Fatigue Failure of Compressor Bolting PDF Print E-mail
Written by Stephen H Shakeshaft   
Thursday, 04 March 2010 20:11
Article Index
Fatigue Failure of Compressor Bolting
Further Investigation
All Pages


A plant operator on a routine walk around his plant taking readings noticed that one of the cylinders on a reciprocating compressor was moving up and down.

Immediately realising something was wrong he shut the unit down straight away and informed the plant management.

Description of Machine

The machine was a single stage double acting compressor with two cylinders.

On this machine, the cylinders were bolted to a distance piece which was in turn bolted to the crankcase of the machine. This is a standard configuration for this type of machine.

The bolted joint where the cylinder bolted to the distance piece was where the operator saw the unusual movement occurring on the machine.

Just one cylinder was affected, the other was still attached.


With the machine made safe one of the plant technicians removed inspection covers in order to gain access to the bolts which at this joint were inside the distance piece.

On this design of machine the cylinder is attached to the distance piece by a ring of 16 stud bolts fitted with “Nyloc” locking nuts.

It was immediately obvious that many of the stud bolts were in fact lying in the bottom of the distance piece. In fact of the 16 bolts, 10 were lying in the bottom of the distance piece, the nuts still in place on the broken studs.

With the cylinder supported the remaining 6 stud bolts were numbered in their positions and then removed.

The broken stud bolts were cleaned up and looked at by a materials laboratory.

In addition, the 6 other stud bolts were sent away for non destructive testing (NDT).

Each of the 10 broken stud bolts was clearly a fatigue failure. The fracture face was smooth across most of its surface where the fatigue cracking had propagated, with a small roughened area where the final tensile failure had occurred.

NDT on the remaining 6 bolts showed that 4 had cracked and only two were found to be free of cracking.

The attachment of the distance piece to the crankcase was inspected and the majority of the bolts were found to have loosened. Some significantly so. The stud bolts from this joint were removed and sent for NDT. Again there were 16 bolts, but this time just two were found to be cracked.

The plant maintenance records were very poor, but the maintenance supervisor who had been on the plant for some time could remember that it was not unusual that when the machines were worked on a bolt may be found broken or a nut found to be vibrated off inside the distance piece.

Apparently, during the early years of operation the nuts vibrated off the cylinder attachment stud bolts so frequently that the manufacturer changed the nuts to 'Nyloc' self locking nuts.

This cured the problem of the nuts coming off, but it had not solved problems with the stud bolts snapping off.

Even though the machines were 190 years old at the time of this failure, they had only ever been worked on by the manufacturers service representatives.

It was said that the service technicians had always been vigilant in torquing up the nuts to the design value. In fact there was no reason to doubt this, each representative had his own calibrated torque wrench issued by the company for precisely this purpose.

Last Updated on Thursday, 04 March 2010 20:22