SANDIA REPORT

SAND2017-11360
Unlimited Release
Printed October 18, 2017

Linear and Nonlinear Constitutive Model

for Piezoelectricity in ALEGRA-FE
Wen Dong

Prepared by
Sandia National Laboratories
Albuquerque, New Mexico 87185 and Livermore, California 94550

Sandia National Laboratories is a multimission laboratory managed and operated by National Technology and
Engineering Solutions of Sandia, LLC., a wholly owned subsidiary of Honeywell International, Inc., for the
U.S. Department of Energy’s National Nuclear Security Administration under contract DE-NA0003525.

Approved for public release; further dissemination unlimited.

Issued by Sandia National Laboratories, operated for the United States Department of Energy
by National Technology and Engineering Solutions of Sandia, LLC.

NOTICE: This report was prepared as an account of work sponsored by an agency of the United
States Government. Neither the United States Government, nor any agency thereof, nor any
of their employees, nor any of their contractors, subcontractors, or their employees, make any
warranty, express or implied, or assume any legal liability or responsibility for the accuracy,
completeness, or usefulness of any information, apparatus, product, or process disclosed, or rep-
resent that its use would not infringe privately owned rights. Reference herein to any specific
commercial product, process, or service by trade name, trademark, manufacturer, or otherwise,
does not necessarily constitute or imply its endorsement, recommendation, or favoring by the
United States Government, any agency thereof, or any of their contractors or subcontractors.
The views and opinions expressed herein do not necessarily state or reflect those of the United
States Government, any agency thereof, or any of their contractors.

Printed in the United States of America. This report has been reproduced directly from the best
available copy.

Available to DOE and DOE contractors from

U.S. Department of Energy
Office of Scientific and Technical Information
P.O. Box 62
Oak Ridge, TN 37831

Telephone: (865) 576-8401

Facsimile: (865) 576-5728
E-Mail: reports@adonis.osti.gov
Online ordering: http://www.osti.gov/bridge

Available to the public from

U.S. Department of Commerce
National Technical Information Service
5285 Port Royal Rd
Springfield, VA 22161

Telephone: (800) 553-6847

Facsimile: (703) 605-6900
E-Mail: orders@ntis.fedworld.gov
Online ordering: http://www.ntis.gov/help/ordermethods.asp?loc=7-4-0#online

NT OF E
ME N
RT
ER
A
DEP

GY

• •
IC A
U NIT

ER
ED

ST M
A TES OF A

2
 SAND2017-11360
Unlimited Release
Printed October 18, 2017

Linear and Nonlinear Constitutive Model for

Piezoelectricity in ALEGRA-FE

Wen Dong
Computational Materials & Data Science Department
Sandia National Laboratories
Albuquerque, NM 87185

Abstract
This report develops and documents linear and nonlinear constitutive relations implemented in
ALEGRA-FE (ferroelectric) [1]. A thermodynamic framework is created to describe the electrome-
chanical system in the form of a free energy functional. Constitutive relations are derived by taking
series expansions of the free energy functional with respect to the independent fields. First order
expansion terms yield linear constitutive relations and higher order expansion terms yield non-
linear constitutive relations. This document serves as supplement to Section 4 of Sandia Report
SAND2013-7363, Rev 3 [2]. Methods for implementation of kinematic relations of piezoelectric
models and rotation of material principal axes are covered in the supplemented report. Additional
discussion on phase velocity calculation is also presented.

3
Acknowledgment

The author would like to thank Allen Robinson and Sharon Petney for discussions and assistance
concerning ALEGRA-FE .

4
Contents

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2 Free Energy Functionals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
3 Constitutive Relations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.1 Linear Piezoelectricity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.2 Nonlinear Piezoelectricity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
3.3 Poled Remanent State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.4 Symmetry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
4 Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
4.1 Implementation Degree . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
4.2 Input File Material Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
4.2.1 Material Axes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
4.2.2 Stiffness and Permittivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
4.2.3 Linear Piezoelectricity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
4.2.4 Nonlinear Piezoelectricity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
5 Benchmark and Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
5.1 Constitutive Law Benchmark . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
6 Phase Velocity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
6.1 Stiffness Tensor in Electromechanical Materials . . . . . . . . . . . . . . . . . . . . . 29
6.1.1 Transversely Isotropic Materials . . . . . . . . . . . . . . . . . . . . . . . . . . 29
6.1.2 Nonlinear Piezoelectric Materials . . . . . . . . . . . . . . . . . . . . . . . . . 31
6.2 Phase Velocity from the Christoffel Equation . . . . . . . . . . . . . . . . . . . . . . . . 31
6.3 Maximum Phase Velocities in Electromechanical Materials . . . . . . . . . . . . 33

References 37

5
List of Figures

1 Stress and electric field response to strain and electric field loading described in
Table 5.1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2 Phase velocity as a function of θ , 0 = x1 and π2 = x3 . . . . . . . . . . . . . . . . . . . . . . . . 36

6
List of Tables

5.1 Strain and electric field loading path . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

5.2 Starting stress and electric displacement values for each loading step from analyt-
ical calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
6.1 Stiffness coefficients used in Figure 2 for PZT-5A from Morgan Electro Ceramics
[4] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

7
This page intentionally left blank.
1 Introduction

This report documents an update to the piezoelectricity model implemented in ALEGRA-FE [1]
during FY2016. The original model was detailed in Sandia Report SAND2013-7363 [2]. This
report covers the derivation of linear and nonhysteretic nonlinear constitutive relations using a
thermodynamic framework as well as the implementation method of the nonlinear model in the
framework of the original linear piezoelectricity model. A discussion of phase velocity calculation
is also presented.

