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.

11 CHAPTER TWO EVALUATION, VALIDATION, AND IMPROVEMENT OF STRUCTURAL DESIGNS INTRODUCTION This chapter discusses APT research to enhance the struc- tural design of pavements. In structural design, the stiffness and thickness of the pavement layers are selected to ensure an adequate support structure such that the bearing capacity of the underlying subgrade is not exceeded. The chapter cites studies that were selected to present a generic overview of APT practice and research on pavement structural designs. Structural designs form the core of pavement engineer- ing. It is therefore not surprising that APT programs focus strongly on this topic. However, as is well known, design cannot be considered in isolation. This is because of its strong interaction with other fields of pavement engineer- ing, such as materials and vehicle–pavement–environment interaction. The net result is that discussions on the topic must take into account the total system, and the process is often iterative to account for changes that take place over time, particularly in the case of materials. The same can be said of changes that take place in vehicle configuration. When considering designs, pavements are normally classi- fied into two broad categories, flexible and rigid (AASHTO 1993). Conventional flexible pavements gener- ally have a composite layered structure with some form of asphaltic material in the upper layers. Full-depth flexible pavements have one or more layers of asphalt directly on the subgrade. Nonasphaltic base and subbase courses gen- erally consist of some type of natural material or crushed stone that may or may not be stabilized. Rigid pavements consist primarily of a layer(s) of concrete separated from the subgrade by a base course layer. This chapter will consider the various aspects of struc- tural design in relation to the composition of the pavement, namely AC, portland cement concrete (PCC), and composite materials. [For this synthesis, hot-mix asphalt (HMA) was considered to be a synonym for AC. Accordingly, the acro- nyms HMA and AC should be read as synonyms throughout the report, as appropriate.] The discussion will focus on guide- lines for evaluating, validating, and improving designs with notes on possible negative features. Unconventional struc- tures such as block pavers will be considered in chapter seven, as will ancillary aspects of pavement design. The results from the survey questionnaire will be pre- sented before discussing the wide variety of applications that were found in the literature. QUESTIONNAIRE SURVEY The responses to Questions 2.1 to 2.12 on structural com- position are reflected in Figures C15 to C26 in Appendix C. This section of the questionnaire was seen as an oppor- tunity for owners and managers to indicate how the capac- ity of their facilities was being deployed for APT studies. The responses were synthesized and the results are con- tained in the following list. The respondents’ views on structural composition are presented in Table D2 in Ap- pendix D. • APT programs are focused on the structural perform- ance of the pavements, as well as functional perform- ance in a ratio of about two to one (Figure C15). • Most of the APT work thus far has been focused on the asphaltic component in the pavement structure. This is not surprising as this material lends itself to APT. However, of equal importance, is that APT has been conducted on granular layers and concrete pavements (Figure C16). • Figure C17 indicates that tests have focused on all forms of distress that occur in surface seals. • Evaluation of performance of pavements with clayey, sandy, and granular materials has focused on perma- nent deformation (Figures C18 and C19). • The primary and not unexpected focus in APT pro- grams on stabilized pavements has been on cracking (Figure C20). • In contrast, Figure C21 shows that the two major forms of distress of interest in asphalt pavements are rutting and fatigue. A few programs have also been focusing on two other important issues, namely mois- ture damage and stripping and aging; however, it is apparent that not much work has been done in these fields. • Cracking is the primary form of distress examined in jointed concrete pavements in APT. Joint failure and load transfer have only been investigated to a limited extent (Figure C22). • Four forms of distress of composite pavements have been investigated; rutting, fatigue, cracking, and debonding (Figure C23). • For functional performance, safety and roughness were the two aspects studied most (Figure C24). • Rutting, skid resistance, and roughness were featured most prominently in the studies on safety (Figure C25).

