Evaluating Mechanical Properties of Earth Material During Intelligent Compaction (2020)

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.

81 Observations from Implementation of Specification Introduction To evaluate and validate the practicality of the developed forward models and backcalculation algorithms under field conditions, four additional test construction sites were visited for actual field implementation of IC. The evaluation and validation processes at these sites were aimed to help understand the variabilities associated with the construction and compaction phases under field conditions. The four sites were located in Minnesota, Ohio, and Texas. Detailed information about the test sites and the rollers used appears in Table 7-1, and the pavement structures of each test section are shown in Figure 7-1. Actual test sections showing representative line passes during proof-mapping and sublots are illustrated in Figures 7-2 through 7-5. Field Testing Program and Test Layout The following activities were undertaken at the construction sites: 1. Identification of Test Section. The research team, along with the contractor and DOT personnel, identified a test section that at least spanned 60 m (200 ft.) in length and had a minimum width of 7.5 m (24 ft.). 2. Setup of GPS. The research team set up a base station or connected to the DOT or contractor’s base station. 3. Setup of IC Roller. The research team set up the IC roller using the data acquisition system described in Chapter 4. The IC roller was checked for proper data collection, including vibration frequency, amplitude, and roller speed. 4. Installation of Instrumentation. Geophones were embedded into the soil at different depths and were connected to a second data acquisition system to monitor the propagation of roller vibration. Figure 7-1 shows the depths of the geophones for each test section. 5. Setup of Grid. At each test section, the team set up a grid of points to define sublots and fine-tune the models. A total of 44 points were marked and latitude/longitude coordinates were measured using a GPS rover at each of the marked points. Four rows of 11 points spanning the test section length were marked. Each row of points was placed under the line pass path to be traversed by the IC roller during mapping operations. Rows were spaced equidistantly based on the roller’s drum length along the test section width, as shown in the field test layout in Figure 7-6. 6. Proof Mapping (following Compaction). An IC roller was used for proof mapping the layer after the completion of compaction. Test section was mapped using one forward pass of the IC roller. 7. Field Testing. Field testing was carried out using LWD and NDG equipment on the compacted surface as soon as proof mapping was completed at each of the marked C H A P T E R 7

82 Evaluating Mechanical Properties of Earth Material During Intelligent Compaction Site Location Length Section Layer Roller * Drum Mass (Weight) 1 El Paso, TX 66 m (220 ft.) Lomaland Park Embankment Hamm H11 ix 5,890 kg (12,985 lb.)300 mm (12 in.) Subgrade 2 Burnsville, MN 75 m (250 ft.) I-35W, North- bound Ramp Subgrade Caterpillar CS74B 5,153 kg (11,360 lb.)300 mm (12 in.) Subbase 3 Springfield, OH 75 m (250 ft.) I-70 West- bound Lane Cement-Treated Subgrade (4% Cement) Caterpillar CS74B 5,153 kg (11,360 lb.) 150 mm (6 in.) Aggregate Base 4 El Paso, TX 66 m (200 ft.) US-62/180, Westbound Lane Embankment Volvo SD75 3,610 kg (8,025 lb.)300 mm (12 in.) Subgrade * These tests were conducted at existing worksites using the contractors’ rollers. One model (the Caterpillar SC74B) matched equipment that had been used in the developmental field tests, but three models (the Hamm H11 ix, Caterpillar CS74B, and Volvo SD75) differed from those used in the earlier tests. Table 7-1. Validation field test sites. Geophone Sites 1 & 4 Site 2 Site 3 -Treated 150 mm. (6 in.) 900 mm (36 in.) 300 mm (12 in.) 300 mm (12 in.) 150 mm (6 in.) 1200 mm (48 in.) 150 mm (6 in.) 150 mm (6 in.) 1200 mm (48 in.) 150 mm (6 in.)Clay Embankment Clay 300 mm (12 in.) Subbase Subgrade 150 mm (6 in.) Base Cement Subgrade Figure 7-1. Pavement structure of test sections and locations of embedded geophones. 44 points. Moisture samples also were collected at each of the spot tests for validation of the NDG data. 8. Collect Sample Material. Samples of the materials used to make up the pavement layers were collected for resilient modulus testing in the laboratory. Nonlinear material parameters were obtained for input into the backcalculation of modulus algorithms. The research team did not interfere with the construction operation, and the IC mapping and spot testing were conducted at a time that was least disruptive to the contractor (Figures 7-7 and 7-8). A Caterpillar CS74B smooth-drum IC roller, similar to the one used at the MnROAD facility, was used at Site 2 and Site 3, whereas a Hamm H 11ix smooth-drum roller was oper- ated at Site 1 and a Volvo SD75 at Site 4. All of the rollers were operated under low-amplitude and low-frequency conditions, as listed in Table 7-2. Data was collected using embedded geophones, starting at least 35 m before and ending 35 m after the geophone locations by synchronizing the roller-mounted GPS units with units located next to the embedded geophones.

