U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
2023664000
Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations
This report is an archived publication and may contain dated technical, contact, and link information 

Publication Number: FHWARD03049
Date: November 2005 

Improving Pavements With LongTerm Pavement Performance: Products for Today and TomorrowPaper 3. Analysis of Influences on AsBuilt Pavement Roughness in Asphalt OverlaysC.M. Raymond^{1}, R. Haas^{2}, S.L. Tighe^{3}, and L. Rothenburg ^{4} ABSTRACTPavement roughness immediately after construction is a key measure of quality. The use of smoothness specifications requires an understanding of the influences on asbuilt roughness for both transportation agencies and contractors. This paper uses data from the LongTerm Pavement Performance (LTPP) program to examine four factors and determine their effects on the asbuilt roughness of a pavement; these factors are: the extent of surface preparation before resurfacing; overlay thickness; type of overlay material; and pavement roughness before resurfacing. Various statistical procedures (including paired data analyses, regression analyses, and a repeated measures analysis) are performed to investigate these effects and any interactive effects. The extent of surface preparation, overlay thickness and pavement roughness before resurfacing are determined to have a statistically significant effect (at a 95 percent significance level) on the asbuilt roughness of a pavement either directly or interactively with another variable. The overlay mix type is determined not to have an influence on asbuilt pavement roughness. Data from the Canadian LongTerm Pavement Performance (CLTPP) program is used to validate the results for overlay thickness and pavement roughness before resurfacing. A series of prediction equations are also developed to allow for estimating the asbuilt roughness of a pavement under various conditions. Pavement designers, construction engineers, and contractors should understand the effects that influence the asbuilt roughness of a pavement so that they can maximize their designs, smoothness specifications, and/or bidding of contracts with smoothness specifications. INTRODUCTIONPavement roughness is the primary measure of public satisfaction with the highway system and, accordingly, the asbuilt roughness of a pavement immediately after construction is a key quality measure. Asbuilt pavement roughness has also been shown to affect the longterm performance of a pavement (Raymond, 2001; TAC, 2002). Consequently, most transportation agencies have incorporated asbuilt roughness requirements into their acceptance criteria. These roughness requirements are generally termed “smoothness specifications.” Although there has been common acceptance of smoothness specifications in Canada and the United States, the pay adjustments and smoothness requirements vary considerably among transportation agencies (Schmitt et al., 1998; TAC, 1999). Smoothness specifications are an effective means of improving the asbuilt smoothness of a pavement (Smith 1997, McGhee 2000). The incentive/disincentive provisions encourage contractors to achieve a smooth pavement surface, which can include purchasing new equipment, improving grade controls and additional training of staff. In developing a smoothness specification, it is important that transportation agencies understand not only the effect of asbuilt pavement roughness on longterm performance but also the factors that affect asbuilt pavement roughness. Contractors must also understand the factors that will affect their ability to construct a smooth pavement. This knowledge is important for competitively bidding contracts with payment adjustments based on asbuilt pavement roughness. This paper examines four design factors for their effect on the asbuilt pavement roughness of asphalt overlays constructed over existing asphalt pavements: the degree of surface preparation before overlay; overlay thickness; type of overlay material; and pavement roughness before resurfacing. The influence of these factors on asbuilt pavement roughness can be an issue of disagreement between transportation agencies and contractors. A number of other factors not considered in the analysis include the use of a material transfer vehicle; the degree of smoothness required by the owner’s smoothness specification; the magnitude of incentives and/or disincentives provided in the owner’s smoothness specification; the contractor’s attitude and capability to achieve a smooth pavement; and the