National Academies Press: OpenBook

Pedestrian and Bicycle Safety Performance Functions (2023)

Chapter: Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology

« Previous: Section 3 - Development of Pedestrian and Bicycle Models Incorporating Available Pedestrian and Bicyclist Exposure Data
Page 201
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 201
Page 202
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 202
Page 203
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 203
Page 204
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 204
Page 205
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 205
Page 206
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 206
Page 207
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 207
Page 208
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 208
Page 209
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 209
Page 210
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 210
Page 211
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 211
Page 212
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 212
Page 213
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 213
Page 214
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 214
Page 215
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 215
Page 216
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 216
Page 217
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 217
Page 218
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 218
Page 219
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 219
Page 220
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 220
Page 221
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 221
Page 222
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 222
Page 223
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 223
Page 224
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 224
Page 225
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 225
Page 226
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 226
Page 227
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 227
Page 228
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 228
Page 229
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 229
Page 230
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 230
Page 231
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 231
Page 232
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 232
Page 233
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 233
Page 234
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 234
Page 235
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 235
Page 236
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 236
Page 237
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 237
Page 238
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 238
Page 239
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 239
Page 240
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 240
Page 241
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 241
Page 242
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 242
Page 243
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 243
Page 244
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 244
Page 245
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 245
Page 246
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 246
Page 247
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 247
Page 248
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 248
Page 249
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 249
Page 250
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 250
Page 251
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 251
Page 252
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 252
Page 253
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 253
Page 254
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 254
Page 255
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 255
Page 256
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 256
Page 257
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 257
Page 258
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 258
Page 259
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 259
Page 260
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 260
Page 261
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 261
Page 262
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 262
Page 263
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 263
Page 264
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 264
Page 265
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 265
Page 266
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 266
Page 267
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 267
Page 268
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 268
Page 269
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 269
Page 270
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 270
Page 271
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 271
Page 272
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 272
Page 273
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 273
Page 274
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 274
Page 275
Suggested Citation:"Section 4 - Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology." National Academies of Sciences, Engineering, and Medicine. 2023. Pedestrian and Bicycle Safety Performance Functions. Washington, DC: The National Academies Press. doi: 10.17226/27294.
×
Page 275

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.

201   This section of the report documents the development of crash prediction methods for pedes- trian and bicycle crashes for potential incorporation into the HSM based on the crash prediction models used by the U.S. Road Assessment Program (usRAP) and its international partner, the International Road Assessment Program (iRAP). These usRAP/iRAP models will be referred to in this report as the Road Assessment Program (RAP) models. It is anticipated that the crash prediction methods for pedestrian and bicycle crashes presented in this section could be consid- ered for potential use in the HSM2 Part C predictive methods chapters. Specifically, predictive models have been developed for use in HSM2 for both pedestrian and bicycle crashes for three roadway facility types: • Two-lane two-way roads for use in HSM Part C, Chapter 10 (or its successor in the HSM2). • Rural multilane highways for use in HSM Part C, Chapter 11 (or its successor in the HSM2). • Urban and suburban arterials for use in HSM Part C, Chapter 12 (or its successor in the HSM2). General information related to pedestrian and bicycle safety from the RAP models, such as contributing factors to pedestrian and bicycle crashes, could also be considered for potential use in the chapter on pedestrians and bicyclists planned for HSM2. The RAP models for pedestrian and bicycle crashes were developed for worldwide application. These models have been adapted in the current research in several ways to better fit with U.S. highway safety practice and terminology and the format for HSM2. 4.1 General Adaptations Made to the RAP Procedures The following general adaptations have been made to the RAP procedures in configuring them for application in HSM2. 4.1.1 Organization of Crash Prediction Procedures The RAP models for pedestrian and bicycle crashes apply to all roadway types. There are no separate procedures by roadway type, as in the HSM. For example, a pedestrian crash on a rural two-lane undivided highway would be analyzed in the RAP models by including the following input variables in the analysis: • Area type = Rural • Number of through lanes = One per direction of travel • Median type = Centerline only S E C T I O N 4 Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology

202 Pedestrian and Bicycle Safety Performance Functions For application in HSM2, the RAP models have been configured separately for two-lane roads, rural multilane highways, and urban and suburban arterials. 4.1.2 Use of Roadway Segment Terminology That Refers Specifically to the Left and Right Sides of the Road The RAP models were developed for worldwide application and, therefore, were developed for application in countries where vehicles are driven on either the left or right side of the road. For this reason, the RAP procedures refer to the two sides of a road as the “driver’s side of the road” or the “passenger’s side of the road.” Since the HSM2 models are intended for application in the United States, where vehicles are driven on the right side of the road, the RAP models for the driver’s side of the road are referred to as applying to the left side of the road, and the RAP models for the passenger’s side of the road are referred to as applying to the right side of the road. Left and right refer to the sides of the road to the left and right with reference to a primary direction of travel on the road designated by the model’s user (often, but not necessarily, the direction of increasing mileposts). 4.1.3 Separate Treatment of Roadway Segments and Intersections The RAP models treat intersections and roadway segments together in the analysis. If a 327-ft (100-m) roadway segment contains an at-grade intersection, then the crash estimates provided by the applicable RAP models include both segment-related and intersection-related crashes that are predicted to occur within the limits of that segment. However, the structure of the RAP models lends itself readily to separating the segment and intersection crashes. Pedestrian Crashes: The RAP model procedures develop estimates for four specific pedestrian- related crash types. These are: • Crashes related to pedestrian movements along the left side of the road. • Crashes related to pedestrian movements along the right side of the road. • Crashes related to pedestrian movements crossing the inspected road. • Crashes related to pedestrian movements crossing the side road. Crashes related to pedestrian movements crossing the inspected road can be separated into two separate categories based on whether an intersection is present within the 327-ft (100-m) road segment in question: • Crashes related to pedestrian movements crossing the inspected road at a midblock location. • Crashes related to pedestrian movements crossing the inspected road at an intersection. By definition, one or the other of these two crash estimates will be zero, depending upon whether or not an intersection is present within the section. The other of these two crash estimates will be zero only if the pedestrian crossing volume is zero. Thus, splitting the category for crashes related to pedestrian movements crossing the inspected road results in five, rather than four, specific pedes- trian crash types: • Crashes related to pedestrian movements along the left side of the road. • Crashes related to pedestrian movements along the right side of the road. • Crashes related to pedestrian movements crossing the inspected road at a midblock location. • Crashes related to pedestrian movements crossing the inspected road at an intersection. • Crashes related to pedestrian movements crossing the side road at an intersection. The first three pedestrian crash types are, by definition, segment-related crashes; the latter two pedestrian crash types are, by definition, intersection-related crashes. Separating the crash types

Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology 203   in this way is equivalent to the following RAP procedure for getting four separate estimates of roadway-segment-related and intersection-related crash estimates. 1. Estimate the annual crash frequency for the 327-ft (100-m) road segment including any intersection that may be present. 2. Estimate the annual crash frequency for the 327-ft (100-m) road segment ignoring any inter- section that may be present, including any pedestrian crossing facility at the intersection. 3. Use the value estimated in Step 2 as the estimate for roadway-segment-related crashes. 4. Use the value estimated in Step 1 minus the value estimated in Step 2 as the estimate for intersection-related crashes. If no intersection is present within the 327-ft (100-m) roadway segment, this value will be zero. The categorization of crash types suggested by the previous five bullet items is suitable for use in the HSM because it uses quantities already quantified in the RAP procedures. Thus, the specific procedure to be used in the HSM differs in form from the RAP procedure but is math- ematically equivalent. Bicycle Crashes: The RAP model procedures develop estimates for three specific bicycle-related crash types. These are: • Crashes related to bicycle movements along the road. • Crashes related to bicycles running off the road. • Crashes related to bicycle movements through intersections. The first two bicycle crash types are, by definition, segment-related crashes; the last bicycle crash type, by definition, consists of intersection-related crashes. Separating the crash types in this way is equivalent to the following alternative procedure for getting separate estimates of roadway-segment-related and intersection-related crash estimates: 1. Estimate the annual crash frequency for the 327-ft (100-m) road segment including any intersection that may be present. 2. Estimate the annual crash frequency for the 327-ft (100-m) road segment ignoring any inter- section that may be present, including any bicycle crossing facility at the intersection. 3. Use the value estimated in Step 2 as the estimate for roadway-segment-related crashes. 4. Use the value estimated in Step 1 minus the value estimated in Step 2 as the estimate for intersection-related-crashes. If no intersection is present within the 327-ft (100-m) roadway segment, this value will be zero. The categorization of crash types suggested by the previous three bullet items is suitable for use in the HSM because it uses quantities already quantified in the RAP procedures. Thus, the specific procedure to be used in the HSM differs in form from the RAP procedure but is math- ematically equivalent. The crashes related to bicycle movements along the road and crashes related to bicycle move- ments through intersections represent two different types of bicycle crashes in the RAP models. The other category of bicycle crashes in the RAP models involves crashes related to bicycles running off the road. This category of crashes does not involve both motor vehicles and bicycles, but rather involves crashes in which a bicycle leaves the roadway, shoulder, or bicycle facility and either overturns or strikes a fixed object, resulting in an injury to the bicyclist. The RAP model for this crash type provides the smallest crash frequency estimates and has the most lim- ited supporting research of any of the RAP models considered in this research. For this reason, a decision was made to drop the models for crashes involving bicycles running off the road and limit the HSM2 procedures to models that predict crashes involving both motor vehicles and bicycles.

204 Pedestrian and Bicycle Safety Performance Functions 4.1.4 Use of Explicit Rather Than Assumed Values for Side-Road Attributes The RAP models are constrained to use only the data included in the RAP core dataset that serves as input to RAP’s software tool, ViDA. The only two variables in the core dataset that explicitly apply to the side road or minor road at an intersection for predicting pedestrian and bicycle crashes are: • Intersecting road volume (the AADT range for the side road). • Pedestrian crossing type – side road. The RAP procedures use default logic to assume the number of lanes on the side road. All other relevant variables for the side road are assumed to have the same values as for the major road. Default logic in the RAP models also assumes that the pedestrian flow crossing the side road will be equal to the pedestrian flow along the inspected road, which may not be the case. This is a reasonable assumption in many cases, but, for example, the operating speed on the side road (on which traffic may be slowing to stop at the major road) seems unlikely to be equal to the operating speed of the inspected road (on which traffic may not be required to stop). In an HSM procedure, there is no need to assume roadway attributes for the minor road. Instead, HSM users can be asked to supply data for these attributes explicitly. Side-road data that HSM users will be asked to supply will include: • Intersecting road volume. • Pedestrian crossing type. • Pedestrian crossing flow. • Bicycle facility type and paved shoulder provision. • Bicycle flow – side road. • Advance visibility of an intersection. • Number of lanes to be crossed. • Median type. • Channelization. • Mean operating speed. • Presence of school zone. • Presence of pedestrian fencing. • Presence of vehicle parking. It is recognized that for some of the side-road variables—particularly intersecting road volume, pedestrian crossing flow, and mean operating speed—counts and measurements may not be available, and estimates based on local knowledge may need to be made. 4.1.5 Use of the Terms “Major Road” and “Minor Road” Rather Than “Inspected Road” and “Side Road” for Intersecting Roads The RAP models were developed for application to extended roadway sections divided into 327-ft (100-m) roadway segments. As a result, the intersecting roadways at an at-grade intersec- tion are referred to in the RAP procedures as the “inspected road” and the “side road.” In most (but not necessarily all) cases, the inspected road is the higher volume of the two intersecting roads, which in HSM terms would be referred to as the “major road.” The side road, as designated by RAP, would generally be referenced in the HSM as the “minor road.” In adapting the RAP models to the HSM, the project team refers to the intersecting roadways as the “major road” and the “minor road” consistent with HSM practice. The major road is, by the existing HSM defini- tion, the intersecting road with the higher motor vehicle traffic volume (AADT).

Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology 205   4.1.6 Explicit Estimation of Fatality Counts The RAP models estimate values for pedestrian or bicyclist fatalities separately for each crash type within each 327-ft (100-m) roadway segment and then sum the estimates across all crash types considered in the procedure to obtain an estimate of total pedestrian or bicyclist fatali- ties for the road segment. Finally, a ratio of serious injuries to fatalities is used to convert the pedestrian or bicyclist fatality total to a pedestrian or bicyclist serious injury total and, finally, to a combined total for pedestrian or bicyclist fatalities and serious injuries total. U.S. safety engi- neers may be uncomfortable with the procedure in this form because it is well known that fatality counts are small and, therefore, highly variable. This is not necessarily the case in the context of a prediction model. Nevertheless, to make U.S. users more comfortable with the procedure, the models have been reorganized so that they estimate the combined totals of pedestrian fatal and serious injury crashes and then break this estimate into (1) fatal and serious injury crashes and (2) fatalities and serious injuries rather than the other way around. This involves no substantive change in the model; it is exactly equivalent mathematically to the current RAP procedure but may be better accepted by U.S. users. 4.1.7 Inclusion of All Injury Severity Levels The RAP models make estimates for fatalities and serious injuries. Serious injuries are defined as crashes in Levels 3, 4, and 5 on the Modified Abbreviated Injury Scale (MAIS), which corre- sponds roughly to injury severity Level A and a portion of injury severity Level B on the KABCO injury scale. Most U.S. highway agencies do not have crash data with injury severities assigned based on the MAIS scale. Therefore, the RAP models have been recast to use the KABCO scale. In addition, the models have been calibrated to provide estimates for total fatal-and-injury crashes, with calibrated tables to break down this estimate into fatal injury (K) crashes, suspected serious injury (A) crashes, suspected minor injury (B) crashes, and possible (C) injury crashes. Calibrated tables are also provided to estimate the number of injured persons (pedestrians or bicyclists) for the fatality (K), suspected serious injury (A), suspected minor injury (B), and possible injury (C) severity levels. For pedestrian and bicycle crashes, the estimated value for property-damage-only crashes is zero. 4.1.8 Varying the Lengths of Roadway Segments The RAP roadway segment models use fixed roadway segment lengths of 327-ft (100-m) for all analyses. These roadway segments are not necessarily homogeneous in their attributes, but, since they are relatively short, the convention followed in applying the RAP models is that, for each attribute of interest, whatever value applies over the majority of the 327-ft (100-m) road- way segment is used in the analysis of that segment as a whole. By contrast, the crash prediction models used in the HSM are based on roadway segments that are reasonably homogeneous in their attributes, with lengths that may vary from shorter to much longer than 327 ft (100 m). For application in the HSM2, the RAP models have been adapted to allow the use of variable roadway segment lengths rather than fixed-length segments. Each crash prediction equation that originally applied only to a 327-ft (100-m) roadway segment now includes an L/0.062 factor, where L represents the length of the roadway segment and 0.062 represents the length of a 327-ft (100-m) roadway segment in miles. 4.1.9 Use of Continuous Functions Where Feasible Where values of factors tabulated in the RAP procedures can be expressed as continuous func- tions, those continuous functions will be used in the HSM procedures in place of table look-up

206 Pedestrian and Bicycle Safety Performance Functions values for the midpoint of a range of values. This will avoid the possibility of large step changes in model predictions for small changes in input values, such as AADTs. 4.1.10 Renaming of Risk Factors The effects of specific roadway characteristics on estimated crash totals are represented in the RAP models by tabulated values that are called risk factors. The risk factors are analogous to the crash modification factors (CMFs) that appear in first edition of the HSM Part C procedures. A decision has been reached to rename the CMFs in HSM Part C to “adjustment factors” in the HSM2. Therefore, in this report, the RAP risk factors are referred to as adjustment factors, abbreviated as AFs in equations. 4.1.11 Review and Revision of Adjustment Factors and CMFs The values of selected adjustment factors obtained from the RAP models have been revised based on: • Research results found in the literature review for the current research that are more recent than the sources used in developing the RAP models. • Research results found in the literature review for the current research that appear more suit- able for U.S. conditions than the sources used by iRAP. • Revised AF values that are more consistent with current U.S. practice (e.g., values currently used in the HSM) than the RAP values. In addition, a review of the FHWA CMF Clearinghouse was undertaken, which identified a few CMFs for use as adjustment factors in the procedures developed in this research. Each case in which an adjustment factors has been changed or a new adjustment factor has been added is described in the remainder of this section. 4.1.12 Facility-Type Factors The RAP models for pedestrian and bicycle crashes apply to all facility types. Specific adjust- ment factors account for the effects of variables that distinguish between facility types, such as the number of travel lanes and median type for roadway segments and number of legs and type of traffic control for intersections. Variables in the portion of the models that are analogous to SPFs—such as motor vehicle, pedestrian, and bicycle flows and motor vehicle speeds—also influ- ence the estimated crash predictions and differ substantively between facility types. When the RAP models are applied, they are calibrated to reproduce networkwide crash totals. This approach used in applying the RAP models raised concern for the application of these models in the HSM2. If HSM users were to use the models without calibrating them, the results would not be meaningful. Therefore, a facility-type factor has been introduced to the crash pre- diction procedures with values for each facility type to account for the roadway geometric, traffic control, and traffic operational characteristics that typically differ between facility types. The facility-type factors were quantified by applying the modified versions of the RAP models developed in the current research with typical input data characteristics of each facility type. The facility-type factor was then computed as the ratio of observed annual average crash frequen- cies for 5 recent years for all sites of a given facility type on the California state highway system to the annual crash frequency for that facility type computed with the predictive models for the typical input data characteristics. California data from the FHWA Highway Safety Information System (HSIS) were used to estimate typical traffic volume levels and roadway characteristics for each facility type as well as observed crash history data. The facility-type factors were developed

Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology 207   separately for the prediction of pedestrian and bicycle crashes. This is equivalent to the calibra- tion of HSM models to a single state as has been done for the HSM models and is planned for the HSM2 models. Table 127 shows the typical characteristics used to represent roadway segments for each facil- ity type. Table 128 shows the typical characteristics used to represent intersections for each facility type. The tabulated values for these characteristics were used to derive the values of the facility-type factors in the recommended HSM procedures. These values were chosen to repre- sent typical conditions for each type of roadway segment or intersection. The tabulated values for major- and minor-road AADTs and proportions of intersection approaches with exclusive left- turn lanes were derived from statewide data for the California state highway system. Variables in the recommended HSM procedures that do not appear in Table 127 and Table 128 were assumed to have values equivalent to the nominal or base condition. The values of the facility-type factors determined from the tabulated values are illustrated in the following discussions. Characteristic Roadway Segment Type R2U RMU RMD U2U UMU UMD Segment length (mi) 1.0 1.0 1.0 1.0 1.0 1.0 Number of through lanes 2 4 4 2 4 4 AADT (veh/day) 4,830 6,270 17,950 12,280 20,450 32,161 Pedestrian flow along left side (ped/hr) 6 to 25 6 to 25 6 to 25 6 to 25 6 to 25 6 to 25 Pedestrian flow along right side (ped/hr) 6 to 25 6 to 25 6 to 25 6 to 25 6 to 25 6 to 25 Pedestrian flow crossing the road (ped/hr) 6 to 25 6 to 25 6 to 25 6 to 25 6 to 25 6 to 25 Bicycle flow along the road (bikes/hr) 6 to 25 6 to 25 6 to 25 6 to 25 6 to 25 6 to 25 Mean traffic speed (mph) 55 55 55 35 40 45 Median type Centerline only Centerline only Raised ≥ 3 ft width Centerline only Centerline only Raised ≥ 3 ft width Proportion of length on horizontal curves 0.25 0.25 0.25 0.25 0.25 0.25 Typical horizontal curvature Moderate Moderate Moderate Moderate Moderate Moderate Proportion of length with lighting 0.10 0.10 0.10 0.10 0.10 0.10 Proportion of length with shoulder rumble strips 0.50 0.50 0.50 0.50 0.50 0.50 Proportion of length with an auxiliary lane 0.1 0.1 0.1 0.1 0.1 0.1 Number of midblock pedestrian crossings 1 1 1 1 1 1 Pedestrian crossing facility type Unsignalized marked crossing without refuge NOTE: Roadway segment types: R2U = Rural two-lane, two-way road. RMU = Rural multilane undivided highway. RMD = Rural multilane divided highway (nonfreeway). U2U = Urban two-lane undivided arterial. UMU = Urban multilane undivided arterial. UMD = Urban multilane divided highway (nonfreeway). AADT is for both directions of travel combined. Table 127. Typical roadway segment characteristics assumed in computing facility-type factors.

