Field Evaluation of the Stiffness of Unbound Aggregate Base Layers in Inverted Flexible Pavements

Transcription

1 Terrell, Cox, Stokoe, Allen and Lewis Word Count 7,448 Field Evaluation of the Stiffness of Unbound Aggregate Base Layers in Inverted Flexible Pavements By Ronald G. Terrell, Graduate Research Assistant Civil Engineering Department, The University of Texas at Austin Austin, Texas 78712, Tel: (512) , Fax: (512) Brady R. Cox, Graduate Research Assistant Civil Engineering Department, The University of Texas at Austin Austin, Texas 78712, Tel: (512) , Fax: (512) Kenneth H. Stokoe, II, Jennie C. and Milton T. Graves Chair in Engineering Department of Civil Engineering, The University of Texas at Austin Austin, Texas 78712, Tel: (512) , Fax: (512) John J. Allen, Associate Director, International Center for Aggregate Research Department of Civil Engineering, The University of Texas at Austin Austin, Texas 78712, Tel: (512) , Fax: (512) Dwayne Lewis, The Georgia Department of Transportation, 15 Kennedy Dr., Forest Park, Georgia Tel: (404) ,

2 Terrell, Cox, Stokoe, Allen and Lewis 2 Field Evaluation of the Stiffness of Unbound Aggregate Base Layers in Inverted Flexible Pavements by Ronald G. Terrell, Brady R. Cox, Kenneth H. Stokoe, II, John J. Allen, and Dwayne Lewis, Abstract: Unbound aggregate base layers in a quarry haul road in Georgia were characterized using embedded sensors and in-situ seismic testing. Two sections of the road were constructed as inverted pavements, one using a South African Roads Board method and the other using a conventional Georgia Department of Transportation method. A third was constructed using a traditional method. Miniaturized versions of traditional crosshole and downhole seismic tests were conducted to determine the stiffness of each base layer. Horizontally propagating compression and shear waves were measured under four different loading conditions to determine Young s moduli and Poisson s ratios of the base. An increase in stiffness with an increase in load was measured. Additionally, it was found that the Georgia and South Africa sections had similar stiffness. Surprisingly, the traditional section was found to be somewhat stiffer than the other sections. This higher stiffness is thought to be due to a prolonged period of compaction prior to construction of the unbound aggregate base layer, essentially transforming the traditional section into an inverted pavement. Using the vertical total normal stresses computed from ILLI-PAVE, a value of 0.3 for the earth pressure coefficient was found to be reasonable for this material in determining the radial total normal stresses. The radial effective normal stresses were calculated from the radial total normal stresses and experimentally determined pore-water pressures. Additionally, the negative pore-water pressures in the partially saturated granular base had a significant impact on the stiffness of the unbound aggregate base layer, especially under small load levels. INTRODUCTION The AASHTO procedure for designing flexible pavements is currently used by state DOTs. In this procedure, the conventional flexible pavement is composed of three components: (1) a prepared subgrade, (2) layers of subbase and base materials, and (3) a surface layer. The surface layer is normally a hot mix asphalt (HMA), and the base and subbase layers are constructed from unbound granular materials. These unbound granular layers can serve as major structural components of the pavement system, but their contribution is often treated rather conservatively in design. The feasibility of relying more on the unbound aggregate base (UAB) for structural support has been raised, especially when used in conjunction with a cemented subbase. Such a pavement system is called an inverted pavement system. In an inverted pavement, the stiffness of one of the lower supporting layers is greater than the stiffness of the upper structural layers. A stabilized, or cemented, subbase layer acts as a platform for the UAB. The stiffness of the supporting layers beneath this cemented subbase is decreased gradually and systematically with depth, creating a relatively deep supportive pavement structure. In addition to acting as a backbone of the pavement, the cemented subbase acts as a firm base on which to build the UAB layer and keeps this base layer in a state of compression when loaded (South Africa National Roads Agency, 1998). The South African Roads Board (SARB) maintains that the inverted pavement system provides such an excellent structural performance that it requires only a thin HMA riding course.

