WF5 Mechanica: new features? 2

WF5 has a lot of new stuff, notably elasto-plastic material laws (with nice material test data fitting), support for snap-through situations in Large Disp, overhauled shell compression definition (including auto detect), "engineering" constraints such as ball/pin/planar, heterogeneous units support, lots of UI enhancements, lots of tools (bearing load at a point, total volumetric heat load, reaction measures at a point, Pro/WELD feature support, much better mesh refinement in contact, new convergence targets in Single Pass Adaptive, element error fringe plots (output), etc... It's a big release.
The combination of LDA + Contact + Plasticity + ... is something being worked on (amongst others) for WF6. Stay tuned!

Thanks, brep, for the quick and informative reply! Yes, those are some very nice enhancements. I'm curious... what makes multi-pass adaptive difficult to do in a LDA? And/or, is there a way to estimate error/accuracy with single-pass adaptive in WF4 or WF5?

Thanks again, David

ps. Sounds like you work at PTC...(just fyi to pass along) on the ptc website, when I search the "Enhancement Details" section for WF5 Mechanica & Advanced Mechanica, it doesn't find anything.

David,

Yes - the what's new details on the website are temporarily missing, the web content people are updating some details so you've caught it at a bad moment!! Thanks for pointing it out though...

OK, there is no *technical* reason why we can't offer multi-pass adaptive with LDA, but hopefully the following sums it up:

First, single-pass has been proven to be faster and more accurate (i.e. efficient) for the vast majority of problems, and has been steadily improved release over release. The new convergence targets in WF5 further bolster this (as it helps overcome a previous objection to using SPA where the area of interest is not necessarily the hottest stress spot..)

Next, for convergence of a nonlinear solution we need to take control over time steps, cutbacks, line searches (Newton tuning) in the solver, and we chose (because of the above point) the single-pass method to optimize for such nonlinear solutions. By just switching to multi-pass the tuned algorithms could be destabilized as the p-levels are successively raised - resulting in long solve times.

Again, there is no technical reason why we could not optimize the multi-pass algorithms separately, but in the interests of getting as much new stuff out to you, and for the fact that single-pass is our 'preference', we have disabled MPA for LDA.

Thanks again, brep, for the helpful reply. And I think I understand how the nonlinear problem would have very different demands on the tuned algorithms.

So for single-pass adaptive, is there a way I can estimate errors in the solution?

Thanks, David

I don't have WF5 installed yet, but what do you mean with LDA...(is single pass adp ??? ) or is it a new convergence mode? ( so SPA, MPa and LDA) ?

MarkCopland,

LDA = Large Displacement Analysis (nonlinear geometry)

When will PTC support LDA for shells? This is an important analysis capability that is currently not present in Mechancia.

Plus, it seems natural that you would want LDA for shells. Given that thin geoemtries are most likely to experience large deformations.

I had a recent project invovling thin geometry being modeled with shell elements and I was prevented from using Mechanica because of its lack of support for LDA with shells.

Steve

I'll hold off committing to a "when" until we have definitive info...

But suffice to say, we are steadily working through a long list of nonlinear enhancements to Mechanica (both Structure and Thermal). You've seen the fruits of some of this labor in WF5 - see the 2nd post in this thread for some details. Unfortunately we have to stage them into the code in order of priority and other technical considerations. I don't want to sound too vague (!!) but what you mention is certainly on this list!

If I posted more direct questions on the PTC User TC portal would be able to see them? I am currently a member of the Mechanica TC.

STeve

Sure