208 Pedestrian and Bicycle Safety Performance Functions Characteristic Roadway Segment Type R2U RMU RMD U2U UMU UMD Three-Leg Intersection with Minor-Road Stop Control Major-road number of lanes to be crossed 2 4 4 2 4 4 Minor-road number of lanes to be crossed 2 2 2 2 2 2 Major-road AADT (veh/day) 4,980 8,550 15,760 10,470 16,060 27,920 Minor-road AADT (veh/day) 1,650 1,240 500 1,010 1,080 910 Pedestrian flow crossing major road (ped/hr) 6 to 25 6 to 25 6 to 25 6 to 25 6 to 25 6 to 25 Pedestrian flow crossing minor road (ped/hr) 6 to 25 6 to 25 6 to 25 6 to 25 6 to 25 6 to 25 Bicycle flow along each approach (bikes/hr) 6 to 25 6 to 25 6 to 25 6 to 25 6 to 25 6 to 25 Mean traffic speed on major road (mph) 55 55 55 35 40 45 Mean traffic speed on minor road (mph) ≤ 20 ≤ 20 ≤ 20 ≤ 20 ≤ 20 ≤ 20 Median type Centerline only Centerline only Raised ≥ 3 ft width Centerline only Centerline only Raised ≥ 3 ft width Proportion of intersection legs with lighting 0.10 0.10 0.10 0.25 0.25 0.25 Proportion of major-road approaches with marked crossings 0.10 0.10 0.10 0.50 0.50 0.50 Proportion of minor-road approaches with marked crossings 0.05 0.05 0.05 0.50 0.50 0.50 Pedestrian crossing facility type Unsignalized marked crossing without refuge or no marked crossing Proportion of major-road approaches with exclusive left-turn lanes 0.26 0.35 0.69 0.39 0.39 0.72 Proportion of minor-road approaches with exclusive left-turn lanes 0.03 0.05 0.04 0.04 0.04 0.02 Three-Leg Signalized Intersection Major-road number of lanes to be crossed 2 4 4 2 4 4 Minor-road number of lanes to be crossed 2 2 2 2 2 2 Major-road AADT (veh/day) 10,280 16,460 16,700 18,360 18,360 32,400 Minor-road AADT (veh/day) 4,730 4,660 3,670 4,630 4,630 6,490 Pedestrian flow crossing major road (ped/hr) 6 to 25 6 to 25 6 to 25 6 to 25 6 to 25 6 to 25 Pedestrian flow crossing minor road (ped/hr) 6 to 25 6 to 25 6 to 25 6 to 25 6 to 25 6 to 25 Bicycle flow along each approach (bikes/hr) 6 to 25 6 to 25 6 to 25 6 to 25 6 to 25 6 to 25 Mean traffic speed on major road (mph) 55 55 55 35 40 45 Mean traffic speed on minor road (mph) ≤ 20 ≤ 20 ≤ 20 ≤ 20 ≤ 20 ≤ 20 Median type Centerline only Centerline only Raised ≥ 3 ft width Centerline only Centerline only Raised ≥ 3 ft width Proportion of intersection legs with lighting 0.10 0.10 0.10 0.25 0.25 0.25 Table 128. Typical intersection characteristics assumed in computing facility-type factors.

Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology 209   Characteristic Roadway Segment Type R2U RMU RMD U2U UMU UMD Proportion of minor-road approaches with marked crossings 0.05 0.05 0.05 0.50 0.50 0.50 Pedestrian crossing facility type Unsignalized marked crossing without refuge or no marked crossing Proportion of major-road approaches with exclusive left-turn lanes 0.88 1.00 0.87 0.83 0.93 0.89 Proportion of minor-road approaches with exclusive left-turn lanes 0.37 0.06 0.63 0.40 0.40 0.54 Four-Leg Intersection with Minor-Road Stop Control Major-road number of lanes to be crossed 2 4 4 2 4 4 Minor-road number of lanes to be crossed 2 2 2 2 2 2 Major-road AADT (veh/day) 5,410 6,820 12,360 11,170 11,170 25,600 Minor-road AADT (veh/day) 1,200 300 660 2,760 2,760 1,750 Pedestrian flow crossing major road (ped/hr) 6 to 25 6 to 25 6 to 25 6 to 25 6 to 25 6 to 25 Pedestrian flow crossing minor road (ped/hr) 6 to 25 6 to 25 6 to 25 6 to 25 6 to 25 6 to 25 Bicycle flow along each approach (bikes/hr) 6 to 25 6 to 25 6 to 25 6 to 25 6 to 25 6 to 25 Mean traffic speed on major road (mph) 55 55 55 35 40 45 Mean traffic speed on minor road (mph) ≤ 20 ≤ 20 ≤ 20 ≤ 20 ≤ 20 ≤ 20 Median type Centerline only Centerline only Raised ≥ 3 ft width Centerline only Centerline only Raised ≥ 3 ft width Proportion of intersection legs with lighting 0.10 0.10 0.10 0.25 0.25 0.25 Proportion of major-road approaches with marked crossings 0.10 0.10 0.10 0.50 0.50 0.50 Proportion of minor-road approaches with marked crossings 0.05 0.05 0.05 0.50 0.50 0.50 Pedestrian crossing facility type Unsignalized marked crossing without refuge or no marked crossing Proportion of major-road approaches with exclusive left-turn lanes 0.26 0.24 0.76 0.26 0.45 0.78 Proportion of minor-road approaches with exclusive left-turn lanes 0.02 0.08 0.01 0.02 0.02 0.05 Four-Leg Signalized Intersection Major-road number of lanes to be crossed 2 4 4 2 4 4 Minor-road number of lanes to be crossed 2 2 2 2 2 2 Major-road AADT (veh/day) 12,290 12,290 18,000 15,600 21,050 31,700 Minor-road AADT (veh/day) 4,740 4,740 2,680 6,840 9,760 10,360 Proportion of major-road approaches with marked crossings 0.10 0.10 0.10 0.50 0.50 0.50 Table 128. (Continued). (continued on next page)

210 Pedestrian and Bicycle Safety Performance Functions It should be noted that no facility-type factor was developed for roundabouts. In addition, the original RAP models have several adjustment factors to address roundabouts; however, the adjustment factors for roundabouts were based on a limited amount of research and were not considered appropriate for inclusion in the HSM. Therefore, the modified RAP models devel- oped here for use in the HSM cannot be used to estimate the frequency of pedestrian and bicycle crashes at roundabouts. 4.1.13 Calibration Calibration of the pedestrian and bicycle models is done in the RAP process at a network level for all sites in a particular road network. Each road network evaluated is calibrated separately. Most networks considered by the RAP models include a mix of roadway types, so there is no separate calibration by roadway type. HSM analyses are typically performed one site at a time. Calibrations are performed by the highway agency applying the HSM separately for each roadway segment and intersection type under their jurisdiction and comparing the results to observed crash frequencies for the same type of roadway segment or intersection. The calibration procedures used with the RAP models can be converted for application in this manner. Mean traffic speed on major road (mph) 55 55 55 35 40 45 Mean traffic speed on minor road (mph) 55 55 55 30 35 40 Median type Centerline only Centerline only Raised ≥ 3 ft width Centerline only Centerline only Raised ≥ 3 ft width Proportion of intersection legs with lighting 0.10 0.10 0.10 0.25 0.25 0.25 Proportion of major-road approaches with marked crossings 0.10 0.10 0.10 0.50 0.50 0.50 Proportion of minor-road approaches with marked crossings 0.05 0.05 0.05 0.50 0.50 0.50 Pedestrian crossing facility type Unsignalized marked crossing without refuge or no marked crossing Proportion of major-road approaches with exclusive left-turn lanes 0.95 0.90 0.97 0.96 0.79 0.96 Proportion of minor-road approaches with exclusive left-turn lanes 0.48 0.70 0.51 0.70 0.57 0.67 NOTE: R2U = Rural two-lane, two-way road. RMU = Rural multilane undivided highway. RMD = Rural multilane divided highway (nonfreeway). U2U = Urban two-lane undivided arterial. UMU = Urban multilane undivided arterial. UMD = Urban multilane divided highway (nonfreeway). AADT is both directions of travel combined. Characteristic Roadway Segment Type R2U RMU RMD U2U UMU UMD Pedestrian flow crossing major road (ped/hr) 6 to 25 6 to 25 6 to 25 6 to 25 6 to 25 6 to 25 Pedestrian flow crossing minor road (ped/hr) 6 to 25 6 to 25 6 to 25 6 to 25 6 to 25 6 to 25 Bicycle flow along each approach (bikes/hr) 6 to 25 6 to 25 6 to 25 6 to 25 6 to 25 6 to 25 Table 128. (Continued).

Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology 211   As with other HSM models, the modified pedestrian and bicycle crash prediction models, including the facility-type factors discussed above, should be calibrated by user agencies for each facility type of interest. The crash prediction models developed in this research include a variable that can be used to represent the value of the calibration factor(s) developed by each agency. It is the research team’s understanding that the calibration procedures that appeared in the first edition of the HSM will be modified for the HSM2. Therefore, there was no attempt to develop a revised calibration procedure, but it is recommended that the new calibration procedure devel- oped for the HSM2 be applied to the pedestrian and bicycle models as well as the other HSM2 models. All equations, tables, and information apply to all facility types (two-lane, two-way roads; rural multilane highways; and urban and suburban arterials) unless otherwise stated. 4.2 Crash Prediction Models for Pedestrian Crashes The following discussion presents the crash prediction models for pedestrian crashes. The discussion notes differences from the RAP models (other than issues discussed above in Sec- tion 4.1) and differences in crash prediction procedures between facility types. 4.2.1 Predictive Method for Pedestrian Crashes on Roadway Segments The calculations for the predictive method for pedestrian crashes on roadway segments are presented below. 4.2.1.1 General Form of Roadway Segment Model for Pedestrian Crashes The general form of the crash prediction model for pedestrian crashes on an individual road- way segment is as follows: (4-1)C#FT#=N N Npedr alongleft ped alongright ped pedi n pedr pedr1 midcrossing + +- - -=Na k/ where: Npedr = predicted number of pedestrian crashes per year for all crash severity levels combined for a specific roadway segment, Nalongleft–ped = predicted number of pedestrian crashes per year involving pedestrian move- ments along the left side of the road for a specific roadway segment, Nalongright–ped = predicted number of pedestrian crashes per year involving pedestrian move- ments along the right side of the road for a specific roadway segment, Nmidcrossing–ped = predicted number of pedestrian crashes per year involving pedestrians crossing the road at a specific midblock location on a specific roadway segment, FTpedr = facility-type factor for pedestrian crashes for a specific roadway segment facility type, Cpedr = calibration factor for pedestrian crashes for a specific roadway segment facility type, i = index variable that identifies a specific midblock crossing location, and n = maximum number of midblock crossing locations within a specific roadway segment. The terms in Equation 4-1 are discussed in Sections 4.2.1.2 through 4.2.1.6. Equation 4-1 is equivalent to the corresponding RAP model except that the facility-type factor term, FTpedr, is new (see Section 4.1.12), and the calibration factor term, Cpedr, applies specifically to roadway segments of specific types and not to a roadway network as a whole (see Section 4.1.13).

212 Pedestrian and Bicycle Safety Performance Functions 4.2.1.2 Pedestrian Crash Prediction Model for Pedestrian Movements Along the Left Side of the Road On an undivided highway, the left side of the road refers to the roadway and roadside in the sec- ondary or decreasing milepost direction of travel on the roadway being evaluated. On a divided highway, the left side of the road refers to the roadway and median roadside in the direction of travel opposite to the primary direction of travel. The pedestrian crash prediction model for pedestrian movements along the left side of a specific roadway segment is as follows: (4-2) . N Likelihood Severity MVTSF MVTFF PFF L 0 062 alongleft ped alongleft ped alongleft ped alongleft ped alongleft ped alongleft # # # # # =- - - - - J L KK N P OO where: Likelihoodalongleft–ped = crash likelihood factor for pedestrian crashes involving pedestrian move ments along the left side of the road for a specific roadway segment, Severityalongleft–ped = crash severity factor for pedestrian crashes involving pedestrian move- ments along the left side of the road for a specific roadway segment, MVTSFalongleft–ped = motor vehicle traffic speed factor for pedestrian crashes involving pedes- trian movements along the left side of the road for a specific roadway segment, MVTFFalongleft–ped = motor vehicle traffic flow factor for pedestrian crashes involving pedes- trian movements along the left side of the road for a specific roadway segment, PFFalongleft = pedestrian flow factor for pedestrian crashes involving pedestrian move- ments along the left side of the road for a specific roadway segment, and L = length (mi) of a specific roadway segment. The terms in Equation 4-2 are discussed in Sections 4.2.1.7, 4.2.1.10, and 4.2.1.13 through 4.2.1.17. The combination of the MVTSFalongleft–ped, MVTFFalonglef–ped, PFFalongleft, and (L/0.062) terms constitute, in effect, an SPF for crashes related to pedestrian movements along the left side of the road. The Likelihoodalongleft–ped and Severityalongleft–ped terms are combinations of adjustment factors. The inclusion of the (L/0.062) term allows application of the model to roadway segments that vary in length, as explained in Section 4.1.8. 4.2.1.3 Pedestrian Crash Prediction Model for Pedestrian Movements Along the Right Side of the Road On an undivided highway, the right side of the road refers to the roadway and roadside in the primary or increasing milepost direction of travel on the roadway being evaluated. On a divided highway, the right side of the road refers to the roadway and median roadside in the primary direction of travel. The pedestrian crash prediction model for pedestrian movements along the right side of a specific roadway segment is as follows: (4-3) . N MVTFF Likelihood Severity MVTSF PFF L 0 062 along ped along ped alongright ped alongright right alongright ped alongright ped right# # # # # =- - - - - J L KK N P OO where: Likelihoodalongright–ped = crash likelihood factor for pedestrian crashes involving pedestrian movements along the right side of the road for a specific roadway segment,

Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology 213   Severityalongright–ped = crash severity factor for pedestrian crashes involving pedestrian move- ments along the right side of the road for a specific roadway segment, MVTSFalongright–ped = motor vehicle traffic speed factor for pedestrian crashes involving pedestrian movements along the right side of the road for a specific roadway segment, MVTFFalongright–ped = motor vehicle traffic flow factor for pedestrian crashes involving pedes- trian movements along the right side of the road for a specific roadway segment, PFFalongright = pedestrian flow factor for pedestrian crashes involving pedestrian move- ments along the right side of the road for a specific roadway segment, and L = length (mi) of a specific roadway segment. The terms in Equation 4-3 are discussed in Sections 4.2.1.8, 4.2.1.11, and 4.2.1.13 through 4.2.1.17. The combination of the MVTSFalongright–ped, MVTFFalongright–ped, PFFalongright, and (L/0.062) terms constitute, in effect, an SPF for crashes related to pedestrian movements along the right side of the road. The Likelihoodalongright–ped and Severityalongright–ped terms are combinations of adjust- ment factors. The inclusion of the (L/0.062) term allows application of the model to roadway segments that vary in length, as explained in Section 4.1.8. 4.2.1.4 Pedestrian Crash Prediction Model for Pedestrian Crossing Movements at Midblock Locations The pedestrian crash prediction model for pedestrian crossing movements at midblock loca- tions along a specific roadway segment is as follows: (4-4) MVTSF#Likelihood Severity#= PFF# N MVTFF ped ped ped ped ped midcrossing midcrossing midcrossing midcrossing midcrossing midcrossing# - - - - - where: Likelihoodmidcrossing–ped = crash likelihood factor for pedestrian crashes involving pedestrian movements crossing the roadway segment at a specific midblock location, Severitymidcrossing–ped = crash severity factor for pedestrian crashes involving pedestrian move- ments crossing the roadway segment at a specific midblock location, MVTSFmidcrossing–ped = motor vehicle traffic speed factor for pedestrian crashes involving pedes- trian movements crossing the roadway segment at a specific midblock location, MVTFFmidcrossing–ped = motor vehicle traffic flow factor for pedestrian crashes involving pedes- trian movements crossing the roadway segment at a specific midblock location, and PFFmidcrossing = pedestrian flow factor for pedestrian crashes involving pedestrian move- ments crossing the roadway segment at a specific midblock location. The terms in Equation 4-4 are discussed in Sections 4.2.1.9 and 4.2.1.12 through 4.2.1.17. The combination of the MVTSFmidcrossing–ped, MVTFFmidcrossing–ped, and PFFmidcrossing terms constitute, in effect, an SPF for crashes related to pedestrian movements crossing the road at midblock locations. The Likelihoodmidcrossing–ped and Severitymidcrossing–ped terms are combinations of adjustment factors. The predictive model in Equation 4-4 should be applied to every individual location within the roadway segment at which pedestrians cross the road. This should include all marked midblock crossings but may also include unmarked midblock locations at which pedestrians are known or presumed to cross the road.

214 Pedestrian and Bicycle Safety Performance Functions 4.2.1.5 Facility-Type Factor for Pedestrian Crashes on Roadway Segments The values of the facility-type factor (FTpedr) developed in this research for pedestrian crashes on roadway segments are presented in Table 129 by HSM Part C chapter and roadway type. The reason for inclusion of this factor is discussed in Section 4.1.12. 4.2.1.6 Calibration Factor for Pedestrian Crashes on Roadway Segments The calibration process is discussed above in Section 4.1.13. 4.2.1.7 Crash Likelihood Factor for Pedestrian Crashes Related to Pedestrian Movements Along the Left Side of the Road The crash likelihood factor for pedestrian crashes related to pedestrian movements along the left side of the road within a roadway segment is determined as follows: (4-5) .Likelihood AF AF AF AF AF AF AF AF AF AF0 075ped LA ped LA ped LA ped LA ped LA ped LA ped LA ped LA ped LA ped alongleft LA ped 9 2 6 7 8 10 11 12 13 1# # # # # # # # # # =- - - - - - - - - - - where: AFLA1–ped = crash likelihood adjustment factor for pedestrian crashes, accounting for side- walk or paved shoulder provision along a specific roadway segment; AFLA2–ped = crash likelihood adjustment factor for pedestrian crashes, accounting for pres- ence of warning signs in school zones along a specific roadway segment; AFLA6–ped = crash likelihood adjustment factor for pedestrian crashes, accounting for lane width along a specific roadway segment; AFLA7–ped = crash likelihood adjustment factor for pedestrian crashes; accounting for hori- zontal curvature along a specific roadway segment; AFLA8–ped = crash likelihood adjustment factor for pedestrian crashes, accounting for advance visibility of a curve along a specific roadway segment; AFLA9–ped = crash likelihood adjustment factor for pedestrian crashes, accounting for percent grade along a specific roadway segment; AFLA10–ped = crash likelihood adjustment factor for pedestrian crashes, accounting for pres- ence and condition of delineation along a specific roadway segment; AFLA11–ped = crash likelihood adjustment factor for pedestrian crashes, accounting for shoulder rumble strips along a specific roadway segment; AFLA12–ped = crash likelihood adjustment factor for pedestrian crashes, accounting for vehicle parking along a specific roadway segment; and AFLA13–ped = crash likelihood adjustment factor for pedestrian crashes, accounting for street lighting along a specific roadway segment. The procedures for determining the values of these adjustment factors and the inclusion of the 0.075 factor in Equation 4-5 are discussed later in Section 4.2.1.16. HSM Part C Chapter Roadway Type Facility-Type Factor (FTpedr) HSM Chapter 10 (rural two-lane, two-way roads) Two-lane, two-way roads 0.00042 HSM Chapter 11 (rural multilane highways) Multilane undivided 0.00069 Multilane divided 0.00062 HSM Chapter 12 (urban and suburban arterials) Two-lane undivided 0.00583 Multilane undivided 0.00724 Multilane divided 0.00480 Table 129. Facility-type factors for pedestrian crashes on roadway segments.

Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology 215   4.2.1.8 Crash Likelihood Factor for Pedestrian Crashes Related to Pedestrian Movements Along the Right Side of the Road The crash likelihood factor for pedestrian crashes related to pedestrian movements along the right side of the road within a roadway segment is determined as follows: (4-6) .Likelihood AF AF AF AF AF AF AF AF AF AF 0 075ped LA ped LA ped LA ped LA ped LA ped LA ped LA ped LA ped LA ped LA ped alongright 1 2 6 7 8 9 10 11 12 13 # # # # # # # # # # =- - - - - - - - - - - where: AFLA1–ped = crash likelihood adjustment factor for pedestrian crashes, accounting for side- walk or paved shoulder provision along a specific roadway segment; AFLA2–ped = crash likelihood adjustment factor for pedestrian crashes, accounting for pres- ence of warning signs in school zones along a specific roadway segment; AFLA6–ped = crash likelihood adjustment factor for pedestrian crashes, accounting for lane width along a specific roadway segment; AFLA7–ped = crash likelihood adjustment factor for pedestrian crashes, accounting for hori- zontal curvature along a specific roadway segment; AFLA8–ped = crash likelihood adjustment factor for pedestrian crashes, accounting for advance visibility of a curve along a specific roadway segment; AFLA9–ped = crash likelihood adjustment factor for pedestrian crashes, accounting for percent grade along a specific roadway segment; AFLA10–ped = crash likelihood adjustment factor for pedestrian crashes, accounting for pres- ence and condition of delineation along a specific roadway segment; AFLA11–ped = crash likelihood adjustment factor for pedestrian crashes, accounting for shoulder rumble strips along a specific roadway segment; AFLA12–ped = crash likelihood adjustment factor for pedestrian crashes, accounting for vehicle parking along a specific roadway segment; and AFLA13–ped = crash likelihood adjustment factor for pedestrian crashes, accounting for street lighting along a specific roadway segment. The procedures for determining the values of these adjustment factors and the inclusion of the 0.075 factor in Equation 4-6 are discussed below in Section 4.2.1.16. 4.2.1.9 Crash Likelihood Factor for Pedestrian Crashes Related to Midblock Crossing Movements by Pedestrians The crash likelihood factor for pedestrian crashes related to pedestrian crossing the road at a midblock location within a roadway segment is determined as follows: (4-7) Likelihood AF AF AF AF AF AF AF AF ped L ped L ped L ped L ped L ped L ped L ped L ped M M M M midcrossing M M M M 1 1 2 3 5 12 13 4 5 4# # # # # # # =- - - - - - - - - where: AFLM2–ped = crash likelihood adjustment factor for pedestrian crashes, accounting for school zone warning at a specific midblock location; AFLM3–ped = crash likelihood adjustment factor for pedestrian crashes, accounting for pedes- trian crossing facility type at a specific midblock location; AFLM4–ped = crash likelihood adjustment factor for pedestrian crashes, accounting for advance visibility of a pedestrian crossing at a specific midblock location;

