A Methodology for Estimating and Comparing the Lifecycles of High-Build and Conventional Waterborne Pavement Markings on Primary and Secondary Roads in South Carolina

Abstract

Highway pavement markings constitute a vital component in roadway transportation systems and provide crucial visual cues for drivers to follow the road. In 2008, the South Carolina Department of Transportation (SCDOT) initiated a study to evaluate pavement markings on noninterstate primary and secondary roads in South Carolina. This article discusses development of a method for estimating and comparing the lifecycles of high-build waterborne and conventional waterborne pavement markings based on retroreflectivity levels. The method was developed using nearly 3 years of field data from 66 sites throughout South Carolina. The article provides an overview of data collection and analysis methods employed in model development. The analysis indicates that high-build markings are predicted to considerably outlast waterborne markings and are also more cost-effective, based on cost per linear foot per year, for two-way Annual Average Daily Traffic (AADT) volume levels up to 2,000 vehicles per day.

Keywords

retroreflectivity pavement markings pavement paint high-build lifecycle

Introduction

Longitudinal pavement markings are essential traffic control devices. Markings are used on roads to delineate travel lanes, channelize traffic in opposing directions, and identify locations on two-lane roads where passing is allowed. Pavement markings have retroreflective properties to increase roadway safety during nighttime conditions. Because of this, it is important for Department of Transportations (DOTs) to provide and maintain pavement markings that remain within acceptable limits of retroreflectivity.

The Manual on Uniform Traffic Control Devices (MUTCD) for streets and highways does not currently stipulate minimum retroreflectivity values for pavement markings (Federal Highway Administration, 2009). The Federal Highway Administration (FHWA) has also not set federal minimum standards, though such standards have been recommended, however, currently remain pending (Debaillon, Carlson, He, Schnell, & Aktan, 2007; Katherine & Paul, 2008). In light of this, it will be very beneficial for South Carolina Department of Transportation (SCDOT) to have policies in place that predict pavement marking degradation and allow for cost-effective management of these markings to conform with forthcoming minimum retroreflectivity values.

In 1998, SCDOT engineers recognized that a formalized system for evaluating pavement markings could provide both safety and economic benefits. In 1999, Clemson University, with assistance from The Citadel, began research with the goal of developing a comprehensive system to quantitatively evaluate pavement marking retroreflectivity at regular time intervals for interstate highways in South Carolina. (Sarasua, Clarke, & Davis, 2002; Sarasua et. al, 2001; Thamizharasan, Sarasua, Clarke, & Davis, 2003). In 2008, Clemson University and The Citadel began a similar study for noninterstate primary and secondary state roadways in South Carolina.

Previous research indicates pavement marking retroreflectivity degradation is affected by a variety of factors, including traffic volumes and composition, weather/climate, quality control in applying marking materials, and type of pavement surface. Prior to these studies, SCDOT pavement marking replacement procedures were not performance based, and many markings were being replaced before the end of their functional life. Similarly, other markings were being allowed to degrade beyond effective limits, posing potential safety concerns.

Furthermore, various marking materials are known to have different lifecycles. A primary research objective is to develop predictive models to estimate the rate of marking retroreflectivity degradation, for use in determining replacement strategies. This article explores the development and comparison of these models for high-build and conventional waterborne markings. The models for high-build markings were developed using approximately 2 years of field data from 15 sites in the upstate of South Carolina, whereas the waterborne models were developed using nearly 3 years of field data from 51 representative sample sites throughout South Carolina. It should be noted that this article focuses only on white edge line markings because of the fact that South Carolina currently limits use of high-build markings to white edge lines. Data used in the analysis were collected during a 34-month period from May 2008 to March 2011 for waterborne markings, and a 25-month period from May 2009 to June 2011 for high-build markings.

Background

The MUTCD (Federal Highway Administration, 2009) defines retroreflectivity as “a property of a surface that allows a large portion of light coming from a point source to be returned directly back to a point near its origin.” More specifically, retroreflectivity values provide an analytic measure of how well a driver can see a particular pavement marking. The retroreflective effect of pavement markings is made possible through the use of small glass beads that are dropped on markings during application of material in liquid form. Measured pavement marking retroreflectivity values depend on several factors, such as bead size, bead type, quantity of beads, angle of bead embedment, and application method, among others.

Conventional waterborne and high-build pavement markings are both latex paints. Waterborne is typically placed at a 15 mil thickness whereas SCDOT specifies a 25 mil thickness for high-build markings (South Carolina Department of Transportation, 2008). Besides marking thickness, primary differences between waterborne and high-build markings are curing rate and glass bead size. According to SCDOT specifications (South Carolina Department of Transportation, 2007), bead types range in size from smallest to largest as Type I to Type IV, respectively. High-build marking specifications (South Carolina Department of Transportation, 2008) require an initial application of the larger Type III or IV beads, followed by an application of Type I beads, whereas waterborne specifications require Type I beads only. Retroreflectivity values of the markings are also dependent on bead factors such as application rate and depth at which beads are embedded into the marking. Glass bead application rates specified by SCDOT are 6 pounds of Type I beads per gallon for waterborne and 6 to 7 pounds of Type III or IV beads per gallon and a second application of 4 to 6 pounds of Type I beads per gallon for high-build. Retroreflectivity values degrade over time as these beads become dislodged from the marking or are worn down. Weather, traffic, snowplowing, and other adverse roadway conditions can also cause degradation of marking retroreflectivity.

