Carrier A (synthetic), 737-800 operator | 3 spare engines | FY2027 maintenance budget ¥3,650M
Step through the months and watch the plan and decisions move
Every month brings news on shop TAT, parts lead times, unscheduled removals and passenger results. Press "Next month" to update the year-end outturn, which scenario reality is closest to, the value of switching plans, the demand vs. schedule comparison and the pack for the monthly review meeting (the engine review). All numbers are synthetic, and the same month always gives the same result.
New information this month
Pulled automatically from quotes, lead-time replies and removal records
Where will this year's maintenance spend land?
Starting from the costs already committed, we add the cost of the remaining shop visits, additional work and unscheduled removals, and estimate the outturn 2,000 ways. Each run's scenario is chosen according to this month's probabilities.
Distribution of the outturn (2,000 runs)
Horizontal: this year's maintenance spend (¥M) · Vertical: number of runs
Probability of going over budget, by month
Horizontal: fiscal month · Vertical: probability (%)
Committed costs
Engines already inducted and unscheduled removals (¥M)
Which scenario is reality closest to? Should we switch plans?
New information updates the probabilities of the three scenarios. We then compare the cost for the rest of the period of staying on the Baseline plan with switching to the plan prepared for each scenario.
24-month shop-visit plan
Which scenario reality is closest to (probability)
Horizontal: fiscal month · Vertical: probability (%)
Cost for the rest of the period (¥M): plan × scenario
Excludes engines already inducted. Alternative plans include the cost of moving booked induction slots and the value of part life that can no longer be used up when a shop visit is brought forward.
Add spare engines, or sign an engine pool agreement?
We simulate this year's removals and turnaround times 2,000 times and count, on the same yardstick for every option, the days aircraft are grounded for lack of an engine.
QCD side by side: what gets better, what gets worse
Q: aircraft-down days. C: added cost this fiscal year. D: passengers turned away (rest of the year). Farther out is better; the dashed line is "as is". No combined score: people decide
Levers that change the triangle itself (catalog)
The options above are knobs inside the plan: for a temporary deviation, use them in the response loop (OODA). For a structural one, pick from here and fix the assumptions and the plan in the planning loop (PDCA). These candidates are not in the calculation
Annual cost vs. aircraft-down days
Horizontal: annual cost of the option (¥M) · Vertical: aircraft-down days (expected per year, log scale)
Results by option
- 3 spare engines owned
- Leasing one more spare: ¥200M per year
- Engine pool agreement: delivery in 3 days, up to 2 engines at a time
- Shortfalls of up to 2 days are absorbed by juggling aircraft assignments
- Loss per aircraft-down day: ¥4M
Review meeting pack and decision log
A planning AI agent runs the numbers and drafts the pack; a review agent checks them against contracts, requirement documents and manuals. Decisions are logged with deadlines.
Agent activity log
Planning agent and review agent (scripted replay)
Decision log
Grows as you step through the months
How far did past plans drift from actuals? What should feed into the next assumptions?
We line up the range each annual plan version forecast against the actual outturn. If the gaps lean one way, the assumption behind them (here, shop turnaround time) is updated with actuals and used in the next plan.
Forecast vs. actual by plan version
Horizontal: annual maintenance spend (¥M). Bands show the 10–90% range; vertical lines the median.
Distribution of shop turnaround time
Horizontal: TAT in months, normalised to a 4-month plan. Dots are actuals for returned engines.
Plan versions
Which aircraft fly which routes
The answer to which aircraft type (737, 767, 787, A350) flies how many of each day rotation, solved from route demand forecasts (fleet assignment). An aircraft that flies ends up at another airport, so each day is built as a rotation that leaves Haneda and comes back. Haneda slots bind the trunk: in busy months widebodies go to the trunk and the 737s are shared with regional routes. How those 737s fly comes back as engine wear.
How the 737s are used (18 months)
Company A's 56 737-800s split into trunk, regional, checks and spare (median demand). Shaded months leave more than ¥30M of passenger revenue behind. Months with more spare aircraft are easier for engine shop visits
Routes and fleet assignment
A day's flying (flight sequences)
Leave Haneda and return to Haneda; each flight departs from the airport the previous one arrived at. Engine changes and spare engines are only at Haneda (the maintenance base)
2028 slot reallocation (airport side)
The government allocates Haneda slots to each airline and reviews them every 5 years (next in 2028). If the trunk slots go up or down, how do the round trips, revenue, 737s and engine shop visits move? Other airlines' flights are held fixed
Airport and policy scenarios
Of the passengers turned away from full flights, the share who rebook on another flight of the same airline (same or adjacent day). The case value is 15%; 30–70% are assumptions. The plan is held as it is
Revenue from passengers left behind (Haneda flights held fixed, demand growing 3.8% a year)
Widen Haneda (Company A, FY2030)
Reallocate slots (change in margin for Companies A and B and the whole market, ¥100M a year, FY2030)
The Linear moves Haneda–Itami passengers to rail and frees Haneda slots
Whether each airline moves the freed slots to other trunk routes (New Chitose, Fukuoka, Naha). Change in margin (¥100M a year; left Company A, right Company B). Three shares of air from the government forecast
Upcoming decisions
Put widebodies on the trunk, share the 737s, lease aircraft, or let passengers go. The deadline is the schedule freeze (3 months ahead)
Demand growth: assumption and actuals
Horizontal axis: fiscal month. Vertical axis: passengers, year on year (%)
Decisions and results
Decisions in the month each schedule was frozen. Aircraft are leased only when the lease costs less than the revenue it wins back
We can run the same calculation on your fleet and contract terms.
This opens the contact form on the RE-X corporate site (in Japanese). Please mention “MRO AI Assisted Solver”.