216 Pedestrian and Bicycle Safety Performance Functions AFLM5–ped = crash likelihood adjustment factor for pedestrian crashes, accounting for pedes- trian fencing at a specific midblock location; AFLM12–ped = crash likelihood adjustment factor for pedestrian crashes, accounting for vehicle parking at a specific midblock location; AFLM13–ped = crash likelihood adjustment factor for pedestrian crashes, accounting for street lighting at a specific midblock location; AFLM14–ped = crash likelihood adjustment factor for pedestrian crashes, accounting for number of traffic lanes to be crossed at a specific midblock location; and AFLM15–ped = crash likelihood adjustment factor for pedestrian crashes, accounting for median type at a specific midblock location. The procedures for determining the values of these adjustment factors are presented in Sec- tion 4.2.1.16. 4.2.1.10 Crash Severity Factor for Pedestrian Crashes Related to Pedestrian Movements Along the Left Side of the Road The crash severity factor for pedestrian crashes related to pedestrian movements along the left side of the road is determined as follows: (4-8)AF=Severityalongleft ped SA ped1- - where: AFSA1–ped = crash severity adjustment factor for pedestrian crashes, accounting for sidewalk or paved shoulder provision along the road. The value of AFSA1–ped for pedestrian movements along the left side of the road should be based on sidewalk or paved shoulder provision along the left side of the road. The values for the adjust- ment factor, AFSA1–ped, are presented in Section 4.2.1.17. 4.2.1.11 Crash Severity Factor for Pedestrian Crashes Related to Pedestrian Movements Along the Right Side of the Road The crash severity factor for pedestrian crashes related to pedestrian movements along the right side of the road is determined as follows: (4-9)AF=Severityalongright ped SA ped1- - where: AFSA1–ped = crash severity adjustment factor for pedestrian crashes, accounting for sidewalk or paved shoulder provision along the road. The value of AFSA1–ped for pedestrian movements along the right side of the road should be based on sidewalk or paved shoulder provision along the right side of the road. The values for the adjustment factor, AFSA1–ped, are presented in Section 4.2.1.17. 4.2.1.12 Crash Severity Factor for Pedestrian Crashes Related to Midblock Crossing Movements by Pedestrians The crash severity factor for pedestrian crashes related to pedestrian crossing the road at mid- block locations is determined as follows: (4-10)AF=Severity ped SM ped3midcrossing- -

Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology 217   where: AFSM3–ped = crash severity adjustment factor for pedestrian crashes, accounting for pedestrian crossing facility type at a specific midblock location. The values for the adjustment factor, AFSM3–ped, are presented in Section 4.2.1.17. 4.2.1.13 Motor Vehicle Traffic Speed Factors for Pedestrian Crashes on Roadway Segments The value of the motor vehicle traffic speed factor for pedestrian crashes, MVTSFpedr, is used to evaluate pedestrian movements along the roadway and pedestrian movements crossing the road at midblock locations. Thus, the values of MVTSFpedr from Table 130, based on the appropriate speed value, are used as values for MVTSFalongleft–ped, MVTSFalongright–ped, and MVTSFmidcrossing–ped. The values shown in Table 130 are illustrated in Figure 48. For speeds below 50 mph, the motor vehicle traffic speed factors have a cubic relationship based on the power curve developed in Sweden (Nilsson 2004). Above 50 mph, the relationship is linear. The mean speed of motor vehicle traffic along the specific roadway segment of interest should be used to determine the values of MVTSFalongleft–ped, MVTSFalongright–ped, and MVTSFmidcrossing–ped. Mean speed should be determined from field measurements of traffic speed on the road segment of interest, field measurements on similar road segments, or the best available estimate based on local knowledge. 4.2.1.14 Motor Vehicle Traffic Flow Factors for Pedestrian Crashes on Roadway Segments The value of the motor vehicle traffic flow factor used to evaluate pedestrian movements along the road or pedestrian movements crossing the road at midblock locations, MVTFFpedr, is deter- mined as follows: (4-11).1 0=. , then MVTFF1 0$if MVTFF= , MVTFF AADT 20 000 pedr lanes pedr pedr N` `j j Mean Speed of Motor Vehicle Traffic on a Specific Roadway S Motor Vehicle Traffic Speed Factor pedr 20 or less 0.017 25 0.051 30 0.105 35 0.183 40 0.290 45 0.429 50 0.536 55 0.590 60 0.643 65 0.697 70 0.751 75 0.804 80 0.858 85 0.912 90 or more 0.966 NOTE: Values of MVTSFpedr based on the appropriate speed are used as values for MVTSFalongleft–ped, MVTSFalongright–ped, or MVTSFmidcrossing-–ped. Table 130. Motor vehicle traffic speed factor for pedestrian crashes on roadway segments (iRAP 2013m).

218 Pedestrian and Bicycle Safety Performance Functions where: MVTFFpedr = motor vehicle traffic flow factor for pedestrian crashes on a specific roadway segment, AADT = annual average daily volume (veh/day) for motor vehicles for both directions of travel combined on a specific roadway segment or intersection approach, and Nlanes = number of travel lanes for through traffic in both directions of travel combined on a specific roadway segment or intersection approach. The motor vehicle traffic flow factors for pedestrian crashes are derived by iRAP (2013a) from the work of Turner, Roozenburg, and Francis (2006). The values of MVTFFpedr determined with Equation 4-11 for the applicable AADT value are used as values for MVTFFalongleft–ped, MVTFFalongright–ped, and MVTFFmidcrossing–ped. Representative values of MVTFFpedr are illustrated in Figure 49. 4.2.1.15 Pedestrian Flow Factors on Roadway Segments The value of the pedestrian flow factor is determined from Table 131. The values shown in Table 131 are illustrated in Figure 50. The PFFr values for pedestrian movements along the road and crossing the road at a midblock location are identical for each level of pedestrian peak-hour flow along or crossing the road. The pedestrian peak-hour flow along the road may be based on pedestrian volume counts or the best available estimate of the peak-hour pedestrian volumes along the road based on local knowledge. While peak-hour pedestrian flow is used in determining the value of the pedestrian flow factor, the model predictions represent annual total pedestrian crash frequencies. Figure 48. Graph of motor vehicle traffic speed factor for pedestrian crashes along the road representing the relative frequency of pedestrian injuries as a function of traffic speed (adapted from iRAP 2013m).

Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology 219   Figure 49. Graph of motor vehicle traffic flow factor as a function of AADT for motor vehicles and number of lanes. Pedestrian Peak-Hour Flow for a Specific Side of the Road or Crossing Location Pedestrian Flow Factor r None 0.000 1 to 5 0.001 6 to 25 0.002 26 to 50 0.002 51 to 100 0.003 101 to 200 0.004 201 to 300 0.005 301 to 400 0.006 401 to 500 0.006 501 to 900 0.007 more than 900 0.009 NOTE: Values of PFFr for the applicable peak-hour pedestrian volume are used to determine values for PFFalongleft, PFFalongright, and PFFmidcrossing. Table 131. Pedestrian flow factors (iRAP 2013a).

220 Pedestrian and Bicycle Safety Performance Functions If the pedestrian flow along a segment or across a given midblock crossing is zero, then zero pedestrian crashes will be predicted by the model. In this case, the pedestrian flow factor can be set to zero, and no further analysis is needed for crashes related to that pedestrian movement. 4.2.1.16 Crash Likelihood Adjustment Factors for Pedestrian Crashes on Roadway Segments This section presents tables to determine the values of the crash likelihood adjustment factors used in Equations 4-5 through 4-7. Crash likelihood adjustment factors for pedestrian crashes related to pedestrian movements along the road have subscripts that begin with the letters LA. Crash likelihood adjustment factors for pedestrian crashes related to pedestrian movements crossing the road at a midblock location have subscripts that begin with the letters LM. AFLA1–ped 2 Sidewalk or Paved Shoulder Provision Sidewalks and paved shoulders reduce the likelihood of pedestrian crashes by providing pedestrians a place to walk outside of the traveled way. The derivation of the crash likelihood adjustment factors for sidewalk or paved shoulder provision (AFLA1–ped) for pedestrian move- ments along the road is illustrated in Table 132 and Table 133 for specific categories of sidewalk and paved shoulder provision. Table 132 shows the table used in the original RAP procedure to represent the effect of side- walk or paved shoulder provision. The table in this form does not appear suitable for use in HSM2 because the table does not have a specific base condition (i.e., an adjustment factor whose value is equal to 1.0). This contrasts to every other crash likelihood adjustment factor that does have one specific value that serves as a base condition. To address this issue, every value of the AF in Table 132 was multiplied by 13.33 to create Table 133, which resulted in a value of 1.0 as a value of AFLA1–ped for a sidewalk with greater than 10 ft separation from the main road with no Figure 50. Graph of pedestrian flow factor for pedestrian crashes as a function of pedestrian peak-hour flow along or crossing the road.

Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology 221   barrier present. To compensate for this change in the values of AFLA1–ped, a constant factor equal to 1/13.33 = 0.075 has been inserted into Equations 4-5 and 4-6. The value of AFLA1–ped for pedestrian movements along the left side of the road is based on a sidewalk or paved shoulder provision along the left side of the road. The value of AFLA1–ped for pedestrian movements along the right side of the road is based on a sidewalk or paved shoulder provision along the right side of the road. The term “physical barrier” in Table 132 and Table 133 refers to a traffic barrier designed so that motor vehicles will be redirected or stopped, rather than passing through the barrier, and installed by a highway agency for that purpose. Physical barriers include steel guardrails, concrete barriers, and cable barriers (as long as the cable barrier would not deflect onto the sidewalk). The term “traveled way” refers to the portion of the roadway intended for through travel by motor vehicles. The rationale for the AF values for a sidewalk and paved shoulder provision is presented in iRAP Road Attribute Risk Factors: Sidewalk Provision (iRAP 2013q). Sidewalks provide the greatest benefits, which vary with the presence of a physical barrier between the sidewalk and the traveled way and the distance from the edge of the traveled way to the sidewalk. The presence of paved shoulders of various widths provides a benefit to pedestrians but a much smaller benefit than sidewalks. Any informal path (i.e., a path created by repeated pedestrian usage, as opposed Sidewalk or Paved Shoulder Provision Adjustment Factor Physical barrier between sidewalk and traveled way 0.00 Sidewalk with > 10 ft separation from traveled way with no barrier present 1.00 Sidewalk with > 3 ft separation from traveled way with no barrier present 1.20 Sidewalk adjacent to traveled way (within 3 ft) 1.33 Paved shoulder present with width ≥ 7.9 ft 186.67 Paved shoulder present with width ≥ 3 ft and < 7.9 ft 200.00 Paved shoulder present with width < 3 ft 240.00 None 266.67 Informal path with > 3 ft separation from road with no barrier 66.67 Informal path with ≤ 3 ft separation from road with no barrier 80.00 NOTE: For roadway segments with unpaved shoulders, use the adjustment factors for informal paths. Choose the appropriate adjustment factor based on whether the unpaved shoulder is ≥ 3 ft wide. Table 133. Crash likelihood adjustment factors for pedestrian crashes along the road, accounting for sidewalk or paved shoulder provision, modified for use in HSM2 (adapted from iRAP 2013q). Sidewalk P Crash Likelihood Risk Factor for Pedestrian Movements Along the Left or Right Side of the Road from the Original RAP Procedure Physical barrier between sidewalk and traveled way 0.000 Sidewalk with > 10 ft separation from traveled way with no barrier present 0.075 Sidewalk with > 3 ft separation from traveled way with no barrier present 0.090 Sidewalk adjacent to traveled way (within 3 ft) 0.100 Paved shoulder present with width ≥ 7.9 ft 14.000 Paved shoulder present with width ≥ 3 ft and ≤ 7.9 ft 15.000 Paved shoulder present with width < 3 ft 18.000 None 20.000 Informal path with ≥ 3 ft separation from road with no barrier 5.000 Informal path with ≥ 3 ft separation from road with no barrier 6.000 Table 132. Likelihood risk factors for sidewalk provision from the original RAP procedure (iRAP 2013q).

222 Pedestrian and Bicycle Safety Performance Functions to a pedestrian facility created by a public agency) provides a benefit to pedestrians between that of a sidewalk and a paved shoulder. The adjustment factors for informal paths may also be used for unpaved shoulders. In the absence of any detailed studies in the literature, iRAP developed these adjustment factors from the first principles of crash initialization considering the likely proportions of pedestrians using the sidewalk or paved shoulder facility provided as opposed to walking in the road, and the relative crash likelihood of using the sidewalk facility provided and walking in the road (iRAP 2013q). Sources used in estimating the AF values include Lynam (2012); Hills, Baguley, and Kirk (2002); Elvik and Vaa (2004); S. Turner, Binder, and Roozenburg (2009); McMahon et al. (2002); Oxley, Corben, and Fildes (2004); Tobey, Shunamen, and Knoblauch (1983); and Harwood et al. (2008). AFLA2–ped and AFLM2–ped 2 School Zone Warning School zone warnings may include flashing beacons or other active warnings, advance sign- ing, messages marked on the pavement surface, and advisory speed limits. The crash likelihood adjustment factors for pedestrian crashes, accounting for school zone warning, for pedestrian movements along the road and crossing the road (AFLA2–ped and AFLM2–ped) are shown in Table 134. The rationale for the adjustment factor values is presented in iRAP Road Attribute Risk Factors: School Zone Warning (iRAP 2013o), based on a literature review by Mead, Zegeer, and Bushell (2013). AFLM3–ped 2 Pedestrian Crossing Facility Type The crash likelihood adjustment factors for pedestrian crashes, accounting for provision of pedestrian crossing facilities of specific types, for pedestrian crossing movements at midblock locations are shown in Table 135. Pedestrian fencing may consist of a conventional fence or other features, such as a landscaping barrier, that prevents pedestrians from entering the traveled way, except at designated points. The rationale for the adjustment factor values is presented in iRAP Road Attribute Risk Factors: Pedestrian Crossing Facilities (iRAP 2014b). A grade-separated crossing facility is assumed to be 100 percent effective in reducing pedestrian crashes only where pedestrian fencing is present to prevent at-grade crossing of the road by pedestrians. The effectiveness of pedestrian refuge areas in the center of a crossing is based on the work of B. Turner et al. (2009). B. Turner et al. (2009) also recommended a factor for the difference in crash likelihood between a signalized crossing and no crossing facility at all. The effectiveness of marked crossings without a refuge is based on a literature review by B. Turner et al. (2012). The effectiveness of school zone crossing guards is based on a literature review by Mead et al. (2013) and consultation by iRAP with other experts. The adjustment factor for a signalized crossing with a leading pedestrian phase is based on work from Goughnour et al. (2018). The adjustment factor for crossings with pedestrian hybrid beacons is a combination of the iRAP adjustment factors for unsignalized marked crossings and the research by Zegeer et al. (2017). School Zone Warning Adjustment Factor School zone flashing beacons or other active warnings present 0.90 School zone static signs or road markings present 0.95 School zone with no school zone warning present 1.00 Not applicable (no school zone at this location) 1.00 Table 134. Crash likelihood adjustment factors for pedestrian crashes on the road, accounting for school zone warning (iRAP 2013o).

Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology 223   AFLM4–ped 2 Advance Visibility of a Pedestrian Crossing Advance visibility of a pedestrian crossing represents an assessment of the ability of approach- ing drivers to see a pedestrian crossing on the roadway ahead under daytime conditions. Advance visibility of a pedestrian crossing considers pavement markings in advance of and at the cross- ing, advance signing, flashing beacons, and sight distance to the crossing. If pavement markings, advance signing, flashing beacons, and/or sight distance to the crossing are such that an approach- ing driver, under daytime conditions, can readily see that a pedestrian crossing is present, advance visibility to a pedestrian crossing should be rated as “substantial.” If pavement markings, advance signing, and/or flashing beacons have lost their reflectivity and/or legibility (e.g., are weathered or faded) or are absent and sight distance is such that an approaching driver is likely to be unaware of the presence of the crossing, then advance visibility of a pedestrian crossing should be rated as “limited.” The crash likelihood adjustment factors for pedestrian crashes, accounting for advance visibility of a pedestrian crossing at midblock locations, are shown in Table 136. The rationale for the adjustment factor values is presented in iRAP Road Attribute Risk Factors: Pedestrian Crossing Quality (2014c). Work by B. Turner et al. (2009) suggested that, where a pedestrian crossing is difficult to see, the likelihood of pedestrian crashes is equivalent to not having a pedestrian crossing facility at all and may be worse in some cases. Pedestrian Crossing Facility Type Adjustment F Locations Other than Schools School Locations with a Crossing Guard School Locations Without a Crossing Guard Grade-separated facility – pedestrian fencing present 0.00 0.00 0.00 Grade-separated facility 0.40 0.30 0.40 Signalized with leading pedestrian phasea 0.85 N/A 0.85 Signalized with refuge 1.00 0.95 1.00 Signalized without refuge 1.25 1.20 1.25 Unsignalized marked crossing with refuge and pedestrian hybrid beacon 1.75 N/A N/A Unsignalized marked crossing without refuge but with pedestrian hybrid beacon 2.15 N/A N/A Unsignalized raised, marked crossing with refugea 2.50 1.00 2.50 Unsignalized raised, marked crossing without refugea 3.20 1.00 3.20 Unsignalized marked crossing with refuge 3.80 1.00 3.80 Unsignalized marked crossing without refuge 4.80 1.25 4.80 Refuge only 5.10 3.80 5.10 No facility 6.70 4.80 6.70 NOTE: N/A = Not applicable. aUsed for urban and suburban arterials only (HSM Chapter 12). Table 135. Crash likelihood adjustment factors for pedestrian crashes at midblock locations, accounting for pedestrian crossing facility types (iRAP 2014b; Zegeer et al. 2017). Advance Visibility of a Pedestrian Crossing Adjustment Factor Substantial 1.00 Limited 1.50 Not applicable (no crossing present) 1.00 Table 136. Crash likelihood adjustment factors for pedestrian crashes at midblock locations, accounting for advance visibility of a pedestrian crossing (iRAP 2014c).

224 Pedestrian and Bicycle Safety Performance Functions AFLM5–ped 2 Pedestrian Fencing Pedestrian fencing can reduce pedestrian crashes by denying pedestrians access to the road- way for crossing movements except at specific crossing facilities. Pedestrian fencing may consist of a conventional fence or other feature, such as a landscaping barrier, that prevents pedestrians from entering the traveled way, except at designated points. The crash likelihood adjustment factors for pedestrian crashes, accounting for continuous pedestrian fencing for the full length of a roadway segment, fencing that limits pedestrian crossing to marked crossing facilities at midblock locations, and absence of pedestrian fencing are shown in Table 137. The rationale for the adjustment factor values is presented in iRAP Road Attribute Risk Factors: Pedestrian Fencing (iRAP 2013m). These factors are based on a review of literature by B. Turner et al. (2012) based on previous studies by Teale (1984), Stewart (1988), Campbell et al. (2004), and Elvik et al. (2009). AFLA6–ped 2 Lane Width As lane widths decrease, motor vehicle drivers may have greater difficulty keeping within their lane. The greater variability of vehicle lateral position and the potential for lane departures increases the likelihood of a collision with a pedestrian. The crash likelihood adjustment factors for pedestrian crashes, accounting for lane width, for pedestrian movements along the road are shown in Table 138. The value of AFLA6–ped for pedestrian movements along the left side of the road should be based on the lane width along the left side of the road. The value of AFLA6–ped for pedestrian movements along the right side of the road should be based on lane width along the right side of the road. The rationale for the adjustment factor values is presented in iRAP Road Attribute Risk Factors: Lane Width (iRAP 2013i). These adjustment factors are based on the work of B. Turner et al. (2009). AFLA7–ped 2 Horizontal Curvature The likelihood that motor vehicles will run off the road is higher on horizontal curves than on tangents and increases as the radius of curvature decreases. As the likelihood that motor vehicles will run off the road increases, so does the likelihood that an errant motor vehicle will strike a pedestrian. The crash likelihood adjustment factors for pedestrian crashes, accounting Pedestrian Fencing Category Description Adjustment Factor Full length of roadway segment Pedestrian fencing present with no crossing facility or grade-separated crossing facility 0.00 At pedestrian crossing Pedestrian fencing and pedestrian crossing both present 1.00 None No pedestrian fencing present 1.25 Table 137. Crash likelihood adjustment factors for pedestrian crashes at midblock locations, accounting for pedestrian fencing (iRAP 2013m). Lane Width Adjustment Factor Wide (≥ 10.6 ft) 1.00 Medium (≥ 9 to < 10.6 ft) 1.20 Narrow (< 9 ft) 1.50 Table 138. Crash likelihood adjustment factors for pedestrian crashes along the road, accounting for lane width (iRAP 2013i).

Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology 225   for horizontal curvature, for pedestrian movements along the road are shown in Table 139. The horizontal curvature categories are defined by advisory speed ranges and corresponding ranges of horizontal curve radius. If a horizontal curve is signed with an advisory speed plate, use of the category in Table 139 corresponding to the signed advisory speed is recommended. If there is no signed advisory speed, the curvature category should be based on the horizontal curve radius. AFLA8–ped 2 Advance Visibility of a Curve Advance visibility of a curve represents an assessment of the ability of approaching drivers to see a horizontal curve on the roadway ahead. Advance visibility of a curve considers pave- ment markings, chevron markers, advance signing, and sight distance to the curve. If pavement markings, chevron markers, advance signing, and sight distance to the curve are such that an approaching driver can readily see that a horizontal curve is present, advance visibility of a curve should be rated as “substantial.” If pavement markings, chevron markers, and advance signing have lost their reflectivity and/or legibility (e.g., are weathered or faded) or are absent and sight distance to the curve is such that an approaching driver is likely to be unaware of the presence of the curve, then advance visibility of a curve should be rated as “limited.” If the advance visibility of a curve is limited, motor vehicles are more likely to run off the road on the curve, and such vehicles are, therefore, more likely to strike a pedestrian. The crash likelihood adjustment factors for pedestrian crashes, accounting for advance visibility of a curve, for pedestrian movements along the road are shown in Table 140. The value of AFLA8–ped for pedestrian movements along the left side of the road should be based on the ability of approaching drivers to see a horizontal curve while driving along the left side of the road. The value of AFLA8–ped for pedestrian move- ments along the right side of the road should be based on the ability of approaching drivers to see a horizontal curve while driving along the right side of the road. AFLA9–ped 2 Percent Grade Motor vehicles are more likely to run off the road on steep grades than on level roadway sec- tions. Motor vehicles that run off the road may strike a pedestrian. The crash likelihood adjust- ment factors for pedestrian crashes, accounting for grade, for pedestrian movements along the road are shown in Table 141. Horizontal Curvature Adjustment Factor Straight or gently curving (advisory speed ≥ 60 mph or curve radius > 2,600 ft) 1.00 Moderate curvature (advisory speed in the range from 45 mph to < 60 mph or curve radius in the range from 1,300 ft to ≤ 2,600 ft) 1.81 Sharp curve (advisory speed in the range from 25 mph to < 45 mph or curve radius in the range from 650 ft to ≤ 1,300 ft) 3.51 Very sharp curve (advisory speed < 25 mph or curve radius < 650 ft) 6.02 Table 139. Crash likelihood adjustment factors for pedestrian crashes along the road, accounting for horizontal curvature (iRAP 2013c). Advance Visibility of a Curve Adjustment Factor Substantial 1.00 Limited 1.25 Not applicable (no horizontal curve present) 1.00 Table 140. Crash likelihood adjustment factors for pedestrian crashes along the road, accounting for advance visibility of a curve (iRAP 2013n).

226 Pedestrian and Bicycle Safety Performance Functions AFLA10–ped 2 Presence and Condition of Delineation Presence and condition of delineation involves the placement of pavement markings, delineators on the roadside or the roadway surface, or other devices purposely placed by a highway agency to help guide drivers along the roadway. Motor vehicles are more likely to run off the road where delineation has lost its reflectivity (e.g., is weathered or faded) or is absent than where delineation is clearly visible to the driver. Motor vehicles that run off the road may strike a pedestrian. The crash likelihood adjustment factors for pedestrian crashes, accounting for the presence and condi- tion of delineation, for pedestrian movements along the road are shown in Table 142. The value of AFLA10–ped for pedestrian movements along the left side of the road should be based on delineation to help guide drivers along the left side of the road. The value of AFLA10–ped for pedestrian move- ments along the right side of the road should be based on delineation to help guide drivers along the right side of the road. The rationale for the adjustment factor values is presented in iRAP Road Attribute Risk Factors: Delineation (iRAP 2013d). The adjustment factor for presence and condition of delineation was based primarily on work by B. Turner et al. (2009). AFLA11–ped 2 Shoulder Rumble Strips Shoulder rumble strips are placed on the edgeline or shoulder of a roadway to alert a driver that their vehicle is leaving the roadway. Since shoulder rumble strips reduce the likelihood that motor vehicles will run off the road, they also reduce the likelihood that errant vehicles will strike pedestrians. The crash likelihood adjustment factors for pedestrian crashes, accounting for shoulder rumble strips, for pedestrian movements along the road are shown in Table 143. The value of AFLA11–ped for pedestrian movements along the left side of the road should be based on the presence or absence of shoulder rumble strips along the left side of the road. The value of AFLA11–ped for pedestrian movements along the right side of the road should be based on the presence or absence of shoulder rumble strips along the right side of the road. Percent Grade Adjustment Factor 0% to < 7.5% 1.00 7.5% to < 10% 1.20 ≥ 10% 1.70 Table 141. Crash likelihood adjustment factors for pedestrian crashes along the road, accounting for percent grade (iRAP 2013e). Presence and Condition of Delineation Adjustment Factor Clearly visible 1.00 Loss of reflectivity (e.g., weathered or faded) or absent 1.20 Table 142. Crash likelihood adjustment factors for pedestrian crashes along the road, accounting for presence and condition of delineation (iRAP 2013d). Shoulder Rumble Strips Adjustment Factor Not present 1.25 Present 1.00 Table 143. Crash likelihood adjustment factors for pedestrian crashes along the road, accounting for shoulder rumble strips (iRAP 2013p).

Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology 227   The rationale for the adjustment factor values is presented in iRAP Road Attribute Risk Factors: Shoulder Rumble Strips (iRAP 2013p). These adjustment factors are based primarily on literature reviewed by B. Turner et al. (2012) and B. Turner et al. (2009). AFLA12–ped and AFLM12–ped 2 Vehicle Parking Vehicle parking along a roadway where no sidewalk is present may increase the likelihood that pedestrians will walk in the traveled way rather than on the roadside or shoulder. If a sidewalk is present but is obstructed by vehicle parking, pedestrians are also more likely to walk in the trav- eled way. Pedestrians walking in the traveled way are more likely to be struck by motor vehicles than pedestrians walking on the roadside or shoulder. The crash likelihood adjustment factors for pedestrian crashes along the road, accounting for vehicle parking (AFLA12–ped), are shown in Table 144. If a sidewalk is present but is regularly obstructed by parked vehicles, proceed as if the pedes- trian facility were not present. The presence of vehicle parking may also increase the frequency of crashes for pedestrians crossing a roadway. Pedestrians emerging from between parked vehicles to cross the roadway are more likely to be struck by moving vehicles than pedestrians crossing the road where no parking is present. The crash likelihood adjustment factors for pedestrian crashes at midblock locations, accounting for vehicle parking (AFLM12–ped), are shown in Table 145. AFLA13–ped and AFLM13–ped 2 Street Lighting The presence of street lighting makes pedestrians walking along the roadway, crossing the roadway, or about to cross the roadway more visible to motor vehicle drivers, thus reducing the likelihood that pedestrians walking along the road or crossing the road will be struck by motor vehicles. The crash likelihood adjustment factors for pedestrian crashes on the road, accounting for street lighting, are shown in Table 146. The rationale for the adjustment factor values is presented in iRAP Road Attribute Risk Factors: Street Lighting (iRAP 2013r). These adjustment factors were based on combined results from a literature review by B. Turner et al. (2012) and studies summarized in the FHWA CMF Vehicle Parking Adjustment Factor None 1.00 One side 1.20 Two sides 1.33 One side (sidewalk present) 1.00 Two sides (sidewalk present) 1.00 Table 144. Crash likelihood adjustment factors for pedestrian crashes along the road, accounting for vehicle parking (iRAP 2013s). Vehicle Parking Adjustment Factor None 1.00 One side 1.20 Two sides 1.33 Table 145. Crash likelihood adjustment factors for pedestrian crashes at midblock locations, accounting for vehicle parking (iRAP 2013s).

228 Pedestrian and Bicycle Safety Performance Functions Clearinghouse (www.cmfclearinghouse.org). Results based on nighttime crashes only were reduced to represent a combination of daytime and nighttime crashes. AFLM14–ped 2 Number of Traffic Lanes The likelihood of pedestrian crashes increases as the number of lanes to be crossed by the pedestrian, and therefore the width of the roadway, increases. The crash likelihood adjustment factors for pedestrian crashes at midblock locations, accounting for number of traffic lanes to be crossed (AFLM14–ped), are shown in Table 147. For pedestrian crossing movements at midblock locations, both through traffic lanes and auxiliary lanes should be considered. While roadways with more than three lanes are not normally present on two-lane, two-way roads, the adjustment factor values for four, five, and six lanes have been retained for use in HSM Chapter 10 to address short roadway segments where additional lanes may be present and pedestrian crossings where auxiliary lanes are present. The rationale for the adjustment factor values is presented in iRAP Road Attribute Risk Factors: Number of Through Lanes (iRAP 2013k), based on research by Corben, Logan, and Oxley (2008). AFLM15–ped 2 Median Type The presence of a median on a road decreases the likelihood of crashes involving pedestrians crossing the road because the median serves as a refuge area that enables pedestrians to make two-stage crossing maneuvers. Where a median is present, pedestrians can wait in the median until a suitable gap in traffic becomes available through which to cross. By contrast, where no median is present, pedestrians must cross the roadway in a single stage. The term “physical median” refers to a raised or depressed median that provides a suitable refuge for pedestrians crossing the roadway. A paved area in the center of the road that is flush with the traveled way pavement surface is not a physical median, although in some cases it might be appropriately clas- sified as a continuous central turning lane or central hatching. The crash likelihood adjustment factors for pedestrian crashes at midblock locations, accounting for median type, are shown in Table 148. While physical medians are not normally used on two-lane, two-way roads, the adjustment factor values for physical medians have been retained for use in HSM Chapter 10 to address short roadway segments where medians may be present. Number of Traffic Lanes for Both Directions of Travel Adjustment Factor Two lanes 1.00 Three lanes 1.80 Four lanes 2.80 Five lanes 4.00 Six lanes 5.20 Seven lanesa 6.60 Eight lanes a 8.00 NOTE: The number of traffic lanes includes both travel lanes and auxiliary lanes crossed by the pedestrian. aNot applicable to rural two-lane, two-way roads (HSM Chapter 10). Table 147. Crash likelihood adjustment factors for pedestrian crashes at midblock locations, accounting for number of traffic lanes to be crossed (iRAP 2013k). Table 146. Crash likelihood adjustment factors for pedestrian crashes on the road, accounting for street lighting (iRAP 2013r). Street Lighting Adjustment Factor Present 1.00 Not present 1.25

Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology 229   4.2.1.17 Crash Severity Adjustment Factors for Pedestrian Crashes on Roadway Segments This section presents the values of the crash severity factors used in Equations 4-8 through 4-10. AFSA1–ped 2 Sidewalk and Paved Shoulder Provision The crash severity adjustment factors for pedestrian crashes along the road, accounting for sidewalk and paved shoulder provision (AFSA1–ped), are shown in Table 149. The value of AFSA1–ped for pedestrian movements along the left side of the road should be based on sidewalk or paved shoulder provision along the left side of the road. The value of AFSA1–ped for pedestrian move- ments along the right side of the road should be based on sidewalk or paved shoulder provision along the right side of the road. AFSM3–ped 2 Pedestrian Crossing Facility Type The crash severity adjustment factors for pedestrian crashes at midblock locations, accounting for pedestrian crossing facility type (AFSM3–ped), are shown in Table 150. 4.2.1.18 Computation of Final Prediction of Pedestrian Crashes on Roadway Segments The total predicted pedestrian crash frequency for all crash severity levels combined for a specific roadway segment during a specific year is Npedr determined with Equation 4-1. This logic is new as the RAP models addressed fatalities and serious injuries only, while the models pre- sented here address four crash severity levels (fatal crashes, A-injury crashes, B-injury crashes, and C-injury crashes) and equivalent numbers of pedestrians fatally injured and injured at the A-, B-, and C-injury levels. Median Type Adjustment Factor Physical median with traffic barrier 1.00 Physical median width ≥ 3 ft 1.00 Continuous central turning lane 3.00 Centerline rumble strip (or flexipost) 2.70 Central hatching or other flush separation (> 3 ft) 2.40 Centerline only (no median) 3.00 Wide centerline (1 to 3 ft) 2.70 Table 148. Crash likelihood adjustment factors for pedestrian crashes at midblock locations, accounting for median type (iRAP 2013j). Sidewalk or Paved Shoulder Provision Adjustment Factor Physical barrier between sidewalk and traveled way 90 Sidewalk with > 10 ft separation from traveled way with no barrier present 90 Sidewalk with > 3 ft separation from traveled way with no barrier present 90 Sidewalk adjacent to the traveled way (within 3 ft) 90 Paved shoulder present with width ≥ 7.9 ft 90 Paved shoulder present with width ≥ 3 ft and < 7.9 ft 90 Paved shoulder present with width < 3 ft 90 None 90 Informal path with > 3 ft separation from road with no barrier 90 Informal path with ≤ 3 ft separation from road with no barrier 90 NOTE: For roadway segments with unpaved shoulders, use the adjustment factors for informal paths. Choose the appropriate adjustment factor based on whether the unpaved shoulder is less than, greater than, or equal to 3 ft wide. Table 149. Crash severity adjustment factors for pedestrian crashes along the road, accounting for sidewalk and paved shoulder provision.

230 Pedestrian and Bicycle Safety Performance Functions Npedr can be divided into individual crash severity levels as follows, where the crash severity level is defined by the most severe pedestrian injury in the crash: (4-12)P#N=N pedr pedr Kpedr K- - (4-13)P#N=N pedr pedrA Apedr- - (4-14)P#N=N pedr pedrB Bpedr- - (4-15)P#N=N pedr pedrC Cpedr- - where: Npedr = predicted number of pedestrian crashes per year for all crash severity levels com- bined for a specific roadway segment, Npedr–K = predicted number of fatal pedestrian crashes per year for a specific roadway segment, Npedr–A = predicted number of A-injury pedestrian crashes per year for a specific roadway segment, Npedr–B = predicted number of B-injury pedestrian crashes per year for a specific roadway segment, Npedr–C = predicted number of C-injury pedestrian crashes per year for a specific roadway segment, Ppedr–K = proportion of fatal pedestrian crashes for specific roadway segment facility types (see Table 151), Ppedr–A = proportion of A-injury pedestrian crashes for specific roadway segment facility types (see Table 151), Ppedr–B = proportion of B-injury pedestrian crashes for specific roadway segment facility types (see Table 151), and Ppedr–C = proportion of C-injury pedestrian crashes for specific roadway segment facility types (see Table 151). The predicted number of injured pedestrians per crash is 1.05 for all roadway segment facility types. Pedestrian Crossing Facility Type Adjustment Factor Grade-separated facility – pedestrian fencing present 90 Grade-separated facility 90 Signalized with leading pedestrian phasea 90 Signalized with refuge 90 Signalized without refuge 90 Unsignalized marked crossing with refuge and pedestrian hybrid beacon 90 Unsignalized marked crossing without refuge but with pedestrian hybrid beacon 90 Unsignalized raised, marked crossing with refugea 90 Unsignalized raised, marked crossing without refugea 90 Unsignalized marked crossing with refuge 90 Unsignalized marked crossing without refuge 90 Refuge only 90 No crossing facility 90 aUsed for urban and suburban arterials only (HSM Chapter 12). Table 150. Crash severity adjustment factors for pedestrian crashes at midblock locations, accounting for pedestrian crossing facility type.

Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology 231   4.2.1.19 Comparison of Predictive Method for Pedestrian Crashes on Roadway Segments with Existing HSM Models Following development of the pedestrian roadway segment model, compatibility testing of the new model was conducted to check the reasonableness of the results across facility types. To gain a better understanding of the potential use of the model within the HSM, output results from the new model were compared to output from existing models in HSM Part C. Note, for these comparisons, the new model and the existing HSM models were not calibrated using a single agency’s data. Unlike most of the existing models in the HSM, pedestrian exposure is accounted for in the new model, so some differences are expected. Nonetheless, output results from the new model were compared to output from existing models in the HSM to gain a sense of the reasonableness of the new pedestrian roadway segment model for application and potential incorporation in HSM2. Figure 51 provides a comparison of the predicted average total pedestrian crash frequency for rural two-lane undivided roads from the pedestrian roadway segment model presented in this section and the predicted average total pedestrian crash frequency from the existing HSM Part C model for rural two-lane undivided roads (i.e., HSM Chapter 10). The predicted average crash frequencies are for the base conditions for both models. Comparisons from the pedestrian roadway segment model are shown for two pedestrian peak-hour volume levels: 1 to 5 ped/hr and 101 to 200 ped/hr. Two midblock crossing locations with crossing volumes of 26 to 50 ped/hr (peak hour) were also assumed within the segment. To provide more perspective, Figure 52 pro- vides the same comparison but also includes total crashes (i.e., multiple-vehicle + single-vehicle + pedestrian + bicycle crashes) from the existing HSM Part C model for rural two-lane undivided roads (i.e., HSM Chapter 10). Based on this assessment, the new pedestrian roadway segment model provides similar results to the existing HSM Chapter 10 model for rural two-lane undivided roads. Thus, the new pedes- trian roadway segment model appears compatible with the existing Part C model for rural two- lane undivided roads in HSM Chapter 10. Figure 53 provides a comparison of the predicted average total pedestrian crash frequency for rural multilane undivided roads (e.g., four-lane undivided) from the pedestrian roadway seg- ment model presented in this section and the predicted average crash frequency for other col- lisions (all severity levels combined) from the existing HSM Part C, Chapter 11 model for rural multilane undivided roads. In Chapter 11, estimates for pedestrian crash frequencies are not directly provided but are included as part of “Other Collisions.” The predicted average crash fre- quencies are for the base conditions for both models. Comparisons from the pedestrian roadway segment model are shown for two pedestrian peak-hour volume levels: 1 to 5 ped/hr and 101 to HSM Chapter Roadway Type Proportion of Pedestrian Crashes by Most Severe Pedestrian Injury Fatal A-Injury B-Injury C-Injury HSM Chapter 10 (rural two-lane, two-way roads) Two-lane, two-way roads 0.306 0.226 0.242 0.226 HSM Chapter 11 (rural multilane highways) Multilane undivided and divided 0.111 0.111 0.556 0.222 HSM Chapter 12 (urban and suburban arterials) Two-lane undivided 0.193 0.250 0.335 0.222 Multilane undivided 0.107 0.280 0.360 0.253 Multilane divided 0.187 0.241 0.331 0.235 Table 151. Proportion of pedestrian crashes by roadway segment facility type and injury severity level.

232 Pedestrian and Bicycle Safety Performance Functions Figure 51. Comparison of predicted average total pedestrian crashes per year from new pedestrian roadway segment model and existing HSM Part C, Chapter 10 model for rural two-lane undivided roads. Figure 52. Comparison of predicted average total pedestrian crashes per year from new pedestrian roadway segment model and existing HSM Part C, Chapter 10 model for rural two-lane undivided roads (2U) (including total crashes from HSM model). 200 ped/hr. Two midblock crossing locations with crossing volumes of 26 to 50 ped/hr (peak hour) were also assumed within the segment. To provide more perspective, Figure 54 provides the same comparison but also includes total crashes (i.e., all crash types combined) from the existing HSM Part C, Chapter 11 model for rural multilane undivided roads. Based on this assessment for rural multilane undivided roads (e.g., four-lane undivided), the new pedestrian roadway segment model predicts fewer pedestrian crashes than the estimates for

Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology 233   Figure 53. Comparison of predicted average total pedestrian crashes per year from new pedestrian roadway segment model and existing HSM Part C, Chapter 11 model for rural four-lane undivided roads (4U). Figure 54. Comparison of predicted average total pedestrian crashes per year from new pedestrian roadway segment model and existing HSM Part C, Chapter 11 model for rural four-lane undivided roads (4U) (including total crashes from HSM model). “Other Collisions” from the existing HSM Part C, Chapter 11 model for rural multilane undi- vided roads, which is to be expected. Thus, the new pedestrian roadway segment model appears compatible with the existing HSM Part C, Chapter 11 model for rural multilane undivided roads. Figure 55 provides a comparison of the predicted average total pedestrian crash frequency for rural multilane divided roads (e.g., four-lane undivided) from the pedestrian roadway segment model presented in this section and the predicted average crash frequency for other collisions

234 Pedestrian and Bicycle Safety Performance Functions Figure 56. Comparison of predicted average total pedestrian crashes per year from pedestrian roadway segment model and existing HSM Part C, Chapter 11 model for rural four-lane divided roads (4D) (including total crashes from HSM model). Figure 55. Comparison of predicted average total pedestrian crashes per year from the new pedestrian roadway segment model and existing HSM Part C, Chapter 11 model for rural four-lane divided roads (4D). (all severity levels combined) from the existing HSM Part C, Chapter 11 model for rural multi- lane divided roads. The predicted average crash frequencies are for the base conditions for both models. Comparisons from the pedestrian roadway segment model are shown for two pedes- trian peak-hour volume levels: 1 to 5 ped/hr and 101 to 200 ped/hr. Two midblock crossing locations with crossing volumes of 26 to 50 ped/hr (peak hour) were also assumed within the segment. To provide more perspective, Figure 56 provides the same comparison but also includes

Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology 235   total crashes (i.e., all crash types combined) from the existing HSM Part C, Chapter 11 model for rural multilane divided roads. Based on this assessment for rural multilane divided roads (e.g., four-lane undivided), the new pedestrian roadway segment model predicts fewer pedestrian crashes than the estimates for “Other Collisions” from the existing HSM Part C model for rural multilane divided roads, which is to be expected. Thus, the new pedestrian roadway segment model appears compatible with the existing HSM Part C, Chapter 11 model for rural multilane divided roads. Similar comparisons of the new pedestrian roadway segment model were performed for urban two-lane undivided roads, urban four-lane undivided roads, and urban four-lane divided roads, and in all cases the new pedestrian roadway segment model appears compatible with the existing models for urban and suburban roadway segments in HSM Part C, Chapter 12. Based on the compatibility testing, it appears that with calibration, the new pedestrian road- way segment model is compatible with the existing HSM Part C roadway segment models. 4.2.2 Predictive Method for Pedestrian Crashes at Intersections The calculations for the predictive method for pedestrian crashes at intersections are pre- sented below. 4.2.2.1 General Form of Intersection Model for Pedestrian Crashes The general form of the crash prediction model for pedestrian crashes at an intersection is based on the sum of the predicted crashes for crossing each intersection leg, as follows: (4-16)= Cintcrossing #ped j- - FT#N Npedi j nl pedi pedi1= b l/ where: Npedi = predicted number of pedestrian crashes per year for all crash severity levels combined for a specific intersection, Nintcrossing–ped–j = predicted number of pedestrian crashes per year involving pedestrian move- ments crossing a specific intersection leg j, FTpedi = facility-type factor for pedestrian crashes for a specific intersection type, Cpedi = calibration factor for pedestrian crashes for a specific intersection type, j = index variable that identifies a specific leg for a specific intersection, and nl = maximum number of legs for a specific intersection. The terms in Equation 4-16 are discussed in Sections 4.2.2.2 through 4.2.2.4. Equation 4-16 is equivalent to the corresponding RAP model except that the facility-type factor term, FTpedi, is new (see Section 4.1.12), and the calibration factor term, Cpedi, applies specifically to intersections of specific types and not to a roadway network as a whole (see Section 4.1.13). 4.2.2.2 Pedestrian Crash Prediction Model for Pedestrians Crossing an Intersection Leg The pedestrian crash prediction model for pedestrian movements crossing a specific intersec- tion leg is as follows: (4-17) Likelihood Severity MVTSF MVTFF PFF intcrossing intcrossing intcrossing intcrossing intcrossing intcrossing # # # # =N ped j ped j ped j ped j ped j j - - - - - - - - - - -

