0 0 FACTORS AFFECTING SNOWPLOWABLE RAISED PAVEMENT MARKER FAILURES IN VIRGINIA Michael D. Fontaine, Ph.D., P.E. Senior Research Scientist Virginia Transportation Research Council 0 Edgemont Road Charlottesville, VA 0 Phone: --0 Fax: --0 E-mail: Michael.Fontaine@VDOT.Virginia.gov Brian K. Diefenderfer, Ph.D., P.E. Research Scientist Virginia Transportation Research Council 0 Edgemont Road Charlottesville, VA 0 Phone: -- Fax: --0 E-mail: Brian.Diefenderfer@VDOT.Virginia.gov Corresponding Author: Michael D. Fontaine Word count: Tables and Figures (0 words) + words = words TRB 0 Annual Meeting
Fontaine and Diefenderfer 0 0 ABSTRACT Snowplowable raised pavement markers (SRPMs) are commonly used in Virginia to supplement lane lines on the interstate and portions of the primary system. While the SRPMs improve lane delineation at night and during inclement weather, the SRPMs can become damaged or detached from the pavement after prolonged exposure to traffic and snowplows. Prior to 00, Virginia had no formal method or schedule for routinely inspecting these markers following initial installation. In April 00, a fragment of an SRPM metal casting became dislodged and injured a motorist. In response to this incident, a statewide study was initiated to determine if there were systemic problems with SRPMs becoming damaged or detached on Virginia s interstates. Casting condition, reflector condition, epoxy condition, and installation adequacy were inspected on -mile segments of interstate pavement containing SRPMs. The study found that nearly percent of all SRPM castings inspected were either missing or damaged. Approximately percent of reflectors inspected were either missing or damaged. An analysis of risk factors showed that total traffic since installation and initial installation adequacy were most strongly correlated with casting failures. These data were used to develop an SRPM casting inspection program and schedule that has been adopted by the Virginia Department of Transportation. The study also recommends increased training for installers and inspectors. TRB 0 Annual Meeting
Fontaine and Diefenderfer 0 0 0 0 INTRODUCTION Snowplowable raised pavement markers (SRPMs) are commonly used in Virginia to supplement lane lines on the interstate and portions of the primary system. The SRPM system consists of a reflective lens housed in a metal snow plow deflector, which is installed into a groove cut into the pavement surface and attached to the pavement with epoxy. The deflector is typically an H- shaped protective steel or cast-iron casting that consists of a pair of parallel keels and a connecting web. The open ends of the H are oriented parallel to the direction of traffic. The ends of the four legs of the H are to be placed just beneath the surface of the pavement. The casting is held in a groove cut into the pavement (by a series of stacked saw blades) with twopart epoxy and is rounded on top to allow a snowplow blade to ride up over the marker without damage. The entire assembly weighs approximately lbs. Past research has shown that SRPMs have significantly longer initial detection distances than many longitudinal markings (,, ). While SRPMs can improve lane delineation at night and during inclement weather, they can become damaged or detached from the pavement after prolonged exposure to traffic and snowplows if not maintained. Prior to 00, there was no formal method or schedule for routinely inspecting SRPMs in Virginia following initial installation. SRPMs were only inspected or replaced following resurfacing. In April 00, a fragment of an SRPM metal casting became dislodged and injured a motorist on I- in Northern Virginia. While the motorist was not seriously injured, this prompted the Virginia Department of Transportation (VDOT) to initiate a study of the section of roadway to determine if there were widespread problems at the site (). This inspection found that 0 to 0 percent of the SRPMs inspected near the incident site were either missing or damaged. The site inspected contained SRPMs installed following pavement resurfacing in under the same contract. Damage consisted primarily of fractured SRPM leading edges. It was noted that the leading edges of the damaged SRPMs were placed either at or slightly above the pavement surface, contrary to manufacturer s installation instructions. Inspection of an adjacent site (with SRPMs installed during pavement resurfacing in 00) showed that all of the SRPMs were in good condition. Thus, it was concluded that the high failure rate at the one location was likely due to improper installation. It was estimated that there are approximately,000 SRPMs in the Northern Virginia District of VDOT alone, and VDOT commonly uses SRPMs in conjunction with -inch wide reflective tape or thermoplastic on interstates and high traffic primary routes. Given the magnitude of SRPM use in the state, a greater understanding of factors impacting the condition of SRPM castings was needed in order to determine if a formal inspection and replacement program was required. As a result, a study was initiated to assess factors that may play a role in the failure of SRPM castings. PURPOSE AND SCOPE The purpose of this study was to investigate the in-place condition of SRPMs on a sample of interstate pavements in Virginia. The objectives of this study were: TRB 0 Annual Meeting
