OPERATIONS RESEARCH IN THE AIRLINE INDUSTRY

INTERNATIONAL SERIES IN OPERATIONS RESEARCH & MANAGEMENT SCIENCE

Frederick S. Hillier, Series Editor Department of Engineering-Economic Systems and

Operations Research Stanford University Stanford, California

Saigal, Romesh The University of Michigan LINEAR PROGRAMMING: A Modem Integrated Analysis

Nagurney, Anna/ Zhang, Ding University of Massachusetts @ Amherst

PROJECTED DYNAMICAL SYSTEMS AND VARIATIONAL INEQUALITIES WITH A PPLICA TIONS

Padberg, Manfred/ Rijal, Minendra P. New York University

LOCATION, SCHEDULING, DESIGN AND INTEGER PROGRAMMING

Vanderbei, Robert J. Princeton University LINEAR PROGRAMMING: Foundations and Extensions

Jaiswal, N . K . Ministry of Defense, INDIA

MILITARY OPERATIONS RESEARCH: Quantitative Decision Making

Gal, Tomas / Greenberg, Harvey J. FernUniversität Hagen/ University of Colorado @ Denver ADVANCES IN SENSITIVITY ANALYSIS AND PARAMETRIC PROGRAMMING

Prabhu, N . U . Cornell University FOUNDATIONS OF QUEUEING THEORY

Fang, S.-C./Rajasekera, J.R./ Tsao, H.-S.J. North Carolina State/ Int'l. University of Japan/ University of California ENTROPY OPTIMIZATION AND MATHEMATICAL PROGRAMMING

OPERATIONS RESEARCH IN THE AIRLINE INDUSTRY

EDITED BY

Gang YU University of Texas at Austin

Austin, Texas, U.S.A.

W Springer Science+Business Media, LLC

Library of Congress Cataloging-in-Publication Data

Operations research in the airline industry / edited by Gang Yu. p. cm. ~ (International series in operations research &

management science; 9) Includes bibliographical references and index. I S B N 978-1-4613-7513-5 I S B N 978-1-4615-5501-8 (eBook) DOI 10.1007/978-1-4615-5501-8 1. Airlines-Management. 2. Airlines-Timetables. 3. Air traffic

control. 4. Aeronautics, Commercial. 5. Operations research. I. Yu, Gang. II. Series. HE9781.063 1998 387.7'0685-dc21 97-35539

CIP

Copyright © 1998 by Springer Science+Business Media New York Originally published by Kluwer Academic Publishers in 1998 Softcover reprint of the hardcover 1st edition 1998 Fourth Printing 2002. A l l rights reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, mechanical, photo­copying, recording, or otherwise, without the prior written permission of the publisher, Springer Science+Business Media, L L C

Printed on acid-free paper.

This printing is a digital duplication of the original edition.

CONTENTS

PREFACE

ACKNOWLEDGMENTS

1 MODELS AND METHODS FOR MANAGING AIRLINE IRREGULAR OPERATIONS Michael F. Arguello, Jonathan F. Bard, and Gang Yu

1 Introduction

2 Problem Definition

3 Mathematical Representation 4 Time-Band Approximation

5 Randomized Search Heuristic

6 Preliminary Results

7 Conclusions REFERENCES

2 A LARGE-SCALE NEURAL NETWORK FOR AIRLINE FORECASTING IN REVENUE MANAGEMENT Xiaoyun (Sherry) Sun, Erik Brauner, and Sharon Homby

1 Introduction

2 Choosing a Forecasting Model

3 The Specific Problem Statement

4 The BANKET Approach

5 BANKET Performance and Implementation

6 Conclusion

REFERENCES

xiii

xix

1 2 5 7

17 29 35 42 44

46 46 48 51 52 61 64 65

VI OPERATIONS RESEARCH IN THE AIRLINE INDUSTRY

3 A TUTORIAL ON OPTIMIZATION IN THE CONTEXT OF PERISHABLE-ASSET REVENUE MANAGEMENT PROBLEMS FOR THE AIRLINE INDUSTRY Lawrence R. Weatherford 68

