency fr,nm of TMnml modes for 300-600-900

right angle triangular patch antenna can be computed using a

set of simple and accurate closed form expressions [9, 11-12]

as

=

c

a eff 3 ε

r , eff

1/ 2

n 2 + nm + m 2

(1)

**where, c is the velocity of light in free space, εr,eff is the
**

effective permittivity of the medium below the patch and aeff

is the effective side length of the patch, m, n, l are integers

which are never zero simultaneously satisfying the condition

n+ m+l =0.

The effective side length of the 300-600-900 RTPA can be

computed as

(2)

a eff = a (1 + q )

In equation (2), q arises due to the fringing fields at the edge

of patch. Here q can be expressed as [13]

computed

simulated [10 ]

4

(GHz)

r , nm

fr,10

f

S

E

I

6

2

0

20

30

40

50

60

70

a (mm)

Fig. 2. Theoretical and software computed variation of dominant mode

resonant frequency of a 300-600-900 right angle triangular patch antenna as a

function of side length with substrate thickness h =h1 = 0.8265 mm and εr =

εr1 = 2.4.

220 of 446

Copyright @ Department of Electrical Engineering, NIT Durgapur

ε re =

ε r1 ε r 2 h

ε r1 h2 + ε r 2 h1

9

(8)

8

computed

simulated [10]

7

TM11

6

RESULT

**Table I shows a comparative study of theoretical and
**

simulated values of resonant frequencies of a 300-600-900 right

angle triangular patch antenna on single substrate operated in

different modes. Here close agreement (avg. % error 1.357) is

revealed with respect to simulation compared to Olaimat [9].

The variation of dominant mode resonant frequency of a 300600-900 RTPA having substrate thickness h =h1 = 0.8265 mm

and dielectric constant εr = εr1 = 2.4 as a function of side

length is visualized in Fig.2. Theoretical curve employing

present model is compared with simulated [10] data and it

reveals good agreements between them.

The variation of resonant frequency of a 300-600-900 RTPA

on single substrate with substrate thickness (h = h1) for

different dielectric constant (εr = εr1) is visualized in Fig. 3.

Here we have compared theoretical curve employing present

model with simulated values [10] and it exhibits close

correlation for total range of substrate thickness.

Theoretical and simulated variation of TM10 & TM11 mode

resonant frequency as a function of suspended substrate

thickness h2 of 300-600-900 RTPA with fixed upper substrate

thickness h1 = 0.8265 mm, r = 40 mm, εr1 = 2.4, εr2 = 1.0 is

shown in Fig.4. Comparison show good proximity between

fr,10(GHz)

III.

2.0

ε =ε = 2.4

r r1

fr,10 (GHz)

1.6

1.2

ε =ε

5

7

9

11

13

IV.

0.6

0.8

1.0

CONCLUSION

**A new, very simple and accurate CAD model based on cavity
**

model analysis has been proposed to predict accurately

resonant frequency of a 300-600-900 RTPA with and without

suspended substrate. An accurate expression for the effective

side length and permittivity of 300-600-900 RTPA is proposed.

Computed resonant frequencies of a 300-600-900 RTPA

employing present model are compared simulated and

theoretical values available in open literature. The present

model shows very close agreement (average % error 1.357)

with simulated values compared to other. The present model is

valid for wide range antenna parameter. For designing the

MIC on semiconductor materials with εr≈ 10, this model will

be widely applicable.

4

1

A

REFERENCES

[1]

15

h = h1 (mm)

(b)

Fig. 3. Theoretical and simulated variation of dominant mode resonant

frequency as a function of substrate thickness h of a 300-600-900 right angle

triangular patch antenna with side length a =60 mm.

0.4

h2 (mm)

[2]

3

0.2

**Fig. 4. Theoretical and simulated variation of TM10 & TM11 mode resonant
**

frequency as a function of air gap thickness h2 of a 300-600-900 right angle

triangular patch antenna with fixed substrate thickness h1 = 0.8265 mm, a =

40 mm εr1 = 2.4, εr2 = 1.0.

0.4

1

TM10

0

0.0

S

E

I

0.0

3

1

r r1 = 10.0

0.8

4

2

2.4

computed

simulated [10]

5

[3]

[4]

[5]

**theoretical curve and simulated values [10].
**

[6]

**J.-S. Hong and M.J. Lancaster, “Theory and experiment of dual-mode
**

microstrip triangular patch resonators and filters,” IEEE Trans.

Microwave Theory Tech., vol. 52, pp. 1237 – 1243, 2004.

J.T.S. Sumantyo, K. Ito, and M. Takahashi, “Dual-band circularly

polarized equilateral triangular-patch array antenna for mobile satellite

communications,” IEEE Trans. Antennas Propagat., vol. 53, pp. 3477 –

3485, 2005.

R. Garg, P. Bhartia, I. Bahl, and A. Ittipiboon, Microstrip Antenna

Design Handbook, Artech House, Canton, MA, 2001.

M. Biswas and M. Dam, “Characteristics of equilateral triangular patch

antenna on suspended and composite substrates,” Electromagnetics, vol.

33, pp. 99–115, 2013.

M. Biswas and M. Dam, “Fast and accurate model of equilateral

triangular patch antennas with and without suspended substrates,”

Microwave Opt. Technol. Lett., vol. 54, pp. 2663-2668, 2012.

A. K. Sharma and B. Bhat, “Analysis of Triangular Microstrip

Resonator,” IEEE Trans. Microwave Theory Tech., Vol. 30, no. 11, pp.

2029 – 2031, Nov. 1982.

221 of 446

Copyright @ Department of Electrical Engineering, NIT Durgapur

**H. R. Hassani and D. Mirshekar Syahkal, “Analysis of triangular patch
**

antennas including radome effects,” IEE Proceedings H., vol. 139, no.

3, pp.251-256, Jun. 1992.

[8] J.R.James , P.S, Hall. The Handbook of Microstrip Antennas. ,Vol. 1 &

2. , Peter Peregrinus Ltd., London,1989.

[9] M. M. Olaimat and N. I. Dib, “A study of 150-750-900 angles triangular

patch antenna,” Progress In Electromagnetics Research Lett., vol. 21, pp.

1-9, 2011.

[10] HFSS 13: Ansoft’s Corp.

**[11] P. L. Overfelt and D. J. White, “TE and TM Modes of some triangular
**

cross-section waveguides using superposition of plane waves,” IEEE

Trans. on Microwave Theory Tech., vol. 34, pp. 161-167, 1986.

[12] J. Zhang, and J. Fu, “Comments on`TE and TM modes of some

triangular cross-section waveguides using superposition of plane

waves,” IEEE Trans. Microwave Theory Tech., vol. 39, pp. 612-613,

1991.

[13] F. Abboud, J. P. Damiano and A. Papiernik, “Simple model for the input

impedance of coax-fed rectangular microstrip patch antenna for CAD,”

IEE Proc. Pt. H.,vol. 35, pp. 323-326, 1988.

[7]

4

1

A

S

E

I

222 of 446

Copyright @ Department of Electrical Engineering, NIT Durgapur

**Model Free Adaptive Control in Industrial Process:
**

An Overview

Khaing Yadana Swe, Lillie Dewan

Electrical Engineering Department

NIT (Kurukshetra)

Kurukshetra, India

khaingyadanaswe18@gmail.com, l_dewan@nitkkr.ac.in

**algorithm can be applied to any single input or multi-input
**

open-loop system which is stable controllable and continuous

or continuous reaction system [4].

**Abstract— This overview presents Model Free Adaptive
**

(MFA) control and its applications in industrial process control.

Model Free Adaptive Control has been widely used in most

process industries since it was invented in 1997. It is the

advanced controller in the market with significant economic

benefits for the users. It can control complex systems without the

need to use first principal model or identification-based models.

It can control single-variable or multi-variable open-loop stable

controllable industrial process without requiring any tuning,

qualitative knowledge of the process or process identifiers.

Keywords—MFA Control; Industrial Process; Applications of

MFA;

I.

INTRODUCTION

**In process industry, it can be clearly seen that conventional
**

controls are widely suited for applications where the processes

can be reasonably described in advance. However, when the

plant’s dynamics are hard to characterize precisely or are

subject to environmental uncertainties, one may encounter

difficulties in applying the conventional controller design

methodologies. In this case, an alternative model-free adaptive

control (MFAC) can be used [1].

