Skip to contentUnited States Department of Transportation - Federal Highway Administration FHWA Home
Research Home   |   Pavements Home
REPORT
This report is an archived publication and may contain dated technical, contact, and link information
Publication Number: FHWA-HRT-10-066
Date: October 2011

 

Impact of Design Features on Pavement Response and Performance in Rehabilitated Flexible and Rigid Pavements

Chapter 5. Rehabilitated Flexible Pavement Analysis and Findings

Introduction

This chapter describes the analysis of rehabilitation alternatives for flexible pavements. The LTPP SPS-5 experiment was the main source of information for this study. The objective of the SPS-5 experiment, "Rehabilitation of Asphalt Concrete Pavements," was to help develop improved methodologies and strategies for the rehabilitation of flexible pavements. Specifically, the experiment evaluated common rehabilitation techniques implemented in the United States and Canada, and it evaluated the effects of climate, structural condition, and material variations on performance of rehabilitated flexible pavements. The design factorial is documented in Rehabilitation of Asphalt Concrete Pavements-Initial Evaluation of the SPS-5 Experiment, and included the following evaluation parameters:(4)

Variation of surface preparation alternatives, overlay material, and overlay thickness led to eight combinations at each SPS-5 project site (see table 32). In addition, one section was assigned as the control and did not receive any overlay, except for routine maintenance, creating nine experimental sections at each SPS-5 project site. As a result, each individual SPS-5 project provided a means for directly comparing rehabilitated HMA pavement performance using different surface preparation intensity, overlay thickness, and type of overlay mixture.

The initial SPS-5 sampling matrix was supposed to include only one subgrade type (fine-grained soil) with a minimum annual traffic of over 85,000 ESALs. Other factors considered in the sampling matrix included structural and functional condition of the pavement before overlay and climate. A total of 18 SPS-5 projects were constructed between 1989 and 1998. Table 33 presents the site location of each experiment according to their experimental design classification and as-built data. As shown, there were at least two projects for each condition except for the wet no-freeze fair condition cell and the dry freeze poor condition cell. A total of 162 test sections were built as part of the core SPS-5 experiment.

Table 32. Core sections of the SPS-5 experiment.

SHRP ID

Overlay Type

0501

Control: No treatment

0502

Thin overlay (1.99 inches (51 mm)): Recycled HMA mix

0503

Thick overlay (4.95 inches (127 mm)): Recycled HMA mix

0504

Thick overlay: Virgin mix

0505

Thin overlay: Virgin mix

0506

Thin overlay: Virgin mix with milling

0507

Thick overlay: Virgin mix with milling

0508

Thick overlay: Recycled mix with milling

0509

Thin overlay: Recycled mix with milling

Table 33. Constructed SPS-5 sites for the experimental factorial.

Pavement Condition

Soil Classification

Climate, Moisture Temperature

Wet Freeze

Wet
No-freeze

Dry Freeze

Dry
No-freeze

Fair

Coarse/fine

Georgia

 

Colorado

 

Coarse

New Jersey

 

Alberta, Canada

New Mexico

Montana

Fine

   

Minnesota

Oklahoma

Texas

Poor

Coarse/fine

   

Manitoba, Canada

California

Coarse

Maine

Florida

 

Arizona

Alabama

Fine

Maryland

Mississippi

   

Missouri

Note: Blank cells indicate data are not available.

One major deviation from the original SPS-5 experimental plan was the subgrade soil type. Originally, the subgrade soils for all SPS-5 projects were supposed to be fine-grained soils; however, only five of the SPS-5 projects actually had fine-grained soils. Four SPS-5 projects had soils that varied between fine- and coarse-grained soils. The subgrade soils for the remaining eight SPS-5 projects were classified as coarse-grained soils.

Additionally, one major deviation from the original SPS-5 sampling matrix was subgrade soil type. Originally, the subgrade soils for all SPS-5 projects were supposed to be fine-grained soils. However, only six of the SPS-5 projects actually had fine-grained soils. Four SPS-5 projects had soils that varied between fine- and coarse-grained soils, while the remaining eight projects were classified as coarse-grained soils. Another deviation from the experimental plan for only a few of the SPS-5 projects was that no control section was left in place. For example, section 0501 for the Colorado project included the placement of a thin overlay during rehabilitation.

Data Analyses

The impact of design features and site conditions on performance and response can be evaluated by looking at the trends in the survey data over time. Statistical tests can be used to verify if there are differences in these trends and if they can be associated with any of the design features in the experiment. Moreover, it is important to establish if any information on performance or response is reproduced in other sites or if they are associated with a particular site characteristic (e.g., climate, traffic, etc.). The best approach to achieve this objective is to consider every site and section available to statistically compare performance and response.

The SPS-5 experimental designed was balanced between design features intended for investigation. With few exceptions, out of 9 sections in each one of the 18 sites, there was 1 control section and 8 sections that combined equally thin and thick overlays with virgin and reclaimed asphalt pavement (RAP) mixes and milling and not milling prior to overlay. This provided an opportunity for a gradual statistical analysis in which information was gained by sequentially analyzing the data from each site first and complemented with a consolidated analysis. The consolidated analysis involved evaluating all sites and sections simultaneously in search of general trends and conclusions about pavement performance and its dependency on design features and site conditions. Figure 13 illustrates the statistical analysis process.

The flowchart consists of a thick gray right-facing upward arrow that shows a progression. There are four silver points on the arrow with labels in bold and bullets below. The first point is labeled Specific Pavement Study (SPS)-5 data." There are two items below: availability and quality control. The second point is labeled Individual sites" with one item below: Analysis of Variance (ANOVA). The third point is labeled Consolidated analysis" with four items below: weighted performance, with short-term and long-term as second-level bullets, and Friedman ANOVA. The final point is at the top right of the arrow and is labeled "Summary of findings."

Figure 13. Illustration. Statistical analysis flow chart.

Statistical Approach and Tests

The approach consisted of analyzing each site individually to initially check its construction history to address possible problems during that phase. ANOVA was then used for repeated measures. ANOVA can be used to explain if different trends in performance exist because of a particular choice of design feature (e.g., milling versus no milling). The repeated measures are the surveys conducted throughout the duration of the experiment. Each section that is part of one site has the same site conditions and traffic volume. The surveys were performed within a short period for sections, which made ANOVA with repeated measures the best option for the statistical analysis of individual sites.

The consolidated analysis was performed using the Friedman test. This is a nonparametric test (distribution-free) used to compare repeated observations on similar subjects. Unlike the more common parametric repeated measures such as ANOVA or a paired t-test, the Friedman test makes no assumptions about the distribution of the data (e.g., normality), and it can be used for multiple comparisons. The Friedman test uses the ranks of the data rather than their raw values to calculate the statistic. The test statistic for the Friedman test is a chi-square with n - 1 degrees of freedom, where n is the number of repeated measures (i.e., the number of sections in each site of the experiment). Statistical significance was defined at 95 percent (p ≤ 0.05 for the chi-square test).

The Friedman test also permits the evaluation of paired statistical significance between two rehabilitation strategies. In some instances, the result of one analysis may indicate that significant differences exist between the rankings of sections (i.e., the performances of these sections are statistically different). However, there might be groups within the sorted ranking with similar performance. The paired statistical analysis feature is important to identify groups of strategies with equivalent performance.

Performance Measures

The analysis of individual sites used distress and response data collected throughout the experiment duration. The data were checked for consistency and reasonableness prior to use in the analysis. Roughness, rutting, and fatigue, longitudinal, and transverse cracking were selected as indicative of performance and maximum deflection as indicative of response.

As described in the previous chapter, the consolidated analysis used WD as a performance measure of various distresses and the pavement response. It is calculated using the equation in figure 14.

WD equals the summation of parenthesis D subscript i plus D subscript i plus 1 end parenthesis multiplied P subscript i plus 1 divided by two divided by the summation of P subscript i plus 1. The numerator has values for i ranging from zero through n minus 1. The denominator has values for i ranging from zero through n.

Figure 14. Equation. Weighted distress.

The weighted average, in reality, represents the total normalized area (per year) under the distress versus time curve. As such, it is a measure of pavement performance relative to the specific distress over the entire monitoring period. The normalization to total time the section was in service provides means for comparing survey periods that may be different, which allowed the comparison of performance for both short term and long term.

The WD parameter is related to pavement performance over the whole analysis period. This concept is comparable to performance originally defined as the area under the serviceability curve.(2) The effect of variability from measurements by different surveyors is reduced when using this procedure and provided a parameter that could be used to compare sections with different survey periods. The WD performance measure proved a viable alternative to access the performance of different sections with individual in situ conditions and in-service ages. WD values were computed for the short term to evaluate performance in less than or equal to 5 years after rehabilitation was executed. Additionally, WD values were computed for the long term to compare performance for a period greater than 5 years.

Effect of Design, Construction Features, and Site Conditions on Performance of Rehabilitated Flexible Pavements

The first task was to analyze the impact of design features on performance for each site separately. The objective of this initial step was to identify trends in performance that could be associated with each design feature in the experiment (thin versus thick overlay, RAP versus virgin mix, and milling versus no milling). ANOVA with repeated measures was the statistical test used for this task. This statistical analysis took advantage of the fact that all sections at each
SPS-5 site had the same underlying pavement structure, traffic, and climate. The distress surveys and profile measurements were taken on the same day in all sections within each site. The survey dates were used as the repeated measures for the ANOVA.

The following section provides a complete example of the ANOVA tests used for the individual evaluation of SPS-5 sites. Each site in the experiment was evaluated following this sequential approach.

Example of Repeated Measures of ANOVA for Evaluation of SPS-5 Sites

Site Description and Data Availability

This site was assigned to LTPP in 1987, marking the initial data collection for the LTPP database. It is located at I-8 eastbound 17 mi (27.37 km) west of the I-10/I-8 interchange in Pinal County, AZ. The subgrade is a silty gravel soil with sand. The base course is a 14-inch (355-mm) soil aggregate mixture. The surface course is 5 inches (127 mm) of HMA. Traffic data are available since 1994, and AADTT for the LTPP lane is approximately 800 trucks. The average growth rate during the period is 8.5 percent. FHWA class 9 trucks account for 80 percent of the total truck traffic.

Data were available for all nine sections of this LTPP site. After being extracted from the LTPP database, performance data were evaluated for completeness and reasonableness. If any outliers were identified, they were removed from the analysis and recorded for reporting. In some instances, there were surveys performed on different days to cover all test sections in one specific date. When this happened, the missing survey was complemented with an interpolation of distresses measured during the previous and next survey dates. The reason for this was the need for equal numbers of surveys required to run ANOVA analysis with repeated measures.

All eight sections were rehabilitated in April 1990 following the selected SPS-5 experimental design. Section 0501 was used as the control section and was left without any rehabilitation. Data collection started after the rehabilitation and continued for 16 years until 2006. Data collection for the control section ended in 1993 probably as a result of the need to rehabilitate this section.

Analysis Approach

The analysis was performed for each type of distress previously described as well as for the maximum deflection. ANOVA with repeated measures was indicated in this case because the evaluation was carried out through a series of performance measurements taken throughout the duration of the experiment. The number of measurements in each section must be equal, and the experiment factorial must be balanced. The core sections of the SPS-5 experiment were balanced within each one of the three design features (overlay thickness, mix type, and surface preparation) and were independent. Without the control section, the eight remaining core sections provided an equal combination of all options in the design features. Therefore, the control section was not used for the statistical analysis of individual sites.

Repeated measures ANOVA is a technique used to test the equality of means. The null hypothesis has no differences between population means. The F-test in ANOVA evaluates the significance of the differences between means. A large F-value yields a correspondingly small p-value. The p-value is the observed significance level, or probability, of a type I error (alpha), which shows that the difference between population means exists when in fact there is no difference. In this study, an acceptable level of significance and probability of a type I error was defined as 0.05. The null hypothesis was rejected if the p-value was

Analysis of Performance

Roughness was measured by the IRI. Figure 15 presents an example showing IRI values for all sections in the site. The vertical dotted line indicates the year the sections were rehabilitated, as the measured roughness drops from the first to the second measurement. When the data were tested with the repeated measures ANOVA, they were grouped by different design features in the experiment.

