Significant Findings from Full-Scale Accelerated Pavement Testing (2004)

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.

37 CHAPTER FOUR EVALUATION OF MATERIALS AND TESTS INTRODUCTION Full-scale APT programs worldwide have produced sig- nificant findings comparing and evaluating pavement ma- terial response and performance. The primary goal of most programs is to evaluate new, innovative, recycled, materi- als, with validation of traditional materials as a secondary goal. Another primary goal of many programs is to validate laboratory material characterization through comparison with response under full-scale loading, recognizing that a strategic approach that incorporates APT, laboratory test- ing, long-term field evaluation, and modeling or analysis, produces meaningful results that allow for the considera- tion of material, loading, and environmental variability. This chapter presents the current knowledge base in these two areas in general, recognizing that the results are not specifically comparable because of differences between programs in terms of loading and environ- mental conditions, measurement and analysis tech- niques, and failure definitions. These differences are ex- pected, given the high cost of operating full-scale APT devices and the corresponding necessity to produce goal-specific results that advance the understanding of pavement materials and behavior for the conditions most relevant to the funding agency for a particular program. Because of the substantial resources required and a focus on multiple goals, experiments that use APT technology in conjunction with laboratory testing and analysis have not provided as many results specific to materials evaluation when compared with experiments that use only laboratory testing. This highlights the importance of sharing signifi- cant findings and underscores the potential for substantial gain through cooperation and coordination of multiple APT programs. Significant findings from applying full-scale APT tech- nology to evaluating materials and tests are organized in this chapter by pavement layer, with the discussion pre- sented by material type. Following an overview of field and laboratory material characterization used in different APT programs, general response and performance results are presented without describing specific instrumentation, the comparison of different measurement techniques, or specific parameters that were monitored and analyzed. All of these results must be qualified in recognition that differ- ences between full-scale APT loading and environmental conditions and those under full-scale traffic in service do exist. Separation of the materials evaluation results from the discussions of design considerations in chapter two and maintenance and rehabilitation techniques in chapter six is incomplete because of the synergy between material selec- tion and structural design. Unconventional pavement mate- rials, including block pavers and ultra-thin whitetopping, and performance not related to primary forms of distress, including drainage effects and curling and warping of con- crete pavements, are discussed subsequently in chapters six and seven. Before discussing the wide variety of applications that were found in the literature, the results from a synthesis of the questionnaire survey is presented. QUESTIONNAIRE SURVEY The responses to Questions 4.1 to 4.9 on materials and tests are reflected in Figures C31 to C38 in Appendix C. These re- sponses were synthesized and the results are contained here. • Asphalt Pavements – HMA was tested the most frequently, followed by granular materials and stabilized materials. There is far less focus on other materials. – In terms of mix type, continuously graded or dense-graded mixes have been tested the most frequently, both for surfacing and base courses. – Asphalt parameters most frequently measured in- clude density, gradation, binder content, and stiff- ness. – Special materials such as geogrids have also been tested. – In the laboratory, the indirect tensile test is the test most frequently used to evaluate the strength of asphalt materials. – Sample preparation was either by gyratory com- paction or by Marshall compaction. – There were essentially two binder tests that were being used, the Dynamic Shear Rheometer and Penetration and Softening Point. The bending beam rheometer and rotational viscometer were also used frequently. • Concrete Pavements – Jointed concrete pavements were the most com- mon rigid structures tested. – PCC was the primary material tested in jointed concrete pavements. – In the laboratory, cylinder compressive tests were the most frequently used.

38 – Flexural strength and stiffness, together with compressive strength, were the primary items that were evaluated for the purposes of controlling the concrete materials. – In the field, the FWD was the primary tool used to gain insight into the characteristics of the pave- ment structure. Density and moisture measure- ments also featured prominently. Other supple- mentary tests were ground penetrating radar and the dynamic cone penetrometer. Deflection meas- urements were also done with the Benkelman beam. Views of the respondents to the survey on materials and tests are presented in Table D4 in Appendix D. MATERIAL CHARACTERIZATION Most APT programs use both field and laboratory material characterization to assess pavement response and perform- ance under full-scale APT loading. Performance monitor- ing at the APT test section generally involves measurement of transverse and longitudinal profiles, deflection and de- formation at the surface or with depth in response to a moving load or falling weight, in situ density, environ- mental conditions including moisture and temperature with depth, visual surface distress, and in situ stresses and strains. For concrete pavements, relative joint movement is also usually monitored. Other field characterization used in some APT programs included SASW to determine stiff- nesses (elastic moduli) and detect damage prior to visual distress identification (Lee et al. 1997), in situ permeability testing, trenching after failure, and measurement of the relative shear resistance of unbound materials. Surface friction or skid resistance is also monitored in some pro- grams. In other programs a scaled APT device was used on a section adjacent to the section tested by the full-scale de- vice to aid in evaluating the effects of different environ- mental conditions including moisture and elevated