Contact algorithm performance?

[an318is]

Hello,

I am running a Dynamic Explicit contact based simulation that contains 2 deformable bodies and 36 rigid bodies. One of the parts will only ever see contact with 18 of the rigid bodies and the other part will have contact with the other 18. The two deformable parts will also be in contact. The analysis is basically simulating a clamping and welding of 2 sheet metal part. The rigid bodies are the clamps and weld tips. Currently I have used:
*CONTACT
*CONTACT INCLUSIONS, ALL EXTERIOR
to define the contact condition. My question is, is there a computationally faster/more efficient method? As I know what contacts there will be? As best I can tell the approach I have taken might be searching for contact with a part that I know it wont come into contact with. My main reason for asking, is that computation time takes around 8 hours for a simple simulation containing 5 steps. There are approximatly 5200 elements(S4R) and 4400 nodes in this model. To run it as quasi-static It required large step times, on the order of 0.5. Maybe there is some way I can improve this?

I have found little in the user manual about performance aspects to consider. Any assistance would be much appreciated.

Thanks
Adam

 

[rstupplebeen]

Using general contact does have a performance hit. If you know the potential contact pairs and deem the extra user work is worth the computational time savings then go for it. I doubt that anyone will be able to provide you a number like 2X as fast because it is very problem specific. Abaqus does usually show in the what's new information how much general contact is gaining ground on contact pairs.

I am sure that there are many places to improve the analysis time however without more details, pictures, your model or a representative model it is difficult to offer suggestions.

I hope this helps.

Rob Stupplebeen

 

[xerf]

You can try to use *contact exclusions or *contact inclusions with direct specification of the surfaces that may (not) come in contact.