Adjoint Solver

  • 141 Views
  • Last Post 15 March 2019
JJia posted this 13 March 2019

Hello there, In my attempt to optimize a shape for drag using the adjoint solver, I keep getting this error when I try to 'modify-mesh' in the design tool (polynomials). After a few iterations of the solver, it gives me an error when I modify mesh, stating that 'negative cell volumes detected' and I believe some floating point errors. I have also noted that this only happens when I select 'calculate-design-change', and the terminal outputs: Convergence tolerance was not reached. (After iterating on the 'bounded by surfaces). I think that it has something to do with the 'bounded-by-surfaces' constraint I have applied, but I do not know what settings to adjust to prevent this. Thanks for the reply.

Order By: Standard | Newest | Votes
rwoolhou posted this 14 March 2019

Looks like the change is too big or the remeshing settings aren't able to accommodate the change in shape. 

JJia posted this 15 March 2019

thanks for the reply.

Do you know anything I can do to mitigate the problem?

And I have another problem:

When I am calculating the adjoint solution, it freezes sometimes, and keeps calculating an iteration without any progression (it get's stuck). There are no errors, and when I try to stop it, nothing happens as I think fluent waits for the last iteration to finish. Do you know why this happens?

thanks

rwoolhou posted this 15 March 2019

Not sure: check RAM usage as it may be doing the adjoint calculation and be running low on RAM. Given the model, "freezing" may just mean you need to give it more time: how long is it stuck for, and how much free RAM do you have at that point?

 

JJia posted this 15 March 2019

I leave it for like 8 hours, and it is still on that 1 iteration. And I have lots of ram. If I load the same case and data file, and run adjoint solver, it'll get stuck at the same iteration. I think it has something to do with the mesh, however not sure. And if I change the amount of iterations fluent solver runs for, the iteration where it gets stuck changes, however it is consistent.

Close