3 Terrell, Cox, Stokoe, Allen and Lewis 3 A cooperative field study was undertaken between the Georgia Department of Transportation (GDOT) and the International Center for Aggregates Research (ICAR) at the University of Texas and Texas A&M University to investigate the characteristics of UAB layers in inverted flexible pavement sections. It is particularly important to know the characteristics of the UAB materials as these materials strongly influence the performance of the flexible pavement. In this study, the GDOT constructed two test sections of inverted pavement in a haul road at the entrance road to a crushed aggregate quarry in Morgan County, Georgia. One inverted section was constructed using the SARB method; the other inverted section was constructed using the conventional GDOT method (Department of Transportation, State of Georgia, 1993). A third control section in the haul road was constructed in the traditional method so that it could be compared with the other two sections. In-situ density and water content measurements were performed in each section by GDOT personnel. In-situ seismic testing was performed to evaluate the stiffnesses of the UAB layer in these sections. Seismic crosshole and downhole tests in the UAB were performed to allow a direct comparison of the stiffness characteristics of the three different sections. Geophones were installed in the UAB layers to measure the travel times of compression and shear waves under different loading conditions ranging from no external load to heavy loads on the pavements.. The haul road is shown in Figure 1. It contains the following sections: 1. Section #1: 300-m length of traditional haul road (prepared subgrade with minimum CBR of 15, 5.1 cm of GAB, 15.2 cm of surge stone, 20.3 cm GAB, and topped with 7.6 cm of HMA); 2. Section #2: 120-m length of inverted pavement with a South Africa base; and 3. Section #3: 120-m length of inverted pavement with a Georgia base. Pavement Sections #2 and #3 include a prepared subgrade, 5.1 cm of unbound GAB, 20.3 cm of GAB with 4% to 5% cement by volume (the backbone layer), 15.2 cm of unbound GAB (UAB), and 7.6 cm of Marshall design HMA. The only difference between the South Africa Section #2 and Georgia Section #3 is that a process known as slushing was used as part of the South Africa compaction process. Slushing involves compacting the UAB under saturated conditions and sweeping off the fines that squeeze to the surface, resulting in a super dense matrix that is 88% of the solid particle density. The GDOT believes that with the cemented layer acting as a compaction platform, the same density can be achieved in the Georgia base using typical compaction methods without slushing. FIELD MEASUREMENTS OF THE UNBOUND AGGREGATE BASE LAYERS As noted above, the haul road was constructed with three test sections. Two sections were inverted pavement sections which were constructed with a South Africa base section and a Georgia base section. The third test section was constructed with a conventional base layer. The characteristics determined for each UAB layer were: (1) total density, (2) water content, (3) small-strain stiffness in terms of Young s modulus, (4) Poisson s ratio, and (5) the variation in Young s modulus with stress state.

4 Terrell, Cox, Stokoe, Allen and Lewis 4 Physical Characteristics of the Unbound Aggregate Base The GAB material used in the UAB layer for all three sections is classified by the GDOT 1993 Specifications as Group II slag, gravel, granitic and gneissic rocks, quartzite, synthetic aggregate, or any combination thereof. It was produced by the Lafarge Morgan County Quarry, located adjacent to the test road. The GDOT determined the theoretical maximum dry density to be 2193 kg/m 3 with an optimum gravimetric water content, w, of 6.7%, as shown in Figure 2. The GAB is non-plastic with a mean grain size, D 50, of 6.5 mm, a D 10 of 0.1 mm, a uniformity coefficient, C u, of 100, and a Unified Soil Classification System Designation of GP-GM. A summary of the index properties is presented in Table 1. The Lafarge quarry performed a sieve analysis on a sample of the stockpile material in accordance with ASTM C It showed 45% passing the #4 (4.75 mm) sieve and 8% passing the #200 (.075 mm) sieve. The fines were non-plastic. The gradation curve, along with upper and lower GDOT limits, is shown in Figure 2. Total Density and Water Contents The GDOT determined the in-situ wet and dry unit weights and the water contents for the two inverted test sections immediately after construction using the Sand Cone Method (ASTM D1556) and a Troxler 3430 Nuclear Density Gauge (NDG) (AASHTO T310-00). The total unit weight and water content for the traditional section were assumed to be 2400 kg/m 3 and 6%, which was supported by the results of the seismic testing. The calibration values used in calibrating the NDG for dry unit weight and water content were calculated using the average of the three sand cone dry unit weights. The GDOT also determined the water contents for all three sections using an NDG. These values are presented in Table 2. It is interesting to note that the Georgia procedure produced a GAB layer with a dry unit weight that is slightly above that produced with the South African procedure. The differences in the unit weights determined from the sand cone and NDG methods for the South African section are due to the large variances in dry unit weights used in calculating the calibration factor. FIELD SEISMIC TESTING TO CHARACTERIZE THE BASE LAYERS Evaluation of Moduli and Poisson s Ratio The primary thrust of this study involved evaluating the stiffness of the base layers. Measurements of the propagation velocities of seismic body waves are a long-practiced method of characterizing the small-strain stiffness of all types of soils in the laboratory and in the field. Specifically, shear wave and constrained compression wave velocities, V s and V p, respectively, are used to determine the shear modulus, G, and constrained modulus, M, respectively. These parameters are measured in the field at strains less than or equal to % where the moduli are independent of strain amplitude and exhibit the largest values; hence, these measurements are called small strain and the moduli are typically denoted as G max and M max, respectively. The equations used to calculate G max and M max are: G max = (ρ) V s 2 (1)

