Road Safety Audits (RSA)

< Previous

Table of Content

Next >

APPENDIX B: RSA CASE STUDIES

Bullhead Parkway: Bullhead City, Arizona

Road: Existing 10.2-mile section of paved, 4-lane, divided, rural roadway
RSA Sites:	4 signalized intersections, 13 unsignalized intersections AADT: 10,000 – 15,000 vehicles per day (2007) 50 mph posted speed limit
Environment:	urban/urbanized suburban rural
Owners:	Bullhead City, Arizona
Road Safety Audit
Date of RSA:	23-25 October 2007
RSA Stage(s):	planning/design stage RSA of existing roads
RSA Team:	5 members representing ADOT Traffic Safety, ADOT Traffic Design, ADOT Kingman District, FHWA, and the City of Yuma
RSA Cost:	$30,000
Implementation Cost:	$839,478
Crash Reduction:	54% total; 50% fatal / incapacitating; 30% intersection-related
Benefit / Cost Ratio:	20:1 total; 16:1 injury; 2:1 intersection-related

Background

An RSA was conducted along a 10.2-mile section of Bullhead Parkway in Bullhead City, Arizona (Figure 2). Bullhead Parkway is a four-lane, divided, rural roadway with a posted speed limit of 50 mph for the entire length. Four signalized intersections are present along the Parkway. At the time of the RSA, the annual average daily traffic along the Parkway was reported to range between 10,000 and 15,000 vehicles per day (vpd).

The RSA was requested by the Bullhead City Department of Public Works. The Parkway appeared on Arizona's 2006 Five Percent Report. The section was also one of the City's top safety priority locations.

Figure 2. Aerial Map of Bullhead Parkway.

The RSA was conducted in October 2007. The RSA team was comprised of five members representing the Arizona Department of Transportation (ADOT) Traffic Safety, ADOT Traffic Design, ADOT Kingman District, City of Yuma, and Federal Highway Administration (FHWA).

Information used by the RSA team to conduct the review included the following:

• Crash data (August 1, 2003 – July 31, 2006) from the ADOT Motor Vehicle Division.
• Traffic data (2007 and 2009) from Bullhead City.
• Roadway data (lane, median, and shoulder widths).

For this RSA, 3.5 hours (total) were required to hold the opening and closing meetings. The RSA team spent 7 hours in the field reviewing the roadway segment. Analysis prior to and following the field review required 10 hours. Documenting the findings of the RSA required 12 hours. Four hours of travel each way were required to complete the RSA. In total, the RSA of Bullhead Parkway took 40.5 hours and cost $30,000.

Key Findings and Suggestions

The key findings and suggestions of the RSA are summarized in Table 8. Several of the measures were initiated in 2007, shortly after the completion of the RSA. A number of measures have since been implemented or are programmed for coming years. Out of all the measures suggested, only the flattening of roadside slopes is not being considered or evaluated further due to a lack of available right-of-way and cost issues.

Table 8. Summary of Safety Issues and Suggestions: RSA of Bullhead Parkway.

The following photos illustrate safety issues observed during the RSA ("before" on left) and the subsequent improvements resulting from the RSA suggestions ("after" on right).

Nontraversable median drains. (left) - Raised traversable median drains. (right)

Out-of-date end terminals. (left) - Current standard end terminals. (right)

T-intersection signing. (left) - Enhanced T-intersection signing. (right)

Short right-turn lane. (left) - Extended right-turn lane. (right)

The cost, to date, of the suggestions implemented based on the findings totaled $839,478 (plus several nominal fee projects). These measures were paid for largely through Bullhead City's Capital Improvement Program. To cover the cost of programmed projects directly related to the RSA, the City completed local government Highway Safety Improvement Program (HSIP) grant applications and submitted to ADOT in the amount of $403,000. An additional $382,000 was also applied for to complete projects along the Parkway that were outside of the RSA suggestions. This amount included $200,000 for signal design and construction at the intersection with N. Oatman Road; $170,000 for the construction of a southbound right-turn lane at the intersection with Silver Creek Road; and $12,000 for the construction of ADA-compliant ramps at the corners of the intersection with Adobe Road.

In general, ADOT and Bullhead City viewed the RSA as a great success. For its efforts in garnering funds and implementing improvements, Bullhead City was presented with an "RSA Champion" award by ADOT and FHWA during the inaugural Arizona Traffic Safety Summit in 2009. ADOT district personnel have seen the value in RSAs, as the representative team member for this RSA has participated in several others since. The City has also accepted the RSA process and its benefits; two additional RSAs have been conducted in Bullhead City since the completion of the one along Bullhead Parkway. The City has also received some educational benefit from this RSA in terms of roadway and roadside hazard safety. For example, the City had previously installed trees along a low-speed facility as landscaping. A similar application was suggested by the City for use along the median of Bullhead Parkway; however, the use of large-diameter trees (i.e., greater than four inches wide) along the roadside of a high-speed facility is not conducive to safety, as the landscaping presents fixed object hazards.

RSA Outcomes

Geometric, traffic volume, and crash data were collected for both treatment and comparison sites in Arizona. The treatment site consisted of two consecutive segments that were divided at the point where the traffic volume changed. Data were provided by direction for the comparison site because certain roadway characteristics differed between the two directions. Similar to the treatment site, the comparison site was split into two segments at the point where the traffic volume changed. Table 9 provides a summary of the treatment and comparison site data.

Table 9. Summary of Arizona Data.

In Arizona, the countermeasures implemented as a result of the RSA were mostly implemented in 2008; however, some countermeasures were implemented in 2009 and after. Thus, the after period conditions were continuously changing for the treated segments. Recognizing the changing conditions and the limited after period, the analysis was run for two after periods. The first after period was based on data from 2008 and the first 11 months of 2009. The second after period only considered the first 11 months of 2009.

The analysis did not account for regression-to-the-mean because the empirical Bayes (EB) approach could not be applied. However, it was possible to investigate the potential for regression-to-the-mean. Specifically, the yearly crash counts for both total and injury crashes were plotted over time for the treatment site, as shown in Figure 3. The period from 2003 to 2007 is relevant to the potential for regression-to-the-mean. If present, one would expect to see a spike in crash counts prior to the RSA being conducted in 2007. The plot indicates the potential for regression-to-the-mean for both total and injury crashes. Notably, the annual crash counts appear to spike for the three-year period from 2004 to 2006 and then decrease in 2007, prior to the implementation of any countermeasures from the RSA.

In this case, the potential regression-to-the-mean bias could result in an overestimation of the expected safety effects of the treatments. As such, the analyses were run with and without the three years of suspect data (2004 to 2006) to investigate the magnitude of the regression-to-the-mean bias.

Figure 3. Treatment Site Crash Totals Over Time.

The following points should also be noted with respect to the analysis:

• For the comparison group method, total crashes in the before and after period at the comparison sites were used to account for general temporal trends in total and intersection-related crashes. Note that total crashes at the comparison sites were used to adjust the intersection-related crashes at the treatment sites due to limited intersection-related crashes at the comparison sites. Injury crashes at the comparison sites were used to adjust for injury crashes at the treatment site.
• When accounting for changes in traffic volume, a linear relationship was assumed between crashes and traffic volume. With a limited variation in traffic volumes it was not possible to develop an SPF or evaluate the suitability of existing SPFs to model the true relationship. Given the inherent uncertainty in other aspects of the analysis, and the relatively small changes in annual average daily traffic (AADT), this assumption was not deemed crucial.
• The suitability of the comparison group could not be rejected using the methods described by Hauer (1997). As such, the comparison site is deemed reliable for use in the comparison group method.

The results of the Arizona analyses are presented in Tables 10 – 13. Note that separate results are provided for each of the following:

1. Comparison-group method, including entire before period.
2. Comparison-group method, excluding years 2004 to 2006 from the before period.
3. Naïve method, including entire before period.
4. Naïve method, excluding years 2004 to 2006 from the before period.

Note that each table includes two sets of results, one considering 2008 in the after period and another excluding 2008 from the after period. Results that are statistically significant at the five percent level are noted in bold.

Table 10. Comparison Group Results from Arizona Including Total Before-Period.

Table 11. Comparison Group Results from Arizona, Excluding 2004 to 2006.

