National Academies Press: OpenBook
« Previous: Chapter 4 - Schedule Blocking
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 206
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 207
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 208
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 209
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 210
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 211
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 212
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 213
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 214
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 215
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 216
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 217
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 218
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 219
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 220
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 221
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 222
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 223
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 224
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 225
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 226
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 227
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 228
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 229
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 230
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 231
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 232
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 233
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 234
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 235
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 236
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 237
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 238
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 239
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 240
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 241
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 242
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 243
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 244
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 245
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 246
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 247
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 248
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 249
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 250
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 251
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 252
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 253
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 254
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 255
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 256
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 257
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 258
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 259
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 260
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 261
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 262
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 263
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 264
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 265
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 266
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 267
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 268
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 269
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 270
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 271
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 272
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 273
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 274
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 275
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 276
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 277
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 278
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 279
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 280
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 281
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 282
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 283
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 284
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 285
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 286
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 287
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 288
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 289
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 290
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 291
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 292
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 293
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 294
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 295
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 296
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 297
Suggested Citation:"Chapter 5 - Runcutting." National Academies of Sciences, Engineering, and Medicine. 2009. Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling. Washington, DC: The National Academies Press. doi: 10.17226/14257.
×
Page 298

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Chapter 5. Runcutting ͱ.ͭ Basic Runcutting (Level ͭ) ͱ.ͮ Intermediate Runcutting (Level ͮ) ͱ.ͯ Advanced Runcutting (Level ͯ)

Chapter 5. Runcutting 5-3 5.1 Basic Runcutting Runcutting is the process of turning blocks into work pieces or “runs” for drivers. The word cutting is used to describe the process of “cutting” blocks into pieces that drivers can actually work. A “run” may consist of all or part of a vehicle block, and may have single or multiple pieces. During the runcutting process blocks may remain intact, be cut, or in some cases be completely rehooked. In practice runcutting is an extremely complex process that takes into account a range of fac- tors, both qualitative and quantitative, to produce a set of outcomes that signifi cantly aff ect transit operations and an agency’s budget. The scheduler is actually creating the work days for the bulk of the agency employees, signifi cantly aff ecting the cost of operation. The sched- ules department is in many ways a transit agency’s budget implementation and enforcement department. We should note up front that runcutting can range from simple solutions to infi nitely complex problems. In fact, most computerized runcut “optimizers” are not true optimizers at all—they cut down the problem into something manageable and then optimize that solution. This shows just how complex the mathematics of runcutting can be—that even the most powerful com- puters are not able to mathematically solve complex runcutting problems. Runcutting Objectives What are we trying to achieve when creating runcut solutions? There can be many objectives, and defi ning the objectives can be an extremely diffi cult task, raising more questions than it answers. But knowing the priorities at your agency is critical because without objectives one cannot measure the success of a runcut. Below we note some of the typical objectives a sched- uler will be trying to achieve when undertaking runcutting. “Legal” Solutions. First and foremost the runcut must be “legal.” This is an expression used at many systems and because labor contracts are often involved, indeed there are times when it carries the force of law. “Legal” means that it must meet all the writ- ten rules of the labor agreement. It must also meet any safety, driving time, or break requirements, some or all of which may be legislated. Effi cient Solutions. In most cases the scheduler will be trying to achieve an effi cient runcutting solution. In bus transit agencies the operator cost typically accounts for more than ͱͬ% of total operating cost. Therefore the impact of runcutting effi ciency is signifi cant in overall operating effi ciencies. Exactly what defi nes “effi ciency” can at 1. 2. LEVEL 1 runcutting The process of converting (or cutting) vehicle blocks into work assignments for operators. run A work assignment for an operator. Most often, run refers to a whole day’s work assignment. block A vehicle (or train) assignment that includes the series of trips operated by each vehicle from the time it pulls out to the time it pulls in. A complete block includes a pull-out trip from the garage followed by one or (usually) more revenue trips and concluding with a pull-in trip back to the garage. rehooking The process of changing how trips are linked into a block. This is done when evaluating blocks and during the runcutting process. The word “run” is often misused. Individual trips, nonstop portions of routes are often called “runs” by laymen. Do not stray from the appropriate defi nition and use. Tip Level ͭ. Basic Runcutting

Level ͭ. Basic RuncuttingChapter 5. Runcutting 5-4 times be somewhat diffi cult to determine and will be discussed in some detail in the fol- lowing sections. “Streetable” Solutions. The scheduler should create runcuts that maximize the op- erational potential of the solution. For the same cost and legality, comparative runcuts could diminish (or alternately enhance) the ease of operation. For example, in systems which have relatively high “leave” benefi ts, i.e., vacation days, personal days, sick days, birthdays off , etc., a streetable solution might mean more runs markedly longer than ʹ hours, often in the ͵- to ͭͬ-hour range (ͱ-day work week). “Friendly” Solutions. In addition to the above objectives the scheduler should be trying to achieve results that allow for a “reasonable” workplace for the operators. Schedulers need to be aware that the runs and rosters developed defi ne an operator’s entire work life, and should look to achieve a balance of operator requests and preferences where possible. Of course some of the above objectives are at times in confl ict or at the very least diffi cult to achieve simultaneously. That is where the skill of the scheduler, able to balance the varying objectives, comes into play. Types of Runs Runcutting, as described above, is creation of operator assignments or runs. But what are these runs? The types of runs operated by public transit agencies generally fall into three broad categories. These are described below. Straight Runs. This type of run implies work paid as “straight through” or continuous time on duty. Traditionally a straight run has consisted of a single piece of work, where the operator stays on the same vehicle for the whole day. A second type of straight run involves a break (often required by labor agreement or legislation) between two pieces of work. This break may be paid or unpaid, and be taken on the vehicle or at the depot, according to site-specifi c labor agreement provisions.ͭ The break in a straight run is often between ͯͬ and Ͳͬ minutes in duration. Below are graphical examples of straight runs. 1 In some systems with both types of these straight runs, only the former is called a “straight,” the latter, even with a break under 60 minutes, is considered at that system to be a split (or swing). For simplicity, this document distinguishes them as one-piece straights and multipiece straights. 3. 4. 1. straight run A run in which trips are consecutive without interruption. Straight runs do not contain any breaks (except for meal breaks at some systems) for the operator. Any break in a straight run is usually less than one hour in length. A straight run with a break is referred to as a multipiece straight. split run A run containing two or more pieces of work separated by a break over one hour in length. Also known as a swing run. At some systems, three-piece split runs are allowed, but one of the breaks (or “swings”) is usually paid whereas in two-piece split runs the break is generally not paid. Split runs tend to be used to allow both peaks to be covered by one operator since the work day would otherwise be too long for a straight run. swing time The elapsed time (usually unpaid) between the pieces of a split run. Also known as “intervening time.” if swing time is paid, it is sometimes called “inside time.” spread time Total time between the start of the fi rst piece and the end of the last piece of a split run with two or more pieces. Also known as “outside time.”

Chapter 5. Runcutting 5-5 Level ͭ. Basic Runcutting tripper A short piece of work whose total time is less than that specifi ed as constituting a full-time run. A tripper is often a piece of work in the AM or PM peak period that cannot be combined with another piece of work to form a split run because of insuffi cient hours, excessive swing time, or excessive spread time. Trippers are often operated by extraboard or part-time operators. Tripper can also refer to a vehicle that pulls out, makes no more than one round-trip, and pulls in. Split Runs. Split (or Swing) runs refer to runs that have two pieces, with a (usually) longer unpaid break between those pieces. The operator is not on duty between the pieces of work, and typically all pieces start and fi nish at the home garage. The break in a split run is characteristically longer than for straight runs (both kinds), greater than ͵ͬ minutes, and may be as long as three or four hours. Split runs tend to be used to allow both peaks to be covered by one operator since the work day would otherwise be too long for a straight run. Below is a graphical example of a typical split run. Trippers. Trippers are almost always short one-piece straight runs, and are often used in peak periods. Trippers are sometimes known as “part-time runs.” In some cases the diff erence between types of runs can become fuzzy—particularly between a split run and a straight run with an unpaid break, or between a tripper with multiple pieces and a split run. Invariably, though, the unique conditions identifying each run type can be found in your labor agreement (and it is likely that your agreement clarifi es the diff erences between very similar runs). These examples are basic in nature and do not cover all types of runs or all types of applications of those runs. For example, a split run does not necessarily cover both peaks, and in fact may be used on weekends or where there is little peak eff ect. Components of Runs Runs consist of a number of distinct components—the actual in-vehicle time is only one of them. Importantly, the actual driving time is only one of several components and may account for as little as Ͳͬ% or ͳͬ% of the total paid time, depending upon the type of run and labor agreement provisions. 2. 3. There are many different issues to juggle when developing split runs. Before you begin, know whether your property has rules regarding the total length of the driving day including the unpaid break times, the length of time a driver can drive, whether runs with more than two work pieces are allowed, i.e., three-piece runs, and the trade-offs between split runs and trippers. Tip

Level ͭ. Basic RuncuttingChapter 5. Runcutting 5-6 The diagram below provides a graphic visualization of the possible components of a run. It is interesting to note than in the split run shown the actual driving time is responsible for only two of eleven unique run components. Total work hours and spread hours are indicated as these are numbers that are needed to calculate penalties such as overtime, spread, and shift premiums. These penalties or premi- ums—which in some systems are collectively known as “collaterals”—are discussed in more detail in later sections. At this stage it is suffi cient to note that understanding the components of runs is required to develop an understanding of the runcutting process. Runcutting Outputs The outputs of any scheduling process can be broken down into two categories—the actual operational outcomes, and the reports required describing those outcomes. In the case of runcutting the outcomes are simply a set of operator assignments that are valid, cover all of the blocks, and meet the objectives of the agency. However there are work units, i.e., departments, divisions, offi ces, etc., of an organization that require a description or sum- mary of those runs—in eff ect, the tools with which certain people will be able to undertake their roles. Some of these are listed in the table below. work hours Total hours worked by an operator, not including fringe benefi t hours such as sick leave, holiday, etc. Work hours include only labor hours associated with the requirements of putting the runs in service and operating the service. collaterals All of the various types of penalties and premiums that might be required to make legal runs. premium pay Pay to an operator that is over and above the straight time pay rate; includes overtime premium, spread premium, shift premium, and any other operating premiums as defi ned by the contract.

Chapter 5. Runcutting 5-7 Level ͭ. Basic Runcutting Output Description Used By Run Guide A summary of runs that describes start/fi nish locations, work hours, and cost element breakdowns Schedulers, garage staff Runs Summary List of runs showing start/ fi nish times, hours worked, and paid hours Payroll systems Dispatch Sheet A list of all runs, sorted by start time Garage staff , to track staff and vehicle movements throughout the day Runcut Statistics Summary of costs, totals, penalties, etc. Schedulers for summaries/ comparisons, Senior Management for costing analysis The Role of Computers You will recall the comment that many runcutting problems are qualitative in nature and are, therefore, too complex for any computer, alone, to optimize completely. This does not, how- ever, mean that computers do not have a signifi cant role to play as tools to signifi cantly assist in developing runcuts. Computerized Scheduling Packages The advent of computerized scheduling packages containing complex algorithms has enabled signifi cant improvements in the ability to generate effi cient runcut solutions. However, since the computer is trying to solve what it sees as a mathematical or quantitative problem, the inputs need to be numeric. So as a scheduler you will need to resolve issues of preference in terms of weights or penalties. And this requires skill and practice. For example, do you think that longer spreads are less preferred than having additional trippers? If so, by how much— ͱͬ%, ͭͬͬ%? The automated algorithms require the user to be able to defi ne such preferences concisely and numerically. The modern ability to run numerous runcut solutions and compare results is a signifi cant ben- efi t generated by automated runcutting programs over previous manual-only runcuts. A key point to remember is that the computer is only a tool to be used by the skilled scheduler, to produce high-quality solutions. Keeping this in mind, here are a few considerations when using computerized systems to generate runcuts: Never accept the automated solution as given. Always consider the need to fi ne tune—for operational preference, runcut desirability, or whatever reason. Very rarely • overtime premium Pay at the rate of 1.5 times (or higher) the normal rate for work performed in excess of daily or weekly thresholds, usually eight or ten hours per day or 40 hours per week. shift premium A premium paid to operators for working during times of the day that are subject to special pay differen- tials, e.g., an owl (late night/early morning) run. spread premium Pay equal to one-half or more of all minutes in excess of a specifi ed maximum spread time, in addition to regular straight pay. The spread premium may be multilayered, e.g., half of all time up to 60 minutes over the specifi ed maximum spread time and three-quarters or all time more than 60 minutes over the specifi ed maximum spread time. Spread premium is separate and distinct from overtime premium.

Level ͭ. Basic RuncuttingChapter 5. Runcutting 5-8 is an automated solution ready for implementation without the need for some, even if minor, manual manipulation. Run multiple solutions, each time changing (preferably) one constraint at a time. The direct impact of this particular constraint can then be measured. Then do the same process with constraints that impact on each other. This way, the best set of inputs can be developed. Experiment, review, change, and adjust. Use the interactive tools to review and ad- just solutions. This may lead you to adjusting parameters, weights, costs or constraints, and resubmitting for a complete new (and improved) solution. Use the tools provided by the system to work interactively. Use the tools provided by the system to double check accuracy of a runcut solution. Familiarize yourself with all of the ways the system can summarize, present, and al- low you to review information. Our experience is that users tend to focus on a subset of the available functionality of comput- erized systems and as a result do not exploit the full capabilities of the software. This in turn aff ects either the effi ciency or quality of runcuts created. Basic Spreadsheet Tools The spreadsheet is not forgotten as a means of developing runcuts either. Spreadsheets pro- vide a range of capabilities. Some typical examples include: Calculation. Use formulas as much as possible to calculate totals, penalties, costs and to generate summaries. For example, if the report allowance is ͭͬ minutes, and you want to show both garage depart and sign-on times, the sign-on time fi eld should equal the garage depart time less ͭͬ minutes (expressed in a formula as ͬ.ͬͬͬͲ͵ͰͰͰ x ͭͬ or as time(:ͬ,:ͭͬ,:ͬ)). If the clear allowance is ͭͬ minutes, then the sign-off time would equal the garage arrival time plus ͭͬ minutes. There are endless examples, at varying levels of sophistication, where calculations can be automatically generated. In addition, times can be converted from time to decimal format if so desired. If the entry in cell Aͭͬ is ʹ:ͮͰ, the formula is: =hour(aͭͬ)+minute(aͭͬ)/Ͳͬ, which yields ʹ.Ͱ. Visualization. Use formatting to show a variety of information. For example, use conditional formatting to show if something is not legal (e.g., if the total spread fi eld is greater than ͭͮ hours and the maximum allowed is ͭͮ hours, highlight the cell with a red background). Conditional formatting in spreadsheets is a powerful tool that can greatly assist the scheduler. Validation. Use formulas to check whether rules and preferences are being broken. These can be applied as checked fi elds, or as formatting changes, highlighting issues. • • • • • • • • report allowance The amount of time paid an opera- tor from sign-in time to pull-out time. During this time, the operator may obtain instructions and supplies pertinent to his/her run, locate the assigned vehicle, and perform a pre-trip inspection. sign-on time (sometimes called sign-in time) The time an operator is assigned to report for duty at the start of each piece of a run. The operator may be required to sign in or may be acknowledged by the dispatcher as having reported. clear allowance The amount of time paid to an operator at the conclusion of the run to turn in transfers, fare media, or other supplies and reports. Learn the things your runcut program does best and use it to your full advantage. Not all programs are equally good at all aspects of runcutting. Tip

Chapter 5. Runcutting 5-9 Level ͭ. Basic Runcutting Finally, and this cannot be stressed highly enough, do not type values into cells unless ab- solutely necessary! The more numbers that are typed, the greater the chance for error. Use formulas, calculations, and formatting as much as possible. Learn to use lookup tables and other functions that can assist to streamline the runcutting process, and can reduce errors. As with computerized scheduling packages there is a tendency for schedulers to use only a small portion of the capacity and power of spreadsheet systems. How to Measure the Success of a Runcut The success of a runcut is always measured against the standards and objectives of the agency. Usually based on agency historic data as a foundation, pay-to-platform ratios are the pri- mary guide (more about pay-to-platform later in this section). Some typical objectives were discussed above. As with many aspects of scheduling, be thorough and methodical. Develop a template for comparing one solution to the next (spreadsheet is preferred), each with a score or summary for the key criteria. Only when the relevant information is laid out can an objective assessment then be made. Remember, as a scheduler you will be asked to provide advice regarding the diff erences or im- pacts from one runcut to the next. And even before you reach this point you are likely to com- pare your initial or preliminary solutions as you consider alternative solutions. The example template below compares some key cost measures between two runcuts. A de- bate about the merits of one solution against another will be put off until later in this chapter. Suffi ce to say, the tables allow us to see quickly that Solution ͮ uses more part-time operators but achieves a lower cost. The agency objectives and experience can then be used to determine whether that is a desirable outcome or not. pay-to-platform ratio The ratio of pay hours to platform time. For example, if an operator receives 9:00 in pay for 8:00 of platform time, the pay-to-platform ratio is 1.125 (9:00/8:00). The pay- to-platform ratio is one of the most widely used methods of measuring runcut effi ciency and is often used to measure the impacts of non- platform items (such as report or travel times) on operator pay hours. Some systems use the inverse, the ratio of platform to pay hours. pay hours The number of hours for which an operator is paid at his/her rate. Pay hours include work hours, make-up time, overtime premium, spread premium, and any other adjust- ments called for in the contract. Linking spreadsheets with “look-up” formulas minimizes direct entry and allows you to keep all of your input variables on a single master sheet. Tip

Level ͭ. Basic RuncuttingChapter 5. Runcutting 5-10 Before beginning the runcut, take a broad view at your blocks. Understand whether the same level of service operates all day, which lends itself to straight runs, or whether service is “peaked” with more vehicles (and drivers) required during limited times of day. This lends itself to a combination of part-time and split runs in addition to a full-time and straight runs. The more variation in block sizes, the more likely you will have a complex runcut. Tip Runcut Comparison Straight Runs 40 73% 38 64% Split Runs 12 22% 9 15% Total FT Runs 52 95% 47 80% Part Time Runs 3 5% 12 20% Total Operators Required 55 59 Hours Breakdown Total Avg % Total Avg % Revenue 420.0 7.6 88% 420.0 7.1 88% Report 22.3 0.4 5% 22.7 0.4 5% Travel 33.5 0.6 7% 34.0 0.6 7% Total Work Hours 475.8 8.7 476.7 8.1 Penalties Spread 14.4 10.8 Overtime 17.9 2.3 Guarantee 6.1 2.3 Total Paid Hours 514.3 492.1 Pay/Plat Ratio 1.2244 1.1717 Solution 1 Solution 2 Note that total operators could potentially be shown in terms of Full-Time Equivalent (FTE) operators. In such cases part-time operators may be substituted on a ͮ for ͭ basis. For Solution ͭ the FTE would be ͱͯ.ͱ, and for Solution ͮ the FTE would be ͱͯ, using this approach. Be care- ful, though, since that approach works well where part time operators are covering the peaks and have shorter average hours, but may be misleading if the part-time runs are longer or not only covering the peaks. The approach to defi nition of FTE, and the applicability of FTE as a measure to compare runcuts, will vary by agency. Some agencies just view each employee as part of a headcount, and some use an FTE approach. Runcutting Inputs The success of any runcut depends on the quality of the information available as inputs to the runcutting process. This applies as much to manual runcutting as it does to computerized run- cutting. If the basic elements of a runcut are not well defi ned, the result will not be eff ective.

Chapter 5. Runcutting 5-11 Level ͭ. Basic Runcutting Scheduling requires absolute accuracy of details. To this end, make sure you have a good meth- odology for ensuring that when you start to runcut, whatever the size of the problem, you have the following pieces of information, in their entirety: A complete set of trips and vehicle blocks; All relevant defi ned rules (usually the labor agreement summarized); Defi ned relief types, relief locations, and travel times; and Known limitations—cost limits, work rule preferences, etc. Most automated scheduling packages have a means for checking that all the required informa- tion is available to start runcutting. Always double check, though. You can simply go back and check your vehicle stats against what they should be. For spreadsheet-based methods, make sure you have a checklist, and a method for viewing both the runs and blocks details. Checklists are important throughout the scheduling process, to ensure work is undertaken thoroughly, in a methodical manner. Trips/Blocks First and foremost we need something to runcut. The runcut process starts with a complete and correct set of vehicle blocks and trips. Where blocks defi ne the assignment for a given vehicle, the run will defi ne the assignment of a given driver. We can, at this stage, consider the nature of the blocks and runs we are working with. In par- ticular the nature of the “service curve” can aff ect how we will approach the runcutting pro- cess. For example, is the service highly “peaked” (i.e., has a lot of blocks with peak-only trips), or do the same frequencies operate pretty much all day? The answer to this question will have a signifi cant bearing on how the runcut results will look and on how we approach the runcut. The Broad Street example will be used to show how to lay this out visually. Remember that while scheduling is a very detailed, numbers-based process, the best way to understand the data is often by looking at things visually, usually with graphs. This allows a simple representa- tion of a great deal of detail that would otherwise be diffi cult to understand. Below is a simple “Vehicles in Operation” graph. This graph simply shows how many buses are in operation anywhere at any given time during the day on Route ͵ͳ Broad Street. The graph can be generated quite simply in a spreadsheet. For computerized scheduling packages the graph can be generated automatically—so check out how to do so as this is an invaluable tool. • • • • travel time Paid time allowed for an operator to travel between the garage and a relief location. If the travel is for relief purposes only and is not part of a pull-in or pull-out, then travel time is not included in platform time. service curve A plot of the number of buses in service by hour. DO NOT attempt a runcut until you know your blocks are complete and correct and all work rule preferences are defi ned. Tip

Level ͭ. Basic RuncuttingChapter 5. Runcutting 5-12 Looking at this graph, we can immediately see that there will be some complexity in cutting the runs during peak times. We can also assume that we may need to use diff erent run types to cover the peaks, including (potentially) split runs or trippers. Some other things we can immediately note in the graph above include: Many of the blocks are around ͭͰ hours which will probably allow us to cut them into two pieces, each as a single run. The peak blocks are shorter in the AM Peak than the PM Peak (this is not unusual). The AM blocks are around ͯ:ͯͬ and the PM blocks around Ͱ:ͯͬ. This may allow us to create ʹ:ͬͬ split runs without too much diffi culty (subject to spread limitations). We will need two additional runs during the peaks to cover the two additional peak vehicles. We will have three runs in operation during the off peak. • • • • vehicles in operation graph A graphical representation of the number of vehicles in operation by time of day, typically by route but also by garage or system.