tem- peratures. Laboratory characterization varies widely from program to program depending on the goals, experience, and avail- able equipment. Stiffness measured by resilient modulus and density of field cores and each layered material are de- termined as part of most APT tests. Shear stiffness meas- urements in frequency sweeps and indirect tensile testing of asphalt materials are also common. Triaxial testing of unbound materials through the use of standard or modified equipment is common. Permeability tests on these materi- als in the laboratory have also been completed. Technology and analysis systems for asphalt materials developed dur- ing SHRP were used in a number of APT tests. Many stud- ies included AC mixture tests to determine resistance to fa- tigue cracking in third point flexural fatigue tests and resistance to rutting in either repeated or simple shear tests. One APT program also included AC mixture testing to en- sure adequate resistance to thermal cracking (Epps et al. 1999). Resistance of these materials to rutting is also commonly assessed through the use of wheel-tracking de- vices or creep tests. In addition, traditional and SHRP per- formance grade binder tests are used to characterize as- phalt binders and their contribution to both fatigue and rutting performance. Indirect tensile fatigue tests and semi- circular bending tests are also occasionally employed to characterize AC. Moisture susceptibility testing of AC through the use of retained indirect tensile strength ratios before and after wet conditioning also contributed to the assessment of APT results. Direct tension testing of both asphalt and concrete materials has also been used, although less frequently. Common characterization for concrete ma- terials includes determination of compressive and flexural strength and stiffness. Other performance indicators for these materials, for example, sulfate resistance and ASR potential, are also measured in some APT tests, for exam- ple, CAL/APT (California Accelerated Pavement Testing) (Harvey et al. 2000). Metcalf (1996) and others provided detailed descrip- tions of instrumentation used to obtain these field and labo- ratory measurements. These measurements are then ana- lyzed using a number of different procedures to determine pavement performance in terms of primary forms of load- related distress. The possible effects of environmental con- ditions were then assessed through the use of laboratory characterization and modeling or analysis. A summary of these general performance results related to the evaluation of materials and tests follows. Surface To date the majority of APT tests have been conducted us- ing AC or a bituminous chip seal as the surface pavement layer. For many of the AC surfaces, modified binders were included and exhibited enhanced performance in terms of resistance to rutting and/or fatigue cracking. The ALF pro- gram in Australia reported this enhanced performance for AC mixtures with a range of modified binders used in rehabilita- tion treatments (Kadar 1991; Sharp et al. 1999a). The im- proved fatigue performance and adequate rutting perform- ance of a rich bottom AC layer; that is, a layer with a high binder content, was also demonstrated. Laboratory fatigue and dynamic creep testing of mixtures to examine per- formance highlighted a need for standardized test methods. The ALF program in Australia also explored the use of modified binders to increase AC mixture resistance to permanent deformation (Oliver 1994; Sharp et al. 1999a). Stone matrix asphalt (SMA) mixtures that incorporated full-scale loading under the ALF and extensive laboratory

39 testing were also included in this study. Conventional mix- tures with two different filler contents and five different binders, including one modified with styrene butadiene styrene (SBS) polymer and one modified with EVA poly- mer, were evaluated, along with a recently developed rut resistant mixture with a conventional binder and an SMA mixture. Laboratory testing included dynamic creep test- ing, resilient modulus determination, laboratory wheel- track testing, and measurement of the Superpave binder rutting parameter G*/sinδ. Mixture creep and resilient modulus testing revealed a significant effect of filler con- tent. Creep test results of field core specimens measured at a representative ALF trafficking temperature were also able to identify a significant effect of binder type; however, these results could not distinguish between mixtures with different aggregate gradations (filler content). Mixture creep properties of field-core specimens measured at ele- vated temperatures did not correctly rank rutting perform- ance under full-scale loading. Laboratory wheel-track re- sults at 60ºC (using the Australian wheel-tracking device) correlated with field performance under the ALF at 50ºC and served as an indicator of mixture resistance to rutting. The binder parameter G*/sinδ was also invalid as an indi- cator of rutting performance for the modified mixtures. Mixtures incorporating the SBS-modified binder and a multigrade binder exhibited reduced temperature suscepti- bility and an increased resistance to rutting when compared with the other mixtures. The CAPTIF program in New Zealand also investigated the effects on performance of modified binders in AC sur- face layers constructed on nominally identical pavement structures (Pidwerbesky 1995b). The performance of a thicker AC layer with a conventional binder was equivalent to that of thinner layers containing either a high-stiffness unmodified binder or binders modified with a plastomer or one of three different elastomers. At the conclusion of the APT tests, all of the structures exhibited minimal surface distress and negligible structural deterioration. The CAL/APT program found that a gap-graded, crumb-rubber-modified AC overlay outperformed a dense- graded mixture with a conventional binder in terms of permanent deformation when the structure is adequate to preclude permanent deformation of the underlying layers (Harvey et al. 2000). Both overlays, discussed as rehabili- tation measures in chapter six, also exhibited substantial resistance to reflective cracking from the underlying failed AC layer. This program also demonstrated the use of the flexural fatigue test for AC mixture design and analysis developed during SHRP. Use of the repeated simple shear test at constant height, also developed during SHRP, was recommended for modified AC mixtures. The SA–HVS program also