12 • Very few respondents reported studies on environ- mental aspects; two had studied noise and one dust pollution (Figure C26). APPLICATIONS OF ACCELERATED PAVEMENT TESTING TO ASPHALT PAVEMENT DESIGNS Odéon et al. (1997) reported on LCPC APT tests in which different asphalt base pavements were evaluated in terms of fatigue performance. The pavements were constructed on a fairly weak subgrade [California bearing ration (CBR) between 5% and 10%] and a subbase consisting of 400 mm of well-graded, untreated granular material. Conventional and modified (BB), improved (GB), and high modulus (EME) AC were used for the base materials. (It should be noted that in this report “stiffness” has been used as a generic term. It should be read as a synonym for modulus and stiffness modulus.) These high-modulus ACs are typically constructed on very stiff subbases, and the researchers found that the high-modulus asphalt mix, in particular, was very sensitive to thickness, especially when placed on a deformable sub- base. Increasing the thickness of the EME from 90 to 110 mm increases the fatigue life 2.5 times; however, rapid degradation of the EME base was apparent with the onset of cracking. Under carousel (circular) loading, modified base pavements outperformed conventional pavements. Harvey et al. (2000) tested pavements conforming to California DOT (Caltrans) specifications. Pavements tested included those with Asphalt-Treated Permeable Base (ATPB), termed “drained” pavements, and those with stan- dard aggregate bases, termed “undrained.” HVS tests have confirmed that the total pavement thickness developed us- ing the Caltrans pavement design procedure is generally ade- quate to prevent rutting through permanent deformation of the subgrade and unbound granular layers. Fatigue cracking of pavements for higher traffic levels with weaker subgrades is a concern. They point out that innovative pavement designs such as the “rich bottom” (high binder content) concept or the use of modified binders significantly improve fatigue per- formance of pavements compared with conventional de- signs. The use of higher binder contents in the lower struc- ture of the pavement is deemed feasible given that deformations and stress levels under loading are greatly reduced with depth. Rut resistant mixes must be used in the “critical zone” for rutting, found to be within 100 to 150 mm of the pavement surface (Harvey et al. 1999). The use of drainage layers (ATPB) in pavements has led to stripping incidents where water may remain trapped within the pavement system because of faulty edge and transverse drains. As an alternative to ATPB, Harvey et al. (2000) recommended that standard asphalt base layers be used. In addition, they emphasized the need for adequate compaction (less than 8% voids in the mix after construc- tion) to reduce permeability. They also proposed that the thickness of these layers be increased to delay the initiation and propagation of cracking. This approach may be further improved by the use of a rich bottom layer. They noted, however, that drainage layers may still be required to re- move water seeping into the pavement from the subgrade. If ATPB layers are required, then the California researchers suggest the use of higher binder content, modified binders such as asphalt rubber, and additives such as lime or anti- stripping agents. Geotextile filters should be used to pre- vent clogging of the ATPB layer and maintenance practices for cleaning edge and transverse drains should be in place. They further recommended raising the “gravel factor” for ATPB from the current 1.4 to 2. Kekwick et al. (1999) outlined the influence of the SA– HVS program on pavement design philosophy. In South Africa, HVS testing has been used to validate the perform- ance of well-balanced, deep pavement structures. These pavements are constructed with materials such that there is a gradual decrease in stiffness with depth in relation to the bearing capacity of the respective layers. HVS testing has demonstrated that poorly balanced, shallow pavements, where most of the stiffness of the structure is concentrated at the top of the pavement, are normally load sensitive. These types of pavements may appear to have adequate bearing capacity but deteriorate rapidly under overloaded conditions. However, they warn against increasing the test wheel load to levels far above those of the standard design load. This may induce failure mechanisms that will never manifest under normal traffic loading conditions, espe- cially in the case of bound layers. The SA–HVS testing program has been instrumental in the development of the South African Mechanistic Design Method for Pavements (Theyse et al. 1996). It is an exam- ple of how APT can benefit pavement engineering overall. They discuss how HVS test results were used to develop transfer functions for the mechanistic–empirical modeling of the permanent deformation of unbound pavement layers in pavements with asphalt and granular base layers as well as granular and stabilized subbase layers. This method was applied to establish standard pavement structures for use in different climatic regions of South Africa and different lev- els of design traffic. These standard pavement structures are cataloged in manuals for implementation by the road industry and have, over the years, been validated and re- fined in the field using HVS testing. The significant amount of data collected during HVS testing of numerous types of pavement structures has allowed confidence limits to be established to assess the reliability of design method- ologies (Structural Design of Interurban and Rural Road Pavements 1980, 1985, 1996). Sharp et al. (1999a) reported Accelerated Loading Facil- ity (ALF) tests on a test section with a high bitumen content

