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.
NCHRP Web-Only Document 291: Development of a Posted Speed Limit Setting Procedure and Tool 8 CHAPTER 3. FINDINGS Following are key findings by each of the major efforts within each phase. PHASE I EFFORTS Within Phase I, the research team grouped its efforts within six major activities. The findings from Phase I influenced the direction proposed and set for Phase II. Phase I. Conduct Project Management The direction set during the kickoff meeting was that the research team would consider the following roadway classes: ï· High-speed rural. ï· Rural-to-city transition zones. ï· Urban/suburban streets. The literature review findings were organized to reflect these three conditions. Occurring concurrently as NCHRP Project 17-76, the parallel NCHRP 17-79 project focused on examining the safety relationship between speed limit, roadway characteristics, and crashes on roads with speed limits of 75 mph or greater. The goal was for NCHRP Project 17-76 to use the findings from NCHRP 17-79 rather than duplicating efforts on higher-speed roads. Because of the recent reports on speeds in transition zones (4, 5), the focus of Phase II research was on city streets. Trends/recent activity that could affect the direction of this project include the following: ï· NACTO 2017 policy (6) that includes the following statement: âState rules or laws that set speed limits at the 85th percentile speed should be repealed.â ï· The National Transportation Safety Boardâs (NTSBâs) publication of a safety study on Reducing Speeding-Related Crashes Involving Passenger Vehicles (7). NTSB provides specific recommendations in its report, such as removing the guidance in the MUTCD (8) that speed limits should be within 5 mph of the 85th percentile speed. Phase I. Identify Existing Procedures for Setting Posted Speed Limit Most, if not all, of the procedures currently used in the United States are based on the 85th percentile speed. Several procedures provide consideration for adjusting the recommended posted speed based on existing conditions, such as existing crashes, roadside development/land use, sight distance, shoulder presence, traffic volume, and so forth. Other countries are using procedures that are based on factors other than the 85th percentile speed to set the posted speed limit. For example, both New Zealand and Canada consider the presence of a pedestrian facility. There is currently an experiment being conducted in Portland, Oregon, for collector streets that is considering roadway conditions specifically associated with pedestrians and bicyclists in addition to conditions for motorists to identify the suggested posted speed limit. Phase I. Review Literature The review of the literature had an emphasis on identifying known relationships among operating speed, safety, and roadway characteristics. Consensus is that higher operating speeds are associated with more severe crashes, as supported by the basic physics of the situation. Few studies are available that examine the relationship between the magnitude of operating speed and the number of crashes, probably primarily due to the difficulties in obtaining actual operating
NCHRP Web-Only Document 291: Development of a Posted Speed Limit Setting Procedure and Tool 9 speed for significant lengths of time and for a significant number of sites. A 2017 study on rural two-lane highways in Israel (9) and a 2016 study on two-lane urban roads in the city of Edmonton, Canada (10), are notable exceptions. The studies found that as operating speed increases, the number of crashes also increase. The roadways, traffic control devices, and traffic variables that have been found to affect speed or crashes were identified and summarized in the interim report and are in Appendices A and B of this document. This knowledge was used to develop a list of variables that should be considered in Phase II. Phase I. Identify Potential Operating Speed Datasets The research team searched for potential sources of speed data. Having a large database of speed data, along with roadway characteristics and crashes, is critical to understanding the complex relationships among speed, roadway design, and safety. Higher quality and quantity of data are available for freeways; however, the focus for NCHRP Project 17-76 was urban/suburban streets. Several sources of speed data were identified; however, each had significant limitations with respect to NCHRP Project 17-76. Samples of the datasets with the most promise were obtained for the effort summarized in the next section. Phase I. Obtain Samples of Speed Data Samples of speed data from several sources were obtained. The review considered the quality and quantity of the speed data with respect to their usability for NCHRP Project 17-76. The traditional data collection technique of using on-road sensors or video cameras could produce usable data but would result in a smaller sample than preferred because of the cost to collect speed data in that