LINK OF THE PROJECT: https://www.simscale.com/workbench/?pid=8941575546584074889&mi=run%3A379%2Csimulation%3A331&mt=SIMULATION_RUN
Hi, I’m doing a transient simulation and I have some troubles, I’m trying to do a cost-effective simulation and I used Hex-parametric meshing with a level 1 of refinement, I had a succesfull run in ‘Transient test 14’ but for accuracy I wanted to increase the number of cells, so i changed the mesh refinement to level 2 in all the body but when I runned the analysis with fine mesh almost with 750 k nodes the run crash with a fatal error nearly this point:
So I don’t have any idea why it is crashing with fatal error, If you can help me I will appreciate it a lot.
Hello,
From the looks of it, the timestep size is just too high for the current configuration. I was comparing the current step size to what you had for the latest successful simulations, and they are fixed at the same value.
However, since we are using a finer mesh now (effectively lowering the \Delta x value for the Courant number), the Courant number will also increase by a lot. Currently we have a maximum of around 200 and a mean very close to 1.
Some ideas:
- The best option is probably setting adjustable time step to true, and limiting the maximum Courant there. Given the PIMPLE algorithm and the number of correctors, it looks like you are pushing for Co > 1, so perhaps 100 for the maximum would be a good initial point
- The other option would be to lower the fixed timestep size
PS: watch out with the number of layers for transient analyses: 5 of them might be a little too much. Have you already tried using less than that for the initial studies?
Cheers