5 Terrell, Cox, Stokoe, Allen and Lewis 5 2 M max = (ρ) V p (2) where γ equals the total unit weight, g equals the acceleration of gravity, and ρ = γ/g. In addition to these moduli, Young s modulus at small strains, E max, and Poisson s ratio, ν calculated by: E H,max V = 2 p ( ρ )( 1+ ν)( 1 2ν) ( 1 ν ) 2 ( )( ) E,max = 2V ρ 1+ υ (4) H s 2 1 V p 1 2 Vs υ = (5) 2 V p 1 Vs (3) Evaluation of Stiffness in Different Directions Besides evaluations of small-strain moduli and Poisson s ratio, measurements of V s and V p along various directions permit one to characterize the anisotropic nature of the UAB layer (Stokoe et al, 1994). Many factors affect the velocities of compression and shear waves, including the degree of saturation, the state of effective stress, inherent anisotropy, and stressinduced anisotropy (Stokoe and Santamarina, 2000). Therefore, seismic waves can also be used to study the change in the base layer under different loading conditions. An attempt was made to evaluate the directional stiffnesses in each of the three test sections shown in Figure 1 in the quarry haul road. Five test sites were chosen along the haul road in which to conduct crosshole and downhole seismic testing in the UAB layer. Two sites were located in the conventional haul road section (denoted as Sites 1 and 2), two sites were located in the South Africa base section (denoted as Sites 3 and 4), and one site was located in the Georgia base section (denoted as Site 5). A second site was not placed in the Georgia base section due to the inability to find a second section of base that was thick enough for the installation of the sensors. Also, the first site had to be disregarded in the end due to unforeseen grading that deposited an unknown quantity of GAB over the sensors. To conduct the in-situ seismic tests in the UAB layer, embedded sensors were installed at each site prior to the placement of the HMA course during a site visit on 17 to 20 December Also, a section of PVC pipe that served as a source casing for crosshole testing was installed at each site. The test setup actually represented miniaturized versions of traditional crosshole and downhole tests. The miniaturized test setups are shown in Figure 3. With this arrangement testing was conducted in an attempt to determine the directional stiffnesses at each of the sites and how these stiffnesses changed under different external loads. Vertical and Radial Normal Stresses As part of the field tests, loading of the base layer was changed by adding three different external loads. The ILLI-PAVE program was run for each of the three external loading conditions. The UHVXOWLQJ YHUWLFDO DQG UDGLDO WRWDO QRUPDO VWUHVVHV v DQG r, respectively) at the middle of the UAB layer, in the center of the loaded area, for a typical site (including the geostatic normal