Retroreflectivity Measurement and Minimum Threshold Values

The most common measure of pavement marking retroreflectivity is the coefficient of retroreflected luminance (R_L). ASTM defines R_L as the ratio of luminance in the direction of observation to normal illuminance, at the surface on a plane normal to incident light, expressed in millicandelas per square meter per lux (mcd/m²/lux) in the standard E 808-01 (re-approved 2009)—Standard Practice for Describing Retroreflection (ASTM Standard E-808-01). The current accepted standard for measurement of retroreflectivity of pavement marking materials using a portable retroreflectometer is ASTM E 1710-05 (ASTM Standard E-1710-05).

Numerous studies have been conducted in an attempt to develop minimum retroreflectivity values. A 2000 Minnesota Department of Transportation (MnDOT) study evaluated acceptable minimum threshold values for retroreflectivity (Loetterle, Beck, & Carlson, 2000). Drivers rated nighttime visibility of edge lines and centerlines along state and county roads. Results identified a threshold retroreflectivity level between 80 and 120 mcd/m²/lux. As a result, MnDOT uses this minimum threshold value for its pavement marking management program. Parker and Meja (2003) performed a study in New Jersey to determine driver visibility of markings on a 32-mile circuit. The study concluded a minimum acceptable level of retroreflectivity appeared to be between 80 and 130 mcd/m²/lux for drivers under 55 years of age and between 120 and 165 mcd/m²/lux for drivers older than 55.

During the summer of 2007, the FHWA held two conferences with the primary goal of finalizing the wording and content of new minimum pavement marking and traffic sign retroreflectivity levels. The new traffic sign minimum levels were put into effect as of January 2008, although pavement marking minimums are still pending (Katherine & Paul, 2008).

An additional report by Debaillon, et al. in October 2007 did recommend minimum values for retroreflectivity to the FHWA. This research took into account pavement type, vehicle type, presence of retroreflective raised pavement markers, marking configuration, and speed. The recommended minimum values ranged from 40 to 90 mcd/m²/lux for fully marked roadways without the presences of Retroreflective Raised Pavement Markers [RRPMs] (centerline, lanelines, and/or edge line) depending on speed.

Although there are currently no minimum threshold standards for marking retroreflectivity, proposed standards have been created and implementation is expected in the near future. In April 2010, a Notice of Proposed Amendments was published in the Federal Register, proposing to revise the 2009 MUTCD by adding Standards, Guidance, Options, and Support information regarding maintaining minimum retroreflectivity of longitudinal pavement markings. The proposed revisions would establish a uniform minimum level of nighttime pavement marking performance based on visibility needs of nighttime drivers, to promote safety, enhance traffic operations, and facilitate comfort and convenience for all drivers, including older drivers. The proposed standard is shown in Table 1 (Federal Highway Administration Proposed Revision, 2011).

Table 1.

MUTCD Proposed Minimum Maintained Retroreflectivity Levels for Longitudinal Pavement Markings (Federal Highway Administration Proposed Revision, 2011).

	Posted speed (mph)
	≤30	35 to 50	≥55
Two-lane roads with center line markings only^a	n/a	100	250
All other roads^a	n/a	50	100

Exceptions: The exceptions are given by FHWA as supplemental information to the table.

A. When RRPMs supplement or substitute for a longitudinal line, minimum pavement marking retroreflectivity levels are not applicable as long as the RRPMs are maintained so that at least 3 are visible from any position along that line during nighttime conditions.

B. When continuous roadway lighting assures that the markings are visible, minimum pavement marking retroreflectivity levels are not applicable.

Predictive Models of Retroreflectivity Degradation

Principal considerations in creation of a predictive model for pavement marking performance include model form and selection of explanatory variables. Although several research efforts have focused on pavement marking performance, only a handful have attempted to develop predictive models. A 1998 study by Perrin, Martin, and Hansen on Utah highways compared three pavement marking materials; paint, epoxy, and tape. Their research evaluated useful life through an investigation of relationships between retroreflectivity, material age, AADT, and pavement type. They found each variable to be significant, and determined a general hyperbolic relationship exists between independent and dependent variables.