This graph shows the roughness from a Specific Pavement Study (SPS)-5 site in Arizona. The x-axis shows the survey date in month/year, and the y-axis shows roughness values in inches per mile. There are nine trends shown in the figure corresponding to various sections (0501, 0502, 0503, 0504, 0505, 0506, 0507, 0508, and 0509). The curve for section 0501 is represented by a dark grey line and grey triangles for data points and shows a slight increase in roughness from about 78 inches/mi (1.23 m/km) to a little under 86 inches/mi (1.36 m/km) over a period of 3 years. The curve for section 0502 is represented by a light grey solid line and light grey squares for data points and starts at a roughness value of about 128 inches/mi (2.02 m/km), decreases to 86.3 inches/mi (1.36 m/km) in 7 months, and increases to a maximum of 232 inches/mi (3.66 m/km) in 16 years. The curve for section 0503 is represented by a dark grey solid line and white triangles for data points and starts at a roughness value of about 107 inches/mi (1.69 m/km), decreases to 60 inches/mi (0.95 m/km) in 7 months, and increases to a maximum of 121 inches/mi (1.91 m/km) in 16 years. The curve for section 0504 is represented by a dark grey solid line and crosses for data points and starts at a roughness value of about 98 inches/mi (1.55 m/km), decreases to 76 inches/mi (1.20 m/km) in 7 months, and increases to a maximum of 95 inches/mi (1.50 m/km) over 16 years. The curve for section 0505 is represented by a thick dark grey solid line and crosses for data points and starts at a roughness value of about 162 inches/mi (2.56 m/km), decreases to 81 inches/mi (1.28 m/km) in 7 months, and increases to a maximum of 120 inches/mi (1.89 m/km) over 16 years. The curve for section 0506 is represented by a light grey solid line and black circles for data points and starts at a roughness value of about 109 inches/mi (1.72 m/km), decreases to 65 inches/mi (1.03 m/km) in 7 months, and increases to a maximum of 97 inches/mi (1.53 m/km) over 16 years. The curve for section 0507 is represented by a light grey solid line and a small perpendicular line for data points and starts at a roughness value of about 116 inches/mi (1.83 m/km), decreases to 82 inches/mi (1.29 m/km) in 7 months, and increases to a maximum of 91 inches/mi (1.44 m/km) over 16 years. The curve for section 0508 is represented by a black solid line and starts at a roughness value of about 98 inches/mi (1.55 m/km), decreases to 60 inches/mi (0.95 m/km) in 7 months, and increases to a maximum of 81 inches/mi (1.28 m/km) over 16 years. The curve for section 0509 is represented by a grey solid line and black triangles for data points and starts at a roughness value of about 150 inches/mi (2.37 m/km), decreases to 65 inches/mi (1.03 m/km) in 7 months, and increases to a maximum of 233 inches/mi (3.68 m/km) over 16 years. The Rehab curve is represented by a dotted line is parallel to the y-axis and corresponds to a survey date of February 1990. It has a minimum roughness of zero inches/mi (zero m/km) and a maximum roughness of 233 inches/mi (3.68 m/km).

1 inch = 25.4 mm
1 mi = 1.61 km

Figure 15. Graph. Roughness from an SPS-5 site in Arizona.

Figure 16 shows IRI over time for both thin and thick overlay sections. The marks are the average IRI for all sections grouped by overlay thickness. The bars represent the range of values with one standard deviation from the average. The p-value in the top of the chart is zero and indicates that there is a significant statistical difference in IRI values over time between sections with thin and thick overlays. The plot also indicates that sections with thick overlay performed better than those with thin overlays, as expected.

Statistically significant differences were also found when comparing sections overlaid with virgin versus recycled mixtures. Figure 17 shows IRI values over time for both mixture types, and the p-value was close to zero. The plot indicates that sections overlaid with virgin mix had better performance than those with recycled asphalt mix. It is interesting to note that soon after rehabilitation, IRI average values were not different between the two types of sections. The great significant difference in performance was more evident later in the pavements' service life.

Surface preparation prior to overlay did not have an impact on roughness performance.
Figure 18 shows IRI values over time for sections that were milled versus not milled prior to receiving the overlay. The p-value was close to 1.0 and indicated that both distributions were statistically similar.

The x-axis shows the survey number, and the y-axis shows International Roughness Index (IRI) values in inches per mile for a Specific Pavement Study (SPS)-5 site in Arizona. The curve for thin overlay is represented by a continuous line and circles for data points, while the curve for thick overlay is represented by dashed lines with black shaded squares for data points. Each data point has bars that represent one standard deviation from the average. The IRI curve for thin overlay starts at 75 inches/mi (1.18 m/km) for survey 1 and constantly increases to about 170 inches/mi (2.68 m/km) for survey 13. The IRI curve for thick overlay starts at 70 inches/mi (1.10 m/km) and increases to about 100 inches/mi (1.58 m/km) at survey 13. The top of the graph displays the following information: SURVEY* overlay; LS Means, Current effect: F (12, 48) = 10.975, p = 0.00000. The plot indicates that sections with thick overlays performed better than the ones with thin overlays.

1 inch = 25.4 mm
1 mi = 1.61 km

Figure 16. Graph. IRI versus overlay thickness (distribution) for an SPS-5 site in Arizona.

The x-axis shows the survey number, and the y-axis shows International Roughness Index (IRI) values in inches per mile for a Specific Pavement Study (SPS)-5 site in Arizona. The curve for reclaimed asphalt pavement (RAP) mix type is represented by a continuous line and circles for data points, while the curve for virgin mix type is represented by dashed lines with black shaded squares for data points. Each data point has bars that represent 1 standard deviation from the average. The IRI curve for RAP mix type starts at 70 inches/mi (1.10 m/km) for survey 1 and constantly increases to about 165 inches/mi (2.60 m/km) for survey 13. The IRI curve for thick overlay starts at about 75 inches/mi (1.18 m/km), remains constant, and then increases to a little over 100 inches/mi (1.58 m/km) at survey 13. The top of the graph displays the following information: SURVEY* mix; LS Means, Current effect: F (12, 48) = 13.512, p = 0.00000.

1 inch = 25.4 mm
1 mi = 1.61 km

Figure 17. Graph. IRI versus mix type (distribution) for an SPS-5 site in Arizona.

 

This trend plot shows International Roughness Index (IRI) values versus surveys for sections that were milled versus not milled prior to receiving the overlay for a Specific Pavement Study (SPS)-5 site in Arizona. The x-axis shows the survey number, and the y-axis shows IRI values in inches per mile. The curve for no milling is represented by a continuous line and circles for data points, while the curve for sections that were milled is represented by dashed lines with black shaded squares for data points. Each data point has bars that represent 1 standard deviation from the average. The IRI curve for sections that were not milled starts at 75 inches/mi (1.18 m/km) for survey 1 and constantly increases to about 140 inches/mi (2.21 m/km) at survey 13. The IRI curve for thick overlay starts at about 70 inches/mi (1.10 m/km) and increases constantly to about 125 inches/mi (1.97 m/km) at survey 13. The top of the graph displays the following information: SURVEY* milling; LS Means, Current effect: F (12, 48) = 0.03921, p = 1.0000. The p-value indicates that both distributions are statistically similar.

1 inch = 25.4 mm
1 mi = 1.61 km

Figure 18. Graph. IRI versus milling (distribution) for an SPS-5 site in Arizona.

The analysis of pavement performance associated with roughness described previously was repeated for rutting, fatigue, longitudinal cracking, and transverse cracking. The results are summarized in table 34. If a statistically significant difference in performance existed, it is marked with a "Y," and the corresponding design feature with the best performance was indicated. When no difference was justified statistically, it is marked with an "N" in the table.
In this case, a qualitative assessment was made of which design feature provided the best performance. Blank cells indicate that no design feature had predominant performance. For example, the first line (roughness) in table 34 indicates statistical differences in performance when evaluating the effect of overlay thickness. In this case, for both short-term and long-term performances, the performance for thick overlays (5-inch (127-mm)) was found to be better than thin overlays (2-inch (51-mm)).

Table 34. Summary of statistical analysis results for SPS-5 site in Arizona.

Distress

Overlay Thickness

Mix Type

Milling

Statistical Difference

Best Performance

Statistical Difference

Best Performance

Statistical Difference

Best Performance

Short

Long

Short

Long

Short

Long

Roughness

Y
(p = 0)

Thick

Thick

Y
(p = 0)

 

Virgin

N

Mill

Mill

Rutting

Y
(p = 0.004)

Thin

Thick

Y
(p = 0.0004)

Virgin

Virgin

N

No mill

 

Fatigue

N

Thick

Thick

N

   

Y
(p = 0.005)

Mill

Mill

Transverse

N

   

N

Virgin

Virgin

N

   

Longitudinal

N

 

Thin

Y
(p = 0.01)

Virgin

Virgin

N

   

Note: Blank cells indicate that no

Summary of Findings

The repeated measures ANOVA results for this SPS-5 site suggest the following:

Conclusions from the Individual Analyses of SPS-5 Sites

The example presented in the previous section illustrates the approach taken to individually evaluate each site in the SPS-5 experiment. The summary tables for each site containing the results of the statistical analysis on performance for roughness, rutting, fatigue cracking, longitudinal cracking, and transverse cracking are presented in appendix A of this report.

The individual analysis of each site provided good qualitative information about the impact of design features on the performance of rehabilitated flexible pavement sections. The differences in performance are best visualized when compiled in plots that explore each design feature investigated in the SPS-5 experiment. These plots and key findings from the individual analyses of SPS-5 sites are presented in this section.

The compilation of results was created by identifying the number of sites in which various design features provided different performance throughout the site's service life. For example, if sections with thick overlays performed better than thin overlaid sections in one site, thick overlay was marked for the site. The process was repeated for all sites and distresses. The plots presented in this section summarize the percentage of sites in which differences in performance were identified. The number of sites with statistically justified differences from the ANOVA repeated measures test was also noted.

Results for roughness are provided in figure 19, suggesting that thick overlays provided the best performance in more than 50 percent of the sites. No differences in short-term performance were found in the majority of sites when comparing the two mix types. When long-term performance was evaluated, virgin mixes were found to perform better than recycled mixes, but no differences were found in 44 percent of sites. Milling improved performance in the majority of sites; however, one-third of sites had no differences in performance between milled and nonmilled sections.

The results in figure 19 were obtained from statistical analysis and engineering judgment when statistically significant differences were not found (p-values higher than 0.05) but differences in trends were clearly visible. The choice to consider both statistically-based and engineering judgment-based results was justified by the expansion of data analysis the two approaches provided combined. In the analysis of thickness, five sites (28 percent of all SPS-5 sites) had statistically significant differences in roughness performance. Four sites (22 percent) had statistically significant differences in the analysis of mix type, and three sites (17 percent) had statistically significant differences in the analysis of milling.

This graph shows a summary of Specific Pavement Study (SPS)-5 sites by best performing design feature for roughness according to repeated measures analysis of variance (ANOVA) results. The y-axis shows the percentage of SPS-5 sites, and the x-axis shows the type of treatment performed on the pavement. The x-axis is divided into three parts: overlay thickness, mix type, and milling. Each one of these design or construction features is divided into three more subdivisions, which, in turn, are further divided as short-term (black bars) and long-term (light gray bars) treatments. While 11 percent of the SPS-5 sites had thin overlay work done for both short-term and long-term treatments, 56 percent accounted for short-term thick overlay work, and 50 percent accounted for long-term thick overlay work. The remaining sites had no differences and consisted of 33 and 39 percent for short- term and long-term treatments, respectively. Reclaimed asphalt pavement mix type accounted for 11 and 17 percent of the SPS-5 sites for short-term and long-term treatments,  with 28 and 39 percent corresponding to virgin mix type while 61 and 44 percent of the sites had no difference in mix type for short-term and long-term treatments, respectively. While 44 and 56 percent of SPS-5 sites had the best short-term and long-term milling work done, 11 percent had no milling done for both short-term and long-term treatments. The remaining sites had no differences and consisted of 44 and 33 percent for short-term and long-term treatments, respectively.

Figure 19. Graph. Summary of SPS-5 sites by best-performing design feature for roughness according to repeated measures ANOVA results.

Results for rutting are provided in figure 20, suggesting that the overlay thickness did not impact performance in the majority of sites. Mix type was also a design feature in which the impact on performance was not observed in the majority of sites. It is important to note that when differences were found, they were observed more in favor of sections overlaid with virgin mixes. The results also suggested that milling did not have an impact on rutting performance, and the majority of sites had milled and nonmilled sections performing similarly.

The results in figure 20 were obtained from statistical analysis and engineering judgment. Statistical significant differences were not found (p-values greater than 0.05), but differences in trends were clearly visible. In the analysis of thickness, five sites (28 percent of all SPS-5 sites) had statistically significant differences in rutting performance. Four sites (22 percent) had statistically significant differences in the analysis of mix type, and only two sites (12 percent) had statistically significant differences in the analysis of milling.

This graph shows a summary of Specific Pavement Study (SPS)-5 sites by best-performing design feature for rutting according to repeated measures analysis of variance (ANOVA) results. The y-axis shows the percentage of SPS-5 sites, and the x-axis shows the type of treatment performed on the pavement. The x-axis is divided into three parts: overlay thickness, mix type, and milling. Each one of these design or construction features is divided into three more subdivisions, which, in turn, are further divided as short-term (black bars) and long-term (gray bars) treatments. While 28 percent of the sites had thin overlay work done for both short-term and long-term treatments, 11 percent accounted for short-term thick overlay work and 28 percent for long-term thick overlay work. The remaining sites had no differences and consisted of 61 and 44 percent for short-term and long-term treatments, respectively. Reclaimed asphalt pavement mix type accounted for 0 and 22 percent of the sites for short-term and long-term treatments, 28 and 33 percent corresponded to virgin mix type, and 72 and 44 percent of the sites had no difference in mix type for short-term and long-term treatments, respectively. While 6 and 11 percent of sites had the best short-term and long-term milling work done, 17 percent of the sites had no short-term milling work done, and 28 percent of sites had no long-term milling treatments. The remaining sites had no differences and consisted of 78 and 61 percent for short-term and long-term treatments, respectively.

