National Academies Press: OpenBook

Practices for Unbound Aggregate Pavement Layers (2013)

Chapter: Chapter Four - Unbound Aggregate Base Characterization for Design

« Previous: Chapter Three - Granular Base and Subbase Construction Practices
Page 55
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 55
Page 56
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 56
Page 57
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 57
Page 58
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 58
Page 59
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 59
Page 60
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 60
Page 61
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 61
Page 62
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 62
Page 63
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 63
Page 64
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 64
Page 65
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 65
Page 66
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 66
Page 67
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 67
Page 68
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 68
Page 69
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 69
Page 70
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 70
Page 71
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 71
Page 72
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 72
Page 73
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 73
Page 74
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 74
Page 75
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 75
Page 76
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 76
Page 77
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 77
Page 78
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 78
Page 79
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 79
Page 80
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 80
Page 81
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 81
Page 82
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 82
Page 83
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 83
Page 84
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 84
Page 85
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 85
Page 86
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 86
Page 87
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 87
Page 88
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 88
Page 89
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 89
Page 90
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 90
Page 91
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 91
Page 92
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 92
Page 93
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 93
Page 94
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 94
Page 95
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 95
Page 96
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 96
Page 97
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 97
Page 98
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 98
Page 99
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 99
Page 100
Suggested Citation:"Chapter Four - Unbound Aggregate Base Characterization for Design ." National Academies of Sciences, Engineering, and Medicine. 2013. Practices for Unbound Aggregate Pavement Layers. Washington, DC: The National Academies Press. doi: 10.17226/22469.
×
Page 100

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.

55 chapter four UNBOUND AGGREGATE BASE CHARACTERIZATION FOR DESIGN INTRODUCTION This chapter presents an overview of unbound aggregate mate- rial characteristics and structural layer behavior as the primary structural component in flexible pavement systems, as well as subbase layers under concrete pavement slabs. A thorough review of different aggregate test procedures and character- ization methods commonly used to model granular pavement layer responses and permanent deformation behavior is imper- ative to facilitate better designs of pavement systems and ulti- mately ensure adequate performance under repeated loading. The main load transfer mechanism governing unbound aggregate structural layer behavior under loading is high- lighted. Important aggregate physical properties affecting granular layer strength, modulus, and permanent deforma- tion behavior are discussed in detail in this chapter. Com- monly used models to characterize the elastic or resilient, as well as permanent, deformation behavior of unbound aggre- gate materials are discussed with a review of typical pave- ment stress states and initial loading conditions affecting the primary features of aggregate repeated-load behavior. Both empirical and M-E methods developed for designing unbound aggregate pavement systems are summarized with histori- cal perspective by listing their advantages and limitations. State-of-the-art approaches, including the stress-dependent and anisotropic models from the recent ICAR research find- ings, are described in detail for proper characterization of unbound aggregate pavement layers. Accordingly, different mechanisms contributing to the failure of pavement systems with unbound aggregate layers are reviewed to emphasize the importance of aggregate material quality governing pavement performance. Finally, permeable and open-graded aggregate layers are discussed to review and summarize the most important climatic (that is, moisture and temperature) conditions influencing designs of unbound aggregate pave- ment systems. LOAD TRANSFER IN GRANULAR MATERIALS The mechanism of load transfer in granular materials was first experimentally studied by Dantu (1957) with the help of photoelastic models. From the experiments performed, it was concluded that the stresses in granular materials were not uniformly distributed but were concentrated along load- carrying particle chains. Later Oda (1974) described other experiments in which photoelastic rods were loaded biaxi- ally. Forces across individual particle contacts were moni- tored by counting the resulting interference fringes. Based on experimental studies, the stresses in particulate media are not transferred in a uniform manner but are con- centrated along continuous columns of particles. The parti- cles in between the columns provide only lateral support but do not carry much load. At a critical load, a column will fail and the internal structure will be rearranged. Formation of a new column takes place if particles in that region are favor- ably oriented. The deformation of a particulate mass under increasing load is then mostly the continual collapse and generation of adjacent chains of load-carrying particles. The predominant orientation of particle contacts is in the direc- tion of the major principal stress. Similar results on the load transfer and deformation char- acteristics of granular materials were obtained by Dobry et al. (1989). Using the discrete element approach (Cundall and Strack 1979), Dobry et al. (1989) modeled granular soil as random arrays of 531 elastic, rough spheres of two different sizes. Numerical simulations of these arrays under monotonic and cyclic loading were compared with typical experimental results from triaxial compression tests on a medium-dense uniform quartz sand. From the numerical simulations, they observed that triaxial deviator stresses were clearly transmit- ted by a limited number of “stiff chains” or irregular columns of grains aligned in generally the vertical direction. Therefore, according to the experimental and numerical findings, the deformation pattern of an unbound aggregate layer is directly related to load transfer by shear in the col- umns of particles supported under confinement. The orien- tation of such columns is primarily in the direction of the principal stresses and also is affected by the assembly of the grains and their shape. UNBOUND GRANULAR MATERIAL BEHAVIOR UNDER REPEATED LOADING Unbound aggregate layers in pavements are subjected to repeated load applications as a result of traffic. They undergo both elastic (commonly known as resilient for pavement applications) as well as plastic (permanent) deformations with every load repetition. Figure 35 presents a schematic of typical unbound aggregate behavior under repeated loading with the help of a stress-strain diagram. Note that the relative

56 magnitudes of elastic and plastic components of the total strain depend on several different factors, including traffic load lev- els and speed of operation, thickness and quality of overlying pavement layers (if any), characteristics of aggregates used in construction of the aggregate layer, and subgrade conditions. In a constructed granular layer, the accumulation of per- manent deformation as a result of each load repetition gradu- ally decreases with increased number of load applications. Once the layer has been well compacted to achieve a densely packed matrix, all the subsequent load applications ideally would result in deformations that are mostly elastic in nature. The resilient and permanent deformations of an unbound granular layer can be attributed to different mechanisms. Werkmeister (2003) summarized the Hertz contact theory and suggested that resilient deformation in granular materials is caused primarily by “temporary” deformation of individual grains, whereas permanent deformation takes place because of relative movement of the particles with respect to each other. The initial rapid accumulation of permanent deformation typically corresponds to the rearrangement of particles dur- ing initial compaction and subsequent loading of the pavement layers. After adequate “shakedown” (that is, particle reorienta- tion and rearrangement into a dense matrix) of the material is reached under this initial loading phase, the pavement layers show predominantly resilient deformation provided that the load levels remain below permissible limits. RESILIENT RESPONSE OF UNBOUND AGGREGATE LAyERS Ideally, pavement layer response under traffic loading should be purely elastic, and thus no accumulation of permanent deformation would occur during its service life. Accordingly, mechanistic-based pavement design approaches traditionally have focused on the elastic or resilient response of unbound aggregate layers to predict the critical pavement responses under traffic loading. Figure 36 shows a schematic of typical hysteretic response exhibited by unbound aggregate materi- als under repeated loading. The most important input prop- erty for characterizing repeated load behavior of unbound aggregate layer in pavement analysis has been the “resilient modulus.” Defined as a secant modulus representing hyster- etic stress-strain behavior of materials, the resilient modulus (MR) is a critical material property needed for M-E pavement design methods (Puppala 2008). As shown in Figure 36, the resilient modulus (MR) of a material is defined as the elastic modulus after the material has accumulated a certain amount of permanent deformation. The difference between elastic or Young’s modulus (E) and the resilient modulus (MR) of a material is clearly highlighted in the figure. Equation 2 can be used to determine the resilient modu- lus of a material from repeated-load triaxial test results. Note that in the equation, sd represents the deviator stress or repeated wheel load stress, and er represents the recoverable strain. = σ ε (2)MR d r Key Lessons • Particle-to-particle interlock is critical to the dissipa- tion of stresses in unbound aggregate layers under loading. • Constructed unbound aggregate layers are best compacted into a densely packed matrix to ensure all deformations under vehicle loading are primarily resilient in nature and no significant permanent defor- mation accumulation occurs. STRESS STATES IN UNBOUND AGGREGATE LAyERS UNDER LOADING Pavement stresses are mainly composed of two parts: initial in situ stresses and stresses resulting from moving wheel loads. The initial in situ stresses, static in nature, are the overburden and residual stresses. The initial stresses typically are lower FIGURE 35 Strains in granular materials during one cycle of load application. FIGURE 36 Resilient modulus defined as the elastic modulus of a deformed material.

57 at shallow depths than at greater depths. Compaction-induced residual stresses that are compressive in nature can often exist in the unbound aggregate layers and contribute to the static stress states (Uzan 1985; Barksdale et al. 1997). On the other hand, traffic loading resulting from moving wheel loads induces much higher dynamic stresses than do the static ones. For example, the dynamic vertical stresses become the high- est underneath the wheel where shear loading is nonexistent on a representative pavement element, but at some radial dis- tance away from the wheel, applied vertical stresses decrease and the shear stresses reach their maximum values. In sum- mary, a pavement element constantly experiences a combi- nation of varying magnitudes of static and dynamic vertical (compressive) and shear stresses, depending on the depth in the pavement layer and the radial offset from the wheel load. A known limitation of repeated-load triaxial tests is that the principal stress rotation and the constantly rotating fields of stresses under moving wheel loads are not possible to sim- ulate in a continuous fashion. However, principal stress rota- tion may cause increased rates of shear and volumetric strains during cyclic loading relative to equivalent stress paths with- out stress rotation. In the case of aggregate bases, the cyclic component of load imposes a change (increment) of stress state, which is not co-axial with the stress state under the static (overburden) load. This is illustrated in Figure 37. The major principal stress caused by overburden is always aligned in the vertical direction, regardless of the location of a moving wheel. However, the incremental stresses imposed by a wheel load are not co-axial with this system, and as a result, the total principal stresses rotate as the wheel load passes. Figure 38 illustrates the concept of stress path loading related to stress path slope (m) and stress path length (L) on a q-p diagram (Kim 2005). Static overburden stresses corre- spond to qmin and pmin, whereas dynamic traffic load reaches to qmax and pmax following a constant stress path slope (m). Analy- ses of test data often require defining geomaterial behavior in terms of these principal stresses considering a mean normal stress component (p) influencing volume change and the devi- ator stress component (q) affecting shear behavior for shape change and distortion (Kim and Tutumluer 2005). In general, the stress path slope (m = ∆q/∆p) for the standard constant con- fining pressure (CCP) tests (characterized by the application of an all-around constant confining pressure while the vertical deviator stress is pulsed) takes a constant value of 3.0. For variable confining pressure (VCP) tests characterized by puls- ing of the confining pressure in phase with the axial deviator stress, the stress slope varies generally from -1.5 to 3. VCP tests offer the capability to apply a wide combination of stress FIGURE 37 Stress states and rotation of principal stresses experienced by the aggregate layer beneath a rolling wheel load. z x : geo-element A Wheel Stresses Wheel Abscissa x principal stress due to overburden principal stress due to wheel load + overburden v h v A h CL CL Mean Normal Stress, p, 1 m L Stress Path Slope = m Stress Path Length = L p min pmax q min q max Deviator Stress, q 3 321p 31q FIGURE 38 Concept of stress path loading showing slope and length (Kim 2005).

58 paths by pulsing both cell pressure, s3, and vertical deviator stress, sd. Various stress paths cause different loading effects on pavement elements, which are not yet fully studied and understood to explain permanent deformation accumulation. Key Lessons • The directions of principal stresses imposed on an unbound aggregate pavement layer undergo con- stant rotation under a moving wheel load. • Such rotation of the principal stress directions as well as the associated loading patterns are best simulated during laboratory testing of aggregates to obtain real- istic estimates of unbound aggregate layer perfor- mance under loading. COMPACTION-INDUCED RESIDUAL STRESSES In the initial stages of new pavement construction, heavy loads are applied to granular layers causing large deforma- tions by compaction equipment. These layers are subjected to larger stresses during construction than they may ever experi- ence during the service life of the pavement structure. The largest vertical and lateral stresses are caused in the upper- most lift as compaction progresses. After the compaction is completed, field measurements indicate compressive residual lateral stresses are locked in the granular bases (Barksdale and Alba 1993). These residual stresses developed as a result of compaction of unbound aggregates should be considered in determining the initial stress state of granular bases. Proper compaction of granular pavement layers is required to ensure adequate strength and stability of the layer. The par- ticles, when subjected to compaction, rearrange themselves by translating and rotating to become locked in a final position. After the externally applied compaction stress is removed, this final stage is not a stress-free state, but rather a residual stress state. The residual stress state includes the effects of both confinement and aggregate interlock. Depending on the pore size distribution in the aggregate matrix, as well as the compaction moisture content, suction-induced negative pore pressures may exist in the newly constructed aggregate layer. The initial stress states used in the analysis of pavements usually is determined only by geostatic stresses attributable to body weight and are ignored in most linear elastic pave- ment analyses. A comprehensive granular base model needs to include both overburden stresses and the horizontal residual stresses. Several researchers have experimentally analyzed the residual stresses produced in granular bases (Stewart et al. 1985; Uzan 1985; Selig 1987; Zeilmaker and Henny 1989; Barksdale and Alba 1993). According to the research performed by Uzan (1985), Stewart et al. (1985), and Zeilmaker and Henny (1989), these horizontal residual stresses were measured to be as high as 2 to 5 psi in cohesionless granular materials. Barksdale and Alba (1993) also reported 3 psi horizontal residual stresses in the upper 6-in. (152-mm) portion of a 12-in. (305-mm) thick granular base obtained from field measurements. Based on experiments, Broms (1971), Ingold (1979) and Uzan (1985) employed a limit equilibrium approach to predict compaction-induced lateral stresses. The vertical stress under the compaction equipment was determined assuming a line loading (Holl 1941) and a semi-infinite homogeneous elas- tic halfspace (Boussinesq 1885). The lateral stresses devel- oped were limited to the maximum compaction loading and unloading conditions applied to a pavement in accordance with the classic earth pressure theory for frictional materials: 1. Under the loading of compaction equipment, horizon- tal stresses start to increase according to the active state when the limit equilibrium is reached and hori- zontal compression develops in the granular layer: σ = σKh a v where sv and sh are the vertical and horizontal stresses, and Ka is the coefficient of active lateral earth pres- sure, which usually is expressed in terms of the friction angle  as: ( )= − φtan 45 22Ka 2. After the compaction is completed, during unloading, the vertical stresses decrease. When the limit equi- librium is reached, horizontal stresses also decrease according to the passive state, and vertical stresses finally reduce down to the overburden stresses: σ = σKh p v where Kp is the coefficient of passive lateral earth pres- sure, which is usually expressed in terms of the friction angle  as: ( )= + φtan 45 2 .2K p Using the method of analysis, Uzan (1985) observed that a maximum vertical stress of 61 psi reached during compac- tion yielded a horizontal residual stress of about 6 psi. This residual stress may be higher depending on the friction angle () and load intensity (Tutumluer and Thompson 1998). The importance of considering compaction-induced residual stresses in the analysis and design of unbound aggregate lay- ers is discussed later in this chapter. CONCEPT OF CROSS-ANISOTROPy The behavior of a granular medium at any point depends on the arrangement of particles, which usually is determined by aggre- gate characteristics, construction methods, and loading condi- tions. In the case of unbound aggregate pavement layers, an

59 apparent anisotropy is induced during construction by aggregate placement and then loading from the compaction equipment. Thus, the granular layer becomes stiffer in the vertical direction than in the horizontal direction, even before the wheel load on the pavement imposes further anisotropic loading. Most geomaterials, such as naturally deposited soils, exhibit a rotational symmetry about their vertical axes called the “axis of symmetry.” The material properties are then the same in all directions on the plane perpendicular to the axis of symmetry. These materials are known as “cross-anisotropic” materials. An isotropic material has the same material prop- erties in all directions. A cross-anisotropic material has dif- ferent properties in the horizontal and vertical directions. The stress-strain conditions in such a material can be defined using the following five material properties (as illustrated in Figure 39): (1) stiffness in the vertical direction MRz, (2) stiff- ness in the radial (horizontal) direction MRr, (3) shear modu- lus in the vertical direction GRz, (4) Poisson’s ratio for strain in the horizontal direction as a result of a vertical direct stress nz, and (5) Poisson’s ratio for strain in any horizontal direc- tion as a result of a horizontal direct stress nr. Equation 3 shows the constitutive relationship for an elas- tic cross-anisotropic material in terms of the five independent material parameters. ε ε ε γ           = − ν − ν − ν − ν − ν − ν       σ σ σ τ           1 0 1 0 1 0 0 0 0 1 (3) E E E E E E E E E G v h h vh v vh h vh h hv v h hh h hv v hh h h vh v h h vh where Eh is the modulus of elasticity in the horizontal direction; Ev is the modulus of elasticity in the vertical direction; Gvh is the shear modulus; nvh and nhv are the out-of-plane Poisson’s ratio; and nhh is the in-plane Poisson’s ratio. The remaining parameters are not independent, as was proven by Love (1944), and is shown in Equations 4 and 5. ν = ν (4)E E hv v vh h ( )= + ν2 1 (5)G E hh h hh Key Lessons • Compaction of unbound aggregate layers results in preferential orientation of individual aggregate par- ticles, which ultimately leads to “cross-anisotropic” behavior. • Such compaction and stress-induced anisotropy is best considered during the design and analysis of pavement systems with UAB and subbase layers. METHODS TO CHARACTERIZE UNBOUND AGGREGATE LAyER BEHAVIOR The recent NCHRP Project 4-23, NCHRP Report 453: Per- formance-Related Tests of Aggregates for Use in Unbound Pavement Layers, summarized the most important tests that relate to the performance of aggregates in unbound pavement layers (Saeed et al. 2001). Among the tests highlighted, the shear strength tests (triaxial tests conducted on wet and dry samples and CBR test) and stiffness test (resilient modulus conducted on wet and dry samples) are the most relevant for characterizing the strength, modulus, and permanent defor- mation behavior of unbound aggregate pavement layers. California Bearing Ratio The CBR test (AASHTO Test Method T-193; ASTM Test Method ASTM D 1883) is an empirical test method. In this test, the aggregate is compacted into a 6-in. diameter mold to form a specimen 4.6 in. high. The maximum particle size permitted is ¾ in. Specimen conditioning usually consists of a 96-hour soaking period to simulate wet pavement conditions. Soaking is particularly important if a significant quantity of fines (passing No. 200 sieve or less than 0.075 mm) material is present. The specimen is then penetrated at a loading rate of 0.05 in./minute with a piston having an end area of 3 square inches. The specimen remains in the mold throughout the testing process. The CBR is calculated by dividing the piston pressure at 0.1 or 0.2 in. penetration by standard reference values of 1,000 psi for 0.1 in. penetration and 1,500 psi for 0.2 in., multiplied by 100 to give the CBR value expressed as a percent. These standard values represent the pressures FIGURE 39 Stratified anisotropic material under axial symmetry.

