Skip to main content
Thesis defences

PhD Oral Exam - Shayan Tavakoli Kafiabad, Industrial Engineering

Design and Planning of Maintenance Logistics Networks

Monday, October 18, 2021 (all day)

This event is free


School of Graduate Studies


Dolly Grewal



When studying for a doctoral degree (PhD), candidates submit a thesis that provides a critical review of the current state of knowledge of the thesis subject as well as the student’s own contributions to the subject. The distinguishing criterion of doctoral graduate research is a significant and original contribution to knowledge.

Once accepted, the candidate presents the thesis orally. This oral exam is open to the public.


The system up-time and availability play an essential role in industrial, maritime, aeronautical, and health care sectors. These sectors utilize, in general, several advanced capital systems, such as gas turbines, radars, airplane engines, and MRI-scanners. Most of these technical devices contain expensive and low-demand repairable parts. The management of maintenance logistics networks deal with decisions both in the design and planning phases. In the design phase, goals such as the allocation of users to maintenance centers and spare-part provisioning are pursued. On the other hand, the planning phase deals with decisions in terms of workforce, capacity, and aggregate planning. Maintenance planning is a hard task due to several conflicting constraints, such as sporadic demand, uncertain repair and inspection time, limited capacity, and availability of resources (inventory and certified operators). Our problem of interest is mainly motivated by maintenance logistics networks in the context of gas turbine engines. The maintenance service providers to these devices are confronted with the interaction of workforce training and operations planning along with demand and repair time uncertainty, that introduce new challenges to the management of these logistics networks.

In the first part of this thesis, we devise a decision model to obtain the optimal size of workforce, training schedule, repair quantity, as well as number of repair jobs to outsource so as to minimize the cost of repair operations, spare part stock, training, outsourcing, and penalties incurred for the delayed delivery of repaired equipment over a planning horizon. Then, we evaluate the impact of integrating workforce training with operational planning decisions in maintenance facilities. Besides, we analyze the role of risk mitigation strategies such as outsourcing of repair jobs to other maintenance centers and borrowing of certified operators in the presence of demand fluctuations by formulating a two-stage stochastic programming model.

The second part of the thesis is an effort to incorporate the repair time uncertainty into the decision model developed in the first contribution. We propose a multi-stage stochastic programming model for integrated production and workforce planning under independent random repair times of faulty components. Then, we develop an approximate decomposition algorithm, based on Lagrangian relaxation approach, to efficiently solve the problem for real-size instances. This algorithm relies on decomposing the MSP model into sub-models corresponding to component scenario trees and coordinating them via a sub-gradient algorithm to obtain a high-quality feasible solution.

In the final part of this thesis, given an MLN that provides maintenance/repair services to geographically dispersed equipment users, we propose a two-stage robust optimization model for collaborative design and planning of maintenance networks under demand uncertainty. The goal of this model is to determine the optimal allocation of customers to each maintenance center along with the initial stock level of different spare parts in each facility so as to minimize the cost of late deliveries under worst-case demand scenarios. We consider component and operator sharing strategies as the recourse actions in this model to hedge against the demand surge. The proposed approach is compared with a deterministic model by the aid of Monte-Carlo simulation on several test instances inspired by a real case study.

Back to top Back to top

© Concordia University