operational constraints related to the project (such as traffic staging, location of intersections, and extent of night paving). RELATED STUDIESPrevious research has provided different conclusions regarding the influences on asbuilt roughness. An earlier study sponsored by the Federal Highway Administration (FHWA) was unable to identify any statistical difference in asbuilt roughness from the four factors mentioned above (surface preparation, overlay thickness, overlay material and roughness before resurfacing) (Perera, Byrum, and Kohn, 1998; Perera, and Kohn, 1999). Research by the Virginia Department of Transportation concluded that asbuilt pavement roughness is influenced by the pavement roughness before resurfacing, the functional classification of the roadway, and the application of a smoothness specification (McGhee, 2000). The same research also reports that the number of additional structural layers, the surface mix type, and pavement milling before overlay do not influence asbuilt pavement roughness. DESCRIPTION OF DATA SOURCESThe data for the analysis is from the LTPP program established as part of the Strategic Highway Research Program (SHRP) that began in 1987. It consists of numerous test sections located throughout the United States and Canada. Of interest to this research are the SPS5 experiments. Each SPS5 site consists of nine 150meter (m) (492feet (ft)) sections (FHWA, 96). Eight sections are experimental, while one section, section 501, is a control section with no specific treatment except for routine maintenance. Because the control section provides no information related to this examination of asbuilt roughness, it is not incorporated into the analysis. The eight experimental sections for each project were setup as shown in table 1 with different treatments for extent of surface preparation, type of overlay material, and overlay thickness (not including replacement of milled material). The primary difference between basic and intensive surface preparation is the amount of milling that was performed for the section. Pavement roughness is measured in terms of International Roughness Index (IRI) using a K.J. Law Profilometer. Data for this analysis are from the DataPave 3.0 database. The tables entitled TST_L05A, SPS5_LAYER, and MON_PROFILE_MASTER were used for this research. Validation of SPS5 analysis was performed based on data from the CLTPP database supplied by the Canadian Strategic Highway Research Program (CSHRP). Table 1. SPS5 experimental test sections
1 mm = 0.039 inch * Design overlay thickness does not include replacement of milled material ASBUILT ROUGHNESS PAVEMENT SECTIONSAsbuilt pavement roughness measurements were taken for each section immediately after resurfacing for the 17 sites and are shown in table 2. Roughness is quantified based on the IRI. The table also provides the average asbuilt IRI for each site, the average asbuilt IRI determined for each section treatment, corresponding standard deviations, and coefficients of variations. Five of the asbuilt IRI values for Maine were missing from the database, requiring extrapolation from the IRI data 2 years following construction. The extrapolation of these asbuilt roughness values is based on the average change in IRI for the other three sections during this time period. An examination of the average asbuilt roughness of the various sections indicates that most averages are close to the overall average of 0.91 m/km (58 inches per mile (inches/mi). Section 502 (basic surface preparation, thin overlay, and recycled material) and section 508 (intensive surface preparation, thick overlay, and recycled material) have the greatest deviation from the overall site average. Section 502 has a high average asbuilt roughness of 0.97 m/km (61 inches/mile); section 508 has a low asbuilt roughness of 0.86 m/km (54 inches/mile). The theory that extensive surface preparation and thick overlays provide the smoothest asbuilt pavements would seem to be consistent with the average roughness values for sections 502 and 508. What remains unexplained is why the corresponding sections with different overlay material do not provide similar asbuilt roughness values. Because the type of overlay material is not considered to affect asbuilt roughness, the asbuilt roughness of sections 505 (basic surface preparation, thin overlay, and virgin material) and 507 (intensive surface preparation, thick overlay, and virgin material) should correspond with the averages for sections 502 and 508. The average asbuilt roughness for section 505 is slightly greater than the overall average and the average asbuilt roughness for section 507 is slightly lower than the overall average. The effect of the various treatments on asbuilt roughness is examined later in this paper. Table 2. Asbuilt IRI measurements
