Professional Documents
Culture Documents
Alloy
by
February, 2013
Seismic Vibration Control of Frame Structure Using Shape Memory
Alloy
by
February, 2013
The thesis titled “Seismic Vibration Control of Frame Structure Using Shape
Memory Alloy” submitted by Md. Golam Rashed, Roll No.: 0411042321, Session:
April 2011 has been accepted as satisfactory in partial fulfilment of the requirement
for the degree of M.Sc. Engineering (Civil and Structural) on 25th February, 2013.
BOARD OF EXAMINERS
______________________________________
Dr. Raquib Ahsan Chairman
Professor (Supervisor)
Department of Civil Engineering
BUET, Dhaka.
______________________________________
Dr. Md. Mujibur Rahman Member
Professor and Head (Ex-officio)
Department of Civil Engineering
BUET, Dhaka.
______________________________________
Dr. Tahsin Reza Hossain Member
Professor
Department of Civil Engineering
BUET, Dhaka.
______________________________________
Dr. Sharmin Reza Chowdhury Member
Associate Professor (External)
Department of Civil Engineering
AUST, Dhaka.
ii
DECLARATION
It is hereby declared that this thesis or any part of it has not been submitted elsewhere
for the award of any degree or diploma.
_________________________________
Md. Golam Rashed
iii
DEDICATION
iv
ACKNOWLEDGEMENT
At first I would like to express my whole hearted gratitude to Almighty Allah for each
and every achievement of my life.
I would like to express my great respect and gratitude to my thesis supervisor, Dr.
Raquib Ahsan, Professor, Department of Civil Engineering, BUET for providing me
continuous support and guideline to perform this research work and to prepare this
concerted dissertation. His contribution to me can only be acknowledged but never be
compensated. His consistent inspiration helped me to work diligently throughout the
completion of this research work and also contributed to my ability to approach and
solve a problem. It was not easy to complete this work successfully without his
invaluable suggestions and continuous help and encouragement. Despite many
difficulties and limitations he tried his best to support the author in every field related
to this study.
I would like to render sincere gratitude to Dr. Toby Kim Parnell, USA and Dr. Furo
Jumbo, UK for providing useful knowledge on SMA simulation and advanced FEA. I
am grateful to Dr. Rafiqul A. Tarefder, UNM, USA for providing the experimental
test data and to Dr. Mehedi Ahmed Ansary, BUET, for providing required
computational facilities.
I would like to convey my gratefulness and thanks to my family, their undying love,
encouragement and support throughout my life and education. Without their
blessings, achieving this goal would have been impossible.
At last I would like to thank my respected supervisor Dr. Raquib Ahsan once again
for giving me such an opportunity, which has obviously enhanced my knowledge and
skills as a structural engineer to a great extent.
v
ABSTRACT
The use of Shape Memory Alloy (SMA) in mitigating the seismic vibration response
of civil infrastructure is gaining momentum. The name “Shape Memory” implies that
it remembers its original formed shape. SMA has two basic properties, Super-
Elasticity and Shape Memory Effect (SME). The “Super-Elastic” behaviour exhibited
by SMA materials, permits a full recovery of strains up to 8% from large cyclic
deformations, while developing a hysteretic loop. SME allows the material to recover
the initial shape or position which in turn can be used as re-centering mechanism. The
mechanism of shape recovery involves two crystallographic phases, Martensite and
Austenite, and the transformations between them. The Austenite phase provides more
stiffness than the Martensite phase. Phase transformation occurs between Martensite
& Austenite depending upon temperature & stress. These unique properties result in
high damping, combined with repeatable re-centering capabilities which can be used
to advantage in several civil infrastructure applications, especially in seismic vibration
control devices.
Super-Elastic response of SMA has historically been the primary mode of interest of
civil engineers as it occurs over a wide-range of temperatures; and also because SMA
reaches activation temperature and becomes Austenite at the ambient temperature of
civil engineering infrastructures. Thus the re-centering capability of SMA by
generation of an activation force is not utilized. The use of high temperature SMA has
enabled the re-centering mechanism to work. The SMA is heated by electrical current
flow and the use of constant current in this purpose will result in greater power
consumption which can be reduced significantly by passing pulsed current through
the SMA using Pulse Width Modulation (PWM) technique.
In this study, both the Super-Elastic and Shape Memory Effect has been taken into
account by using SMA with high activation temperature. A Thermo-Mechanical SMA
phenomenological constitutive model is used to simulate the SMA behaviour. The
dynamic response data of a frame structure has been obtained from FE analysis by
using the nonlinear FE software program MSC Marc. Then the frame is braced and
reanalyzed; first using standard steel wire and then later using SMA wire, the seismic
response of both the braced frames were measured. The SMA bracing is activated by
joule-heating due to electrical current flow. The SMA is first activated by constant
current, later using pulsed current. In this research work, From the FE solutions, the
effectiveness of SMA braces as a seismic vibration control device and guidelines to
optimum electrical input, considering appropriate stiffness and damping
characteristics; is established. From the simulation result, it is evident that the use of
pulsed current resulted in reduced energy consumption by the SMA, as well as
mitigating the seismic vibrations on the frame structure.
vi
CONTENTS
Page
No.
DECLARATION iii
DEDICATION iv
ACKNOWLEDGEMENT v
ABSTRACT vi
LIST OF FIGURES x
NOTATIONS xviii
CHAPTER 1 INTRODUCTION
1.1 General 19
2.2.2 Pseudo-Elasticity 27
vii
2.3.1 Phenomenological Modeling 29
2.5 Limitations 33
3.4 Implementation 45
CHAPTER 4 VERIFICATION
4.1 General 49
4.5 Verification 58
viii
CHAPTER 5 RESULTS AND DISCUSSIONS
5.1 General 64
REFERENCES 106
Appendix - A 112
ix
LIST OF FI GU RE S
Page No.
Figure 2.9 (a) Schematic of the SMA isolation system for buildings 31
x
Figure 2.10 (b) Schematic of the SMA damper for a stay-cable bridge 32
Figure 2.10 (c) Schematic of the SMA isolation device for elevated 32
highway bridges
Figure 3.1 Austenite to Martensite and Martensite to Austenite 37
Decomposition
Figure 3.2 Thermal History 39
xi
Figure 4.10 Acceleration at the top from simulation with 5% 60
damping in frame
Figure 4.11 Acceleration at the top from simulation with 10% 61
damping in frame
Figure 4.12 Acceleration at the top from experimental data 61
Figure 4.13 Acceleration at the top from simulation with no damping 62
in frame
Figure 4.14 Acceleration at the top from simulation with 5% 62
damping in frame
Figure 4.15 Acceleration at the top from simulation with 10% 63
damping in frame
Figure 5.1 Acceleration time history of El Centro earthquake 65
xii
Figure 5.16 Pulsed current input for 1A current 74
xiii
Figure 5.36 Predicted displacement at the top of the frame for 2.2A 84
current
Figure 5.37 Predicted maximum temperature history of the SMA 85
brace for 2.3A current
Figure 5.38 Predicted displacement at the top of the frame for 2.3A 85
current
Figure 5.39 Predicted maximum temperature history of the SMA 86
brace for 2.4A current
Figure 5.40 Predicted displacement at the top of the frame for 2.4A 86
current
Figure 5.41 Optimum constant current 87
xiv
Figure 5.56 Predicted maximum temperature history of the SMA 95
brace for 4.8A current
Figure 5.57 Predicted displacement at the top of the frame for 4.8A 96
current
Figure 5.58 Predicted maximum temperature history of the SMA 96
brace for 4.9A current
Figure 5.59 Predicted displacement at the top of the frame for 4.9A 97
current
Figure 5.60 Predicted maximum temperature history of the SMA 97
brace for 5.0A current
Figure 5.61 Predicted displacement at the top of the frame for 5.0A 98
current
Figure 5.62 Predicted maximum temperature history of the SMA 98
brace for 5.1A current
Figure 5.63 Predicted displacement at the top of the frame for 5.1A 99
current
Figure 5.64 Predicted maximum temperature history of the SMA 99
brace for 5.2A current
Figure 5.65 Predicted displacement at the top of the frame for 5.2A 100
current
Figure 5.66 Predicted maximum temperature history of the SMA 100
brace for 5.3A current
Figure 5.67 Predicted displacement at the top of the frame for 5.3A 101
current
Figure 5.68 Predicted maximum temperature history of the SMA 101
brace for 5.4A current
Figure 5.69 Predicted displacement at the top of the frame for 5.4A 102
current
Figure 5.70 Optimum pulsed current 102
xv
LIST OF TABLES
Page No.
xvi
LIST OF ABBREVIATION
FE Finite Element
NiTi Nickel-Titanium
PE Pseudo-Elasticity
xvii
NOTATIONS
𝐴𝑓 Austenite Finish temperature
𝐴𝑠 Austenite Start temperature
𝑀𝑓 Martensite Finish temperature
𝑀𝑠 Martensite Start temperature
𝜎𝑒𝑞 Von Mises equivalent stress
𝐶𝑚 Austenite to Martensite slope
𝐶𝑎 Martensite to Austenite slope
∆𝜺𝐸𝑙 Elastic strain
∆𝜺𝑇 Thermal strain
∆𝜺𝑃 Phase transformation strain
∆𝜺𝑃𝑙 Plastic strain
∆𝜺𝑇𝑟𝑖𝑝 Trip strain
∆𝜺𝑇𝑤𝑖𝑛 Twin strain
∆𝑓 Volume fraction Martensite
∆𝜆 Equivalent plastic strain increment
𝛼 Hardening coefficient
𝑲 Bulk modulus
𝜗 Poisson’s ratio of the aggregate
𝛼𝑎 Coefficient of thermal expansion for the Austenite
𝛼𝑚 Coefficient of thermal expansion for the Martensite
𝜇 Shear modulus
𝑇
𝜀𝑒𝑞 Deviatoric part of transformation strain
𝜀𝑣𝑇 Volumetric part of the transformation strain
I Identity tensor
𝑇 Temperature
𝑔 G function
xviii
CHAPTER 1
INTRODUCTION
1.1 General
In recent times, several researchers have explored the application possibility of SMA
in civil infrastructures. Fugazza (2003) proposed a uniaxial constitutive model for
Superelastic SMA’s that are candidates for installation into civil engineering
structures. Auricchio (1995) developed one and three dimensional thermo-mechanical
constitutive models for SMA materials. Both of these research studies included
numerical simulations to show the ability of their models and exemplify the dynamic
behavior of SMA’s. More recently, Penar (2005) studied NiTi shape memory material
with the goal of developing a re-centering beam-column connection for steel frames.