I

**Model Free Adaptive Control was proposed in 1994 by
**

HOU Zhongsheng in his doctoral thesis. His method is based

on a new pseudogradient vector and pseudo-order, and uses a

series of dynamic linear time-varying model (tight format,

partial format, wide format linear model) to replace the discrete

time nonlinear systems in the vicinity of the controlled system,

this model uses the I/O data to estimate the pseudo-gradient

vector of the process, for the MFA control [3].

In 1997, Dr. George Cheng introduced MFA control

technology based on neural network. His idea approaches

model free adaptive control with neural network. This

4

1

A

S

E

**Model Free Adaptive (MFA) Control, as its name implies,
**

is an adaptive control which does not need any information of

system model within the corresponding controller and it uses

I/O data of the controlled system to design controller, even

without any information of the system. It is novel and effective

idea and method to solve complex industrial process control

and it is totally different from PID and self-tuning PID control,

fuzzy control, neural network control, expert system control

[2].

**MFA control avoids the fundamental problems (such as the
**

headache of off-line training that might be required, the

tradeoff between the persistent excitation of signals for correct

identification and the steady state system response for control

performance, model convergence and local minimum problems

and system stability issues) by not using any identification

mechanism in the system. Once an MFA controller is launched,

no controller parameter tuning is typically required. The MFA

algorithms used to update the weighting factors are based on a

sole objective, which is to minimize the error between Set

Point and Process Variable. That means, when the process is in

a steady state where error is close to zero, there is no need to

update the MFA weighting factors [2].

**Model-Free controller is not a panacea, it is based on the
**

principle of feedback control, and applicable to the object with

clear controlled variables, operation variables and settings, and

can constitute a closed-loop regulation loop structure. ModelFree control technology has proven the applicability of the

linear and non-linear open-loop stability process in theory. (Not

suitable for open-loop instability process) [5].

**A Model-Free Adaptive control system can be defined with
**

the following properties:

• Precise quantitative knowledge of the process is not

necessary;

• Process identification mechanism or identifier is not

included in the system;

• Controller design for a specific process is not needed;

• Manual tuning of controller parameters is not required;

and

• Closed-loop system stability analysis and criteria are

available to guarantee the system stability.

II.

**DIFFERENCES BETWEEN MFA AND OTHER
**

CONTROLLERS

**By using most of the traditional adaptive control, robust
**

control, predictive control, and intelligent control methods, the

control system needs to design with high level expertise to

223 of 446

Copyright @ Department of Electrical Engineering, NIT Durgapur

**which average users do not have access. Because of the
**

difficulty for implementing these methods, practical control of

complex systems is very difficult and expensive [4].

Model free adaptive control technique has excellent

performance in the aspect of adaptiveness, robustness,

convergence and stability. It is easy to use and require

minimal investment and risk [6].

Candidate for

next generation

mainstream

controller

III.

U kk−−1m

**B. Traditional Adaptive Control
**

For Traditional adaptive control methods, if the quantitative

knowledge of the process is not available, an online or off-line

identifier is required to obtain the process dynamics. For these

reasons, the system identification is difficult, and its accuracy

cannot be guaranteed. They have two main problems, the

dependence on the mathematical model structure of the

controlled plant and mismatch plant dynamics.

TABLE I.

I

**DIFFERENCE BETWEEN CONVENTIONAL AND MODEL-FREE
**

ADAPTIVE CONTROL [6]

Compared

Item

PID

Model

Predictive

Robust

Control

N

Modelbased

Adaptive

N

ModelFree

Adaptive

Y

General

Purpose

Adaptive

Capability

No process

model

No

identification

No controller

design

No controller

manual tuning

Stability

criteria

available

Easy to use and

maintain

Y

N

N

N

N

Y

Y

Y

N

N

N

Y

Y

Y

Y

N

Y

Y

N

N

N

Y

N

Y

Y

Y

Y

Y

Y/N

Y

Y/N

Y

N

N

N

N

Y

N

N

Y

DESIGN OF MFA CONTROL

f [Ykk − n , u (k ), U kk−−1m , k + 1]

(1)

**= one dimensional state output for k = 0, 1,…
**

= input variable for k= 0, 1, …

= discrete time

= unknown order of output y(k)

= unknown order of input u(k)

= general Non-linear function

= {y (k), …., y (k-n)}, n = positive integer

= {u (l), …, u (k-m)}, m = positive integer

The system in (1) needs to meet the following assumptions:

4

1

**l. It is observable and controllable.
**

2. The partial derivative of f(.) about u(k) is continuous.

A

S

E

**Differences between conventional controls and model free
**

adaptive control are mentioned in TABLE 1.

N

**A. Universal Process Model
**

Consider the general discrete-time Non-linear system,

A. PID

Most industrial processes are being controlled by PID

=

y (k + 1)

(proportional–integral-derivative) controllers till now. PID is a

simple general-purpose automatic controller that is useful for

where,

controlling simple processes. PID works the process with

y (k)

linear, time-invariant, small or no dynamic. Accordingly, it

cannot work effectively for the complex system with nonu (k)

linear, time-variant, coupled and which have parameter or

k

structure uncertainties. And it also requires frequent manual

tuning of its parameters when the process dynamics change [4].

m

The typical characteristic of MFA is model-free, which is

n

suitable for the system of nonlinear, strong coupling, strong

f()

interference, and time-varying nature [6]. It does not need

complicated manual tuning, quantitative knowledge of the

Ykk − n

process and learning process [4].

**MFA Control does not need mathematical model of
**

controlled system and any identifier. And it can solve the two

most important modern control theory problems: un-modeled

dynamics, the dependant problem of controlled system model

and parameters [7].

N

**3. It is generalized Lipschitz, that is to say, Δy(k+1) ≤ b,
**

Δu(k) is reasonable for any k and Δu(k) ≠0. Where Δy

(k +1) = y(k+1) - y(k), Δu(k) = u(k) - u(k −1), b is a

constant.

Suppose

that y (k + 1) f [Ykk − n , u (k ), U kk−−1m , k + 1] has a

=

continuous gradient with respect to u(k). When the system is in

the steady state, because of the condition ǁu(k)-u(k-1)ǁ = 0, we

have y(k+1) = y(k). By using these assumptions, we have

∆y (k +=

1) ϕ (k )T ∆u (k )

(

**Equation (2) is known as Universal Model, which is to avoid
**

modeling before controller design and ϕ(k) is pseudo gradient

of f [U kk−−1m , Ykk − n ] [1, 9, 10, 11, 12, 13].

B. Model Free Adaptive Control Algorithm

For the control law algorithm, consider a weighted one-step

ahead control input cost function is adopted as :

J (u (k=

)) | y * (k + 1) − y (k + 1) |2 + λ | u (k ) − u (k − 1) |2

(3)

where

y*(k + 1) = desired output

λ

= penalty/ weighted factor

**In (3), | y * (k + 1) − y (k + 1) |2 can reduce the steady tracking
**

error and λ [u(k)-u(k-1)]2 can restrict the change in control

output [12].

224 of 446

Copyright @ Department of Electrical Engineering, NIT Durgapur

**If the case Δu(k) = 0 comes forth at certain sampling time,
**

equation (1) can be transformed into Compact Form Dynamic

Linearization (CFDL) model as:

y (k + 1) − y (k −=

σ + 1) ϕ (k )[u (k ) − u (k − σ )] and

y (k + 1)= y (k ) + ϕ (k )∆u (k )

(4)

**By substituting equation (4) to equation (3) and solving
**

∂J (u (k ))

= 0 , the control law u(k) is obtained as follows:

∂u (k )

∧

u (k )= u (k − 1) +

ρ ϕ (k )

Fig.1 Principle of MFA Controller

∧

ϕ (k )[ y * (k + 1) − y (k )

∧

IV.

(5)

λ + || ϕ (k ) ||2

A. Types of MFA

A series of MFA controller and related products have been

developed to solve many kinds of control problems. The

followings are the developed MFA controllers and their

purposes [2, 6].

**where ρ is the step factor. This control law (5) thus obtained is
**

model-free, order-free and only I/O data-related.

In the control law defined by (5), the only unclear

parameter is the characteristic parameters ϕ(k), so the main

task is to find ϕ(k) [9]. There are several ways for estimating

ϕ(k) such as recursive least square approach, recursive gradient

algorithm, the least-squares method with time varying factors,

the least-squares with variance re-set, the time-varying

parameter estimation method with Kalman filter, and so on [9],

[5]. The necessary condition that the universal model (2) could

be used in practice is that the estimation of ϕ(k), denoted as

ϕ

�(k), is available in real-time, and is sufficiently accurate.