Figure 20. Graph. Summary of SPS-5 sites by best-performing design feature for rutting according to repeated measures ANOVA results.

Results for fatigue cracking are summarized in figure 21. None of the design features evaluated had an impact on short-term fatigue cracking performance. Sections with thick overlays had better long-term performances than thin overlays, although there were still 44 percent of sites in which no differences were found. Sections with virgin mix overlay had better long-term performances in half of the sites. Surprisingly, milled and nonmilled sections had equivalent long-term performances in the majority of sites.

The results in figure 21 were obtained from statistical analysis and engineering judgment. Statistically significant differences were not found (p-values higher than 0.05), but differences in trends were clearly visible. In the thickness analysis, four sites (22 percent of all SPS-5 sites) had statistically significant differences in fatigue cracking performance. Five sites (28 percent) had statistically significant differences in the analysis of mix type, and two sites (12 percent) had statistically significant differences in the analysis of milling.

This graph shows a summary of Specific Pavement Study (SPS)-5 sites by best-performing design feature for fatigue cracking according to repeated measures analysis of variance (ANOVA) results. The y-axis shows the percentage of SPS-5 sites, and the x-axis shows the type of treatment performed on the pavement. The x-axis is divided into three parts: overlay thickness, mix type, and milling. Each one of these design or construction features is further divided into three more subdivisions, which, in turn, are further divided as short-term (black bars) and long-term (gray bars) treatments. While none of the SPS-5 sites had short-term thin overlay work done, 11 percent had long-term thin overlay work performed. A total of 28 percent of sites accounted for short-term thick overlay work, and 44 percent accounted for long-term thick overlay work. The remaining sites had no differences and consisted of 72 and 44 percent for short-term and long-term treatments, respectively. Reclaimed asphalt pavement mix type accounted for 0 and 17 percent of the sites for short-term and long-term treatments, 28 and 50 percent corresponding to virgin mix type, and 72 and 33 percent of the sites had no differences in mix type for short-term and long-term treatments, respectively. While 22 and 28 percent of the sites had the best short-term and long-term milling work done, 0 percent of the sites had no short-term milling work done, and 17 percent of sites had no long-term milling treatments. The remaining sites had no differences and consisted of 78 and 56 percent for short-term and long-term treatments, respectively.

Figure 21. Graph. Summary of SPS-5 sites by best-performing design feature for fatigue cracking according to repeated measures ANOVA results.

Results for transverse cracking in figure 22 suggest that none of the design features evaluated in the SPS-5 experiment had any impact on performance in the majority of sites. It is worth noting that thick overlays had better transverse cracking long-term performance in 22 percent of sites, and milling prior to overlay had better transverse cracking long-term performance in 39 percent of sites.

The results in figure 22 were obtained from statistical analysis and engineering judgment. Statistically significant differences were not found (p-values greater than 0.05), but differences in trends were visible. In the thickness analysis, five sites (28 percent of all SPS-5 sites) had statistically significant differences in transverse cracking performance. Five sites (28 percent) had statistically significant differences in the analysis of mix type, and three sites (18 percent) had statistically significant differences in the analysis of milling.

This graph shows summary of Specific Pavement Study (SPS)-5 sites by best-performing design feature for transverse cracking according to repeated measures analysis of variance (ANOVA) results. The y-axis shows the percentage of SPS-5 sites, and the x-axis shows the type of treatment performed on the pavement. The x-axis is divided into three parts: overlay thickness, mix type, and milling. Each one of these design or construction features is further divided into three more subdivisions, which, in turn, are further divided as short-term (black bars) and long-term (gray bars) treatments. While none of the SPS-5 sites had short-term thin overlay work done, 6 percent had long-term thin overlay work performed. Additionally, 17 percent of sites accounted for short-term thick overlay work, and 22 percent accounted for long-term thick overlay work. The remaining sites had no differences in overlay thickness and consisted of 83 and 72 percent for short-term and long-term treatments, respectively. Reclaimed asphalt pavement mix type accounted for 17 percent of the sites for short-term and long-term treatments, while 6 and 11 percent corresponded to the virgin mix type. A total of 72 percent of the sites had no difference in mix type for both short-term and long-term treatments. While 39 percent of the sites had both short-term and long-term milling work done, only 6 percent had no long-term milling treatment. The remaining sites had no differences in milling treatment and consisted of 61 and 56 percent for short-term and long-term treatments, respectively.

Figure 22. Graph. Summary of SPS-5 sites by best-performing design feature for transverse cracking according to repeated measures ANOVA results.

Results for longitudinal cracking are provided in figure 23 and suggest that for the majority of sites, none of the design features evaluated in the SPS-5 experiment had an impact on performance. It is worth noting that thin overlays had better longitudinal cracking long-term performance in 22 percent of sites, and milling prior to overlay had better longitudinal cracking long-term performance in 39 percent of sites.

The results in figure 23 were obtained from statistical analysis and engineering judgment. Statistically significant differences were not found (p-values greater than 0.05), but differences in trends were visible. In the thickness analysis, three sites (18 percent of all SPS-5 sites) had statistically significant differences in longitudinal cracking performance. Three sites (18 percent) had statistically significant differences in the mix type analysis, and four sites (22 percent) had statistically significant differences in the milling analysis.

This graph shows a summary of Specific Pavement Study (SPS)-5 sites by best-performing design feature for longitudinal cracking according to repeated measures analysis of variance (ANOVA) results. The y-axis shows the percentage of SPS-5 sites, and the x-axis shows the type of treatment performed on the pavement. The x-axis is divided into three parts: overlay thickness, mix type, and milling. Each one of these design or construction features is divided into three more subdivisions, which, in turn, are further divided as short-term (black bars) and long-term (gray bars) treatments. A total of 11 percent SPS-5 sites had short-term thin overlay work done, and thin overlay work was performed on 22 percent of the sites. Additionally, 6 percent of sites accounted for short-term thick overlay work, and 17 percent accounted for long-term thick overlay work. The remaining sites had no differences in overlay thickness and consisted of 83 and 61 percent for short-term and long-term treatments, respectively. Reclaimed asphalt pavement mix type accounted for 6 percent of the sites for short-term treatments, while long-term treatments were performed on 11 percent of the sites. A total of 22 percent of the sites accounted for virgin mix type corresponding to both short-term and long-term treatments, while 72 and 67 percent of the sites accounted for no differences in mix type for the short-term and long-term treatments. Short-term milling was performed on 28 percent of the sites, and long-term milling was performed on 39 percent of the sites. The remaining sites had no differences and consisted of 72 and 61 percent for short-term and long-term treatments, respectively.

Figure 23. Graph. Summary of SPS-5 sites by best-performing design feature for longitudinal cracking according to repeated measures ANOVA results.

The key findings from the assessment of individual sites are listed below.

Roughness results were as follows:

Rutting results were as follows:

Fatigue cracking results were as follows:

Transverse cracking results were as follows:

Longitudinal cracking results were as follows:

Consolidated Analysis

The consolidated analysis involved compiling all sites in the SPS-5 experiment and simultaneously evaluating the impact of design features and site conditions for short-term and long-term performance. WD was the parameter selected for the comparisons, and it allowed the analysis to be carried across different conditions observed in each site of the experiment (more specifically, the different periods of monitoring data).

After the data were processed and verified for quality and existing outliers were corrected after LTPP analysis or removed, WD was computed for short-term and long-term performance. For simplicity, only the values for long-term performance are shown in table 35 through table 39. The remaining results are available in appendix A of this report. The Friedman test used the distress-associated WD to create a ranking of performance from the lowest value of WD (best performance) to the highest value (worst performance) for each site in the dataset. Ranking statistics for each type of section were then used to calculate the Friedman chi-square value used to determine if statistical differences existed among the performance rankings of the sections.

The WD-distress represents the overall performance of the section. It is better understood as an index computed based on the entire performance at a given period, as illustrated in figure 24. Therefore, it is intended for comparative analyses. The higher the WD value, the more distressed the pavement section is compared to sections with lower WD values.

This graph shows an example of weight distress (WD) values in comparative performance analysis. The x-axis shows the number of years after the pavement is repaired starting from -5 (i.e., 5 years before rehabilitation work was performed on the pavement) to 20 years. The y-axis shows International Roughness Index (IRI) values in inches per mile. The curve for IRI is represented by a dark blue line with blue diamonds for data points. The IRI value starts at about 125 inches/mi (1.97 m/km) a few months before year zero and decreases to 60 inches/mi (0.95 m/km) a few months after year zero. Next, there is a consistent pattern of an alternate increase and decrease about every 2 years. The IRI increases from 60 to 118 inches/mi (0.95 to 1.86 m/km) over 20 years. The area below the curve is hatched and has the following text: "Area is related to pavement performance relative to the specific distress." There are two blue arrows, the first has the word "Area" next to it and is pointing down, and the second has the word "Performance" next to it and is pointing up, indicating that as area decreases, pavement performance increases.

1 inch = 25.4 mm
1 mi = 1.61 km

Figure 24. Graph. Example of WD-distress values in comparative performance analysis for IRI trend after rehab.

Table 35. Long-term average WD-IRI values for SPS-5 sites.

Section

Experimental Design

Sites (State Codes)/Average WD-IRI Values (m/km)

Mill

Mix

Thickness (mm)

1

4

6

8

12

13

23

24

27

28

29

30

34

35

40

48

81

83

0501

No

None

   

80

108

70

   

92

97

193

88

130

 

134

39

113

 

127

98

0502

No

RAP

51

55

134

130

65

49

40

42

80

90

96

72

74

70

44

86

82

93

104

0503

No

RAP

127

53

77

75

50

49

40

54

71

87

114

62

62

45

33

66

76

89

69

0504

No

Virgin

127

57

82

74

56

42

40

56

86

95

87

71

48

51

37

70

93

101

71

0505

No

Virgin

51

58

92

102

55

36

40

45

90

101

110

69

57

57

39

64

95

83

107

0506

Yes

Virgin

51

48

71

81

87

32

36

52

59

92

101

69

56

51

37

66

90

72

117

0507

Yes

Virgin

127

56

88

73

65

37

39

55

63

72

83

83

61

52

42

62

83

93

59

0508

Yes

RAP

127

65

64

64

52

46

49

48

53

80

92

62

48

48

35

61

74

78

63

0509

Yes

RAP

51

55

115

142

62

37

40

60

76

87

108

84

62

49

37

64

78

94

84

1 inch = 25.4 mm
1 ft = 0.305 m
1 mi = 1.61 km
Note: Higher WD values indicate rougher pavement over time. The blank cells indicate data are not available

Table 36. Long-term average WD-rutting values for SPS-5 sites.

Section

Experimental Design

Sites (State Codes)/Average WD-Rutting Values (mm)

Mill

Mix

Thickness (mm)

1

4

6

8

12

13

23

24

27

28

29

30

34

35

40

48

81

83

0501

No

None

   

0.36

0.15

0.33

   

0.56

0.36

0.27

0.55

0.33

 

0.31

0.14

0.41

 

0.36

0.36

0502

No

RAP

51

0.10

0.19

0.16

0.13

0.16

0.13

0.27

0.17

0.10

0.36

0.14

0.17

0.11

0.12

0.11

0.23

0.25

0.14

0503

No

RAP

127

0.13

0.15

0.10

0.11

0.17

0.13

0.27

0.25

0.08

0.43

0.19

0.12

0.09

0.15

0.17

0.16

0.33

0.17

0504

No

Virgin

127

0.14

0.11

0.16

0.09

0.15

0.14

0.31

0.21

0.07

0.60

0.12

0.17

0.11

0.15

0.12

0.21

0.28

0.13

0505

No

Virgin

51

0.12

0.12

0.15

0.12

0.13

0.12

0.25

0.15

0.09

0.33

0.12

0.13

0.10

0.12

0.16

0.18

0.18

0.16

0506

Yes

Virgin

51

0.09

0.11

0.12

0.14

0.11

0.12

0.35

0.12

0.09

0.36

0.12

0.20

0.13

0.15

0.15

0.25

0.25

0.18

0507

Yes

Virgin

127

0.13

0.21

0.20

0.17

0.14

0.13

0.33

0.22

0.09

0.59

0.08

0.17

0.12

0.18

0.15

0.25

0.22

0.21

0508

Yes

RAP

127

0.21

0.14

0.11

0.13

0.16

0.12

0.32

0.19

0.08

0.56

0.17

0.11

0.10

0.16

0.11

0.18

0.23

0.22

0509

Yes

RAP

51

0.13

0.16

0.13

0.09

0.13

0.13

0.30

0.43

0.11

0.33

0.11

0.17

0.13

0.15

0.09

0.16

0.26

0.15

1 inch = 25.4 mm
Note: Higher WD values indicate the pavement is more rutted over time. The blank cells indicate data are not available.

Table 37. Long-term average WD-fatigue cracking values for SPS-5 sites.

