[crn]

Here's my problem explained very simply...
At this moment, I’m developing a new air condition duct for a bus.
As you can see on the picture attached, we have an evaporator on the top / middle of the duct.
The model I created converges well if I use a simple "mixing length" turbulence model
But as soon as I start using a k-e model, it doesn't converge:
I think that I calculated correctly the turbulence length scale and energy using the formulas found on this very site.
So my questions are:
- Is the Mixing length model really not good enough for this kind of study?
- Is the k-e model appropriate for such kind of AC duct study or not? (I saw that some people were raising doubts about this model usage to study the flow into ducts)
- Is there another more appropriate model?
I would like to add that I tried various mesh sizes and velocity speed for the inlet and the only reason It doesn't converge seem to be the k-e model.
I have some doubts also regarding the type of wall condition that I should use to work with this k-e model.
Could the problem be related to the definition of wall boundary I am using?

[ztdep]

Originally Posted by crn

Here's my problem explained very simply...
At this moment, I’m developing a new air condition duct for a bus.
As you can see on the picture attached, we have an evaporator on the top / middle of the duct.
The model I created converges well if I use a simple "mixing length" turbulence model
But as soon as I start using a k-e model, it doesn't converge:
I think that I calculated correctly the turbulence length scale and energy using the formulas found on this very site.
So my questions are:
- Is the Mixing length model really not good enough for this kind of study?
- Is the k-e model appropriate for such kind of AC duct study or not? (I saw that some people were raising doubts about this model usage to study the flow into ducts)
- Is there another more appropriate model?
I would like to add that I tried various mesh sizes and velocity speed for the inlet and the only reason It doesn't converge seem to be the k-e model.
I have some doubts also regarding the type of wall condition that I should use to work with this k-e model.
Could the problem be related to the definition of wall boundary I am using?

1. try to use fully coupled solver
2. adjust your wall function method.
3. what is your grid density and quality.
4. improve the orthogonality of grid near the wall.