Handbook of Operations Research Applications at Railroads (eBook)

Dr. Bruce Patty is the Vice President of Transportation Analytics at Veritec Solutions.  For over 25 years, Dr. Patty has provided guidance and insights for leading transportation companies such as American Airlines, CSX Transportation, and Pacer Stacktrain.  His academic background in Network Optimization and his experience in passenger and freight transportation has provided Dr. Patty with considerable expertise regarding a wide range of issues in these industries.While at American Airlines, Bruce developed their pioneering automated flight routing tool that scheduled aircraft so that they satisfied overnight maintenance requirements.  He participated on an Edelman Prize finalist team in the area of Crew Scheduling and then led rail consulting projects at Conrail.   He went on from AA to be a Director of Operations Research at CSX Transportation where he developed workload planning models of classification yards.    Dr. Patty left CSX to develop the transportation practice at MCGI, where he led or participated on consulting engagements with companies such as Norfolk Southern, CSX, BAX Global, and Delta Air Lines.  Prior to joining Veritec, Bruce was the AVP of Equipment Strategy and Planning at Pacer Stacktrain.  As part of his responsibilities, he managed an annual budget of $70 million, providing containers and chassis for use by Pacer’s customers.  He pioneered the use of analytics to manage the chassis fleet and keep the chassis-to-container ratio at an industry-wide low.Dr. Patty received his BS in Systems Engineering from the University of Arizona, and then a MS and Ph.D. in Operations Research from Southern Methodist University.   He served on the faculty at the University of Southern California, has served on the Board of Directors for INFORMS, and has published several papers and presented at a multitude of conferences.  He has lived in Marin County, California with his wife, Paula, for over 20 years and enjoys travel, golf, and relentlessly rooting on the University of Arizona Wildcats.

Dedication 6
Preface 8
Contents 12
Chapter 1: Train Scheduling 13
1.1 Introduction and Background 13
1.2 Role of Trains in the Railroad Operations Research Landscape 14
1.3 Types of Trains and Related Definitions 17
1.4 Specifying Road Trains 19
1.5 OR Challenges: Designing the Road Train Plan 22
1.5.1 Road Train Design Problem 23
1.5.2 Single Versus Multi-Block Trains 23
1.6 Train Routing/Block-to-Train Assignment Problems 25
1.6.1 Example Problem 29
1.6.2 Feasible Solution 31
1.7 Train Scheduling (Timing) Problem 34
1.7.1 Key Assumptions 35
1.7.2 Scheduling Variables 36
1.7.3 Scheduling Constraints 36
1.7.4 Cost Parameters 37
1.7.5 Observations on Solution Strategies 38
1.7.6 Special Cases 39
1.7.7 Problem Examples 41
1.8 Specifying Unit Trains 45
1.9 Local Service Specification Strategies 46
1.10 Train Plan Design Versus Real-Time Operations 49
1.11 Opportunities 52
References 53
Chapter 2: Locomotive Scheduling Problem 55
2.1 Introduction 55
2.2 Background on Locomotive Scheduling 56
2.2.1 Hard Constraints 57
2.2.2 Soft Constraints 58
2.2.3 Objective Function 58
2.3 Mathematical Models for Locomotive Scheduling 58
2.3.1 Space–Time Network Construction 58
2.3.2 Problem Size and Stage-Wise Solution Approach 59
2.3.3 Consist Flow Formulation for the LPP 61
2.3.3.1 Notation 62
2.3.3.2 Decision Variables 62
2.3.3.3 Objective Function 63
2.3.3.4 Constraints 63
2.4 Incorporating Practical Requirements 64
2.4.1 Cab-Signal Requirements 64
2.4.2 Foreign Power Requirements 65
2.5 Applications of the Model 66
2.5.1 Quantifying the Impact of Varying Minimum Connection Time 66
2.5.2 Quantifying the Effect of Changing Transport Volume on Key Performance Characteristics 67
References 68
Chapter 3: Simulation of Line of Road Operations 69
3.1 Introduction 69
3.2 Fundamental Elements for a Dispatching Algorithm 76
3.3 Developing a Dispatching Algorithm 78
3.3.1 Overview 78
3.3.2 Example 79
3.3.3 Simplified Assumptions 89
3.4 Future Directions 90
Chapter 4: Car Scheduling/Trip Planning 91
4.1 Introduction and Background 91
4.2 Car Scheduling/Trip Planning Systems in Context 93