Table 12. Naïve Results from Arizona, Including all Before-Period Years.

Table 13. Naïve Results from Arizona, Excluding 2004 to 2006.

The results indicate a general safety benefit of conducting the RSA and implementing the treatments suggested in the RSA report. For total and injury crashes, the estimated crash modification factor (CMF) is consistently less than 1.0 for all variations of the analysis, indicating a reduction in expected crashes. Further, the reductions in total and injury crashes are substantial and highly significant. With respect to intersection-related crashes, the results generally indicate a safety benefit with two exceptions, which are not statistically significant.

While there are several variations of the analysis, the results from the comparison group method are preferred over the naïve before-after results. The estimated crash reductions are generally larger for the comparison group method than for the naïve method. This is due to the observed increase in crashes at the comparison site in accounting for time trends. However, the naïve method does appear to corroborate the results of the comparison group method. The remainder of the discussion is focused on the results of the comparison group method but generally applies to the naïve results, as well.

The estimated crash reductions are generally smaller when the 2004 to 2006 data are not included in the before period. This confirms the suspicion that regression-to-the-mean may be present. Therefore, the results from Table 11 (Comparison Group Results Excluding 2004 to 2006) provide a more conservative estimate of the safety effects than Table 10. Within each table, the estimated crash reductions are larger when 2008 is excluded from the after period. This is not surprising, as countermeasures were implemented throughout 2008 and 2009. As such, the analyses that include 2008 in the after period provide conservative estimates of the safety effect.

Table 14 presents a summary of the safety benefits estimated from the treatment site in Arizona. The conservative estimates are based on the comparison group method, excluding 2004 to 2006 from the before period and including 2008 in the after period. The range of results from both methods and all variations is also presented. The results indicate a 54 percent reduction in total crashes, a 50 percent reduction in injury crashes, and a 30 percent reduction in intersection-related crashes.

Table 14. Summary of Results from Arizona.

The annualized costs and benefits were then compared. Based on the comparison group method, there was an annual reduction of 21.1 total crashes, 7.7 injury crashes, and 2.4 intersection-related crashes. Using the FHWA crash costs document, Crash Cost Estimates by Maximum Police-Reported Injury Severity within Selected Crash Geometries (Publication Number FHWA-HRT-05-051), the annual benefits are $1,973,395, $1,582,195, and $222,025 for total, injury, and intersection-related crashes, respectively. Assuming a total cost of $869,478 ($30,000 for the RSA and $839,478 for the improvements), a discount rate of 2.2 percent, and a service life of 10 years, the annualized cost is $97,812. Therefore, the B/C ratio of the improvements made as a result of the RSA is 20:1, 16:1, and 2:1 for total, injury, and intersection-related crashes, respectively.

Program Review

The Arizona DOT (ADOT) uses its RSA program to build collaborative partnerships that proactively improve safety. RSAs have brought together State, county, and tribal transportation practitioners to solve complex safety issues. This has led to long-standing partnerships between ADOT, counties, and the tribes where each agency has a better understanding of each agencies concerns, limitations, and abilities. This collaborative approach applied through RSAs has also been helpful in educating local agencies about the risks associated with various crash types and bringing attention to safety issues at locations that may be lacking crash data. For example, a pedestrian crash may be a rare occurrence, but when one occurs, there is often a high probability of severe injury or death. ADOT staff brings perspectives to local agencies to help provide context on the types of operational and geometric characteristics that can adversely impact pedestrian safety. In an example of safety issues at locations lacking crash data, most non-fatal crashes occurring on tribal reservations are not entered into the state DOT crash database. This skews the crash frequencies and rates on reservation roads, many of which are state-owned and maintained, making it appear that these roads do not have a crash problem. The proactive nature of the RSA provides an opportunity to identify safety issues without the benefit of crash data and develop appropriate countermeasures to improve road safety.

Over time, the RSA program has successfully communicated to design engineers that merely following standards does not necessarily produce a safe design. The collaborative and independent approach of the RSA teams to improving safety is changing how design engineers initially viewed RSAs—from a strict adherence to design standards to a more open discussion about the safety implications of various designs. The findings from the RSA program have been used to update the DOT's policies and standards. The ADOT experience has shown a need for better upfront coordination with the road owner as a key element to obtain better data and reports and to determine who should be invited to the RSA meetings. ADOT also recommends a nighttime site review, even if the crash data does not indicate any issues, and to obtain the law enforcement perspective in each RSA.

State Route 101 (Peavine Road): Cumberland County, Tennessee

Background

An RSA was conducted along a 3.84-mile section of State Route 101 (Peavine Road) in Cumberland County, Tennessee (Figure 4). Throughout this section, State Route 101 is a two-lane, rural roadway with a posted speed limit of 45 mph. The AADT was reported as 12,860 vpd in 2007.

The area of concern was brought to attention by local citizens who perceived the section of roadway to be dangerous. Consequently, Tennessee DOT (TDOT) Planning staff were asked to evaluate this section of State Route 101 to determine the need for appropriate safety measures. Upon reviewing the crash data, the number of reported crashes supported the public's concern.

The RSA was conducted along State Route 101 in November 2005 from Peavine Firetower Road (L.M. 18.31) to Tuttle Lane (L.M. 22.15). The RSA team was comprised of nine members representing the FHWA, TDOT Headquarters, TDOT Traffic Engineering, and TDOT Planning. The efforts of the team were led by the TDOT Regional Engineer.

Figure 4. Aerial Map of State Route 101.

Information used by the RSA team to conduct the review included roadway data (i.e., geometrics, functional class, route features, etc.), traffic data, and crash data. A comprehensive list of the available information is as follows:

• Crash data and ambulance call records.
• General statements from citizens.
• Cumberland County general highway map.
• Tennessee Roadway Information Management System (TRIMS) Route Feature
• Description Listing.
• TRIMS Photolog.
• TRIMS Crash Data (2001 – 2003, and any available from 2004 and 2005).
• TRIMS Crash Rate Summary Report (01/01/2001 – partial 2005).
• TRIMS Crash Summary Report (01/01/2001 – partial 2005).
• Maps pinpointing crash locations (01/01/2001 – partial 2005).

While no cost information (i.e., consultant fees, time for public officials, etc.) was recorded with regard to completing the RSA on State Route 101, a typical RSA of this extent costs in the range of $10,000 to $12,000. For RSAs on the scale of five miles or more, the cost to complete the entire RSA process increases.

Key Findings and Suggestions

The key findings and suggestions of the RSA are summarized in Table 15. TDOT responded that all of the suggested measures in the RSA report were installed in May 2006, six months after the RSA was completed and documented. There were no additional improvements made beyond those suggested in the RSA report. The cost of the measures implemented based on the findings of the RSA totaled $23,000. There were no locations along the study corridor that were eligible for funding by the Hazard Elimination Safety (HES) Program. However, the segment from L.M. 18.50 to L.M. 18.70 and the location at L.M. 20.81 were eligible for funding by the High Risk Rural Roads (HRRR) Program, as fatal and/or injury crash rates were in excess of statewide averages.

Table 15. Summary of Safety Issues and Suggestions: RSA of State Route 101.

(Left) Horizontal curvature looking south at L.M. 18.50 to L.M. 18.70. (Right) Vertical alignment looking north at L.M. 18.50 to L.M. 18.70.

(Left) Limited sight distance looking south at L.M. 20.81. (Right) Limited sight distance looking north at L.M. 20.81.

In general, the suggestions emerging from the RSA were well-received by TDOT, local transportation personnel, and the public. While the RSA was well-received, TDOT indicated that they did not observe a noticeable reduction in crashes after the measures were implemented. The cursory review did not, however, account for other potential changes over time. TDOT also stated that the effect of the implementation measures may be altered since the time of the study (i.e., signs are no longer in their suggested locations, vegetation has grown back, etc.). It was clear that a more formal analysis would be necessary to determine the safety impacts of the RSA and implemented strategies.