236 Pedestrian and Bicycle Safety Performance Functions where: Likelihoodintcrossing–ped–j = crash likelihood factor for pedestrian crashes involving pedestrian movements crossing a specific intersection leg j, Severityintcrossing–ped–j = crash severity factor for pedestrian crashes involving pedestrian movements crossing a specific intersection leg j, MVTSFintcrossing–ped–j = motor vehicle traffic speed factor for pedestrian crashes involving pedestrian movements crossing a specific intersection leg j, MVTFFintcrossing–ped–j = motor vehicle traffic flow factor for pedestrian crashes involving pedestrian movements crossing a specific intersection leg j, and PFFintcrossing–j = pedestrian flow factor for pedestrian crashes involving pedestrian movements crossing a specific intersection leg j. The terms in Equation 4-17 are discussed in Sections 4.2.2.5 through 4.2.2.9. The combination of the MVTSFintcrossing–ped–j, MVTFFintcrossing–ped–j, and PFFintcrossing–j terms constitute, in effect, an SPF for crashes related to pedestrian movements crossing the road on a specific intersection leg j. The Likelihoodintcrossing–ped–j and Severityintcrossing–ped–j terms are combinations of adjustment factors. 4.2.2.3 Facility-Type Factor for Pedestrian Crashes at Intersections The values of the facility-type factor (FTpedi) developed in the research for pedestrian crashes at intersections are presented in Table 152 by HSM chapter and intersection type. The reason for inclusion of this factor is discussed in Section 4.1.12. 4.2.2.4 Calibration Factor for Pedestrian Crashes at Intersections The calibration process is discussed in Section 4.1.13. 4.2.2.5 Crash Likelihood Factor for Pedestrian Crashes Involving Pedestrians Crossing an Intersection Leg The crash likelihood factor for pedestrian crashes involving pedestrian movements crossing a specific intersection leg has subscripts that begin with LI and is determined as follows: (4-18) Likelihood AF AF AF AF AF AF AF AF AF LI LI LI LI LI LI LI LI LI 2 3 4 5 2 3 14 15 16 1 1 intcrossing # # # # # # # # =ped j ped ped ped ped ped ped ped ped ped - - - - - - - - - - - HSM Part C Chapter Intersection Type Facility-Type F HSM Chapter 10 (rural two-lane, two-way roads) 3ST, 3SG 0.00167 4ST, 4SG 0.00366 HSM Chapter 11 (rural multilane highways) 3ST, 3SG 0.00478 4ST, 4SG 0.00316 HSM Chapter 12 (urban and suburban arterials) 3ST 0.09550 3SG 0.14700 4ST 0.09570 4SG 0.03870 NOTE: 3ST = Three-leg intersection with minor-road stop control. 3SG = Three-leg signalized intersection. 4ST = Four-leg intersection with minor-road stop control. 4SG = Four-leg signalized intersection. Table 152. Facility-type factors for pedestrian crashes at intersections.

Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology 237   where: AFLI2–ped = crash likelihood adjustment factor for pedestrian crashes on a specific intersection leg, accounting for school zone warning; AFLI3–ped = crash likelihood adjustment factor for pedestrian crashes on a specific intersection leg, accounting for pedestrian crossing facility type; AFLI4–ped = crash likelihood adjustment factor for pedestrian crashes on a specific intersection leg, accounting for advance visibility of a pedestrian crossing; AFLI5–ped = crash likelihood adjustment factor for pedestrian crashes on a specific intersection leg, accounting for pedestrian fencing; AFLI12–ped = crash likelihood adjustment factor for pedestrian crashes on a specific intersection leg, accounting for vehicle parking; AFLI13–ped = crash likelihood adjustment factor for pedestrian crashes on a specific intersection leg, accounting for street lighting; AFLI14–ped = crash likelihood adjustment factor for pedestrian crashes on a specific intersection leg, accounting for number of lanes to be crossed; AFLI15–ped = crash likelihood adjustment factor for pedestrian crashes for a specific intersection leg, accounting for median type; and AFLI16–ped = crash likelihood adjustment factor for pedestrian crashes for a specific intersection leg, accounting for intersection type. The procedures for determining the values of these adjustment factors are presented in Section 4.2.2.10. 4.2.2.6 Crash Severity Factor for Pedestrian Crashes Involving Pedestrians Crossing an Intersection Leg The crash severity factor for pedestrian movements crossing a specific intersection leg is deter- mined as follows: (4-19)Severity AFped SI ped3intcrossing =- - where: AFSI3–ped = crash severity adjustment factor for pedestrian crashes on a specific intersection leg, accounting for pedestrian crossing facility type. The values for the adjustment factor AFSI3–ped are presented in Section 4.2.2.11. 4.2.2.7 Motor Vehicle Traffic Speed Factors for Pedestrian Crashes Involving Pedestrians Crossing an Intersection Leg The value of the motor vehicle traffic speed factor, MVTSFintcrossing–ped–j, is determined from Table 130 using the value of MVTSFpedr for the appropriate speed for the roadway segment that includes the intersection leg. For major-road legs at an intersection, use the major-road mean traffic speed (e.g., the speed used to determine MVTSFpedr for the roadway segment that contains the intersection leg) to determine MVTSFintcrossing–ped–j from Table 130, unless stop-sign control is present on the major road; in the latter case, the speed used to determine MVTSFintcrossing–ped–j from Table 130 should be 20 mph or less. For minor-road legs at an intersection, use the best avail- able estimate of the mean traffic speed on the minor road to determine MVTSFintcrossing–ped–j from Table 130. If stop-sign control is present on the minor road or if the majority of minor-road traffic makes a turning maneuver onto the major road, the speed used to determine MVTSFintcrossing–ped–j from Table 130 should be 20 mph or less.

238 Pedestrian and Bicycle Safety Performance Functions 4.2.2.8 Motor Vehicle Traffic Flow Factors for Pedestrian Crashes Involving Pedestrians Crossing an Intersection Leg The value of the motor vehicle traffic flow factor used to evaluate pedestrian movements cross- ing an intersection leg, MVTFFintcrossing–ped–j is determined in the same manner as MVTFFalongleft–ped and MVTFFalongright–ped for the roadway segment containing the intersection leg, using Equation 4-11. Thus, at intersections, the value of MVTFFintcrossing–ped–j for major-road approaches should be set equal to the value of MVTFFalongleft–ped or MVTFFalongright–ped based on the AADT for the road segment that contains the major-road approach. The value of MVTFFintcrossing–ped–j for minor-road approaches should be set equal to the value of MVTFFalongleft–ped or MVTFFalongright–ped based on the AADT for the road segment that contains the minor-road approach. 4.2.2.9 Pedestrian Flow Factors for Pedestrian Crashes Involving Pedestrians Crossing an Intersection Leg The value of the pedestrian flow factor should be determined from Table 131. The value of the pedestrian flow factor for the pedestrian movement crossing an intersection leg, PFFintcrossing–ped–j, should be determined from Table 131 as equal to the value of PFFr for the peak-hour pedestrian flow crossing the road. The peak-hour pedestrian flow crossing an intersection leg may be based on pedestrian volume counts or the best available estimate of the peak-hour pedestrian volumes crossing an inter- section leg based on local knowledge. While peak-hour pedestrian flow is used in determining the value of the pedestrian flow factor, the model predictions represent annual total pedestrian crash frequencies. If the pedestrian flow crossing of a given intersection leg is zero, then zero pedestrian crashes will be predicted by the model for that pedestrian crossing movement. In this case, the pedestrian flow factor can be set to zero, and no further analysis is needed for crashes related to that pedes- trian movement. 4.2.2.10 Crash Likelihood Adjustment Factors for Pedestrian Crashes Involving Pedestrians Crossing an Intersection Leg This section presents tables to determine the values of the crash likelihood adjustment factors used in Equation 4-18. Crash likelihood adjustment factors for pedestrian crashes involving pedestrians crossing an intersection leg have subscripts that begin with the letters LI. In all the adjustment factor tables, if more than one category of adjustment factor applies, use the adjust- ment factor that appears first in the table. AFLI2–ped 2 School Zone Warning School zone warnings may include flashing beacons or other active warnings, advance sign- ing, messages marked on the pavement surface, and advisory speed limits. The adjustment factor for pedestrian crashes involving pedestrian movements crossing intersection legs, accounting for the effect of school zone warnings (AFLI2–ped), is determined from Table 134 in the same manner as the adjustment factor applicable to pedestrian crossing movements at midblock crossings, AFLM2–ped. AFLI3–ped 2 Pedestrian Crossing Facility Type The adjustment factor for pedestrian crashes involving pedestrian movements crossing inter- section legs, accounting for the effect of pedestrian crossing facility type (AFLI3–ped), is determined from Table 135 in the same manner applicable to pedestrian crossing movements at midblock crossings, AFLM3–ped.

Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology 239   AFLI4–ped 2 Advance Visibility of a Pedestrian Crossing Advance visibility of a pedestrian crossing represents an assessment of the ability of approach- ing drivers to see a pedestrian crossing at the intersection ahead. Selecting the appropriate adjust- ment factor for advance visibility of a pedestrian crossing involves an assessment made from a street-level photograph or a field visit as to whether the pedestrian crossing is visible to and likely to be seen by a driver approaching the pedestrian crossing. Advance visibility of a pedestrian crossing considers pavement markings and signing in advance of and at the crossing and sight distance to the crossing. If pavement markings, signing, and sight distance to the crossing are such that an approaching driver can readily see that a pedestrian crossing is present, advance vis- ibility of a pedestrian crossing should be rated as substantial. If pavement markings and signing have lost their reflectivity and/or legibility (e.g., are weathered or faded) or are absent and sight distance is such that an approaching driver is likely to be unaware of the presence of the crossing, then advance visibility of a pedestrian crossing should be rated as limited. The adjustment factor for pedestrian crashes involving pedestrian movements crossing intersection legs, accounting for the effect of advance visibility of a pedestrian crossing (AFLI4–ped), is determined from Table 136 in the same manner as the adjustment factor applicable to pedestrian crossing movements at midblock crossings, AFLM4–ped. AFLI5–ped 2 Pedestrian Fencing Pedestrian fencing can reduce pedestrian crashes by denying pedestrians access to the inter- section for crossing movements except at specific crossing facilities. Pedestrian fencing may con- sist of a conventional fence or other feature, such as a landscaping barrier, that prevents pedestrians from entering the traveled way, except at designated points. The adjustment factor for pedestrian crashes involving pedestrian movements crossing intersection legs, accounting for the effect of pedestrian fencing (AFLI5–ped), is determined from Table 137 in the same manner as the adjustment factor applicable to pedestrian crossing movements at midblock crossings, AFLM5–ped. AFLI12–ped 2 Vehicle Parking The presence of vehicle parking may increase the frequency of pedestrian crashes for pedes- trians crossing an intersection leg. Pedestrians emerging from between parked vehicles to cross the roadway are more likely to be struck by moving vehicles than pedestrians crossing the road where no parking is present. The adjustment factor for pedestrian crashes involving pedestrian movements crossing intersection legs, accounting for the effect of vehicle parking (AFLI12–ped), is determined from Table 145 in the same manner as the adjustment factor applicable to pedestrian crossing movements at midblock crossings, AFLM12–ped. AFLI13–ped 2 Street Lighting The presence of street lighting makes pedestrians walking along the roadway, crossing the roadway, or about to cross the roadway more visible to motor vehicle drivers, thus reducing the likelihood that pedestrians crossing an intersection leg will be struck by motor vehicles. The adjustment factor for pedestrian crashes involving pedestrian movements crossing intersection legs, accounting for the effect of street lighting (AFLI13–ped), is determined from Table 146 in the same manner as the adjustment factor applicable to pedestrian crossing movements at midblock crossings, AFLM13–ped. AFLI14–ped 2 Number of Traffic Lanes The likelihood of pedestrian crashes increases as the number of lanes to be crossed by the pedestrian, and therefore the width of the roadway, increases. The adjustment factor for pedestrian crashes involving pedestrian movements crossing intersection legs, accounting for the effect of the number of traffic lanes to be crossed (AFLI14–ped), is determined from Table 147 in the same

240 Pedestrian and Bicycle Safety Performance Functions manner as the adjustment factor applicable to pedestrian crossing movements at midblock cross- ings, AFLM14–ped. For pedestrian crossing movements at intersections, both through and auxiliary lanes should be considered. While roadways with more than three lanes are not normally present on two-lane, two-way roads, the adjustment factor values for four to six lanes have been retained for use in HSM Chapter 10 to address intersection crossings where, because of short sections of added through lanes or the presence of auxiliary lanes, up to six lanes may need to be crossed. AFLI15–ped 2 Median Type The presence of a median at an intersection decreases the likelihood of crashes involving pedestrians crossing an intersection leg because the median can serve as a refuge area that enables pedestrians to make two-stage crossing maneuvers. Where a median is present, pedestrians can wait in the median until a suitable gap in traffic becomes available through which to cross. By contrast, where no median is present, pedestrians must cross the roadway in a single stage. The adjustment factor for pedestrian crashes involving pedestrian movements crossing intersection legs, accounting for the effect of median type (AFLI15–ped), is determined from Table 148 in the same manner as the adjustment factor applicable to pedestrian crossing movements at midblock crossings, AFLM15–ped. AFLI16–ped 2 Intersection Type Intersection-type factors for consideration include the number of intersection legs, the pres- ence or absence of traffic signals, and the presence or absence of exclusive left-turn lanes. The adjustment factor for pedestrian crashes involving pedestrian movements crossing intersection legs, accounting for intersection type (AFLI16–ped), is shown in Table 153. 4.2.2.11 Crash Severity Adjustment Factors for Pedestrian Crashes Involving Pedestrians Crossing an Intersection Leg AFSI3–ped 2 Pedestrian Crossing Facility Type There is only one crash severity adjustment factor for pedestrian crashes involving pedestrians crossing intersection legs, AFSI3–ped. The adjustment factor, AFSI3–ped, is determined from Table 150 in the same manner as the adjustment factor applicable to pedestrian crossing movements at midblock crossings, AFSM3–ped. 4.2.2.12 Computation of Final Prediction of Pedestrian Crashes at Intersections The total predicted pedestrian crash frequency for all crash severity levels combined for a specific intersection during a specific year is Npedi determined with Equation 4-16. Npedi can be Intersection Type Adjustment Factor Three-leg unsignalized with exclusive left-turn lane 1.10 Three-leg unsignalized with no exclusive left-turn lane 1.10 Three-leg signalized with exclusive left-turn lane 1.10 Three-leg signalized with no exclusive left-turn lane 1.10 Four-leg unsignalized with exclusive left-turn lane 1.20 Four-leg unsignalized with no exclusive left-turn lane 1.20 Four-leg signalized with exclusive left-turn lane 1.20 Four-leg signalized with no exclusive left-turn lane 1.20 Table 153. Crash likelihood adjustment factors for pedestrian crashes involving pedestrian movements crossing intersection legs, accounting for intersection type (iRAP 2013h).

Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology 241   divided into individual crash severity levels as follows, where the crash severity level is defined by the most severe pedestrian injury in the crash: (4-20)P#N=Nped K ped ped Ki i i- - (4-21)P#N=Nped ped pedi A i i A- - (4-22)P#N=Nped ped pedi B i i B- - (4-23)P#N=Nped ped pedi C i i C- - where: Npedi = predicted number of pedestrian crashes per year for all crash severity levels com- bined for a specific intersection, Npedi–K = predicted number of fatal pedestrian crashes per year for a specific intersection, Npedi–A = predicted number of A-injury pedestrian crashes per year for a specific intersection, Npedi–B = predicted number of B-injury pedestrian crashes per year for a specific intersection, Npedi–C = predicted number of C-injury pedestrian crashes per year for a specific intersection, Ppedi–K = proportion of fatal pedestrian crashes for specific intersection facility types (see Table 154), Ppedi–A = proportion of A-injury pedestrian crashes for specific intersection facility types (see Table 154), Ppedi–B = proportion of B-injury pedestrian crashes for specific intersection facility types (see Table 154), and Ppedi–C = proportion of C-injury pedestrian crashes for specific intersection facility types (see Table 154). The predicted number of injured pedestrians per crash is 1.21 for all facility types. HSM Chapter Road Type Intersection Type Proportion of Pedestrian Crashes by Most Severe Pedestrian Injury Fatal A-Injury B-Injury C-Injury HSM Chapter 10 (rural two-lane, two- way roads) Two-lane, two-way roads 3ST, 3SG 0.125 0.375 0.250 0.250 4ST, 4SG 0.076 0.154 0.462 0.308 HSM Chapter 11 (rural multilane highways) Multilane undivided and divided 3ST, 3SG 0.031 0.031 0.313 0.625 4ST, 4SG 0.045 0.455 0.455 0.045 HSM Chapter 12 (urban and suburban arterials) Two-lane undivided 3ST 0.193 0.250 0.335 0.222 Two-lane undivided 3SG 0.142 0.286 0.286 0.286 Two-lane undivided 4ST 0.167 0.200 0.267 0.366 Two-lane undivided 4SG 0.034 0.138 0.449 0.379 Multilane undivided 3ST 0.138 0.138 0.431 0.293 Multilane undivided 3SG 0.142 0.286 0.286 0.286 Multilane undivided 4ST 0.059 0.118 0.618 0.206 Multilane undivided 4SG 0.138 0.138 0.431 0.293 Multilane divided 3ST 0.051 0.154 0.436 0.359 Multilane divided 3SG 0.030 0.258 0.364 0.348 Multilane divided 4ST 0.036 0.234 0.376 0.325 Multilane divided 4SG 0.041 0.164 0.448 0.347 NOTE: 3ST = Three-leg intersection with minor-road stop control. 3SG = Three-leg signalized intersection. 4ST = Four-leg intersection with minor-road stop control. 4SG = Four-leg signalized intersection. Table 154. Proportion of pedestrian crashes by intersection type and injury severity level.

242 Pedestrian and Bicycle Safety Performance Functions 4.2.2.13 Comparison of Predictive Method for Pedestrian Crashes at Intersections with Existing HSM Models Following development of the pedestrian intersection model, compatibility testing of the new model was conducted to check the reasonableness of the results across intersection types. To gain a better understanding of the potential use of the model within the HSM, output results from the new model were compared to output from existing models in HSM Part C. Note, for these comparisons, the new model and the existing HSM models were not calibrated using a single agency’s data; and unlike most of the existing models in the HSM, pedestrian exposure is accounted for in the new model so some differences are expected. Nonetheless, output results from the new model were compared to output from existing models in the HSM to gain a sense of the reasonableness of the new pedestrian intersection model for application and potential incorporation in the HSM. Figure 57 provides a comparison of the predicted average total pedestrian crash frequency for three-leg stop control intersections on rural two-lane roads from the pedestrian intersection model presented in this section and the predicted average total pedestrian crash frequency from the existing HSM Part C, Chapter 10 model for three-leg stop control intersections on rural two-lane roads. The predicted average crash frequencies are for the base conditions for both models. Comparisons from the pedestrian intersection model are shown for two pedestrian peak-hour crossing volume levels: 1 to 5 ped/hr and 101 to 200 ped/hr. Two levels of minor- road traffic volumes were also assumed for the comparison: 1,000 and 3,000 veh/day. To provide more perspective, Figure 58 provides the same comparison but also includes estimates for total crashes (i.e., multiple-vehicle + single-vehicle + pedestrian + bicycle crashes) from the existing HSM Part C, Chapter 10 model for three-leg stop control intersections on rural two-lane roads. Based on this assessment, the new pedestrian intersection model provides similar estimates compared to the existing HSM Part C, Chapter 10 model for three-leg stop control intersections Figure 57. Comparison of predicted average total pedestrian crashes per year from the new pedestrian intersection model and existing HSM Part C, Chapter 10 model for three-leg stop control intersections (3ST) on rural two-lane undivided roads.

Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology 243   Figure 58. Comparison of predicted average total pedestrian crashes per year from the new pedestrian intersection model and existing HSM Part C, Chapter 10 model for three-leg stop control intersections (3ST) on rural two-lane undivided roads (including total crashes from HSM model). on rural two-lane roads. Thus, the new pedestrian intersection model appears compatible with the existing Chapter 10 model for three-leg stop control intersections on rural two-lane roads. Figure 59 provides a comparison of the predicted average total pedestrian crash frequency for four-leg signal control intersections on rural two-lane roads from the pedestrian intersection model presented in this section and the predicted average total pedestrian crash frequency from the existing HSM Chapter 10 model for four-leg signal control intersections on rural two-lane roads. The predicted average crash frequencies are for the base conditions for both models. Comparisons from the pedestrian intersection model are shown for two pedestrian peak-hour crossing volume levels: 26 to 50 ped/hr and 201 to 300 ped/hr. Two levels of minor-road traf- fic volumes were also assumed for the comparison: 4,000 and 10,000 veh/day. To provide more perspective, Figure 60 provides the same comparison but also includes estimates for total crashes (i.e., multiple-vehicle + single-vehicle + pedestrian + bicycle crashes) from the existing HSM Chapter 10 model for four-leg signal control intersections on rural two-lane roads. Based on this assessment, the new pedestrian intersection model provides similar estimates compared to the existing HSM Chapter 10 model for four-leg signal control intersections on rural two-lane roads. Thus, the new pedestrian intersection model appears compatible with the existing HSM Chapter 10 model for four-leg signal control intersections on rural two-lane roads. Figure 61 provides a comparison of the predicted average total pedestrian crash frequency for four-leg stop control intersections on urban and suburban arterials from the pedestrian intersec- tion model presented in this section and the predicted average total pedestrian crash frequency from the existing HSM Part C, Chapter 12 model for four-leg signal control intersections. The predicted average crash frequencies are for the base conditions for both models. Comparisons from the pedestrian intersection model are shown for two pedestrian peak-hour crossing volume levels: 26 to 50 ped/hr and 201 to 300 ped/hr. Two levels of daily pedestrian volumes crossing all

244 Pedestrian and Bicycle Safety Performance Functions Figure 59. Comparison of predicted average total pedestrian crashes per year from the new pedestrian intersection model and existing HSM Part C, Chapter 10 model for four-leg signal control intersections (4SG) on rural two-lane undivided roads. Figure 60. Comparison of predicted average pedestrian crashes per year from the new pedestrian intersection model and existing HSM Part C, Chapter 10 model for four-leg signal control intersections (4SG) on rural two- lane undivided roads (including total crashes from HSM model).

Figure 61. Comparison of pedestrian crashes per year from the new pedestrian intersection model (4SG) and existing HSM Part C, Chapter 12 model for four-leg signal control intersections on urban and suburban arterials.

246 Pedestrian and Bicycle Safety Performance Functions intersection legs are shown from the existing HSM Part C, Chapter 12 model: 300 ped/day and 1,000 ped/day. To provide more perspective, Figure 62 provides the same comparison but also includes estimates for total crashes (i.e., multiple-vehicle + single-vehicle + pedestrian + bicycle crashes) from the existing Chapter 12 model for four-leg signal control intersections. Based on this assessment, the new pedestrian intersection model predicts higher estimates of pedestrian crash frequencies compared to the existing HSM Part C, Chapter 12 model for four- leg signal control intersections, but with calibration the new pedestrian intersection model will likely be compatible with the existing Chapter 12 model for four-leg signal control intersections. Similar comparisons of the new pedestrian intersection model were performed for additional intersection types in HSM Part C chapters Based on the compatibility testing, it appears that with calibration the new pedestrian inter- section model is compatible with the existing HSM Part C intersection models. 4.3 Crash Prediction Method for Bicycle Crashes The following discussion presents the crash prediction models for bicycle crashes. The discus- sion notes differences from the RAP models (other than issues discussed previously in Section 4.1) and differences in crash prediction procedures between facility types. 4.3.1 Predictive Method for Bicycle Crashes on Roadway Segments The calculations for the predictive method for bicycle crashes on roadway segments are pre- sented in the following sections. 4.3.1.1 General Form of Roadway Segment Model for Bicycle Crashes The general form of the crash prediction model for bicycle crashes on an individual roadway segment is as follows: (4-24)#N=N FT Cbi bi bibi unadjusted ker ker kerker #- where: Nbiker = predicted number of bicycle crashes per year for all crash severity levels com- bined for a specific roadway segment, Nunadjusted-biker = unadjusted predicted number of bicycle crashes per year involving bicycle movements along a specific roadway segment, FTbiker = facility-type factor for bicycle crashes for a specific roadway segment facility type, and Cbiker = calibration factor for bicycle crashes for a specific roadway segment facility type. The terms in Equation 4-24 are discussed in Sections 4.3.1.2 through 4.3.1.4. 4.3.1.2 Bicycle Crash Prediction Model for Bicycle Movements Along the Road The crash prediction model for bicycle movements along a roadway segment estimates the fre- quency of crashes involving bicycle movements along a roadway segment. The bicycle crash pre- diction model for bicycle movements along the road for a specific roadway segment is as follows: (4-25) N = . Severity MVTSF MVTFF BFF L Likelihood 0 062unadjusted bi bi bi bi bi rker ker ker ker ker # # # # #- J L KK N P OO

Figure 62. Comparison of predicted average pedestrian crashes per year from pedestrian intersection model and existing HSM Part C, Chapter 12 model for four-leg signal control intersections (4SG) on urban and suburban arterials (including total crashes from HSM model).

248 Pedestrian and Bicycle Safety Performance Functions where: Likelihoodbiker = crash likelihood factor for bicycle crashes involving bicycle movements along the road for a specific roadway segment, Severitybiker = crash severity factor for bicycle crashes involving bicycle movements along the road for a specific roadway segment, MVTSFbiker = motor vehicle traffic speed factor for bicycle crashes involving bicycle move- ments along the road for a specific roadway segment, MVTFFbiker = motor vehicle traffic flow factor for bicycle crashes involving bicycle move- ments along the road for a specific roadway segment, BFFr = bicycle flow factor for bicycle crashes involving bicycle movements along the road for a specific roadway segment, and L = length (mi) of a specific roadway segment. The terms in Equation 4-25 are discussed in Sections 4.3.1.5 through 4.3.1.9. The combina- tion of the MVTSFbiker, MVTFFbiker, BFFr, and (L/0.062) terms constitute, in effect, an SPF for crashes related to bicycle movements along the road. The Likelihoodbiker and Severitybiker terms are combinations of adjustment factors. 4.3.1.3 Facility-Type Factor for Bicycle Crashes on Roadway Segments The values of the facility-type factor (FTbiker) developed in this research for bicycle crashes on roadway segments are presented in Table 155 by HSM chapter and roadway type. 4.3.1.4 Calibration Factor for Bicycle Crashes on Roadway Segments The calibration process is discussed in Section 4.1.13. 4.3.1.5 Crash Likelihood Factor for Bicycle Crashes Along the Road The crash likelihood factor for bicycle crashes related to bicycle movements along the road within a roadway segment is determined as follows: (4-26) .Likelihood AF AF AF AF AF AF AF AF AF0 1bi LA bike LA bike LA bike LA bike LA bike LA bike LA bike LA bike LA bike 1 1 1 1 6 7 8 9 4 5 6 8 1ker # # # # # # # # # = - - - - - - - - - where: AFLA1–bike = crash likelihood adjustment factor for bicycle crashes along a specific roadway segment, accounting for bicycle facilities or paved shoulder provision; AFLA6–bike = crash likelihood adjustment factor for bicycle crashes along a specific roadway segment, accounting for lane width; AFLA7–bike = crash likelihood adjustment factor for bicycle crashes along a specific roadway segment, accounting for horizontal curvature; AFLA8–bike = crash likelihood adjustment factor for bicycle crashes along a specific roadway segment, accounting for advance visibility of a curve; HSM Part C Chapter Roadway Type Facility Type F HSM Chapter 10 (rural two-lane, two-way roads) Two-lane roads 0.00826 HSM Chapter 11 (rural multilane highways) Multilane undivided 0.02690 Multilane divided 0.00606 HSM Chapter 12 (urban and suburban arterials) Two-lane undivided 0.19100 Multilane undivided 0.55700 Multilane divided 0.35000 Table 155. Facility-type factors for bicycle crashes on roadway segments.

Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology 249   AFLA9–bike = crash likelihood adjustment factor for bicycle crashes along a specific roadway segment, accounting for percent grade; AFLA14–bike = crash likelihood adjustment factor for bicycle crashes along a specific roadway segment, accounting for presence and condition of delineation; AFLA15–bike = crash likelihood adjustment factor for bicycle crashes along a specific roadway segment, accounting for shoulder rumble strips; AFLA16–bike = crash likelihood adjustment factor for bicycle crashes along a specific roadway segment, accounting for vehicle parking; and AFLA18–bike = crash likelihood adjustment factor for bicycle crashes along a specific roadway segment, accounting for street lighting. The values of these adjustment factors are presented in Section 4.3.1.10. The constant factor of 0.1 was not in the original RAP model but was incorporated in Equation 4-26 for the reason explained in the discussion of adjustment factor AFLA1-bike in Section 4.3.1.10. 4.3.1.6 Crash Severity Factor for Bicycle Crashes Along the Road The crash severity factor for bicycle crashes related to bicycle movements along the road (Severitybiker) is a function of bicycle facilities and paved shoulder provision, as follows: (4-27)Severity AFSA1 bikebiker = - where: AFSA1–bike = crash severity adjustment factor for bicycle crashes along a specific roadway seg- ment, accounting for bicycle facilities and paved shoulder provision. The values for the adjustment factor AFSA1–bike are presented in Section 4.3.1.11. 4.3.1.7 Motor Vehicle Traffic Speed Factors for Bicycle Crashes Along the Road The value of the motor vehicle traffic speed factor for bicycle crashes along the road, MVTSF- biker, should be determined from Table 156. The values shown in Table 156 are illustrated in Fig- ure 63. These factors account for the increased likelihood of severe injury to bicyclists in bicycle Mean Speed of Motor Vehicle Traffic on a Specific Motor Vehicle Traffic Speed Factor 20 or less 0.011 25 0.031 30 0.064 35 0.112 40 0.178 45 0.264 50 0.372 55 0.505 60 0.643 65 0.697 70 0.751 75 0.804 80 0.858 85 0.912 90 or more 0.966 Table 156. Motor vehicle traffic speed factor for bicycle crashes along the road (iRAP 2013l).

250 Pedestrian and Bicycle Safety Performance Functions collisions as the speed of motor vehicle traffic, which is typically substantially higher than the speed of bicyclists, increases. The mean speed of motor vehicle traffic along the specific roadway segment of interest should be used to determine the value of MVTSFbiker. Mean speed should be determined from field measurements of traffic speed on the road segment of interest away from intersections, field measurements on similar road segments, or on the best available estimate based on local knowledge. 4.3.1.8 Motor Vehicle Traffic Flow Factors for Bicycle Crashes Along the Road The value of the motor vehicle traffic flow factor for bicycle crashes along the road related to bicycle movements along the road, MVTFFbiker, is computed with the following equation derived from S. Turner, Roozenburg, and Francis (2006): (4-28).0 000166MVTFF AADTPerLane .0 65biker midpoint#= ` j where: AADTPerLanemidpoint = midpoint of the applicable AADT per lane range applicable to the road segment being analyzed (veh/day). The value of MVTFFbiker is used in calculations for bicycle movements along the road. AADT per lane is computed as the total AADT divided by the number of through lanes on the road. The motor vehicle traffic flow factor essentially represents the level of saturation in the flow on the roadway, where 18,000 veh/day/lane represents a fully saturated roadway. Figure 63. Graph of motor vehicle traffic speed factor for bicycle crashes along the road representing the relative frequency of bicyclist injuries as a function of traffic speed (adapted from iRAP 2013l).

Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology 251   The procedure uses categories for the traffic flow on the roadway segment being analyzed that represent ranges of AADT per lane. The AADT ranges and midpoints applicable to MVTFFbiker are shown in Table 157. The table also shows the computed values of the motor vehicle traffic flow factors. The values are plotted as a continuous function in Figure 64. 4.3.1.9 Bicycle Flow Factor for Bicycle Crashes Along the Road The value of the bicycle flow factor for bicycle crashes along the road is determined from Table 158. The values shown in Table 158 are illustrated in Figure 65. The bicycle peak-hour flow along the road represents the combined bicycle flow for both direc- tions of travel. The bicycle peak-hour flow may be based on bicycle volume counts or the best available estimate of the peak-hour bicycle volumes based on local knowledge. While peak-hour Ranges for Motor Vehicle AADT per L Midpoint of AADT per Lane Computed Motor Vehicle Traffic Flow less than 1,999 1,000 0.015 2,000–3,999 3,000 0.030 4,000–5,999 5,000 0.042 6,000–7,999 7,000 0.052 8,000–9,999 9,000 0.062 10,000–11,999 11,000 0.070 12,000–13,999 13,000 0.078 14,000–15,999 15,000 0.086 16,000–17,999 17,000 0.093 18,000 or more 19,000 0.100 Table 157. Motor vehicle traffic flow factors for bicycle crashes along the road as a function of AADT ranges and midpoints (adapted from iRAP 2013a). Figure 64. Graph of motor vehicle traffic flow factors for bicycle crashes along the road as a function of AADT (adapted from iRAP 2013a).

252 Pedestrian and Bicycle Safety Performance Functions Figure 65. Graph of bicycle flow factor for bicycle crashes along the road as a function of bicycle peak-hour flow. Bicycle Peak-Hour Flow Along the Roada Bicycle Flow Factor None 0.000 1 to 5 0.001 6 to 25 0.002 26 to 50 0.002 51 to 100 0.003 101 to 200 0.004 201 to 300 0.005 301 to 400 0.006 401 to 500 0.006 501 to 900 0.007 more than 900 0.009 aIncluding both directions of travel combined. Table 158. Bicycle flow factors for bicycle crashes along the road. bicycle flow is used in determining the value of the bicycle flow factor, the model predictions represent annual total bicycle crash frequencies. If the bicycle flow along a roadway segment is zero, then zero bicycle crashes will be predicted by the model. In this case, the bicycle flow factor can be set to zero, and no further analysis is needed for crashes related to that bicycle movement. 4.3.1.10 Crash Likelihood Adjustment Factors for Bicycle Crashes Along the Road This section presents tables to determine the values of the crash likelihood adjustment factors used in Equation 4-26. Crash likelihood adjustment factors for bicycle crashes along the road, related to bicycle movements along the road, have subscripts that begin with the letters LA.

Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology 253   AFLA1-bike 2 Bicycle Facilities and Paved Shoulder Provision The derivation of the crash likelihood adjustment factor for bicycle crashes along the road, accounting for bicycle facilities and paved shoulder provision (AFLA1–bike), is illustrated in Table 159 and Table 160. Table 159 illustrates the factors used in the original RAP procedure. This table appeared suitable for use in the HSM2, except that the predicted crash frequency for a shared-use path used by both bicyclists and pedestrians was shown to exceed the crash fre- quency for a separated bicycle path without a barrier by a factor of 10. This difference in crash frequency in the original RAP procedure, based on research from New Zealand and Australia, appears to result from pedestrian/bicycle conflicts. The factor of 10 for this effect seems high for U.S. conditions, but the project team is not aware of any U.S. research that quantifies this effect. The team does not believe that the HSM should suggest an effect that does not appear applicable to U.S. conditions. Since there is no U.S. research to substitute for this effect, the project team has chosen to delete the shared-use path effect from Table 159 and subsequent tables that originally incorporated shared-use paths. It is believed the HSM should remain silent on this issue until such time as relevant U.S. research is available. A revised table of adjustment factors for bicycle facilities and paved shoulder provision for use in HSM2 is shown in Table 160 with the adjustment factor for shared-use paths deleted. Since a shared-use path originally represented the only adjustment factor corresponding to the base condition of 1.0, the adjustment factor values in Table 160 were recalculated to make “separated bicycle path without barrier” the base condition. To do this, the values of all of the remaining adjustment factors in Table 159 were multiplied by 10 to create Table 160. To compensate, so that Bicycle Facilities and Paved Shoulder Provision Adjustment Factor Separated bicycle path with barrier 0.0 Separated bicycle path without barrier 1.0 Dedicated bicycle lane on roadway 120.0 None 200.0 Extra wide outside lane ≥ 14 ft 170.0 Signed or marked shared roadway 190.0 Paved shoulder present with width ≥ 7.9 ft 160.0 Paved shoulder present with width ≥ 3 ft and < 7.9 ft 170.0 Paved shoulder present with width < 3 ft 180.0 Table 160. Crash likelihood adjustment factors for bicycle crashes along the road, accounting for bicycle facilities and paved shoulder provision, modified for use in HSM2 (adapted from iRAP 2014a). Bicycle Facilities and Paved Shoulder Provision Risk Factors for Bicycle Crashes Along the Road from Original RAP Procedure Separated bicycle path with barrier 0.0 Separated bicycle path without barrier 0.1 Dedicated bicycle lane on roadway 12.0 None 20.0 Extra wide outside lane ≥ 14 ft 17.0 Signed or marked shared roadway 19.0 Shared-use path 1.0 Paved shoulder present with width ≥ 7.9 ft 16.0 Paved shoulder present with width ≥ 3 ft and < 7.9 ft 17.0 Paved shoulder present with width < 3 ft 18.0 Table 159. Crash likelihood risk factors for bicycle crashes along the road, accounting for bicycle facilities and paved shoulder provision (iRAP 2014a).

254 Pedestrian and Bicycle Safety Performance Functions the crash predictions for each type of bicycle facility is unchanged, a constant factor of 0.1 was incorporated in Equation 4-26. If the bicycle facility or shoulder types differ for the two directions of travel on an undivided road, average the two applicable adjustment factor values to determine the value of AFLA1–bike. For divided roads, the procedures address all features for the roadways in each direction of travel separately. In Table 159 and Table 160, the term “separated bicycle path” is used to indicate any bicycle facility that is separated from the motor vehicle traveled lanes by a raised or depressed divider or by a traffic barrier. The term “dedicated bicycle lane” on roadway is used to indicate a bicycle facility that is flush with travel lanes used by motor vehicles; a dedicated bicycle lane may be either immediately adjacent to or separated by a flush paved buffer from the travel lanes used by motor vehicles. The rationale for these adjustment factors for bicycle facilities and paved shoulder provision is presented in iRAP Methodology Fact Sheet #10: Facilities for Bicycles (iRAP 2014a). Research suggests that there are, at most, modest crash reduction benefits for bicyclists from provision of bicycle facilities, except where bicyclists are completely separated from motor vehicle traffic. Elvik and Vaa (2004) state that some studies indicate that bicycle crashes may remain the same, or even increase, where bicycle facilities are provided. However, the results of such studies may be confounded because bicycling activity may increase where bicycle facilities are provided. Bicycle collisions are clearly very severe, such that any collision between a motor vehicle and a bicyclist has a substantial likelihood of death or serious injury; there is no evidence that the crash severity varies by bicycle facility type. Based on their own review of the literature, includ- ing work from the Netherlands and other reviews including S. Turner, Roozenburg, and Francis (2006) and B. Turner et al. (2009), iRAP developed the adjustment factors for bicycle facilities and paved shoulders shown in Table 159. AFLA6-bike 2 Lane Width The crash likelihood adjustment factors for bicycle crashes along the road, accounting for lane width (AFLA6–bike), are shown in Table 161. If the lane widths differ for the two directions of travel on an undivided road, average the two applicable adjustment factor values to determine the value of AFLA6–bike. For divided roads, the procedures address all features for the roadways in each direction of travel separately. The rationale for the adjustment factor values is presented in iRAP Road Attribute Risk Factors: Lane Width (iRAP 2013i). These adjustment factors were based on the work of B. Turner et al. (2009). AFLA7-bike 2 Horizontal Curvature The likelihood that motor vehicles will run off the road is higher on horizontal curves than on tangents and increases as the radius of curvature decreases. As the likelihood that motor vehicles will run off the road increases so does the likelihood that an errant motor vehicle will strike a bicyclist. The crash likelihood adjustment factors for bicycle crashes, accounting for horizontal Lane Width Adjustment Factor LA6–bike Wide (≥ 10.6 ft) 1.00 Medium (≥ 9 to < 10.6 ft) 1.20 Narrow (< 9 ft) 1.50 Table 161. Crash likelihood adjustment factors for bicycle crashes along the road, accounting for lane width (iRAP 2013i).

Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology 255   curvature for bicycle movements along the road (AFLA7–bike) are shown in Table 162. The hori- zontal curvature categories are defined by advisory speed ranges and corresponding ranges of horizontal curve radius. If a horizontal curve is signed with an advisory speed plate, use of the category in Table 162 corresponding to the signed advisory speed is recommended. If there is no signed advisory speed, the curvature category should be based on the horizontal curve radius. The rationale for the adjustment factor values is presented in iRAP Road Attribute Risk Factors: Curvature (iRAP 2013c). AFLA8–bike 2 Advance Visibility of a Curve Advance visibility of a curve represents an assessment of the ability of approaching drivers to see a horizontal curve on the roadway ahead. Advance visibility of a curve considers pave- ment markings, chevron markers, advance signing, and sight distance to the curve. If pavement markings, chevron markers, advance signing, and sight distance to the curve are such that an approaching driver can readily see that a horizontal curve is present, advance visibility of a curve should be rated as substantial. If pavement markings, chevron markers, and advance signing have lost their reflectivity and/or legibility (e.g., are weathered or faded) or are absent, and sight distance to the curve is such that an approaching driver is likely to be unaware of the presence of the curve, then advance visibility of a curve should be rated as limited. If the advance visibility of a curve is limited, motor vehicles are more likely to run off the road on the curve, and such vehicles are, therefore, more likely to strike a bicyclist. The crash likelihood adjustment factors for bicycle crashes along the road, accounting for advance visibility of a curve (AFLA8–bike), are shown in Table 163. The rationale for the adjustment factor values is presented in iRAP Road Attribute Risk Factors: Quality of Curve (iRAP 2013n). AFLA9–bike 2 Percent Grade Motor vehicles are more likely to lose control on steep grades than on level roadways. Motor vehicles that lose control may strike a bicyclist. The crash likelihood adjustment factors for bicycle crashes along the road, accounting for grade (AFLA9–bike), are shown in Table 164. Horizontal Curvature Adjustment Factor Straight or gently curving (advisory speed ≥ 60 mph or curve radius > 2,600 ft) 1.00 Moderate curvature (advisory speed in the range from 45 mph to < 60 mph or curve radius in the range from 1,300 ft to ≤ 2,600 ft) 1.81 Sharp curve (advisory speed in the range from 25 mph to < 45 mph or curve radius in the range from 650 ft to ≤ 1,300 ft) 3.51 Very sharp curve (advisory speed < 25 mph or curve radius < 650 ft) 6.02 Table 162. Crash likelihood adjustment factors for bicycle crashes along the road, accounting for horizontal curvature (iRAP 2013c). Advance Visibility of a Curve Adjustment Factor Substantial 1.00 Limited 1.40 Not applicable 1.00 Table 163. Crash likelihood adjustment factors for bicycle crashes along the road, accounting for advance visibility of a curve (iRAP 2013n).