13 (0.5% above optimum) mix in the lower base that did not show excessive deformation in that material. This indicated that these layers could be used in thinner asphalt structures than previously thought necessary and/or that the addition of more bitumen was possible to further improve fatigue life. ALF testing was used to validate fatigue transfer func- tions for asphalt and cement-treated crushed rock (CTCR) in the Austroads Pavement Design Guide (Austroads, for- merly NAASRA—the National Association of Australian State Road Authorities). These are primarily based on em- pirical relationships derived from overseas data (e.g., Shell) and field-performance data. Hugo et al. (1999c) reported on Texas Mobile Load Simulator (TxMLS) tests completed in Victoria, Texas, to evaluate the widely used district pavement design using lo- cal siliceous river gravel flexbase with thin asphalt surfac- ing. They reported that high construction variability and high asphalt void content led to early fatigue failure of the AC in the test sections. Stabilizing subgrade layers en- hanced the structural capacity of the pavement. Deep- seated variability in the pavement foundation, however, in- fluenced its performance, and lenses of poor materials af- fected the pavement surface profile. Bhairo et al. (1998a,b) reported on LINTRACK (a full- scale HVS for APT) experiments that evaluated the fatigue performance of two full-depth asphalt pavements with varying asphalt thickness on a sand subgrade. One of the structures had a total asphalt thickness of 150 mm consist- ing of two layers, an 80-mm bottom layer and a 70-mm top layer. The second structure had a single 75 mm layer. For the thinner structure, LINTRACK loading led to structural fatigue cracking in the asphalt (bottom-to-top) and surface cracking (top-to-bottom). The researchers concluded that the Shell subgrade strain criterion appeared to be very ap- plicable for subgrade sands as used in The Netherlands. Addis (1989) reported on trials undertaken in the Pave- ment Test Facility at the TRL to evaluate the relative performance of dense bitumen macadam (DBM) and heavy duty macadam (HDM). In general in Great Britain, mac- adam consists of a high-quality aggregate with large, sin- gle-sized particles (37–53 mm), which is stabilized by fill- ing the voids with a suitable material. Typically, the macadam is defined more specifically in relation to the ma- terial used for filling the voids; for example, waterbound macadam has a filler of natural material with a low plastic- ity, whereas slurry-bound macadam has a filling of slurry. The research team reported that the results of the acceler- ated tests showed little difference in the overall perform- ance of the two materials. The initial rate of rutting of the materials was different, that of the HDM being lower than the conventional DBM. The team found, however, that the HDM generally weakened more under the influence of very heavy wheel loads. They concluded that both the ob- served and measured variability in the compacted quality of the bituminous materials, in particular that associated with the HDM, could have been a major contributor to some of the later life performance differences. APPLICATIONS OF ACCELERATED PAVEMENT TESTING TO CONCRETE PAVEMENTS Caltrans testing of concrete pavements (Harvey et al. 2000) has indicated the importance of the use of dowels and non- erodable bases for heavily trafficked, jointed concrete pavements. The Caltrans researchers point out that dowels are effective in restricting the curling of concrete slabs along transverse joints. In the same way, tie bars were found to be useful in restricting curling along longitudinal joints. They suggest seeking higher than currently required flexural strengths together with material having low coeffi- cients of thermal expansion to reduce the thickness of con- crete slabs. Harvey et al. (2000) further stated that shorter slab lengths are required for high-shrinkage hydraulic cement to prevent premature top-to-bottom cracking in the slabs. Fur- thermore, they suggest that joint spacing requirements be made a function of climate. According to Roesler (1998), joint spacing should be less than 4 m for a slab thickness of 200 mm. From the same test program, Roesler et al. (1999) reported that the performance of fast-setting hydraulic ce- ment concrete (FSHCC) pavements was very similar to that observed for PCC pavements. Vuong et al. (2001) reported on plain concrete pave- ments tested using the ALF at Goulburn, New South Wales in Australia. Four pavements were tested to assess fatigue performance, and five pavement sections were tested to as- sess erosion performance. The site chosen had a high diur- nal temperature change, on the order of 20ºC, which pro- duced significant interaction of loading and slab curl. The influence of dowels, shoulder ties, and slab thickness (150 mm, 175 mm, and 200 mm), as well as erosion of unbound and bound subbases, was investigated. Erosion was inves- tigated by wetting of the pavement before and during traf- ficking. Because of the effect of the shading of the pave- ment under the ALF on curling of the concrete slabs, conventional (rigid) trucks were also used to evaluate load- ing response. The following findings are relevant to struc- tural design: • No fatigue failure had been induced in the concrete slab after 170,000 load applications of an 80-kN ALF axle load. • When a slab 150 mm thick with undoweled trans- verse joints was tested with ALF 40-kN, 60-kN, and 80-kN dual-wheel loads, the movement at the center of the slab was the same for all three wheel loads.