Chapter 5. Runcutting 5-13 Level ͭ. Basic Runcutting So by simple visual representation, without looking at any tables of numbers, we can discern a great deal of information about the vehicle blocks, their impacts on the runcutting to follow, and some potential outcomes. Of course not all graphs will have this form—some have unbal- anced peaks, some are fl at, or have higher peaks, or wider/shorter spans. But in each case one can immediately consider implications for the runcutting process. Naturally, as we get into the actual cutting of blocks we will need to work at a more detailed level. Rules, Constraints, and Practices We now have a set of blocks to generate runs from. What else do we need? One thing you will hear about often when discussing runcutting is the word “constraints.” In the context of creat- ing a runcut there are numerous constraints, each limiting how we can cut the blocks into runs. Constraints include the following types of categories: Hard Rules. “Hard” rules are typically defi ned in the labor contract. They include spe- cifi c fi xed limits such as how many hours can be worked in a day, or how many of a type of run is allowed. Sometimes hard rules are expressed as a fi xed number (e.g., a ͭͬ-hour maximum work time in a day). In some cases they are expressed as percentages (e.g., no less than ͮͬ% of full-time runs may be straight runs). At times the language and defi nition of hard rules can be extremely complex. In this context hard rules cannot be broken and are nonnegotiable. Soft Rules. “Soft” rules may be in the labor contract or may fall into the category of “operating practices” (see below). Soft rules tend to be preferences or “wants,” rather than nonnegotiable requirements. Typical soft rules tend to relate to not scheduling excessively to hard rule requirements. For example, while you may be allowed to sched- ule up to a ͭͯ-hour spread, there is a preference to keep as many as possible within ͭͮ hours. This may be written into the labor agreement or may just be an understanding with operators. Operating Practices. Some constraints, both hard and soft, are not written into the labor agreement, or even necessarily required by operators, but are applied based on either historical precedents or by preference. For example, it may be an agency pref- erence that any operator who works on a certain route always gets a break every two hours (if that is a busy route). Or perhaps, going back to the spread example, the labor agreement allows ͭͰ-hour spreads but Management sees that as excessive and prefers to schedule to a ͭͯ-hour maximum. • • •

Level ͭ. Basic RuncuttingChapter 5. Runcutting 5-14 The labor agreement will be unique to each agency and therefore we do not propose here to prescribe what it will or will not include. We would, however, urge every scheduler to consider each constraint they are given and consider if it is: required by the labor agreement; in line with agency objectives; in keeping with good scheduling practices; and properly defi ned or implemented. The table below lists some typical types of constraints or rules found in labor agreements, or in operating practices, of transit systems across the United States. Work Rule/ Constraint Typical Source Notes Minimum/Maximum Work Times Labor contract Can either be expressed in terms of platform time or work time. Run Types Allowed Labor contract Typically will discuss straight runs, split runs, and trippers. Nearly all run types are some derivative of these. Run Type Requirements Labor contract Relates to minimum or maximum allowed of specifi c run types. Spread Time Labor contract, operating practice Maximum spread time is usually specifi ed in the labor contract. Meal and Break Requirements Labor contract, operating practice, state regulations Will dictate whether straights are one-piece or multipiece and if so, are breaks paid/unpaid, and how long/short must they be. In some states break requirements are legislated and may override the agency-specifi c labor agreement. Breaks may also be driven by agency preferences. Report/Clear Allowances Labor contract, operating practice Typically mandate allowances at start and end of runs, and possibly around breaks. Usually expressed as time (e.g., ͭͱ minutes report allowance at the start of a run). Relief Requirements Labor contract, operating practice Labor contract may describe where and when reliefs can occur, how they must be undertaken, time allowed for travel, etc. Alternatively, may be totally based on agency operating practice, often by verbal agreement with operators. See following section for discussion of reliefs. a. b. c. d.

Chapter 5. Runcutting 5-15 Level ͭ. Basic Runcutting Computers and Constraints It is one thing to understand a labor rule or constraint. It can be an entirely diff erent thing to be able to succinctly defi ne that constraint in a way that a computer can meaningfully process. In this context we are considering two types of computers—the spreadsheet and sophisticated proprietary computerized scheduling package. For a spreadsheet we may be considering simple things such as conditional formatting to highlight where runs are longer than allowed. For a computerized system the range of input approaches varies considerably, from screens that specifi cally ask for minimums/maximums, to the requirement to write a line of “almost computer code” to describe a complex preference. Consider the following rule in the labor agreement at an agency: At least one of the following constraints must be met in straight runs over ʹ hours: ͭ x ͮͬ minute break and ͭ x ͭͬ minute break; or ͯ x ͭͱ minute breaks; or Ͳͬ minutes total layover (ͭͱ% of runs minimum) The agreement may further state that not only must one of these constraints be applied in each run, but also that ʹͬ% must be achieved in actual operation. In fact, though, all sched- uled break requirements are expected to be achieved in operations. Schedules should never be knowingly written that are inoperable. Runs are further limited by the requirement that no more than ͮͯ% of runs can involve a vehicle change. This limits options to create multipiece runs to deal with the breaks. To model this as a spreadsheet is virtually impossible, and so when the runs are created it will require a great deal of management by the scheduler. Even in an automated system there will be diff erent ways to approach and model the rule. In this case there are a mix of hard rules, per- centage limits, and preferences. The important thing to understand is that when providing input to a computer (be it spread- sheet or automated system) the user must be able to clearly understand and describe the rule/constraint and, just as importantly, must understand the intent of the rule. Only by under- standing the intent of the constraint will the user be able to control the outcome to achieve that desired result. In most computerized systems users will be required to weight rules as the system strives for an optimal mathematical result. Understanding this is very important—the computer will be considering every aspect of the runcut as a mathematical problem, to which it is trying to • • •

Level ͭ. Basic RuncuttingChapter 5. Runcutting 5-16 provide an optimal mathematical solution. Therefore in most instances the scheduler will be required to express the rules, constraints, and preferences in terms of numbers and weights. Penalties and Costs At this stage we have understood that blocks and rules are key inputs to the runcutting pro- cess. But in the end, our runcuts will be measured by their cost. Therefore we must understand the range of costs and penalties that will aff ect the cost of the runcut. Penalties and costs will be specifi cally described in the labor agreement. The scheduler needs to be aware of all relevant penalties. When reviewing or defi ning penalties as part of the run- cutting process, the scheduler should consider a range of questions. The key issues to consider are noted in the table below. Bonus/Penalty Issues To Be Aware Of Hourly Rate What is the base hourly rate? Is it the same for all run types? Does it vary by day of week (particularly on weekends)? Are there time-of-day pay rate diff erences (e.g., late night or overnight pay rates)? • • • • Overtime At what stage is overtime paid, daily or weekly? What is the overtime factor (e.g., time and a half, double time, etc.)? Is the rate the same across all day types? • • • Guarantee What is the daily guarantee? Is guarantee paid for all run types? Is guarantee paid on a daily or weekly basis? (depending on how rosters are cre- ated this can have signifi cant cost impacts) • • • Spread Is there a spread penalty? At what stage does it apply? What is the penalty rate (e.g., time and a half, double time, etc.)? Does the penalty increase for longer spreads (e.g., ͭ.ͱ over ͭͬ hours but ͮ.ͬ over ͭͮ hours)? Does the spread penalty apply only to split runs? • • • •

Chapter 5. Runcutting 5-17 Level ͭ. Basic Runcutting Bonus/Penalty Issues To Be Aware Of Others Are all aspects of a run paid at the regular rate (e.g. are report/clear and relief allowances paid)? Are penalties cumulative? If a run goes into overtime and has a long spread, are both penalties paid? If so, are they cumulative or multiplicative? If not, which takes precedence? Are penalties based on all aspects of the work day, or do some parts (such as travel or report/clear allowances) not incur penalties? • • • In developing runcuts it is imperative to understand the range of penalties that apply and how they interact. For example, if spread penalties are at a higher premium than overtime, it may be more effi cient to create more long straight runs. This will incur an overtime cost but avoid split runs and the accompanying (higher) spread premium. It is also necessary to intimately understand the range of costs and penalties in order to pro- vide an accurate assessment of the cost of a particular runcut solution. Computers and Costs Fortunately this is one area where the need for everything to be numerically based is not so dif- fi cult. Even at a spreadsheet level it is not diffi cult to cost runs automatically, based on known costing or penalty rates. For example, a formula to calculate a spread penalty, where the rate is time-and-a-half for all time worked over ͭͬ hours, is quite simple and is shown below: Type Run Start Piece 1 End Piece 2 Start Run Finish Total Spread Total Work Spread Penalty Overtime Str 6:00 14:30 8:30 8:30 0:00 0:45 9:15 $115.63 Spl 6:05 9:15 14:15 19:00 12:55 7:55 4:22 0:00 12:17 $153.65 Overtime 8:00 Overtime Rate 1.5 Spread Penalty 10:00 Spread Penalty Rate 1.5 Base Hourly Rate $12.50 Total Cost Now already in this simple example we are asking the spreadsheet to calculate diff erent types of penalties based on two diff erent variables (spread and overtime). And then how worked hours are calculated is based upon whether the run is straight or split. In order to calculate the total dollar cost it is best to express all costs and penalties as hours, and then at the end simply multiply by the base hourly rate to get an actual cost. Below are the

Level ͭ. Basic RuncuttingChapter 5. Runcutting 5-18 formulas used to derive the penalty and cost impacts. As always with schedule spreadsheets, the scheduler should avoid any fi xed numbers where possible, and revert to using formulas as often as possible. This reduces the chance of error and ensures values are automatically updat- ed as base inputs (in this case, start and fi nish times) are changed. Total Spread Total Work Spread Penalty Overtime =E8-B8 =IF(D8>0,(C8-B8)+(E8-D8),(E8- =IF(F8>$C$16,(F8-$C$16)*$C$17,0=IF(G8>$C$13,(G8-$C$13)*$C$ =G8+H8+I8 =(J8/0.00069444/60)*$ =E8-B8 =IF(D9>0,(C9-B9)+(E9-D9),(E9- =IF(F9>$C$16,(F9-$C$16)*$C$17,0=IF(G9>$C$13,(G9-$C$13)*$C$ =G9+H9+I9 =(J9/0.00069444/60)*$ Total Cost When using computerized systems to evaluate penalties and costs the user should be able to answer the questions in the table above. These can then be converted to formulas and calcu- lated automatically by the system. In the above example, the totals are all calculated, and only the start/fi nish times or penalty factors are adjusted. Operator Reliefs Reliefs are a key part of the basic underlying data required to develop a runcut, however simple or complex they may be. All agencies will have reliefs defi ned in some form, whether it be informal and based upon existing practices, or defi ned in detail within a computerized schedul- ing package. Do not take existing reliefs as a fi xed input. Always question why the current locations are used, the types of reliefs applied, and the relief travel times. This is part of the critical ongoing process of optimizing key scheduling inputs. Reliefs—An Introduction What are reliefs? The term is pretty much self-explanatory in that it refers to one operator get- ting off the vehicle, typically to go on a break or end his or her daily run. The operator is then “relieved” by another operator who takes over operation of the vehicle. Reliefs can occur at several times of day, including: Start or end of a run Start or end of a meal break Between two pieces of a multipiece run These actions could be best described as “activities,” and the point at which they start or end “activity locations.” • • • relief The replacement of one operator on a vehicle by another operator on the same vehicle. The fi rst operator may be going on a break or may be ending his/her work day. The second operator may be starting his/her work day or coming back from a break.

Chapter 5. Runcutting 5-19 Level ͭ. Basic Runcutting Defi ning Reliefs There are three key considerations in defi ning relief opportunities. Some of the key factors to be considered are listed below. 1. Relief Locations Below are some of the key considerations that should be applied when deciding on relief loca- tions: Relief locations should be strategically placed to minimize travel times and maximize relief opportunities. Proximity to facilities for operators is important. Be mindful of operational complexities for street supervisors where there are exces- sive, multiple relief locations. Consider appropriateness of relief locations, including safety issues. May need to limit reliefs at certain locations to daylight hours only, for example. Consider the availability of relief cars or buses. 2. Relief Types There are numerous ways for reliefs to occur. The “type” of relief primarily refers to the means of transport to that on-street relief location from the garage, start point, break location, or other relief location. The types of reliefs typically used are listed below: Walk. Where the operator simple walks between the relief location and their next activ- ity location. The activity location may be the garage (for a break or for the start or end of day), another relief location, or other break location. Typically used for short distanc- es only. Car. Similar to walking, but instead an offi cial agency car is used to ferry crews between relief locations and their previous/next activity point. Many agencies will have fl eets of cars used for this purpose. The concept often leads to car pooling to relief loca- tions, where multiple crews may share a car travelling to the same or proximate relief location(s). Bus as “taxi.” For this type of relief buses are used to ferry one or multiple operators instead of cars. The buses used for this type of relief would be “spares,” notably unused buses in the midday, when most street reliefs occur. • • • • • • • • relief location A designated point on a route where operators or crews may be scheduled to begin or end their run or a piece of their run. This can include the garage itself. car pooling The use of an automobile to ferry more than one operator between the garage and a relief location, or even between two relief locations.

Level ͭ. Basic RuncuttingChapter 5. Runcutting 5-20 Pull. Pull reliefs also use the bus itself as a means to travel to/from relief locations. But this type of relief actually cuts the block into two separate blocks, rather than bus as taxi, where the in-service block remains intact. Travel as Passenger. Here the operator travels on an in-service route to or from the relief location. This often works best where a high-frequency service operates past the garage, but can also be utilized to travel between relief locations. Which relief type is best? As with so many scheduling decisions there is no “best” solution, or simple answer. The types of reliefs used will depend upon a range of factors. The table below provides an indication of the pros and cons of each. In addition the types of reliefs may be combined. For example one operator may return to the garage using a “pull” relief. On that same bus it is possible to have other operators travelling as passengers back to the garage. Type of Relief Pros Cons Walk Simplest solution for reliefs, re- quires no vehicle mileage or car fl eet Least likely to result in operational complications • • Limited to locations near the garage, or near to other relief loca- tions • Car Fastest and most fl exible way to get to remote relief locations Car mileage is less expensive than bus mileage Can car pool to achieve multiple reliefs with one car • • • Maintaining a car fl eet has its own set of logistical and maintenance issues Complicates scheduling, requires tracking/scheduling of “car blocks” in some cases • • Bus as taxi Uses available off -peak vehicles for relief trips, avoiding need to pur- chase a car fl eet • Expensive compared to using car for same purpose Buses lack fl exibility for turning movements needed in some loca- tions Complicates scheduling of buses, requires “travel trips” to be sched- uled into “blocks” Need to be careful to avoid during peak times or vehicle requirements may increase • • • • • •

Chapter 5. Runcutting 5-21 Level ͭ. Basic Runcutting Type of Relief Pros Cons Pull Operational simplicity as operator changeovers are avoided Allow better vehicle ownership (can stay with the vehicle all day, even on a split or multipiece run) Avoids need to cover the layover between trips • • • Can increase bus mileage, which is more expensive than car mileage Need to be careful to avoid pulls during peak times or vehicle re- quirements may increase • • Travel as Passenger Avoids need for additional buses, bus mileage, or a car fl eet • Can have reliability implications, since service reliability issues will cause delayed changeovers, result- ing in a cascading eff ect Expensive compared to using car for same purpose (usually results in longer time) Complicates scheduling processes if actual scheduled times are used as the travel schedule is built around fi xed trip departures (and possibly walk-travel time to a con- necting bus). Also requires sched- ule changes on routes dependent on “travel-in-service” routes used for reliefs when their schedule changes. • • • Unattended Avoids the cost of covering layover time at either side of a relief • Few agencies (justifi ably) are pre- pared to leave vehicles unattended at relief locations • 3. Travel Times The key message here is to look for precision. Greater accuracy in travel time defi nition allows two often confl icting key scheduling outcomes (operational reliability and scheduling effi - ciency) to be better achieved. Some key considerations that should be considered when setting travel times include: Diff erentiate travel times according to the mode used (from the above list).•

Level ͭ. Basic RuncuttingChapter 5. Runcutting 5-22 Consider time requirements at diff erent times of day. In many cases travel times will dif- fer greatly across diff erent time periods. The same may apply to diff erent service types. Understand the time requirement associated with vehicle positioning at relief location bus stops. In some locations this may be signifi cantly diff erent according to which cor- ner the stop is on and which direction it faces. When using “travel in service,” use actual times instead of average times, as much as possible. Average times tend to generally overstate the time required and result in ef- fi ciency losses. Alternatively, there will be some cases where operators are not given enough time to get to the relief location, simply because there is not a trip scheduled to get them there at the right time. At most (but not at all) agencies the time taken to travel between relief locations and an operator’s garage, home, or other relief location will be paid. The times are often defi ned by schedulers, at times in collaboration with the union. Most computerized scheduling tools require that scheduling times be specifi cally defi ned. This is a good thing as it enforces some consistency in how scheduling data is maintained. Most also allow for exceptions (e.g., for certain times of day). A word of warning on travel times: times must be set at levels that will allow robust operations. If there is not enough time given to get from a garage to a relief location, the service could run late or have a cascading reliability impact. However, on the reverse side of the equation, exces- sive travel time can result in simply wasting money. This applies particularly to agencies who schedule longer runs with more overtime, since the end-of-day travel time will be at time-and-a half in many cases. The above points imply a need to consistently review, measure, and refi ne travel times to en- sure the right balance between operational reliability and scheduling effi ciency. Relief Opportunities The concept of relief opportunities is critical in understanding how to create a runcut. One of the diffi culties presented by creating runs (or daily shifts) for public transportation is the fact that you cannot simply decide when or where to have staff start work, end work, or take meal breaks. These times are all constrained by the blocks or trip themselves, plus where and how reliefs can be enacted. • • • relief opportunities Times within a block when reliefs could be scheduled, typically at the end of a trip or when the vehicle passes a specifi ed relief location.

Chapter 5. Runcutting 5-23 Level ͭ. Basic Runcutting Scheduled reliefs are only known after the runs have been created. Actual reliefs are a subset of the total potential reliefs, or relief opportunities, presented by the vehicle blocks. The dia- gram below provides an example of the potential relief opportunities for a single uncut block. In this example there is a single relief location at the end of a trip. The double lines indicate the potential for an operator to be relieved at either the end or the start of a trip. Conventional wisdom suggests that the end of a trip is the best time for a relief, to allow the layover time to cover for unforeseen issues, to accommodate operator changeover time, and to avoid custom- er inconvenience. Relief locations may also be at trip midpoints. And in some cases routes may have multiple relief locations. The location and number of relief locations will govern the number of potential relief opportu- nities. Why is this worth considering? In short, because the number and location of relief op- portunities place one of the most signifi cant mathematical constraints on the runcutting process. Again, this applies to both spreadsheet and computerized runcutting. Simply put, more relief opportunities provide more potential options for creating runs. And as noted above, minimizing constraints is a sure-fi re way to improve runcutting effi ciency. In an extreme example, if you were to have relief locations every minute along a route you would then be able to cut or create pieces at any given minute. As a result it would be simple to create runs of optimum length (optimal being a nebulous term in this example, but assume it means most effi cient for now)—just cut the block into pieces of any length as required. But the reality of transit operations excludes this as an option. The occurrence of reliefs pres- ents a range of operational issues and can aff ect service quality—driver changeovers can take between one and fi ve minutes. Any scheduled relief has an inherent operational risk due to un- reliability of travel times, traffi c conditions, or human factors such as lateness for work. There- fore the scheduling of reliefs needs to be undertaken with operational impacts in mind. One of the key tenets of scheduling is to make for operations that are as simple as possible: the less complex the operation, the less risk of issues occurring. However, as with all things sched- uling-related, the tradeoff of complexity needs to be balanced against effi ciency.

