Material models not saved correctly when switching between them?
I have found something that I think is annoying or possible a bug. When switching between material models in the "Calibrated models" window the latest version is not presented/saved. To reproduce it do this:
1. New file
2. Make a virtual test to run.
3. Choose abaqus linear elastic. change E to 15. Run test to save it in "Calibrated models"
4. Choose abaqus hyperelastic model. Run test to save it in "Calibrated models"
5. Run the linear model from the "Calibrated models" to make it show as the current model.
6. Change E to 20. Run again (I assume the updated model should be saved to "Calibrated models")
7. Run the hyperelastic model.
8. Run the linear model. E is now 15, not 20.
If I load the linear model from "Set material model..." E is 20. Maybe it works as intended.
That behavior is working as intended. Let me try to explain the logic for the behavior:
- When you click "run once" (or "run calibration") in the toolbar in the main window, the current active material model and the the currently active load cases will be evaluated. Once the calculation has been performed, MCalibration checks if the current material model already exists in the "Calibrated Models" tab. If the material model type is not already there, then the current material model is saved in the table of calibrated material models. If there is already a model of the same type in the "Calibrated Models" table, then the current model is only saved there if it has a better fitness than the model that is already in that table.
- When you run a single virtual experiment the fitness value is set to a value of 9e9, so in your case when you run the linear elastic model again, it will get the same fitness as already saved in the table, and the new model is not overwriting the previous linear elastic model.
- If you want to save a material model with the same or worse fitness than what is already in the "calibrated models" tab, then you need to click on the "Locked" check box in the corresponding row in the table of calibrated models. Any material model that is "locked" will not be considered when MCalibration examines if a new material model should be added to the table.
Let me know if this answers your question.
Thank you for the clarification. I have often started with a single tension test to get the general shape correct. And then I have added relaxation tests. The fitness of the tension test could never be beaten and my model was never updated.
One quick way to fix that issue is to click on the following button once you have a material model that you want to store. This will reset the fitness of the selected model to 9e9.