investigated the use of bi- tumen–rubber asphalt in a relatively thin, open-graded AC overlay of a cracked concrete pavement (Viljoen et al. 1987). This overlay exhibited good performance in terms of reflection cracking when compared with many other re- habilitation overlays, including those with different types of interlayers, as discussed in chapter six. The LA ALF program also examined crumb-rubber- modified AC in a surface layer and found no significant improvement in mixture rutting performance as compared with conventional AC (Mohammad et al. 2000). This finding was consistent with laboratory mixture characterization, in- cluding indirect tensile strength and resilient modulus, in- direct and axial creep results, and Superpave shear tests. Modified asphalt binders were also used at the LCPC facility in France in two experiments aimed at determining the binder effect on both rutting and fatigue of AC (Corté et al. 1994, 1997; De la Roche and Rivière 1997). For the rutting tests, surface layers with seven different binders were placed on nominally identical pavement structures. The binders evaluated included one conventional unmodi- fied material, two with low-temperature susceptibilities, one modified with SBS elastomer, one modified with EVA plastomer, one hard binder (20/30 Pen), and one modified with low-density polyethylene waste. The hard binder was used in a high-modulus mixture that had an additional very thin AC layer at the surface. When compared with the con- ventional mixture, mixtures with modified binders exhib- ited increased resistance to rutting in laboratory wheel- tracking tests with the French device. In addition, all mix- tures incorporating new and innovative materials exhibited better rutting performance under full-scale loading. The high-modulus mixture with its very thin AC surface layer also retained surface texture under trafficking. Other labo- ratory mixture test results, including those from repeated triaxial and static and dynamic creep tests, provided per- formance rankings equivalent to those based on perform- ance under full-scale loading or in laboratory wheel- tracking tests. For the LCPC fatigue tests, surface layers with five different binders and three thicknesses were constructed on nominally identical pavement structures (De la Roche et al. 1994). These binders included two conventional unmodi- fied materials of the same grade, but obtained from differ- ent sources; two hard binders (20/30 Pen) of the same grade from different sources used in a high-modulus mix- ture and one binder modified with SBS polymer. For one of three experiments included in this study, very thin AC surface layers were used; these thin surface layers were omitted from the other experiments. The two mixtures with conventional binders from different sources exhibited simi- lar fatigue performance under full-scale loading, but very different fatigue behavior was measured in the laboratory. When compared with a conventional mixture, the high- modulus mixture is expected to increase fatigue life, but

40 this material must be used on relatively stiff supporting layers. On a deformable supporting layer, this type of mix- ture in a thin layer offered no improvement in perform- ance, but performance benefits increased rapidly with thickness. When properly used, a high-modulus mixture can offer performance benefits in terms of resistance to both fatigue cracking and rutting. The polymer-modified mixture also increased fatigue life, but only slightly. As part of this study, extensive laboratory fatigue testing was also completed. Laboratory test results indicated that rela- tive fatigue performance depends on the specific test, the presence or absence of rest periods, and the testing mode. Relative behavior of the mixtures evaluated in this study was successfully predicted using controlled-stress fatigue testing in the laboratory. Modified asphalt binders were also evaluated as part of the FHWA ALF program to validate Superpave binder pa- rameters controlling rutting and fatigue cracking of AC (Stuart et al. 1995, 2000; Stuart and Mogawer 1997; Romero et al. 1998, 2000; Sherwood et al. 1998, 1999). For the rutting tests, surface layers with two aggregate gradations and maximum sizes and five different binders, including one modified with low-density polyethylene and one modified with styrene–butadiene polymer, were placed on nominally identical pavement structures. When compared with con- ventional binder mixtures, modified binder mixtures exhib- ited more resistance to rutting as measured under full-scale loading and indicated by high binder G*/sinδ values. Mix- tures with larger nominal maximum size aggregate also exhibited decreased susceptibility to permanent deforma- tion. The Superpave binder specification test results corre- lated well with rutting performance in the APT tests for conventional mixtures, as did test results with three labora- tory wheel-tracking devices. Modified mixture perform- ance was less sensitive to the binder parameter G*/sinδ, and this parameter did not successfully predict the per- formance of these mixtures under the ALF or in laboratory wheel-tracking tests. This study highlights the need for ad- ditional characterization of modified binders and mixture testing of modified mixtures to capture their improved per- formance. As part of this same study, laboratory shear test- ing and wheel-tracking tests of two unmodified mixtures were not able to capture the effect of aggregate gradation on rutting performance, although they were able to capture the effect of the binder on rutting performance. For the FHWA ALF fatigue tests, surface layers with two thicknesses and five different binders, including one modified with low-density polyethylene and one modified with styrene–butadiene polymer, were placed on nominally identical pavement structures. The Superpave binder speci- fication test results correlated with fatigue performance in the APT tests only for the thin AC layers when binder test- ing was conducted at a higher frequency (Sherwood et al. 1999). This behavior is expected because flexibility in thin layers increases fatigue life, and the Superpave specifica- tion sets a maximum G*/sinδ value at intermediate tem- peratures. This parameter was derived for strain-controlled conditions in thin AC layers, and thus the specification only applies to thin layers where these conditions are ap- propriate (Stuart et al. 2000). The binder effect on fatigue performance also reflected this