Also, McCormick (2006) investigated cyclical properties of large diameter shape
memory alloys for structural applications.
20
Despite the fact that recent studies have contributed significantly to comprehension of
the potential for using shape memory alloys in civil engineering applications, there
are still many questions to be answered before full-scale applications can be
implemented. To this end, this thesis adopts a thermo-mechanical model fully capable
of simulating the Super elasticity and shape memory effect. A framework has been
suggested on how to reduce the seismic vibration using high temperature SMA wire
as bracing system.
Shape memory alloys (SMA) are widely used in different disciplines and it has
substantial potential for civil engineering applications (Alam et al., 2007, Song et al.,
2006, Janke et al., 2005). The name shape memory implies that it remembers its
original formed shape. The super-elastic behavior exhibited by SMA’s help material
to totally recover from large cyclic deformations, while developing a hysteretic loop
(Lagoudas, 2008). The mechanism of shape regaining works in two phases,
Martensite and Austenite. Austenite phase provides more stiffness than that of
Martensite and civil engineers can leverage the variation in stiffness depending upon
temperature and stress. These unique properties result in high damping, combined
with repeatable re-centering capabilities which can be used in civil infrastructures,
especially in seismic vibration control devices (Saadat et al., 2002).
Super-elasticity of SMA is primary interest of civil engineers till now. As the SMA
reaches activation temperature at the ambient temperature of civil engineering
infrastructures, shape memory effect does not take place. Thus the re-centering
capability of SMA is left unused (Ozbulut et al., 2011). Both the super-elasticity and
shape memory effect can be taken into account by using SMA of high activation
temperature (Churchill & Shaw, 2008). Proper numerical simulation of SMA can be
done by taking both the super-elasticity & shape memory effect into consideration
(Saeedvafa & Asaro, 1995).
To overview the state of the art of Shape Memory Alloy (SMA) usage in civil
engineering infrastructures.
To study the available constitutive theories and suitable numerical
implementation of Shape Memory Alloy (SMA).
To simulate the behavior of frame structure with Shape Memory Alloy (SMA)
considering earthquake oscillations.
Shake table vibration response data of a frame structure has been obtained from
literature where the response of the frame structure subjected to seismic vibration on
shake table was measured. The frame structure was braced using Shape Memory
Alloy (SMA) and later again using standard steel. The response of both the braced
frames was measured accordingly. The SMA bracing was activated by joule-heating
due to electrical current flow. In this study, a coupled Thermo-Mechanical SMA
phenomenological constitutive model will be used to simulate the SMA response. For
this purpose a program, MSC Marc based on nonlinear Finite Element Method (FEM)
will be used to analyze the braced frames. The FEM solutions will be compared with
the experimental data and thus the Thermo-Mechanical SMA phenomenological
model used in numerical simulation will be verified. Further guidelines to optimum
electrical input considering appropriate stiffness and damping characteristics will be
established, also the effectiveness of SMA braces as a seismic vibration control
device will be studied from numerical simulation data.
Apart from this chapter, the remainder of the thesis has been divided into five
chapters.
22
Chapter 4 presents the experimental test setup with results obtained from literature
and verification of the presented SMA phenomenological constitutive model.
Finally, Chapter 6 presents the major conclusions of the study and also provides
recommendations for future study.
CHAPTER 2
LITERATURE REVIEW
2.1 General
The name “shape memory” implies that it remembers its original formed shape. The
super-elastic behavior exhibited by shape-memory alloys help material to totally
recover from large cyclic deformations, while developing a hysteretic loop. Due to its
hysteretic behavior and excellent re-centering capability, SMA can be used in a wide
variety of civil engineering applications. The other key features of SMA’s include
high strength, good fatigue and corrosion resistance, high damping capacity,
temperature-dependent Young’s modulus, ability to undergo large deformations, and
availability in many possible shapes and configurations.
Figure 2.1: SMA phases: (a) Austenite; (b) twinned Martensite; and (c) detwinned or
deformed Martensite (Key to Metals AG, 2008).
SMA’s have two main phases which have different crystal structures. One is called
Martensite that is stable at low temperatures and/or high stresses and the other
Austenite, which is stable at high temperatures and/or low stresses. Austenite, also
named as the parent phase, generally has a cubic crystal structure while Martensite
has a less-ordered crystal structure (Song et al., 2006). The un-deformed Martensite
phase is the same size and shape as the cubic austenite phase on a macroscopic scale,
so that no change in size or shape is visible in shape memory alloys until the
24
Martensite is deformed. Martensite also has two forms that are termed twinned and
detwinned (Figure 2.1). SMAs owe their peculiar characteristics to the solid-to-solid
transformations between these two phases. Austenite phase provides more stiffness
than that of Martensite and civil engineers can leverage the variation in stiffness
depending upon temperature & stress.
Table 2.1: Typical properties of NiTi SMA with comparison to Structural steel
(Santos, 2011).
Over the past two decades, SMA’s have been widely investigated for their possible
application in civil engineering structures (Rashed et al., 2012). This chapter tries to
cover some latest applications of SMA’s in civil engineering along with established
examples while also focusing on the limitations in application.
25
SMA’s have two unique properties, The Shape Memory Effect (SME) which is the
phenomenon that the material returns back to their original shape upon heating and
the Pseudo-Elasticity (PE) which is the phenomenon that the material can undergo a
large amount of inelastic deformation and recover after unloading. These properties
are the result of reversible phase transformations between the Austenite phase and the
Martensite phase.
Figure 2.3: Schematic of the shape memory effect of an SMA showing the detwinning
of the material with an applied stress (Lagoudas, 2008).
Figure 2.4: Schematic of the shape memory effect of an SMA showing the unloading
and subsequent heating to austenite under no load condition (Lagoudas, 2008).
Shape memory effect (SME) is a unique characteristic of SMA that exhibits thermo-
elastic Martensitic phase transformation. It is the ability of SMA material to recover
its original shape after being deformed through a thermal cycling. Through training,
the material has the ability to memorize a very specific physical configuration or
shape in either the Martensite or Austenite phase, which is called one-way shape
memory (Figure 2.6a). Also it is possible to train the material, such that it memorizes
two different configurations or shapes in Martensite and Austenite phases, which is
called two-way shape memory.
The key to the SME is the build-up of residual stress fields within the SMA, by
deforming the material plastically, and then these stress fields control the phase
transformation (Saadat et al., 2002).
2.2.2 Pseudo-Elasticity
Also, if the temperature in the Austenite phase exceeds the maximum temperature at
which Martensite occurs, Md, the material is stabilized in the Austenite phase and the
Martensitic transformations cannot be induced by an applied load, thus the PE of
SMA is completely lost (Figure 2.6c).
SMA used for damping can be both Martensitic as well as Austenitic. The damping
comes from either Martensite variations reorientation in the Martensitic material or
from stress-induced Martensite in Austenitic material. When an SMA specimen is
subjected to a cycle of deformation within its Superelastic strain range, it dissipates a
certain amount of energy without permanent deformation (Figure 2.7a). This results
from the phase transformation from Austenite to Martensite during loading and the
reverse transformation during unloading, ensuring a net release of energy. When an
SMA is loaded in the Martensite phase, it yields at a nearly constant stress after initial
elastic deformation and displays strain hardening at larger strains. When unloaded,
there remains some residual strain at zero stress. This Martensitic composition of
SMA’s generates a full hysteresis loop around the origin (Figure 2.7b). Thus,
Martensite SMA dissipates a much higher amount of energy compared with that of
Austenite SMA because of its larger hysteresis loop. But it has no re-centering
capability like the Austenitic SMA. In the Martensite phase under tension–
compression cycles, the maximum stress attained in compression has been found to be
approximately twice that in tension (Figure 2.7b). Although Superelastic SMA
dissipates less energy than Martensitic SMA, its advantage is that it can still dissipate
a considerable amount of energy under repeated load cycles with negligible residual
strain (Alam et al., 2007).
(a) (b)
Figure 2.7: Typical stress–strain curve of SMA under cyclic axial stresses: (a)
Superelastic SMA; (b) Martensite SMA; (Alam et al., 2007).
29
The modeling of SMA behavior such as SME and PE has been an active area of
research. The modeling approaches can be categorized into phenomenological and
thermo-dynamical approaches.
Several studies have considered the use of SMA’s as diagonal braces in frame
structures (Saadat et al., 2001; Tarefder et al., 2006). The frame structures deform
under excitation, SMA braces dissipate energy through stress-induced Martensite
transformation (in the Superelastic SMA case) or Martensite reorientation (in the
Martensite SMA case) as shown in Figure 2.8a. Auricchio et al. (2006) investigated
the effectiveness of using large diameter NiTi bars as a bracing system for steel
structures and compared the SMA braces with buckling-restrained steel braces. The
outcome of numerical studies showed that SMA bracing systems can satisfactorily
limit the inter-story drifts in steel buildings and significantly reduce the residual drifts
(McCormick et al., 2007).
