[roksonkmathew]

Hi,
I have been doing fluid analysis on pumps and turbines. I have always used prism layer giving y+ <2. What would the variations in the total result be if I do not add prisms and use only tetra mesh. Other than the flow through the boundary layer, what could be the total variation in other parameters say total pressure and torque be?

Regards
Rokson

[ghorrocks]

That is dependant on what you are modelling. It could make the results completely wrong by a massive amount, or it could have no effect. It all depends.

The best thing to do is to just try it on your case and find out for yourself.

[Opaque]

You mesh count will increase drastically because of the poor resolution a high aspect ratio tetrahedral mesh provides.

Recall that flow in a boundary layer has very sharp/high gradients normal to the boundary; therefore, you need very thin tetrahedral elements normal to the boundary and to satisfy a non-drastic aspect ratio changes along the boundary, you will need A LOT more elements than hexahedral, or prismatic elements.

Hope the above helps,

[ghorrocks]

Hi Opaque,

Your comments are correct for mid to high Reynolds number flows where wall friction is important. But in very low Re flows a tet mesh to the wall is fine (eg MEMS flows), or in inviscid flows it does not matter either. While it is true most simulations are mid to high Reynolds numbers where inflation layers are important it is not the case for all simulations - hence my vague answer.