اشكان تيموري

Revenue management and dynamic pricing for airline fare classes are considered one of the most important decisions of airlines. A systematic approach requires the obtaining of cutting-edge optimization models and techniques. Considering the increasing need of airlines for these decision support systems (DSS), this thesis discusses the decision-making process in airline revenue management and pricing at strategic and operational levels by presenting models and methods of mathematical programming and optimization. The thesis mainly focuses on providing optimization models for dynamic pricing and seat inventory management. Due to this purpose, a model for pricing and seat inventory control in multi-fare classes presents (MFC-PSIC model). Then, by considering customer choice behavior (CCB) to choose different flights during the day, which is quantified by the multinomial logit (MNL) model, we develop our initial model to a model for dynamic pricing considering customer choice behavior (MFC-PSICCCB model). In another extension of our original model, we introduce two new sales mechanisms, named customer price selling (CPS) and temporary general selling (TPS), and address the dynamic airline pricing problem by considering multiple sales channels (model MFC-PSICMCS), we model. Finally, the main model of this research in operating revenue management and airline pricing is presented with the abbreviated MFC-PSIC®, in which: 1) seat inventory control and ticket pricing are integrated, 2) customer choice behavior is considered, 3) sales channels are developed, 4) service level control and ticket refund modules are considered, and finally, 5) demand uncertainty is controlled using the adaptive multistage robust optimization (MSARO) technique. To solve the mentioned models, two solution methods based on the progressive hedging algorithm (PHA) and the combined rolling horizon-benders decomposition (RHBD) method are provided to solve the problem in larger dimensions with more computational efficiency than CPLEX Solver. In the implementation phase of the MFC-PSIC® model, if a disruption occurs in the initial flight plan, we provide a model for rescheduling the crew and aircraft so that by considering options such as substitution and flight cancellation, the profit of airlines under disruption is maximized. In addition to a model based on the robust fuzzy-stochastic programming (RFSP) approach, a meta-heuristic search method based on the adaptive large neighborhood search (ALNS) is developed to solve this problem. Finally, to optimize airlines strategic decisions and policies in different states of the sales and revenue management system and determine the optimal strategic price level by considering assumptions such as stationary and memoryless, an optimization model based on the Markov decision process (MDF) is presented. This model determines the optimal action of airlines in each state of revenue management. In the end, all the models and solution methods proposed in this research are eva‎luated using several numerical studies, and the analysis of the numerical results is done so that in addition to validating them, several management insights in airline revenue management and pricing can be extracted and presented to the industry.

مديريت درآمد خطوط هوايي , قيمت‌گذاري پويا , بهينه‌سازي استوار و تطبيق‌پذير , رفتار انتخابي مشتريان , مدل لاجيت چندجمله‌اي , الگوريتم پوشش پيش‌رونده , الگوريتم افق غلطان , الگوريتم جستجوي همسايگي تطبيق‌پذير , فرآيند تصميم‌گيري ماركوف

Airline Revenue Management , Dynamic Pricing , Multi-Stage Adaptive Robust Optimization , Customer Choice Behavior , Multinomial Logit Models , Progressive Hedging Algorithm , Rolling Horizon Algorithm , Adaptive Large Neighborhood Search , Markov Decision Process

Ashkan Teymouri

Hadi Sahebi