Consider the below estimation criterion function as

∧

J (ϕ (=

k )) | y * (k ) − y (k − 1) − ϕ (k )∆u (k ) |2 + µ | ϕ (k ) − φ (k − 1) |2

**It can be estimated as given by
**

∧

∧

ϕ (k )= ϕ (k − 1) +

∧

∧

∧

η∆u (k − 1)

µ + ∆u (k − 1)

∧

2

[∆y (k ) − ϕ (k − 1)∆u (k − 1)]

I

ϕ (k ) = ϕ (1) if | ϕ (k ) |≤ ε or | ∆u (k − 1) |≤ ε

(7)

**SISO MFA : to replace PID so that manual controller
**

tuning is eliminated

•

MIMO MFA : to control multivariable processes

•

**Non-Linear MFA : to control extremely nonlinear
**

processes

•

**Anti-delay MFA : to control processes with large time
**

delays

•

**Feed Forward MFA : to deal with large measurable
**

disturbances

The principle of MFA controller is as shown in Fig.1

4

1

•

**Robust MFA : to protect the process variable from
**

running outside a bound

•

Time-varying MFA : to control time varying processes

•

MFA pH Controller : to control pH processes

•

**Anti-delay MFA pH Controller : to control pH
**

processes with varying time delays

•

**Flex-phase MFA : to control open-loop oscillating
**

processes

•

**MFA XRT controller : to control exothermal Reactor
**

Temperature

•

MFA Optimizer

applications

(8)

**The equation (5), (6), (7) are the Model Free Adaptive
**

Control laws which do not need to specify a particular

controlled system, are unrelated with the mathematical model

and the order of the controlled system [1, 5,9,10,11,12].

[2].

•

A

S

E

(6)

DIFFERENT TYPES OF MFA AND IT’S APPLICATION

:

for

dynamic

optimization

**It categorizes various processes into a few types to make it
**

convenient for user to meet different requirements of processes.

B. Applications of MFA

Due to its adaptive, robust and easy to use, MFA control

has been successfully implemented to solve many industrywide control problems with minimal investment and risk. The

followings are some applications of different MFA controller

in process industry [6]:

•

**Oil recovery boilers use Anti-delay MFA for
**

controlling steam dryness and SISO MFA for

combustion.

225 of 446

Copyright @ Department of Electrical Engineering, NIT Durgapur

696-699. It has been successfully used in industrial process. Jiangsu. Cheng. issue 3. Third International Workshop on Advanced Computational Intelligence August 25-27.Suzhou. 2007. Cheng Yinman and Wang Yingfei. IEEE Industry Applications 1998. 2010 Second International Conference on Computer Modeling and Simulation. Harbin. Yaqiong Liu and Bin yin. PR China 2004.333-339.” Hydrocarbon Processing. S. Robust MFA for level and Non-linear MFA for pressure and flow [14]. Comparing with other PID and model-based controllers. Chaos. F. Aug 2010. Washington: US Patent & Trademark Office of the United States Department of Commerce. 28 – 33. power generation. Cheng. MFA control is used by combining with other controllers like PI. “Design of glass furnace control system based on model-free adaptive controller” . and Ovidiu Marin (2001). Guiwen Zheng and Yuguang Niu. • Air separation unit uses Robust MFA for controlling HP column reflux level [17]. Putman Media. 1998. the user can simply choose the suitable type of controller for the application. MFA process control. Cheng Qiming. “The study of model-free adaptive controller based on dSPACE”. Cheng. [10] • Water treatment plant use MFA pH or Anti-delay MFA pH for pH control [19]. and Zhang. Expert Systems with Applications. Min He. As controller design procedures are not needed for MFA.. MFA-PI control is used in nuclear steam turbine water level control [20] and MFA-PID cascade controller is used in thermal power plant to adjust main steam temperature control system [8]. S. and Hongwei Zhangm “Model-dree adaptive control algorithm with data dropout compensation”. May 2001. oil refining."Control. “Model-free adaptive control design using evolutionary-neural compensator”. 226 of 446 Copyright @ Department of Electrical Engineering. Zhongsheng Hou. Solitons and Fractals. pp. W. “Model-Free adaptive control of evaporator. Control and Decision Conference (CCDC). October 2004. August 2008. pp. pp 1168-1171. 243-248. • Tomato process uses SISO MFA for controlling temperature of tomato hotbreaks and temperature of sterilization system and MIMO MFA for evaporator density and level [15]. China. NIT Durgapur . Marcelo Wicthoff Pessôa . Beijing. “Mode-free adaptive technology improves distillation column chain control. June 2004. • Catalyst or powder conveyor process use Anti-delay MFA for controlling vacuum pressure. especially in chemical.” Control Engineering Europe. Wang Mingmei. 14 pages. Ren Dexue.. 2001–2009. and S. “Simulation study on model-free control method in asynchronous motor control system”.• Distillation Columns use 2x1MFA for controlling plate and bottom temperature. S. pp. IEEE International Conference on Industrial Imformatic. M.” Dynamic Modeling Control Applications for Industry Workshop. Aug 2009.. Houston TX. Shengyang: North-easten University. pp. “The application of model free adaptive control”.D Thesis].608-611. Qi Jianling and Ma Guang. April 2004. “Modelfree adaptive control for model mismatch power converters”. June 2010. 16-19. IEEE. vol 15. S. and Zhang. Rodrigo Rodrigues Sumar and Antonio Augusto Rodrigues Coelho. Dong Chao and Jia Zefang. pp. Gulf Publishing Company. 2000. “Tomato Process Improvements Using Model-Free Adaptive Controllers. Seiver.W. Gao Qiang. It can not only replace the traditional PID control but also solve tough loop control problems. Cheng. 2010 . Chen. Major. Houston TX. United State Patent. Xuhui Bu.” Tomato News. G. “Model-free adaptive control optimization using a chaotic particle swarm approach”. MFA-PID control is used in glass furnace temperature control system [2]. China. MFA shows great advantages in control process. October 4-6. S. April 25. Lv. Cheng. Ping MA. 605524[P].. Dave. G. [7] [8] [11] [12] In some applications. May 2011. [1] [13] Cheng. Springer-Verlag Berlin Heidelberg. 2010 International Conference on Intelligent Computation Technology and Automation. Zhi-Gang Han and Xinghuo Yu. glass etc. issue 4. Science Direct . [14] [15] [16] [17] [18] [19] [20] Leandro dos Santos Coelho. December 1999.B. China. 1994. pp. Zhongsheng Hou. PID and Fuzzy. Multiconference on Computational Engineering in systems Applications (CESA). G.. “ Model-Free Adaptive Control in Water Treatment. “Model-free adaptive control method for nuclear steam generator water level”. 2006. I REFERENCES [2] [3] 4 1 A S E CONCLUSION Model Free Adaptive control becomes a unique technology which can be used as general-purpose solution for various control applications in process industries. Gulf Publishing Company.L.. For example. Article ID 329186. Fashan Yu. Leandro dos Santos Coelho.D. and De-Lin Li (1999). 37. September 2004. IEEE. vol 41. Yang Haitao. January 2010. W. • Coking Furnace use MIMO MFA for oil outlet temperature and SISO MFA for combustion [16]. Antonio Augusto Rodrigues Coelho." Hydrocarbon Processing. Hindawi Publishing Corporation Mathematical Problems in Engineering Volume 2012. configure with certain parameter and just launch the controller [6]. S. Jianhua Guo. G. Science Direct. “An adaptive model free control design and its applications”. G. "Model-Free Coking Furnace Adaptive Control. pp. IEE Computing & Control Engineering. IEEE. issue 1. Second International Symposium on Intelligent Information Technology Application. “A model-free adaptive control of pulsed GTAW”.. "Air Separation Advances with MFA Control. Haitao Yang. vol. Wu Jianhua. 2012.130-133. [9] • Evaporators use MIMO MFA or FB/FF MFA for controlling density and level [18]. Wei Li. G. [4] [5] [6] • Gas mixing plant use Anti-delay MFA for controlling heating value and Robust MFA for gas pressure and gas flow. “Model free adaptive control (MFAC)”. pp 499–508”. Dai. Zhang Haixin and Zhu Mingguang. pp. Jianhua Wu. S. adaptive control and model free learning adaptive control for non-linear system [Ph. V. “The parameter identification. Z.

