MNT-FLEET — Learn from maintenance history

• Situation: One pump group has fewer recorded failures, a proposed monitoring service promises improvement, and an upcoming task already lacks an applicable spare.
• Question: Which programme, task or support change does the available evidence warrant?
• First useful result or blocker: A comparable fleet account and supported retain or change decision, or the event, exposure or applicability discrepancy that prevents the relevant conclusion.
• Start with: MNT.13 - Coordinate the Maintenance Programme and Fleet Learning, using MNT.3 and MNT.12 only where the failure or history basis needs work.
• Stop or return: Retain an adequate current policy or fix a known support deficiency on its existing basis. Further inquiry is selected only for a worthwhile attainable answer.

1. Make the comparison mean the same thing. In the 48-pump case, twenty pumps have four relevant failures over 20,000 operating hours; 28 have six over 56,000 hours. The crude rates are 0.20 and approximately 0.107 events per 1,000 hours. Counts favor the first group, rates the second. MNT.3 establishes what counts as a failure, and MNT.12 attaches events and exposure to the actual components and intervals.
2. Propagate a corrected history before changing the policy. If two of the second group’s six records are preventive removals, correct that event classification and the affected rate. Duty, component histories and selection into the groups still limit any causal policy comparison.
3. Keep a policy decision separate from a support decision. MNT.2 - Select and Reopen the Maintenance Policy can retain the current task where its existing failure-and-response basis remains adequate. The weak comparative data do not establish that adequacy. MNT.13 can separately select repair of the known spare deficiency, using MNT.5 to establish readiness for the upcoming task. A broader redesign of recurring support returns to the relevant enabling-arrangement practice; it need not wait for a causal fleet study.
4. Compare the proposed method at the response it can improve. MNT.14 - Compare Maintenance Methods for Retention or Change compares the online predictor with current periodic interpretation at the same maintenance use. In its case, existing provider information shows no improvement in feasible response: the same specialist and part constraints dominate, while integration and interpretation add work. Retain the present method for that use and address support. Another family with rapidly changing condition and an available response can reopen the comparison for that family.

The policy alternatives also depend on the function. An obvious, low-consequence indicator failure may justify deliberate run-to-failure; a hidden protective-function failure needs an applicable means of finding it.