4.13. Shape Memory Alloy (SMA)

A shape memory alloy (SMA) is a metallic alloy that returns to its original shape after deformation. Upon loading and unloading cycles, an SMA can undergo large deformation without showing residual strains (pseudoelasticity effect, also often called superelasticity), and can recover its original shape through thermal cycles (the shape memory effect).

Such distinct material behavior is due to the material microstructure in which there exists two different crystallographic structures, one characterized by austenite (A), and another one by martensite (M). Austenite is the crystallographically more-ordered phase, and martensite is the crystallographically less-ordered phase. The key characteristic of an SMA is the occurrence of a martensitic phase transformation.

Typically, the austenite is stable at high temperatures and low stress, while the martensite is stable at low temperatures and high stress. The reversible martensitic phase transformation results in unique effects: the pseudoelasticity (PE) and the shape memory effect (SME).

As shown by (a) in the following figure, whenever σL is positive, the specimen recovers its original shape completely and returns to a stress-free configuration (PE).

Figure 4.42:  Pseudoelasticity (PE) and Shape Memory Effect (SME)

Pseudoelasticity (PE) and Shape Memory Effect (SME)
(a) PE -- High Temperature
Pseudoelasticity (PE) and Shape Memory Effect (SME)
(b) SME -- Low Temperature

As shown by (b) in the figure, when σL is negative, residual strains (E and E') can be observed after unloading into a stress-free configuration. If the material is heated, then eventually σL becomes positive; however, the admissible configuration under a stress-free state points to A. The material therefore undergoes an inverse transformation process (SME).

Two SMA material model options (accessed via TB,SMA) are available, one for simulating superelastic behavior and the other for simulating the shape memory effect behavior of shape memory alloys.

The material option for superelasticity is based on Auricchio et al. [1] in which the material undergoes large-deformation without showing permanent deformation under isothermal conditions, as shown by (a) in Figure 4.42: Pseudoelasticity (PE) and Shape Memory Effect (SME). The material option for the shape memory effect is based on the 3-D thermomechanical model for stress-induced solid phase transformations [2] [3] [4].

The following shape memory alloy topics are available:

4.13.1. SMA Model for Superelasticity

The following topics are available for the SMA superelasticity option:

4.13.1.1. Constitutive Model for Superelasticity

From a macroscopic perspective, the phase-transformation mechanisms involved in superelastic behavior are:

  1. Austenite to martensite (A->S)

  2. Martensite to austenite (S->A)

  3. Martensite reorientation (S->S)

Figure 4.43:  Typical Superelasticity Behavior

Typical Superelasticity Behavior

Two of the phase transformations are considered here: A->S and S->A. The material is composed of two phases, the austenite (A) and the martensite (S). Two internal variables, the martensite fraction (ξS) and the austenite fraction (ξA), are introduced. One of them is a dependent variable, and they are assumed to satisfy the relation expressed as:

The independent internal variable chosen here is ξS.

The material behavior is assumed to be isotropic. The pressure dependency of the phase transformation is modeled by introducing the Drucker-Prager loading function, as follows:

where α is the material parameter, σ is the stress, and 1 is the identity tensor.

The evolution of the martensite fraction, ξS, is then defined as follows:

where:

where are the material parameters shown in the following figure:

Figure 4.44:  Idealized Stress-Strain Diagram of Superelastic Behavior

Idealized Stress-Strain Diagram of Superelastic Behavior

where are the material parameters shown in Figure 4.44: Idealized Stress-Strain Diagram of Superelastic Behavior.

The material parameter α characterizes the material response in tension and compression. If tensile and compressive behaviors are the same, then α = 0. For a uniaxial tension-compression test, α can be related to the initial value of austenite to martensite phase transformation in tension and compression (, respectively) as:

The stress-strain relation is:

where D is the elastic stiffness tensor, is the transformation strain tensor, and is the material parameter shown in Figure 4.44: Idealized Stress-Strain Diagram of Superelastic Behavior.

4.13.1.2. Material Parameters for the Superelastic SMA Material Model

To model the superelastic behavior of shape memory alloys, initialize the data table via the TB,SMA command's SUPE option.

