Description
Hi!
I'm trying to restart a unsteady cfd simulation at a certain time iteration using a new grid obtained by deforming the geometry used in previous time iterations with predefined new coordinates (using SU2_DEF), but the problem is that once the simulation is restarted, the new geometry is not used and the simulation proceeds with the undeformed geometry. This seems odd, and I guess that, since the deformation is only changing the coordinates of nodes of the original grid while keeping the same connectivity and number of elements, SU2_CFD doesn't detect that there's a new grid for some reason and might be proceeding to use the coordinates in the solution_flow_xxxxxx.dat file upon restart.
In a different scenario in which a new grid is built using the new boundary (then a new mesh with different connectivity and different number of elements than the initial, undeformed grid is used), SU2 detects that there's a difference in the new grid and invokes an interpolation routine to build the initial flow field in the new grid. This doesn't have to occur if the deformation is obtained using SU2_DEF because the connectivity and number of elements are the same (which is my objective), but for some reason, SU2_CFD is bypassing the mesh update and restarts the simulation with the old, undeformed geometry (even though the filename in MESH_FILENAME is different and refers to the new grid with the deformed boundary) .
I'm attaching a compressed folder with dummy files that can help reproduce the issue. In there, you'll find 3 2D grid files: grid_0.su2 (the undeformed cylinder), grid_4.su2 (first deformation obtained using SU2_DEF), and grid_8.su2 (second deformation obtained using SU2_DEF). Also, there are corresponding config_0.cfg, config_4.cfd, and config_8.cfg files. In the first config, the unsteady simulation is started at the second time iteration using solution_flow_00000.dat and solution_flow_00001.dat, and 10 time steps are run using the initial geometry (MESH_FILENAME=grid_0.su2). Then, config_4.cfg is meant to restart the CFD at the 10th time iteration using the flow fields in solution_flow_00008.dat and solution_flow_00009.dat (obtained using the geometry in grid_0.cfg) while updating the grid file (MESH_FILENAME=grid_4.su2), and it is observed in the results that the new flow_xxxxx.vtu have the undeformed boundary even though the grid filename has been uptaded. Similarly, the attempt is to restart the CFD at the 20th time iteration with an updated geometry (MESH_FILENAME=grid_8.su2), but the simulation continues using the undeformed geometry in grid_0.su2.
If you run the simulations, you'll observe that when the unsteady cfd is restarted, the flow_xxxxxx.vtu files show that the initial geometry is kept even though different files are used in MESH_FILENAME upon restart. These simulations correspond to unsteady RANS with a rigid body motion dynamic mesh with a pitching cycle defined in the config.cfg.
This doesn't occur if, instead, a steady CFD simulation is restarted using the new deformed grids with a solution_flow.dat obtained from a simulation of the original geometry. In this case, it can be observed that the geometry is updated in the solution upon restart.
Thank you in advance for the help!