LJDCRF,
Obviously the failure is due to the geometry of the splines and how these transition into the shaft. Increasing the diameter of the splines while maintaining the same shaft diameter while blending the two sections with a large radius would be the first step to improve the design. Having said that, I will address the follow-on items that you requested regarding specifying this shaft as a forging.
First, as you already know there is a quite large difference between the shaft diameter and the drum diameter, which complicates the forging process. One option would be to manufacture the drum and shaft separately and then join them by friction welding. The following link shows a part manufactured in this manner:
Now the next thing I should clear up is the thermal processing that should take place after forging and prior to machining. Normalizing is frequently used for lower alloyed and lower carbon steels. For 4340Mod/300M normalizing to a ferrite + pearlite microstucture is not a huge improvement in machinability-- I think the hardness is still around 350 HB in this condition. Annealing or spheroidize annealing to a hardness in the range of 187-241 HB will result in optimum machinability. Now, keep in mind that this type of processing is based on conventional machining, and does not take into account hard turning/milling concepts that have become more popular recently, primarily for precision tools and dies.
If this were my component to develop, I would either make it in two pieces as I mentioned above, or I would look at making it on hot extrusion equipment like the Hatebur HM75XL. The maximum billet diameter for this machine is 90 mm, and this can be upset to a maximum diameter of 145-165 mm and a maximum length of 180 mm. This would create a preform in four die stations that would be roughly the proper drum diameter, but the shaft would be larger and shorter than required. The next step would be to cold extrude this part to the final near-net shape, only requiring final grinding to the finished diameter. Turning would be limited to the drum end and for the grooves. The splines could be extruded as well. Obviously this entails a lot of forging tools to be created, so if the number of parts to be manufactured is not very high, then this will be overly expensive.
Using this process, the part would be annealed after hot forging and prior to cold extrusion. The splines could be finish extruded, and the annealed + cold worked microstructure would be adequate for machining. The part would the quenched and tempered to the desired bulk hardness, probably in the range of 42-47 HRC. Finish grinding would create the final contours, accounting for distortion. Induction hardening of the splines to a hardness of ~ 53-60 HRC and a depth of 0.5-0.8 mm would probably be sufficient. Tempering for 1 hour at ~ 180 C would improve the toughness without deteriorating the hardness much.
The last item I would investigate is shot peening. You may want to do this as the very last process, making off areas that you don't want a change in surface roughness. Alternatively, this could be done prior to induction hardening, and tailoring the process to achieve the desired residual hardness, surface roughness, etc. Metal Improvement has a nice website and a lot of technical information that can be downloaded. The Forging Industry Association also has a nice summary of the questions and details that should addressed regarding specifying forgings (
Good luck and feel free to ask any further questions.