dependence on AC layer thickness. Mixtures whose binders have high G*/sinδ val- ues exhibited increased resistance to fatigue cracking in thin layers and decreased resistance in thick layers. These effects are also expected for mixtures with polymer- modified binders with anticipated high G*/sinδ values. As part of this same study, flexural fatigue testing of AC mix- tures was conducted. When comparing mixtures with dif- ferent binders, results from the laboratory were in agree- ment with the full-scale APT results. The flexural fatigue test was recommended to determine relative fatigue per- formance; however, its limitations when used as a stand- alone test to account for pavement structure and represen- tative temperature fluctuations were recognized (Romero et al. 2000). As developed in the SHRP program, this test must be used as part of a design and analysis system. The Texas DOT program evaluated a thin AC surface layer and found early fatigue failure resulting from high air void contents and construction variability (Hugo et al. 1997). Early and unexpected rutting failure occurred in coarse-graded AC mixtures at WesTrack (Epps 1998; Epps et al. 1999). In this APT program, coarse- and fine-graded AC mixtures at three air void contents and three binder contents were placed on nominally identical pavement structures, and performance was monitored toward devel- opment of performance-related specifications for AC and field verification of Superpave volumetric mix design. Ma- terials evaluation involved both field performance in terms of fatigue cracking, rutting, moisture damage, and thermal cracking, and corresponding mixture characterization tests in the laboratory. Coarse- and fine-graded mixtures per- formed differently in both field and laboratory conditions. Moisture sensitivity was indicated for one-half of the mix- tures according to laboratory determination of the ratio of indirect tensile strength before and after moisture condi- tioning; however, moisture damage was not detected under full-scale loading as designed. This result highlights the need to reevaluate this laboratory testing process and the correspondence of laboratory results with field perform- ance. All mixtures were also designed to preclude thermal cracking. Laboratory test results and field performance both indicated adequate resistance to this form of distress. Similar results in both the laboratory and field performance were also demonstrated with the flexural fatigue test and corresponding fatigue cracking. As expected, mixtures with lower air void contents and higher binder contents ex- hibited increased resistance to fatigue. These effects were

41 amplified for coarse-graded mixtures. For rutting perform- ance, field and laboratory mixture assessment again agreed in terms of the effects of air void content, binder content, temperature, and gradation. Laboratory assessment in- cluded the use of repeated shear tests, shear frequency sweeps, laboratory wheel-tracking devices, and a scaled APT device (Ruiz and Romero 1999; Williams and Prowell 1999; Epps et al. 2002). The unexpected rutting perform- ance displayed during this APT test highlighted the need for including a mixture performance test at a critical high temperature in the Superpave mix design process. Other recommendations for improving the design of these mix- tures are discussed in chapter five. Coarse-graded Superpave AC mixtures also exhibited less resistance to permanent deformation when compared with fine-graded mixtures during APT tests conducted as part of the Indiana DOT/Purdue program (Galal and White 1999; White et al. 1999). Laboratory characterization using a scaled wheel-tracking device and shear frequency sweeps validated these results under full-scale loading. In a sepa- rate warranty study, a Superpave AC mixture demonstrated improved rutting performance compared with a conven- tional mixture. A third study in the Indiana DOT/Purdue program emphasized the need for a minimum percentage of crushed aggregate to ensure adequate rutting perform- ance of AC mixtures. This study also examined the effect of coarse aggregate type on rutting performance and found that slag and limestone aggregates that are traditionally crushed produced mixtures with rutting performance sub- stantially better than mixtures with more rounded, un- crushed gravel aggregate. To validate Superpave mixtures for implementation in Kansas, the Kansas Accelerated Testing Laboratory (K–ATL) program used the agency’s APT facility to traffic two fine-graded Superpave mixtures with two different percentages of natural sand (Wu et al. 2000). These mix- tures were constructed on nominally identical pavement structures. Both mixtures exhibited severe rutting after relatively few load applications, from shear flow of the mixture with a high natural sand content and from consoli- dation of the other mixture. At this point, no visible fatigue cracking was present in either mixture. Flexural fatigue testing in the laboratory indicated that the mixture with a lower natural sand content is expected to exhibit better re- sistance to fatigue cracking; however, this result was not validated under full-scale loading as a result of the rutting failure. This program also found that a Superpave mixture provides better rutting performance as an overlay of con- crete pavement, when compared with a commonly used mixture designed by the more traditional Marshall method. Based on relatively few full-scale load applications, the HVS–NORDIC program in Finland estimated the relative performance of a traditional AC structure with an SMA surface layer and a conventional AC base and an innova- tive structure with a stiff AC surface layer and a flexible, fatigue-resistant AC base layer (Huhtala et al. 1999). The innovative structure was estimated to substantially increase fatigue life based on pavement response data measured in terms of strain at the bottom of the AC layers and labora- tory stiffness and fatigue testing. The Texas DOT program reported on the comparison of two rehabilitation processes under APT (Hugo et al. 1999a,b; Smit et al. 1999; Walubita et al. 2000, 2002). The tests are also discussed in chapter six. It was found that the recycling of a thicker layer of lightweight aggregate as- phalt concrete (LWAC) resulted in improved rutting per- formance and decreased layer deflections as compared with the recycling of a thinner layer of the same material in situ and overlaying with a new AC mixture. Details of the pavement and