Define the elastic behavior in the austenite state (MP).

The superelastic SMA option is described by six constants that define the stress-strain behavior in loading and unloading for the uniaxial stress-state.

For each data set, define the temperature (TBTEMP), then define constants C1 through C6 (TBDATA). You can define up to 99 sets of temperature-dependent constants in this manner.

Table 4.23:  Superelastic Option Constants

Constant Meaning Property
C1 Starting stress value for the forward phase transformation
C2 Final stress value for the forward phase transformation
C3 Starting stress value for the reverse phase transformation
C4 Final stress value for the reverse phase transformation
C5 Maximum residual strain
C6 α Parameter measuring the difference between material responses in tension and compression


Example 4.27:  Defining Elastic Properties of the Austenite Phase
MP,EX,1,60000.0
MP,NUXY,1,0.36
Define SMA material properties
TB,SMA,1,,,SUPE
TBDATA,1, 520, 600, 300, 200, 0.07, 0.0


4.13.2. SMA Material Model with Shape Memory Effect

The following topics concerning SMA and the shape memory effect are available:

4.13.2.1. The Constitutive Model for Shape Memory Effect

The shape memory effect was based on a 3-D thermomechanical model for stress-induced solid phase transformations that was presented in [2] [3][4]. Within the framework of classical irreversible thermodynamics, the model is able to reproduce all of the primary features relative to shape memory materials in a 3-D stress state. The free energy potential is set to:

where:

D = material elastic stiffness tensor
= total strain
= total transformation strain
= deviatoric transformation strain
τM(T) = a positive and monotonically increasing function of the temperature as 〈β(T - T0)〉+ in which 〈∙〉+ is the positive part of the argument (also known as Maxwell stress).
β = material parameter
T = temperature
T0 = temperature below which no austenite is observed in a stress-free state
h = material parameter related to the hardening of the material during the phase transformation
= indicator function introduced to satisfy the constraint on the transformation norm [1] in which

from which we have

where X tr is defined as the transformation stress.

Stresses, strains, and the transformation strains are then related as follows:

Splitting the stress into deviatoric and volumetric components, we have

where S is the deviatoric stress and p is the volumetric stress (also called hydrostatic pressure)

The transformation stress is given as follows:

where γ is defined by

where is a maximum transformation strain.

Numerous experimental tests show an asymmetric behavior of SMA in tension and compression, and suggest describing SMA as an isotropic material with a Prager-Lode-type limit surface. Accordingly, the following yield function is assumed:

where X tr is the transformation stress, J2 and J3 are the second and third invariants of transformation stress, m is a material parameter related to Lode dependency, and R is the elastic domain radius.

J2 and J3 are defined as follows:

The evolution of transformation strain is defined as:

where ξ is an internal variable and is called as transformation strain multiplier. ξ and F(X tr) must satisfy the classical Kuhn-Tucker conditions, as follows:

which also reduces the problem to a constrained optimization problem.

The elastic properties of austenite and martensite phase differ. During the transformation phase, the elastic stiffness tensor of material varies with the deformation. The elastic stiffness tensor is therefore assumed to be a function of the transformation strain , defined as:

where D A is the elastic stiffness tensor of austenite phase, and D S is the elastic stiffness tensor of martensite phase. The Poisson’s ratio of the austenite phase is assumed to be the same as the martensite phase. When the material is in its austenite phase, D = D A, and when the material undergoes full transformation (martensite phase), D = D S.

The following figure illustrates a number of the mechanical model features:

Figure 4.45:  Admissible Paths for Elastic Behavior and Phase Transformations

Admissible Paths for Elastic Behavior and Phase Transformations

The austenite phase is associated with the horizontal region abcd. Mixtures of phases are related to the surface cdef. The martensite phase is represented by the horizontal region efgh. Point c corresponds to the nucleation of the martensite phase. Phase transformations take place only along line cf, where . Saturated phase transformations are represented by paths on line fg. The horizontal region efgh contains elastic processes except, of course, those on line fg.

