PFMD,
Detailed analysis of splines is not that simple. And how you approach the analysis is very dependent upon what conditions the spline operates in and how carefully the spline and its mate are manufactured. What is the torque the spline must transmit, what is it made from, and how many load cycles does it see?
If the spline is well supported, accurately manufactured, has a L/D ratio below 1.0, has adequate back-up shaft structure, and is properly lubed, then its torque capacity will likely be determined by its resistance to fretting at the tooth contact. Unless your spline geometries are very unusual, then the spline will not fail in root shear, or fracture failure propagating from the fillet area. It will fail from surface initiated fractures propagating out from the contact faces. The classical spline failure mode is due to these fractures originating from tooth surface pits caused by trapped metallic debris producing surface fretting or corrosion. That is not a condition that you can evaluate for with an FEA model. And most likely, your FEA model would show big, fat margins for both root fillet stresses and tooth shear stress for a spline designed for contact limits.
If your spline is susceptible to misalignment during operation, or has a L/D ratio greater than 1.0, then you will need to manufacture the spline with some degree of lead correction or crowning to compensate for torsional wind-up and edge loading.
I design lots of high-performance involute splines for aircraft that are carburized and require unlimited fatigue life. For these oil lubed splines, the spline is designed so that it has a constant flow of oil along the teeth to flush away any debris, has an L/D below 1.0, and the spline is designed for a simple face contact stress (ext. spline OD - int. spline ID X the number of teeth) below 5 KSI. This limit may seem overly conservative, but experience has shown that oil lubed splines designed to this limit will last forever.
Regards,
Terry