Section 2 introduces a thermodynamic description of the material through free energy functionals.
Section 3 takes a series expansion of the free energy functional with respect to independent fields to
arrive at the linear and nonlinear constitutive relations, a discussion of material independent sym-
metry arguments is also presented. The implementation of the nonlinear model within the existing
linear framework is described in Section 4. A benchmark confirming correct material behavior is
presented in Section 5. Section 6 provides a discussion on computing the maximum phase velocity
which is used to determine the automatic time steps.

This document will use indicial notation. Lowercase Latin indices range from 1 to 3 and are used
to denote cartesian coordinate axes. Lowercase Greek indices range from 1 to 6 and are used to
denote indices in Voight notation. Greek indices of 1, 2, 3, 4, 5, 6, correspond to Latin indicial
pairs of 11, 22, 33, 23, 13, 12 respectively. Scalar variables are indicated by normal weight font.
Vector and tensor variables are indicated by bold weight font.

9
2 Free Energy Functionals

A phenomenological approach to describing the material response of a system can be achieved by

taking a series expansion of the free energy density that describes the material and fitting the ex-
pansion coefficients to measured values. The simplest free energy for a electromechanical system
is the internal energy u with independent field variables of strain S and electric displacement D and
dependent field variables of stress T and electric field E.

u (S, D) = Ti j (S, D) Si j + Ea (S, D) Da (2.1)

The internal energy density functional would lead to a Strain-Charge constitutive model, however,
ALEGRA-FE is more suited to handle a Strain-Voltage constitutive relation. A free energy density
functional must therefore be developed with independent variables of S and E. This can be accom-
plished through a Legendre transformation of u to produce the new free energy density functional g.

g = u − Ea Da (2.2)

and dg is the infinitesimal form of g.

dg (S, E) = Ti j (S, E) dSi j − Da (S, E) dEa (2.3)

From Equation 2.3, the T and D quantities can be defined as follows:

∂g
Ti j = ∂ Si j
(2.4)
Da = − ∂∂Ega

10
3 Constitutive Relations

The free energy density functional g can be approximated with a phenomenological approach by
taking a series expansion about a state r and fitting the expansion coefficients to observed values.
For piezoelectric materials the state of interest is the poled zero load (T = 0 and E = 0) state.
Expanded to third order terms, the approximate free energy density functional g̃ becomes



g̃ = g̃
r
   
1 ∂ g̃  ∂ g̃ 
 
+ 1! ∂ Si j  ∆Si j + ∂ Ea  ∆Ea ...
r r
    
1 ∂ 2 g̃  ∂ 2 g̃  ∂ 2 g̃ 
  
+ 2! ∂ Si j ∂ Skl  ∆S i j ∆Skl + 2 ∂ Si j ∂ Ea  ∆Si j ∆E a + ∂ Ea ∂ Eb  ∆Ea ∆Eb ... (3.1)
r r r
   
∂ 3 g̃ ∂ 3 g̃
 
∂ Si j ∂ Skl ∂ Smn  ∆Si j ∆Skl ∆Smn + 3 ∂ Si j ∂ Skl ∂ Ea  ∆Si j ∆Skl ∆Ea
 
 
r r
1 
 
+ 3!   
...

3 3
3 ∂ Si j ∂∂ Eg̃a ∂ E  ∆Si j ∆Ea ∆Eb + ∂ Ea ∂∂ Eg̃ ∂ Ec  ∆Ea ∆Eb ∆Ec
   
b b
r r


Where ∆ implies a deviation from the state R (∆A = A − Ar ). From Equation 3.4, the expressions
for T and D become

11
 
∂ g̃ 

Ti j = ∂ Si j  . . .
r
 
∂ 2 g̃  ∂ 2 g̃ 
 
+ ∂ Si j ∂ Skl  ∆Skl + ∂ Si j ∂ Ea  ∆Ea . . .
r r
  
1 ∂ 3 g̃ ∂ 3 g̃ 1 ∂ 3 g̃
  
+ 2 ∂ Si j ∂ Skl ∂ Smn  ∆Skl ∆Smn + ∂ Si j ∂ Skl ∂ Ea  ∆Skl ∆Ea + 2 ∂ Si j ∂ Ea ∂ Eb  ∆Ea ∆Eb . . .
  
r r r
 (3.2)
∂ g̃ 

Ea = − ∂ Ea  . . .
r
 
∂ 2 g̃  ∂ 2 g̃ 
 
− ∂ Si j ∂ Ea  ∆Si j − ∂ Ea ∂ Eb  ∆Eb . . .
r r
 
1 ∂ 3 g̃ ∂ 3 g̃  ∆Si j ∆Eb − 1 ∂ 3 g̃  ∆Eb ∆Ec . . .
 
− 2 ∂ Si j ∂ Skl ∂ Ea ∆S i j ∆Skl − ∂ Si j ∂ Ea ∂ Eb  2 ∂ Ea ∂ Eb ∂ Ec 
r r

3.1 Linear Piezoelectricity

In Equation 3.2, the first and second order expansion coefficients can be reconciled with familiar
terms from linear piezoelectric constitutive equations.