Level ͭ. Basic RuncuttingChapter 5. Runcutting 5-24 So we have a classic scheduling conundrum—mathematically it will almost certainly be cheap- er to allow numerous relief locations and relief types. But operationally we would like to limit the number of relief locations and opportunities. In general agencies tend to defi ne relief locations at places where travel time will be mini- mized, access for buses and/or cars is good, the location is safe, or there are other strategic benefi ts (such as it is near an acceptable bathroom, meal location, or a junction of several routes). Basic Runcutting Defi ning Our Objectives Before the runcutting process can begin, the scheduler needs to be sure that all the necessary inputs are in place. These inputs were described in some detail in the previous section and included: trips and vehicle blocks; rules and constraints; and relief types and travel times. Runcutting is like a puzzle with many solutions. There is no single “right answer,” but there is an answer that most closely matches the objectives of your agency. Before doing any detailed work we need to clearly understand our objectives. What are we try- ing to achieve when creating this runcut? It could be any combination of the following: To minimize operating costs To minimize the number of split runs To minimize the number of trippers To evenly distribute the work among the operators To ensure all reliefs are at one particular location To maximize long runs for AM starts To make the runcuts more “operations-friendly” To ensure all runs are legal (this is often overlooked) To achieve a target distribution of full-time and part-time work based upon labor agree- ments and/or current manpower availability These are only a few of the possible objectives that could defi ne what we are trying to achieve. There are many more, and we’ll discuss some of those later. For now just be sure to jot down • • • • • • • • • • • •

Chapter 5. Runcutting 5-25 Level ͭ. Basic Runcutting what the key objectives are before you start, and know that at times these may even confl ict with one another. But the key thing here is to be clear on your objectives. Work Rules One of the objectives of any runcut is to ensure that the runs are “legal”—meaning they con- form both to the employment laws in your jurisdiction and to the requirements of your labor agreement. The table below lists some fairly typical types of work rules that we will use in cre- ating this fi rst runcut. Before you begin a runcut for your property, make sure you are familiar with all of the requirements that aff ect legal runs. Work Rule Requirement Minimum Platform Time (full-time run) Ͳ:ͬͬ Maximum Platform Time ͭͬ:ͬͬ Minimum Platform Time (tripper) ͮ:ͬͬ Maximum Platform Time (tripper) ͱ:ͱ͵ Report Allowance (start of run) ͬ:ͭͱ Clear Allowance (end of run) ͬ:ͭͱ Clear Allowance (end of fi rst half of split run) ͬ:ͬͱ Report Allowance (start of second half of split run) ͬ:ͭͱ Maximum Spread ͭͯ:ͬͬ Run Type Limits ͱͬ% minimum straight runs ͮͱ% maximum split runs ͮͱ% maximum trippers Guarantee (Daily) ʹ:ͬͬ Overtime (Daily) Time and a half over ʹ:ͬͬ Spread Penalty Time and a half over ͭͬ:ͬͬ Reliefs Must be at “Point A” All reliefs are taken as travels using a car ͬ:ͭͬ travel time from garage to Point A It is essential for the runcutter to both under- stand the “rules” for your property and to know which ones can never be violated, and which ones may be “bent” when there are no solutions that meet all of the requirements. Often bending the rules requires management approval, and the runcutter must understand the process before beginning. Tip To get a quick estimate of the total runs required: Runs = vehicle hours/ desired run length . Tip

Level ͭ. Basic RuncuttingChapter 5. Runcutting 5-26 Knowing the Answer So now we’re ready to jump in and create the runs, yes? No! Because runcutting has many possible solutions, but only one optimal result, it is essential that we get as much information as possible before we begin cutting. We begin with a number of techniques that allow us to ap- proximate the solution to the runcutting puzzle before we begin the cut itself. Let’s start with our basic Route ͵ͳ-Broad Street schedule example—the solution with a ͭͱ-min- ute peak and ͯͬ-minute off peak service. There are a few simple things we can look at. Firstly we could simply divide total vehicle hours by a desired average platform time per run to come up with an estimate of total runs. To come up with an average platform time you could look at your existing operating information—on average how much travel, sign, or “non-platform” time does each run have (just stick with averages here). In our case we know the garage is ͭͬ minutes from Point “A.” We will assume a ͭͱ-minute report allowance at the start of a run and a clear allowance of ͭͱ minutes at the end. Therefore any run will have at least Ͱͬ minutes of non-platform time (ͭͱ sign on allowance, ͭͱ sign off allowance, and probably at least one ͭͬ-minute travel). This means we would want to cut runs with somewhere around ͳ:ͮͬ platform time to reach our desired eight hours. Below are the blocks we produced earlier. For many transit operators, the desired run length will be eight hours minus time allowed for sign in and sign out and travel time. Hours Summary Block Garage Depart Garage Arrive Hours 1 5:46 9:19 3:33 2 6:01 18:38 12:37 3 5:50 18:19 12:29 4 6:05 9:38 3:33 5 6:20 19:53 13:33 6 15:01 19:49 4:48 7 15:20 19:19 3:59 Total 54:31 This gives us a total of ͱͰ hours. If we assume ͳ:ͮͬ hours of platform time per run we come up with an estimate of ͳ.ͱ runs. Of course ͳ.ͱ isn’t a viable number of operators so we’ll assume that we are looking at either ͳ or ʹ runs.

Chapter 5. Runcutting 5-27 Level ͭ. Basic Runcutting What else can our blocks tell us about the optimal runcutting solution? Let’s go back and de- construct the ”Vehicles in Operation Graph” from the previous section. What does it tell us? We can see there are three vehicles operating during the day and fi ve during the peaks. Immediately that suggests we will have either two split or four trippers, to cover those addi- tional peak blocks. That leaves us with three all-day blocks, each operating well in excess of the allowable platform time for a straight run. These blocks will therefore need to be cut, potential- ly into two straight runs each. But as these blocks are only between ͭͮ and ͭͯ hours in length we can also expect shorter straight runs, possibly with some guarantee time. By this stage of our analysis we now have an idea of not only how many runs but also of the likely breakdown of those runs by run type and the potential for guarantee time—all without doing any runcutting at all! This is what we mean by “knowing the answer” before starting.

Level ͭ. Basic RuncuttingChapter 5. Runcutting 5-28 As we’ve stated elsewhere, many scheduling tasks are made easier by simple graphical repre- sentation of the information rather than detailed numerical tables. This applies equally to run- cutting tasks. A good visualization tool is to lay out the blocks in a graphical manner as shown below. Block 1 2 3 4 5 6 7 17 18 19 20 2111 12 13 14 15 165 6 7 8 9 10 While the times are rounded to ͭͱ-minute segments, it enables us to pretty quickly see block lengths and suitability for run types. For example blocks one, four, six, and seven appear suit- able for split run or tripper pieces, and blocks two, three, and fi ve appear suitable for straight runs. Creating the Runs The hard work is far from done. Although we have an idea of how the runcut will look, we need to work through the detail of that runcut and create the runs. Looking back at our graphical blocks diagram we have a good idea that blocks one and four will combine with blocks six and seven to create two split runs. And we expect to cut blocks two, three, and fi ve into two straight runs each. However the graphic does not provide enough detail about specifi c times and locations. So we turn back to the schedule. Let’s start by trying to cut Block ͮ into straight runs. Block ͮ starts at Ͳ:ͬͭ and ends at ͭʹ:ͯʹ, totaling ͭͮ:ͯʹ. Our run gets ͭͱ minutes report allowance and so signs on at ͱ:ͰͲ. We’d like to be as close to eight hours as possible (to avoid guarantee or overtime) and so ideally we would like to fi nish at around ͭͯ:ͰͲ. Allowing for ͭͱ minutes of clear allowance, and ͭͬ minutes of travel time, that means we would like to cut the run at around ͭͯ:ͭͭ. But the only relief opportunities (i.e., the times that the bus passes Point A) in this time vicinity are at ͭͮ:ͬ͵ or ͭͯ:ͯ͵. So we have a judgment call to make, and it has implications for not only the AM run but of course the PM run also. The best thing is to look at the two options and see the implications as below.

Chapter 5. Runcutting 5-29 Level ͭ. Basic Runcutting Option 1 - Cut at 12:09 Time Place Type Time Place Type 1 Str 2 5:46 6:01 Garage Pull 12:09 A Street 12:34 6:08 0:30 0:10 6:48 6:48 1:12 0:00 0:00 8:00 1.304 2 Str 2 11:44 12:09 A Street 18:38 Garage Pull 18:53 6:29 0:30 0:10 7:09 7:09 0:51 0:00 0:00 8:00 1.234 12:37 1:00 0:20 13:57 2:03 0:00 0:00 16:00 1.268 Option 2 - Cut at 13:39 Time Place Type Time Place Type 1 Str 2 5:46 6:01 Garage Pull 13:39 A Street 14:04 7:38 0:30 0:10 8:18 8:18 0:00 0:09 0:00 8:27 1.107 2 Str 2 13:14 13:39 A Street 18:38 Garage Pull 18:53 4:59 0:30 0:10 5:39 5:39 2:21 0:00 0:00 8:00 1.605 12:37 1:00 0:20 13:57 2:21 0:09 0:00 16:27 1.304 Type Report Hours Travel Hours Report Hours Travel Hours Type Report Time Sign Off Time Plat Hours Total Hours Total Hours Spread Guarantee Run # Block # Report Time Start End Pay/ Plat Pay/ Plat Start EndBlock # Plat Hours Run # Overtime Spread Penalty Paid Hours Sign Off Time Spread Guarantee Overtime Spread Penalty Paid Hours The answer in both cases is “not good!” Both options result in substantial guarantee time. Why? Block ͮ is only ͭͮ:ͯͳ, which allows an average of Ͳ:ͭʹ of platform time. Even adding the travel, report, and clear allowances this only gets us to an average run length of ͳ:ͬͬ, mean- ing somewhere around an hour of guarantee time for each run. In this case we choose to cut the blocks at ͭͮ:ͬ͵ since this incurs the minimum total cost. However we would be looking for other solutions as we cut the runs since we have incurred some serious ineffi ciency in the fi rst two runs. In the end we will choose Option ͭ based upon the fact that it incurs ͮͳ less minutes of paid time. Is ͮͳ minutes important? The answer is an emphatic “yes!” In scheduling every minute matters. To illustrate this point let’s consider the costing impacts of that ͮͳ minutes. Over a year (as- suming ͮͱͱ weekdays) that comes to a total of ͭͭͰ.ͳͱ hours. At a base hourly rate of ͈ͮͬ that comes to over ͈ͮ,ͬͬͬ, which will end up more once variable benefi ts are applied. From this example we can use a rule of thumb that every hour of paid time equates to over ͈ͱ,ͬͬͬ annu- ally—that’s why not wasting any paid time in every situation is critical. We will push ahead and cut straight runs from Blocks ͯ and ͱ (the other two all day blocks) in the same manner as we cut Block ͮ. Again we need to decide when to cut the block to mini- mize the overall paid time while keeping the runs legal. In this case if we cut Block ͯ at ͭͭ:ͬ͵ the fi rst run becomes less than six hours, so we are forced to cut it at ͭͮ:ͯ͵. We cut Block ͱ at ͭͯ:ͬ͵ as that option best balances the work hours of the AM and PM runs. Our runs summary now looks like the following: Runcutting, scheduling, and planning are often iterative processes. When you begin your runcut, you may notice that adjusting the schedule slightly can either add service at no cost, or could save signifi cantly with a slight trim. Good schedulers do not just cut what’s in front of them, but make sure everyone responsible understands the implications of the schedule. Tip

Level ͭ. Basic RuncuttingChapter 5. Runcutting 5-30 Option 1 - All Long Blocks Now Cut Time Place Type Time Place Type 1 Str 2 5:46 6:01 Garage Pull 12:09 A Street 12:34 6:08 6:48 6:48 1:12 0:00 0:00 8:00 1.304 2 Str 2 11:44 12:09 A Street 18:38 Garage Pull 18:53 6:29 7:09 7:09 0:51 0:00 0:00 8:00 1.234 3 Str 3 5:35 5:50 Garage Pull 12:39 A Street 13:04 6:49 7:29 7:29 0:31 0:00 0:00 8:00 1.174 4 Str 3 12:14 12:39 A Street 18:19 Garage Pull 18:34 5:40 6:20 6:20 1:40 0:00 0:00 8:00 1.412 5 Str 5 6:05 6:20 Garage Pull 13:09 A Street 13:34 6:49 7:29 7:29 0:31 0:00 0:00 8:00 1.174 6 Str 5 12:44 13:09 A Street 19:53 Garage Pull 20:08 6:44 7:24 7:24 0:36 0:00 0:00 8:00 1.188 38:39 42:39 5:21 0:00 0:00 48:00 1.242 Spread Penalty Paid Hours Pay/ Plat Type Sign Off Time Plat Hours Total Hours Spread Guarantee Overtime Run # Block # Report Time Start End Again we can see that the runs are generally ineffi cient, each resulting in guarantee time. At about this stage we may want to go back, rethink the schedules, and note that it would be possible to extend the hours of operation, by one round trip for each vehicle, with minimal cost impact. The cost impact would be minimal as the additional hours would be absorbed by the available guarantee time, meaning the additional paid hours would be negligible. The real cost would be the variable vehicle costs, and these tend to form only a small percentage of total operating costs. But we should leave such decisions until we have worked through all the runs, both straights and splits. Next we might create split runs from the smaller peak blocks. We have four peak blocks (Blocks ͭ, Ͱ, Ͳ, and ͳ), two in each peak. These should be able to form two split runs. We can simply list the appropriate blocks as follows to get an idea of how the runs will look. Block Garage Depart Garage Arrive Hours 1 5:46 9:19 3:33 6 15:01 19:49 4:48 Spread 14:33 Work Hours 9:11 4 6:05 9:38 3:33 7 15:20 19:19 3:59 Spread 13:44 Work Hours 8:22

Chapter 5. Runcutting 5-31 Level ͭ. Basic Runcutting Now we have two split runs. The block lengths result in these runs being longer than the straights. We’ll assume that is acceptable and push ahead. This gives us two splits and six straights, a total of eight runs. All blocks are covered and we must be done, right? Wrong! Go back and look at the split runs—they fail the very fi rst test of meeting our scheduling criteria—they are not legal! The two split runs are well in excess of the ͭͯ-hour maximum spread mandated by our labor rules. This leaves us some potentially diffi cult decisions. The simplest solution here could be to use some trippers. But the Labor Agreement only allows a maximum of ͮͱ%, in eff ect meaning two of our eight. That would still leave us with two blocks not yet assigned to runs. One of the lessons we are learning here is that runcutting small solutions is often the most dif- fi cult of the runcutting tasks that can be undertaken. Going back to the math, there are simply fewer options for cutting and hooking pieces in a manner that will allow a feasible solution. For some small runcutting problems there may in fact only be one feasible solution possible. So what are our options at this point? Let’s note them and discuss the pros and cons: Create four trippers, one for each of the peak blocks . As noted above this is the sim- plest procedure but fails to meet the ͮͱ% maximum rule. There may be an opportunity to negotiate with the union and allow some variation to the labor rules, or to let one or two runs through that don’t meet the rules. This is not the preferable option but at times may be the most pragmatic, particularly if faced with tight timeframes. Revisit the schedules and blocks, potentially reducing the spread of some of the blocks to allow the split runs to fall within the ͭͯ-hour spread limit. Or at least adjust one of them this way and leave the other two peak blocks as shorter part-time pieces. This can be an expedient approach for runcut purposes but may aff ect desired service level outcomes. Revisit the entire process and recut the pieces to allow “better” runs to be cut, and a legal solution to be created. This approach, assuming it allows us to fi nd a solution, is often preferred as it allows us to meet our runcut objectives without aff ecting the schedule or blocks. Revisit the blocking to allow blocks that will cut into runs in an effi cient manner. This is of course the approach that will take the most time and could aff ect the schedule itself. Another key lesson we are learning is that the schedules and blocks cannot be written in iso- lation from the runcut process. To do so is to ignore the fact that scheduling is an integrated and holistic process that requires all aspects to be considered. The scheduler who undertook the blocking would have given an indication of the costing impacts of the proposed schedule, • • • •

Level ͭ. Basic RuncuttingChapter 5. Runcutting 5-32 based on the blocks developed. Right now those costs are probably completely inaccurate as the runcuts are not able to be provided effi ciently with the initially developed blocks. For the sake of expediency we will assume that in this case we are allowed one split run to go over the ͭͯ-hour spread maximum. In practice this is not an option for the scheduler but for now we will make this assumption as it allows us to see how the process unfolds. This will allow us to create one split run and two trippers from the four peak blocks. How does one choose which is the split and which will be the part-time blocks? In this case we choose the combination of blocks that minimizes the spread of our split run—Blocks Ͱ and ͳ. Blocks ͭ and Ͳ then become two trippers. Well done—we now have a completed runcut. Our run guide now has a completed runcut and is provided below. Option ͭ - Completed Runcut Time Place Type Time Place Type 1 Str 2 5:46 6:01 Garage Pull 12:09 A Street 12:34 6:08 0:30 0:10 6:48 6:48 1:12 0:00 0:00 2 Str 2 11:44 12:09 A Street 18:38 Garage Pull 18:53 6:29 0:30 0:10 7:09 7:09 0:51 0:00 0:00 3 Str 3 5:35 5:50 Garage Pull 12:39 A Street 13:04 6:49 0:30 0:10 7:29 7:29 0:31 0:00 0:00 4 Str 3 12:14 12:39 A Street 18:19 Garage Pull 18:34 5:40 0:30 0:10 6:20 6:20 1:40 0:00 0:00 5 Str 5 6:05 6:20 Garage Pull 13:09 A Street 13:34 6:49 0:30 0:10 7:29 7:29 0:31 0:00 0:00 6 Str 5 12:44 13:09 A Street 19:53 Garage Pull 20:08 6:44 0:30 0:10 7:24 7:24 0:36 0:00 0:00 1 5:31 5:46 Garage Pull 9:19 Garage Pull 9:24 6 14:46 15:01 Garage Pull 19:49 Garage Pull 20:04 8 Pt 4 5:50 6:05 Garage Pull 9:38 Garage Pull 9:53 3:33 0:30 0:00 4:03 4:03 0:00 0:00 0:00 9 Pt 7 15:05 15:20 Garage Pull 19:19 Garage Pull 19:34 3:59 0:30 0:00 4:29 4:29 0:00 0:00 0:00 54:32 4:50 1:00 60:21 5:21 0:35 2:09 0:50 0:00 2:09 Spread Penalty 7 Spl 8:21 9:11 14:18 0:00 0:35 Sign Off Time Plat Hours Total Hours Spread Guarantee Overtime Report Hours Travel Hours Run # Type Block # Report Time Start End Note that in the above spreadsheet there are as many formulas as possible, holding true to our previously stated aim of avoiding typing numbers that can be calculated in every possible circumstance, no matter how complex the formula. run guide A summary of runs that describes start/fi nish locations, work hours, and cost element breakdowns. The Run Guide is the principal docu- ment that describes all of the runs available for bid.

Chapter 5. Runcutting 5-33 Level ͭ. Basic Runcutting Now we need to carefully check that we have all trips covered, all runs are legal (except for the agreed Split Run), that our numbers add up, and that the totals are calculated correctly. You should have a checklist and templates to assist with this process. One solution is to go back and review our graphical depiction of the blocks. For each block, “color in” or trace over the pieces and runs. This way we can be clear that the runcut covers the blocks entirely. The table below summarizes the total cost of the runcut. Runcut Summary Straight Runs 6 67% Split Runs 1 11% Total FT Runs 7 78% Part Time Runs 2 22% Total Operators Required 9 Hours Breakdown Total % Avg Platform 54.5 90% 6.1 Report 4.8 8% 0.5 Travel 1.0 2% 0.1 Total Work Hours 60.4 6.7 Penalties % Avg Spread 2.2 3% 0.2 Overtime 0.6 1% 0.1 Guarantee 5.4 8% 0.6 Total Paid Hours 68.5 Pay/Plat Ratio 1.2553 Solution 1

Level ͭ. Basic RuncuttingChapter 5. Runcutting 5-34 As the table shows, we have seven full time and two trippers. The total weekday paid hours are Ͳʹ.Ͳ, and we have a ratio of driver pay hours to total platform hours (pay/plat ratio) of ͭ.ͮͱͱ. So is this a “good” runcut? To answer this question go back and review against the original stat- ed objectives. Clearly there is some ineffi ciency in the runs—over ͭͮ% of the cost is incurred in guarantee, spread, and overtime penalties. But these are a function of the constraints of the runcut (the rules and the blocks we started with) more than a refl ection of the competency of the runcut itself. We would also need to look at historical trends for the route and the agency to understand if the solution fi ts in within normal bounds. Reliefs No job is over until the details are done. There are still a few things that need to be completed before the runcut can be considered fi nal (assuming the agency decides that this is the runcut to implement). A key task relates to reliefs. The rules stated that all travels to relief locations must use a car. Generally this will mean that an agency has a car or fl eet of cars that operators use to get to and from relief locations. We need to be sure we have enough cars for the travels dictated by the runcut. To be sure this is the case each travel required by the runcut should be listed—in ef- fect a list of “trips” that need to be scheduled into a “block” to be operated by the car (or cars). Car Trips 3 5 Driven By Run # 2 4 6 12:29 12:39 12:39 12:49 12:59 13:09 13:09 13:19 11:59 12:091 12:09 12:19 Leave Garage Arrive Relief Leave Relief Arrive Garage Driven By Run # In this case the one car will be able to undertake each of the relief trips required. However, there may be cases when multiple reliefs occur at the same time, or at diff erent locations, and more than one car could be required. In such cases a separate scheduling task is required to match cars to reliefs (in the above manner) to obtain car vehicle requirements.

Chapter 5. Runcutting 5-35 Level ͭ. Basic Runcutting Information and Outputs Now it is a matter of generating the appropriate outputs and distributing them accordingly. Most transit agencies have their own specifi c set of reports or outputs for distribution. There- fore we will not provide samples here, but just list those that are typical. Run Guide. The run guide displays a summary of each run—where it starts, fi nishes etc., along with information about the length of runs. It may also include a breakdown of hours and penalties. This report is used across the organization as a means of quickly reviewing runs. Paddle. The paddle comes in many shapes, sizes, and formats. The aim of the paddle is to provide the operator with information regarding his or her workday—what time the work day starts/ends, how to get to/from relief locations, and the trips to be operated (complete with times at all timepoints). Headway Sheet. We include this in runcutting as it may be helpful to operations staff at this stage if the run number for each trip is included. Automating the Runcut The above process used spreadsheets to develop a runcut. However the majority of transit agencies now use computerized scheduling software packages to generate runcuts automati- cally. Why bother with manual techniques then? As stated previously it is necessary to understand the manual techniques in order to understand, and be able to review, the solutions provided by automated scheduling systems. Many schedulers make the mistake of setting up basic param- eters and then “pressing the button” to produce a completed automatic runcut. Our experience is that it is extremely diffi cult to model all of the subtleties in a format (i.e., a logical or mathematical statement) that will allow a computerized system to fully automate. The more likely outcome is that solutions will be produced that achieve most of the aims of a runcut but require some manual intervention or tweaking to fi nalize. Great care must be taken to defi ne all rules, constraints, and preferences to meet the runcut objectives. Take our above example as a good case. Here we were able to agree to allow one “bad” run with a spread above the maximum allowed. However if we used an automated sys- tem, with a hard limit of ͭͯ hours set, it would not have been capable of generating the solu- tion we created manually. This is an example of where the skill and experience of the scheduler are required to assist the automated tool to generate an outcome. In this example we may • • • paddle An output of the scheduling process that provides the operator with information regarding his or her workday—what time the work day starts/ends, how to get to/from relief locations, the trips to be operated, and times at all timepoints. If an operator drives on more than one route in the day, the paddle will have all trips shown sequentially, as well as travel paths between routes if needed. The paddle may also include a list of route turns, route maps, farebox, headsign, and radio codes, and key intersections and stops that must be announced. run number The unique number assigned to each work assignment on a specifi c day.* At some systems, the run number is unique only when used in combination with a designator for the garage or the route or route group number. *Throughout this chapter, we have simplifi ed the examples by refer- ring to Run 1, Run 2, etc. Obviously, these numbers are not unique. In practice, agencies typically use multi-digit run numbers that can include the route or route group number, a garage identifi er, a code for weekday/Saturday/Sunday, and a code for time of day. Each of these run numbers is unique on any given day.