Sasidharan, Karwa, & Donnell, (2009) developed pavement marking degradation models using data collected on 88 roadway segments over a 1-year period from May 2007 through May 2008 by the Pennsylvania Department of Transportation. They found pavement marking retroreflectivity decreases as the age of both epoxy and waterborne paint pavement markings increase. White pavement markings were shown to have higher estimated service lives than yellow pavement markings. Traffic exposure was found to be negatively correlated with pavement marking retroreflectivity in the waterborne paint analysis, but traffic exposure was not statistically significant in the epoxy pavement marking retroreflectivity model (Sasidharan et. al., 2009).

A 2009 study by Rasdorf, Hummer, Zhang, & Sitzabee, (2009a), developed models to predict life cycles for waterborne and thermoplastic markings. The independent variables validated by the models included time, initial R_L reading, AADT, color, and lateral location. Their research concluded that AADT had a small but significant affect on thermoplastic marking degradation. The study also conducted a correlation study between pavement marking retroreflectivity and glass bead density, which determined that higher bead densities resulted in higher retroreflectivity values throughout pavement marking life.

Cumulative number of Traffic Passages (CTP) is a model variable used to represent the cumulative exposure of vehicles since marking installation as the product of AADT and time, measured in millions of vehicle passages per lane. A 2001 study by Migletz, Graham, Harwood, Bauer, & Sterner, covering 19 states evaluated pavement marking durability for a variety of marking materials. CTP was used as the primary variable to quantify a relationship between coefficient of retroreflectivity (R_L) and CTP. Using minimum threshold values between 85 to 150 mcd/m²/lux for white lines, service life for white waterborne markings on freeways was determined to range from 4.1 to 18.4 months. A 2003 study by Lindly and Wijesundera found that CTP had a better correlation with retroreflectivity than marking age alone.

Thamizharasan et. al., (2003) presented research findings on durability of interstate pavement markings and determined when markings are first applied, retroreflectivity values initially increase until glass beads become exposed, after which retroreflectivity decreases linearly over time. The study also found traffic volume was not statistically significant for retroreflectivity degradation along straight alignment sections of road.

Several additional factors are relevant to a thorough analysis of pavement marking performance including directionality of pavement markings (yellow markings; Rasdorf, Zhang, & Hummer, 2009b), performance of wet markings (Gibbons, Anderson, & Hankey, 2005; Schnell & Aktan, 2004), roadway geometry (Tsyganov, Machemehl, Warrenchuk, & Wang, 2006), shoulder properties (Van Driel, Davidse, & van Marseveen, 2004), climatic conditions, and quality control of marking placement.

The authors were not able to identify any available literature related to performance of high-build pavement markings including modeling degradation.

Cost Variability for Various Marking Types

Cost is always a major factor in determining which type of marking to install on a roadway. Over the years, many cost-benefit analyses have been performed for pavement markings. These analyses depend on many different factors including estimates for marking cost per linear foot, marking lifespan, traffic volume, pavement type, and so forth. A 2000 report by Montebello and Shroeder gave estimated costs and marking lifespans for various marking types. For this article, approximate marking prices were obtained from SCDOT personnel who indicated that, in general, marking prices have nearly doubled since 2000 (Boozer, 2011). Table 2 shows a comparison of various marking types including approximate cost and estimated lifespan, assuming that the markings are applied to new pavement.

Table 2.

Estimated Pavement Marking Costs and Lifespans.

Marking type	Estimated cost per linear foot^a (US$)	Estimated lifespan [from Boozer (2011)]
Waterborne	0.06 to 0.09	9 to 36 months
High-Build	0.12 to 0.18	3 years
Epoxy	0.25 to 0.31	4 years
Thermoplastic	0.25 to 0.30	3 to 6 years
Tape	3.00 to 5.00	4 to 8 years

Costs are given in 2011 dollars

Data Collection

Data collection sites were established along representative roads within the primary and secondary road network throughout South Carolina and geographically distributed to reflect conditions across the state. Initially, 44 white edge line waterborne pavement marking sites were established during the summer of 2008. One year later, seven waterborne sites were added to provide better coverage of coastal regions, providing a total of 51 sites. In addition, 15 high-build sites in the upstate of South Carolina were added in an attempt to research and model the performance of high-build markings. Over time, quite a few sites were abandoned because of repaving, remarking, or chip seal application; however, the data for these sites up until the time of abandonment was still included in the analysis. All of the high-build sites were a single brand with similar installation specifications by 2 different contractors. The brands for waterborne markings varied by site. These installations were done either in-house or by a contractor. Recycled glass beads were used in all sites.

Data collection sites were established along tangent roadway sections with consideration to a variety of safety related criteria such as sight distance and shoulder width. Creation of edge line sites consisted of five specific data collection sample points, measured at 25-foot increments, and identified using template markings equivalent in size to the base of the two Delta LTL-X retroreflectometers used in this research. This method of establishing data collection sites allowed readings to be obtained at precisely the same location for subsequent rounds. Data collection sites were established and initial values of retroreflectivity recorded within 30 days of pavement marking application. Global Positioning System (GPS) coordinates were recorded for use in mapping and locating sites. Follow-on retroreflectivity readings were collected approximately four times a year over a 3-year performance period.