60 observed for a high-quality, well-graded, crushed stone ref- erence material. Accordingly, high-quality, dense-graded crushed stone commonly has CBR values in excess of 80, whereas well-graded gravel (AASHTO classification A-1-a; Unified Classification GW) may have CBR values ranging from 30 to 80. Note that testing of angular crushed stones in the laboratory often results in CBR values significantly higher than 100. Note that many base course aggregate specifications require CBR values in excess of 80 and subbase specifications require minimum CBR values in the range of 20% to 50%. CBR values often are presented together with moisture-density test results to indicate the change in CBR behavior above and below the OMC. Swell measurements if taken for the sample also pre- cede the CBR testing. Figure 40 shows typical CBR and swell test results obtained from an unsoaked molded sample and a sample of the same material that was allowed to soak for 96 hours. Note that unsoaked specimens are likely to give high CBR values, often higher than 100, on the dry side of OMC. CBR is not a fundamental material property and thus is unsuitable for direct use in mechanistic and M-E design pro- cedures. However, it is a relatively easy and inexpensive test to perform, it has a long history in pavement design, and it is reasonably well correlated with more fundamental properties such as resilient modulus. Consequently, it continues to be used in practice. Most CBR testing is laboratory-based; thus, the results will be highly dependent on the representativeness of the samples tested. It is also important that the testing conditions be clearly stated: for example, CBR values measured from as-compacted samples at optimum moisture and density conditions can be significantly greater than CBR values measured from sim- ilar samples after soaking. For field measurement, care is to be taken to make certain that the deflection dial is anchored well outside the loaded area. Field measurement is made at the field moisture content, whereas laboratory testing typically is performed for soaked conditions, so soil-specific correlations between field and laboratory CBR values are often required. Table 5 lists typical field CBR values for different Unified Soil Classification System classifications as obtained from Christopher et al. (2010) with reference to original work by the U.S. Army Corps of Engineers (1953). Static Triaxial Testing Strength is defined as the maximum level of stress that material can sustain before it fails or excessively deforms. Strength properties of a granular material can be best determined from static triaxial testing with monotonically increasing loading. A cylindrical test specimen is prepared at a target density and moisture content and is then encased in a membrane. The specimen is subjected to a constant all- around confining pressure (s3) and then loaded under an increasing axial stress until failure. Because the axial stress is in addition to the confining stress already on the speci- men, it is called the deviator stress: sd = s1 - s3. The total axial stress is called s1. Usually three triaxial tests are con- ducted over a range of confining pressure levels representa- tive of probable in-service conditions. Confining pressures used typically vary from 3 to 40 psi. Axial strain rates used in triaxial testing are typically 1% to 2% strain per min- ute. Triaxial test data are then interpreted to determine the cohesion (c) and angle of internal friction (f) of the material tested. The parameters c and f define the shear strength of the material, which is given by the Mohr-Coulomb equation: τ = + σ φtan (6)max c n where tmax = Shear strength c = Cohesion sn = Normal stress on specimen failure plane f = Angle of internal friction. C B R (% ) Moisture Content (%) CBR AS MOLDED OMC OMC + 3 Sw el l ( % )CBR AFTER SOAKED SWELL FIGURE 40 Presentation of CBR and swell mea- surements in relation to specimen moisture content. TABLE 5 TYPICAL FIELD CALIFORNIA BEARING RATIO (CBR) VALUES FOR DIFFERENT SOIL CLASSES Unified Soil Classification System (USCS) Soil Class Field CBR (%) GW 60–80 GP 35–60 GM 40–80 GC 20–40 SW 20–40 SP 15–25 SM 20–40 SC 10–20 ML 5–15 CL 5–15 OL 4–8 MH 4–8 CH 3–5 OH 3–5 Source: Christopher et al. (2010).

61 Considering vehicles usually move across a pavement very quickly, triaxial shear tests at the University of Illinois were performed at a rapid shearing rate, which is more repre- sentative of usual loading conditions than is the conventional slow triaxial shear test. Three different samples are tested at confining pressures of 5, 10, and 15 psi to determine the shear strength properties, friction angle, and cohesion of the aggregate materials. In rapid shear tests, a high loading rate of 1.5 in./s is applied, instantly causing 12.5% deformation in a 12-in. high specimen; the loading rates in such tests are higher than those in conventional triaxial shear tests. Fig- ure 41 shows the deformed shape of an aggregate specimen after completion of the test. Because of the high loading rate, the University of Illinois rapid shear strength test gives slightly higher peak stress results than do the conventional shear strength tests (see Figure 42). Although not conservative, the rapid shear tests are believed to better simulate the conditions of the actual pavement layer under the dynamic application of a moving wheel load. Repeated Load Triaxial Testing Repeated load triaxial testing has received major emphasis as a means for evaluating in the laboratory the modulus–deformation characteristics of granular materials and subgrade soils. Both resilient modulus and permanent deformation accumulation can be quantified based on the appropriate repeated-load test- ing data. Resilient modulus testing requires pneumatic or servohydraulic loading, a data acquisition system with feed- back control, a personal computer with an integrated soft- ware package, modern equipment, a good technician, and careful equipment calibration. The equipment must be capa- ble of producing load pulse duration of approximately 25 to 150 ms. The load pulse is generally repeated 15 to 30 times a minute. Specimen deformation over the entire length (or in some cases a portion of the specimen) typically is measured FIGURE 41 Deformed sample after completion of University of Illinois Rapid Shear Monotonic Triaxial Strength Test. FIGURE 42 Conventional slow and rapid shear strength test results on a crushed stone. 0 50 100 150 200 250 300 0 50 100 150 200 250 300 350 Normal Stress, psi S h ea r S tr es s, p si c= 19.2 61.7 c= 6.8 57.6 Slow, monotonic 1%/minute UI Rapid Shear: 12.5%/second max = c + n*tan 6.9 kPa = 1 psi

62 with “externally” or “internally” mounted linear variable dif- ferential transformers (LVDTs). The modulus and permanent deformation tests are per- formed on a cylindrical shaped solid specimen subjected to repeated axial compressive stresses. The specimen is subjected to a constant or variable (pulsed) all-around confining pressure to simulate the field stress condition. The cyclic application of the deviator stress (sd = s1 - s3) distinguishes the modulus and permanent deformation tests from the static triaxial tests. Air or fluid is commonly used to provide the all-around con- fining stress, and the vertical deviator stress is applied with a servopneumatic or servohydraulic actuator onto a specimen placed between top and bottom platens (see Figure 43). The conventional repeated-load triaxial tests use a simple and convenient arrangement to apply stresses in the vertical and horizontal directions. Little friction between the loading system and specimen is generated during the test. The stress state in a sample remains fairly uniform, especially in the middle one-third section. Most importantly, the simplicity and the lower cost are the reasons the conventional repeated- load triaxial test setup is widely used for aggregate character- ization. Yet, in this type of a conventional triaxial test, only an all-around CCP is applied while the vertical deviator stresses are pulsed. Special triaxial testing devices with the capabil- ity of pulsing confining stresses offer an advanced material characterization by simulating various dynamic stress states under experienced moving wheel loads. Modulus and perma- nent deformation tests that consider the application of such realistic stress states often are referred to as VCP tests. Figure 44 shows the principal stresses applied in a repeated load triaxial test apparatus. The typical stress states applied on the specimen are according to the CCP condi- tions with cell pressure not pulsed in the triaxial chamber. The VCP-type, repeated-load triaxial tests, on the other hand, offer much wider loading possibilities by better simu- lating actual field conditions because in the pavement struc- ture the confining stress acting on the material is cyclic in nature. The inherent differences between the CCP and VCP tests are such that in the VCP tests (1) the confining pres- sure is also cycled in phase with the axial deviator stress and (2) the axial specimen deformations generally are larger owing to the lack of a constant all-around confinement on the specimen. A Strategic Highway Research Program (SHRP) Testing Protocol (P46—Resilient Modulus of Unbound Granular Base/Subbase Materials and Subgrade Soils) was developed in the United States for conducting standard MR tests on unbound aggregate materials. SHRP P46 was used in test- ing the various granular material and subgrade soil samples collected in support of the SHRP (FHWA) LTPP program. A “round robin” type evaluation was conducted with the SHRP P46 Protocol. The results were very helpful in a priori M-E design activities. SHRP P46 was first approved as an AASHTO Interim Method of Test (AASHTO T 294-92 I, Resilient Modulus of Unbound Granular Base/Subbase Materials and Subgrade Soils—SHRP Protocol P46), then carried the designation T294-94 in the 1995 AASHTO speci- fications. Another standard for resilient modulus (AASHTO T 292), which was originally developed in 1991, was still active until 2003. A new test standard (AASHTO T 307) was introduced in 1999, leading to the existence of two resilient modulus test specifications (AASHTO T 292 and AASHTO T 307) until 2003. AASHTO T 292 was withdrawn in 2003, and AASHTO T 307 became the only standard for resilient modulus testing. AASHTO recommends the T307-99 (2003) repeated load triaxial test as a standard test for resilient char- acterization of pavement materials in the United States. In general, CCP-type triaxial test conditions are used for MR testing of granular materials in the United States accord-FIGURE 43 Repeated load triaxial testing apparatus. FIGURE 44 Stresses applied on a cylindrical specimen in repeated load triaxial testing.

63 ing to the SHRP P46 and AASHTO T307-99 test protocols. The test specimens are subjected to 15 stress states in which the pulsed dynamic stresses (sd) range from 21 to 276 kPa in the axial direction, and the confining pressures (s3) range from 21 to 136 kPa. All applied stress states are in general below the failure stress conditions and applied following a CCP test condition with a loading stress path slope of m = 3.0. The conditioning stage requires applying a confining pres- sure of 103.4 kPa and a minimum of 500 (up to 1,000) repeti- tions of a load equivalent to a deviator (maximum axial) stress of 103.4 kPa. Considering that the conditioning test data are often used for permanent deformation characterization, this stress state, which only corresponds to a conditioning stress ratio (s1/s3) of 2, may not be high enough to properly shake down granular materials before MR testing. The MR testing stage requires applying 100 repetitions of the corresponding cyclic stress using a haversine-shaped load pulse and record- ing the average recovered vertical deformations for each LVDT separately for the last five cycles. No doubt the findings from the NCHRP 1-28 project (Barksdale et al. 1997), SHRP LTPP studies (LTPP Materials Characterization: Resilient Modulus of Unbound Materials— LTPP Protocol P46 Laboratory Startup and Quality Control Procedures, FHWA-RD-96-176), and the recent NCHRP 1-28A study on “Harmonized Test Methods for Laboratory Determination of Resilient Modulus for Flexible Pavement Design” greatly helped in preparing and updat- ing the current SHRP TP P46 and the AASHTO T307-99 (2003) test protocols, which are adopted for routine use in the MEPDG. It is also apparent that resilient testing proce- dures for granular materials are still undergoing develop- ment and refinement. In Europe, the final draft European standard for MR test- ing by the European Committee for Standardization (CEN) is the EN 13286-7 (2004), “Unbound and Hydraulically Bound Mixtures—Test Methods—Part 7: Cyclic Load Triaxial Test for Unbound Mixtures. This European Standard specifies test procedures for determining the resilient and permanent behavior of unbound mixtures under conditions that simulate the physical conditions and stress states of these materials in pavement layers subjected to moving loads. These proce- dures allow determining mechanical properties that can be used for performance ranking of materials and for calculat- ing the structural responses of pavement structures. Testing procedures similar to those of EN 13286-7 adopted in the United Kingdom can be found in the British standard BS EN 13286-7 (2004). Note that the European standard specifies test methods to characterize both the resilient and perma- nent deformation behavior of unbound aggregates, whereas the AASHTO T 307 focuses only on evaluating the resilient behavior. Moreover, the European standard incorporates both VCP (method A) and CCP (method B) loading condi- tions, whereas AASHTO T 307 uses only CCP conditions. Need for Permanent Deformation Testing Rutting or accumulation of permanent deformation is the pri- mary damage/distress mechanism of UAB/subbase layers in pavements. Accordingly, rutting resistance is a major perfor- mance measure for designing pavements with granular base/ subbase layers. Granular base/subbase permanent deforma- tion may contribute significantly to the overall flexible pave- ment surface ruts. Low-strength granular materials generally are more susceptible to higher permanent deformation accu- mulation. However, a properly compacted UAB/subbase layer comprising crushed particles adequately prevents set- tlement and any lateral movement in the layer through high shearing resistance and contributes significantly to the dissi- pation of wheel load stresses. The NCHRP 4-23 study identi- fied shear strength of unbound aggregates as one of the most significant mechanistic properties influencing pavement per- formance (Saeed et al. 2001). Moreover, shear strength prop- erty, rather than “resilient modulus” (MR), always has been shown to better correlate with unbound aggregate permanent deformation behavior for predicting field rutting perfor- mance (Thompson 1998; Tao et al. 2010; Xiao et al. 2012). Although the influence of stress state on unbound aggregate resilient modulus is relatively well understood, its influence on the actual performance—rutting, cracking, roughness— of flexible pavements is less clearly known in practice. The design domains in which the influence of stress state is sig- nificant are also poorly defined. Note that it is not uncommon to have two different aggregate materials with very poor and excellent rutting characteristics possess similar high modulus properties from laboratory MR testing (Mishra and Tutumluer 2011). Accordingly, it is never possible to evaluate aggregate base course rutting performances from just the MR tests con- ducted on aggregates for modulus characterization and mech- anistic pavement analysis. This is because computed elastic responses, such as the vertical resilient strain (ev) within an aggregate base/subbase layer, can never be properly corre- lated with their permanent strain/deformation independent of the material’s shear strength. Furthermore, permanent defor- mation accumulation of a particular layer also depends signifi- cantly on the level of wheel load stress applied in relation to the aggregate material’s strength under confinement, which is often represented by the stress/strength ratio (the percentage of strength that is reached upon loading at that same layer confine- ment) and closely linked to the material’s “shakedown” limits (Werkmeister et al. 2004). Accordingly, repeated-load triaxial tests need to be con- ducted on unbound aggregate materials to study the accu- mulation of permanent deformation under loading. Such tests can be used for different purposes, such as ranking of materials, evaluation of maximum allowable stress levels, and predicting permanent deformation accumulation in pave- ment layers (CEN 2004). The European Standard (CEN 2004) makes use of three different “shakedown zones,” as defined by Werkmeister (2003), to rank unbound aggregate materials

64 based on the permanent deformation test results. These shake- down zones correspond to certain load-deformation behavior trends identified by significantly different permanent strain accumulation rates under repeated loading. For a detailed discussion on shakedown zones, the reader is directed to the work by Werkmeister (2003). Note that none of the test pro- cedures currently available in the United States (AASHTO T 307 or NCHRP 1-28A) covers permanent deformation testing and characterization of unbound aggregates. Key Lessons • The CBR test is a commonly used index test to esti- mate the shear strength of unbound aggregates. • Although triaxial tests for shear strength, resilient modulus, and permanent deformation behavior give more realistic estimates of unbound aggregate behav- ior under loading, conducting such tests requires significant investments in equipment and personnel training. • Test procedures for conducting resilient modulus tests on aggregates, AASHTO T 307 and NCHRP 1-28A, have been available for more than a decade. These specifications can adequately capture the stress-dependent nature of unbound aggregates and are ready to be implemented in practice. • Although permanent deformation behavior has been established as a more direct indicator of pavement performance compared with resilient modulus, no standard test procedure is available in the United States or Canada governing the testing of aggregates for permanent deformation. • New research efforts should be focused on developing harmonized protocols for quantifying the permanent deformation behavior of aggregates. Innovative Devices for Advanced Triaxial Characterization of Unbound Aggregates Traditional triaxial testing equipment, operating under CCP conditions, cannot simulate the rotation of principal stress directions experienced by a pavement element under moving wheel loads. Such equipment is only capable of applying one constant stress path representing the stress states immediately under the wheel loading. However, as discussed, because of the moving nature of the wheel load, the major principal stress often is not aligned in the vertical direction and rotates in the direction of the applied load, as shown in Figure 45a. Thus, the total principal stress on a pavement element rotates as the wheel passes. Advanced triaxial test devices operating under VCP con- ditions offer the capability to apply different combinations of stress paths by pulsing both the confining (cell) pressure and the vertical deviator stress. Such stress path loading tests better simulate actual field conditions because the confining pressures acting on a representative soil element in a pave- ment structure also are cyclic in nature (see Figure 45b). Typ- ically, at a distance away from the centerline of loading, the horizontal component of dynamic wheel load can become greater in magnitude than the vertical component. In that event, an extension type of loading is more critical on top of the base. Advanced triaxial testing devices can simulate such loading conditions because of their ability to apply confin- ing pressures (s3 for a cylindrical specimen) that are larger in magnitude that the vertical stresses (s1 for a cylindrical specimen). Several researchers have found that the resilient strains measured from CCP tests are smaller than those from VCP tests even under similar peak stress levels (Shackel 1974; Allen and Thompson 1974). Allen and Thompson (1974) also found that the Poisson’s ratio values obtained from CCP tests were significantly higher than those obtained from VCP tests. Brown and Hyde (1975) suggested that similar resil- ient moduli could be obtained from CCP and VCP tests by ensuring that the confining stress in the CCP test was equal to the mean value of stress used in the VCP test. However, Nataatmadja (1989) observed that the resilient moduli obtained from CCP tests were higher than those from VCP conditions, even when the stress states followed a pattern similar to that proposed by Brown and Hyde (1975). Thus, it is apparent that characterization of unbound granular materials under CCP conditions may only overestimate the resilient moduli, thus leading to inadequate structural design of pavements. Similarly, repeated load triaxial testing of unbound aggre- gates incorporating stress path rotation usually results in higher permanent deformation accumulations compared with tests conducted under constant stress path loading. Acceler- ated loading tests carried out at the University of Nottingham in the United Kingdom showed that moving wheel loading or actual trafficking is more damaging to a pavement system FIGURE 45 Stress conditions in a granular base to consider for advanced aggregate characterization: (a) Rotation of principal stress directions. (b) Stresses in extension loading (Seyhan and Tutumluer 1999). (b)(a)

65 than is running a repeated plate loading test on the same pave- ment system (Brown and Brodrick 1999). They also observed that bidirectional loading causes more severe permanent deformation development when shear stress reversals owing to oscillating wheel loads are considered. Similar findings were also reported from a full-scale pavement experiment undertaken in France to study the behavior and performances of unbound granular materials as pavement granular layers (Hornych et al. 2000). The permanent strains accumulated in the granular layers under moving wheel loading were about three times as large as those under cyclic plate loads. Tutumluer and Kim (2003) studied typical airport granu- lar base/subbase course materials at various densities through single and multiple stress path laboratory tests. They observed that multiple stress path tests always resulted in much higher permanent volumetric and shear strains than those of the single path tests. Thus, their findings indicate that actual traf- fic loading, simulated by the multiple path tests, can cause greater permanent deformations or rutting damage, especially in the loose base/subbase, than can dynamic plate loading or a constant confining pressure-type laboratory test. Figure 46 shows the stress states applied by Tutumluer and Kim (2003) to evaluate the effects of multiple stress path tests on per- manent deformation accumulation in unbound aggregate specimens. From advanced triaxial testing incorporating loading of the aggregate specimen using multiple stress paths, they observed that the specimen axial strains (e1) were consider- ably lower in magnitude than the radial (e3) strains owing to the proper specimen compaction effort in the vertical direc- tion and the VCP type multidirectional stress pulsing. They also observed that both the volumetric and deviatoric strains obtained from the multiple stress path tests were consistently higher than the ones from the single path tests. The maxi- mum or peak values of all multiple stress path permanent strains at the elevated load cycles were significantly higher than those from the single path test procedure. Their findings suggested that moving wheel load effects should be properly accounted for in laboratory testing to better predict unbound aggregate layer performance under typical highway and air- port pavement loads (Tutumluer and Kim 2003). It is therefore important to use advanced triaxial testing devices capable of incorporating variable confining pres- sure conditions to properly account for the effects of moving wheel loads on unbound aggregate behavior. However, such test devices are usually very expensive, so their use has been limited. Some of the most well-known advanced triaxial test- ing devices, developed through research, capable of better simulating the stress conditions in a pavement structure are discussed here. FIGURE 46 Concept map of multiple stress path tests compared to single path tests (Tutumluer and Kim 2003).

