Direct Sparse solver recommended for contact assemblies, but why?

[321GO]

Hello guys,

So, why is the Direct Sparse solver recommended for (contact)assemblies?

My (contact)assembly seems to solve fine with the default FFE solver, meanwhile requiring only half the memory and solve time.
So why use Direct Sparse, why is it prefered for contact assemblies?

Thank's in advance to you all, greatly appreciated

 

[EngAddict]

This is how I believe it works. The FFE solver uses the penalty method for calculating contact problems, it is a simple method but not very accurate, and the iteration process can make mistakes depending on where the iteration points lie. Direct sparse solver uses the Lagrange multiplier, this is a much better method and it solves the equations directly.
I use direct sparse almost all the time since I can reply on the results (have had some dodgy looking stress results when using FFE). In some applications, even large models, it can be quicker than FFE, depending on the setup.

 

[321GO]

Appreciated, thank you.

 

[kellnerp]

FFE is for static linear problems. To solve contact additional assumptions are made.

Direct Sparse is capable of iteration and so can solve non-linear problems.

TOP
CSWP, BSSE

http://www.engtran.com

http://www.niswug.org

"Node news is good news."

 

[321GO]

Thank you EngAddict/kellnerp,

kellnerp,

is this basically the same as using FFE with "use large displacement" checked? Or is then the whole simulation non-linear, not only the contact(as opposed to using
linear with Direct Sparse)?

 

[EngAddict]

Large displacement is not a proper non-linear solver. It applies a percentage of the load and runs the simulation, then applies a bit more and runs again, etc. It adds to the previous results in order to simulate a non-linear analysis.