RSA Outcomes

Geometric, traffic volume, and crash data were collected for both treatment and reference sites in Tennessee. The treatment site consisted of a 3.84 mile corridor, and the specific treatments included the following:

• Segment treatments: add chevron signs, advance curve warning signs, extra-wide (8-inch) thermoplastic edge lines, and new thermoplastic for a portion of the project.
• Intersection treatments: move stop bar to increase sight distance, remove trees from the right-of-way, and relocate sign ("Beach & Marina →").

The reference sites included 21.46 miles of similar corridors that were selected to match the treatment site by number of lanes, speed limit, illumination, and shoulder width. Table 16 provides a summary of the treatment and reference site data.

The initial before period included the years 2000 to 2005 and a relatively large multi-jurisdictional reference group. Based on an initial examination of the data, it was determined that some of the years in both the treatment and reference group exhibited an unusual change in crashes. There was also a wide variation in the annual crash trends among the jurisdictions from which the reference group was selected. The DOT confirmed this finding and noted that in 2001 they changed the structure of their crash database, which resulted in a loss of data due to trouble reconciling the two systems. In addition, they developed a new paper report and implemented it during the period from 2001 to 2002. As a result, it was determined that years 2004 to 2005 would be more suitable for the before period as there were no changes during this timeframe and the previous changes had taken effect. Due to inconsistencies among the jurisdictions, it was also determined that the reference group would only include sites from Cumberland County, the same jurisdiction as the treatment site.

There was some concern regarding the integrity of the data in the after period. In the period from 2006 to 2007, TDOT switched to an electronic crash reporting system, and some of the location information was lost in the report processing (i.e., transferring the data entered by the police officers to the central database). There are some crash data missing for 2006 and approximately 10 percent of crashes missing statewide for 2007. Additional crash data were obtained for 2008 to 2009, and these data are more complete than those from 2006 to 2007.

Table 16. Summary of Tennessee Treatment and Reference Site Data.

The analysis considered the entire corridor. The reference group of similar corridors was used to develop SPFs for use in the EB process. Generalized linear modeling was used to estimate model coefficients, assuming a negative binomial error distribution, which is consistent with the state of research. Separate SPFs were calibrated for total and injury crashes.

The general model form for all models is given by Equation 1. Note that annual multipliers were estimated and added to each model for the appropriate year. The coefficients for each SPF are presented in Table 17.

Equation 1: Crashes/mile/year = exp^ln(a)(AADT)^b exp^c(dd)

Where:

AADT = annual average daily traffic (vehicles per day).

dd = driveway density (total number of driveways and intersections divided by the segment length in miles).

a, b, and c = coefficients estimated from the model.

Table 17. Estimated SPFs for Tennessee.

The results of the Tennessee analyses are presented in Table 18 for total and injury crashes.

Table 18. EB Corridor Results for Tennessee.

The results indicate a general safety benefit of conducting the RSA and implementing the treatments suggested in the RSA report. The estimated CMF is consistently less than 1.0 for both total and injury crashes, indicating a reduction in expected crashes. The reduction in injury crashes is statistically significant at the 95 percent confidence level, but the reduction in total crashes is statistically insignificant. Due to the limited crash counts for other crash types, it was not possible to conduct separate EB analyses for additional target crash types. Instead, a cursory analysis of the proportion of crash types was undertaken to see if any insights could be gleaned. Recall that the treatments included both segment – and intersection-related improvements.

Given the nature of the treatments, the intersection improvements may target rear-end and angle crashes, and the segment improvements may target run-off-road crashes. The crash data do not identify run-off-road crashes specifically but do indicate single – vehicle crashes that were analyzed as a substitute. Table 19 presents the analysis of crash proportions. There are no apparent insights in terms of changes in proportions of the various target crash types. Although there are reductions in the proportion of rear-end and angle crashes, the apparent effects are based on numbers that are much too small from which to draw definitive conclusions.

Table 19. Analysis of Proportion of Crash Types in Tennessee.

The annualized costs and benefits were then compared. Based on the EB methodology for observational before-after studies, this study indicated a 13.7 percent reduction in total crashes and a 31.3 percent reduction in injury crashes over a three-year "after" period. Using the FHWA crash costs document, Crash Cost Estimates by Maximum Police-Reported Injury Severity within Selected Crash Geometries (Publication Number FHWA-HRT-05-051), the annual reduction of 6.23 crashes results in an annual benefit of $200,830. Assuming a total cost of $35,000 ($12,000 for the RSA and $23,000 for the improvements), a discount rate of 2.2 percent, and a service life of 10 years, the estimated annual cost is $3,937. Therefore, the B/C of the improvements made as a result of the RSA is 51:1 based on the estimated change in total crashes.

The annual reduction in injury crashes is 4.97 crashes per year. This results in a range of annual benefits, from $199,301 per injury crash (assuming all Type C injuries) to $968,432 per injury crash (assuming all Type A injuries). The estimated cost of a generic injury crash is $367,668. Based on the same annual cost from above, the injury B/C ranges from 51:1 (Type C) to 246:1 (Type A). The B/C is estimated as 116:1 for generic injury crashes. These results indicate a positive effect on safety as a result of the RSA performed on State Route 101.

Program Review Selection Process

Each year, TDOT's safety section generates a list of candidate projects. On average, approximately 60 projects are selected statewide annually. These projects qualify as either a Hazard Elimination Safety Project (HESP) or an HRRR project. Three conditions must be met to qualify as a HESP: 1) locations must have experienced fatal and/or incapacitating injury crashes; 2) a total of seven crashes must have occurred in a three-year period; and 3) the actual crash rate must be at least four times the critical rate. (NOTE: crash rates are typically calculated on a per-million-vehicle-miles-traveled basis.) Instances that may remove projects from the HESP list include an unsupportive crash history due to a miscoding of incidents or other projects already programmed for the location in which a review has been or will be conducted.

HRRR projects consist of intersection, spot, or short segment locations. Like HESPs, these locations must have experienced fatal and/or incapacitating injury crashes, three total crashes must have occurred in a three-year period, and the severe crash rate must exceed the statewide average severe crash rate for similar locations.

In addition to the approximate 60 projects each year, additional localized projects are request-based. Requests are typically submitted via the comments section on TDOT's website, through locally-elected officials, or by public contact with TDOT regional offices. To be considered for review, crash data must indicate that the location has experienced a fatal and/or incapacitating injury crash, a total of three crashes must have occurred during a three-year period, the total crash rate must exceed the statewide average crash rate, and the severe crash rate must exceed the statewide average severe crash rate for similar sections. Although the statistics are calculated for a three-year period, it is recommended that data be studied for at least a six-year period to get a more thorough understanding of the severe crash history. The qualifying statistics can be developed based on any consecutive three years of the six-year period. On average, approximately 10 request-based projects are selected each year. Those locations that do not meet the requirements or are not selected are passed on to the regional engineer for a maintenance review or to be included in the "Spot Safety" program.

RSA Process

TDOT conducts spot or section RSAs – called Road Safety Audit Reviews (RSARs) by TDOT – on Interstates, State roads, and local functionally-classified roadways. For non-classified roadways, TDOT depends on the county or city to provide the appropriate data. RSAs on these facilities are generally request-based. The total number of RSAs conducted throughout the State typically varies from year to year.

Since the RSA on State Route 101 in 2005, the TDOT safety team has become more defined; thus, RSAs are conducted in a more efficient and cost effective manner. In-house RSAs are typically led and written by the project planning team. When consultants are involved, the team manages consultant activities. TDOT conceptual and National Environmental Policy Act (NEPA) planning personnel handle the RSA improvement cost estimates and development of layout sheets (design plans scaled down for field review purposes).
The size of an RSA team is typically 10 to 12 people. The team members represent TDOT Traffic Engineering, Planning, Design, Maintenance, Right-of-way, and Construction. Other agencies represented include regional metropolitan planning organizations (MPOs), rural planning organizations (RPOs), and law enforcement. The TDOT Bike and Pedestrian Coordinator is also generally contacted for input. Generally, the RSA team is led by the TDOT Regional Engineer.

Information typically requested for purposes of conducting RSAs includes crash data, traffic data, and design plans. In terms of crashes, data from the most recent five years are used for HRRR projects, and the most recent three years of crash data are used for HESP projects. The most recent year of traffic data is used. When applicable, turning movement counts are analyzed or may be conducted. Lastly, design plans are used to generate layout sheets to show problematic areas/issues and locations for potential improvements (e.g., sign placement, geometric/alignment changes, etc.).