Figure 2.8: (a) SMA braced frame (Tarefder et al., 2006); (b) Schematic of SMA
connector for steel structures (Song et al., 2006); (c) Steel beam-column connection
using SMA tendons (Qian et al., 2010).
Several studies have been conducted on SMA beam-column connectors (Qian et al.,
2010). SMA connectors have been designed to provide damping and tolerate
relatively large deformations and found to be most effective in controlling structural
response under high levels of seismic intensity. Ocel et al. (2004) experimentally
31
evaluated the performance of partially restrained steel beam-column (Figure 2.8b &
2.8c) connections using Martensitic SMA’s. It was observed that the SMA
connections were able to recover 76% of the beam tip displacement.
Figure 2.9: (a) Schematic of the SMA isolation system for buildings (Qian et al.,
2010); (b) Schematic of the SMA spring isolation device (Song et al., 2006); (c)
Schematic of a bell tower using SMA anchorage retrofitting (Song et al., 2006).
Several studies have been made on SMA based isolation devices for seismic
protection of building structures by performing shake table tests (Song et al., 2006).
But majority of them are Superelastic SMA due to its zero residual strain after
unloading. Martensite SMA’s can be used to help dissipate more energy and further
improving the damping effect of the Superelastic SMA isolation devices. The re-
centering device by Dolce et al. (2001) is a good example of combining the
Superelastic and Martensitic SMA’s (Figure 2.9a).
SMA’s have been used to retrofit existing or damaged structures (Islam et al., 2012).
The San Giorgio Church Bell-Tower, which was damaged in 1996 by earthquake, was
retrofitted using SMA tie bars. As shown in Figure 2.9c, the SMA tie bars, which run
through the height of the tower and are anchored at its foundation, reinforce the
structure and increase its modal frequencies. That tower stood intact after a similar
earthquake in 2000 (Qian et al., 2010).
32
Figure 2.10: (a) Schematic of the setup of SMA restrainer for a simple-supported
bridge; (b) Schematic of the SMA damper for a stay-cable bridge; (c) Schematic of
the SMA isolation device for elevated highway bridges; (Song et al., 2006).
Several studies have been carried out to investigate the possibility of using SMA as
unseating prevention devices on multiple span bridges to overcome some of the
limitations of traditional devices such as steel cable restrainers, steel rods, and shock
transmission units (Qian et al., 2010). The schematic of SMA restrainer setup for
simple supported bridge is shown in Figure 2.10a.
Both Superelastic and Martensite SMA’s can be used as damper elements for bridges.
Li et al. (2004) theoretically studied the vibration mitigation of a combined cable-
SMA damper system which can be used on a stay-cable bridge (shown in Figure
2.10b). The dynamic responses of the SMA damped cable were simulated as it
vibrated at its first mode or at its first few modes respectively. They stated that the
proposed Superelastic SMA damper can suppress the cable’s vibration in both cases
(Song et al., 2006).
For highway bridges, Comparative simulations of the SMA isolation system (shown
in Figure 2.10c) and a conventional isolation system were conducted with three
excitation levels. For small excitation level, the SMA isolation system firmly links the
pier and the deck, while the relative motion emerges in the case of the conventional
system. For a medium excitation level, the SMA bar undergoes a stress-induced
Martensitic transformation so that the soft stiffness allows a relative displacement
comparable to that of the conventional isolation system. At severe loading, the SMA
bar enters an elastic range of Martensite and the maximum displacement is one-fifth
as much as that of the conventional isolation system. The comparison shows that the
33
damage energy of the bridge with the SMA isolation system is smaller than with the
conventional system (Song et al., 2006).
2.5 Limitations
The price of the SMA’s is high in comparison to the conventional civil engineering
construction materials. However, a significant reduction in the price of SMA has
occurred over the last decade, recently developed Iron based SMA's are quite cheaper
than traditional NiTi, sometimes by ten folds (Alam et al., 2007). Fe-Mn-Si-X alloys
are an example of a potentially low cost SMA.
SMA’s can be heated by using electric current for actuation. But short activation
times in the range of seconds are not possible for large cross sections. A high capacity
power supply with a current of several hundred Ampere can reduce activation times
considerably. This may incur greater costs when setting up the actuator and keeping
up the high temperature state for long time (Janke et al., 2005).
This chapter presents a review of the basic properties of SMA such as the SME and
the PE. The ability to change shape by application of heat can be used to move
objects. If recovery is resisted, the SMA generates force that is useful as an actuator
which can be activated by electrical Joule heating. Thus many applications are
possible. Numerous analytical and experimental studies point toward the feasibility
and superiority of SMA based devices over conventional methods for seismic
protection. The main characteristics of such devices are high energy dissipation and
re-centering capabilities.
This chapter also reviews the constitutive models developed to carry out numerical
simulation of SMA devices. These constitutive characteristics of SMA’s greatly
depend on variations in the alloy’s component and manufacturing method to begin
with. From discussion it is observed that phenomenological models are more adequate
34
for civil engineering applications because they are simple and easy to incorporate in
finite element programs and are not computationally demanding.
The high price of SMA is a burden for their use as construction materials. Hopefully,
with the advent of new manufacturing and processing techniques, the price of SMA is
reducing over the years.
CHAPTER 3
CONSTITUTIVE MODEL
3.1 General
The behavior of SMA’s shows a high level of complexity, since it depends on stress
and temperature, and is closely connected with the crystallographic phase of the
material and the thermodynamics underlying the transformation processes. For these
complications, any simulation taking advantage of the peculiar properties of the SMA
has to be based on a proper constitutive model. In this chapter, modeling of SMA by
various researchers has been discussed, followed by describing a thermo-mechanical
SMA phenomenological model. Finally, the implementation of the SMA model in the
non-linear FE program MSC Marc is discussed.
The modeling of SMA behavior such as SME and PE has been an active area of
research for the last two decades. The modeling approaches can be categorized into
phenomenological and thermo-dynamical approaches.
Liang & Rogers (1990) reviewed the early phenomenological theories. They
presented a one-dimensional model and does not specifically account for the dual
effects of stress induced transformation and Martensite reorientation. The model
presented by Barrett & Sullivan (1995) have generalized one-dimensional constitutive
description of the shape memory phenomena, using a mixture of phenomenological
relations to describe the effects of multi-axial states of stress on the evolution of
Martensite volume fraction. But it does not lead to a full set of integrable constitutive
equations that may be used for numerical simulation.
The Brinson’s (1993) model was used by Brinson & Lammering (1993) in 3D version
to perform FEM simulation. The limitations of this model are, it describes the
transformation-induced plasticity as a form of non-linear elastic behavior, and it does
not distinguish between deformation caused by the transformation itself and
reorientation events.
36
Based on the above discussion, while there has been considerable attention paid to the
constitutive modeling of shape memory alloys, no model yet exists that accounts for
the various phenomena of interest to properly simulate and at the same time is simple
enough form to be practicably implementable (Saeedvafa, 2001).
The curves indicate that upon cooling, the material transformation from Austenite to
Martensite begins once the temperature is reached. Upon further cooling, the volume
fraction on Martensite is a given function of temperature; the volume fraction
becomes 100% Martensite when the temperature is reached. Upon heating,
transformation from Martensite to Austenite begins only after 𝐴𝑠 temperature is
reached. This re-transformation is complete when the 𝐴𝑓 temperature is reached.
Finally, note that the four transformation temperatures are stress dependent. The
38
𝜎
𝑀𝑠 = 𝑀𝑠0 + 𝐶𝑒𝑞
𝑚
𝜎
𝑀𝑓 = 𝑀𝑓0 + 𝐶𝑒𝑞 , and
𝑚
𝜎𝑒𝑞
𝐴𝑠 = 𝐴0𝑠 + 𝐶𝑎
𝜎𝑒𝑞
𝐴𝑓 = 𝐴𝑓0 + 𝐶𝑎
Where 𝜎𝑒𝑞 is the Von Mises equivalent stress. At a sufficiently high temperature,
often called the 𝑀𝑑 temperature, transformation to Martensite does not occur at any
level of stress. The transformation characteristics such as the transformation
temperatures depend sensitively on alloy composition and heat treatment.
For the discussion of the thermo-mechanical response of NiTi, the data of (Miyazaki
et al., 1981) is shown in Figure 3.2. Following this thermal history, it is observed that,
when unstrained specimens with fully austenitic microstructures are cooled, the
transformation to Martensite begins at a temperature of 190K; the transformation is
complete at 128K. This established the so-called Martensite start (𝑀𝑠0 ) and Martensite
finish (𝑀𝑓0 ) temperatures at 190K and 128K, respectively. With the imposition of an
applied uniaxial tensile stress, the low temperature Martensite is favored and the 𝑀𝑠0
and 𝑀𝑓0 temperatures increase. Upon heating a specimen with fully martensitic
microstructure, the reverse transformation is observed to begin at a temperature of
188K and to be complete at 221K. These define the Austenite start (𝐴0𝑠 ) and Austenite
finish 𝐴𝑓0 temperatures, respectively. Uniaxial tension tests are carried out in
temperature ranges where T<𝑀𝑠 , 𝑀𝑠 < 𝑇 < 𝐴𝑓 , and 𝐴𝑓 < 𝑇 < 𝑇𝑐 where 𝑇𝑐 is defined
as the temperature above which the yield strength of the austenitic phase is lower than
the stress required to induce the Austenite-to-Martensite transformation.