A backward Euler integration scheme is used to solve the stress update and the consistent tangent stiffness matrix required by the finite element solution for obtaining a robust nonlinear solution. Because the material tangent stiffness matrix is generally unsymmetric, use the unsymmetric Newton-Raphson option (NROPT,UNSYM) to avoid convergence problems.

4.13.2.2. Material Parameters for the Shape Memory Effect Option

To model the shape memory effect behavior of shape memory alloys, initialize the data table via the TB,SMA command’s MEFF option.

Define the elastic behavior in the austenite state (MP).

The shape memory effect option is described by seven constants that define the stress-strain behavior of material in loading and unloading cycles for the uniaxial stress-state and thermal loading.

For each data set, define the temperature (TBTEMP), then define constants C1 through C7 (TBDATA). You can define up to 99 sets of temperature-dependent constants in this manner.

Table 4.24:  Shape Memory Effect Option Constants

Constant Meaning Property

C1

h

Hardening parameter

C2

To

Reference temperature

C3

R

Elastic limit

C4

β

Temperature scaling parameter

C5

Maximum transformation strain

C6

Em

Martensite modulus

C7

m

Lode dependency parameter


Example 4.28:  Defining Shape Memory Effect Properties of the Austenite Phase
MP,EX,1,60000.0
MP,NUXY,1,0.36
Define SMA material properties
TB,SMA,1,,,MEFF
TBDATA,1,1000, 223, 50, 2.1, 0.04, 45000
TBDATA,7,0.05

4.13.3. Result Output of Solution Variables

For postprocessing, solution output is as follows:

  • Stresses are output as S.

  • Elastic strains are output as EPEL

  • Transformation strains, εtr, are output as plastic strain EPPL

  • The ratio of the equivalent transformation strain to maximum transformation strain, , is available as part of nonlinear solution record NL, and can be processed as component EPEQ of NL.

  • Elastic strain energy density is available as part of the strain energy density record SEND (ELASTIC).

4.13.4. Element Support for SMA

Support for SMA material models with the superplasticity option (TB,SMA,,,,SUPE) is available with current-technology plane, solid, and solid-shell elements where 3-D stress states are applicable (including 3-D solid elements, solid-shell elements, 2-D plane strain, axisymmetric elements, and solid pipe elements).

Support for SMA material models with the memory-effect option (TB,SMA,,,,MEFF) is available with current-technology beam, shell, plane, solid, and solid-shell elements (including 3-D solid elements, solid-shell elements, 2-D plane stress and strain, axisymmetric elements, and solid pipe elements).

For specific element support for SMA, see Material Model Element Support.

4.13.5. Learning More About Shape Memory Alloy

A considerable body of literature exists concerning shape memory alloy material models. The following list of resources offers a wealth of information but is by no means exhaustive:

  1. Auricchio, F. “A Robust Integration-Algorithm for a Finite-Strain Shape-Memory-Alloy.” International Journal of Plasticity. 17 (2001): 971-990.

  2. Souza, A. C., E. N. Mamiya, N. Zouain. “Three-Dimensional Model for Solids Undergoing Stress-Induced Phase Transformations.” European Journal of Mechanics-A/Solids . 17 (1998): 789-806.

  3. Auricchio, F., R. L. Taylor, J. Lubliner. “Shape-Memory Alloys: Macromodeling and Numerical Simulations of the Superelastic Behavior.” Computational Methods in Applied Mechanical Engineering. 146, 1 (1997): 281-312.

  4. Auricchio, F., L. Petrini. “Improvements and Algorithmical Considerations on a Recent Three-Dimensional Model Describing Stress-Induced Solid Phase Transformations.” International Journal for Numerical Methods in Engineering. 55 (2005): 1255-1284.

  5. Auricchio, F., D. Fugazza, R. DesRoches. “Numerical and Experimental Evaluation of the Damping Properties of Shape-Memory Alloys.” Journal of Engineering Materials and Technology. 128:3 (2006): 312-319.

For an example analysis, see Shape Memory Alloy (SMA) with Thermal Effect in the Technology Demonstration Guide.


Release 18.2 - © ANSYS, Inc. All rights reserved.