## Introduction

Santopreneโข is a thermoplastic vulcanizate from Celanese. This interesting material consists of EPDM rubber particles embedded in a PP matrix. This gives the material a rubber-like mechanical properties, but it can still be processed similar to a thermoplastic. In this article I will not focus on micromechanical simulations, but instead I will wear my engineering hat and “simply” discuss how to accurately capture the response of this material using a material model. My goal is to give you the right tools to win the FE simulation game.

## Experimental Data for Santoprene

The experimental data consists of monotonic uniaxial tension at 3 different strain rates, and cyclic stress relaxation with built-in stress relaxation segments. These are pretty smart tests. If you look at the data carefully you will see that the material exhibits clear evidence of Mullins damage. It is also clear that the material is viscoelastic. This insight help reduce the number of material models to explore.

*Figure 1. Experimental data of the Santoprene. (Click the image for high-res).*

I calibrated the following material models using the MCalibration software. Get your own free trial license.

## Results: Ansys TNM

Figure 2 compares the predictions from the Ansys TNM to the experimental data. The average error in the model predictions (NMAD) is 7.9%. The reason this material model is not more accurate is that it was developed for thermoplastics, and therefore does not allow for predictions of Mullins damage.

*Figure 2. Predictions from the Ansys TNM. (Click an image for high-res).*

## Results: Ansys Bergstrom-Boyce Mullins

Figure 3 compares the predictions from the Ansys Bergstrom-Boyce model with Mullins damage to the experimental data. The average error in the model predictions (NMAD) is 5.2%. This is the material model that I developed as part of my Ph.D work at M.I.T, and as is shown in the figure, the model works well for this Santoprene.

*Figure 3. Predictions from the Bergstrom-Boyce model with Mullins damage. (Click an image for high-res).*

## Results: Abaqus 3 Network PRF (Yeoh + Mullins + Power Flow)

Figure 4 shows that a 3 network Abaqus PRF model can predict the response of the Santoprene accurately. The average error in the model predictions is 4.2%. This model is good, but not as good as the winner…

*Figure 4. Predictions from the Abaqus PRF model (3 networks, Yeoh + Power flow). (Click an image for high-res).*

## Results: PolyUMod TNV

The most accurate material model for the Santoprene is the PolyUMod TNV model. The average error in the predictions of this model is 3.6%. Very nice!

*Figure 5. Predictions from the PolyUMod TNV model. (Click an image for high-res).*

## Summary

The following figure shows the predictive accuracy of a number of different material models. The TNV model is the most accurate model for Santoprene.