1. Prediction of Crash Frequency for Suburban/Urban Multilane Streets
HSM Practitioners Guide to Rural Multilane Highways and Urban Suburban Multilane Streets
Prediction of Crash Frequency for Suburban/Urban Multilane Streets
- Session #5
Session #5 – Prediction of Crash Frequency for suburban and urban multilane streets

2. Predicting Crash Frequency of Suburban/Urban Multilane Streets
Learning Outcomes:
Describe the models to Predict Crash Frequency for Multilane Suburban/Urban Streets
Describe Crash Modification Factors for Multilane Suburban/Urban Streets
Apply Crash Modification Factors (CMF’s) to Predicted Crash Frequency for Multilane Suburban/Urban Streets
Learning Outcomes for Session #5 on Suburban Urban multilane streets SPF’s

3. Defining Urban Multilane Highways
HSM Methodology applies to arterial four-lane undivided and divided urban and suburban highways.
Urban and Suburban areas are defined as areas within the urban and urbanized area boundaries established by FHWA. These include all areas with populations of 5,000 or more.
Some areas beyond the FHWA boundaries may be treated as urban or suburban if the boundaries have not been adjusted to include recent development.
The boundary dividing rural and urban areas can at times be difficult to determine, especially since most multilane rural highways are located on the outskirts of urban agglomerations.
These procedures may be used for any multilane road in which the general design features and land use setting are urban or suburban in nature rather than rural.
Instructor: From Chapter 12 of the Highway Safety Manual
12.1. introDuction
This chapter presents the predictive method for urban and suburban arterial facilities. A general introduction to the Highway Safety Manual (HSM) predictive method is provided in the Part C—Introduction and Applications Guidance.
The predictive method for urban or suburban arterial facilities provides a structured methodology to estimate the expected average crash frequency, crash severity, and collision types for facilities with known characteristics. All types of crashes involving vehicles of all types, bicycles, and pedestrians are included, with the exception of crashes between bicycles and pedestrians. The predictive method can be applied to existing sites, design alternatives to existing sites, new sites, or for alternative traffic volume projections. An estimate can be made for crash frequency in a period of time that occurred in the past (i.e., what did or would have occurred) or in the future (i.e., what is expected to occur). The development of the SPFs in Chapter 12 is documented by Harwood et al. (8, 9). The CMFs used in this chapter have been reviewed and updated by Harkey et al. (6) and in related work by Srinivasan et al. (13). The SPF coefficients, default collision type distributions, and default nighttime crash proportions have been adjusted to a consistent basis by Srinivasan et al. (14).
Session 4 – Predicting Highway Safety for Multilane Urban Streets

4. Predicting Crash Frequency of Suburban/Urban Multilane Streets
Separate Prediction Models for:
Homogeneous highway segments
Intersections
Sum of Individual Intersection Calculations
Key Message: The Analysis divides the highway into homogeneous analysis sections.
Additional Info: Analysis sections include both (1) homogeneous highway segments, and (2) individual intersections. Each analysis section is homogenous with respect to geometry and traffic conditions.
Homogeneous highway segments have uniform horizontal, vertical, cross section, traffic characteristics, and roadside geometry. At any location where there is a change in geometry (e.g., changing from a horizontal curve to a tangent or a change in shoulder width) or a change in traffic volume, a new highway segment begins.
For suburban/urban, the HSM does not apply CMF for horizontal curves
Each intersection is also defined as a separate, homogenous analysis section.
Question/Interactivity: Ask participants to study the roadway plans for IHSDM Pike and to identify the first few homogenous highway segments. Show how the first tangent and horizontal curve would be different homogeneous segments. (refer to sheet 2 of the plan/profile)
Reference:
Session 4 – Predicting Highway Safety for Multilane Urban Streets

5. Definition of Segments and Intersections
From Final HSM
Section 12.4 provides an explanation of the predictive method. Sections 12.5 through 12.8 provide the specific detail necessary to apply the predictive method steps. Detail regarding the procedure for determining a calibration factor to apply in Step 11 is provided in the Part C, Appendix A.1. Detail regarding the EB Method, which is applied in Steps 6, 13, and 15, is provided in the Part C, Appendix A.2.
In Step 5 of the predictive method, the roadway within the defined limits is divided into individual sites, which are homogenous roadway segments and intersections. A facility consists of a contiguous set of individual intersections and roadway segments, referred to as “sites.” A roadway network consists of a number of contiguous facilities.
Predictive models have been developed to estimate crash frequencies separately for roadway segments and intersections. The definitions of roadway segments and intersections presented below are the same as those used in the FHWA Interactive Highway Safety Design Model (IHSDM) (4).
Roadway segments begin at the center of an intersection and end at either the center of the next intersection or where there is a change from one homogeneous roadway segment to another homogenous segment. The roadway segment model estimates the frequency of roadway-segment-related crashes which occur in Region B in Figure When a roadway segment begins or ends at an intersection, the length of the roadway segment is measured from the center of the intersection.
Chapter 12 provides predictive models for stop-controlled (three- and four-leg) and signalized (three- and four-leg) intersections. The intersection models estimate the predicted average frequency of crashes that occur within the limits of an intersection (Region A of Figure 12-2) and intersection-related crashes that occur on the intersection legs (Region B in Figure 12-2).
A - All crashes that occur within this region are classified as intersection crashes
B – Crashes in this region may be segment or intersection related, depending on the characteristics of the crash
Session 6 – Predicting Highway Safety for Intersections

6. Subdividing Roadway Segments
Before applying the safety prediction methodology to an existing or proposed rural segment facility, the roadway must be divided into analysis units consisting of individual homogeneous roadway segments and intersections.
A new analysis section begins at each location where the value of one of the following variables changes (alternatively a section is defined as homogenous if none of these variables changes within the section):
• Annual Average daily traffic (AADT) volume (veh/day)
• Number of through lanes
• Presence/Type of a median
Presence/Type of Parking
Roadside Fixed Object density
Presence of Lighting
• Speed category
Instructor:
Instructor should review carefully with workshop participants so they have a precise understanding.
Divide the roadway network or facility into individual homogenous roadway segments and intersections, which are referred to as sites.
The segmentation process produces a set of roadway segments of varying length, each of which is homogeneous with respect to characteristics such as traffic volumes and key roadway design characteristics and traffic control features. Figure 12-2 shows the segment length, L, for a single homogenous roadway segment occurring between two intersections. However, several homogenous roadway segments can occur between two intersections. A new (unique) homogeneous segment begins at the center of each intersection and where there is a change in at least one of the following characteristics of the roadway:
■■ ■Annual average daily traffic volume (AADT) (vehicles/day)
■■ ■Number of through lanes
■■ ■Presence/type of median
■■ ■Presence/type of on-street parking
■■ ■Roadside fixed object density
■■ ■Presence of lighting
■■ ■Speed category (based on actual traffic speed or posted speed limit)
Session 4 – Predicting Highway Safety for Multilane Urban Streets

7. Subdividing Roadway Segments
homogeneous roadway segments – Median Width:
Instructor: From Chapter 12 of the Highway Safety Manual
The above are rounded widths for medians (without barriers) that are recommended before determining homogeneous segments.
Session 4 – Predicting Highway Safety for Multilane Urban Streets

