Conveyor Drive Shaft Failure at a Shoulder Fillet

The failure is almost certainly a bending-fatigue crack initiating at the 50 mm shoulder fillet, not at the keyway. The shoulder is the geometric discontinuity: a 1.5 mm radius on a 60-to-50 mm step gives a sharp change in stiffness, and the rotating shaft applies fully reversed bending at that plane every revolution. The keyway is 8 mm away and is a secondary stress raiser; the fracture location would confirm which dominated.

Nominal stresses at d = 50 mm:

• Bending moment: M = 4.2 kN × 0.120 m = 504 N·m
• Section modulus: Z = πd³/32 = π(0.050)³/32 = 1.227 × 10⁻⁵ m³
• Nominal bending stress: σ = M/Z = 504/1.227 × 10⁻⁵ = 41.1 MPa, fully reversed
• Polar section modulus: J/c = πd³/16 = 2.454 × 10⁻⁵ m³
• Nominal torsional stress: τ = T/(J/c) = 850/2.454 × 10⁻⁵ = 34.6 MPa, steady

The bending stress is alternating (σa = 41.1 MPa, σm = 0); the torsion is steady (τm = 34.6 MPa, τa = 0). The notch has its effect through Kf = 2.1, so the local alternating stress is about 86 MPa before any mean-stress interaction.

First correct the endurance limit:

Se = Se′ × ka × kb × kc = 340 × 0.90 × 0.85 × 0.814 = 211.8 MPa

Alternating stress with the notch factor applied to the nominal bending stress:

σa = Kf × 41.1 = 86.3 MPa

For the steady torque, use the von Mises equivalent mean stress:

σm = √3 × 34.6 = 60.0 MPa

Goodman line:

1/n = σa/Se + σm/Sut = 86.3/211.8 + 60.0/630 = 0.4075 + 0.0952 = 0.5027

n ≈ 1.99

The calculated factor is about 2.0, below the 2.5 target even before accounting for the field history. That is the key judgement: a calculated 2.0 is not reassuring for a conveyor drive that experiences start-up torque, possible misalignment, and a coupling change that increased the overhung load. The actual load may now exceed 4.2 kN, so the real factor is probably below 2.0. The 14-month life is consistent with high-cycle fatigue at a notch rather than a single overload. Also note that Kf = 2.1 is the fatigue notch factor; if it were applied to both alternating and mean stress the result would be more conservative, and if the keyway also carries load the local stress is higher still.

The root cause is the combination of a sharp shoulder and a load path that was never checked after the coupling change. Priority actions:

• Increase the fillet radius substantially, for example to 3–4 mm, or use a relief groove or tapered shoulder. This directly reduces Kf and is the single most effective change at the initiation site.
• Increase the coupling-end diameter from 50 to 55–60 mm, or use a larger shaft with a shrink-disc coupling that eliminates the keyway. Stress falls with d³, so this is a strong lever.
• Check the new coupling: mass, overhung moment, balance, and alignment. A heavier hub increases both the steady load and the alternating bending. Require laser alignment records at installation and after thermal growth.
• Specify a flexible coupling and verify that the shaft does not see the coupling's misalignment moment.
• Consider induction hardening or shot peening of the fillet, but only after the geometry is fixed; surface treatment does not compensate for a poor notch.
• Review start-up and jam torque. If the drive can stall against a jam, add a torque limiter and check the shaft for peak load, not just running load.

Maintenance changes: record alignment, coupling condition, and vibration at each service; inspect the fillet at the first opportunity; and treat a coupling replacement as an engineering change requiring a load check, not a like-for-like spare-part swap. Geometry and load control address the root cause; hardening or a larger safety factor alone would merely delay the same failure.