The answer to this question, as with many questions about Maintenance, is “it depends”.
I’ll illustrate with an example:
In a large multi-line pulp and paper mill, a 20″ knife-gate valve on the main effluent header failed, shutting down the entire operation for three days. As manufactured, the stainless steel valve yoke had been attached to the valve body with stitch welds and these welds failed, probably from fatigue. The pressure forced the spade out of the valve, and the escaping 180 degree effluent submerged the effluent lift pumps which fed the treatment system.
The question was asked – could this failure have been avoided if the preventive maintenance system had been better designed? My response was that it could not, at least not economically in a process plant like a paper mill. My reasoning was that if the scope of the PM inspection programme included regular checks on items such as manufacturers’ shop welds on all plant equipment, and all the other possible checks at that level of detail, then it would be unrealistically expensive. PM inspections for that plant should be limited to probable failure modes on most equipment, and this event was not the result of a probable failure mode.
After analyzing the downtime in that plant over a number of years, it turned out that the total losses from “probable” and “unlikely” failure modes were about equal. The unlikely failures were less frequent but each one resulted in a large production loss.
Of course, the depth to which it is reasonable (and economic) to search for potential failures depends on the criticality of the equipment, defined as the consequence of failure. During the design and maintenance of commercial aircraft, for example, each and every possible failure mode for each component is examined and actions are taken to either prevent failure or to prevent a failure from affecting the safety of the aircraft (e.g. by providing back-up systems). Such analysis in a large manufacturing operation is usually neither economic nor necessary, although for very critical systems, such as those containing hazardous chemicals, a much more rigorous design process and a correspondingly higher level of inspection detail than normal is certainly justified (see “Reducing the risk of unspared critical components“)
So what action should be taken when an unlikely failure occurs? One possibility is to extend the PM programme to include that mode of failure on all similar equipment. This is only practical if the failure development period for each failure mode is long enough to allow regular inspections to detect a potential failure with enough warning for repairs to be made before a breakdown occurs.
The best way to deal with unlikely failures is to conduct a root cause analysis when one occurs and take whatever action is required to prevent any future similar failures. This often involves some redesign, and in the case of the valve example, the attachment of the valve yoke to the valve body was redesigned to prevent further failures.
PM Programme Design
If the PM programme is well-designed, most of the “probable” failure modes with a reasonable failure development time can be avoided.
Good records of downtime are of great value when setting up a PM programme (see “Measuring Reliability“). In one plant which recorded all production losses against the equipment which caused them a Pareto analysis showed that 80% of all unscheduled maintenance downtime could be attributed to 87, or less than 1%, of over 12,000 equipment items. A new PM programme which focused on these 87 items, and similar equipment, resulted in a reduction in unscheduled maintenance downtime of over 50% within 18 months.
In a process plant, such as a pulp and paper mill, PM inspections should consume no more than 15-20% of available maintenance manpower. It is a very good “system check” to occasionally compare the total hours of work generated by your PM programme each year with your total available manpower.
To return to the Articles index click here.
© Veleda Services Ltd
Don Armstrong, P.Eng, President
250-655-8267 Pacific Time