In general, TDOT indicates that the time from when crash data reports are first pulled to the time when the RSA report is officially signed is typically between 2 and 2-1/2 months. The following is a list of the order of events and the associated time requirements:

1. Candidate projects are ranked in order of their severity index (from the most severe to the least severe), as calculated from the HSIP list.
2. A packet of crash reports and crash rate summary sheets are received. Crash rate sheets are supplied for in-house RSAs only; consultants generate their own. Information is distributed to the RSA team prior to the pre-brief meeting.
3. The pre-brief/background meeting is scheduled and the RSA time and location are set. Usually the RSA is conducted within one week of the pre-brief meeting.
4. For an RSA of one location, the draft report is developed within two weeks of the site visit. For an RSA conducted at multiple locations or along a corridor, the draft report is developed within three to four weeks of the site visit.
5. The draft report is reviewed over a one – to two-week period.
6. The draft report is edited within a two-week period.
7. The project manager (typically the TDOT Regional Engineer) verifies that changes have been made and then submits the RSA report to the director for review. At this point, the director may accept and sign the report; however, it is typical that the director will have comments to be addressed.
8. The report is typically finalized within one week of the Director's review.
9. The report is distributed to all RSA team members, both electronically and by hard copy.
10. The report is submitted to the TDOT environmental section, which typically takes one to two weeks to issue a "programmatic categorical exclusion (CE)."
11. The report is submitted to the TDOT funding section, where a project number is received. The project is then submitted for implementation.

Figure 5 illustrates TDOT's RSA process through the development phase.

Figure 5. RSAR Review Flow Chart: Development Phase.

Implementation

On average, implementation occurs approximately 18 months from the start of the RSA process. TDOT indicates that this time period is highly variable, as "political motivation" occasionally drives implementation scheduling. Figure 6 illustrates TDOT's RSA process through the implementation phase.

Figure 6. RSAR Review Flow Chart: Implementation Phase.

Funding for implementing RSA improvement measures may come from HRRR funds or HESP funds. For HRRR projects, TDOT has a self-imposed "soft cap" of $50,000 per project. For HESP projects, TDOT has a self-imposed "soft cap" of $1,000,000 per project. This self-imposed cap is intended to limit the project extents and the type of improvement measures to be considered. Funding in excess of the cap will be approved when necessary, but these occurrences are very rare. Local officials generally exhibit an interest in RSAs due to funding obligations. For projects not on TDOT's 100 percent funding list, locals have to match 10 percent of the total project cost.

At the time of the State Route 101 RSA, the implementation was somewhat rapid, being only six months from approval. This quick turnaround time was likely due to TDOT Maintenance involvement. However, since that time, all projects (including signing and marking) are let out to contract. Projects with a lower dollar amount are bundled together, adding time to implementation because of the need to wait for multiple projects to accumulate. Projects with a higher dollar amount may be implemented in a shorter timeframe, unless right-of-way acquisition is involved. In either case, contracting the work rather than using State forces has added significantly to implementation time. TDOT Maintenance divisions—which are responsible for implementation—are trying to reestablish the means to use on-call contractors rather than let out every project to contract.

TDOT is not aware of any problems with implementing RSA suggestions. Typical questions of "concern" include, "Why aren't we spending more (in terms of improvements)?" and "Why can't we (TDOT Maintenance) pay to do the work now and get reimbursed later?" In general, suggestions and improvements at all RSAs are well received. All stakeholders (i.e., TDOT, local agencies, and the public) seem pleased to see that strides are being made to improve roadway safety.

TDOT monitors each RSA location for a three-year period from the notice of completion. If there is a recurring problem within the three-year period, TDOT reevaluates the improvement measures and makes adjustments, as necessary. Each year, the locations from the HRRR and HSIP lists are reentered into TDOT's database to see if they come up again as repeat critical locations.

Since the time of this project in 2005, the RSA process has become more formalized and uniform. TDOT has developed a manual to guide the RSA process, to explain project and data needs, and to establish a "boilerplate" report format. TDOT has described the RSA as a "living process"; TDOT is consistently finding ways to improve the RSA process, making it better along the way.

 

Collier Boulevard at Golden Gate Parkway: Collier County, Florida

Road: Existing intersections along paved, four-lane, divided roadway
RSA Sites:	Three-legged, signalized intersection Adjacent unsignalized intersection at Collier Boulevard / 25th Avenue AADT: 18,000 – 24,000 vehicles per day (2005) 45 mph posted speed limit
Environment:	urban/urbanizedsuburban rural
Owners:	Florida Department of Transportation (FDOT); Collier County
Road Safety Audit
Date of RSA:	12 July 2005
RSA Stage(s):	planning/design stage RSA of existing roads
RSA Team:	2 teams of 4 – 5 members representing the consultant, County engineers, non – engineers (i.e., technician), and law enforcement (County Sheriff Department)
RSA Cost:	$15,000
Implementation Cost:	$250,000
Crash Reduction:	11% total; – 10% injury (note the "-" sign denotes an increase)
Benefit / Cost Ratio:	5:1 total; – 3:1 injury

Background

An RSA was conducted at the intersection of Collier Boulevard and Golden Gate Parkway in Collier County, Florida (Figure 7). The intersection was chosen based on a definitive crash issue. Consequently, Collier County decided to evaluate this intersection to determine the need for appropriate safety measures. Additionally, upgrades desired at this particular location were wrapped into the RSA process. Due to its close proximity, the RSA team also evaluated the adjacent unsignalized intersection at Collier Boulevard/25th Avenue. The County used the RSA for an in-house design of median closures and directional turns.

Figure 7. Aerial Map of Collier Boulevard/Golden Gate Parkway.

The RSA was conducted by two teams, with each having four to five members. The teams were comprised of consulting staff, County engineers, non-engineers (technicians), and law enforcement (County sheriff's department). The efforts of the teams were led by the consultant with direct oversight by County personnel.

Information used by the RSA teams to conduct the review included the following:

• Crash data (2002 – 2004, potentially partial 2005).
• Traffic data (2002 – 2004, potentially partial 2005).
• Aerial imagery.

Overhead maps were provided for use by the consultant to indicate problem areas and issues and to draw potential improvements (e.g., sign placement, geometric/alignment changes, etc.). If design plans were used, County personnel indicate that they were probably very old and out of date.

For this RSA, three hours (total) were required to hold the opening and closing meetings. The two RSA teams spent two days in the field reviewing the study locations. Both teams reviewed the same locations: one team started on Collier Boulevard, while the other started on Golden Gate Parkway. A night review was also conducted by members of each team, although not all members from each participated. Each team completed RSA reports immediately following the site visits. The consultant compiled the reports written by the RSA teams. If one team felt more strongly about a particular issue or observation than the other team, the team's remarks were identified as "Team A" or "Team B" in the report. The draft report was submitted by the consultant to the County within one week after completion of the field reviews. In addition to meeting and field review time, the consultant traveled four hours each way to perform the RSA. Total cost to complete the RSA (including consultant fees) was $15,000.

Key Findings and Suggestions

The key findings and suggestions of the RSA are summarized in Table 20. County personnel responded that all of the suggested improvement measures in the RSA report were installed between January and March 2006, approximately six months after the RSA was completed and documented. There were no additional improvements made beyond those suggested in the RSA report. The costs of the measures implemented based on the findings of the RSA totaled $250,000. This amount covered design through construction activities. These improvements were funded by gas tax revenues. County personnel indicated that the tax source may be County and State, as well as Federal.

Table 20. Summary of Safety Issues and Suggestions: RSA of Collier Boulevard & Golden Gate Parkway.