In the temperature range where 𝑇 < 𝑀𝑓 , the microstructures are all Martensitic. The
stress versus strain curves display a smooth parabolic type of behavior which is
consistent with deformation caused by the movement of defects such as twin
39
boundaries and the boundaries between variants. Note that unloading occurs nearly
elastically and that the accumulated deformation, caused by the reorientation of the
existing Martensite and the transformation of any pre-existing Austenite, remains
after the specimen is completely unloaded. Note also that the accumulated
deformation is entirely due to oriented Martensite and this would be recoverable upon
heating to temperatures above the (𝐴𝑠 − 𝐴𝑓) range. This would, then, display the
shape memory effect.
As expected, the stress levels rise with increasing temperature. In this range, the
transformation induced deformation is nearly all reversible upon unloading. At
temperatures where 𝑇 > 𝑇𝑐 , plastic deformation appears to precede the formation of
stress induced Martensite. The unloading part of the stress versus strain behavior
displays nonlinearity and the unloading is now associated with permanent (plastic)
deformation. Permanent deformation, due to plastic deformation of the Austenite, is
Non-recoverable and, if such deformation is large, shape memory behavior is lost.
In Eq. (1), the superscripts, “El”, “Th”, “Ph” and “Pl” mean Elastic, Thermal, Phase
transformation and Plastic, respectively.
The elastic strain is taken to be simply related to as set of elastic modulus, L, and the
Cauchy stress as:
Also, the thermal strain increment is related to a volume average thermal expansion
coefficient, α through
In the work, it is assumed that the phase transformation strain is composed of trip
strain and twin strain as follows:
In Eq. (4), TRIP strain is the deformation by the formation of oriented stress-induced
Martensite and TWIN strain is the deformation by the reorientation of randomly
oriented thermally induced Martensite.
41
For TRIP strain formulation, it is assumed that the extent to which Martensite forms
with preferred variants depends on the intensity of the deviatoric stress state. In
addition, under the assumption of isotropy, it is assumed that the deviatoric part of
∆𝜺𝑃 is co-axial with the deviatoric stress (Saeedvafa, 2001).
2 𝜎
∆𝜺𝑇𝑟𝑖𝑝 = ∆𝑓 𝑀 𝑔 𝜎𝑒𝑞 𝜀𝑒𝑞
𝑇
+ ∆𝑓 𝑀 𝜀𝑣𝑇 𝐼 + ∆𝑓 𝐴 𝜀 𝑃 (5)
3𝜎 𝑒𝑞
In Eq. (5), ∆𝑓 𝑀 ≥ 0 represents the increment at which Martensite is formed. Also, 𝜀𝑒𝑞
𝑇
is deviatoric part of transformation in uniaxial tension and 𝜀𝑣𝑇 is the volumetric part of
the transformation strain. In addition, 𝜎𝑒𝑞 is the effective stress defined with Von-
3
Mises yield function as 𝜎𝑒𝑞 = 𝜎: 𝜎 and I is the identity tensor. The function
2
2 𝜎 𝑔
∆𝜺𝑇𝑤𝑖𝑛 = 𝑓∆𝑔 𝜎𝑒𝑞 𝜀𝑒𝑞
𝑇
∆𝜎𝑒𝑞 {𝜎𝑒𝑞 − 𝜎𝑒𝑓𝑓 } (6)
3 𝜎𝑒𝑞
In Eq. (6) the modified form of McCauley’s bracket is used. We also note that there is
no dilatational contribution to TWIN strain and the twinning strain increment is zero
𝑔
when 𝜎𝑒𝑞 is less than 𝜎𝑒𝑓𝑓 or when the magnitude of the stress decreases (𝜎𝑒𝑞 < 0). It
is also often observed that there exists a threshold Equivalent stress level below which
𝑔
de-twinning does not occur; this stress is referred as 𝜎𝑒𝑓𝑓 . The function 𝑔 in Eq. (5)
42
and (6) has the range of 0 ≤ 𝑔 𝜎𝑒𝑞 ≤ 1. Furthermore, the value of 𝑔 at the stress of
𝑔 𝑔
𝜎𝑒𝑓𝑓 is 𝑔 𝜎𝑒𝑓𝑓 = 0. Also, in practice the function 𝑔 tends to approach unity at a
𝑔 𝑔
finite equivalent stress level, called 𝜎0 . Thus 𝑔(𝜎0 ) = 1, and the trial function is
chosen to be of the form:
𝑔 𝑛
𝑔 𝜎𝑒𝑞 = 1 − 𝑒𝑥𝑝𝑎 𝜎𝑒𝑞 /𝜎0 (7)
With a < 0. Figure 1 shows the plot of a = - 4.0 and n =2.0 under simple uniaxial
tension and f = 1.0. Then, Eq. (7) integrates to 𝜀 𝑇𝑤𝑖𝑛 = 𝑔 𝜎𝑒𝑞 𝜀 𝑇𝑒𝑞 .
A better fit is achieved by replacing the term in the bracket with sum of polynomial of
stress and a general form of the function 𝑔 has been implemented in the present work:
𝜎 𝜎 𝜎
𝑔 𝜎𝑒𝑞 = 1 − 𝑒𝑥𝑝 𝑔𝑎 ( 𝜎𝑒𝑞𝑔 )𝑔 𝑏 + 𝑔𝑐 ( 𝜎𝑒𝑞𝑔 )𝑔 𝑑 + 𝑔𝑒 ( 𝜎𝑒𝑞𝑔 )𝑔 𝑓 (8)
0 0 0
The plastic strain increment of the two-phased aggregate as well as that of each phase
is assumed to be governed by J2 -flow theory, that is
3 𝜎
∆𝜀 𝑃𝑙 = ∆𝜆 2 𝜎 (9)
𝑒𝑞
43
Where ∆𝜆 is the equivalent plastic strain increment. For the case to use von-Mises
𝜎𝛼
∆𝜆𝛼 = ∆𝜆𝛼0 ( 𝜎𝑒𝑞𝛼 )𝑚 (10)
𝑌
Where 𝜎𝑌𝛼 is the flow stress of the 𝛼 − 𝑡 phase and can be described by 𝜎𝑌𝛼 =
𝜎𝑌𝛼 𝑇, 𝜎𝑒𝑞 , 𝜆 … . Also 𝛼 means 𝑀 (Martensite) and 𝐴 (Austenite). The current value
of 𝜎𝑌𝛼 for each phase can be calculated from
5𝛤
∆𝜆𝛼 = 3𝛤+2𝛤 𝛼 ∆𝜆 (12a)
5𝛤 𝛼
𝜎 𝛼 = 3𝛤+2𝛤 𝛼 𝜎 (12b)
𝛼
Let 𝑥 𝛼 = ∆𝜆𝛼 /∆𝜆, = 𝜎𝑌𝛼 /𝜎𝑌 , then by rearranging the above two equations using
Eq. (10), the following equations are obtained as
1
𝑥𝛼 = 5 3 − 2 3 𝛼
(𝑥 𝛼 ) 𝑚 (13)
1= 𝛼 𝑥 𝛼 𝑓𝛼 (14)
Eq. (13) and Eq. (14) can be expressed as the equivalent form as follows:
1
𝑥𝐴 = 5 3 − 2 3 𝐴
(𝑥 𝐴 ) 𝑚 (15a)
44
1
𝑥𝑀 = 5 3 − 2 3 𝑀
(𝑥 𝑀 ) 𝑚 (15b)
1 = 1 − 𝑓 𝑥 𝐴 + 𝑓𝑥 𝑀 (15c)
By substituting the relation, 𝑥 𝛼 = ∆𝜆𝛼 /∆𝜆 along with Eq. (10) in Eq. (15c), we get:
−𝑚 1 𝑚
(𝜎𝑌 ) = 1 − 𝑓 𝜎𝑌𝐴 −𝑚
+ 𝑓 𝜎𝑌𝑀 (16)
If we assume that 𝜎𝑌 = 𝜎𝑌 (𝑇, 𝜎𝑒𝑞 ) and 𝑓 = 𝑓(𝑇, 𝜎𝑒𝑞 ), the following variations can be
derived:
1 𝜕𝜎𝑌 1 𝜕 𝜎𝑌
∆𝜎𝑌 = 𝜎𝑌 ∆𝑇 + 𝜎 ∆𝜎𝑒𝑞 (17a)
𝜎𝑌 𝜕𝑇 𝑌 𝜕 𝜎𝑒𝑞
= 𝜎𝑌 (𝐻𝑇 ∆𝑇 + 𝐻𝑞 ∆𝜎𝑒𝑞 )
𝜕𝑓 𝜕𝑓
∆𝑓 = ∆𝑓 𝑀 +∆𝑓 𝐴 = 𝜕𝑇 ∆𝑇 + 𝜕 𝜎 ∆𝜎𝑒𝑞 (17b)
𝑒𝑞
𝐹𝑀 𝐹𝐴
= 𝐹 𝑀 + 𝐹 𝐴 ∆𝑇 − (𝐶 +𝐶 )∆𝜎𝑒𝑞
𝑀 𝐴
In Eq. (17a), 𝐻𝑇 and 𝐻𝑞 can be rewritten using the chain rule as:
1 𝜕 𝜎𝑌 𝜕𝑓
𝐻𝑇 = 𝜎 (18a)
𝑌 𝜕𝑓 𝜕𝑇
1 𝜕 𝜎𝑌 𝜕𝑓
𝐻𝑞 = 𝜎 (18b)
𝑌 𝜕𝑓 𝜕𝜎𝑒𝑞
In general, the elastic moduli of Austenite and Martensite are quite different from
each other. In fact, the Young’s modulus of Austenite is approximately 75 GPa, while
that of the Martensite is about 40 GPa. Thus, another scheme is required to compute
the elastic moduli in Eq. 2 from the elastic properties of each phase. Assuming that
each phase as well as the total aggregate is elastically isotropic, 𝑳 can be expressed as
𝐿𝐸 = 2𝜇𝑰𝑰 + 𝜆𝑰 (19)
2
Where 𝑰𝑰 is the fourth order identity tensor. 𝜇 is the shear modulus and 𝜆 = 𝐾 − 3 𝜇,
𝐸 𝐸
𝜇 = 2(1−𝜗 ) and 𝐾 = 3(1−2𝜗 ) (20)
Then, the shear and bulk moduli of the total aggregate can be derived from those of
each phase following Eshelby’s (1957) solution as
𝜇
𝜇 1−𝑆1 (1− 𝑚 )
𝜇𝑎
= 1−(𝑆 𝜇𝑚 (21a)
𝜇𝑎 1 −𝑓)(1− 𝜇 )
𝑎
𝐾
𝐾 1−𝑆2 (1− 𝑚 )
𝐾𝑎
= 𝐾 (21b)
𝐾𝑎 1−(𝑆2 −𝑓)(1− 𝑚 )
𝐾𝑎
Where
2 4−5𝜗
𝑆1 = 15 (22a)
1−𝜗
2 1+𝜗
𝑆2 = 15 (22b)
1−𝜗
3𝐾−2𝜇
𝜗 = 6𝐾+2𝜇 (23)
𝛼 = 1 − 𝑓 𝛼𝑎 + 𝑓𝛼𝑚 (24)
3.4 Implementation
The above constitutive model has been implemented in MSC Marc (Marc User
Documentation, 2012). Figure 3.4 shows the Saeedvafa thermo-mechnical SMA
model implemented in the pre-processor of MSC Marc, Marc Mentat. The Young’s
Modulus & Poisson’s ratio for both the Austenite and Martensite phase are entered in
the overview window. Also there exits the options to input Austenite to Martensite
(Figure 3.5), Martensite to Austenite (Figure 3.6), Transformation Strains (Figure
3.7), Plasticity (Figure 3.8) and Thermal Expansion properties (Figure 3.9).