rehab structure are shown in Figure 11 Both surface layers exhibited substantial resistance to permanent deformation, but the remaining underlying LWAC was less resistant to rutting and susceptible to mois- ture damage. The process that involved recycling of the thinner layer in situ and placement of an overlay was also more susceptible to moisture damage; however, these ma- terials exhibited improved fatigue performance under hot and dry conditions. Both sections showed reduced fatigue lives after wet trafficking based on indirect tensile fatigue testing. These results highlighted the importance of consid- ering degradation and deterioration that are the result of the combined effects of trafficking and moisture. Use of seis- mic analysis of surface waves to detect decreasing stiffness during wet trafficking was also demonstrated (Walubita et al. 2002). Material evaluation results for this type of APT test are difficult to assess without the extensive use of labo- ratory characterization because the original pavement structures play a large role in performance, as discussed in chapter six. In this case, the underlying LWAC lay on a stiff structure, and therefore pavement performance was controlled by the rehabilitated layers and the underlying LWAC. AC overlays of nominally identical pavement structures were also evaluated in the SA–HVS program (Kong Kam Wa et al. 1997). This APT program has traditionally tested in-service pavements, and therefore the evaluation of reha- bilitation techniques, as discussed subsequently in chapter six is common. The performance of a dense-graded AC mixture with a conventional unmodified binder was com- pared with that of a more open-graded mixture containing styrene–butadiene–rubber binder. Two different overlay thicknesses were examined for each mixture, with smaller values used for the polymer-modified mixture with the ex- pectation of equivalent or improved performance. Im- proved fatigue performance of the modified mixtures was demonstrated in flexural fatigue tests in the laboratory.

42 FIGURE 11 Texas US-281 pavement structures tested with the TxMLS and the MMLS3 (Walubita et al. 2002). Traditional stiffness prediction methods were not applica- ble to the modified mixtures; these methods could not pre- dict stiffness values measured in indirect tensile testing and determined from deflection values measured by layer. APT has also been used to evaluate bituminous materi- als other than AC as surface layers, but less frequently. The ALF program in Australia demonstrated adequate perform- ance of geotextile-reinforced chip seals over clay sub- grades for low-volume roads (Sharp et al. 1999a). This re- sult validated the use of local materials that produced cost savings by removing the need to import higher quality ma- terials. The ALF was also used to evaluate deep-lift, in situ recycling (Sharp et al. 1999a). Slag/lime binder was used at three different recycling depths and performance was compared with an unbound granular material. Adequate fa- tigue performance was obtained for all stabilized sections. Modulus and unconfined compressive strength values of laboratory-compacted specimens and field cores did not match because of differences in preparation techniques. Recently, interest has revived for using APT to evaluate concrete materials. The CAL/APT program examined FSHCC in a jointed concrete pavement (Roesler et al. 1999; Harvey et al. 2000). The fatigue of this material was simi- lar to ordinary type II PCC tested in the laboratory. Rec- ommendations included enhanced laboratory characterization of the sulfate resistance and ASR potential of these concrete materials to improve performance prediction. These were based on laboratory tests to study those aspects that were run to complement the accelerated load testing. The use of a nonerodable, flexible support material to ensure adequate performance was also suggested. According to the ques- tionnaire survey, the K–ATL program found no improve- ment in performance from fiber reinforcement of a plain concrete overlay without dowels. As discussed earlier in chapter two, the ALF program in Australia recently com- pleted APT tests of plain concrete pavements to quantify the effects of different design elements (Vuong et al. 2001). Base/Subbase Numerous APT tests have been conducted to examine the performance of unbound and stabilized granular materials used as base or subbase pavement layers. These studies are particularly prevalent in international APT programs be- cause of the role and importance of these layers in many

43 low-volume road networks. On these networks, thin AC layers or chip seals are usually only providing a surface that waterproofs the underlying base and/or subbases. The first APT test in the CAPTIF program in New Zea- land validated the use of well-compacted, dense-graded, crushed, and unbound materials as base layers beneath chip seal surface layers (Pidwerbesky 1995b). The first APT test in the Australian ALF program investigated the perform- ance of unbound and stabilized base materials beneath a thin surface seal (Kadar and Walter 1989). The use of base materials stabilized with slag in place of high-quality crushed rock was validated under full-scale loading when adequate support from the subgrade and protection from a surface layer are provided. The second ALF test verified adequate performance of crushed rock pavements surfaced with chip seals, provided that the seal is kept intact (Sharp et al. 1999a). The third ALF test demonstrated that deterio- ration of a CTCR base could be duplicated under ALF loading (Kadar et al. 1989; Sharp et al. 1999a). This mate- rial failed by debonding at the interfaces of the multiple lifts placed during construction and by erosion of the bot- tom of the top layer. These results primarily influenced construction practices as discussed in chapter seven. The CAPTIF program also investigated the effect of ag- gregate size and shape in unbound base layers beneath open-graded AC surface layers (Pidwerbesky 1995a). Ag- gregate shape had the most pronounced effect on perform- ance in terms of deflection and deformation measured at the surface under full-scale loading, validating the aggre- gate angularity specification for unbound base materials. The more coarse-graded base material did exhibit im- proved performance when compared with the more fine- graded material. There were also differences in the per- formance of base materials with the same gradation. Possi- ble reasons for this