6 Terrell, Cox, Stokoe, Allen and Lewis 6 stresses for zero external load, assuming a K-value of 0.4) were calculated. The loaded area used in the analysis was determined by measuring the footprint of the wheel used to load the test site. Utilizing the vertical total normal stresses computed by ILLI-PAVE and the pore-water pressures determined using the soil-water characteristic curve (shown in Figure 4), the vertical HIIHFWLYH QRUPDO VWUHVV v ) for each loading condition was calculated. Due to unrealistically low values from ILLI-PAVE for the radial total normal stresses, the radial total normal stresses were recalculated using Eq. 7 σ = σ u (6) v v w in which a value of 0.3 was used for K. σ = σ K (7) r v * 7KHUDGLDOHIIHFWLYHQRUPDOVWUHVVHVZHUHFDOFXODWHGE\XVLQJWKH r-value calculated in Eq. 7 and then subtracting the pore-water pressure, as shown in Eq. 8 In this case, the water is assumed to have a K-value of The resulting vertical and radial effective normal stresses at the middle of the UAB layer for a typical site are presented in Table 3. σ = σ * K u (8) r v w FIELD STIFFNESS RESULTS As discussed above miniaturized versions of crosshole and downhole seismic tests were conducted in the UAB layer in an attempt to evaluate stiffnesses in the different directions. Unfortunately, the only readily identifiable seismic waves were the PH and SHH waves measured in the crosshole test as depicted in Figures 3a and 3b. (A PH wave is a P wave propagating in the horizontal direction and a SHH wave is a S wave with the directions of particle motion and propagation polarized in the horizontal plane.) The other measurements (SHV waves in the crosshole and PV waves in the downhole) were complicated by reflected and refracted waves and by problems associated the very short distance in the downhole test. Therefore, only results from the PH- and SHH-wave testing are presented below. Representative plots of PH- and SHH-wave velocities versus radial effective normal stresses determined for Sites 2, 4, and 5 are shown in Figure 5. Also shown in the figure are best-fit lines to the data and associated equations. The equations for the best-fit lines are of the form: n σ r V = C (9) 1atm where V = the stress wave velocity (y in the equations in the figures)

7 Terrell, Cox, Stokoe, Allen and Lewis 7 C = the stress wave velocity at 1 atm, r = the radial effective normal stress in atmospheres, and n = the exponent of the power function. The relationship has been shown over the past four decades to correctly represent the relationship between seismic wave velocity and stress state by numerous studies using unbound granular soils. It is important to see that the best-fit line for the measured data agrees closely with the results from those previous studies; that is, the log wave velocity versus log normal stress relationship is linear. Also, in this analysis, the radial normal effective stress was assumed to have the same value in all horizontal directions. A comparison of the PH-wave velocities for each of the four sites is shown in Figure 6. As can be seen in this figure, the slopes of the best-fit lines for the PH-wave velocities fall closely in line with each other, with the exponent varying between 0.26 and Also, in only one case (Site 3) was it necessary to interpret the PH results by placing by eye a line through the four data points that is slightly different than the best-fit line which fits the data and also agrees well with the slopes of the other PH-wave results. The logic used in this interpretation is that slight unknown variations in the measurements caused the measured data to differ slightly but the general trend was still captured in the measured data. The values of the PH-wave velocities show that the stiffnesses in the horizontal direction of all base layers are reasonably close, with the traditional section (Site 2) being the stiffest. The relatively high stiffness of Site 2 is thought to be due to the long period of compaction of the surge stone subbase before construction of the UAB layer (about four months of use of the road), essentially transforming this section into an inverted pavement. The South Africa and Georgia sections are nearly the same, with the two South Africa sites (Sites 3 and 4) bracketing the Georgia site (Site 5). This relative comparison is also supported by the SHH waves as discussed below. It is obvious by reviewing Figure 5 that the SHH-wave velocities exhibit much more scatter than the PH-wave velocities, requiring a bit more interpretation to find the appropriate line for the data. This variance in the SHH-wave velocity profiles is due to the significant amount of interpretation that was sometimes required by the individual analyzing the travel time records. This difficulty in interpretation was certainly increased by the short travel times, reflecting and refracting boundary surfaces, and distribution in the states of stress in the vertical direction throughout the UAB layer. A comparison of the SHH-wave velocities for each of the four sites is shown in Figure 7. As with the PH-wave measurements, the traditional section (Site 2) is shown to be stiffest, with the South African sites (Sites 3 and 4) bracketing the Georgia site (Site 5). Poisson s Ratio Values of Poisson s ratio are also shown in Figure 5. These values were calculated by using Eq. 5. In all cases, the interpreted relationships were used when they were present. The values for ν range from 0.11 to 0.17 and average 0.14.