	Safety Issues	Suggestions
Signing and Marking	Object markers obscured by signal poles on EB Golden Gate Pkwy End of Golden Gate Pkwy not clearly defined. Need to enhance signing NB approach not well marked Route markers on all approaches poorly placed and obscured Street name signs poorly placed. NB visibility is particularly poor	Relocate obscured object markers Install overhead signing Add supplemental plaque for 25th Ave. SW to cross street sign on NB approach Review and revise placement of route marker and destination signing Improve placement of street name signs
Intersection Geometry	EB left-turn lane confusing: (1) opportunity to make two left – turn movements out of a single lane (41st St. SW, alley to east); (2) possible misinterpretation of lane use U-turns observed to cross over island at plaza just east of 41st St. SW Provision for SB left turns to 25th Ave. SW impacts queue space on Collier Blvd WB to SB left turn can be blocked by 2-stage left-turn movements	Cut median nose back to facilitate U-turns just east of 41st St. SW. Modify median opening providing access to 25th Ave. SW. Consider median closure
Visibility, Sight Distance	SB sight distance issues from inside lane with NB through. Problem compounded when left-turning vehicle from 25th Ave. SW to SB. Median hedges impact visibility for NB and SB left – turn and U-turn movements	Trim vegetation in median
Pedestrian Accommodations	No sidewalk on south side	Review sidewalk needs Confirm pedestrian phasing will be installed

The following photos illustrate safety issues observed during the RSA ("before" on left) and the subsequent improvements resulting from the RSA suggestions ("after" on right).

SB left-turn lane to 25th Ave. SW (left/before). - SB left-turn lane to 25th Ave. SW (right/after).

EB median nose at plaza just east of 41st SW (left/before). - EB median nose at plaza just east of 41st SW (right/after).

In general, the suggestions resulting from the RSA were well-received by County personnel, law enforcement, and the public. Residents near the study intersection were satisfied with the changes, as access into the residential area across the canal at 25th Street was improved. Overall, a qualitative improvement in safety has been observed. County personnel also indicate that, in the long term, the measures implemented have eased traffic flow through the intersection. Drivers are now more expectant of the maneuvers to be made, enhancing the traffic flow and overall intersection safety.

Those involved with this RSA were also satisfied with the process. In the end, the County decided that they probably didn't need two teams to complete this RSA. County personnel indicated that one team could complete the same amount of work required and still come up with the same conclusions.

RSA Outcomes

Geometric, traffic volume, and crash data were collected for both treatment and reference sites in Florida. The treatment sites consisted of one signalized intersection (Collier Boulevard/Golden Gate Parkway) and the nearby unsignalized intersection (Collier Boulevard 25th Avenue Southwest). The RSA considered both locations because the improvements at 25th Avenue Southwest included the prohibition of left-turns from one direction, which now requires that those vehicles first turn right and make a U-turn at the intersection with Golden Gate Parkway. The reference sites consisted of 22 untreated signalized intersections that were otherwise similar to the treated signalized intersection. A specific reference group was not available for the unsignalized treatment site, but the signalized reference group was utilized to adjust for specific trends. (This is discussed further in analysis section.) Table 21 provides a summary of the treatment and reference site data.

Table 21. Summary of Florida (Collier Boulevard) Data.

The signalized reference group was used to develop SPFs for use in the EB process. Generalized linear modeling was used to estimate model coefficients, assuming a negative binomial error distribution, which is consistent with the state of research. Separate SPFs were calibrated for each target crash type. Since minor road AADTs were only available for some of the reference sites, only the major road AADT was considered as an exposure variable in the models.

The general model form for all models is given by Equation 2. Note that annual multipliers were estimated and added to each model for the appropriate year. The coefficients for each SPF are presented in Table 22.

Equation 2: Crashes/year = exp^(a)(MajAADT)^b

Where:

MajAADT = major road AADT

a and b = coefficients estimated from the model

Table 22. Estimated SPFs for Florida.

The EB method could not be applied to estimate the number of crashes without treatment at the unsignalized intersection because a suitable reference group was not provided. For this location, the expected number of crashes without treatment was estimated by the before period crash counts and several factors. The first factor was used to adjust for the different lengths of the before and after periods and was computed as the ratio of the duration of the after period to the before period. The second factor was used to adjust for the change in traffic volume between the before and after period and was computed as the ratio of AADT in the after period to before period. (Note that this adjustment assumes a linear relationship between traffic volume and crashes, which may not be valid for large changes from the before to the after period). The final factor was used to adjust for general temporal trends and was computed as the ratio of SPF estimates in the after period to the before period.

The results of the Florida analyses for Collier Boulevard are presented in Table 23. Note that specific results are provided for total, injury, left-turn, rear-end, and sideswipe crashes. Results are presented separately for the unsignalized treatment site, the signalized treatment site, and the two sites combined.

The results indicate a general safety benefit of implementing the treatments at the unsignalized intersection. As expected, the elimination of westbound left-turns resulted in a substantial and significant reduction in several crash types, including total, injury, left-turn, and sideswipe crashes. There was also a reduction in rear-end crashes, but the result was not statistically significant.

As might be expected, there was an increase in several crash types at the signalized intersection, potentially due to the increase in turning movements after the westbound left-turns were eliminated at the adjacent unsignalized intersection. Specifically, there was an increase in total, injury, left-turn, and sideswipe crashes at the signalized intersection; however, these increases are highly insignificant. There does appear to be a reduction in rear-end crashes at the signalized intersection, but, again, the result is highly insignificant.

Table 23. EB Results from Florida.

Overall, there appears to be a reduction in several crash types based on the results for both sites combined. The reduction in crashes at the unsignalized intersection is partly offset by an increase in crashes at the signalized intersection, but the results still show an overall reduction in total crashes. Specifically, the net effect is an 11 percent reduction in total crashes over the two intersections, which is significant at the 58 percent confidence level. The combined results also indicate a reduction in left-turn, rear-end, and sideswipe crashes. These results are highly insignificant, as is the increase in injury crashes.

The annualized costs and benefits were then compared based on the results of the two intersections combined and focusing on total crashes. Based on the analysis, this study indicated an 11 percent reduction in total crashes over a 2.75-year after period. Using the FHWA crash costs document, Crash Cost Estimates by Maximum Police-Reported Injury Severity within Selected Crash Geometries (Publication Number FHWA-HRT-05-051), the annual reduction of 4.45 total crashes results in an annual benefit of approximately $143,597. The analysis was conducted based on a total cost of $265,000 ($15,000 for the RSA and $250,000 for the improvements), a discount rate of 2.2 percent, and a service life of 10 years, resulting in an annual cost of $29,811. Therefore, the collective B/C of the improvements made as a result of the RSA is 5:1, indicating an overall positive effect on safety in terms of total crashes.

Program Review

The Board of Collier County Commissioners in the State of Florida adopted an RSA Policy that included RSAs for 30 percent and 60 percent design plans for their capital improvement projects, as well as a predetermined number of existing high-crash locations or roadways. RSAs are also required for new site development permits. The County utilizes an annual professional services contract to deliver the RSA program. The consultant serves as the team leader with agency staff and other experts serving as the RSA team. The contract is negotiated for each specific RSA project with typical consultant costs between $6,000 and $15,000. The County has found that the life-cycle savings far outweigh the RSA cost. The services contract is an effective method to train county staff on how to perform an RSA. The contract has also institutionalized the RSA process, which has adapted to meet local needs and focuses on what is implementable.

The County has also trained locally-based groups of highway safety advocates to form Community Traffic Safety Teams (CTSTs) to complete RSAs. An example of CTST involvement was a pre-design RSA on a greenway. Members of the CTST were avid cyclists who provide a safety perspective from a cyclist's standpoint. The CTST is not always called upon to assist in intersection or corridor reviews. The County has also had members of RSA teams in wheelchairs. This adds another perspective to RSA reviews, as a team member can enhance the RSA team's understanding of issues that affect people with mobility restrictions.

Immokalee Road: Collier County, Florida

Background

An RSA was conducted along Immokalee Road from Interstate 75 to Collier Boulevard in Collier County, Florida (Figure 8). To accommodate current and forecasted future demand, the County was considering widening the road. The proposed upgrade entailed widening the road from a four-lane to a six-lane divided urban arterial. At the time of the RSA, the widening project was at a conceptual stage. Design documentation consisted of State design practices and the County's typical section for a six-lane divided urban arterial.

The RSA was conducted by a team of experts representing staff from the consultant, Collier County, and the FHWA Resource Center. The efforts of the teams were led by the consultant, with direct oversight by County personnel.

Figure 8. Aerial Map of Immokalee Road RSA Location.