Fontaine and Diefenderfer 0 0 0 0. Determine practices in other states for the inspection and maintenance of SRPMs. Assess the physical condition of SRPM castings and reflectors on a sample of interstate pavements in Virginia. Determine which site characteristics are correlated with an increased likelihood of SRPM casting failures.. If a problem with SRPM condition exists, develop recommendations for an SRPM inspection program. The scope of this project was limited to a review of a sample of Virginia interstate segments that were at least years old and carried 00 traffic volumes of at least,000 vehicles per day. These thresholds were focused inspections on older roads with higher traffic volumes since they were thought to be more prone to casting failures. No primary roads were evaluated. Only the physical condition of the SRPM casting, reflector, and epoxy were assessed, and retroreflectivity and visibility were not measured. LITERATURE REVIEW There are two types of RPM in common use in the United States: non-snowplowable and snowplowable. Non-snowplowable RPMs consist of a rounded or square reflector housing that is epoxied to the surface of the pavement, and is most often used in southern and western states in areas where snowplow use is infrequent. Snowplowable RPMs have a metal housing that is designed to protect the reflector from snowplow hits, and are most often used in northern states. Snowplowable markers are often substantially more costly than non-snowplowable markers, with recent VDOT bid data showing a unit cost to install a SRPM of over $ each. This cost is sharply higher than recently cited Texas DOT bid costs for non-snowplowable markers of between $.00 and $.0 each (). This section will briefly review installation guidance for SRPMs, maintenance practices for SRPMs, and past studies on the visibility and retroreflectivity of SRPMs. State Practices Relating to the Installation and Maintenance of SRPMs The Manual on Uniform Traffic Control Devices (MUTCD) provides guidance on spacing between RPMs when used to supplement longitudinal pavement markings (). The MUTCD specifies an 0-ft spacing for SRPMs supplementing broken lane lines, assuming that the broken lane lines are 0 ft long with a 0-ft gap between lines (Sections B. and B.). Other spacing guidance on usage of SRPMs can be found in sections B. and B. in the MUTCD. State installation practices for RPMs were reviewed in 00 as part of National Cooperative Highway Research Program (NCHRP) Report (). That project surveyed states with known RPM installations. Of these states, installed SRPMs, and the remainder installed non-snowplowable RPMs. In NCHRP Report, RPM usage was classified as either non-selective (RPMs installed on all facilities of a certain functional classification) or selective TRB 0 Annual Meeting
Fontaine and Diefenderfer 0 0 0 0 (RPMs installed on a subset of roads based on their characteristics) (). Ohio, Texas, and California indicated that they installed RPMs on all state-maintained roads. Of the states that provided information on RPM usage criteria, all states installed RPMs non-selectively on the interstate system. Several other states also indicated that RPMs were installed on certain other facility types based on combinations of AADT, speed limit, and geometric considerations. NCHRP Report also summarized maintenance practices for RPMs (). The researchers reported that Pennsylvania and Ohio replace RPM reflectors on fixed - to -year cycles. Indiana defined SRPM lens replacement cycles as a function of the average daily traffic (ADT) on a road and the number of lanes present, with lenses on higher volume roads being replaced more frequently (). NCHRP Report also found that Colorado and Iowa had removed all RPMs and stopped future installations because of high maintenance costs. In 00, an electronic survey on SRPMs was conducted by the Missouri DOT (). Twenty U.S. states and two Canadian provinces responded to the survey. Of these, agencies reported using SRPMs, with of these respondents indicating that their use was experimental. When asked if there were any problems with the SRPMs coming loose from the pavement, of the agencies whose use of SRPMs was non-experimental responded that they were aware of a single or occasional occurrence. When asked if the cause of the failure was determined, hits from snowplow blades, pavement failures, or improper installation were the most often cited reasons. Three northern states (Alaska, Montana, and Colorado) stated that they do not use SRPMs because of heavy snowplow operations. New York DOT stated that they were using SRPMs less often, with wet-night reflective tape used as an alternative (). Durability of SRPMs Several studies have provided generalized estimates of the service life of SRPM castings and lenses (0,,). Estimates of casting service life ranged from years (0) to 0 years (), and lens service life was estimated at to years (0, ). Depending on the study, these estimates were developed based on field tests, opinions of knowledgeable practitioners, or models that predicted retroreflective performance. Several other studies have reviewed the durability of SRPMs on isolated test segments over a relatively short period of time. In 00, the North Dakota DOT conducted a limited durability study of SRPMs at two locations over a year period (). In 00, the Alberta DOT evaluated low-profile SRPMs from two manufacturers (). The researchers did not find any problems with the SRPM castings over the -year inspection period, but they did find numerous failures of the SRPM lenses, particularly on the shoulder installations. Overall average lens failure rates at the end of the -year period were. percent for centerlines and. percent for shoulder installations. A 00 study by the Vermont Agency of Transportation examined the durability of four types of SRPM made by three manufacturers over approximately. years (). Castings remained in generally good condition, but lens damage was much more prevalent. Between and percent of all SRPM lenses were missing or damaged at the end of 0 months. TRB 0 Annual Meeting