Level ͭ. Basic RuncuttingChapter 5. Runcutting 5-36 have needed to manually create and then force or fi x the run before resubmitting the runcut. Only then could the computerized system have generated the same solution. The computerized system will signifi cantly enhance the ability of the scheduler to produce runcuts if the scheduler is aware of how to make the most of the system as a tool, and not as a solution itself. Computerized systems also signifi cantly enhance the fi nal parts of the pro- cess—error checking and report production—to provide a level of automation and accuracy not otherwise possible. Other Factors Before ending the basic section, it is worthwhile to cover a few qualitative issues that can sometimes get lost in the quest for effi ciency. The fi rst issue is operator fatigue. Even in the absence of rules limiting spread time, it is a good idea to ensure that an operator’s day is not too long. The scheduler’s biggest concern regard- ing fatigue is to make sure that rostered operators get at least eight hours off between work shifts, whether or not this is spelled out in the labor agreement. Another issue is how “optimal” runcuts are defi ned. Cost, number of drivers, and number of buses are all extremely important factors, but poor run structure (and poor rostering) can cre- ate morale problems, leading to higher turnover rates and increased recruitment and training costs. These factors could eventually outweigh any cost effi ciency benefi ts. A good scheduler will balance working conditions and effi ciency factors in the runcutting process.

Chapter 5. Runcutting 5-37 Level ͭ. Basic Runcutting LEVEL 1 End of Basic Runcutting. The Intermediate Section of Runcutting continues on the next page. To jump to Rostering, go to page Ͳ-ͭ.

Level ͮ. Intermediate RuncuttingChapter 5. Runcutting 5-38 5.2 Intermediate Runcutting In the previous section we looked at some basic runcutting concepts and worked through a relatively straightforward example. However even in a “straightfor- ward” example we ran into some diffi cult issues and had to make a number of decisions. This only serves to highlight the complexities of runcutting. It should be noted that every agency, and every scheduler, approaches runcutting slightly diff erently. There are no universal approaches that can be applied to every scenario, since the inputs (described below) will rarely be the same from one agency to the next—most agencies will have at the minimum subtle diff erences in policies or work rules. And of course the blocks, relief locations, travel times etc. will always be diff erent. Runcutting Inputs In the previous section we defi ned the required runcutting inputs to include a complete set of trips and vehicle blocks, all relevant defi ned rules, defi ned relief types and travel times, and known limitations, such as cost limits or work rule preferences. To produce eff ective runcuts the scheduler needs to work with a complete picture of what is required, and with a full set of the necessary input data. This applies to runcutting solutions of any complexity, size, or nature. As runcutting gets more complicated it is important to have several additional documents available to the scheduler: The Schedule. As we are creating detailed runs with specifi c times at relief locations we obviously need to see the detail of the schedule. Runcutting seldom changes the actual service schedule, which is often designed around service considerations. Blocks Summary. In order to match up starts and fi nishes, and to view the blocks in a summary form, we need to generate a blocks summary. This summary can simply show, for each block, the start and fi nish times, total length or duration, vehicle type, and garage. Blocks Diagram. This is the key tool in developing the runcut at a conceptual level. The blocks diagram simply shows one line or bar per block. Use of a spreadsheet for creat- ing the blocks diagram is critical, including eff ective use of color-coding (as we will see throughout this section). If a blocks diagram is not a routine output of the run blocking process the runcutter will need to create a blocks diagram before beginning the runcut. • • • LEVEL 2 A block summary and block diagram are often generated during the blocking step of scheduling. However, blocks may be altered by the runcutting process, making these important inputs AND outputs of the runcutting process. Tip

Chapter 5. Runcutting 5-39 Level ͮ. Intermediate Runcutting Runcutting Outputs It is important that the outputs of the runcutting process be standardized as well. The follow- ing outputs are an integral part of developing runcuts. Run Guide. The run guide provides information about each run—start/fi nish time and location, total hours, type of relief, etc. It is built as the runs are created. As with all tools, if a spreadsheet is being used, maximize the use of formulas and minimize the use of “typed in” values. Runs Summary. The runs summary, produced at the end of the basic runcutting sec- tion, provides an overview of the totals of the runcut, including a breakdown of the components of runs (platform time, travel time, etc.). In some situations the Run Guide and the Run Summary may in fact be the same document or spreadsheet. When using a computerized scheduling system, many of the above items are provided auto- matically and can be viewed interactively (e.g., looking at the schedule while cutting runs). When using other approaches, spreadsheets and/or databases should be utilized as much as possible to minimize manual calculations, reduce errors, and improve the overall runcut qual- ity. Trips and Blocks The basic section of the blocking chapter provided an important tip: “Blocking is not done in isolation—it is an intermediate step between writing a schedule and developing driver assign- ments, and must be done with the ultimate goal of developing effi cient and legal driver work pieces.” However a typical blocking solution will focus more on the needs of the blocking pro- cess, and will not necessarily provide blocks that will cut neatly or effi ciently into a fi nal solu- tion. With this in mind we can revisit our runcut for Line ͵ͳ. Again we start with a completed set of trips and vehicles blocks. But this time we will start with a signifi cantly diff erent view—that the blocks can potentially be adjusted as necessary to allow a more eff ective runcut to be pro- duced. Let us explore this a little further by using the example from the previous runcut. The key issue we had was the length of the splits, or the start/fi nish times of the peak pieces that we assumed would form those splits. In the previous section, the beginning runcutter was told to consider the blocks as fi xed, which left some signifi cant issues that resulted in a poten- tially less effi cient runcut. The solution developed in that example required adding an operator (we used two part-time runs instead of one split). • •

Level ͮ. Intermediate RuncuttingChapter 5. Runcutting 5-40 Looking at the diagram of our blocks we can see that there are actually some starts and fi nish- es that, if combined, would allow us to create split runs within the prescribed ͭͯ-hour spread. Block 1 2 3 4 5 6 7 17 18 19 20 2111 12 13 14 15 165 6 7 8 9 10 In particular we are referring to matching the start times of Blocks Ͱ and ͱ with the end times of Blocks ͮ and ͯ. Looking at the start/fi nish times of our blocks below we can get a sense of whether this would allow split runs to be created within the ͭͯ-hour spread limit. Hours Summary Block Garage Depart Garage Arrive Hours 1 5:46 9:19 3:33 2 6:01 18:38 12:37 3 5:50 18:19 12:29 4 6:05 9:38 3:33 5 6:20 19:53 13:33 6 15:01 19:49 4:48 7 15:20 19:19 3:59 Total 54:31 Block Ͱ starts at Ͳ:ͬͱ (meaning a sign-on time of ͱ:ͱͬ) and Block ͯ ends at ͭʹ:ͭ͵ (meaning a sign off time of ͭʹ:ͯͰ). Block ͱ starts at Ͳ:ͮͬ (Ͳ:ͬͱ sign on) and Block ͮ ends at ͭʹ:ͯʹ (sign off of ͭʹ:ͱͯ). This means we have sign-ons at ͱ:ͱͬ and Ͳ:ͬͱ to match with sign off s at ͭʹ:ͯͰ and ͭʹ:ͱͯ. Logic tells us that the ͱ:ͱͬ sign on could not be paired with the ͭʹ:ͱͯ sign off as the spread would be ͭͯ:ͬͯ. So we need to match the ͱ:ͱͬ sign on (Block Ͱ) with the ͭʹ:ͯͰ sign off (Block ͯ), and match the Ͳ:ͬͱ sign on (Block ͱ) with the ͭʹ:ͱͯ sign off (Block ͮ). This would give us two split runs with spreads of ͭͮ:ͰͰ and ͭͮ:Ͱʹ, respectively. The trick is to now generate pieces of work that will allow us to create two split runs with the above start and fi nish times. In essence we are trying to achieve what is shown in the following

Chapter 5. Runcutting 5-41 Level ͮ. Intermediate Runcutting diagram. The two white pieces represent the matching of Blocks Ͱ and ͯ into one split run. The black pieces represent the matching of Blocks ͱ and ͮ into another split run. Block 1 2 3 4 5 6 7 17 18 19 20 2111 12 13 14 15 165 6 7 8 9 10 There are several ways to achieve this outcome. These are described below. 1. Recut the Pieces We cut the blocks into pieces as shown, using street or car reliefs, and create the runs accord- ingly. If this approach was used we would be left with one piece of Block ͭ and a piece of Block ͱ (from ͵:ͯͬ AM onwards) to be matched together to form a straight run. In eff ect the operator would cover Block ͭ, pull into the garage, and then travel back out to make the relief on Block ͱ at around ͵:ͯͬ AM. This is known as a two-piece or a multipiece run. The swing time between the two pieces would be paid. In practice, it is much more likely with a short swing time that the bus would remain on the street, either laying over or deadheading to the starting point of the next trip. Assuming Block ͱ was then cut at ͭͯ:ͯͰ (the same as our previous runcut) it would look something like the following diagram—the black run. Block 1 2 3 4 5 6 7 We then follow the same process for Blocks ͮ and ͯ (the PM pieces of our two split runs). The cuts are made at around ͭͱ:ͬͬ, leaving the PM pieces of Blocks ͮ and ͯ, both of which start around ͭͮ:ͭͱ, to be matched with Blocks Ͳ and ͳ to create two multipiece straight runs. Again this is presented diagrammatically in the following fi gure. We have added two more multipiece straight runs, the white and black runs.

Level ͮ. Intermediate RuncuttingChapter 5. Runcutting 5-42 Block 1 2 3 4 5 6 7 This leaves us with three more straight runs to create—the AM pieces of Blocks ͮ and ͯ, and the PM piece of Block ͱ. When created our completed diagram looks like the following. Block 1 2 3 4 5 6 7 At this stage of course we have only created the runs conceptually and assumed a relief op- portunity will occur somewhere around the time required. Before we complete this runcut we need to optimize relief times based on the actual schedule. On the diagram below, we make a subtle change where the PM split piece on Block ͯ (white) is pushed forward a little earlier (to around ͭͰ:ͯͬ) to allow enough time for the relief to be made, and our straight white run enough time to make the relief, return to the garage, and begin Block ͳ. Again, with a short swing time like this, it is much more likely that the bus would remain on the street. On Block ͱ for example we assumed a possible relief around ͭͬ AM. Fortunately, looking at the schedule, we see that there is a relief opportunity at ͭͬ:ͬ͵. It pays to look at specifi c potential relief times as you create the conceptual runs in the diagram. Next we work through the details of specifi c relief times and create the runs summary. The fi nished product is presented below. Block 1 2 3 4 5 6 7 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21

Chapter 5. Runcutting 5-43 Level ͮ. Intermediate Runcutting We also made some changes to the conceptual runcut to minimize any overtime costs. If you compare with the previous diagram you can see we cut Block ͱ earlier (at ͭͯ:ͬ͵). Below is our updated runs summary, color-coded to match the conceptual run diagrams developed above. We also need to consider the operational impacts of the revised runcut. Note in particular that Run Ͳ has an issue. The cut at ͭͱ:ͬ͵ means an arrival back at the garage at ͭͱ:ͭ͵. The run is then supposed to operate Block ͳ, a pull out, at ͭͱ:ͮͬ. This leaves only ͭ minute and that assumes the ͭͱ:ͬ͵ relief is made on time and the ͭͬ-minute travel time back to the garage is achievable. Clearly this solution would create the potential for the ͭͱ:ͯͬ trip from Point A (the fi rst in-service trip on Block ͳ) to run late on a consistent basis. It is likely that the bus would remain on the street in this situation as well. Importantly we have now eliminated the split run of greater than ͭͯ hours—we had initially allowed this to illustrate the spread issues resulting from initial vehicle blocks. In reality this run, as shown in the initial solution, invalidated the entire run cut. Here we have found a more innovative approach and dealt with this problem successfully. We have also added some general complexity and potential service instability with the mul- tipiece runs and multiple reliefs. In general, simple is better for ease of operation. We need to be mindful as to whether the potential effi ciency gains warrant the potential service quality impacts. This is a question that the scheduler must continually be asking as solutions are devel- oped and is central to the production of good scheduling outcomes.

Level ͮ. Intermediate RuncuttingChapter 5. Runcutting 5-44 Completed Runcut - Pieces ReCut Time Place Type Time Place Type 1 5:31 5:46 Garage Pull 9:19 Garage Pull 5 10:09 A Street 13:09 A Street 13:34 2 Str 3 5:35 5:50 Garage Pull 12:39 A Street 13:04 6:49 0:30 0:10 7:29 7:29 0:31 0:00 0:00 8:00 1.174 3 Str 2 5:46 6:01 Garage Pull 12:09 A Street 12:34 6:08 0:30 0:10 6:48 6:48 1:12 0:00 0:00 8:00 1.304 4 5:50 6:05 Garage Pull 9:38 Garage Pull 9:43 3 13:44 14:09 A Street 18:19 Garage Pull 18:34 5 6:05 6:20 Garage Pull 10:09 A Street 10:24 2 14:44 15:09 A Street 18:38 Garage Pull 18:53 2 11:44 12:09 A Street 15:09 A Street 7 15:20 Garage Pull 19:19 Garage Pull 19:34 3 12:14 12:39 A Street 14:09 A Street 6 15:01 Garage Pull 19:49 Garage Pull 20:04 8 Str 5 12:44 13:09 A Street 19:53 Garage Pull 20:08 6:44 0:30 0:10 7:24 7:24 0:36 0:00 0:00 8:00 1.188 54:32 4:40 1:23 2:00 62:35 2:39 0:37 2:46 68:37 1.258 7:18 10:06 0:10 8:43 0:30 0:42 0:01 0:00 8:04 1.233 0:14 1:24 1:22 10:26 1 Str-M 6:33 0:30 Overtime 0:00 End Sign Off Time Plat Hours 0:00 Paid Hours Pay/ Plat 1.384 Spread Penalty Run # Type Block # Report Time Start 0:20 Spread Guarantee 0:20 8:03 8:03 Report Hours 7:43 0:50 12:44 Paid Break 0:40 Travel Hours Total Hours 8:007:50 7:50 4 7 Str-M 6:18 5 Spl 0:20 7:50 0:20 1.353Spl 0:00 0:21 8:28 12:480:50 0:10 0:00 0:006 Str-M 6:59 0:30 0:01 7:50 0:10 1.146 8:00 1.2700:00 0:00 The result of this revised runcut is basically no change in total paid hours (Ͳʹ:ͯͳ as opposed to Ͳʹ:ͮͳ in the previous cut). But we have reduced total operators required by one—in eff ect replacing two part time runs with one full-time split run. The benefi ts and costs of full-time and part-time runs will be discussed further in the next section. However, generally fewer runs mean reduced travels and report allowances. It may also reduce benefi t costs when part-time operators are given paid benefi ts by the agency. We have also reduced the make-up time from over fi ve hours to ͮ:ͯ͵ (make-up time is one of the least productive components of paid time). This is off set to some extent by the introduc- tion of ͭ:ͮͯ of paid break—the intervening time between two pieces of the multipiece straight runs. Overall we could consider this to be a good result with no paid hours increase but a manpower reduction. And we achieved this without any change to the blocks or scheduled trips. A further refi nement step could be undertaken to reconsider where the cuts were made with a view to off setting any guarantee and overtime costs.

Chapter 5. Runcutting 5-45 Level ͮ. Intermediate Runcutting 2. Rehook the Trips The above section provided an example of how a runcut can be created using multiple pieces. But there can be limitations on the ability to create such runs (many labor agreements prohibit multipiece straights), or the operational complexity may be considered undesirable. In this case we can throw out the philosophy that runs are cut from blocks. We can consider that runs are simply a combination of trips that comply with labor agreement rules and meet the objectives of the blocks. Many schedulers keep the blocks mostly intact out of habit. How- ever if runs can be created as a series of trips, and the resultant blocks meet the original block- ing objectives, then why take the constrained starting point of the existing blocks? In our example the opportunity presents itself to revisit the blocks and create driver runs. In this case we only need to do some additional cutting and rehooking of blocks and don’t pro- pose to start from scratch. So let’s take the starting point of where we left off with our basic runcut—unable to create the split runs due to spread issues. The blocks are set out below. Block 1 2 3 4 5 6 7 17 18 19 20 2111 12 13 14 15 165 6 7 8 9 10 A word of caution here. Any cutting and rehooking of blocks during the peaks runs the risk of increasing the number of vehicles required. This is not an option, under almost any circum- stance. There is no reason for the runcut to result in an increase in peak vehicle numbers, regardless of the number of vehicles in operation, the labor rules, or any other constraints. So as we rehook we simply avoid cutting any blocks during the peaks. Again we’re focused on the issue of making our split runs work eff ectively by hooking the start of Block Ͱ with the fi nish of Block ͯ and the start of Block ͱ with the end of Block ͮ The generic approach is to look at what happens with the blocks and link up around the edges of the peaks. Once the daytime off -peak pattern has been established (by ͭͬ AM in this case) there is little opportunity to rehook the blocks without aff ecting the number of vehicles being utilized.

Level ͮ. Intermediate RuncuttingChapter 5. Runcutting 5-46 We could simply take the pieces generated by our previous cutting process and turn those into our new blocks. Alternatively, we could revisit the linkup, particularly around the edges of the peaks, to generate blocks that meet our runcutting requirements. AM Blocks Example. Block Ͱ currently arrives at Point D at ͵:ͭʹ. As the service reduces from ͭͱ to ͯͬ minutes around this time there is no trip for it link to, and it pulls into the Garage at ͵.ͯʹ. Here we take the ͵:ͭʹ arrival and link it to the ͵:ͯͲ departure. The Block ͱ ͵:ͯͯ arrival (currently forms the ͵:ͯͲ departure) pulls into the garage instead. PM Blocks Example. In the PM we take the ͭͱ:ͬ͵ arrival on Block ͮ and link it to the ͭͱ:ͯͬ Block ͳ departure. This then leaves us with the remainder of Block ͮ, departing Point A at ͭͱ:ͭͱ and now requiring a pull out. The schedule at this time of day would then look like the follow- ing. Original - AM Old New Pull Out A B C D D C B A Next Trip Pull In 1 5:46 6:06 6:17 6:31 6:39 6:45 2 6:01 6:21 6:32 6:46 6:54 7:00 3 5:50 6:00 6:08 6:22 6:33 6:36 6:47 7:01 7:09 7:15 4 6:05 6:15 6:23 6:37 6:48 6:51 7:02 7:16 7:24 7:30 5 6:20 6:30 6:38 6:52 7:03 7:06 7:17 7:31 7:39 7:45 1 6:45 6:53 7:07 7:18 7:21 7:32 7:46 7:54 8:00 2 7:00 7:08 7:22 7:33 7:36 7:47 8:01 8:09 8:15 3 7:15 7:23 7:37 7:48 7:51 8:02 8:16 8:24 8:30 4 7:30 7:38 7:52 8:03 8:06 8:17 8:31 8:39 8:45 5 7:45 7:53 8:07 8:18 8:21 8:32 8:46 8:54 9:00 1 8:00 8:08 8:22 8:33 8:36 8:47 9:01 9:09 9:19 2 8:15 8:23 8:37 8:48 8:51 9:02 9:16 9:24 9:30 3 8:30 8:38 8:52 9:03 9:06 9:17 9:31 9:39 10:00 4 8:45 8:53 9:07 9:18 9:38 5 9:00 9:08 9:22 9:33 9:36 9:47 10:01 10:09 10:30 2 9:30 9:38 9:52 10:03 10:06 10:17 10:31 10:39 11:00 3 10:00 10:08 10:22 10:33 10:36 10:47 11:01 11:09 11:30 Revised - AM Old New Pull Out A B C D D C B A Next Trip Pull In 1 5:46 6:06 6:17 6:31 6:39 6:45 2 6:01 6:21 6:32 6:46 6:54 7:00 3 5:50 6:00 6:08 6:22 6:33 6:36 6:47 7:01 7:09 7:15 4 6:05 6:15 6:23 6:37 6:48 6:51 7:02 7:16 7:24 7:30 5 6:20 6:30 6:38 6:52 7:03 7:06 7:17 7:31 7:39 7:45 1 6:45 6:53 7:07 7:18 7:21 7:32 7:46 7:54 8:00 2 7:00 7:08 7:22 7:33 7:36 7:47 8:01 8:09 8:15 3 7:15 7:23 7:37 7:48 7:51 8:02 8:16 8:24 8:30 4 7:30 7:38 7:52 8:03 8:06 8:17 8:31 8:39 8:45 5 7:45 7:53 8:07 8:18 8:21 8:32 8:46 8:54 9:00 1 8:00 8:08 8:22 8:33 8:36 8:47 9:01 9:09 9:19 2 8:15 8:23 8:37 8:48 8:51 9:02 9:16 9:24 9:30 3 8:30 8:38 8:52 9:03 9:06 9:17 9:31 9:39 10:00 4 8:45 8:53 9:07 9:18 9:36 9:47 10:01 10:09 10:30 5 9:00 9:08 9:22 9:33 9:53 2 9:30 9:38 9:52 10:03 10:06 10:17 10:31 10:39 11:00 3 10:00 10:08 10:22 10:33 10:36 10:47 11:01 11:09 11:30 5 4 10:30 10:38 10:52 11:03 11:06 11:17 11:31 11:39 12:00 2 11:00 11:08 11:22 11:33 11:36 11:47 12:01 12:09 12:30 Block # Eastbound Westbound Block # Eastbound Westbound