8. Predicting Crash Frequency of Suburban/Urban Multilane Streets
Procedure for safety prediction for a roadway segment:
Combine base models, CMFs, and calibration factor
Nspf rs = Nbrmv + Nbrsv + Nbrdwy
Nbr = Nspf rs (CMF1r x CMF2r x … CMFnr)
Instructor: from Chapter 12 of the Highway Safety Manual
Roadway Segments
The safety prediction procedure for roadway segments estimates the expected annual nonintersection-related accident frequency for each of the segments that make up a highway facility or improvement project. Nonintersection-related accidents include accidents that occur on roadway segments between intersections and accidents that occur near an intersection, but are not related to the intersection. The general procedure for safety prediction for a roadway segment, combining the base models, CMFs, and calibration factor, is presented below:
Npredicted rs = (Nbr + Npedr + Nbiker) Cr Equation 12-2
Nbr = Nspf rs (CMF1r CMF2r CMF3r) Equation 12-3
Where,
Npredicted rs= predicted number of total roadway segment accidents per year;
Nbr= predicted number of total roadway segment accidents per year for base conditions (excluding vehicle-pedestrian and vehicle-bicycle collisions);
Nspf rs = predicted number of total roadway segment accidents per year for base conditions (excluding vehicle-pedestrian and vehicle-bicycle collisions);
Npedr = predicted number of vehicle-pedestrian collisions per year;
Nbiker = predicted number of vehicle-bicycle collisions per year;
CMF1r CMFnr = accident modification factors for roadway segments; and
Cr = calibration factor for roadway segments of a specific type developed for use for a particular geographical area.
Equation 12-2 shows that roadway segment accident frequency is estimated as the sum of three components: Nbr, Npedr, and Nbiker. The following equation shows that the base model portion of Nbr, designated as Nbrbase, is further broken down into three components by collision type:
Nspf rs = Nbrmv + Nbrsv + Nbrdwy (Equation 12-4)
Nbrmv = predicted number of multiple-vehicle nondriveway collisions per year for nominal or base conditions;
Nbrsv = predicted number of single-vehicle collision and noncollision accidents per year for nominal or base conditions; and
Nbrdwy = predicted number of multiple-vehicle driveway-related collisions per year.
Npredicted rs = (Nbr + Npedr + Nbiker) Cr
Session 4 – Predicting Highway Safety for Multilane Urban Streets

9. Crash Frequency Prediction Models for Urban/Suburban Roadway Segments
Five types of Collisions are considered:
Multiple-vehicle nondriveway crashes
Single-vehicle crashes
Multiple-vehicle driveway related crashes
Vehicle-pedestrian crashes
Vehicle-bicycle collisions
Instructor: from Chapter 12 of the Highway Safety Manual
The procedures addresses five types of collisions noted here.
Safety performance functions for urban and Suburban arterial roadway Segments
The predictive model for predicting average crash frequency on a particular urban or suburban arterial roadway segment was presented in Equation The effect of traffic volume (AADT) on crash frequency is incorporated through the SPF, while the effects of geometric design and traffic control features are incorporated through the CMFs. The SPF for urban and suburban arterial roadway segments is presented in this section. Urban and suburban arterial roadway segments are defined in Section 12.3.
SPFs and adjustment factors are provided for five types of roadway segments on urban and suburban arterials:
■■ ■Two-lane undivided arterials (2U)
■■ ■Three-lane arterials including a center two-way left-turn lane (TWLTL) (3T)
■■ ■Four-lane undivided arterials (4U)
■■ ■Four-lane divided arterials (i.e., including a raised or depressed median) (4D)
■■ ■Five-lane arterials including a center TWLTL (5T)
Guidance on the estimation of traffic volumes for roadway segments for use in the SPFs is presented in Step 3 of the predictive method described in Section The SPFs for roadway segments on urban and suburban arterials are applicable to the following AADT ranges:
■■ ■2U: 0 to 32,600 vehicles per day
■■ ■3T : 0 to 32,900 vehicles per day
■■ ■4U: 0 to 40,100 vehicles per day
■■ ■4D: 0 to 66,000 vehicles per day
■■ ■5T: 0 to 53,800 vehicles per day
Application to sites with AADTs substantially outside these ranges may not provide reliable results.
Other types of roadway segments may be found on urban and suburban arterials but are not addressed by the predictive model in Chapter 12.
The procedure addresses five types of collisions. The corresponding equations, tables, and figures are indicated in Table 12-2 above:
■■ ■multiple-vehicle nondriveway collisions
■■ ■single-vehicle crashes
■■ ■multiple-vehicle driveway-related collisions
■■ ■vehicle-pedestrian collisions
■■ ■vehicle-bicycle collisions
The predictive model for estimating average crash frequency on roadway segments is shown in Equations 12-2 through The effect of traffic volume on predicted crash frequency is incorporated through the SPFs, while the effects of geometric design and traffic control features are incorporated through the CMFs. SPFs are provided for multiple-vehicle nondriveway collisions and single-vehicle crashes. Adjustment factors are provided for multi-vehicle driveway-related, vehicle-pedestrian, and vehicle-bicycle collisions.
Session 4 – Predicting Highway Safety for Multilane Urban Streets

10. Predicting Crash Frequency of Suburban/Urban Multilane Streets
Nspf rs = Nbrmv + Nbrsv + Nbrdwy
Where:
Nbrmv = Predicted number of multiple-vehicle non-driveway crashes per year for base conditions
Nbrsv = Predicted number of single-vehicle collision and non-collision crashes per year for base conditions
Instructor: from Chapter 12 of the Highway Safety Manual
The following equation shows that the SPF portion of Nbr, designated as Nspf rs, is further separated into three components by collision type shown in Equation 12-4:
Nspf rs = Nbrmv + Nbrsv + Nbrdwy
Where,
Nbrmv = predicted number of multiple-vehicle nondriveway collisions per year for nominal or base conditions;
Nbrsv = predicted number of single-vehicle collision and noncollision accidents per year for nominal or base conditions; and
Nbrdwy = predicted number of multiple-vehicle driveway-related collisions per year.
Lead discussion as to just what are segment single vehicle crashes on an urban/suburban street; although much fewer in number, they are still part of the total crashes; these crashes involve a single vehicle striking the median or outside curb or roadside objects.
Nbrdwy = Predicted number of multiple-vehicle driveway related crashes per year
Session 4 – Predicting Highway Safety for Multilane Urban Streets

11. Predicting Crash Frequency of Suburban/Urban Multilane Streets
Nbr = Nspf rs x (CMF1r x CMF2r x .. CMFnr)
Where:
Nbr = Predicted number of total roadway segment crashes per year with CMFs applied (excluding ped and bike collisions)
Nspf rs = Predicted number of total roadway segment crashes per year for base conditions
Instructor: from Chapter 12 of the Highway Safety Manual
Nbr is comprised of the predicted total roadway segment crashes per year for base conditions (excluding vehicle-pedestrian and vehicle-bicycle collisions).
Lead discussion as to just what are segment single vehicle crashes on an urban/suburban street; although much fewer in number, they are still part of the total crashes; these crashes involve a single vehicle striking the median or outside curb or roadside objects.
CMF1r CMF2r, .. CMFnr = Crash Modification Factors for roadway segments
Session 4 – Predicting Highway Safety for Multilane Urban Streets