1 m/km = 63.36 inches/mile PRIOR ROUGHNESS OF PAVEMENT SECTIONSRoughness measurements for each section before resurfacing are shown in table 3. The table also provides the average prior IRI for each site, the average prior IRI determined for each section treatment, and the corresponding standard deviations and coefficients of variations. Roughness measurements before resurfacing are not available for the Manitoba site. The New Mexico site has prior roughness measurements for only two sections. Because there is considerable variation in the two roughness values (1.74 m/km (110 inches/mile) and 3.04 m/km (193 inches/mile), the measurements were not incorporated into the analyses (i.e., the New Mexico site was evaluated without prior roughness measurements). Two other sites, Colorado and Minnesota, were missing roughness measurements before resurfacing for at least one section. These sites were evaluated based on the information available, with the average site roughness being substituted for missing values. Table 3. IRI measurements before resurfacing
1 m/km = 63.36 inches/mile The average roughness before resurfacing for the sites is 1.68 m/km (106 inches/mile). Section 506 has the lowest average roughness before resurfacing of 1.51 m/km (96 inches/mile) and section 502 has the highest, of 1.83 m/km (116 inches/mile). If the IRI before resurfacing is related to the asbuilt roughness of a pavement, the low prior IRI for sections 505 and 508 may help explain why their average asbuilt roughness values are lower than for their corresponding sections with different overlay material (i.e., sections 502 and 507). Similarly, the high prior IRI of section 509 may help explain why its average asbuilt roughness is higher than section 506, the corresponding section with different overlay material. INVESTIGATION OF INFLUENCES ON ASBUILT ROUGHNESSThe influences on asbuilt pavement roughness are investigated using several statistical techniques. The effects of surface preparation, overlay thickness, and overlay material on asbuilt roughness are investigated based on a series of paired analyses. The paired analyses consider the various effects based on three data sets: pavement sites with a high pavement roughness before resurfacing (i.e., IRI greater than 1.5 m/km (95 inches/mile)); pavement sites with a low pavement roughness before resurfacing (i.e., IRI less than 1.5 m/km (95 inches/mile)); and all sites. Following the paired data analysis, a regression analysis is preformed to examine the influence of pavement roughness before resurfacing on the asbuilt roughness of a pavement. This analysis is required to examine pavement roughness before resurfacing as a quantitative variable. A third statistical tool, a repeated measures analysis, is used to examine the interactive effects of surface preparation, overlay thickness, overlay material and pavement roughness before resurfacing on the asbuilt roughness of a pavement. Next, regression analyses are performed on data from the CLTPP program to validate some of the conclusions reached from the SPS5 data. Lastly, regression analyses are performed on the SPS5 data to develop four prediction equations for estimating the asbuilt roughness of a pavement. A 95 percent statistical significance level was selected for all analyses as the criteria to identify the presence of a significant relationship. SURFACE PREPARATIONThe SPS5 test sections involved two types of surface preparation, basic and intensive (Perera, Byrum, and Kohn, 1998; Perera, and Kohn, 1999). Intensive surface preparation consists of milling theexisting asphalt followed by patching distressed areas and crack sealing. Basic surface preparation is intended to consist of patching severely distressed areas and potholes, and placement of a leveling course for ruts greater than 12 mm (0.47 inches) in depth, although the construction information indicates that 5 of the 17 sites had milling performed for the sections designated for basic surface preparation. Where milling was performed, the depth of replacement material was not counted as the part of the overlay thickness specified for the section (Perera, Byrum, and Kohn, 1998; Perera, and Kohn, 1999). The primary difference in the two extents of surface preparation is whether or not milling of the existing asphalt pavement is