1 Definition and Taxonomy of Perishable Asset Revenue Man-agement Problems 69

2 Taxonomy 3 Seat Allocation Decisions for the Fare Classes 4 Overbooking

72 82 85

5 Heuristics that Deal with Customer Diversion or Sell-up 86 6 Single Flight Leg vs. Network Origin and Destination Approach 87 7 Optimal Pricing 94 8 Summary 97 REFERENCES 98

4 A SELECTIVE MULTICOMMODITY NETWORK FLOW ALGORITHM FOR AIR TRAFFIC CONTROL Marcia P. Helme 101 1 Introduction 101 2 Problem Description 102 3 Assumptions of the Basic Space-Time Network Model 108 4 Solution Method 114 5 Conclusions 121 REFERENCES

5 A METHOD FOR OPTIMALLY SOLVING THE ROSTERING PROBLEM Michel Gamache and Fran~ois Soumis

1 INTRODUCTION 2 The Rostering Problem 3 Proposed Model 4 Computational Performance of the Algorithm 5 Productivity and Solution Quality REFERENCES

121

124 125 127 131 140 149 155

Contents vn

6 AN APPROACH FOR JUST-IN-TIME AIRLINE SCHEDULING Ira GershkoJJ 158 1 The Unsung Cost Problem 158 2 Airline Demand Economics 161 3 Approach to Just-in-Time Scheduling 164 4 Implementation Problems 178 5 Strategic Benefits of JIT Capacity Adjustments 181 6 Setting Up a Just-in-Time Scheduling Operation 184 7 Conclusions 187

7 AIRCRAFT GROUND MOVEMENT SIMULATION Dennis F.X. Mathaisel and Husni !dris 189 1 Background 190 2 The GMS Host 193 3 Human Interfaces 212 4 The GMS Graphics Display 218 5 Underlying Network for Taxiways and Runways 221 6 Modes of Operation 225 7 Hardware and Software Considerations 226 REFERENCES 226

8 CREW PAIRING OPTIMIZATION Erik Andersson, Efthymios Housos, Niklas Kohl, and Dag Wede/in 228 1 Introduction to Crew Pairing Optimization 229 2 Review of Existing Optimization Methods 233 3 The Carmen Pairing Construction System 244 4 Practical Experience with the Carmen System 253 5 Conclusions and Future Developments 256 REFERENCES 256

9 A DECISION SUPPORT FRAMEWORK FOR CREW MANAGEMENT DURING AffiLINE IRREGULAR OPERATIONS Mark Song, Guo Wei, and Gang Yu 259

Vlll OPERATIONS RESEARCH IN THE AIRLINE INDUSTRY

1 Introduction 260 2 Crew Legalities and Crew Pairing Repair 264 3 Model and Mathematical Formulation 266 4 Solution Methodology 271 5 Computational Experiences 277 6 Conclusion 285 REFERENCES 286

10 THE USE OF OPTIMIZATION TO PERFORM AIR TRAFFIC FLOW MANAGEMENT Kenneth Lindsay, E. Andrew Boyd, George Booth, and Charles

Harvey 287 1 Introduction 288 2 The Traffic Flow Management (TFM) Problem 289 3 Recent TFM Optimization Models 292 4 The Time Assignment Model (TAM) 302 5 Summary and Conclusions 307 REFERENCES 309

11 THE PROCESSES OF AIRLINE SYSTEM OPERATIONS CONTROL Seth C. Grandeau, Michael D. Clarke, and Dennis F.X. Mathaisel 312 1 Introduction 313 2 The Four Phases of Airline Schedule Development 315 3 The Airline Operations Control Center (OCC) 320 4 Analysis of Operational Problems 331 5 Areas For Improvement 352 6 Case Study: PT Garuda Indonesia Airlines 357 REFERENCES 368

12 THE COMPLEX CONFIGURATION MODEL Bruce W. Patty and Jim Diamond 370 1 Introduction 370 2 Problem Description 371 3 Problem Formulation 375 4 Model Implementation 379