12. Predicting Crash Frequency of Suburban/Urban Multilane Streets
Npredicted rs = (Nbr + Npedr + Nbiker) Cr
Where:
Npredicted rs = Predicted number of total roadway segment crashes per year
Nbr = Predicted number of total roadway segment
crashes per year with CMFs applied
Npedr = Predicted number of vehicle-pedestrian collisions per year
Nbiker = Predicted number of vehicle-bicycle
collisions per year
Cr = calibration factor for a particular geograhical area
Instructor: from Chapter 12 of theHighway Safety Manual
Npredicted rs is the predicted number of total roadway segment crashes per year and is the sum of N base + predicted ped crashes + predicted cyclist crashes all times a calibration factor
Roadway Segments
The safety prediction procedure for roadway segments estimates the expected annual nonintersection-related accident frequency for each of the segments that make up a highway facility or improvement project. Nonintersection-related accidents include accidents that occur on roadway segments between intersections and accidents that occur near an intersection, but are not related to the intersection. The general procedure for safety prediction for a roadway segment, combining the base models, CMFs, and calibration factor, is presented below:
Npredicted rs = (Nbr + Npedr + Nbiker) Cr Equation 12-2
Nbr = Nspf rs (CMF1r CMF2r CMF3r) Equation 12-3
Where,
Npredicted rs= predicted number of total roadway segment accidents per year;
Nbr= predicted number of total roadway segment accidents per year for base conditions (excluding vehicle-pedestrian and vehicle-bicycle collisions);
Nspf rs = predicted number of total roadway segment accidents per year for base conditions (excluding vehicle-pedestrian and vehicle-bicycle collisions);
Npedr = predicted number of vehicle-pedestrian collisions per year;
Nbiker = predicted number of vehicle-bicycle collisions per year;
CMF1r CMFir = accident modification factors for roadway segments; and
Cr = calibration factor for roadway segments of a specific type developed for use for a particular geographical area.
Session 4 – Predicting Highway Safety for Multilane Urban Streets

13. Combining Safety Predictions for an Entire Series of Segments
Ntotal predicted = Sum Nrs + Sum Nint
Where:
Ntotal predicted = Predicted crash frequency for the entire arterial street
Nrs = Predicted number of total roadway segment crashes
Nint = Predicted number of total intersection- related crashes
Instructor: from Chapter 12 of the Highway Safety Manual
Grand total predicted number of crashes per year is the sum of the segment + the sum of the intersections
Combining Safety Predictions for an Entire Arterial Facility or Project
The total estimated number of crashes within the network or facility limits during a study period of n years is calculated using
Ntotal = Σ Nrs Σ Nint
all rdwy all intersections
segments
Where,
Ntotal= total expected number of crashes within the limits of a rural two-lane two-way road facility for the period of interest. Or, the sum of the expected average crash frequency for each year for each site within the defined roadway limits within the study period;
Nrs= expected average crash frequency for a roadway segment using the predictive method for one specific year; and
Nint= expected average crash frequency for an intersection using the predictive method for one specific year.
Equation 12-8 represents the total expected number of crashes estimated to occur during the study period.
Session 4 – Predicting Highway Safety for Multilane Urban Streets

14. Crash Frequency Prediction Models for Urban/Suburban Roadway Segments
Base Models and Adjustment Factors addresses five types of Roadway Segments:
(2U) Two-lane undivided arterials
(3T) Three-lane arterials including a center two-way Left Turn Lane
(4U) Four-lane undivided arterials
(4D) Four-lane divided arterials (including a raised or depressed median)
(5T) Five-lane arterials including a center TWLTL
Instructor: from Chapter 12 of the Highway Safety Manual
For multilane urban/suburban streets, there are five (5) different crash prediction sets of model coefficients as a function of the number of lanes and median type (i.e., 5 types of roadway segments on urban and suburban arterials).
Within the models are the separate prediction of multiple vehicle crashes, single vehicle crashes, driveway related crashes, ped, and cyclist crashes
Session 4 – Predicting Highway Safety for Multilane Urban Streets

15. Limitations as to AADT for Urban/Suburban Roadway Models
Instructor: from Chapter 12 of the Highway Safety Manual
Guidance on the estimation of traffic volumes for roadway segments for use in the SPFs is presented in Step 3 of the
predictive method described in Section The SPFs for roadway segments on urban and suburban arterials are
applicable to the following AADT ranges:
■■ ■2U: 0 to 32,600 vehicles per day
■■ ■3T : 0 to 32,900 vehicles per day
■■ ■4U: 0 to 40,100 vehicles per day
12-18 HIGHWAy SAFeTy MAnuAL
■■ ■4D: 0 to 66,000 vehicles per day
■■ ■5T: 0 to 53,800 vehicles per day
Application to sites with AADTs substantially outside these ranges may not provide reliable results.
Session 4 – Predicting Highway Safety for Multilane Urban Streets

16. Crash Frequency Prediction Models for Urban/Suburban Roadway Segments
No procedure has been developed for application to six-lane undivided (6U) nor for six-lane divided (6D) arterials.
- Until such procedures are developed:
The procedures for 4U arterials may be applied to 6U arterials and for 4D arterials to 6D arterials.
These procedures should be applied cautiously to 6U and 6D arterials because this application is not based on data for 6U and 6D arterials.
Instructor: from Chapter 12 of the Highway Safety Manual .
Session 4 – Predicting Highway Safety for Multilane Urban Streets

17. Crash Frequency Prediction Models for Urban/Suburban Roadway Segments
Multiple-Vehicle NonDriveway Crashes
Nbrmv = exp(a + b ln(AADT) + ln(L))
Where:
AADT = Annual Average Daily Traffic (veh/day)
L = Length of roadway segment (mi)
a & b = regression coefficients (Table 12-3)
Instructor: from Chapter 12 of the Highway Safety Manual
Multiple-Vehicle Nondriveway Collisions
The SPF for multiple-vehicle nondriveway collisions is applied as follows:
Nbrmv = exp(a + b × In(AADT) + In(L)) (12-10)
Where:
AADT = average annual daily traffic volume (vehicles/day) on roadway segment;
L = length of roadway segment (mi); and
a, b = regression coefficients.
Table 12-3 presents the values of the coefficients a and b used in applying Equation The overdispersion parameter, k, is also presented in Table 12-3.
Session 4 – Predicting Highway Safety for Multilane Urban Streets

18. Multiple-Vehicle NonDriveway Crashes
Nbrmv = exp(a + b ln(AADT) + ln(L))
Instructor: from Chapter 12 of the Highway Safety Manual
Please note that the form of the crash prediction model (equation) does not change, but rather the coefficients change as a function of the number of lanes and median type
A larger size of this table is on the following slide that can be used to compute only fatal & injury crashes, or PDO (property damage only) crashes.
Session 4 – Predicting Highway Safety for Multilane Urban Streets

19. Instructor: from Chapter 12 of the Highway Safety Manual
Instructor: from Chapter 12 of the Highway Safety Manual. Full table of Table 12-3 with a & b coefficients for Total, Fatal-Injury, and PDO crashes.
Session 4 – Predicting Highway Safety for Multilane Urban Streets

20. Predicting Crash Frequency for a Suburban Street – Example:
4-lane Undivided commercial Suburban Street:
AADT = 24,000
Length = 3.6 miles
1st, Calculate Predicted Crash Frequency for Multiple-Vehicle NonDriveway Crashes - use 4U coefficients
Nbrmv = exp(a + b ln(AADT) + ln(L))
Instructor: example calculation of predicted number of multiple vehicle segment crashes
A, b coefficients from Table 12-3 for 4U model
= exp( ln(24,000) + ln(3.6))
= exp(3.065)
= 21.4 crashes/yr
Session 4 – Predicting Highway Safety for Multilane Urban Streets