8 Terrell, Cox, Stokoe, Allen and Lewis 8 Laboratory Dynamic Testing of the Base Material Resonant column testing was also conducted on a sample of GAB at a gravimetric water content of 6% and a total density of 2400 kg/m 3 (Menq, 2002). The variation in SHV-wave velocities with effective confining pressures is also presented in Figure 7. The linear relationship in log-log space based on Eq. 9 is shown in Figure 8 and shows the same behavior as found for other granular materials in previous studies (Menq, 2002). As can be seen in Figure 7, the SHV-wave velocities measured during resonant column testing compare well with the SHH-wave velocities measured in the field. The fact that the bestfit line to the resonant column data has nearly the same slope (n = 0.27) as the slope of the fitted lines associated with the field measurements (n = 0.26 to 0.27) indicates that the radial effective stresses for the field measurements were most likely well predicted under the field loading conditions. EVALUATION OF YOUNG S MODULI Using Eq. 3, the horizontal maximum Young s modulus, E H,max was calculated for each site and loading condition utilizing the Poisson s ratios shown in Figure 5. A comparison of E H,max for each of the four sites is presented in Figure 8. As with the PH-wave velocities, Site 2 is shown to be stiffer than the other three sites, with the South Africa sites (Sites 3 and 4) again bracketing the Georgia site (Site 5). It should also be noted that E H,max, equals the small-strain value of the resilient modulus, M r, because there is no permanent deformation in the small-strain range. The importance of pore-water pressure as related to the value of E H,max is illustrated in Figure 9, where the plot of E H,max versus radial effective normal stress for Site 5 is shown. The measured data is shown by the open diamonds. The best-fit line to the data is also shown. The values for E H,max for two other conditions are also shown. These conditions are (1) zero porewater pressure and no external load and (2) E H,max under saturated conditions with a fully-loaded truck inducing a positive pore-water pressure of 0.7 atm in the UAB layer. As can be seen in Figure 9, negative pore-water pressures have a significant impact on the stiffness of the UAB layer, with the value of Young s modulus under zero loading (geostatic) being reduced significantly (by 75%) with the pore-water pressure increasing from a 0.1 atm to zero. Equally disturbing is the loading of the pavement under saturated conditions, where the increase in pore-water pressure from 1 atm to 0.7 atm under a fully-loaded truck can result in the stiffness reducing by 75% which could result in pavement failure after repeated loading. ESTIMATION OF THE ANISOTROPIC STATE OF THE UAB LAYER IN TERMS OF E MAX An estimation of the vertical maximum Young s modulus, E V,max, from these measurements can be done by using the following: (1) the vertical effective normal stresses, (2) an assumption of structural anisotropy, and (3) the same value of n (exponent of the power best-fit line) as found for E H,max versus the radial effective normal stresses. The resulting equation can be expressed as: E = 0.95C (10) ( ) n V, max σ v

9 Terrell, Cox, Stokoe, Allen and Lewis 9 where 0.95C = the estimation of structural anisotropy (C is the coefficient of the best fit line from the plot of E H,max YHUVXV r ) (Belloti, et al, 1996) and n = the exponent of the power best-fit line for E H,max YHUVXVWKH r. The plot of the values for E V,max and E H,max versus the loading condition is shown in Figure 10. As can be seen in this figure, there is an increase in E V,max over E H,max with loading, which is due WR v EHLQJJUHDWHUWKDQ r with increased loading. At the condition with no external load, the values of E are essentially the same because the state of stress is dominated by the value of negative pore-water pressure that has the same value in all directions (K = 1.0 for u). CONCLUSIONS Measurements of stiffness of the UAB layer in compression and shear were successfully conducted. These measurements involved horizontally propagating P and S waves. It was found that the Georgia section was as good as the South Africa section, but that the traditional section was slightly stiffer. The higher stiffness of the traditional section was thought to result from the subbase being subjected to a prolonged period of compaction before construction of the UAB layer, in essence transforming the traditional section into an inverted pavement. It was found that negative pore-water pressures in the partially saturated granular base had a significant impact on the stiffness of the UAB layer in the unloaded and van-load conditions. It was important to take the pore-water pressure (u) into account if the stiffness was to be correctly predicted. An increase in stiffness with an increase in load was measured. Using the vertical total normal stresses computed from ILLI-PAVE, a value of 0.3 for K was found to be reasonable for this material in determining the associathg UDGLDO WRWDO QRUPDO VWUHVV r. The radial effective QRUPDOVWUHVV r ZDVFDOFXODWHGIURP r and u. The values of horizontal Young s modulus at small strains, maximum horizontal Young s modulus (E H,max ), were determined to range from a low of 180,000 kpa for the UAB layer in the unloaded condition to a high of 660,000 kpa for the fully loaded condition. A summary of the results in terms of E are given in Table 4. REFERENCE Belloti, R., Jamiolkowski, M., Lo Presti, D.C.F., and O Neill, D.A. (1996), Anisotropy of Small Strain Stiffness of Ticino Sand, Geotechnique, Vol. 46, No. 1, pp Department of Transportation, State of Georgia, Standard Specifications Construction of Roads and Bridges, Menq, F. Y. (2002), Dynamic Properties of Gravelly and Sandy Materials, Ph.D. dissertation, University of Texas at Austin.