manner. Therefore, Phase I efforts were focused on exploring emerging big datasets to see if they could serve the needs of NCHRP Project 17-76. Vendor speed data by way of the National Performance Management Research Data Set (NPMRDS) is available to states and metropolitan planning organizations (MPOs); however, NCHRP would have had to pay to access the data for use in this research project. A review of a sample of the NPMRDS data, however, revealed several limitations with respect to arterial streets including lack of roadway vehicle volume, long segments that include several signalized intersections, and gaps in the speed data for a large portion of the 5-minute time bins. Bluetooth data are available for several roadway segments in Austin, Texas. While the dataset is open source (i.e., the 17-76 project did not have to pay for the data), similar limitations as with other vendor data are present. The BluetoothÂ® data do not include roadway volume, represent long segments with signalized intersections in many cases, and are missing speed data for several 15-minute time bins. The Texas A&M Transportation Institute (TTI) recently explored a dataset of binned speed data available for streets in Austin, Texas. This dataset formed the core of the Texas data used in the related analysis in Phase II. Shapefiles of the existing street network within the City of Ann Arbor and the surrounding suburban townships were available to the research team and formed the core of the Michigan data used in the related analysis in Phase II. Phase I. Review Other Techniques for Selecting Posted Speed Limits The review of USLIMITS2 (11), Portland (12), New Zealand (13), and Canadaâs procedures showed that several of the variables identified in the literature review are also being considered in their procedures. In some cases, the consideration is specific; for example, in
NCHRP Web-Only Document 291: Development of a Posted Speed Limit Setting Procedure and Tool 10 USLIMITS2, a precise value for signal or access density (e.g., 4 signals per mile) would change the recommendation. In other cases, the value for the variable is based on engineering judgment (e.g., is parking activity high or not). The values and criteria being used in these other procedures could help guide the recommendations for a revised or newly developed procedure. PHASE II EFFORTS Based upon the Phase I key findings and with due consideration of the available funds, the research team conducted activities within six major areas. The following describes the findings by research effort. Phase II. Build and Analyze Austin, Texas, Database The findings from the database built using data for Austin, Texas, are in Appendix D. The database was used to analyze the relationships among roadway characteristics, posted speed limit, operating speed, and crash data. The analysis was conducted within three rounds using different approaches or subsets of the database. Key findings from the evaluations supported the following decision rules for the SLS-Tool: ï· Inclusion of the following variables: o Number of signals or signal density. o On-street parking. ï· Addition of the following variable: o Median type. Raised medians were associated with fewer fatal and injury (KABC) crashes compared to no median or two-way left-turn lane (TWLTL), while TWLTLs were associated with more KABC crashes compared to no median. The findings from the path analysis support the consideration of the 50th percentile speed in identifying a suggested speed limit. Phase II. Build and Analyze Washtenaw County, Michigan, Database The Washtenaw County, Michigan, database was used to analyze the relationships among roadway characteristics, posted speed limit, and crash data. The findings from this effort supported the following decision rules for the SLS-Tool: ï· Inclusion of the following variables: o Signal density. o Access density, along with the break points at 40 and 60 access points per mile. ï· Addition of the following variable: o Median type (raised medians were associated with fewer KABCO crashes compared to none or TWLTL, while TWLTLs were associated with fewer KABCO crashes compared to no median). Phase II. Develop SLS-Procedure Selecting the Base Format for SLS-Procedure Currently, the predominant method for setting speed limits is with the use of the 85th percentile speed. It is viewed as being representative of a safe speed that will minimize crashes, and the 1964 Solomon study (14) is frequently quoted as being the source to justify the
