Simple method: pressure increases after some steps

Hermano · December 23, 2012, 18:24

Hi,

I have implemented the simple method on a collocated grid (Rhie-Cow pressure correction) for solving the NS-equation. However, after more then 10 steps in the outer iteration (outer iteration = next time step) the pressure correction p' calculated in the inner iteration increases drastically (>1e5). In the steps before the pressure correction p' stays smaller then 1e-4.

I don't see anything strange in the velocity field what can be the reason for this sudden pressure increase of p'.

The problem I want to solve is a channel flow with the pressure specified at the in- and outlet.

Any idea what can cause the problem?

cdegroot · December 23, 2012, 18:44

Specifying pressure at inlet and outlet is very unstable. You need to specify velocity (mass flow) somewhere. The most stable thing to do is specify pressure at the outlet with zero velocity gradient (assuming domain is long enough) and specify velocity at the inlet with pressure extrapolated.

Hermano · December 23, 2012, 19:04

Chris do you really think that this can be the reason?

If I use the guessed starting pressure in the outer loop instead of the corrected pressure my calculation converges to the solution.

But it's right that I have to use the updated pressure in the outer iteration (next time step) to solve the momentum equations, not? Otherwise there is no mass conservation.

cdegroot · December 23, 2012, 21:28

I think it is pretty likely this is the problem. This setting pressure on all boundaries is really unreliable. Sometimes if you have a good enough initial guess it will work but I'd say this is more the exception than the rule. It's interesting that it converges with the guessed pressure. I can't say why with any certainty.

I'm curious if you have tested your code on any other problems yet? If so, what were the results like?

sutiut · February 3, 2015, 12:07

I had the same problem
but thanks to this Article, I could realize my problem. this article is really helpful.

one should notice to put Pprime=0 after each outer loop of SIMPLE algorithm.
this was my problem.

December 23, 2012, 18:24	Simple method: pressure increases after some steps	#1
Hermano New Member Join Date: Jul 2010 Posts: 10 Rep Power: 16	Hi, I have implemented the simple method on a collocated grid (Rhie-Cow pressure correction) for solving the NS-equation. However, after more then 10 steps in the outer iteration (outer iteration = next time step) the pressure correction p' calculated in the inner iteration increases drastically (>1e5). In the steps before the pressure correction p' stays smaller then 1e-4. I don't see anything strange in the velocity field what can be the reason for this sudden pressure increase of p'. The problem I want to solve is a channel flow with the pressure specified at the in- and outlet. Any idea what can cause the problem?

February 3, 2015, 12:07	Solved	#5
sutiut New Member aah Join Date: Mar 2013 Posts: 14 Rep Power: 13	I had the same problem but thanks to this Article, I could realize my problem. this article is really helpful. one should notice to put Pprime=0 after each outer loop of SIMPLE algorithm. this was my problem.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Timestep and Pressure Correction Relationship in SIMPLE	rks171	Main CFD Forum	23	May 4, 2012 02:04
Pressure constant B.C. in fractional step method	amin144	Main CFD Forum	10	April 10, 2012 18:58
compressible SIMPLE method	fakor	Main CFD Forum	1	August 30, 2010 12:21
Streamfunction-vorticity method - Pressure find	Perumal	Main CFD Forum	1	September 23, 2008 15:43
pressure driven flow by pressure correction method	justentered	Main CFD Forum	0	December 30, 2003 00:52

December 23, 2012, 18:44		#2
cdegroot Senior Member Chris DeGroot Join Date: Nov 2011 Location: Canada Posts: 414 Rep Power: 18	Specifying pressure at inlet and outlet is very unstable. You need to specify velocity (mass flow) somewhere. The most stable thing to do is specify pressure at the outlet with zero velocity gradient (assuming domain is long enough) and specify velocity at the inlet with pressure extrapolated.

December 23, 2012, 19:04		#3
Hermano New Member Join Date: Jul 2010 Posts: 10 Rep Power: 16	Chris do you really think that this can be the reason? If I use the guessed starting pressure in the outer loop instead of the corrected pressure my calculation converges to the solution. But it's right that I have to use the updated pressure in the outer iteration (next time step) to solve the momentum equations, not? Otherwise there is no mass conservation.

December 23, 2012, 21:28		#4
cdegroot Senior Member Chris DeGroot Join Date: Nov 2011 Location: Canada Posts: 414 Rep Power: 18	I think it is pretty likely this is the problem. This setting pressure on all boundaries is really unreliable. Sometimes if you have a good enough initial guess it will work but I'd say this is more the exception than the rule. It's interesting that it converges with the guessed pressure. I can't say why with any certainty. I'm curious if you have tested your code on any other problems yet? If so, what were the results like?