Section

Experimental Design

Sites (State Codes)/Average WD-Fatigue Cracking Values (m2)

Mill

Mix

Thickness (mm)

1

4

6

8

12

13

23

24

27

28

29

30

34

35

40

48

81

83

0501

No

None

 

2,305

692

   

13

810

1

51

1,927

 

2,475

11

0

 

46

377

2,305

692

0502

No

RAP

51

2,566

966

38

0

0

25

0

458

0

1,263

463

1

1

3

1,480

916

2,566

966

0503

No

RAP

127

513

137

1

0

0

5

0

46

0

993

151

0

10

8

1,052

734

513

137

0504

No

Virgin

127

475

111

0

0

0

96

0

4

2

0

178

2

1

0

384

544

475

111

0505

No

Virgin

51

1,467

674

1

0

0

183

0

94

0

16

165

1

4

0

700

681

1,467

674

0506

Yes

Virgin

51

474

1,261

0

0

0

9

0

198

3

1

7

6

1

0

765

822

474

1261

0507

Yes

Virgin

127

528

840

0

1

0

0

0

0

108

0

45

4

1

5

273

489

528

840

0508

Yes

RAP

127

85

180

0

0

0

84

0

258

0

725

47

0

1

8

399

498

85

180

0509

Yes

RAP

51

2,204

16

0

1

0

0

0

705

2

1,511

650

2

0

29

1,272

1068

2,204

16

1 ft = 0.305 m
1 inch = 25.4 mm
Note: Higher WD values indicate increased cracking in the pavement over time. The blank cells indicate data are not available.

Table 38. Long-term average WD-transverse cracking values for SPS-5 sites.

Section

Experimental Design

Sites (State Codes)/Average WD-Transverse Cracking Values (m)

Mill

Mix

Thickness (mm)

1

4

6

8

12

13

23

24

27

28

29

30

34

35

40

48

81

83

0501

No

None

   

0

154

63

   

38

140

245

201

32

 

270

33

91

 

30

32

0502

No

RAP

51

110

87

153

69

5

0

0

134

262

96

0

26

199

47

62

220

96

188

0503

No

RAP

127

4

339

194

11

0

0

0

54

196

136

0

14

69

44

24

98

88

151

0504

No

Virgin

127

0

40

83

43

0

0

0

32

200

3

0

35

50

3

5

4

61

250

0505

No

Virgin

51

54

216

197

64

8

0

0

152

327

50

0

57

155

56

52

186

288

81

0506

Yes

Virgin

51

1

50

152

90

2

0

0

129

294

65

1

51

9

2

33

4

130

39

0507

Yes

Virgin

127

1

2

100

10

0

0

0

4

149

1

4

10

11

0

0

2

143

208

0508

Yes

RAP

127

1

207

257

9

0

0

0

59

217

80

0

10

51

2

0

73

54

225

0509

Yes

RAP

51

13

339

209

40

0

0

0

5

230

32

1

0

47

9

30

155

19

110

1 ft = 0.305 m
1 inch = 25.4 mm
Note: Higher WD values indicate increased cracking in the pavement over time. The blank cells indicate data are not available.

Table 39. Long-term average WD-longitudinal cracking values for SPS-5 sites.

Section

Experimental Design

Sites (State Codes)/Average WD-Longitudinal Cracking Values (m)

Mill

Mix

Thickness (mm)

1

4

6

8

12

13

23

24

27

28

29

30

34

35

40

48

81

83

0501

No

None

   

0

364

842

   

1,039

939

894

239

563

 

670

380

53

 

170

250

0502

No

RAP

51

95

47

268

722

2

304

277

795

843

209

52

377

862

405

252

797

575

741

0503

No

RAP

127

143

337

430

552

2

209

292

624

549

101

96

186

949

508

113

646

579

848

0504

No

Virgin

127

86

26

362

715

9

132

302

429

677

52

33

176

827

103

83

80

704

669

0505

No

Virgin

51

92

130

342

996

27

293

277

486

894

47

465

270

585

268

28

797

319

932

0506

Yes

Virgin

51

0

78

485

673

0

144

214

627

737

52

88

205

599

158

36

337

84

574

0507

Yes

Virgin

127

19

3

459

410

0

129

277

651

649

17

74

89

753

60

65

19

456

578

0508

Yes

RAP

127

69

285

497

152

73

135

218

745

369

170

116

266

926

559

0

631

578

921

0509

Yes

RAP

51

100

329

492

152

8

206

138

99

572

146

54

322

826

411

48

639

544

392

1 ft = 0.305 m
1 inch = 25.4 mm
Note: Higher WD values indicate increased cracking in the pavement over time. The blank cells indicate data are not available.

As noted earlier, the Friedman null hypothesis states that there are no differences between the ranking of sections (i.e., all sections have similar performances). The null hypothesis is rejected if the p-value is lower than 0.05, which represents a 95 percent confidence level that at least two sections have statistically different rankings. Examples of Friedman test outputs are provided in figure 25 and figure 26. In the figures, the average WD value for IRI found for each rehabilitation strategy among all sites was analyzed. The vertical bars represent the interval between the mean value ±1 standard deviation as an illustration of the variability of the measurements. The results in figure 25 indicate that for short-term roughness performance, there were at least two sections with statistically different performance (p < 0.0001, ANOVA chi-square = 48.8889). A similar result was found for long-term performance, as shown in figure 26 (p < 0.0001, ANOVA chi-square = 44.5667).

This graph shows a bar plot of weighted distress (WD) average International Roughness Index (IRI) for jointed plain concrete pavement in Specific Pavement Study (SPS)-5 sections for short-term analysis. The x-axis shows nine SPS-5 sections (0501, 0502, 0503, 0504, 0505, 0506, 0507, 0508, and 0509), and the y-axis shows IRI values in meters per kilometers. Mean values are represented by grey dots, and the nine corresponding black vertical bars represent the mean plus or minus 1 times the standard deviation. The nine have the following mean and high and low values of IRI for the short term: 5.91, 8.13, and 3.70 ft/mi (1.12, 1.54, and 0.7 m/km); 4.17, 6.65, and 1.70 ft/mi (0.79, 1.26, and 0.32 m/km); 3.22, 4.75, and 1.69 ft/mi (0.61, 0.9, and 0.32 m/km); 3.17, 4.17, and 2.16 ft/mi (0.6, 0.79, and 0.41 m/km); 3.80, 5.86, and 1.74 ft/mi (0.72, 1.11, and 0.33 m/km); 3.64, 5.65, and 1.64 ft/mi (0.69, 1.07, and 0.31 m/km); 2.80, 3.64, and 2.01 ft/mi (0.53, 0.69, and 0.38 m/km); 2.90, 3.75, and 1.79 ft/mi (0.53, 0.71, and 0.34 m/km); and 4.17, 7.50, and 0.90 ft/mi (0.79, 1.42, and 0.17 m/km).

1 ft = 0.305 m
1 mi = 1.61 km

Figure 25. Graph. WD-IRI short-term values in SPS-5 sites.

This graph shows a bar plot of weighted distress (WD) average International Roughness Index (IRI) for jointed plain concrete pavement in Specific Pavement Study (SPS)-5 sections for long-term analysis. The x-axis shows nine SPS-5 sections (0501, 0502, 0503, 0504, 0505, 0506, 0507, 0508, and 0509), and the y-axis shows IRI values in meters per kilometers. Mean values are represented by grey dots, and the nine corresponding black vertical bars are used to represent the mean plus or minus 1 times the standard deviation. The nine sections have the following mean and standard deviation values of IRI for the long term: 8.24, 11.19, and 5.28 ft/mi (1.56, 2.12, and 1 m/km); 6.55, 8.81, and 
4.22 ft/mi (1.24, 1.67, and 0.8 m/km); 5.44, 7.07,and 3.69 ft/mi (1.03, 1.34, and 0.7 m/km); 5.65, 7.34, and 3.96 ft/mi (1.07, 1.39, and 0.75 m/km); 6.02, 8.13, and 3.91 ft/mi (1.14, 1.54, and 0.74 m/km); 5.65, 7.60, and 3.64 ft/mi (1.07, 1.44, and 0.69 m/km); 5.39, 6.81, and 4.01 ft/mi (1.02, 1.29, and 0.76 m/km); 5.02, 6.23, and 3.80 ft/mi (0.95, 1.18 and 0.72 m/km); and 6.18, 8.55, and 3.75 ft/mi (1.17, 1.62, and 0.71 m/km).

1 ft = 0.305 m
1 mi = 1.61 km

Figure 26. Graph. WD-IRI long-term values in SPS-5 sites.

When the result of the Friedman test indicated the existence of at least two strategies with statistically different rankings, the next step was to identify which sections were different and to build the rankings of best-performing strategies based on the statistical analysis. Table 40 and table 41 provide the p-values for each Friedman test paired analysis for short-term and long-term roughness performance rankings.

Table 40. Friedman test paired analysis of rehabilitation strategies for short-term roughness performance ranking for SPS-5 sites.

Paired Analysis

p-Value

0501 and 0502

-

0501 and 0503

< 0.05

0501 and 0504

< 0.05

0501 and 0505

-

0501 and 0506

< 0.05

0501 and 0507

< 0.05

0501 and 0508

< 0.05

0501 and 0509

< 0.05

0502 and 0503

-

0502 and 0504

-

0502 and 0505

-

0502 and 0506

-

0502 and 0507

< 0.05

0502 and 0508

< 0.05

0502 and 0509

-

0503 and 0504

-

0503 and 0505

-

0503 and 0506

-

0503 and 0507

-

0503 and  0508

-

0503 and 0509

-

0504 and 0505

-

0504 and 0506

-

0504 and 0507

-

0504 and 0508

-

0504 and 0509

-

0505 and 0506

-

0505 and 0507

-

0505 and 0508

< 0.05

0505 and 0509

-

0506 and 0507

-

0506 and 0508

-

0506 and 0509

-

0507 and 0508

-

0507 and 0509

-

0508 and 0509

-

- Indicates pair analysis with no
statistical significance.

Table 41. Friedman test paired analysis of rehabilitation strategies for long-term roughness performance ranking for SPS-5 sites.

Paired Analysis

p-Value

0501 and 0502

-

0501 and 0503

< 0.05

0501 and 0504

-

0501 and 0505

-

0501 and 0506

< 0.05

0501 and 0507

< 0.05

0501 and 0508

< 0.05

0501 and 0509

-

0502 and 0503

-

0502 and 0504

-

0502 and 0505

-

0502 and 0506

-

0502 and 0507

-

0502 and 0508

< 0.05

0502 and 0509

-

0503 and 0504

-

0503 and 0505

-

0503 and 0506

-

0503 and 0507

-

0503 and 0508

-

0503 and 0509

-

0504 and 0505

-

0504 and 0506

-

0504 and 0507

-

0504 and 0508

-

0504 and 0509

-

0505 and 0506

-

0505 and 0507

-

0505 and 0508

-

0505 and 0509

-

0506 and 0507

-

0506 and 0508

-

0506 and 0509

-

0507 and 0508

-

0507 and 0509

-

0508 and 0509

-

- Indicates pair analysis with no
statistical significance.

The paired analysis results were used to create a practical ranking of roughness performance based on the statistical differences that were identified. The tables were intended to help users select the best alternatives given the specific conditions that they may want to evaluate. Based on the results presented in table 40 and table 41, the final ranking for evaluating roughness performance was created for the short term and long term (see table 42). Sections were ordered from best to worst performance, and sections with equivalent performance were grouped under the same rank.

Table 42. Ranking of rehabilitation strategies for roughness, SPS-5 sites.

Statistical Relevance (Y/N)

Roughness

Short-Term

Long-Term

Y

(p < 0.0001)

Y

(p < 0.0001)

Ranking
(if relevant)

Ranking

Strategy

Ranking

Strategy

1

Mill, thick, RAP

1

Mill, thick, RAP

2

Mill, thick, virgin

2

No mill, thick, RAP

3

No mill, thick, RAP

2

Mill, thin, virgin

3

No mill, thick, virgin

2

Mill, thick, virgin

3

Mill, thin, virgin

5

No mill, thick, virgin

3

Mill, thin, RAP

5

Mill, thin, RAP

3

No mill, thin, virgin

5

No mill, thin, virgin

8

No mill, thin, RAP

8

No mill, thin, RAP

9

Control

9

Control


The results in table 42 suggest that rehabilitations with milling and thick recycled overlays provided smoother pavements in both the short-term and long-term performance. Strategies with milling and virgin thick overlay were the second best for short-term performance. For long-term roughness performance, three strategies had equivalent second best performances. A broader analysis of both rankings suggests that thick overlays provided better performance over the short term and long term. Overall, differences between performance of recycled asphalt and virgin overlays were difficult to identify, suggesting that roughness performance was not significantly affected by the overlay mix type. Both rankings also suggested that strategies with milling were more likely to provide better short-term and long-term roughness performance. These results agree with the conclusions drawn from the analysis of individual sites.

The same approach described for the analysis of roughness was applied to all distresses (rutting, fatigue, transverse, and longitudinal cracking). Figure 27 and figure 28 describe the Friedman test for rutting. Based on the same test statistics, the ranking of best-performing rehabilitation strategies for rutting was created (see table 43). The results suggest that thin overlays performed better at early stages for short-term performance (see figure 27, p < 0.0001, ANOVA chi-square = 40.2), while no significant differences were identified for long-term performance (see figure 28, p < 0.0001, ANOVA chi-square = 38.9778). The top rankings also were equally distributed among sections overlaid with virgin and RAP mixes and among sections previously milled and not milled. These findings agree with the observations from the analysis of individual sites.

