[gentodin]

Hi all, I am currently trying to run interfoam modified solver using MPIrun openfoam's parallel simulation command.

Currently running on OF8. Using decomposepar, simple 4 subdomains (2 2 1).

Run command: mpirun -np 4 interfoam -parallel >> log &

I have 4 subdomains and is testing out if there is any discrepancy for alpha.water in solving with parallel simulation vs serial.

I realised if I run dambreak tutorial under interfoam for parallel vs serial, I get two slightly different values at about 8.45% difference [cell ID 499]

As shown in the picture are the cell values comparison.

Does anyone know why this happen?
How do I prevent this or
How do I improve the accuracy of parallel simulation?
I saw online it maybe some floatTransfer issue i.e. WritePrecision?

I have come across functions like .correctBoundaryConditions() that should help for volScalarField but what about surfaceScalarField?

Do I need to correct them also for parallel simulation?

Please help me if possible

Thank you all,
Stan