Large, Rapid loads causing times steps to be way too small. Any solutions to this?

[SwimBikeRun4342]

Hello,

So I am running a dynamic implicit analysis. The time dependent loads are defined by a tabular amplitude. My typical procedure is:
I receive data for a time-dependent load
I input that as a tabular amplitude
I run a several dynamic implicit simulations for several different sets of data sent to me.


However, there is now a problem with this simple procedure. The data now represents a time-dependent load that has a very large ampilitde and acts immediately over a very short time span. As you would expect, this results in extremely small time steps for my dynamic implicit analysis. This is a huge problem because simulations that are suppose to take 20 minutes now will take days. IS there any way I could increase the size of the time steps(even if I must compromise accuracy slightly)? Is there a mass scaling or tiem scaling option for dynamic implicit analysis?

Any help would be great - I really need to speed up these time steps.


Note - the only reason the simulation is running slow is due to a large, rapid load that is no applied to the object. The mesh, material properties, etc are fine.




Thanks!

 

[IceBreakerSours]

Have you considered Explicit analysis?

Are you new to this forum? If so, please read these FAQ:

http://www.eng-tips.com/faqs.cfm?fid=376

http://www.eng-tips.com/faqs.cfm?fid=1083

 

[SwimBikeRun4342]

IceBreakerSours, thanks for the suggestion. Would explicit analysis work better for this type of dilemma? Also how does mass scaling effect the accuracy of the results? I am only concerned about the stress distribution, not the dynamic properties.

Again, it is very important that I find a way to speed up the simulation. Since I need to run multiple simulations, the automatic time steps of approx. 1e-8 seconds are not acceptable.

Any help and/or experience with this type of problem would be appreciated.

Thanks again!

 

[IceBreakerSours]

As you might know, implicit and explicit are just two different numerical schemes of solving a model. So, if you are careful, both solvers should give you the same solution.

As a thumb rule, if a LOT is happening in a very short span of time, implicit solver can not handle it very well. Or, if there are too many nonlinearities, then the implicit solver can spend forever trying to converge; it may be worthwhile to try the Explicit solver. However, one has to be careful in interpreting the results and making sure the energies balance out, the artificial energies are low compared with strain energy, etc. There are examples in the documentation dealing with such issues.

I have not used mass scaling but if an artificially high mass is assigned to the model, then it will have a direct impact on the mechanics.

Are you new to this forum? If so, please read these FAQ:

http://www.eng-tips.com/faqs.cfm?fid=376

http://www.eng-tips.com/faqs.cfm?fid=1083