2010
AGIFORS Scheduling and Strategic Planning Study Group Meeting
Lausanne, Switzerland
PRELIMINARY AGENDA, as of 8 June 2010.
All times and presentations described are tentative, and are subject to
change.
Meeting
Location: Starling Hotel
24 June 2010
0830 – 0900: Registration and light breakfast
0900 – 0910: Welcome: Roger Parker, Tim Jacobs
0910 – 0930: Welcome and overview of EPFL
0930 – 1015: Round the room
1015—1045: Break
1045—1130: Key Note Presentation (TBD)
1130—1215: AirVM:
The Airline Passenger Virtual Market Simulator. Part I:
R.
A. Parker, Virtual M1nds.
This presentation will demonstrate the global agent-based model of worldÕs airline passenger market that has been developed by Parker and his colleagues at Virtual M1nds, SA. The system architecture, underlying models and data issues will be described. How the simulator can assist in service change evaluations, pricing studies, strategic analysis and other aspects of schedule and service planning will be illustrated with live demonstrations of the simulation, using a small, hypothetical network of some 500,000 passengers and 14,000 flight legs. This presentation will be in two parts. The main discussion will be this one, with a brief additional presentation after lunch. The discussion is wrapped around the lunch hour to allow time for a Monte Carlo simulation to execute.
1215—1330: Lunch
1330—1400: AirVM: Part II
R.
A. Parker, Virtual M1nds.
Continuation of the AirVM discussion from before lunch.
1515—1530: Heuristics of Estimating the Effects of Codesharing (tentative title)
Marc-Alexandre Laroche, Ecole Polytechnique de
Montreal
Codesharing is a widespread practice that allows an airline to put one of its flight numbers on a flight operated by a partner airline and consider it in its own network. Even though it has been repeatedly demonstrated that this type of partnership can generate important additional revenues, it is very complex for an airline to choose which flights to share with which partners. We propose a heuristic method that maximizes the revenues of an airline considering two aspects that were not explicitly considered in the literature: the impact of adding a shared flight in the network and the interactions that exist between flights that are chosen to be codeshared. At each iteration, the demand is distributed on the network considering a selection of codeshare flights. Based on the resulting network flow, this selection is modified in order to increase the profitability of the carrier. A mathematical model of demand distribution is built and validated using existing software before being used in the heuristic. This method was used to optimize the codeshare flight selection of Air Canada with one and two partners. In comparison with two methods currently used, these algorithms propose codeshare flight selections that generate greater revenues.
1530—1600: Break
1600—1645: Gate Assignment Modeling at United Airlines
A gate assignment model is developed that will be used by both the
schedule planning group and station personnel for assigning gates to aircraft
turns. There are a variety of constraints that the model needs to adhere to:
gates can handle only certain aircraft types, some gates are unavailable for
certain periods in the day, physical layout of neighboring gates does not allow
aircrafts of certain types to be gated at the same time, separation is required
between two departures from certain neighboring gates, rest periods are required
between two consecutive turns at the same gate to absorb delays and tows have
to be minimized given limited tow-off sites and personnel/equipment. The model
can gate multiple stations at a time. The model can also retime flights to
create feasible solutions while accounting for network effects and station
specific constraints across the network. An integer programming model is
formulated to solve this problem. Since the run time of the entire problem is
prohibitive, we develop a heuristic decomposition based approach that solves
the problem an order of magnitude faster giving good quality solutions.
1645—1730: Abstract: Robust
Scheduling for Airlines
Niklaus Eggenberg, TRANSP-OR, EPFL
Due to economic pressure
industries, when planning, tend to focus on optimizing the expected profit or
the yield. The consequence of highly optimized solutions is an increased
sensitivity to uncertainty. This generates additional "operational"
costs, incurred by possible modifications of the original plan to be performed
when reality does not reflect what was expected in the planning phase. The modern research trend focuses on
"robustness" of solutions instead of yield or profit. Although robust
solutions have a lower expected profit, they are less sensitive to noisy data
and hence generate less operational costs. In this talk, we focus on the
robustness of airline schedules. We compare different existing methods for
"robust scheduling" on simulated data in order to analyze their
performance. In particular, we analyze the consequences of erroneous prediction
models on the performance of robust solutions.
1730—1800: Pre-dinner break
1800 – 2330: Dinner
25 June 2010
0830 – 0900: Registration and light breakfast
0900—0930: AirVM: Part III
R.
A. Parker, Virtual M1nds.
A look at the Monte Carlo results started the previous
night, if the attendees decide to execute a large sample simulation
0930—1015: ClipAir: A Modular Multi-modal
Transportation System
Bilge Kucuk,
Claudio Leonardi, Matteo Salani, Prem Kumar Viswanathan
TRANSP-OR,
EPFL
Demanding
reduction in CO2 emissions and continuous pressure on ticket prices are pushing
towards radical modifications in future objectives for the air transport
industry. Operators are asked to consider fundamental structure change and new
approaches to fleet management. We present part of a feasibility study for a
new modular multi-modal transportation system called "ClipAir". In this work, we focus on the
integrated schedule generation and fleet assignment problem and compare the
performances of a regular operator and a fictional operator running a ClipAir fleet. Comparison is made in terms of expected
operating costs and demand satisfaction.
1015—1045: Break
1045—1130: Incorporating Customer
Choice Methodology into Airline Network Optimization
Sergey Shebalov, Sabre
Holdings, Diego Klabjan, Northwestern University, Sumit
Kunnumkal, Indian School of Business, Milind Sohoni, Indian School of
Business
We present an airline network developing framework
that includes automation of several decision making processes. One of these
processes is flight retiming that allows obtaining additional revenue by
improving network connectivity without violating any of the resource related
constraints. Another process is codeshare agreements
optimization that identifies the most profitable way to extend airlines network
via marketing flights operated by its partners. Finally we consider the problem
of operated network expansion through introduction of services for new markets
and frequency increase. We use a well-established customer choice modeling
methodology to model OD and recapture effects associated with retiming and codesharing. Our approach allows
incorporating these effects directly into the optimization formulation and
therefore eliminating any approximation for demand interaction. We tested our
model on several real world scenarios and initial computational results
indicate significant revenue increase.
1130—1215: Presentation (TBD)
1215—1330: Lunch
1330—1415: Presentation (TBD)
1415—1500: Presentation (TBD)
1500—1515 Break
1515—1530: Business Meeting and Close. R. Parker and T. Jacobs