Data Editing and Management Prior to Preliminary Analysis

Field collected data were entered into a comprehensive database and checked for logical consistency. Anomalous readings typically attributed to tire marks, scraping, excess moisture, physical abrasion and ground in dirt, debris, and so forth, were identified and removed from the analysis. To best preserve sample size, only individual anomalous sample points (of the five points taken for each site) were removed from the analysis rather than entire site. Furthermore, in processing specific sample points, retroreflectivity values greater than twice the standard deviation of the five site readings were considered anomalous and were omitted in the determination of a mean site value in any given round of data. Median values of measured retroreflectivity were determined along with average values for every data collection site. The average difference observed between means and medians was 4 mcd/m²/lux for high-build markings and 3 mcd/m²/lux for waterborne markings. These negligible differences provide a reliable indication that site collected data were not skewed. For use in the pavement marking degradation and retroreflectivity analysis, median values for each site were used because these values are less sensitive to outliers.

A graph of high-build marking performance is shown in Figure 1, where sites with AADT less than 500 are displayed with a solid line, AADTs from 500 to 1000 a dashed line, and AADTs greater than 1000 a dotted line. The graph shows that the initial median retroreflectivity for high-build markings range between 300 to 500 use mcd/m2/lux with relatively small changes in retroreflectivity levels more than 2 years after placement.

Figure 1.

High-build white edge line marking performance.

Analysis and Modeling of Pavement Marking Degradation

This section focuses on developing pavement marking degradation models for waterborne and high-build white edge markings using field-collected data on primary and secondary roads in South Carolina.

Use of Absolute Differences and Percentage Differences in Retroreflectivity Modeling

Retroreflectivity values measured at different sites with similar characteristics over a similar time period were observed to vary considerably. One reason for this range in values is that initial readings of newly placed pavement markings were rarely consistent. Although specific values of site measured retroreflectivity varied noticeably during any particular data collection round, the theoretical rate of degradation is expected to be similar across all sites for a given marking material. As an alternative to producing models that incorporate initial values as a constant, the researchers decided to produce models that predict actual degradation rather than predict retroreflectivity. Applying these models would require subtracting the modeled difference in degradation from an initial value to determine a predicted retroreflectivity value after a period of time or total amount of traffic passages.

Furthermore, many sites were observed to have similar values of percent difference in retroreflectivity, whereas the magnitude of the initial measured value varied greatly across data collection sites. For example, a retroreflectivity difference of 10 mcd/m²/lux would constitute a substantial change for a site with an initial retroreflectivity value of 100 mcd/m2/lux and conversely, the same change would be essentially irrelevant for a site with an initial value of 600 mcd/m²/lux. The use of percentage differences from the initial measured value over time would successfully account for this discrepancy. As a result, pavement marking degradation models were analyzed and developed using both percent and absolute difference as dependent variables for the benefit of comparative purposes and to provide insight in variable sensitivity.

Pavement marking age was taken as number of days since the marking was applied to the pavement. For any given site, retroreflectivity values measured in the first collection round were taken as the initial value. Differences between this initial value and values measured during subsequent collection rounds were calculated for every site. Absolute differences in measured values were multiplied by 100 and divided by initial values to obtain percent difference.

Based on the literature review, experience from previous similar studies, and discussion with SCDOT officials, a final list of independent analysis variables was identified that included: initial retroreflectivity, days after application, AADT, CTP, temperature, humidity, lane width, and shoulder width. These variables were used in stepwise regression analysis for white edge lines for both high-build and waterborne markings. In model formulation, CTP was not used in the same model as days after application or AADT because CTP is highly correlated with both of these variables.

Stepwise Regression Analyses of Significant Variables

Stepwise regression analysis was used to determine which variables were significant in predicting measured values of pavement marking retroreflectivity. For waterborne white edge markings (N = 261), days after application, traffic volume, lane width, and shoulder width were significant for the percent difference model, although only days after application was significant for the absolute difference model. However, both the difference and percent difference analyses found both traffic volume and days when used together and CTP when used alone to be significant for explaining the variation of the high-build white edge line model (N = 111). Temperature and humidity were not determined to be significant in the modeling of either marking type for South Carolina.

Further Examination of Variables for Analyses

Predictive model variables analytically identified through stepwise regression as being meaningful in estimating pavement marking retroreflective degradation were further evaluated from a practicality standpoint and specifically from an agency application perspective. Results from regression analysis consider a variable mathematically useful if inclusion contributes significantly to explaining model variance. However, in reality, additional variables needed to be carefully examined to determine the overarching usefulness to model results. For agency application, a variable was deemed useful if its contribution to the model outweighed reasonable additional costs to accurately collect the data for all primary and secondary roads.