66 K-Mould Semmelink and De Beer (1995) introduced a sophisticated laboratory test system, called the K-Mould device, for the rapid determination of the elastic and shear properties of pavement materials as developed by CSIR in South Africa. In the K-Mould device, an axial load is applied to soil/ aggregate specimen contained in a segmented, thick-walled cylinder (see Figure 47). The segments are held in place with springs whose stiffnesses are chosen to simulate typical lat- eral stiffnesses of, for example, aggregates in granular bases. Lateral stress is mobilized by an elastic support system with a stiffness that can be varied between 15 and 60 MPa and designed to simulate in situ conditions. This provides a state somewhere between K0 (zero lateral strain) and unconfined. Thus, the lateral stress depends on the total lateral strain mobilized by the applied vertical stress against the horizon- tally mounted radial disc springs. It was shown that K-mold was relatively rapid and cost-effective for determining the engineering properties, such as the modulus, friction angle f, and the cohesion intercept, c. Both elastic and permanent deformation properties of unbound granular materials and soils were studied, including quantification of the stress sen- sitivity and anisotropic nature of these materials. According to an assessment by the European COST Action 337 (2002) research program study, the K-Mould possibly may offer an alternative test to the cyclic/repeated load triaxial test, but considerable developmental work is required before this would be suitable for routine applica- tion. However, it has the potential to assess modulus and permanent deformation behavior characteristics, and sample preparation may be simpler. University of Illinois FastCell An innovative laboratory cyclic/repeated load triaxial testing device, the University of Illinois FastCell (UI-FastCell), was introduced by Tutumluer and Seyhan (1999) to have provi- sions for applying static and dynamic stresses in both verti- cal and radial direction by the use of the two independently controlled stress channels. This advanced triaxial setup was a custom-designed system superior to the commercially avail- able rapid triaxial tester equipment, RattCell, manufactured by the Australian Industrial Process Controls (IPC), Ltd. company. With UI-FastCell, higher magnitudes of radial stresses can be pulsed by the use of a triaxial chamber filled with hydraulic oil. The UI-FastCell was designed mainly for the purpose of determining in the laboratory the anisotropic and dynamic properties of unbound aggregates through stress path testing in VCP conditions. Because it is not possible to reorient the granular samples in the triaxial cell, applying and switching of the various stress states on the same specimen allowed for adequately determining the inherent and load- induced anisotropy. The device is also suitable for simulating field stress conditions in the laboratory and for studying the effects of principal stress rotation as a result of moving wheel loads that involve a change in total shear stress direction. The UI-FastCell uses a fluid/air interface to minimize compressibility effects when conducting tests in which the horizontal stress on a specimen must be cycled. This is use- ful for investigating anisotropic effects and the response to loading in which a 90° rotation of planes of principal stress is important. The cell also provides a capability for on-specimen displacement measurements, which eliminate problems asso- ciated with compliance of the machine used to load the speci- men. When on-specimen vertical displacements are used as well, end effects are eliminated. Figure 48a shows an unbound aggregate specimen (6-in. diameter, 6-in. height) being prepared in a split mold for testing using the University of Illinois FastCell. Figure 48b shows the specimen setup under the loading ram of the UI- FastCell, with the confining chamber in a raised position. Figure 48c shows a picture of the confinement cell lowered down around the specimen for the testing position. An air actuator applies the axial pressure, and the confining pres- sures are cycled through a hydraulic fluid within the rubber membrane. The driving cylinders on the back of the confining cell (not shown in the picture) include an air-fluid interface that provides fast application and switching of the dynamic loading. Some of the advanced features of the UI-FastCell, as discussed by Tutumluer and Seyhan (1999), are as follows: • Measurement of on-specimen vertical and radial dis- placements and axial force and displacement external to cell; • A bladder-type horizontal confinement chamber with a built-in membrane that is inflated to apply variable con- fining pressures during vertical cyclic loading; • Ability to independently cycle either vertical or radial loading/confining pressures in phase or out of phase, in compression or extension type loading; • Ability to reverse principal loading direction on the same specimen with applied radial pulse stresses exceeding the vertical ones. The UI-FastCell and the IPC RattCell advanced triaxial devices are fundamental research tools when compared with FIGURE 47 Schematic of K-Mould testing equipment ( Semmelink and De Beer 1995).

67 the conventional repeated load/cyclic triaxial testing equip- ment. Using the UI-FastCell, the following important labo- ratory testing considerations can be addressed (Tutumluer and Seyhan 1999): (1) the aggregate specimen can be anisotropically consolidated (K0 condition in the field); (2) various stress paths experienced under a rolling wheel load can be adequately applied; (3) anisotropic aggregate resilient moduli can be conveniently obtained by pulsing vertical and radial stresses; and (4) different orientations of principal stresses can be achieved by independently applying vertical and radial stresses (that is, major principal stress direction is not limited to only 0 or 90 degrees with the horizontal). Springbox The Springbox equipment (Edwards et al. 2005) is a suitable tool for testing unbound granular and some weak hydraulically bound mixtures (see Figure 49). It consists of a steel box containing a cubical sample of unbound aggregate material, of edge dimen- sion 170 mm, to which a repeated load can be applied over the full upper surface. One pair of the box sides is fully restrained, and the other is restrained through elastic springs, giving a wall stiffness of 10–20 kN per mm. The equipment enables a realistic level of compaction to be applied to the test material by means of a vibrating hammer and also includes a facility to introduce water to the sample or drain water from its underside. Loading takes the form of repeated vertical load applications of controlled magnitude at a frequency of at least 1 Hz and no greater than 5 Hz. The load capacity is equivalent to a vertical stress of at least 150 kPa. Measurements of both vertical and horizontal (spring restrained) deflections can be made with two measurement trans- ducers for each measure. In the case of vertical deflection mea- surement, the equipment allows the transducers to make direct contact with the specimen through holes in the loading platen. The stiffness modulus of the material can be calculated from the averaged deflections measured over a series of loading patterns. FIGURE 48 University of Illinois—FastCell. LVDT Springbox Mould Adjustable Bolt Springbox Mould Spring Housing Plates Inner Liner Detachable Sides Inner Liner Base Low Friction Bearing Specimen (170 by 170 mm in size) Loading Platen Spring Inner Liner Locking Bolt FIGURE 49 Schematic of the Springbox testing equipment (Edwards et al. 2005).

68 The Springbox device provides a relatively rapid and economic test method for determining modulus behavior of aggregate materials in an accelerated fashion, ranking materials susceptibility to permanent deformation, and assess- ing materials durability and moisture sensitivity. The Spring- box test method is currently included in the first revision of the 2009 Design Guidance for Road Pavement Foundations (draft HD25) in the United Kingdom. Key Lessons • Several researchers have developed innovative devices to simulate the actual stress conditions induced in an unbound aggregate layer under mov- ing wheel loads. • Despite the ability of these devices to better predict pavement behavior under loading, their use in prac- tice is not feasible because of equipment cost and personnel training requirement concerns. Interpretation of Repeated Load Triaxial Test Data Data collected during repeated load testing of unbound aggre- gates can be analyzed to calculate the resilient modulus (MR) and permanent deformation (dp) values as indicators of aggre- gate layer performance under loading. Figure 50 presents a schematic of typical deformation behavior of unbound granu- lar materials under repeated loading. As can be seen from the figure, with increasing number of load applications, the mate- rial rapidly accumulates permanent deformation under the first few cycles. This can be attributed to the rearrangement of indi- vidual particles in the aggregate matrix. However, as the num- ber of load applications increases, the rate of accumulation of permanent deformation gradually decreases, and all the defor- mation corresponding to each loading cycles is resilient (recov- erable) in nature. The initial reorientation of particles often is said to correspond to the compaction and construction phases in a pavement layer. Thus, for an in-service pavement, all the deformation under vehicle loading ideally would be recover- able in nature. The resilient modulus of a material is usually calculated after all the particle reorientation has taken place. In the case of conventional triaxial tests (CCP conditions), the resilient modulus (MR) and the Poisson’s ratio (n) can be obtained from the measured recoverable strains using axi- symmetric stress-strain relations as follows: ε = σ (7)1 M d R ε = −ν σ (8)3 M d R In the case of advanced triaxial tests conducted under VCP conditions, both the vertical and horizontal stresses are pulsed. Thus, the resilient modulus (MR) and the Poisson’s ratio (n) need to be obtained from the measured recoverable strains using axisymmetric stress-strain relations as follows: ε = σ − ν σ2 (9)1 1 3M M d R d R ( )ε = −ν σ + − ν σ1 (10)3 1 3M M d R d R where s1d and s3d are the pulsed stresses in the vertical and horizontal directions, respectively; and e1 and e3 are the recov- erable strains in the axial and radial directions, respectively. In using the earlier equations to obtain resilient parameters as constants, an assumption is made that the material behaves linearly and elastically for any individual stress state. In addition, interpretations of anisotropic moduli with both s1d and s3d are that the pulsed stresses require the con- sideration of vertical modulus (MRV), horizontal modulus (MRh), in-plane Poisson’s ratio (nh), and out-of-plane Pois- son’s ratio (nV) to be obtained from the measured recoverable strains using axisymmetric stress-strain relations as follows: ε = σ − ν σ2 (11)1 1 3M M d R V V d R h ( )ε = −ν σ + − ν σ1 (12)3 1 3M MV d R V h d R h Current Resilient Modulus Models Resilient moduli of granular materials increase with increas- ing stress states (stress-hardening), especially with confining pressure and/or bulk stress and slightly with deviator stress (Lekarp et al. 2000a). Resilient behavior of unbound aggre- gate materials can be reasonably characterized by using stress- dependent models that express the modulus as nonlinear functions of stress states. Such a characterization model must include in the formulation the two triaxial stress conditions (that is, the confining pressure s3 and the deviator stress sd or the applied mean pressure p and the deviator stress q) to FIGURE 50 Behavior of unbound granular materials under repeated loading.

69 account for the effects of both confinement and shear loading. The model parameters traditionally are obtained from the mul- tiple regression analyses of the repeated load triaxial test data. Currently available models to predict the resilient modulus of granular materials are extensively discussed in Appendix D. Current Permanent Deformation Models Constitutive relationships often need to be developed to prop- erly describe permanent deformation accumulation in unbound granular materials with number of load applications. A sum- mary of the different models proposed by many researchers to predict permanent strain as a function of load and material property-related factors appears in Appendix E. HISTORICAL DEVELOPMENT IN UNBOUND AGGREGATE CHARACTERIZATION FOR PAVEMENT DESIGN More and more sophisticated geotechnical concepts have been introduced into AASHTO pavement design guides with the release of each successive version. Christopher et al. (2010) presented an extensive overview of the geotechnical inputs used in different AASHTO pavement design methods. The following sections present a summary of the discussion pre- sented by Christopher et al. (2010). 1961 Interim Pavement Design Guide The AASHO 1961 Interim Design Guide used the concept of structural number (SN) to account for the contribution of individual layers to pavement structural capacity. = + + (13)1 1 2 2 3 3SN a D a D a D where D1, D2, and D3 are the thicknesses (inches) of the sur- face, base, and subbase layers, respectively, and a1, a2, and a3 are the corresponding layer coefficients. For the materials used in the AASHO road test, the values for the layer coefficients were fixed at 0.44, 0.14, and 0.11, respectively. Because the parameters in the design equation were primarily based on the materials used in the AASHO road test, there was no scope for geotechnical material input in the design procedure. 1972 Interim Pavement Design Guide The 1972 Interim Design Guide (AASHO 1972) attempted to extend findings from the AASHO Road Test to foundation, material, and environmental conditions different from those at the test site. Several new features for the flexible and rigid pavement design were introduced, along with a rudimentary overlay design procedure. Guidelines were given for estimating the structural layer coefficients a1, a2, and a3 for materials other than those used in the AASHTO road test. These guidelines were developed based on a survey of state highway agencies regarding the val- ues for the layer coefficients that they were using in design for various materials. For example, the recommended range of a2 values (for untreated base layers) was from 0.05 to 0.18. Sim- ilarly, the structural layer coefficients for subbase (a3) could range from 0.05 to 0.14. Each agency was recommended to rely on past experience to establish appropriate layer coef- ficient values. The 1972 Guide also introduced an empirical soil support scale to account for the different environmental conditions experienced based on geographical locations. 1986 Pavement Design Guide The 1986 Pavement Design Guide introduced the concept of resilient modulus in a rational attempt to better charac- terize subgrade soil and unbound aggregate materials. The structural layer coefficients for base (a2) and subbase (a3) were estimated through correlations with resilient modulus. However, these relations for the structural layer coefficients were largely empirical and based primarily on engineering judgment with only limited amounts of data. Drainage coefficients were also introduced into the struc- tural number (SN) expression to accommodate different drainage conditions. Accordingly, the SN expression given in Equation 13 was modified to incorporate drainage coef- ficients (mi), as given here: = + + (13 )1 1 2 2 2 3 3 3SN a D a D m a D m a where m2 and m3 are the drainage coefficients for the base and subbase layers, respectively. The empirical values for mi, which are specified in terms of quality of drainage and the estimated percentage of time the layer will be near satura- tion, range from 0.4 to 1.4. Similarly, the rigid pavement design procedure in the 1986 Pavement Design Guide incorporated seasonal adjust- ments to the effective modulus of subgrade reaction. This effective modulus of subgrade reaction was a function of sea- sonally adjusted values for the subgrade and subbase resil- ient modulus. A drainage coefficient (Cd) was also introduced to account for drainage conditions under rigid pavements. 1993 Pavement Design Guide The 1993 design guide was similar to the 1986 Pavement Design Guide as far as the design of new flexible and rigid pavement structures were considered. The primary emphasis of this design guide was on rehabilitation design. NCHRP 1-37A Pavement Design Guide The MEPDG developed through NCHRP Project 1-37A (2004) incorporated new models of material behavior with the recognition that all pavement materials are exposed to and significantly affected by environmental, or climatic, factors.

70 Input Levels A refinement in the MEPDG was the use of a hierarchical design approach. Such an approach provided the designer with several levels of “design efficacy” that could be related to the class of highway under consideration or to the level of reliability of design desired. A chosen higher level of design output implied that the inputs were also of a higher level. In keeping with the hierarchical approach, materials character- ization was comprised of three levels, with Level 1 indicative of a design approach philosophy of the highest practically achievable reliability and Levels 2 and 3 of successively lower reliability. The Level 1 analysis was a “first-class” or advanced design procedure to provide for the highest practically achievable level of reliability. For the unbound aggregate resilient mod- ulus (MR) characterization, the laboratory-determined inputs (k1, k2, and k3 parameters) also were of the highest practically achievable level and generally required site-specific data col- lection or testing. Level 2 inputs provided an intermediate level of accu- racy and were closest to the typical procedures used with earlier editions of the AASHTO Guide. This level was rec- ommended when resources or testing equipment were not available for tests required for Level 1. Level 2 inputs typi- cally were user-selected, possibly from an agency database, could be derived from a limited testing program, or could be estimated through empirical correlations, such as resilient modulus and CBR relationships. Level 3 inputs provided the lowest level of accuracy. This level was recommended for design where there were minimal consequences of early failure (e.g., lower-volume roads). Inputs typically were user-selected values or typical averages for the region. One example would be the use of default unbound material resilient modulus values identified from aggregate material soil classification. The MEPDG used resilient modulus (MR), obtained from the NCHRP 1-28A and AASHTO T307, as the primary material property for all unbound pavement layers and sub- grade soils. Furthermore, the Level 1 inputs for the granular base, subbase, and subgrade required the characterization of the nonlinear, stress-dependent MR behavior for these layers. MEPDG Level 2 presented a correlation for determining modulus as a function of the layer coefficient as follows: MR (psi) = 30,000 (ai/0.14)3. A major disadvantage of this equation was that the layer coefficient has to be determined or estimated before estimating the resilient modulus. Other empirical correlations used for Level 2 design were: = +1155 555( ) psi (14)M RR ( )= 2555 (15)0.64M CBRR MEPDG Level 3 resilient modulus inputs for a new design were obtained from soil property correlations and nomographs DARWin-ME The recent release of the DARWin-ME, the software package implementing the Mechanistic Empirical Design Procedure, does not consider the stress dependence of unbound aggre- gate resilient modulus. The earlier implementation of the M-E pavement design procedure in the public domain MEPDG software explicitly included stress dependence of unbound resilient moduli as Level 1 inputs, but this capability has been removed from the new DARWin-ME software implementa- tion. Instead, DARWin-ME uses empirical correlations cor- responding to Level 2 of the MEPDG to compute the layer resilient modulus values. To ensure proper characterization of unbound aggregate layers and accurate prediction of pave- ment base/subbase layer performance under loading, stress dependence of unbound aggregate modulus needs to be incor- porated into future versions of DARWin-ME. Key Lessons • Resilient modulus of aggregates is used as a critical input in M-E pavement design methods. • Current AASHTO pavement design methods do not consider the stress-dependence of unbound aggre- gate resilient modulus. • It would be useful to incorporate the stress depen- dence of unbound aggregate materials into future releases of DARWin-ME, the current AASHTO mechanistic-empirical pavement design procedure. STATE OF THE PRACTICE IN UNBOUND AGGREGATE CHARACTERIZATION AND DESIGN Background The survey of state and Canadian provincial transportation agencies conducted under the scope of this synthesis study aimed to assess the state of the practice in unbound aggre- gate material characterization for design of pavement layers. Important findings from the survey are discussed in this section. When asked about the personnel responsible for the testing/ characterization of unbound aggregate materials for use in pavement base/subbase layers, 39 of 46 (~85%) respondents indicated that the design was done by the agency geotechnical/ materials laboratory. Only one agency (Wisconsin) delegated the characterization of such materials to a university labora- tory under a research subcontract. Several agencies specify a constant modulus value and AASHTO 1993 layer coefficient

71 for the aggregate materials irrespective of their source. For example, the practice in Oregon is to use a constant aggregate layer modulus of 20,000 psi and a layer coefficient (AASHTO 1993) of 0.10. The Alberta (Canada) transportation agency also reported a practice of specifying a constant layer coefficient for UAB/subbase layers irrespective of the aggregate sources. Figure 51 shows the distribution of personnel responsible for aggregate material characterization for design, in different agencies surveyed under the scope of this synthesis effort. Fourteen of 46 respondents reported the use of repeated- load triaxial tests for resilient modulus characterization. Twelve agencies do not conduct any test to characterize the strength, modulus, and permanent deformation character- istics of unbound aggregate materials. The use of strength index tests such as CBR or Hveem stabilometer appears to be a common practice among transportation agencies (41% of respondents, as shown in Figure 52). FWD testing appears to be the most common practice among transportation agencies for strength, deformation, and modulus characterization of in-service unbound aggregate pavement layers; more than 65% of respondents reported its use (see Figure 53). Only one agency (Maryland) indicated the use of GeoGauge, whereas four states (Maryland, Louisiana, Oklahoma, and Indiana) reported the use of light weight deflec- tometers (LWD). Seventeen (17) agencies reported not mea- suring the strength, modulus, or deformation characteristics of in-service aggregate layers. Several agencies use density as the only indicator of constructed aggregate layer quality. Twenty-seven of 46 respondents conduct laboratory/field tests to characterize aggregate materials for use in granular base and subbase layers on a project-need basis (see Fig- ure 54). Six agencies do not conduct any laboratory/field tests on aggregates. When asked about the method used to design pavements with UAB/subbase layers, 28 agencies (~61%, see Figure 55) reported using the AASHTO 1993 procedure. Four agen- cies use the AASHTO 1972 design guide, whereas 14 have adopted the MEPDG into their agency specifications. 85% 2% 17% 26% (39) (1) (8) (12) 0 10 20 30 40 0% 20% 40% 60% 80% 100% Geotechnical/Materials laboratory University laboratory (under research subcontract) Field Engineer Other (pavement engineer, consultant, fixed value, etc.) Number of Responses Percentage of Respondents 46 survey respondents FIGURE 51 Personnel responsible for characterization and design of unbound aggregate pavement layers in the surveyed transportation agencies. 41% (19) 13% (6) 30% (14) 0 48% (22) 0 5 10 15 20 25 0 20 40 60 80 100 Strength index tests (e.g. CBR, Hveem stabilometer, etc.) Triaxial shear strength tests Repeated load triaxial tests for resilient modulus (such as AASHTO T 307,… Repeated load triaxial tests for permanent deformation behavior Other (FWD, standard layer coeff., etc) Number of Responses Percentage of Survey Respondents FIGURE 52 Different test methods conducted by transportation agencies for strength, deformation, and modulus characterization of unbound aggregate materials used in base and subbase course applications.

