MRO manpower scheduling: the weekly roster your annual plan never built
Your man-hour plan says the year is staffed: 12 engineers against 4,400 forecast man-hours, coverage ratio 1.05, auditor satisfied. Then week 47 arrives — a 600-hour check in the hangar, an A-check due, line coverage to hold — and the roster fails by half. Nothing was wrong with the annual math. The failure is the missing middle: nobody converted the annual plan into weekly demand. Here is the four-step bridge, with the numbers.
01Why annual averages and weekly reality disagree
An AMC1 145.A.30(d) man-hour plan is deliberately annual: forecast workload, divide by productive hours per engineer, get a headcount. But maintenance doesn't arrive annually — it arrives in lumps. A 1,500 MH check is 1.2% of a big annual programme and 145% of a single week's capacity. Roster-level scheduling is where the plan is proven or exposed, and it's the step most organisations skip between "plan approved" and "who's on shift Tuesday."
02Step 1–2: build weekly demand, then real supply
Take every committed workstream landing in the week and price it in man-hours — not heads, not shifts, hours:
| Week 47 workstream | Basis | Man-hours |
|---|---|---|
| Line & daily coverage | 2 engineers × 8h × 6 days | 96 |
| A-check event (AW139) | 1 full package | 207 |
| 600 FH check (week 2 of 3) | 1,500 MH ÷ 3 weeks | 500 |
| Unscheduled / findings allowance | 15% of scheduled | 120 |
| Weekly demand | 923 |
Now the supply side — with the actual availability factor, not the contractual one. Twelve engineers at 40 hours is 480 gross; after leave, training, sickness and meetings at a realistic 28%, you have 346 productive hours. Coverage ratio: 923 ÷ 346 = 2.7× demand over supply. And the category check is worse: roughly 28% of package hours need a certifying signature — 258 certifying MH against five certifying staff supplying 144. The week fails on signatures even harder than on hands.
check it per license category — B1 and B2 separately
03Step 3: level-load the peak — four moves, in order
- Stretch the check. The 600h package goes from 3 weeks to 5: 500 → 300 MH/week. Most packages tolerate a slower input if the delivery date moves — negotiate it with the forecast, not from the hangar floor.
- Slide events inside their interval window. The A-check's ±10% interval tolerance lets it fly one more week into the gap after the check input. Demand drops 207 MH.
- Re-price the allowance. Unscheduled allowance recalculates on the new scheduled base: 15% of 396 = 59 MH.
- Buy exactly the shortfall. New week-47 demand: 96 + 300 + 59 = 455 MH against 346 — still 109 short. That is not "hire someone"; that is 2.5 contractor-weeks, booked for specific dates, at a known cost.
Certifying check after smoothing: 28% of 455 ≈ 127 certifying MH against 144 available — passes, which is why the check crew can include contractor mechanics but your own certifying staff must stay on the critical path. Four moves turned a 2.7× failure into a manageable 109-hour purchase. That's the entire discipline: the annual plan tells you that you're staffed; the weekly build tells you when you're not.
04Step 4: lock it, then feed the variance back up
Once a week is level-loaded, freeze it as the roster baseline and track actual hours against it. Those weekly variances are not just ops data — they are the raw material of your quarterly AMC1 145.A.30(d) plan review and the monthly deviation monitoring behind the 25% rule. An organisation that schedules weekly never gets surprised by its own quarterly review: the deviation trend is visible eight weeks before the auditor's spreadsheet shows it. Build next quarter's thirteen weekly demand rows now; it takes an afternoon with the calculator and saves you the crisis meeting.
Turn your annual plan into weekly demand rows: add your fleet, set the check schedule, and see which weeks break — workload vs. coverage with an auditor-ready PDF.
Build the weekly model →