Location of Test Section (a) (c) (b) (d) Figure 7-2. Overview of Lomaland Recreational Center construction in El Paso, Texas, showing (a) location of test site, (b) line passes during proof mapping of embankment, (c) test section with sublots, and (d) IC roller used at this location. (b) (c) (d) (a) Location of Test Section Figure 7-3. Overview of I-35W northbound reconstruction in Burnsville, Minnesota, showing (a) location of test site, (b) line passes during proof mapping of subgrade, (c) test section with sublots, and (d) roller during proof-mapping of subgrade.

84 Evaluating Mechanical Properties of Earth Material During Intelligent Compaction (a) Location of Test Section (b) (c) (d) Figure 7-4. Overview of I-70 westbound reconstruction in Springfield, Ohio, showing (a) location of test site, (b) line passes during proof mapping of subgrade, (c) test section with sublots, and (d) roller during proof mapping of subgrade.

Observations from Implementation of Specification 85 (b) (c) (a) Location of Test Section Figure 7-5. Overview of US-62/180 westbound construction of Montana Expressway in El Paso, Texas, showing (a) location of test site, (b) line passes during proof mapping of subgrade, and (c) test section with sublots. (a) (b) (c) (d) Figure 7-6. Installation process of embedded geophones: (a) GPS localization, (b) drilling with auger, (c) embedding ground sensors, and (d) wiring of geophones to data acquisition system.

86 Evaluating Mechanical Properties of Earth Material During Intelligent Compaction Vendor/Manufacturer Model Width (m) Drum Mass (kg) Centrifugal Force (kN) Frequency (Hz) Where Used HAMM H 11ix 2.1 5,890 136 30 Site 1 Caterpillar CS74B 2.1 5,153 166 23 Sites 2 and 3 Volvo SD75 1.7 3,610 121 30 Site 4 Table 7-2. Specifications of instrumented IC rollers. 1.8 m (6 ft.) spacing 7.5 m (24 ft.)Spot test A B C D 66 m (2 20 ft .) 11 sp ot te sts @ 6 m (2 0 ft. ) s pa ci ng Figure 7-7. Schematic of spot-test layout. (a) (b) Figure 7-8. Smooth-drum IC rollers used for mapping: (a) Hamm H 11ix, used at Site 1, and (b) Caterpillar CS74B, used at Site 2 and Site 3.