21. Safety Performance Function (SPF)
Highway Safety Manual Approach:
Average Crash Rate
“one rate”
Instructor:
Rather than “one” rate as with Crash Rate, the Highway Safety Manual SPF’s better conform to the observed Crash Frequency based upon exposure. 5T performs the worst and 4D performs the best.
Session 2 – Predicting Highway Safety for Multilane Rural Highway Segments

22. “Is this a Higher Crash Frequency Site?”
Highway Safety Manual Approach:
“Substantive Crash Frequency”
17 crashes/yr
“Difference”
“Predicted Crash Frequency”
Instructor: discussion as the meaning of being “below” the predicted crash Frequency curve or being “above” the curve
Lowest crash frequency is for the 4D model (spf). Highest is for 5T.
2.5 crashes/yr
Session 2 – Predicting Highway Safety for Multilane Rural Highway Segments

23. Crash Frequency Prediction Models for Urban/Suburban Roadway Segments
Single Vehicle Crashes:
Nbrsv = exp(a + b ln(AADT) + ln(L))
Where:
AADT = Annual Average Daily Traffic (veh/day)
L = Length of roadway segment (mi)
a & b = regression coefficients (Table 12- 5)
Instructor: from Chapter 12 of the Draft Highway Safety Manual
Single-Vehicle Crashes
SPFs for single-vehicle crashes for roadway segments are applied as follows:
Nbrsv = exp(a + b × In(AADT) + In(L)) (12-13)
Table 12-5 presents the values of the coefficients and factors used in Equation for each roadway type.
Equation is first applied to determine Nbrsv using the coefficients for total crashes in Table Nbrsv is then divided into components by severity level; Nbrsv(FI) for fatal-and-injury crashes and Nbrsv(PDO) for property-damage only crashes. Preliminary values of Nbrsv(FI) and Nbrsv(PDO), designated as N’brsv(FI) and N’brsv(PDO) in Equation 12-14, are determined with Equation using the coefficients for fatal-and-injury and property-damage-only crashes, respectively, in Table The following adjustments are then made to assure that Nbrsv(FI) and Nbrsv(PDO) sum to Nbrsv:
Session 4 – Predicting Highway Safety for Multilane Urban Streets

24. Single Vehicle NonDriveway Crashes
Nbrsv = exp(a + b ln(AADT) + ln(L))
Instructor: from Chapter 12 of the Draft Highway Safety Manual
Table of the coefficients for prediction of single vehicle crashes
Session 4 – Predicting Highway Safety for Multilane Urban Streets

25. Instructor: from Chapter 12 of the Highway Safety Manual
larger view of Table 12-5 for the coefficients for single vehicle crash prediction model that include a & b coefficients for calculating fatal-injury and pdo crashes.
Session 4 – Predicting Highway Safety for Multilane Urban Streets

26. Predicting Crash Frequency for a Suburban Street – Example:
4-lane Undivided commercial Suburban Street:
AADT = 24,000
Length = 3.6 miles
- Predicted Crash Frequency for Single-Vehicle NonDriveway Crashes - use 4U coefficients
Nbrsv = exp(a + b ln(AADT) + ln(L))
Instructor: from Chapter 12 of the Draft Highway Safety Manual
Example calculation for single vehicle crashes using the coefficients from Table12-5 for the 4U model
= exp( ln(24,000) + ln(3.6))
= exp(1.46)
= 4.3 crashes/yr
Session 4 – Predicting Highway Safety for Multilane Urban Streets

27. Driveway Related Crashes
72% of driveway related crashes involve a left turning vehicle – either into, or out of, the driveway
FHWA HAS Intersection outreach Brochure by Kittleson and Assoc
*FHWA-SA Access Management in the Vicinity of Intersections
Session 4 – Predicting Highway Safety for Multilane Urban Streets

28. Predicting Crash Frequency of Suburban/Urban Multilane Streets
Multiple-Vehicle Driveway Related Crashes
Major driveways are those that serve 50 or more parking spaces
Minor driveways serve sites with less than 50 parking spaces
Major residential driveways have AADT greater than 900 vpd
Minor residential driveways have “AADT less than 900 vpd”
Instructor: from Chapter 12 of the Highway Safety Manual
The criteria of less than 900 vpd for minor residential driveways is not in Chapter 4 but developed by Fred Ranck from his experience.
Major or minor – functional choice based upon # of parking spaces for the business or residential apt/condo buildings
Per Fred Ranck, research has no specific threshold to distinguish between residential and nonresidential roadways, however, residential traffic volumes range from ,100 vpd, therefore use 900 vpd as the threshold. Multiple-Vehicle Driveway-Related Collisions
The model presented above for multiple-vehicle collisions addressed only collisions that are not related to driveways.
Driveway-related collisions also generally involve multiple vehicles, but are addressed separately because the frequency of driveway-related collisions on a roadway segment depends on the number and type of driveways. Only unsignalized driveways are considered; signalized driveways are analyzed as signalized intersections.
The total number of multiple-vehicle driveway-related collisions within a roadway segment is determined as:
Where:
Nj = Number of driveway-related collisions per driveway per year for driveway type j from Table 12-7;
nj = number of driveways within roadway segment of driveway type j including all driveways on both sides of the
road; and
t = coefficient for traffic volume adjustment from Table 12-7.
The number of driveways of a specific type, nj, is the sum of the number of driveways of that type for both sides of the road combined. The number of driveways is determined separately for each side of the road and then added together.
Seven specific driveway types have been considered in modeling. These are:
■■ ■Major commercial driveways
■■ ■Minor commercial driveways
■■ ■Major industrial/institutional driveways
■■ ■Minor industrial/institutional driveways
■■ ■Major residential driveways
■■ ■Minor residential driveways
■■ ■Other driveways
Major driveways are those that serve sites with 50 or more parking spaces. Minor driveways are those that serve sites with less than 50 parking spaces. It is not intended that an exact count of the number of parking spaces be made for each site. Driveways can be readily classified as major or minor from a quick review of aerial photographs that show parking areas or through user judgment based on the character of the establishment served by the driveway.
Commercial driveways provide access to establishments that serve retail customers. Residential driveways serve single- and multiple-family dwellings. Industrial/institutional driveways serve factories, warehouses, schools, hospitals, churches, offices, public facilities, and other places of employment. Commercial sites with no restriction on access along an entire property frontage are generally counted as two driveways.
Session 4 – Predicting Highway Safety for Multilane Urban Streets