Contents ix

5 Summary 383 REFERENCES 383

13 INTEGRATED AIRLINE SCHEDULE PLANNING Cynthia Barnhart, Fang Lu, and Rajesh Shenoi 384 1 Introduction 385 2 Fleet Assignment and Crew Pairing Problems: Existing Mod-

els and Algorithms 388 3 An Integrated Approximate Fleet Assignment and Crew Pair-

ing Model 393 4 An Advanced Integrated Solution Approach 395 5 Case Study 396 6 Conclusions and Future Research Directions 399 REFERENCES 401

14 AIRLINE SCHEDULE PERTURBATION PROBLEM: LANDING AND TAKEOFF WITH NONSPLITABLE RESOURCE FOR THE GROUND DELAY PROGRAM Songjun Luo and Gang Yu 404 1 Introduction 405 2 An Overview of the Airline Schedule Perturbation Problem 406 3 Literature Review 412 4 Preliminaries 414 5 Landing Assignment 418 6 Landing and Takeoff: Single Resource 425 7 Conclusions 429 REFERENCES 431

15 AIRLINE SCHEDULE PERTURBATION PROBLEM: GROUND DELAY PROGRAM WITH SPLITABLE RESOURCES Songjun Luo and Gang Yu 433 1 Introduction 434 2 Minimizing Maximum Delay Among Out-flights 435

x OPERATIONS RESEARCH IN THE AIRLINE INDUSTRY

3 4 5

Minimizing the Number of Delayed Flights Model Simplification, Valid Inequalities, and a Heuristic Computational Experiments

6 Conclusions REFERENCES

437 446 449 454 459

Erik Anderson Carmen Systems AB Gothenburg, Sweden

Michael F. Arguello Department of Mechanical Engineering The University of Texas at Austin

Jonathan F. Bard Department of Mechanical Engineering The University of Texas at Austin

Cynthia Barnhart Center for Transportation Studies Massachusetts Institute of Technology

George Booth Research and Development Service Federal Aviation Administration

E. Andrew Boyd Department of Industrial Engineering Texas A&M University

Erik Brauner BehavHeuristics, Inc.

Michael D. Clarke Department of Aeronau tics and Astronautics Massachusetts Institute of Technology

JIm Diamond Sabre Decision Technologies

CONTRIBUTORS

Michel Gamache Groupe d'Etude et de Recherche en Analyse des Decisions (GERAD) Ecole des HEC

Ira Gershkoft' Sabre Decision Technologies

Seth C. Grandeau Simat, Helliesen and Eichner Cambridge, MA

Charles Harvey Department of Decision and Information Sciences The University of Houston

Marcia P. Helme Resolved Communications

Sharon Hormby Behav Heuristics, Inc.

Efthymios Housos University of Patras Greece

Hasni Idris Department of Aeronautics and Astlonautics Massachusetts Institute of Technology

Niklas Kohl University of Copenhagen and COWlconsult Denmark

xii

Kenneth Lindsay The MITRE Corporation

Fang Lu Sabre Decision Technologies

Songjun Luo ZS Associates

Dennis F .X. Matbaisel Department of Aeronautics and Astronautics Massachusetts Institute of Technology and Babson College

Bruce W. Patty Mykytyn Consulting Group

Rajesh Shenoi McKinsey and Company

Mark Soog CALEB Technologies Corporation

Franc;ois Soumis Departement de Mathematiques Appliquees et de Genie Industriel Ecole Poly technique de Montreal

Xiaoyun (Sherry) Sun BehavHeuristics, Inc.

Lawrence R. Weatherford Department of Management and Marketing University of Wyoming

Dag Wedelin Chalmer University of Technology Sweden

Guo Wei Department of Management Science and Information Systems The University of Texas at Austin

CONTRIBUTORS

Gang Yu Department of Management Science and Information Systems Center for Management of Operations and Logistics The University of Texas at Austin

PREFACE

The field of operations research has had a tremendous impact on the manage­ment of today's air transportation. Driven by enormous demand from man­agement to gain a competitive advantage in the market, airlines are turning to advanced optimization techniques to develop mission-critical decision support systems for management and control of airline operations.