14 However, when an 80-kN standard axle was intro- duced using a rigid truck, movement at the slab cen- ter was six times greater than that produced by ALF. This reflects the importance of slab curl, and the ef- fects of pavement shading and loading configuration. • Deflection data showed that slabs lost support at the corners and edges during the night and at the center of the slab in daytime because of curling. • The presence of tied shoulders significantly reduced the curling behavior of the slab during the night (up to 80%). • The presence of dowels in transverse joints signifi- cantly reduced curling behavior, which raised slab centers during the daytime and hence loading deflec- tions (up to 47%); however, the dowels allowed higher movement at corners without a shoulder dur- ing the night. • Increasing slab thickness reduced curling of the slab during the daytime and hence reduced bending stresses. • Deflections under load increase rapidly for a slab in a curled state until contact is made between the base and subbase, whereupon there is little further increase. • For slabs with and without dowels, erosion occurred in the subbase of unbound granular material. This material is unsuitable for subbases under plain con- crete pavements subjected to heavy loading. • A very small amount of erosion occurred in a heavily bound subbase of a concrete pavement without PCC pavement dowels. • Erosion did not occur with a lean mix concrete sub- base, and there was no clear evidence of erosion in a heavily bound subbase with dowels. Lean mix con- crete subbase and heavily bound subbase with dowels may be suitable for plain concrete pavements sub- jected to heavy loading. Vuong et al. (2001) emphasized that for APT testing of concrete pavements consideration must be given to long- term environmental effects and the possibility of fines moving between the base and subbase, which may change loading stresses arising with slab curl. Draining of this in- terface is considered essential. They concluded and rec- ommended that • Tied shoulders need to be retained in pavement de- sign, • A minimum slab thickness to reduce curling and the effects of curling on pavement performance needs to be specified for pavements subject to heavy loading, and • Unbound subbases are unsuitable under plain con- crete pavements subject to heavy loading. Balay et al. (1992) reported on LCPC APT tests on con- crete pavements aimed at validating the thickness designs in the French design catalogue of new pavement structures. The goal was to determine whether three concrete pave- ment structures proposed in the French design catalogue were equivalent with regard to their performance under traffic. The following three structures from the catalog were tested: • Short slabs with dowels built on a treated subbase, • Short slabs without dowels built on a treated subbase, and • Short slabs built on an untreated subbase. For the third structure, slabs with normal and lean con- crete (300 kg/m3 cement vs. 140 kg/m3) were tested. A comprehensive paper on the numerical analysis of the test track was presented by Balay and Goux (1994). They con- cluded that the APT results accurately reproduced modes of functioning and distress of actual concrete pavements. They found that the functioning of the pavements was re- produced sufficiently realistically through their Finite Ele- ment (FE) analysis to be useful. Failure of the pavements was characterized by cracking, joint failures, and pumping of fines. As expected, the slabs with dowels built on the treated subbase performed the best. The lean concrete slabs on the untreated subbase failed completely halfway through completion of the tests, necessitating repair. Strengthening of the subbase signifi- cantly improved the performance of the concrete pave- ments. The researchers found that the thickness of some standard designs could be reduced slightly when the sub- surface conditions were favorable. This required good effi- cient drainage with a nonerodible soil surface under the concrete slab. Paved shoulders were also considered neces- sary. A number of tests have been completed at the NAPTF facility in Atlantic City, New Jersey. Guo and Marsey (2002) presented some important details relating to the ef- fect of curling of the slabs that need to be taken in to ac- count during APT. • Measured deflections at the center of the slab re- mained effectively constant, whereas the deflections at the joints and corners varied significantly during testing. • Deflections at joints and corners are significantly lar- ger in winter compared with summer. Joint load transfer capability was also lower in winter. • Analysis indicated that slabs were always curled up in winter and this was more significant on a stronger subgrade. • The sum of deflections on both sides of joints, re- main almost unchanged when traffic direction is re- versed. However, sides of joints vary significantly from summer to winter.

