[feixiangniao]

hi,everbody!
now, i am modeling conjugate heat transfer.
i am modeling cooling plate which is used to cool the electric devices.and i optimize the cooling plate structure. so i must simulate the different boundary conditions.e.g. the massflow.And i write the patch file to solve the differnt massflow.
i have three questions:
1.the Reynolds number for the flow at the inlet is 1600 ,and the Reyonld number for the flow at the internal flow is 1000. Because the structure of the flow is very complex , i think there is vortex flow, and the flow is modeled as being turbulent (SST).
Am i right ?
2.i write the patch file to solve the differnt massflow to optimize the cooling plate.
Fistly the result is good. and then i see that setting the initial values from an existing results file will well converge. so i write the patch file by using a set of results from a previous run as the basis for an initial guess.
But the results is not good, and is not similar to the previous results without using a set of results from a previous run at the same boundary condition.
So what is the problem ?
Do the initial values change the results?
Do the results file at the low-Reyonlds can initialize the high-Reyonlds model?

3.because the flow structure is complicated, i can not mesh the the boundary layer.In the cfx-post, the areaAve-yplus is 3.6 at the flow wall. In the CFX, i can use the wall function to impose suitable conditions near to the wall without resolving the boundary layer.
so i don't need mesh the boundary layer.
Am i right ?or how i mesh the near-well region and what is the law ?

[ghorrocks]

the Reynolds number for the flow at the inlet is 1600 ,and the Reyonld number for the flow at the internal flow is 1000. Because the structure of the flow is very complex , i think there is vortex flow, and the flow is modeled as being turbulent (SST). Am i right ?

Unlikely. Sounds like a laminar flow to me so you should model it laminar. Even if the flow does have low levels of turbulence, even then a laminar model is often better than a turbulence model for extremely low levels of turbulence.

i write the patch file ... initialize the high-Reyonlds model?

Sorry, I don't understand this question.

because the flow structure is complicated, ... and what is the law?

I don't think y+ is a parameter of interest as I don't think the flow is turbulent. When you run the model as laminar you will not have y+ as an available variable. Just do a normal mesh refinement analysis to see if your mesh is fine enough.