The airline business is very unique. The product offered by airlines is repre­sented by flights that carry passengers or cargo from various origins to various targeted destinations. In general, the marketability of the product is judged by the timeliness, accuracy, functionality, quality, and price of the service. As perceived by air transportation customers, these criteria translate into flexible schedules, on-time flights, safety, satisfactory in-flight services, proper baggage handling, and convenient ticket purchases. The direct resources that are needed to build the product include aircraft, crew, and airport facilities (runways and gates). Additional supporting resources such as maintenance bases, fuel ser­vices, food services, and crew training facilities are also required.

The air transportation market is very competitive. This competition comes not only from peer air carriers, but also from ground transportation companies such as buses, trains, ships, and rental cars. The increase of personally-owned vehicles also poses some threat. Competition in the United States airline in­dustry" is particularly fierce. This is largely due to the 1978 deregulation of the U.S. airline industry. Consequently, airlines have been allowed to choose their own market segments, decide their own routes, and set their own fares as long as they comply with the safety and security regulations enforced by the Federal Aviation Administration (FAA). The quality of airline services is monitored and reported to the public by the FAA. Deregulation has profoundly impacted the basic 'aspects of airline operations, including fares, services, quality, and safety. As B consequence, new markets and services have proliferated. In fact, eigh­teen months after deregulation, 106,000 city-pair authorizations were issued, in contrast to the 24,000 authorizations granted during the eighteen month pe­riod immediately preceding deregulation. The computer reservation system is another milestone in air transportation history that notably promoted compe-

XIV OPERATIONS RESEARCH IN THE AIRLINE INDUSTRY

tition. Customers, through travel agencies and/or airlines' reservation agents, can freely access all published routes and fares, and thereby make their selec­tion of services. Competition brings remarkable benefits to the customers; in turn, the informed customers issue more challenges to the airlines by demanding better services and lower fares.

To meet these challenges, and to provide a product with high quality and low cost, airlines spend a tremendous amount of resources and effort to generate profitable and cost-effective fare classes, flight schedules, fleet plans, aircraft routes, crew pairings, gate assignments, maintenance schedules, food service plans, training schedules, and baggage handling procedures. However, the air­lines' business is far more complex than it seems. The complexity comes from the very nature of the airlines' business and can be illustrated via the following facts:

• As a result of economic considerations in generating the plans and sched­ules, all the resources are tightly coupled and utilized to the extent of the airlines' knowledge. Not much slack is left to maneuver and to respond to changes.

• The environment airlines are working in is extremely dynamic and uncer­tain. Many causes may lead to disruption of an original plan, resulting in contingency handling or irregular operations. The major causes are inclement weather, aircraft mechanical problems, crew sickness, and fuel shortages. A small disruption in one place will snowball through the net­work. The absence of ample slack in the schedule makes the absorption of disruption a difficult task.

• Tight FAA restrictions on aircraft maintenance, crew legality, runway availability, security, etc. add many more constraints, as well as another level of complexity to planning and contro!'

• The large operational scale in terms of involved resources and components and operational scope in terms of geographic regions and market size and diversity make solving airlines' operational and planning problems a fear­some task.

How to gain and keep a leading edge in the competitive air transportation market and how to run the airlines effectively and efficiently to respond to cus­tomers' needs are the issues faced by top management of every airline. The key to attaining these objectives is the successful employment of advanced com­puter and optimization technology exemplified in the form of decision support

Preface xv

systems, and the time is ripe for applying decision support technologies in the airline industry. This point can be made from the following viewpoints.

From a technological perspective, computing speed is now fast enough to solve complex, large-scale, real-time problems; computer storage is now large enough to accommodate corporate data; the network is reliable enough to host mission­critical applications; graphical user interface (GUI) is friendly enough to facili­tate rapid application development and users' learning; and, furthermore, cost is low enough to justify return on investment.

From the methodological side, the last decade has enjoyed significant advance­ments in algorithms and solution techniques development. Optimization mod­els are more practi"al and yet more solvable. Much larger and more complex problems have been conquered with much shorter computation time. Research related to airline applications is much more active, with hundreds of publica­tions dedicated to air transportation each year.

After conducting interviews at major and regional airlines with hundreds of ex­ecutives, operations managers, crew coordinators, dispatchers, fleet planners, aircraft routers, crew planning managers, customer service managers, mete­orologists, pilots, and other operational personnel, the operations researchers and practitioners are more confident than ever in applying the right techniques, modeling the right problems, and acquiring the right answers. This has led to the successful development of decision support systems which save companies millions of dollars.

