I am doing flow simulation of an existing centrifugal pump with measured known output values. The simulation converges nicely and gives somewhat sensible results, but they're still very far off from the measured ones. Both the pressure drop (Outlet pressure - Inlet pressure) and motor power (Torque on impeller * rotation speed radian/s) are nearly double to what they should be in reality. Is this a common problem? Are there some usual causes for it? I can't think of any other reasons but insufficient mesh resolution. I have been running 600k cell mesh, which is not enough to obtain really accurate data, but could this big of a difference be due to the mesh? Tinkering with wall roughness had no major impact.
The odd thing about it is that the pump efficiency is still quite close to the real value. The errors in pressure drop and motor power are about the same in relation to the value and they cancel eachother out in the efficiency calculation.
Any hints or tips are much appreciated.