CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   Mass flow rate through a valve (http://www.cfd-online.com/Forums/fluent/27403-mass-flow-rate-through-valve.html)

Martin Sigrist December 8, 1999 05:37

Mass flow rate through a valve
 
Hello,

I try to simulate a slidegate valve with Fluent 5.1. Befor, I have made some measurements in the laboratory. I have measured the static pressure behind the valve at several points and the mass flow rate. The pressures from the Fluent simulation are very close to the measured pressures. But the value of the mass flow rate isn't very good. My question is now, which parameter should I change to get better results for the mass flow rate?

I calculate with the folowing models: - 3D, second order and doubble precision - k-e-model, standart - standart wall function - air, density: ideal gas law, all other parameters are constant - Boundary conditions: pressure inlet and pressure outlet

Thanks

Martin Sigrist

Biel School of Engineering and Architecture University of Applied Sciences Bern Switzerland

John C. Chien December 8, 1999 16:00

Re: Mass flow rate through a valve
 
(1).I can not answer the questions related to the code. ( A better way to do is to try the voice input device , so that you can talk directly to the computer and the code. You can also train the computer to recognize your command. ) (2). Your problem is : a) internal flow, b) internal turbulent flow, c) 3-D, internal turbulent flow, d) 3-D, internal, turbulent, separated flow. (3). It is a very difficult problem. (4). For separated flows, the use of the wall function is not accurate. (5). A low Reynolds number model is likely needed to handle the separated flow. As a result, fine mesh is required (Y+=1 criterion) near the wall, which may be very expensive in 3-D in terms of memory usage. (6). The flow separation regions could be located ahead of, around and behind the valve gate. The separation region also is sensitive to the exact geometry shape. (7). There is no simple cure. But, you sure can try the low Reynolds number model (if there is an equivalent one in the code, a two-layer model is all right to use.) along with fine mesh near the wall to improve your solution accuracy. (8). By the way, the general rule is to use a low Reynolds number model for separated flow calculations. (if you are not getting good results with the standard k-epsilon model/with wall function.) (9). Improvement in turbulence modeling is another area, but I can't tell you what to do.


All times are GMT -4. The time now is 20:31.