performed. Milling is a technique than planes off the existing asphalt to a predetermined depth. The Asphalt Institute (AI) reports that milling can often remove roughness better than an asphalt paver because of its greater ability to remove a variable thickness of material (AI, 1989). A paired comparison of the asbuilt roughness of pavement sections with different extents of surface preparation is shown in table 4, which is separated into three rows. The first row presents data from pavement sites with a high pavement roughness before resurfacing (i.e., IRI greater than 1.5 m/km (95 inches/mile)); the second row presents data from pavement sites with a low pavement roughness before resurfacing (i.e., IRI less than 1.5 m/km (95 inches/mile)); and the third row presents the complete set of data. The results indicate that there is an overall significant difference (at a twotailed significance level of 95 percent) in the asbuilt roughness of pavement sections with basic and extensive surface preparation. This statistical difference is evident with both the complete data set and for the data with a high roughness before resurfacing. No significant statistical difference was found for the data from sites with a low roughness level before resurfacing. In fact, the mean difference in asbuilt roughness for these sites is essentially zero, indicating no difference in asbuilt roughness for the different levels of surface preparation. This lack of difference may be because a low prior pavement roughness does not provide an adequate opportunity for the effect of surface preparation to be demonstrated. The paired ttest analyses indicates that on average, there is a 0.040 m/km (2.5 inches/mile) lower asbuilt IRI for pavements constructed with intensive surface preparation as compared with pavements constructed with basic surface preparation (i.e., intensive surface preparation contributes to smoother asbuilt pavements). When only pavements with a high level of roughness before resurfacing are examined, the average difference in asbuilt IRI increases to 0.081 m/km (5.1 inches/mile). Table 4. Paired data analysis of the effect of surface preparation on asbuilt IRI
1 m/km = 63.36 inches/mile OVERLAY THICKNESSOverlay thickness is considered to be a contributing factor to the asbuilt roughness of a pavement. Thicker overlays provide a contractor more opportunity to reduce the roughness of a pavement section. This improvement can be attributed to the fact that thicker overlays typically involve more lifts, which allow for incremental improvements in pavement smoothness. The opportunity for improving smoothness with more overlay lifts is related to the operational constraints of an asphalt paver. One objective of an asphalt paver is to produce a smooth asphalt mat behind the screed through the use of proper paving techniques and grade controls. Should a perfectly smooth asphalt layer be placed over a previous pavement deviation, the subsequent compaction of the asphalt mix by rollers will “reflect” a portion of the roughness into the new asphalt layer, as illustrated in figure 1. Asphalt behind a paving screed is at approximately 70 to 80 percent of theoretical maximum density, while the final asphalt compaction is generally around 94 percent of theoretical maximum density (USACE, 1991). This additional consolidation can result in approximately 20 percent of the previous deviations being reflected into the new asphalt lift. It should be noted that the overlay thickness values presented earlier do not include the replacement of milled material for sections where pavement milling was performed. Figure 1. Limitation on achieving a smooth pavement with a single lift of asphalt A comparison of the asbuilt roughness of thin overlays in relation to thick overlays is shown in table 5. The results of this analysis indicate no statistically significant differences in asbuilt roughness compared to overlay thickness. Although the mean difference in IRI between thin and thick overlays is 0.033 m/km (2.1 inches/mile), the twotailed pvalue is 35 percent, which indicates no statistically significant relationship (at a twotailed significance level of 95 percent) is present. The mean difference in IRI between thin and thick overlays is 0.055 m/km (3.5 inches/mile) for pavements with a high pavement roughness before resurfacing with a twotailed pvalue of 12.1 percent. Although not statistically significant at a twotailed significance level of 95 percent, when the analysis is considered from a onesided approach, the pvalue becomes 6.05 percent, which is