could be differences in compacted den- sity and the percentage of angular particles in the mix. The higher density appeared to outweigh the effect of the per- centage of angular particles, because the former materials had less deformation. The ALF program in Australia also investigated the use of marginal sandstone by stabilizing in situ with slag/lime or bitumen/cement binders (Sharp et al. 1999a; Yeo et al. 1999). Performance of stabilized sections was compared with that of a section with the same material in an unbound state. All sections had a thin AC surface layer. Both stabi- lized sections exhibited good performance in terms of fa- tigue cracking and subgrade deformation. Failures of the thin surface layer were addressed by the placement of a prime coat before constructing this layer and recommenda- tions to seal the stabilized layers to prevent cracking. As for the deep-lift, in situ recycling study, modulus and un- confined compressive strength values of laboratory- compacted specimens and field cores did not agree. Marginal sandstone was also evaluated in a second study conducted by the ALF program in Australia (Vuong et al. 1996; Sharp et al. 1999a). In this study, performance of a base layer consisting of a high-quality marginal sand- stone was compared with that of a more abundant, lower- quality marginal sandstone. In addition, performance of an unbound sandstone base was compared with that of one composed of sandstone stabilized with a bitumen/cement binder. With a stiff underlying subgrade, any of the mar- ginal materials tested exhibited adequate performance in support of a waterproof surface seal. These materials did not add to the pavement structural capacity; permanent de- formation manifested in these layers and the surface seal. Reconstructed sandstone bases on these types of subgrades only improved performance up to a specific thickness, re- sulting in recommendations for thinner structures, as dis- cussed in chapter five. The stabilized base exhibited good performance in terms of increased stiffness, decreased permanent deformation, and resistance to water infiltration. Stabilization with a bitumen/cement binder was also effec- tive in improving the performance of a high-quality recon- structed crushed-rock base. The Australian ALF program also conducted additional APT tests to establish a simple test to characterize unbound granular materials in terms of resilient modulus and per- manent deformation (Sharp et al. 1999a). High-quality crushed-rock bases beneath thin AC surface layers were used in this study and a standardized laboratory repeated load triaxial test method was proposed. Both good and poor quality lateritic gravels or ferricrete beneath sealed surfaces also performed adequately under dry conditions in another Australian ALF study (Sharp et al. 1999b). After moisture infiltration, both of these materi- als failed. As part of the CAPTIF program in New Zealand the performance of lime-stabilized subbase materials was compared with that of unbound crushed aggregate materi- als in nominally identical pavement structures (Pidwer- besky 1995a). Laboratory testing was used to determine optimum stabilizer content, and the stabilized materials outperformed the unbound materials in terms of deflection and deformation measured at the surface under full-scale loading. Increasing the thickness of the stabilized subbase layer also substantially improved performance. Stiffness values measured in the laboratory and those determined based on deflections measured in the field did not agree because of compaction problems on a weak subgrade. The SA–HVS program has traditionally tested in- service pavements; therefore, evaluation of rehabilitation techniques as discussed in chapter six is common. In terms of materials evaluation, this program examined the per- formance of labor-intensively constructed bases under full-

44 scale loading and in static and dynamic triaxial tests (Theyse 1999). Emulsion-treated natural gravel, water- bound and composite macadams, and an untreated and emulsion-treated ash waste material were compared to a machine-constructed, crushed-stone base. Each base mate- rial was supported by a cement-treated sandstone base and either imported sandstone or ferricrete. The crushed-stone base exhibited the best performance under full-scale load- ing in terms of rate of permanent deformation and bearing capacity (defined as the number of load repetitions to a specific level of permanent deformation). Static triaxial test results produced the opposite results for the ash waste ma- terial; however, a valuable link between dynamic triaxial test results and performance under full-scale loading was realized in this study. The best labor-intensively con- structed base material was the emulsion-treated natural gravel. Waterbound macadams were also recommended for heavier traffic loads on light pavements, although the ash waste material with adequate compaction is appropriate for lower traffic loads. The SA–HVS program has also used APT tests in the development of guidelines for the use of new base mate- rials, specifically large aggregate mixes for bases (LAMBs) and granular emulsion mixes (GEMs) (De Beer and Grobler 1993). LAMBs were tested under full-scale loading to validate an extensive laboratory testing program that demonstrated adequate performance of these materials in heavy-duty pavements. Dynamic creep modulus was also correlated with deformation under full-scale loading. The performance of GEMs that upgrade marginal in situ materials was comparable to an imported crushed aggre- gate base under full-scale loading. Because of reduced transportation and material costs, this finding again di- rectly results in cost savings as discussed in chapter nine. Other base materials evaluated in the CSIR program in South Africa included roller-compacted concrete, slag, re- cycled AC, and emulsion-treated recycled granular mate- rial (Horak et al. 1992; Rust et al. 1997). Design and usage guidelines for all of these materials were developed based on their performance under full-scale loading. For exam- ple, in rehabilitating untreated or cement-treated bases, the addition of cement and lime to emulsion-treatment im- proves strength and durability. This type of base material exhibited decreased fatigue performance when compared with AC, but resistance to fatigue was greater than for ce- ment-treated bases (Horak and Rust 1992). Guidelines for the stabilization of marginal natural aggregate materials were