Introduction
In this article I will give my recommendations for what experiments to run if you want to calibrate a material model for a thermoplastic. I have found a lot of misinformation about this, and here I will try to give a modern answer to this. To make my recommendations are useful as possible, I divided them into 2 different applications:
Slow Monotonic Loading
- Time domain simulations
- Monotonic loading at one strain rate
- Finite strains
General Loading
- Time domain simulations
- Any strain history
- Finite strains
In the following sections I will propose test plans for each of these applications.
Slow Monotonic Loading
Run the following monotonic uniaxial tension tests:
- Uniaxial tension using a dog-bone shaped specimen (e.g. ASTM D638 Type IV).
- Deform the specimen to large strain, insert at least one unloading-reloading segment somewhere in the test.
- Repeat the test 2 to 3 times using different specimens (in order to determine the experimental variability).
This data can be used to calibrate an rate-independent plasticity model.
General Loading
Run the following cyclic uniaxial tension tests:
- Load to 1% strain, hold the strain for 1 min, unload to a force that is close to 0;
- then continue to load to 5% strain, hold the strain for 1 to 10 min, unload to a force that is close to 0;
- then continue to load to 10% strain, hold the strain for 1 to 10 min, unload to a force that is close to 0;
- finally either load to failure in tension, or unload to zero.
- Repeat this experiment 2-3 times (using different specimens) in order to determine the experimental repeatability.
- The testing should be performed using ASTM D638 type IV (or similar) dogbone-shaped specimens.
- All loading and unloading should be performed at a constant engineering strain rate of about 0.01/s to 0.1/s.
This data can be used to calibrate an I1-based non-linear viscoplastic material model (like the PolyUMod TNV model)
- Repeat the testing at multiple temperatures if needed.
- Use MCalibration to calibrate a suitable material model to experimental data.