15 APPLICATIONS OF ACCELERATED PAVEMENT TESTING TO COMPOSITE STRUCTURES HVS testing has been instrumental in validating the effec- tiveness of inverted pavement structures, which are now used extensively throughout South Africa. These structures incorporate stabilized or lightly cemented (<4%) subbase layers that provide support to granular or asphaltic base layers. The stiffness of these stabilized subbase layers, while intact, are higher than that of the base layers. This al- lows adequate compaction of the base layer, and in the case of asphaltic base layers, reduces the development of hori- zontal tensile strains beneath the layer, hence extending the fatigue performance of the pavement structure. In the case of high-quality granular bases, the stiff subbase layer confines the base, and this “sandwich” effect has been shown to sig- nificantly increase the shear strength of high-quality granular bases. The influence of climate as well as traffic level is ac- counted for in the structural design of pavements. Further validation of this phenomenon was reported by Gramsammer et al. (1999) in France and Harvey et al. (2000). The APT tests at LCPC in France were done to de- termine the optimum thickness of the unbound granular material on top of a cemented subbase and below the as- phalt surfacing layers. A specific optimum thickness was not reported. The Australian Road Research Board (ARRB) (Sharp et al. 1999a) reported on ALF tests aimed at the evaluation and improvement of structural designs. Testing included trials to investigate the influence of thickness on the per- formance of CTCR pavements constructed to similar stan- dards. Pavements with CTCR layers of 200 mm and 300 mm were tested. These experiments were complemented by the testing of similar pavements with and without a bi- tumen heavy-cure coat interlayer and a section constructed in one lift instead of the usual two or three lifts. ALF test- ing confirmed that the typical failure mode was the result of the debonding of the CTCR base layers, followed by the ingress of water at the interfaces and the subsequent ero- sion of the bottom of the upper layer, leading to a failure of the top layer. As a result of these tests, construction prac- tice was changed to allow the construction of cement- treated bases in single layers rather than in multiple lifts. ALF testing programs were undertaken to evaluate the in situ stabilization of marginal sandstone material. The importance of curing stabilized materials was emphasized. A lack of curing, especially of the slag/lime material, re- sulted in some drying out of the top of the bound material, and a considerable number of shrinkage cracks were ob- served before the application of the prime and asphalt sur- facing. Crushing of the bound material was apparent under ALF loading, which led to erosion of the base and subse- quent pumping of fines. In view of the results, the re- searchers recommended that stabilized pavements required 7 days moist curing or that they be sealed immediately with an approved curing compound. Alternatively, the next layer of the pavement should be constructed to prevent excessive dry- ing of the stabilized surface that may lead to cracking. De Beer (1990) and De Beer et al. (1991) also reported on the performance of cemented base and subbase layers under APT. The program covered a period of 6 years. Spe- cific failure mechanisms were identified and guidelines were developed for the use of cementitious layers in pave- ment structures. Sharp et al. (1999a) reported on ARRB–ALF tests con- ducted to evaluate geotextile-reinforced seal pavements. Geotextiles were used to strengthen pavements with clay subgrades in regions where gravels are scarce and of low quality. They indicated that the geotextile-reinforced pave- ments performed satisfactorily, with little distress observed af- ter the design traffic loading had been applied, even when testing was conducted near the edge of the pavement adja- cent to a filled dam. Guidelines for the design, construction, maintenance, and management of geotextile-reinforced seal pavements were prepared and issued. Sharp et al. (1999b) discussed ALF trials undertaken to validate the ERDC (formerly USACE Waterways Experi- ment Station) tentative classification scheme for lateritic gravels for road and airfield pavement construction. For these trials the APT test objectives included • Establishing the performance of gap-graded ridge gravels and a relative measure of the performance of “good” and “poor” lateritic gravels, • Comparing the performance of the lateritic gravels when they are constructed to two compaction levels, and • Comparing the performance of lateritic gravels when they were constructed “full depth” (300 mm in two layers on a clay subgrade) and 150 mm in one layer on a cement-treated subbase (CTSB). ALF testing indicated that in a dry state there was not a significant difference between the performance of the good and poor lateritic gravels. In addition, the APT testing indi- cated that the level of compaction was not a major factor affecting the performance of the lateritic gravels. The per- formance of the two full-depth lateritic layers compared with the thinner layer on a CTSB was similar. Higher de- flections were apparent in the full-depth lateritic layers; however, it was suggested that this was acceptable given the cost savings inherent in constructing an unbound sub- base layer rather than a CTSB. Kadar et al. (1989) reported extensively on the perform- ance of slag road bases under APT. Their findings indi-

