SimScale CAE Forum

"The machine run out of memory" meshing error

Hey guys,
For the past few weeks i was remaking some half FSAE car simulations (“Mesh test 4” with “w/boundary surface - region” mesh ) of our team and now i came across a new bug that i never seen before. My goal here is to get some good residuals, by making a good mesh and not chaging to much the numerics or simulation control, in order to get some more reliable coefficients. For that i did some research about good meshing and the post “Optimizing mesh quality and Y+ formation on an FSAE car” its like a bible for me and my tests. (I would like to leave my acknowledgment to the PowerUsers who participate on this beautiful and usefull post). In the post they talked a lot about the “Inflate Boundary Layer” refinement and about the “Layer Size” configuration. For me using layer size was better and to get some good results i had to make some spreadsheets of surface leveling to get a good 0.4 - 0.5 inner layer thikness ratio. And thats where my difficulty begins. After testing some confirgurations of levels and bounding box resolution i´ve come to this numbers that worked a lot for me:
Size bounding box resolution
With this input i started to make level testing to get good surface on the most important parts of my project (Sidepods, Tires and Sidewings) and decided that the best levels are:

min/máx 6 for all of then. With that decided i did some Geometry primitives for all the parts of my car, to get a smooth transition between surface - boundary - region and used the “Region Refinement” to combine the level of surface - and region to get the same ratio for the boundary.

So the problem is that everytime i choose to use this config (even if i already did some pretty heavy meshes, like this one, and never had any issue) the “machine run out of memory” keep showing up. I already tried to increase the number of processors for 32 and it didnt work. I cannot get it even higher because i want to save core hours, but i dont know how to make it lighter without losing mesh quality and still trying to implement the ideas that i´m trying here. After solving this i will probably create a new topic about numerics, simulation control and boundary conditions to get some good residuals, but firts i need to solve this unexpected issue.
Here´s the link of the project and the mesh is “w/boundary surface - region” inside “Mesh test 4” :
Thank you all for the attention,

I forgot to mention that i´m completely open to new ideas and changing my mesh, the goal here its just get a good simulation ,thats it. Even if i have to change everything to make it work in a good way. I´m just a newbie who wants to try his ideas here :slightly_smiling_face: :slightly_smiling_face:.


I had a quick look at the mesh settings. A few pointers:

  • Take care with the level of fineness defined for region refinements. As it’s a volume-based refinement, the number of cells can skyrocket if you don’t take care. Try to estimate how many cells each of the region refinements will take.

For example, let’s take the wake:

Total volume: 4.65 m³
Level of refinement 4. Cell size (edge length): 0.0078125 mm
Total number of cells: 9751756

This is almost 10M cells just for one of your 4 region refinements.

There’s a second one prescribed with level 6 refinement (this one is taking about 125M cells).

  • You can try to “escalonate” the region refinements. E.g. you can create a small box around the car with level 4 and then a larger box with level 3 or 2. Applying a very fine discretization in large volumes will cause issues with mesh size!
1 Like

Hey, Ricardo!
Thank you to answer me so quickly and your help. I really don’t know where my head was when i choose this numbers. But the problem here, that made me abandon The “escalonate” is that i want focus my computitional time on the tire - sidepod - tire and beak - sidepod flows. Thats why im creating some unusual boxes.