Through discussions with SCDOT and a review of the availability and completeness of their road characteristics database, “days after application” is the easiest to use in a model, followed by “lane width,” “AADT/CTP,” and “shoulder width.” CTP represents the cumulative exposure of the marking to vehicle travel since its application. Through rational examination of expected model results and data collection concerns, variables for inclusion in final degradation models were determined for white edge high-build and waterborne pavement markings.

The data trends for retroreflectivity values for absolute differences are shown in Figure 2. Figure 2a shows the trend of absolute difference versus time for waterborne markings (R² = 0.24). The trend of absolute difference versus time for high-build markings shown in Figure 2b indicates that retroreflectivity does not change much with time as indicated by the very flat slope. The R² value is low (0.03) because models that are near horizontal typically have low R² values because of the magnitude of unexplained error compared to explained error. The R² increases to 0.15 when ADT is included as a variable with time. The best absolute difference model for high-build as reflected in the R² value 0.29 uses CTP as the lone independent variable as shown in Figure 2c.

Figure 2.

Descriptive graphs of absolute differences.

Figure 3 shows the data trends for retroreflectivity values for percent differences. Figure 3a shows the trend for conventional waterborne (R² = 0.30). A slightly improved R² (0.34) can be achieved by using volume, lane width, and shoulder width along with time, however, the added data requirements do not justify the improved model. Figures 2b and 2c show high-build model graphs. The R² value of 0.33 when CTP is used as the predictor for retroreflectivity is a vast improvement over using time (R² = 0.03). Figure 2c indicates that there is roughly a 30% reduction in retroreflectivity after 2 million traffic passages (total of both directions).

Figure 3.

Descriptive graphs of percent differences.

Final Retroreflectivity Degradation Models

Table 3 includes the final pavement marking degradation models for white edge waterborne and high-build markings along with their corresponding R² values. A single variable model using time for waterborne white edge lines is included for both the difference and percent difference for application simplification. An additional percent difference model that includes 4 significant variables is included for comparison. Note that only the CTP models for difference and percent difference are included in the high-build models because of the model performance. It is worth noting that R² improved slightly for both marking types using percent difference.

Table 3.

Final Degradation Models.

Material	Variables used	Notes	Model	R ²
White edge waterborne	Days (Difference and percent difference models)	For agency application	Difference = −0.1447 (D)	0.24
			% Difference = −0.0554 (D)	0.30
	Days, lane width, traffic volume, shoulder width (Percent difference model)	For research comparison	% Difference = −94.0653 - 0.0490 (D) - 0.0021(T) + 9.8371 (L) - 2.6999 (S)	0.34
White edge high-build	CTP (Difference and percent difference models)	For agency application	Difference = −51.2835(C)	0.29
			% Difference = −14.2987(C)	0.33
Identification for model variables:
D—days after pavement marking application. Units are in days.		L—Width of travel lane. Units are in feet.
T—Two-way traffic volume using roadway. Units are in annual average daily traffic (AADT).		S—Width of adjacent shoulder. Units are in feet.
C—Cumulative traffic passages. Units are in million vehicles.

Retroreflectivity Degradation Model Performance

Model performance was evaluated by comparing model predicted values with field data from high-build and waterborne white edge sites. The evaluation indicated the magnitude of the residuals is pretty consistent as time progresses, which is a desirable characteristic from a modeling standpoint. Performance of the models as based on the mean absolute percent error (MAPE) and through indication of the percentage of measured values that would fall within identified error ranges is shown in Table 4. The difference degradation models performed better than the percent difference models for both the high-build and conventional waterborne models. The MAPE value is 16.7% for the waterborne difference model but increases to over 42.2% for the waterborne percent difference model. The MAPE values for both high-build models is less than 5% indicating that the average difference between actual and predicted retroreflectivity values for these models is very small. The column to the left of the MAPE column shows the percentage of sites with less than ± 20% error, which is equal to the sum of the first two error columns. Generally speaking, the difference degradation models developed from this research predict retroreflectivity values within a 20% error for 69% of the measured pavement marking values for waterborne white edge lines and approximately 93% of the measured values for high-build white edge line markings.

Table 4.

Overall Model Performance and Overprediction of Retroreflectivity.