close to being statistically significant at a 95 percent significance level. The mean difference in IRI between thin and thick overlays is 0.063 m/km (4.0 inches/mile) for pavements with a low pavement roughness before resurfacing. The difference is not statistically significant. However, the fact that pavements with a high pavement roughness before resurfacing have a positive difference in IRI and pavements with a low pavement roughness before resurfacing have a negative difference suggests a possible interactive effect with prior roughness. This interactive effect is examined later in this paper. Table 5. Paired data analysis of the effect of design overlay thickness on asbuilt IRI
1 m/km = 63.36 inches/mile COMPARISON OF OVERLAY MATERIALThe type of overlay material (i.e., recycled or virgin) has not traditionally been considered an influencing factor in pavement roughness, but is investigated to confirm its lack of effect. The results of a comparison of the effect of overlay material on asbuilt roughness are shown in table 6. The results indicate that there are essentially no statistical differences in asbuilt roughness between recycled and virgin mixtures. All of the pvalues are close to a value of one, indicating no relationship with asbuilt roughness. Table 6. Paired data analysis of the effect of overlay material on asbuilt IRI
1 m/km = 63.36 inches/mile COMPARISON OF PAVEMENT ROUGHNESS BEFORE RESURFACINGAs outlined previously, the SPS5 experimental sections were setup as a factorial experiment based on extent of surface preparation, overlay thickness, and type of overlay material. Because the effect of pavement roughness before resurfacing cannot be investigated in the same manner as the factorial variables, a regression analysis was performed to determine whether the asbuilt roughness of a pavement is affected by the pavement roughness before resurfacing. A plot of the regression analysis is shown in figure 2, and the analysis of variance statistics are shown in table 7. The pvalue is 3.5 percent, which indicates that the pavement roughness before resurfacing has a statistically significant effect (at a twotailed 95 percent significance level) on asbuilt pavement roughness. Pavements with a high pavement roughness before resurfacing will tend to have a higher asbuilt roughness after resurfacing. The estimate of the regression slope is 0.29, which indicates that for every 1.0 m/km (63 inches/mile) increase in pavement IRI before resurfacing there will be an expected 0.29 m/km (18 inches/mile) increase in asbuilt IRI. The regression line is based on the average roughness for each site before resurfacing and the average asbuilt roughness for each site. The data exclude two sites, New Mexico and Manitoba, because limited information is available on the roughness of these sites before resurfacing. Figure 2. Asbuilt roughness versus prior roughness for SPS5 data. Table 7. Results of analysis of variance for logarithm of prior roughness for SPS5 sites
INVESTIGATION OF INTERACTIVE EFFECTSAs noted earlier in this paper, the effects of surface preparation and pavement roughness before resurfacing statistically influence the asbuilt roughness of a pavement, and the effect of overlay thickness has a marginally statistical influence on the asbuilt roughness of a pavement. To fully understand the factors that influence the asbuilt roughness of a pavement, it is also necessary to investigate the interactive effects. The best method for examining the interactive effects is to perform a repeatedmeasures analysis on the effects of surface preparation, overlay thickness, overlay material, and pavement roughness before resurfacing. The limited amount of data requires that pavement roughness before resurfacing be categorized into high and low pavement roughness. An IRI level of 1.5 m/km (95 inches/mile) was selected as an appropriate level to separate low and high prior pavement roughness. The withinsubjects effects for repeated measured analysis are presented in table 8 and the betweensubjects effects are presented in table 9. The results for the repeated measures analysis confirm the results of the previous paireddata analysis related to the individual effects of surface preparation, overlay thickness, and overlay material. The analysis also indicates that there are statistically significant (at a 95 percent significance level) interactive effects for surface preparation and pavement roughness before resurfacing as well as for overlay thickness and pavement roughness before resurfacing. The pvalues for these interactive effects are 1.8 percent and 2.7 percent, respectively. The presence of these interactive effects means that when estimating the effect of surface preparation or overlay thickness on asbuilt roughness, the effect should be considered in combination with the pavement roughness before resurfacing. No other statistically significant interactive effects are apparent from the analysis. Table 8. Withinsubjects effects of repeated measures analysis