16 cated that the blast furnace slag could be used in the place of high-quality, crushed-rock road bases. The slag and sta- bilized slag materials proved suitable for use as base mate- rial provided they are protected from excessive tensile stresses. The findings of the APT tests provided a sound basis for developing guidelines for the design of pavement structures with this type of material. In particular, they gave insight into the manner in which the structural behav- ior changes as the material characteristics change. The im- portance of the uniformity of the mixing of stabilized ma- terials was again demonstrated. The researchers suggested increasing the content of the binder (blends of granulated slag and lime) used for stabilizing to improve the uniform- ity of the mix. The increase was not needed for strength purposes. Another problem that was identified was the negative effect that thin leveling layers had on the per- formance of the pavement. It was emphasized that these should be avoided under any circumstances, because they lead to delamination. Vuong et al. (1996) reported on the Beerburrum II ALF trials, which contributed to the state of the art for the de- sign and construction of stabilized and unstabilized granu- lar pavements. These trials consisted of 34 experiments completed in Australia, where 3 million light-load cycles were applied to each of 10 pavement types. They found that pavements with high-quality crushed-rock base layers benefit significantly from increased compaction because of the reduced influence of moisture. It was found that the degree of saturation is of paramount importance. For granular materials with low plasticity, the degree of satura- tion is considered to be a better indicator than optimum moisture content of the aggregate. In situ stabilization us- ing 2% bitumen and 2% cement was found to be more ef- fective than other treatments that had been explored as re- placements for the sandstone and high-quality crushed rock. This included stabilization with bitumen only and different proportions of bitumen and cement. Saarelainen et al. (1999) report HVS–Nordic tests done to evaluate a thawing and frost-susceptible subgrade. Three pavement structures were tested, each consisting of a thin asphalt surfacing (50 mm), a base layer of crushed rock (200 mm), and a subbase layer of sand (250 mm). In one of the structures, a reinforcing steel mesh was installed in the base layer. The subgrade was constructed using natural lean clay, referred to locally as dry crust clay. The pave- ment was cooled by exposure to atmospheric frost tem- peratures until there was a frost penetration of approxi- mately 1.2 m. The pore pressure development was monitored. The resulting frost heave was typically about 50 to 70 mm. Before HVS testing, the pavement was thawed to a depth of 0.9 m. The decrease in pavement stiffness with thawing was confirmed using FWD testing. It was found that, with the development of rutting in the re- inforced pavement, the steel mesh significantly increased the strength of the base layer and ultimately the perform- ance of this pavement. The moving test wheel caused fast stress pulses to the soil and this resulted in increased pore pressure. Ruiz and Romero (1999) outlined the Spanish CEDEX APT program, the main objective of which is to improve pavement structures detailed in a pavement design cata- logue. The catalogue lists pavement options depending on subgrade quality and traffic level, including flexible, semirigid (flexible with some stabilized or bound layers in the composite structure), and rigid pavements. These struc- tures were determined based on experience and analytical design approaches. The APT facility is being using as a tool for validating the pavements in the catalogue, but also to evaluate the performance of the pavements in the Span- ish road network. The first CEDEX test evaluated two pavement options for traffic levels of 50 to 200 trucks per day and subgrades with a CBR of between 10% and 20%. One of the sections consisted of a 150-mm asphalt surfacing and base over 500 mm of granular material. The other section consisted of 180 mm of asphalt over 250 mm of granular