Model Performance—proportion of predicted values with specified error margin
Material	Model	<±10% Error	±10% to 20% Error	>±20% Error	<±20% Error	MAPE (%)
Waterborne white edge	Difference	55%	14%	31%	69%	16.7%
(N = 261)	% Difference	46%	16%	37%	62%	42.2%
High-build white edge	Difference	74%	19%	7%	93%	1.3%
(N = 111)	% Difference	71%	21%	9%	91%	4.2%

Model performance—proportion of values over predicted specified margin.
Material	Model	<10% Over	10% to 20% Over	>20% Over		<20% Over

Waterborne white edge	Difference	76%	5%	19%		81%
(N = 261)	% Difference	54%	13%	33%		67%
High-build white edge	Difference	83%	10%	7%		93%
(N = 111)	% Difference	76%	15%	9%		91%

Model error can result from either underpredicting or overpredicting actual measured values of retroreflectivity. Underpredicted values could lead to premature pavement marking replacement, but is not a safety issue. However, overpredicted values are safety issues in that pavement markings could exhibit low levels of retroreflectivity before the model identifies the need for replacement using threshold minimums. Taking these factors into consideration, the lower portion of Table 4 provides a tabulation of the percentage of sites that were classified as overpredicted in various error ranges as determined from the models created for waterborne and high-build white edge pavement markings evaluated in this research.

Note that the columns labeled “<10% Over” and “<20% Over” include all under predicted values. On further examination, difference models were more accurate predictors of retroreflectivity, and in both cases difference models produced a higher percentage of sites predicted at less than 20% over actual retroreflectivity values. This observation serves to support the assumption that all similar type pavement markings deteriorate at the same rate, regardless of the initial retroreflectivity value. Based on this analytical insight, difference models are recommended as the most suitable means for predicting retroreflectivity degradation of pavement markings.

As indicated in Table 4, there is a likelihood that the degradation models will overpredict actual retroreflectivity in some cases. To account for this concern, a margin of safety should be considered to decrease the chance of this occurring, particularly as pavement markings begin to reach minimum threshold values of retroreflectivity.

In this section, the service lives of waterborne and high-build pavement markings are estimated from a retroreflectivity standpoint using the models developed in the research. For comparison purposes, the average initial values of all sites determined for both waterborne and high-build markings were used. Based on literature, a minimum threshold of 100 mcd/m²/lux was used as the lowest acceptable retroreflectivity value. Estimates of marking life for high-build and waterborne markings were calculated using the difference and percent difference models. The results are shown in Table 5. Note that this is only one example of predicted lifecycles using initial values based on the average of field data. The actual future average lifespans may actually be longer in South Carolina in the future because SCDOT is currently considering instituting performance based contracts for pavement markings that may include threshold values for minimum initial and retroreflectivity values.

Table 5.

Prediction of White Edge Pavement Marking Life.

Material	Model	R ²	Average initial value	Estimated marking lives
Waterborne (N = 261)	Difference = −0.1447 (D)	0.24	307	1431 Days	3.91 Years
Waterborne (N = 261)	% Difference = −0.0554(D)	0.34		1217 Days	3.33 Years
High-build (N = 111)	Difference = −51.2835(C)	0.29	404	5.92 CTP
High-build (N = 111)	% Difference = −14.2987(C)	0.33		5.26 CTP
Identification for model variables:
D—Days after pavement marking application. Units are in days; C—Cumulative traffic passages. Units are in million vehicles

Because retroreflectivity for waterborne markings is dependent on marking age only and volume only for high-build markings, a sensitivity analysis was conducted using the models to compare the lifespans of the two marking types while varying time and volume. It was determined that the average two-way AADT for the waterborne sites in this analysis was around 2000 veh/day, so AADTs up to 2000 were used for the high-build calculations. The analysis was conducted using the average initial values from Table 5. Retroreflecitivity values were then calculated at one year increments until they were less than or equal to the proposed minimum threshold value of 100 mcd/m2/lux. The results are shown in Table 6. Once the estimated life spans were determined, the installation costs per linear foot were divided by the number of years in the lifespan to calculate normalized costs for comparative purposes.

Table 6.

Model Predicted White Edge Pavement Marking Retroreflectivity Lifespans.

		Time (Years)
Marking type		1	2	3	4	5	6	7	8	9	17	Estimated cost (US$/LF)	Cost (US$/LF/year)
WB		254	201	149	96							0.09	0.023
HB	ADT
	1000	385	367	348	329	310	292	273	254	236	86	0.18	0.011
	2000	367	329	292	254	217	179	142	105	67		0.18	0.020

Note: WB - waterborne; HB - high-build; ADT - average daily traffic; LF - linear foot.

As shown in Table 6, high-build markings are predicted to outlast waterborne markings and are also more cost-effective for the two common rural ADT volume levels. Although not replacing a high-build pavement marking for 17 years (as shown in the Table 6 for 1000 ADT) may be unrealistic, it shows the potential performance of high-build as compared to waterborne markings when only retroreflectivity is considered. Because the waterborne and high-build models created were based on 34 and 25 months of data collection, respectively, it is recommended that these models be used with caution for time periods greater than these specified periods after marking placement. It should also be noted that the volumes at all of the high-build sites range between 200 and 3,500 veh/day. For further research purposes, it would be beneficial to conduct this analysis on high-build markings with higher volumes and compare the results. Additional data collection is necessary to verify these models for the remainder of pavement marking life.