Fontaine and Diefenderfer 0 0 0 0 METHODOLOGY The literature review shows that while states have developed guidelines for inspection and maintenance of SRPMs, available durability data is often based on either expert opinions or relatively short-term evaluations. The methodology developed in this project was used to assess the condition of SRPMs under a variety of traffic, pavement, and environmental conditions and then define specific factors correlated with SRPM maintenance problems. This involved conducting a series of field inspections, and then analyzing that data to determine factors correlated with failures. An inspection cycle for SRPM castings was then developed to reduce the likelihood of castings coming loose and injuring motorists. Field Inspections A plan for inspecting the in-situ condition of SRPMs was developed to determine whether systemic problems in SRPM conditions exist. The following factors were used to select locations for inspection: Located on an interstate segment Directional annual average daily traffic (AADT) of at least,000 vehicles per day Homogeneous pavement surface of at least mile in length, with both rigid and flexible pavements considered Pavement surface between and 0 years old Sites were initially identified by using VDOT pavement condition and roadway inventory databases. The candidate site lists were developed to try to ensure that a range of pavement ages and types were inspected. Sites were also distributed geographically across the state to capture differences in snowplow operations, including differences in magnitude of snowfall, blade type, and salt application. The researchers initially developed a list of 0 -mile sites to be inspected. These sites were selected to ensure a good distribution of pavement types (concrete, dense graded hot-mix asphalt (HMA), and stone matrix asphalt (SMA)) and pavement ages ( to 0 years old, and more than ten years old). Sites represented a mixture of traffic volumes and snowfall amounts. Field personnel screened the initial list, and ultimately -mile segments of interstate were inspected. The distribution of sites by VDOT construction district, pavement type, and age of installation are shown in Table. Some of the initially identified sites were removed for a variety of reasons. In some cases, the identified section included bridges, tunnels, or active work zones where inspections would have been difficult to perform safely. In other cases, there were discrepancies between the actual pavement that was in place and what was recorded within HTRIS. The screening by the field personnel removed a large proportion of the SMA and concrete pavements that were originally identified for inspection. Although this was not ideal, the -mile sample still represented a wide range of traffic, pavement, and climactic conditions. TRB 0 Annual Meeting
Fontaine and Diefenderfer 0 0 0 Table. Summary of Sites Inspected. District -0 Years Old > 0 Years Old Dense Graded HMA SMA Concrete Dense Graded HMA SMA Concrete Bristol 0 0 0 0 Culpeper 0 0 0 0 Fredericksburg 0 0 0 0 Hampton Roads 0 0 0 0 0 Northern Virginia 0 0 Richmond 0 0 0 0 Salem 0 0 0 Staunton 0 0 Statewide 0 Total SRPMs Inspected 0 A series of lane closures were set up at each location so that inspection teams could view each SRPM close up. Inspection teams recorded the following information for each SRPM in the section inspected: Casting Condition: Castings were rated as either Good (no visible damage), Missing (SRPM no longer attached to pavement), or Damaged (casting cracked, chipped, or has pieces missing), Epoxy Condition: The SRPM epoxy was coded as either Epoxy well-mixed (uniform gray color) and intact or Epoxy discolored (black and white swirl) or missing. While cases where epoxy was well-mixed but damaged were seen in the field, they were not recorded separately. The categorization was intended to strictly define epoxy as either being totally acceptable or unacceptable in some way. Reflector Condition: Reflectors were rated as either Missing or Damaged/Cracked, or No visible damage. Installation Adequacy: Inspectors checked three aspects of the SRPM installation: Side tabs resting on pavement surface, Front/leading edge keel tips below or flush with pavement surface, and Rear keel tips below or flush with pavement surface. If all three of these were checked, the installation was adequate. Following the completion of inspections, the data were reviewed to identify any inconsistencies or errors in inspection coding. Cases where inconsistencies or errors were present were removed from further analysis to be conservative. This resulted in a removal of approximately. percent of SRPM casting inspections. Data Analysis A three step process was undertaken to analyze the field inspection data. First, high level summaries of the inspection results were produced. Next, potential factors that were correlated TRB 0 Annual Meeting
