Error while coupling Atmospheric Boundarly Solver PALM with an In-house CFD solver

Using preCICE
1

precice-config.xml (2.1 KB)

FLOWer_Log.txt (1.4 KB)

debug1.log (150.4 KB)

PALM_log.txt (9.2 KB)

Dear preCICE team
I am trying to couple an incompressible Atmospheric Fluid Solver called PALM, with an In-house FORTRAN based compressible CFD solver for carrying out unsteady simulations. PALM is used to carry out simulations on a bigger domain (D1), whereas the CFD solver is utilized for a smaller domain (D2) embedded within D1. The task of the coupling framework is to inject values obtained from D1 at at the interface boundary planes of the smaller domain D2 that is the inflow, lateral planes and the top planes of D2. The process of developing the required adapters were initiated by testing the respective solvers to exchange variable values using dummy solver. In the second stage, the same adapters are being utilised for coupling PALM and my in-house CFD solver. In the current state of this coupling framework, the two solvers are able to connect, initialise and also start the coupled simulations. In the first stage of this implementation, only one-way coupling is implemented, wherein velocity components will be injected from PALM to the second solver’s interface planes (mentioned above).
I have tested this framework using two different grid configurations of D2, called as Fine D2 and Coarse D2.

The coupled simulations get started and data exchange begins successfully for both the solvers and both the grid configurations:

  1. In the case of Coarse D2, PALM and CFD solver completes the simulations for the entire prescribed time steps (tested till 900 time steps).

  2. In the case of Fine D2, PALM runs its complete set of time steps (which seems correct as there are no changes in its overall mesh configuration). On the other hand, CFD solver with the finer Mesh, runs for only limited time steps and just stops without presenting any error. I tried some debugging and noticed that CFD solver stops at the advance(dt) phase. In the meantime simulations in PALM complete and has exchanged data with preCICE buffer using write_data. When I run the CFD simulations using mpirun, I notice that only one of the processes goes to sleep mode, whereas the other processes are just waiting.

    One more thing to be noted is that, PALM simulations run in single core and run very fast, due to which it ends very early and also seems to be closing the connection. The second solver performs more comprehensive calculations and proceeds at its own slow speed beyond PALM’s termination and stops after a while.

    Here I am attaching the the precice-config file, snippets from the log files of the two solvers as well as snippet of the debug.log file generated using preCICE debugging tools. I am using precice-version v3.3.1-77 for this coupling.

    Your expert suggestions on this matter would be extremely helpful.

    Summary

    This text will be hidden

2

Dear All

This is just an update on my current problem.

Once I restarted the system, this problem seemed to have solved and the coupled simulations are now running the entire prescribed time duration.

Thus, the problem is solved.