Chapter 5. Runcutting 5-47 Level ͮ. Intermediate Runcutting Original - PM Old New Pull Out A B C D D C B A Next Trip Pull In 13:00 13:08 13:22 13:33 13:36 13:47 14:01 14:09 14:30 13:30 13:38 13:52 14:03 14:06 14:17 14:31 14:39 15:00 14:00 14:08 14:22 14:33 14:36 14:47 15:01 15:09 15:15 3 14:30 14:38 14:52 15:03 15:06 15:17 15:31 15:39 15:45 6 15:01 15:21 15:32 15:46 15:54 16:00 5 15:00 15:08 15:22 15:33 15:36 15:47 16:01 16:09 16:15 2 15:15 15:23 15:37 15:48 15:51 16:02 16:16 16:24 16:30 7 15:20 15:30 15:38 15:52 16:03 16:06 16:17 16:31 16:39 16:45 3 15:45 15:53 16:07 16:18 16:21 16:32 16:46 16:54 17:00 6 16:00 16:08 16:22 16:33 16:36 16:47 17:01 17:09 17:15 5 16:15 16:23 16:37 16:48 16:51 17:02 17:16 17:24 17:30 2 16:30 16:38 16:52 17:03 17:06 17:17 17:31 17:39 17:45 7 16:45 16:53 17:07 17:18 17:21 17:32 17:46 17:54 18:00 3 17:00 17:08 17:22 17:33 17:36 17:47 18:01 18:09 18:19 6 17:15 17:23 17:37 17:48 17:51 18:02 18:16 18:24 18:30 5 17:30 17:38 17:52 18:03 18:06 18:17 18:31 18:39 19:00 2 17:45 17:53 18:07 18:18 18:38 7 18:00 18:08 18:22 18:33 18:36 18:47 19:01 19:09 19:19 6 18:30 18:38 18:52 19:03 19:06 19:17 19:31 19:39 19:49 5 19:00 19:08 19:22 19:33 19:53 Revised - PM Old New Pull Out A B C D D C B A Next Trip Pull In 13:00 13:08 13:22 13:33 13:36 13:47 14:01 14:09 14:30 13:30 13:38 13:52 14:03 14:06 14:17 14:31 14:39 15:00 14:00 14:08 14:22 14:33 14:36 14:47 15:01 15:09 15:15 3 14:30 14:38 14:52 15:03 15:06 15:17 15:31 15:39 15:45 6 15:01 15:21 15:32 15:46 15:54 16:00 5 15:00 15:08 15:22 15:33 15:36 15:47 16:01 16:09 16:15 2 7 15:05 15:15 15:23 15:37 15:48 15:51 16:02 16:16 16:24 16:30 7 2 15:20 15:30 15:38 15:52 16:03 16:06 16:17 16:31 16:39 16:45 3 15:45 15:53 16:07 16:18 16:21 16:32 16:46 16:54 17:00 6 16:00 16:08 16:22 16:33 16:36 16:47 17:01 17:09 17:15 5 16:15 16:23 16:37 16:48 16:51 17:02 17:16 17:24 17:30 2 7 16:30 16:38 16:52 17:03 17:06 17:17 17:31 17:39 17:45 7 2 16:45 16:53 17:07 17:18 17:21 17:32 17:46 17:54 18:00 3 17:00 17:08 17:22 17:33 17:36 17:47 18:01 18:09 18:19 6 17:15 17:23 17:37 17:48 17:51 18:02 18:16 18:24 18:30 5 17:30 17:38 17:52 18:03 18:06 18:17 18:31 18:39 19:00 2 7 17:45 17:53 18:07 18:18 18:38 Block # Block # Eastbound Westbound Eastbound Westbound The blocks are then updated in the diagram to show the new Blocks ͮ and ͳ (the PM change) and Blocks Ͱ and ͱ (the AM change). The end result is two peak-only blocks (Blocks ͱ and ͳ) that can form a split run within the spread requirements. Block 1 2 3 4 5 6 7 17 18 19 20 2111 12 13 14 15 165 6 7 8 9 10

Level ͮ. Intermediate RuncuttingChapter 5. Runcutting 5-48 We then repeat the process to generate two more blocks (based on the start of Block Ͱ and the end of Block ͯ) that will allow another split run to be created from peak blocks. The process is the same. Looking at the above diagram we simply want to have Block Ͱ fi nish after the AM Peak and have Block ͭ operate all day. In the PM Peak we want to have Block ͯ (the earliest fi nish) be a standalone PM Block, with Block Ͳ taking the rest of Block ͯ’s day. One must be extremely careful during this process to keep track of the block numbers and changes through what is an iterative process. After making the changes as described above, the run guide is updated, and the blocks and runs are renumbered. The fi gures below provide the outputs of the process with a new runcut solution. The solution is provided including the fi nalized headway sheet (with revised block numbers and pull ins/pill outs), blocking diagram, and run guide.

Chapter 5. Runcutting 5-49 Level Í®. Intermediate Runcutting Revised blocks for RunCut ROUTE 97 Broad Street DAY Weekday Block # Pull Out A B C D D C B A Next Trip Pull In 1 5:46 6:06 6:17 6:31 6:39 6:45 3 6:01 6:21 6:32 6:46 6:54 7:00 2 5:50 6:00 6:08 6:22 6:33 6:36 6:47 7:01 7:09 7:15 4 6:05 6:15 6:23 6:37 6:48 6:51 7:02 7:16 7:24 7:30 5 6:20 6:30 6:38 6:52 7:03 7:06 7:17 7:31 7:39 7:45 1 6:45 6:53 7:07 7:18 7:21 7:32 7:46 7:54 8:00 3 7:00 7:08 7:22 7:33 7:36 7:47 8:01 8:09 8:15 2 7:15 7:23 7:37 7:48 7:51 8:02 8:16 8:24 8:30 4 7:30 7:38 7:52 8:03 8:06 8:17 8:31 8:39 8:45 5 7:45 7:53 8:07 8:18 8:21 8:32 8:46 8:54 9:00 1 8:00 8:08 8:22 8:33 8:36 8:47 9:01 9:09 10:30 3 8:15 8:23 8:37 8:48 8:51 9:02 9:16 9:24 9:30 2 8:30 8:38 8:52 9:03 9:06 9:17 9:31 9:39 10:00 4 8:45 8:53 9:07 9:18 9:36 9:47 10:01 10:09 10:19 5 9:00 9:08 9:22 9:33 9:53 3 9:30 9:38 9:52 10:03 10:06 10:17 10:31 10:39 11:00 2 10:00 10:08 10:22 10:33 10:36 10:47 11:01 11:09 11:30 1 10:30 10:38 10:52 11:03 11:06 11:17 11:31 11:39 12:00 3 11:00 11:08 11:22 11:33 11:36 11:47 12:01 12:09 12:30 2 11:30 11:38 11:52 12:03 12:06 12:17 12:31 12:39 13:00 1 12:00 12:08 12:22 12:33 12:36 12:47 13:01 13:09 13:30 3 12:30 12:38 12:52 13:03 13:06 13:17 13:31 13:39 14:00 2 13:00 13:08 13:22 13:33 13:36 13:47 14:01 14:09 14:30 1 13:30 13:38 13:52 14:03 14:06 14:17 14:31 14:39 15:00 3 14:00 14:08 14:22 14:33 14:36 14:47 15:01 15:09 15:30 6 14:46 15:06 15:17 15:31 15:39 15:45 2 14:30 14:38 14:52 15:03 15:21 15:32 15:46 15:54 16:00 1 15:00 15:08 15:22 15:33 15:36 15:47 16:01 16:09 16:15 7 15:05 15:15 15:23 15:37 15:48 15:51 16:02 16:16 16:24 16:30 3 15:30 15:38 15:52 16:03 16:06 16:17 16:31 16:39 16:45 6 15:45 15:53 16:07 16:18 16:21 16:32 16:46 16:54 17:00 2 16:00 16:08 16:22 16:33 16:36 16:47 17:01 17:09 17:15 1 16:15 16:23 16:37 16:48 16:51 17:02 17:16 17:24 17:30 7 16:30 16:38 16:52 17:03 17:06 17:17 17:31 17:39 17:45 3 16:45 16:53 17:07 17:18 17:21 17:32 17:46 17:54 18:00 6 17:00 17:08 17:22 17:33 17:36 17:47 18:01 18:09 18:19 2 17:15 17:23 17:37 17:48 17:51 18:02 18:16 18:24 18:30 1 17:30 17:38 17:52 18:03 18:06 18:17 18:31 18:39 19:00 7 17:45 17:53 18:07 18:18 18:38 3 18:00 18:08 18:22 18:33 18:36 18:47 19:01 19:09 19:19 2 18:30 18:38 18:52 19:03 19:06 19:17 19:31 19:39 19:49 1 19:00 19:08 19:22 19:33 19:53 Eastbound Westbound

Level ͮ. Intermediate RuncuttingChapter 5. Runcutting 5-50 Block 1 2 3 4 5 6 7 17 18 19 20 211 12 13 14 15 165 6 7 8 9 10 Completed Runcut - Blocks Re-Hooked Time Place Type Time Place Type 101 Str 1 5:31 5:46 Garage Pull 13:09 A Street 13:34 7:23 0:30 0:10 8:03 8:03 0:00 0:01:30 0:00:00 8:04:30 1.094 102 Str 1 12:44 13:09 A Street 19:53 Garage Pull 20:08 6:44 0:30 0:10 7:24 7:24 0:36 0:00:00 0:00:00 8:00:00 1.188 103 Str 2 5:35 5:50 Garage Pull 12:39 A Street 13:04 6:49 0:30 0:10 7:29 7:29 0:31 0:00:00 0:00:00 8:00:00 1.174 104 Str 2 12:14 12:39 A Street 19:49 Garage Pull 20:04 7:10 0:30 0:10 7:50 7:50 0:10 0:00:00 0:00:00 8:00:00 1.116 105 Str 3 5:46 6:01 Garage Pull 12:09 A Street 12:34 6:08 0:30 0:10 6:48 6:48 1:12 0:00:00 0:00:00 8:00:00 1.304 106 Str 3 11:44 12:09 A Street 19:19 Garage Pull 19:34 7:10 0:30 0:10 7:50 7:50 0:10 0:00:00 0:00:00 8:00:00 1.116 4 5:50 6:05 Garage Pull 10:19 Garage Pull 10:24 6 14:31 14:46 Garage Pull 18:19 Garage Pull 18:34 5 6:05 6:20 Garage Pull 9:53 Garage Pull 9:58 7 14:50 15:05 Garage Pull 18:38 Garage Pull 18:53 56:17 4:40 0:00 1:00 61:57 2:43 0:20 2:46 67:46 1.204 Start End Spread Penalty Paid Hours Run # Type Block # Report Time Spread Make-up Pay/ Plat Sign Off Time Plat Hours Report Hours Paid Break Travel Hours Total Work Hours Overtime 0:18:30107 Spl 7:47 0:50 0:00 8:37 12:44 0:00 1:22:00 10:17:30 1.322 108 Spl 7:06 0:50 0:00 1.3241:24:00 9:24:007:56 12:48 0:04 0:00:00 In terms of effi ciencies this solution is the lowest cost of the three developed thus far—a total of Ͳͳ:ͰͲ paid time compared to Ͳʹ:ͮͳ and Ͳʹ:ͯͳ for the previous two solutions. It retains the six straight runs and four split runs of the previous solution, ensuring manpower requirements are minimized. Operationally the outcome is a completed set of legal runs, with the two full-time splits, and with no multipiece or tight connection issues. In fact Block ͭ could run an additional round trip departing Point A at ͵:ͭͱ and Point D at ͵:ͱͭ—thus keeping the ͭͱ-minute frequency out for an additional ͯͬ minutes with no labor cost implications. This is a classic example of benefi ts obtained when the runcut and blocking interaction is iterative in nature.

Chapter 5. Runcutting 5-51 Level ͮ. Intermediate Runcutting Before this (or any) solution is completed we must go back and recreate the car “runs,” ensur- ing that we know when the car(s) are used, who will be driving them, and when they arrive and depart the garage. The above example allowed us to work through some reasonably straightforward examples of how revisiting the blocks can assist in producing a better runcut solution. But there are nu- merous other situations where revisiting the blocks can enhance the runcut solution. Some of these are discussed below. Rehooking a School Extra In some transit systems school trips are blocked separately. The reasons for this are numer- ous, and for now we simply accept that our blocks may have such occurrences. Usually this will be evident as soon as we look at our block graph. The example below will show, again without needing to look at numbers or details, a block that could potentially be rehooked during the runcut. Block 1 2 3 4 5 6 7 8 17 18 19 20 2111 12 13 14 15 BLOCKS 7 AND 8 165 6 7 8 9 10 Here we have added a new block (block ʹ) to cover a short school trip. It then pulls back into the garage. In this example the block has only ͭͱ minutes of revenue time but Ͱͱ minutes of platform time—one short school trip with a garage pull either side. By rehooking the blocks to add this trip to the front of Block ͳ we could potentially create a more effi cient run and even reduce total platform time. The reduction would be achieved by removing the pull in on Block ʹ and pull out on Block ͳ, and replacing them with a direct deadhead between the end of Block ʹ and the start of Block ͳ, and possibly some layover. Of course this would only be achievable if there is enough time for the bus to get from where the school tripper ends to the starting point of Block ͳ. However if this achievable we are likely to produce a better runcut solution, and deal eff ectively with a block that would otherwise be diffi cult to assign. But should we do this? It could depend on several factors. Is it agency policy to keep school trippers separate? Sometimes school trips are off ered as additional overtime and are intentionally kept separate. 1. Exercise caution when hooking school trips to regular daily runs. School schedules are often variable with schedule changes that may not be compatible with run assignments. Tip

Level ͮ. Intermediate RuncuttingChapter 5. Runcutting 5-52 Often school trips are kept separate to deal with changing bell times—for example, if the school gets out ͯͬ minutes later on one day per week our nice rehook suddenly does not work on that day. You will need to consider such questions and understand how your agency approaches such issues. If the runcutter is not the same person who develops schedules and blocks at your agency, it is important to maintain good communication between all to understand the intent behind how the blocks were created. As with many aspects of scheduling the decision includes some form of value judgment and needs to be informed by as much relevant information as possible. If in the end such changes are made, the blocking sheets, run guides, and any other documents need to be updated accordingly. A Recap In working through these examples we have actually dealt with many of the important con- cepts relating to runcutting. Let’s list a few below: Types of runs. The basic run types, which are most suitable in specifi c cases, and some of the impacts of each. Relief types. The types of reliefs that can be used and how to create runs with those relief types. Cutting pieces. How we decide when and where to cut pieces that will be used to form runs. Matching pieces. How we match pieces (in our examples, of split runs) to create full runs. Shifting pieces. Changing which relief opportunities we used to cut pieces to create the most effi cient and eff ective overall solution. Reliefs. How to choose which relief opportunities to cut pieces. Revisiting blocks. Some instances where we may want to rehook some blocks to achieve better runcut outcomes. Improving runcut effi ciencies. Looking at combinations of the above methods to re- duce the overall runcut cost. Congratulations, you’re now doing most of the things that sophisticated computerized sys- tems attempt to mimic! Initially we presented runcutting as a complex mathematical problem with myriad inputs, requirements, preferences, and outcomes. However in working through the above examples we have actually considered much of what is required to produce even more complex runcuts. 2. • • • • • • • •

Chapter 5. Runcutting 5-53 Level ͮ. Intermediate Runcutting LEVEL 2 End of Intermediate Runcutting. The Advanced Section of Runcutting continues on the next page. To jump to Rostering, go to page Ͳ-ͭ.

Level ͯ. Advanced RuncuttingChapter 5. Runcutting 5-54 5.3 Advanced Runcutting In the previous sections we have developed runcut solutions for a scenario con- taining a relatively basic schedule, simple work rules, and minimal constraints. The Line ͵ͳ example is of course simple in that it involves one route, one garage, and one relief location (using one type of relief). In those examples, the work rules used were limited to a basic set of hard rules, with no implied limits based on preferences or unwritten practices. The reality of scheduling is that many situations presented to the scheduler are more complex than this, with multiples in many of these categories—mul- tiple routes, garages, relief locations, and types, for example. However the Line ͵ͳ schedule presents many constraints that often do not exist in many run- cuts. The scale or size of the runcut is a signifi cant factor in the diffi culty of creating a solution. Cutting just a single run often results in a less optimal solution than cutting an entire system. However, contrary to many expectations, this is often an inversely proportional relationship, i.e., the larger the runcut the easier it is to fi nd a solution! Looking back at any of the block diagrams for the Line ͵ͳ runcuts we see limited options to cut pieces. For example there were only two shorter peak blocks in each peak, and those blocks did not allow us to form runs within the ͭͯ-hour spread limitation. In another situation, with Ͱͬ or ͱͬ vehicles we might have had a range of shorter blocks to choose appropriate starts and fi nishes to match pieces into legal split runs. For this reason there is less need to work through further examples of larger problems. Just simply apply the same approaches, philosophies, and tools to dealing with larger runcuts. In this section we will look less at the mechanics of the runcuts (although there will be numerous examples), and consider more of the discussion around runcut approaches, and how more dif- fi cult issues can be resolved. The later parts of this section revolve around a discussion of some key policy issues relating to runcuts. Larger or more complex problems can add the following elements to the examples we have worked through to this point: Volume of trips, blocks, and routes If runs are cut at a route level (i.e., each run only operates on one route) the approach to cut- ting a route with more service or longer service span is exactly the same as in the previous examples. There will of course be increased options, more runs, additional pieces, etc., but the approach remains the same. LEVEL 3

Chapter 5. Runcutting 5-55 Level ͯ. Advanced Runcutting Additional Garages Creating runs for multiple-garage solutions adds complexity to the scheduler’s task, both at a blocking and runcutting level. Where there are multiple garages, another level of consideration is required: At which garage should the piece start/endt? Can the block be adjusted to change garages? What are the travel times between the relief locations and each garage? Can meals occur at a diff erent garage than the “home” garage for a run? Are compatible bus types assigned to each garage? While these issues do add complexity to the runcut process, it tends to be more an issue of in- creased options rather than placing constraints on the scheduler. However in considering those options the impact of optimizing start/end locations is critical to minimizing overall deadhead- ing or travelling costs. Two key considerations are: Vehicles to be balanced at the start and end of the day, i.e., the same number of ve- hicles must pull out of a garage that pull in at the end of the day Generally operators must start and end their work day at the same garage These two factors must be taken into account, along with the considerations noted above, when undertaking multi-garage runcutting. Additional Relief Types and Locations Our Line ͵ͳ example had one relief location and one method of traveling to that location. In many cases routes will have multiple relief locations and possibly other methods of travel. Additional relief opportunities actually only help to make the problem more readily solvable— through presenting the scheduler with more options for cutting pieces. The extension of concepts for relief types is discussed in more depth later in this section. More Complex Work Rules Work rules present the most diffi cult and constraining factor in runcutting and also represent the most variability from one operator to another. Earlier sections of this manual have already discussed concepts of hard rules, soft rules, constraints, and practices. At times these can con- tradict each other, or be diffi cult to understand or explain. • • • • • • •

Level ͯ. Advanced RuncuttingChapter 5. Runcutting 5-56 Take the work rules relating to meal breaks for one transit system: At least one of the following constraints must be met in straight runs over ʹ hours: ͭ x ͮͬ minute break and ͭ x ͭͬ minute break; or ͯ x ͭͱ minute breaks; or Ͳͬ minutes total layover (ͭͱ% of runs maximum) The agreement states that not only must one of these be scheduled in each run, but also that ʹͬ% must be achieved in actual operation. In actuality, virtually all scheduled break require- ments are expected to be achieved in operations. Runs are further limited by the requirement that no more than ͮͯ% of runs can involve a ve- hicle change. This limits options to create multipiece runs to deal with the breaks. Just reading and understanding these interrelated work rules is a complex task. And these con- straints apply to only a small part of the overall work rules at the agency (i.e., the meal breaks). The limitations applied to scheduling processes by more constrained work rules or practices present the biggest challenge to the scheduler. This manual cannot even begin to attempt to explain myriad work rules and practices adopted or applied by transit systems. We have noted some of the “typical” kinds of rules applied and discussed how to create runs with these rules in mind. We would simply recommend that strong consideration is given to work rules when schedul- ing, and that the rules be appropriately applied. By “appropriate” here, we mean as intended by the agency to fi t within the overall agency’s objectives and policies. Using a computerized system actually provides some good checks and balances in this respect, as the scheduler must actually write down the rules, constraints, limits, and preferences. These must be succinctly defi ned and expressly stated for the system to be able to create legal runs. When using other scheduling techniques, this approach should be mimicked—write down the specifi c rules, not in the words of the labor agreement, but as input into runcutting. Summarize these rules or practices and circulate them to gain wider approval. On the other hand, the computer is not as good at making nuanced judgments as a human scheduler can be. Only an expert scheduler knows which rules can be “bent” and how often— the computer will “do as you say” but will not make the complex trade-off s and judgments that an experienced scheduler can make. This is why, even at properties that utilize up-to-date computerized scheduling packages, an experienced scheduler should always review the end result and see if further optimization could take place with a few reasonable adjustments. • • •

Chapter 5. Runcutting 5-57 Level ͯ. Advanced Runcutting Runcutting Multiple Routes The number of variables increases dramatically when cutting multiple routes. The classic ex- ample of multiple-route scheduling involves hooking together peak blocks to form split runs. In our original Line ͵ͳ example, we had four standalone peak blocks that had a spread too long to fi t within spread limits, or at least to avoid incurring high spread premiums. If we add a second route into the runcutting mix we can see some potential for immediate ben- efi ts, as per the diagram below: Line 97 Blocks Block 1 Run 1a 2 3 4 Run 2a 5 6 Run 2b 7 Run 1b Line 98 Blocks Block 1 2 Run 3a 3 Run 4a 4 Run 4b 5 Run 3b 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 On Line ͵ʹ we have a much shorter peak, with ͳ:ͬͬ and ͳ:ͯͬ starts, and two ͭʹ:ͬͬ fi nishes. If Line ͵ʹ was blocked standalone we would have two split runs (the white Runs ͯ and Ͱ) with Ͳ:Ͱͱ and ͳ:ͭͱ report times, and ͭʹ:ͭͱ sign off times. At the same time we have split runs on Line ͵ͳ that are either illegal (spread too long) or have long spreads with high premium paid. If we matched the late starts on Line ͵ʹ with the late fi nishes on Line ͵ͳ, and the early starts on Line ͵ͳ with the early fi nishes on Line ͵ʹ, we get four legal split runs with a reduction in total spread premium. The following diagram represents the newly created multiroute split runs.

Level ͯ. Advanced RuncuttingChapter 5. Runcutting 5-58 Line 97 Blocks Block 1 Run 1a 2 3 4 Run 2a 5 6 Run 4b 7 Run 3b Line 98 Blocks Block 1 2 Run 3a 3 Run 4a 4 Run 1b 5 Run 2b 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 This is a simple example where the pieces matched easily. In practice such outcomes are not always the case. However the concept and approach remains the same—that the introduction of multiple routes allows the potential for increased options and better runcuts. As with larger problems, the issue is one of size, and the more options present more potential solutions. Trips, Blocks, and Runs In our earlier examples, we learned that it is sometimes possible to make small changes to blocks to improve the runcut solution. But in those examples, the changes made were only minor, aff ecting only two or three blocks. Let’s now go back to a statement made in a preceding section—that the runs are, in eff ect, a group of trips that form an operator’s work day. Often schedulers limit the ability to produce eff ective runcuts by assuming the blocks are fi xed, or by limiting changes Note here that to some extent we are playing with fi re. The primary reason to begin runcutting with a set of blocks is that blocking is designed to minimize the number of vehicles required to operate a given schedule. By essentially throwing out the blocks and starting with a list of trips, we are optimizing for driver hours, but not necessarily for peak vehicles. Peak vehicle availability is often fi xed at a transit operator and may be an even more rigid limitation than the number of driver hours.

