Fluent accuracy and boundary conditions

Paolo Lampitella

Hi to everyone,

i'm running some test cases to asses the accuracy of fluent and i'm getting in troubles.

The first problem is about bi-periodic boundary conditions; in a simple turbulent incompressible steady channel flow, with longitudinal and transverse periodic boundary conditions with top and bottom wall, there are non physical pressure peaks along the edges common to the periodic faces (that is along the four vertical edges) and in the same place i have a very high divergence (about 0.2, calculated via custom field function). I used standard and presto pressure interpolant with node based gradient and simplec.

The second problem is also concerned with this. I run some cases for the taylor vortex solution, that is a double array of vortex (2D) in a square decaying in time with bi-periodic or dirichlet boundary conditions. I performed several runs on several grids with different time steps and i always got a first order accuracy (in space, in time and with costant courant number; i obviously used the second order solver). In this case i used the nita fractional step method with presto/standard and central/2nd. order upwind spatial scheme with the same results. In this case the maximum error is concentrated along the edges for dirichlet boundary conditions and in the corners of the square for the bi-periodic ones.

I would like to know if there is someone with some experience about the error accuracy of Fluent and/or the boundary conditions, which actually seems to be only first order accurate and i think this depends on the pressure boundary conditions.

Thanks a lot.