Keywords— Half bridge Series Resonant Inverter. The AC main (220V. The simulated results have been compared with analytical result. ZVS and ZCS have been made by Llorente et al [1]. The series resonant based multi-inverter used for multiple induction heaters is described by Lucia et al. 50 Hz 1-Ph RX CX Harmonic filter N C IGBT2 M Rsn2 C2 B O P Signal for detection of cooking vessel Fig. Cohen [10]. [16] developed a series resonant inverter based induction cooker with two heating zone. PSPICE Simulation II. The principle of inverter operation has been presented and different waveforms are shown at different modes of the inverter. half bridge. And results are coming almost similar in nature. [15] built a model based on series resonant ZVS inverter to supply several 220 V. cost saving and simple circuit configuration half bridge series resonant inverter is most popular. Two single switch inverter topologies ZVS and ZCS are described by Omori et al. Burdio et al. The heating coil is made up of litz wire and is connected with a high frequency (4kHz to 50kHz) power source. The coupling between heating coil and cooking pan is modeled as the series connection of an inductor and resistor based on transformer analogy. The circuit is also simulated using PSPICE software. [17]. [19]. Then an inverter supplies high frequency alternating current to the heating coil. Jung described dual bridge series resonant inverter for 2 loads [13]. Durgapur. INTRODUCTION Demand of domestic induction cooker increases day-byday because of its inherent advantages.Roy Engg. The results are validated with real time experimental model simulation. Durgapur. sterilization plant and drier for surgical instruments by Sadhu et al. Now-a-days resonant inverter topologies are commonly used for induction cooker to produce high frequency resonance loss at the cooking pan. In this paper an attempt is made to analyze the circuit of half bridge series resonant based mirror inverter analytically and using PSPICE software simulation.5 [1].1. The capacitor C is of small capacity (5uF) so that the DC voltage (Vdc) across C does not get leveled. Cooking vessel Vessel support Choke Q R IGBT1 A Rsn1 C1 Bridge rectifier I. The return path of the high frequency current is through this capacitor C as at high frequency C offers negligible capacitive reactance 227 of 446 Copyright @ Department of Electrical Engineering. inverter is operated in Zero Voltage Switching (ZVS) or Zero Current Switching (ZCS) condition. The internal resistance of the cooking pan causes heat to be dissipated following Joules effect. The output of the rectifier is passed through an inductor and a capacitor C. Forest et al. The load power factor is usually considered around 0. In the domestic induction cooker copper made heating coil is placed beneath the ferromagnetic cooking pan.F.Simulation of IGBT fed Mirror Inverter based H. B. This in turn also helps to improve the overall power factor of the system. Mostly used inverter topologies are full bridge [2-3] or half bridge [4-8]. The circuit of half bridge inverter using the principle of positive negative phase shift control under ZVS and non-ZVS operation for small size and low voltage induction cooker is analysed by Achara et al.1 illustrates the mirror inverter circuit. the alternating magnetic flux is induced at cooking pan and produce eddy current in it. and Leisten & Hobson [11]. Abstract— This paper deals with the circuit analysis of a half bridge series resonant IGBT-fed mirror inverter based high frequency induction cooker. Induction Cooker Dola Sinha Department of Electrical Engineering Dr. Induction cooker. 50Hz) is routed through EMI / EMC filter before being fed to the bridge rectifier. Mirror Inverter. A bus filter is designed to allow a high voltage ripple and the resultant power factor close to one. India resonant loads. [12]. From the literature it can be concluded that due to robustness. [14] used hybrid inverter for induction heating using ZVS and ZCS condition.C. Circuit of Mirror inverter Fig. To reduce the switching loss. College. Sadhu et al. 4 1 CIRCUIT CONFIGURATION AND ANALYSIS OF MIRROR INVERTER S E I A Mirror inverters are basically half bridge series resonant inverter and commonly used for medium power induction heating applications [18]. The series-resonant radio-frequency mirror inverter system has been introduced as a new conception for induction-heated pipeline or vessel fluid heating in medicinal plant. [9]. The overall comparison considering full bridge. The induction cooker takes the energy from the mains voltage and this voltage is then rectified by a bridge rectifier. NIT Durgapur . At high frequency.

. these four modes will repeat for continuous conduction. Mode–4: When both the IGBTs are OFF.e. hence. Leq. The operating modes of mirror inverter are separately analyzed below. therefore. IGBT1 and C2 and the next resonant cycle is due to Req. 1 C c L eq cos t tan 1 R eq can be calculated from the initial conditions. Mode–1: When IGBT -1 is ON and IGBT-2 is OFF. The main supply voltage is 220V and applied operating frequency at the mains is 50Hz. Fig. There has been no conduction through IGBTs. (4) RESULTS AND DISCUSSIONS In this study. It also acts as higher order harmonic filter at the same cost. A1 and A2 A S E I 2 C R eq 2 The principle of operation of the circuit is described by dividing it into four different modes. In this mode C1 discharges from Vdc to zero through the path 2 QRMNQ. 2 shown that charging current of C1 and discharging current of C2 both flow in the same path N to M. Rsn2 C2 B O P Analyzing all the modes. 2. As the values of snubber resistors are very high (470kohm). respectively. Between these two cycles there is an OFF period of IGBTs. C1 and IGBT2. METHOD 1: Analytical Solution 228 of 446 Copyright @ Department of Electrical Engineering. series current flowing through the heating coil is expressed as: Fig. the charge on capacitor C2 will act as a source of energy to drive current and thus charge C1 from zero to Vdc . maximum current flows through the capacitors. VC1 charged through the path AQNMPOBA. The high frequency alternating current is flowing through capacitor C because at high frequency the capacitive reactance offered by C is negligible hence the capacitor acts as a short circuit and allowing the high frequency current to flow through it. currents and voltages have been compared across the heating coil. A small voltage drop appears across the coil impedance and the rest voltage is equally shared by the capacitor C1 and C2 as initial charge voltages VC1 and VC2 respectively and the value of this voltage is almost Vdc . where f is in KHz range. IGBT is used as the power semiconductor switch for its superiority for domestic induction cooker operating below the frequency range of 50 kHz [20]. those two parameters are obtained using the expressions (1) and (3). One resonant cycle is due to combination of Req. After the end of this mode Leq Req Q A R IGBT 1 C1 C Vdc 2 Rsn1 both C1 and C2 store same voltage i. In the first method. a P-SPICE-based simulation has been carried out. Both the cycles get repeated alternatively to push resonant current through the heating coil. the circuit current flows through the snubber resistors Rsn1 and Rsn2 and capacitors C1 and C2. 2 M N IGBT 2 Mode 1 to Mode 4. In the next method. the capacitor C acts as a short circuit and allows high frequency current to flow. 2 c 1 2 L eq Exp k1t A1 cos k 2 t A2 sin k 2 t (1) R where. used according to the operation of C1 or C2. The equivalent circuit of mirror inverter The Operating principal: Vdc i1 (t)= Initial mode: When both the IGBTs are OFF and capacitors C1 and C2 are not initially charged.67. NIT Durgapur . the charge on capacitor C1 will now act as a source of energy to drive current and thus charge C2 from zero to Vdc . Mode–2: When both the IGBTs are OFF. It is ( X c 1 2 fC ).Voltage across heating coil can be obtained using the expression Vcoil Req i1 (t ) Leq di1 (t ) dt (3) The voltage stored in capacitors C1 and C2 during charging will be expressed as: t 1 VC 1 VC 2 i1 (t )dt Cc 0 III.99 and 48. These two cycles are in ON period of IGBTs.2 shows the equivalent circuit of mirror inverter. VC2 charged through the path AQRMNOBA. Two different methods have been 1. adopted during this study. It is shown that charging current of C2 and discharging current of C1 both follow the same path M to N. In this mode C2 discharges from Vdc to zero through the path NMPON. k1 and k 2 ( Leq Cc ) 1 k12 2 Leq 4 1 (2) Where Cc = C1 or C2. Values of A1 and A2 are calculated from initial values of voltage stored in the capacitors and obtained as 1. Leq.Mode–3: When IGBT -1 is OFF and IGBT-2 is ON. Vdc .

