Supporting Materials for NCHRP Report 626 (2009)

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.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report         PART II—SUMMARY OF FINDINGS    37

Page intentionally left blank.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report CHAPTER 2 MATERIAL PROPERTIES FOR CONTROL AND ACCEPTANCE 2.1 Quality Characteristics Used in Previous and Current QA Programs The specific mixture and/or mat properties used to accept HMA construction, i.e. the AQCs vary from agency to agency. Table 13 lists those material or layer properties that were used by many state agencies prior to adoption of the Performance Graded (PG) asphalt specification and Superpave volumetric mixture design method. The material properties listed in Table 13 do not include those properties that are typically specified by an agency for material source approval and mixture design. Density, air voids, and fluids content are important volumetric properties related to flexible pavement and HMA overlay performance, which is why the majority of state and federal agencies use these properties as quality characteristics for accepting the pavement layers and materials (refer to Table 13). Many state agencies have incorporated or are in the process of incorporating smoothness and void in mineral aggregate (VMA) into their acceptance plan and have removed stability and penetration or viscosity after adopting the PG asphalt specification and Superpave mixture design method. Density, air voids, fluids content, and gradation are still included in most agencies’ acceptance plans, while engineering properties are excluded from nearly all QA programs in use by state and federal agencies to-date. With the exception of smoothness, all of the properties are determined at a point⎯either on the roadway or at a specific point in time during mixture production. 2.2 Control and Acceptance Procedures Of the many process control procedures that can be used in highway construction, process control charts, particularly statistical control charts, are most commonly used by contractors and material producers for verifying that their process is under control. Although there are different approaches that can be taken in implementing NDT technologies to verify that the process is in control, statistical control charts were used within this project. As a result, the NDT test methods must produce results that can be adapted to existing AASHTO procedures in pavement construction. The ASTM Manual on Presentation of Data and Control Chart Analysis was used for preparing practical procedures that contractors can use in deciding whether their process is in control (ASTM, 1992). 39

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report Table 13. Material Properties Used by State and Federal Agencies for Accepting Flexible Pavement Construction (Von Quintus et al., 1998; Darter et al., 1997) Pavement Layer Material-Layer Property Number of Agencies Mat Density/Air Voids 26 Asphalt Binder Content 18 Joint Density 14 Gradation 10 Smoothness; IRI, Profile Index 10 Mat Thickness 8 Mix Density – Laboratory Compacted 5 Voids in Mineral Aggregate 4 Stability – Laboratory Compacted 10 Asphalt Binder Properties; Viscosity or Penetration 3 Voids Filled with Asphalt 0 Strength – Indirect Tensile 0 HMA Layers; Dense-Graded Mixes Modulus; Dynamic, Resilient, Creep 0 Dry Density 26 Gradation 12 Minus 200 Material 8 Strength; Dynamic Cone Penetrometer, CBR, R-value 6 Atterberg Limits 0 Moisture Content 0 Unbound Layers; Aggregate Base, Embankment Resilient Modulus 0 NOTE: The material properties that are in italics may not represent the in place materials, if measured on laboratory compacted specimens, asphalt recovered from the storage tank at the plant, or have been replaced by other more recent tests. Similarly, there are different acceptance procedures that are used in judging whether the pavement material meets the required specifications. Two methods that have been adopted by most agencies are Percent Within Limits (PWL) and Average Absolute Deviation (AAD). PWL is the procedure used by over 75 percent of the agencies that have adopted statistical- based acceptance specifications. As a result, American Association of State and Highway Transportation Officials (AASHTO) R9 entitled Acceptance Sampling Plans for Highway Construction was used for preparing practical but effective procedures that agencies can use in deciding whether the product meets their specifications (AASHTO, 2003). In summary, statistical control charts is the recommended method for determining whether the construction is in-control or out-of-control, and PWL will be the method used for judging construction quality or acceptability. 2.3 Sampling Plans for Measuring Quality Characteristics The definition of a lot varies from agency to agency, but generally is a day’s production or an amount of material. The sampling frequency for QA tests within a lot is determined assuming that the material property being measured has a normal distribution. The assumption of 40

