Do you have a model which can be used to verify the theoretical skin depth? (or one that demonstrates that ANSYS calculates the skin depth correctly)? At one skin depth below the surface of a conductor, the volumetric joule heating (W/m^3) should be (1/e^2) times the value on the surface of the conductor (e = natural log base).
For the purposes of this investigation, I adapted an input file from our documentation (Coupled Field Guide, Section 4.6: Sample Induction Heating Analysis of a Circular Billet). I simplified the model to perform a field calculation at a single temperature and frequency. I added APDL to obtain the maximum volumetric joule heating at the surface and that at one nominal skin depth below the surface. I created a path 5 skin depths in length extending from the surface and map volumetric joule heating onto this path. I evaluate the following parameter: RATIO = (joule heat @ 1 skin depth)/(surface joule heat/e^2) I get RATIO = 1.028 which isn`t too terribly bad. Note that the joule heat read into the general postprocessor is the time averaged value for harmonic response analyses such as this one. Note, too, that I use the following commands: esel,s,mat,,2 ! SELECT ONLY BILLET ELEMENTS TO SUPPRESS AVERAGING AT SURFACE nsle prior to reading JHEAT into the element table and leave this set selected before plotting JHEAT along a path with PLPATH command. This suppresses undesirable data averaging at the surface of the work piece. 