This graph shows a bar plot of short-term weighted distress (WD) average rutting for jointed plain concrete pavement in Specific Pavement Study (SPS)-5 sections. The x-axis shows nine SPS-5 sections (0501, 0502, 0503, 0504, 0505, 0506, 0507, 0508, and 0509), and the y-axis shows rutting values in millimeters. Mean values are represented by grey dots, while the nine corresponding black vertical bars are used to represent the mean plus or minus 1 times the standard deviation. The nine sections have the following mean and high and low values of rutting for the short term: 0.24, 0.33, and 0.15 inches (6.09, 8.39, and 3.79 mm); 0.12, 0.20, and 0.04 inches (3.03, 5.04, and 1.01 mm); 0.13, 0.25, and 0.02 inches (3.45, 6.3, and 0.61 mm); 0.13, 0.22, and 0.04 inches (3.26, 5.53, and 1.0 mm); 0.10, 0.14, and 0.06 inches (2.57, 3.71, and 1.43 mm); 0.12, 0.19, and 0.04 inches (2.95, 4.76, and 1.15 mm); 0.13, 0.22, and 0.05 inches (3.36, 5.55, and 1.17 mm); 0.13, 0.23, and 0.02 inches (3.27, 5.92, and 0.63 mm); and 0.12, 0.19, and 0.04 inches (3.03, 4.94, and 1.12 mm).

1 inch = 25.4 mm

Figure 27. Graph. WD-rutting short-term values for SPS-5 sites.

This graph shows a bar plot of long-term weighted distress (WD) average rutting for jointed plain concrete pavement in Specific Pavement Study (SPS)-5 sections. The x-axis shows nine SPS-5 sections (0501, 0502, 0503, 0504, 0505, 0506, 0507, 0508, and 0509), and the y-axis shows rutting values in millimeters. Mean values are represented by grey dots, while the nine corresponding black vertical bars are used to represent the mean plus or minus 1 times the standard deviation. The nine sections have the following mean, high, and low values of rutting for the long term: 0.33, 0.44, and 0.23 inches (8.54, 11.29, and 5.8 mm); 0.17, 0.24, and 0.10 inches (4.29, 6.05, and 2.53 mm); 0.18, 0.27, and 0.09 inches (4.53, 6.83, and 2.24 mm); 0.18, 0.30, and 0.06 inches (4.60, 7.67, and 1.53 mm); 0.15, 0.21, and 0.09 inches (3.82, 5.27, and 2.37 mm); 0.17, 0.25, and 0.09 inches (4.3, 6.41, and 2.2 mm); 0.20, 0.31, and 0.08 inches (5.09, 8.02, and 2.16 mm); 0.18, 0.29, and 0.07 inches (4.63, 7.46, and 1.79 mm); and 0.17, 0.26, and 0.08 inches (4.45, 6.8, and 2.1 mm).

1 inch = 25.4 mm

Figure 28. Graph. WD-rutting long-term values for SPS-5 sites.

Table 43. Ranking of rehabilitation strategies for rutting for SPS-5 sites.

Statistical Relevance (Y/N)

Rutting

Short-Term

Long-Term

Y

(p < 0.0001)

Y

(p < 0.0001)

Ranking
(if relevant)

Ranking

Strategy

Ranking

Strategy

1

No mill, thin, virgin

1

No mill, thin, virgin

1

No mill, thin, RAP

1

Mill, thick, RAP

1

Mill, thin, RAP

1

No mill, thin, RAP

1

Mill, thick, RAP

1

Mill, thin, virgin

1

Mill, thin, virgin

1

No mill, thick, RAP

1

No mill, thick, RAP

1

Mill, thin, RAP

1

No mill, thick, virgin

1

No mill, thick, virgin

8

Mill, thick, virgin

8

Mill, thick, virgin

9

Control

9

Control


Design features were found to have an impact on only long-term performance associated with fatigue, longitudinal, and transverse cracking. Figure 29 through figure 31 present descriptive statistics for performance rankings among all sites. Based on the Friedman test, the ranking of best-performing rehabilitation strategies for cracking was created (see table 44 through table 46). Figure 29 (p = 0.0009, ANOVA chi-square = 26.5074) and table 44 suggest that thick overlays performed better in the long term for fatigue cracking. As expected, it was evident from the figure that the no treatment control alternative performed the poorest in regards to fatigue cracking. The ranking of alternatives was more equally distributed when comparing mix types; however, out of the top three alternatives, two were virgin mix overlays. Overall, mix type had limited influence on long-term fatigue cracking performance. The results also suggested that milling prior to overlay improved performance. Two of the top three alternatives included milling prior to the overlay.

The results described in figure 30 (p < 0.0001, ANOVA chi-square = 33.6741) and table 45 suggested that thick overlays were better to mitigate transverse cracking. The two best-ranked sections had virgin mix overlays, but overall the performances of virgin and RAP mix overlays were similar. Sections that were milled prior to overlay consistently performed better than nonmilled ones.

The results for longitudinal cracking in figure 31 (p = 0.0011, ANOVA chi-square = 25.8407) and table 46 suggested that none of the design features had a significant influence on performance. Although the best alternative was milling and overlaying with a thick virgin mix, the remaining alternatives that ranked second consisted of different combinations of design features with no clear trend to which one provided better performance associated with longitudinal cracking.

This graph shows a bar plot of long-term weighted distress (WD) fatigue cracking for jointed plain concrete pavement in Specific Pavement Study (SPS)-5 sections. The x-axis shows nine SPS-5 sections (0501, 0502, 0503, 0504, 0505, 0506, 0507, 0508, and 0509), and the y-axis shows WD fatigue cracking values in square meters. Mean values are represented by grey dots, while the nine corresponding black vertical bars are used to represent the mean plus or minus 1 times the standard deviation. The nine sections have the following mean, high, and low values of rutting for fatigue cracking: 1,205.51, 3,369.13, and zero ft2 (111.96, 313, and zero m2); 604.61, 1,438.82, and 0 ft2 (56.17, 133.67, and 0 m2); 204.19, 562.96, and zero ft2 (18.97, 52.3, and 0 m2); 99.99, 278.86, and 0 ft2 (9.29, 25.88, and 0 m2); 315.28, 815.37, and 0 ft2 (29.29, 75.75, and 0 m2); 197.84, 574.37, and 0 ft2 (18.38, 53.36, and 0 m2); 128.52, 372.86, and 0 ft2 (11.94, 34.64, and 0 m2); 133.26, 342.08, and 0 ft2 (12.38, 31.78, and 0 m2); and 422.16, 1,091.79, and 0 ft2 (39.22, 101.43, and 0 m2).

1 ft2 = 0.093 m2

Figure 29. Graph. Fatigue cracking WD values for long-term performance of SPS-5 sites.

This graph shows a bar plot of long-term weighted distress (WD) average transverse cracking for jointed plain concrete pavement in Specific Pavement Study (SPS)-5 sections. The x-axis shows nine SPS-5 sections (0501, 0502, 0503, 0504, 0505, 0506, 0507, 0508, and 0509), and the y-axis shows transverse cracking values in meters. Mean values are represented by grey dots, and the black vertical bars represent the mean plus or minus 1 times the standard deviation. The nine sections have the following mean, high, and low values of rutting for transverse cracking: 111.65, 228.19, and 0 ft (34.04, 69.57, and 0 m); 97.42, 178.60, and 16.27 ft (29.7, 54.45, and 4.96 m); 78.98, 172.2, and 0 ft (24.08, 52.5, and 0 m); 44.94, 115.65, and 0 ft (13.7, 35.26, and 0 m); 107.91, 208.64, and 7.18 ft (32.9, 63.61, and 2.19 m); 58.42, 135.76, and 0 ft (17.81, 41.39, and 0 m); 35.82, 101.42, and zero ft (10.92, 30.92, and 0 m); 6.91, 160.33 ft (21.08, 48.88, and 0 m); and 68.81, 168.39, and 0 ft (20.98, 51.34, and 0 m).

1 ft = 0.305 m

Figure 30. Graph. Transverse cracking WD values for long-term performance of SPS-5 sites.

 

This graph shows a bar plot of long-term weighted distress (WD) average longitudinal cracking for jointed plain concrete pavement in Specific Pavement Study (SPS)-5 sections. The x-axis shows nine SPS-5 sections (0501, 0502, 0503, 0504, 0505, 0506, 0507, 0508, and 0509), and the y-axis shows longitudinal cracking values in meters. Mean values are represented by grey dots, and the black vertical bars represent the mean plus or minus 1 times the standard deviation. The nine sections have the following mean, high, and low values of rutting for longitudinal cracking: 442.67, 781.59, and 103.71 ft (134.96, 238.29, and 31.62 m); 423.38, 726.42, and 120.38 ft (129.08, 221.47, and 36.7 m); 397.90, 673.22, and 122.57 ft (121.31, 205.25, and 37.37 m); 303.53, 593.29, and 13.78 ft (92.54, 180.88, and 4.2 m); 402.55, 721.14, and 83.97 ft (122.73, 219.86, and 25.6 m); 282.77, 542.18, and 23.32 ft (86.21, 165.3, and 7.11 m); 261.48, 528.99, and 0 ft (79.72, 161.28, and 0 m); 438.27, 667.71, and 77.64 ft (133.62, 203.57, and 23.67 m); and 304.25, 541.10, and 67.40 ft (92.76, 164.97, and 20.55 m).

1 ft = 0.305 m

Figure 31. Graph. Longitudinal cracking WD values for long-term performance of SPS-5 sites.

Table 44. Ranking of rehabilitation strategies for fatigue cracking at SPS-5 sites.

Statistical Relevance (Y/N)

Long-Term

Y

p = 0.0009

Ranking
(if relevant)

Ranking

Strategy

1

No mill, thick, virgin

1

Mill, thick, RAP

1

Mill, thick, virgin

4

Mill, thin, virgin

4

No mill, thick, RAP

4

No mill, thin, virgin

4

Mill, thin, RAP

4

No mill, thin, RAP

9

Control


Table 45. Ranking of rehabilitation strategies for transverse cracking at SPS-5 sites.

Statistical
Relevance (Y/N)

Long-Term

Y

p < 0.0001

Ranking
(if relevant)

Ranking

Strategy

1

Mill, thick, virgin

2

No mill, thick, virgin

3

Mill, thick, RAP

3

Mill, thin, RAP

3

Mill, thin, virgin

3

No mill, thick, RAP

7

No mill, thin, RAP

8

No mill, thin, virgin

8

Control

Table 46. Ranking of rehabilitation strategies for longitudinal cracking at SPS-5 sites.

Statistical
Relevance (Y/N)

Long-Term

Y

p = 0.0011

Ranking
(if relevant)

Ranking

Strategy

1

Mill, thick, virgin

2

Mill, thin, virgin

2

No mill, thick, virgin

2

Mill, thin, RAP

2

No mill, thin, virgin

2

Mill, thick, RAP

7

No mill, thick, RAP

7

Control

7

No mill, thin, RAP

The results obtained in the consolidated analysis agreed for the most part with the results found in the individual site analysis. Overlay thickness was the most influential design feature. Thick overlays consistently performed better, as expected. The impact of thickness on performance was more evident in the long term (more than 5 years) rather than the short term for most of the distresses used as performance measures. The exception was rutting, for which no evidence was found suggesting that either thin or thick overlays provided less rutted pavements.

The majority of sites did not show significant differences in performance between sections overlaid with virgin and RAP mixes. However, when differences existed, they were mostly in favor of virgin mixes.

The analysis of milling prior to overlay suggested that replacing the distressed portion of the surface layer improved the performance for the majority of distresses commonly observed in flexible pavements.

Influence of Site Condition

The influence of site condition was determined by three variables: (1) pavement surface condition prior to rehabilitation, (2) climate, and (3) traffic levels. These three conditions were determined for each site, and the Friedman test was repeated by grouping the sites according to each of the following variables:

The designation of fair versus poor was assigned by the owner agency nominating the SPS-5 project. These ratings were purely subjective and not based on the actual level of existing distresses prior to rehabilitation. They were used only to ensure a range of surface conditions of the original pavement before rehabilitation. However, the assessment of distresses prior to overlay indicated that, on average, fair pavements had IRI values of 9.50 ft/mi (1.8 m/km) with 0.39 inches (10 mm) or less of rutting and up to 1,237.86 ft2 (115 m2) of fatigue cracking per section. Poor pavements had roughness of 8.71 ft/mi (1.65 m/km) with 0.59 inches (15 mm) of rutting and up to 1,937.52 ft2 (180 m2) of fatigue cracking per section.

Climate condition was defined based on the freeze index and average rainfall for each site. Sites with an average annual rainfall greater than 39 inches (1,000 mm) were classified as wet, and sites with less than 39 inches (1,000 mm) of rain were classified as dry. Similarly, sites with a freeze index greater than 140 °F (60 °C) were classified as a freezing climate, and sites with less than 140 °F (60 °C) were designated as a no-freeze climate. These classifications are part of the LTPP experiment definition.

