How to speed up Radioss computation?

Allen Lai_22108

Hi all,

I'm doing bicycle brake frictional heat test. Test scenario is a bicycle has constant 12.5 km/hr velocity continued 120 second. In the same time, brake pad press on the brake drum. We want to know during the time the brake drum displacement & temperature. Unit system is "mm kg ms GPa".  I also use solver keyword "-mpi -np 4 " but computation time still need a lot of time. Please someone help me to correct this case.

Thanks!

Adriano A. Koga

Allen Lai_22108 said:

Thank you sir. It's really helpful to me. I had use AMS card to enforce my model timestep and Parith/off card to speed up my computation. AMS card parameter is Tscale 0.9 and Tmin 50 times natural  timesteps. In this case, what timestep set up are you recommend?

i don't know what else can be done here, as your analysis is really long, indeed.

50x increase in timestep looks a lot already.

Check the added mass, anyway.

 

Adriano A. Koga

There is a similar example supplied with the Help for Radioss. It might be helpful, maybe.

 

I'm not Radioss expert, but you could try to accelerate your run by enforcing a fixed timestep, maybe, using the DT/CST solver card inside the engine cards. Trying maybe to enforce a 1e-04 timestep.

Be careful with the added mass (DM/M), in order to keep the model realistic, specially as there's lots of rotation here.

 

Allen Lai_22108

Thank you sir. It's really helpful to me. I had use AMS card to enforce my model timestep and Parith/off card to speed up my computation. AMS card parameter is Tscale 0.9 and Tmin 50 times natural  timesteps. In this case, what timestep set up are you recommend?

Adriano A. Koga

Allen Lai_22108 said:

Thank you sir. It's really helpful to me. I had use AMS card to enforce my model timestep and Parith/off card to speed up my computation. AMS card parameter is Tscale 0.9 and Tmin 50 times natural  timesteps. In this case, what timestep set up are you recommend?

i don't know what else can be done here, as your analysis is really long, indeed.

50x increase in timestep looks a lot already.

Check the added mass, anyway.

 

Allen Lai_22108

Adriano A. Koga_21884 said:

i don't know what else can be done here, as your analysis is really long, indeed.

50x increase in timestep looks a lot already.

Check the added mass, anyway.

 

Thanks sir. You did a big favor !