NCHRP Web-Only Document 291: Development of a Posted Speed Limit Setting Procedure and Tool 11 use of the 85th percentile speed. The use of the 85th percentile speed has been supported with the following: ï· Represents a safe speed that minimizes crashes. ï· Promotes uniform traffic flow along a corridor. ï· Is a fair way to set the speed limit based on the driving behavior of most drivers (i.e., 85 percent). ï· Represents reasonable and prudent drivers since the fastest 15 percent of drivers are excluded. ï· Is enforceable in that it is fair to ticket the small percentage (15 percent) of drivers that are exceeding the posted speed limit. Criticisms of the 85th percentile speed method have included the following: ï· Setting the posted speed limit based on existing driver behavior may create unsafe road conditions because drivers may not see or be aware of all the conditions present within the corridor. ï· Setting the posted speed limit on existing driver behavior rather than the roadway context may not adequately consider vulnerable roadway users such as pedestrians and bicyclists. ï· Drivers are not always reasonable and prudent, or they only consider what is reasonable and prudent for themselves and not for all users of the system. ï· Using measured operating speeds could cause operating speeds to increase over time (i.e., speed creep). Drivers frequently select speeds a certain increment above the posted speed limit, anticipating that they will not receive a ticket if they are not above that assume enforcement speed tolerance. In this case, the resulting operating speed would be above the posted speed limit. Using the 85th percentile speed approach in this situation would result in recommending a posted speed limit that is higher than the existing posted speed limit. Posting that higher speed limit would set up the cycle that the next spot speed study may again find a higher operating speed because of drivers using the assumed speed enforcement tolerance to select their speed. ï· Most of the early research justifying the use of the 85th percentile speed was conducted on rural roads, so it may not be appropriate for urban roads. As documented in Appendix C, a National Committee on Uniform Traffic Control Devices (NCUTCD) task force conducted a survey on speed limits. One of the questions from the NCUTCD task force survey was âHow would you set speed limits if given the choice?â The provided responses included rounding to the nearest 5 mph of the 85th percentile, or rounding up or down, and so forth. Half of the survey participants selected âotherâ and typed a response, with the word âcontextâ being used more than any other word. Within the design community, there is greater emphasis on designing roadways to fit the context of the site. Given the increased emphasis on context within the profession, the research team decided that the SLS-Procedure should also be sensitive to context. The Expanded Functional Classification System available in NCHRP Research Report 855 (3) was used to develop SLSGs that reflect logical groups with respect to setting speed limits. For example, freeways, which have very specific geometric design criteria, are present within several roadway type and roadway context combinations. Those roadway type and context combinations were grouped into a Limited-Access SLSG. The other SLSGs identified were Undeveloped, Developed, and Full Access.
NCHRP Web-Only Document 291: Development of a Posted Speed Limit Setting Procedure and Tool 12 Within each of these SLSGs, a unique set of decision rules was developed. For the Limited-Access and Undeveloped SLSGs with their higher operating speed and greater emphasis on mobility, retaining a connection to measured operating speed was deemed appropriate. After much debate among the research team, panel, and other subject matter experts, the research team also decided to retain the connection with measured operating speed for the Developed SLSGâ with the knowledge that which measured operating speed would serve as the starting point (e.g., 85th percentile or 50th percentile), along with whether to use the closest speed or rounded down to nearest 5-mph increment, would be influenced by consideration of safety through the use of decision rules. Extensive debate was then engaged in about how to set the decision rules for the Full-Access SLSG, which included local streets and the urban core. The research team initially considered having set speed limits (e.g., 25 mph) for a set of conditions (e.g., specific combinations of roadway characteristics such as the number of lanes, average lane width, median presence, sidewalk presence, etc.). After additional extensive discussion among the team, panel, and subject matter experts, the final decision by the research team was to also have the Full- Access SLSG use measured operating speed; however, the measured operating speed would only consider the 50th percentile rather than the 85th percentile to provide greater consideration for the anticipated other users of the street within these settings. In summary, for the SLS-Procedure, the research team recommended considering the measured operating speed as the starting point for selecting a posted speed limit, but also recommended adjusting the measured operating speed based on roadway conditions and crash experience on the segment. The NCHRP Project 17-76 SLS-Procedure was developed based on this key decision. Guiding Principles The guiding principles developed by the research team