I think that your idea of using a radiused shaft is a good approach.  There is a good section on calculating the kt factor in the Machinerys handbook 25th edition page 98.
I am wondering where you came up with the 2.8 stress concentration factor?
I will keep looking for you.

A good book for Stress Concetrations is a book by Peterson. You can always build a FEA model. Depending on the application you may be over designing the part. Stress concentrations are typically looked at during LCF, I am not sure how you margins of safety are being calculated.

You might also want to check into using a different material, i.e., one with more strength, and also look into hardening the material. I realize these are more exspensive approaches, but sometimes when you have dimensional constraints, these are the only alternatives.

Don Leffingwell PE

dleffingwell@snet.net

To reply to ecampos and stress I used the stress concentration factor from Peterson's Stress Concentration Factors for axial tension in a thread. I tried the FEA model and after the 4 th time it crashed I realized the FEA software I have is pretty lame.

Don,
I tried the material approach on the customer and he about croaked when I quoted him prices on the harder and stronger materials.

I guess the next question I need to ask is there a resonably priced FEA program out there that can handle threads?

Drive shafts typically experience cyclical loading so materials with an endurance limit are needed, otherwise no matter what you do in terms of design, the shaft will eventually fail.  Stress concentration factors are typically related to the part geometry and loading condition only, no affect of material properties is included.  The constraints that you have for geometry will force your design problem into a material or loading issue.  Hardened materials are stronger but they also have no ductility so they crack when lower strength materials deform.  A hardened material might be worse than a lower strength but ductile material.  Can anything be done to reduce the load?  Also, http://www.cadreanalytic.com offers a low price FE product that is relatively easy to use.  It uses beam and plate elements.  You might be able to mesh up the drive shaft cross section and look at the state of stress at the thread form.  Such analysis gives you a "sharper pencil," that is, a more accurate evaluation of the true stress state than can be obtained from "closed form," or handbook calculations.  Such accuracy allows you to relax safety factors.  Be warned however, this can be dangerous to those who use the equipment and testing must be done to verify that you did not make a mistake during analysis.

Mike Van Voorhis
MJVanVoorhis@CS.com