The history of PODS

Chronology

  • 1994: Conceived at Boeing to support airline planning; relationship quickly developed with MIT researchers
  • 1999: MIT/PODS Research Consortium formed; major airlines support and influence PODS development and academic research
  • 2000-2011: PODS code develops; airlines benefit from insights; more than 30 MIT dissertations based on PODS simulations and findings
  • 2012: PODS Research, LLC formed to provide airline consulting, support further PODS development, and ensure continued access to PODS for the MIT/PODS Consortium

Founding and development

In 1994, Craig Hopperstad began development of PODS for Boeing’s purposes, to move beyond the practice of validating revenue management algorithms with simulations that made exactly the same assumptions as the optimization schemes being tested. The core idea was to build a model in two parts – generated passengers who choose amongst available path/classes and revenue management systems that set path/class availability based on the airline’s history of passenger passenger bookings.

The objective of this integrated simulation was to replicate both passenger choice behavior and the actual functions of airline revenue management (RM) systems. The simulation models the interaction between RM booking limits and passenger decisions with respect to flights and fares, and tests the impacts of RM practices on the market shares, traffic and revenues of the different competitors in airline networks.

In 1995, Boeing joined with MIT (and Dr. Belobaba) to improve and expand the representation of revenue management forecasters and optimizers within PODS. MIT began investigations of some basic issues in revenue management using PODS (e.g., is RM a zero-sum game?). Early evaluations of simple scenarios in which competing airlines have differing RM systems in a single market found interactions between the revenue gains of RM and the relative sophistication of seat inventory control practiced by different airlines. Evaluations of RM demand forecasting and unconstraining models showed that the choice of RM demand estimation techniques can yield as many revenue benefits as the choice of booking limit optimizers.

The next big technology step in revenue management arose from path-based (ODIF) control rather than the leg-based control that was incorporated initially in PODS. Substantial industry involvement guided the further development of RM capabilities in PODS, and thus, in 1996, a high level steering group on research capabilities was constituted consisting of Bill Brunger (Continental Airlines), Erik Tilanus (KLM), Jorn-Peter Petersen (SAS) and Louis Busittil (Cathay Pacific).

In 1999, Hopperstad retired from Boeing and founded Hopperstad Consulting. Boeing granted a license for PODS to Hopperstad Consulting, allowing it to take over responsibility for PODS programming and development. In turn, MIT researchers took the primary role in the large-scale testing of alternative scenarios and evaluation of the impacts of airline revenue management systems under competitive market conditions. Building on the PODS research at MIT, Belobaba proposed and created a formal MIT research consortium. The MIT/PODS Revenue Management Research Consortium was launched with an initial membership of Continental, Northwest, KLM, SAS, Lufthansa and Swissair.

From 2000 to 2011, the Consortium flourished, with PODS demonstrating its capability to simulate and evaluate changes in the airline industry (e.g. the advent of low cost carriers with fare products with few restrictions). In this period, the Consortium also made several significant contributions to the industry itself, including the original formulation of “Q” forecasting. At the same time, Hopperstad Consulting also engaged in multiple consulting assignments developing client proprietary versions of PODS.

In 2012, PODS Research, LLC was formed with Hopperstad, Belobaba and Brunger as the principal founders, to continue developing PODS and to provide Consortium support, as well as providing consulting services to airlines.