University of Washington

COVID-19 has reshaped the global airline industry. Travel demands are volatile, and passengers have more flexibility in bookings and cancellations. More than ever, airlines have to be agile and adaptive in terms of their operational decisions. In this talk, we introduce a new dynamic model for airlines to make adaptive scheduling and fleeting decisions, based on evolving demand and booking signals. We derive interesting theoretical, algorithmic, and managerial insights from this model. We present computational experiments based on real-world scenarios to demonstrate potential benefits of this approach.

Mike Zoutendijk

Delft University of Technology

Flight delay is commonly predicted as a class or value. However, the aviation industry can benefit from probabilistic individual delay predictions, as these give insight into the uncertainty of the predictions. Therefore, in this study, two probabilistic forecasting algorithms, Mixture Density Networks and Random Forest regression, are applied to predict flight delays at a European airport. The algorithms estimate the delay distribution with a Mean Absolute Error under 15 minutes. To illustrate the utility of the estimated delay distributions, we use these probabilistic predictions to increase the robustness of flight-to-gate assignments. Considering probabilistic delay predictions, our flight-to-gate assignment model reduces the number of conflicted aircraft by up to 74% when compared to a deterministic assignment model. In general, the results illustrate the utility of considering probabilistic forecasting for robust airport operations’ optimization.

Pranav Gupta - Amadeus IT Group

Luicano Donoso - LATAM Airlines

Schedule planning problem for is solved with little or no stochasticity in the inputs with the objective to maximize profitability. Because of this, the highly optimized schedules might show an opposite trend when compared against operational metrics. Having an approach that evaluates the schedule for reliability can help airlines to select schedules that are both profitable and reliable. Besides, COVID has changed the schedule structures and operational efficiency for several airlines. Traditional methods to improve reliability by adding block and are sub-optimal and costly, hence, smart allocation of buffer based on the schedule structure can improve the reliability in an efficient manner. LATAM Airlines (LA), in partnership with Amadeus, has been using Simulation technology to measure their reliability and explore different operational and schedule strategies to create robust schedules

Khaled Abdelghany -  Southern Methodist University

This study presents a framework for flight block time design for airline schedule planning while considering schedule reliability. Historical actual block time (ABT) data is collected for several U.S. airlines, and the variability of the block time is presented. The difference between the scheduled block time (SBT) and the ABT is investigated to understand how airlines use padding minutes to achieve the desired on-time performance. An optimization model is developed to allocate the padding minutes in the schedule to achieve a user-defined on-time performance. The schedule padding strategies and the potential block time savings are compared for several U.S. airlines.

Fabio Ghielmetti, Timo Koch, Dominique Schmitt-Bohlender

Lufthansa Group

Optimizing ones airlines network is challenging enough. Now imagine trying to optimize six hubs. A key process of Lufthansa’s network planning and scheduling teams is the network and hub optimization process, which optimizes the total network. How to choose the right destinations, design the optimal bank structures, decide on frequencies, departure times and aircraft equipment across the multi-hub network. During the last years process improvements have been applied, but the optimization algorithm stayed mainly unchanged. To make use of modern technology Lufthansa Group started to develop a new optimizer to replace the existing solution already before the Covid crisis. With the help of the new optimization solution Lufthansa Group targets to gain increase network profitability.

Pranav Gupta - Amadeus IT Group

David Hagerhjelm, John Gatu - Southwest Airlines

Airline network planning and scheduling is one of the most critical functions for any airline as it defines its revenue potential. Due to the combinatorial nature of this problem, airlines have been using advanced OR techniques and models to optimize their schedules for the last three decades. Solving the schedule planning problem involves finding an optimal solution that meets their commercial objectives (network of flights, differing aircraft types, uncertain demand, competition, etc.) and operational constraints (maintenance requirements, gate feasibility, airport slots, curfews, crew legalities, etc.). Airlines need to solve this problem every season and have traditionally relied on incremental optimization techniques which rely heavily on a previous schedule. Post-COVID, airlines have had to re-think their schedule structures to handle new and changing market conditions.

Eric Dowty, James Crasta, Matt McElfresh

Boom Supersonic

Commercial supersonic travel is set to return to the skies at the end of this decade. The return of faster than sound travel will present unique challenges and opportunities. This presentation will review the latest approaches to forecasting the operational and economic performance of supersonic travel on various routes. It will include detailed reviews of supersonic optimized routing & approaches to forecasting flight times, fuel burn, & operating costs.

Thomas Billet - KLM Royal Dutch Airlines

Willem-Jozef Van Goethem - Aviation Decision Sciences

Paul Roling - TU Delft

Operations Research has focused on long-term strategic fleet planning for commercial airlines. Recent global events show that humanitarian air operations require shorter-term fleet decisions due to the urgent nature of their operation. This method combines fleet planning and flight schedulling at a tactical time frame, weeks to months. The method uses a multi-commodity network flow model to size an initial fleet. Next, the output is used in a Fleet Size and Mix Vehicle Routing Problem model to increase the accuracy of vehicle routes. The results show a weekly routing costs reduction of 40% compared to expert flight planners who schedule and route humanitarian requests on a daily basis and reduce the fleet size by 60% from 14 air assets to 6. The model demonstrates that OR can be effectively used as a decision support tool for aircraft contracting, flight routing, and scheduling to increase the efficiency and effectiveness of humanitarian air operations.

Mihaela Mitici, Madalena Pereira, Fabrizio Oliviero

Delft University of Technology

Electric aircraft are expected to serve soon as a replacement for conventional, short-range aircraft. In this presentation we address the main operational challenges for short-range flights operated with electric aircraft: determining the investment needs for a fleet of electric aircraft, and the logistics of charging stations and swap batteries. A mixed-integer linear program with two phases is proposed. In the first phase, a schedule for flight and battery recharge is developed for a fleet of electric aircraft. In the second phase, optimal times for battery charging are determined, together with an optimal sizing of the number of charging stations and swap batteries. We illustrate our model for short-range flights to and from a large European airport and for an electric aircraft designed based on the operational characteristics of a conventional, narrow-body aircraft.

Alexey Tarasov, Max Andreev

Rosterize Inc

It happens that a slight adjustment of the flight departure time allows the airline to reduce the cost of crew positioning and accommodation, increases crew utilization and minimizes the risk of delays due to unavailability of crews. We have made a tool that automatically allows flight planners to find such flight schedule adjustments. This algorithm provides the most significant effect during the peak seasons when the crew demand is higher than the crew supply, and the number of crew sets is the primary constraint of the airline. Based on data from several airlines, we have shown a reduction of the crew demand by 10%, which allows an increase in the number of flights. The algorithm also helps airlines introduce new flights more profitably, optimize existing schedules, evaluate the convenience of an airport time slot for crew planning, and simplify negotiations with the airport.