I have a shaft model that is subject to torsional loading only. I used surface forces on the splines of the shaft normal to the spline faces. The model is constrained at the opposite end with beam elements that converge to one node so that I can measure the resulting moment directly. The results look normal until I do a slice plane through the center of the shaft. Then I see the inner part of the shaft appears to have been meshed "double". See the attached images. I had difficulty getting the model to run with the iterative solver and in the end resorted to the sparse solver.I will try a new mesh with different mesh size to see what happens.