also developed based on APT test results. High- quality granular materials were also tested under full-scale loading to verify their use in heavy-duty pavements. The HVS–NORDIC program in Finland tested two crushed-rock base materials of different quality underneath AC surface layers (Huhtala et al. 1999). Definitive results were not provided in this first study because of abbreviated tests and incomplete analysis; however, performance was better than expected. The CAL/APT program compared the performance of a high-quality unbound aggregate base to a commonly re- quired ATPB (Harvey et al. 2000). Both pavement struc- tures tested contained a dense-graded AC surface layer, and performance was evaluated in terms of response under full- scale loading and in standard and repeated triaxial testing in the laboratory under dry and saturated conditions. Per- meability testing in the field and the laboratory was also conducted. The ATPB layer was determined to be unneces- sary if the permeability of the AC surface layer is de- creased and the fatigue resistance is increased through adequate compaction, increased binder content, and in- creased layer thickness. Improved performance in terms of resistance to fatigue and increased structural capacity was demonstrated for the ATPB in comparison to the unbound aggregate base. However, if ATPB layers are used, strip- ping and intrusion of fines was shown to be likely in the wet condition. To preclude the failure of this layer, recom- mendations were made for periodic maintenance of the drainage system, increased binder contents and the use of modified binders, geotextile filters, and additives to guard against moisture damage. Moisture sensitivity evaluation of ATPB in the laboratory was also suggested. The LA ALF program also investigated base and sub- base materials with AC surface layers (Metcalf et al. 1999). Crushed-stone and stabilized soil–cement materials were combined in nine different base/subbase structures and evaluated in terms of performance under full-scale loading. When the stabilized soil cement was used as a base layer, the AC surface layer cracked as a result of reflection shrinkage cracks and top-down cracks. All stabilized base structures failed because of softening and erosion of these materials and subsequent loss of support. The researchers noted that the source material for the soil–cement was silty and prone to erosion. The modes of distress were probably related to the nature of the material. Structures with crushed-stone bases failed owing to permanent deforma- tion of this material. The combination of the two materials with the crushed stone as the base layer in an inverted structure provided improved performance over standard structures containing only one of these materials. In-plant cement mixing and plastic fibers did not improve perform- ance, but improved performance was demonstrated for ma- terials with increased cement content and increased thick- ness of the stabilized soil–cement layer. The CEDEX program conducted a third set of APT tests in Spain to examine the performance of base and subbase materials beneath AC surface layers in terms of load- related distress (Romero et al. 1992; Ruiz and Romero 1999). Each structure was designed for the same level of

45 traffic; therefore, layer thicknesses of the different materi- als varied. Granular and soil–cement base materials were compared with a third combination of a gravel–cement base with a soil–cement subbase. Two different subgrade materials were also used. Results under full-scale loading indicated that the cement-treated materials provided im- proved resistance to rutting and exhibited no cracking as compared with the unbound granular material. An older APT facility at Washington State University examined the fatigue performance of sulfur-modified AC used as a base layer (Mahoney and Terrel 1982). Material performance under full-scale loading and in laboratory wheel-tracking tests indicated that the modified mixture was more sensitive to strain level than a conventional AC mixture. As a result, longer fatigue lives were demon- strated for these modified materials at larger strain levels. The ALF program in Australia recently completed APT tests to assess erosion of three subbase layers constructed beneath a lean concrete base and a concrete surface layer (Vuong et al. 2001). The performances of unbound and bound crushed granular materials were compared with that of a lean concrete subbase. Results under full-scale loading indicated that unbound subbases provide adequate resis- tance to erosion. The SA–HVS program also examined the performance of three rehabilitation options for lightly cemented pavements (LCP) (Steyn et al. 1997). The performance of a double seal, a thin AC overlay, and a crushed-stone base with a double seal surface were compared under full-scale loading to relatively deep and shallow new LCP pavements. The material evalua- tion results associated with this study included equivalent performance of the rehabilitated and new pavements in terms of permanent deformation at the surface. The reha- bilitated pavements failed only in terms of bleeding of the double seal and surface deformation and pumping of fines in the wet condition for the other options. Further results from this study are discussed in related chapters. The SA–HVS program examined other rehabilitation options through full-scale loading and laboratory charac- terization of recycled asphalt pavement (RAP) used as a base layer (Servas et al. 1987). Based on the laboratory re- sults, RAP contents from 30% to 70% did not affect perform- ance in terms of indirect tensile strength, resistance to rutting, or resistance to indirect tensile fatigue. Full-scale loading tests confirmed these results and established RAP as a viable ma- terial with performance comparable to conventional AC. Subgrade Relatively few APT tests have been conducted specifically to evaluate subgrade materials. In chapter two, LINTRACK ex- periments in The Netherlands on a sand subgrade were dis- cussed (Bhairo et al. 1998a,b). The researchers concluded that the Shell subgrade strain criterion appeared to be applicable for subgrade sands that are prevalent in that country. One study conducted as part of the HVS–CRREL pro- gram used full-scale loading of different pavement struc- tures in developing an understanding of response in terms of subgrade strain as a function of subgrade soil type and moisture content (Lynch et al. 1999; Odermatt et al. 1999). FWD