Hello CFD-friends. A short problem description: flow through a baffle, simple geometry like an elbow divided in several channels, Newtonian, incompressible,2D, inlet-velocity = 24.9 m/s, pressure outlet, fine and good mesh,structured in near wall regions, discretisation scheme: only first-order accurate, nu = 2,8e-5. Analysing the solution brought out big differences in the calculated pressure loss between the steady-state and the transient solution. Has anyone ever experienced huge differences (pressure loss nearly twice as high) between steady-state and transient simulation although the final residuals are of order 1e-7. The flow is obviously unsteady.

100% is a large difference. I am assuming the unsteady loss is the higher one?

No, the unsteady solution reports a pressure loss of 220 Pa, the steady-state simulation gives a value of 330 Pa. Meanwhile I changed discretisation scheme to second order, but the great differences remain. Software: OpenFoam, solver: simpleFoam - unsteady; turbFoam - transient. Any advice?

It's hard to give advice other than to make sure that the computations are not just converged, but have also been checked for accuracy (experimental data would be invaluable!). Try increasing grid resolution and temporal resolution (unsteady) to the point where the solution doesn't change any more, and make sure your boundary conditions are suitable. Closely examine the unsteady solution (what actually happens over time -- is there some kind of flow instability... some kind of oscillation... ).

One thing that's suspect is the fact that the steady state gives higher losses. I can imagine the opposite situation: if there really is unsteadiness such as a vortex instability, then the creation of vorticity over time will likely be seen as additional loss as compared to the steady state. However, in your case I couldn't trust either steady state or unsteady result. A closer look at both solutions should help to see if either of them actually makes sense.