29. Predicting Crash Frequency of Suburban/Urban Multilane Streets
Multiple-Vehicle Driveway Related Crashes
Nbrdwy = SUM (nj Nj (AADT/15,000)t )
Where:
nj = number of driveways within roadway segment of driveway type j
Nj = Number of crashes per year for an individual driveway of driveway type j
from Table 12-7
t = coefficient for traffic volume adjustment
AADT = Annual Average Daily Traffic (veh/day)
Instructor: from Chapter 12 of the Draft Highway Safety Manual
Multiple-Vehicle Driveway-Related Collisions
The model presented above for multiple-vehicle collisions addressed only collisions that are not related to driveways.
Driveway-related collisions also generally involve multiple vehicles, but are addressed separately because the frequency of driveway-related collisions on a roadway segment depends on the number and type of driveways. Only unsignalized driveways are considered; signalized driveways are analyzed as signalized intersections.
The total number of multiple-vehicle driveway-related collisions within a roadway segment is determined as:
Where:
Nj = Number of driveway-related collisions per driveway per year for driveway type j from Table 12-7;
nj = number of driveways within roadway segment of driveway type j including all driveways on both sides of the
road; and
t = coefficient for traffic volume adjustment from Table 12-7.
The number of driveways of a specific type, nj, is the sum of the number of driveways of that type for both sides of the road combined. The number of driveways is determined separately for each side of the road and then added together.
Session 4 – Predicting Highway Safety for Multilane Urban Streets

30. Multiple-Vehicle Driveway Crashes
Nbrdwy = SUM (nj Nj (AADT/15,000)t ) Nj
Instructor: from Chapter 12 of the Highway Safety Manual
From previous slide: nj = number of driveways within the roadway segment of driveway type j (i.e., Major Commercial, Minor Commercial, Major Industrial, Minor Industrial, etc). t
Session 4 – Predicting Highway Safety for Multilane Urban Streets

31. Predicting Crash Frequency for a Suburban Street – Example:
Nbrdwy = SUM (nj Nj (AADT/15,000)t )
4-lane Undivided commercial Suburban Street:
AADT = 24,000
Length = 3.6 miles
3 major commercial driveways
42 minor commercial driveways
2 major industrial/institutional driveways
5 major residential driveways
2 minor residential driveways
7 other
61 total driveways
Instructor: from Chapter 12 of the Draft Highway Safety Manual
Example calculation for predicted segment driveway crashes for a total of 61 driveways in 3.6 miles at 24,000 adt
From previous slide: nj = number of driveways within the roadway segment of driveway type j (i.e., Major Commercial, Minor Commercial, Major Industrial, Minor Industrial, etc).
Session 4 – Predicting Highway Safety for Multilane Urban Streets

32. Predicting Crash Frequency for a Suburban Street – Example:
4-lane Undivided commercial Suburban Street:
Nbrdwy = SUM (nj Nj (AADT/15,000)t )
= 3 x (24,000/15,000)1.172
+ 42 x (24,000/15,000)1.172
+ 2 x (24,000/15,000)1.172
+ 0 x (24,000/15,000)1.172
+ 5 x (24,000/15,000)1.172
+ 2 x (24,000/15,000)1.172
+ 7 x (24,000/15,000)1.172
Instructor: from Chapter 12 of the Highway Safety Manual
Example Calculation for Driveway Crash prediction for the following driveway numbers by type:
3 major commercial driveways
42 minor commercial driveways
2 major industrial/institutional driveways
0 minor industrial/institutional driveways
5 major residential driveways
2 minor residential driveways
7 other
61 total driveways
Coefficients for Nj and b values are from Exhibit 12-11
Result is 7.1 segment driveway related crashes for the 61 driveways in 3.6 miles
= 7.1 crashes/yr
Session 4 – Predicting Highway Safety for Multilane Urban Streets

33. Predicting Crash Frequency of Suburban/Urban Multilane Streets
Nspf rs = Nbrmv + Nbrsv + Nbrdwy
Where:
Nspf rs = Predicted number of total roadway segment crashes per year for base conditions for suburban 4-Lane Undivided (4U) of 24,000 AADT for 3.6 miles
Nbrmv =
Nbrsv =
Nbrdwy =
Instructor: from Chapter 12 of the Highway Safety Manual
Nspf rs = Nbrmv + Nbrsv + Nbrdwy (Equation 12-4)
Where:
Nbrmv = predicted average crash frequency of multiple-vehicle nondriveway collisions for base conditions;
Nbrsv = predicted average crash frequency of single-vehicle crashes for base conditions; and
Nbrdwy = predicted average crash frequency of multiple-vehicle driveway-related collisions.
Thus, the SPFs and adjustment factors are applied to determine five components: Nbrmv, Nbrsv, Nbrdwy, Npedr, and Nbiker, which together provide a prediction of total average crash frequency for a roadway segment.
Equations 12-2 through 12-4 are applied to estimate roadway segment crash frequencies for all crash severity levels combined (i.e., total crashes) or for fatal-and-injury or property-damage-only crashes.
Nspf rs =
= 32.8 crashes per year
Session 4 – Predicting Highway Safety for Multilane Urban Streets

34. Applying Severity Index to Urban Suburban Multilane Streets
Example: Suburban Four Lane Undivided Segment (4U) street of 24,000 AADT for 3.6 miles;
Fatal and Injury crashes are 15 of 40 total crashes
a. Compute the actual Severity Index (SI)
SI4sg = Fatal + Injury Crashes = 15/40 = 0.375
Total Crashes
Instructor:
Compute the severity index for actual crash frequency performance of 15 inj and fatal crashes out of a total of 40 crashes = 0.375
SI4sg = 4-leg signalized intersection
Session 3 –Exercise I

35. Applying Severity Index to Urban Suburban Multilane Streets
Instructor:
b. Compute Predicted Fatal + Injury Crashes
Nbrmv = exp( ln( 24,000) + ln(3.6))
= 6.1
Predicting Highway Safety for Intersections on 2-Lane Rural Highways

36. Applying Severity Index to Urban Suburban Multilane Streets
Instructor:
b. Compute Predicted Fatal + Injury Crashes
Nbrsv = exp( ln( 24,000) + ln(3.6))
= 1.1
Predicting Highway Safety for Intersections on 2-Lane Rural Highways

37. Applying Severity Index to Urban Suburban Multilane Streets
Instructor:
b. Compute Predicted Fatal + Injury Crashes
Nbrdwy = Nbrdwy x Coefficient = 7.1 x 0.342
= crashes per year
Predicting Highway Safety for Intersections on 2-Lane Rural Highways

38. Applying Severity Index to Urban Suburban Multilane Intersections
Example: Suburban Four Lane Undivided Segment (4U) street of 24,000 ADT for 3.6 miles; Fatal and Injury crashes are 15 of 40 total crashes
a. Compute the actual Severity Index (SI)
SI = Fatal + Injury Crashes = 15/40 = 0.375
Total Crashes
b. Compute the Predicted Severity Index (SI)
SI = Fatal + Injury Crashes = ( )/32.8
Total Crashes =
Example of comparing Actual SI to Predicted SI. In this case, the Actual is higher than predicted which means more Fatal and Injury crashes are occuring than predicted.
Actual Severity is greater than Predicted Severity
Session 3 –Exercise I