72 FIGURE 53 Field tests conducted by transportation agencies for strength, deformation, and modulus characterization of in-service unbound aggregate pavement layers. 0% 26% 65% 9% 2% 37% (0) (12) (30) (4) (1) (17) 0 10 20 30 40 0% 20% 40% 60% 80% 100% Plate Load tests Dynamic Cone Penetrometer (DCP) tests Falling Weight Deflectometer (FWD) tests Light Weight Deflectometer (LWD) tests Soil Stiffness Gauge tests (GeoGauge, etc.) Other Number of Responses Percentage of Survey Respondents 46 survey respondents FIGURE 54 Frequency of laboratory/field tests conducted to characterize aggregate materials for use in granular base and subbase layers. 11% (5) 59% (27) 0 (0%) 7% (3) 24% (11) 0 5 10 15 20 25 30 0 20 40 60 80 100 Once on limited aggregate types/materials commonly used by the agency On a project-need basis Once on all agency-approved aggregate sources At regular intervals on agency-approved aggregate sources to establish a database Other Number of Responses Percentage of Survey Respondents 46 survey respondents FIGURE 55 Different methods used by agencies to design pavements with UAB and subbase layers. 9% (4) 2% (1) 61% (28) 30% (14) 11% (5) 11% (5) 22% (10) 0 10 20 30 0 50 100 1972 AASHTO Design Guide 1986 AASHTO Design Guide 1993 AASHTO Design Guide Mechanistic-Empirical Pavement… Agency-Specific Mechanistic Procedure: Agency-Specific Empirical Procedure: Other Number of Responses Percentage of Survey Respondents 46 survey respondents

73 Resilient modulus appears to be the most commonly used aggregate property (used by 21 of 46 respondents) serving as an input for the design of pavement structures (see Fig- ure 56). Only two agencies (Utah and Saskatchewan province in Canada) reported using aggregate shear strength as an input for pavement design. Twenty-two agencies reported other practices, such as the use of AASHTO-specified layer coeffi- cients for designing pavements with unbound aggregate lay- ers without using any material specific property. Twenty-six of 46 respondents assign a single modulus value to the entire aggregate layer without considering the stress-dependency of aggregate materials (see Figure 57). Ten agencies do not use modulus in the pavement design process, whereas only one agency (Oklahoma) incorporates the anisotropy of aggre- gate layers into its pavement design procedure. Only 10 agencies conduct resilient modulus testing in the laboratory to determine the modulus of unbound aggregate materials for use in granular base and subbase layers. Twenty- two use empirical correlations to predict the resilient modulus from index properties such as CBR or aggregate gradation parameters. Although the use of in-place modulus measure- ment using FWD and LWD appears to be fairly common (used by 30% of the respondents, as shown in Figure 58), several agencies do not test unbound aggregate materials for resilient modulus and adopt generic values during their pave- ment design procedures. Conclusions from Survey of Transportation Agencies The survey of state and Canadian provincial transportation agencies indicated that there is a wide variety in agency prac- tices as far as unbound aggregate material characterization and layer design is concerned. A significant gap appears to exist between the state of the art and state of the practice concerning unbound aggregate material characterization and pavement layer design procedures. Although it is widely recognized that aggregate shear strength, resilient modulus, and permanent deformation characteristics affect the performance of base and subbase layers in pavement systems, several agencies do not use these properties in their pavement design procedures. 39% 20% 26% 4% 46% 48% (18) (9) (12) (2) (21) (22) 0 10 20 30 40 0% 20% 40% 60% 80% 100% Percent passing sieve sizes (gradation) and/or maximum aggregate particle size Particle shape and angularity (crushed or uncrushed) Compaction characteristics, i.e., optimum moisture content and maximum dry density Shear strength properties (e.g. friction angle, CBR, etc.) Resilient modulus Other Number of Responses Percentage of Respondents 46 survey respondents FIGURE 56 Aggregate properties/characteristics used by agencies as inputs for the design of pavements with UAB and subbase layers. 22% (10) 57% (26) 0% (0) 2% (1) 20% (9) 0 5 10 15 20 25 30 0 20 40 60 80 100 Modulus is not used in pavement design Single modulus is assigned to the entire layer Stress-dependency of aggregate layer modulus is considered Anisotropy (directional dependency-ICAR models) of aggregate layer modulus is… Other (default value, etc.) 46 Survey Respondents Number of Responses Percentage of Survey Respondents FIGURE 57 Different approaches used by agencies for assigning resilient modulus values to UAB and subbase layers.

74 The initial survey of transportation agencies conducted during this synthesis study reflected that 14 of the responding agencies conducted resilient modulus (MR) testing on unbound aggregates before using them in pavement base/subbase layers. A follow-up survey of these agencies was conducted to gather further information on the prevalent state of the practice as far as MR testing is concerned. Conversations with agency person- nel indicated that although these agencies have the equipment to conduct MR testing on unbound aggregates, performing this test is not a common practice. Although some of these agen- cies frequently conduct MR testing on subgrade soils, using a constant modulus value for the base/subbase layer appears to be the preferred alternative. Difficulties associated with speci- men preparation, obtaining reliable data, and personnel train- ing appeared to be the most common factors responsible for the agencies not conducting MR tests of unbound aggregates. The agencies that do occasionally conduct MR tests on aggre- gates appeared to be unsure regarding making use of the test data in their respective pavement design procedures. Detailed findings from the follow-up survey of these 14 agencies are documented in Appendix F of this report. A harmonized approach may need to be developed that will recommend to state and provincial transportation agen- cies appropriate laboratory tests and design methods for unbound aggregate pavement layers. Key Lessons • A significant gap appears to exist between the state of the art and the state of the practice concerning unbound aggregate material characterization and pavement layer design procedures. • Resilient modulus testing of unbound aggregates is not a common practice among U.S. state and Cana- dian provincial transportation agencies. • Several agencies possess the required equipment for conducting resilient modulus tests on aggregates. However, difficulties associated with specimen prep- aration, obtaining reliable “good quality” data, and personnel training have resulted in agencies often assigning constant modulus values to UAB/subbase layers during pavement design. • Agencies would benefit from conducting resilient modulus testing on locally available aggregates for better consideration of unbound aggregate proper- ties in pavement design. STATE-OF-THE-ART METHODS FOR UNBOUND AGGREGATE LAyER CHARACTERIZATION AND DESIGN Proper consideration of compaction-induced residual stresses in granular materials would no doubt more appropriately model the behavior of unbound aggregate layers in pave- ments. The stress path approach by Uzan (1985) and experi- ments performed by Selig (1987) and Zeilmaker and Henny (1989) are useful for estimating the magnitudes of residual stresses existing in the granular layers because of compaction or preloading of the pavement layers. Knowing these residual (locked-in) horizontal stresses is essential for determining the most accurate initial stress state to evaluate correctly the resilient modulus values used in any mechanistic pavement analysis. Stress Path Testing To better characterize unbound aggregate layer modulus and deformation behavior, it is important to properly simulate in the laboratory actual field loading conditions. The pavement in the field usually is loaded by moving wheel loads, which at any time impose varying magnitudes of normal and shear FIGURE 58 Methods to determine the resilient modulus of unbound aggregate materials for use in granular base and subbase layers. 22% 48% 30% 33% (10) (22) (14) (15) 0 10 20 30 40 0% 20% 40% 60% 80% 100% Resilient modulus testing in the laboratory Empirical correlations with index properties like CBR, gradation parameters, etc. In-place modulus measurement of constructed layers by deflection-based methods such as FWD, LWD, etc. Other Number of Responses Percentage of Respondents 46 survey respondents

75 stresses in the aggregate layer, as reflected by the rotation of the principal stresses. This type of loading cannot be ideally simulated in the laboratory by the CCP-type repeated-load triaxial tests, such as the AASHTO T307-99. It is possible to apply only one constant stress path (m = ∆q/∆p = 3; see Fig- ure 59) in the CCP tests. However, the VCP-type repeated-load triaxial tests offer the capability to apply a wide combination of stress paths by pulsing both cell pressure and deviator stress (see Figure 59). Such stress path loading tests better simulate actual field conditions because in the pavement structure the confining stresses acting on the material are cyclic in nature. Stress path tests can be performed by switching stress states from CCP to VCP, depending on the stress path slope that is subject to be studied. The VCP tests require pulsing of the stresses in both vertical and horizontal directions. For example, to test a specimen with a stress path slope of m = 1.5, horizontal dynamic stresses should be one-fourth of the ver- tical dynamic stresses in magnitude; that is, s1d = 4*s3d (see Figure 43). Furthermore, extension stress states correspond to conditions of an approaching or departing wheel load at a distance with horizontal dynamic stresses exceeding the vertical ones. Kim and Tutumluer (2005) showed that under a moving wheel load, an extension-compression-extension stress cycling occurs to involve shear stress reversals. Directional (Anisotropic) Modulus Testing Unbound aggregate pavement layers exhibit higher stiffness characteristics in the vertical wheel loading direction than in the horizontal direction. This directional dependency is a special type of anisotropy, known as cross-anisotropy, caused by the preferred orientation of the aggregate to which both the shape characteristics of the aggregate and the compaction and traffic loading contribute. A cross-anisotropic represen- tation has different material properties (i.e., elastic modulus and Poisson’s ratio) assigned in the horizontal and vertical directions. Thus, for a cylindrical triaxial sample, a realistic assignment of in-plane and out-of-plane moduli is achieved under axial symmetry with the axial modulus increasing rela- tive to the radial one. Tutumluer and Seyhan (1999) considered the extreme stress conditions that may exist in the base layer of a flexible pavement structure under a moving wheel load. Then, con- sidering these extreme compression and extension loading conditions, deviator stresses are pulsed either in the vertical or horizontal directions only (see Figure 43). If the tested specimen is made up of a material that is truly isotropic in behavior, the moduli determined from the two-extreme load- ing conditions should be similar in magnitude. Accordingly, the laboratory findings of Tutumluer and Seyhan (1999) from four aggregates tested using UI-FastCell indicated definite directional dependency (anisotropy) of aggregate moduli. The resilient moduli computed in the vertical and radial puls- ing directions using a consistent set of isotropic stress-strain equations varied pronouncedly with the applied stress states. The vertical moduli typically were higher than the horizontal moduli for most aggregates tested, except for sandy gravel having a significant amount of P200 fines. The research project 502 conducted at the ICAR focused on determining structural issues of unbound aggregate lay- ers for a proper representation in a mechanistic-based design of flexible pavements (Adu-Osei et al. 2001; Tutumluer et al. 2001). The research team developed models for the resil- ient and permanent deformation behavior from the results of advanced triaxial tests conducted at the Texas Transportation Institute (TTI) using the IPC RattCell and at the University of Illinois using the UI-FastCell. The ICAR research team also developed a resilient modulus testing protocol, which although significantly different from the AASHTO T307-99 protocol, is not more complicated. The studies mainly indi- cated that the UAB material is to be modeled as nonlinear and cross-anisotropic to account for stress sensitivity and q 3d 1d 3d p 1 - q 3 Pulsed Dynamic Stresses: p = ( 1d+2 3d)/3 q = 1d- 3d m = q / p = slope of stress path CCP: Constant Confining Pressure VCP: Variable Confining Pressure p0 CCP VCP 3d = 0 1d = 0 m -3 2 Compression Extension Vertical pulsing only Horizontal pulsing only FIGURE 59 Stress paths that can be studied in an advanced triaxial setup such as UI-FastCell.

76 the significant differences between vertical and horizontal moduli and Poisson’s ratios. With anisotropic modeling, a more realistic stress distribution could be achieved in UABs. A recent state-of-the-art paper summarized the most sig- nificant work accomplished in the past 15 years in the area of anisotropic and stress-dependent modulus behavior of UABs used in flexible pavements (Tutumluer 2009). Findings of past research studies on both the laboratory and field valida- tions of the anisotropic aggregate behavior were discussed in detail. The most important result of properly accounting for the anisotropic stiffnesses of compacted granular base/ subbase layers is that critical pavement design parameters, such as vertical deviator stress and strain on top of the base course and the subgrade, are predicted to be typically higher than those computed when traditional isotropic pavement models are used. Note that these critical pavement responses are directly related to the degree and rate of permanent defor- mation in the base course and subgrade layers, and this is the substantial proportion of the overall pavement rutting in low- to medium-volume roads with thin asphalt surfaces. There- fore, traditional isotropic design approaches run the risk of under-designing flexible pavements or over-estimating the number of design axle loads the pavement can withstand. Effect of Cross-Anisotropy on Pavement Analysis and Design Traditional mechanistic pavement design methods use linear elastic programs that consider only isotropic material proper- ties in granular base layers to predict deflections, stresses, and strains in the pavement structure. However, the assignment of a single modulus to the entire layer does not correctly model base stiffness owing to stress variations in both vertical and hori- zontal directions. This is one of the reasons the linear elastic programs predict significant tensile stresses at the bottom of the base layer in most cases. However, because unbound aggregate layers are not capable of withstanding tensile stresses, such pre- dicted stress states are not realistic representations of the actual stress states in an unbound aggregate pavement layer. Barksdale et al. (1989) observed from instrumented test sections that a linear cross-anisotropic model of an UAB was at least equal to, and perhaps better for, predicting general pavement response than the simplified contour model pro- posed by Brown and Pappin (1981), which requires elaborate testing. Tutumluer and Barksdale (1995) modeled the same test sections employing cross-anisotropic resilient properties in the base layer and using the nonlinear model proposed by Uzan (1985) to represent resilient modulus. Considerably lower horizontal tensile stresses were predicted in the gran- ular base when the horizontal resilient modulus was equal to 15% of the vertical resilient modulus. Using this aniso- tropic modeling approach, reasonably good agreement was achieved with measured values of the resilient behavior for as many as eight response variables at the same time. Karasahin et al. (1993) also reported results of a study in which the applicability of various resilient constitutive mod- els of granular material was investigated for use in unbound base layers. An anisotropic volumetric-deviatoric model by Elhannani (1991) was found to give the best results for modeling the resilient behavior for the following two load- ing conditions: (1) only the deviator stress was cycled, and (2) both deviatoric and confining pressures were cycled in a triaxial test. Early work in characterizing the anisotropic modulus prop- erties of unbound aggregate layers used in flexible pavements was carried out at the Georgia Institute of Technology and the University of Illinois (Tutumluer 1995; Tutumluer and Thompson 1997a). Anisotropic modeling of a typical flexible pavement resulted in the magnitudes of both the horizontal and shear stiffnesses throughout the base being only small fractions of the vertical stiffness (Tutumluer 1995; Tutumluer and Thompson 1997a). Unlike the results of the isotropic-type analysis, the horizontal stiffnesses were found to be much lower when compared with the vertical values. These stiff- nesses were not assumed in the base layer but predicted by the nonlinear stress-dependent models obtained directly from the triaxial specimen behavior. Both the important effects of load-induced directional stiffening and the dilative behav- ior of granular materials under applied wheel loading were successfully modeled using a cross-anisotropic approach (Tutumluer 1995; Tutumluer and Thompson 1997a). Tutumluer and Thompson (1997b) modeled conven- tional flexible pavements using the GT-PAVE FE program and observed that, unlike the findings of isotropic analyses, a certain set of aggregate types and properties used in the granular layer typically resulted in horizontal stiffnesses varying between 3% and 21% of the vertical and the shear stiffnesses between 18% and 35% of the vertical through- out the base. As shown in Figure 60, the horizontal stiffness ratios (MRh/MRV) were low under the wheel load, 0.08 to 0.12 from the contour lines near the centerline, and increased radially away from the centerline to reach a value of 1 at approximately 6 load radii, which corresponds to the isotro- pic case. These stiffnesses were not assumed in the base layer but predicted by the anisotropic, nonlinear stress-dependent models developed from triaxial test data. The effects of compaction-induced residual stresses locked in granular bases also were of significance, especially when calculating horizontal stiffnesses. Such stresses offset any low-magnitude tensile stresses and provided adequate confinement radially away from the wheel load (Tutumluer and Thompson 1997a, b; Garg et al. 1998). A procedure was also established for estimating cross-anisotropic properties from repeated load triaxial tests with only vertical deformation measurements (Tutumluer and Thompson 1997a; Tutumluer 1998). Using the UI-FastCell, a large stress excursion analy- sis was conducted to characterize unbound aggregate layer behavior under various stress path loadings. Seyhan et al.

77 (2005) presented a new methodology for determining cross-anisotropic aggregate base properties (i.e., directional dependency of moduli and Poisson’s ratios as inputs into mechanistic pavement analysis considering effects of actual traffic or moving wheel loading). The proposed materials characterization requires conducting constant stress path triaxial tests and incrementally varying loading stress path slopes at similar stress states that are representative of vari- ous moving wheel loading conditions in the laboratory. In accordance, cross-anisotropic aggregate properties were determined by varying slightly the stress path slopes dur- ing testing and then by employing an error minimization approach to interpret the test results. Crushed aggregate spec- imens were prepared and tested to obtain cross-anisotropic properties at five different stress path slopes representative of various moving wheel-load–induced compression and exten- sion pavement stress states. Vertical resilient moduli were commonly found to be larger than horizontal ones, and criti- cally low vertical resilient moduli were obtained for some extension states (Seyhan et al. 2005). Simplified Procedure for Determining Anisotropic Model Parameters Based on the data presented by Hicks (1970), Allen (1973), and Crockford et al. (1990), Tutumluer and Thompson (1998) established a procedure for estimating cross-anisotropic properties from repeated load triaxial tests in which only ver- tical deformations were measured (the standard procedure, i.e., AASHTO T307-99). To characterize the typical varia- tions of horizontal and shear stiffness ratios, they analyzed a conventional flexible pavement section with anisotropic stiffness models used in a 203-mm thick granular base. The models were obtained from the multiple regression analy- ses of 50 triaxial test results on different aggregates obtained from the works of Hicks (1970), Allen (1973), and Crockford et al. (1990). Three stress-dependent MR models were used to completely define the resilient granular material behavior in vertical, horizontal, and shear planes, as follows: M K p I p pR A a a K a KB C =    1 16τoct ( ) where MR is the resilient modulus; I1 = σ1 + σ2 + σ3 = θ = first stress invariant or bulk stress; τoct = 1/3 ( ) ( ) ( )σ − σ + σ − σ + σ − σ1 2 2 2 3 2 3 1 2 = octahedral shear stress; pa = atmospheric pressure (100 kPa or 14.7 psi); KA, KB, and KC are material constants obtained from regression analyses of repeated-load triaxial test data. The three cross-anisotropic moduli (MRV, MRh, and GR) were modeled using the same formu- lation, and the model parameters used were as follows: Coefficient I1 Exponent τoct Exponent Horizontal resilient modulus (MRh) K1 K2 K3 Vertical resilient modulus (MRv) K4 K5 K6 Resilient shear modulus (GR) K7 K8 K9 Therefore, the stiffness ratios, (MRh/MRv) and (GR/MRv), could be expressed in terms of the coefficients, (K1/K4) and (K7/K4), respectively. Tutumluer and Thompson (1998) observed that the constant terms in the stiffness ratio models (K1/K4 or K7/ K4) were good approximations for the horizontal and shear stiffness ratios (MRh/MRv and GR/MRv) predicted by the FE analy- ses under the wheel load. Figure 61 shows the variations of the constant terms in the shear (K7/K4) and horizontal (K1/K4) stiffness ratio models obtained from tests performed on a vari- ety of crushed and partially crushed aggregates and gravel. Although somewhat scattered, the data points plotted at vari- ous saturation levels clearly indicated an increasing trend of K7/K4 (thus GR/MRv) with K1/K4 (thus MRh/MRv). The dotted lines plotted around the data define the lower and upper bounds for FIGURE 60 Horizontal stiffness ratio (MRh/MRV ) distribution throughout the base in the presence of 20.7 kPa horizontal residual stresses (after Tutumluer and Thompson 1997b).

78 a typical variation of K7/K4 with K1/K4 from triaxial test results for which the horizontal and shear stiffnesses proportionally increase or decrease. Accordingly, a granular material with high shear and horizontal stiffnesses would have a reduced tendency to lateral spreading under wheel loads. Figure 61 (Tutumluer and Thompson 1998) also shows a linear relationship found to exist between the constant shear ratio K7/K4 and the constant horizontal ratio K1/K4 for a con- sistent set of nine test results reported by Allen (1973). The standard estimated error (SEE) in the equation (see Figure 61) was given as 0.00636 for K7/K4. To estimate horizontal and shear model parameters, Tutumluer and Thompson (1998) proposed an additional equation relating the shear model constant parameter K7 with the vertical model parameters K4, K5, and K6, as follows (1 psi = 6.89 kPa): ( ) = − + + + = = ( ) 90.92 0.27 305.34 158.22 R 0.94, SEE 178 psi (17) 7 4 5 6 2 K psi K K K Figure 62 shows for the 50 test results the deviator stress exponents (K3-K6 or K9-K6) plotted with the bulk stress expo- nents (K2-K5 or K8-K5) as obtained from the horizontal and shear stiffness ratio models. In both plots, the data points are generally centered on the equality line, indicating that they are equal in magnitude but opposite in sign. Overall, these plots indicate that when the deviator and bulk stresses take similar values, K1/K4 and K7/K4 primarily determine the stiffness ratios. According to the earlier outlined simplified procedure by Tutumluer and Thompson (1998), these steps can be fol- lowed to estimate the shear and horizontal model parameters when the experimentally determined vertical modulus mod- els (i.e., K4, K5, and K6 are established from conventional repeated load triaxial test results) are known: 1. Use Equation 17 to compute K7; 2. Compute the constant ratio K7/K4; FIGURE 61 Variation of constant ratios in horizontal and shear stiffness ratio models (after Tutumluer and Thompson 1998). C = crushed; PC = partially crushed. FIGURE 62 Variation of stress exponents in the horizontal and shear stiffness ratio models (after Tutumluer and Thompson 1998).