Table 9. Betweensubjects effects of repeated measures analysis
The effect of pavement roughness before resurfacing has a pvalue of 12.5 percent, which does not indicate a statistically significant relationship (at a 95 percent significance level). Pavement roughness before resurfacing should still be considered as a statistically significant influence on asbuilt roughness because it was determined to have a statistically significant effect when examined quantitatively instead of qualitatively as high and low roughness. A second reason to consider pavement roughness before resurfacing as a statistical influence is that it has statistically significant interactive effects with surface preparation and overlay thickness. VALIDATION WITH CLTPP DATATo validate the findings of the analyses performed on the SPS5 sites, data from the CLTPP program was examined. Regression analyses were performed to validate the individual effects of prior roughness and overlay thickness. Insufficient data were available to validate the effects of surface preparation and overlay material. Average prior roughness and overlay thickness values for each site were analyzed for their relationship with the average asbuilt roughness for each site. The CLTPP measurements represent 14 data points. VALIDATION OF THE EFFECT OF SURFACE PREPARATIONThe regression graph for the effect of pavement roughness before resurfacing on asbuilt roughness is shown in figure 3. This graph is similar to the corresponding SPS5 graph in that the data indicate a positive relationship between prior roughness and asbuilt roughness. Pavements with a high roughness before resurfacing correspond to a high asbuilt roughness. One difference between the data sets is that the CLTPP data contain generally higher prior roughness and higher asbuilt roughness measurements than the SPS5 data. The analysis of variance for the relationship is shown in table 10. The pvalue is 5.2 percent and indicates that the relationship is very close to being statistically significant at a 95 percent significance level. Examining the analysis as a validation of the SPS5 relationship, the pvalue becomes 2.6 percent and indicates a statistically significant relationship at a 95 percent confidence level. The results from the CLTPP data confirm the findings of the LTPP SPS5 analysis. Figure 3. Asbuilt roughness versus prior roughness for CLTPP data Table 10. Results of analysis of variance for prior roughness with CLTPP data
VALIDATION OF THE EFFECT OF OVERLAY THICKNESSA regression plot of overlay thickness and asbuilt roughness for the CLTPP data is shown in figure 4. The data indicate lower asbuilt roughness with thicker overlay thickness. This is consistent with the analyses of the SPS5 data. The analysis of variance for the CLTPP data is shown in table 11. The pvalue for the relationship is 58 percent, indicating a weak relationship. It should be noted that the SPS5 relationship was also weak and not statistically significant at a 95 percent significance level. Figure 4. Asbuilt roughness versus overlay thickness for CLTPP data Table 11. Results of analysis of variance for overlay thickness with CLTPP data
PREDICTION EQUATIONS FOR ASBUILT ROUGHNESSThe previous analyses have shown that four factors have a statistically significant effect on asbuilt roughness at a 95 percent significance level:
The presence of these statistical effects identifies the need to quantify the expected asbuilt roughness that will occur for these factors, thus four regression analyses were performed to determine prediction equations that account for the influence of these factors. These prediction equations can be used by designers to evaluate the effect on asbuilt roughness of their design alternatives and by contractors to estimate the asbuilt roughness that will be achieved under a particular combination of design factors. It should be noted that there is a considerable amount of variability in the prediction equations, and the actual asbuilt roughness of a pavement may be influenced by other factors not incorporated into the equations. The equations are intended to provide typical asbuilt roughness values for different levels of surface preparation, overlay thickness, and prior pavement roughness and to provide a method of quantifying the expected difference resulting from a change in one or more of these factors. For example, both a designer and contractor could estimate the additional smoothness that would be expected by increasing a pavement overlay thickness. The designer would be interested in the longer pavement life that would result from the lower asbuilt roughness and the contractor would be interested in the administrative impact related to the achieving the requirements of a smoothness specification. The prediction equations are considered valid for the range of the regression data, which is for a range of prior roughness of approximately 1.0 m/km (63 inches/mile) to 2.75 m/km (174 inches/mile). BASIC SURFACE PREPARATION AND THIN OVERLAYThe regression graph for basic surface preparation and thin overlays is shown in figure 5 and shows an Rsquared of 0.28. The relationship has a pvalue of 4.2 percent as shown in table 12. The prediction equation determined for basic surface preparation and a thin overlay is as follows: IRI_{AsBuilt} = 0.44 + 0.31 × IRI_{Prior} (1) where IRI_{AsBuilt} = asBuilt IRI in
m/km For example, the expected asbuilt roughness for a pavement with a prior roughness of 2.00 m/km (127 inches/mile) undergoing basic surface preparation and a thin overlay is 1.06 m/km (67 inches/mile). Table 12. Results of analysis of variance for basic surface preparation and thin overlay
Figure 5. Asbuilt roughness versus prior roughness for SPS5 data with basic surface preparation and thin overlay BASIC SURFACE PREPARATION AND THICK OVERLAYThe regression graph for basic surface preparation and thick overlays is shown in figure 6 and shows an Rsquared of 0.28. The relationship has a pvalue of 4.1 percent as shown in table 13. The prediction equation for basic surface preparation and a thick overlay is as follows: IRI_{AsBuilt} = 0.52 + 0.25 × IRI_{Prior} (2) For example, the expected asbuilt roughness for a pavement with a prior roughness of 2.00 m/km (127 inches/mile) undergoing basic surface preparation and a thick overlay is 1.01 m/km (64 inches/mile). This value is approximately 0.05 m/km lower than the previous value determined for a basic surface preparation and thin overlay, and is in line with the mean difference determined earlier in the paired data analysis. Another difference in the prediction equation for basic surface preparation and thick overlay is that the constant is larger and the slope is smaller than in the previous equation. The smaller slope is an indication that the thicker overlay reduces the effect of pavement roughness before resurfacing. Figure 6. AsBuilt roughness versus prior roughness for SPS5 data with basic surface preparation and thick overlay Table 13. Results of analysis of variance for basic surface preparation and thick overlay
INTENSIVE SURFACE PREPARATION AND THIN OVERLAYThe regression graph for intensive surface preparation and thin overlays is shown in figure 7 and shows an Rsquared of 0.29. This Rsquared value is the highest of the four regression analyses for specific surface preparation and overlay thickness treatments. The relationship has a pvalue of 3.8 percent as shown in table 14. This is the strongest statistical relationship of the four regression analyses. The prediction equation for intensive surface preparation and a thin overlay is as follows: IRI_{AsBuilt}= 0.40 + 0.29 × IRI_{Prior} (3) For example, the expected asbuilt roughness for a pavement with a prior roughness of 2.00 m/km (127 inches/mile) undergoing an intensive surface preparation and a thin overlay is 1.01 m/km (64 inches/mile). This value is approximately 0.05 m/km (3.2 inches/mile) lower than the value determined for a basic surface preparation and thin overlay. This value is similar to the mean difference determined earlier in the paired data analyses. Figure 7. Asbuilt roughness versus prior roughness for SPS5 data with intensive surface preparation and thin overlay Table 14. Results of analysis of variance for intensive surface preparation and thin overlay
