CFD Online Discussion Forums

CFD Online Discussion Forums (
-   Main CFD Forum (
-   -   Open boundary conditions? (

John November 22, 2000 12:46

Open boundary conditions?
Dear All,

I am trying to solve a 2D free jet (from a wall) problem. I am wondering how should I define the open boundary conditions in terms of velocity and pressure. Ideally, this open boundary should be transparent and the fluid is free to leave or come back to the computational domanin.

Thank you very much.


kalyan November 22, 2000 14:22

Re: Open boundary conditions?
If the flow is leaving the domain, a simple extrapolation or convective boundary condition should work for the velocity components. It is customary to prescribe the pressure at the outflow boundaries.

If the flow is entering your computational domain which does indeed happen due to entrainment, the problem is more difficult to fix. There are many CFD problems in which the flow tries to enter the domain at what was meant to be an outflow boundary (reverse flow at the outflow boundary) though these problems may mathematically be considered as ill-posed. In such cases, you extend the domain far away into the ambient that a simple extrapolation from inside should give you the entrainment velocity. I know that this isn't an elegant solution but atleast it got me closer to experimental solutions in some cases.

Chidu November 22, 2000 17:13

Re: Open boundary conditions?
I agree with what kalyan says. Another thing that might help is if you coarsen the grids near the outflow. This will dampen the effect of boundary condition errors. Ofcourse it will also dampen the validity of the solution near the boundary. But it might help with numerical reflections of pressure etc.


Abhijit Tilak November 23, 2000 05:45

Re: Open boundary conditions?
hi Jhon,

one way is u could consider jet symmetric about centerline ( it will grow in symmetric fashion in absence of body forces/buoyancy). you can refer to patankar-spalding genmix program for 2-d flows. the coordinate transformation to x-w is applied. the program tells u how to deal with various types of boundary conditions. for exit boundary du/dx=constant ie asymptotic boundary condition is sufficient. The velocity at entrainment boundary is updated at each iteration by taking into account the diffusion of momentum. i hope this helps

Abhijit Tilak

Peter Young November 23, 2000 07:01

Re: Open boundary conditions?
The pressure boundary is the best option for open boundary condition because it allows flow to go out as well as come in based on the solution. For large open boundary, environmental pressure boundary condition is required (such as a large window in a room): this condition simply means that when flow goes out, the user-specified pressure is treated as static pressure; and when flow comes into the domain, the specified pressure is treated as total pressure.

Ghanshyam Singh December 10, 2000 10:38

Re: Open boundary conditions?
Dear John,

I am agree with Kalyan that "convective" B.C. at the exit plane and fixing the "pressure" will do the job. Regarding the inflow or on the boundary where from entrainment is taking place "tangential component" of velocity can be neglected. For details, you can refer to my paper:

"Entrainment and mixing studies for a variable density confined jet", Numerical heat transfer, Part A Vol. 35, pp 205-223, 1999.


All times are GMT -4. The time now is 16:13.