[zyy95900]

Hi everyone,

I am trying to simulate in duct flow with surface roughness. I am using realizable k-e model with enhanced wall treatment now. I know k-e model not good for boundary simulation, but how about the realizable k-e model with enhanced wall treatment. Does realizable k-e model with enhanced wall treatment work well with surface roughness simulation? Is this correct model. If not, which turbulence model I should use? I have try SST K-W model but the result seems not good.

Thanks.

[LuckyTran]

Originally Posted by zyy95900

Hi everyone,

I am trying to simulate in duct flow with surface roughness. I am using realizable k-e model with enhanced wall treatment now. I know k-e model not good for boundary simulation, but how about the realizable k-e model with enhanced wall treatment. Does realizable k-e model with enhanced wall treatment work well with surface roughness simulation? Is this correct model. If not, which turbulence model I should use? I have try SST K-W model but the result seems not good.

Thanks.

In Fluent, the rough wall treatment is not compatible with the enhanced wall function approach for the epsilon based models. You must use the standard wall functions or use an omega based model.

Wall/surface roughness is a wall modelling problem, not strictly a turbulence modeling problem. Also, recall that SST k-w is just a hybrid standard k-epsilon with omega treatment near walls.

I would imagine that neither model should have much difficulty simulation rough walls for simple duct flows.

[zyy95900]

Originally Posted by LuckyTran

In Fluent, the rough wall treatment is not compatible with the enhanced wall function approach for the epsilon based models. You must use the standard wall functions or use an omega based model.

Wall/surface roughness is a wall modelling problem, not strictly a turbulence modeling problem. Also, recall that SST k-w is just a hybrid standard k-epsilon with omega treatment near walls.

I would imagine that neither model should have much difficulty simulation rough walls for simple duct flows.

Thanks for reply.
In my case, the intruduced roughness is for increasing the turbulence in duct, which increasing the reactant mass transfer to reaction surface (wall). So I think I should use a turbulence model. I have try the standard wall function (k-e model), but the turbulence intensity on duct wall go infinit. I think turbulence intensity on duct wall should be zero.
Can you tell me which model is best in my case? k-e with standard wall treatment or k-e with enhanced wall treatment or SST k-w?

Thanks.

[LuckyTran]

Originally Posted by zyy95900

I have try the standard wall function (k-e model), but the turbulence intensity on duct wall go infinit. I think turbulence intensity on duct wall should be zero.

Turbulence intensity increases as you approach the wall and reaches a maximum value before being damped very close to the wall. That is a well known fact. It depends on your near wall grid resolution whether or not you will see the resulting decrease but what you stated is consistent with what is known about wall-bounded turbulence.

Originally Posted by zyy95900

Can you tell me which model is best in my case? k-e with standard wall treatment or k-e with enhanced wall treatment or SST k-w?

You cannot use enhanced wall treatment, see my last post. Both models are good, you can use either. I wouldn't use the standard k-epsilon model though, I'd go for realizable k-epsilon or SST k-omega model as they are superior.

[zyy95900]

Thank you very much.

[randolph]

Originally Posted by LuckyTran

Turbulence intensity increases as you approach the wall and reaches a maximum value before being damped very close to the wall. That is a well known fact. It depends on your near wall grid resolution whether or not you will see the resulting decrease but what you stated is consistent with what is known about wall-bounded turbulence.



You cannot use enhanced wall treatment, see my last post. Both models are good, you can use either. I wouldn't use the standard k-epsilon model though, I'd go for realizable k-epsilon or SST k-omega model as they are superior.

When you use enhanced wall treatment with y+<1 mesh, low Re K-epsilon is used near the wall. There is some stability issue with the damping function. But it still not a really bad option