[Ajmit]

I performed steady-state CFD analysis (3D conjugate heat transfer analysis) of a gas turbine blade using Ansys CFX 2020 R1. The domain include hot gas domain, solid blade domain and cold gas domain (cooling channels). I used three turbulence model available in CFX in order to get best match with experimental values available on the mid span of blade (both pressure & suction side).

1. k epsilon Turbulence Model with scalable wall function (Default)
2. SST Turbulence Model with automatic wall function (Default)
3. SST Turbulence Model with automatic wall function and with Transitional turbulence Gamma-Theta.


These turbulence model calculate wall heat transfer coefficient based on turbulent wall function. I have gone through this page (https://www.cfd-online.com/Wiki/Ansys_FAQ) and also read other posts. I exported the results (Wall Heat Transfer Coefficient, Wall Adjacent temperature and Wall heat flux) in Microsoft Excel and then I used tbulk (bulk temperature) to calculate my new heat transfer coefficient (htc) as mentioned in the FAQ. When I calculate/plot heat transfer coefficient using 1st & 2nd turbulence model then the obtained values appears near the experimental values. Other Research paper shows similar trends. When I calculate/plot htc with 3rd turbulence model then the values changed too much. There is little change in Temperature distribution over mid span as compared to other two turbulence model. I have attached the images of plot for clear understanding.


Why it is happening like that? Does above mentioned 3rd turbulence model require new method/equations to calculate HTC?

[ghorrocks]

Turbulence models are very complex things and nothing is simple with them.

There could be many things causing this. You appear to have an idea of what is going on in CFD, so it will be a research task for you to work out the issue. Some ideas:
* Is the turbulence transition model predicting turbulence transition in the correct place?
* Is your mesh fine enough to integrate to the wall (ie: y+ = ~1)? The turbulence transition model requires a mesh with y+= ~1 to work. Your models 1 and 2 will work OK on meshes coarser than that with wall functions - but heat transfer modelling may mean wall functions do not work well and integration to the wall is required.
* I bet the turbulence transition model predicts a turbulent transition separation bubble and that might affect heat transfer.
* Have you had a detailed look at the flow fields of the different models? Do they have any apparent differences in separations, wall shear, or anything else?