Root Cause Analysis of Hydrogen Compressor Failure

Recently a major chemical company asked that a root cause analysis be conducted on the failure of a hydrogen compressor. The situation was that the plant had two moderate sized, slow speed (900 rpm) compressors used to increase the pressure of hydrogen so it could be used in the plant fuel system. The pair of compressors was roughly thirty-five years old and each of the two had been used almost equally in the lead (or operating) role and stand-by (or idle) role.

At about the thirty-five year point, the first of the two compressors experienced a broken shaft. The shaft was approximately four and one-half inches in diameter and made of mild steel. After the first failure, no investigation was done and the shaft was replaced with a material that was thought to be slightly higher quality.

Within a few months the shaft on the second compressor failed in much the same manner. In both cases, had the shaft been released from the compressor case, there would be a four-plus diameter hole releasing almost pure hydrogen into the plant. Had the gas found a source of ignition, an explosion or fire could have resulted. As a result, a root cause investigation was initiated.

While a variety of possible causes were investigated, the findings were fairly straightforward. Despite the fact that the compressors ran at fairly low speed, the number of cycles added up to several billion over seventeen and one-half years of operation. Because of the relative low speed, it was felt that the couplings did not need to be balanced and the machines did not need to be accurately aligned. Misalignment and imbalance of a large gear coupling had caused the shaft to operate beyond its fatigue limit. The fact that the two failures had occurred so close in time was a testament to the fact that the two machines had been used almost equally over their lives.

Expert Analysis:

With this root cause, the solution is not so simple. The next failure (in fact the next two failures) is likely to occur in another thirty-five years. The present value of a failure that will occur thirty-five years our can be amortized back to a very small present value. As a result there is very little economic justification for doing anything. On the other hand, if a small material flaw or an unexpected combination of forces from imbalance and misalignment leads to earlier than expected failure, the results could be costly.

As a result, it was recommended that the few extra dollars be spent to balance the coupling and conduct at least a crude alignment.

To see the resume of the expert associated with this case study, see the link below.