Static vs dynamic application of load

[jball1]

I typically evaluate large steel structures subjected to severe, dynamic loads. The loading is severe enough that it is not uncommon for our analyses to predict significant plasticity.

Because the structures are large and complex, we typically use a global/local approach for evaluation of material nonlinearity. We build a large, less detailed linear-elastic model that we analyze dynamically to produce loads. The peak predicted dynamic loads are then applied to a much more detailed submodel which incorporates material nonlinearity. The submodel is analyzed statically. Our elastic-plastic analyses are typically performed in Abaqus.

I am trying to wrap my mind around whether or not this approach is conservative. The time scale for our dynamic analyses is typically about 100 msec, with load being applied for 5-10 msec. My hunch is that for some of our problems, this approach is actually overly conservative, because the load is not applied long enough for significant plastic strains to develop. The structures we are analyzing are massive. My hunch is that the short duration of the load may not be enough to get the mass moving enough to result in the same plastic strains that we predict when the load is applied statically. Not sure if that makes sense? I’m kind of thinking out loud here.

The reason I am rambling on about this is that we now have the computing power to run a much more detailed model dynamically. I would like to make the case to do this for a particular project where our results from the static analysis are bad enough that it could kill the project. My hunch is that dynamically applying the load may significantly reduce the predicted plastic strains. However, it will take me some time to get that model running, and so if I’m off track here, I don’t want to go on a wild goose chase. I’d appreciate any insight you have.

 

[FEA way]

When deciding whether to use static or dynamic analysis, typically the lowest natural frequency is calculated to obtain the longest vibration period. If the load rate is significantly slower than that, static analysis can be carried out.

Of course, even highly dynamic problems are often analyzed using equivalent (conservative) static loads due to software/time limitations but if you can carry out a fully dynamic study then I would highly recommend it. You won't know for sure that your assumption is valid until you do such a comparison. It may cost you some time but otherwise, it's pretty much free (unlike physical experiments).