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report normality has been found to be adequate for most quality characteristics in use to-date, assuming proper construction procedures are followed. Localized deficiencies, however, can and do occur that are difficult to detect using sampling frequencies defined using the assumption of normality. Most agencies use 4 to 10 tests per lot for acceptance. The higher frequencies are used for density determinations. These sampling frequencies generally are adequate for estimating the mean and standard deviation of material properties of the population, except when local defects or anomalies occur at random or on a consistent basis. Most acceptance and control sampling and testing frequencies are insufficient to identify these conditions without adequate inspection. It is expected that inspection will be just as important when using NDT methods as for current QA plans. 2.4 Integration of Design and Acceptance Procedures Historically, most of the pavement design procedures that have been used by state highway agencies in the U.S. require the use of structural layer coefficients. These layer coefficients can be estimated from resilient modulus using charts included in the 1993 AASHTO Guide for the Design of Pavement Structures (AASHTO, 1993). Currently, there is a move to use the MEPDG structural design procedures that are based on M-E methods (ARA, 2004, 2006). An Interim Manual of Practice for the MEPDG (ARA, 2007) has been prepared and as of data, the MEPDG has been adopted as the AASHTO Interim Design Procedure.2 M-E procedures use fundamental pavement material properties such as modulus and strength. Few agencies, however, actually measure resilient modulus or other modulus values of HMA and unbound materials (Darter et al., 1997). Some agencies do not have confidence in these values, while others simply do not have the laboratory equipment. Using the same mixture properties for accepting the pavement layer that were used for structural and mixture design allows the agency to more precisely estimate the impact that deficient materials and pavement layers have on performance. The focus and approach of NCHRP Project 09-22 is to develop PRS and associated software that is directly tied to the MEPDG by using quality characteristics that are input to that design procedure (Killingsworth, 2003). The material tests that are needed for structural and mixture design using the newer procedures are listed in Table 14. 2.5 Issues with Existing QA Tests for Measuring Quality Characteristics To improve procedures for measuring the quality of flexible pavement construction and to detect deficiencies in the final product, one must identify some of the limitations, problems, and issues with the current procedures that are used to control and accept construction. This section summarizes the more common issues with current QA programs. 2 The Manual of Practice for the MEPDG was prepared under NCHRP Project 1-40B and was the official AASHTO ballot for voting on the procedure or MEPDG software. 41

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report Most of the QA tests in use by state and federal agencies for flexible pavement construction use location or time specific tests for judging the quality of construction. As noted above, the frequencies of the samples and tests are determined assuming that the layer properties have a normal distribution. This assumption is incorrect in some cases. Anomalies can and do occur during construction. Cold spots and segregation in the HMA mat can be difficult to detect under typical QA programs. In general, the sampling frequencies used by many agencies are inadequate to detect localized areas with defects. Table 14. Summary of Material and Layer Properties Used for Design and Acceptance of Flexible Pavements and HMA Overlays Property Needed for: Pavement Layer Material-Layer Property Structural Design Mixture Design Acceptance Density – Air Voids at Construction Yes Yes Yes Voids in Mineral Aggregate Yes Yes Yes Effective Asphalt Binder Content Yes Yes Yes Voids Filled with Asphalt Yes Gradation Yes Yes Yes Asphalt Binder Properties Yes Yes Indirect Tensile Strength and Creep Compliance Yes Yes Dynamic Modulus Yes Yes Flow Time or Flow Number Yes HMA Layers; Dense-Graded Mixtures Smoothness, Initial Yes Yes Density Yes Yes Yes Moisture Content Yes Yes Gradation Yes Yes Yes Minus 200 Material Yes Yes Yes Resilient Modulus Yes Yes Unbound Layers; Dense Graded Granular Base, Embankment Soils Strength, Dynamic Cone Penetrometer Yes Yes Another major issue related to acceptance tests is the time required for obtaining the acceptance test result. Contractors can place a large amount of mixture and materials with today’s production and construction equipment. The longer it takes to obtain the results for acceptance criteria, the greater the amount of material and dollars that can be in dispute or penalty. Minimizing the time to obtain accurate test results for acceptance should reduce the magnitude and number of dispute claims. The following summarizes the three major issues with the current QA procedures and tests that were considered in selecting NDT methods for use within this project. 1. Inferior mixtures are defined as those that do not meet the assumptions used in the structural and mixture design process. Density, air voids, and other volumetric properties by themselves do not always detect inferior mixtures or materials. To improve on the process of detecting inferior materials, QA tests need to measure the 42