two IGBTs of HGTP6N50E1D are used.4uF C9 5uF -0.161 0. Total time period is taken as 26µsec where IGBT on time is 12μsec and off time is 14µsec.01 0. Voltage through heating coil The complete waveform of current and voltage across the heating coil including ON and OFF time of each switch at high frequency (38. therefore heat penetration is also low. Current through heating coil Operating high 38512Hz 12 usec and IGBT ON/OFF frequency 600 18 usec timing 400 Voltage across coil (V) The four modes (Mode 1 to Mode 4) will repeat according to specified IGBT ON time and OFF time. The circuit of mirror inverter is analytically analyzed by MS Excel 2007 and different graphs are shown below at different frequency.4uF -20 5uF Capacitor C 0 0. Applied and capacitor voltages with the time at low frequency.03 METHOD 2: PSPICE Simulation VOFF = 0 0.015 Time (sec) 0. Snubber resistors 15 10 Coil Current (A) TABLE I. Series current at low frequency.4uF HGTP6N50E1D 0 0 V15 V16 V1 = -5V V2 = 5V TD = 20. when both switches are OFF 0.04 0. 20 INPUT PARAMETERS OF MIRROR INVERTER 470kohm Mains Rsn1 & Rsn2 supply 220V 5 0 -5 voltage -10 Coil inductance (L) 119uH Internal resistance 0. 6 respectively.001 R9 100uH 2 2 -0.081 0.101 0.69 D21 1 Z5 C10 0.121 0. So.041 0. For high frequency inverter. The depth of heat penetration on cooking pan is inversely proportional to operating frequency and the operating frequency is inversed of operating time period.015 0. 5 and Fig.02 0.02 0.02 0.021 0. A bridge rectifier is modeled using four diodes of 1N6392 type.03 4 1 Time (sec.2. by changing the IGBT ON-OFF time operating frequency can be changed and thus the heat penetration on cooking pan can be controlled.The main equivalent circuit of the mirror inverter is shown in Fig.01us TR = 2us TF = 2us PW = 12us PER = 26us Z6 C11 0. The parameters are considered for the mirror inverter is shown in Table I. Since the current is very low at low frequency. 200 0 -200 -400 -600 0 0. 7.03 -0. 5.025 0.025 0. when both the switches are OFF D22 1N6392 V1 = -5V V2 = 5V TD = 0.005 C2 0.02 119uH 0. 4.512 kHz) is shown in Fig.01 1 Current (Amp) 0. 3 and 4 show voltage and current across the heating coil at low frequency. A S E I Fig.01 L9 L8 0 1N6392 D20 -0. D23 1N6392 1 1 Fig.01 0.03 Fig.04 The developed PSPICE schematic circuit diagram is shown in Fig.181 V14 R10 470k HGTP6N50E1D VAMPL = 220V F REQ = 50Hz 2 2 Time (sec) Figs. NIT Durgapur R11 470k .061 0. The circuit diagram for PSPICE simulation. 7.69 ohm -15 (R) Capacitors C1 and 0. 6.01us TR = 2us TF = 2us PW = 12us PER = 26us 0 Fig.) Fig. 0. 229 of 446 Copyright @ Department of Electrical Engineering. 3.141 0. The circuit simulation has been done with alternating supply voltage of 220 V.005 0.

M. A comparative study of resonant inverter topologies used in induction cooker. 2005. Karan. 9. IEEE Power Electronics Specialists Conf.. So by changing these parameters the current and voltage of the heating coil will be changed. A 50-150 kHz half bridge inverter for induction heating application.. Design of the half bridge series resonant converter for induction cooking. vol XL. [18] [19] [20] 230 of 446 Copyright @ Department of Electrical Engineering. New Delhi. 60-64. vol.. vol. (Req=R9)= 0. Costa and I. H Yamasita.M.2. The waveform of current through heating coil..a tool to reduce switching & conduction losses of high frequency hybrid resonant inverter fed induction cooker International Journal of Computer and Electrical Engineering. 1995. vol. R.W. IEEE trans. 729-735.J. 1998. 265-270. Y. Dual element induction cooking unit using power MOSFETs. J.6. A two-out put series resonant inverter for induction-heating cooking appliance. H. Patent Office – Government of India.A. 157-163. systems and computers.A. S. Principle of a multi load single converter system for low power induction heating.D. Conf. Millan.24.4. 1999. C10 and C11 = 0. Int. Hobson. Koertzen. pp. S. Kamli. L. IEEE Applied Power Electronics Conf. P. Yoo and D. no. J.59. J. I. pp.S. pp.The parameters are considered for the PSpice simulation of mirror inverter is shown in Table II. pp. D. Martinez.N. pp. Nakaoka and T. F. pp. S.M. vol. Viriya and K Matsuse. Vol.. F. C9 = 5µF [5] IGBT ON and OFF timing = 12 µs and 14 µs [6] [7] [8] [9] [10] Fig. D. IEEE Trans.M. Feb. pp. 1985. 1993. L. Tebb and F.K. pp. Firstly. pp. P.. Tebb. Hyun. [11] [12] L. Journal Electronics. 1991. 430-441. I.35. World Scientific journal of circuits. Sinha and A. Hobson and D. each Supply Voltage = 220V [4] Coil resistance. The series current flowing through heating coil and the voltage across load depends on the resistance and inductance of the heating coil. 223-230. A novel type induction heating single ended resonant inverter using new bipolar Darlington-transistors. Cohen. Half bridge series resonant inverter for induction heating applications with load adaptive PFM control strategy. A comparison of load commutated inverter system for induction heating and melting applications. Results obtained through all these methods are found to be similar and must be feasible for real implementation. Analysis of a half bridge inverter for a small size induction cooker using positive-negative phase shift control under ZVS and non-ZVS operation. R. R10 and R11 = 470kohm. Power Electronics.K. IEEE Trans Industry Applications. 575-581. (APEC) Rec.. IEEE Power Electronics Specialists Conf. The simulated current and voltage waveforms are plotted in Fig. Llorente. European Power Electronics Conf. J. Barragan. Omori.P. Dual half bridge series resonant inverter for induction heating applications with two loads.P.1996. 2. vol. IEEE Applied Power Electronics Conf. 224-230. analytical expressions have been developed for different modes of operation.. Electronics letters. IEEE Trans. vol. Wang et al. H. Lucica. F. S.59. Series resonant multi-inverter for multiple induction heaters. S.4µF.16. (PESC) Rec. A new generation energy efficient sterilization plant for surgical instruments – The Indian Journal of Hospital Pharmacy. 2010. INPUT PARAMETERS OF MIRROR INVERTER FOR PSPICE [3] SIMULATION Operating high frequency = 38512Hz Coil inductance. (PESC) Rec. pp. “Induction heated cooking appliance using new quasiresonant ZVS-PWM inverter with power factor correction”. Evaluation and comparison of power conversion topologies. vol.69 ohm Capacitor. it is simulated through P-SPICE software. no. 3.43.G. Chowdhury. 1985 F. M. J. Listen and L. no. S. Gaspard. 8 and Fig. [17] CONCLUSIONS In this paper. E. Power Electronics. pp.13451346.W. N.N.11. pp. P. P.K. Bandyopadhyay. An improved inverter circuit arrangement. P. pp. Patent Number 69/cal/2001. M.K. Turnbull. 590-599. no. Sadhu. 6. no. N. 2005. 15.533-542. Burdio.R. IEEE Trans Industrial Electronics. Laboure Forest.4. IEEE Int. pp. Y. Jung. A parallel resonant power supply for induction cooking using a GTO. Pal.. Selection of power semiconductor switches .A.163-172. (Leq=L9) = 119µH Capacitors. on Magnetics.3. 1990. (APEC) Rec. Abe. Maruhashi. F. Chakrabarti. Journal of Electronics. 2003. Sadhu. pp. April.M. no. pp. Achara. 2. Mitra. 815-822. pp. O. Dawson and P. no.C. Garcia. Transistorised power supply for induction heating. no.14. J. P. Chakrabarti. Hobson. 4. L. Ferreira. The waveform of voltage across heating coil. 2002 P. Power Electronics. Acero. 1999 H. Monterde. vol. Jain. vol 34. K. R. Burdio and J. NIT Durgapur . Monterde.705-712. 2860-2868. mirror inverter-fed induction cooker has been analyzed in three ways. no. and A.W. 1985. 8. Chakrabarti and D. each Snubber resistors. In the next case. 4 1 A [13] S E I [14] [15] [16] Fig. no. (EPE) Rec. Kwon. Sadhu. REFERENCES [1] [2] TABLE II.1.20. 9 respectively. 2011. 2007. 747-757. Van Wyk and J. Int. on Power Electronics and variable Speed Drivers (PEVSD) Rec. Chowdhury and B. Sadhu. Acero. 9-16. PEDS. A unique induction heated cooking appliances range using hybrid resonant converter. 2000. Barragan. IEEE Trans. 1168-1174. Burdio. No. vol. Jana. J. Yamamoto and M. pp.