39. Applying CMF’s for Conditions other than “Base”
- Next Step is:
Nbr = Nspf rs(CMF1r x CMF2r x .. CMFnr)
Where:
Nbr = Predicted number of total roadway segment crashes per year with CMFs applied
Nspf rs = Predicted number of total roadway segment crashes per year for base conditions
Instructor: from Chapter 12 of the Highway Safety Manual
The predictive models for roadway segments are presented in Equations 12-2 and 12-3 below.
Npredicted rs = Cr × (Nbr + Npedr + Nbiker) (Equatioin 12-2)
Nbr = Nspf rs × (CMF1r × CMF2r × … × CMFnr) (Equation 12-3)
Where:
Npredicted rs = predicted average crash frequency of an individual roadway segment for the selected year;
Nbr = predicted average crash frequency of an individual roadway segment (excluding vehiclepedestrian and vehicle-bicycle collisions);
Nspf rs = predicted total average crash frequency of an individual roadway segment for base conditions (excluding vehicle-pedestrian and vehicle-bicycle collisions);
Npedr = predicted average crash frequency of vehicle-pedestrian collisions for an individual roadway segment;
Nbiker = predicted average crash frequency of vehicle-bicycle collisions for an individual roadway segment;
CMF1r … CMFnr = crash modification factors for roadway segments; and
Cr = calibration factor for roadway segments of a specific type developed for use for a particular geographical area.
Equation 12-2 shows that roadway segment crash frequency is estimated as the sum of three components: Nbr, Npedr, and Nbiker. The following equation shows that the SPF portion of Nbr, designated as Nspf rs, is further separated into three components by collision type shown in Equation 12-4:
CMF1r CMF2r, .. CMFnr = Crash modification factors for roadway segments
Session 4 – Predicting Highway Safety for Multilane Urban Streets

40. Applying CMF’s for Conditions other than “Base”
Instructor: from Chapter 12 of the Highway Safety Manual
Table lists the CMFs to account for conditions that vary from the base conditions.
Session 4 – Predicting Highway Safety for Multilane Urban Streets

41. CMF for Curb Parking on Urban Streets
CMF1r = 1 + Ppk* (fpk -1.0)
Instructor: from Chapter 12 of the Highway Safety Manual
Trends indicate that crash frequency increases linearly with the percentage of business or office land use. Crashes are also more frequent on two-lane streets than multilane streets.
CMF1r—On-Street Parking
The CMF for on-street parking, where present, is based on research by Bonneson (1). The base condition is the absence of on-street parking on a roadway segment. The CMF is determined as:
CMF1r = 1 + ppk × (fpk − 1.0) (12-32)
Where:
CMF1r = crash modification factor for the effect of on-street parking on total crashes;
fpk = factor from Table 12-19;
ppk = proportion of curb length with on-street parking = (0.5 Lpk/L); and
Lpk = sum of curb length with on-street parking for both sides of the road combined (miles); and
L = length of roadway segment (miles).
This CMF applies to total roadway segment crashes.
The sum of curb length with on-street parking (Lpk) can be determined from field measurements or video log review to verify parking regulations. Estimates can be made by deducting from twice the roadway segment length allowances for intersection widths, crosswalks, and driveway widths.
Where:
Ppk = Proportion of curb length with parking, = (0.5Lpk/L)
Lpk = curb length with on-street parking, both sides (mi) combined
fpk = factor from Table 12-19
Session 4 – Predicting Highway Safety for Multilane Urban Streets

42. CMF for Curb Parking on Urban Streets
Example: For 4-Ln Urban commercial street (4U), angle parking one side 3.12 miles of 3.6 mile length, commercial area:
CMF1r = 1 + Ppk (fpk -1.0)
CMF1r = 1 + (0.50 (Lpk/L)1 ) x (fpk-1)
= 1 + (0.50 (3.12/3.6)1) x ( )
Instructor: from Chapter 12 of the Highway Safety Manual
Example calculation for parking one-side of 3.12 miles of a 3.6 mile length in a commercial area.
For one side use 1 x the 0.50 factor for parking; the comes from Table for angle parking commercial area
= 1 + (0.50(0.867)) x 2.999
= 1 + (0.43 x 2.999)
= 2.30
Session 4 – Predicting Highway Safety for Multilane Urban Streets

43. CMF for Curb Parking Urban Streets: Example
For 4-Ln Urban commercial street (4U), parallel parking both sides 3.12 miles of 3.6 mile length, commercial area:
CMF1r = 1 + Ppk (fpk -1.0)
CMF1r = 1 + (0.50(3.12/3.6)2) x ( ))
Instructor: from Chapter 12 of the Highway Safety Manual
Ask participants to calculate the CMF for parking two-side of 3.12 miles of a 3.6 mile length in a commercial area but as parallel parking
For two sides use 2 x the 0.50 factor for parking; the comes from Exhibit for parallel parking commercial area
= 1 + (0.5(0.867)2) x 0.709
= 1 + (0.867 x 0.709)
= 1.614
Session 4 – Predicting Highway Safety for Multilane Urban Streets

44. CMF for Roadside Fixed Objects
CMF2r = foffset * Dfo * pfo + (1 – pfo)
Where:
foffset = fixed object offset factor from Table 12-20
Dfo = fixed object density (fixed objects/mi)
pfo = fixed-object collisions as a proportion of
total crashes, Table 12-21
Instructor: from Chapter 12 of the Highway Safety Manual
CMF2r—Roadside Fixed Objects
The base condition is the absence of roadside fixed objects on a roadway segment. The CMF for roadside fixed objects, where present, has been adapted from the work of Zegeer and Cynecki (15) on predicting utility pole crashes.
The CMF is determined with the following equation:
CMF2r = foffset × Dfo × pfo + (1.0 − pfo) (12-33)
CHAPTER 12— Predictive Method for Urban and Suburban Arterials:
Where:
CMF2r = crash modification factor for the effect of roadside fixed objects on total crashes;
foffset = fixed-object offset factor from Table 12-20;
Dfo = fixed-object density (fixed objects/mi) for both sides of the road combined; and
pfo = fixed-object collisions as a proportion of total crashes from Table
This CMF applies to total roadway segment crashes. If the computed value of CMF2r is less than 1.00, it is set equal to This can only occur for very low fixed object densities.
In estimating the density of fixed objects (Dfo), only point objects that are 4 inches or more in diameter and do not have breakaway design are considered. Point objects that are within 70 ft of one another longitudinally along the road are counted as a single object. Continuous objects that are not behind point objects are counted as one point object for each 70 ft of length. The offset distance (Ofo) shown in Table is an estimate of the average distance from the edge of the traveled way to roadside objects over an extended roadway segment. If the average offset to fixed objects exceeds 30 ft, use the value of foffset for 30 ft. Only fixed objects on the roadside on the right side of the roadway in each direction of travel are considered; fixed objects in the roadway median on divided arterials are not considered.
Only point objects that are 4inches or more in diameter and do not have a breakaway design are considered.
Point objects that are within 70 feet of each other longitudinally are considered as a single object
Session 4 – Predicting Highway Safety for Multilane Urban Streets

45. CMF for Roadside Fixed Objects
Example: For 4-Ln Urban undivided street (4U) with power poles at 2 ft offset
foffset = 0.232
pfo = 0.037
Offset is measured from edge of travel way
Instructor: from Chapter 12 of the Highway Safety Manual
Note that the proportion for fixed object crashes changes as a function of the road type
Roadside Fixed Objects (CMF2r)
The base condition is the absence of roadside fixed objects on a roadway segment. The CMF for roadside fixed objects, where present, has been adapted from the work of Zegeer and Cynecki (3) on predicting utility pole crashes. The CMF is determined with the following equation:
CMF2r = foffset Dfo pfo + (1 – pfo) (Equation 12-33)
Where,
CMF2r = accident modification factor for the effect of roadside fixed objects on total crashes;
foffset = fixed-object offset factor from Table 12-20;
Dfo = fixed-object density (fixed objects/mi) for both sides of the road combined; and
pfo = fixed-object collisions as a proportion of total crashes from Table 12-21
This CMF applies to total roadway segment crashes. If the computed value of CMF2r is less than 1.0, it should be set equal to 1.0. This can only occur for very low fixed object densities.
In estimating the density of fixed objects (Dfo), only point objects that are 4 inches or more in diameter and do not have breakaway design are considered. Point objects that are within 70 feet of one another longitudinally along the road should be counted as a single object. Continuous objects that are not behind point objects should be counted as one point object for each 70 ft of length. The offset distance (Ofo) shown in Table is an estimate of the average distance from the edge of the traveled way to roadside objects over an extended roadway segment. If the average offset to fixed objects exceeds 30 ft, use the value of foffset for 30 ft.
Session 4 – Predicting Highway Safety for Multilane Urban Streets