for the SLS-Procedure included the following: ï· Use a data-driven approach with research-based decision rules. ï· Produce consistent results for a given set of conditions. ï· Incorporate contemporary policies, guidelines, and practices. ï· Consider driversâ speed choice and roadway safety. ï· Provide transparency in the decision process. ï· Consider all roadway types and roadway contexts. ï· Vary the decision rules to account for the diverse characteristics of each speed limit setting group. ï· Consider agency data and human resource constraints. ï· Include inputs and outputs on the same screen to demonstrate the relationship between each roadway characteristic and selection of the suggested speed limit. ï· Allow for future modifications to accommodate new knowledge. ï· Create efficiencies in the decision process, where possible. Overview of SLS-Procedure With consideration of the issues identified along with research into the relationships among roadway characteristics including posted speed limit, operating speed, and safety, the NCHRP Project 17-76 team developed a procedure to calculate a suggested speed limit (SSL). The SLS-Procedure starts with identifying the roadway segment context and type. Next, the
NCHRP Web-Only Document 291: Development of a Posted Speed Limit Setting Procedure and Tool 13 speed distribution of drivers on that segment is used to identify a potential SSL that is then adjusted with consideration of the crash potential for the segment. Figure 1 illustrates the steps for the procedure. Additional details are provided in the user guide (1). The SLS-Procedure was designed to result in an unbiased and objective method for identifying an SSL value. The following reference documents were directly considered during the research effort: MUTCD (8), USLIMITS2 original study (11), USLIMTS2 revised user guide (15), and Highway Safety Manual (HSM) (16). Several other references were also considered, as documented in Appendix F. Previous knowledge and engineering judgment along with feedback from the panel were used to develop the decision rule recommendations on setting of posted speed limits with respect to freeways and rural highways. The team also explored the applicability of other speed limit guidance tools for establishing speed limits on rural freeways and other highways. This included knowledge-based expert systems like USLIMITS2, which uses detailed site-specific characteristics to determine appropriate speed limits. The findings from the analyses conducted within NCHRP Project 17-76, in addition to previous knowledge, engineering judgment, and subject matter expertsâ suggestions, were used to develop decision rule recommendations on the setting of the posted speed limit for urban/suburban streets. Within NCHRP Project 17-76, the research team focused Phase II on collecting data for suburban and urban roads to be able to investigate the relationships among crashes, roadway characteristics, and posted speed limit to fill the known research gap for city streets. The team found that crashes were lowest when the operating speed was within 5 mph of the average operating speed (see Appendix D). Therefore, the research team recommended that the 50th percentile speed also be a consideration within the SLS-Procedure. Phase II. Develop SLS-Tool The SLS-Procedure was automated into an SLS-Tool using a spreadsheet as the base format. Along with the SLS-Tool is a stand-alone document, NCHRP Research Report 966: Posted Speed Limit Setting Procedure and Tool: User Guide (1), that provides information regarding the variables used in the spreadsheet tool along with general information about the setting of speed limits. Information on the SLS-Procedure and a copy of the SLS-Tool are available in NCHRP Research Report 966 (1). The research team established the following guiding principles for the SLS-Tool development: ï· Most or all data should be on one screen. ï· Colors should indicate what the user should enter, what is being calculated, and warning/advisory notes. ï· Data input should be organized by type (e.g., site description, speed data, site characteristics, and crashes). ï· Only needed site characteristics for the particular SLSG should be shown by hiding rows using spreadsheet macros. Phase II. Conduct Technology Transfer Activities Presenting a workshop was a requirement of the research. Members of the research team conducted several workshops and presentations during the development of the SLS-Procedure, and these presentations provided opportunities to obtain feedback on the potential format of the
NCHRP Web-Only Document 291: Development of a Posted Speed Limit Setting Procedure and Tool 14 procedure. The presentations with the panel were especially influential in setting the direction for the SLS-Procedure and SLS-Tool. A copy of the slides used at one of the workshops is provided in Appendix G. Phase II. Document Research The final effort within the project was to document the methodology and findings from the research.