material. The influence of thickness variations of the asphalt and granu- lar materials were investigated. CEDEX APT testing, to- gether with an analysis of the performance of in-service pavements, led to the elimination of the second section (180 mm) asphalt from the design catalogue. These pave- ments did not perform as expected. An equivalency be- tween granular materials and asphalt mixtures was estab- lished (10 mm of asphalt mixture being equivalent to 30 mm of granular material). Subsequent testing was done to evaluate additional catalog pavement structures with dif- ferent types of base materials; that is, granular, soil– cement, and gravel–cement. The influence of subgrade strength was also investigated. CEDEX testing validated the performance of the catalogue designs; all the pave- ments performed adequately under the conditions consid- ered in the design. The flexible pavements exhibited crack- ing after trafficking was continued beyond the level to which in-service pavements would have been subjected. Pavements with cement-treated layers did not crack during the tests and performed better than flexible pavements. Pavements with soil–cement bases performed similarly to the pavements with gravel–cement layers. This was unex- pected and contrary to results of the analytical design of these pavements. Metcalf et al. (1999) discussed ALF testing done in Louisiana to evaluate nine different soil–cement base courses under accelerated loading to failure. In-place ce- ment-stabilized select soils are the primary base material for the vast majority of noninterstate pavements con- structed in Louisiana. Such pavements are usually surfaced with 90 mm of AC. Metcalf points out that factors influ-

17 encing the performance of these base types are nonuniform cement distribution, inadequate mixing of the cement and soil, and the high probability of shrinkage cracking. This leads to nonuniform support of the pavement, which results in isolated pavement failures and marked variability of the pavement performance. Cracking of the cement-stabilized bases generally results in block cracks at the pavement sur- face, which allows moisture to infiltrate the pavement structure and negates the rideability and performance. The following findings were found to be relevant: • The crushed-stone base structure outperformed the soil–cement structure. • All stabilized base structures failed because of sof- tening and erosion of the materials and loss of sup- port under the asphalt layer. Shrinkage cracks in the stabilized base generated reflection cracks in the as- phalt surface layer. • The higher cement content (10%) in the in-plant- mixed soil–cement only slightly increased the life of the structure when compared with the low cement content (4%) of the in-plant-mixed stabilized base. • In-place-mixed soil–cement performs similar to the in-plant-mixed soil–cement. • Plastic fibers do not significantly improve the per- formance of the soil–cement base. • At the same cement content, increasing the thickness of the soil–cement improves the performance of the road structure. • Under high moisture conditions, an inverted pave- ment outperforms the soil–cement base pavement, as well as the conventional flexible pavement. Based on the findings, Metcalf et al. (1999) concluded that consideration should be given to • The use of an AASHTO layer coefficient of 0.10 for stone-stabilized base, • The use of the inverted pavement configuration, • The use of thicker cement-stabilized bases with lower cement contents, and • Ending the use of fiber reinforcement in cement- stabilized layers and of geogrids in unbound bases. Meng et al. (1999) reported on ALF testing in China to evaluate stabilized base pavement options for heavy design traffic. The options included cement-stabilized soil, lime- stabilized soil, cement-stabilized crushed-stone, fly ash, and lime-stabilized crushed-stone bases. The stabilized base pavements tested were found to fail because of the disintegration of the surface–base interface and not from fatigue cracking of the bases. The performance of the fly ash gravel base pavement was considerably better than the pavements with cement-stabilized crushed-stone and ce- ment-stabilized soil bases. The researchers reported that a relative thick asphalt surface over a stabilized base can significantly improve the pavement rutting resistance. Based on these findings, they recommended that the qual- ity of the materials and the selection of the layers be con- sidered in the pavement design stage to alleviate the ero- sion of the stabilized base at the interface of the asphalt and base layers. They suggested using a waterproofing layer between the asphalt surface and the stabilized base to alleviate the effect of water on pavement failure. Further- more, special consideration should be given to enhancing the bond between surfacing and base