Hi everyone,
I am struggling with some tough problem so come here seeking help. I am writing a self developed explicit dynamic solver (central difference method)code. it is based on the previous static procedure, so basically i didnot change much on the element formulations(Keep them fully integration). when i run cases to validate, just figure out that for some cases it is not stable or need a really small step camparing the typical defined dt=L/c (c: sound speed in the medium). Like, the case in high temperature. The element i m using is a traditional Tim*** beam-column element, and all the material-based integration(stress,strain,ect.) has been proved working well in previous static procedure.
so can any one help me to identify the reason of this instability and cause of the over-tiny step? or is it not compatible to use full integration in explicit integration solver? or there is something i missed?
Thanksssss
Re: Explicit Integration
We'll need a few more details to diagnose the problem.
1) What is L? Is it the size of the element or the length of the beam?
2) What happens at high temperatures in your calculation? Does the modulus change? In that case the value of c changes too.
A CFL number of 1 will not necessarily lead to a stable solution. I've found that, for stability, one needs to go to 0.5 or lower in many situations.
-- Biswajit
In reply to Re: Explicit Integration by Biswajit Banerjee
Hi,Biswajit, Thanks
Hi,Biswajit,
Thanks a lot for reply. yes, the L upthere means the lenght of element as i am just using the bending beam element.
Under the high temperature, the modulous is decreasing, so as a consequance, the c should be reducing as well, I think. But as the material steps into plastic and nonlinear story, then how the critical step change would be a issue. Is there any robustness self adaptive time step control for explicit scheme?? I have using a CFL number as 0.4, actually.
Looking forwards to you reply,
thanksssss