testing during a thawing cycle was conducted to de- termine stiffness reduction factors for use in design and analysis. The results are discussed in chapter five and elsewhere. The HVS–NORDIC program also conducted APT tests at a Finnish site during a thawing cycle, with the goal of setting deformation limits for subgrade materials for this critical environmental condition (Saarelainen et al. 1999). Three pavement structures with thin AC surface layers, crushed-rock base layers, and sand subbase layers of equivalent thicknesses were constructed on a frost- susceptible lean clay subgrade. One of the structures con- tained a reinforcing steel mesh at mid-depth of the base layer. A sand filter layer lay beneath the subgrade layer to control the groundwater level during the thawing cycle. All three structures failed as the result of cracking in the sur- face layer and permanent deformation from the underlying subgrade layer. The reinforcing steel mesh reduced defor- mation by 50% in the unbound layer. However, as has been found generally with reinforcement, the pavement first had to deform before the reinforcement became effective. CURRENT RESEARCH Ongoing testing at the NCAT test track in Alabama will produce further performance results for AC surface layers (Brown and Powell 2001). Rutting is the expected mode of distress for the 46 mixtures constructed on nominally iden- tical pavement structures. These mixtures include coarse- and fine-graded Superpave mixtures and SMA mixtures. The effects of aggregate type and binder on performance will be evaluated through the use of full-scale loading and performance monitoring and laboratory testing. At the con- clusion of the trafficking of phase 1, these mixtures exhib- ited no visible fatigue cracking and limited, but measur- able, permanent deformation. An extensive laboratory mixture testing program to determine the best performance test for assessing rutting performance is also ongoing. The full-scale trafficking facility at Mn/ROAD has also collected performance monitoring data for AC, concrete, and aggregate surface layers in 40 pavement sections con- structed on two different subgrades (Newcomb et al. 1999). This facility uses actual truck trafficking on a mainline fa-

46 cility and full-scale trucks on a low-volume road. This study focuses on the development of a mechanistic– empirical pavement design procedure and guidelines for truck load restrictions during spring thaws. The impact of material properties on performance is being explored using the same performance data used in achieving the primary goals. An aircraft load simulator at ERDC–GSL in Atlantic City is also currently being used to verify and validate a 3- D pavement design and evaluation program (Lynch et al. 1999). Data required for reaching this goal can also be used to assess the impact of material properties on per- formance. The LA ALF program is currently using full-scale load- ing to determine the effectiveness of using RAP as a base material in an inverted pavement structure (Metcalf et al. 1999). The performance of the RAP base layer is being compared with that of a crushed-stone base layer in nomi- nally identical pavement structures. An international cooperative program involving 11 European countries is also ongoing (Hildebrand et al. 2001). This program builds on previous efforts organized under the OECD (Road Transport Research). The new pro- gram, organized under the European Cooperation in the Field of Scientific and Technical Research (COST), con- sists of several tasks aimed at improving pavement re- search with accelerated load testing. One of these tasks mirrors the efforts of this synthesis to summarize previous and current research in APT. Identification of new and in- novative future research is also planned. Evaluation of ma- terials and tests from previous efforts will be documented, and coordination of research in this area among the differ- ent APT facilities should produce significant findings. SUMMARY The primary objective of many full-scale APT programs is to evaluate pavement material response and performance. APT programs have produced significant findings that al- low for validation of existing materials and implementation of new and innovative materials. APT testing programs al- low for performance-based evaluation of these materials, which is often related to material characterization pro- grams and testing in the laboratory. This chapter has em- phasized the importance of considering differences be- tween APT and laboratory characterization in terms of loading and environmental conditions, measurement and analysis techniques, and failure definitions. The following are a selection of lessons learned through APT in the field of materials and tests that provide evi- dence of the wide scope of applications that were discussed in this chapter. • Guidelines for the use of marginal base course mate- rials were established. This is a direct result of the application of APT for understanding the mecha- nisms that affect pavement performance. • The effect of soil type and moisture on the perform- ance of subgrade under freeze–thaw conditions is be- ing quantified. • The ability to monitor the change in stiffness of pavement layers by means of SASW during traffick- ing is being used to evaluate the performance of bound materials in pavements, particularly under wet trafficking conditions. With respect to the latter, the use of scaled APT in conjunction with full-scale APT has been reported to be of value. • A wide range of materials have been evaluated for use in the various layers of the pavement structure. In the process, information has been collected that will enable the evaluation of newly defined laboratory test guidelines such as Superpave specifications for char- acterizing materials in terms of performance. The mate- rials include stiff modified binders (SBS and EVA), ce- ment-modified base course, emulsion-treated natural gravel, ATPB, and RAP, to name only a few. • Conditions conducive to the improved fatigue perform- ance of HMA were identified in several programs, lead- ing to changes in structural configuration. • There was clear evidence that modified binders out- performed conventional binders in terms of resis- tance to fatigue and permanent deformation. The new APT programs that have been initiated are ex- pected to increase the already wide range of applications in materials and tests. These programs should also benefit from the COST 347 study that is underway in Europe.