79 3. Use the upper and lower band as well as Allen’s linear fit indicated in Figure 62 to obtain the corresponding K1/K4 constant ratio; 4. From Figure 63, select values equal in magnitude but opposite in sign for the stress exponents K2-K5 and K3-K6 to be used in the horizontal stiffness ratio model (an approximate value of 2.5 has been used as sug- gested by Tutumluer and Thompson, 1998); and finally 5. From Figure 63, select values equal in magnitude but opposite in sign for the stress exponents K8-K5 and K9-K6 to be used in the shear stiffness ratio model (an approximate value of 0.2 has been used as suggested by Tutumluer and Thompson 1998). Note that because of the low to nonexistent horizontal compressive con- fining pressures under the wheel load, approximating these stress exponents does not have any significant effect in the overall anisotropic dilative behavior of granular bases. Recent ICAR Procedure for Determining Anisotropic Model Parameters Based on the ICAR test protocol established for determin- ing stress-dependent anisotropic MR properties of unbound aggregate materials (Adu-Osei et al. 2001; Tutumluer et al. 2001), Ashtiani and Little (2009) developed a methodology for designing aggregate mixtures for base courses. A com- prehensive aggregate database was developed to identify the contribution level of different aggregate materials and base course features to the directional dependency of material properties. Accordingly, to characterize the level of anisot- ropy in unbound aggregate systems, the fitting parameters in material models (k-values) were used as characterizers of the level of anisotropy, which can vary considerably depend- ing on aggregate base properties such as gradation, satura- tion level, and the geometry (that is, shape properties of the aggregate particles). Three aggregate sizes for each of the 10 aggregate sources were tested for angularity, form, and tex- ture using Aggregate Image Measurement System (AIMS), and the distributions were fitted to two parameter cumulative Weibull distributions. Three gradations (fine, intermediate, and coarse) of the aggregate materials were used to determine dry density and moisture states of aggregate systems used in the aggregate database to account for the effects of optimum, dry of optimum, and wet of moisture conditions on directional dependency of material properties. From anisotropic modulus testing, the k model parameters were determined to capture the stress sensitivity, nonlinearity, and anisotropic behavior of the aggregate systems studied in the laboratory. Among the particle geometry features in the aggregate database, the ver- tical to horizontal modular ratio (Eh/Ev) was found to be most sensitive to the degree of elongation of the aggregate particles or how cubical the aggregate particles were. In their study, Ashtiani and Little (2009) also developed a new mechanistic performance protocol based on plasticity theory to ensure the stability of the pavement foundations under traffic loads. Field Validations As part of the ICAR 502 research project, field validation data were collected from two previous full-scale pavement test studies: the TTI and Georgia Tech studies (Tutumluer et al. 2001). The validation of the nonlinear anisotropic behav- ior of UABs was accomplished by analyzing these full-scale pavement test sections using TTI-PAVE and GT-PAVE FE analysis programs, predicting UAB responses and comparing them with the measured ones. The TTI project dealt with two flexible pavement test sec- tions, one with a thin and the other with a thick asphalt sur- face layer, built at the TTI Research Annex. The base course in each pavement was a crushed Texas limestone meeting the Texas DOT Grade 1, Item 248, aggregate base specifica- tions. The test sections were instrumented with multidepth deflectometers (MDDs), and an FWD was positioned directly over the MDDs and at several different positions away from MDD and the pavement responses (deflections) collected. FWD data were used to backcalculate material properties of the two pavement sections. For validation of the anisotropic resilient behavior, the limestone was characterized in the lab- oratory according to the ICAR testing protocol. Based on the FWD surface deflections and MDD depth deflections, several computer runs were made using the TTI-PAVE FE program. The linear elastic analyses had much higher errors between the measured and the predicted when compared with those obtained from the nonlinear isotropic and cross-anisotropic analyses. The nonlinear cross-anisotropic material models used in the base layer predicted vertical deflections closest to field deflections (Tutumluer et al. 2001). The Georgia Tech full-scale pavement test study (Barksdale and Todres 1983) had provided the original field data for the anisotropic base modeling study conducted by Tutumluer (1995). The pavements studied consisted of three conventional sections and two inverted sections, which had an UAB sand- wiched between an upper asphalt concrete surfacing and a lower cement-stabilized subbase. A total of eight response parameters, stresses, and strains at different locations in the test sections and surface deflections were measured in each test using strain gages, pressure cells, and LVDTs. After char- acterizing the crushed granitic gneiss used in the test sections for cross-anisotropic properties through advanced laboratory tests, Tutumluer et al. (2001, 2003) further analyzed the Geor- gia Tech test sections using the GT-PAVE FE program at differ- ent locations in the test sections considering several methods of UAB characterization for comparison and field validation. These methods included (1) a linear elastic, isotropic analysis; (2) a linear elastic, cross-anisotropic analysis; (3) a nonlinear, stress-sensitive isotropic analysis; (4) characterization of the vertical resilient modulus as nonlinear stress sensitive accord- ing to a model similar to that of Uzan (1985) and then assum- ing that the horizontal modulus is some percentage of the vertical modulus (work done by Tutumluer 1995); (5) a non- linear, stress-sensitive, cross-anisotropic analysis using modu- lus models developed following the laboratory SID approach

80 (Adu-Osei et al. 2001); and (6) a nonlinear, stress-sensitive, cross-anisotropic analysis with model parameters obtained from a simplified procedure that uses AASHTO T307-99 resil- ient modulus test results and was adopted earlier by Tutumluer and Thompson (1998). The accuracy of the overall modeling of resilient behavior of both the conventional and inverted sec- tions was related to how well the measured response variables were predicted at the same time. Only when a nonlinear cross- anisotropic model was used in the UAB (either method 4 or method 6), were the resilient behaviors of five pavement test sections predicted reasonably accurately for as many as eight response variables (i.e., displacements, stresses, and strains) from the same analysis. The resilient moduli computed in the horizontal direction, typically in the range of 12% to 27% of the vertical, were shown to correctly predict the horizontal and vertical measured strains in the UAB (Tutumluer et al. 2003). More recent field validations of anisotropic UAB behav- ior have been reported by Masad et al. (2006), Steven et al. (2007), and Kwon et al. (2008). Masad et al. (2006) success- fully demonstrated the efficacy of using anisotropic aggregate properties to represent unbound layers by comparing AASHO road test pavement surface deflection measurements under wheel loads to FE predictions based on models that incor- porated isotropic and anisotropic properties for the unbound base and subbase layers. The surface deflections in the flexible pavements of the AASHO road test were selected for this com- parison because the AASHO road test is such a widely used database and because of the tight control of traffic, pavement cross sections, and material quality at the road test (Masad et al. 2006). The deflection predictions correlated best with the experimental measurements when the horizontal moduli were about 30% of the vertical moduli in the UAB layers. Steven et al. (2007) performed elastic nonlinear FE analy- ses of a flexible pavement section, which was instrumented and tested in the New Zealand CAPTIF full-scale pavement test facility subjected to varying FWD loads. An inductive coil soil strain system was installed in the test section to measure vertical compressive strains within the granular and subgrade layers, and pressure cells were used to measure the vertical compressive stresses. The measured values of stress and strain at the top of the subgrade were used to give an indication of the stiffness. In an effort to match the measured FWD deflections and the vertical strain profile in the pave- ment section with the FE predictions, a nonlinear anisotropic modulus model with n = MRh/MRv as low as 0.15 had to be assigned in the granular layer. Kwon et al. (2008) reported on the resilient response pre- dictions of instrumented full-scale pavement test sections, both geogrid base reinforced and control sections, studied under single and dual wheel loadings at the University of Illinois. A mechanistic FE model, which considers the nonlinear, stress- dependent pavement foundation as well as the isotropic and anisotropic layer stiffness behavior of the granular base/subbase materials, was used to predict the field measured responses needed for the FE model validation. The cross-anisotropic modulus model parameters for the resilient moduli in vertical and horizontal directions (MRv and MRh) and shear modulus (GR) were characterized from laboratory testing in accordance with the approach by Tutumluer and Thompson (1997a, 1997b). Figure 63 shows for the unreinforced B1 test section (76-mm asphalt concrete underlain by 305-mm UAB) comparisons of the measured pavement responses and the initial response pre- dictions as a result of the different magnitudes of dual wheel loading with 689 kPa tire pressure. The cross-anisotropic base characterization gave much better predictions for the vertical LVDT displacements on top of subgrade and the radial LVDT displacements at the bottom of base course (see Figure 64). In the design of future full-scale pavement test studies, the performance prediction parameters, such as deflection basin shape and magnitude, degraded stiffnesses, rutting in the base course and subgrade, and other manifestations of distress, should be monitored during accelerated testing for develop- ing transfer functions (or distress models) to adequately relate pavement response variables to pavement performance. Masad et al. (2006) nicely pointed out that the performance models originally developed using isotropic material properties would require refinement and calibration for use with anisotropic material properties. Such a refinement would lead to smaller 0 0.2 0.4 0.6 0.8 1 1.2 1.4 26.7 35.6 44.5 53.4 62.3 Load (kN) Su bg ra de V er tic al L VD T (m m ) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 26.7 35.6 44.5 53.4 62.3 Load (kN) B as e Ve rti ca l L V D T (m m ) 0 0.1 0.2 0.3 0.4 0.5 26.7 35.6 44.5 53.4 62.3 Load (kN) B as e R ad ia l L VD T (m m ) Measured Prediction (Anisotropic) Prediction (Isotropic) FIGURE 63 Comparisons of measured and initial pavement response predictions from B1 unreinforced section (tire pressure of 689 kPa) (after Kwon et al. 2008).

81 shift factors and calibration coefficients owing to the improved match between the actual anisotropic material behavior and the response mode. The periodic monitoring and testing of pavement test sections also should help incorporate anisotropy and material nonlinearity in backcalculation methods to better account for the behavior of flexible pavements with unbound granular layers and estimate remaining life and performance. Anisotropy as Aggregate Quality Indicator Tutumluer and Seyhan (2000) evaluated the anisotropic resil- ient properties of aggregate systems through advanced labo- ratory tests and reported that the aggregate matrix showed significant softening behavior as the percentage of P200 fines (materials smaller than 75 µm or passing the No. 200 sieve) exceeded 12%. Research by Kim et al. (2005) has shown that aggregate type, gradation, and particle shape, texture and angularity significantly affect the level of anisotropy: that is, the ratio of horizontal to vertical aggregate layer moduli n = MRh/MRv. The anisotropy levels of aggregate base (the horizontal and shear moduli model parameters) could be approximated from regression analyses based on the model parameters of the vertical resilient moduli (K4 to K6) and some fitting parameters developed for aggregate physical properties, such as grain size distribution, form, angularity, and surface texture. Typically, higher values of moduli and modulus ratios were obtained when aggregate particles were well-graded, less elongated, and more angular with rougher surface texture. Later, Kim et al. (2007) successfully used a similar anisotropy level assessment technique to estimate in situ resilient modulus properties of sandy subgrade soils from FWD test results based on gradation properties, granular- base-to-asphalt-concrete-pavement thickness ratios, and the applied surface loading. An extension of the approach by Kim et al. (2005) was adopted recently by Ashtiani et al. (2007), who evaluated the impact of increasing fines content on the performance of unbound (unstabilized) and lightly cement-stabilized aggre- gate systems. It was found that with the proper design of fines content, cement content, and moisture, the performance of the stabilized systems with high fines content could per- form in a manner equivalent to or even better than could the systems with standard fines content. Ashtiani et al. (2007) also reported that by enhancing the resilient properties (increase in stiffness and decrease in anisotropy), compres- sive strength and permanent deformation properties could be improved in lightly cemented aggregate systems. Recently, Ashtiani and Little (2009) developed a compre- hensive aggregate database to identify the contribution level of different aggregate material types and properties as well as base course features to the directional dependency of non- linear, stress-dependent MR properties. Figure 64 demonstrates the impact of particle texture and aggregate angularity on the level of anisotropy characterized by vertical-to-horizontal modular ratios (i.e., Ex/Ey). Aggregate systems containing par- ticles with rougher texture and more crushed surfaces (more angular) result in much higher Ex/Ey (= Eh/Ev) ratios to more efficiently distribute load with greater aggregate interlock and friction in the unbound aggregate layer and thus to become less prone to plastic deformation under traffic (Ashtiani and Little 2009). Note that an isotropic system would correspond to a modulus ratio (Ex/Ey) of 1.0. FIGURE 64 Impact of aggregate angularity and texture on anisotropy level assessed using the axial modulus ratio (Ex/Ey) (Ashtiani and Little 2009).

82 Improving aggregate properties, such as by using well- graded cubical shaped and crushed aggregates with rough surface texture and reducing the amount of fines decreases the level of anisotropy while keeping MRv constant. Table 6 illustrates the relationship between aggregate quality and the level of anisotropy affecting pavement response and perfor- mance from GT-PAVE FE analyses of a conventional flex- ible pavement. The results given in Table 6 agree quite well with the known effective practices of the transportation agencies that pay attention to aggregate properties for building long-lasting pavements with deep UABs/subbases (Beatty et al. 2002). Thus, properly accounting for stress sensitivity and modulus anisotropy of unbound aggregate structural layers will be essential to the optimized use of available aggregate resources, building pavements with deep aggregate base/subbase courses, and accurately pre- dicting their expected field performances. Note that the MRh and MRv values shown in Table 6 are similar in definition to the Ex and Ey values indicated in Figure 65; as for pavement design purposes, resilient modulus values often are used as the layer moduli. Key Lessons • Several research studies and field validation projects have highlighted the benefits of considering aggre- gate anisotropy in pavement design; traditional pave- ment design methods based on the assumption that the unbound aggregate layer is isotropic may under- design flexible pavements and overestimate the pave- ment design life. • For adequately considering the behavior of UAB/ subbase pavement layers under loading, the cross- anisotropic properties of such layers are best incor- porated into pavement design. A simplified approach (see Simplified Procedure for Determining Aniso- tropic Model Parameters) is available for agencies to incorporate such cross-anisotropy of unbound aggregate layers into pavement design without the need to conduct advanced triaxial tests. CONSIDERATION OF DRAINAGE IN UNBOUND AGGREGATE LAyER DESIGN As a factor well-known for significantly affecting strength and modulus characteristics of pavement base/subbase materials and subgrade soils, excessive field moisture content is among the primary causes of pavement deterioration and premature failure. Among the primary challenges associated with maintaining pavement structures in adequate serviceable conditions, one aspect is to adequately cope with increasing moisture contents and minimize the effects of seasonal spring thaw by removing all transient water from the pavement structure as rapidly as possible before excessive accumulation (Birgisson and Roberson 2000). Such requirements imply the following two objectives of pavement foundation design: (1) proper selection/utilization of base/subbase materials retaining low average moisture contents and/or moisture sus- ceptibility, and (2) efficient subsurface drainage design. Many field forensic studies have demonstrated that good drainage and subsurface layer grading are essential for long-term satis- factory performance (Khazanovich et al. 1998; Chen and Lin 2010). In this section, previous findings on the mechanisms of moisture-related deterioration and corresponding mitigation measures are reviewed. NCHRP 4-23 project listed physical and mechanical properties of base/subbase materials in relation to major con- crete pavement distresses for performance evaluation (Saeed et al. 2001). In concrete pavements, subbase and subgrade are required to be stable under the application of traffic and environmental loadings; however, mechanical properties of unbound granular materials, such as resilient modulus (MR), shear strength, and rutting resistance, have long been found to be affected significantly by moisture content of the materials. Saarenketo and Scullion (1996) reported the dielectric value and electrical conductivity related to both the strength and deformation properties and frost-susceptibility of base course aggregates. The substantial effect of saturation hysteresis on base strength was also among important findings. Toros and Hiltunen (2008) characterized time-dependent changes in strength and stiffness of Florida base materials. These are changes that can be explained with unsaturated soil mechan- ics framework, such as changes in moisture or moisture dis- AC SG ( ) d 295 652 ( ) 0.15 284 6220.2 268 5650.3 253 5180.4 227 4780.5 n = M R h / M R v Anisotropy Decreases Agg. Quality Increases Pavement Responses Decrease Performance Increases Bottom AC Top Subgrade 25.5 24.1 22.0 20.7 18.6 (kPa) TABLE 6 ANISOTROPY AS AGGREGATE QUALITY INDICATOR AFFECTING PAVEMENT RESPONSE AND PERFORMANCE

83 tribution as a result of changes in internal pore pressure to influence the effective confining pressure constraining the material. Such changes in moisture content anticipated in the base/subbase during the service life of a pavement struc- ture should be properly accounted for in a pavement design process. Moisture-Related Deterioration Base/Subbase Erosion Erosion mechanism base/subbase layers below concrete pavements should be properly designed to provide a stable construction platform, uniform slab support, and erosion resistance. Different base/subbase materials below concrete pavements have been applied over the years, including dense- and open-graded unbound materials, cement- and asphalt- treated bases, lean concrete bases, and the combinations of such layers (sometimes with asphalt concrete interlayer for debonding). The degree of erosion resistance, called erod- ibility, depends on the type and nature of the base/subbase materials. In general, base/subbase erodibility increases with higher fines content and lower admixture stabilizer content (cement or asphalt). Figure 65 illustrates the subbase erosion process with which the pavement deteriorates (Jung et al. 2009). Any water infiltrated into pavement structure through joints/ cracks becomes trapped if it cannot flow out on a timely basis. Water then is injected as pressurized water into the pores between granular particles under dynamic moving wheel loads, causing the migration and accumulation of fine materials at the top surface of base/subbase layers. As this process evolves, distresses such as pumping and joint fault- ing could be initiated and accelerated in combination with other mutual effects responsible for pavement deterioration [for example, concrete slab curling and warping resulting from temperature and moisture variations and decreasing load transfer efficiency (LTE) along joints]. Both field and laboratory observations have indicated that the combination of traffic loading, trapped subsurface water, and erosion sus- ceptible base/subbase materials is the primary cause of base/ subbase erosion and subsequent joint faulting. According to Jung and Zollinger (2011), shear stress along the slab-subbase interface resulting from a moving load is the main contributor for the subbase erosion pro- cess. As shown in Figure 65, fines generated by such shear stress along the interface can be dislocated by pumping and then deposited into the void beneath concrete slab created by pumping, which eventually results in further loss of joint stiffness and joint faulting. As joint stiffness and the inter- face bonding keep decreasing, the magnitude of the interface shear stress increases and the erosion process becomes exac- erbated. Such hydraulic erosion was modeled by Jung and Zollinger (2011) for predicting erosion depth and joint fault- ing as a function of the number of wheel load applications. Evaluation of Base/Subbase Erodibility The need for a non-erodible subbase to maintain uniform sup- port under concrete pavements and thus ensure satisfactory service performance has long been recognized. The erosion resistance of materials beneath a concrete slab is an important performance-related property. As summarized in Table 7 by Jung et al. (2009), despite the existence of several empirical and subjective test methods, no well-accepted laboratory test methods are available for characterizing the erosion resistance of base/subbase materials using a mechanistic approach. Tables 8 and 9 summarize existing erosion models and design procedures that take erosion into account (Jung FIGURE 65 Subbase erosion and pavement deterioration processes (Jung et al. 2009).

