Vacation-season staffing: why July breaks your man-hour plan — and the January fix

Every October, someone on my courses laughs and says "we'll sort leave in spring." Every August, the same person is cancelling weekend duty rosters because four engineers are at the beach and a 600-hour check is mid-package. Vacation-season staffing is not an HR admin task — it is a man-hour plan input, and AMC1 145.A.30(d) expects your plan to account for it. Here is the arithmetic, with numbers you can lift into your January planning meeting.

01Why leave clusters whether you plan it or not

School holidays, family logistics and the flying season all pull in the same direction: in most European and offshore operations, 60–75% of annual leave is requested inside the same eight weeks (roughly mid-June to mid-August). That is also precisely when utilisation peaks — offshore crews fly longer days, and charter demand fills the hangar with A-checks and unscheduled inputs. Your demand curve and your absence curve point the same way at the same moment.

• Leave is paid workload you must subtract. An engineer with 25 days' leave gives you roughly 11% fewer productive hours than the calendar suggests — before training and sickness.
• Clustering converts a small average into a deep weekly hole. Spread evenly, 8 engineers' leave is invisible. Clustered, it removes half the crew in specific weeks.
• First-come-first-served is a coverage policy — just a bad one. Whoever submits leave first wins, regardless of the check schedule.

02The worked example: eight engineers, one July

Base maintenance team of 8 engineers, each with 25 days' annual leave. Each engineer works ~220 days/year. If leave lands wherever people want it, a realistic worst week in July looks like this — while a 600-hour check package is in input:

And this is the optimistic version: it assumes no findings, no sick days, and no second aircraft. Add one rectification of 40 man-hours in W3 and the slip doubles — because with four engineers away, there is nobody to flex.

03The January fix: leave density cap, approved once a year

Three mechanisms, in order of what actually works:

• Cap leave density, not leave totals. Max 1 of every 5 certifying staff (20%) on leave in any week that touches a committed check input. For an 8-engineer team that means max 2 away — never 4. Publish the cap in January with the check schedule attached, so approval and coverage are decided in the same meeting.
• Price the shoulder season. If June and September leave costs the operation nothing, say so explicitly — "June and September requests approved same-day." You are steering demand with information, which is free.
• Book leave into the man-hour plan as a workload line. 8 engineers × 25 days = 200 engineer-days ≈ 1,600 man-hours that simply do not exist. Your plan should show them as a reduction, not discover them in July. That is also your audit answer when an inspector asks how leave was considered in the workload assessment.

04The carry-over trap nobody prices

There is a second failure mode in the same arithmetic. Engineers who cannot take summer leave carry it into Q4 — and in many jurisdictions, untaken statutory leave becomes a liability the company must pay out or grant at the worst possible time. If your summer plan only "works" by refusing leave, you have not solved the problem; you have moved it into November, where it collides with winter check season and training-recency deadlines. A leave plan that survives contact with reality must clear ~85% of annual leave by 30 September, every year, measured as a number — not as a vibe.