Learning Forum

[Ansyshs]

Hello,

Please i need your help !!

I have a bolt connection from stainless steel( the bolt with pre-load) and i want to use Chboche and a creep model ( i am using modified time hardening). i have a good fitting for chaboche and for the creep curve !! . I didn't get any solution and i don't know what is the problem.

could you help please. i would be grateful for your answer.

[John Doyle]

It looks like you are trying to perform a restart at LS1 with an endtime of 1e-6 seconds. Beyond that, I am not able to tell what the problem is. What is the solver output reporting about the failed convergence?
Regards John

[Ansyshs]

Hello John thank u for ur answer. The problem is that no results . Ansys remained at 20% progress for a long time . if i leave , it may solving for a an hour or more !!

[Ansyshs]

i have tried Time hardening and modified time hardening . the same thing !! no results and 1 % for a long time

[John Doyle]

So there is no solver output at all? It sound like a remote connection issue, but I might be misunderstanding what you are reporting.
Two questions:
When you highlight Solution Information and Solver Output (in details window of Solution information) do you see anything or is it a blank page? If blank page, then it sounds like nothing is being set to the MAPDL solver and perhaps there is some kind of remote connection issue. If you do have something recording in the solver output, what is it saying?
Is this happening only when you introduce nonlinear materials? Does a simple linear elastic test model run to completion successfully?

[Ansyshs]

1- yes it is recording in solver output, but for a long time 1% , see the photo.
2 no its the same thing , just 1% for linear elastic

[Ansyshs]

It is till this time 1 %

[John Doyle]

There is nothing in this solver output that suggests plasticity or creep are the cause of nonconvergence.
I am still not clear about question 2. What happens when you suppress the nonlinear materials? Does it not converge?
The convergence tread on force residual actually looks very good. I do not know why it is taking so long.

[Ansyshs]

Oh man sorry , for the question 2, i have just tried creep with linear elastic, and its done but with a lot of Warnings like thos in the Photo...( Sorry i Cant change the german languege this computer for my university)

[Ansyshs]

[Ansyshs]

I hope you could have some time sir..it's so important :(

[John Doyle]

Sorry for the delay. Thanks for the feedback.
The error message "...creep integration algorithm does not converge..." suggests that the creep strain that is calculated for the elements referenced cannot be reconciled with the stress state that the model is in. It is probably a very large number. If you turn off all the nonlinearities in the model and just run this with the loads as defined, does the model converge? If it does, can you take the stress result at one of the elements referred in the error messages and do a hand calculation to predict the expected creep strain for a given time increment? That might give you some insight as to what the code is struggling with. If your hand calc of creep strain is very high because the peak stress is very high, you need to change something. Are the stress values real or fictitious? I see some sharp corners. Also, is there temperature dependency in the model. Keep in mind that we linearly interpolate the creep coefficients between temperature values and this linear interpolation is not always appropriate. I am not sure if that is an issue here or not, but it might be something that needs to be investigated. There is a way to switch to a logarithmic interpolation with the TBIN command. It can be done in WB with a command object if it appears to be necessary. I hope this helps.