Currently I am carrying out the simulations without transferring EM forces to mechanical. In the stress analysis setup, I am checking the maximum stress (Von Mises)  at only selected edges of the air pocket instead of the entire rotor body. When I manually run some combinations with change in two radiuses of the air pocket I obtain a maximum stress of around 585MPa.

When I run the combinations with the design of experiments based on response surface block in the Workbench /optiSlang DoE, I obtain much reduced stress maximum stress values (185MPa). On verifying manually the same design combination in mechanical I obtain the correct stress of around 550-600Mpa. Is there some other way to take the stress for a particular area or is there some mistake in this method?