PSO and local search method (CST). new hybrid approach for non convex algorithm [13]. The intelligent algorithms have different characteristics and it has some advantages and disadvantages. has been widely used to solve the optimization problems. While. A population based algorithm HIA collects a group of solutions to reach the optimum solution. operating limit constraint are considered. However. to combine global search and local search methods.com Electrical Engineering Deptt. Differential evolution algorithm (DE) [6]. for power producing. is tested in 40 generators system with NOX emission. SOX. since 1980s due to pressure of public demand and implementation of several pollution control acts. SOX. a hybrid intelligent algorithm used for solving EELD problems by combining global search method i.B. NIT. HIA is initialized by a set of random particles which travel through the search space. fine tuning operation in DE is unsuccessful to give better performance. multi-objective evolutionary algorithm [4]. the CST integrated into PSO to improves the diversity of the particle swarm so as to avoid PSO trapping the local optima.A Hybrid Intelligent Algorithm Applied in Economic Emission Load Dispatch Problems Partha Pratim Das. I. Simultaneously. The population-based optimization techniques. this hybridizing technique is an efficient and make best of the advantages and avoid the disadvantages. GA has parallel search capabilities but it takes enormous computational time and concentrate in local minima. power demand equality constraint.C. As a result the concept of economic emission load dispatch (EELD) has come up where cost and emission both are equally considered as multi-objective function. etc. particle swarm optimization (PSO) [5]. COX. Agartala Agartala. India ani_bhatta@rediffmail. NIT Durgapur . To compare with other complicated optimization problem. In the present paper Hybrid Intelligent Algorithm (HIA). different classical methods are not able to solve EELD problems. Kuntal Bhattacharjee Aniruddha Bhattacharya Electrical Engineering Deptt. Particle swarm optimization (PSO). This technique is hybridizing different search methods i. which can handle all the constraints like uni-modal and multi-modal system. chaos searing techniques (CST) [17] can easily skip local optimal solution with the property of non repeatedly traversal all the states according to its own character in a certain range. Due to some disadvantages of popular optimization technique PSO for consumption of taken a lot of time and efforts for appropriate adjustment of the parameters. COX.e. the power generating station has to consider the environmental impact like NOX.Roy Engineering College Durgapur. Gravitational search alogorithm [16] etc. Based on this. The proposed algorithm.com kunti_10@yahoo. Simulation results bear out that the proposed algorithm superior than other several existing optimization techniques. Dr. modified particle swarm optimization [11]. Multiobjective quasi-oppositional teaching learning based optimization [9] are introduced in recent past. This hybrid algorithm is also used to solve the bilevel programming problems. Though PSO has good convergence performance and faster than GA but it is easily trapping in local minima. CST integrated in the PSO to improve the worse particles for overcoming the weakness that PSO may be trapped in local minima. In the earlier age numerous approaches have been developed to reduce the emission such as goal programming technique [1]. Economic emission load dispatch. INTRODUCTION Economic load dispatch (ELD) problem is used to allocating generation of all available generating units to optimum the cost of generation with fulfillment of different equality and inequality constraints of different thermal power plants. Different hybridization and modification have been adopted like niched pareto genetic algorithm (NPGA) [10]. modified NSGA-II algorithm [14]. DE has comparatively better and faster solution using its different crossover strategies.com Abstract— This paper presents hybrid intelligent algorithm (HIA). bee colony optimization [8].e. Opposition-based harmony search algorithm [15]. modified bacterial foraging algorithm [12]. biogeographybased optimization [7]. combine with particle swarm optimization (PSO) and chaos searching technique (CST) to solve the economic and emission load dispatch (EELD) problems of thermal generators of power systems with considering valve point loading effect. But when complexity and size of the system increases. India partha_pratim85@gmail. dynamic programming [2]. Therefore. CST is also judge the point is 231 of 446 Copyright @ Department of Electrical Engineering. Emission substances like NOX. But the common disadvantages of these algorithms are very complicated and using many parameters. Valve point loading ability such as stochastic search technique [3]. But EELD problem has many local optimum solutions and need to consider a huge number of complex constraints. which can overcome these problems due to their efficient searching 4 1 A S E I Keywords—Chaos searching technique (CST).

Economic Emission Dispatch Economic emission dispatch problem may be expressed as: MATHEMATICAL FORMULATION OF EELD PROBLEMS (6) Minimize C = [ F ( P ). It is supervise practically that when fuel cost increases emission consequently decreases and vice versa.. Recently BCS method is proposed [7]. bi and ci are the cost coefficients of the ith generator. NIT Durgapur .. feasible or not by solving problem. The total transmission loss (PL) is expressed using B-coefficients as: N N N PL = ∑ ∑ Pi Bij P j + ∑ B0i Pi + B00 i =1j =1 i =1 (3) 4 1 Where Ei(Pi) is the value of emission and Fi(Pi) represents the total cost of generations. II. The economic emission dispatch reduces the total emission from the system. Numerical result is presented and discussed in Section VI. Section III has brief Idea of PSO. so this problem may be considered as a multiobjective optimization problem. Pimin ≤ Pi ≤ Pimax i = 1. So it is very difficult to make a solution that will give the best best solution. PD is the total system active power demand. The above equation is minimized subject to the following constraints mentioned in (2) and (4). Section II of the paper provides mathematical formulation of different types of EELD problems. (2) The amount by which best compromising solutions served from the global best fuel cost and emissions are calculated using the following indices: ∑n F (P ) − F min i =1 i i FCPI = × 100 Fmax − F min 2) The Generating capacity constraint The real power generated by each generator shall be restricted within its respective lower and upper operating limits. α i . without any anxiety about the rate of emission.. Economic Dispatch The may be described by minimizing the fuel cost function (FT) of EcD problem is presented as: { ( N N FT = ∑ Fi (Pi ) = ∑ ai +biPi +ciPi2 + ei ×sin fi × Pimin−Pi i=1 i=1 )} (1) Where. where E(P) is total amount of NOX released from the system in (kg/h or ton/h). Section IV has brief introduction of Chaos Searching Technique (CST).2. N is the number of generators. The modified overall multi-objective function bi-objective of EELD problem is represented as: The following objectives and constraints are considered for EELD problem. ξ i and λi are the emission coefficients of ith generator. The conclusion is drawn in Section VII. which generally increase the system operating cost. As economic emission load dispatch (EELD) try to find a balance between the fuel cost and emission simultaneously. A. The objective function for economic emission load dispatch is used as (7) and it is used for multi-objective optimization subject to the constraints of (2) and (4).N (4) B. Section V has short description the proposed Hybrid Intelligent Algorithm. Emission Dispatch The emission dispatch for NOx gases may be defined as: () EP = N [ ( N −2 2 E P = ∑10 αi + βi Pi + γ i Pi + ξi exp λi Pi i i i=1 i =1 ∑ ECPI )] (5) N ∑ E (P ) − E min × 100 i =1 i i = E max − E min (8) (9) However. for finding best compromising solutions. C. βi . Ei(Pi) is the ith generator’s emission function 232 of 446 Copyright @ Department of Electrical Engineering. The objective function of (1) is minimized subject to following constraints: 1) Real power balance constraint N ∑ P − ( PD + PL ) = 0 i =1 i N∑ F (P ) − F N∑ E (P ) − E i i min min i=1 + (1 − w)i=1 i i Min C = w F −F E −E (7) max min max min A (2) S E I Where. ei and fi are the coefficients of the ith generator reflecting the valvepoint effects. is the ith generator cost function for Pi output. The aim of this new develop algorithm is to combines the benefit of good convergence performance of PSO with the advantage of easier jumping local optimal solution of CST to overcome the limitations of the EELD problems. E ( P )] i i The economic load dispatch reduces the total fuel cost of the system.for Pi output. γ i . the relative significance between fuel cost and emissions can be varied by changing w in between 0 and 1 in the multi objective function of (7). ai.. Fi(Pi)..

