You need to determine what failure modes you are experiencing. For each failure mode, determine the P-F interval. This is the time interval between when the failure can be detected using vibration monitoring equipment and when the machine fails to provide the needed function. For a rolling element bearing with a well-designed and maintained lubrication system, this interval should be several weeks. In this case, portable vibration monitoring once per month should be able to detect the failure.
We have approximately 1400 centrifugal pumps. Most of them use oil lubricated ball bearings. We take vibration data on all of these pumps once per month. This vibration program detects approximately 80 percent of all detectable failures before the operators know that there is a problem.
In general, I don’t recommend continuous vibration monitoring for any pump that has a full spare. For pumps without a full spare, I would do a cost benefit analysis. If the pump uses rolling element bearings, the result is likely to be that the cost cannot be justified. Portable vibration analysis once per month is sufficient. For pumps with hydro-dynamic bearings, the answer is often the same. The money could be better spent on a very good oil system with redundancy, monitoring, alarms, trips and auto-starts.
We have continuous vibration monitoring on four pumps out of 1400. All of these are critical, unspared and greater than 1000 HP. We are going to add wireless monitoring to some high speed pumps which are known to fail very quickly. These pumps are integrally geared and have internal, shaft driven oil pumps. The lube oil system has no redundancy, no monitoring, no alarming, no trips and known failure modes which can occur with little or no detectable symptoms. In addition, the pumps are in light hydrocarbon service with a potential for a very serious event if there is a large leak. Even with all of these shortcomings, the monitoring is justified by the risk of a large release, not maintenance cost savings.
Johnny Pellin