INTENSIVE SURFACE PREPARATION AND THICK OVERLAYThe regression graph for intensive surface preparation and thick overlays is shown in figure 8 and shows an Rsquared of 0.13. This is lowest Rsquared value of all four regression analyses. The relationship has the lowest statistical strength with a pvalue of 18.9 percent as shown in table 15. One possible reason for the weak relationship is the low influence in the relationship of pavement roughness before resurfacing has in the relationship. The slope of the regression line is the lowest of all the regression lines. Although the relationship is weak, the prediction equation for asbuilt roughness based on pavement roughness before resurfacing with intensive surface preparation and a thick overlay is as follows: IRI_{AsBuilt}= 0.63 + 0.17 × IRI_{Prior} (4) For example, the expected asbuilt roughness for a pavement with a prior roughness of 2.00 m/km (127 inches/mile) undergoing intensive surface preparation and a thick overlay is 0.97 m/km (61 inches/mile). As expected, this is the lowest expected asbuilt roughness of the four combinations. The expected IRI is 0.05 m/km (3.2 inches/mile) lower than the treatment of basic surface preparation and a thick overlay and 0.03 m/km (1.9 inches/mile) less than the value expected for a intensive surface preparation and thin overlay. These values compare well with the expected differences from the paired analyses. The slope of the prediction equation for basic surface preparation and thick overlay is the lowest of all four prediction equations. The low slope is an indication that asbuilt roughness is least sensitive to prior roughness when intensive surface preparation and thick overlay treatment are used. This is as expected because this treatment is considered the most extensive for improving the asbuilt roughness of a pavement. Figure 8. Asbuilt roughness versus prior roughness for SPS5 data with intensive surface preparation and thick overlay Table 15. Results of analysis of variance for intensive surface preparation and thick overlay
CONCLUSIONSVarious statistical analyses, including paired analyses, regression analyses, and a repeated measures analysis, were performed to examine the effects of surface preparation, overlay thickness, type of overlay material, pavement roughness before resurfacing, and the interactive effects of these factors on the asbuilt roughness of a pavement. The following conclusions are provided based on these analyses:
RECOMMENDATIONSBased on the findings of this paper, the following recommendations are provided:
REFERENCESAsphalt Institute, 1989 The Asphalt Handbook Manual Series No. 4 (MS4), 1989 Edition, Lexington, KY. Federal Highway Administration, 1996 LongTerm Pavement Performance Program Reference Guide McLean, VA. McGhee, K.K., 2000. “Factors Affecting Overlay Ride Quality. ” Transportation Research Record 1712,TRB, Washington, DC, National Research Council pp. 5865. Perera, R.W., C. Byrum, and S.D. Kohn, 1998. Investigation of Development of Pavement Roughness. FHWA Report FHWARD97147, Washington, DC, Federal Highway Administration. Perera, R.W. and S.D. Kohn, 1999 “IRI of Asphalt Concrete Overlays: Analysis of Data from LTPP SPS5 Projects.“ Transportation Research Record 1655 TRB, Washington, DC, National Research Council, pp. 100109. Raymond, C.M., 2001. An Investigation of Roughness Trends in Asphalt Pavement OverlaysDoctoral Thesis, Waterloo, Ontario, Canada, University of Waterloo. Schmitt, R.L., J.S. Russell, A.S. Hanna, H.U. Bahia, and G.A. Jung, 1998. “Summary of Current Quality Control/Quality Assurance Practices for HotMix Asphalt Construction ” Transportation Research Record 1632,TRB, Washington, DC, National Research Council, pp. 2231. Smith, K.L., K.D. Smith, L. Evans, T. Hoerner, M. Darter, and J. Woodstrom, 1997. NCHRP Web Doc 1 Smoothness Specifications for Pavements: Final Report, www.nap.edu/books/nch001/html/, TRB, Washington, DC, National Research Council. Transportation Association of Canada, 1999. Summary of Pavement Smoothness Specifications in Canada and Around the World. Canadian Strategic Highway Research Program Technical Brief #16, www.cshrp.org, CSHRP, Ottawa, Canada, CSHRP. TAC, Transportation Association of Canada, 2002. The Effect of AsBuilt Pavement Smoothness on LongTerm Roughness Progression: Some Findings from CLTPP and USLTPP Studies.Canadian Strategic Highway Research Program Technical Brief #23, CSHRP, Ottawa, Canada, CSHRP. United States Army Corps of Engineers, 1991 Hot Mix Paving Handbook, Lanham, MD, National Asphalt Paving Association.

Topics: research, infrastructure, pavements and materials Keywords: research, infrastructure, pavements and materials,LTPP, pavement performance, DataPave contest, DataPave TRT Terms: PavementsPerformanceUnited States, PavementsUnited StatesDesign and construction, LongTerm Pavement Performance Program (U.S.), Pavement performance, Pavement distress, Bituminous overlays, Climate Updated: 03/08/2016
