UPDATE:nSo I seem to have fixed the issue thanks to a combination of all the advice given. Adjusting the reference value for the drag coefficient mad a significant difference in increasing the accuracy of the solution. It was a little tricky at first but I just had to find the frontal area of the wheel. The biggest change however was that I rotated all the bodies 90 degrees so that the inlet flow was in the +ve X-direction. Once I did that and used the default vector directions of 0,0,1 for the rotation axis of the rotating domain and 1,0,0 for the force reports the solution was instantly fixed. It was no longer negative and the plots were normal, meaning there were no large fluctuations in drag coefficient or force. I am not entirely sure why rotating all the bodies helped so much, I am assuming there was something wrong with the direction of the vectors before but I could not seem to figure that out.Now my remaining question is how I can better refine the solution? My mesh quality seems to be quite good as shown here, but reducing the face sizing on the wheel comes at a huge computational cost. Reducing the size of the far field domain also does not impact the resulting drag force/coefficient that much. would I need to apply some sort of inflation on the wheel surface? nnWould I also be better of using shared topology for the rotating domain as well? If did, would I then have to define the interface manually in the fluent setup? I just see there is a very harsh transition between the rotating domain and the fluid domain as shown.nn