Fontaine and Diefenderfer 0 0 0 0 with the number of failures were examined. Finally, thresholds for inspecting the SRPM castings were developed. Summary of Inspection Results The first task in the analysis was to examine the screened inspection data to determine if there were any broad trends in the data. The goal of this analysis was to assess whether there were large problems with SRPM castings, reflectors, or installation methods at the sites inspected. These summaries did not attempt to determine the causes for any observed problems. Rather, they were only used to identify potential areas of concern. Identification of Factors Correlated with Casting and Lens Failures An odds ratio analysis was used to assess whether a particular factor increased the risk of an SRPM casting or reflector failure (). Odds ratio analysis is commonly used in medical studies to assess whether certain risk factors increase the chance of a person contracting a disease. Those studies have a number of structural similarities to this analysis, so it was selected as a relatively simple way to identify which factors are correlated with SRPM failures. The odds ratio is interpreted as The odds of an SRPM/reflector failing are (odds ratio) higher if (condition A) exists than if (condition B) exists. Cases where castings were missing or cracked were combined together as failures. Likewise, missing and damaged reflectors were combined together. Odds ratios are independent and cannot be added or multiplied together. The following factors were examined using the odds ratio technique: Age of the SRPM installation Total traffic that has traveled over the road segment since the SRPM was installed through the end of the analysis year Total snowfall (in inches) that has fallen on the road since the SRPM was installed through the end of the analysis year. This acted as a surrogate measure for the number of snowplow hits that an SRPM experienced. Mean International Roughness Index (IRI) for the road segment. This acted as a surrogate measure for the overall pavement condition. Condition of epoxy Installation adequacy The age of installation, total traffic, and pavement condition were determined from HTRIS. The condition of the epoxy and installation adequacy was determined from the inspection results. The total snowfall data was collected from the weather station closest to the segment being inspected. That data was obtained from the University of Virginia Climatology Office. In the TRB 0 Annual Meeting
Fontaine and Diefenderfer 0 0 0 0 case of continuous variables (e.g., age, IRI, snowfall, and total traffic), an iterative search was performed to define breakpoints where the odds ratio was maximized. Development of Casting Inspection Thresholds The odds ratio analysis focused on determining which factors are correlated with the failure of SRPMs. While this is informative, it does not provide guidance as to when a DOT should consider performing inspections of SRPM castings to detect failures before they become widespread. To address this need, regression models were developed that predicted the proportion of casting failures as a function of specific variables that were shown to be correlated with failures. The regression model building process focused on developing descriptive models where the data were readily available in existing databases. In other words, data that could only be obtained through detailed field inspection (such as installation adequacy) were excluded from this analysis. The SPSS statistical modeling software curve fitting algorithm was used to examine a variety of factors and model forms. SPSS was used to examine linear, quadratic, cubic, logarithmic, inverse, power, S, compound, growth, and exponential model forms for each variable examined in the odds ratio analysis (both separately and in combination). The model form and variable(s) with the highest adjusted coefficient of determination (R a ) was selected as the best fit to the data. The adjusted coefficient of determination (R a ) is a measure of the variability in the independent variable accounted for in the model, with higher values indicating a better model fit. The adjusted R accounts for the number of variables in the model along with the fit of the data, with simpler models given preference over more complex ones. This is in contrast to traditional R values which always increase as more terms are added to the model. RESULTS Summary of Inspection Results Table summarizes the aggregate results of the inspections, stratified by surface type. The data available shows that there appears to be a problem with the SRPM castings becoming damaged or dislodged. About percent of all SRPM assemblies inspected were missing entirely, and almost percent were cracked or damaged. This represents separate instances where the SRPM casting had the potential to produce a hazard. Figure shows several examples of different types of casting failures that were observed. The reflector condition inspection is also shown in Table. Again, the data available shows that a significant number of SRPM reflectors were cracked, damaged, or missing. Almost percent of the SRPMs inspected possessed either damaged or missing reflectors. Many manufacturers recommend that reflectors be inspected every two years. Given that these installations were all at least years old, it is not surprising that such high levels of failure were observed. Missing or damaged reflectors can obviously have a significant impact on nighttime visibility. TRB 0 Annual Meeting