layers. Núñez et al. (1999) report on APT research in Brazil on weathered basalts to reduce the costs of low-volume roads. Tests were done on pavements with weathered basalts as base and/or subbase layers. In some of the tests only weathered basalts were used as aggregate, whereas in oth- ers densely graded, sound crushed stone was included as the base layer over the basalt subbase. Two distress mechanisms that affect weathered basalts were identified; crushing in weaker aggregates and lateral displacement in stronger ones. They concluded that weathered basalts may be used as base layers for pavements of low-volume roads. Lynch et al. (1999) discussed APT research by the USACE in Vicksburg, Mississippi. The original California Highway Department design curves for light and medium- heavy highway traffic were used as a basis for airfield de- sign curves. The first of these curves were for 31-kN and 53.4-kN single-wheel aircraft loads. Subsequently, design curves were established for 890-kN single-wheel loads with tire pressures up to 1378 kPa. The concept of an equivalent single-wheel load also stems from this research. Accelerated tests conducted to investigate environmental effects led to refinement of the CBR design procedure to include thaw weakening. Gramsammer et al. (1999) reported on LCPC APT tests in France to evaluate a foam-bound aggregate-graded course (cold mix) inserted between two high-modulus as- phalt courses. After the application of 4.3 million 13-t axle loads, no deterioration was visible on the surface, although core boring revealed a horizontal crack within the foam- bound course a few centimeters above the bottom of this layer. Gramsammer reported that the horizontal crack was the result of shear fatigue rather than flexural fa- tigue. Corte et al. (1997) reported on LCPC APT experiments designed to evaluate the fatigue performance of pavement structures incorporating hot and cold mixes. The purpose of the program was to study the behavior of cold mixes on a deformable (flexible) base. The experiments were initial- ized to evaluate a newly developed emulsion-bound granu- lar material using a modified binder and a “double-coated” cold mix. Tests were also done on conventional hot-mix asphalt and an emulsion-bound material, both serving as

18 reference or benchmarks to compare the relative perform- ance of the experimental materials. The bituminous materi- als were compacted to a thickness of 100 mm on top of a 300-mm, untreated, well-graded crushed-aggregate sub- base, purposely designed to allow deformation in the base. No rutting was observed in the cold mixes after comple- tion of the testing. Each of the bituminous materials tested cracked, although the cracking of the cold mixes was de- scribed as being finer than that found in the hot mix. Major debonding was observed on the conventional emulsion- bound granular material and, to a lesser extent, on the emulsion-bound granular material with the modified binder. The double-coated cold mix did not debond. Gram- sammer et al. (1999) concluded that for cold mixes the nonlinearity of these materials must be taken into account during mechanistic analysis and that the cold mixes mani- fested less cracking. SUMMARY This chapter has covered a number of issues related to the structural design of pavements. Structural designs are typi- cally tested at fixed-site test facilities, whereas rehabilita- tion designs are usually evaluated in the field, although in a few cases rehabilitation designs are evaluated on fixed-site lanes that had previously been tested to failure. For asphalt pavements, APT has demonstrated the bene- fits of using very stiff, high-modulus asphalt for bases, rut- resistant mixes in the upper structure of the pavement, and “rich bottom” layers to increase fatigue resistance, and has shown the importance of well-balanced, deep-structure pavements. For concrete pavements, the influence of thickness, dowels, and tie bars has been investigated, as well as the influence of subgrade support and curling and warping. For composite structures, the effectiveness of in- verted structures has been illustrated. APT has contributed to advances in the field of stabilization of marginal materi- als to strengthen pavements and the use of geofabrics for reinforcement. APT has also been instrumental in validating and refin- ing agency structural design guidelines. In addition, im- provements in structural design have been brought about by the insight gained on the effect of a number of factors on pavement performance, including • The influence of concrete slab configuration, • The influence of support under concrete slabs, • The influence of water on performance and related failure mechanisms, • The importance of bond between layers and the quantification of the effect, and • The interaction between structural composition and material characteristics.