Observations from Implementation of Specification 87 Laboratory Testing Table 7-3 summarizes the index properties of all sampled materials, including the classifica- tion of each geomaterial, as per the Unified Soil Classification System (USCS). The optimum moisture contents and maximum dry unit weights, obtained as per standard Proctor tests (AASHTO T99) for the subgrades and as per modified Proctor tests (AASHTO T180) for base materials, also are reported in the table. Resilient modulus tests were performed in the laboratory as per AASHTO T307-03 to determine the resilient modulus and nonlinear parameters of geomaterials sampled at the test sections. Representative specimens prepared for resilient modulus testing are shown in Figure 7-9. The results of resilient modulus tests are summarized in Table 7-4 at the optimum moisture content (OMC). Validation of Approaches to Extract Modulus The outcomes of the field test implementation of the two approaches discussed in Chapter 6 are presented in the balance of this chapter. Extracting Modulus Using ANN Inverse Solvers (Approach 1) The inverse solvers make use of the input variables containing the nonlinear parameters obtained from the combination of both laboratory and field measurements, layer thickness of the top layer, and surface deflection from the accelerometer measurements. Using the geo- phone displacements obtained from the test sites, the inverse solvers were further fine-tuned to Site 1 1 2 2 3 4 Layer Embankment Subgrade Subgrade Subbase Base Embankment Subgrade Material Type Sand Sand/Clay Poorly Graded Sand Poorly Graded Sand Well-Graded Gravel Sand/Clay USCS Classification CL CL SP SP GW CL Gradation (%) Gravel 0.0 0.0 16.1 16.1 42.8 0.0 Coarse Sand 4.0 4.8 62.2 62.2 35.3 3.8 Fine Sand 46.0 45.2 17.4 17.4 12.0 48.2 Fines 50.0 50.0 2.4 2.4 1.2 48.0 Atterberg Limits (AASHTO T-89 and T-90) Liquid Limit (LL) 20 25 Non- Plastic Non- Plastic Non- Plastic 22 Plastic Limit (PL) 12 15 13 Plasticity Index (PI) 8 10 8 Moisture/Density (AASHTO T-99) Optimum Moisture Content (%) 13.8 17.1 7.1 7.1 -- 13.8 Maximum Dry Density 1,853 kg/m3 (115.7 pcf) 1,802 kg/m3 (112.5 pcf) 2,142 kg/m3 (133.7 pcf) 2,142 kg/m3 (133.7 pcf) -- 1,899 kg/m3 (118.5 pcf) Moisture/Density (AASHTO T-188) Optimum Moisture Content (%) -- -- -- -- 6.5 -- Maximum Dry Density (MDD) -- -- -- -- 2,289 kg/m3 (142.9 pcf) -- Table 7-3. Summary of index properties of construction site materials.

88 Evaluating Mechanical Properties of Earth Material During Intelligent Compaction improve the predictive accuracy following the process described in Chapter 6 and illustrated in Figure 6-6. The surface displacement data obtained for each sublot during the proof-mapping process was input into the appropriate ANN inverse solver. Also input were the nonlinear k ′2 and k ′3 parameters obtained from the resilient modulus tests, along with the nonlinear k ′1 parameter adjusted using the LWD modulus at representative sublots. The comparison of the measured and extracted moduli for each sublot obtained from the ANN inverse solver for the single-layer systems of the three test sites is shown in Figure 7-10. The inverse solver can predict the modulus of the subgrade with a variability of up to 25%. This level of variability was deemed acceptable given the variability of the earthwork in each sublot. Similarly, the top-layer moduli for the two-layer systems were extracted using the appropriate ANN inverse solver. In addition to the nonlinear parameters obtained from the laboratory resilient modulus tests, the surface displacements measured on top of both the single-layer and (b) (c) (a) Figure 7-9. Preparation of soil for testing: (a) soil mixture based on the sieving analysis, (b) prepared samples for resilient modulus test, and (c) MTS® Load Unit System. Sites Layer Two Thickness h (mm) Nonlinear Parameters for Layer One * Nonlinear Parameters for Layer Two * Modulus * (MPa) Surface Displacement on Top of ELWD MR Layer 1 d1 (mm) Layer 2 d2 (mm) k′1-back k′2 k′3 k′1-back k′2 k′3 Layer 1 Layer 2 Layer 1 Layer 2 1 300 217 1.57 -2.04 313 1.24 -3.00 39 40 40 39 1.07 1.11 2 300 598 1.69 -2.16 267 1.69 -2.16 130 58 115 51 1.26 1.52 3 150 1581 0.61 -0.05 329 0.57 -0.05 231 91 230 103 1.08 1.05 4 300 213 1.76 -2.6 214 1.76 -2.6 40 29 37 37 1.07 1.11 *ESUBG is the subgrade modulus, in this case using the LWD modulus determined on top of subgrade. MRSUBG is the resilient modulus of subgrade material as obtained from the resilient modulus test as per AASHTO T-307, and k′1-back is the backcalculated k′1 value using the LWD modulus ELWD of the corresponding layer in Equation 3-2 following the model from Ooi et al. (2004). Table 7-4. Geomaterial properties of test sections used for validation of inverse models.