The classification of traffic was defined based on volume and commercial vehicle distribution. These characteristics were simple to evaluate and, at the same time, most influential on pavement performance predictions estimated with MEPDG. The combination of criteria generated two groups of sites: low traffic and high traffic. Table 47 describes the characteristics of both groups used in this study. Georgia and Texas did not have any traffic information.

Table 47. Criteria for evaluating traffic characteristics of SPS-5 sites.

Traffic Characteristics

Low Traffic

High Traffic

AADTT

340–950

750–2,750

Vehicle class 5
(percent in volume)

25–75

5–20

Vehicle class 9
(percent in volume)

10–50

40–85

SPS-5 sites

Alabama, Florida, Maine, Maryland, Minnesota, Missouri, and Oklahoma

Arizona, California, Colorado, Mississippi, Montana, New Jersey, New Mexico, Alberta, and Manitoba


The analysis followed the same steps presented in the previous section. Rankings of rehabilitation strategies were developed for each group of sites using descriptive statistics and the paired analyses from the Friedman test when statistical differences in performance were found. The results are summarized in the tables presented in appendix C.

Table 48 and table 49 provide examples of how the data were summarized. These tables show the ranking of best-performing sections based on long-term performance for roughness and rutting in sections with fair and poor surface conditions prior to rehabilitation.

The examples illustrate the impact of site conditions on performance of rehabilitated flexible pavements. They suggest that rehabilitation strategies with milling and virgin mix overlays were better to improve roughness performance in pavements with poor surface condition. If surface condition was fair, RAP mixes provided a slight advantage in terms of roughness performance.

According to the ranking for rutting performance, rehabilitation strategies with milling and thin overlays with virgin mixes were the best alternatives when pavements had poor surface condition before the overlay. When surface conditions were fair, the impact of design features was not as significant. In fact, rehabilitation strategies with milling prior to overlay were among the worst ranked for rutting performance.

Table 48. Summary of rankings for long-term roughness and rutting performance of SPS-5 sites in fair surface condition prior to overlay.

Statistical Relevance (Y/N)

Distress

Roughness

Rutting

Y

p = 0.0001

Y

p = 0.0044

Ranking
(if relevant)

Ranking

Strategy

Ranking

Strategy

1 (Best)

Mill, thick, RAP

1 (Best)

Mill, thick, RAP

2

No mill, thick, RAP

1

No mill, thin, virgin

2

Mill, thin, virgin

1

No mill, thick, RAP

2

Mill, thin, RAP

4

No mill, thick, virgin

5

Mill, thick, virgin

4

No mill, thin, RAP

5

No mill, thick, virgin

4

Mill, thin, RAP

5

No mill, thin, virgin

4

Mill, thin, virgin

8

No mill, thin, RAP

4

Mill, thick, virgin

9 (Worst)

Control

9 (Worst)

Control

Table 49. Summary of rankings for long-term roughness and rutting performance of SPS-5 sites in poor surface condition prior to overlay.

Statistical Relevance (Y/N)

Distress

Roughness

Rutting

Y

p = 0.019

Y

p = 0.0015

Ranking
(if relevant)

Ranking

Strategy

Ranking

Strategy

1

Mill, thick, RAP

1

No mill, thin, virgin

2

Mill, thin, virgin

1

Mill, thin, virgin

2

Mill, thick, virgin

3

Mill, thin, RAP

2

No mill, thick, RAP

3

No mill, thin, RAP

2

No mill, thick, virgin

3

No mill, thick, virgin

2

No mill, thin, virgin

3

No mill, thick, RAP

2

No mill, thin, RAP

3

Mill, thick, RAP

2

Mill, thin, RAP

3

Mill, thick, virgin

9

None

9

None

A detailed assessment of the combined results for each of the analyses performed was assembled in tables for better visualization and interpretation of results. These tables were created for each distress and performance period (short-term and long-term performance) with the exception of fatigue and longitudinal cracking, which only presented statistically significant differences for long-term performance data . Table 50 and table 51 present the results for short-term and long-term roughness performance.

Table 52 and table 53 summarize the results for rutting. Table 54 shows the results for long-term fatigue cracking, and table 55 and table 56 present the results for short-term and long-term transverse cracking performance. Finally, table 57 summarizes the results for long-term longitudinal cracking. The best alternatives with statistical relevance are shown in each cell. The number before the treatment indicates its ranking among all alternatives.

The summary of best-performing strategies can be used as a practical guide to help select the best rehabilitation option based on performance. For example, if the section is located in a wet freeze region, and the pavement is in fair surface condition with low traffic levels, based on long-term roughness performance, three alternatives in table 51 provide equivalent best performance (mill, thick, and RAP; mill, thin, and virgin; and no mill, thick, and RAP). This performance-based selection can be further improved by evaluating material availability, costs, and other relevant issues.

These summary tables provide clear information for choosing the best rehabilitation treatment based on distress type and site condition. Moreover, the influence of different site conditions can be determined by observing the best treatments for each condition.

Table 50. Summary based on short-term roughness performance of SPS-5 pavement structures.

Climate

Traffic/Surface Condition

High

Low

Poor

Fair

Poor

Fair

Wet

Freeze

1: Mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thick, virgin

2: Mill, thick, RAP

2: Mill, thick, RAP

2: Mill thick, RAP

2: Mill, thick, RAP

3: No mill, thick, RAP

3: No mill, thick, RAP

3: No mill, thick, RAP

2: No mill, thick, RAP

4: Mill, thin, virgin

4: Mill, thin, RAP

4: Mill, thin, virgin

4: Mill, thin, RAP

No-freeze

1: Mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thick, virgin

2: Mill, thick, RAP

2: Mill thick, RAP

2: Mill, thick, RAP

2: Mill, thick, RAP

3: Mill, thin, virgin

3: No mill, thick, RAP

3: No mill, thick, RAP

3: No mill, thick, RAP

3: No mill, thick, RAP

4: Mill, thin, RAP

4: Mill, thin, virgin

4: Mill, thin, RAP

Dry

Freeze

1: Mill, thick, RAP

1: Mill, thick, RAP

1: Mill, thick, RAP

1: Mill, thick, RAP

1: Mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thick, virgin

3: No mill, thick, RAP

3: No mill, thick RAP

3: No mill, thick, RAP

3: No mill, thick, RAP

4: Mill, thin, RAP

4: Mill, thin, RAP

No-freeze

1: Mill, thick RAP

1: Mill, thick, RAP

1: Mill, thick, RAP

1: Mill, thick, RAP

1: Mill, thick, virgin

1: Mill thick, virgin

1: Mill, thick, virgin

1: Mill, thick, virgin

3: Mill thin, RAP

3: No mill, thick, RAP

3: No mill, thick, RAP

3: No mill, thick, RAP

4: Mill, thin, RAP

Indicates that no preferred treatment was statistically found.

Table 51. Summary based on long-term roughness performance of SPS-5 pavement structures.

Climate

Traffic Surface/Condition

High

Low

Poor

Fair

Poor

Fair

Wet

Freeze

1: Mill, thick, RAP

1: Mill, thick, RAP

1: Mill, thick, RAP

1: Mill, thick, RAP

1: Mill, thick, virgin

1: Mill, thin, virgin

1: Mill, thick, virgin

1: Mill, thin, virgin

1: Mill, thin, virgin

1: No mill, thick, RAP

1: Mill, thin, virgin

1: No mill, thick, RAP

1: No mill, thick, RAP

4: Mill, thick, virgin

1: No mill, thick, RAP

4: Mill, thick, virgin

No-freeze

1: Mill, thick, RAP

1: Mill, thick, RAP

1: Mill, thick, RAP

1: Mill, thick, RAP

1: Mill, thick, virgin

1: Mill, thin, virgin

1: Mill, thick, virgin

1: Mill, thin, virgin

1: Mill, thin, virgin

1: No mill, thick, RAP

1: Mill, thin, virgin

1: No mill, thick, RAP

1: No mill, thick, RAP

4: Mill, thick, virgin

1: No mill, thick, RAP

4: Mill, thick, virgin

Dry

Freeze

1: Mill, thick, RAP

1: Mill, thick, RAP

1: Mill, thick, RAP

1: Mill, thick, RAP

1: No mill, thick, RAP

1: No mill, thick, RAP

1: No mill, thick, RAP

1: No mill, thick, RAP

3: Mill, thick, virgin

3: Mill, thin, virgin

3: Mill, thick, virgin

3: Mill, thin, virgin

3: Mill, thin, virgin

4: Mill, thick, virgin

3: Mill, thin, virgin

4: Mill, thick, virgin

No-freeze

1: Mill, thick, RAP

1: Mill, thick, RAP

1: Mill, thick, RAP

1: Mill, thick, RAP

1: No mill, thick, RAP

1: No mill, thick, RAP

1: No mill, thick, RAP

1: No mill, thick, RAP

3: Mill, thick, virgin

3: Mill, thin, virgin

3:Mill, thick, virgin

3: Mill, thin, virgin

3: Mill, thin, virgin

4: Mill, thick, virgin

3: Mill, thin, virgin

4: Mill, thick, virgin

Table 52. Summary based on short-term rutting performance of SPS-5 pavement structures.

Climate

Traffic Surface/Condition

High

Low

Poor

Fair

Poor

Fair

Wet

Freeze

1: No mill, thin, virgin

1:No mill, thin, RAP

1: No mill, thin, virgin

1: No mill, thin, virgin

2: No mill, thin, RAP

1: No mill, thin, virgin

2: Mill, thin, virgin

2: No mill, thin, RAP

3: Mill, thick, RAP

3: Mill, thick, RAP

3: No mill, thin, RAP

3: Mill, thick, RAP

3: Mill, thin, virgin

4: Mill, thin, RAP

4: Mill, thick, RAP

3: Mill, thin, RAP

No-freeze

1: No mill, thin, virgin

1: No mill, thin, virgin

1: Mill, thin, virgin

1: No mill, thin, virgin

2: Mill, thin, virgin

2: No mill, thin, RAP

1: No mill, thin, virgin

2: Mill, thin, virgin

3: No mill, thin, RAP

3: Mill, thick, RAP

3: Mill, thin, RAP

3: Mill, thin, RAP

4: Mill, thick, RAP

3: Mill, thin, RAP

3: No mill, thin, RAP

3: No mill, thin, RAP

Dry

Freeze

1: No mill, thin, virgin

1: Mill, thick, RAP

1: No mill, thin, virgin

1: No mill, thin, virgin

2: Mill, thick, RAP

1: No mill, thin, RAP

2: Mill, thick, RAP

2: Mill, thick, RAP

2: No mill, thin, RAP

1: No mill, thin, virgin

2: Mill, thin, RAP

2: Mill, thin, RAP

4: Mill, thin, RAP

4: Mill, thin, RAP

2: Mill, thin, virgin

2: No mill, thick, RAP

No-freeze

1: No mill, thin, virgin

1: No mill, thin, virgin

1: No mill, thin, virgin

1: No mill, thin, virgin

2: Mill, thick, RAP

2: Mill, thick, RAP

2: Mill, thin, virgin

2: Mill, thin, RAP

2: Mill, thin, RAP

2: Mill, thin, RAP

3: Mill, thin, RAP

3: Mill, thick, RAP

2: Mill, thin, virgin

2: No mill, thin, RAP

4: Mill, thick, RAP

3: Mill, thin, virgin

Table 53. Summary based on long-term rutting performance of SPS-5 pavement structures.

Climate

Traffic/Surface Condition

High

Low

Poor

Fair

Poor

Fair

Wet

Freeze

1: No mill, thin, virgin

1: No mill, thin, virgin

1: No mill, thin, virgin

1: No mill, thin, virgin

2: Mill, thin, virgin

2: Mill, thick, RAP

2: Mill, thin, virgin

2: Mill, thick, RAP

3: Mill, thick, RAP

2: No mill, thick, RAP

3: Mill, thick, RAP

2: Mill, thin, virgin

3: No mill, thick, RAP

4: Mill, thin, virgin

3: No mill, thick, RAP

2: No mill, thick, RAP

No-freeze

1: Mill, thin, virgin

1: No mill, thin, virgin

1: Mill, thin, virgin

1: No mill, thin, virgin

1: No mill, thin, virgin

2: Mill, thin, virgin

1: No mill, thin, virgin

2: Mill, thin, virgin

3: Mill, thin, RAP

3: Mill, thick, RAP

3: Mill, thin, RAP

3: Mill, thick, RAP

4: Mill, thick, RAP

3: Mill, thin, RAP

4: Mill, thick, RAP

3: Mill, thin, RAP

Dry

Freeze

1: No mill, thin, virgin

1: Mill, thick, RAP

1: No mill, thin, virgin

1: No mill, thin, virgin

2: Mill, thick, RAP

1: No mill, thin, virgin

2: Mill, thick, RAP

2: Mill, thick, RAP

2: No mill, thick, virgin

3: No mill, thick, RAP

2: Mill, thin, virgin

3: No mill, thick, RAP

4: Mill, thin, virgin

3: No mill, thick, virgin

2: No mill, thick, virgin

3: No mill, thick, virgin

No-freeze

1: No mill, thin, virgin

1:No mill, thin, virgin

1: No mill, thin, virgin

1: No mill, thin, virgin

2: Mill, thin, virgin

2: Mill, thick, RAP

2: Mill, thin, virgin

2: Mill, thick, RAP

3: Mill, thick, RAP

3: Mill, thin, RAP

3: Mill, thick, RAP

2: Mill, thin, virgin

3: Mill, thin, RAP

3: Mill, thin, virgin

3: Mill, thin, RAP

4: Mill, thin, RAP

Table 54. Summary based on long-term fatigue cracking performance of SPS-5 pavement structures.

