Estimated clock time (Dyna)

[Phoenix1986]

Hi, I'm running a series of problems regarding collision model and I've hit a bit of a snag, was hopping someone could provide some assistance.

I've set up a scenario to run in LS-Dyna to test out a base model from which I intended to develop with more relevent data, this model compiled and returned results in approximatly 30 minutes. The issue is that upon completion I altered the material densities and slightly adjusted the impact speed of the vehicle, this has increased the clock time to almost 23 hours. I understand that this initial estimate is rarely acurate however my second analysis is well into its second hour now.

I have about 12 analysis to run and do not have time to wait essentially a day for each one, the models are not particularly complicated so I have no idea why they are suddenly taking so long with such minimal alterations.

Just to be totally clear, there were no additional loads or alterations to geometry, hence my confusion.

T

 

[Drej]

Changing the densities will affect the critical time step for the analysis. If the densities have decreased then you would expect the time step to become smaller, and therefore the analysis run times would extend (assuming the same total model time).

Check the critical time step size for the initial run, and then compare this to later runs with the densities changed. Check also that your slight change in impact speed is actually a slight change and not an order of magnitude.


------------
See faq569-1083 for details on how to make best use of Eng-Tips.com

 

[Phoenix1986]

Thanks, you were right the timestep has been signifficantly increased. The velocity only rose from 2m/s up to 2.25352m/s (5mph), without the altereations in density this adds 11 minutes.

I'm pretty inexperienced with dyna and altering the density just seemed the easiest way to go. For example one of the bodies in the simulation is supposed to be 6600kgs, obtaining the total volume of the model via VSUM I wasa ble to calculate the required density to achieve this overall mass. Surely there must be a way to achieve this desired mass without such prolonged computational time though?

T

 

[Drej]

If you're certain that inertial effects aren't affected you can artificially increase the critical time step to a value to reduce run times ie you can force the code to use a different critical time step. This artificial increase in time step will internally modify (increase) the density of elements to ensure the new time step is met.

However the first check I should do is look at your mesh, and particularly check the shape and size of your elements. The critical time step is based on the smallest element cross diagonal size (for shells, solids) so check that you don't have some very small and/or badly shaped elements (eg tri's) in your model as these will control the time step.

Hope this helps.


------------
See faq569-1083 for details on how to make best use of Eng-Tips.com