4.3 Plan Compliance and the Value of Trip Plans 95
4.4 Current Industry Practices: Basic Car Scheduling/ Trip Planning Concepts 96
4.4.1 Current Industry Practices: Block Selection Logic 101
4.4.2 Current Industry Practices: Train Selection Logic 101
4.4.3 Current Industry Practices: Other Special Considerations 104
4.5 OR Challenge: Typical Reasons of Trip Plan Failures 106
4.6 Trip Plan Output Usages 107
4.7 OR Challenges: Alternate Approaches to Car Scheduling and Special Cases 108
4.8 Capacitation and Reservations 113
4.8.1 Specifying Capacities 114
4.8.2 Managing Reservations 118
4.9 Planning and Optimization 119
4.10 Time-Space Network Solutions 121
4.10.1 Dynamic Car Scheduling 123
4.11 Opportunities 127
References 129
Chapter 5: Railway Blocking Process 131
5.1 Introduction and Background 131
5.1.1 Impact of Blocking on System Efficiency and Service 132
5.1.2 Specifying the Blocking Plan 134
5.1.3 Plan Complexity 135
5.2 Current Industry Practices: The Blocking Rules Concept 136
5.2.1 Yard-Blocks, Train-Blocks, Class Codes, and Block Swaps 139
5.2.2 Local Service 141
5.3 The Table-Based Blocking Systems OR Challenge 142
5.4 Algorithmic Blocking 144
5.5 Examples of Areas Presenting OR Challenges 146
5.6 Semi-manual Blocking Plan Design Techniques 148
5.6.1 Incremental Blocking Plan Design Techniques 148
5.6.2 Tuning an Existing Plan 148
5.6.3 Checking Circuity and Excessive Handlings 150
5.6.4 Change Traffic Volume at a Yard 150
5.6.5 Designing Blocking Plans Using a Clean-Sheet Approach 151
5.6.6 Tuning Table-Based, Traffic Destination Attribute Rules Using Relaxation 152
5.6.7 Additional Methods for Testing Plans 155
5.6.8 Triplet Analysis for Blocking Plan Comparisons 155
5.6.9 Tree View Analysis 157
5.7 Specialized Blocking Situations 157
5.8 Blocking Plan Optimization 161
5.8.1 Considerations That Automated Blocking Optimization Techniques Should Consider 162
5.8.2 Mathematical Representation of the Block Design Optimization Problem 163
5.8.2.1 Data 164
5.8.2.2 Variables 164
5.8.2.3 Constraints 165
5.8.2.4 Objective 166
Optimization Techniques 166
Heuristic Approach 167
Initial Blocking Plan 167
Iteratively Improve the Plan 168
Resequencing Quickly 168
Finding Global Optimum 169
Changing Yard Penalties 169
Advanced Mathematical Programming 169
5.9 Additional Considerations 171
5.10 Opportunities 172
References 173
Chapter 6: Crew Scheduling Problem 175
6.1 Introduction 175
6.2 Background on Crew Scheduling 176
6.2.1 Terminology 176
6.2.2 Regulatory and Contractual Requirements 178
6.3 Mathematical Models for Crew Scheduling 179
6.3.1 Model Inputs 179
6.3.2 Space–Time Network Construction 179
6.3.3 Mathematical Formulation 181
6.3.4 Solution Methods 183
6.3.4.1 Successive Constraint Generation (SCG) 183
6.3.4.2 Quadratic Cost-Perturbation (QCP) Algorithm 183
6.4 Applications of the Model 185
6.4.1 Tactical Benefits 185
6.4.2 Planning Benefits 186
6.4.3 Strategic Benefits 186
References 187
Chapter 7: Empty Railcar Distribution 188
7.1 Introduction 188
7.2 Background on Empty Railcar Distribution 189
7.2.1 Local Distribution and Shipper Pools 189
7.2.2 Rules-Based Transaction Processing Systems 189
7.2.3 Nonintegrated Optimization Systems 190
7.3 Current Day Integrated Real-Time Optimization Systems 190
7.3.1 Model Inputs 190
7.3.1.1 Car Supply: Actual and Predicted 191
7.3.1.2 Car Orders: Actual and Predicted 191
7.3.1.3 Shipper Preferences 191
7.3.1.4 Cost Parameters 191
7.3.1.5 Operational Information 192
7.3.2 Model Framework 192
7.3.2.1 Model Preprocessing 192
7.3.2.2 Model Formulation 192
7.3.3 Model Output Post Processing 194
7.3.4 Systems Integration 194
7.3.4.1 Optimization Engine: Customer Car Order System 195
7.3.4.2 Optimization Engine-Transactional Equipment Distribution System 195
7.3.4.3 Transactional Equipment Distribution System: Car Movement Management and Tracking System 195
7.3.4.4 Optimization Model: Operational Systems: Decision Making Process Integration 196
7.3.5 Reported Benefits 197
7.3.6 Other Implementation Considerations 197
7.3.6.1 User Acceptance 197
7.3.6.2 Model Thrashing 197
7.3.7 Other Modeling Considerations 198
7.3.7.1 Endogenizing Stochasticity 198
7.3.7.2 Including Blocking Costs in Empty Car Assignment 198