Test Method Features Strengths Weaknesses Rotational shear device Stabilized test samples are eroded by application of hydraulic shear stress. The critical shear stress is recommended as an index of erosion resistance. Easy and precise to control shear stress No consideration of crushing or compressive failure. Overestimation of weight loss by coarse aggregates loss. Jetting device Pressurized water at an angle to the upper surface of unstabilized samples generating weight loss over time. Easy to test Shear stress is not uniform and inaccurate. Overestimation of weight loss by coarse aggregates loss. Brush test device Rotational brush abrasions generate fines. An erosion index, IE, is defined as the ratio of the weight loss to that of a reference material. Easy to test; consider durability of wet and dry cycle; relative erodibility of each material is defined using an erosion index, IE Long test time and overestimation of weight loss by coarse aggregate loss. Rolling wheel erosion test device Wheel movements over a friction pad on sample induce erosion. Measure average erosion depth after 5,000 wheel load applications. Simulate field conditions for flexible pavement; no coarse aggregate loss Voiding of the subbase under concrete slab cannot be considered. Sample saw-cut can damage sample surface. Source: Jung et al. (2009). TABLE 7 SUMMARY OF EXISTING EROSION TEST METHODS Erosion Model Features Strengths Weaknesses Rauhut Empirical model using COPES data Include many erosion related factors Rough categories for each material factor Markow Empirical model using AASHTO data: traffic, slab thickness, drainage Consider more detail drainage condition Subbase material properties are ignored Larralde Empirical model using AASHTO data: traffic, slab thickness Normalized pumping index to eliminate the effect of slab length and reinforcement No consideration about many erosion-related factors Van Wijk Fusion model of Rauhut and Larralde models with more field data Consider various erosion- related factors and four types of climates Rough categories for each material factor Portland Cement Association Mechanistic-empirical model using AASHO data Significant advancement in the mechanistic analysis Application of the model is limited to subbase types used in AASHO test Jeong and Zollinger Mechanistic model using theoretical hydraulic shear stress model Predict erosion depth based on feasible mechanistic equations Calibration required through lab tests and field performances Source: Jung et al. (2009). TABLE 8 SUMMARY OF EXISTING EROSION PREDICTION MODELS Design Guide Features Strengths Weaknesses Portland Cement Association Provide erosion factor as a function of the slab thickness, composite k value, dowel, and shoulder type Consider erosion analysis in design procedures as the most critical distress in rigid pavement performance Proposed composite design k-values for treated bases are overestimated and need discrimination for different stabilization levels 1993 AASHTO Composite modulus of subgrade reaction considers the loss of support (LS) caused by the foundation erosion Accounting structural degradation of support caused by erosion using LS factor k-value obtained from the chart is overestimated and LS is insensitive to various stabilized materials NCHRP 1-37A MEPDG Classified erodibility of subbase materials is utilized in jointed concrete pavement faulting prediction models as well as erosion width estimation of continuously reinforced concrete pavement Employed the erodibility class based on the type and level of stabilization along with compressive strength Erodibility class is determined based on dry brush test results and strength even though erosion occurs mostly under saturated condition Texas Department of Transportation Select one from two types of stabilized subbase and require minimum 7-day compressive strength Historical performances and erosion resistance are demonstrated as good Costly excessive design regardless of subgrade and environmental condition Source: Jung et al. (2009). TABLE 9 SUMMARY OF CURRENT SUBBASE DESIGN PRACTICES AND GUIDELINES

85 et al. 2009). The first concrete pavement design procedure addressing erosion may be the Portland Cement Associa­ tion procedure, which relates subbase erosion with pave­ ment deflection (at the slab corner) owing to axle loading. The equations for estimating percent erosion damage, together with the erosion criterion, were developed primar­ ily from the AASHO Road Test results. Note that only one highly erodible subbase type was used during the AASHO Road Test, so extending the application of these equations to other different subbase types for mechanistic analysis of pavement support conditions may be problematic. The AASHTO 1993 Guide relates foundation erosion to the potential loss of support (LS), which is numerically cate­ gorized into four different contact conditions (i.e., LS = 0, 1, 2, and 3). Each contact condition is associated with an effective reduction of the modulus of subgrade reaction in the thickness design procedure. An LS value of 0 represents the best contact condition when the concrete slab and the foundation are in full contact, whereas the value of 3 rep­ resents the worst case when the concrete slab is completely separated from the foundation. The major limitation of this method, as pointed out by Jung et al. (2010), is that it is too subjective to be sensitive to material factors causing ero­ sion, which may lead to inconsistency and limiting applica­ bility. The MEPDG recommends five different erodibility classes (from 1 to 5) for assessing the erosion potential of treated and untreated base materials on the basis of material type and stabilizer percent. The erodibility factor of base/ subbase materials is incorporated as an input parameter for modeling maximum and minimum transverse joint faulting. Note that none of those widely used analysis and design procedures explicitly include base/subbase erosion in a mechanistic approach. After reviewing previous erosion test methods and mod­ els, Jung et al. (2010) proposed a new test configuration that uses a rapid triaxial test and a Hamburg wheel­tracking device for evaluating the erodibility of various base/subbase materials under dry and wet conditions, respectively. Fig­ ure 66 shows the schematic diagrams of those test setups. During the rapid triaxial test, shear stresses of varying mag­ nitudes result from different combinations of deviator, and confining pressures are applied on the interface between concrete and subbase samples. The erodibility of subbase materials is measured from the percent weight loss caused by shear­induced interfacial abrasion. By integrating this new erosion test scheme with the theoretical hydraulic shear stress model (Jung and Zollinger 2011), they also developed a new laboratory­based M­E model for faulting in jointed concrete pavement and calibrated it using lab test results and LTPP field performance data. Key Lessons • Several pavement distresses may result from the presence of excessive moisture in unbound aggre- gate layers. • Base/subbase erosion leads to pavement distress in the original PCC slab and in (repair) patches installed on the original slab. If initial pavement dis- tresses indicate the presence of pumping, base/ subbase repairs are required to eliminate chances of the (repair) patch failing by the same mechanism as the original slab. (a) Rapid Triaxial Test for Erodibility of Dry Sample (b) Hamburg Wheel-Tracking Test for Erodibility of Wet Sample FIGURE 66 Schematic diagrams of the new erodibility tests by Jung and Zollinger (2011).

86 • Rapid removal of excessive moisture from unbound aggregate layers can be achieved through selection of aggregate materials with low water-retaining ten- dencies and design of suitable subsurface drainage systems. • Aggregate materials are best tested for erosion poten- tial or “erodibility” before being used in unbound base/ subbase layers. STATE OF THE PRACTICE REGARDING THE CONSIDERATION OF DRAINAGE AND CLIMATIC EFFECTS ON UNBOUND AGGREGATE BASE/SUBBASE LAyERS The survey of U.S. state and Canadian provincial transporta- tion agencies included questions regarding the state of the practice as far as the effects of climatic conditions and drain- age on UAB and subbase layer design is concerned. Only 11 of 46 respondents indicated that the structural con- tribution of open-graded aggregate drainage layers is consid- ered during the pavement design procedure (see Figure 67). Eleven agencies reported not considering the structural con- tribution of drainage layers, whereas 19 agencies do not use open-graded aggregate drainage layers in pavement systems. Twenty-seven of 46 respondents do not consider the effects of climate changes on the performance of unbound aggregate pavement layers (see Figure 68). Only nine agen- cies consider the effects of climatic conditions on unbound aggregate layer performance, whereas 10 respondents were not certain about prevalent agency practices. Of the nine agen- cies that do consider the effects of climate changes, seven do so by adjusting the aggregate layer resilient modulus value. Four agencies reported changing the layer structural coef- ficients, whereas only one agency (Virginia Department of Transportation) reported adjusting the aggregate layer shear strength, as shown in Figure 69. Note that Virginia DOT also reported adjustments to layer resilient modulus values and structural coefficients as common practice while considering the effects of climate change. Only 19 agencies currently specify different gradations for drainable and low-permeability applications of unbound aggregate layers, and 24 agencies reported no such practice 24% (11) 24% (11) 41% (19) 11% (5) 0 10 20 30 40 0% 20% 40% 60% 80% 100% Yes No Open Graded Aggregate Drainage Layers are Not Used Other 46 survey respondents Number of Responses Percentage of Survey Respondents FIGURE 67 Consideration of structural contribution of open-graded aggregate drainage layers in pavement thickness design. 20% 59% 22% (9) (27) (10) 0 10 20 30 40 0% 20% 40% 60% 80% 100% Yes No Other Number of Responses Percentage of Survey Respondents 46 survey respondents FIGURE 68 Agency response to whether the effects of climatic changes on unbound aggregate layer performance are considered.

87 (see Figure 70). As shown in Figure 71, 23 agencies do not consider drainage to be one of the primary functions of flexible pavement UAB/subbase layers. Of the ones that do consider drainage to be one of the primary functions, 17 facilitate the drainability of such layers by limiting the maximum allowable percent fines (material passing No. 200 sieve). Six agencies reported adjusting the material gradation to construct a more open-graded layer. As far as measuring the effectiveness of open-graded aggregate drainage layers is concerned, eight agencies use laboratory tests to measure the permeability of aggregate samples. Nine agencies use empirical correlations to esti- mate layer permeability from aggregate physical properties. Only two agencies (Maryland and Utah) reported using in situ permeability measurements, as shown in Figure 72. Twelve agencies use a geosynthetic filter layer to prevent OGDLs from clogging. Five agencies construct the filter layers using open-graded aggregates, whereas nine agencies do not construct any extra layer for filtration purposes (see Figure 73). Finally, as shown in Figure 74, construction sub- surface drainage systems, such as edge drains, are frequently constructed by 12 agencies. Twenty-one others construct such subsurface drainage systems only for specific projects when required by the design. Key Lessons • Only nine of 46 responding agencies consider the effects of climatic conditions on unbound aggregate pavement layer performance. • Agencies that do consider the effects of climatic con- ditions on unbound aggregate layer performance do so by adjusting the resilient modulus or layer struc- tural coefficient values. 44% 78% 11% 22% (4) (7) (1) (2) 0 2 4 6 8 0% 20% 40% 60% 80% 100% Layer structural coefficients Resilient modulus Shear strength Other Number of Responses Percentage of Respondents 9 survey respondents FIGURE 69 Agency response to unbound aggregate layer properties that are adjusted to account for climatic effects. 41% 52% 7% (19) (24) (3) 0 10 20 30 40 0% 20% 40% 60% 80% 100% Yes No Other Number of Responses Percentage of Respondents 46 survey respondents FIGURE 70 Agency response to whether different gradations are specified for unbound aggregate applications targeting drainable versus Low-permeability layers.

88 37% (17) 4% (2) 13% (6) 17% (8) 50% (23) 0 5 10 15 20 25 0% 10% 20% 30% 40% 50% 60% Limiting the maximum allowable percent fines (material passing sieve No. 200) Increasing the maximum aggregate size Adjusting the constructed layer gradation towards a more open-graded layer Other Drainage is NOT one of the primary functions of flexible pavement unbound aggregate… Number of Responses Percentage of Survey Respondents FIGURE 71 Different approaches adopted by agencies to facilitate the drainage of dense-graded base courses. 4% 17% 20% 50% 20% (2) (8) (9) (23) (9) 0 10 20 30 40 0% 20% 40% 60% 80% 100% In-situ permeability measurements Laboratory tests to measure the permeability of aggregate samples Empirical correlations to estimate the permeability from aggregate physical properties like gradation, dry density, specific surface, and void ratio (or porosity) Open-graded drainage layers are not used Other Number of Responses Percentage of Survey Respondents 46 survey respondents FIGURE 72 Different methods used by agencies to measure the effectiveness of open-graded aggregate drainage layers. 11% 26% 20% 48% (5) (12) (9) (22) 0 10 20 30 40 0% 20% 40% 60% 80% 100% Yes (open graded aggregates commonly used to construct the filter layer) Yes (geosynthetics commonly used as the mode of filtration) No (no extra layer constructed for filtration purposes) Open-graded drainage layers are not used Number of Responses Percentage of Respondents 46 survey respondents FIGURE 73 Agency response to whether filter layers are used to prevent the clogging of open-graded drainage layers.

89 • Nineteen of 46 responding agencies specify different gradations for permeable versus low-permeability unbound aggregate pavement layers. • The permeability of drainage layers usually is esti- mated from laboratory test results or empirical corre- lations. Only two agencies conduct in situ permeability tests on unbound aggregate drainage layers. EFFECTS OF AGGREGATE MATERIAL PROPERTIES ON LAyER PERMEABILITy Permeable Base Designs The use of permeable bases, open-graded drainage layers (OGDLs) and installation of edge drains are the primary options for restoring drainage efficiency. The hydraulic con- ductivity k (or the coefficient of permeability) is an essential parameter needed for analysis and design of the subsurface drainage systems, along with other engineering properties of materials, such as the grain size distribution (gradation), packing, degree of saturation, and frost susceptibility. Labo- ratory determination of the coefficient of permeability is very important for permeable base/subbase designs, fill materials, and other drainage layers, and test methods for this purpose are quite mature. Obtaining in situ measurements of perme- ability for base/subbase layers is desired from the standpoint of adequate engineering design. With the advent of portable gas permeameter test (GPT) devices, such as that introduced by White et al. (2010), this becomes possible and widely used in QC/QA. In cases in which field measurement and laboratory testing of permeability are not possible, empirical approaches can be used with caution to estimate permeability from readily available material properties, such as gradation and moisture and density conditions. Two main permeable base materials are widely used: stabilized and unbound. For example, in Minnesota, grada- tion specifications for an unstabilized base and a stabilized base are represented by the following percent passing values (Arika et al. 2009): Unstabilized Base • 1 inch: 100%; • ¾ inch: 65% to 100%; • ³⁄8 inch: 35% to 70%; • 4.75 mm (#4): 20% to 45%; • 2 mm (#10): 8% to 25%; • 0.425 mm (#40): 2% to 10%; • 0.075 mm (#200): 0% to 3% Stabilized Base • 1½ inch: 100%; • 1 inch: 95% to 100; • ½ inch: 25% to 60%; • 4.75 mm (#4): 0% to 10%; • 2.38 mm (#8): 0% to 5%. With the addition of a stabilizer (asphalt or cement), the gradation of a stabilized granular material becomes much coarser; however, unstabilized materials require consid- erable amounts of finer-size aggregates to achieve better packing (low voids content) and stability through aggregate interlock. Typical permeability values are 6,800 ft/day for stabilized bases and 1,000 to approximately 3,000 ft/day for unbound granular ones. Note that permeability values as low 26% 46% 30% (12) (21) (14) 0 10 20 30 40 0% 20% 40% 60% 80% 100% Yes, very common Yes, for specific projects when required by the design No, not common at all Number of Responses Percentage of Respondents 46 survey respondents FIGURE 74 Agency response to whether the construction of subsurface drainage systems, such as edge drains, is a common practice.

90 as 1,000 ft/day are considered acceptable as far as pavement layer drainage requirements are concerned. Table 10 shows the MnDOT guidelines for selecting permeable aggregate base. In addition to maintaining adequate permeability, these layers are required to remain stable during construction and future rehabilitation activities over the design service life. Hagen and Cochran (1995) described and evaluated in a study (MD-RD-95-28) the drainage characteristics and pave- ment performance of four drainage systems under jointed concrete pavement: MnDOT standard dense-graded base, two dense-graded base sections incorporating transverse drains placed under the transverse joints, and permeable asphalt- stabilized base reflecting the MnDOT drainable base concept at that time. Moisture sensors were placed in the pavement to assist in evaluating the relative performance of the tradi- tional and new drainage systems and their effects on pave- ment performance. Longitudinal edge drains were installed in all sections. Drainage flows, percent of rainfall drained, time to drain, base and subgrade moisture content, and pave- ment and joint durability were the variables studied. The fol- lowing important observations were reported: (1) although all systems appeared capable of removing drainable water from the pavement base, the permeable asphalt-stabilized base commonly drained the most water within 2 hours after rainfall ended while providing the driest pavement founda- tion with the least early pavement distress; (2) approximately 40% of the rain infiltrated into the concrete pavement; (3) the open-graded and geocomposite systems removed water most rapidly; (4) spring thaw flows were roughly equivalent to a major rain event; and (5) all rain inflow was reduced temporarily by sealing the longitudinal and transverse joints but resumed after approximately 2 weeks, despite the joint sealants appearing to be intact. In Illinois, Dhamrait and Schwartz (1979) evaluated four types of subbases (4-in. thick cement-aggregate mixture, 4-in. thick bituminous-aggregate mixture, 8-in. thick lime- stabilized soil mixture, and 4-in. thick granular materials) and three types of subsurface drainage systems, such as shoulder drainage. Pavement behaviors in terms of transverse crack- ing and deflections were analyzed and correlated with the type of subbase and type of subsurface drainage system. The lime-stabilized soil mixture as subbase was reported to offer the potential for reducing construction costs. The subsur- face drainage system with longitudinal underdrains placed at the edge of the stabilized subbase was the most efficient in removing free water from beneath the pavement structure; the system was adopted by the Illinois DOT as the standard treatment for interstate highways. Winkelman (2004) investi- gated OGDL performance in Illinois. The OGDL consisted of a uniform size aggregate that may be bound together as a lean concrete mixture or low asphalt cement content bituminous mixture. Pavement performance was monitored in terms of FWD measurements, International Roughness Index values, visual distress surveys, and condition rating survey values. Despite the OGDL being more costly than a stabilized base, its use under a continuously reinforced concrete pavement was not recommended according to the findings. This study suggested that a geotextile fabric or dense-graded aggregate filter be used under the OGDL to prevent the intrusion of subgrade fines. As compared with UABs, asphalt- or cement-treated bases become more expensive solutions and thus less desirable for some roadways, especially for low- to medium-volume ones. In these situations, it is worth exploring if the use of a properly graded unbound aggregate can maintain adequate drainability and structural stability during construction and the expected service lifetime after the roadway is open to traffic. In Louisiana, Tao and Abu-Farsakh (2008) studied typical permeable base materials for their drainage benefits, includ- ing asphalt- and cement-treated aggregates, open-graded aggregates, and dense-graded unbound aggregates. The per- meability of unbound aggregate was quantified by its satu- rated hydraulic conductivity, whereas its structural stability was characterized by the results of various laboratory tests for strength, stiffness, and permanent deformation of the material. A trade-off between structural stability and perme- ability of unbound aggregates was observed; the increase of Legend: AR = As Recommended (A) NA = Not Applicable (B) NR = Not Recommended R = Recommended R/AR = (C) Subgrade Soil Traffic Level Plastic / Non-Granular VH H M L M L Granular (D) VH Interstate Non-Interstate R R R R R R R/AR R/AR NA R NA AR NA NR NA NR Traffic Level: VH = (Very High) H = (High) M = (Medium) L = (Low) 35-yr Design Lane CESALs > 30 million 9 – 30 million 3 - million < 3 million TABLE 10 MINNESOTA DEPARTMENT OF TRANSPORTATION CONCRETE PAVEMENT PERMEABLE AGGREGATE BASE APPLICATION GUIDELINES

91 permeability was often at the cost of structural stability or vice versa. Therefore, the criteria for selecting such an opti- mum gradation were: (1) an adequate permeability to drain the infiltrated-water from the pavement as rapidly as pos- sible; and (2) a sufficient structural stability to support the traffic loading. Laboratory tests were conducted on a Mexi- can limestone (commonly used in Louisiana highways) with different gradations, including constant-head permeability, CBR, Dynamic Cone Penetrometer (DCP), tube suction test (TST), monotonic load triaxial tests, and repeated load tri- axial tests. The gradations under investigation included coarse and fine branches of Louisiana class II gradation, New Jersey gradation medium, and an optimum gradation (fine and coarse branches). As a result, Tao and Abu-Farsakh (2008) determined a proper/optimum gradation for perme- able base materials. For a detailed discussion of the proposed proper/optimum gradation, the reader is directed to the origi- nal publication (Tao and Abu-Farsakh, 2008). Figure 75 shows the gradation curves of different aggregate materials tested by Tao and Abu-Farsakh (2008). Key Lessons • Permeability values as low as 1,000 ft/day are con- sidered acceptable as far as pavement layer drain- age requirements are concerned. • Stable open-graded or gap-graded (a gradation with some intermediate-size particles missing) aggregates with low fines contents (P200) are best selected for use in unbound aggregate drainage layers. CONSIDERATION OF CLIMATIC CONDITIONS IN UNBOUND AGGREGATE BASE DESIGN The life cycles of most pavements are significantly shorter than the time span over which climatic effects will have a statistically significant effect on pavement performance (Dawson 2010). However, the effect of climate change on UAB/subbase performance can be manifested primar- ily through changes in moisture content, effect of freeze- thaw cycles, and depth of frost penetration. As discussed in chapter two, moisture often has been identified by research- ers and practitioners as one of the most important factors affecting unbound aggregate layer performance. Permanent deformation is more likely to occur in an unbound aggre- gate layer during wet spring months when the modulus and strength properties are greatly reduced, especially in the northern climates with wet freeze and thaw conditions. Unbound pavement layers are most likely to reach equilib- rium moisture contents, often on the wet side of compacted optimum moisture conditions, and this can drastically affect the long-term modulus and permanent deformation behavior. The seasonal variation in unbound pavement material moduli is widely recognized as contributing to decreased load-carrying capacity and pavement failure. Factors influ- encing layer moduli are stress state, moisture, suction, den- sity, and material characteristics. Climate factors such as precipitation and temperature contribute to seasonal varia- tion in layer moduli. These seasonal variations are mainly the result of variations in moisture/suction. Depending on the magnitude of the load or applied stress state in rela- tion to the strength, modulus and permanent deformation FIGURE 75 Particle size distributions of aggregate types tested by Tao and Abu-Farsakh (2008).