   
∂ g̃ 
− ∂∂Eg̃a 
   
∂ Si j  = Ti j  , = Ea 
r r r r
 
∂ 2 g̃  ∂ 2 g̃ 
 E
 S
∂ Si j ∂ Skl  = ci jkl , − ∂ Ea ∂ E  = κab (3.3)
b
r r

2
− ∂ S∂i j ∂g̃Ea 

= eai j
r

where cEijkl , κab

S , and e
ai j are respectively the stiffness tensor at constant electric field, the permit-
tivity tensor at constant strain, and the strain-voltage piezoelectric tensor. Therefore, the first and
second order expansion terms yield the linear piezoelectric constitutive relations for the strain-
voltage form.

12
 

Ti j − Ti j  = ∆Ti j = cEijkl ∆Skl − ebi j ∆Eb + . . .
r

 (3.4)
Da − Da  = ∆Di j = S ∆E + . . .
eakl ∆Skl + κab b
r

3.2 Nonlinear Piezoelectricity

The third order expansion terms describe nonlinear electromechanical behavior and can be clas-
3
sified by their function. Non-cross terms are easily identified: ∂ Si j ∂∂S g̃∂ Smn r describes the strain

kl
3
induced nonlinearity in the stiffness tensor ĉEijklmn ; similarly, − ∂ Ea ∂∂ Eg̃ ∂ Ec r describes the electric

b
S . Circumflex ˆ and caron ˇ variable accents are respec-
field induced nonlinearity in permittivity κ̌abc
tively used to denote nonlinearities in strain and electric field.

Third order cross terms

 can be identified as nonlinearities in more than one constitutive Erelation.
∂ 3 g̃
The term ∂ Si j ∂ S ∂ Ea r describes both the electric field induced nonlinearity in stiffness čai jkl and
kl
∂ 3 g̃

the negative strain induced nonlinearity in the piezoelectric tensor −êai jkl . Similarly, 
∂ Si j ∂ Ea ∂ Eb r
describes both the negative electric field induced nonlinearity in the piezoelectric tensor −ěabi j and
S .
the negative strain induced nonlinearity in the permittivity −κ̂abi j

Although the third order expansion cross terms can be identified as nonlinearities in either consti-
tutive relation, this document will refer to them as nonlinearities in the piezoelectric behavior. The
nonlinear constitutive relation for a piezoelectric material derived from a third order expansion of
the free energy g becomes:

   
E 1 E 1
∆Ti j = ci jkl + ĉi jklmn ∆Smn ∆Skl − ebi j + êbi jmn ∆Smn + ěbci j ∆Ec ∆Eb
2 2
| {z } | {z }
c̄i jkl ēTbi j

    (3.5)
1 S 1 S
∆Da = eakl + êaklmn ∆Smn + ěabkl ∆Eb ∆Skl + κab + κ̌abc ∆Ec ∆Eb
2 2
| {z } | {z }
ēD
akl
κ̄ab

Where given a ∆S and ∆E, the quantities c̄ (∆S), ēT (∆S, ∆E), ēD (∆S, ∆E), and κ̄ (∆E) are respec-
tively the tangential stiffness tensor, piezoelectric tensor for stress, piezoelectric tensor for electric

13
displacement, and permittivity tensor.

3.3 Poled Remanent State


About the poled state, a piezoelectric material has zero remanent stress and electric field (Tr = 0
and Er = 0), which yields ∆T = T and ∆E = E. If the material
 is ferroelectric,
 the material posses
a nonzero remanent strain and electric displacement (S r 6= 0 and D r 6= 0). However, if the fer-
 
roelectric material is loaded within the linear range the remanent strain and electric displacement
will not change. In experiments, the remanent values can therefore be zeroed out in the instrumen-
tation and ignored in the simulation as long as the material behaves linearly. With this assumption,
∆S = S and ∆D = D. Equation 3.5 is simplified to

   
E 1 E 1
Ti j = ci jkl + ĉi jklmn Smn Skl − ebi j + êbi jmn Smn + ěbci j Ec Eb
2 2
| {z } | {z }
c̄i jkl ēTbi j

    (3.6)
1 S 1 S
Da = eakl + êaklmn Smn + ěabkl Eb Skl + κab + κ̌abc Ec Eb
2 2
| {z } | {z }
ēD
akl
κ̄ab

3.4 Symmetry

The nonlinear coefficient tensors ĉi jklmn , êai jkl , ěabi j , and κ̌abc respectively have 36 , 35 , 34 , and 33
elements. However, very few of these elements are independent and nonzero. Without considering
material specific symmetries, minor symmetry can be applied to the mechanical indices (Ai j =
A ji → Aα ) and a single Voight notation index can be used to replace two tensor indicial pairs.
Equation 3.6 takes the form of

   
E 1 E 1
Tα = cαβ + ĉαβ γ Sγ Sβ − ebα + êbαγ Sγ + ěbcα Ec Eb
2 2
| {z } | {z }
c̄αβ ēTbα

    (3.7)
1 S 1 S
Da = eaβ + êaβ γ Sγ + ěabβ Eb Sβ + κab + κ̌abc Ec Eb
2 2
| {z } | {z }
ēD
aβ
κ̄ab

14
with reductions in the element count of the coefficient tensors ĉαβ γ , êaαβ , and ěabα to 63 , 3 62 ,

32 (6). Major symmetry exhibited in the mechanical material (between mechanical indices Aαβ =
Aβ α ) and electrical material (between electrical indices Aab = Aba ) reduces the number of unique
elements within ĉαβ γ , êaαβ , ěabα and κ̌abc respectively to 56, 63, 36, and 10.