Observations from Implementation of Specification 89 two-layer systems of each sublot were used as inputs. Figure 7-11 compares the extracted moduli of the top layer with the corresponding backcalculated LWD moduli. Comparing Figure 7-10 with Figure 7-11, the two-layer inverse model extracts the moduli with less accuracy than the single-layer inverse model but still falls within the ±25% uncertainty bounds (a variability of up to 25%). Variability of Extracted Modulus Due to Number of Spot Tests Given that an extensive testing program is impractical in day-to-day operations, the research team also assessed the minimum number of spot tests with LWD. The sublots exhibiting considerable nonuniformity were first excluded because they introduced significant uncertainty in the process. Based on a substantial field database, Tirado et al. (2019) defined sublots with COV of CMVs 25% and less as “uniform.” Figure 7-12 compares the extracted moduli and the LWD moduli averaged per station for the subgrade layers of Site 1 through Site 4. The error bars represent one standard deviation of moduli at each station. The extracted moduli shown in the figures exhibit little variability y = 1.08x R² = 0.98 SEE = 14.0 MPa 0 100 200 300 400 0 100 200 300 400 LWD Modulus (MPa) Site 1 Site 2 Site 3 Site 4 Line of Equality +/- 25% Uncertainty Bounds E xt ra ct ed M od ul us (M Pa ) Figure 7-10. Comparison of LWD and extracted moduli for single-layer systems for Sites 1–4. y = 0.94x R² = 0.80 SEE = 9.0 MPa 0 30 60 90 120 150 0 30 60 90 120 150 LWD Modulus (MPa) Site 1 Site 2 Site 3 Site 4 Line of Equality +/- 25 Uncertainty Bounds E xt ra ct ed M od ul us (M Pa ) Figure 7-11. Comparison of LWD and extracted moduli for two-layer systems for Sites 1–4.

90 Evaluating Mechanical Properties of Earth Material During Intelligent Compaction 20 40 60 80 STA +000 STA +020 STA +040 STA +080 STA +100 STA +120 STA +140 STA +160 STA +180 M od ul us (M Pa ) LWD Modulus Extracted Modulus (a) Site 1 0 100 200 300 STA +025 STA +050 STA +100 STA +125 STA +150 STA +175 STA +200 STA +225 STA +250 M od ul us (M Pa ) LWD Modulus Extracted Modulus 0 150 300 450 STA +000 STA +025 STA +050 STA +075 STA +100 STA +125 STA +150 STA +175 STA +200 STA +225 STA +250 M od ul us (M Pa ) LWD Modulus Extracted Modulus 0 20 40 60 80 100 STA +000 STA +025 STA +050 STA +075 STA +100 STA +125 STA +150 STA +175 STA +200 STA +225 M od ul us (M Pa ) LWD Modulus Extracted Modulus (b) Site 2 (c) Site 3 (d) Site 4 Figure 7-12. Relationship between LWD measured and extracted moduli per station for single-layer systems in Sites 1–4.

Observations from Implementation of Specification 91 because each point was calculated using an average of up to 50 surface displacement values measured in each sublot, and also because the single k ′1 value has been adjusted and assigned to represent a station. The LWD and extracted moduli exhibit similar trends along the length of the lot for all four sites. Similarly, for two-layer systems, the backcalculated LWD moduli and extracted top layer moduli using the inverse solver exhibited similar trends along the lots, as shown in Figure 7-13. Figures 7-14 and 7-15 compare the color-coded maps of the LWD moduli and the moduli extracted from the inverse solvers for single-layer and two-layer systems at Site 1, respectively, after implementing this procedure. The LWD modulus and backcalculated modulus maps show some resemblance. Blank sublots in the extracted modulus map correspond to sublots with considerable variability in their ICMVs (COVs greater than 35%), and thus were removed from the verification process. Figure 7-16 shows the percent difference between the extracted and the LWD moduli per site. A mean percent error (MPE) of 10% exists when using 11 spot-test measurements. Maps of the estimated moduli for the single-layer and two-layer systems, calibrated with five LWD tests, are also shown in Figures 7-14 and 7-15, respectively. Reducing the number of LWD tests to five in sublots with COV of CMVs of less than 25% results in an increase in MPE to 23%. Thus, using five spot tests still provides extracted moduli within ±25% accuracy, which is appropriate given the inherent variability of soils. Using Drum Force to Extract Modulus (Approach 2) The second approach for the extraction of modulus, described in Chapter 6, employs an inverse model that uses the roller’s drum force and LWD spot measurements at selected locations with COVs of CMV of 25% and less. Figure 7-17 shows the COVs of CMV for the drum force and LWD modulus at every sublot of the embankment of Site 1. Sublots with COVs of 25% and less are shown in green. Selected sublots for conducting LWD testing are highlighted in the maps. The same process was applied to the 300 mm (12 in.) layer of clay material laid on top of the embankment for identifying potential sublots for conducting LWD. Values within these sublots were used to develop a localized calibration factor between the drum force and LWD modulus (see Figure 7-18). Figure 7-19 compares the estimated moduli from the drum force with the LWD-measured moduli on top of the embankment and the additional 300 mm (12 in.) layer in areas with COVs of CMV of 25% and less. The extracted moduli are in general agreement with the LWD moduli for both layers as judged by the number of cases falling within the ±25% uncertainty bounds. The uncertainty in the estimation of modulus using the drum inertial force can be closely related to the level of uniformity achieved for the compacted section. In other words, the more uniform the section is, the more certain the modulus estimation will be. Validation of Approach 2 Using Expanded Database To further validate the proposed approach, a database assembled from an extensive IC testing program that was acquired as part of a Texas Department of Transportation research project (FHWA/TX-19/0-6903-1) was used. Table 7-5 summarizes the general information about the test sites. The data consisted of CMV, surface deflection, and drum force measure- ments. LWD testing was conducted as part of the field evaluation of the single-layer and two-layer systems at a 7.5 m (25 ft.) spacing along the roller line passes at the center of each sublot.