256 Pedestrian and Bicycle Safety Performance Functions The rationale for the adjustment factor values is presented in iRAP Road Attribute Risk Factors: Grade (iRAP 2013e). These adjustment factors represent the potential for loss of control on grades by motor vehicles, which could then strike bicyclists. AFLA14–bike 2 Presence and Condition of Delineation Presence and condition of delineation involves the placement of pavement markings and delineators to help guide motor vehicle drivers along the roadway. Motor vehicles are more likely to run off the road where delineation has lost its reflectivity (e.g., is weathered or faded) or is absent than where delineation is clearly visible to the driver. Motor vehicles that run off the road may strike a bicycle. The crash likelihood adjustment factors for bicycle crashes along the road, accounting for the presence and condition of delineation (AFLA14–bike), are shown in Table 165. If the delineation differs for the two directions of travel on an undivided road, average the two applicable adjustment factor values to determine the value of AFLA14–bike. For divided roads, the procedures address all features for the roadways in each direction of travel separately. The rationale for the adjustment factor values is presented in iRAP Road Attribute Risk Factors: Delineation (iRAP 2013d). The adjustment factor for presence and condition of delineation was based primarily on work by B. Turner et al. (2009). AFLA15–bike 2 Shoulder Rumble Strips Shoulder rumble strips are placed on the edgeline or shoulder of a roadway to alert a driver that their vehicle is leaving the roadway. Since shoulder rumble strips reduce the likelihood that motor vehicles will run off the road, they also reduce the likelihood that errant vehicles will strike bicyclists. The crash likelihood adjustment factors for bicycle crashes along the road, accounting for shoulder rumble strips (AFLA15–bike), are shown in Table 166. If the installation of shoulder rumble strips differs for the two directions of travel on an undivided road, average the two applicable adjustment factor values to determine the value of AFLA15–bike. For divided roads, the procedures address all features for the roadways in each direction of travel separately. The rationale for the adjustment factor values is presented in iRAP Road Attribute Risk Factors: Shoulder Rumble Strips (iRAP 2013p). These adjustment factors are based primarily on literature review by S. Turner, Singh, and Nates (2012) and B. Turner et al. (2009). AFLA16–bike 2 Vehicle Parking Vehicle parking along a roadway may increase the likelihood that bicyclists will ride in the traveled way rather than on the roadside or shoulder. If a bicycle facility is present but obstructed by vehicle parking, bicyclists are also more likely to ride in the traveled way. Bicyclists riding Percent Grade Adjustment Factor 0% to < 7.5% 1.00 7.5% to < 10% 1.20 ≥ 10% 1.70 Table 164. Crash likelihood adjustment factors for bicycle crashes along the road, accounting for percent grade (iRAP 2013e). Presence and Condition of Delineation Adjustment Factor Clearly visible 1.00 Loss of reflectivity (e.g., weathered or faded) or absent 1.20 Table 165. Crash likelihood adjustment factors for bicycle crashes along the road, accounting for the presence and condition of delineation (iRAP 2013d).

Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology 257   in the traveled way are more likely to be struck by motor vehicles than bicyclists riding on the roadside or shoulder. The crash likelihood adjustment factors for bicycle crashes along the road, accounting for vehicle parking (AFLR16–bike), are shown in Table 167. If a bicycle facility is present but the facility is regularly obstructed by parked motor vehicles, apply the adjustment factors in Table 167 as if the bicycle facility were not present. The rationale for the adjustment factor values is presented in iRAP Road Attribute Risk Factors: Vehicle Parking (iRAP 2013s). These adjustment factors assume that vehicle parking may block bicycle facilities and, thus, reduce the benefits of bicycle facilities presented above. AFLA18–bike 2 Street Lighting The presence of street lighting makes the roadway and objects on the roadway more visible to motorists moving along the roadway, thus reducing the likelihood that bicyclists will be struck by motor vehicles. The crash likelihood adjustment factors for bicycle crashes along the road, accounting for street lighting (AFLA18–bike), are shown in Table 168. The rationale for the adjustment factor values is presented in iRAP Road Attribute Risk Factors: Street Lighting (iRAP 2013r). These adjustment factors were based on combined results from a literature review by B. Turner et al. (2012) and studies summarized in the FHWA CMF Clearing- house (www.cmfclearinghouse.org). Results based on nighttime crashes only were reduced to represent the effect on a combination of daytime and nighttime crashes. 4.3.1.11 Crash Severity Adjustment Factors for Bicycle Crashes Along the Road Equation 4-27 includes a single crash severity adjustment factor for bicycle crashes on road- way segments. This factor represents the effect on bicycle crash severity for bicycle facilities and paved shoulder provision for bicycle movements along the road (AFSA1-bike), and the values are shown in Table 169. In Table 169, the term “separated bicycle path” is used to indicate any bicycle facility that is separated from the motor vehicle travel lanes by a raised or depressed divider or by a traffic barrier. The term “dedicated bicycle lane on roadway” is used to indicate a bicycle facility that is flush with travel lanes used by motor vehicles; a dedicated bicycle lane may be either immediately adjacent to or separated by a flush, paved buffer from the travel lanes used by motor vehicles. Shoulder Rumble Strips Adjustment Factor LA15–bike Not present 1.25 Present 1.00 Table 166. Crash likelihood adjustment factors for bicycle crashes along the road, accounting for shoulder rumble strips (iRAP 2013p). Vehicle Parking Adjustment Factor None 1.00 One side 1.20 Two sides 1.33 One side (bicycle facility present) 1.00 Two sides (bicycle facility present) 1.00 Table 167. Crash likelihood adjustment factors for bicycle crashes along the road, accounting for vehicle parking (iRAP 2013s).

258 Pedestrian and Bicycle Safety Performance Functions If the bicycle facility or shoulder types differ for the two directions of travel on an undivided road, average the two applicable adjustment factor values to determine the value of AFSA1–bike. For divided roads, the procedures address all features for the roadways in each direction of travel separately. The table corresponding to Table 169 in the original RAP model included an adjustment factor for shared-use paths. This adjustment factor for shared-use paths was dropped from Table 169 for the reason explained in the discussion of the adjustment factor AFLA1–bike in Section 4.3.1.10. The rationale for these adjustment factors is presented in iRAP Methodology Fact Sheet #10: Facilities for Bicycles (iRAP 2014a). All but one of the adjustment factors have the relatively high value of 90, indicating that, when a motor vehicle strikes a bicyclist, a severe crash is likely to result. The adjustment factor for a separated bicycle path with a barrier is zero, indicating that, where a traffic barrier is present between the roadway and the bicycle path, no severe crashes involving bicyclists will result. This zero factor should only be used where the traffic barrier is sufficiently strong that it will not be penetrated by a motor vehicle traveling at typical roadway speeds. 4.3.1.12 Computation of Final Prediction of Bicycle Crashes on Roadway Segments The total predicted bicycle crash frequency for all crash severity levels combined for a specific roadway segment during a specific year is Nbiker determined with Equation 4-24. Nbiker can be divided into individual crash severity levels as follows, where the crash severity level is defined by the most severe bicyclist injury in the crash: (4-29)P#N=N K Kbiker biker biker- - (4-30)P#N=N A Abiker biker biker- - P#N=N B Bbiker biker biker- - (4-31) #N=N PC Cbiker biker biker- - (4-32) Street Lighting Adjustment Factor Present 1.00 Not present 1.25 Table 168. Crash likelihood adjustment factors for bicycle crashes along the road, accounting for street lighting (iRAP 2013r). Table 169. Crash severity adjustment factors for bicycle crashes along the road, accounting for bicycle facilities and paved shoulder provision (iRAP 2014a). Bicycle Facilities and Paved Shoulder Provision Adjustment Factor Separated bicycle path with barrier 0 Separated bicycle path without barrier 90 Dedicated bicycle lane on roadway 90 None 90 Extra wide outside lane ≥ 14 ft 90 Signed or marked shared roadway 90 Paved shoulder present with width ≥ 7.9 ft 90 Paved shoulder present with width ≥ 3 ft and < 7.9 ft 90 Paved shoulder present with width < 3 ft 90

Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology 259   where: Nbiker = predicted number of bicycle crashes per year for all crash severity levels combined for a specific roadway segment, Nbiker–K = predicted number of fatal bicycle crashes per year for a specific roadway segment, Nbiker–A = predicted number of A-injury bicycle crashes per year for a specific roadway segment, Nbiker–B = predicted number of B-injury bicycle crashes per year for a specific roadway segment, Nbiker–C = predicted number of C-injury bicycle crashes per year for a specific roadway segment, Pbiker–K = proportion of fatal bicycle crashes for specific roadway segment facility types (see Table 170), Pbiker–A = proportion of A-injury bicycle crashes for specific roadway segment facility types (see Table 170), Pbiker –B = proportion of B-injury bicycle crashes for specific roadway segment facility types (see Table 170), and Pbiker–C = proportion of C-injury bicycle crashes for specific roadway segment facility types (see Table 170). The predicted number of injured bicyclists per crash is 1.05 for two-lane, two-way roads and urban and suburban arterials. The predicted number of injured bicyclists per crash is 1.01 for rural multilane highways. 4.3.1.13 Comparison of Predictive Method for Bicycle Crashes on Roadway Segments with Existing HSM Models Following development of the bicycle roadway segment model, compatibility testing of the new model was conducted to check the reasonableness of the results across facility types. To gain a better understanding of the potential use of the model within the HSM, output results from the new model were compared to output from existing models in HSM Part C. Note, for these comparisons, the new model and the existing HSM models were not calibrated using a single agency’s data, and, unlike existing models in the HSM, bicycle exposure is accounted for in the new model so some differences are expected. Nonetheless, output results from the new model were compared to output from existing models in the HSM to gain a sense of the reasonableness of the new bicycle roadway segment model for application and potential incorporation in HSM2. Figure 66 provides a comparison of the predicted average total bicycle crash frequency for rural two-lane undivided roads from the bicycle roadway segment model presented in this sec- tion and the predicted average total bicycle crash frequency from the existing HSM Part C, Chap- ter 10 model for rural two-lane undivided roads. The predicted average crash frequencies are for the base conditions for both models. Comparisons from the bicycle roadway segment model are shown for two bicycle peak-hour volume levels: 1 to 5 bike/hr and 101 to 200 bike/hr. To provide more perspective, Figure 67 provides the same comparison but also includes estimates for total HSM Chapter Roadway Type Proportion of Bicycle Crashes by Most Severe Bicyclist Injury Fatal A-Injury B-Injury C-Injury HSM Chapter 10 (rural two- lane, two-way roads) Two-lane, two-way roads 0.045 0.315 0.4470 0.193 HSM Chapter 11 (rural multilane highways) Multilane undivided 0.045 0.182 0.5910 0.182 Multilane divided 0.056 0.111 0.6110 0.222 HSM Chapter 12 (urban and suburban arterials) Two-lane undivided 0.033 0.138 0.0571 0.258 Multilane undivided 0.013 0.102 0.5030 0.382 Multilane divided 0.016 0.082 0.5500 0.387 Table 170. Proportion of bicycle crashes by roadway segment facility type and injury severity level.

260 Pedestrian and Bicycle Safety Performance Functions crashes (i.e., multiple-vehicle + single-vehicle + pedestrian + bicycle crashes) from the existing HSM Part C, Chapter 10 model for rural two-lane undivided roads. Based on this assessment, the new bicycle roadway segment model provides similar results to the existing HSM Part C, Chapter 10 model for rural two-lane undivided roads. Thus, the new bicycle roadway segment model appears compatible with the existing Chapter 10 model for rural two-lane undivided roads. Figure 67. Comparison of predicted average total bicycle crashes per year from new bicycle roadway segment model and existing HSM Part C, Chapter 10 model for rural two-lane undivided roads (including total crashes from HSM model). Figure 66. Comparison of predicted average total bicycle crashes per year from new bicycle roadway segment model and existing HSM Part C, Chapter 10 model for rural two-lane undivided roads.

Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology 261   Figure 68 provides a comparison of the predicted average total bicycle crash frequency for urban multilane undivided roads (e.g., four-lane undivided) from the bicycle roadway segment model presented in this section and the predicted average total bicycle crash frequency from the existing HSM Part C, Chapter 12 model for urban and suburban four-lane undivided arterials. The predicted average crash frequencies are for the base conditions for both models. Comparisons from the bicycle roadway segment model are shown for two bicycle peak-hour volume levels: 6 to 25 bike/hr and 201 to 300 bike/hr. To provide more perspective, Figure 69 provides the same Figure 69. Comparison of predicted average total bicycle crashes per year from new bicycle roadway segment model and existing HSM Part C, Chapter 12 model for urban four-lane undivided roads (including total crashes from HSM model). Figure 68. Comparison of predicted average total bicycle crashes per year from new bicycle roadway segment model and existing HSM Part C, Chapter 12 model for urban four-lane undivided roads.

262 Pedestrian and Bicycle Safety Performance Functions comparison but also includes estimates for total crashes (i.e., multiple-vehicle + single-vehicle + pedestrian + bicycle crashes) from the existing HSM Part C model for urban and suburban four- lane undivided roads. Based on this assessment, the new bicycle roadway segment model provides similar results to the existing HSM Part C, Chapter 12 model for urban and suburban four-lane undivided roads. Thus, the new bicycle roadway segment model appears compatible with the existing Chapter 12 model for four-lane undivided roads. Similar comparisons of the new bicycle roadway segment model were performed for addi- tional road types in HSM Part C, and in all cases, the new bicycle roadway segment model appears compatible with the existing HSM Part C models for roadway segments. Based on the compatibility testing, it appears that with calibration the new bicycle roadway segment model is compatible with the existing HSM Part C roadway segment models. 4.3.2 Predictive Method for Bicycle Crashes at Intersections The calculations for the predictive method for bicycle crashes related to bicycle movements through intersections are presented below. 4.3.2.1 General Form of Intersection Model for Bicycle Crashes The general form of the crash prediction model for bicycle crashes related to bicycle move- ments through intersections is as follows: #N=N CFT bikeii unadjusted i ibike bikebike #- (4-33) where: Nbikei = predicted number of bicycle crashes per year for all crash severity levels com- bined for bicycle movements at a specific intersection, Nunadjusted-bikei = unadjusted prediction for the number of bicycle crashes per year involving bicycle movements through a specific intersection, FTbikei = facility-type factor for bicycle crashes for a specific intersection type, and Cbikei = calibration factor for bicycle crashes for a specific intersection type. The terms in Equation 4-33 are discussed in Sections 4.3.2.2 through 4.3.2.4. For an inter- section to be analyzed as a whole, Nbikei needs to be calculated for bicycle movements through the intersection from the major road and for bicycle movements through the intersection from the minor road. 4.3.2.2 Bicycle Crash Prediction Model Involving Bicycle Movements Through an Intersection The bicycle crash prediction model involving bicycle movements through an intersection from the major road and the minor road is as follows: = Likelihood Severity MVTSF MVTFF Severity MVTSF MVTFF BFF BFF Likelihood N major major major major or or or bikei or or i i i i i i imajor unadjusted i bike bike bike bike bike m bike min bike min min min in bike # # # # # # # # + - - - - - - - - - ` `j j (4-34) where: Likelihoodbikei–major = crash likelihood factor for bicycle crashes involving bicycle movements through a specific intersection on the major road,

Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology 263   Severitybikei–major = crash severity factor for bicycle crashes involving bicycle movements through a specific intersection on the major road, MVTSFbikei–major = motor vehicle traffic speed factor for bicycle crashes involving bicycle movements through a specific intersection on the major road, MVTFFbikei–major = motor vehicle traffic flow factor for bicycle crashes involving bicycle movements through a specific intersection on the major road, BFFmajor = bicycle flow factor for bicycle crashes involving bicycle movements through a specific intersection on the major road, Likelihoodbikei–minor = crash likelihood factor for bicycle crashes involving bicycle movements through a specific intersection on the minor road, Severitybikei–minor = crash severity factor for bicycle crashes involving bicycle movements through a specific intersection on the minor road, MVTSFbikei–minor = motor vehicle traffic speed factor for bicycle crashes involving bicycle movements through a specific intersection on the minor road, MVTFFbikei–minor = motor vehicle traffic flow factor for bicycle crashes involving bicycle movements through a specific intersection on the minor road, and BFFminor = bicycle flow factor for bicycle crashes involving bicycle movements through a specific intersection on the minor road. The terms in Equation 4-34 are discussed in Sections 4.3.2.5 through 4.3.2.9. The combination of the MVTSFbikei, MVTFFbikei, and BFF terms for major and minor roads constitute, in effect, an SPF for crashes related to bicycle movements through a specific intersection. The Likelihoodbikei and Severitybikei terms for major and minor roads are combinations of adjustment factors. 4.3.2.3 Facility-Type Factor for Bicycle Crashes at Intersections The values of the facility-type factor (FTbikei) developed in this research for bicycle crashes at intersections are presented in Table 171 by HSM chapter and intersection type. 4.3.2.4 Calibration Factor for Bicycle Crashes at Intersections The calibration process is discussed above in Section 4.1.13. 4.3.2.5 Crash Likelihood Factor for Bicycle Crashes Involving Bicycle Movements Through an Intersection The crash likelihood factor for bicycle crashes involving bicycle movements through an inter- section is determined as follows: Likelihood AF AF AF AFAF AFLI bike LI bike LI bike LI bike LI bikei LI bike1 2 3 4 5 18bike # # # # #= - - - - - - (4-35) HSM Part C Chapter Intersection Type HSM Chapter 10 (rural two-lane, two-way roads) 3ST, 3SG 0.03520 4ST, 4SG 0.04240 HSM Chapter 11 (rural multilane highways) 3ST, 3SG, 4ST, 4SG 0.02410 HSM Chapter 12 (urban and suburban arterials) 3ST 0.52700 3SG 0.59600 4ST 0.64900 4SG 0.61900 NOTE: 3ST = Three-leg intersection with minor-road stop control. 3SG = Three-leg signalized intersection. 4ST = Four-leg intersection with minor-road stop control. 4SG = Four-leg signalized intersection. Table 171. Facility-type factors for bicycle crashes at intersections.

264 Pedestrian and Bicycle Safety Performance Functions where: AFLI1–bike = crash likelihood adjustment factor for bicycle crashes through a specific intersection, accounting for bicycle facilities and paved shoulder provision; AFLI2–bike = crash likelihood adjustment factor for bicycle crashes through a specific intersection, accounting for bicycle path and pedestrian crossing type; AFLI3–bike = crash likelihood adjustment factor for bicycle crashes through a specific intersection, accounting for intersection type; AFLI4–bike = crash likelihood adjustment factor for bicycle crashes through a specific intersection, accounting for advance visibility of an intersection; AFLI5–bike = crash likelihood adjustment factor for bicycle crashes through a specific intersection, accounting for intersection channelization; and AFLI18–bike = crash likelihood adjustment factor for bicycle crashes through a specific intersection, accounting for street lighting. Equation 4-35 can be applied to determine either Likelihoodbikei–major or Likelihoodbikei–minor. The values of the adjustment factors are presented in Section 4.3.2.10. 4.3.2.6 Crash Severity Factor for Bicycle Crashes Involving Bicycle Movements Through an Intersection The crash severity factor for bicycle crashes involving bicycle movements through an inter- section (Severitybikei) is a function of intersection type as follows: Severity AFi I bikeS 3bike = - (4-36) where: AFSI3–bike = crash severity adjustment factor for bicycle crashes through a specific intersection, accounting for intersection type. Equation 4-36 can be applied to determine either Severitybikei–major or Severitybikei–minor. The values of these adjustment factors are presented in Section 4.3.2.11. 4.3.2.7 Motor Vehicle Traffic Speed Factor for Bicycle Crashes Involving Bicycle Movements Through an Intersection The value of the motor vehicle traffic speed factor for bicycle crashes involving bicycle move- ments through intersections, MVTSFbikei–major or MVTSFbikei–minor, is determined in the same manner as the motor vehicle traffic speed factor for bicycle crashes along a roadway segment, MVTSFbiker, as shown in Table 156. These factors account for the increased likelihood of severe injury to bicyclists in bicycle collisions as the speed of motor vehicle traffic, which is typically substantially higher than the speed of bicyclists, increases. For bicycle movements through an intersection, use the mean traffic speed on either the major or minor road (e.g., the speed used to determine MVTSFbiker), as applicable, to determine the value of MVTSFbikei, unless stop-sign control is present at the intersection on the road being ana- lyzed; where stop-sign control is present on the road being analyzed, the speed used to determine the value of MVTSFbikei–major or MVTSFbikei–minor should be 20 mph or less. 4.3.2.8 Motor Vehicle Traffic Flow Factor for Bicycle Crashes Involving Bicycle Movements Through an Intersection The values of the motor vehicle traffic flow factor for bicycle crashes involving bicycle move- ments on a specific road through an intersection (i.e., crossing the side road) (MVTFFbikei–major or MVTFFbikei–minor) are shown in Table 172. These are the same values as the motor vehicle traffic

Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology 265   flow factor for bicycle movements along a roadway segment shown in Table 157, except Table 172 is slightly adapted. For bicycle movements through an intersection along the major road, the side road is the minor road. For bicycle movements through an intersection along the minor road, the side road is the major road. 4.3.2.9 Bicycle Flow Factor for Bicycle Crashes Involving Bicycle Movements Through an Intersection The values of the bicycle flow factor for bicycle crashes involving bicycle movements through an intersection (BFFmajor or BFFminor) are the same as the bicycle flow factor values for bicycle crashes along a roadway section (BFFr) shown in Table 158. The bicycle peak-hour flow through an intersection on a particular roadway represents the combined bicycle flow for both directions of travel. The bicycle peak-hour flow may be based on bicycle volume counts or the best available estimate of the peak-hour bicycle volumes based on local knowledge. While peak-hour bicycle flow is used in determining the value of the bicycle flow factor, the model predictions represent annual total bicycle crash frequencies. If the bicycle flow through an intersection is zero, then zero bicycle crashes will be predicted by the model. In this case, the bicycle flow factor can be set to zero, and no further analysis is needed for crashes related to that bicycle movement. 4.3.2.10 Crash Likelihood Adjustment Factors for Bicycle Crashes Involving Bicycle Movements Through an Intersection This section presents tables to determine the values of the crash likelihood adjustment factors used in Equation 4-35. Crash likelihood adjustment factors for bicycle crashes involving bicycle movements through an intersection have subscripts that begin with the letters LI. AFLI1–bike 2 Bicycle Facilities and Paved Shoulder Provision The crash likelihood adjustment factors for bicycle crashes, accounting for bicycle facilities and paved shoulder provision (AFLI1–bike), for bicycle movements through an intersection are shown in Table 173. The value of AFLI1–bike for bicycle movements through an intersection along the major road should be based on bicycle facilities or paved shoulder provision along the major road. The value of AFLI1–bike for bicycle movements through an intersection along the minor road should be based on bicycle facilities or paved shoulder provision along the minor road. Ranges for Motor Vehicle AADT per Lane for the Side Road Entering the Intersection Midpoint of Side-Road AADT per Computed Motor Vehicle Traffic Flow Factor less than 1,999 1,000 0.015 2,000–3,999 3,000 0.030 4,000–5,999 5,000 0.042 6,000–7,999 7,000 0.052 8,000–9,999 9,000 0.062 10,000–11,999 11,000 0.070 12,000–13,999 13,000 0.078 14,000–15,999 15,000 0.086 16,000–17,999 17,000 0.093 18,000 or more 19,000 0.100 Table 172. Motor vehicle traffic flow factors for bicycle crashes involving bicycle movements through an intersection as a function of AADT ranges and midpoints (adapted from iRAP 2013a).

