RCM Scope Decision for a Boiler Feedwater System
A plant operates three boiler feedwater pump trains (two running, one standby) feeding a continuous process. Historical records over four years show: pump seal leaks (11 events, detected by operator rounds, average 3 days to repair), motor bearing degradation (4 events, vibration-monitored, planned replacement), and one catastrophic coupling failure that caused a 9-hour unplanned outage. A consultant proposes full preventive replacement of seals every 6 months and bearings every 12 months on all three trains. Maintenance cost is already 22% over budget. Operations argues that condition monitoring is cheaper; maintenance argues that condition monitoring has already missed the coupling failure. The plant wants a defensible maintenance strategy per train without increasing total cost.
1. Apply reliability-centered maintenance logic to decide the maintenance approach for each failure mode. Justify each choice.
Read the worked answer
Seal leaks (11 events, operator-detectable, 3-day repair). This is a hidden-to-failure? No - it is an evident failure with a clear precursor (leak) and a low consequence per event. RCM logic favors on-condition/operator inspection rather than time-based replacement, because the failure is detectable and the consequence is modest. Scheduled seal replacement every 6 months on all trains would replace good seals and add cost without addressing the dominant driver (installation quality, alignment, or operating point).
Motor bearing degradation (4 events, vibration-monitored, planned replacement). Condition monitoring is working: failures were caught and converted to planned work. RCM logic supports continuing condition-based maintenance with a defined alarm-to-action interval. A 12-month blanket replacement is redundant where monitoring already provides adequate warning.
Catastrophic coupling failure (1 event, 9-hour outage). This is the one case where the consequence is high and the precursor may be subtle. RCM would ask whether there is a detectable precursor (vibration, coupling inspection) and whether the interval is short enough. The right response is to strengthen the condition-monitoring task specifically for couplings (add coupling inspection to rounds, set vibration alarm bands for coupling frequencies) rather than to conclude that condition monitoring as a whole failed. A single missed event is a task-design problem, not proof that the strategy is wrong.
Per-train conclusion: Apply the same mode-based logic to each train, but let criticality drive depth: the standby train may need a different inspection cadence because it runs less and its failure is hidden until demand.
2. How would you respond to the maintenance team's claim that condition monitoring 'already missed the coupling failure,' and how would you keep total cost from rising?
Read the worked answer
Reframe the claim. One missed event does not invalidate condition monitoring; it points to a gap in the monitoring task (wrong parameter, wrong interval, wrong alarm threshold, or an inspection that was skipped). Ask: was the coupling on the route? Was the alarm band set for coupling frequencies? Was the last inspection actually performed and signed off? The answer determines whether the fix is task design or execution discipline.
Cost-neutral design.
- Remove blanket 6-month seal replacement and 12-month bearing replacement where condition monitoring already provides warning; this frees budget.
- Reinvest that budget in coupling-specific monitoring and in root-cause work on seal leaks (alignment, piping strain, operating point), which attacks the 11-event driver.
- Use criticality to tier tasks: full monitoring on running trains, reduced cadence on standby, with a startup check before switching.
- Track cost per train and failure consequence per train so the trade is visible.
Governance. Define alarm-to-action times, audit route completion, and review missed events in a formal failure review so the strategy is adjusted on evidence rather than abandoned.