Help.. Boundary conditions for external flow...

suresh1483 · April 13, 2011, 15:47

Hi,
I am solving 3D euler equation on aircraft wing, I am using FVM in unstructured grid with cell centred method. I use Lax-Friedrich scheme to discretize the equation. My results are not correct, temperatures, pressure and density are higher than the stag. values at wall boundaries. I think my boundary conditions are wrong.

I am using a circular domain of 15 times of root chord as diameter, my BC are as follows,

Far field: Free stream values. (since the domain is circular, no seperate BC for inlet and exit)

Wall boundary: reflective velocity method is used for velocity, and pressure and temperature are extrapolated from boundary cell to ghost cell.

Can any one suggest me for solution?

gems · April 13, 2011, 22:12

what is reflective velocity? Does that refer to slipping?

suresh1483 · April 14, 2011, 04:00

I am reflecting normal component of velocity from the boundary cell to the ghost cell to make the velocity normal to the surface zero. (i.e) I implemented v.n = 0, by this method. I have used the expression
vg=vb-2(vb.n)n, where g refers to ghost cell and b refers to boundary cell. Above expression is from the book Blazek

gems · April 14, 2011, 11:19

This method should work, actually it is exactly what I use. How is your convergence? I use Roe, but should not make much difference.

suresh1483 · April 14, 2011, 12:11

I am calculating norms by taking residual per time step. (i.e) Residual / t, I a using local time step. My norm converges in the order of 10-4, not increasing more than that.

gems · April 14, 2011, 12:26

If residual does not drop to machine zero, 10e-14, there is bug in your code, especially for Euler problem.

April 13, 2011, 15:47	Help.. Boundary conditions for external flow...	#1
suresh1483 New Member Suresh Join Date: Apr 2011 Posts: 3 Rep Power: 15	Hi, I am solving 3D euler equation on aircraft wing, I am using FVM in unstructured grid with cell centred method. I use Lax-Friedrich scheme to discretize the equation. My results are not correct, temperatures, pressure and density are higher than the stag. values at wall boundaries. I think my boundary conditions are wrong. I am using a circular domain of 15 times of root chord as diameter, my BC are as follows, Far field: Free stream values. (since the domain is circular, no seperate BC for inlet and exit) Wall boundary: reflective velocity method is used for velocity, and pressure and temperature are extrapolated from boundary cell to ghost cell. Can any one suggest me for solution?

April 13, 2011, 22:12	what is reflective velocity? Does that refer to slipping?	#2
gems New Member goldling Join Date: Apr 2011 Posts: 5 Rep Power: 15	what is reflective velocity? Does that refer to slipping?

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Boundary conditions low Mach number flow	lost.identity	Main CFD Forum	0	November 28, 2010 05:44
3-D Compressible Flow Boundary Conditions	Tyler	FLUENT	4	February 5, 2009 20:58
boundary conditions for boundary layer flow	A. Al-zoubi	CFX	0	November 3, 2007 08:11
Internal flow simulation boundary conditions	Kishore	FLUENT	1	July 10, 2007 12:42
Please help with flow around car modelling!	Tudor Miron	CFX	17	March 19, 2004 20:23

April 14, 2011, 04:00		#3
suresh1483 New Member Suresh Join Date: Apr 2011 Posts: 3 Rep Power: 15	I am reflecting normal component of velocity from the boundary cell to the ghost cell to make the velocity normal to the surface zero. (i.e) I implemented v.n = 0, by this method. I have used the expression vg=vb-2(vb.n)n, where g refers to ghost cell and b refers to boundary cell. Above expression is from the book Blazek

April 14, 2011, 11:19		#4
gems New Member goldling Join Date: Apr 2011 Posts: 5 Rep Power: 15	This method should work, actually it is exactly what I use. How is your convergence? I use Roe, but should not make much difference.

April 14, 2011, 12:11		#5
suresh1483 New Member Suresh Join Date: Apr 2011 Posts: 3 Rep Power: 15	I am calculating norms by taking residual per time step. (i.e) Residual / t, I a using local time step. My norm converges in the order of 10-4, not increasing more than that.

April 14, 2011, 12:26		#6
gems New Member goldling Join Date: Apr 2011 Posts: 5 Rep Power: 15	If residual does not drop to machine zero, 10e-14, there is bug in your code, especially for Euler problem.