The engine which suffered damage was an eight-cylinder two-stroke marine diesel engine arranged to drive a large alternator. The first indications that it had a problem were loud knocking sounds. Unfortunately, before the engine could be brought to a stop, one of its connecting rods separated from the crank pin and emerged from the crankcase, causing substantial damage. The final state of the connecting rod can be judged from the picture opposite. As the various damaged parts became accessible, possible alternative causes were tested for credibility. By this process, attention was increasingly drawn to the need to find the crank bearing bolts, which were not initially accessible. When they were retrieved, one of them was found to be virtually intact, but without its nut. This nut, its locking device and locking screw were all found separately. The nut threads for about one and a half turns nearest to the lower face were seen to have been damaged by axial shear.

About one quarter of the first  turn, although present, had been separated from the nut around its circumference. The other bolt had broken under severe combined tensile and bending load and its nut, with locking device and screw, was still present on the broken end. Furthermore, a bolt in a different connecting rod was found to be 1 mm slack, with its nut locking device and locking screws correctly fitted.

This evidence clearly showed that the slackening and loss of the intact bolt was the initiating cause of the engine damage. In searching for the reason why the bolt became slack, the effects on the bolt of possible cylinder malfunctions were calculated and physical tests of similar bolts were conducted. The key finding from this work was that if the bolt had been properly tightened, the probability that sufficient torque could have been generated to loosen the nut against the frictional resistance of the threads and of the nut face was minimal. The torque due to tensile load, which normally acts to unscrew a nut, was found to be one eighth of the frictional torque resisting such movement. It was concluded, therefore, that the damage resulted because the bolt had not been properly tightened to the correct tensile load.

The routine maintenance procedures adopted for this installation involved periodic checks of the bolt tension. The method used was to slacken and re-tension the bolts and, as such, this procedure had the potential to introduce error.