Chapter 5. Runcutting 5-59 Level ͯ. Advanced Runcutting Runcutting Multipiece Runs Many transit systems avoid multipiece runs and many labor agreements forbid or constrain them. The reason appears to be primarily one of simplicity and control. A set of runs comprised of long blocks cut into straights, and peak blocks matched to splits (or assigned to part-time runs) is a simple and coherent runcut option. However as we have discussed and seen in our examples, the limitations of relief opportunities make it diffi cult to cut straight runs, and/or match pieces into split runs, in an optimized man- ner. We can go back and reconsider our defi nition of “optimized” here. Our Line ͵ͳ example was possibly “optimal” with the constraints placed on it. But was it the “optimal” solution in terms of minimized paid hours for the given set of trips? The answer is almost certainly not. A primary aim of creating more and smaller pieces is to be able to combine and match them into runs that better improve the runcut solution. This can potentially result in reduced overall cost and a lower pay-to-platform ratio. The pieces used in multipiece runs can be cut at the same relief location, or perhaps at diff erent locations with a travel in between. The aim is to create pieces of work that can be best matched together across the whole runcut. In the previous section we did this to improve our split runs and then rehooked the blocks to avoid the “multipiece run” name. But in eff ect that is what happened—we created runs with multiple pieces to improve the runcut solution. We are treading here on some prevailing scheduling theory in that multipiece runs are often frowned upon as “bad” or “wasteful.” This can be true for one or two runs, particularly if there is a paid break or long travel between the pieces. However in the context of the total runcut the creation of some smaller pieces can provide signifi cantly enhanced overall effi ciencies. So how do multipiece runs work? The concept is simple—cut smaller pieces and match them together to form runs, in much the same way that we matched pieces together for split runs in earlier examples. But in this case we are trying to match pieces with smaller breaks (so as not to incur paid break time), while meeting other runcut objectives. The advent of computerized scheduling makes it possible to test diff erent types of runcutting solutions. These systems allow run types such as multipiece runs (of any sort—straight, split, part-time) to be tested and reviewed, within an overall optimized runcut. One reason multi- piece runs are not always embraced is the sheer complexity they cause for schedulers. If mul- tiple piece runs are allowed, the potential number of pieces and cuts increases dramatically. Under no circumstance should the runcut result in increased peak vehicle requirements over the blocking solution. Tip multipiece runs Runs made up of pieces from multiple blocks. In most cases, split runs are inherently multipiece runs. But either half of a split run could itself could have multiple pieces, cut from multiple blocks.

Level ͯ. Advanced RuncuttingChapter 5. Runcutting 5-60 Pull Reliefs and Multipiece Runs We can go back to our example in the intermediate section (“Rehook the Trips”) to see multi- piece runs with pull reliefs which were then rehooked to provide updated blocks. A key feature of this example was that the blocks were rehooked outside the peak periods. This is a very important point. Allowing runcutting processes to rehook trips during the peaks (when peak vehicle requirements are defi ned) can result in increased vehicle requirements. Outside of the peak, the use of pull reliefs can increase vehicle needs with little consequence (an exception being if you have a very low peak-to-base ratio, in which case off -peak vehicle increases can aff ect other operational requirements such as availability of vehicles for maintenance). When using computerized systems, this requirement must be carefully monitored, as the runcut modules at times fail to consider the cost of peak vehicles in their cost optimization process. The potential costing impacts of pull reliefs need to be carefully considered by the scheduler before any widespread application. Greater use of pull reliefs is almost guaranteed to increase vehicle mileage, often at the expense of cheaper car mileage. As noted, the evaluation of use of diff erent relief types can be undertaken quickly and eff ectively through use of computerized scheduling systems. If multipiece runs have been created, using pull reliefs, it usually is a simple matter of recreat- ing the blocks after the runs have been created. Again, be careful not to overlook some consid- erations taken into account by the scheduler when blocking such as: Congestion/capacity at terminals; Vehicle type limitations; Reliability on certain routes; Interlining limitations; and Maintaining adequate recovery time. Runcutting for Meals So far, our examples have included only three types of run—straight runs, split runs, and part- time runs. Our straight runs have been one-piece runs, and the splits have had a break back at the garage. Many agencies now have labor rules mandating some form of operator breaks. In many parts of the United States new rules mandating meal and rest break requirements are being devel- oped and implemented. Rules relating to meal breaks can diff er signifi cantly in complexity and in how they aff ect effi ciencies. These may change from one transit agency to another. • • • • •

Chapter 5. Runcutting 5-61 Level ͯ. Advanced Runcutting The types of breaks required may include: Minimum layover requirements. Sometimes expressed as a percentage of layover time during a work day, or at least one layover of x minutes. Stipulated meal breaks. Meal breaks come in many forms with varying requirements. The breaks may be paid or unpaid, taken back at the garage or at relief locations (or in-vehicle), may have minimum/maximum lengths (above the maximum length the run typically becomes a split) , and may be required to be spaced at certain times during the work day. Other rest breaks. These may be required in addition to scheduled meal breaks (such as one ͭͬ-minute layover every two hours). Many agencies view meal break requirements as immediately resulting in cost increases. However, as we will see as we explore some of the detail, this is not necessarily the case. In many parts of the world transit systems operate with meal break requirements while creating extremely effi cient run cuts. Depending upon the combination of rule requirements, the scheduling of meal breaks can make the runcutting task signifi cantly more complex. If a piece of work can only be fi ve hours long, for example, it means that all straight runs must have at least two pieces. This doubles the number of reliefs that must be considered, exponentially increasing the mathematical complexity of the problem to be solved. In the case of minimum layover requirements, it may be possible to build the blocks with these already built in. In our Line ͵ͳ example we built the schedule with ͮͭ minutes of layover at Point A. When we hooked our trips into blocks, we considered that breaks might be required. If we go back and review our runs in detail we will see that the straight runs (those that are most likely to require meal or rest breaks) have a ͮͬ-minute break every ͵ͬ minutes. In this case our thinking ahead allowed the requirement to be met. But what happens when there is a more complex or restrictive rule—a typical meal break requirement such as “the operator must have a meal break of between ͯͬ and Ͳͬ minutes, be- tween the second and fi fth hour”? The break may be paid or unpaid; in this case we will assume it is unpaid. If we assume everything else remains unchanged we can go back and rebuild our runs. At this stage we can consider two potential approaches to runcutting to meet the meal break requirements. The fi rst involves cutting longer pieces, across multiple vehicles, and assigning operators two pieces of work in a run. The second involves operator drop-backs at terminals. • • • Many agencies struggle with providing meal breaks, due to the increased complexity and potential constraints. However, meal breaks, if the right labor rules exist, need not add signifi cantly to the cost of operating service. This holds true in particular where meal breaks are unpaid and can be taken at the terminal rather than returning to the garage. Tip

Level ͯ. Advanced RuncuttingChapter 5. Runcutting 5-62 Approach 1: Recutting the Runs This approach to cutting the runs changes the mechanics of the process signifi cantly, since we now must have run “portions” (the work either side of a meal break) no longer than fi ve hours. And those pieces must be less than fi ve hours, given that we will possibly need to add travel time and report time to the actual platform time of the piece. In this case we’ll go back and use our original Line ͵ͳ blocks and assume all street reliefs at Point A once again. We will assume that meal break must be taken back at the garage, requir- ing travels to and from the depot. We will also assume the meal break will also be unpaid. As we will discuss later, these are two key factors in the cost impacts of meal breaks. In this exam- ple we have taken a less restrictive view of one factor (paid meals) and a more restrictive view as to where meal breaks can be taken (in this case at the garage and not at trip terminals). Where do we start? As always the scheduler should make some macro observations about the number of runs (by type), the nature of the blocks, and potential pieces. We know from the previous examples that we had eight full-time runs. However with ͱͰ hours of platform time we could be looking to create as few as seven runs. And the runcuts produced had almost three hours of guarantee due to the runs being generally shorter than we would have preferred. We also need to think about how many pieces we will cut—we will assume all runs are now two pieces for a straight or split, and one piece for a part-time run. If we are looking for eight-hour runs, and a piece cannot be more than fi ve hours, that means we’d ideally like all pieces to be between three and fi ve hours. Before starting to create pieces or runs, we should review the blocks and consider how often we will need to cut. This is shown simply below. Block Length Pieces 1 3:33 1 2 12:37 3 3 12:29 3 4 3:33 1 5 13:33 3 6 4:48 1 7 3:59 1 Total 54:31 13 Average 4:11

Chapter 5. Runcutting 5-63 Level ͯ. Advanced Runcutting The table gives us a very rough idea of the number of pieces we are aiming for and the average length of each piece. The actual number of work pieces and run length will be determined in large part by the opportunities we have for reliefs. The number of pieces will also be aff ected by the need to combine pieces into straight runs. The odd number of pieces suggests we may end up with either a part-time run or a longer straight run with more than two pieces (and pos- sibly two meals). But for now it is enough to know how many pieces and runs we are aiming for. So the ͭͯ-piece indication is only an approximation at this point and could change. A general rule of thumb is to begin with the early start times as candidates for straight runs, leaving the splits to be the later starts and earlier fi nishes. This minimizes spread premium costs and is a generally accepted preference for operators. If we review our blocks again this means that we are looking for the starts of blocks ͭ, ͮ, and ͯ to be the fi rst piece of our straight runs. In the PM we want our early fi nishes to be the splits and later fi nishes to be PM Straights. In eff ect we are ensuring our shorter splits, with starts at Ͳ:ͬͱ (Block Ͱ) and Ͳ:ͮͬ (Block ͱ), and fi nishes at ͭʹ:ͭ͵ (Block ͯ) and ͭʹ:ͯʹ (Block ͮ). This is the same matching process we undertook when rehooking blocks to arrive at our preferred splits, with minimized spread premium, in the earlier example. In eff ect we have already created the split pieces. We just need to know when to fi nish the two AM pieces and start the two PM pieces. The fi nishing time of Block Ͱ is mandated by the fact that it is a smaller peak-only block. So we assume Block Ͱ is the fi rst half of one of our split runs. To minimize our spreads we want to match this with the piece on Block ͯ ending at ͭʹ:ͭ͵. Assuming we are looking for around an ʹ-hour run, this dictates a cut at before ͭͰ:ͯͬ. A process of cutting and matching is then undertaken to create and match pieces to form legal and effi cient runs. Again we turn to a graphical representation of the runs to provide an over- As a starting point, look to the blocks with the earliest starts as opportunities for straight runs, and blocks with later starts and early fi nishes as possibili- ties for split runs. Tip

Level ͯ. Advanced RuncuttingChapter 5. Runcutting 5-64 view of how the blocks have been cut into pieces. Pieces with the same run number (e.g., run ͱ is pieces of Blocks ͯ and Ͳ) form a Straight Run (Run Ͳ—see Run guide below). Block 1 Run 1 2 Run 3a Run 2b Run 3b Run 5b 3 Run 2a Run 6a Run 4b 4 Run 4a 5 Run 5a Run 7a Run 8 6 Run 6b 7 Run 7b 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 As the diagram depicts, the additional pieces required by creating runs with meal breaks adds signifi cant complexity. The process requires consideration of numerous combinations of poten- tial pieces. However some basic rules can be applied: Look at the start and end pieces initially to provide an idea of the pieces and runs. Work- ing forward from the block starts (to a maximum of ͱ hours) and backwards from block ends will provide an indication of how many of each type of run will be required. Avoid cutting for meals during the peaks where possible, since this in eff ect means two operators are covering the one peak. Cut the AM straight pieces as the early starts. PM straight pieces should generally be as long as possible to ensure operators can cover both the PM peak and late night services. This of course depends on how late the ser- vice runs. In this case the night service is minimal. Split pieces should be cut with spread limitations in mind. The number of straight runs will be a function of the level of early, midday, and night service levels. With only three midday blocks in our example, and no night service, the opportunities to cut straight runs are reduced. Now let’s move on to the solution. The run guide is again provided, this time with some addi- tional information to explain the straight runs. There are eight total runs. Of these, we have four straights, two splits, and two part-time. Note the total paid time has increased from Ͳͳ:ͰͲ in our previous solution to Ͳ͵:ͬͲ. However this is probably a good outcome given the additional travel time required to have all straight runs return to the depot for meals (in eff ect adding an additional ͭ:ͮͬ of time into the runs). • • • • • •

Chapter 5. Runcutting 5-65 Level ͯ. Advanced Runcutting Meal Breaks - Completed Run Cut Time Place Type Time Place Type 1 Pt 1 5:31 5:46 Garage Pull 9:19 Garage Pull 9:34 4:03 3:33 0:30 0:00 4:03 4:03 0:00 0:00 0:00 4:03 1.141 3 5:35 5:50 Garage Pull 9:39 A Street 9:49 4:14 2 10:29 10:39 A Street 12:09 A Street 12:34 2:05 2 5:46 6:01 Garage Pull 10:39 A Street 10:49 5:03 2 11:59 12:09 A Street 15:09 A Street 15:34 3:35 4 5:50 6:05 Garage Pull 9:38 Garage Pull 9:43 3:53 3 13:44 14:09 A Street 18:19 Garage Pull 18:34 4:50 5 6:05 6:20 Garage Pull 11:39 A Street 11:54 5:49 2 14:44 15:09 A Street 18:38 Garage Pull 18:53 4:09 3 9:14 9:39 A Street 14:09 A Street 14:19 5:05 6 15:01 15:01 Garage Pull 19:49 Garage Pull 20:04 5:03 5 11:14 11:39 A Street 14:39 A Street 14:49 3:35 7 15:20 15:20 Garage Pull 19:19 Garage Pull 19:34 4:14 8 Pt 5 14:14 14:39 A Street 19:53 Garage Pull 20:08 5:54 5:14 0:30 0:10 5:54 5:54 0:00 0:00 0:00 5:54 1.127 54:32 4:40 2:20 0:10 61:42 1:41 2:48 2:46 68:57 1.265 11:12 1.204 Piece Hours Paid Hours Pay/ Plat 6 Str-p 9:18 0:30 12:21 1.403 1:04 0:00 5 Spl 8:48 0:50 0:20 9:58 12:48 0:00 1.119 0:00 0:59 1:24 8:43 12:44 10:26 1.353 7:490:00 4:01 2:50 0:42 0:31 0:20 10:08 10:50 0:00 0:40 6:59 0:00 8:00 1.505 6:59 0:30 0:20 7:49 8:20 1:41 0:00 0:00 0:24 6:190:00 8:480:10 0:00 9:12 1.205 2 Str-a 5:19 0:30 0:30 1:10 9:48 4 Spl 0:10 7:38 0:30 0:303 Str-a Overtime Spread Penalty Paid Break Break Length Travel Hours Total Hours Spread Report Hours Run # Type Block # Piece Start Start End Piece End Plat Hours Guarantee 1:22 0:00 0:00 0:007 Str-p 0:00 0:217:43 0:50 Is this a “good” runcut? The results can be compared to our most recent solution. Runcut Comparison Straight Runs 6 75% 4 50% Split Runs 2 25% 2 25% Total FT Runs 8 6 Part Time Runs 0 0% 2 25% Total Operators Required 8 8 Hours Breakdown Total Avg % Total Avg % Platform 56.3 7.0 91% 54.5 6.8 88% Report 4.7 0.6 8% 4.7 0.6 8% Travel 1.0 0.1 2% 2.3 0.3 4% Paid Break 0.0 0.0 0% 0.2 0.0 0% Total Work Hours 62.0 7.8 61.7 7.7 Penalties Spread 2.8 2.8 Overtime 0.3 2.8 Guarantee 2.6 1.7 Total Paid Hours 67.7 69.0 Pay/Plat Ratio 1.2025 1.2661 Solution 1 Solution 2

Level ͯ. Advanced RuncuttingChapter 5. Runcutting 5-66 The major structural change is that two full-time straight runs are replaced by two part-time runs. The total runs remain unchanged, and the total worked and paid hours are only marginal- ly diff erent. A careful review of the runs themselves, including operational impacts and opera- tor preference, would be needed before a realistic answer could be provided. However, at fi rst glance the change does not appear to be signifi cant. Note that this example allowed meal breaks to be unpaid. This is not necessarily the case at all agencies, where labor rules may require breaks to be paid. Meal breaks are likely to be costly if paid and/or if required to be taken at the garage. Later in this chapter we note the potential impacts of such restrictions and discuss the operational desirability of these solutions. Approach 2: Operator Drop-backs Cutting runs for meal breaks using operator drop-backs (also known as fall-backs) is similar to the process described above. But in this case the operators are not required to return to the depot for their meals, and so have the meal break at the terminal. The procedure is relatively straightforward. Looking at the headway sheet, we can see that Block ͮ passes the relief location (Point A) at ͭͬ:ͯ͵. We cut the piece here. Then Block ͯ passes through at ͭͭ:ͬ͵. We cut the start of the next piece here. That gives a run with a drop-back (or break) from ͭͬ:ͯ͵ through to ͭͭ:ͬ͵ at Point A—shown as the black “run” in the diagram below. The same process can then be applied to Blocks ͯ and ͱ, providing a meal break between ͵:ͯ͵ and ͭͬ:ͬ͵ (the white “run” below). This process then cascades as a series of drop-backs are generated, each allowing the operator a ͯͬ-minute break at Point A. At the end of the process we would (hopefully) have a more effi cient runcut, as we have avoided travel to and from the depot associated with our previous runcut. Block 1 2 3 4 5 6 7 20 2111 12 13 14 15 16 17 18 195 6 7 8 9 10 However Block ͭ presents us with a more complex situation. It ends at Point A at ͵:ͬ͵ and then pulls back in to the depot. If we create a “run” shown in black on the diagram below we are left with a pull-in to cover, since the drop-back starts at ͵:ͬ͵ at Point A, and the operator gets off the bus. We could have a split run with a drop-back on Block Ͱ at Point A between ʹ:ͯ͵ and drop-back A technique where the operator or train crew gets off an arriving vehicle at a terminal, takes layover, and as- sumes operation of the next vehicle to arrive. Most common on frequent rail lines where close headways do not allow suffi cient layover time for the train crew, this technique is also used for special events to maximize the number of trains in service. If service is very frequent, the train crew may not board the next train but instead the train after that; this is called a “double drop-back.” Some agencies use the term “fall-back” instead.

Chapter 5. Runcutting 5-67 Level ͯ. Advanced Runcutting ͵:ͬ͵, with the run then covering the pull-in. However this may cascade and result in an infea- sible solution. Block 1 2 3 4 5 6 7 5 6 7 8 9 10 11 12 19 20 2113 14 15 16 17 18 At this stage we could go back to the approach used in the previous section and possibly rehook the blocks, much as we did then. In undertaking this approach we could eff ectively rehook the blocks and create a set of long layovers, each ͯͬ minutes, at point A. But we would not do this for all links at Point A, just for those that require a meal break in our runs. The op- erator would then have a ͯͬ-minute meal break in his or her vehicle. But would the break be ͯͬ minutes? In fact it would not—the operator would get the ͯͬ-minute break plus the layover before the next trip of ͮͭ minutes, resulting in a ͱͭ-minute break. Meal Break Runcut Issues The runcuts above indicate that meal breaks can be achieved for our revised Line ͵ͳ runcut with only a minor change in paid hours. Again, this was based on relatively liberal rules around the meal break requirements. As initially noted, many transit agencies assume that meal breaks will automatically result in signifi cant cost impacts. The likelihood of meal breaks resulting in increased or decreased run cut effi ciencies depends on a variety of rules and approaches. These include the following. Meal breaks are more likely to result in reduced effi ciencies if the breaks are paid. In the example above paid breaks would have added three hours, or Ͱ%, to the overall runcut cost. A Ͱ% cost impact would be at the lower end of expectations, based upon general experience. Multipiece runs (i.e., those with meal breaks) allow improved control over the length of one runs. Take the example of a one-piece run from an earlier solution. The vehicle only passed the relief location once every ͵ͬ minutes. Therefore if it passed at around ͳ:ͯͬ into the block, we are left with either an expensive (due to guarantee) short run, or increased overtime (taking the ͵:ͬͬ relief option). Multipiece runs, however, often allow runs to be created at a preferred length, according to the transit system’s preferences. This advantage is often overlooked. • •

Level ͯ. Advanced RuncuttingChapter 5. Runcutting 5-68 The effi ciency impact of meal breaks can be reduced if meals are not required to be taken at the garage. In the example above, the cost would have reduced by almost two hours and the “with meal break” runcut would have been less expensive than the “no meal break” runcut had the operators not been required to return to the garage for meal breaks. An additional benefi t of meal breaks is they obviate the need for consistently long lay- overs in blocks (which act as de facto meal breaks, but potentially occur several times a day and are paid). In one-piece runs with no prescribed meal break, there is a natural ex- pectation of layover to allow operator rest periods, often in the form of longer layovers. The potential impacts are both additional vehicles (through increased cycle times) and longer total nonproductive times, as the longer layovers tend to result in several rest periods per day. The implementation of meal breaks can therefore allow an opportunity to reduce those longer layovers that exist as a surrogate for meals and reduce overall operating costs. Pull reliefs can be used in either type of runcutting approach to meal breaks. However, this needs to be carefully handled in order to avoid increasing peak vehicle require- ments or excessive mileage. The table below summarizes some specifi c existing conditions where implementation of meal breaks and rest breaks can impact on runcutting effi ciencies. Rule/Parameter Low Cost Impact High Cost Impact Meal Break Paid/Unpaid Unpaid meal breaks. If unpaid, the meal breaks do not add unproductive time. Paid meal breaks. Adds (typically) ͯͬ-plus minutes of unproductive time, reducing the productive time in a work day. Likely to result in increased paid hours and increased operator requirements. Meal Break Location Away from Garage (can be either at the terminal or some other nearby location). Allows the meal breaks to occur without signifi cant increase in travel requirements. At the garage or limited locations. Causes increased travel times around the meal breaks. Relief Types Street reliefs allow more fl exibility in cutting runs with meal breaks, without necessarily increasing operational costs. Pull or car reliefs can result in higher mileage costs where meal breaks are required, depending on meal break location requirements. • • •

Chapter 5. Runcutting 5-69 Level ͯ. Advanced Runcutting Rule/Parameter Low Cost Impact High Cost Impact Overtime, Guarantee, and Run Length Where runs must be cut close to a predetermined length (e.g., eight hours), the ability to match long and shorter pieces together to get close to the preferred run length can actually reduce overtime and/ or guarantee costs. Where run length is not an issue, e.g., where overtime is paid weekly and not daily, the matching of shorter/longer pieces will have less benefi t. Relief Frequency Where reliefs occur infrequently (e.g., long routes with one relief location), one-piece straights tend to have higher guarantee and/or layover, as relief opportunities around eight hours are limited. Again the ability to match shorter/longer pieces presents an opportunity to reduce costs where meal breaks are utilized. Where relief locations are frequent and existing one-piece runs can be cut at around eight hours, the matching of shorter/longer pieces will have less benefi t. Layover Requirements Where existing layovers act as pseudo-meal breaks (where no offi cial meal break exists). In eff ect this means a meal break replaces an unoffi cial paid break, and has the potential for a net reduction in costs. Where existing layovers do not provide built-in breaks and meal breaks need to be created in addition to existing layovers. Run Cut Approach Pieces are recut into multipiece runs with a meal break in between. The example above showed how meals can be created without any signifi cant costing impact using this approach The approach using operator fall- backs could reduce fl exibility in cutting and matching pieces to create runs, where meal breaks have been applied. Relief Types Revisited When discussing scheduling inputs we covered the basics of relief types. However, these were presented as the basic types and only a limited discussion was provided. At this point, after undertaking several runcut solutions, with diff erent types of reliefs involved, further explora- tion is warranted.