46. CMF for Roadside Fixed Objects: Example
For one mile of 4-Ln Urban undivided commercial curbed street (4U) with power poles on one side on 150 foot spacing 2 feet from edge of travel way:
CMF2r = foffset x Dfo x pfo + (1 – pfo)
= (5280/150)(1)(0.037)+ (1 – 0.037)
= x 35.2 x (0.963)
Instructor: from Chapter 12 of the Highway Safety Manual
Example calculation for 2 feet back of curb for power poles for poles on one side of a Commercial 4U =
= 1.265
Session 4 – Predicting Highway Safety for Multilane Urban Streets

47. CMF for Roadside Fixed Objects: Example
For one mile of 4-Ln Urban undivided commercial curbed street (4U) with power poles on both sides on 150 foot spacing 2 feet from edge of travel way:
CMF2r = foffset x Dfo x pfo + (1 – pfo)
= (5280/150)(2)(0.037))+(1 – 0.037)
Instructor: from Chapter 12 of the Highway Safety Manual
Example calculation for 2 feet back of curb for power poles for poles on both sides of a Commercial 4U
= x 70.4 x (0.963)
= 1.567
Session 4 – Predicting Highway Safety for Multilane Urban Streets

48. CMF3r for Median Width – Urban/Suburban Multilane Streets
This CMF applies only to divided roadway segments with traversable medians without barrier.
The effect of traffic barriers on safety would be expected to be a function of barrier type and offset, rather than the median width; however, the effects of these factors on safety have not been quantified. Until better information is available, an CMF value of 1.00 is used for medians with traffic barriers.
Instructor: from Chapter 12 of the Highway Safety Manual
CMF3r - Median Width
An CMF for median widths on divided roadway segments of urban and suburban arterials is presented in Table based on the work of Harkey et al.(6). The base condition for this CMF is a median width of 15-ft. The CMF applies to total crashes and represents the effect of median width in reducing cross-median collisions; the CMF assumes that nonintersection collision types other than cross median collusions are not affected by median width.
This CMF applies only to traversable medians without traffic barriers. The effect of traffic barriers on safety would be expected to be a function of barrier type and offset, rather than the median width; however, the effects of these factors on safety have not been quantified. Until better information is available, an CMF value of 1.00 is used for medians with traffic barriers. The value of this CMF is 1.00 for undivided facilities.
Session 4 – Predicting Highway Safety for Multilane Urban Streets

49. CMF4r = 1- (pnr x (1.0 – 0.72 pinr – 0.83 ppnr ))
CMF for Lighting
CMF4r = 1- (pnr x (1.0 – 0.72 pinr – 0.83 ppnr ))
Where:
pinr = proportion of total nighttime crashes for unlighted roadway segments that involve a nonfatal injury
ppnr = proportion of total nighttime crashes for unlighted roadway segments that involve PDO crashes only
pnr = proportion of total crashes for unlighted roadway segments that occur at night
Instructor: from Chapter 12 of the Highway Safety Manual
CMF4r—Lighting
The base condition for lighting is the absence of roadway segment lighting (CMF4r = 1.00). The CMF for lighted
roadway segments is determined, based on the work of Elvik and Vaa (3), as:
CMF4r = 1.0 − (pnr × (1.0 − 0.72 × pinr − 0.83 × ppnr)) (12-34)
Where:
CMF4r = crash modification factor for the effect of roadway segment lighting on total crashes;
pinr = proportion of total nighttime crashes for unlighted roadway segments that involve a fatality or injury;
ppnr = proportion of total nighttime crashes for unlighted roadway segments that involve property damage only; and
pnr = proportion of total crashes for unlighted roadway segments that occur at night.
CMF4r applies to total roadway segment crashes. Table presents default values for the nighttime crash proportions
pinr, ppnr, and pnr. Replacement of the estimates in Table with locally derived values is encouraged. If
lighting installation increases the density of roadside fixed objects, the value of CMF2r is adjusted accordingly.
Session 4 – Predicting Highway Safety for Multilane Urban Streets

50. CMF for Lighting CMF4r = 1- [pnr x (1.0 – 0.72 pinr – 0.83 ppnr ) ]
Instructor: from Chapter 12 of the Draft Highway Safety Manual
Using the default values you will get a predicted value that can be compared against the actual value based on local information.
CMF4r—Lighting
The base condition for lighting is the absence of roadway segment lighting (CMF4r = 1.00). The CMF for lighted roadway segments is determined, based on the work of Elvik and Vaa (3), as:
CMF4r = 1.0 − (pnr × (1.0 − 0.72 × pinr − 0.83 × ppnr)) (12-34)
Where:
CMF4r = crash modification factor for the effect of roadway segment lighting on total crashes;
pinr = proportion of total nighttime crashes for unlighted roadway segments that involve a fatality or injury;
ppnr = proportion of total nighttime crashes for unlighted roadway segments that involve property damage only; and
pnr = proportion of total crashes for unlighted roadway segments that occur at night.
CMF4r applies to total roadway segment crashes. Table presents default values for the nighttime crash proportions
pinr, ppnr, and pnr. Replacement of the estimates in Table with locally derived values is encouraged. If
lighting installation increases the density of roadside fixed objects, the value of CMF2r is adjusted accordingly.
These are default values for nighttime crash proportions; replace with local information
If light installation increases the density of roadside fixed objects, adjust CMF2r
Session 4 – Predicting Highway Safety for Multilane Urban Streets

51. CMF for Lighting: Example
For 4-Ln Urban undivided commercial curbed street (4U) with power poles on 150 foot spacing 2 feet from edge of travel way on one-side– Add Lighting
CMF3r = 1- [pnr x (1.0 – 0.72 pinr – 0.83 ppnr )]
= 1- [0.365 x (1.0– 0.72(0.517) – 0.83 x 0.483) ]
= 0.917
Lighting adds light poles at 160 foot spacing on one side (the other side) set back 2 feet from back of curb
Recompute CMF2r
Instructor: from Chapter 12 of the Highway Safety Manual
Example calculation for CMF for lighting – no lighting is CMF=1.00
When add light poles, now need to go back and recomputed CMF for horizontal clearance for the light poles
Session 4 – Predicting Highway Safety for Multilane Urban Streets