92 Evaluating Mechanical Properties of Earth Material During Intelligent Compaction 0 20 40 60 80 STA +020 STA +040 STA +060 STA +080 STA +100 STA +120 STA +140 STA +160 STA +180 STA +200 M od ul us (M Pa ) LWD Modulus Extracted Modulus (a) Site 1 0 20 40 60 80 100 120 STA +000 STA +025 STA +050 STA +075 STA +100 STA +125 STA +150 STA +175 STA +200 STA +225 STA +250 M od ul us (M Pa ) LWD Modulus Extracted Modulus (b) Site 2 0 40 80 120 160 200 STA +000 STA +025 STA +050 STA +075 STA +100 STA +125 STA +150 STA +175 STA +200 STA +225 STA +250 M od ul us (M Pa ) LWD Modulus Extracted Modulus 0 20 40 60 80 STA +025 STA +050 STA +075 STA +100 STA +125 STA +150 STA +175 STA +200 STA +225 STA +250 M od ul us (M Pa ) LWD Modulus Extracted Modulus (c) Site 3 (d) Site 4 Figure 7-13. Relationship between averaged rectangular buffered areas measured and extracted moduli for two-layer systems in Sites 1–4.

Observations from Implementation of Specification 93 Units in MPa (c) (a) (b) Figure 7-14. Comparison of mapping of (a) LWD modulus and extracted modulus for a single layer in Site 1, calculated using (b) 11 spot tests and (c) 5 spot tests. Units in MPa (c) (a) (b) Figure 7-15. Comparison of mapping of (a) LWD modulus and extracted modulus for two-layer system in Site 1, calculated using (b) 11 spot tests and (c) 5 spot tests.

94 Evaluating Mechanical Properties of Earth Material During Intelligent Compaction 0 20 40 60 80 100 Pe rc en t E rr or (% ) Eleven Spots Five Spots Figure 7-16. Comparison of variability of inverse solver extracted modulus with respect to LWD modulus for all sites and for single-layer and two-layer systems. (a) COV of CMV (%) (b) Drum Force (kN) (c) LWD Modulus (MPa) Selected sublots for LWD testing Figure 7-17. Mapping after compaction of embankment showing (a) COV of CMV, (b) drum force, and (c) LWD modulus.

