Converting a *HYSTERESIS model to PRF form

The keyword *HYSTERESIS was added to Abaqus/Standard over a decade ago to implement a nonlinear viscoelastic network model published by Jorgen Bergström and Mary Boyce.  This model was not heavily used by Abaqus users and was never implemented in Abaqus/Explicit.  However, Verification guide section 2.2.4  shows us four different materials and their material parameters taken from Bergström and Boyce (1998). 

In Abaqus 6.12 we first added the PRF model (Parallel Rheological Framework), which is a generalized Maxwell model intended to capture nonlinear viscoelastic behavior.  The nonlinear viscoelastic behavior of the model is stipulated with the *VISCOELASTIC, NONLINEAR keyword.  In Abaqus 6.13 the LAW=BERGSTROM-BOYCE parameter was added to choose a Bergstrom-Boyce law. The PRF model was extended to Abaqus/Explicit in 6.13 as well.   Since version 6.13 the PRF model with *VISCOELASTIC, NONLINEAR, LAW=BERGSTROM-BOYCE has become a superset of the *HYSTERESIS capability. 

It can be a challenge to determine a set of material model parameters for the PRF model.  Sometimes it can even be a challenge to know the order of magnitude for some of the the material parameters! For those cases where one chooses to use the Bergström-Boyce viscous behavior in the PRF model, it can be helpful to understand how to convert a *HYSTERESIS material model to its equivalent PRF form.   The *HYSTERESIS model (really a combination of *HYPERELASTIC plus *HYSTERESIS), is limited to two networks - an elastic network A and a viscoelastic network B.   

We will use Verification guide section 2.2.4, problem  mbbcdo3mcy_ps_1e_2_cl65.inp as an example (Chloroprene rubber, 16 pph carbon black).  In this example the material model is given as:

*hyperelastic, arruda-boyce, MODULI=LONG TERM
1.71,2.8284,0.01
*hysteresis
4.0,0.03135,5.,-0.9

As we convert this to a PRF material model, the hyperelastic part remains exactly the same.

*hyperelastic, arruda-boyce, MODULI=LONG TERM
1.71,2.8284,0.01
**hysteresis
**   S,          A     ,  m,    C,    E
** 4.0, 0.03135,  5.,  -0.9
*viscoelastic, nonlinear, NetworkID=1, SRatio=0.80, Law=Bergstrom-Boyce
**  A,        m,    C,     E
0.047025, 5.,  -0.9

The *Hysteresis parameters S (stress scaling factor) plays the same role as the PRF parameter SRatio. These parameters tell us how to apportion the elastic stiffness into each of the two networks.  But they are defined differently.  In each case the *Hyperelastic part of the model defines the total stiffness.  The *Hysteresis parameter S is defined as  S =  Sb/Sa and the PRF model parameter SRatio is defined as SRatio = Sb/Stotal  ;  Stotal = Sa+Sb.   So, the PRF SRatio = S / (1+S).  In our example problem the *Hysteresis S=4 becomes PRF SRatio = 4 / (1+4) = 4/5 = 0.80

From the Abaqus documentation, section 22.8.2 we note that the A parameter in each model differs by a factor of 3/2.  The PRF Bergstrom-Boyce parameter A = (3/2) * the A parameter in the Hysteresis model.  In our example problem the *Hysteresis A=0.03135 becomes PRF A = (3/2)*0.3135 = 0.047025    The parameters m, C and E are exactly the same for each model. 

As we convert the verification problem mbbcdo3mcy_ps_1e_2_cl65.inp to a PRF form, we must also realize that we need to change the *STATIC procedure to a *VISCO procedure.  Also, to help in post-processing we added a NSET called N34 (nodes 3 and 4) and history output for the RF2 and U2 at that NSET.   We compare the RF2 and U2 at node 3 in the image at the top of this post.  These runs were performed with Abaqus/Standard 2017.  The input files are included in the attached zip file.

mbbcdo3mcy_ps_1e_2_cl65.inp  - original verification problem file

mbbcdo3mcy_hysteresis.inp  - slightly altered input file (add NSET and history output)

mbbcdo3mcy_PRF.inp  -  PRF version of the problem.

 

Bergström,  J. S., and M. C. Boyce, “Constitutive Modeling of the Large Strain Time-Dependent Behavior of Elastomers,” Journal of the Mechanics and Physics of Solids, vol. 46, pp. 931–954, 1998

 

Comments 0