Conclusions

The objective of this research was to develop, compare, and evaluate degradation models for high-build and conventional waterborne pavement markings to determine how often to replace pavement markings on primary and secondary roads in South Carolina. Conclusions from this research are summarized as follows:

As shown in Table 3, for waterborne pavement markings, number of days after application is the most significant variable in the retroreflectivity degradation model. Traffic volume was found to be marginally important in waterborne white edge lines, however such a slight model improvement does not justify adding more data and increasing the effort required to apply the model on a statewide basis. The single variable model indicated that the waterborne markings analyzed in this study degraded at 53 mcd/m²/lux annually. This degradation rate is about 30% lower than the rate determined by Rasdorf, Hummer, Zhang, & Sitzabee, 2009a for waterborne markings.

As summarized in Table 3 for high-build pavement markings, CTP is the most significant variable in the retroreflectivity degradation model. CTP was found to be more significant than marking age alone, as well as volume alone. Implementing a model that relies on CTP may pose some problems on a statewide basis, however with relatively accurate volume estimations, a reliable model can be created and used to predict high-build pavement marking performance.

As depicted in Figure 3, and estimated in corresponding degradation models, high-build white edge line markings are predicted to last considerably longer than waterborne markings for comparable locations. Although both marking types may have similar initial retroreflectivity values, predictive models indicate that high-build edge line markings degrade at a much lower rate than waterborne edge lines. The performance of high-build is most likely because of the larger glass beads and the ability of the thicker high-build paint to hold the beads over time.

As summarized in Table 2, even though high-build marking costs are approximately double that of waterborne markings, observed durability and lifespan of high-build markings appear more desirable based on retroreflectivity degradation comparisons.

As tabulated in Table 4, an evaluation of model performance indicated that the likelihood of pavement marking degradation models to produce overpredicted retroreflectivity values, as compared to actual measured values, was roughly 30% of the time for waterborne markings, and less than 10% for high-build markings. This concern could be effectively addressed through use of margin of safety factors.

As summarized in Table 5, model estimated pavement-marking life was determined to be 3.33 years for waterborne markings and 5.26 million vehicles for high build markings, using percent difference retroreflectivy models and 3.91 years and 5.92 million vehicles using absolute difference retroreflectivity models. This assumes a minimum threshold value of 100 mcd/m²/lux. Rasdorf et al., 2009a determined an average lifespan of 34.2 months for waterborne markings. Montebello and Shroeder estimated that the lifespan ranged from 9 to 36 months for conventional waterborne markings and 36 months, on average, for high-build markings.

As shown in Table 6, high-build markings are predicted to considerably outlast waterborne markings and are also more cost-effective, based on cost per linear foot per year, for two-way ADT volume levels up to 2,000 vehicles per day.

Additional research should focus on the long-term durability of high-build markings as well as testing on higher volume roads. It should also be noted that South Carolina does not experience the weather extremes of many of the northern states. Snow plowing is a somewhat rare event. Past studies have shown that snowplow activities can significantly degrade pavement marking retroreflectivity. The performance of high-build waterborne markings should be further analyzed for the effect of extreme weather.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The authors gratefully acknowledge the role of the South Carolina Department of Transportation in funding this project and providing valuable technical guidance.

Author Biographies

Joseph Robertson is a Transportation Engineering Analyst at Kimley-Horn and Associates, Inc. in Tallahassee, FL. He received his BS degree in civil engineering from The Citadel in 2010 and his MS degree in civil engineering from Clemson University in 2011. His work includes a variety of transportation projects ranging from traffic impact studies to interstate master plans.

Wayne Sarasua is an associate professor in Civil Engineering at Clemson University. His specialties are in planning and design of transportation systems, transportation asset management, and spatial data analysis. His courses include geomatics, transportation engineering, traffic engineering, and travel demand forecasting.

Joshua Johnson is a project engineer at Stantec Consulting in North Charleston, SC. He received his MS degree in civil engineering from Clemson University in 2010. His work includes a variety of transportation projects ranging from safety studies to signal timing.

William Davis is a professor in Civil & Environmental Engineering at The Citadel in Charleston, SC. His research interests focus on transportation infrastructure planning and design, highway safety, and active living by design. He teaches courses in transportation engineering, geographic information systems, concrete and asphalt design, and engineering economy. He has served on numerous technical committees for ASCE, TRB, ITE and ASEE.

References

ASTM Standard E—1710-05. (2009). Standard test method for measurement of retroreflective pavement marking materials with CEN prescribed geometry using a portable retroreflectometer. American Society for Testing and Materials International, West Conshohocken, PA, USA.

ASTM Standard E—808-0, Re-approved. (2009). Standard for describing retroreflection. American Society for Testing and Materials International, West Conshohocken, PA, USA.

Boozer

(2011). Unpublished data, South Carolina Department of Transportation.