46
Figure 3.7: Saeedvafa model implemented in MSC Marc 2012 – Phase transformation
parameters.
Figure 3.9: Saeedvafa model implemented in MSC Marc 2012 – Thermal expansion.
48
In this chapter the various components of the simulation of the present study are
described. The various components are experimental setup & geometric properties,
material properties, the experimental procedure and assumptions in simulation. Shake
table vibration response data of a SMA braced frame structure has been obtained from
literature where the response of the frame structure subjected to seismic vibration on
shake table was measured (Ma et al., 2004). Then, using the Saeedvafa (2001) SMA
phenomenological model, the experiment is simulated. The experimental results were
then compared to the FEA predictions for the verification of the used SMA model.
The experimental setup consisted of a single storey frame structure with the option to
install various braces including SMA braces, a shake table with its control system, a
separate control system for controlling current applied to SMA braces, a
programmable power supply to actuate SMA braces, and various sensors (Figure 4.1).
A picture of the experimental setup is shown in Figure 4.2.
The frame was 22.86-cm (9-inch) wide and 58.42-cm (23-inch) high. A 3.288 kg
mass was attached to the top floor of the frame. With a feedback control, the shake
table simulated various earthquakes. The accelerations of the ground and the top
levels of the frame were measured by two accelerometers. The dimension of the frame
members are summarized in Table 4.1. The SMA braces used three 0.384 mm (0.015-
inch)-diameter Nitinol wires that were in a diagonal configuration (Ma et al., 2004).
The beam and column member consisted of the most common grade type 304
stainless steel. 90C Flexinol SMA wire from Dynalloy, Inc. was used as the material
for the SMA brace. The term 90C means the temperature region where SMA activates
or fully converts to Austenite phase (Dynalloy, 2012). The material properties of 304
steel and 90C Flexinol SMA wire are presented in Table 4.2 and Table 4.3
respectively.
304 Steel
Table 4.3: Material properties for 90C Flexinol SMA wire (Dynalloy, 2012; Churchill
& Shaw, 2008 and Harvey, 2010).
90C Flexinol
General Properties Austenite Martensite
Mass density (Kg/mm3) 6.45e-6
Young’s modulus (MPa) 83000 28000
Poisson’s ratio 0.33
Yield strength (MPa) 415 140
Melting temperature (°C) 1300
Thermal expansion co-efficient (mm/mm°C) 11e-6 6.6e-6
Thermal conductivity (W/mm °C) 0.018 0.0086
Specific heat (Joule/Kg °C) 837
Latent Heat (Joule/Kg) 24200
Convective cooling coefficient (W/mm2 °C) 0.0006
Emissivity 0.5
Electrical Resistivity (Ohm-mm) 0.001 0.0008
Saeedvafa Model Properties Austenite Martensite
Start temperature (°C) 88 72
Finish temperature (°C) 98 62
Austenite to Martensite slope (MPa/°C) 7.9
Martensite to Austenite slope (MPa/°C) 8.2
Deviatoric transformation strain 0.085
Volumetric transformation strain 0
Twinning stress (MPa) 100
g-A -4.0
g-B 2.0
g-C 0.0
g-D 2.8
g-0 (MPa) 300
g-max 1.0
Stress at g-max 1.00e+20
53
The SMA braced frame has been simulated using the Non-linear FE software MSC
Marc, the FE model of the frame is presented in Figure 4.4.
Figure 4.5: SMA brace consisting of three wires (Ma et al., 2004).
54
Because of the complex experimental procedure carried out by Ma et al. (2004) and
the presence of drawbacks, some assumptions have to be made in simulating the SMA
braced frame. The drawbacks & the corresponding assumptions are explained below-
1. The most significant drawback is using three SMA wires in the brace to get
larger cross sectional area. From Figure 4.5, the wires are touching each other
at various points. This may lead to no current at all in some wires because
electrical current will flow through the shortest distance and least resistance
path. So, there may be some wires which are not actuating at all, hence leaving
over-stressed situation in other wires. Also, it is a complex problem to model
the three wires independently because there is chance of electrical conduction
and short circuit among the wires. In order to simplify the model, the three
wires were replaced by a wire of equivalent cross-sectional area (0.3474 mm2)
and the input voltage adjusted to achieve an equivalent electrical resistance.
For this large Dia wire to achieve the same resistance (which results in same
amount of temperature), the input current has to be increased by three times.
The theoretical proof regarding this assumption is presented in Appendix A
and has also been verified numerically by MSC Marc.
2. No material property data was directly available in Ma et al., 2004. All
material properties have been obtained from other literature.
3. The beam-column connection is assumed to be 100% moment carrying.
4. The mass at the top is assumed to be a point mass at the middle of the beam.
5. The pre-stress load of the brace is unknown. An axial load of 35N is assumed.
6. The damping character of the steel frame is unknown. No external damping
has been applied in the simulation (e.g. numerical damping).
7. The resistance of the SMA wire increases with the increase of temperature.
But due to limitation in the constitutive model, temperature dependent
resistance is not considered here.
The simulation of the SMA braced frame has been carried out in two steps. First, only
the brace has been analyzed for coupled Electrical-Thermal analysis case (Figure 4.6).
The electrical problem has been solved first for the nodal voltages. Next, the thermal
problem has been solved to obtain the nodal temperatures. Then, the SMA braced
frame has been analyzed for Structural analysis case using the Saeedvafa SMA model
55
and importing the temperature of the brace from the first step (Figure 4.7). Here, the
structural problem is solved for the nodal displacements.
Copper wire
SMA brace
Figure 4.6: Step one - Coupled Electrical-Thermal FE model of SMA brace only.
Steel beam
SMA brace
problem in a Joule heating analysis is due to the fact that the resistance in the electric
problem is dependent on temperatures, and the internal heat generation in the thermal
problem is a function of the electrical flow. However, temperature-dependent
resistivity (instantaneous resistivity at any given temperature) has not been used in
this thesis. A weak coupling between the electrical and thermal problems has been
assumed in the coupled thermo-electrical analysis, such that the distributions of the
voltages and the temperatures of the structure can be solved separately within a time
increment. A steady-state solution of the electrical problem (in terms of nodal
voltages) has been calculated first within each time step. The heat generation due to
electrical flow is included in the thermal analysis as an additional heat input. The
temperature distribution of the structure (obtained from the thermal analysis) is used
to evaluate the temperature-dependent resistivity, which in turn is used for the
electrical analysis in the next time increment. For output, voltage, current density, and
heat generation are available as integration point values. The Current density is the
electric current per unit area of cross section, while Ohmic current is the current going
through the total area (Marc User Documentation, 2012).
Temperature loss occurs to the environment which is a heat transfer problem. There
are three types of heat transfer - conduction (easy), convection (less easy) and
radiation (difficult). The first two are linear problems, but radiation is non-linear. A
temperature difference must exist for heat transfer to occur. Heat is always transferred
in the direction of decreasing temperature. Temperature is a scalar, but heat flux is a
vector quantity. Conduction takes place within the boundaries of a body by the
diffusion of its internal energy. Convection occurs in a fluid by mixing. Here we will
consider only free convection from the surface of a body to the surrounding fluid.
Radiation heat transfer occurs by electromagnetic radiation between the surfaces of a
body and the surrounding medium. It is a highly nonlinear function of the absolute
temperatures of the body and medium.