7.3.8 Other Areas of Application in Rail 198
References 199
Chapter 8: Network Analysis and Simulation 201
8.1 Introduction and Background 201
8.1.1 Planning and Simulation 202
8.1.2 Other Types of Simulations 203
8.2 Types of Network Level Simulations 203
8.2.1 Uncapacitated Deterministic Simulations with Fixed Plans 204
8.2.2 Uncapacitated Deterministic Simulations with Probabilistic Connections 205
8.2.3 Capacitated Simulations with Fixed Plans 205
8.2.4 Capacitated Simulations with Dynamic Plan Elements 207
8.2.5 Full Monte-Carlo Capacitated Simulations 208
8.3 Resource Estimation 208
8.3.1 Estimation of Crews 209
8.3.2 Estimation of Locomotives 210
8.3.3 Estimation of Railcar Requirements 211
8.3.4 Estimation of Yard Workloads 213
8.4 Roles of Network Simulation 214
8.4.1 Mergers 214
8.4.2 Network Modifications 215
8.4.3 Emergency Situations or Special Circumstances 216
8.5 Average Day Analysis 216
8.5.1 Uncapacitated Average Day Analysis 217
8.5.2 Capacitated Average Day Analysis 218
8.5.2.1 Achieving a Robust Train Volume Formulation 222
8.5.2.2 Train-Block Prioritization 223
8.5.2.3 Fill Blocks 224
8.5.2.4 Capacitation by Length and Gross Weight 224
8.6 Future Directions and Opportunities 225
References 226
Chapter 9: Simulation of Yard and Terminal Operations 228
9.1 Introduction 228
9.2 Reasons to Simulate 229
9.3 The Problem 231
9.3.1 Train Arrival 231
9.3.2 Handling the Inbound Crew and Power 232
9.3.3 Inbound Car Inspection 232
9.3.4 Switch (Classify) Cars 232
9.3.5 Train Assembly 234
9.3.6 Final Train Assembly 234
9.3.7 Train Departure 235
9.4 Matching the Analytic Approach with Study Requirements 235
9.5 Building a Yard Simulation 237
9.5.1 Conceptual Design 237
9.5.1.1 Simulation Engine 238
9.5.1.2 Decision Engine 239
9.5.1.3 Inbound Process 240
9.5.1.4 Switching Process 240
9.5.1.5 Train Assembly Process 241
9.5.1.6 Departure Process 241
9.5.2 Data for Simulation 245
9.5.3 Other Issues to Be Resolved 246
9.6 Recent Past to Current State of the Art 247
9.7 Future Directions 249
Chapter 10: Operations Research in Rail Pricing and Revenue Management 252
10.1 Introduction 252
10.1.1 U.S. Freight Rail Pricing History 252
10.1.2 Revenue Management for Rail: Importance 253
10.1.3 Revenue Management for Rail: Challenges 253
10.1.4 Revenue Management for Rail: Recent Opportunities 254
10.2 Analytical Techniques in Freight Revenue Management 255
10.3 Characterizing Customer Behavior: Estimating Product Demand 255
10.3.1 Forecasting Demand Levels 256
10.3.2 Predicting Customer Price Sensitivity 257
10.4 Research in Revenue Management Models 258
10.4.1 Train and Block-Based Capacity Approaches 258
10.4.2 Service-Based Pricing Strategies 260
10.4.3 Container-Centric Yield Management 261
10.5 Future Directions and Opportunities for Revenue Management and Freight Rail 262
References 262
Chapter 11: Intermodal Rail 264
11.1 Introduction and Background Information 264
11.1.1 Definition of Intermodal 264
11.1.2 Brief History of Intermodal 265
11.1.3 Equipment Variations 267
11.1.4 Role of Railroads and IMCs 268
11.1.5 Chassis Pools, Both Domestic and International 268
11.2 Examples of Decisions to Be Made Where OR Models Can Be Used 269
11.2.1 Pricing 269
11.2.2 Container Fleet Sizing 270
11.2.3 Demand Forecasting 270
11.2.4 Assignment of Equipment to Customers 271
11.2.5 Chassis Fleet Sizing and Positioning 271
11.3 Detailed Examples of Actual Model Implementations 272
11.3.1 Empty Container Repositioning 272
11.3.1.1 Background on Problem 272
11.3.1.2 Typical Decision Making Approach 272
11.3.1.3 Optimization Approach 273
11.3.1.4 Network Construction 274
11.3.1.5 Objective Function Components 275
11.3.1.6 Constraints 275
11.3.1.7 Solution Approach 276
11.3.1.8 Results 276
11.3.2 Chassis Pool Sizing 277
11.3.2.1 Overview 277
11.3.2.2 Approach 277
11.3.3 Container Selection Process 278
11.3.3.1 Background 278
11.3.3.2 Approach 279
11.3.3.3 Supply and Demand Forecasting 279
11.3.3.4 Capacity Valuation 280
11.3.3.5 Fleet Inventory Targeting 280
11.3.3.6 Load Accept Optimization (LAO) 281
11.3.3.7 Load Routing Optimization 281
11.3.3.8 Results 282
11.4 Opportunities 282
11.4.1 Forecasting 282
11.4.2 Tactical Equipment Matching 283
Index 284