With the competition in the air transportation arena increasing, customers expect variety in terms of more flights and better routes, as well as better service quality in terms of on-time performance, food, and handling of baggage, and yet, a lower cost. The key is optimization.

What constitutes a desirable decision support system for the airlines industry? The basic criteria are:

• A good decision support system should produce effective decisions in a timely manner. The quality of generated decisions is important since it directly affects the company's revenue, cost, performance, and customers' satisfaction. However, if the decisions are not provided in time, the en­vironment changes, and the opportunities and resources may no longer be available. During the lengthy solution process, the occurrence of new problems complicates the situation.

XVI OPERATIONS RESEARCH IN THE AIRLINE INDUSTRY

• A good decision support system should not be a black box. It is not de­signed to replace decision makers, but to help them make better decisions. It should provide multiple alternatives so that the decision makers can make their choice based on their knowledge and many years of experience. It frees the decision makers from the tedious and time-consuming solution generation and validation process, and allows them to efficiently use their valuable time in evaluating the choices, framing problems, and anticipating and preventing future problems.

• A good decision support system should incorporate decision makers' knowl­edge into the system. It should make proper usage of decision makers' fine intuition and rules-of-thumb which have been proven effective. This will not only accelerate the generation of good solutions, but also enable ef­fective handling of many soft issues that are difficult to embody in the optimization model.

• A good decision support system should take into account all scenarios and all options. Very often in the real-world, a complete feasible solution may not exist. Under these situations, the system should provide a partial solution that resolves the problem as much as it can or solves the most urgent problem and leaves less important problems to a later time. This "buy-time" strategy has been widely practiced by airline managers.

• A good decision support system should facilitate interactions among deci­sion makers. This requires single instance of data, common graphical user interface, message-oriented, event-driven, and distributed components ar­chitecture. The interactions can be achieved through "what-if' function­ality. Thus, before committing to a final decision, it lets all involved deci­sion makers review the options, check their own resource availability and implementation feasibility, and pass their feedback through the common interface for collaborative and coordinated decision-making.

• A good decision support system should permit users to set parameters for limiting the scope of impact. It should also embed the functionality for fixing or protecting partial and local solutions. This is very important for protecting certain markets, specific aircraft routes for maintenance and other reasons, and particular crew pairings assigned to high priority flights.

• A good decision support system should have single instance of data, com­mon GUI, message oriented, event-driven, and distributed components ar­chitecture. This guarantees data integrity, real-time update of data, the same look and feel to all system users, and easy integration with existing systems.

Preface XVll

The purpose of this book is to show some recent advances in optimization techniques and decision support systems applications in air transportation. It covers a wide variety of operations research topics in the air transportation industry including:

• Demand forecasting

• Revenue management

• Airline schedule planning

• Integrated scheduling

• Crew paring optimization

• Crew rostering

• Network design

• Route planning

• Irregular operations aircraft routing

• Real-time crew management

• Air traffic flow management

• Airport traffic simulation and control

• Coping with the FAA's traffic control programs

Due to limited space and time, additional topics of interest to airlines as well as to air transportation researchers could not be covered. Hopefully, time will be available to edit succeeding volumes in this area in the future.

Gang Yu The University of Texas at Austin Austin, Texas, USA

ACKNOWLEDGMENTS

I would like to take this opportunity to thank all the contributors who made this book possible. Their dedication to making their papers of high quality is truly admired and recognized. The anonymous referees of the book deserve special acknowledgment. Their timely and careful reviews expedited the turn-around process and ensured that the high standard set for this work was met. My special appreciation goes to Jian Yang, Xiaowei Wang, and Caryn Cluiss who helped me tremendously in typing, editing, formatting, and proofreading the book. I was fortunate enough to work with a pleasant and supportive editor, Gary Folven, who patiently waited for the birth of this book. Finally, I am greatly indebted to my family, especially my wife, Xiaomei Song, and my boy, Ray. Preparing the book took an enormous amount of my time and kept me from being with them. I hope to make it up in the future.

Gang Yu Editor June 1997