when solving solid rocket motor flowfields, the boundary conditions for mass injection is typically a function of the local static pressure against the wall. As far as specifying the turbulence parameters at the boundary, what are some suggestions for intensity, length scale, dissipation, etc? What is the best turbulence model to use? What is the best value to use for limiting the ratio of turbulent to molecular viscosity ?

thanks

(1). The velocity at the propellant surface normally is relatively small when compared with the cross flow velocity or the velocity entering the nozzle. So, the turbulent kinetic energy at the surface is probably not going to have important effect on the flow field development. (2). So, you can specify any reasonable number there and check the flow field results. (3). Two-equation model is a good choice because you don't have to worry too much about the details, and it is also good for complex geometry. (algebraic models require much more attention when the geometry is complex) (4). In general you have accelerating internal flow where the two-equation model should be able to perform all right. (5).Limiting the turbulent viscosity ratio? I don't know. This is a modeling issue, so, you are free to input any number you want. But, remember to check the results. (6).A very complex problem, but normally, it is handled in a very simple way.

regarding the turbulent viscosity: that limit is critical in determining the right velocity profile (radial) near the aft end of the motor, and also for determining the pressure drop for large L/D ratio motors. It seems though that the turbulent viscosity can easily get to very high magnitudes, so limiting it is essential.

thanks

(1). I thought that if you use a two-equation turbulence model to compute the eddy viscosity from the solution of k and epsilon, then you have the eddy viscosity values everywhere in the flow field. (2). In general, the turbulent Reynolds number is of order O(100), and the molecular Reynolds number can be very high O(1.0E+06). So, in this example, the eddy viscosity is 10000 times larger than the molecular viscosity. For this reason, the molecular viscosity is largely ignored when using the high Reynolds number model. (3). I think, if you use a two-equation model for the turbulent eddy viscosity, you will have the whole flow field solution. And there is no need to limit the turbulent eddy viscosity level. I am not aware of the need to limit the level of the eddy viscosity in a two-equation turbulence model. So, I guess you must have a special reason to do so.

Your need to limit the turbulent viscosity is probably due to the strong accelaration that you have in your flow. This can trigger an instability in the classical k-epsilon model which leads to very high values of k and eddy viscosity. Limiting the eddy-viscosity can "hide" this problem. The question then is which limit you should use? This depends on your case and I think that you'll have to test a few "known cases" first to develop some experience.

A better solution is to use a model which does not have this problem. There are several "realizability" corrections which will improve this behaviour siginificantly - Look for papers by Durbin and Shih-Lumley. You can also use the variable C_mu approach based on the old works by Rodi. Let me know if you want references on these and I'll dig it out. A simpler fix is to use the Kato-Launder modification. If your free-stream turbulence is negligible then this might be your best choice. k-omega is less affected by this problem, although you sometimes can see similar things also with k-omega. For rocket-nossle flows we use a k-omega model with a realizability fix.

Jonas,

I have this instability problem in k-omega I believe so references would be welcomed.

Thanks. Sergei

that's some good information. I would appreciate getting those references.

thanks