improve ansys workbench performance

[tec151]

Hello Ansys Workbench users!

I am actually running a structural calculation with ansys workbench and I was wondering if anyone has some tips to give me in order to improve the performances of ansys in terms of calculation time. In fact ansys is installed on 64 bit system, with 8 Go of RAM end 140 Go of available space on the hard disk. The model has only 157000 element and seams to me that it takes a lot of time to calculate (2 hours since I start the calculation and still running).

many thanks if you can help me with this.

 

[Dave442]

Hi tec151,

If you have more than one processor available you could try running you analysis in parallel. This will require a specific licence though. Depending on your geometry, loads and constraints, you could reduce the cost of your analysis by exploiting symmetry and/or plane stress/strain scenarios. You should also perform a mesh convergence study to determine the minimum number of elements required to ensure reasonable accuracy.

If you describe your analysis in greater detail somebody might provide more specific advice.

Good luck,
Dave

 

[tec151]

Hi Dave,

In fact I need to calculate an assembly made of 2 cylindrical bodys which are hold together via a flange. the flange presents two symmetrical lugs on which I have to apply 10 kN (1st lug)and 15 kN (2nd lug). As I mentioned the model has 157000 elements (I used solid quad elements with mide side nodes). The PC has 6 processor; I specified 4 in the process resolution parameters (via ansys tools). anyway the calculation time is too long; I am wondering if I should specify a different solver type (actually is default) and increase virtual memory.

Thanks again.

 

[rickfischer51]

What kind of model is this? Any nonlinearities? Plasticiity? Large deflections? Sliding contact? How are your components attached (bonded contact, multibody part)?

Take a look at the Solver Output in Solution Information and see what solver WB picked and if it is using multiple substeps. I've seen WB do wacky stuff, like change from sparse to iterative solver by opening the job in a different version, start using multiple substeps because it thinks bonded contact is a nonlinearity, and set a nonlinear material run to one equilbrium iteration and then send me a warning that this is a bad idea. I just ran a model with 156000 elements and it took 3 minutes. Linear static, 18 components connected with bonded contact, all default settings. It chose sparse and one substep.

Hope this helps.

Rick Fischer
Principal Engineer
Argonne National Laboratory