52. CMF for Roadside Fixed Objects: Example
For one mile of 4-Ln Urban undivided commercial curbed street (4U) with power poles on one side on 150 foot spacing 2 feet from edge of travel way + street lighting on other side on 160 foot spacing 2 feet from edge of travel way:
CMF2r = foffset x Dfo x pfo + (1 – pfo)
= 0.232((5280/150)(1)+ (5280/160)(1))(0.037) + (1 – 0.037)
Instructor: from Chapter 12 of the Highway Safety Manual
Example calculation for 2 feet back of curb for power poles for poles on one side of a Commercial 4U + added light poles on other side at 2 feet back of curb spaced at 160 feet.
= x ( ) x (0.963)
= x 68.2 x
= =
1.548
Session 4 – Predicting Highway Safety for Multilane Urban Streets

53. CMF for Automated Speed Enforcement
CMF5r is:
1.00 for no automated speed enforcement;
0.95 for automated speed enforcement
Instructor: from Chapter 12 of the Highway Safety Manual
CMF5r—Automated Speed Enforcement
Automated speed enforcement systems use video or photographic identification in conjunction with radar or lasers to detect speeding drivers. These systems automatically record vehicle identification information without the need for police officers at the scene. The base condition for automated speed enforcement is that it is absent. Chapter 17 presents a CMF of 0.83 for the reduction of all types of fatal-and-injury crashes from implementation of automated speed enforcement. This CMF is assumed to apply to roadway segments between intersections with fixed camera sites where the camera is always present or where drivers have no way of knowing whether the camera is present or not. No information is available on the effect of automated speed enforcement on noninjury crashes. With the conservative assumption that automated speed enforcement has no effect on noninjury crashes, the value of the CMF for automated speed enforcement would be 0.95.
Session 4 – Predicting Highway Safety for Multilane Urban Streets

54. Nbr = Nspf rs (CMF1r x CMF2r……..CMFnr)
Applying Crash Modification Factors to Prediction of Crash Frequency for Urban/Suburban Roadway Segments
Nbr = Nspf rs (CMF1r x CMF2r……..CMFnr)
Where:
Nbr = Predicted number of total roadway segment crashes per year with effects of conditions other than base conditions
Instructor: from Chapter 12 of the Highway Safety Manual
The next step is to apply the CMFs to the SPF.
Now we take the predicted segment crashes of 32.8 (Sum of Segment MV =21.4, Segment SV = 4.3, Segment Driveways = 7.1, see previous slides) and multiply the CMFs for parking, roadside fixed objects, and lighting to obtain 93.9 predicted crashes per year
Session 4 – Predicting Highway Safety for Multilane Urban Streets

55. Commercial on-street parallel parking both sides
Applying Crash Modification Factors to Prediction of Crash Frequency for Urban/Suburban Roadway Segments
Example:
Commercial on-street parallel parking both sides
Roadside Fixed Objects and non-breakaway light both sides)
Lighting
Transversible 15 foot wide median
No speed enforcement
CMF = 1.613
CMF = 1.937
CMF = 0.917
CMF = 1.00
Instructor: from Chapter 12 of the Highway Safety Manual
Now we take the predicted segment crashes of 32.8 (Sum of Segment MV =21.4, Segment SV = 4.3, Segment Driveways = 7.1, see previous slides) and multiply the CMFs for parking, roadside fixed objects, and lighting to obtain 93.9 predicted crashes per year
Nbr = Nspf rs (CMF1r x CMF2r x CMFnr)
= (1.613 x x x 1.00 x 1.00)
= 93.9 crashes per year
Session 4 – Predicting Highway Safety for Multilane Urban Streets

56. Predicting Crash Frequency for Peds + Bikes on Urban/Suburban Streets
Npredicted rs = (Nbr + Npedr + Nbiker) Cr
Where:
N predicted rs = predicted average crash frequency of an
individual roadway segment for the selected year
Nbr = predicted average crash frequency of an individual roadway segment excluding vehicle-pedestrian and vehicle-bicycle crashes
Npedr = predicted average crash frequency of vehicle-pedestrian crashes for an individual roadway segment
Nbiker = predicted average crash frequency of vehicle-bicycle crashes for an individual roadway segment
Cr = calibration factor for roadway segments of a specific type developed for use for a particular geographical area
Instructor:
OK let’s get back to our formula. We now have Nbr which really is the majority of the work. You’ll find in the next few slides that calculating for the crash frequency of vehicle pedestrian and vehicle bicycle crashes is fairly quick.

57. Predicting Crash Frequency for Peds + Bikes on Urban/Suburban Streets
Npedr = Nbr x fpedr
Instructor:
The predicted average crash frequency of vehicle-pedestrian crashes represented by Npedr is calculated by multiplying the predicted average crash frequency of an individual roadway segment Nbr by the pedestrian accident adjustment factor represented by fpedr
All vehicle pedestrian collisions are considered to be fatal-and-injury crashes. The values of fpedr are likely to depend on the climate and the walking environment in particular states or communities. HSM users are encouraged to replace the values in Table 12-8 with suitable values for their own state or community through the calibration process which can be found in the Appendix to Part C

58. Predicting Crash Frequency for Peds + Bikes on Urban/Suburban Streets
From continued Example:
4-Ln Undivided
40 mph
Instructor:
So for our example we with a 4 lane undivided roadway with a speed limit of 40 mph so the pedestrian accident adjustment factor is multiply that with Nbr and we get 0.86 crashes per year for the predicted pedestrian crashes.
Npedr = Nbr x fpedr
= 93.9 crashes per year x 0.009
= 0.85 crashes per year

59. Predicting Crash Frequency for Peds + Bikes on Urban/Suburban Streets
Nbiker = Nbr x fbiker
Instructor:
In the same manner the predicted average crash frequency of vehicle-bicycle crashes represented by Nbiker is calculated by multiplying Nbr by the bicycle accident adjustment factor represented by fbiker
Once again all vehicle bicycle collisions are considered to be fatal-and-injury crashes and the values of fbiker are likely to depend on the climate and bicycling environment in particular states or communities and they can be calibrated as well for local areas

60. Predicting Crash Frequency for Peds + Bikes on Urban/Suburban Streets
From continued Example:
Nbiker = Nbr x fbiker
Instructor:
So the bicycle accident adjustment factor for our example is
When we multiply that with Nbr we get a predicted bicycle crash frequency of 0.19 crashes per year
4-Ln Undivided
40 mph = high speed
= 93.9 crashes per year x 0.002
= 0.19 crashes per year

61. Predicting Crash Frequency for Peds + Bikes on Urban/Suburban Streets
Combining Segment, Ped and Bike crashes:
Npredicted rs = (Nbr + Npedr + Nbiker) Cr
Npredicted rs = ( ) x 1
= crashes per year
Instructor:
Letting Cr the calibration factor =1.0, adding predicted values together and multiplying the answer by Cr we come up with a predicted crash frequency of 94.9 crashes per year.
Now you might be thinking the pedestrian and bicycle crash values are so small compared to just vehicle crashes, but this isn’t surprising since we know that walking along the road crashes are about 10 to 15 percent of all pedestrian crashes and the bicycle numbers are even less.

62. Predicting Crash Frequency of Suburban/Urban Multilane Streets
Learning Outcomes:
Described the models to Predict Crash Frequency for Multilane Suburban/Urban Streets
Described Crash Modification Factors for Multilane Suburban/Urban Streets
Applied Crash Modification Factors (CMF’s) to Predicted Crash Frequency for Multilane Suburban/Urban Streets
Learning Objectives for Session #5 on Suburban Urban Multilane Streets prediction of Crash Frequency

63. Questions and Discussion:
Introduction and Background
Questions and Discussion: