Effects of Subsurface Drainage on Performance of Asphalt and Concrete Pavements (2003)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

48 CHAPTER 4 CONCLUSIONS AND RECOMMENDATIONS CONCLUSIONS In this study, the potential effects of subsurface drainage on the performance of asphalt and concrete pavements in the SPS-1 and SPS-2 experiments were assessed using available data on IRI for both pavement types, as well as rutting and alligator/longitudinal cracking for asphalt pavements, and faulting, transverse cracking, and longitudinal cracking for concrete pavements. The results of the preliminary analyses conducted in this study suggest that the SPS-1 and SPS-2 experiments are begin- ning to manifest differences in some measures of performance that are related to the base type/subdrainage experimental design factor. However, base type and subdrainage presence are confounded in both the SPS-1 and SPS-2 experiments; and it is difficult to differentiate, on the basis of the informa- tion that was available for this study, between performance differences due to the presence or absence of subdrainage and performance differences due to base type. The SPS-1 and SPS-2 experiments suffer from some limi- tations. First, some cells in the site matrix (i.e., combinations of climate and subgrade type) of each experiment are empty. This is an obstacle to analysis of the SPS-1 and SPS-2 per- formance data in the manner that was envisioned in the orig- inal design of the experiments. Empty cells in an experimen- tal design matrix can be accommodated by using measured data from other cells to estimate values for the missing cells, but only when the factors in the experiment can be considered independent. It is clearly inappropriate to estimate values to fill missing cells when there is reason to believe, a priori, that interactions may exist among the experimental factors. Second, base type and subdrainage presence are confounded in both the SPS-1 and SPS-2 experiments. It is difficult, per- haps impossible, to separate and quantify the effects of these two factors on the basis of roughness and distress data alone. Third, discrepancies between the as-designed and as- constructed drainage features appear to exist for many sec- tions. Subdrains were not located and inspected in several SPS-1 and SPS-2 test sections designed to be drained, but were located and inspected in several SPS-1 and SPS-2 sections designed to not be drained. Both types of discrepancies are an obstacle to analysis of the SPS-1 and SPS-2 performance data in the manner that was envisioned in the original exper- iment design. The original objectives of NCHRP Project 1-34C were to assess the feasibility of using the SPS1 and SPS-2 data to eval- uate the effects of subsurface drainage on pavement perfor- mance, to develop an analysis plan, and to identify additional data collection needs. The scope of the study was subsequently expanded to include a preliminary analysis of the SPS-1 and SPS-2 data. This study has demonstrated that appropriate sta- tistical methods can be applied to circumvent the limitations of the incomplete experimental design matrices and the appar- ent discrepancies in as-designed versus as-constructed drain- age features. The analysis results presented in this report must, how- ever, be considered preliminary, because they are based on data contained in the LTPP database only through mid-June 2001, and because the effects of truck traffic, climate (i.e., different degrees of need for subsurface drainage at differ- ent sites), and structural capacity (i.e., different structural contributions of different base types) were not analyzed in depth. In a follow-up study, already underway, more defin- itive findings are expected to be obtained from analysis of: • Μore recent performance data, • Deflection data (to quantify the structural contributions of different base types), • Drainage system flow time measurements (to quantify the effectiveness of the drains), and • Need for drainage as a function of truck traffic level and climate. Observations from Preliminary Analysis of SPS-1 Performance Data The preliminary analysis suggests that undrained asphalt pavement sections in the SPS-1 experiment with dense-graded aggregate (AGG) bases may develop roughness, rutting, and cracking more rapidly than drained asphalt-pavement sections with dense asphalt-treated base (ATB) material over perme- able ATB material. However, in most cases, the differences detected between the two base types were slight and not sta- tistically significant. The preliminary analysis suggests that undrained asphalt pavement sections in the SPS-1 experiment with undrained

