Is it a bug?

[benliangxiao]

Hi,

I'm running time history analysis for a large model. It's getting stuck at one step without showing any error or convergence message and it lasts forever.
It seems that it's running the command "set ok [analyze 1 $dt]" forever.
This problem also happened to my other relatively simpler models before. With different integrators or time steps, the problems were overcome.
But for this large model, it always has this problem even if I have tried other integrators.

Thanks,
Xiao.

[benliangxiao]

BTW, I'm using the latest version 2.4.4.

[fmk]

Ben,

are you using the force beam column, i.e. element type "nonlinearBeamColumn", "forceBEamColumn" , or "beamWIthHinges"? if so you are getting stuck in the internal element iterations, use forceBeamColumn command and look at the -iter option.

[benliangxiao]

Thank you. Yes, the beamWithHinges is used.
The problem was solved by changing constraints from Transformation to Penalty.
Did you mean change to other hinge integration methods?

fmk wrote:
> Ben,
>
> are you using the force beam column, i.e. element type
> "nonlinearBeamColumn", "forceBEamColumn" , or
> "beamWIthHinges"? if so you are getting stuck in the internal
> element iterations, use forceBeamColumn command and look at the -iter
> option.

[fmk]

no .. it was change the element command to use the forceBeamElement and use the -iter option.

[benliangxiao]

Did you mean in this case I should use the command "element forceBeamElement" instead of "element beamWithHinges"? For the -iter option, the default maximum number of iteration is 1 and the default tolerance is 1e-16, should I change to other values.

Many thanks,
Xiao.


fmk wrote:
> no .. it was change the element command to use the forceBeamElement and use
> the -iter option.