Fontaine and Diefenderfer 0 0 A summary of the epoxy conditions of the SRPMs inspected is also shown in Table. The data shows that almost percent of SRPMs had epoxy that was either damaged or obviously not well mixed. The epoxy plays a critical role in adhering the SRPM casting to the pavement, so damaged or poorly mixed epoxy may be an important factor in determining why SRPMs are not adequately anchored. Also, it is reasonable to expect that epoxy damage could be the result of an SRPM becoming dislodged or damaged by something like a snowplow. As a result, it is difficult to ascertain whether epoxy condition is the result of the SRPM becoming damaged or the cause of it. The adequacy of installations was also assessed, and is summarized in Table. There also appears to be some problems with the installation of SRPMs, with about percent of SRPMs not being installed adequately in some way. The installation problems appeared to occur with different frequencies throughout the state. One urban district, in particular, reported many installation problems. Only. percent of SRPMs in that district were installed properly. In fact, it accounted for percent of all SRPMs inspected statewide that were installed incorrectly. This suggests that there are potential training or inspection problems that need to be addressed in specific regions. Table. Summary of Inspection Results, Stratified by Surface Type. SRPM Count Factor Rating Dense Graded Concrete HMA SMA Total (%) SRPM in good condition with no Casting damage (.%) Condition SRPM missing (.0%) SRPM cracked or damaged (.%) Reflector in good condition with no Reflector damage (.%) Condition Reflector missing (.%) Reflector cracked or damaged (.%) Epoxy Epoxy well-mixed and intact 0 0 (.%) Condition Epoxy discolored or missing 0 (.%) Side tabs on pavement surface 00 (.%) Leading edge at or below pavement Installation surface (.%) Adequacy Trailing edge at or below pavement surface (0.0%) Installation completely correct (.%) TRB 0 Annual Meeting
Fontaine and Diefenderfer 0 (a) (b) 0 0 (c) (d) Figure. Examples of missing or damaged SRPMs: (a) epoxy failure, (b) pavement failure, (c) damaged casting with damaged reflector, (d) damaged casting with missing reflector. Analysis of Factors Correlated with SRPM Failures The summary statistics presented in the preceding section show that there were some significant problems with SRPM casting and reflector failures. Likewise, it showed problems with epoxy condition and overall SRPM installation. The next major task was to try to identify specific factors that appeared to be correlated with increased risk of an SRPM casting or reflector failure. SRPM Casting Failure First, various factors that might be correlated with SRPM casting failing were examined. The odds ratios that were significant are shown in Table. Once again, the odds ratios are interpreted as The odds of an SRPM failing are (odds ratio) higher if (condition A) exists than if (condition B) exists. The odds ratios would be interpreted follows: TRB 0 Annual Meeting
Fontaine and Diefenderfer 0 0 0 For cases where two distinct qualitative conditions exist, condition A represents the first condition and condition B represents the second condition. An example would be that The odds of an SRPM failing are. times higher if the epoxy is damaged or cracked than if the epoxy is well mixed and in good condition. For cases where there is a single numerical breakpoint, condition A represents values above the breakpoint and condition B represents values below the breakpoint. An example would be The odds of an SRPM failing are. times higher on roads that have carried more than 0 million vehicles since the pavement was installed than on roads that have carried less than 0 million vehicles since the pavement was installed. The th % confidence interval is provided for the odds ratios as well. For a factor to be significant, the confidence interval should not contain.0. The confidence intervals serve to provide an indication of uncertainty due to variation in sample sizes in the different categories. Table. Odds Ratio Analysis for Casting Failures. Factor Breakpoint Odds Ratio th % Confidence Interval for Odds Ratio Epoxy Condition Epoxy damaged or not well mixed vs. well mixed.0 [.0, 0.] Total Traffic Volume > 0,000,000 vehicles since installation. [.,.] SRPM Installation Some element of SRPM not installed properly. [.,.] Accumulated Snowfall > inches of snowfall since installation. [.,.0] Pavement Condition Mean IRI > 0. [.,.00] Pavement Type Compared SMA to Dense Graded HMA SMA is higher. Concrete not. [.,.] analyzed due to limited data Pavement Age 0 years old or more.0 [.,.0] The factor that showed the strongest relationship to casting failures was epoxy condition. Damaged or poorly mixed epoxy increased the odds of an SRPM failing by a factor of.. While this finding seems to indicate that epoxy condition is an important factor in causing failures, it may simply reflect the result of an SRPM being damaged by passing traffic. Thus, the role of epoxy condition cannot be fully quantified at this time. The next factor that was significant was the total traffic volume from installation through the analysis year. Roads that have carried more than 0 million vehicles since installation were. times more likely to have failures than roads below this threshold. Poor SRPM installation was also shown to increase the likelihood of an SRPM failing. SRPMs that were not installed correctly were. times more likely to fail than those that were installed properly. Again, percent of improperly installed SRPMs were on high traffic volume roads in an urban district, so this finding may also be influenced by total traffic volume. Furthermore, the installation could not be assessed for SRPMs that were missing. It is possible that poor installations contributed to those SRPMs becoming detached from the pavement. TRB 0 Annual Meeting