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report same property used in structural and mixture design or at least measure a parameter that is highly correlated to those properties used in design. As previously noted, one of the objectives of NCHRP Project 09-22 is to develop HMA PRS using quality characteristics that are tied to the structural and HMA mixture design procedures. Thus, QA needs to have a similar focus and approach—use fundamental parameters needed to predict performance to assist in developing payment penalties and incentives.3 2. The sampling and testing frequency of most location specific tests are inadequate to detect and locate deficiencies in HMA mixtures and other materials. These deficiencies can include random and longitudinal segregation of HMA mixtures and segregated and wet areas of unbound materials. To improve on the detection process, higher sampling and testing frequencies or larger sample sizes within the population are needed, in combination with proper inspection. Another major issue facing state and federal agencies is the reduction in personnel resulting in decreased inspection of construction activities. 3. A large amount of material and mixture can be placed by the contractor within a day’s production. The time for obtaining and interpreting the test results for acceptance should be as short as possible. In addition, the tests for accepting the final product are being performed by the contractor for an increasing number of agencies. In other words, using a contractor’s QC test results for acceptance is becoming more popular. Thus, the acceptance tests should be applicable for use by the contractor during day-to-day production and the results from the acceptance tests need to be available on a real-time basis (within a day) to minimize the amount of material or funds in dispute. 2.6 Summary of Material Properties Used for QA The approach taken for this project was to use fundamental properties that are needed for both mixture and structural design for both control and acceptance of flexible pavements and HMA overlays (see table 15). The NDT technologies were evaluated for their ability to estimate these properties accurately. Table 15 lists those properties that were considered for use in the field evaluation study, while Figure 2 illustrates this systems approach. The properties were grouped into three areas—volumetric, structural, and functional. The material properties included traditional QA test methods, fundamental engineering properties needed to predict performance using the MEPDG, and simple performance tests to assist in volumetric mixture design. Structural and mixture design properties are needed to ensure that the NDT technology can quantify construction quality accurately, and that the assumptions used in structural and mixture design have been met. The structural properties that are critical for predicting the performance of flexible pavements and HMA overlays are modulus and thickness. 3 Although density and air voids are not commonly used in most structural design procedures, both are key material properties for HMA as well as unbound aggregates and soils, as they have a significant effect on the material’s strength and modulus. 43

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report Field permeability tests have become important since agencies started using coarse-graded Superpave mixtures at the surface and can be considered a nondestructive test. However, none of the agencies contacted or the QA programs reviewed within this project require the use of permeability tests for control or acceptance. The Florida Department of Transportation (DOT) is just about the only agency that has seriously considered using field permeability tests for acceptance. In addition, the field permeability tests take time to perform and require that the roadway be closed to traffic. Although a field permeability test measures an important property of compacted HMA mixtures, the test was not considered in the field evaluation portion of this project because of the time needed to perform the test; furthermore, the results are not needed for both structural and mixture design procedures. Table 15. Summary of Layer Properties Typically Used in Traditional QA Programs and Material/Pavement Properties Estimated from NDT Methods Traditional QC/QA Test Methods NDT Test Methods Type of Property HMA Unbound Materials HMA Unbound Materials Density – Mat & joint densities Density Density – Mat & joint densities Density Asphalt Binder Content Moisture Content Asphalt Binder Content Moisture Content Gradation Gradation Gradation: Segregation --- Volumetric VMA Minus 200 Material Air Voids, VMA Percent Saturation Thickness Thickness Elastic (Dynamic & Resilient) Modulus Resilient Modulus Indirect Tensile Creep Compliance Shear Strength Structural Thickness Thickness Adequate bond or adhesion between HMA layers NA IRI & Profile Noise Functional Profile: IRI or Profile Index NA Friction NA VMA – Voids in Mineral Aggregate IRI – International Roughness Index 44