Climate

Traffic Surface/Condition

High

Low

Poor

Fair

Poor

Fair

Wet

Freeze

1: Mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thick, virgin

2: Mill, thick, RAP

2: Mill, thick, RAP

2: Mill, thick, RAP

2: Mill, thick, RAP

2: No mill, thick, virgin

2: No mill, thick, virgin

2: Mill, thin, virgin

2: Mill, thin, virgin

4: Mill, thin, virgin

4: Mill, thin, virgin

2: No mill, thick, RAP

2: No mill, thick, RAP

No-freeze

1: Mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thick, virgin

2: Mill, thick, RAP

2: Mill, thick, RAP

2: Mill, thick, RAP

2: Mill, thick, RAP

2: No mill, thick, virgin

2: No mill, thick, virgin

2: Mill, thin, virgin

2: Mill, thin, virgin

4: Mill, thin, virgin

4: Mill, thin, virgin

2: No mill, thick, RAP

2: No mill, thick, RAP

Dry

Freeze

1: Mill, thick, RAP

1: Mill, thick, RAP

1: Mill, thick, RAP

1: Mill, thick, RAP

1: Mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thick, virgin

1: No mill, thick, virgin

1: No mill, thick, virgin

1: No mill, thick, virgin

1: No mill, thick, virgin

-

-

-

-

No-freeze

1: Mill, thick, RAP

1: Mill, thick, RAP

1: Mill, thick, RAP

1: Mill, thick, RAP

1: Mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thick, virgin

1: No mill, thick, virgin

1: No mill, thick, virgin

1: No mill, thick, virgin

1: No mill, thick, virgin

4: Mill, thin, RAP

4: Mill, thin, RAP

4: Mill, thin, RAP

4: Mill, thin, RAP

- Indicates that no preferred treatment was statistically found.

Table 55. Summary based on short-term transverse cracking performance of SPS-5 pavement structures.

Climate

Traffic/Surface Condition

High

Low

Poor

Fair

Poor

Fair

Wet

Freeze

No-freeze

1: Mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thick, virgin

1: No mill, thick, virgin

1: No mill, thick, virgin

1: No mill, thick, virgin

1: No mill, thick, virgin

Dry

Freeze

No-freeze

1: Mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thick, virgin

1: No mill, thick, virgin

1: No mill, thick, virgin

1: No mill, thick, virgin

1: No mill, thick, virgin

- Indicates that no preferred treatment was statistically found.

Table 56. Summary based on long-term transverse cracking performance of SPS-5 pavement structures.

Climate

Traffic/Surface Condition

High

Low

Poor

Fair

Poor

Fair

Wet

Freeze

1: Mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thick, virgin

1: No mill, thick, virgin

2: No mill, thick, virgin

1: No mill, thick, virgin

2: No mill, thick, virgin

3: Mill, thick, RAP

3: Mill, thick, RAP

3: Mill, thick, RAP

3: No mill, thick, RAP

4: No mill, thick, RAP

No-freeze

1: Mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thick, virgin

1: No mill, thick, virgin

2: No mill, thick, virgin

1: No mill, thick, virgin

2: No mill, thick, virgin

 - 

3: Mill, thick, RAP

3: Mill, thick, RAP

3: Mill, thick, RAP

 - 

3: No mill, thick, RAP

4: No mill, thick, RAP

Dry

Freeze

1: Mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thick, virgin

1: No mill, thick, virgin

2: Mill, thick, RAP

1: No mill, thick, virgin

2: Mill, thick, RAP

3: Mill, thick, RAP

2: No mill, thick, virgin

3: Mill, thick, RAP

2: No mill, thick, virgin

3: Mill, thin, RAP

4: Mill, thin, RAP

3: No mill, thick, RAP

4: No mill, thick, RAP

No-freeze

1: Mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thick, virgin

1: No mill, thick, virgin

2: No mill, thick, virgin

1: No mill, thick, virgin

2: No mill, thick, virgin

3: Mill, thick, RAP

3: Mill, thick, RAP

3: Mill, thick, RAP

3: Mill, thick, RAP

3: Mill, thin, RAP

4: Mill, thin, RAP

3: No mill, thick, RAP

4: No mill, thick, RAP

- Indicates that no preferred treatment was statistically found.

Table 57. Summary based on long-term longitudinal cracking performance of SPS-5 pavement structures.

Climate

Traffic/Surface Condition

High

Low

Poor

Fair

Poor

Fair

Wet

Freeze

1: Mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thin, virgin

1: Mill, thin, virgin

1: Mill, thin, virgin

1: Mill, thin, virgin

3: Mill, thick, RAP

3: Mill, thick, RAP

3: Mill, thick, RAP

3: Mill, thick, RAP

3: Mill, thin, RAP

3: Mill, thin, RAP

3: Mill, thin, RAP

3: Mill, thin, RAP

No-freeze

1: Mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thin, virgin

1: Mill, thin, virgin

1: Mill, thin, virgin

1: Mill, thin, virgin

3: Mill, thick, RAP

3: Mill, thick, RAP

3: Mill, thick, RAP

3: Mill, thick, RAP

3: Mill, thin, RAP

3: Mill, thin, RAP

3: Mill, thin, RAP

3: Mill, thin, RAP

Dry

Freeze

1: Mill, thick, RAP

1: Mill, thick, virgin

1: Mill, thick, RAP

1: Mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thin, virgin

1: Mill, thick, virgin

1: Mill, thin, virgin

1: Mill, thin, RAP

3: Mill, thick, RAP

1: Mill, thin, RAP

3: Mill, thick, RAP

1: Mill, thin, virgin

3: Mill, thin, RAP

1: Mill, thin, virgin

3: Mill, thin, RAP

No-freeze

1: Mill, thick, RAP

1: Mill, thick, virgin

1: Mill, thick, RAP

1: Mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thin, virgin

1: Mill, thick, virgin

1: Mill, thin, virgin

1: Mill, thin, RAP

3: Mill, thick, RAP

1: Mill, thin, RAP

3: Mill, thick, RAP

1: Mill, thin, virgin

3: Mill, thin, RAP

1: Mill, thin, virgin

3: Mill, thin, RAP

The summary tables show only alternatives in which statistically significant differences in performance were identified with the Friedman test analysis. They were created by considering the best alternatives from each analysis performed in the dataset. These selected rehabilitation alternatives were then grouped for each combination of site conditions.

According to the summary tables, the following conclusions were made when roughness performance was considered:

The analysis of rutting indicated the following:

The analysis of fatigue cracking indicated the following:

The analysis of transverse cracking indicated the following:

The analysis of longitudinal cracking indicated the following:

The summary tables presented in this section, along with a complete set of tables with descriptive statistics for all the analyses conducted, are provided in appendix C.

Effect of Design and Contruction Features and Site Conditions on Response of Rehabilitated Flexible Pavements

The study to evaluate the impact of design features and site conditions on response of rehabilitated flexible pavements followed the same approach used in the analysis of performance. FWD deflection data are available in the LTPP database and used as the response measure of the pavement structure. Deflections in the wheel path of the LTPP lane were used as indicators of structural response.

The same analysis sequence applied to the study of performance was applied to the study of response. The first task was to evaluate individual sites and how the pavement structural responses were affected by each rehabilitation strategy and its design features. Since the analysis was carried out for each site separately, the deflection values were not adjusted for temperature, and the FWD survey was assumed to take place under similar conditions for all sections of a given site. There were small variations in surface temperature throughout the day, but they were considered minor and unlikely to significantly affect the total pavement stiffness. The variation of deflections over time and over the seasons throughout the year was used as input for the repeated ANOVA.

Individual Analysis of SPS-5 Sites

Summary tables containing the results of the statistical analysis for pavement response for each site are presented in appendix A.

A compilation of results was created by identifying the number of sites in which design features had provided differences in response throughout the site's service life. For example, if sections with thick overlays had lower deflection measurements than thin overlay sections in one site, the thick overlay was marked for this site, and the process was repeated for all sites. Figure 32 summarizes the percentage of sites in which differences in response were identified.

This bar graph shows a comparison of percentages of Specific Pavement Study (SPS)-5 sections with lower maximum deflection based on a repeated measures analysis of variance (ANOVA) experiment. The design/construction feature is shown on the x-axis, and the percentage of SPS-5 sites is shown on the y-axis. The x-axis is divided into three parts: overlay thickness, mix type, and milling. Each one of these parts is further divided into three subdivisions. While none of the sites had thin overlay work done, 61 percent accounted for thick overlays, and the remaining 39 percent had no difference in overlay thickness. Reclaimed asphalt pavement mix type accounted for 6 percent of the sites, 17 percent corresponded to the virgin mix type, and 78 percent of the sites had no difference in mix type. A total of 28 percent of the sites had milling work done, 6 percent had no milling work done, and the remaining 67 percent had no differences in milling.

Figure 32. Graph. Percentage of SPS-5 sections with lower maximum deflection based on repeated measures ANOVA results.

The key findings from the assessment of individual sites were as follows:

Consolidated Analysis

The consolidated analysis involved the simultaneous analysis of all sites in the SPS-5 experiment using the Friedman test. A follow-up analysis was performed by grouping the data by site condition to evaluate the impact of design features and site conditions for short-term and
long-term performance. WD was the parameter used for this analysis.

After the data were processed and verified for quality, WD was computed for each section in each SPS-5 site. The values are shown in table 58. The Friedman test used WD to create the ranking of performance from the lowest to the highest value for each site in the dataset. Ranking statistics for each type of section were then used to calculate the chi-square value, and this parameter was used to determine if statistical differences existed among the performance rankings of the sections.

The Friedman test output for maximum deflection in the wheel path is provided in figure 33, and it presents the average and standard deviation values for the WD maximum deflection for each type of rehabilitation strategy among all sites analyzed. The results indicate that there were at least two sections with statistically different structural responses (p < 0.0001, ANOVA chi-square = 78.737). Therefore, the ranking of sections (i.e., rehabilitation strategies) for structural response can be created using Friedman paired analyses. The results of the paired analyses were used to define a practical ranking for structural response (see table 59).

Table 58. WD-deflection values at the center of the load for SPS-5 sites.

Section

Experiment Design

Sites (State Codes)/Average Deflection WD Values (microns)

Mill

Mix

Thickness (mm)

1

4

6

8

12

13

23

24

27

28

29

30

34

35

40

48

81

83

0501

No

None

   

374

271

249

   

174

164

323

202

285

 

198

333

213

 

394

374

0502

No

RAP

51

264

292

319

260

190

102

134

218

250

270

207

465

214

241

168

139

465

257

0503

No

RAP

127

115

196

151

197

144

62

107

157

189

209

112

383

33

139

158

149

272

182

0504

No

Virgin

127

130

77

131

155

134

69

106

164

197

143

109

317

104

175

203

148

272

186

0505

No

Virgin

51

286

256

348

275

186

85

132

265

261

198

139

389

131

271

205

112

273

304

0506

Yes

Virgin

51

156

112

271

319

182

75

141

161

243

209

144

420

37

294

259

134

384

293

0507

Yes

Virgin

127

168

56

167

288

125

63

106

162

177

126

116

292

36

222

145

110

292

180

0508

Yes

RAP

127

162

99

164

205

128

58

108

168

177

179

84

321

30

157

232

122

217

207

0509

Yes

RAP

51

179

202

330

240

166

90

145

190

255

212

153

540

35

217

245

138

324

320

1 m = 0.039 mil
1 inch = 25.4 mm
Note: Higher WD values indicate higher maximum deflection for FWD measurements for pavements. Blank cells indicate data are not available.

This graph shows a bar plot of weighted distress (WD) maximum deflection at a wheel path of a Long-Term Pavement Performance (LTPP) lane in Specific Pavement Study (SPS)-5 sections for short-term analysis. The x-axis shows nine SPS-5 sections ((0501, 0502, 0503, 0504, 0505, 0506, 0507, 0508, and 0509), and the y-axis shows deflection values in microns. Mean values are represented by grey dots, and black vertical bars represent the mean plus or minus 1 times the standard deviation. The nine sections have the following mean, high, and low values of maximum deflection: 275, 386, and 165 microns (10.73, 15.05, and 6.44 mil); 248, 345, and 150 microns (9.67, 12.46, and 5.89 mil); 164, 242, and 87 microns (6.40, 9.44, and 3.39 mil); 157, 220, and 94 microns (6.12, 8.58, and 3.67 mil); 229, 314, and 144 microns (8.93, 12.25, and 5.62 mil); 213, 318, and 108 microns (8.31, 12.40, and 4.21 mil); 157, 234, and 80 microns (6.12, 9.12, and 3.12 mil); 157, 226, and 87 microns; and 221, 333, and 110 microns (8.62, 12.99, and 4.29 mil).