10 Terrell, Cox, Stokoe, Allen and Lewis 10 South African National Roads Agency (1998), High Performance Crushed Stone Bases (G1) Used On South African Highways A General Overview. Stokoe, K.H., II, Hwang, S.K., Lee, J.N.K., and Andrus, R.D. (1994), Effects of Various Parameters on the Stiffness and Damping of Soils at Small to Medium Strains, International Symposium on Prefailure Deformation Characteristics of Geomaterials, Saporo, Japan, September Stokoe, K.H., II and Santamarina, J.C. (2000), Seismic-Wave-Based Testing in Geotechnical Engineering, International Conference on Geotechnical and Geological Engineering, GeoEng 2000, Melbourne, Australia, Nov

11 Terrell, Cox, Stokoe, Allen and Lewis 11 LIST OF TABLES Table 1 Table 2 Index Properties of GAB from the Lafarge Morgan County Quarry Measured Density and Moisture Contents for Unbound Aggregate Base Layer Table 3 Computed Effective Normal Stresses in the Middle of the UAB Layer at Site 5 Table 4 Summary of Results in Terms of E

12 Terrell, Cox, Stokoe, Allen and Lewis 12 LIST OF FIGURES Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Locations of Test Sites in Haul Road Compaction and Gradation Curves of the Granular Aggregate Base Material Miniaturized Crosshole and Downhole Arrangements that were used to Evaluate the UAB Layers Soil-Water Characteristic Curve for UAB Layer Typical Variations in Wave Velocities with Radial Stress State Measured in the UAB Layer Comparison of the Variation of PH-Wave Velocity with Stress State for all Four Sites Comparison of Field SHH-Wave Velocities Measured in the UAB Layer with Shear-Wave Velocities Measured in the Laboratory by Resonant Column Testing with a Reconstituted UAB Specimen Variation of E H,max with Stress State for all Four Sites Figure 9 Variation of E H,max with Stress State for Site 5 Figure 10 Estimated Relationship between E V,max and E H,max vs. the Loading Condition for Site 5

13 Terrell, Cox, Stokoe, Allen and Lewis 13 Table 1 Index Properties of GAB from the Lafarge Morgan County Quarry Liquid Limit Plastic Limit Non-Plastic Non-Plastic Optimum w 6.7% ASTM D 1557 (Modified Proctor) Max. Dry Density 2193 kg/m 3 Mean Grain Size, D 50 D mm 0.1 mm Uniformity Coefficient, C u = D 60 /D Unified Soil Classification System Designation GP-GM Table 2 Measured Density and Moisture Contents for Unbound Aggregate Base Layer Table 3 Computed Effective Normal Stresses in the Middle of the UAB Layer at Site 5 External Loading (kg) Vert. Total Normal Stress (atm) Rad. Total Normal Stress (atm) Pore-water Pressure (atm) Vert. Eff. Normal Stress (atm) Rad. Eff. Normal Stress (atm) * *note: calculated using a K-value of 0.4

14 Terrell, Cox, Stokoe, Allen and Lewis 14 Table 4 Summary of Results in Terms of E External Load v r E H,max (lbs) (atm) (atm) (kpa) ,000 1, ,000 5, ,000 10, ,000 External Load v r E H,max (lbs) (atm) (atm) (kpa) ,000 1, ,000 5, ,000 10, ,000 External Load v r E H,max (lbs) (atm) (atm) (kpa) ,000 1, ,000 5, ,000 8, ,000 External Load v r E H,max (lbs) (atm) (atm) (kpa) ,000 1, ,000 5, ,000 8, ,000 *note: Actual value was out of range. Reported number is an interpolation.