Level ͯ. Advanced RuncuttingChapter 5. Runcutting 5-70 We may want to change our approach to reliefs to improve the effi ciency of the runcut, make the solution more operationally robust, or both. These are key goals that the scheduler should be striving to achieve for each and every runcut. Car Pooling There are cases when a car can be used to ferry more than one operator between the garage and a relief location, or even between two relief locations. This is known as car pooling. Take an example where we have a relief at Point A at ͭͭ:ͱ͵ (relief for Runs ͭ and ͮ) and ͭͮ:ͬͱ (relief for Runs ͯ and Ͱ). With a travel time of ten minutes this cannot be achieved by the same car, as it will get back to the garage at ͭͮ:ͬ͵ from the fi rst relief, but needs to have left at ͭͭ:ͱͱ to make the ͭͮ:ͬͱ relief. A second car would be required to perform this function. However it would be possible for one car to be used to accommodate both reliefs. If the car leaves the garage at ͭͭ:Ͱ͵ (to make the ͭͭ:ͱ͵ relief) it could carry the operators starting at both ͭͭ:ͱ͵ and ͭͮ:ͬͱ). And on the return journey it would leave at ͭͮ:ͬͱ and arrive back at the garage at ͭͮ:ͭͱ. The only cost implication is that Run ͯ now leaves the garage Ͳ minutes ear- lier (at ͭͭ:Ͱ͵ instead of ͭͭ:ͱͱ to make the ͭͮ:ͬͱ relief) and Run ͮ arrives back at the garage Ͳ minutes later (ͭͮ:ͭͱ instead of ͭͮ:ͬ͵ after making the ͭͭ:ͱ͵ relief). In both cases the aff ected operators are delayed by Ͱ minutes at the relief location Point A. The diagram below depicts this simple example of car pooling.

Chapter 5. Runcutting 5-71 Level ͯ. Advanced Runcutting The total runcut cost of this change is eight minutes of paid time (potentially ͭͮ minutes if there is overtime or spread premium involved). However this is off set against the cost of pur- chasing, managing, and maintaining an additional car. The calculation would clearly indicate a signifi cant cost saving. The scheduling of car pooling can become quite complex as multiple relief locations, cars, and operators are considered. For example, one car could drop an operator at one location, collect the driver relieved, drop a second operator at a second relief location, and return to the garage with the two returning operators. The approach to car pooling will be based upon how much additional time can be added to the runcut to enable a reduction in car requirements and mile- age. Most computerized systems have a means for displaying these car blocks and indicating which operators are car pooling. As with vehicle trips and blocks, car trips and blocks need to be carefully managed and well documented. For a relief to occur, the relief driver needs to be able to get to the right place at the right time. The car trips in the example above are shown in the following diagram. From Garage To Garage Driven By (Run #) Passenger (Run #) Driven By (Run #) Passenger (Run #) 1 3 2 411:49 11:59 12:05 12:15 Leave Garage Arrive Relief Leave Relief Arrive Garage Travelling on the Service Many agencies allow operators to travel as a passenger on an in-service vehicle to and from relief locations. The benefi t of this approach is to avoid the need for cars (and their accompany- ing costs), or pull or deadhead mileage. This approach can only be used where a route actually passes close to the depot and the relief location. Many agencies tend to use average travel times for such reliefs. This approach is fl awed and risks several poor outcomes. Typically the costs will be higher as agencies have to make a conservative estimate of the travel time required to avoid missed connections or reliefs. It also risks missed reliefs if the allowance is not high enough at times when running times are high- est. A more realistic approach is to use the actual scheduled times, perhaps with a late running buff er, to generate the travel requirements. This ensures the appropriate time is used for each

Level ͯ. Advanced RuncuttingChapter 5. Runcutting 5-72 individual relief, inclusive of running time variations during the day. The calculation is quite simple: Work back from the relief time and location. If a relief occurs at ͭͮ:ͯͬ at Point A, look at the schedule for the arrival at that location nearest to ͭͮ:ͯͬ, but with enough time to cover late running. If an in-service bus arrives at ͭͮ:ͮͱ or earlier, that would be ideal. Working back from the ͭͮ:ͮͱ arrival look at the schedule to ascertain when the bus passes the garage. Add any walk time needed to get to the bus stop from the garage, and that is the time for the operator to leave the garage. The diagram below depicts the calculation of travels using service trips. The assumptions are: Report/sign off times: ͭͱ minutes Walking time to stop outside garage: ͯ minutes Scheduled trip run time of ͭͬ minutes to Point A (relief location) Service frequency of ͭͬ minutes on the route used for travel ͱ minutes minimum buff er required either side of relief 1. 2. • • • • •

Chapter 5. Runcutting 5-73 Level ͯ. Advanced Runcutting Note that Run ͭ could have potentially caught a ͭͮ:ͯͬ trip back to the garage. However a zero- minute connection was ignored as we required a fi ve-minute buff er between the relief time and travel trip. Some computerized scheduling systems have features that enable a matrix of “travel in ser- vice” trips, inclusive of using other modes or even taking two routes to get to the relief loca- tion. With these features, if you are lucky enough to have a system with this feature, all you need to do is provide the computer with the “rule” on how tight a connection you will allow and identify any time of day restrictions. Reliefs During Layovers Thus far, all of the runcut examples have used the arrival time at a trip terminal as the relief time. This is the normal approach used in most cases because it allows a buff er (the layover) for late arrival at the relief location by the new operator. It also allows the relieving operator the time to get onto the vehicle, adjust the seat and mirrors, and key in any necessary information to the farebox. However there are cases when reliefs are not possible or preferred at the time of arrival at the terminal. For example one piece may extend beyond the maximum allowed before a meal, or run length. There are numerous reasons that a relief at a specifi c time may not work or be pre- ferred. Some of these are noted below. In our Line ͵ͳ example we have ͮͭ minutes of layover time at Point A, meaning there are a number of alternate options available. In the case of our drop-backs problem, cutting Block ͯ during the layover (͵:ͬ͵ to ͵:ͯͬ) may have allowed pieces from Blocks ͭ and ͯ to be matches to form a straight run, whereas the relief time of ͵:ͬ͵ did not. There may be cases where paid time can be reduced by reliefs during layovers. If a block is cut into two straight runs, one ͳ:Ͱͱ and the other ʹ:ͭͱ, it may be possible to move the relief ͭͱ minutes (assuming at least a ͭͱ-minute layover) and save ͮͮ minutes of paid time. The use of car pooling for reliefs may need some relief times to be shifted through layovers to allow a car to carry multiple operators. In these cases the relief can simply be made at any minute during the layover time. However there would be a loss of simplicity in the runcut, as the rule of thumb approach that “reliefs occur upon vehicle arrival at a terminal” has been violated. Finally, this option is only possible where a layover exists. At a midpoint relief location where there is no dwell or layover time, the time at that point must be the relief time.

Level ͯ. Advanced RuncuttingChapter 5. Runcutting 5-74 Pull Reliefs, Blocks, and Runs Pull reliefs present another option for meeting the aims of the scheduling process. Going again back to our runcut for Line ͵ͳ we can take an example Block and apply a pull relief. This means that the ͭͮ:ͬ͵ relief becomes a pull relief and the runs look like the following: Original - Sreet (Car) Relief Time Place Type Time Place Type 1 Str 2 5:46 6:01 Garage Pul l 12:09 A Street 12:34 6:08 0:30 0:10 6:48 2 Str 2 11:44 12:09 A Street 18:38 Garage Pul l 18:53 6:29 0:30 0:10 7:09 With Pull Relief Time Place Type Time Place Type 1 Str 2 5:46 6:01 Garage Pul l 12:19 Garage Pul l 12:34 6:18 0:30 0:10 6:48 2 Str 2 12:05 12:20 Garage Pul l 18:38 Garage Pul l 18:53 6:18 0:30 0:10 6:48 Total Hours Run # Type Block # Report Time Start End Sign Off Time Plat Hours Report Hours Travel Hours Plat Hours Report Hours Travel Hours Total Hours Run # Type Block # Report Time Start End Sign Off Time As the example shows, by using a pull relief we create a later start time for Run ͮ. How is this possible? Simply by the fact that for pull reliefs the layover does not need to be covered. The fi rst bus fi nishes the trip at ͭͮ:ͬ͵ and pulls straight back to the garage, and the second bus pulls out to arrive in time for the ͭͮ:ͯͬ departure. If this same approach is applied for all three reliefs of the straight runs the net cost outcome can be summarized as follows: Costs: Additional ͮͭ vehicle miles (ͯ.ͱ miles per trip x ͯ round trips) Savings: ͭ:ͬͯ worked time (may not translate to paid time if the run is below ʹ hours) ͮͭ car miles ͭ car In this case it would appear that the savings clearly outweigh the costs. There is however an impact of the vehicles being used at any one time, which now looks like the following. • • • • pull reliefs Reliefs made by pulling out one vehicle and pulling in another vehicle.

Chapter 5. Runcutting 5-75 Level ͯ. Advanced Runcutting We see a spike in the vehicles required from three to four at times when the pull reliefs occur. This assumes (reasonably) that the ͭͮ:ͭ͵ arriving bus cannot be the same bus as the departing ͭͮ:ͮͬ bus. A strong benefi t of this approach is that on-time performance of some trips is no longer de- pendent on the trips that preceded it, but rather is dependent only on the pull-out trip arriving on time. Since pull-out trips are easier to manage than time delays associated with in-service trips, the outcome should be improved service reliability, something all runcut solutions must look to achieve. There are of course circumstances where use of pull reliefs does not provide savings but in- stead incurs increased paid hours and increased mileage—relief locations farther from the garage or with minimal scheduled layover for example. And a pull relief during the peak, with an accompanying additional vehicle, is not acceptable scheduling practice. Limit the use of pull reliefs to midday when a pull relief will not increase vehicle requirements beyond the peak requirement for the route. Like operator hours, the maximum vehicle count is important and needs to be minimized. Tip

Level ͯ. Advanced RuncuttingChapter 5. Runcutting 5-76 There may be other reasons, such as maintenance issues, to avoid pull reliefs. In the example above there are now fewer vehicles available between ͭͬ AM and ͯ PM, a preferred mainte- nance window. Schedulers need to consult regularly with operations and maintenance staff s, especially if they are changing historic practice with reliefs, conditions for operators, or vehicle requirements. Relief Simplicity There is at times a tradeoff between simplicity and minimizing costs. Often the solution that would result in the lowest overall costs is too complex to run reliably in the fi eld. Use of reliefs is a classic example of an opportunity to achieve the right balance. Mathematically the more relief points there are, the better the chance of fi nding an optimized runcut solution. In fact the great constraint of transit runcutting over creating shifts with vari- able start/fi nish times is the limitation of when operators can get on and off the vehicle, to start or end their run. So from a mathematical viewpoint it would be ideal to have a potential relief at every bus stop. However excessive use of relief locations and types can complicate operations to the point of making the service less reliable. Car pooling is an example where there are clear effi ciencies to be gained. However, as soon as the schedule relies on a second operator’s trip arriving at the relief location to allow the fi rst operator to travel to the garage on time, the likelihood of service problems increases. The same approach to defi ning relief locations applies. While it is possible to have relief loca- tions at numerous timepoints throughout a route, operational simplicity suggests limiting this number. If operators know that Point A is the terminal and relief location for Line ͵ͳ, there is certainty. If occasionally Point B is used for reliefs, we have instantly introduced the chance of operator error (e.g., missing the relief at Point B and continuing to Point A, leaving the relieving operator stranded). The same concept applies to relief types—if relief to Point A is a car travel some times, travel in service at other times, and pull yet others, the chance for errors, resulting in service reliability impacts, increases. This is not to say that everything needs to be kept as simple as possible in all instances. It is up to the scheduler to work with operations staff and develop an understanding of the cost im- pacts of various approaches. Then the agency itself can consider the tradeoff as a policy deci- sion, based upon facts.

Chapter 5. Runcutting 5-77 Level ͯ. Advanced Runcutting Exception/Events Scheduling Scheduling for ever-changing circumstances places a signifi cant work burden on the scheduler. It also has a range of effi ciency and complexity impacts. There are primarily two types of “events” that aff ect schedules in a way that can be anticipated and written into the runcut. One is the special event such as a sports event or concert. These may require additional service on an occasional or semi-regular basis. The second type relates to changes to regular trips—either by day of week or perhaps time of year (school time/non- school time, for example). In both cases, there are well-utilized scheduling techniques that can be employed to minimize operational complexity while maintaining scheduling effi ciencies. Creating Base Solutions to Fit Known Events In order to accommodate known events, depending on the frequency of the event, it may be wise to create a runcut solution (and blocking solution for that matter) that will allow the re- vised or extra services to be readily incorporated. For instance this may mean cutting pieces in the regular runcut that can be extended for later service. If we take a ”typical” ͭͮ-hour block, to be cut into four-hour split piece and an eight- hour straight piece, we have a choice when to cut it—as a morning straight with a PM split (or part-time) piece, or as a morning split piece with a PM straight. If an agency has a semi-regular night event, it may be wise to schedule as an AM straight with a PM piece left as a part-time run. This piece can then be readily extended to cover the event. The approach to such schedules must be tailored to suit the specifi c requirements of a transit system. Forward planning in conjunction with operations staff can allow additional services to be accommodated eff ectively and effi ciently. Dealing with Day Exceptions Day exceptions relate typically to school services, or to services that operate diff erently on dif- ferent days. Examples include: School holiday versus school day schedules, where extra or altered services may oper- ate; School day exceptions for early or later bell times; or Additional services running later on a Friday or Saturday night. • • •

Level ͯ. Advanced RuncuttingChapter 5. Runcutting 5-78 A good approach is to maintain, as much as is reasonably possible, consistent schedules, blocks, and runs. What do we mean by this? Take a block with extra (later) Friday night trips, as below. The regular block is ͭͱ hours long whereas the Friday block is ͭͳ hours long. When cutting the block into two straights for Monday–Thursday the scheduler should consider how the run will fi nish on Fridays. This way the blocks and run can be kept the same for all days with a later fi n- ish on Friday (but the same block number and run number). School exceptions provide a more diffi cult task. One solution is to keep the school blocks as separate as possible. This can however aff ect effi ciency, both in terms of extra mileage and even potentially additional peak vehicles. Where the school blocks are kept separate, the PM Peak can look something like the following. Here we have a very short one-trip school block (Block ͱ) that operates as a stand-alone ve- hicle. We are left with two choices for blocking and runcutting this trip. At the runcutting stage we see that it has been blocked stand alone. We also see an obvious opportunity to hook the end of Block ͱ with the start of Block Ͱ to create a more effi cient run. But should we make this change? The answer depends on the amount of exceptions that apply to the school trip. If the changes only tend to be earlier bell times then we get to keep a consis- tent run across all exceptions by hooking the two pieces into a single block and part (or all) of a run. If this is the second part of a split run with a suffi ciently long AM piece, this run would oper- ate during both school and vacation periods, as well as during exception days. Again we have achieved a desirable operational outcome—simple and effi cient.

Chapter 5. Runcutting 5-79 Level ͯ. Advanced Runcutting The use of computerized systems to manage exceptions and events enables the scheduler to more readily create integrated solutions across regular and exception schedules. Owl Scheduling Scheduling for “owl” or overnight services presents the scheduler with an unusual question — when does one day end and the next one begin? There are two basic approaches to scheduling for owl services, and these are discussed below. Continued Blocks and Runs The most effi cient way to schedule owl blocks and runs is for the vehicle to continue on dur- ing the AM Peak and fi nish after the peak. Under this approach it is best to have the all-night blocks being the pre-PM peak pull-outs to avoid excessive block and run length. However this means that in eff ect the block covers both “today” and “tomorrow,” and the scheduler must understand this to be able to create runs. The diagram below depicts this approach to scheduling all-night services. Block 1 2 3 4 5 6 To Block 2 7 To Block 1 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 Note that Block ͭ is eff ectively the end of Block ͳ, and Block ͮ is eff ectively the end of Block Ͳ. The blocks operate from around ͯ PM until around ͵ AM the following morning. How can we cut these runs, present the information, and avoid mistakes? In the example pro- vided we have two blocks of ͭͳ and ͭ͵ hours duration. Assuming we have one-piece straights creating the runs is simply a case of cutting them somewhere in the middle, to create two straight runs per block. In the case of Block ͳ this would mean a cut at around ͮͯ:ͬͬ. This leaves us with a run commencing around ͭͭ PM and working through until ͵ AM the following day. Conceptually this is not diffi cult to handle. But in practice this eff ectively means that Blocks ͭ and ͮ are already covered and in eff ect don’t exist, or are just “dummy” blocks and pieces. This is shown below: Block 1 2 3 4 5 6 A To Block 2 7 B To Block 1 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34

Level ͯ. Advanced RuncuttingChapter 5. Runcutting 5-80 Cut Blocks and Runs An alternative is to cut the blocks by use of pull reliefs. The same concept as the above example can be applied, but the buses instead pull out and in at around midnight, giving ͮ blocks from ͮ or ͯ PM until midnight, and another two from midnight until after the AM peak. The blocks could be cut at a number of diff erent times—for example after the PM peak, and before the AM peak. This would leave shorter peak blocks, with longer blocks from around ͳ or ʹ PM through until ͱ AM—in eff ect a single-piece owl run with a stand alone vehicle block. Impacts of Diff erent Service Days One of the diffi culties lies in transitioning between diff erent service types. In the example above, on a Friday night, Block Ͳ cannot connect to Block ͭ, since Block ͭ does not necessar- ily exist, certainly in the same form as weekdays, on Saturday. And the same problem occurs transitioning from Saturday to Sunday, and Sunday to weekdays. This means that the scheduling becomes quite complex if the “continuous blocks” approach is applied, since it means having a separate Friday runcut with a separate “dummy” type run for the Saturday AM continuation. The alternative approach, cutting the blocks through pull reliefs into independent runs, is more operationally expedient and keeps the scheduling more streamlined. However there is a cost associated with the additional mileage resulting from the extra pull trips. Under either approach, issues as to when one day starts and another ends, and when the owl runs actually begins, need to be resolved. Runcutting Issues This section describes several policy issues related to scheduling of labor resources. The dis- cussion is by no means exhaustive, either in the number of issues that are discussed, or in the discussion of each issue. The aim here is to make schedulers and others within transit systems aware of some of the more important issues around runcutting. The diff erences in work rules, policies, and approaches across agencies make it impossible to provide solutions that will work at every property. We hope that readers use the discussion to better understand the issues and review their thinking in light of specifi c requirements at their particular system. A. Overtime Optimization The normal work week is thought of as eight hours a day, fi ve days per week, totaling Ͱͬ hours. While there are variations (Ͱ/ͭͬ is a common example), the weekly total for the typical full-time

Chapter 5. Runcutting 5-81 Level ͯ. Advanced Runcutting American hourly worker is Ͱͬ hours. Yet as everyone involved with scheduling and operating transit service soon realizes, transit operators rarely fall into what offi ce workers take for grant- ed as a “standard” work week. If a run does pay exactly Ͱͬ hours it is almost certainly because make-up time has been added to a shorter run to bring the run up to a required Ͱͬ hours. Why is this? Simply stated, it is the rare route where one-way running time, plus recovery/lay- over is exactly Ͳͬ minutes. And then, of course, there are pull-out and pull-in time to consider. So schedulers are constantly balancing between paying extra “make-up time” to bring full-time runs up to a required Ͱͬ hours and paying overtime for work in excess of Ͱͬ hours per week. Runcutting is in essence a balancing act, fi nding the optimal point where the number of over- time hours is optimized to produce the lowest overall cost to the property, within the restric- tions of your labor agreement. The optimal amount of overtime is not always the lowest amount. Take for example a schedule with ͳͮ hours of work to be divided into daily runs. In the simplest example, this work could be divided into nine ʹ-hour runs or eight ͵-hour runs. Actually, in the simple convention of com- puting overtime on a daily basis (this topic—daily versus weekly overtime—is discussed further in Chapter Ͳ: Rostering) eight ͵-hour runs will pay ͳͲ hours (including Ͱ hours overtime pay, as- suming a time-and-a-half rate for overtime), or ͮͬ hours more per week than nine ʹ-hour runs. However, we should not assume that the runs with the least overtime actually cost less. It may seem counterintuitive but paying out some overtime can actually SAVE money. The reason is simple; it may be cheaper in total payroll costs, overall. When there were ͳͮ hours paid daily, rather than ͳͲ, there were nine people payrolled. With ͳͲ daily pay hours, there were eight people payrolled, or one less. That is one less set of benefi ts: medical benefi ts, pension benefi ts, insurance, etc. It also is, from a risk management standpoint, one less risk of/ exposure to disability, long-term absence, etc. Further, depending on the agency’s wage rates and expenditures for unscheduled overtime (when an operator works a second run in the same day, etc.), Social Security (FICA) costs may be avoided. Taken together, these types of benefi ts are known as “welfare” benefi ts, as they are focused on the health and welfare of the operator. There are also “leave” benefi ts. These are the ones that infl uence and impact an operator’s daily attendance such as vacation, holidays, personal days, etc. Also included in this group are the days that cause an operator’s absence from the seat of the bus but are not “away from the offi ce,” so to speak. This includes those days when the operator is at work but not driving in revenue service such as training, fi lling in for a road supervisor, court appearances (due to an accident witnessed, for instance), etc. If the combined value of these benefi ts is actuarially ͱͬ.ͭ% or greater than the average sys- tem base wage rate, then paying overtime will almost always be less expensive for the agency