Fontaine and Diefenderfer 0 0 0 Finally, total accumulated snowfall, mean IRI, and pavement age were correlated with a higher likelihood of casting failures. It should be noted, however, that snowfall and pavement age were also highly related with total traffic volume. Older pavements tended to have greater accumulated snowfall and greater total traffic volume. Since the confidence intervals show that that total traffic volume has an odds ratio that is larger by a statistically significant margin, it would appear to be a more important factor in failures. Mean IRI was not highly correlated with those other measures, but also did not appear to play as strong of a role in failures as total traffic volume. Pavement type also emerged as a significant factor, with SMA pavements being. times more likely to have a casting failure than dense graded HMA pavements. The pavement types were evenly distributed by volume, so it appears that there is a significant difference primarily attributable to pavement type. Only SMA sites were available for analysis, however, so this finding should be considered preliminary and worthy of future research. Reflector Failure While a SRPM casting failure could create a road hazard, a failure of the SRPM reflector represents a reduction in nighttime visibility of the markings. It should again be emphasized that only the physical failure of the reflector was examined, and loss of retroreflectivity was not specifically assessed. Reflector failures were also examined to assess which conditions may influence reflector failures, and the odds ratios that were significant are shown in the Table. Table. Odds Ratio Analysis for Reflector Condition. Factor Breakpoint Odds Ratio th % Confidence Interval for Odds Ratio Epoxy Condition Epoxy damaged or not well mixed vs. well mixed. [.,.] Pavement Age years old or more. [.,.] Total Traffic Volume > 0,000,000 vehicles since installation. [.0,.] Pavement Condition Mean IRI > 0. [.,.] Accumulated Snowfall > inches of snowfall since installation. [.,.0] SRPM Installation Some element of SRPM not installed properly. [.,.] Pavement Type Compared SMA to Dense Graded HMA SMA is higher. Concrete not analyzed due to limited data. [.,.] As Table shows, the reflectors that were inspected were much more prone to failure than SRPM castings. When the levels of failures were examined across different factor levels in Table, results were reasonably consistent for most factors. This was likely because only pavements that were years old or more were examined. If lenses had not been replaced within those five years, all of the lenses inspected would have been beyond their recommended service life. Some of the findings include: TRB 0 Annual Meeting
Fontaine and Diefenderfer 0 0 0 0 Epoxy and installation condition appeared to play a role in reflector failure, likely related to SRPM casting failure. In other words, when a casting failed, it likely damaged the reflector as well. The relationship between epoxy condition and installation adequacy to reflector failures were not as strong as for SRPM casting failures, however. The duration of exposure appears to play an important role in reflector failure, with the age and accumulated traffic volume since installation having odds ratios that were statistically the same. Pavement condition also appears to play a role in creating problems with reflector failures, likely due to overall casting failures. The type of pavement appears to play a role, with the odds of an SRPM reflector failing on an SMA pavement being almost. times greater than failure on a dense graded HMA pavement. Once again, this is finding is correlated with overall casting failures and is based on limited data. To develop more detailed results for the reflectors, it would be necessary to inspect newer pavements. Since only pavements that were at least years old were inspected, many locations already had significant numbers of reflectors that were damaged or missing. Discussion of Risk Factor Analysis A major challenge of the SRPM inspection data set is that accumulated traffic volume, accumulated snowfall, and pavement age are all showing up as significant factors, but these criteria tend to be correlated for a given site. For example, sections of I- in Northern Virginia and Richmond Districts generally had the worst SRPM casting and reflector performance. All of those sites tended to have high traffic volumes and older pavements. Generally speaking, several trends were observed. First, installation appears to play a significant role in casting failures. Both the condition of the epoxy and the proper installation of the SRPM are highly correlated with failures. The duration of exposure appears to play an important role in SRPM failure, with the total traffic since installation having a stronger relationship than pavement age or accumulated snowfall. Casting Failure Regression Model The previous analysis focused on determining which factors influence the overall number of failures of SRPMs. While this is informative, it does not provide guidance as to when inspections of SRPM castings should be performed to detect failures before they become widespread. To address this need, a regression model was developed that predicted the proportion of casting failures as a function of specific variables that were shown to have the strongest relationship to failures. Some of the factors that were shown to be highly significant in SRPM casting failures, like epoxy condition and installation adequacy, could not be detected without performing a detailed inspection in the first place. As a result, the regression model TRB 0 Annual Meeting
