Le 02/01/2022 par houssin :
Stage M2 à l'ISAE-SUPAERO (Toulouse)
Encadrants : Alain Haït, Laurent Houssin, Ronald McGarvey
Title: Repair prioritization for a service support contractor
Service support contractors for high-value equipment, such as aircraft, increasingly support performance-based contracting (PBC) strategies, under which contracts are designed to “optimize system readiness" allowing for a purchaser who “obtains a comprehensive performance package, not individual parts, transactions, or spares [and] repairs actions"
One of the key decisions facing such a contractor is determining, in real time, which failed components to induct into repair, given that multiple components can be repaired with a common pool of workforce and resources
Given the desire of PBC approaches to optimize system readiness, the impact of any individual component failure or repair should be stated in terms of its effect on aircraft availability
Assuming that once a repair is started, it cannot be interrupted (and thus must run to completion), it might be preferable to defer the induction of failed low-priority component, in the event that a failed higher-priority component might require repair in the future.
This could apply in instances where the repair of the low-priority component requires a long time duration.
Alternatively, if the budget available for repairs is nearly exhausted, it might be preferable to defer repair of the low-priority component to a future budget period
The main analytic challenges are (1) the future failure of components is unknown, (2) the actual repair time for components is unknown, (3) the relative “priority” of any given component, in terms of its effect on aircraft availability, is dynamic and cannot be readily specified a priori.
One approach to address these uncertainties is to model the repair enterprise as a stochastic system. Given the extremely strong desire to avoid high values of unavailable aircraft, instead of making decisions based on an expected future availability, we propose to make decisions based on a conditional value-at-risk (CVaR) measure of aircraft availability.
The objective of this research will be to develop optimization models that identify a repair versus defer decision for each failed component, on a rolling horizon, such that aircraft availability CVaR is maximized, subject to constraints on repair capacity and repair budget.
Contexte de l'étude :
Le stage est prévu au sein du groupe Systèmes Décisionnels (SD) du Département d’Ingénierie des Systèmes Complexes (DISC) de l'ISAE (10, avenue Édouard-Belin, 31055 Toulouse)
Le candidat recherché est en cours de Master 2 ou en dernière année d’école d’ingénieur (bac+5).
De bonnes connaissances en recherche opérationnelle, optimisation combinatoire et optimisation stochastique seront appréciées.
Documents à fournir : CV détaillé, lettre de motivation, notes et rang sur les 2 dernières années.
Contact : Alain.Hait@isae.fr, Laurent.Houssin@isae.fr