Level ͯ. Advanced RuncuttingChapter 5. Runcutting 5-82 than adding additional payrolled bodies. Indeed, theoretically under such circumstances there should be no limit on overtime. But other contract limitations/collaterals (maximum spread)— and the humaneness and safety implications—dictate not going overboard. How does an agency go about determining its optimum scheduled overtime? It is indeed an actuarial eff ort because the key factor, the linchpin so to speak, in making the determination is its typical operator (employee) availability, i.e., what remains after the leave benefi ts are used. In a normal work year, an employee is paid for ͮͲͬ days (fi ve day work-week multiplied by ͱͮ weeks per year). If the average employee took ͭͬ days off , while paid for ͮͲͬ, s/he would have ͮͱͬ days of person-day-availability, or PDA. With its mean PDA of ͮͱͬ (͵Ͳ.ͮ%), the agency knows that based on typical availability, for every ͮͱ operators payrolled, it needs another one. That’s because for every day the operator doesn’t work his/her run, another operator needs to back fi ll to cover the work. Only ͭͬ days off per year—or not working in revenue service—is highly unrealistic. In addition to vacation, there are holidays, personal days, training, sick leave, long-term disability leave, an operator fi lling in for a vacationing road supervisor, and other reasons for absences. When computed on a systemwide basis—particularly with an older work force—ͮͯͬ PDA is consid- ered quite good (ʹʹ.ͱ%); ͮͰͬ (͵ͮ.ͯ%) extraordinary. In the case of the ͮͯͬ PDA, for every ͮͱ operators payrolled, another ͯ.ͯ are required. There are numerous systems—particularly larger ones with an older complement of operators—where ͮͬͬ PDA is not uncommon. That ͳͳ% availability translates into ͳ.ͱ additional operators for every ͮͱ payrolled to fi ll runs. (This is, of course, also one of the fundamental building blocks in determining extraboard size.) Thus, in trying to determine the optimum amount of overtime, in addition to the welfare ben- efi ts listed above (medical insurance, FICA, etc.), the cost of leave benefi ts to the agency—va- cation, sick, personal days, average number of days in training, etc.—on an individualized basis must be ascertained. Adding operators to the payroll adds both the cost of welfare benefi ts and the cost of leave benefi ts to the payroll. In eff ect, it is hard to add only one operator since you also have to add enough bodies, either through extraboard or other means to cover for operator leaves. To determine the value of leave benefi ts attributable to an individual, an analysis of the agen- cy’s overall operating budget/costs is required. Simply stated, this analysis takes all leave costs and apportions those costs to each operator budget position. This is a daunting eff ort involving many diff erent departments (human resources, labor relations, payroll, treasury and fi nance in addition to transportation). It is certainly not something that is done with every sign-up but should be done occasionally to revisit the policy on overtime and to provide guidance to the scheduler when deciding how much overtime to put into runs.

Chapter 5. Runcutting 5-83 Level ͯ. Advanced Runcutting It is after that cost determination is made—with the realization that each operator brings to the payroll not just salary costs but the mathematic probability of exposure to collateral long- term costs such as disability payments in addition to those of his/her “benefi ts package”—that the size of the optimum run that includes built-in, scheduled overtime can be established. B. Part-time Operators The previous section considered some of the key factors in optimizing overtime costs. This initial discussion however failed to take into account alternatives for the makeup of the runcut. Use of part-time operators can aff ect the types of runs used, the cost of the runcuts, the use of full-time operators, and a range of subtle outcomes. Again this document will not attempt to provide a prescriptive assessment of the benefi ts of part-time operators, but instead discuss some of the more important scheduling-related con- cepts TCRP Report Ͳʹ indicated three key cost saving infl uences of part-time operators: They permit transit agencies to reduce the amount of premium and guaranteed pay entailed by a service schedule staff ed entirely by full-time operators; Part-time operator wages are generally lower; and Part-time operators generally have lower fringe benefi ts. Which Situations Best Fit Part-time Runs? Traditional scheduling approaches suggest that part-time runs are best suited to dealing with highly peaked situations, or where rules are complex and the number of constraints is high. The traditional industry model for part-time operators is to create runs of up to six hours in length, allowing up to a fi ve-day ͯͬ-hour workweek. The fi rst situation allows part-time runs to be substituted for split runs, thereby resulting in re- duced spread premiums and potentially avoiding guarantee or overtime costs. The second case allows smaller pieces to be created that allow the overall runcut solution to be optimized. If we go back and look at our simple Line ͵ͳ example we see potential for some blocks to form either split or part-time runs. In this case there are two blocks in either peak that were used as a basis to form split runs in an initial runcut. • • •

Level ͯ. Advanced RuncuttingChapter 5. Runcutting 5-84 If we simply take those four pieces of work assigned to split runs and convert them to part-time runs the cost impacts can be easily reviewed, and are provided below. Split Part Time Total Runs 2 4 Platform Hours 15:01 15:01 Travel Hours 0:30 0:30 Spread Premium 2:46 Overtime 0:35 Guarantee 0:00 Total Paid Hours 18:52 15:31

Chapter 5. Runcutting 5-85 Level ͯ. Advanced Runcutting In this example the savings of using four part-time runs to replace the two split runs is over three hours and equates to almost ͭʹ%. If the hourly rate for the part-time operator is lower than for full time the savings become even more signifi cant. It may also be possible in other cases for part-time operators to work two small peak pieces (assuming total work time is within labor rule limits), thereby replacing some split runs on a one-to-one basis. Appreciable gains in effi ciency can be achieved by this strategy. This is a basic example, and the conditions for splits and spreads made the split runs relatively ineffi cient to begin with. However, it illustrates the potential savings that creation of part-time runs can provide. Part-time runs may also be useful where there is an imbalance between the peaks and the peak blocks cannot all be assigned to split runs. Chapter Ͳ: Rostering notes an- other potential benefi t of the use of part-time operators: to fi ll open run imbalances on certain days of the week. Use of computerized scheduling packages can allow more sophisticated modeling of the impacts of varying levels of part-time runs, where total paid hours can be compared and multi- plied by actual wage rates. More Complex Runcutting The example presented described a basic situation with relatively simple rules and few con- straints or limitations. There are many more situations where part-time runs can be eff ectively used to produce an effi cient runcut. The complexity of work rules around all run types can result in pieces of work that are diffi cult to assign to full-time runs. For example, in the meal breaks section we created a runcut that required two smaller pieces of work, which were then assigned as part-time runs. A set of more highly constrained work rules is more likely to result in a need for shorter run types ideally suited to part-time operators. As the complexity of the rules and number of con- straints increase, it becomes more diffi cult to cover all of the blocks with full-time runs. Typically there will be a limitation on the number of part-time runs that can be created within a runcut or assigned in a roster. In many cases the “optimal” number of part-time runs, from a runcut perspective, will be close to or above this limitation.

Level ͯ. Advanced RuncuttingChapter 5. Runcutting 5-86 Operational and External Impacts The above discussion indicates the well-known potential effi ciency benefi ts in application of part-time runs and operators. However there are a number of other issues relating to part-time operator deployment that can aff ect the overall benefi ts. The relative competence, training requirements, and retention rates of part-time operators all aff ect the overall benefi ts of their use. These factors can translate into measurable cost impacts such as higher accident rates or increased unscheduled overtime (through either increased absenteeism or unfi lled operator positions). The specifi c impacts of such factors tend to be local in nature, and our research indicated diff erent viewpoints as to the benefi ts or costs relating to part-time operators across diff erent transit agencies. Ideally historical data at an agency can be obtained, and any related factors can have costs or values applied to allow a reasonable estimation of the total impacts (i.e., measure the runcut or roster savings against any operational or external cost impacts). C. 10-Hour Runs Ten-hour runs and rostering of four ͭͬ-hour days for a driver work week off ers another tool to schedulers. As with the discussion of balancing overtime and part-time operators, introducing ͭͬ-hour runs can be considered as purely a runcut issue, or as a runcut and rostering issue. Returning to some of the examples presented above, creating longer runs at the runcut level is likely to result in reduced travel, report, and guarantee. Taking another simple example we will look at ͭͬͬ hours of platform time and assume the runs can be cut as either all eight-hour or all ͭͬ-hour. We assume that there are Ͱͬ minutes of report and travel time for each run, and that we are looking for an average of ʹ:ͬͬ and ͭͬ:ͬͬ, respectively. 8-Hour 10-Hour Total Runs 13.3 10.5 Platform Hours 100.0 100.0 Report Hours 6.7 5.3 Travel Hours 2.2 1.8 Total Paid Hours 122.2 117.5 -4.7 -3.8%

Chapter 5. Runcutting 5-87 Level ͯ. Advanced Runcutting The idea here is to demonstrate that, as fewer runs are created, the total cost of travel time and report time is likely to decrease because these cost factors are related to the number of runs. Rules Aff ecting 10-Hour Runs There are of course many diff erent implementations and rules governing the use of ͭͬ-hour runs, diff ering by agency. However some general types of rules exist and may specify: The minimum and/or maximum number of ͭͬ-hour runs allowed, usually expressed as a percentage of total runs Overtime and guarantee will tend to apply at the ͭͬ-hour mark and not the usual ʹ- hour mark Spread penalties will apply, but probably only to split runs Other rules will be applied to ͭͬ-hour runs as they are to eight-hour runs Workforce Factors While the example above would suggest that ͭͬ-hour runs are a tool for saving money, the re- ality is not that clear. The assumption with ͭͬ-hour runs is that they are tied to four-day work- weeks. If a ͭͬ-hour solution requires more operators to be hired, it may actually be a higher cost solution since every operator comes at a fi xed cost for benefi ts. At this stage we need to take a step back and consider the workforce impacts of our two options. We take our runcut numbers from above and use them to estimate the number of operators that are required. 8-Hour 10-Hour Total Weekly Runs 66.7 52.6 Days Worked Per Week 5 4 Operators Required 13.3 13.2 And the answer is the same! Taking a step back like this makes sense, as with either ͱ-ʹ hour or Ͱ-ͭͬ hour workweeks the total number of worked hours is Ͱͬ. The table does show a mathe- matical reduction of ͬ.ͭ operators required (less than ͭ%), but this does not necessarily trans- late into any real workforce reductions. The implication here is that Ͱ-ͭͬ workweeks do not result in reduced labor requirements. The basic tenet is that the total number of operators required is a function of how much work is undertaken in a week, not in a given runcut. The reduced numbers in a runcut therefore need to be considered against the total days worked. • • • •

Level ͯ. Advanced RuncuttingChapter 5. Runcutting 5-88 Runcut Impacts Further consideration is required into how ͭͬ-hour (or any longer) runs aff ect the runcut. As with many runcut issues, this is a function of many inputs. Let’s return to our original Line ͵ͳ blocks and consider the potential to create ͭͬ-hour runs. Block 1 2 3 4 5 6 7 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 Here there are three ͭͬ-hour straight runs highlighted. But notice the eff ect on the rest of the runcut. We are left with seven shorter pieces to make into split or part-time runs. In this case, spread limits constrain the number of split runs to at most two, leaving us with three part-time runs. The impact of the ͭͬ-hour runs in this case is to increase the number of shorter part-time runs to three. This may or may not be a desirable outcome. Our total number of runs remains at eight (three straight, two split, and three part-time), yet when it comes to rostering/bidding there will actually be an additional operator required (as the ͭͬ-hour runs will be covered by operators working only four days). In this case the answer is probably that ͭͬ-hour runs, combined with four-day workweeks, would result in an effi ciency loss. However there are many cases where ͭͬ-hour runs and Ͱ-day workweeks may be appropriate. Again the key driver of effi ciencies through this approach is the reduction in nonproductive time in the runcut (primarily travel and sign-on/clear time) through fewer total runs. Issues with 10-Hour Runs The discussion above has covered some of the impacts of ͭͬ-hour runs, whether combined with Ͱ-day workweeks or not. Below are some further issues to consider when creating ͭͬ-hour runs. A longer span of service, with a higher level of off -peak service, tends to lessen the runcut impact of ͭͬ-hour runs. For example if all blocks are only ͭͰ-hours long, cutting ͭͬ-hour pieces limits the number of straight runs that can be created (as in the above example). However longer blocks reduce this impact. Eighteen-hour blocks allow one ͭͬ-hour run and one eight-hour straight run to be created, for example. •

Chapter 5. Runcutting 5-89 Level ͯ. Advanced Runcutting A larger runcut generally provides more options for “the rest of the runcut” if ͭͬ- hour runs are applied. Simply put, higher volume in the runcut allows more options for the remaining pieces. Typically there will be limitations on the number of ͭͬ-hour runs that can be cre- ated. As a rule of thumb it would be rare for more than ͮͬ% of runs to be of the ͭͬ-hour type. ͭͬ-hour runs should incorporate the same philosophy as eight-hour runs, where possible. That is, there should be a balance between runs over ͭͬ hours and under ͭͬ hours. However in some circumstances driving time limits (in some cases set at ͭͬ hours) limit this capability. In this case guarantee time should be avoided and runs cut as close to ͭͬ hours as possible. Split runs are diffi cult to increase in length at their extremities (i.e., before the AM Peak and after the PM peak) due to spread constraints, meaning that ͭͬ-hour splits will tend to creep into the off -peak periods between ͵ AM and ͭͭ AM and ͭ PM and ͯ PM. This becomes less feasible if the off -peak service levels are low and can also reduce the number of straight runs. However, it is generally diffi cult to build longer split runs, and use of ͭͬ-hour splits often leads to an increase in make-up time (assuming a guarantee of ͭͬ hours). Work rules at many systems allow for the creation of longer (up to ͭͬ-hour) runs as part of a normal fi ve-day workweek. Whether such runs are effi cient depends upon the overtime approach adopted (see the earlier discussion on optimized overtime). The decision to apply ͭͬ-hour runs to four-day workweeks is as much a rostering and work- force utilization issue as it is a runcut issue. Longer runs can have operational and absence coverage impacts. For example an operator who works a Ͳ AM to ͮ PM straight run may be available to cover a school trip- per or short PM block as overtime, possibly to cover an unscheduled absence. However if the work day is longer there may be less opportunity for such coverage, or safety is- sues through longer driving times may arise. There are potential safety impacts of operators working longer runs. Your agency should have some data on any link between longer daily driving hours and accidents, for example. D. Workforce Utilization The three preceding sections have each covered aspects of workforce utilization. The runcut can aff ect workforce utilization through the mix of run types applied, by the length of runs cre- ated, and by creating a runcut that allows for rosters to be built eff ectively. While each of the • • • • • • •

Level ͯ. Advanced RuncuttingChapter 5. Runcutting 5-90 three discussions was presented in isolation, the tradeoff s between overtime, part-time opera- tors, and ͭͬ-hour runs are made interactively by schedulers when creating runcuts. Workforce utilization from a scheduler’s perspective relates to how the runs and rosters are built to apply the available labor resources most effi ciently, within the constraints provided by contracts and work rules. In this sense it relates not only to the scheduled cost of the runcut but also how this translates to extraboard utilization, absenteeism management, and unscheduled overtime. In the preceding discussions we have noted a strong link between the makeup of the runcut/ rosters and the management of labor in terms of levels of absenteeism and unscheduled over- time. There are certainly no prescriptive answers here. Below we note some of the factors that infl uence how eff ectively the planned runcut translates into effi cient operations: The current labor market Agency-specifi c wage rates How the roster is constructed, e.g., is weekend work rostered or left as uncovered for overtime? Are trippers created? If so, are they left uncovered for overtime, or assigned to the extraboard? What is the size of the extraboard? E. Cost, Complexity, and Quality A good scheduler understands that at times there are tradeoff s between the overall cost of a runcut and the complexity of the runs created. We have touched on this topic repeatedly throughout the manual, as we covered issues including: Interlining. There can be a tradeoff between the amount of interlining and the cost of the solution. Garage assignments. Keeping garage assignments simple may result in additional deadhead mileage and hours. Run Types. In particular the use of part-time and ͭͬ-hour runs can aff ect effi ciency lev- els and operational complexity. Reliefs. We have covered in detail the potential to use diff erent relief types, at diff ering levels of complexity, to achieve a range of outcomes. • • • • • • • • •

Chapter 5. Runcutting 5-91 Level ͯ. Advanced Runcutting Multipiece runs, which allow potential for improved effi ciencies but represent a more challenging operational outcome. Meals. Having meal breaks in runs, however handled, makes the runcut and the opera- tion more complex. Meals can be created in a manner that reduces this complexity. Events and exceptions, which can be scheduled in a manner conducive to consistency of operation from one period to the next. In all aspects of scheduling the scheduler is attempting to create operationally sound and effi cient runcuts. The important consideration here is that these are not mutually exclusive outcomes. The section below notes the ability of computerized systems to produce improved quality out- comes for the same effi ciency levels. What does this mean? It means that the scheduler should not only model cost constraints and work rules but also quality and operational preferences. These can then be considered as part of the runcutting solution as well. Consider a case where we have two split runs with spreads of ͭͯ:ͬͬ and ͭͭ:ͬͬ, and maximum of ͭͯ:ͬͬ allowed in the labor agreement. By swapping the PM pieces we could end up with two ͭͮ-hour spreads. The cost of the spread premium is the same but now we have a potentially better outcome. The computerized system can be told this—that we prefer equalized spreads or fewer spreads close to the maximum limit. We can then run an option with this limit and without this limit, and see if there is any discernable change in effi ciency levels. F. Computerized Scheduling Revisited During the course of previous sections we have noted the capacity of computerized scheduling systems to assist in and improve the scheduling process. Our overall observation is that the widespread introduction of computerized systems has resulted in effi ciency gains, particularly in larger systems where the size of the runcutting prob- lem means the sheer computational power of the computer provides better solutions. When undertaking runcutting, there are several approaches that need to be applied in order to best harness the computational power of the system and work within the needs of operational outcomes. Some of these are noted below. Understand the solution that the system should be trying to achieve. Consider some of the approaches we have discussed here and get to know what the solution should be, before the computer generates one. • • • •

Level ͯ. Advanced RuncuttingChapter 5. Runcutting 5-92 Test the rules by manually creating, within the system, various types of runs. Make sure these are legal and are costed correctly. The system must be able to accurately create and cost runs according to the rules provided, and this calibration step is critical. The rules should be tested and refi ned with a “one-at-a-time” approach to ensure each aspect of the labor agreement is accurately modeled. The fi rst solution is not the best solution—ever. “Push button” scheduling is never an option for producing quality schedules, runcuts, or rosters. In every scheduling solution there is potential for improvement, whether created manually or by the computer. Cre- ate solutions, tweak the rules or parameters, fi x some preferred runs—do any of these and recreate the solution, always looking for improvement. Running ͮͬ or ͱͬ versions of a solution is not uncommon, nor is it unreasonable. Use the interactive tools provided. Some systems provide excellent tools for a sched- uler to create a runcut interactively, without submitting an automatic solution. The abil- ity to cut pieces, form runs, and see results as the runcut develops is extremely powerful in the hands of an experienced scheduler. Reconsider the rules/parameters/constraints for each new runcut. Schedulers can fall into the habit of a “set and forget” approach to rules as defi ned in a computerized sys- tem. For every new sign-up, there are likely to be changes to the trips, blocks, or even work rules. Consider how these have changed and adjust the system accordingly. Recall objectives that could not be achieved last time and attempt to achieve them this time. Start with simple rules. Do not overconstrain a system in the fi rst instance. Start by putting in only basic “hard” rules—durations, counts, limits. Do not decide things for the system if you are not sure of the outcome. Then start to add in constraints and changes to penalties or weightings, preferably one at a time. Build up a more complex set of rules as you go. Carefully consider every rule, parameter, and constraint. Each rule is eff ectively a line of code that will have a value assigned during an automated runcut. Each rule is there- fore like writing a line of computer code. Make sure the rule has the desired impact by changing only that rule and recreating the solution. Look for solutions that allow improved quality within the same resource require- ments or total costs. Once you are satisfi ed with the costing levels (number of runs, paid hours, etc.) look to improve the quality of the runcut. This can be both the opera- tional quality (simplicity is generally better) and the quality of life for operators. These outcomes are not mutually exclusive, and a good scheduler will be aware of how far a system can be used to provide these multiple outcomes. • • • • • • •

Chapter 5. Runcutting 5-93 Level ͯ. Advanced Runcutting Note that many of these suggestions hold true for both computerized and noncomputerized scheduling processes. Even the most sophisticated computer system still requires review and hands-on involvement of the scheduler, who can make adjustments throughout the automated process to enhance both the quality and the effi ciency of the outcome.

Level ͯ. Advanced RuncuttingChapter 5. Runcutting 5-94 LEVEL 3 End of Advanced Runcutting. Rostering continues on the next page.

Next: Chapter 6 - Rostering »
Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

TRB’s Transit Cooperative Research Program (TCRP) Report 135: Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling explores information on available scheduling tools and techniques and their capabilities. Also, the report provides guidance to transit agencies on a variety of scheduling issues typically faced in a transit operating environment.

TCRP Report 135 is an update to TCRP Report 30: Transit Scheduling: Basic and Advanced Manuals and addresses contemporary issues in transit scheduling. The appendixes to TCRP Report 135 were published as TCRP Web-Only Document 45: Appendixes to TCRP Report 135: Controlling System Costs: Basic and Advanced Scheduling Manuals and Contemporary Issues in Transit Scheduling.

An interactive scheduling manual is available as an ISO image. Links to the ISO image and instructions for burning a CD-ROM from an ISO image are provided below. Once a CD-ROM has been made with the ISO image, open the folder on the CD-ROM called Interactive Scheduling Manual and click on Transit_Scheduling_Lessons.pps.

Help on Burning an .ISO CD-ROM Image

Download the .ISO CD-ROM Image

(Warning: This is a large and may take some time to download using a high-speed connection.)

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!