Here is a story that took me three visits to understand. A plant had a process pump, 40 mm discharge, transferring 80°C condensate. The shaft kept breaking at the bearing journal. They went through three shafts in a year. First was C45, broke at 3000 hours. Then 42CrMo4, broke at 5000 hours. Then stainless 316, broke at 2000 hours. Nobody could explain it. The shaft diameter was only 25 mm, and the customer kept asking us to make it thicker. We went to 30 mm. It broke sooner.
The mistake was obvious in hindsight but took me a while. We were looking at the shaft as a strength problem. It wasn’t. The pump was coupled to a motor by a short spacer coupling. The alignment was off by about 0.4 mm parallel. Every rotation, the shaft saw a reversing bending stress. The stress amplitude was tiny — maybe 30 MPa — but it was at 2900 RPM, 24 hours a day. That’s 127 million cycles a year. Even a ductile shaft has a fatigue limit, and 30 MPa reversing at that frequency is right at the edge for C45 in a corrosive condensate environment.
We tried the stronger material because the failure looked like a strength failure. The fracture surface was smooth and flat, classic fatigue. But the fatigue origin wasn’t at a keyway or a fillet. It was right under the bearing inner ring. The bearing was press-fit with k5. The inner ring press created a residual tensile stress at the surface. Combined with the bending from misalignment, the local stress was about 80 MPa. Over a year, that’s enough to initiate a crack. The stainless was worse because it work-hardened — the press fit created more residual stress. The 42CrMo4 lasted longer because it was tougher, but it still cracked.
The fix wasn’t a stronger shaft. It was two things. First, we re-aligned the pump and motor within 0.05 mm. The bending stress dropped to near zero. Second, we switched the shaft fit from k5 to h6 (clearance). The bearing inner ring sits on a shaft with a slight clearance and is axially retained by a circlip. The residual tensile stress from the press disappeared. The shaft didn’t break for the next four years. We didn’t need a thicker shaft. We needed to stop abusing the existing one.
I still have the broken shaft on my bench. The fracture surface shows the beach marks, curving from a point about two-thirds of the way through the shaft. That initiation point is where the bearing inner ring started. The shaft itself was fine. The failure was in how it was mounted.
If you’re reading this because you keep breaking pump shafts, check the alignment first. Then check the fit. Don’t go straight to a better material. Better materials solve the wrong problem when the problem is mechanical abuse. The customer was relieved when we told them the 30 mm shaft wasn’t needed — they had paid extra for it. We went back to 25 mm C45 with a h6 fit and proper alignment. It outlasted everything we tried before.
