About tesla valve stl file EB with AMR build coarse level failed

[pyrimidine]

Tesla2.zip

Hi~ o(￣▽￣)ブ
I am new to AMReX. The case I want to calculate is a Tesla valve.
For the non-refinement situation, the program is OK. However, when the refinement level is 2, the program keeps saying: amrex::Abort::0::Failed to build required coarse EB level 1 !!!

My initial input file is :
geometry.prob_lo = 0.0 0.0 0.1
geometry.prob_hi = 1.0 1.0 0.2
amr.n_cell = 160 160 16

I tried a finer mesh without AMR, and it works well, like the following inputs:
amr.n_cell = 320 320 32
and
amr.n_cell = 640 640 64

I think, since the finer mehs works well, the amr should also work perfectly.

It is possible to tell about the limitation on EB with AMR, is it highly restricted with sharp geometry or multi-body geometry?

Based on the previous discussion shown on the board, I changed the eb2.maxiter to 128, but nothing changed.

Looking forward to your help.

[WeiqunZhang]

It failed because the coarsening of fine EB data in this case resulted in multi-cut or multi-valued cells. Depending what your code does, you might be able to work around it by building the coarse level EB directly. There is an option in EB2::Build, build_coarse_level_by_coarsening. You can set it to false.

https://amrex-codes.github.io/amrex/doxygen/namespaceamrex_1_1EB2.html#a027cac750cb881076f90fdcf0e9c5adc