Further reduction in the number of unique elements of the nonlinear coefficients can be made
through the application of material specific symmetry. The analytical method of determining the
relations and independence of the material property coefficients is to equate the material prop-
erty tensor to the same material property tensor that is rotated by the material’s crystal symmetry
operations and solving for the system of equations.

fi0jklm = Aai Ab j Ack Adl Aem fabcde (3.8)

For the 5th rank material tensor f in Equation 3.8, if the material symmetry class is 6, then the
rotation operator A could take the form described in Equation 3.9 below.

cos 60◦ sin 60◦ 0

 
 
◦ cos 60◦ 0 
 
A=
 − sin 60  (3.9)
 
0 0 1

In the case of the material possessing transverse isotropic symmetry (commonly observed in piezo-
electric materials), the number of independent material parameters for the nonlinear piezoelec-
tric tensors can be greatly reduced by applying isotropic symmetry arguments. For a transverse
isotropic material with the isotropic plane normal to the 3 axis, symmetry class 4 or 6 operators
can be used in the 1- and 2-axes plane. While material is technically hexagonal and 6-fold sym-
metry should be used, using 4-fold symmetry is much easier and can be done using the inspection
method, otherwise the method of Equation 3.8 must be used. For ê and ě respectively, the method
of 3.8 would require solving systems of up to 108 and 54 linear equations.

Applying a 4-fold symmetry operation in the 1- and 2-axes plane, the transformation of the Carte-
sian and Voigt indices are as follows:

15
    
 1   −2 
a= 2 → a0 = 1
3 3
   

   
1 2
(3.10)

 
 
 

2 1

 
 
 


   
   
3 3
  
α= 0
→ α =

 4 
 
 5 

5 −4

 
 
 


 
 
 

6 −6
   

Under 4-fold symmetry, the array of equivalent indices for the nonlinear piezoelectric tensors with
components êaαβ and ěabα are:

  


 −222 −221(−212) −223 −225 224 226

symm −211 −213 −215 214 216

  

  
 
symm −233 −235


  symm 234 236 

a = 1,  
symm symm −255 254(245)





 symm 256 


−244 −246
 
symm symm symm symm



  

symm symm symm symm symm −266









  



 122 121(112) 123 125 −124 −126
−114 −116

symm 111 113 115

  

  
 
−134 −136



a = 2, 
 symm symm 133 135 

êaαβ =  symm symm symm 155 −154(−145) 156 
   (3.11)
144 −146
 
symm symm symm symm



  




 symm symm symm symm symm 166




  
−324 −326



 322 321(312) 323 325

−314 −316

symm 311 313 315

  

  
 
symm symm 333 335 −334 −336

  


a = 3,  
symm symm symm 355 −354(−345) −356

  

  
  



  symm symm symm symm 344 346 




 symm symm symm symm symm 366



16
   


 222 −212(−122) −232
α = 1,  symm

112 132
  

 

symm symm 332









  
221 −211(−121) −231





α = 2,  symm 111 131

  

 




 symm symm 331




  
223 −213(−123) −233





α = 3,  symm 113 −133

  

 

symm symm 333






ě =  

 225 −215(−125) −235 (3.12)

α = 4,  symm

115 135
  

 

symm symm 335









  
−224 214(124) 234





α = 5,  symm −114 −134 

  






 symm symm −334




  
−226 216(126) 236





α = 6,  symm −116 −136 

  



symm symm −336







From Equations 3.11 and 3.12, the values and relationships of the coefficients can be deduced
through direct observation. For example, ê111 = −ê222 but ê222 = ê111 therefore the only possible
solution is ê111 = ê222 = 0. Similarly, ê311 = ê322 and ê322 = ê311 , therefore ê311 = ê322 . Lastly,
ê316 = −ê326 and ê326 = −ê316 , therefore ê316 = −ê326 . Using this method, the relationships of the
coefficients can be determined as shown below.

17
   


 0 0 0 ♦  0

symm 0 0   0

  

  
  

  symm symm 0 0 0 0 


a = 1,  




 symm symm symm 0 0 0 

  
symm symm symm symm 0 0



  

symm symm symm symm symm 0









  



 0 0 0  () 0

symm 0 0  (♦) 0

  

  
  


a = 2, 
 symm symm 0 0 0 0 

êaαβ =  symm symm symm 0 0 0 
 


 (3.13)
symm symm symm symm 0 0



  




 symm symm symm symm symm 0




  
4



 N H 0 0 I

4 H (I) 

symm 0 0

  

 
 
symm symm O 0 0 0 

 


a = 3,  
symm symm symm C 0 0 

 

  
  



  symm symm symm symm C 0 




 symm symm symm symm symm B



18
   


 ◦ • 0
α = 1,  symm ×

0 
  



symm symm ?









  
× (•) 0





α = 2,  symm ◦ 0 

  






 symm symm ?




  
 0 





α = 3,  symm  () 

  



symm symm







ěabα =  

 0 0  (3.14)

α = 4,  symm

0 } 
  



symm symm 0









  
0 0 }





α = 5,  symm 0 () 

  






 symm symm 0




  
♣ ♠ 0





α = 6,  symm (♣) 0 

  



symm symm 0







In Equations 3.13 and 3.14, each symbol represents a unique value and symbols in parentheses
represent equal magnitude but opposite sign ((♥) = −♥). Using 4-fold symmetry, the number of
independent coefficients drops to 11 for êaαβ and 11 for ěabα . Since the symmetry operation used
was 4-fold instead of 6, the actual number of nonzero parameters could be fewer.