Conceptual design plans (less than 30 percent complete), the State's design standards, and aerial photos were used to conduct this RSA.

Three hours (total) were required to hold the opening and closing meetings. The RSA team spent two days in the field reviewing the study locations. A night review was also conducted, although not all members participated.

Key Findings and Suggestions

The RSA findings were divided into the following two parts:

1. Issues arising from the proposed upgrade: These issues reflected concerns arising from the proposed upgrading from a four-lane mixed urban/rural arterial to a six-lane urban arterial cross section and dealt with properties or specific conditions of the Immokalee Road widening project that could influence design details or require the modification of design standards.
2. Further opportunities for improvement: Additional opportunities for improvements that could be incorporated during the upgrade were also identified. Although these improvements were not specifically related to the proposed widening, they responded to issues observed along the corridor by the RSA team. These improvements could be "piggybacked" onto the upgrade project at relatively limited effort and expense.

The key findings and suggestions as a result of the RSA are summarized in Table 24. County personnel indicated that most of the suggested improvement measures in the RSA report were installed between March 2006 and March 2009.

Table 24. Summary of Safety Issues and Suggestions: RSA of Immokalee Road.

The following photos illustrate safety issues observed during the RSA.

Immokalee Road near Valewood Drive. (left) - Immokalee Road in front of schools. (right)

In general, the suggestions resulting from the RSA were well-received. In design-stage RSAs, a wider scope for safety enhancements exists. This opportunity is even more pronounced for designs that are early in the planning process (e.g., a conceptual design).

This RSA was conducted very early in the planning stage, when the only design documentation was the County's typical section for a six-lane divided urban roadway. As a result, the RSA was able to not only consider issues related specifically to the proposed widening from four to six lanes but could also suggest additional opportunities to enhance road safety that were not necessarily related to the specified widening. Of particular significance was the opportunity to include several measures to accommodate older drivers, which will help the County to implement elements of the Florida DOT's "Elder Road User Program" on County roads.

RSA Outcomes

Geometric, traffic volume, and crash data were collected for both treatment and comparison sites in Florida. The treatment site consisted of 3.25-mile corridor that included six signalized and several unsignalized intersections and driveways. The comparison sites consisted of six untreated corridors that were otherwise similar to the treated site. The comparison group was utilized to adjust for specific trends. (This is discussed further in the analysis section.) Table 25 provides a summary of the treatment and reference site data.

Table 25. Summary of Florida (Immokalee Road) Data.

The analysis considered the corridor in its entirety, including the signalized and unsignalized intersections. A naïve before-after analysis was conducted but included adjustments from a comparison group and also employed an EB technique to assess the potential for regression-to-the-mean bias. It was only possible to analyze total crashes because data were not provided for other crash categories.

The EB method was used to assess the potential for regression-to-the-mean bias. Based on a relatively limited reference group, an SPF was developed for total crashes. Generalized linear modeling was used to estimate model coefficients, assuming a negative binomial error distribution, which is consistent with the state of research. The general model form for the SPF is given by Equation 3 and the model coefficients are presented in Table 26.

Equation 3: Crashes/year = Length*exp^(a+b*lanes)(AADT)^c

Where:

AADT = total two-way traffic volume (vehicles per day)

Lanes = indicator variable for the number of lanes (1 if 4-lanes; 0 if 6-lanes)

Length = segment length (miles)

a, b, and c = coefficients estimated from the model

Table 26. Estimated SPFs for Florida.

While the coefficients from the SPF were not statistically significant, it was possible to investigate the potential for regression-to-the-mean. The EB-adjusted expected crash count in the before period was 70.74 as opposed to 68 observed crashes indicating that regression-to-the-mean is not a significant consideration. Based on the lack of evidence to suggest the potential for regression-to-the-mean and the fact that the SPF and Collier County reference data are suspect for some locations, it was decided to use the naïve before-after method with observed crash counts in the analysis.

The naïve before-after method was used to estimate the overall change in safety at the treatment site. However, there were other improvements made to the treatment site in addition to the suggestions from the RSA. Specifically, the corridor was widened from a four-lane to a six-lane cross section. As such, it was necessary to adjust for the lane conversion to isolate the effects associated with the RSA. An adjustment factor was derived from the most relevant CMF information in the Highway Safety Manual (HSM) (3). The HSM provides separate CMFs for four – to five-lane conversions and five – to six-lane conversions as follows:

CMF for 4-lane to 5-lane conversion = 1.10 (standard error =0.07)

CMF for 5-lane to 6-lane conversion = 1.04 (standard error = 0.11)

The estimated effect of converting from four to six lanes is obtained by multiplying the two CMFs (CMF = 1.10*1.04 = 1.14).

A final factor was used to adjust for general temporal trends and was computed as the ratio of number of crashes in the after period to number of crashes occurring in the before period for the comparison group. The time trend factor is 1.05 and represents the change from 2004 to 2009. It was assumed that this calculated time trend was applicable to the treatment site for which the before period is 2005 and the after period is January 2010 to July 2011.

The results of the Florida analyses for Immokalee Road are presented in Table 27. Note that Table 27 contains three sets of results, including 1) the naïve before-after analysis without adjustment, 2) adjustment for the lane conversion, and 3) adjustment for the lane conversion and time trend.

Table 27. Corridor Results for Florida (Immokalee Road).

The results indicate a general safety benefit of implementing the suggestions in the RSA report.

While the unadjusted naïve analysis indicates an insignificant increase in total crashes, the results show an expected reduction in crashes once adjustments are made to account for the effects of the four – to six-lane conversion and time trends. Specifically, the analysis indicates an expected 10.8 percent reduction in total crashes as a result of the RSA and associated improvements after adjusting for time trends and the lane conversion. Note that this result is statistically insignificant at the 95 percent confidence level. Using the FHWA crash costs document, Crash Cost Estimates by Maximum Police-Reported Injury Severity within Selected Crash Geometries (Publication Number FHWA-HRT-05-051), the annual reduction of 12.33 total crashes results in an annual benefit of approximately $397,470. It was not possible to compute the B/C ratio for the improvements made as a result of the RSA because cost data were not provided, but there appears to be an overall positive effect on safety in terms of total crashes. As long as the annual costs are less than approximately $400,000, there will also be a positive return on investment.

Program Review

See information in the previous section for Collier Boulevard.

 

Ninth Street: Ocean City, New Jersey

Road: Existing 0.47-mile section of paved, four-lane arterial roadway
RSA Sites:	5 signalized, 1 unsignalized "major" intersections 6 unsignalized other intersections and alleyways AADT: 13,870 vehicles per day (2007) 25 mph posted speed limit
Environment:	urban/urbanizedsuburbanrural
Owners:	Ocean City, New Jersey
Road Safety Audit
Date of RSA:	28 May 2004
RSA Stage(s):	planning/design stage RSA of existing roads
RSA Team:	14 members representing Cape May County, FHWA, NJ Division of Highway Traffic Safety, New Jersey Department of Transportation (NJDOT), Ocean City, South Jersey Transportation Planning Organization (SJTPO), and Orth-Rodgers & Associates, Inc.
RSA Cost:	$21,000
Implementation Cost:	$900,000
Crash Reduction:	25.6% total
Benefit / Cost Ratio:	1.2:1 total

Background

An RSA was conducted along a 0.47-mile section of Ninth Street in Ocean City, New Jersey from Bay Avenue to Atlantic Avenue (Figure 9). Throughout this section, Ninth Street is a four-lane arterial roadway with a posted speed limit of 25 mph. Five signalized intersections and one "major" unsignalized intersection are present within the study area; six other unsignalized intersections/alleyways are also present. The AADT along Ninth Street was reported as 13,870 vehicles per day in 2007. Ninth Street is owned and maintained by Ocean City.

Figure 9. Aerial Map of Ninth Street.