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 45 Select Strategy: Trial cross-sections for pavement structural design or rehabilitation. Complete structural design using mixture specifications; Select structural properties to minimize distress. Material selection & certification; Source approval. Material Specifications: Aggregate, Asphalt binder, Additives, etc. Volumetric Mixture Design; Superpave Gyratory Compactor Feedback from monitoring pavement performance, NCHRP Project 9-30 Confirmation – Adjustment of volumetric mixture design. Prepare specimens over range of volumetric conditions. Perform NDT QA tests on laboratory test specimens. Mix Design Tests, NCHRP Project 9-33 QA Tests; NCHRP Project 10-65 Select final mixture design & measure E* master curve. Field Verification of Mixture Design Select/Establish QA criteria for measuring quality; determine seismic design modulus. Agency Acceptance Plan & Specifications Contractor Quality Control Plan Pavement Management: Monitoring projects – Functional, Structural, Volumetric Properties & Surface Distress. Confirmation of structural design assumptions & performance expectations Perform torture test(s) (APA, Hamburg, etc.) or Dynamic modulus, fracture, permanent deformation tests, etc.; NCHRP Project 9-19. Site Features & Inputs: • Climate • Traffic • Foundation Structural Design, NCHRP Project 1- 37A Calibrate NDT QA tests; control strips, NCHRP Project 10-65. PRS, NCHRP Project 9- 22 Figure 2. Example Flow Chart for the Systems Approach for Specifying, Designing, and Placing Quality HMA Mixtures

Page intentionally left blank.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report CHAPTER 3 REVIEW OF NONDESTRUCTIVE TESTING TECHNOLOGIES Materials are seldom free of defects. Some defects can be difficult to identify, but they can form the basis for rejection or penalty when they occur more frequently or in larger areas. NDT can provide a reliable and cost-effective means to assess the extent of defects in the material or product under evaluation without changing its physical properties. In the past, NDT has been an after-the-fact technology, used primarily for inspection of parts and components. In contrast, NDT today plays an important role in the examination and evaluation of materials and structures. A number of NDT technologies and inspection systems have been developed that provide data on the quality of the material that do not alter or damage the materials being tested. This chapter provides a discussion and overview on the state-of-practice of selected NDT technologies that have been used for measuring the properties and features of pavements. 3.1 Nondestructive Testing and Evaluation – Definitions and Terminology NDT is a rather general term used by engineers today to refer to any form of testing that aids in evaluating the strength or perfection of a material without causing any damage or detriment to the material. NDT encompasses several evaluation techniques that are broadly classified as active and passive. An active NDT technique involves the measurement of a response caused by the application of an external force or energy on or through a material or structure without changing the physical character of that material or structure. Ultrasonics, falling weight deflectometer (FWD), and penetration techniques are active NDT techniques. On the other hand, a passive NDT technique involves monitoring or observing the test object in its typical load environment or proof cycle and detecting the presence of a defect by comparing the observed and expected reactions. For instance, visual inspections, acoustic emission, and strain gages are passive techniques. Active NDT techniques are most commonly used in pavement engineering, and all discussion pertaining to NDT techniques in this report refer to active testing methods. NDT has been in use for a long time in many industries involving the evaluation, inspection, and QC of materials and/or constructed products. In fact, many of the nondestructive examination (NDE) procedures have been standardized. Agencies actively involved in NDT and NDE standardization issues include ASTM (Committee E-7 on Nondestructive Testing) and American Society for Nondestructive Testing (ASNT). ASTM E 1316-95a (Standard Terminology for Nondestructive Examinations) includes definitions for those testing technologies and terms typically used. As a result, ASTM E 1316 will be used as the standard for the terms used within this report. ASTM defines NDT as: "the development and application of technical methods to examine materials or components in ways that do not impair future usefulness and serviceability in order to detect, locate, measure and evaluate 47