266 Pedestrian and Bicycle Safety Performance Functions In Table 173, the term “separated bicycle path” is used to indicate any bicycle facility that is separated from the motor vehicle travel lanes by a raised or depressed divider or by a traffic barrier. The term “dedicated bicycle lane on roadway” is used to indicate a bicycle facility that is flush with travel lanes used by motor vehicles; a dedicated bicycle lane may be either immediately adjacent to or separated by a flush, paved buffer from the travel lanes used by motor vehicles. If the bicycle facility or shoulder types differ for the two directions of travel on the major road or the minor road, average the two applicable adjustment factor values to determine the value of AFLI1–bike. The table corresponding to Table 173 in the original RAP model included an adjustment factor for shared-use paths. This adjustment factor for shared-use paths was dropped from Table 173 for the reason explained in the discussion of the adjustment factor AFLA1–bike in Section 4.3.1.10. AFLI2–bike 2 Bicycle Path Presence and Pedestrian Crossing Facility Type The crash likelihood adjustment factors for bicycle crashes involving bicycle movements through an intersection, accounting for the presence of bicycle paths and pedestrian crossing facilities of specific types (AFLI2–bike), are shown in Table 174. The value of AFLI2–bike for bicycle movements through an intersection along the major road should be based on presence of a bicycle path and type of pedestrian crossing along the major road. The value of AFLI2–bike for bicycle movements through an intersection along the minor road should be based on presence of a bicycle path and type of pedestrian crossing along the minor road. Bicycle Facilities and Paved Shoulder Provision Adjustment Factor Separated bicycle path with barrier 1.0 Separated bicycle path without barrier 1.0 Dedicated bicycle lane on roadway 1.0 None 1.2 Extra wide outside lane ≥ 14 ft 1.2 Signed or marked shared roadway 1.0 Paved shoulder present with width ≥ 7.9 ft 1.2 Paved shoulder present with width ≥ 3 ft and < 7.9 ft 1.2 Paved shoulder present with width < 3 ft 1.2 Table 173. Crash likelihood adjustment factors for bicycle crashes through an intersection, accounting for bicycle facilities and paved shoulder provision (iRAP 2014a). Combination of Bicycle Path Presence and Pedestrian Crossing Facility Type Adjustment Factor Bicycle path and grade-separated facility 0.00 Bicycle path and signalized with refuge 0.15 Bicycle path and signalized without refuge 0.19 Bicycle path and unsignalized marked crossing with refuge 0.57 Bicycle path and unsignalized marked crossing without refuge 0.72 Bicycle path and refuge only 0.76 No bicycle path, no crossing facility, or no intersection 1.00 NOTE: This table combines data for bicycle facility type for the road being analyzed with data for pedestrian crossing facility type for the side road in determining crash likelihood where a bicycle path crosses the side road at an intersection. Table 174. Crash likelihood adjustment factors for bicycle crashes through an intersection, accounting for the presence of bicycle paths and pedestrian crossing facility types (iRAP 2014a; iRAP 2014b).

Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology 267   The rationale for these adjustment factors combines information from iRAP Methodology Fact Sheet #10: Facilities for Bicycles (iRAP 2014a) and from iRAP Road Attribute Risk Factors: Pedestrian Crossing Facilities (iRAP 2014b). AFLI3-bike 2 Intersection Type Intersection type includes consideration of factors including the presence of an at-grade inter- section, the number of intersection legs, the presence or absence of traffic signals, and the pres- ence or absence of exclusive left-turn lanes. The crash likelihood adjustment factors for bicycle crashes involving bicycle movements through an intersection, accounting for intersection type (AFLI3–bike), are shown in Table 175. The rationale for the adjustment factor values is presented in iRAP Road Attribute Risk Factors: Intersection Type (iRAP 2013h). Most of the factors are based on a literature review by iRAP. AFLI4–bike 2 Advance Visibility of an Intersection Advance visibility of an intersection represents an assessment of the ability of approaching drivers to see an at-grade intersection on the roadway ahead. Advance visibility of an intersection considers pavement markings and signing in advance of and at the intersection and sight dis- tance to the intersection on the road being analyzed. The crash likelihood adjustment factors for bicycle crashes for bicycle movements through an intersection, accounting for advance visibility of an intersection (AFLI4-bike), are shown in Table 176. If pavement markings, signing, and sight distance to the intersection are such that an approaching driver can readily see that an intersec- tion is present, advance visibility of an intersection should be rated as substantial. If pavement markings and signing have lost their reflectivity and/or legibility (e.g., are weathered or faded) or are absent and sight distance is such that an approaching driver is likely to be unaware of the presence of the intersection, then advance visibility of an intersection should be rated as limited. The value of AFLI4–bike for bicycle movements through an intersection along the major road should be based on the ability of drivers to see the at-grade intersection along the major road. The value Intersection Type Adjustment Factor Three-leg unsignalized with exclusive left-turn lane 45 Three-leg unsignalized with no exclusive left-turn lane 55 Three-leg signalized with exclusive left-turn lane 30 Three-leg signalized with no exclusive left-turn lane 40 Four-leg unsignalized with exclusive left-turn lane 55 Four-leg unsignalized with no exclusive left-turn lane 80 Four-leg signalized with exclusive left-turn lane 35 Four-leg signalized with no exclusive left-turn lane 50 Table 175. Crash likelihood adjustment factors for bicycle crashes through an intersection, accounting for intersection type (iRAP 2013h). Advance Visibility of an Intersection Adjustment Factor Substantial 1.00 Limited 1.20 Not applicable 1.00 Table 176. Crash likelihood adjustment factors for bicycle crashes through an intersection, accounting for advance visibility of an intersection (iRAP 2013g).

268 Pedestrian and Bicycle Safety Performance Functions of AFLI4–bike for bicycle movements through an intersection along the minor road should be based on the ability of drivers to see the at-grade intersection along the minor road. The rationale for the adjustment factor values is presented in iRAP Road Attribute Risk Factors: Intersection Quality (iRAP 2013g). iRAP found that research evidence concerning the advance visibility of an intersection adjustment factor is minimal. Elvik and Vaa (2004) suggest that improvements in pavement markings and channelization reduce intersection crashes by 15 percent. B. Turner et al. (2009) found a 20 percent crash reduction for delineation, and iRAP chose to use this value for the effect of advance visibility of an intersection on crash likelihood. AFLI5-bike 2 Intersection Channelization Intersection channelization separates motor vehicle movements at intersections by direction of travel and/or turning maneuver. Examples of intersection channelization include splitter islands and islands separating turn lanes from through lanes. The crash likelihood adjustment factors for bicycle crashes for bicycle movements through an intersection, accounting for intersection channelization (AFLI5–bike), are shown in Table 177. The value of AFLI5–bike for bicycle movements through an intersection along the major road should be based on intersection channelization along the major road. The value of AFLI5–bike for bicycle movements through an intersection along the minor road should be based on intersection channelization along the minor road. The rationale for the adjustment factor values is presented in iRAP Road Attribute Risk Factors: Intersection Channelization (iRAP 2013f) and based on a literature review by B. Turner et al. (2012). AFLI18–bike 2 Street Lighting The crash likelihood adjustment factors for bicycle crashes for bicycle movements through an intersection, accounting for street lighting (AFLI18-bike), are determined in the same manner as the crash likelihood adjustment factors for bicycle crashes along the road (AFLA18–bike) shown in Table 168. 4.3.2.11 Crash Severity Adjustment Factors for Bicycle Crashes Involving Bicycle Movements Through an Intersection Equation 4-36 includes a single crash severity adjustment factor for bicycle crashes involv- ing bicycle movements through intersections. This factor represents the effect on crash severity for intersection type, as shown in Table 178. Intersection type includes consideration of factors including the presence of an at-grade intersection, the number of intersection legs, the presence or absence of traffic signals, and the presence or absence of exclusive left-turn lanes. The rationale for the adjustment factor values is presented in iRAP Road Attribute Risk Factors: Intersection Type (iRAP 2013h). Most of the factors are based on a literature review by iRAP. 4.3.2.12 Computation of Final Prediction of Bicycle Crashes at Intersections The total predicted bicycle crash frequency for all crash severity levels combined for a specific intersection during a specific year is Nbikei determined with Equation 4-33. Nbikei can be divided Intersection Channelization Adjustment Factor Not present 1.20 Present 1.00 Table 177. Crash likelihood adjustment factors for bicycle crashes through an intersection, accounting for intersection channelization (iRAP 2013f).

Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology 269   into individual crash severity levels as follows, where the crash severity level is defined by the most severe bicyclist injury in the crash: P#N=N K Ki i ibike bike bike- - (4-37) P#N=N i A i i Abike bike bike- - (4-38) P#N=N i B i i Bbike bike bike- - (4-39) P#N=N i C i i Cbike bike bike- - (4-40) where: Nbikei = predicted number of bicycle crashes per year for all crash severity levels combined for a specific intersection, Nbikei–K = predicted number of fatal bicycle crashes per year for a specific intersection, Nbikei–A = predicted number of A-injury bicycle crashes per year for a specific intersection, Nbikei–B = predicted number of B-injury bicycle crashes per year for a specific intersection, Nbikei–C = predicted number of C-injury bicycle crashes per year for a specific intersection, Pbikei–K = proportion of fatal bicycle crashes for specific intersection facility types (see Table 179), Pbikei–A = proportion of A-injury bicycle crashes for specific intersection facility types (see Table 179), Pbikei–B = proportion of B-injury bicycle crashes for specific intersection facility types (see Table 179), and Pbikei–C = proportion of C-injury bicycle crashes for specific intersection facility types (see Table 179). The predicted number of injured bicyclists per crash is 1.01 for two-lane, two-way roads and rural multilane highways. The predicted number of injured bicyclists per crash is 1.02 for urban and suburban arterials. 4.3.2.13 Comparison of Predictive Method for Bicycle Crashes at Intersections with Existing HSM Models Following development of the bicycle intersection model, compatibility testing of the new model was conducted to check the reasonableness of the results across intersection types. To gain a better understanding of the potential use of the model within the HSM, output results from the new model were compared to output from existing models in HSM Part C. Note, for these Intersection Type Adjustment Factor Three-leg unsignalized with exclusive left-turn lane 45 Three-leg unsignalized with no exclusive left-turn lane 45 Three-leg signalized with exclusive left-turn lane 45 Three-leg signalized with no exclusive left-turn lane 45 Four-leg unsignalized with exclusive left-turn lane 50 Four-leg unsignalized with no exclusive left-turn lane 50 Four-leg signalized with exclusive left-turn lane 50 Four-leg signalized with no exclusive left-turn lane 50 Table 178. Crash severity adjustment factors for bicycle crashes through an intersection, accounting for intersection type (iRAP 2013h).

270 Pedestrian and Bicycle Safety Performance Functions comparisons, the new model and the existing HSM models were not calibrated using a single agency’s data; and unlike the existing models in the HSM, bicycle exposure is accounted for in the new model so some differences are expected. Nonetheless, output results from the new model were compared to output from existing models in the HSM to gain a sense of the reasonableness of the new bicycle intersection model for application and potential incorporation in HSM2. Figure 70 provides a comparison of the predicted average total bicycle crash frequency for three-leg stop control intersections on rural two-lane roads from the bicycle intersection model presented in this section and the predicted average total bicycle crash frequency from the exist- ing HSM Part C, Chapter 10 model for three-leg stop control intersections. The predicted average crash frequencies are for the base conditions for both models. Comparisons from the bicycle intersection model are shown for two bicycle peak-hour crossing volume levels: 1 to 5 bike/hr and 101 to 200 bike/hr. Two levels of minor-road traffic volumes were also assumed for the com- parison: 1,000 and 3,000 veh/day. To provide more perspective, Figure 71 provides the same comparison but also includes estimates for total crashes (i.e., multiple-vehicle + single-vehicle + pedestrian + bicycle crashes) from the existing HSM Part C, Chapter 10 model for three-leg stop control intersections. Based on this assessment, the new bicycle intersection model provides nearly constant esti- mates for bicycle crashes as the major-road volume increases as compared to the existing HSM Part C, Chapter 10 model where bicycle crashes increase as the major-road volume increases. However, as Figure 71 illustrates, the magnitude of the estimates for bicycle crashes from the new bicycle intersection model and existing Chapter 10 model for three-leg stop control intersections are similar compared to total crashes. Figure 72 provides a comparison of the predicted average total bicycle crash frequency for four-leg signal control intersections on urban and suburban arterials from the bicycle intersection HSM Chapter Roadway Type Intersection Type Proportion of Bicycle Crashes by Most Severe Bicycle Injury Fatal A-Injury B-Injury C-Injury HSM Chapter 10 (rural two- lane, two-way roads) Two-lane, two-way roads 3ST, 3SG 0.077 0.077 0.769 0.077 4ST, 4SG 0.083 0.167 0.583 0.167 HSM Chapter 11 (rural multilane highways) Multilane undivided and divided 3ST, 3SG 0.077 0.077 0.769 0.077 4ST, 4SG 0.083 0.167 0.583 0.167 HSM Chapter 12 (urban and suburban arterials) Two-lane undivided 3ST 0.056 0.111 0.500 0.333 Two-lane undivided 3SG 0.005 0.005 0.643 0.347 Two-lane undivided 4ST 0.077 0.077 0.615 0.231 Two-lane undivided 4SG 0.009 0.009 0.446 0.536 Multilane undivided 3ST 0.019 0.019 0.385 0.577 Multilane undivided 3SG 0.005 0.005 0.643 0.347 Multilane undivided 4ST 0.011 0.011 0.761 0.217 Multilane undivided 4SG 0.077 0.077 0.308 0.308 Multilane divided 3ST 0.022 0.109 0.348 0.522 Multilane divided 3SG 0.005 0.005 0.643 0.347 Multilane divided 4ST 0.049 0.049 0.536 0.366 Multilane divided 4SG 0.016 0.041 0.468 0.475 NOTE: 3ST = Three-leg intersection with minor-road stop control. 3SG = Three-leg signalized intersection. 4ST = Four-leg intersection with minor-road stop control. 4SG = Four-leg signalized intersection. Table 179. Proportion of bicycle crashes by intersection type and injury severity level.

Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology 271   Figure 71. Comparison of predicted average bicycle crashes per year from new bicycle intersection model and existing HSM Part C, Chapter 10 model for three-leg stop control intersections on rural two-lane undivided roads (including total crashes from HSM model). Figure 70. Comparison of predicted average bicycle crashes per year from new bicycle intersection model and existing HSM Part C, Chapter 10 model for three-leg stop control intersections on rural two-lane undivided roads.

Figure 72. Comparison of predicted average bicycle crashes per year from new bicycle intersection model and existing HSM Part C, Chapter 12 model for four-leg signal control intersections on urban and suburban arterials.

Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology 273   model presented in this section and the predicted average total bicycle crash frequency from the existing HSM Part C, Chapter 12 model for four-leg signal control intersections. The predicted average crash frequencies are for the base conditions for both models. Comparisons from the bicycle intersection model are shown for two bicycle peak-hour crossing volume levels: 26 to 50 bike/hr and 201 to 300 bike/hr. Two levels of minor-road traffic volumes were assumed for the comparison: 10,000 and 20,000 veh/day. A daily pedestrian crossing volume of 1,000 ped/day was also assumed for use in the HSM model. To provide more perspective, Figure 73 provides the same comparison but also includes estimates for total crashes (i.e., multiple-vehicle + single-vehicle + pedestrian + bicycle crashes) from the existing HSM Part C, Chapter 12 model for four-leg signal control intersections. Based on this assessment, the new bicycle intersection model provides similar estimates for bicycle crashes as compared to the existing Chapter 12 model, although bicycle crashes increase with increases in the major-road volume at a slightly higher rate for the HSM model as compared to the new bicycle intersection model. Similar comparisons of the new bicycle intersection model were performed for additional intersection types in HSM Part C, and in all cases the new bicycle intersection segment model appears compatible with the existing HSM Part C models for intersections. Based on the compatibility testing, it appears that with calibration the new bicycle intersection model is compatible with the existing HSM Part C intersection models. 4.4 Summary and Recommendations Section 4 presents several crash prediction models to estimate pedestrian and bicycle crashes for potential incorporation into the HSM based on the crash prediction models initially devel- oped by usRAP and its international partner iRAP. The RAP models for pedestrian and bicycle crashes were developed for worldwide application and have been adapted in the current research in several ways to better fit within the structure and format of the HSM. Comparisons of the pedestrian and bicycle SPFs were developed based on the RAP models with existing HSM Part C models and demonstrated that the new models appear compatible with existing HSM models. However, even with the introduction of a facility-type factor to account for the roadway geo- metric, traffic control, and traffic operational characteristics that typically differ between facility types, additional calibration of the models will likely be necessary for the new pedestrian and bicycle SPFs to be compatible with the existing HSM models. Based on the compatibility testing of the new models to check the reasonableness of their results, it is recommended that the pedestrian and bicycle SPFs developed herein, based on the RAP models, be considered for potential integration into the following: • Rural two-lane, two-way roads chapter: HSM Part C, Chapter 10 or its successor in the HSM2. • Rural multilane highways chapter: HSM Part C, Chapter 11 or its successor in the HSM2. • Urban and suburban arterials chapter: HSM Part C, Chapter 12 or its successor in the HSM2. Because the pedestrian and bicycle SPFs from the RAP models and modified herein were developed from findings in the literature and not from negative binomial regression analysis, there are no overdispersion parameters for use with these SPFs. As a result, these SPFs cannot be applied directly with the EB method; however, they can still be incorporated into the overall predictive method of HSM Part C. Another limitation of the intersection models is that the models do not address roundabouts.

Figure 73. Comparison of predicted average bicycle crashes per year from new bicycle intersection model and existing HSM Part C, Chapter 12 model for four-leg signal control intersections on urban and suburban arterials (including total crashes from HSM model).

Development of Pedestrian and Bicycle Models Based on Road Assessment Program Methodology 275   Another notable way that the pedestrian and bicycle SPFs developed herein are different from existing HSM Part C models is that the exposure measures for pedestrians and bicycles (i.e., the pedestrian and bicycle flow factors) are based on peak-hour volumes. All existing HSM Part C predictive models are based on daily volumes, either motor vehicle traffic volumes or pedestrian volumes. While peak-hour pedestrian and bicycle flows are used in determining the values of the pedestrian and bicycle flow factors, respectively, model predictions still represent annual crash frequencies.

Next: Section 5 - Development of Pedestrian and Bicycle Models in the Absence of Pedestrian and Bicyclist Exposure Data »
Pedestrian and Bicycle Safety Performance Functions Get This Book
×
 Pedestrian and Bicycle Safety Performance Functions
Buy Paperback | $121.00
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

Each year, national crash studies have estimated that while overall traffic fatalities are decreasing, the percentages of those fatalities among pedestrians and cyclists are increasing.

NCHRP Research Report 1064: Pedestrian and Bicycle Safety Performance Functions, from TRB's National Cooperative Highway Research Program, presents state departments of transportation and other transportation professionals with an update of pedestrian and bicycle safety performance functions (SPFs).

Supplemental to the report are three spreadsheet tools that address SPFs on rural multilane roads, rural two-lane roads, and urban/suburban arterials.

READ FREE ONLINE

  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!