Mentat (and other post processor) may plot negative values for Von Mises stress due to extrapolation.
By default the stress, which are calculated and stored in the result file (t16 or h5) by Marc, are extrapolated to nodes by Mentat. This extrapolation may produce negative values :
After extrapolation, averaging is done : each nodes connected to, for example, 4 elements will get 4 different values. Mentat computes the average at each nodes and then creates the fringe plot according to these averaged values.
The extrapolation and averaging settings are managed by the following menu :
Using Translate as extrapolation method will not produce negative values because in that case, integration point values are affected directly to node values :
Thank you for your quick and comprehensive answer. Now it all makes sense
To prevent negative values, one could work with a finer mesh. In this case, the calculated values of the integration points would no longer differ so greatly and the risk of extrapolation to negative values is reduced.