19
4 Implementation

4.1 Implementation Degree

The current implementation of nonlinear piezoelectricity in the piezoelectricity model of ALEGRA-

FE does not consider the effects of ĉαβ γ and κ̌abc . Only the nonlinear effects of êaαβ and ěabα
are implemented. The constitutive equation solved by the material model is a modified form of
Equation 3.7.

 
1
Tα = cEαβ Sβ − ebα + êbαγ Sγ + ěbcα Ec Eb
2
| {z }
ēTbα

  (4.1)
1 S E
Da = eaβ + êaβ γ Sγ + ěabβ Eb Sβ + κab b
2
| {z }
ēD
aβ

Given the S and E of a load step, tangential ēT and ēD coefficient tensors are computed for a frame-
work identical to a linear piezoelectric constitutive model.

The ê and ě tensors are respectively of dimensions 3×6×6 and 3×3×6 elements and all non-zero
elements must be specified by the user in the input file. The major symmetries described in section
3.4 are not automated, instead the user must take care to insure the coefficient tensors satisfy the
major symmetries shown in Equation 4.2.

êaαβ = êaβ α
(4.2)
ěabα = ěbaα

4.2 Input File Material Parameters

4.2.1 Material Axes

The orientation of the material axes to the global axes is specified by a rotation matrix Ri j . The
columns of the matrix R are composed of unit vectors of the material coordinate axes in terms of

20
the global reference frame. The model requires an input of 2 material coordinate axes m1i and m2j
with the third m3k computed as the cross product of m1 and m2 .

m3k = εki j m1i m2j (4.3)

Where εi jk is the Levi-Civita tensor. Vectors components of m1i and m2j are respectively denoted
by input parameters of M1# and M2#, where the # corresponds to tensor indices.

Material axis m1i input parameters:

M11, M12, M13

Material axis m2j input parameters:

M21, M22, M23

4.2.2 Stiffness and Permittivity

Only non-zero components for the upper triangular portion of the symmetric 6 × 6 stiffness tensor
cαβ are required as input and are denoted by input parameters of the form C##. Similarly only
non-zero components of the upper triangular portion of the symmetric 3 × 3 permittivity tensor κab
are required as input and are denoted by input parameters of the form EP##.

Stiffness tensor cαβ input parameters:

C11, C12, C13, C14, C15, C16,

C22, C23, C24, C25, C26,
C33, C34, C35, C36,
C44, C45, C46,
C55, C56,
C66

Permittivity tensor κab input parameters:

EP11, EP12, EP13,

EP22, EP23,
EP33

21
4.2.3 Linear Piezoelectricity

Non-zero components of the 3 × 6 full linear piezoelectric tensor eaα are required as input and are
denoted by input parameters of the form E###. Despite having 3 indices, the last two of E### are
Voight pairs.

Linear piezoelectric tensor eaα input parameters

E111, E122, E133, E123, E113, E112,

E211, E222, E233, E223, E213, E212,
E311, E322, E333, E323, E313, E312

4.2.4 Nonlinear Piezoelectricity

Non-zero components of the full 3 × 6 × 6 nonlinear strain dependent piezoelectric tensor êaαβ
and the full 3 × 3 × 6 nonlinear electric field dependent piezoelectric tensor ěabα are required as
input and are respectively denoted by input parameters of the forms ES### and EE###. The #
corresponding to Voight (lower case greek) indices range from 1-6.

Nonlinear strain dependent piezoelectric tensor êaαβ input parameters:

ES111, ES121, ES131, ES141, ES151, ES161,

ES211, ES221, ES231, ES241, ES251, ES261,
ES311, ES321, ES331, ES341, ES351, ES361,

ES112, ES122, ES132, ES142, ES152, ES162,

ES212, ES222, ES232, ES242, ES252, ES262,
ES312, ES322, ES332, ES342, ES352, ES362,

ES113, ES123, ES133, ES143, ES153, ES163,

ES213, ES223, ES233, ES243, ES253, ES263,
ES313, ES323, ES333, ES343, ES353, ES363,

ES114, ES124, ES134, ES144, ES154, ES164,

ES214, ES224, ES234, ES244, ES254, ES264,
ES314, ES324, ES334, ES344, ES354, ES364,

ES115, ES125, ES135, ES145, ES155, ES165,

ES215, ES225, ES235, ES245, ES255, ES265,
ES315, ES325, ES335, ES345, ES355, ES365,

ES116, ES126, ES136, ES146, ES156, ES166,

ES216, ES226, ES236, ES246, ES256, ES266,
ES316, ES326, ES336, ES346, ES356, ES366

22
Nonlinear electric field dependent piezoelectric tensor ěabα input parameters:

EE111, EE112, EE113, EE114, EE115, EE116,

EE211, EE212, EE213, EE214, EE215, EE216,
EE311, EE312, EE313, EE314, EE315, EE316,

EE121, EE122, EE123, EE124, EE125, EE126,

EE221, EE222, EE223, EE224, EE225, EE226,
EE321, EE322, EE323, EE324, EE325, EE326,

EE131, EE132, EE133, EE134, EE135, EE136,

EE231, EE232, EE233, EE234, EE235, EE236,
EE331, EE332, EE333, EE334, EE335, EE336

Using symmetry in the 1-2 plane for a transverse isotropic material and 4-fold symmetry argu-
ments, the nonzero coefficient values from Equations 3.13 and 3.14 are limited to