The structural analysis is a non-linear dynamic analysis. There are three sources of
nonlinearity - material, geometric, and nonlinear boundary conditions. Contact
problem is present in this case between the two diagonal SMA brace. However, the
contact has been assumed to be frictionless and there is no electrical or heat transfer
between the SMA wires. The steel beam-column has been simulated using Von Mises
57
yield criterion taking elastic-plastic material behavior. The SMA brace has been
simulated using Saeedvafa (2001) SMA model. The details of the FE model for both
the analysis cases are summarized in Table 4.4.
Structural FE Analysis
A mesh convergence study has been carried out to determine optimum element
number for the simulation. The coarse mesh is consisted of Table 4.4 element data.
The fine mesh is created by doubling the number of elements. The study is
summarized in Table 4.5. The Fine mesh provides 0.118% of variation with respect to
the coarse mesh but consumes higher computational time and resource. So the coarse
mesh is selected for further analysis.
4.5 Verification
The presented constitutive model has been previously verified by several authors.
Choudhry & Yoon (2004) verified it for two cases. First, they simulated a one-
dimensional test of cubic specimen. Later, they simulated a stent used in medical
applications. For both cases, the Saeedvafa (2001) model provided good result.
Harvey (2010) accurately simulated coupled Electrical-Thermal-Structural behavior
of a SMA actuator. The simulation results were compared with physically obtained
experimental data of a small scale actuator. The results show that the finite element
simulations are in good agreement with measured test results.
For verification purpose of the SMA braced frame, two simulation cases are
performed. From the experimental data obtained from Ma et al., 2004, the SMA
braced frame is actuated by 1.8A current for a scaled down El Centro (1940)
earthquake and 2.2A for a scaled down Northridge (1994) earthquake. The shake table
acceleration time history for each earthquake case has been converted to displacement
time history using Digital signal Processing (DSP) techniques. For this purpose, a
strong ground motion processing software, SeismoSignal has been used
59
(SeismoSignal, 2012). First the input acceleration time history is corrected for base
line and then filtering has been applied to root out unwanted frequencies. Finally,
corrected displacement time history has been applied at the base as boundary
condition. Subsequently, the predicted absolute acceleration at the top of the frame
was compared with the corresponding experimental result.
Figure 4.8 represent the acceleration at the top from experimental data and Figures 4.9
to 4.11 represent the acceleration at the top from simulation for 0%, 5% & 10%
damping in the frame respectively for the El Centro earthquake case. The initial
acceleration value matches but after some time they differ. The experimental data
shows high damping after around 20 seconds while the simulation data shows that the
frame is under free vibration. The reason the system has high damping is because of
the presence of clear plastic fixtures to hold the frame together (Figure 4.4), which is
absent in the FE model, and assumption for the simulation is that those joints are
perfectly rigid in moment transfer. Also, no external damping such as numerical
damping has been applied in the simulation.
Figure 4.9: Acceleration at the top from simulation with no damping in frame.
Figure 4.10: Acceleration at the top from simulation with 5% damping in frame.
61
Figure 4.11: Acceleration at the top from simulation with 10% damping in frame.
Figure 4.12 represent the acceleration at the top from experimental data and Figures
4.13 to 4.15 represent the acceleration at the top from simulation for 0%, 5% & 10%
damping in the frame respectively for the Northridge earthquake case. This system
seems to be highly damped. Of the two earthquake cases, the El Centro earthqauke is
chosen for the parametric study to determine optimum current input in the next
chapter.
Figure 4.13: Acceleration at the top from simulation with no damping in frame.
Figure 4.14: Acceleration at the top from simulation with 5% damping in frame.
63
Figure 4.15: Acceleration at the top from simulation with 10% damping in frame.
A procedure of the SMA braced frame analysis is detailed in this chapter. The
Analysis was done in two steps for simplicity. The SMA brace is heated by electric
current due to joule heating.
In this chapter, a framework has been presented about how semi-active vibration
control system works using electric current as the driving variable. The predicted
dynamic responses of three configurations of the frame (unbraced, steel braced and
SMA braced) were obtained using MSC Marc. The effectiveness of SMA as a
vibration control device is proved by the use of constant current input. However, the
use of constant current for the entire period of earthquake results in high electrical
power consumption. Then the SMA is activated by pulsed current and both constant
& pulsed current input is compared to determine the optimum solution for SMA
vibration control. It is evident that the use of pulsed current resulted in reduced
energy consumption by the SMA, as well as mitigating the seismic vibrations on the
frame structure. Guidelines to optimum electrical input, considering appropriate
stiffness and damping characteristics; are established for both the constant current and
pulsed current case by parametric study.
The same frame structure from chapter 4 has been considered for this chapter. All
geometric and material properties apply here from chapter 4. The ground motion
selected is the El Centro earthquake recorded at the Imperial Valley Irrigation District
substation in El Centro, California, during May 18, 1940, is shown in Figure 5.1. The
earthquake acceleration time history was scaled down and converted to displacement
time history using Digital signal Processing (DSP) techniques. For this purpose, a
strong ground motion processing software, SeismoSignal has been used
(SeismoSignal, 2012). First the input acceleration time history is corrected for base
line and then filtering has been applied to root out unwanted frequencies.
Acceleration time history is band-pass filtered between 0.10 and 25.00 Hz using
Butterworth filter. The corrected displacement time history for El Centro earthquake
is shown in Figure 5.2. Finally, corrected displacement time history has been applied
65
at the base of the frame as displacement boundary condition. The predicted vibration
response of the frame is obtained from dynamic FE structural analysis.
Figures 5.3 and 5.4 illustrate the Fourier Amplitude Spectrum and Power Spectrum of
El Centro earthquake respectively. Fast Fourier Transformation (FFT) and the Power
Spectrum are useful for measuring the frequency content of stationary or transient
signals. FFTs produce the average frequency content of a signal over the entire time
66
that the signal was acquired. From the spectrums, the dominant frequency of the
earthquake vibration is estimated.
Table 5.1: Predicted modal frequencies of the unbraced and steel braced frame
Figure 5.5: Predicted displacement at the top of the unbraced frame, Earthquake
scaled down by a factor of 10
68
Figure 5.6: Predicted displacement at the top of the steel braced frame, Earthquake
scaled down by a factor of 10
Figure 5.7: Predicted displacement at the top of the unbraced frame, Earthquake
scaled down by a factor of 20
69
Figure 5.8: Predicted displacement at the top of the steel braced frame, Earthquake
scaled down by a factor of 20
Figure 5.9: Predicted displacement at the top of the unbraced frame, Earthquake
scaled down by a factor of 30
70
Figure 5.10: Predicted displacement at the top of the steel braced frame, Earthquake
scaled down by a factor of 30
Figure 5.11: Predicted displacement at the top of the unbraced frame, Earthquake
scaled down by a factor of 40
71
Figure 5.12: Predicted displacement at the top of the steel braced frame, Earthquake
scaled down by a factor of 40
Figure 5.13: Maximum displacement of unbraced and steel braced frame at different
scaling factor.
72
0.1
0.09
0.08
Plastic strain 0.07
0.06
0.05
0.04
0.03
0.02
0.01
0
0 10 20 30 40 50
Scaling Factor
Figures 5.5 to 5.12 show the predicted vibration response of the unbraced and steel
braced frame at different earthquake scaling factor. The unbraced frame is cross
braced with steel to reduce the earthquake vibration response, but the steel braced
frame experiences high displacement at the top than that of the unbraced frame for
high intensity earthquake vibration. With the lower intensity earthquake vibrations,
the steel braced frame succeeds in reducing the displacement at the top of the frame.
This phenomenon is summarized in Figure 5.13 which illustrates the maximum
displacement at the top of the frame for both the unbraced and the steel braced frame
at different scaling factor. The reason for the steel braced frame experiencing larger
displacement is, at the high intensity earthquake the steel brace experiences high
plastic deformation while for low intensity earthquake the steel brace frame
experiences low plastic deformation. Figure 5.14 illustrates the plastic strain for
different scaling factors and complements the observation. Due to plastic
deformation, the brace requires more displacement to develop enough tension force
on the brace. Also, due to elongated length of the brace because of the permanent
plastic deformation, the brace provides compressive force for small displacement,
resulting in amplifying the displacement at the top. Also, from Table 5.1 and from
Figures 5.3 & 5.4, it is evident that the dynamic amplification is also at play here
73
because of the closeness of natural frequency of the unbraced & steel braced frame
with the dominant frequency of the earthquake.
This parametric study indicates the weakness of steel brace system due to early
yielding in earthquake scenarios (for 304 steel). The predictions show that the steel
brace system works best when within the elastic limit throughout the entire
earthquake. As a result, the steel brace is pre-stressed by no more than 5% of the yield
stress of 215 MPa.
The SMA brace is actuated by electric current due to joule heating. First the brace is
heated by constant current. Later, the brace is activated by pulsed current using
Pulsed Width Modulation (PWM) technique and the optimization gain using pulsed
current input is established. The El Centro earthquake acceleration time history is
scaled down by a factor of 30, processed and the displacement time history is applied
at the base of the SMA braced frame for both the constant and pulsed current input
case. The predicted displacement at the top of the frame is obtained from dynamic FE
structural analysis. The SMA brace is pre-stressed with 100 MPa. The pre-stressing is
simulated by applying a 1 second static load-case before the dynamic load-case. Point
to be noted here is that the SMA brace can be pre-stressed by far higher value than
the steel brace, and be still within the elastic limit. The constant current and pulsed
current input profile for 1A current is illustrated in Figures 5.15 & 5.16 respectively.
The constant current is kept on for the entire earthquake duration of 53.74 seconds.