92 properties vary considerably with moisture/suction and tem- perature, which in turn depend on the weather conditions. Through investigation of several pavements in Wash- ington State, Newcomb et al. (1989) observed that seasonal variations in the moduli of the subgrade materials were much less significant than those observed in granular base materials. Uhlmeyer et al. (1996) reported that the effect of seasonal variations on base layer performance were greater than those for the subgrade. Moreover, they observed that the seasonal effect on unbound aggregate layer performance was reduced significantly when the stress sensitivity of unbound aggregate materials was considered, rather than treating the layers as linearly elastic. Kolisoja et al. (2002) conducted cyclic-loading triaxial tests on base course aggregates to simulate seasonal condi- tions of dryness, moisture, and the period after a freeze-thaw cycle. From the test results they reported that even though the permanent deformation behavior of aggregates were signifi- cantly affected by freeze-thaw cycles, no significant changes in the resilient modulus values were observed even during the spring thaw phase. Werkmeister et al. (2003) observed that even a 1% increase in moisture had a significant effect on the perma- nent deformation behavior of unbound aggregates. Although the increase in moisture content did not result in significant changes in the resilient modulus values (and thus the stress levels), it was reflected through drastic changes in the perma- nent deformation behavior. Carrera et al. (2009) listed the following factors as sig- nificantly affecting the moisture sensitivity of unbound aggregates: (1) compaction properties, (2) amount of deg- radation, (3) grain size composition, and (4) quality of P200 fines (plasticity and swelling index). Mishra (2012) studied the effects of material quality on aggregate behavior using aggregate specimens containing different amounts of non- plastic and plastic P200 fines (material finer than 0.075 mm) compacted to different moisture-density conditions. Con- ducting laboratory tests to characterize the shear strength, resilient modulus, and permanent deformation behavior, Mishra concluded that the effect of moisture content on aggregate behavior was particularly severe for specimens containing high amounts of plastic fines. Moreover, the quality of fines (plastic or nonplastic) was significant only for specimens with intermediate to high fines contents (8% or higher for crushed aggregates, 6% or higher for uncrushed gravel). A shallow depth to GWT decreases suction and critically affects the long-term modulus and permanent deformation behavior of aggregate base/subbase, especially in regions with a moisture surplus, as measured by the Thorntwaithe Moisture Index when equilibrium moisture contents are on the wet side of compacted optimum conditions (Zapata and Houston 2008). Figure 76a shows increased permanent deformations in all of the subgrade, subbase, and base lay- ers during accelerated pavement testing at 330,000 cycles when the water table was raised to 30 cm below the top of sand subgrade (Erlingsson and Ingason 2004). Such an effect of wetting from the water table up is depicted in Fig- ure 76b, where the initial moisture contents are indicated at the compacted optimum moisture condition. This increase in moisture content in excess of initial compaction value, primarily as a result of capillary rise from the water table, was indicated to be more critical in the long term than was the seasonal variation in layer moduli (AASHTO 2004, Appendix DD). Approaches for predicting the seasonal variation in pave- ment layer moduli have evolved from models that rely on regional adjustment factors that do not directly address sea- sonal variations in pavement structure to climate models that relate the changes in modulus to key factors affecting those changes, such as suction (Larson and Dempsey 1997). Explicit consideration of seasonal variation came with the introduction of mechanistically based design methods (Richter 2006). The MEPDG approach produces a design section that includes required thicknesses and elastic moduli for UAB and subbase for flexible and rigid pavements. The resil- ient modulus (MR) of the unbound layer materials used in MEPDG may be specified by means of stress-dependent ki parameters determined from lab testing (Level 1 MEPDG) or as a single average value determined per lab testing/field nondestructive FWD testing (Level 1 MEPDG), through correlation (Level 2 MEPDG) or estimated with typical val- ues (Level 3 MEPDG). The parameters required to estimate MR for Level 1 MEPDG by means of laboratory testing are derived from samples compacted at optimum moisture and maximum dry unit weight (standard or modified Proctor). For Level 2 MEPDG, MR is estimated by means of corre- lation from laboratory measured parameters such as CBR, Hveem stabilometer R-value, and so forth. For rehabilitation Level 1 MEPDG, MR is estimated through FWD backcal- culation. Through the Enhanced Integrated Climatic Model (EICM), MEPDG seasonally adjusts the subgrade and unbound layer moduli when performing fatigue and perma- nent deformation analyses. The EICM provides MR seasonal adjustment through suction model parameters and soil water characteristic curves (SWCCs). The EICM used in the MEPDG takes into account unsatu- rated soil mechanics concepts through the climatic-materials- structure and two-dimensional-drainage-infiltration models to calculate coupled heat-moisture flows in pavement struc- tures and predict pavement temperature (AASHTO 2004). The model evaluates the expected changes in moisture condition from the initial or reference condition (gener-

93 ally optimum moisture condition and maximum dry den- sity) as the unbound materials reach equilibrium moisture condition. Seasonal variation in modulus is determined by (1) computing the environmental effects such as layer mois- ture condition, (2) translating the computed layer moisture into suction through the Fredlund and Xing (1994) SWCC, and (3) predicting a seasonal modulus value from a modulus- suction relationship. The model also evaluates the sea- sonal changes in moisture condition and consequently the changes in resilient modulus, MR. This often is done on a biweekly basis for flexible pavements. The model also com- putes moisture and temperature in the middle of sublayers (established as finite difference node points in the EICM), calculates the effects of freezing, thawing, and recovery on MR, and uses the new MR values, corrected for envi- ronmental conditions, for calculation of critical pavement response parameters and damage at various points within the pavement system. The effects of varying moisture, freezing, thawing, and recovery on MR are reflected in the calculation of critical pavement responses and in the dam- age accumulation within the pavement system (AASHTO 2004, Appendix GG). However, a major concern exists in the way permanent deformation damage is computed using the unbound base/subbase rutting model adopted; the individual rutting amounts in the UAB/subbase layers are computed by incorporating only the changes in moisture (a) (b) FIGURE 76 Effect of shallow groundwater table on permanent deformation accumulation (Erlingsson and Ingason 2004).

94 content and MR but no applied stress state in relation to the strength properties. Furthermore, the EICM does not per- mit the use of models other than that of Fredlund and Xing (1994) to predict SWCC. The state of the practice in the United States regarding M-E design, field measurement of modulus, and its connec- tion to design modulus was summarized in NCHRP Synthe- sis 382 (Puppala 2008). Of the 41 states that responded to NCHRP Synthesis 382, 24 used the 1993 AASHTO Design Guide, seven states used the 1972 AASHTO Design Guide, five states (including California and Minnesota) used inter- nally developed mechanistic procedures, four states used internally developed empirical procedures, and one state used the 2002 AASHTO MEPDG. Regarding the use of sub- grade and aggregate base design moduli, 22 of the 41 states used MR in pavement design. Fourteen states determined MR through correlation to CBR, R-value, and so forth, and nine directly measured MR in the laboratory. Twenty states used FWD-based backcalculation of subgrade and sometimes base modulus for design of rehabilitation projects. In addi- tion, 12 states used FWD results to determine layer coeffi- cients for the 1993 Design Guide. Twenty-two states took the seasonal variation of modulus into account during pavement design in a variety of ways. The majority of states, including California, did not take seasonal variation into consideration. Arkansas chose the lowest modulus value from saturated lab testing, and Minnesota used internally developed charts. In the current synthesis study, 30 of 46 responding agen- cies indicated that climatic effects were a major concern as far as pavement subgrade performance was concerned (see Figure 77). Upon further investigation, a change in subgrade soil properties resulting from seasonal fluctuations was iden- tified as the primary concern. In addition, nearly 79% of the respondents indicated that the presence of fine-grained soils in areas susceptible to upward movement of the GWT was responsible for adverse climatic effects on pavement performance. Thirty-nine of 46 responding agencies do not conduct any testing to evaluate the aggregate materials selected for use in granular base/subbase applications for effects of adverse cli- matic conditions. Twenty-seven of 46 respondents indicated that effects of climatic changes on unbound aggregate layer performance were not considered in the pavement design procedure. Ten other agencies indicated that the approach adopted by the pavement design procedure to incorporate the effects of climatic changes on unbound aggregate layer performance was not clearly defined. Of the nine agencies accounting for the effect of climatic conditions on unbound aggregate layer performance, four adjusted the layer structural coefficients, seven modified the resilient modulus of unbound aggregate layers under different climatic conditions, and one adjusted the shear strength of the unbound aggregate layer. Moreover, one agency indicated that the drainage coefficients of unbound aggregate layers were changed under different climatic conditions. One more agency indicated that the mini- mum thickness requirements for unbound aggregate layers were modified according to climatic conditions. Note that the survey results reported in this synthesis reflect state practices as of May 2012. 57% 79% 46% 86% 61% 68% 61% 57% 16 (22) (13) (24) (17) (19) (17) (16) 0 5 10 15 20 25 0% 20% 40% 60% 80% 100% Ground water table (GWT) is often shallow (can be less than 5 ft. deep) under the pavements Native soils primarily fine-grained (e.g. silts, clays, etc.) and may get wet of optimum due to upward movement of moisture from the GWT In-service pavement subgrades are often under "wet of optimum" moisture conditions Seasonal fluctuations cause significant changes in subgrade soil properties Subgrades stay frozen for extended periods (one month or longer) More than 10 freeze-thaw cycles per year are experienced at the subgrade level Spring thaw weakening and timing of spring load restrictions Subgrade soils are primarily frost-susceptible (i.e. silty soils) Number of Responses Percentage of Respondents 28 survey respondents FIGURE 77 Different factors identified by state and Canadian provincial transportation agencies as responsible for affecting pavement performance under adverse climatic conditions.

95 Freeze-Thaw and Frost Penetration Frost susceptibility refers to the degree to which an unbound aggregate layer is affected by the action of freeze-thaw in the presence of water. In many northern states and Canadian prov- inces, the pavement, base, subbase, and subgrade materials experience one or more freeze-thaw cycles during each year, leading to frost-associated pavement distresses. Pavement dis- tresses associated mainly with frost heaving and thaw weak- ening can be commonly encountered given the presence of three factors: freezing temperatures, availability of moisture, and presence of frost-susceptible soils. The pavement failure mechanism associated with freeze-thaw involves nonuniform heave, which is destructive in terms of causing uneven sup- port, whereas thaw weakening causes deformation in the base or subgrade and eventually damages pavement surface. For example, in Minnesota, frost depth typically ranges between 40 and 70 in., greatly exceeding the thicknesses of the nonfrost susceptible bases and subbases (anecdotal evidence suggests that frost depths to 96 in. have been measured in northern Minnesota). If base/subbase layers have low fines (passing No. 200 sieve or smaller than 0.075 mm) content, treating frost-susceptible subgrade soils is the emphasis of mitigat- ing freeze-thaw damages; otherwise, both base/subbase and subgrade may need to be properly addressed. The existing methods of mitigating frost damage in flexible pavements can be costly and sometimes cumbersome (Khan 2008). Current MnDOT concrete pavement design practices require that a certain thickness of nonfrost-susceptible or frost-free materials be incorporated into pavement designs. Frost-free materials may include aggregate base (MnDOT’s Specification 3138, Classes 3, 4, 5, 6, and 7) and select gran- ular borrow (MnDOT’s Specification 3149.2B2) containing less than 12% passing the No. 200 sieve (0.075 mm). The minimum thickness of the frost-free materials is a function of the 20-year design lane ESALs and varies between 30 and 36 in. for most bituminous designs. To examine the ade- quacy of existing design standards for frost protection, bet- ter understanding of thermophysical properties of aggregate base/subbase materials and accurate modeling of pavement temperature-depth profile are required. Other existing methods of preventing or minimizing frost damage are briefly reviewed as follows: (1) simply increasing the pavement thickness to account for the damage and loss of support caused by frost action as the AASHTO 1993 Guide implies; (2) reducing the depth of frost-impacted subgrade under the pavement (between the bottom of the pavement structure and frost depth) by extending the pavement sec- tions well into the frost depth; (3) replacing the frost suscep- tible subgrade with nonfrost susceptible material; (4) using an insulation layer between the pavement and subgrade; (5) preventing free water from infiltrating into pavement structures; (6) providing a capillary break in the subgrade water flow path; (7) using alternative insulation materials (sawdust, sand/tire chips mix, extruded Styrofoam) for pre- venting frost action; (8) using a peat layer above the sub- grade soils; and (9) engineering a pavement structure with reduced heat conductivity using lightweight aggregate, as proposed by Khan (2008). According to Saeed (2008), the frost susceptibility of aggregates can be determined in terms of the USACE “F” categories and from the results of the TST (Saarenketo and Scullion 1996). The USACE method categorizes soils into several categories based on their degree of susceptibility, from F1 (least susceptible) to F4 (most susceptible). The F categories are based on general soil type and the amount of material finer than 0.02 mm. The TST measures the amount of free water that exists within an aggregate sample. The asymptotic dielectric constant value (DCV) at the end of the test can be used to characterize an aggregate as a poor (>16), marginal (10 to 16), or good (<10) performer in terms of its moisture susceptibility and frost resistance. Key Lessons • With proper consideration of the effects of climatic conditions, such as frost penetration and freeze-thaw cycles, on UAB/subbase layer performance, pre- mature pavement failures can be prevented. • TSTs are conducted to evaluate the frost susceptibil- ity of aggregates before their application in unbound base/subbase layers in areas experiencing significant frost penetration. REFERENCES American Association of State Highway Officials (AASHO), AASHO Interim Guide for the Design of Rigid and Flex- ible Pavements, AASHO, Washington, D.C., 1961. American Association of State Highway and Transportation Officials (AASHTO), Pavement Design Guide, Section 2.0, Fundamental Design Parameters, AASHTO, Wash- ington, D.C., 1998. American Association of State Highway and Transportation Officials (AASHTO), Resilient Modulus of Unbound Granular Base/Subbase Materials and Subgrade Soils— SHRP Protocol P46, Standard Specifications for Trans- portation Materials and Methods of Sampling and Testing, AASHTO T294-94, 17th ed., AASHTO, Wash- ington, D.C., 1995. American Association of State Highway and Transportation Officials (AASHTO), Standard Test Method for Deter- mining the Resilient Modulus of Soils and Aggregate Materials, AASHTO T307-99, AASHTO, Washington, D.C., 2003.

96 American Association of State Highway and Transportation Officials (AASHTO), Guide for Mechanistic-Empirical Design of New and Rehabilitated Pavement Structures, Appendix GG-1: Calibration of Permanent Deformation Models for Flexible Pavements, prepared for NCHRP by ARA, Inc., AASHTO, Washington, D.C., Feb. 2004. Adu-Osei, A., D.N. Little, and R.L. Lytton, Structural Char- acteristics Of Unbound Aggregate Bases To Meet AAS- HTO 2002 Design Requirements, International Center for Aggregates Research (ICAR) Report 502-1, Texas Trans- portation Institute, The Texas A&M University System, College Station, 2001. Allen, J.J., The Effects of Non-Constant Lateral Pressures on The Resilient Response of Granular Materials, Ph.D. Dis- sertation, Department of Civil Engineering, University of Illinois, Urbana, 1973. Allen, J.J. and M.R. Thompson, “Resilient Response of Granular Materials Subjected to Time-Dependent Lateral Stresses,” Transportation Research Record 510, Trans- portation Research Board, National Research Council, Washington, D.C., 1974, pp. 1–13. Arika, C.N., D.J. Canelon, and J.L. Nieber, Subsurface Drainage Manual for Pavements in Minnesota, Final Report MN/RC 2009-17, Minnesota Department of Transportation, St. Paul, 2009. Ashtiani, R.S. and D. Little, Acceptability Criteria for High Fines Content Aggregate Pavement Layers, Report No ICAR 401131, International Center for Aggregates Research (ICAR), College Station, Tex., 2007, 100 pp. Ashtiani, R.S. and D. Little, Methodology For Designing Aggregate Mixtures For Base Courses, Report No 508-1, International Center for Aggregates Research (ICAR), Col- lege Station, Tex., 2009, 336 pp. Barksdale, R.D. and H.A. Todres, A Study of Factors Affecting Crushed Stone Base Performance, School of Civil Engi- neering, Georgia Institute of Technology, Atlanta, 1983. Barksdale, R.D., S.F. Brown, and F. Chan, NCHRP Report 315: Potential Benefits of Geosynthetics in Flexible Pave- ments, Transportation Research Board, National Research Council, Washington, D.C., 1989. Barksdale, R.D. and J.L. Alba, Laboratory Determina- tion of Resilient Modulus for Flexible Pavement Design, Interim Report No. 2, prepared for NCHRP, Transportation Research Board, National Research Council, Washington, D.C., 1993. Barksdale, R.D., J. Alba, P.N. Khosla, R. Kim, P.C. Lambe, and M.S. Rahman, NCHRP Web Document 14: Laboratory Determination of Resilient Modulus for Flexible Pavement Design, Transportation Research Board, National Research Council, Washington, D.C., 1997, 486 pp. Beatty, T.L., et al., Pavement Preservation Technology in France, South Africa, and Australia, Office of Interna- tional Programs, Federal Highway Administration, U.S. Department of Transportation, and the American Associa- tion of State Highway and Transportation Officials, Alex- andria, Va., 2002. Birgisson, B. and R. Roberson, “Drainage of Pavement Base Material: Design and Construction Issues,” Trans- portation Research Record: Journal of the Transporta- tion Research Board, No. 1709, Transportation Research Board of the National Academies, Washington, D.C., 2000, pp. 11–18. Boussinesq, J., “Application des Potentiels a l’etude de l’equilibre et du Mouvement des Solids Elastiques,” Gauthier-Villars, Paris, 1885. British Standard BS EN 13286-7, Unbound and Hydrauli- cally Bound Mixtures. Cyclic Load Triaxial Test for Unbound Mixtures, British-Adopted European Standard, Feb. 2004, 40 pp. Broms, B., “Lateral Earth Pressure Due to Compaction of Cohesionless Soils,” Proceedings, 4th Budapest Confer- ence on Soil Mechanics and Foundation Engineering, 1971, pp. 373–383. Brown, S.F. and A.F.L. Hyde, “Significance of Cyclic Con- fining Stress in Repeated-Load Triaxial Testing of Gran- ular Material,” Transportation Research Record 537, Transportation Research Board, National Research Coun- cil, Washington, D.C., 1975, pp. 49–58. Brown, S.F. and J.W. Pappin, “Analysis of Pavements with Granular Bases,” Transportation Research Record 810, Transportation Research Board, National Research Coun- cil, Washington, D.C., 1981, pp. 17–23. Brown, S.F. and B.V. Brodrick, “25 Years’ Experience with the Pilot-Scale Nottingham Pavement Test Facility,” Pro- ceedings of the International Conference on Accelerated Pavement Testing, Reno, Nev., Oct. 18–20, 1999. Carrera, A., A. Dawson, and J. Steger, State of the Art Materials’ Sensitivity to Moisture Content Change: Report Nr. 2, P2R2C2, Pavement Performance and Remediation Require- ments due to Climate Change, Road ERA Net, 2009. CEN EN 13286-7, Unbound and Hydraulically Bound Mix- tures—Test Methods—Part 7: Cyclic Load Triaxial Test for Unbound Mixtures, CEN, European Committee for Standardization, 2004. COST Action 337, Unbound Granular Materials for Road Pavements, Final Report, Chapter 8, European Commis- sion, Directorate General for Research and Technological Development, Brussels, Belgium, 2002. Chen, D.H. and H.H. Lin, “Effect of Base Support and Load Transfer Efficiency (LTE) on Portland Concrete Pave- ment Performance,” Journal of Testing and Evaluation, No. 38, No. 1, 2010, pp. 1–10. Christopher, B.R., C. Schwartz, and R. Boudreau, Geo- technical Aspects of Pavements, Report FHWA NHI- 10-092, National Highway Institute, Federal Highway Administration, Washington, D.C., 2010. Crockford, W.W., L.J. Bendana, W.S. Yang, S.K. Rhee, and S.P. Senadheera, Modeling Stress and Strain States In Pavement Structures Incorporating Thick Granular Layers, Final Report, Contract F08635-87-C-0039, The Texas Transportation Institute, Texas A&M University, College Station, 1990.