15 N Location of Test Site Quarry Georgia Base South African Base Conventional Haul Road Sta Sta Sta Figure 2 Locations of test sites in haul road Figure 1 Entrance Seven Islands Rd. Sta 0+00 Terrell, Cox, Stokoe, Allen and Lewis 15

16 Terrell, Cox, Stokoe, Allen and Lewis 16 ρ d,max = 2193 kg/m 3 w opt = 6.7 % a. Maximum dry density (Charles Smith, Georgia Department of Transportation) b. Gradation curve for the GAB from the Lafarge Morgan County Quarry (Carol Cook, Lafarge North America) Figure 2 Compaction and Gradation Curves of the Granular Aggregate Base Material

17 Terrell, Cox, Stokoe, Allen and Lewis 17 In-line, hor. (x) geophone (for PH) Orthogonal, vert. (z) geoph one (for SHV) Source hole Ray path Ray path Orthogonal, hor. (y) geophone (for SHH) a. Plan view of crosshole testing using 3-D embedded sensor HMA Source hole Wave ray path Orthogonal, vert. (z) geophone R1 or R cm UAB In-line, hor. (x) geophone Orthogonal, hor. (y) geophone Cemented Subbase b. Side view of crosshole testing using 3-D embedded sensor Source Wave ray path HMA UAB R3 R4 1-D sensors Cemented Subbase Figure 3 c. Plan view of downhole testing using two, 1-D sensorsã Miniaturized Crosshole and Downhole Arrangements that were used to Evaluate the UAB Layers

18 Terrell, Cox, Stokoe, Allen and Lewis 18 Figure 4 Soil-water characteristic curve for UAB layer

19 Terrell, Cox, Stokoe, Allen and Lewis y = 579.9x 0.26 Wave Velocity (m/s) y = x 0.25 PH Poisson s Ratio = 0.16 SHH Power (PH) Assumed K = 0.3 Power (SHH) Radial Effective Normal Stress (atm) a. Variation of PH- and SHH-wave Velocities with stress state for Site Wave Velocity (m/s) y = x 0.26 y = x 0.18 PH SHH Interpreted SHH: y = 290x.26 Interpreted SHH Poisson s Ratio = 0.12 Power (PH) Power (SHH) Assumed K = Radial Effective Normal Stress (atm) b. Variation of PH- and SHH-wave velocities with stress state for Site y = x 0.26 Figure 5 Wave Velocity (m/s) y = x 0.37 PH Interpreted SHH: y = 320x.27 SHH Interpreted SHH Poisson s Ratio = 0.11 Power (PH) Assumed K = 0.3 Power (SHH) Radial Effective Normal Stress (atm) c. Variation of PH- and SHH-wave velocities with stress state for Site 5 Typical Variations in Wave Velocities with Radial Stress State Measured in the UAB Layer

20 Terrell, Cox, Stokoe, Allen and Lewis y = 579.9x 0.26 PH-Wave Velocity (m/s) y = x 0.26 y = x 0.26 y = x 0.26 Site 2 Site 3 - Interpreted Site 4 Site 5 Power (Site 2) Power (Site 5) Power (Site 4) Power (Site 3 - Interpreted) Radial Effective Normal Stress (atm) Figure 6 Comparison of the Variation of PH-Wave Velocity with Stress State for all Four Sites 1000 SHH-Wave Velocity (m/s) y = x 0.27 Site 2 Site 3 - Interpreted Site 4 - Interpreted Site 5 - Interpreted Res. Col. Power (Site 2) Power (Res. Col.) Radial Effective Normal Stress (atm) Figure 7 Comparison of Field SHH-wave Velocities Measured in the UAB Layer with Shear- Wave Velocities Measured in the Laboratory by Resonant Column Testing with a Reconstituted UAB Specimen

21 y = x 0.53 y = x 0.5 Terrell, Cox, Stokoe, Allen and Lewis Young s Modulus (kpa) y = x 0.45 y = x Radial Effective Normal Stress (atm) Site 2 Site 3 Site 4 Site 5 Site 3 adusted Site 4 adjusted Power (Site 2) Power (Site 3 adusted) Power (Site 5) Power (Site 4 adjusted) Figure 8 Variation of E H,max with stress state for all four sites Figure 9 Variation of E H,max with stress state for Site 5

22 Terrell, Cox, Stokoe, Allen and Lewis 22 Figure 10 Estimated Relationship between E V,max and E H,max vs. the loading condition for Site 5