The SMA brace temperature increases because of joule heating. Due to conduction at
both ends of the SMA brace, the minimum temperature is always 23°C which is the
room temperature. The maximum temperature depends on the applied current. The
SMA will actuate over the length of the brace having varying temperature depending
on the phase transformation temperature. Figures 5.17 & 5.18 illustrate the predicted
maximum temperature and displacement at the top of the frame respectively for 0A
current. In this case, the SMA brace is in fully Martensite phase and there is no phase
transformation. Then constant current of 1.4A is applied and subsequently increased
by 0.1A till 2.4A. The simulated maximum temperature history and displacement at
the top of the frame is illustrated in Figures 5.19 to 5.40. The maximum displacement
of the frame is plotted against each constant current input case in Figure 5.41 and the
optimum constant current which results in minimum displacement at the top is
determined. The optimum constant current is 2.0A from Figure 5.41. For this
earthquake vibration, the maximum displacement for unbraced frame is 12.0121 mm,
for steel braced frame is 9.0833 mm and for SMA braced frame is 6.1644 mm with
2.0A constant current as the driving force. The SMA brace experiences no plastic
deformation while the steel brace suffers maximum plastic strain of 0.0038. The
result proves the effectiveness of SMA brace over conventional steel brace in
mitigating the seismic vibration response.
75
Figure 5.17: Predicted maximum temperature history of the SMA brace for 0A
current.
Figure 5.18: Predicted displacement at the top of the frame for 0A current.
76
Figure 5.19: Predicted maximum temperature history of the SMA brace for 1.4A
current.
Figure 5.20: Predicted displacement at the top of the frame for 1.4A current.
77
Figure 5.21: Predicted maximum temperature history of the SMA brace for 1.5A
current.
Figure 5.22: Predicted displacement at the top of the frame for 1.5A current.
78
Figure 5.23: Predicted maximum temperature history of the SMA brace for 1.6A
current.
Figure 5.24: Predicted displacement at the top of the frame for 1.6A current.
79
Figure 5.25: Predicted maximum temperature history of the SMA brace for 1.7A
current.
Figure 5.26: Predicted displacement at the top of the frame for 1.7A current.
80
Figure 5.27: Predicted maximum temperature history of the SMA brace for 1.8A
current.
Figure 5.28: Predicted displacement at the top of the frame for 1.8A current.
81
Figure 5.29: Predicted maximum temperature history of the SMA brace for 1.9A
current.
Figure 5.30: Predicted displacement at the top of the frame for 1.9A current.
82
Figure 5.31: Predicted maximum temperature history of the SMA brace for 2.0A
current.
Figure 5.32: Predicted displacement at the top of the frame for 2.0A current.
83
Figure 5.33: Predicted maximum temperature history of the SMA brace for 2.1A
current.
Figure 5.34: Predicted displacement at the top of the frame for 2.1A current.
84
Figure 5.35: Predicted maximum temperature history of the SMA brace for 2.2A
current.
Figure 5.36: Predicted displacement at the top of the frame for 2.2A current.
85
Figure 5.37: Predicted maximum temperature history of the SMA brace for 2.3A
current.
Figure 5.38: Predicted displacement at the top of the frame for 2.3A current.
86
Figure 5.39: Predicted maximum temperature history of the SMA brace for 2.4A
current.
Figure 5.40: Predicted displacement at the top of the frame for 2.4A current.
87
Ma & Song (2003) demonstrated that the use of discrete pulse signal to drive SMA
actuators has the advantage of reducing power consumption without sacrificing
performance. In this study, the pulsed current is kept on for a cycle of 1.5 second
period with a 0.15 second duty time, for the entire earthquake duration of 53.74
seconds. The reason for using the pulse driving mode in this study is to achieve a
larger damping while maintaining a wide range for frequency adjustment. The
periodically applied pulse current ensures shape recovery if the SMA braces
experience a large strain. In a sense, pulsed actuation combines the advantages of
high stiffness and high damping (Ma et al., 2004). Due to conduction at both ends of
the SMA brace, the minimum temperature is always 23°C which is the room
temperature. The maximum temperature depends on the amplitude of the applied
pulsed current. The SMA brace temperature increases because of joule heating for the
0.15 second duty time and decreases due to heat loss to the environment for the rest
1.35 second off duty time. The on and off duty cycle occurs throughout the entire
earthquake duration. This ultimately increases the temperature of the SMA brace and
creates a fairly uniform temperature band of 30°C. The SMA actuates over the entire
length of the brace having varying temperature depending on the phase
transformation temperature.
88
Pulsed current with 4.1A amplitude is applied and subsequently increased by 0.1A till
5.4A. The simulated maximum temperature history and displacement at the top of the
frame is illustrated in Figure 5.42 to Figure 5.69. The maximum displacement of the
frame is plotted against each pulsed current input case in Figure 5.70 and the
optimum pulsed current which results in minimum displacement at the top is
determined. The optimum pulsed current is 5.2A from Figure 5.70. For this
earthquake vibration, the maximum displacement for optimum constant current
actuated SMA braced frame is 6.1644 mm and for optimum pulsed current actuated
SMA braced frame is 8.3641 mm.
Figure 5.42: Predicted maximum temperature history of the SMA brace for 4.1A
current.
89
Figure 5.43: Predicted displacement at the top of the frame for 4.1A current.
Figure 5.44: Predicted maximum temperature history of the SMA brace for 4.2A
current.
90
Figure 5.45: Predicted displacement at the top of the frame for 4.2A current.
Figure 5.46: Predicted maximum temperature history of the SMA brace for 4.3A
current.
91
Figure 5.47: Predicted displacement at the top of the frame for 4.3A current.
Figure 5.48: Predicted maximum temperature history of the SMA brace for 4.4A
current.
92
Figure 5.49: Predicted displacement at the top of the frame for 4.4A current.
Figure 5.50: Predicted maximum temperature history of the SMA brace for 4.5A
current.
93
Figure 5.51: Predicted displacement at the top of the frame for 4.5A current.
Figure 5.52: Predicted maximum temperature history of the SMA brace for 4.6A
current.
94
Figure 5.53: Predicted displacement at the top of the frame for 4.6A current.
Figure 5.54: Predicted maximum temperature history of the SMA brace for 4.7A
current.
95
Figure 5.55: Predicted displacement at the top of the frame for 4.7A current.
Figure 5.56: Predicted maximum temperature history of the SMA brace for 4.8A
current.
96
Figure 5.57: Predicted displacement at the top of the frame for 4.8A current.
Figure 5.58: Predicted maximum temperature history of the SMA brace for 4.9A
current.
97
Figure 5.59: Predicted displacement at the top of the frame for 4.9A current.
Figure 5.60: Predicted maximum temperature history of the SMA brace for 5.0A
current.
98
Figure 5.61: Predicted displacement at the top of the frame for 5.0A current.
Figure 5.62: Predicted maximum temperature history of the SMA brace for 5.1A
current.
99
Figure 5.63: Predicted displacement at the top of the frame for 5.1A current.
Figure 5.64: Predicted maximum temperature history of the SMA brace for 5.2A
current.
100
Figure 5.65: Predicted displacement at the top of the frame for 5.2A current.
Figure 5.66: Predicted maximum temperature history of the SMA brace for 5.3A
current.
101
Figure 5.67: Predicted displacement at the top of the frame for 5.3A current.
Figure 5.68: Predicted maximum temperature history of the SMA brace for 5.4A
current.
102
Figure 5.69: Predicted displacement at the top of the frame for 5.4A current.
The constant current is on duty for the whole 53.74 seconds while the pulsed current
is 35.3266 cycles with 36 on duty peaks resulting in 5.40 second effective on duty
duration. The total electric energy is calculated by Eq. 5.1.
𝐸 = 𝐼𝑉𝑡 (5.1)
The total electric energy consumption by both the optimum constant and optimum
pulsed current is summarized in Table 5.2. Pulsed current actuated SMA brace has
39.91% less electric energy consumption than that of constant current actuated SMA
brace. From the result it is evident that the use of pulsed current not only resulted in
better seismic response control, but also reduced electric energy consumption.
CHAPTER 6
CONCLUSIONS AND RECOMMENDATIONS
6.1 Conclusions
The present research work describes the various aspects of Shape Memory Alloy
(SMA) properties, usage of SMA in civil engineering structures and the constitutive
modeling approaches. An effective thermo-mechanical model is presented which can
be used in the numerical simulation of element having any dimensionality. Two new
approaches in vibration reduction of a frame structure using active SMA braces are
presented. The first approach uses a constant electrical current to actuate the SMA
braces to ensure partial phase transformation and the co-existence of both Martensite
and Austenite phases. The second approach uses specially designed pulsed current to
actuate the SMA braces. Both of these currents increase damping by using SMAs in
Martensite phase, which has better damping than its austenite phase. Both approaches
use electrical current to ensure the shape recovery after the SMA braces experience
large strain. Simulation results show that the two approaches using SMA braces are
effective in reducing structural vibrations induced by earthquake excitations, as
compared with the case of steel brace.
Under the scope of the present study, following conclusions can be made:
1. Unbraced frame can only be effectively stiffened using steel brace against
earthquake vibration when the intensity is low, ensuring that the steel brace
stays in elastic limit.
2. The displacement of the two dimensional (2D) frame under earthquake
loading can be reduced significantly using SMA bracing which steel brace
fails to achieve under severe earthquake scenarios.
3. For El Centro earthquake, the optimum value of the constant current is 2.0A
for minimum displacement of the system. The optimum current is believed to
result in a partial phase transformation, a co-existence of both Martensite and
austenite phases.
