Indeed (from Abaqus documentation example):
An initial geometric imperfection is used to induce necking in the specimen analyzed with ABAQUS/Standard. In ABAQUS/Explicit the imperfection is not needed because stress wave effects induce necking at the center of the bar.
Same material, BC, Automatic time step, and everything except for the Amplitude shape gives a PEEQ of 1.744. Thatâs a 13% less than the âofficial resultâ.
ÂżWhich is the right one?. Iâm experimenting with some amplitude shapes , but I have no reference to decide which one could be more appropriate.
If you share 2 complete input files sufficient to make such a comparison, I can run them in Abaqus and check the results.
Oh thanks.!! that would be nice.
Do you have Abaqus explicit?.
Linear Amplitude=0.3t [0,1] PEEQ =2.003
Original with a a half wave of a sinus Amplitude=0.3sin(pi()*t/2) [0,1] PEEQ=1.752
Thank you very much for your time.
Abaqus makes more sense. If plasticity in ccx is not strain rate dependent that should be the results. The same for both amplitudes.
ÂżThis looks like a bug in the PEEQ? Âżisnât it?
There shouldnât be such a huge difference.
I think I found an explanation.
When the analysis is set up with automatic time step and without a maximum time step, the time step can be increased automatically if things go fine which can make PEEQ to lose accuracy (PEEQ is an accumulated value over the whole time period).
If one provides a nonlinear custom step function by means of an amplitude (like Sinus), the automatic time step is not free to increase so fast, the time increments are smaller and the final PEEQ is more accurate.
This makes look like the result is rate dependent when it really is Time step dependent. Nonlinear Plasticity requires a convergence study not only in the mesh but also in time.
Thanks again Calc_em. Your reference help me a lot.
Whatâs interesting regarding this, from Abaqus documentation:
If the amplitude varies rapidly (âŚ) you must ensure that the time increment used in the analysis is small enough to pick up the amplitude variation accurately since Abaqus samples the amplitude definition only at the times corresponding to the increments being used.
But time points can be used to make sure that increments will fit the amplitude definitions:
By default, Abaqus/Standard adjusts the time increment (in some cases Abaqus/Standard might violate the minimum time increment specified) to ensure that data are written at the exact time points specified.
I would say that this is for Direct where one defines a time increment, but if one set up an automatic time stepping without maximum time step , the number of points of the amplitude also conditions the way the time increment advances.
Every example Iâm testing confirms there is an amplitude shape dependence of the solution.
I have been using the sinus function for a long time to help me with the nonlinear convergence when linear was failing. It was in an intuitive manner . I didnât really understand that it was not only allowing the convergence but increasing the accuracy of the result.
This time step dependece is caracteristic of explicit solvers if iâm not wrong . I understand from the manual that structural computations in Mecway are implicit.
