RCA Close-Loop Checklist

Every RCA produces an action list. Ninety percent of plants never implement half of the actions. This checklist is how the other ten percent do it.

Companion to: Fix the System, Chapter 16. Who uses it: the RCA owner (the reliability engineer, senior tech, or planner who led the session). When: filled out at the end of each RCA session. Follow-up verifications at 60 days and at 6 months. Why this exists: the single largest failure mode of RCA programs is running good sessions and then never implementing the corrective actions. The list sits in a folder for six months and dies. This checklist is the gate between "RCA completed" and "RCA closed." Nothing closes without the 60-day and 6-month verifications.

Part A — At the session (day 0)

Fill out at the end of the RCA facilitation session, before anyone leaves the room.

RCA ID: ____________________

Asset tag: ____________________

Failure code under investigation: ____________________

Session date: ____________________

Owner: ____________________

Facilitator: ____________________

Attendees: (name + role)

1. _______________________________________________ 2. _______________________________________________ 3. _______________________________________________ 4. _______________________________________________ 5. _______________________________________________ 6. _______________________________________________ 7. _______________________________________________

Required roles (check that each is represented)

• Technician who has done the repair most often
• Operator who runs the asset
• Planner who schedules the PM
• Reliability engineer
• Area supervisor

If any required role is missing, the session should have been rescheduled. Document why it proceeded anyway: _______________________________

Root causes identified (from the 5 Whys chain, Fishbone, or FMEA output)

1. _______________________________________________ 2. _______________________________________________ 3. _______________________________________________ 4. _______________________________________________

Corrective actions assigned

Every corrective action falls into one of four channels. Log each action with its channel.

Channel key:

• PM — PM procedure update (rewrite, new task, frequency change). Feeds into Chapter 15 PM Compliance Recovery.
• CBM — Condition-based monitoring install (sensor, threshold, alert). Feeds into Chapter 22 / Playbook 8.
• SOP — Operating procedure change, training, operator action. Operations signs off.
• Design — Design change, spec change, spare-part substitution. Engineering owns.

Verification dates (assign at session)

• 60-day check date: ____________________
• 6-month check date: ____________________
• Verification owner: ____________________ (often the RCA owner; may be delegated to the planner)

Sign-off at session close

Part B — 60-day verification check

Completed on the 60-day check date.

Check completed on: ____________________ By: ____________________

For each corrective action from Part A, mark the implementation status:

Summary:

• Actions implemented: ______ of ______
• Actions in-progress with revised deadline: ______ (list action numbers: __________)
• Actions blocked — escalated to plant manager: ______ (list action numbers: __________)
• Actions abandoned — documented reason: ______ (list action numbers: __________)

Asset status at 60-day check

• Asset has not failed under the same failure code since the RCA.
• Asset has failed under the same failure code since the RCA. → Re-open the RCA. Document on next page. Root cause was not addressed.
• Asset has failed under a different failure code since the RCA. → Log as new entry for future watch-list build.

Sign-off (60-day):

Blocked actions — plant-manager escalation:

Any action that was blocked at 60-days gets escalated to the plant manager. Attach this page to the next weekly report.

Part C — 6-month recurrence verification

Completed on the 6-month check date.

Check completed on: ____________________ By: ____________________

Asset recurrence check

Pull the WO history for the asset since the RCA session date.

• Total WOs since RCA: ______
• WOs under the same failure code: ______
• WOs under related failure codes (same subsystem): ______

Verdict

• No recurrence. Zero WOs under the same failure code. The root cause was correctly identified and the corrective actions held. RCA closes.
• One recurrence, cause identified. One WO under the same code, investigation shows a different root cause than the one we addressed (e.g., a new failure mode emerged). Log as new watch-list entry. RCA closes on the original cause; new RCA scheduled on the new cause.
• Multiple recurrences — corrective action failed. Two or more WOs under the same code. Root cause was mis-identified or actions were insufficient. RCA does NOT close. Re-open for a second RCA session with the benefit of the additional data.

Supporting evidence

• All Part A actions listed as Implemented in Part B have been verified as still in effect (the procedure revision still shows in the CMMS, the CBM sensor still reports, the SOP is in the current binder).
• MTBF trend on the asset since the RCA: improving / flat / degrading (circle one).
• Operator and tech interview at 6 months: has the fix held from the floor's perspective? Y / N. Notes: ____________________

Sign-off (6-month close)

Final disposition:

• CLOSED — no recurrence. RCA filed as complete. Asset removed from watch list.
• CLOSED — new cause logged. Asset stays on watch list under the new failure code.
• RE-OPENED. Return to Part A for a second investigation.

What to avoid

• Closing the RCA at 60 days. The 6-month check is where half the recurrences show up. Premature close is the #2 failure mode of RCA programs (after "never implementing the actions"). Hold the close.
• Treating "Implemented" as the finish line. Implementation is necessary but not sufficient. The 6-month recurrence check is what verifies the implementation actually addressed the cause.
• Letting abandoned actions slide without documentation. If an action was abandoned, the reason goes in writing. This is how the program learns which channels work and which do not.
• Skipping the operator interview at 6 months. Data alone can lie. The operator knows whether the fix held on the floor. Ask.
• Accepting "we think it's better" as close. MTBF data, WO count data, or both. No vague confirmations.

End of checklist. File with the RCA session packet. Archive after final close.