4. For El Centro earthquake, the optimum value of the pulsed current for
minimization of the displacement of the building frame is found to be 5.2A,
105
It is recommended that the study can be extended further in the following fields
focusing on both the shape memory effect and the super-elasticity properties:
Auricchio, F., & Taylor, R. (1996). Shape memory alloy super-elastic behavior: 3D
finite-element simulations. In P. Gobin, & J. Tatibouet (Ed.), Proceedings of the 3rd
International Conference on Intelligent Materials and the 3rd European Conference
on Smart Structures and Materials. 2779, pp. 487–492. Lyon, France: SPIE.
Auricchio, F., & Taylor, R. (1996). Shape memory alloys: Modeling and numerical
simulations of the finite-strain superelastic behavior. Comp. Meth. In Appl. Mech.
Eng. , 143, 175-194.
Auricchio, F., Fugazza, D., & Desroches, R. (2006). Earthquake Performance of Steel
Frames With Nitinol Braces. Journal of Earthquake Engineering , 10 (Special 1), 45–
66.
Boyd, J., & Lagoudas, D. (1996). A thermodynamical constitutive model for shape
memory materials. Part I. The monolithic shape-memory alloy. Int. J. Platsicity , 129,
805-842.
CESMD. (2013). CESMD Internet Data Report. Retrieved 01 25, 2013, from Center
for Engineering Strong Motion Data: http://www.strongmotioncenter.org/cgi-
bin/CESMD/iqr_dist_DM2.pl?IQRID=ElCentro_18May1940
Desroches, R., & Smith, B. (2003). Shape memory alloys in seismic resistant design
and retrofit: a critical review of their potential and limitations. Journal of Earthquake
Engineering , 7 (3), 1-15.
Dolce, M., Cardone, D., & Marnetto, R. (2001). SMA re-centering devices for seismic
isolation of civil structures. Proceedings of SPIE 2001. 4330, pp. 238–249. SPIE.
Dynalloy. (2012). Dynalloy, Inc. Technical Sheets. Retrieved 01 25, 2012, from
Dynalloy: http://www.dynalloy.com/TechSheets.php
108
Huang, M., & Brinson, L. (1998). Multivariant model for single crystal shape-
memory-alloy behaviour. Journal of the Mechanics and Physics of Solids , 46, 1379–
1409.
Islam, B., Ahsan, R., & Rashed, M. G. (2012). Retrofit Techniques of Structures
Associated With Soft Story. Proceedings of the 1st International Conference on
Advances in Civil Engineering. Chittagong, Bangladesh: CUET.
Janke, L., Czaderski, C., Motavalli, M., & Ruth, J. (2005). Applications of shape
memory alloys in civil engineering structures—Overview, limits and new ideas.
Materials and Structures , 38 (5), 578-592.
Key to Metals AG. (2008, 11). Shape Memory Alloys: Part One. Retrieved 01 25,
2013, from Key to Metals:
http://www.keytometals.com/page.aspx?ID=CheckArticle&site=ktn&NM=207
Li, H., Liu, M., & Ou, J. (2004). Vibration mitigation of a stay cable with one shape
memory alloy damper. Structural Control and Health Monitoring , 11, 1–36.
Lubliner, J., & Auricchio, F. (1996). Generalized plasticity and shape memory alloys.
Int. J. of Solid Struct. , 33, 991-1003.
Ma, N., & Song, G. (2003). Control of shape memory alloy actuator using pulse width
modulation. Smart Mater. Struct. , 12, 712–719.
Ma, N., Song, G., & Tarefder, R. (2004). Vibration control of a frame structure using
shape memory alloy braces. Proceedings of the Third International Conference on
Earthquake Engineering : new frontier and research transformation. Nanjing, China.
Marc User Documentation. (2012). Volume A: Theory and User Information. Santa
Ana: MSC.Software Corporation.
MatWeb. (2012). 304 Stainless Steel. Retrieved 01 25, 2012, from MatWeb:
http://www.matweb.com/search/datasheet.aspx?MatGUID=abc4415b0f8b490387e3c9
22237098da
McCormick, J., DesRoches, R., Fugazza, D., & Auricchio, F. (2007). Seismic
Assessment of Concentrically Braced Steel Frames with Shape Memory Alloy
Braces. Journal of Structural Engineering , 133 (6), 862-870.
Miyazaki, S., Otsuka, K., & Suzuki, Y. (1981). Transformation Pseudoelasticity and
Deformation Behavior in a Ti-50.6at%Ni Alloy. Scripta Metallurgica , 287-292.
Ocel, J., DesRoches, R., Leon, R., Hess, W., Krumme, R., Hayes, J., et al. (2004).
Steel Beam-Column Connections using Shape Memory Alloys. Journal of Structural
Engineering , 130, 732-740.
110
Ozbulut, O. E., Hurlebaus, S., & Desroches, R. (2011). Seismic Response Control
Using Shape Memory Alloys: A Review. Journal of Intelligent Material Systems and
Structures , 22 (14), 1531-1549.
Patoor, E., Eberhardt, A., & Berveiller, M. (1994). Micromechanical modelling of the
shape memory behaviour. In L. Brinson, & B. Moran (Ed.), Mechanics of Phase
Transformations and Shape Memory Alloys: Proceedings of the ASME International
Congress and Exposition. 189, pp. 23–37. Chicago, Ill.: American Society of
Mechanical Engineers (ASME).
Qian, H., Li, H. N., Song, G., Chen, H., Ren, W. J., & Zhang, S. (2010). Seismic
Vibration Control of Civil Structures Using Shape Memory Alloys: A Review. Earth
and Space 2010 : Engineering, Science, Construction, and Operations in Challenging
Environments (pp. 3377-3395). Honolulu, Hawaii: American Society of Civil
Engineers.
Rashed, M. G., Ahsan, R., & Islam, B. (2012). Civil Engineering Application of
Shape Memory Alloys. Proceedings of the 1st International Conference on Advances
in Civil Engineering. Chittagong, Bangladesh: CUET.
Saadat, S., Noori, M., Davoodi, H., Hou, Z., Suzuki, Y., & Masuda, A. (2001). Using
NiTi SMA tendons for vibration control of coastal structures. Smart Materials and
Structures , 10 (4), 695.
Saadat, S., Salichs, J., Noori, M., Hou, Z., Davoodi, H., Bar-on, I., et al. (2002). An
overview of vibration and seismic applications of NiTi shape memory alloy. Smart
Materials and Structures , 11 (2), 218–229.
Song, G., Ma, N., & Li, H.-N. (2006). Applications of shape memory alloys in civil
structures. Engineering Structures , 28 (9), 1266–1274.
Tarefder, R. A., Ma, N., & Song, G. (2006). Dynamic Behavior of a Two-Story
Building Frame Braced with SMA for Vibration Control. Earth & Space 2006 :
Engineering, Construction, and Operations in Challenging Environment (pp. 1-9).
Houston, Texas: American Society of Civil Engineers.
Torchu, F., & Qian, Y. (1997). Nonlinear finite element simulation of superplastic
shape-memory alloy parts. Computers and Structures , 62, 799-810.
APPENDIX-A
Proof of Equivalent Cross Sectional Area SMA Wire
Figure A1 shows three wires connected in parallel, having cross sectional area of
0.1158 mm2 each and resistivity of 0.0008 Ohm-mm will be replaced by 1 wire
having cross sectional area of 0.3474 mm2.
Here,
𝜌 = 𝑅𝑒𝑠𝑖𝑠𝑡𝑖𝑣𝑖𝑡𝑦
𝑅 𝜌 0.0008
The linear resistance of each wire is = 𝐴 = 0.1158 = 0.00691 𝑂𝑚/𝑚𝑚
𝐿
𝑅𝑇 1 1 1 1
= = = =
𝐿 1 1 1 1 1 1 1 1
𝑅1 + 𝑅2 + 𝑅3 𝑅 +𝑅 +𝑅 3×𝑅 3 × 0.00691
𝐿 𝐿 𝐿 𝐿 𝐿 𝐿 𝐿
= 0.0023 𝑂𝑚/𝑚𝑚
The parallel circuit will be replaced by a Large Dia wire. For the large Dia wire,
𝑅𝑇 𝜌
=
𝐿 𝐴
𝜌
𝑜𝑟, 0.0023 =
𝐴
So, Resistivity will remain same for the Large Dia wire.
𝑉
𝐼1 = 𝐼2 = 𝐼3 =
𝑅
𝑉1 = 𝑉2 = 𝑉3 = 𝑉
And, 𝐼𝑇 = 𝐼1 + 𝐼2 + 𝐼3
𝑉
Now, 𝐼𝑇 =
𝑅𝑇
𝑉
𝑜𝑟, 𝐼1 + 𝐼2 + 𝐼3 =
𝑅𝑇
𝑉 𝑉
𝑜𝑟, 3 × =
𝑅 𝑅𝑇
𝑜𝑟, 𝑅 = 3 × 𝑅𝑇
114
𝐴𝑔𝑎𝑖𝑛, 𝑉 = 𝐼 × 𝑅
𝑜𝑟, 𝑉 = 𝐼 × 3 × 𝑅𝑇
𝑜𝑟, 𝑉 = 3 × 𝐼 × 𝑅𝑇
So, to achieve same amount of resistance, the input current for the large Dia wire has
to be increased by 3 times or by the number of wires connected in parallel
configuration.
Figure A2: Comparative study of equivalent and single wire maximum temperature.
Coupled Electrical-Thermal FE analysis is conducted for both the single wire and
equivalent cross section wire and the maximum temperature against constant current
input cases for both the wire is illustrated in Figure A2. The resultant graph ensures
the correctness of the input current assumptions for the equivalent cross section.