This question is from Mark in the Petrochemical Industry:

For average criticality machinery monitored periodically with portable vibration data collectors, how often should the data be collected and should that interval be changed when a problem is observed?

Bob: 

Between 25 and 30 years ago, the typical portable vibration data collection interval was quarterly.  This found plenty of problems on operating machinery in advance of their functional failure which led to the concept that we “saved” those machines.  Unfortunately, many machines that did not show problems in the quarterly data failed before the next quarterly collection cycle.  This led to monthly being the norm, starting between 20 and 25 years ago, and resulted in many more problems being identified so that they could be corrected before the machine failed.  Monthly remains the norm for routine portable vibration data collection.  Anything that requires data more frequently over the long term is more cost effectively handles with a permanently installed on-line monitoring system.

When portable data is collected that shows the first indication(s) of a developing problem – but not so severely that it needs to be removed from service immediately – the prudent thing is to collect data on it much more frequently, generally starting within the next day or two.  This is because once the initial degradation has been identified, the next thing that is needed is to determine how quickly it is progressing.  The actual problem could have started just yesterday, or it could have started a mere couple of hours after the previous collection.  This leads to determining the urgency of the need for corrective action and supports planning and scheduling prioritization.  In many cases, once the problem is discovered, it could require daily or even shiftly monitoring to help nurse it along until it can be safely and cost effectively removed from service to implement the appropriate corrective action.

Brian: 

What I have found over the past years in the industry is that a monthly collection will give you the best picture of machine health for most standard machine setups.  Collecting data on monthly intervals gives you a good picture of machine health and rate of degradation over time once an anomaly is detected.

As far as adjusting collection cycle once a fault is detected, in my opinion would be decided by a few factors.

1. What fault is detected?  Does this fault type typically cause rapid failure or will the unit run without losing operational capability for an extended period of time.  Typical failure modes detected through vibration analysis will give ample time prior to failure to allow a planned and scheduled repair without increased data collection intervals.  Some failures can occur within hours of initial detection, at which point online systems would typically be used to monitor.
2. What is the scheduled window of time for repair?  If an available time for repair is not readily available, increasing collection frequency can help determine rate of progression and whether the unit will make it to the next available repair window.
3. What is the availability of replacement parts?  Similar to the response above, replacement parts are not always readily available and knowing rate of progression can help determine whether parts need to be expedited or not.  Increased frequency of collection in these scenarios will once again assist in determining timeline to failure.

Like most things, one answer won’t cover every scenario.  Looking at multiple factors can help determine the proper monitoring frequency for the specific scenario, whether it be continue at current schedule or increase collection to weekly, or even daily; it all depends on the answers to the above questions.