4 Presentations

• 
  01:30 PM - 01:55 PM

  Willingness-to-Pay Estimation and Pricing Optimization for Airline Seat Assignment

  • Yury Sambale, Air Canada
  • Sajad Aliakbari Sai, Air Canada
  • Adam Bockelie, Air Canada
  • Liu Tianjiao, Air Canada
  • Alan Regis, Air Canada
  • Cindy Yao, Air Canada
  • Mohsen Dastpak, presenter, Air Canada
  • Qianzhe Wang, Air Canada
  • Kejun Li, Air Canada
  • Aldair Alvarez, Ivado Labs
  • Teodora Dan, Ivado Labs
  • Carl Perreault-Lafleur, Ivado Labs
  • Emma Frejinger, Ivado Labs
  • Andrea Lodi, Ivado Labs
  • Guillaume Rabusseau, Ivado Labs

  Air Canada allows customers to purchase a preassigned seat on each segment of their travel journey, as an "à la carte” product during the booking flow. Multiple dimensions impact a customer’s willingness-to-pay (WTP) like the seat product and the chosen branded fare; a bundle that complement the airfare with additional ancillaries, such as refundability, seat reservation, and checked baggage. Decision makers have traditionally relied on a trial-and-error method to price preselected seats aiming to maximize revenue while considering these dimensions. With very few seat assignment pricing optimization models in the literature that can be applied in a practical setting, we present a bi-level optimization approach, that leverages machine learning to estimate the customer’s WTP in the lower-level problem, and an optimization model that generate optimal prices in the upper-level problem. We also shed light on a field experiment methodology implemented to improve data quality and enable a data-driven approach.

• 
  01:55 PM - 02:20 PM

  Optimization of a Multi-echelon Inventory Management Problem

  • Jean-François Landry, presenter, Ivado Labs
  • Carlos Zetina, CIRRELT
  • Louis-Philippe Bigras, Ivado Labs
  • Jean-François Cordeau, HEC Montréal, GERAD, CIRRELT
  • Yossiri Adulyasak, HEC Montréal

  Multi-echelon inventory planning remains a challenging task, especially when faced with uncertainty at various levels and the need for high robustness. In this context, Ivado Labs has been contracted to develop a generic solution that offers significant improvements over the classical safety/cycle stock methods currently employed by our client in managing uncertainty. This work proposes a model for managing inventory across a multi-echelon network for finished goods, addressing uncertainties in both lead time and demand. The approach begins with solving an initial optimization problem through a stochastic formulation that considers demand uncertainty. This is followed by a simulation phase aimed at refining the solution by applying a grid search to address lead time uncertainty. Key Performance Indicators (KPIs), including fill rate, holding costs, aged inventory, and stock outs, are closely monitored. Finally, we detail results demonstrating substantial enhancements over the existing policies, as observed in steady-state operations for a dataset encompassing over 2,800 individual products over a 12-month planning period.

• 
  02:20 PM - 02:45 PM

  School Bus Trips Forecasting for EV Charging Plan Optimization

  • Augustin Parjadis de Larivière, presenter, Ivado Labs
  • Maëlle Zimmermann, Ivado Labs
  • Teodora Dan, Ivado Labs
  • Pierre-Luc Bacon, Ivado Labs
  • Emma Frejinger, Ivado Labs
  • Andrea Lodi, Ivado Labs
  • Jorge Mendoza, Ivado Labs

  Cleo is a Hydro-Québec subsidiary providing solutions for transportation companies to manage their electrical vehicle (EV) fleet. In particular, they help companies plan and schedule charging operations of their EV fleet to ensure vehicles are charged at the right level, at the right time, while minimizing energy costs. Throughout our collaboration with Cleo, we created several AI-driven components aimed at enhancing their solution: 1) a charge plan optimizer which seeks to minimize energy cost and maximum power demand, outputting how much power should be injected into each vehicle at each point in time; 2) a prediction module which predicts the duty cycles of EVs based on telemetry data, thus automatically providing the inputs needed by the optimizer and bypassing the need for manual inputs. The latter focuses on the case of electric school buses only and will be generalized to other business segments in a subsequent project.

• 
  02:45 PM - 03:10 PM

  Optimizing pickup and delivery for sustainable chemical solutions in the water treatment industry

  • Tu-San Pham, presenter, Ivado Labs
  • Caroline Rocha, Ivado Labs
  • Rodrigo Alves Randel, Ivado Labs
  • Camilo Ortiz, Ivado Labs
  • Jean-François Cordeau, HEC Montréal, GERAD, CIRRELT
  • Guy Desaulniers, GERAD - Polytechnique Montréal
  • Jorge Mendoza, HEC Montréal

  In this work, we address the optimization challenges of a multifaceted pickup and delivery problem encountered by a global leader in sustainable chemical solutions for the water treatment industry. The main objective is to adeptly incorporate various complex operational constraints, enabling an adaptable decision-making framework for fleet management, route planning, resource allocation, inventory balance, and service maintenance across North America. Through the application of route enumeration and hierarchical optimization techniques, our solutions demonstrate significant improvements in both cost-effectiveness and operational efficiency.