and add the particle to the infeasible list. r1 and r2 are random variables in the range [0. The particle will then modify its direction such that it has additional components towards its own best position. power demand. Step 1: For initialization. The particles. For the first m particles. Vmax. which can force the infeasible particle to be feasible. The feasible weighting value is set up to replace the fitness value of the particle when it is infeasible. pibest and gbest are the positions with the best objective value found so far by particle i and the entire population.e. which are in the feasible list are ranked in ascending order. The best previously visited position of the ith particle. and global best position. go to Step 8. and the particle is feasible. 1]. easy to skip out of the locally minimum value. CST has two steps: first one is to search all the points in turn within the changing range of variables and taking the better point as the current optimum point. The following more convenient equation is used to generate the chaos sequence: zi+1 = µzi (1 − zi ) (12) V. (3) Evaluate the fitness values of the each particle of CST. Then it regards the current optimum point as the centre. zi ϵ [0.1] (i=1. gbest. the velocity of particle and its new position will be assigned according to Eq. Vmax. compute and output the upper level and lower level’s objective function values. W is the inertia weight factor. Compare the first particle’s fitness evaluation with the global best position. PARTICLE SWARM OPTIMIZATION (PSO) Particle swarm optimization (PSO) is very well known and efficient optimization technique was inspired by the social behavior of animals such as fish schooling and bird flocking. The chaos searching technique (CST) is a modern searching method [17]. respectively. Population size is set M =m+n. The position update equation is: ( k +1) ( k +1) k xi = x i + Vi IV. gbest. 1] are set. Then. c1 and c2. compare particle’s fitness assessment with its Pibest. A tiny chaos disturbance is imposed and carefully searches to find out the optimum point. HYBRID INTELLIGENT ALGORITHM The basic idea of this new algorithm is to embed CST into PSO for solving the multi objective EELD problems. k is the number of iteration. set the population size M. the particle is infeasible. Then it add to the feasible list and set the upper level’s objective function value as the fitness value of the particle.. (6) Update the particles. Each element of a given solution in the space should satisfy the equality 233 of 446 Copyright @ Department of Electrical Engineering. is initialized as Xi. If current value is better than Pibest. gbest.III. A. (11) BRIEF INTRODUCTION OF CHAOS SEARCHING TECHNIQUE (CST) Chaos is a kind of non periodic moving style which exists widely in the nonlinear system and is unique to the system. (5) Update the local best position. t = t+1. If current value is better than gbest. respectively. 2. pbest and towards the overall best position. M) particle randomly with initial position. Specify maximum and minimum capacity of each generator. which can avoid the situation that the 4 1 A S E I where. (7) Terminal conditions. the other particles in the infeasible list are ranked in ascending order. then set the current value to gbest. CST also improves the diversity of the particle swarm to avoid PSO trapping the local optima. choose the no of generator units. r1. HIA algorithm for economic emission load dispatch problem: This subsection describes the procedure for implementation of the HIA algorithm for solving the EELD problems.. r2 ϵ [0.2. Using (7) Cost and emission are calculated for different values of w. two learning factors. Xi. otherwise. then set the current value to Pibest. The calculations then continue for a specific number of iterations. in the range of maximum speed. it is calculated using (1) for the system having valve point loading. The maximum number of iterations (T) is initialized. NIT Durgapur . c1 and c2 are acceleration constants. The velocity update equation is: (k +1) Vi k k k = W × Vi + c1 × r1 ( pbesti − xi ) + c 2 × r2 ( g best − xi ) (10) where xi & Vi are the current position and velocity of particle i.. If the solution exist the limit then the objective function value equals zero. infeasible particle is denoted as the best particle although it is the best of all when only the fitness value is used by all particles. each set of M is represented by the total fuel cost of generation or emission for all the generators of that given population set. In case of EED problems it is calculated using (5) for the system having complex emission characteristic. fast searching velocity and global gradual convergence. If the number of iterations is larger than the maximum number of iterations (T). In the early stage of the algorithm. otherwise go to Step 3. Pi best. Pibest. Output the optimal particle. (2) Randomly initialize each of the ith (i = 1. (4) Rank the particles. The CST embedded in the PSO to improve the worse particles. Here the upper level’s decision variable is encoding and the lower level’s decision variable is computed according to the upper level’s decision variable. within the pre-specified range and velocity. The upper level and lower level’s decision variables are all randomly generated and updated by PSO or CST. i. (8) Output the results. The sequential steps of the HIA algorithm applied to solve EELD problem are as follows: In the case of EELD problem. Maximum velocity. Vi. in case of EELD problems. the feasible weighting value is used to rank the infeasible particles. The stepwise have to be taken: (1) At initial stage choose the parameters arbitrarily.…) and µ is the control parameter generated to replace that frog. In case of ELD problems. The chaos search technique has many advantages such as not sensitive to the initial value. For the ith particle. (8-9) and the n particles in the end of the list are updated using the CST with the initial chaos variable Xi. and two random variables. where m and n are the particles updated by PSO and CST respectedly.

Find the power output of that best set of the “Optimal Set”.9209 412.1716 413. Vmax.0000 150.4210 298.2761 523.96 199.1145 111.2739 489.1590 161366 .constraints.99 110.0000 200.7841 439. 12425 229310.29 394.22 110.8159 110.0000 119. (9). Average & Maximum Best Compromise 234 of 446 Copyright @ Department of Electrical Engineering.00 109. 57.4352 97.8035 164.3671 97.2794 394.5961 130.8085 140. Vi.4132 439. Step 4: Update the population set.6002 284. Store the best power outputs obtained as “Optimal Set”. otherwise repeat the steps 3 to 6.9934 511.2132 33. Step 6: If the new value of each individual is better than previous pbest.90 10.4388 280.1587 413.113 285823 .0000 10.2794 523.00 123638 0 356424.0000 120.4587 439.27 489..8072 140.8362 439.2780 394.8356 93.7598 304.Calculate the value of fuel cost of generation and emission for each solution sets.9989 189. Evaluate the absolute value of difference between FCPI and ECPI for each fuel cost of generation and emission set.000 139.5195 394.6006 130.003 10. for each fuel cost of generation and emission set.0002 10.0001 10.0011 214.9828 97.0000 190.8292 110.9083 412.2681 100 25.00 243.1677 413. 299.2794 523.8548 280. The minimum fuel cost. NUMERICAL RESULTS Proposed Hybrid Intelligent algorithm has been applied in a test system and its performance has been compared with modified bacterial foraging algorithm (MBFA) [12] for verifying its feasibility.1602 413.93 97.3624 199.0000 93. Step 5: Compute the fitness values of each population set. For the first m particles.8377 189.70 97.2688 523.8441 100.99 199.24 433.2794 489.0000 164.1650 413.5590 421.8026 164.1607 413.139 0 28.2796 523.7921 97.9945 109. Av g. Initialize the ith (i = 1.00029 87. Step 8) Increment the value of ‘w’ in step of 0.7697 440.0000 169.0000 94.56 433.55 10.7998 214.9149 0 100 0 6.1198 158.0000 110.64 130.19 124257 .7360 422.96 4 1 259. Xi.7333 87.0000 511.8080 194.0 000 Test System: This test system using 40 generators system having fuel cost function and the effects of valve-point loading on emission level functions are considered.2794 489.2939 10.4002 179.1219 28. of e hits to Simulat optimu ion m Avg.27 489. Calculate FCPI and ECPI using the equations (8-9).8856 280.27 511. Step 9) Best Compromising Solution: .61 284.6004 284.0000 158.013 96.2121 100 100 259.7394 394.4078 439.97 189.0000 116. Set its starting value as w=0. until the value of ‘w’ reaches to 1.4830 200.8074 114.) n 1242 2293 124257. Step 3: Calculate the objective function for each population set.8445 279.0004 94. If the best pbest is better than gbest. Then each pbest values are compared with previous pbest.2868 511.2790 523.61 243.54 433.00 110. 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 Cost ($/hr) Emission (ton/hr) TABLE 1: COMPARE WITH OTHER OPTIMIZATION METHODS: Optimal Generation Gen no 1 2 3 4 5 6 FCPI 0 ECPI 100 Differenc e 100 HIA Generation EcD [12] EmD [12] BCS of [12] EcD EmD BCS of PM 114..99 511.0000 114.2630 - 110.27 394.6931 297.0000 124.2729 412.4155 439.0000 110.0000 172.49 176682.4845 158. The calculated values of FCPI and ECPI with respect to minimum cost and emission results of HIA and MBFA are also presented in Table 1.3348 172.21 0. 229310.2794 511. The load demand is 10500MW.3327 172.2796 10. those are obtained for different values w and stored as “Optimal Set”.0000 97. MAXIMUM BEST COMPROMISE SOLUTION OBTAINED BY HIA OVER 50 TRIALS Total Cost ($/hr.2771 511.994 189. The fuel cost of generation and emission values associated with that set represent the best compromising fuel cost and emission.0000 298. 259.0000 168.2741 121414 .9194 413.2871 489. if current iteration is greater than or equal to the maximum iteration.) Methods Max.8346 100.3985 179.63 284. Minimum.4421 439.3868 129955.27 394.26 188963 7 9 A S E I VI.0264 433. M) population sets are randomly with its initial position. NIT Durgapur .60 433. The input data have been taken from [12].51 433.9955 109.8568 413.1619 413. the current value is set to be pbest.2375 280.9043 2132 2132 43 32 50 Sta nda rd Dev iati on 0.7822 439.2795 121415 .96 189.1191 440. the velocity of particle and its new position will be assigned by PSO according to Eq. After that it’s verify the feasibility of each newly generated population set of the modified with CST. in the range of maximum speed.0000 114.60 394. Averag No. 2. AVERAGE.9043 229310 . within the pre-specified range and velocity.4829 158. Min.7884 422.9093 297.653 HIA Min.2785 523.7330 87..2728 523.0000 200. the value is set to be gbest.2795 523.2795 511. Here the objective functions have been taken as fuel cost for ELD and Emission for EED. Total Emission (Ib/hr.0000 150.57 9043 7.8624 280.4150 0 TABLE 2: MINIMUM.0000 200.0002 89.2795 523.1732 150.2578 130.) Max.0000 110.2786 523.004 10.2794 523. Step 2: Initialize the value of w.9989 189.2794 394.1607 413.7028 100 0 100 19.489 179.97 199.0000 10.9948 109.8349 413. Use (1) and (5) to calculate fuel cost of generation and emission respectively for each set.8035 97. Step 7: Terminate the iterative process.0000 110. The set that attains minimum absolute value of difference between FCPI and ECPI is chosen as the best compromising solution.0000 190.05 and repeat the steps starting from step 3 to step 8.724 1 4.615 176682 . The best value among the p-bests is denoted as gbest.6214 284.9991 164.1625 413.0000 200.3262 200.0000 100.7999 168.6066 284.5070 190.22 511. Time solutio (Sec.0007 28.54 433.4111 439.8366 93. minimum NOX emission results and Comparisons of best compromising results obtained by HIA and MBFA [12] has been presented in Table 1. (7-8) and the n particles in the end of the list are updated using the CST with the initial chaos variable Xi followed the Eq.9738 29.