ES114 = ES141 = -ES225 = -ES252

ES124 = ES142 = -ES215 = -ES251
ES115 = ES151 = ES224 = ES242
ES125 = ES152 = ES214 = ES241
ES311 = ES322
ES312 = ES321
ES313 = ES331 = ES323 = ES332
ES333
ES344 = ES355
ES366
ES316 = ES361 = -ES326 = -ES362

EE111 = EE222
EE121 = EE211 = -EE122 = -EE212
EE221 = EE112
EE331 = EE332
EE113 = EE223
EE133 = EE313 = -EE233 = -EE323
EE333
EE134 = EE314 = -EE235 = -EE325
EE234 = EE324 = EE135 = EE315
EE116 = -EE226
EE126 = EE216

23
5 Benchmark and Tests

This section covers benchmark tests.

5.1 Constitutive Law Benchmark

This benchmark tests a single cube element of 1 by 1 by 1 meter domain. For linear elements, the
single element removes inertial factors. This allows the verification that the constitutive parameters
are working correctly in the constitutive relations. In this case, the effects of nonlinearity to e33 is
explored.

Input material parameters are:

$Stiffness coefficients
C11 = 100.0E9
C22 = 100.0E9
C33 = 80.0E9
C12 = 70.0E9
C13 = 70.0E9
C23 = 60.E9
C44 = 15.0E9
C55 = 15.0E9
C66 = 20.0E9

$Piezoelectric coupling coefficients

E311 = -5.00
E322 = -5.00
E113 = 10.0
E223 = 10.00
E333 = 50.0

$Nonlinear piezoelectric coefficients

ES333 = -1000000
EE333 = 5e-5

$ Permittivities
EP11 = 8E-9
EP22 = 8E-9
EP33 = 7E-9
$ Direction cosines
M11 = 1.
M12 = 0.
M13 = 0.

24
 M21 = 0.
M22 = 1.
M23 = 0.
density = 7750

In this benchmark, the material is loaded in strain and electric field along the same axis (3-axis) in
four steps as shown in Table 5.1. All other strain and electric field components are zero.

Table 5.1. Strain and electric field loading path

Step S3 (µ) E3 (MV /m)
1 0 → −50 0
2 −50 0→1
3 −50 → 0 1
4 0 1→0

The electric displacement and stress response to the loading conditions and along the loading axis
can be seen in Figure 1.

The constitutive relations from Equation 4.1, given the loading conditions (E1 = E2 = S11 = S1 =
S2 = S4 = S5 = S6 = 0), for T3 and D3 is shown below:

T3 = cE33 S3 − e33 E3 − ê333 S3 E3 − 12 ě333 E3 E3

(5.1)
1 S
D3 = e33 S3 + 2 ê333 S3 S3 + ě333 S3 E3 + κ33 E3

Figures 1.a and 1.b highlight the effects of loading steps 1 and 3. In steps 1 and 3, the electric field
E3 is held fixed, while the strain S3 is changed. Since the E3 is fixed, the terms associated with
only the electric field can be grouped as constants.

1
T3 = cE33 S3 − ê333 S3 E3 −e33 E3 − ě333 E3 E3
| {z2 }
Constant
(5.2)
S
D3 = e33 S3 + 12 ê333 S3 S3 + ě333 S3 E3 +κ33 E3
| {z }
Constant

Similarly, Figures 1.c and 1.d highlight the effects of loading steps 2 and 4, where the S3 is held
fixed and the E3 is varied. The terms associated with only the strain can be grouped as con-
stants.

25
 (a) (c)

(b) (d)
Figure 1. Stress and electric field response to strain and electric
field loading described in Table 5.1.

T3 = cE33 S3 −e33 E3 − ê333 S3 E3 − 12 ě333 E3 E3

| {z }
Constant
(5.3)
1 S E
D3 = e33 S3 + ê333 S3 S3 +ě333 S3 E3 + κ33 3
| 2
{z }
Constant

Using Equations 5.2 and 5.3 the constitutive relations for each of the loading steps become:

26
 

 T3 = 8e10S3
= −4e6





Step 1: S3 = 0 → −50µ E3 = 0

 D3 = 50S3 − 5e5S32
= −3.75e − 3








 T3 = −4e6 − 100E3 − 2.5e − 5E32
= −129e6



Step 2: S3 = −50µ E3 = 0 → 1e6
D = 9.5e − 9E3 − 3.75e − 3

 3



= 0.75e − 3


 T3 = 1.08e12S3 − 7.5e7 (5.4)
= −75e6





Step 3: S3 = −50µ → 0 E3 = 1e6

 D3 = 100S3 − 5e5S32 + (7e − 3)
= 7e − 3






T3 = −50E3 − 2.5e − 5E32




=0





Step 4: S3 = 0 E3 = 1e6 → 0

 D3 = 7e − 9E3
=0






From Equation 5.4, it is expected that the T3 vs. S3 and D3 vs E3 are linear relations. The D3 vs.
S3 relations are nonlinear in steps 1 and 3 and the T3 and E3 relations are nonlinear in steps 2, and
4. This is observed in the simulation results shown in Figure 1. The effect of ê333 and E3 on the
effective C33 can be seen in Figure 1.a and the effect of ě333 and S3 on the effective κ33 can be seen
in Figure 1.d. The nonlinearity of the effective piezoelectric coefficient due to ê333 and ě333 can be
observed in Figure 1.b and 1.c.

