I am modeling 3D CT specimen on Abaqus. The assigned material is elastic-plastic and damage is defined by separation laws. I tried to check in the results if the elements behaviour is the one I assigned. I plotted stress vs. strain (respective components as well as principal directions) of integration points and centroids but the response does not match the elementary response I implemented. I saw that Abaqus gives stress and strain in points (integration/nodal/centroid) and I guess these outcomes have to combined in somehow in order to find the material response, but I do not have idea about how to do this. Can someone help me, please?
Hi. I'm not sure I can answer your question, but there are two points that you should check in order to compare :
- Results are at integration points and not interpolated to have correct values
- Material behavior you implemented is uniaxial tension maybe I believe but if you are doing 3D load on CT specimen, the material behavior does not fit the model conditions. Hence no direct correlation.
Regards.
For plasticity, you need to plot (aslways at integration points as Daniel explained) PEEQ vs Mises stress. If this does not agree with your input curve (only look at the data points, they may be connected differently), there is something wrong.
Thanks a lot. Very helpful aid.
As Daniel guessed, I assigned uniaxial behaviour to the material, but if I understand right, there is not a direct correlation between the stress and strain plots (which considers the global response) and the uniaxial response implemented. Do you have suggestions about checking the simulation elementary (or uniaxial) behaviour of the material?
Dear Martin, shoulg the plotting you suggested be fitting the elementary response (I mean plastic true strain and related strength) I assigned, even if I gave the uniaxial behaviour?
Moreover, what are the possible checks to see if the simulation runned well, or rather if the assigned behaviour (material, constrains, among others) has been correctly implemented?
Many thanks again.
The only way to correlate expe with numerics is to have the model correspond to the same load condition as expe to obtain same curve. Else, finding the right point load that correspond. Other way is to calculate theoretical 3D plastic strain from expe curve and compare with numerics, but it is not trivial.
Best.
@Danilo
I'm not sure I understood your question. What I suggest is to create an xy-plot of PEEQ vs Mises for some of your deformed elements. If the datapoints match your theoretical curve, plasticity works fine. PEEQ and von Mises stress are defined so that the uniaxial plasticity curve is matched (or rather the other way round: Plasticity calculation in Abaqus is performed under the assumption that you can for any stress state calculate the von Mises stress and then use the stress-strain curve to get the strain increment). This is explained in some detail in the Theory manual (it's a bit heavy reading, though).
Concerning your general question: There is no unique way of testing a complex simulation. I always suggest to make it as simple as possible in the beginning and then add one feature after another. Always look at the outcome and check whether what you see is what you expect. If it is not, try to find whether your intuition or the model is wrong. Only proceed until you are happy that you understand the simulation results.
I will plot as you suggested. I have to perform cyclic analysis but I started with Static. Many thanks for your suggestion.
- Badges
- Science topic