Is it normal to have fluctuations of the flow rate of ~2% throughout a pipe flow? I am measuring the flow rate via bulk data (plane perpendicular to flow direction). I dont have backflow, so can this error come simply due to the unstructured mesh?
Are you sure all the information has propogated spatially, ie. is the solution fully converged? In case of very fine mesh, it takes a while till this happens.
Unstructured mesh does have its demonds of diffusion. Am hoping you are using second order discretization (ADV5) and if possible, extruded mesh. These should pacify diffusion and hence inaccuracy.