We are modeling flow over a cylinder with a hemispherical head at various angles to the flow direction. We have a Reynold's number of around 9000 if consider the cylinder diameter as the characteristic length or 90,000 if we consider the cylinder length. We are using unstructured tetrahedral 3D grid. We have created concentric shells around the cylinder to increase the grid density.

We need to measure pressure at 3 points on the cylinder head, which we can validate by comparing the Cp values at these points with the experimental values.

The problem is we have been getting Cp > 1 with Laminar or any k-e model with first order upwind method which is not practically possible.

The problem seems to be in the laminar regime, but we are getting good results by using Reynolds stress model for turbulence which needs a lot of computational time.

Another thing that we are stumped with is if we use velocity inlet instead of pressure inlet then the residuals dropped quickly for velocity inlet as compared to pressure inlet. But Cp was > 1.

The thing is we identified that Cp is > 1 only at very few cells adjacent to the stagnation point. We even tried adapting the grid in regions of high velocity gradient to see if the grid was not sufficiently fine.

1. What is normally the characteristic length in such problems?

2. Is Cp > 1 a gridding problem or turbulence model problem and how do we solve it ?

Thanks in advance.

well friend the solution is easier that you assumed, just use prism cells near the cell, with yplus near one, and after that you will see that fluent has cp = at the cylinder had, otherwise you will get cp>1, there is nothing wrong with fluent.