## Introduction

ABS (acrylonitrile butadiene styrene) is a thermoplastic material with many industrial applications. It is used, for example, to make all Lego pieces! Go Lego! In this article I will examine the accuracy of different material models when it comes to predicting the mechanical response of ABS at large deformations, cyclic loading, at room temperature. As I have shown before, traditional plasticity model are not accurate. Fortunately there are accurate viscoplastic material models that can be used for ABS. The details are given below.

## Experimental Data Used for the ABS Material Model Study

In this study I used monotonic uniaxial tension and compression data at many different strain rates. The following two figures show the experimental stress-strain data. All results shown here were analyzed using MCalibration. The experimental data shows that the material is strain-rate dependent, and the yield stress is higher in compression than in tension.

*Figure 1*. Uniaxial tension data at many different strain rates.

*Figure 2*. Uniaxial compression data at different strain rates.

In the following sections I will go through 8 different candidate material models in order from worst to best!

## Results #8: Ansys MISO Plasticity with Creep

*Figure 3*. The Ansys MISO plasticity model with creep has an average error of 27.6% Not good.This model cannot predict a different yield stress in compression vs tension, and the unloading predictions are bad. As expected, a metal plasticity model is not good here.

## Results #7: Abaqus Johnson-Cook

*Figure 4*. The Johnson-Cook model a somewhat popular material model that is sometimes used for polymers. It can handle different strain rates, but is not very good at predicting the unloading response. The average error in this case is 26.5%.

## Results #6: Bergstrom-Boyce

*Figure 5*. The Bergstrom-Boyce model is one of my favorites (since I developed it!), but it works way better for elastomers than thermoplastics. I do not recommend using it for ABS. The average prediction error is 23.9%.

## Results #5: Abaqus Elastic-Plastic Combined Hardening

*Figure 6*. The elastic-plastic material model with combined hardening does not allow for strain rate dependence or different yield stress in compression. Not good. The average error is 22.9%.

## Results #4: Abaqus PRF 3Net-Yeoh-Power

*Figure 7*. The average error for the Abaqus PRF model with 3 networks (Yeoh + Power flow) is 16.6%. That is sometimes acceptable, but it is also a 100% larger error than the best model…

I also calibrated a 4 network PRF model. the lowest error that I could get with that model was 15.9%.

## Results #3: PolyUMod FEN

*Figure 8*. I don’t usually recommend the PolyUMod FEN model any more. In this case it is more accurate than the Abaqus PRF model, but predictions look a bit odd. We can do better!

## Results #2: Ansys TNM

*Figure 9*. The Ansys TNM model is a good material model for many thermoplastics! In this case the average error in the model predictions is 11.7%. Not too bad.

## Results #1: PolyUMod TNV

*Figure 10*. Once again, the PolyUMod TNV model is the most accurate material model for a thermoplastic. In this case the average error is 8.8%. This is excellent.

## Summary ABS Material Model Study

Here is a comparison between the different material models. The PolyUMod TNV model is the most accurate model for ABS. Note that I have previously shown that the TNV model is also the most accurate model for thermoplastic-elastomers and PEEK). If you have not tried the TNV model, then request a free trial license.