[Mehar]

Hi
I am simulating air flow around a building (dome like) using OpenFoam (Re ~10^7).
The problem is with the drag coefficent.
The drag coefficent calculated while using upwind scheme for convective term is positive.
But drag coefficient calculated while using linearUpwind/limitedLinear1 convective is negative.
Everything else is kept same for both the cases.

The turbulance model is SA.
The domain BCs are freestream.
The walls are with nutUspaldingWallFunction.
I have also run so many cases separately with different domain boundary conditions, mesh, and 2nd order schemes, but the problem persist and I get -ve drag each time I use 2nd order scheme
Can somebody help to identify the cause and the solution.
It will be a great help.
Thanks

[sergiu tofanel]

I really doubt that your solution is converging. One way to tell is to plot the Cd value vs the number of iterations over, let's say, 10000 iterations. Toward the end of the simulation (8000+ iterations) the Cd value should be flat. If not, you have a convergence problem. This can be due to mesh resolution, bad initial BC's and/or using the wrong solver.

[Mehar]

Originally Posted by sergiu tofanel

I really doubt that your solution is converging. One way to tell is to plot the Cd value vs the number of iterations over, let's say, 10000 iterations. Toward the end of the simulation (8000+ iterations) the Cd value should be flat. If not, you have a convergence problem. This can be due to mesh resolution, bad initial BC's and/or using the wrong solver.

Thank you sergiu tofanel for the interest.
The problem is not of the convergence, my solution is converged and all the forces and the coefficients are stable.
The problem is with the negative Cd in 2nd order discretization of the same case having positive Cd with 1st order scheme.
Any comments on this issue?