Another possibility for the bad results, unrelated to the hourglass control, is that the lack of a mid-point strain calculation can also create stresses in a rotating system. To use a mid-point strain calculation requires that the
*CONTROL_ACCURACY input be included with a one in the first field. In this case the strain-displacement matrix is computed twice: at times n+1/2 and n+1. This increases the computational cost by 25%, but is generally not required in explicit calculations since the mid-point configuration is nearly identical to the n+1 configuration. If you created a non-physical test problem, where the rotation is significant each time step, this could also cause non-physical stresses. For implicit problems, where the time step sizes are 1000 times greater, the *CONTROL_ACCURACY option is important. If your are analyzing spinning bodies this option must be used.