[grjones10]

Dear All,

I am having problems with the reference pressure when I use pimpleDyMFoam.

I am using pure Crank-Nicolson in time and central differencing in space. I use GAMG for pressure and mesh motion and PBiCG for velocity. Furthermore, my BCs are as follows:

front: p = cyclic, U = cyclic, pointDisplacment = fixedValue
back: p = cyclic, U = cyclic, pointDisplacment = fixedValue
inlet: p = zeroGradient, U = fixedValue , pointDisplacment = fixedValue
outlet: p = zeroGradient, U = advective, pointDisplacment = fixedValue
aerofoil: p = zeroGradient, U = fixedValue, pointDisplacment = fixedValue
controlSurfaces: p = zeroGradient, U = movingWallVelocity, pointDisplacment = oscillatingDisplacement

The simulation runs and converges within 3 pimple iterations however each time step is taking a very long time and the pressure field is incorrect. When I say the pressure field is incorrect I mean that, despite setting pRefValue = 0, the pressure in the reference cell is not zero. In fact, the pressure is not zero anywhere in the domain.

Please find some figure attached of the instantaneous pressure in my entire domain, in the vicinity of the pRefPoint and the trace of instantaneous pressure at pRefPoint as the grid moves.

Has anybody encountered this sort of thing before or have any ideas how to solve it? I would be extremely grateful for any advice.

I should also mention that I run the simulation in parallel and have tested 96 - 1500 cpus for my grid of 45 million cells. While in all cases the simulation is very slow (best case scenario 50 seconds per times step), it does scale linearly from 96 - 800 cpus.

Thank you for your time,
Gareth

motion_pressure_domain.jpg

motion_pressure_pPoint.jpg

Pref_movingWallVelocity.jpg

[grjones10]

Also, it appears that the reference pressure follows the acceleration of the grid? See figure attached.

Many Thanks,
Gareth

pRef.jpg