In November 2003, a preliminary list of six candidate locations for RSAs was submitted by the Cape May County Engineer and forwarded to the NJDOT Bureau of Data Development for further analysis. At this time, the Local Safety Program provided only $0.5 million to each NJ metropolitan planning organization annually. Due to time constraints for using the available metropolitan planning (PL) funding, all RSA work needed to be completed by June 30, 2004. Data-driven approaches to the RSA process had not yet been well developed; therefore, the SJTPO used outreach, local knowledge, and limited data analysis to identify Ninth Street as a high-crash corridor. As a result, Ninth Street was one of two locations selected for an initial pilot project by the SJTPO for a formal RSA to determine safety deficiencies and appropriate countermeasures. SJTPO proceeded with the expectation that the RSA was likely to yield low-cost, quick-turnaround improvements, as opposed to high-cost, construction-intensive projects requiring extensive NEPA processing.

A kickoff meeting was held in March 2004 and attended by 24 individuals. RSAs were relatively new to New Jersey, and there was a strong desire to see the process expanded to other parts of South Jersey. In May 2004, the RSA was conducted by a team of 14 individuals from various agencies, including representatives from Ocean City, Cape May County, SJTPO, NJDOT, the New Jersey Division of Transportation Safety, and FHWA.

The team was comprised of consulting staff, engineers, planners, and law enforcement. The efforts of the team were led by the consultant with direct oversight by County personnel.

Information used by the RSA team to conduct the review included the following:

• Aerial imagery and maps.
• Traffic volume data (2000).
• Bicycle and pedestrian counts.
• Level of Service (LOS) analysis.
• Crash data (2001-2004).
• Traffic signal plans.
• Video logs.

Aerial maps (including an 11-inch by 14-inch sheet scaled at 1 inch =500 ft and additional maps at 1 inch =60 ft) were provided for use by the County to indicate problem areas and issues and to draw potential improvements (e.g., sign placement, traffic signal improvements, etc.). Traffic volumes indicated a large presence of non-motorized traffic, with peak demand occurring during the midday period. A capacity analysis demonstrated that all intersections were operating at acceptable levels of service, except for the Bay Avenue intersection. Approximately 200 crashes were submitted for review; 128 were identified as having occurred within the study area (which included an influence area around the intersections as determined by the consultant).

In the morning, after a brief introduction, a three-hour brainstorming session allowed participants to review the existing data and discuss several operational and congestion-related issues. In the afternoon, the team conducted the field review. It was estimated that five hours were utilized to complete the field review. The total cost to complete the RSA (including consultant fees) was approximately $21,000.

Key Findings and Suggestions

The consultant prepared a draft report to document the RSA, the key findings, and suggestions. In total, 42 issues were identified with 53 safety measures suggested. The measures cover a wide range of actions, from maintenance-level tasks to capital-intensive projects. Each measure was ranked in priority high to low based upon the level of effort required and the potential safety benefit. Of the 53 suggestions, 33 (62 percent) were projected to have a high safety benefit, and 13 (25 percent) were projected to have a medium safety benefit. Fifteen of the proposed suggestions could be implemented within one year, whereas the remaining 38 suggestions were intended to be incorporated into longer-range design plans. Those suggestions that were implemented are summarized in Table 28.

Table 28. Summary of Safety Issues, Suggestions, and Implementation Status: RSA of Ninth Street.

Traffic signal mast arms and signal display alignments prior to improvements.

One of the major changes suggested as a result of the RSA was to implement a roadway diet, converting Ninth Street from a four-lane section to a three-lane section with a two-way left-turn lane and bicycle lanes. Another suggestion resulting from the RSA was to install a traffic signal at the intersection of Ninth Street and West Avenue. The improvements included an uninterruptible power supply, all new signal hardware, 12-inch LED lenses, accessible countdown pedestrian signals with ramps, sidewalk improvements, and signing and pavement marking enhancements. A preliminary before-after study indicated a 50 percent reduction in crashes and injuries at the intersection. This study provides the results of a more rigorous evaluation. The cost of this improvement was $172,000 and was completed on June 15, 2005.

New traffic signal mast arms and signal display alignments resulting from the RSA.

Further traffic signal improvement work was implemented based on the RSA suggestions. The existing traffic signals were outdated and did not provide pedestrian accommodations. The work consisted of installing countdown pedestrian traffic signals at the intersections of Asbury, Central, Wesley, and Ocean Avenues. The work also interconnected signal controller communications with all five traffic signals between West Avenue and Ocean Avenue. This was a joint project between Cape May County and Ocean City and was funded by the HSIP program. The Ninth Street Signals Improvement Project was completed in May 2007 at a cost of $728,000.

In addition to the traffic signal upgrades, several other safety measures will be implemented with the Route 52 Causeway construction project to the immediate west of the study corridor. The Route 52 project will replace two drawbridges and tie into the Ninth Street corridor at West Avenue. Included within this project is the raising of portions of Ninth Street where flooding causes periodic closures, new raised medians, and new pedestrian sidewalk along the south side of the roadway (Figure 10). Construction began in August 2006 and is scheduled for completion in December 2012 at a cost of $400 million. Ocean City was aware that the Route 52 construction would affect the western portion of the corridor, particularly in terms of significantly altering traffic patterns. Additional improvements will be made beyond those suggested in the RSA report, as the causeway is a significant gateway to Ocean City.

Figure 10. Route 52 Construction Changes to Ninth Street.

In general, the suggestions resulting from the RSA were well-received by County personnel, law enforcement, and the public. Those involved with this RSA were pleased with the process and sought to expand this pilot into a full RSA program. Additionally, residents near the study area were happy with the outcomes. According to local officials, the public was especially receptive of the traffic signal improvements.

RSA Outcomes

Geometric, traffic volume, and crash data were collected for both treatment and comparison sites in New Jersey. The treatment site consisted of a 0.47-mile corridor that was divided into two consecutive segments based on geometric changes. Data were obtained for each segment of the corridor and the signalized intersections within the corridor. Several comparison sites were identified for the corridor and signalized intersection analyses. A total of 4.32 miles of corridor were identified for the comparison group. Tables 29 and 30 provide a summary of the treatment and comparison site data for the corridors and signalized intersections, respectively. Note that two before periods were analyzed separately: 2003 to 2004 and 2003 to 2007. This approach was based on the implementation period in which most changes were implemented in 2008, but the intersection of Ninth Street/West Street was also modified in 2005. It is preferable to include more than two years of data in the before period; hence, the analysis considers a before period of 2003 to 2007. The additional years in the before period provide a more stable estimate of the safety performance prior to the treatment's installation. The period from 2003 to 2004 was used to account for changes due to the improvements made in 2005.

Table 29. Summary of New Jersey Corridor Data.

Table 30. Summary of New Jersey Signalized Intersection Data.

In New Jersey, the countermeasures implemented as a result of the RSA were mostly implemented in 2008; however, one intersection was modified in 2005. Thus, the after period conditions changed at two different times for the treated segments. Recognizing the changing conditions, the analysis was run for two after periods. The first after period was based on data from 2003 to 2007 (i.e., the period before most of the major changes occurred). The second after period only considered the period from 2003 to 2004 (i.e., the period before any changes occurred).

The analysis considered the corridor in its entirety, including the signalized intersections, and separately considered the signalized intersections within the corridor. Comparison group and naïve before-after analyses were conducted for both scenarios. It was only possible to analyze total crashes for each scenario because there were too few crashes in the other categories.

The analysis did not account for regression-to-the-mean because the EB approach could not be applied. However, it was possible to investigate the potential for regression-to-the-mean. Specifically, total crashes were plotted over time for the corridor and signalized intersections, as shown in Figures 11 and 12. If regression-to-the-mean is present, one would expect to see a spike in crashes just prior to 2005 and/or 2008 (recall that the majority of changes occurred in 2008, with modifications to a signalized intersection in 2005). Figures 11 and 12 do not indicate such a spike in crashes but rather reflect a consistent oscillation around some mean value. Thus, for the purposes of the analysis, it is concluded that regression-to-the-mean is not likely to introduce significant bias.

Figure 11. Corridor-Level Crash Totals Over Time.

Figure 12. Signalized Intersection Crash Totals Over Time.

The following points should also be noted with respect to the analysis.

• A linear relationship was assumed between crashes and traffic volume when accounting for changes in traffic volume. With a limited variation in traffic volumes, it was not possible to develop an SPF or evaluate the suitability of existing SPFs to model the true relationship. Given the inherent uncertainty in other aspects of the analysis and the relatively small changes in AADT, this assumption was not deemed crucial.
• The suitability of the comparison group could not be rejected using the methods described by Hauer (1997). As such, the comparison site is deemed reliable for use in the comparison group method.

