topic Re: Best method of solving the system of equations in Robot Structural Analysis Forum

Best method of solving the system of equations

Anonymous — Wed, 18 Apr 2018 08:15:14 GMT

Hi,

I have steel tube (column) modeled as shell elements. Model have 8496 nodes and 9120 elements.

Bottom part have steel stiffeners and also nodal supports as bolts. Planar support is also nonlinear (uplift is allowed). I want to perform geometrical nonlinear analysis using large displacement and check von Mises stresses..

So, my calculation got stucked on first load case and not converge. I have only 3 load cases.

I am using iterative algoritham, relative code tolerance already increased to 0.001. Can someone advice me what to do. Robot is known as a softwer who have god solvers, perhaps I am using wrong one.

Also, I already watched webinar....

Regards

Re: Best method of solving the system of equations

Artur.Kosakowski — Wed, 18 Apr 2018 13:24:15 GMT

Hi @Anonymous

Have you tried to set the set the full Newton-Raphson method?

If I managed to answer your question(s) press the Accept as Solution button please. This will help other users to find solution(s) much faster. Thank you.

Re: Best method of solving the system of equations

Anonymous — Wed, 18 Apr 2018 13:43:56 GMT

Hi Artur,

Yes I set the full Newton-Raphson method, used iterative solver and increased relative tolerance to 0.005.

Then passed !

Do you think this tolerance give unreliable results ?

Also, when I include geometrical nonlinearity (large dispalcements - also include p-delta) model don't converge ?

Regards

Re: Best method of solving the system of equations

Artur.Kosakowski — Wed, 18 Apr 2018 13:51:01 GMT

Hi @Anonymous

Yes I set the full Newton-Raphson method, used iterative solver and increased relative tolerance to 0.005.

Then passed !

Is it the same if you set the solver as Automatic?

Also, when I include geometrical nonlinearity (large dispalcements - also include p-delta) model don't converge ?

Does it converge for P-Delta alone? If it is a building type structure the large displacement is most likely not needed at all.

Re: Best method of solving the system of equations

Anonymous — Wed, 18 Apr 2018 14:05:33 GMT

Hi Artur,

Is it the same if you set the solver as Automatic?

Yes, it passed.

Does it converge for P-Delta alone? If it is a building type structure the large displacement is most likely not needed at all.

Yes, it passed even I choose P-delta.

Structure type is shell because it is steel column 1200mm diameter, thickness 14mm.

Height of column is 20m. I modeled complete with shell elements, including support stiffeners.

That is reason why I took shell type of structure.

I tried also ''large displacement''+ automatic solver ---- many type 3 instabilities showed nad still calculating... probably won't work..