Observations from Implementation of Specification 95 y = 0.19x R² = 0.43 25 30 35 40 45 50 55 60 160 180 200 220 240 Drum Force (kN) (a) Embankment of Sandy Material L W D M od ul us (M Pa ) y = 0.11x R² = 0.47 0 20 40 60 80 200 300 400 500 Drum Force (kN) (b) 300 mm (12 in.) Clay Material L W D M od ul us (M Pa ) Figure 7-18. Development of local transfer functions at Site 1 for (a) embankment of sandy material and (b) 300 mm (12 in.) clayey material. y = 0.95x 0 20 40 60 80 0 10 20 30 40 50 60 70 80 R et ri ev ed M od ul us (M Pa ) LWD Modulus (MPa) Embankment of Sandy Material 300mm (12 in.) Silty-Sand Material Line of Equality +/- 25% Uncertainty Bounds Figure 7-19. Relationship between retrieved modulus and LWD modulus of sublots with COV of CMV less than or equal to 25% for embankment of sandy material and 300 mm (12 in.) silty-sand material. Site Length Layer* Roller Drum Mass (Weight) 5 75 m (250 ft.) 150 mm (6 in.) Cement-Treated Subgrade (CTS) Bomag BW 211 D-40 5,750 kg (12,677 lb.)150 mm (6 in.) Cement-Treated Base (CTB) on Top of CTS 6 75 m (250 ft.) 300 mm (12 in.) Lime-Treated Subgrade (LTS) CAT CS87B 12,960 kg (28,572 lb.) 7 75 m (250 ft.) 450 mm (18 in.) Reclaimed Asphalt (RAP) and Base Material as Subgrade Ingersoll Rand DD-110 6,075 kg (13,393 lb.) 150 mm (6 in.) Flexible Base on Top of RAP/Base Material Subgrade 8 75 m (250 ft.) 300 mm (12 in.) Cement-Treated Subgrade (CTS) CASE SV 212 7,354 kg (16,213 lb.)300 mm (12 in.) Cement-Treated Base on Top of CTS Table 7-5. Validation field test sites.

96 Evaluating Mechanical Properties of Earth Material During Intelligent Compaction Figure 7-20 compares the retrieved layer moduli with their corresponding LWD moduli after implementing the proposed approach for subgrade and base materials at those test sections. Data points represent average of extracted layer moduli of the sublots with the COV of CMV less than or equal to 25% and LWD moduli as obtained for each of the evaluated test sites. The error bars show the bound between the first and third quartiles of the measurements per test site. Some extracted moduli shown in the figures exhibit little variability because each point was calculated using an average of up to 50 drum force values measured in each sublot and because these sites were more uniform in terms of ICMVs as compared to LWD moduli. The layer modulus can be predicted using the drum force within a level of uncertainty of less than 30%. These analyses show that the modulus retrieved using the dynamic drum force can be more representative and reliable when compaction uniformity is achieved. The compaction uniformity plays a key role on retrieving the modulus of geomaterials with certainty. In other words, when the uniformity in compaction is not achieved, an LWD spot test cannot appropriately represent the quality of compaction for a sublot with approximate size of 45 m2 (500 ft.2). Two approaches to using inverse solvers with differing numbers of input variables were evaluated. The first approach used an inverse solver that was developed using an extensive database that had been assembled from responses of a wide range of pavement properties and layer thicknesses using a calibrated FE model. The use of this approach required a laboratory effort to determine the needed input variables. The second approach needed fewer inputs than the first approach because it made use of the dynamic drum force. Given that compaction uniformity affects the extraction of modulus, it was found that both approaches benefit from the use of sublots exhibiting uniform compaction. This observation points toward the useful- ness of developing a local calibration to reduce the variability of the model output. Determining Target Field Values for Quality Acceptance For a robust acceptance process, the target field values should be set. The target value can be the stiffness for the vibratory IC rollers. Most of the deflection-based devices measure the stiffness of the pavement system, and the reported stiffness is based on an elastic half space Site 5: Subgrade Site 5: Base Site 6: Subgrade Site 6: Base Site 7: Subgrade Site 7: Base Site 8: Subgrade Site 1: One Layer Site 1: Two Layer Site 2: One Layer Site 2: Two Layer Site 3: Subgrade Site 3: Base Site 4: One Layer Site 4: Two Layer y = 1.00x R² = 0.99 0 50 100 150 200 250 300 350 400 450 0 50 100 150 200 250 300 350 400 450 LWD Modulus (MPa) Line of Equality +/- 30% Uncertainty Bounds R et ri ev ed M od ul us (M Pa ) Figure 7-20. Relationship between averaged retrieved modulus and LWD modulus.