To verify the material model is computing the correct response, the T3 and D3 starting points
of each loading step are calculated in Table 5.1 below and shows agreement with the simulated
results.

27
Table 5.2. Starting stress and electric displacement values for
each loading step from analytical calculations
Starting T3 (MPa) Starting D3 mC/m2


Step
1 0 0
2 -4 -3.75
3 -129 0.75
4 -75 7

28
6 Phase Velocity

Correct automatic time stepping in ALEGRA-FE is dependent on the maximum phase velocity
within the material in any direction. The following section walks through the process of determin-
ing the maximum phase velocity given constitutive parameters.

6.1 Stiffness Tensor in Electromechanical Materials

The phase velocity of a material is directly dependent on the stiffness tensor. In electromechanical
systems, the effective stiffness for a material is dependent on the electrical boundary conditions.
For finite element solvers, the constitutive relation of an electromechanical material is typically
expressed in the Stress-Charge form. Shown in Cartesian indicial notation, the Stress-Charge
constitutive relation is:

Ti j = cEijkl Skl − ebi j Eb

S E (6.1)
Da = eakl Skl + κab b

The constitutive relation can also be described in the Stress-Voltage form:

Ti j = cD
i jkl Skl − qbi j Db

S

−1 (6.2)
Ea = −qakl Skl + κab Db

Where qakl = (κab )−1 ebkl . The stiffness tensors cE and cD describe the electrically unstiffened and
stiffened stiffness tensors respectively. cE describes the stiffness of the material under constant
electric field (constant voltage boundary) conditions and cD describes the stiffness of the material
under constant electric displacement (constant charge boundary) conditions. The relation between
the two is:

−1
cD E
i jkl = ci jkl + ebi j (κab ) eakl (6.3)

6.1.1 Transversely Isotropic Materials

If expressed in the principal material coordinates of a transversely isotropic electromechanical

material, the material parameters are (in Voight notation):

29
 cE11 cE12 cE13
 
0 0 0
 
cE12 cE11 cE13
 

 0 0 0 

 
 

 cE13 cE13 cE33 0 0 0 

cE = 




 0 0 0 cE44 0 0 

 
 

 0 0 0 0 cE44 0 

 
0 0 0 0 0 cE66
 
0 0 0 0 e15 0
 
 
 0 0 0 e15 0 0 
e = 



 
 e31 e31 e33 0 0 0 

 
κ11 0 0
 
 
 0 κ11 0  (6.4)
κ = 



 
 0 0 κ33 

cD = cE + et κ −1 e

e2 e2
 
cE11 + κ31
33
cE12 + κ31
33
cE13 + e31κ33e33 0 0 0
 
 
 cE + e231 e2
 
 12 κ33 cE11 + κ31
33
cE13 + e31κ33e33 0 0 0 

 
 
 cE + e31 e33 cE + e31 e33 cE + e233
 
 13 κ33 13 κ33 33 κ33 0 0 0 

 
=  
 e2 

 0 0 0 cE44 + κ15
11
0 0 

 
 
2
E + e15
 

 0 0 0 0 c44 κ11 0 

 
 
 0 0 0 0 0 cE66 

30
Since the maximum phase velocities of of bulk and shear waves scale with the c of the material,
the phase velocities from the electrically stiffened material will always be higher. Therefore, cD
should be used to calculate the upper bound for phase velocities.

6.1.2 Nonlinear Piezoelectric Materials

For models with nonlinear electromechanical coupling as described by Equation 4.1 of section 4.1,
the cD takes the form of

t
cD = cE + eT (κ)−1 eD (6.5)

6.2 Phase Velocity from the Christoffel Equation

This section will walk through the derivation of the Christoffel equation (adapted from Lane, C.
2014 “The Development of a 2D Ultrasonic Array Inspection for Single Crystal Turbine Blades”)
[3] which calculates the phase velocities for a specified direction.

The equilibrium equation for a for a material with mass density ρ is given by:

Ti j, j + ρ fi = ρ üi
(6.6)

and in the absence of a body force

Ti j, j = ρ üi
(6.7)

Where, given a displacement field u

Ti j = ci jkl Skl
(6.8)
1


Skl = 2 uk,l + ul,k

Substituting Equation 6.8 into 6.7 yields:

ci jkl Skl, j = ρ üi

(6.9)
ci jkl 12 uk,l + ul,k


,j
= ρ üi

31
Due to minor symmetry within the c tensor, this becomes

ci jkl uk, jl = ρ üi

(6.10)

The solution for u takes the form of a harmonic function

uk = Apk ei(Kk xk +ωt)

(6.11)

Where pk is the polarization of the wave function. The second derivatives with respect to xi and t
are

uk, jl = −K j Kl Apk ei(Kk xk +ωt) = −K j Kl uk

(6.12)
ük = −ω 2 Apk ei(Kk xk +ωt) = −ω 2 uk

Substituting Equation 6.12 into 6.10,

−ci jkl K j Kl uk = −ρω 2 ui

(6.13)