vol. pp.3.K. 1988. K.33. G.” in: Canadian Conference on Electrical and Computer Engineering. 2010. 1.) 1. Ghoshal. pp: 47-57. vol. 1. Telecommunications and Computer Science. P. 2013. Maranino. 2012. CONCLUSION [16] In this present work. Fig. no. S.65.) 4 x 10 5 [15] Fig. Dhillon. “A novel multiobjective evolutionary algorithm for environmental/economic power dispatch. Hota. Therefore. vol. Bhui.” Swarm and Evolutionary Computation. “Economic emission load dispatch through fuzzy based bacterial foraging algorithm.Aravindhababu. S.3 1. Khotari. A.10. P. Barisal.2149 [2] 1.A. Wang. Analyses of all the simulation results reveal that the performance of HIA in all respect is better in comparison with the previously developed several optimization techniques. 1.) x 10 REFERENCES 1. B.28 1.6 0 J. R. P. vol. 2012.36. S Özyön. pp55– 64. R Chakrabarti. Characteristics of Total Fuel cost obtained by HIA 3 x 10 5 [6] Emission (ton/hr.2 1. “Pareto genetic algorithm for multi-objective environmental/economic dispatch. pp:992–1002.S.” Int J Electr Power Energy Syst. A. 1.Solution values obtained by HIA over 50 trials are presented in Table 2. vol.R. pp.25. 2002. Pasini.5 2 2. D.” Energy. Abido. vol.” Electr Pow Syst Res.K.5 3 3. “Application of modified NSGA-II algorithm to combined economic and emission dispatch problem. pp:5838–5845.3. S.A. Nanda. “Multi-objective quasi-oppositional teaching learning based optimization for economic emission load dispatch problem. 2.C. pp26–32.P.P.Ghoshal. Roy. pp. vol 24. 2003.4. Dhanalakshmi. S.214 0 [4] 100 200 300 400 500 600 Iterations 700 800 900 1000 [5] Fig. MA Abido.K.” Int. Ottawa. 1993 M. “Emission constrained dynamic dispatch.K.32 x 10 5 Best compromising solution w=0.” in: Proceedings of the International Conference on Modern Problems of Radio Engineering. pp:937–948.P. July 1992 J. Kothari. 97–98. vol.” Electric Power System Research. Abou El Ela. J Elect Power Energy Syst. pp71–81.E. G. K. vol. A. pp179–186. no. A.) Emission = 4141. M.2 shows the Fuel cost and Emission cost curve respectively for HIA.5 Cost= 113181. “Economic emission dispatch through goal programming technique. S. no. R.” Electrical Power and Energy Systems.2011. “Differential evolution algorithm for emission constrained economic power dispatch problem. Yasar. All these results denote the robustness and supremacy of HIA.26 [10] [11] [12] [13] 1. no. Total Fuel Cost of Generation ($/hr.24 [14] 1. 2008. M.2146 [3] 1.8 2. “Solution of combined economic and emission dispatch problems of power systems by an opposition-based harmony search algorithm. vol. S. AlRashidi. Singh. Panda. “Chaos searching algorithm for second order oscillatory system. “Emission-economic dispatch using a novel constraint handling particle swarm optimization strategy.Rajasomashekar. NIT Durgapur .) Total Fuel Cost ($/hr. 2011. 664–669 7–10 May. A Niched.” Electrical Power and Energy Systems. L.6 [7] 2. no.P. 21-33.S. Mukherjee. Baskar. Mukherjee. 2010 .R. Spea. HIA can be considered as one of the strongest tools to solve complex EELD problems.L.7.2 [8] 2 [9] 100 200 Iterations 300 400 S E I Fig. vol. pp1286–1292. vol. 9-20. vol. C. El-Hawary. M. V. S. “Biogeography Based Optimization Technique for Best Compromise Solution of Economic Emission Dispatch.” Electr Power Syst Res. 39.5 Emission (ton/hr. “A new hybrid approach for nonconvex economic dispatch problem with valve-point effect.8 1. “Stochastic economic emission load dispatch. S.4 2. P. D.” Int J Electr Power Energy Syst. Characteristics of Emission obtained by HIA 1. April 2003. Zayats. Kannan. “A novel multi-objective directed bee colony optimization algorithm for multi-objective emission constrained economic power dispatch. “A novel opposition-based gravitational search algorithm for combined economic and emission dispatch problems of power systems.53.” IEEE Trans Energy Conver. Shaw.32. pp:794–803. Mahadevan. V.8. M. Canada.80. ON. vol. no.78. Granelli. pp:1241–1250.10. “Stochastic economic emission load dispatch through a modified particle swarm optimization algorithm. Kumar. Montagna. Parti. 2012. The 4 1 A 1. no. Trade-off curve obtained by HIA VII. 5 HIA achieves superior quality solutions near global solutions with high convergence speed and robustness in a much vigorous way compared to other methods.1 and Fig. pp. 3. 2006. [17] 235 of 446 Copyright @ Department of Electrical Engineering.” Electric Power Systems Research.2143 1. V. pp:79–105. vol.3013587636 ($/hr.) 2.” Electric Power System Research.Chatterjee.A. 2.” Int J Elect Power Energy Syst. S. Trade-off curve obtained by HIA for different values of w using objective function of (12) is shown in Fig. no. 35. Kumar.26.22 1. HIA method seems to become an important tool for solving more complex optimization problems.P. pp:1466–1476. C. S. In future.43.3521969296 (lb/hr.” Int J Electr Power Energy Syst. Abido. Sadu. 2012.2152 [1] 1. Lingamurthy. a newly developed Hybrid Intelligent Algorithm has been successfully implemented to solve different non-convex large scale EELD problems.

India Abstract—Due to the continuous increment of the load demand. In the recent literature review of voltage collapse revealed that the various analytical tools had been proposed to predict voltage collapse [5]-[6]. In [8]. [7]-[11]. the quantum genetic algorithm (QGA) was suggested for the optimal placing and sizing of the DG. In [10].T Durgapur Durgapur. Because of the increasing importance of the stability. The multi-objective PSO is used to find the optimal location and size of DG and SVC. reduction of line loss and improvement of voltage stability. The suitable location for the reactive power compensating device for improving the voltage profile and steady state voltage stability of the system is the weak bus [5]. improvement of voltage profile and power losses in the context of the voltage stability problems become one of the major concerns for the larger. The power quality related problems can be alleviated to some extent by using FACTS devices such as SVC. The integration of DG into the system reduces overall energy loss and improves supply quality. power losses in the context of steady state stability considering weak buses of the network. Remarkable improvement in the system performance is observed by integrating SVC and DG in the test systems like IEEE 30 interconnected and 12 bus radial systems. Several techniques were suggested for th