Debaillon

Carlson

Schnell

Aktan

(2007). Updates to research on recommended minimum levels for pavement marking retroreflectivity to meet driver night visibility needs, (Report FHWA-HRT-07-059). Federal Highway Administration, Washington, DC.

Federal Highway Administration. (2003). Manual on uniform traffic control devices (Appendix A1 Section 406 (a), A1-1). Washington, DC.

Federal Highway Administration. (2009). Manual on uniform traffic control devices: For streets and highways. Washington, DC.

Federal Highway Administration. (2009). Proposed pavement marking retroreflectivity, Revision 1, Manual on uniform traffic control devices, maintaining minimum retroreflectivity of longitudinal pavement markings. Retrieved from http://mutcd.fhwa.dot.gov/knowledge/proposed09mutcdrev1/propmretromutcdxtxt.pdf

Gibbons

R. B.

Anderson

Hankey

(2005). Wet night visibility of pavement markings: A static experiment. Transportation Research Record: Journal of Transportation Research Board, 1911, 113-122.

Katherine

W. F.

Paul

J. C.

(2008). Pavement marking retroreflectivity workshops summary report, (Report FHWA-SA-08-003). Federal Highway Administration, Washington, DC.

10.

Lindly

Wijesundera

(2003). Evaluation of profiled pavement markings, (Report Number 01465). University Transportation Center for Alabama, Univeristy of Alabama, Tuscaloosa, AL.

11.

Loetterle

F. E.

Beck

R. A.

Carlson

(2000). Public perception of pavement marking brightness. Transportation Research Record: Journal of Transportation Research Board, 1715, 51-59.

12.

Migletz

Graham

Harwood

Bauer

Sterner

(2001). Service life of durable pavement markings. Transportation Research Record: Journal of Transportation Research Board, 1749, 13-21.

13.

Migletz

Graham

J. R.

(2002). Long term pavement marking practices. National Cooperative Highway Research Program: A Synthesis of Highway Practice, 306, pp. 13-27.

14.

Montebello

Shroeder

(2000). Cost of pavement marking materials. (Report 2000-11). Minnesota Local Road Research Board, Minnesota Department of Transportation, St. Paul, MN.

15.

Parker

N. A.

Meja

J. S. M.

(2003). Evaluation of the performance of permanent pavement markings. Transportation Research Record: Journal of Transportation Research Board, 1824, 123-132.

16.

Perrin

Martin

Hansen

(1998). A Comparative Analysis of Pavement Marking Materials, Transportation Research Board Annual Conference, Washington DC.

17.

Rasdorf

W. J.

Hummer

J. E.

Zhang

Sitzabee

W. E.

(2009a). Pavement marking performance analysis. project: 2008-05 final report. North Carolina Department of Transportation, Research and Development Group. Raleigh, NC.

18.

Rasdorf

W. J.

Zhang

Hummer

J. E.

(2009b), The Impact of directionality on paint pavement marking retroreflectivity. Public Works Management & Policy, 13, 3, 265-277.

19.

Sarasua

W. A.

Clark

D. B.

Collins

Tropea

Pfaller

Davis

J. W.

(2001) Measurement and evaluation of pavement marking retroreflectivity: South Carolina experience, Proceedings of the 80th Annual Transportation Research Board Meeting, Washington, DC.

20.

Sarasua

W. A.

Clarke

D. B.

Davis

W. J.

(2002). Methods and considerations for evaluating the retroreflectivity of pavement markings, Final report, South Carolina Department of Transportation, Columbia, SC.

21.

Sasidharan

Karwa

Donnell

E. T.

(2009) Use of pavement marking degradation models to develop a pavement marking management system, Public Works Management & Policy, 14, 2, 148-173.

22.

Schnell

Aktan

(2004). Performance evaluation of pavement markings under dry, wet, and rainy conditions in the field. Transportation Research Record: Journal of Transportation Research Board, 1877, 38-49.

23.

South Carolina Department of Transportation. (2007). Standard specifications for highway construction. (Section 626, 408,416). Columbia, SC: Author.

24.

South Carolina Department of Transportation. (2008). Permanent pavement markings fast dry, high build, high durability waterborne paint. (Section 628). Columbia, SC: Author.

25.

Thamizharasan

Sarasua

Clarke

D. B.

Davis

W. J.

(2003) A methodology for estimating the lifecycle of interstate highway pavement marking retroreflectivity, Proceedings of the 82nd Annual Transportation Research Board Meeting, Washington, DC.

26.

Tsyganov

A. R.

Machemehl

R. B.

Warrenchuk

N. M.

Wang

(2006). Before-after comparison of edgeline effects on rural two-lane highway. (Report FHWA/TX-07 /0-5090-2). Federal Highway Administration, Washington, DC.

27.

Van Driel

C. J. G.

Davidse

R. J.

van Marseveen

M. F. A. M.

(2004). The effects of an edgeline on speed and lateral position: A meta-analysis. Accident Analysis & Prevention, 36, 671-682.