Observations from Implementation of Specification 97 Boussinesq theory. This limitation is particularly critical for a multi-layered system being tested with deflection. For this study, however, an inverse model was used to determine the target stiffness (as described in Chapter 3). Table 7-6 provides the target stiffness values that were determined from the inverse model constructed for single-layer and two-layer systems for Sites 1 through 4. The target stiffness was set using the operating features of the roller (see Table 7-2), the layer thicknesses (see Table 7-1), and the nonlinear parameters obtained from the resilient modulus test. A fair and equitable acceptance process requires appropriate tolerances based on the uncer- tainties in establishing the target modulus and the measuring device (in this case, the IC roller). In this study, 75% of the target stiffness served as the boundary for marginal acceptance. Figure 7-21 compares the averaged measured stiffness per station with the target stiffness obtained from the inverse solver for single-layer systems for all four sites. The error bars show the bounds between the first and third quantiles of the stiffness measurements per station. Site 3 was the only site for which the testing of the subgrade yielded stiffness measurements that marginally passed. Likewise, Figure 7-22 shows the comparison of the measured stiffness with the corresponding target stiffness obtained from the inverse solver for the two-layered systems for Site 1 through Site 4. For the two-layer systems, the measured stiffness marginally passed for all the visited sites except Site 4, where roller measurements indicated that this site did not meet the design stiffness. With all of these steps taken into account, the vibratory IC rollers can be considered as rigorous stiffness-based devices for quality acceptance of the compacted geomaterials to replace density-measured approaches, as stiffness parameters are more relevant to and employed in pavement design even though the use of IC for quality control and acceptance might be challenging. Whenever high variability exists in the underlying ground strata, the methodology developed in this study can overcome the limitations mentioned in the literature and is capable of more realistically representing the quality and uniformity of compaction in a continuous manner. Site Target Stiffness, ks-Target (MN/m) Single-Layer Two-Layer 1 38 30 2 61 47 3 198 171 4 37 37 Table 7-6. Target stiffness values.

98 Evaluating Mechanical Properties of Earth Material During Intelligent Compaction 0 50 100 150 200 STA +000 STA +025 STA +050 STA +075 STA +100 STA +125 STA +150 STA +175 STA +200 STA +225 STA +250 St iff ne ss (M N /m ) Target Stiffness 75% Marginal Acceptance 0 25 50 75 100 St iff ne ss (M N /m ) Target Stiffness 75% Marginal Acceptance 0 100 200 300 400 St iff ne ss (M N /m ) Target Stiffness 75% Marginal Acceptance 0 20 40 60 80 100 St iff ne ss (M N /m ) Target Stiffness 75% Marginal Acceptance (a) Site 1 (b) Site 2 (c) Site 3 (d) Site 4 STA +000 STA +025 STA +050 STA +075 STA +100 STA +125 STA +150 STA +175 STA +200 STA +225 STA +250 STA +000 STA +025 STA +050 STA +075 STA +100 STA +125 STA +150 STA +175 STA +200 STA +225 STA +250 STA +000 STA +025 STA +050 STA +075 STA +100 STA +125 STA +150 STA +175 STA +200 STA +225 STA +250 Figure 7-21. Averaged stiffness per station, with corresponding target stiffness for single-layer systems at Sites 1–4.

Observations from Implementation of Specification 99 0 50 100 150 St iff ne ss (M N /m ) Target Stiffness 75% Marginal Acceptance 0 25 50 75 100 125 150 St iff ne ss (M N /m ) Target Stiffness 75% Marginal Acceptance 0 100 200 300 400 St iff ne ss (M N /m ) Target Stiffness 75% Marginal Acceptance 0 25 50 75 100 St iff ne ss (M N /m ) Target Stiffness 75% Marginal Acceptance (a) Site 1 (b) Site 2 (c) Site 3 (d) Site 4 STA +000 STA +025 STA +050 STA +075 STA +100 STA +125 STA +150 STA +175 STA +200 STA +225 STA +250 STA +000 STA +025 STA +050 STA +075 STA +100 STA +125 STA +150 STA +175 STA +200 STA +225 STA +250 STA +000 STA +025 STA +050 STA +075 STA +100 STA +125 STA +150 STA +175 STA +200 STA +225 STA +250 STA +000 STA +025 STA +050 STA +075 STA +100 STA +125 STA +150 STA +175 STA +200 STA +225 STA +250 Figure 7-22. Averaged stiffness per station, with corresponding target stiffness for two-layered systems at Sites 1–4.