Pressure Correction

peatmac · April 1, 2012, 14:46

Hello,

I am modelling a cylinder oscillating in a uniform current due to VIV. The simulation will run up to about 25 seconds flow time, but then it fails due to a pressure correction.
I have had a read through all the other threads available about trying to solve this problem. I am running the simulation with PISO pressure-velocity coupling with skewness-Neighbour Coupling disabled. My URF values are 0.3,1,1 and 0.7 - Pressure, Density, Body Forces and Momentum respectively. Under the Discretization tab the pressure is set to Standard and Momentum as Second Order Upwind.
I have tried to change all these values but with still no success.

However.

When I change my time step from 0.001 s to 0.003 s the analysis will keep iterating. If I let it run at this time step I run into a negative cell volume in my mesh, but If I quickly change the time step back to 0.001 s after 1 or 2 iterations the analysis will run for about 10 more time steps then fail due to a negative cell volume.

Can anyone please advise me on how to resolve this problem?

Many thanks.

P.

peatmac · April 3, 2012, 06:30

Could it be anything to do with the number of iterations per time step? Currently I run the simulation at 1 iteration per time step, and plan to crank it up to 20-30 when the cylinder settles into a regular motion.

Cheers

peatmac · April 3, 2012, 06:45

Currently I am running Fluent in 2d, would running it in 2ddp be better?

Regards.

neville · April 3, 2012, 08:33

The double precision solver does help in getting improved results however I don't think its the solution to your problem. Check your mesh. Also try using the green gauss node based scheme for gradient. Maybe this is your problem.

peatmac · April 3, 2012, 10:05

Neville,

Thanks for your advise. I am running the simulations as we speak, they take out 24 hours to run so I will post how I get on.

Thanks!

P

peatmac · April 4, 2012, 02:22

Neville,

This may seem trivial. But when you say check the mesh, What am I checking for?

Peter

peatmac · April 5, 2012, 05:27

Sorry to flag his up again people.

Can anyone explain to me what I should be checking for in my mesh to try stop getting these pressure convergence errors?

Thanks for your help

Peter

neville · April 14, 2012, 05:52

Hello Peter,
Sorry for the late reply but By mesh check I mean you should see whether your mesh is uniform and that cells are not clustered in particular region unless necessary refinement is required in that region.

Also the min cell volume for the mesh should be positive. I had an experience the mesh being discontinuous in my work. this is due to the fact that the mashing algorithm failed in that region but I was lucky to notice it visually.

peatmac · April 14, 2012, 08:48

Neville,

Thanks very much for that info there.

One thing I have found with my problem seems pretty weird. If I reduce the mass of my cylinder, the simulations run for a lot longer. I ran a simulation a few months ago with an extremely light cylinder that was critically damped. And it seemed to settle into a steady state.

Any idea why this could be?

Thanks again,

Peter

neville · April 16, 2012, 08:43

Sorry Peter, but I can't help you on this as I have not worked on vibrations.

April 1, 2012, 14:46	Pressure Correction	#1
peatmac New Member Peter Join Date: Jan 2012 Posts: 29 Rep Power: 14	Hello, I am modelling a cylinder oscillating in a uniform current due to VIV. The simulation will run up to about 25 seconds flow time, but then it fails due to a pressure correction. I have had a read through all the other threads available about trying to solve this problem. I am running the simulation with PISO pressure-velocity coupling with skewness-Neighbour Coupling disabled. My URF values are 0.3,1,1 and 0.7 - Pressure, Density, Body Forces and Momentum respectively. Under the Discretization tab the pressure is set to Standard and Momentum as Second Order Upwind. I have tried to change all these values but with still no success. However. When I change my time step from 0.001 s to 0.003 s the analysis will keep iterating. If I let it run at this time step I run into a negative cell volume in my mesh, but If I quickly change the time step back to 0.001 s after 1 or 2 iterations the analysis will run for about 10 more time steps then fail due to a negative cell volume. Can anyone please advise me on how to resolve this problem? Many thanks. P.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
changing the coefficients of pressure correction	Noel	Phoenics	1	April 7, 2009 08:54
Does star cd takes reference pressure?	monica	Siemens	1	April 19, 2007 11:26
Pressure correction scheme in axissymmetric model	Adam Wu	Main CFD Forum	0	September 30, 2005 11:45
residual in the pressure correction	George	Main CFD Forum	2	July 28, 2005 04:43
a problem about pressure correction method	tommewang	Main CFD Forum	2	May 15, 2003 21:18

April 3, 2012, 06:30		#2
peatmac New Member Peter Join Date: Jan 2012 Posts: 29 Rep Power: 14	Could it be anything to do with the number of iterations per time step? Currently I run the simulation at 1 iteration per time step, and plan to crank it up to 20-30 when the cylinder settles into a regular motion. Cheers

April 3, 2012, 06:45		#3
peatmac New Member Peter Join Date: Jan 2012 Posts: 29 Rep Power: 14	Currently I am running Fluent in 2d, would running it in 2ddp be better? Regards.

April 3, 2012, 08:33		#4
neville New Member neville Join Date: May 2011 Posts: 12 Rep Power: 14	The double precision solver does help in getting improved results however I don't think its the solution to your problem. Check your mesh. Also try using the green gauss node based scheme for gradient. Maybe this is your problem.

April 3, 2012, 10:05		#5
peatmac New Member Peter Join Date: Jan 2012 Posts: 29 Rep Power: 14	Neville, Thanks for your advise. I am running the simulations as we speak, they take out 24 hours to run so I will post how I get on. Thanks! P

April 4, 2012, 02:22		#6
peatmac New Member Peter Join Date: Jan 2012 Posts: 29 Rep Power: 14	Neville, This may seem trivial. But when you say check the mesh, What am I checking for? Peter

April 5, 2012, 05:27		#7
peatmac New Member Peter Join Date: Jan 2012 Posts: 29 Rep Power: 14	Sorry to flag his up again people. Can anyone explain to me what I should be checking for in my mesh to try stop getting these pressure convergence errors? Thanks for your help Peter

April 14, 2012, 05:52		#8
neville New Member neville Join Date: May 2011 Posts: 12 Rep Power: 14	Hello Peter, Sorry for the late reply but By mesh check I mean you should see whether your mesh is uniform and that cells are not clustered in particular region unless necessary refinement is required in that region. Also the min cell volume for the mesh should be positive. I had an experience the mesh being discontinuous in my work. this is due to the fact that the mashing algorithm failed in that region but I was lucky to notice it visually.

April 14, 2012, 08:48		#9
peatmac New Member Peter Join Date: Jan 2012 Posts: 29 Rep Power: 14	Neville, Thanks very much for that info there. One thing I have found with my problem seems pretty weird. If I reduce the mass of my cylinder, the simulations run for a lot longer. I ran a simulation a few months ago with an extremely light cylinder that was critically damped. And it seemed to settle into a steady state. Any idea why this could be? Thanks again, Peter

April 16, 2012, 08:43		#10
neville New Member neville Join Date: May 2011 Posts: 12 Rep Power: 14	Sorry Peter, but I can't help you on this as I have not worked on vibrations.