97 Cundall, P.A. and O.D.L. Strack, “A Discrete Numerical Model for Granular Assemblies,” Geotechnique, Vol. 29, 1979, 47 pp. Dantu, P., “Contribution a l’Etude Mechanique et Geometrique des Milieux Pulverulents,” Proceedings, 4th International Conference on Soil Mechanics, Vol. 1, London, U.K., 1957. Dawson, A., “Pavement Performance and Remediation Requirements Following Climate Change,” 2010 [Online]. Available: http://www.nottingham.ac.uk/~evzard/P2R2C2/ Brochure.pdf. Dhamrait, J.S. and D.R. Schwartz, Effect of Subbase Type and Subsurface Drainage on Behavior of CRC Pave- ments, Interim Report IHR-36, Illinois Department of Transportation, Springfield, 1979, 70 pp.. Dobry, R., T.T. Ng, and E. Petrakis, “Deformation Character- istics of Granular Soil in the Light Of Particulate Mechan- ics,” Proceedings, 14th Conference on Geotechnics, Italian Geotechnical Association, Torino, Nov. 28–30, 1989. Edwards, J.P., N.H. Thom, P.R. Fleming, and J. Williams, “Accelerated Laboratory Based Mechanistic Testing of Unbound Materials within the Newly Developed NAT Springbox,” Transportation Research Record: Journal of the Transportation Research Board, No. 1913, Transpor- tation Research Board of the National Academies, Wash- ington, D.C., 2005. Elhannani, M., Modelisation et Simulation Numerique des Chaussees Souples, Ph.D. Thesis, University of Nantes, France, 1991. Erlingsson, S. and T. Ingason, “Performance of Two Thin Pavement Structures during Accelerated Pavement Test- ing Using a Heavy Vehicle Simulator,” Proceedings of the Second International Conference on Accelerated Pavement Testing, University of Minnesota, Minneapolis, 2004. Federal Highway Administration (FHWA), “Transportation Applications of Recycled Concrete Aggregate,” FHWA State of the Practice National Review, FHWA, Washing- ton, D.C., Sep. 2004. Fledlund, D.G. and A. Xing, “Equations for the Soil-Water Characteristic Curve,” Canadian Geotechnical Journal, Vol. 31, No. 4, 1994, pp. 533–546. Garg, N., E. Tutumluer, and M.R. Thompson, “Structural Modeling Concepts for the Design of Airport Pavements for Heavy Aircraft,” Proceedings, 5th International Con- ference on the Bearing Capacity of Roads and Airfields, Trondheim, Norway, July 1998, pp. 115–124. Hagen, M.G. and G.R. Cochran, “Comparison of Pavement Drainage Systems.” Transportation Research Record 1519, Transportation Research Board, National Research Council, Washington, D.C., 1995, pp. 1–10. Hicks, R.G., Factors Influencing the Resilient Properties of Granular Materials, PhD Dissertation, Institute of Trans- portation and Traffic Engineering, University of California, Berkeley, 1970. Holl, D.L., “Plane Strain Distribution of Stress in Elastic Media,” Iowa Engineering Experimentation Station Bul- letin, 1941, pp. 148–163. Hornych, P., A. Kazai, and A. Quibel, “Modeling a Full- Scale Experiment of Two Flexible Pavement Structures with Unbound Granular Bases,” Unbound Aggregates in Road Construction, A.R. Dawson, Ed., A.A. Balkema Publishers, Proceedings of the Unbound Aggregates in Roads (UNBAR5) Symposium, University of Nottingham, England, June 21–23, 2000, pp. 359–367. Ingold, T.S., “The Effect of Compaction on Retaining Walls,” Geotechnique, Vol. 29, No. 3, 1979, pp. 265–283. Jung, Y.S., D.G. Zollinger, M. Won, and A.J. Wimsatt, Sub- base and Subgrade Performance Investigation for Con- crete Pavement, Final Report FHWA/TX-09/0-6037-1, Texas Department of Transportation, Austin, 2009. Jung, Y.S., D.G. Zollinger, and A.J. Wimsatt, “Test Method and Model Development of Subbase Erosion for Con- crete Pavement Design,” Transportation Research Record; Journal of the Transportation Research Board, No. 2154, Transportation Research Board of the National Academies, Washington, D.C., 2010, pp. 22–31. Jung, Y.S. and D.G. Zollinger, “New Laboratory-Based Mechanistic-Empirical Model for Faulting in Jointed Concrete Pavement,” Transportation Research Record; Journal of the Transportation Research Board, No. 2226, Transportation Research Board of the National Acad- emies, Washington, D.C., 2011, pp. 60–70. Karasahin, M., A.R. Dawson, and J.T. Holden, “Applicability of Resilient Constitutive Models of Granular Material for Unbound Base Layers,” Transportation Research Record 1406, Transportation Research Board, National Research Council, Washington, D.C., 1993, pp. 98–107. Khan, A.A., Feasibility of Using Lightweight Aggregate to Mitigating Frost Damage in Flexible Pavements, PhD Thesis, Department of Civil Engineering, University of New Brunswick, Canada, 2008. Khazanovich, L., M. Darter, R. Bartlett, and T. McPeak, Common Characteristics of Good and Poorly Performing PCC Pavements, Final Report FHWA-RD-97-131, Fed- eral Highway Administration, McLean, Va., 1998. Kim, I.T. and E. Tutumluer, “Unbound Aggregate Rutting Models for Stress Rotation and Effects of Moving Wheel Loads,” Transportation Research Record: Journal of the Transportation Research Board, No. 1913, Transportation Research Board of the National Academies, Washington, D.C., 2005, pp. 41–52. Kim, I.T., Permanent Deformation Behavior of Airport Flexible Pavement Base and Subbase Courses, PhD Dis- sertation, University of Illinois at Urbana–Champaign, Urbana, 2005. Kim, S.-H., D.N. Little, and E. Masad, “Simple Methods to Estimate Inherent and Stress-Induced Anisotropy of Aggre- gate Base,” Transportation Research Record: Journal of the Transportation Research Board, No. 1913, Transporta- tion Research Board of the National Academies, Washing- ton, D.C., 2005, pp. 24–31. Kim, S.-H., E. Tutumluer, D.N. Little, and N. Kim, “Effect of Gradation on Nonlinear Stress-Dependent Behavior of

98 a Sandy Flexible Pavement Subgrade,” Journal of Trans- portation Engineering, ASCE, Vol. 133, No. 10, Oct. 2007, pp. 590–598. Kolisoja, P., T. Saarenketo, H. Peltoniemi, and N. Vuorimies, “Laboratory Testing of Suction and Deformation Proper- ties of Base Course Aggregates,” Transportation Research Record: Journal of the Transportation Research Board, No. 1787, Transportation Research Board of the National Academies, Washington, D.C., 2002, pp. 83–89. Kwon, J., E. Tutumluer, I.L. Al-Qadi, and S. Dessouky, “Mechanistic Model Response Predictions of Geogrid Base Reinforced Flexible Pavements,” Proceedings, Geo- Americas 2008, The First Pan American Geosynthetics Conference, Cancun, Mexico, Mar. 2–5, 2008. Larson, G. and B. Dempsey, Enhanced Integrated Climatic Model, Version 2.0, Report DTFA MN/DOT 72114, Department of Civil Engineering, University of Illinois at Urbana–Champaign, Urbana, 1997. Lekarp, F., U. Isacsson, and A. Dawson, “State of the Art. I: Resilient Response of Unbound Aggregates,” Journal of Transportation Engineering, Vol. 126, No. 1, 2000a, pp. 66–75. Lekarp, F., U. Isacsson, and A. Dawson, “State of the Art. II: Permanent Strain Response of Unbound Aggregates,” Journal of Transportation Engineering, Vol. 126, No. 1, 2000b, pp. 76–83. Love, A.E.H., A Treatise on the Mathematical Theory of Elasticity, 4th ed., Dover Publications, New York, N.Y., 1944, 643 pp. Masad, S., D.N. Little, and E. Masad, “Analysis of Flexible Pavement Response and Performance Using Isotropic and Anisotropic Material Properties,” Journal of Trans- portation Engineering, ASCE, Vol. 132, No. 4, 2006, pp. 342–349. Mishra, D., Aggregate Characteristics Affecting Response and Performance of Unsurfaced Pavements on Weak Subgrades, PhD Thesis, University of Illinois at Urbana– Champaign, Urbana, 2012. Mishra, D. and E. Tutumluer, “Aggregate Physical Proper- ties Affecting Modulus and Deformation Characteristics of Unsurfaced Pavements,” Paper 11-3486, 90th Annual Meeting of the Transportation Research Board, Jan. 2011, Washington, D.C., 16 pp. Nataatmadja, A., Variability of Pavement Material Param- eters under Repeated Loading, PhD Dissertation, Monash University, Australia, 1989. NCHRP 1-37A, Mechanistic-Empirical Pavement Design Guide, Draft Report, Part 2, Design Inputs, Transporta- tion Research Board of the National Academies, Wash- ington, D.C., 2004. Newcomb, D.E., S.W. Lee, J.P. Mahoney, and N.C. Jackson, “The Use of Falling Weight Deflectometer Data in Moni- toring Flexible Pavement Systems,” Nondestructive Test- ing of Pavements and Backcalculation of Moduli, ASTM STP 1026, A.J. Bush III and G.Y. Baladi, Eds., American Society for Testing and Materials, Philadelphia, Pa, 1989, pp. 470–486. Oda, M., “A Mechanical and Statistical Model of Granular Materials,” Soils and Foundations, Japanese Society of Soil Mechanics and Foundation Engineering, Vol. 14, 1974. Puppala, A.J., Synthesis of Highway Practice 382: Estimat- ing Stiffness of Subgrade and Unbound Materials for Pavement Design, Transportation Research Board of the National Academies, Washington, D.C., 2008. Richter, C.A., Seasonal Variations in the Moduli of Unbound Pavement Layers, Report FHWA-HRT-04-079, Federal Highway Administration, McLean, Va., 2006. Saarenketo, T. and T. Scullion, Using Electrical Properties to Classify the Strength Properties of Base Course Aggregates, Final Report FHWA/TX-97/1341-2, Texas Department of Transportation, Austin, 1996. Saeed, A., J.W. Hall, Jr. and W. Barker, NCHRP Report 453: Performance-Related Tests of Aggregates for Use in Unbound Pavement Layers, Transportation Research Board, National Research Council, Washington, D.C., 2001, 55 pp. Saeed, A., NCHRP Report 598: Performance-Related Tests of Recycled Aggregates for Use in Unbound Pavement Layers, Transportation Research Board of the National Academies, Washington, D.C., 2008, 53 p. Selig, E.T., “Tensile Zone Effects on Performance of Lay- ered Systems,” Geotechnique, Vol. 37, No. 3, 1987, pp. 247–254. Semmelink, C.J. and M. de Beer, “Rapid Determination of Elastic and Shear Properties of Road-Building Materials with the K-Mould,” Proceedings of the UNBAR4 Sym- posium, A.R. Dawson and R.H. Jones, Eds., July 17–19, 1995, Nottingham, U.K., pp. 151–162. Seyhan, U. and E. Tutumluer, “Unbound Granular Mate- rial Characterization from Stress Path Loading Tests,” Geotechnical Special Publication (GSP) No. 89, Recent Advances in the Characterization of Transportation Geo- materials, E. Tutumluer and A.T. Papagiannakis, Eds., 1999, pp. 49–60. Seyhan, U., E. Tutumluer, and H. Yesilyurt, “Anisotropic Aggregate Base Inputs for Mechanistic Pavement Analy- sis Considering Effects of Moving Wheel Loads,” ASCE Journal of Materials in Civil Engineering, Vol. 17, No. 5, Oct. 2005, pp. 505–512. Shackel, B., “Repeated Loading of Soils–A Review,” Austra- lian Road Research, Vol. 5, No. 3, 1973, pp. 22–49. Shackel, B., Some Effects of Repeated Triaxial Stresses on Road Pavement Materials, PhD Thesis, The University of New South Wales, 1974. Steven, B., D. Alabaster, and J. de Pont, “Elastic Nonlinear Finite Element Analysis of a Flexible Pavement Sub- jected to Varying Falling Weight Deflectometer Loads,” Transportation Research Record: Journal of the Trans- portation Research Board, No. 2016, Transportation Research Board of the National Academies, Washington, D.C., 2007, pp. 31–38.

99 Stewart, E.H., E.T. Selig, and G.H. Norman-Gregory, “Fail- ure Criteria and Lateral Stresses in Track Foundations,” Transportation Research Record 1022, Transportation Research Board, National Research Council, Washing- ton, D.C., 1985, pp. 59–64. Tao, M. and M.Y. Abu-Farsakh, Effect of Drainage in Unbound Aggregate Bases on Flexible Pavement Perfor- mance, Final Report FHWA/LA.07/429, Louisiana Trans- portation Research Center, Baton Rouge, 2008, 48 pp. Tao, M., M.N. Louay, N.D. Munir, Z. Zhang, and Z. Wu, “Application of Shakedown Theory in Characterizing Traditional and Recycled Pavement Base Materials,” Journal of Transportation Engineering, Vol. 136, No. 3, 2010, pp. 214–222. Thompson, M.R., “State-of-the-Art: Unbound Base Per- formance,” Proceedings of the 6th Annual Symposium of International Center for Aggregate Research (ICAR), Austin, Tex., 1998. Toros, U. and D.R. Hiltunen, “Effects of Moisture and Time on Stiffness of Unbound Aggregate Base Course Mate- rials,” Transportation Research Record: Journal of the Transportation Research Board, No. 2016, Transporta- tion Research Board of the National Academies, Wash- ington, D.C., 2008, pp. 41–51. Tutumluer, E., Predicting Behavior of Flexible Pavements with Granular Bases, PhD Dissertation, School of Civil and Environmental Engineering, Georgia Institute of Technology, Atlanta, Ga., 1995. Tutumluer, E., “Anisotropic Behavior of Unbound Aggre- gate Bases—State of the Art Summary.” Invited Paper, Proceedings, 6th Annual Symposium of the International Center for Aggregate Research (ICAR), St. Louis, Mo., Apr. 19–21, 1998, pp. 11–33. Tutumluer, E. and R.D. Barksdale, “Behavior of Pavements with Granular Bases—Prediction and Performance.” Proceedings of the International Symposium on Unbound Aggregates in Roads (UNBAR4), Nottingham, U.K., July 1995. Tutumluer, E. and M.R. Thompson, “Granular Base Mod- uli for Mechanistic Pavement Design,” Proceedings, ASCE Airfield Pavement Conference, Seattle, Wash., Aug. 17–20, 1997a, pp. 33–47. Tutumluer, E. and M.R. Thompson, “Anisotropic Modeling of Granular Bases in Flexible Pavements,” Transporta- tion Research Record 1577, Transportation Research Board, National Research Council, Washington, D.C., 1997b, pp. 18–26. Tutumluer, E. and M.R. Thompson, Anisotropic Modeling of Granular Bases, Technical Research Report Submit- ted to the Federal Aviation Administration, Department of Civil Engineering, University of Illinois at Urbana– Champaign, Urbana, 1998. Tutumluer, E. and U. Seyhan, “Laboratory Determination of Anisotropic Aggregate Resilient Moduli Using An Inno- vative Test Device,” Transportation Research Record: Journal of the Transportation Research Board 1687, Transportation Research Board, National Research Coun- cil, Washington, D.C., 1999, pp. 13–21. Tutumluer, E. and I.T. Kim, “Permanent Deformation Behav- ior of Airport Pavement Granular Layers,” Proceedings of the 3rd International Conference on Deformation Char- acteristics of Geomaterials (IS LYON 2003),” H. DiBene- detto, T. Doanh, H. Geffroy, and C. Sauzeat, Eds., Lyon, France, Sep. 22–24, 2003, A.A. Balkema Publishers, the Netherlands, pp. 345–352. Tutumluer, E. and U. Seyhan, “Effects of Fines Content on the Anisotropic Response and Characterization of Unbound Aggregate Bases,” Proceedings, 5th International Con- ference on Unbound Aggregates in Roads (UNBAR5), Nottingham, England, 2000, pp. 153–161. Tutumluer, E., A. Adu-Osei, D.N. Little, and R.L. Lytton, Field Validation of the Cross-Anisotropic Behavior of Unbound Aggregate Bases, International Center for Aggregates Research (ICAR) Report 502-2, Texas Trans- portation Institute, The Texas A&M University System, College Station, 2001. Tutumluer, E., D.N. Little, and S.H. Kim, “Validated Model for Predicting Field Performance of Aggregate Base Courses,” Transportation Research Record: Journal of the Transportation research Board, No. 1837, Transpor- tation Research Board of the National Academies, Wash- ington, D.C., 2003, pp. 41–49. Tutumluer, E., “State of the Art: Anisotropic Characterization of Unbound Aggregate Layers in Flexible Pavements,” ASCE Geotechnical Special Publication No. 184, Pave- ments and Materials—Modeling, Testing and Performance, Z. You, A.R. Abbas, and L. Wang, Eds., 2009, pp. 1–16. Uhlmeyer, J.S., J.P. Mahoney, G. Hanek, G. Wang, R.L. Cop- stead, and D.J. Janssen, Estimation of Seasonal Effects for Pavement Design and Performance—Volume 1, Report FHWA-FLP-95-006, Federal Highway Administration, Washington, D.C., 1996. U.S. Army Corps of Engineers, “The Unified Soil Classifica- tion System,” Technical Memorandum 3-357, Waterways Experiment Station, Vicksburg, Miss., 1953. Uzan, J., “Characterization of Granular Material,” Transpor- tation Research Record 1022, Transportation Research Board, National Research Council, Washington, D.C., 1985, pp. 52–59. Werkmeister, S., Permanent Deformation Behaviour of Unbound Granular Materials in Pavement Construc- tions, PhD Thesis, Dresden University of Technology, Dresden, Germany, 2003. Werkmeister, S., R. Numrich, A.R. Dawson, and F. Well- ner, “Design of Granular Pavement Layers Considering Climatic Conditions,” Transportation Research Record: Journal of the Transportation Research Board, No. 1837, Transportation Research Board of the National Acad- emies, Washington, D.C., 2003, pp. 61–70. Werkmeister, S., A.R. Dawson, and F. Wellner, “Pavement Design Model for Unbound Granular Materials,” Journal

100 of Transportation Engineering, ASCE, Vol. 130, No. 5, 2004, pp. 665–674. White, D.J., P.K.R. Vennapusa, D. Eichner, H. Gieselman, L. Zhao, and C. Jahren, Rapid, Self-Contained In situ Permeameter for Field QA/QC of Pavement Base/Subbase Materials, Final Report for Highway IDEA Project 130, Transportation Research Board of the National Academies, Washington, D.C., 2010, 50 pp. Winkelman, T.J., Open Graded Drainage Layer Perfor- mance in Illinois, Final Report FHWA/IL/PRR 147, Illi- nois Department of Transportation, Springfield, 2004. Xiao, Y., E. Tutumluer, Y. Qian, and J. Siekmeier, “Grada- tion Effects Influencing Mechanical Properties of Aggre- gate Base and Granular Subbase Materials in Minnesota,” Transportation Research Record: Journal of the Trans- portation Research Board, No. 2267, Transportation Research Board of the National Academies, Washington, D.C., 2012, pp. 14–26. Zapata, C.E. and W.N. Houston, NCHRP Report 602: Calibration and Validation of the Enhanced Integrated Climate Model for Pavement Design, Transportation Research Board of the National Academies, Washington, D.C., 2008. Zeilmaker, J, and R.J. Henny, “The Measurement of Resid- ual Stresses Due to Compaction in Granular Materials,” In Proceedings, International Symposium on Unbound Aggregates on Roads (UNBAR3), Unbound Aggregates in Roads, R.H. Jones and A.R. Dawson, Eds., Reed International Publishing, United Kingdom, Apr. 1989, pp. 159–168.

Next: Chapter Five - Compaction, Quality Control, and Field Performance »
Practices for Unbound Aggregate Pavement Layers Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

TRB’s National Cooperative Highway Research Program (NCHRP) Synthesis 445: Practices for Unbound Aggregate Pavement Layers consolidates information on the state-of-the-art and state-of-the-practice of designing and constructing unbound aggregate pavement layers. The report summarizes effective practices related to material selection, design, and construction of unbound aggregate layers to potentially improve pavement performance and longevity.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!