49 dense-graded asphalt-treated base (ATB) may develop rough- ness and cracking more slowly than drained permeable asphalt- treated base over aggregate subbase (PATB/AGG), while the undrained sections may develop rutting more rapidly. In no case, however, were the differences detected statistically significant. The preliminary trends suggest that undrained asphalt pave- ment sections in the SPS-1 experiment with undrained dense- graded asphalt-treated base over aggregate subbase (ATB/ AGG) may develop roughness and rutting more quickly than drained permeable asphalt-treated base over aggregate sub- base (PATB/AGG), while the undrained sections may develop cracking more slowly. In no case, however, were the differ- ences detected the statistically significant. When SPS-1 sections with drainage functioning rated as good were analyzed separately from sections with drainage functioning rated as poor, the results obtained suggest that drainage functioning does play some role in the performance differences observed between drained and undrained sections. In most cases, either a larger positive mean difference or a smaller negative mean difference was calculated for the sec- tions with good drainage functioning. Again, most of the dif- ferences detected were not statistically significant. On the other hand, the preliminary trends suggest that dif- ferences in the structural contributions of the different base types may be as important or more important to the perfor- mance differences noted between drained and undrained sections. In general, drained permeable asphalt-treated base (PATB) sections have performed slightly better, in some respects, than undrained dense-graded aggregate (AGG) base sections, but slightly worse, in some respects, than undrained dense-graded asphalt-treated base (ATB) sections. Observations from Preliminary Analysis of SPS-2 Performance Data The preliminary analysis suggests that undrained concrete pavement sections in the SPS-2 experiment with either an aggregate base (AGG) or a lean concrete base (LCB) may develop roughness, transverse cracking, and longitudinal cracking more rapidly than drained concrete pavement sec- tions with a permeable asphalt-treated base (PATB) and other- wise like design. However, in most cases, the differences detected between the drained and undrained sections were slight and not statistically significant. The SPS-2 faulting data available through mid-June 2001 were too erratic to support meaningful statistical analysis. This may be largely due to section-wide faulting averages having been calculated from many small-negative, zero, and small- positive faulting measurements. When larger magnitudes of faulting develop—such that the averages are consistently being calculated from positive measurements—trends in fault- ing with time and traffic may become more evident. In the analysis of IRI change in drained versus undrained SPS-2 sections, larger mean differences were detected for the PATB sections with drainage functioning subjectively rated as poor than for the PATB sections with drainage functioning rated as good. This was true in comparison to both undrained base types, aggregate and lean concrete. This—to the extent that the subjective ratings of drainage functioning are accu- rate—suggests that quality of drainage is not a significant fac- tor in the differences observed in IRI increase. On the other hand, in the analyses of transverse and longi- tudinal cracking in drained versus undrained SPS-2 sections, larger mean differences were detected for PATB sections with good drainage functioning than for those with poor drainage functioning. This was true in comparison to both undrained base types, aggregate and lean concrete. RECOMMENDATIONS Recommendations for Further Monitoring and Analysis of SPS-1 and SPS-2 The subjective assessments of subsurface drainage func- tioning used in this study should be checked by testing the functioning of the drainage systems. It should be kept in mind that measurement of the permeability of the base material does not shed light on the ability of longitudinal drains and out- lets to effectively remove water, nor does visual assessment of the longitudinal drains and outlets shed light on the per- meability of the base material. Tests that will take both base permeability and drain/outlet adequacy into consideration are recommended. Given the confounding of subdrainage presence and base type in the SPS-1 and SPS-2 experiments, deflection data should be analyzed to quantify the relative structural contribu- tions of the different base types used. This information should be used together with the measured drainage flow time data to distinguish, as much as possible, between the effect of base type and the effect of drainage quality in any performance dif- ferences observed. Some inconsistencies exist from site to site with respect to the presence of filter fabric below the permeable asphalt- treated base layer in both the SPS-1 and SPS-2 experiments. Whether or not filter fabric is needed to protect the PATB from infiltration of fines—or whether, for example, a granular sub- base serves the same purpose—is a question on which opinion is divided. Ultimately, this question should be resolved by measuring flow times through permeable bases with and with- out filter fabric protection and analyzing this drainage flow time data together with long-term performance data. The performance data analyses demonstrated in this study should be repeated with more recent data. By mid-2003, the different SPS-1 sites will have been in service about 5 to 11 years and will have carried between about one and five million flexible pavement ESALs. The different SPS-2 sites will also have been in service about 5 to 11 years and will have carried between about one and fifteen million rigid pavement ESALs.

50 Subdrains were not located and inspected in 24 percent (26 of 108) of the SPS-1 test sections designed to be drained nor in 14 percent (8 of 56) of the SPS-2 test sections designed to be drained. The presence of subsurface drains and lateral out- lets should be confirmed for these test sections. Subdrains were located and inspected in 16 percent (17 of 108) of the SPS-1 test sections designed not to be drained and in 12 per- cent (14 of 112) of the SPS-2 test sections designed not to be drained. The presence of drains in these sections that should not be drained should be confirmed and must be taken into proper account in future analysis of performance data from the SPS-1 and SPS-2 experiments. Recommendations for Future Field Experiments to Assess Drainage Effects The SPS-1 and SPS-2 experiments offer lessons for the design and construction of future field experiments to assess the effects of subsurface drainage on asphalt and concrete pavement performance. One lesson is that the effects of drain- age presence and base type should be separated by including in the experimental design both drained and undrained sections of the same base types. Where site conditions are suitable, drainage of the same base type or types via daylighting ver- sus longitudinal edgedrains and outlets is also recommended. Another lesson is that considerable care must be taken in the construction of drainage test sections to ensure that sections designed to be drained have adequate edgedrains and outlets installed and that sections designed not to be drained do not have access to edgedrains and outlets. Sites selected for future field experiments in subsurface drainage effectiveness should be ones with truck traffic lev- els sufficiently high to yield clear indications, within ten years, of whether or not subsurface drainage significantly influences pavement performance. Experiments constructed on routes with truck traffic volumes so low that subsurface drainage would not normally be used will not shed much meaningful light on whether or not subsurface drainage does or does not have significant beneficial effects on higher-volume routes. Uniformity of vertical and horizontal alignments should also be considered in selection of sites for future field exper- iments in subsurface drainage effectiveness, given that drain- age is just one of several factors in the SPS-1 and SPS-2 experiments, the SPS-1 and SPS-2 sites cover considerable lengths, and at some sites, the longitudinal grade varies con- siderably along the length. Also at some SPS-1 and SPS-2 sites, horizontal alignment changes result in some test sec- tions having an outward transverse grade and being drained toward the outer foreslope (beyond the outer shoulder), with other test sections having an inward transverse grade and being drained to the inner foreslope (under the inner traffic lane and beyond the inner shoulder). In experiments in which the outer traffic lane is the monitored lane, test sections should be located so that the transverse grade and subsurface drainage is consistently outward. Video inspection of longitudinal edgedrains and their out- lets provides some useful information on blockage, sags, and accumulation of water and soil in drains, but does not provide a quantitative measure of the quality of subsurface drainage. Measurement of drainage flow times is recommended for monitoring of future field experiments in subsurface drain- age effectiveness. Finally, analysis of drainage flow time data together with deflection data is recommended to differentiate appropriately, and quantitatively, the effects of base type and drainage quality.