1 Mum = 0.039 mil

Figure 33. Graph. Maximum deflection (wheel path of LTPP lane) WD values at SPS-5 sites.

Table 59. Ranking of rehabilitation strategies for structural response at SPS-5 sites.

Statistical Relevance (Y/N)

Deflection

Y

p < 0.0001

Ranking
(if relevant)

Ranking

Strategy

1 (Lowest)

Mill, thick, virgin

2

Mill, thick, RAP

2

No mill, thick, RAP

2

No mill, thick, virgin

5

Mill, thin, virgin

6

No mill, thin, virgin

6

Mill, thin, RAP

6

No mill, thin, RAP

6

None


The results in table 59 and figure 29 support the findings from the analysis of individual sites that suggest that thick overlays provided lower maximum deflections compared to alternatives with thin overlays. The ranking also shows that there were no significant differences in response for strategies using either virgin or RAP mix overlays, and they are intertwined in the ranking. The results further suggest that milling prior to overlay may reduce the level of pavement response, as strategies with this design feature were ranked among those with lower deflections for FWD measurements.

Influence of Site Condition

Similar to the pavement performance analysis, the influence of site conditions was determined by three variables: (1) pavement surface condition prior to rehabilitation, (2) climate, and (3) traffic levels. The analysis followed the same steps described in the previous section. Rankings of rehabilitation strategies were developed for each group of sites using descriptive statistics and the paired analyses from the Friedman test when statistical differences in performance were found in the first step. The results were summarized in tables and are presented in appendix C.

After a careful assessment of the results from the analyses performed for each group, summary tables including all possible combinations of site conditions were assembled. Table 60 presents the results, and the alternatives with lower deflections with statistical relevance are shown in each cell. The number before the treatment indicates its ranking among all alternatives.

Table 60 can be used as a practical guide to help select rehabilitations strategies that provide low deflection values. For example, if the section is located in a wet freeze region and the pavement is in fair surface condition with low traffic levels, an alternative with a thick RAP overlay without milling is indicated. However, if the same section is in poor condition, the alternative indicated is one with milling and a thick virgin overlay.

These summary tables provide clear information to identify the rehabilitation treatment with the lowest and highest pavement responses based on site conditions, and the influence of different site condition can be determined by observing the ranking of treatments for each condition.

Table 60. Summary based on structural response (maximum deflection in the wheel path of the LTPP lane) of SPS-5 pavement structures.

Climate

Traffic/Surface Condition

High

Low

Poor

Fair

Poor

Fair

Wet

Freeze

1: Mill, thick, RAP

1: Mill, thick, RAP

1: Mill, thick, virgin

1: No mill, thick, RAP

1: No mill, thick, RAP

1: No mill, thick, RAP

1: No mill, thick, RAP

2: Mill, thick, RAP

1: No mill, thick, virgin

3: No mill, thick, virgin

1: No mill, thick, virgin

2: Mill, thick, virgin

4: Mill, thick, virgin

4: Mill, thick, virgin

4: Mill, thick, RAP

2: No mill, thick, virgin

No-freeze

1: No mill, thick, virgin

1: No mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thick, virgin

2: Mill, thick, virgin

2: Mill, thick, RAP

1: No mill, thick, virgin

1: No mill, thick, virgin

3: Mill, thick, RAP

2: Mill, thick, virgin

3: No mill, thick, RAP

3: No mill, thick, RAP

3: No mill, thick, RAP

2: No mill, thick, RAP

4: Mill, thick, RAP

4: Mill, thick, RAP

Dry

Freeze

1: Mill, thick, RAP

1: Mill, thick, RAP

1: Mill, thick, virgin

1: No mill, thick, RAP

1: No mill, thick, RAP

1: No mill, thick, RAP

1: No mill, thick, RAP

2: Mill, thick, RAP

1: No mill, thick, virgin

3: No mill, thick, virgin

1: No mill, thick, virgin

2: Mill, thick, virgin

4: Mill, thick, virgin

4: Mill, thick, virgin

4: Mill, thick, RAP

2: No mill, thick, virgin

No-freeze

1: No mill, thick, virgin

1: No mill, thick, virgin

1: Mill, thick, virgin

1: Mill, thick, virgin

2: Mill, thick, virgin

2: Mill, thick, RAP

1: No mill, thick, virgin

1: No mill, thick, virgin

3: Mill, thick, RAP

2: Mill, thick, virgin

3: No mill, thick, RAP

3: No mill, thick, RAP

3: No mill, thick, RAP

2: No mill, thick, RAP

4: Mill, thick, RAP

4: Mill, thick, RAP


Table 60 shows only alternatives in which statistically significant differences in WD-maximum deflection were found. The table was created by considering the alternatives with the lowest responses for each analysis performed with the dataset. These selected rehabilitation alternatives were then grouped for each combination of site conditions. According to the summary table, the following conclusions were made with respect to structural response:

The summary table presented in this section, along with a complete set of tables with descriptive statistics for all the analyses performed, is provided in appendix C.

Relationhip Between Structural Respones Immediately After Rehabilitation and Future Performance

There have been many attempts to find direct relationships or models to predict performance based on the structural response of the pavement to loading immediately after construction or rehabilitation. These relationships are not completely straightforward, and deriving them accurately using mechanistic theory can be difficult. The objective of this study was to identify trends in the relationship between response measured after the rehabilitation and the observed performance in subsequent years of the pavement's service life. If these trends can be identified, they will provide important information and guidance on what to expect for a pavement's performance as a result of rehabilitation strategies that yield to a certain level of structural response.

This study concentrated on evaluating FWD maximum deflections measured under the center of the load against the average pavement performance during the service life of SPS-5 sites. WD was once again used as the performance measure. Each response was evaluated against all distresses previously used in this study. Only long-term performance was used for this analysis, indicating performance data of 5 years or more.

The SPS-5 experiment had sites across the United States under different climatic zones, subgrade types, and traffic loads. Consequently, the results from this study could be impacted by in situ conditions. The alternative to circumvent this problem was to normalize the data in each site by a common factor. The rehabilitation strategy selected as the normalization factor had a thin virgin mix overlay without milling, and it was selected because the control section was not available or was eliminated from the surveys for some of the SPS-5 sites. Response and performance measured for thin virgin mix overlaid sections were used to normalize the data of the remaining sections in each site. Normalized values were computed according to the equation in figure 34.

Parameter subscript Normalized equals Parameter subscript section divided by Parameter subscript 0602.

Figure 34. Equation. Parameter subscript normalized.

The relationship between performance and deflection was verified using Pearson's linear coefficient, r. The t-Student's test was used to test the null hypothesis stating that no correlation existed between performance and deflection (r = 0). A potential relationship between performance and deflection was confirmed if the null hypothesis was rejected with 95 percent confidence (p ≤ 0.05).

The trend between roughness and maximum deflection is shown in figure 35. It suggests that roughness, measured by IRI, is poorly related to the deflection values measured after the rehabilitation of the pavement structure (p = 0.029).

This scatter plot shows normalized long-term weighted distress (WD)-roughness versus normalized maximum deflection. Normalized maximum deflection is on the x-axis, and the normalized roughness is on the y-axis. A solid line with a positive slope of 9 percent connects the lowest and highest normalized maximum deflection at 0.4 and 2.75. The individual points are represented by white diamond markers and are spread evenly around the solid line at the center of the plot with the majority of normalized roughness and deflection values ranging from 0.5 to 1.5. In addition, there is a high concentration of data points near the solid line. The trend between roughness and maximum deflection suggests that roughness, as measured by International Roughness Index values, is poorly related to the deflection values measured after the rehabilitation of the pavement structure.

Figure 35. Graph. Normalized long-term WD-roughness versus normalized maximum deflection measured after rehabilitation in SPS-5 sites.

Similarly, the results in figure 36 suggest that maximum deflection after the pavement's rehabilitation cannot provide good qualitative information about the rutting performance predictions (p = 0.296). This observation may be contrary to what is expected. Rutting is a load-related distress, and because deflection measures the pavement's response to load applications, it seems intuitive that a positive trend might exist. However, instant deflections as measured by FWD tests were more likely to capture instantaneous elastic response of the pavement structure. Rutting is a plastic deformation more likely to occur in unbound aggregate layers and AC at warm temperatures. In rehabilitated flexible pavements, most of the permanent deformation occurred in the overlay. As temperature increases, AC behaves less like a time-dependent elastic material and more like a time-dependent plastic material. The instantaneous FWD deflections cannot be associated with the material's behavior impacting rutting performance.

This scatter plot shows normalized long-term weighted distress (WD)-rutting versus normalized maximum deflection. Normalized maximum deflection is on the x-axis, and the normalized rutting is on the y-axis. A solid line with a negative slope of 10 percent connects the lowest and highest normalized maximum deflection at 0.4 and 2.75. The individual points are represented by white diamond markers and are spread evenly around the solid line with the majority of normalized roughness and deflection values ranging from 0.5 to 1.3. In addition, there is a high concentration of data points near the solid line. The trend suggests that the maximum deflection after the pavement’s rehabilitation cannot provide good qualitative predictive information about the rutting performance.

Figure 36. Graph. Normalized long-term WD-rutting versus normalized maximum deflection measured after rehabilitation in SPS-5 sections.

Contrary to expectations, the data in figure 37 indicate that no statistically significant trend was found between fatigue cracking performance and maximum deflection (p = 0.565). Conversely, the data in figure 38 suggest that higher values of transverse cracking are expected when the pavement has higher deflections (p = 0.001). The trend in figure 39 suggests that longitudinal cracking is not related to deflections measured under the center of the load on FWD tests but just marginally (p = 0.058).

This scatter plot shows normalized long-term weighted distress (WD)-fatigue cracking versus normalized maximum deflection. Normalized maximum deflection is on the x-axis, and the normalized fatigue cracking is on the y-axis. A solid line with a positive slope of 60 percent connects the lowest and highest normalized maximum deflection at 0.4 and 2.75. The individual points are represented by white diamond markers, and the majority of the points are below the solid line, having normalized fatigue cracking values ranging from approximately 0 to 0.7 and deflection values ranging from 0.5 to 1.5. In addition, there is a high concentration of data points near the x-axis at deflection values ranging from 0.5 to 0.9.

Figure 37. Graph. Normalized long-term WD-fatigue cracking versus normalized maximum deflection measured after rehabilitation in SPS-5 sections.

This scatter plot shows normalized long-term weighted distress (WD)-transverse cracking versus normalized maximum deflection. Normalized maximum deflection is on the x-axis, and the normalized transverse cracking is on the y-axis. A solid line with a positive slope of 47 percent connects the lowest and highest normalized maximum deflection at 0.4 and 2.75. The individual points are represented by white diamond markers, and they are distributed both above and below the solid line. The majority of points have normalized transverse cracking values ranging from approximately 0 to 1.5 and deflection values ranging from 0.5 to 1.5. In addition, there is a high concentration of data points near the x-axis at deflection values ranging from 0.5 to 1.0.

Figure 38. Graph. Normalized long-term WD-transverse cracking versus normalized maximum deflection measured after rehabilitation in SPS-5 sections.

This scatter plot shows normalized long-term weighted distress (WD)-longitudinal cracking versus normalized maximum deflection. Normalized maximum deflection is on the x-axis, and the normalized longitudinal cracking is on the y-axis. A solid line with a negative slope of 42 percent connects the lowest and highest normalized maximum deflection at 0.4 and 2.2. The individual points are represented by white diamond markers, and they are evenly distributed both above and below the solid line with the majority of points having normalized longitudinal cracking values ranging from approximately 0 to 1.5 and deflection values ranging from 0.5 to 1.5. In addition, there is a high concentration of data points near the x-axis at deflection values ranging from 0.5 to 1.0.

Figure 39. Graph. Normalized long-term WD-longitudinal cracking versus normalized maximum deflection measured after rehabilitation in SPS-5 sections.

Some of the distresses had clear trends with responses that agreed with the conventional wisdom and expectations of relationships between performance and response. However, none of the trends were strong enough to suggest a direct correlation between performance and response as measured by maximum deflection. These plots only suggested that deflection could be used to infer qualitative assessments of future performance of some distresses but not to quantitatively predict performance.

 


The Federal Highway Administration (FHWA) is a part of the U.S. Department of Transportation and is headquartered in Washington, D.C., with field offices across the United States. is a major agency of the U.S. Department of Transportation (DOT).
The Federal Highway Administration (FHWA) is a part of the U.S. Department of Transportation and is headquartered in Washington, D.C., with field offices across the United States. is a major agency of the U.S. Department of Transportation (DOT). Provide leadership and technology for the delivery of long life pavements that meet our customers needs and are safe, cost effective, and can be effectively maintained. Federal Highway Administration's (FHWA) R&T Web site portal, which provides access to or information about the Agency’s R&T program, projects, partnerships, publications, and results.
FHWA
United States Department of Transportation - Federal Highway Administration