Boundary conditions for pressure-correction method

moshe · May 22, 2018, 03:19

i want to learn about boundary conditions for a compact Poisson stencil. after searching through this forum it seems that people are using ghost points when handling pressure BCs for a collocated arrangement (cf. pressure boundary conditions for collocated grid for Navier-Stokes) is there a way that avoids having a layer of ghost points? this ghost point approach seems to have difficulties. After some google searching I found some slides that talks about pressure boundary conditions for a compact stencil, but it seems to be incorrect (slide 5 of https://www3.nd.edu/~gtryggva/CFD-Co...re-15-2017.pdf) The professor explicit sets the bc rather than modifying the PPE at the boundary.

FMDenaro · May 22, 2018, 04:25

There is no need to prescribe ghost nodes, you can find similar post in this forum

selig5576 · May 22, 2018, 13:47

Attached is a complete pressure bc based on Dr. Denaros approach. I have not tested this as I have switched to a staggered based solver since I was unfortunately never able to get my collocated solver to work properly.

FMDenaro · May 22, 2018, 13:56

On staggered grid this issue is also more simple, a good explanation is given in the JCP paper of Kim and Moin on the fractional step method.

moshe · May 22, 2018, 14:02

i think this formulation is wrong as you only have 1 diagonal entry in the iteration routine, no?

moshe · May 25, 2018, 13:47

don't mean to bump this, but when we write the laplacian operator at i = 2, we iterate at i = 1, which means we only have 1 diagonal term since the other diagonal term dpdy|j=2 is moved to the right-hand side.

May 22, 2018, 03:19	Boundary conditions for pressure-correction method	#1
moshe Member Moshe De Leon Join Date: Nov 2017 Location: Portugal Posts: 31 Rep Power: 8	i want to learn about boundary conditions for a compact Poisson stencil. after searching through this forum it seems that people are using ghost points when handling pressure BCs for a collocated arrangement (cf. pressure boundary conditions for collocated grid for Navier-Stokes) is there a way that avoids having a layer of ghost points? this ghost point approach seems to have difficulties. After some google searching I found some slides that talks about pressure boundary conditions for a compact stencil, but it seems to be incorrect (slide 5 of https://www3.nd.edu/~gtryggva/CFD-Co...re-15-2017.pdf) The professor explicit sets the bc rather than modifying the PPE at the boundary.

May 22, 2018, 04:25		#2
FMDenaro Senior Member Filippo Maria Denaro Join Date: Jul 2010 Posts: 6,849 Rep Power: 73	There is no need to prescribe ghost nodes, you can find similar post in this forum moshe likes this.

May 22, 2018, 13:56		#4
FMDenaro Senior Member Filippo Maria Denaro Join Date: Jul 2010 Posts: 6,849 Rep Power: 73	On staggered grid this issue is also more simple, a good explanation is given in the JCP paper of Kim and Moin on the fractional step method. selig5576 likes this.

May 22, 2018, 14:02	pressure bc	#5
moshe Member Moshe De Leon Join Date: Nov 2017 Location: Portugal Posts: 31 Rep Power: 8	i think this formulation is wrong as you only have 1 diagonal entry in the iteration routine, no?

May 25, 2018, 13:47	thread bump	#6
moshe Member Moshe De Leon Join Date: Nov 2017 Location: Portugal Posts: 31 Rep Power: 8	don't mean to bump this, but when we write the laplacian operator at i = 2, we iterate at i = 1, which means we only have 1 diagonal term since the other diagonal term dpdy\|j=2 is moved to the right-hand side.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
CFD analaysis of Pelton turbine	amodpanthee	CFX	31	April 19, 2018 19:02
Out File does not show Imbalance in %	Mmaragann	CFX	5	January 20, 2017 11:20
Wrong flow in ratating domain problem	Sanyo	CFX	17	August 15, 2015 07:20
Water subcooled boiling	Attesz	CFX	7	January 5, 2013 04:32
New topic on same subject - Flow around race car	Tudor Miron	CFX	15	April 2, 2004 07:18