Fontaine and Diefenderfer 0 0 0 0 building focused on developing models where macroscopic level data were readily available in existing databases. The SPSS curve fitting algorithm was used to examine a variety of factors and model forms. The following factors were tested for the SRPM castings: Total traffic volume since installation Total accumulated snowfall since installation Age of installation Average segment IRI Maximum segment IRI The model form and variable(s) with the highest adjusted R was selected as the best fit to the data. The adjusted R accounts for the number of variables in the model along with the fit of the data, with simpler models given preference over more complex ones. A cubic model using the total accumulated traffic since installation was found to produce the best model of SRPM casting failures. Total snowfall was investigated as an independent variable, but it was not founds to perform as well as total traffic. This may have been due to several limitations in the snowfall data. For example, snowfall measurements may not be a strong surrogate for the number of snowplow passes since it is unclear weather a inch snowfall fell as one event or as a series of or discrete events. Weather stations also were not always located immediately adjacent to a roadway, and may not have fully captured conditions on that route. If data on actual snowplow passes were available, it may produce a stronger model. The model produced using total traffic was: F T. 0 T.0..0 T where, T = the total traffic that has traveled over the section of roadway since the SRPM was installed, and F = the percentage of SRPMs that were damaged or missing below the level of traffic, T. This model was created using data points from different sections of roadway inspected, and had an adjusted R of 0.. All factors were significant to at least 0.0. The model is valid only for the range of casting failures from approximately to percent. Although the odds ratio analysis showed different failure rates between SMA and dense graded HMA pavements, the small number of data points for SMA and concrete pavements made it impossible to produce models with a strong explanatory power for the different pavement types. As a result, this model applies to all pavement types. Figure plots the data points as well as the model. TRB 0 Annual Meeting
Fontaine and Diefenderfer 0 0 Total % RPMs Damaged or Missing per Mile 0 Data Model 0 00 00 00 00 Total Traffic Since Installation (Millions of Vehicles) Figure. Failure Model for SRPM Castings. This model can be used to define traffic thresholds where the SRPM castings should be subjected to a detailed inspected process, provided that an acceptable level of failures is defined. Realistically speaking, the level of effort required to ensure that 0 or SRPMs on a -mile section were damaged or missing would be difficult to maintain. Table shows the critical traffic volume thresholds that are analogous to,, and SRPMs failing within a -mile, twolane directional segment of interstate. An estimated elapsed time to achieve this amount of traffic is also provided given different current year annual average daily traffic (AADT). This time is estimated given current year traffic volumes and assumption of a percent annual growth rate in traffic. The actual sites inspected had AADTs that ranged from,000 to 0,000 vehicles per day, so estimates for current year volumes between 0,000 and 0,000 are provided. Table summarizes the results of this analysis. % of SRPMs Missing or Damaged a Table. Possible Inspection/Replacement Intervals for SRPM Castings. Traffic Threshold Estimated Years to Achieve Traffic Threshold, Given Current AADT (vehicles) 0,000 vpd 0,000 vpd 0,000 vpd 0,000 vpd.0%. million.% 0. million >.%. million > a Analogous to,, and SRPMs missing or damaged per mile of -lane directional interstate, respectively. TRB 0 Annual Meeting
Fontaine and Diefenderfer 0 0 0 0 If a fairly strict requirement of having no more than percent of SRPMs fail is adopted, then SRPMs would need to be thoroughly inspected between every and years, depending on traffic volume. These findings are consistent with the to 0 year casting service life cited in the literature. If higher levels of SRPM failures are deemed to be acceptable, then the time periods between replacement and inspections would increase. Any potential casting inspection would require a lane closure so that inspectors could examine the SRPM closely. The inspector would need to be able to see if the SRPM is still well-bonded to the pavement, and physically attempt to see if the casting can be dislodged. This would obviously require significant personnel time and create user delays, especially if done regularly on high volume roads. DISCUSSION AND CONCLUSIONS The results of this study produced several notable conclusions. Major findings of this project include: A significant portion of the SRPM castings inspected were damaged or dislodged. About percent of SRPMs were completely missing, while almost percent were damaged or cracked. These damaged and missing SRPMs could become potential hazards. Night time visibility may be significantly degraded on some of the sections that were inspected. The inspection results show that over