The vector Ki can be expressed as Kni where ni is the unit vector in the direction of wave propaga-
tion and K is the wave number. Equation 6.13 can be simplified to

ci jkl K 2 n j nl uk = ρω 2 ui
(6.14)

The phase velocity C is defined as

ω
C = K (6.15)

Defining δi j as the Kronecker delta, Equation 6.14 can be expressed as

ci jkl n j nl uk = ρC2 δik uk

(6.16)
ci jkl n j nl − ρC2 δik uk =
 
0

32
Which is the Christoffel equation. For any given wave propagation direction n, if Aik = ci jkl n j nl
and λ = ρC2 , it is possible to see that the Christoffel equation is a simple eigen-value prob-
lem

[A − λ I] u = 0 (6.17)

Where the phase velocities are

q
λ (6.18)
C= ρ

There are up to three unique eigen-values and eigen-vectors of A. The eigen-value associated with
the eigen-vector parallel to the wave direction is related to the longitudinal phase velocity. The two
eigen-values associated with eigen-vectors perpendicular to the wave direction are the shear phase
velocities.

6.3 Maximum Phase Velocities in Electromechanical Materials

In the material coordinates of a transversely isotropic piezoelectric material, the stiffness tensor
takes the form described in Equation 6.4. Since c11 , c33 > c12 , c13 > c44 , c66 , the longitudinal mode
eigen-values of A are maximized when n is perpendicular [1, 0, 0] or parallel [0, 0, 1] to the direction
of polarization. Conversely, shear mode eigen-values of A are maximized when n is near θ ≈ π4
where θ is the angle from the isotropic plane.

33
  q 
c11
ρ ,
   0   
c11 0 0 
 Cl 









  
 
  
    q 

  0 c12
n = [1, 0, 0] , A=
 0 c12 0 , Cs1 = ρ ,

    
  
 
 
 

 0 
   
0 0 c13 Cs2
 q 
c13

 

 
ρ

 q 
 π  c33
Cl2 ρ ,
   
c13 0 0

 
 
 


 
 
 

  
 
 
 

 π   q 
  c13
n = [0, 0, 1] , A=
 0 c13 0 , Cs21 = ρ ,

    
  
 
 
 


 π   
0 0 c33
  q 
c13

 C2 
   

s2 
ρ


 
c11 + c13 0 c44
 
n = [1, 0, 1] √12 1
 
, A= 2 0 c12 + c13 0 
 ,
 
c44 0 c13 + c33
 v   
c11 + 2c13 + c33 +
 u 

 u 

 u   

 u1  , 


 t 4ρ q 

+ c211 + 2c11 c33 + c233 + 4c244

 

 π  
 

Cl4
  
 


 
 
 


 
 
 q 

1
   
(c12 + c13 ),
 π
  
4 = 2ρ
 C s1   

 
 
 v  

    
 C π4 c11 + 2c13 + c33 +

 
 
 u 

  u 
s2 
 


u 
u1   


 t 4ρ q  


− c211 + 2c11 c33 + c233 + 4c244

 


 


 

 

(6.19)

The longitudinal, shear 1, and shear 2 modes are respectively denoted by subscripts of l, s1 , and
s2 . Phase velocity plots are included below in Figure 2 for a material with PZT-5A parameters
described in Morgan Electro Ceramics report TP226 [4] and reproduced below in Table 6.3:

As long as the material parameters are not such that Cl falls below Cs near θ ≈ π4 , the maximum
r r
cD
11 cD
33
phase velocity of any transversely isotropic material is the larger of ρ and ρ . Note, in the

34
 Table 6.1. Stiffness coefficients used in Figure 2 for PZT-5A
from Morgan Electro Ceramics [4]
Stiffness Coefficient Value (GPa)
c11 121
c33 111
c12 75.4
c13 75.2
c44 21.1
c66 22.8

typical piezoelectric ceramic, cE11 > cE33 , but e33 > e31 . Therefore cD
11 could be larger or smaller
D
than c33 depending on the piezoelectric constants. If multiple material types are present, the upper
bound for phase velocity should be the highest phase velocity of any material.

35
 π
Figure 2. Phase velocity as a function of θ , 0 = x1 and 2 = x3

36
References

[1] M. Wong and A. Robinson, “User Manual for Ferroelectric (FE) Modeling in ALEGRA,”
Technical report SAND2015-5529, Sandia National Laboratories, Albuquerque, New Mexico
87185 and Livermore, California 94550, July 2015.

[2] J. Mitchell and T. Fuller, “Nonlinear Kinematics for Piezoelectricity in ALEGRA-EMMA,”

Technical report SAND2013-7363, Sandia National Laboratories, Albuquerque, New Mexico
87185 and Livermore, California 94550, April 2014.

[3] C. Lane, The Development of a 2D Ultrasonic Array Inspection for Single Crystal Turbine
Blades. Springer International Publishing, 1 ed., 2014.

[4] D. Berlincourt, H. Krueger, and C. Near, “Properties of Morgan Electro Ceramic Ceramics,”
Technical Publication TP-226, Morgan Electro Ceramics.

37
DISTRIBUTION:
1 MS 1323 Allen Robinson, 01443
1 MS 1323 Sharon Petney, 01443
1 MS 1321 James Carleton, 01446
1 MS 1323 John Niederhaus, 01446
1 MS 0889 Wen Dong, 01864
1 MS 0899 Technical Library, 9536 (electronic copy)

38
 v1.40

39
40