Segmentation fault when using BEP urban parametrization

Hello,
We are running a simulation in the WRF model for the Vitória/ES - Brasil, for ten days time-period.
The model is forced with boundary conditions from GFS with spatial resolution of 1 º. In the simulations, we are testing the single-layer and the multi-layer urban model as the urban physics scheme, however, we find the segmentation fault error when using the BEP model on a nested grid.
Three grids are used with 27km, 9km and 3km end the error ocours only when we acctivate the BEP model in 3km grid. Errors are not found when running witc UCM model or BEP active only on two bigger grids.

When running the model, the pre-processing programs i.e. geogrid, ungrib, metgrid are performed successfully. The real.exe ran successfully but the error occurred on the first time-step when running wrf.exe.
The WRF version is 4.0.

The namelist are attached.

Can you please advise how we can resolve this problem?
Thanks!

If the model crashed at the right beginning, it often indicates either a data issue or a memory issue. However, your namelist.input indicates this is a small case with only 70x70 grids in each domain.
(1) Let's first figure out whether this is a memory issue.
Please tell me how you compile and run WRF model. Also, how large the available memory is in your machine.
if you reduce the grid number but still run triple nesting case, can this case work?
(2) I notice that you use some outdated scheme such as cu_physics =93. Is there any special reason for using this scheme?
(3) Please turn off cumulus scheme for D03 (3km)
(4) Please try the YSU PBL scheme and let me know whether you have the same problem.