percent of SRPM reflectors were missing and over percent were damaged in some way. Although retroreflectivity was not explicitly assessed, these physical inspection results show that prolonged use of SRPMs without regular lens replacement can significantly compromise the visibility benefits of using SRPMs. Many SRPMs were not installed properly, which appears to be a contributing factor in SRPM failures. About percent of SRPMs had epoxy that was not mixed properly or was damaged. About percent of SRPMs were not correctly installed. The odds ratio analysis showed that poor epoxy condition and improper installation were significantly correlated with increased chances of an SRPM failing at a percent confidence level. Total traffic volume was related to SRPM casting failures. The odds ratio analysis shows that both reflector and casting failures appear to be correlated to the total accumulated traffic volume since installation. Likewise, the age of the installation and the total accumulated snowfall were correlated with failures, although not as highly. Age of installation, total traffic, and total snowfall are all collinear, though, since they all increase over time. Pavement type and condition may play a role in SRPM failures, but these findings are still preliminary. SMA pavements appear to be more prone to casting and reflector failures than dense graded HMA pavements, but this is based on limited data. Given that this research found several problems with SRPM casting condition, it strongly indicates that DOTs need to pay careful attention to SRPM installation and maintenance practices. Since the quality of installation appears to be strongly correlated with casting and lens TRB 0 Annual Meeting
Fontaine and Diefenderfer 0 condition, inspectors and installers should be well-trained in SRPM installation techniques. Regular programs to inspect and replace castings and reflectors are also needed. A potential replacement timetable for castings was shown in Table ; however reflectors should be independently replaced according to manufacturer specifications or based on the results of visibility inspections to maintain adequate reflector performance. Since the completion of this research, the VDOT has embarked on a program to improve training to inspectors and contractors on SRPM installation. A statewide policy on the installation and regular maintenance of SRPMs has also been implemented to ensure that the reflectors and castings are maintained in acceptable condition. The effectiveness of these policies is still being assessed. REFERENCES. Gibbons, R.B., J.M. Hankey, I. Pashaj, B.H. Cottrell, and C. Anderson. Wet Night Visibility of Pavement Markings. VTRC 0-CR. Virginia Transportation Research Council, Charlottesville, 00.. Gibbons, R.B. Pavement Marking Visibility Requirements During Wet Conditions. VTRC 0-CR. Virginia Transportation Research Council, Charlottesville, 00.. Carlson, P., J. Miles, A. Pike, and E. Park. Evaluation of Wet Weather Pavement Markings: First Year Report. Report 0-00-. Texas Transportation Institute, College Station, 00.. Shiells, D. P. Report of Field Inspection of Snow Plowable Pavement Markers on I- in Prince William County. Virginia Department of Transportation, Chantilly, 00.. Carlson, P., J. Miles, A. Pike, and E. Park. Evaluation of Wet Weather and Contrast Pavement Marking Applications: Final Report. Report 0-00-. Texas Transportation Institute, College Station, 00.. Federal Highway Administration. Manual on Uniform Traffic Control Devices. Washington, D.C., 00.. Bahar, G., Mollett, C., Persuad, B., Lyon, C., Smiley, A., Smahel, T., and McGee, H. Safety Evaluation of Permanent Raised Pavement Markers. NCHRP Report. Transportation Research Board, Washington, D.C., 00.. American Association of State Highway and Transportation Officials, Research Advisory Committee. AASHTO RAC Email Survey Results: Snowplowable Raised Pavement Markers. Conducted by the Missouri Department of Transportation. January, 00. http://cms.transportation.org/?siteid=&pageid=0. Accessed January, 00.. New York State Department of Transportation. Special Specification.0XX- Preformed, Wet-Reflective Tape (Grooved Pavement Method). May, 00. https://www.nysdot.gov/portal/page/portal/main/business-center/consultants/forms- TRB 0 Annual Meeting
Fontaine and Diefenderfer publications-and-instructions/engineering-information-issuance-system/eirepository/ei000.pdf. Accessed January, 00. 0. Bryden, J.E. Long-Term Performance of Grooved Stripe-reflective Markers. New York State Department of Transportation, Albany,.. Cottrell, B.H. Evaluation of Pavement Markings for Improved Visibility during Wet Night Conditions. VTRC -R. Virginia Transportation Research Council, Charlottesville,.. Markow, M.J. Managing Selected Transportation Assets: Signals, Lighting, Signs, Pavement Markings, Culverts, and Sidewalks. NCHRP Synthesis. Transportation Research Board, Washington, D.C., 00.. Doerr, G.L., R. Walker, and S. Henrichs. Evaluation of Snowplowable Reflective Pavement Markers for Effective Delineation. North Dakota Department of Transportation, Bismark, 00.. Filice, J. Study of Snowplowable Raised Pavement Markers. Alberta Transportation. Edmonton, Alberta, Canada, 00.. Patterson, K., and J. Fitch. Evaluation and Comparison of Snowplowable Raised Pavement Markings (SRPMs). Vermont Agency of Transportation, Montpelier, 00.. Neter, J., M.H. Kutner, C.J. Nachtsheim, and W. Wasserman. Applied Linear Statistical Models, th Edition. McGraw-Hill, New York,. TRB 0 Annual Meeting