CBL - Campus del Baix Llobregat

Projecte llegit

Títol: Viabilidad operativa del mantenimiento ecualizado en una flota Airbus A330 de largo radio: modelo de optimización y simulación frente a la operación real en bloque

Estudiants que han llegit aquest projecte:

Director/a: FORNÉS MARTÍNEZ, HECTOR

Departament: FIS

Títol: Viabilidad operativa del mantenimiento ecualizado en una flota Airbus A330 de largo radio: modelo de optimización y simulación frente a la operación real en bloque

Data inici oferta: 31-05-2025     Data finalització oferta: 31-01-2026


Estudis d'assignació del projecte:
    GR ENG SIST AEROESP
Tipus: Individual
 
Lloc de realització: EETAC
 
Paraules clau:
CAMO, A330, Manteniment aeronàutic, manteniment equalitzat, aeronavegabilitat continuada, flota de llarg radi, programació lineal entera mixta, MILP, Monte Carlo, algoritme genètic, GA
 
Descripció del contingut i pla d'activitats:
Estudi del procés de consolidació d'aerolínies al mercat europeu a través d'adquisicions. Anàlisi financera d'una adquisició hipotètica. S'examinarà l'impacte sobre el compte de resultats, cash flow i balanç consolidat. Es valorarà l'evolució del deute i la capacitat financera resultant. L'estudiant haurà de construir escenaris realistes amb dades públiques o estimades. Treball orientat a finances corporatives i gestió econòmica d'empreses del sector. Es recomana tenir curiositat per aprendre models financers i coneixement del sector aeri.
 
Overview (resum en anglès):
This project evaluates, from the perspective of a CAMO organisation, the operational feasibility of an equalized maintenance strategy for a fleet of seven long-haul Airbus A330 aircraft, and contrasts it with the reference operator's actual block-check operation. The central question is whether equalized maintenance, which breaks long maintenance groundings into short visits spread across the calendar, can reduce hangar occupancy without degrading fleet utilisation.
The methodology chains three phases built on real operator data: fleet status, route network, approved maintenance programme and defect history. The first phase builds the flight plan with a mixed-integer linear programming (MILP) model that optimises a cyclic week with continuous departure times, balances flight hours across aircraft and guarantees a permanent backup aircraft. The second phase distributes the complete maintenance programme over that plan with a flight-anchored temporal-coverage MILP: every flight must depart with all its tasks within limits, heavy maintenance is restricted to a single hangar bay, and maintenance windows are created by cancelling flights when no natural ground time exists. Task durations without historical records are extrapolated with Random Forest algorithm. The third phase subjects the plan to one year of stochastic non-routine maintenance and aircraft-on-ground (AOG) events, calibrated against the real defect history, through discrete-event simulation and Monte Carlo with 2000 replications, where each delay-or-cancel decision emerges from cost functions based on public references. A genetic algorithm solves the second phase as a methodological benchmark.
Results show that the equalized scenario is operationally viable: the plans execute 100% of the maintenance programme with no flight departing with expired tasks, cancelling between 3.4% and 4.5% of planned flights at the planning stage. Exposed to random events, the plan retains around 92.5% of planned flight hours, which places simulated utilisation between 99.1% and 100.4% of the fleet's actual flight hours: a tie with the block operation, tilting against equalized maintenance once the non-technical disruptions outside the model are considered. The most actionable finding concerns the hangar: past the start-up transient, a single bay sustains heavy maintenance for all seven aircraft, matching the operational indicators of the uncapped scenario with 43% less aircraft time inside the hangar. The genetic algorithm covers the programme but schedules 17% to 22% more task executions and never reaches the exact method's hard feasibility, which confirms the choice of MILP.

© CBLTIC Campus del Baix Llobregat - UPC