PhD Oral Exam - Yasser Ghamary, Industrial Engineering

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.

Abstract

This thesis addresses the complex and computationally challenging problem of scheduling aircraft maintenance operations at the operational level. These maintenance activities are critical for ensuring aircraft safety and reliability but involve intricate interdependencies, resource constraints, and sequencing requirements that render the scheduling problem NP-hard.

To tackle this, we develop a mixed-integer linear programming (MILP) model that captures the key constraints of the problem, including task durations, inter-component dependencies, resource capacity limitations, and intra-task activity sequencing. Due to the computational intractability of solving the full MILP for real-world instances, we propose a hybrid solution approach combining a tailored Iterative Local Search (ILS) heuristic and a Lagrangian Relaxation (LR) framework.

The ILS heuristic is designed to efficiently generate high-quality feasible schedules by exploring the search space through domain-specific neighborhood moves and restart strategies. It serves as both a practical scheduling tool and a source of upper bounds for evaluating solution quality.

Complementing this, the LR framework decomposes the original problem by relaxing combinations of resource, dependency, and sequencing constraints. We employ a projected subgradient optimization method to solve the resulting dual problems, using adaptive step-size strategies to improve convergence stability. The LR approach provides lower bounds for the optimal solution, allowing us to assess the quality of heuristic solutions and gain insight into the impact of different constraint sets.

Comprehensive computational experiments on small and large problem instances demonstrate the effectiveness of the proposed methods. The ILS produces near-optimal solutions in reasonable time, while the LR method yields informative bounds and highlights structural properties of the problem. Results show that adaptive step-size strategies enhance convergence behavior without increasing computational effort.

This thesis contributes a robust, scalable framework for operational maintenance scheduling and offers a foundation for future work in stochastic modeling, parallel computing, and real-time decision support for aviation maintenance operations.