The results of the New Jersey analyses are presented in Table 31 and Table 32 for the corridor and signalized intersections, respectively. Results that are statistically significant at the five percent level are noted in bold. Note that Table 31 contains results for the comparison group and naïve before-after analyses; separate results are provided for each of the following:

1. Comparison group and naïve before-after methods, including entire before period.
2. Comparison group and naïve before-after methods, excluding years 2005 to 2007 from the before period.

Table 32 contains results for the comparison group and naïve before-after analyses, and separate results are provided for each of the following:

1. Comparison group and naïve before-after methods, including the signalized intersection of Ninth Street and West Avenue.
2. Comparison group and naïve before-after methods, excluding the signalized intersection of Ninth Street and West Avenue.

Table 31. Corridor Results for New Jersey.

Table 32. Signalized Intersection Results for New Jersey.

The results indicate a general safety benefit of conducting the RSA and implementing the treatments suggested in the RSA report. The estimated CMF is consistently less than 1.0 for all variations of the analysis, indicating a reduction in expected crashes. While the estimated reductions in total crashes for the corridor are substantial, they are insignificant for the comparison group method. With respect to the treated signalized intersections, the results indicate a safety benefit, although not statistically significant when the intersection of Ninth Street and West Avenue is included.

While there are several variations of the analysis, the results from the comparison group method are preferred over the naïve before-after results. The estimated crash reductions are consistently smaller for the comparison-group method than for the naïve method and could be considered conservative. However, the naïve method does appear to corroborate the results of the comparison group method.

While the EB methodology could not be applied in this evaluation, the results are consistent among the various approaches, even when considering the potential for regression-to-the-mean. The estimated crash reductions are also large enough to justify confidence in the RSA process, especially considering that the evaluation was based on only one project and on relatively low crash counts. The confidence level in the RSA process will be increased as the results of other projects developed from RSAs are amalgamated with the results in this report.

The annualized costs and benefits were then compared based on the results of the comparison group before-after study. For the corridor as a whole, the analysis indicated a 25.6 percent reduction in total crashes over a one-year "after" period. Using the FHWA crash costs document, Crash Cost Estimates by Maximum Police-Reported Injury Severity within Selected Crash Geometries (Publication Number FHWA-HRT-05-051), the annual reduction of 3.8 total crashes results in an annual benefit of $122,497. Assuming a total cost of $921,000 ($21,000 for the RSA and $900,000 for the improvements), a discount rate of 2.2 percent, and a service life of 10 years, the annualized cost is $103,608. Therefore, the B/C ratio of the improvements made as a result of the RSA is 1.2:1 for the corridor, indicating an overall positive effect on safety in terms of total crashes.

Program Review

Transportation safety is one of the SJTPO's top priorities. Since completing its first RSA on Ninth Street in 2004 without the benefit of rigorous corridor-selection criteria, the SJTPO created a comprehensive process for the evaluation of future candidate RSA locations. In its fiscal year 2005 Planning Work Program, SJTPO began to gear the RSA program towards identifying locations where quick, low-cost safety improvements could be implemented. A process was established to program approximately $1 million in HSIP funding towards efforts that would have the highest reduction on severe crashes.

In 2005, two procedures were adopted as a means to select sites for RSAs.

Nomination Selection Process

The nomination selection process uses a "bottom up" approach. County engineers and officials qualitatively nominate sites based on their knowledge of high-crash locations. Geographic considerations, local control and cooperation, and available funding sources are several factors used to determine suitable locations. The following questions also help to identify locations that would likely have a high return on the investment of conducting an RSA and quickly implementing safety improvements:

• Are there few State road intersections in the corridor?
• Are there few municipalities involved, and are they responsive to safety projects?
• Is the corridor length compact enough to make for a good concentrated RSA?
• Are planning funds available for this project?

Crash Data Selection Process

The crash data selection process uses a "top down" approach. The process began with the SJTPO reviewing two years of crash data for every road in the region and identifying high crash-density locations as suitable sites. Crash data were reviewed based on crash density per mile, prior high-crash locations, and high-crash segments versus intersections. Of the three, crash density was given the most weight.

In 2006, the selection process became more refined. Fiscal year 2006 RSA corridors were selected using a balance of qualitative and quantitative elements. The SJTPO engaged local experts to nominate locations that could be improved quickly with inexpensive countermeasures. The South Jersey Traffic Safety Alliance and the SJTPO's Technical Advisory Committee solicited nominations from engineers, planners, and traffic officers who possessed first-hand knowledge of the local roadways, driving habits, and crash documentation. Once nominations were received, four factors were used as a filter:

• Geographic compactness of the corridor.
• High degree of local control.
• Good local cooperation.
• Potential for safety improvements.

The sites that met the above criteria were considered as top candidates and continued to SJTPO's quantitative analysis. A high-crash corridor was considered to experience at least 15 crashes per mile per year and a minimum of 100 total crashes. Injuries and fatalities were also considered. The top candidates from this step of the process were compared to all routes in the region on the basis of crash density. A sliding window technique was utilized to find the crash clusters. An overview flowchart of the process is illustrated in Figure 13.

Figure 13. Overview of the South Jersey Transportation Planning Organization's RSA Site Selection Process.

NJDOT is very supportive of the SJTPO RSA program, working to enable the authorization of HSIP funds and assisting with unobligated HSIP funding issues. The NJ Division of Transportation Safety has also devoted a considerable amount of time to partnering with the SJTPO to conduct RSAs. Over the past several years, this RSA program has been highlighted by FHWA. In 2005, the program received an Honorable Mention in the National Roadway Safety Awards:

"It is one of the first local programs of its kind, utilizing Federal planning funds to systemically identify local road segments of concern, organizing a team of independent specialists under the auspices of a metropolitan planning organization, engaging a consultant team for the audits and securing Federal funding for the resulting recommended improvement packages."

Additionally, in 2007, the SJTPO Director received the Dave Powell Excellence Award. This award is given to recognize partnerships with the FHWA New Jersey Division Office.

Transportation Safety Planning Results

In 2005, the results yielded four corridors from the nomination selection process and one corridor from the crash data selection process. The corridors ranged from 1.2 miles to 5.3 miles in length. In 2006, the results yielded five corridors from the qualitative and quantitative selection process. These corridors ranged from 1.9 miles to 4.3 miles in length.

In recent years, the SJTPO has shifted from using consultant support for the actual RSA to more project development tasks. The RSAs are now conducted by the localities as the SJTPO flexes regional Surface Transportation Program (STP) funds to metropolitan planning funds to conduct the RSA. In the near future, to ensure independent teams are being used, localities will begin a pilot effort to conduct peer expert exchanges in neighboring counties. For example, an established RSA team based in one county would conduct an independent safety assessment in a neighboring county in return for in-kind services.

Roadway Safety Scans

Since 2007, the SJTPO has also been using a "Road Safety Scan" (RSS) process. An RSS is a scaled-down version of an RSA and is more comparable to a traditional safety review. The scans allow experienced highway safety engineers to cover significantly more road mileage than a formal RSA. Local participation is also encouraged.

For an RSS, each section of study roadway is first driven at normal operating speeds and then at slower speeds to stop as necessary to take pictures and measurements. Particular items the engineers are looking for include traffic control device deficiencies, clear zone issues, sight distance issues, pavement drop-offs, guardrail end treatment deficiencies, etc. Voice-over video-recording devices are also used extensively for documentation and further office review.
There are three parts to the scan:

• Data collection – crash data printouts are reviewed by engineers to identify potential safety issues.
• Field review – conducted by the traffic engineer with local participation; each location is documented as accurately as possible.
• Preparation of the report and findings – documentation of findings and any recommended actions.

In 2007, a scan was completed for 31 sections (totaling 122 miles) in Cumberland, NJ and 13 sections (totaling 46 miles) in Atlantic, NJ. Each section had several safety issues identified, proposed remedial actions, an estimate of the level of effort required to implement the measures, and the potential safety benefit. A map with additional photographs and the reviewed crash data were also included to complete the scan.

< Previous

Table of Content

Next >