Failure of an M24 Engine Mounting Bolt

In this article, Bill Eccles of Bolt Science presents a real-world failure investigation involving an M24 property class 8.8 bolt used to secure engine mounts on a bus chassis. The case study illustrates how inadequate preload sets off a chain of events — joint movement, bending fatigue, and eventually self-loosening — that no secondary locking device can reliably stop once it has started.

This is a failure mode that plays out in stages, which is part of what makes it difficult to diagnose. The first sign was bolts found loose in service. A split pin was added to prevent nut back-off. The split pins started shearing. Bolts began fracturing. Each intervention addressed a symptom rather than the root cause, and the failures continued until the joint design itself was revisited.

What the article covers:

• Why the specified torque wasn’t being achieved — The design department specified 660 Nm (487 lb-ft). That torque was achievable on prototype vehicles, but space constraints and inadequate tooling meant production assembly and field maintenance were consistently achieving closer to 400 Nm (295 lb-ft) — roughly 60% of the required value. The gap between specified and actual torque was the origin of every subsequent problem.
• How a multi-layer joint compounds embedding loss — The joint consisted of several alternating steel and aluminum sections. Each interface contributes to preload loss through embedding, and the joint analysis showed that when tightening scatter was factored in alongside embedding losses, a proportion of joints would reliably fall below the minimum preload needed to resist the applied loads. The preload requirement chart from the analysis makes this visible in a way that a single torque value does not.
• The two consequences of insufficient preload — Once preload dropped below the threshold needed to maintain friction grip, lateral joint movement became inevitable under braking, acceleration, and cornering loads. That movement produced two parallel failure mechanisms: bending stresses induced into the bolt at the thread runout region — an area already prone to stress concentration due to incomplete thread form — leading to fatigue fracture; and repeated transverse slip driving progressive self-loosening, consistent with Junker’s findings on the primary cause of rotational loosening.
• Why the split pin fix failed — Adding a split pin below the nut addresses nut rotation, not joint movement. With insufficient preload, the joint continued to move laterally, the bolt continued to experience bending fatigue, and the mechanical loads on the split pin were sufficient to shear it entirely in some instances. The fix treated the symptom and left the cause intact.
• The redesigned solution — Rather than attempting to enforce the original torque value in an environment where it couldn’t be consistently achieved, the solution was to change the fastener. An M16 flange-headed property class 12.9 bolt was specified at 380 Nm (280 lb-ft). The higher strength class and improved length-to-diameter ratio reduced both the torque requirement and the sensitivity to embedding loss. The resulting preload range of 86–137 kN comfortably exceeded the 78 kN minimum requirement, and the preload requirement chart for the modified design confirmed adequate margin across the full assembly scatter range.

The case demonstrates a principle that applies well beyond bus chassis engineering: when a tightening specification cannot be reliably executed in the field, the answer is usually a better joint design, not a more aggressive enforcement of the original torque value.

About Bolt Science 

Bolt Science was founded in 1992 to provide independent technical expertise in bolted joint technology. The company offers bolted joint analysis software, consulting and problem-solving services, fastener and joint testing, and training on bolting technology. Their client list includes many of the world’s major engineering organizations. Learn more at boltscience.com.