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report discontinuities, defects and other imperfections; to assess integrity, properties and composition; and to measure geometrical characteristics." 3.2 Overview of Nondestructive Evaluation Technologies Using well established, and sometimes fundamental, physical principles, a number of NDT/NDE and/or inspection systems have been developed which provide information on the quality of the material or component. The main techniques used in NDE of pavements are: • Ultrasonic Techniques • Acoustic Emission Testing • Imaging Techniques, including Thermography • Radar and Microwave Techniques • Magnetic Flux Leakage Techniques • Radiographic Techniques All these NDE systems coexist and, depending on the application, may be used individually or in combination with one another. There obviously is some overlap between the various test methods, but they can be complementary to one another. For example, the fact that ultrasonic testing can reveal both internal and surface flaws and material properties does not necessarily mean that it will be the best method for all inspection applications. Much depends upon the type of flaw or quality characteristic to be measured. Table 16 is a listing of NDE techniques used for evaluating selected non-metallic materials and components, while Table 17 (portions of which were taken from ASTM E 543-02, Standard Practice for Agencies Performing Nondestructive Testing) provides a basic comparison of these selected NDE methods. 3.3 NDE Technologies for Evaluating Flexible Pavements Pavement engineers traditionally have relied on standardized laboratory test procedures to assess material properties and strength parameters. Unfortunately, laboratory tests may not represent the in-place condition of the material and may not provide an accurate condition of the performance characteristics of selected materials under in-service conditions. Furthermore, in pavement diagnostic studies or rehabilitation projects, it becomes imperative to evaluate the existing condition of the pavement, and laboratory tests that are “destructive” are not the preferred option. A number of NDE technologies and/or inspection systems have been developed that provide information on the quality of the material that do not alter or damage the material being tested. A detailed evaluation of these test methods was completed by Von Quintus et al. (1996) and Saeed et al. (2001) for measuring critical pavement properties and features. Although NDT has been the subject of research for several decades, it is only recently that NDT has seen wide applications in evaluating pavement construction quality. The specific application, the material property to be evaluated, and the obstacles to be overcome in each application dictate the choice of the NDE technique utilized for evaluating a specific material or layer. 48

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report Table 16. Key NDE Techniques for Selected Non-Metallic Materials, such as Ceramics and Composites NDT Techniques Methods Example Applications Dynamic Vibration Impact & Resonance Damage, Elastic Moduli Scanning Ultrasonics Contact & Non-Contact Defect Distributions & Damage Assessment Acoustic Microscopy Laser & Pulse Echo Micro-Flaw Detection & Characterization Analytical Ultrasonics Velocity & Attenuation Elastic Moduli, Distributed Flaw Population Statistics Radiography Film Radiography Micro-Flaw Detection, Porosity, Inclusions X-Ray Tomography Computed Tomography Fiber Architecture, Micro-Flaws, Density Variations Thermography Thermal Wave Thermal Conductivity, Densification Visual-Optical Laser/Fiber Optics Stress & Strain, Surface Finish & Roughness NDT methods saw their first applications in pavement engineering more as a diagnostic and forensic tool rather than a QA tool. NDT for pavement diagnostic studies has been confined primarily to the measurement of peak deflections on the pavement's surface for estimating pavement stiffness for overlay design or for backcalculation of elastic layer modulus. The FWD, the dynamic cone penetrometer (DCP), and more recently, the portable FWD (PFWD) have been the most commonly used deflection-based NDT devices in the pavement industry. An increased awareness of NDT applications in other fields drew significant attention from pavement engineers, resulting in the adoption of several other NDT techniques for evaluating pavements. Seismic and ground penetrating radar (GPR) methods have been used for determining the modulus and thickness of individual layers in flexible pavements, respectively. The efficiency of infrared thermography, impact-echo, impulse response, and other test methods also has been evaluated for use in pavement engineering (Nazarian et al., 1993; Maser, 1990, 1994, Willoughby, et al., 2003). Acoustic emission has been used to evaluate moisture damage in HMA pavements, while nuclear magnetic resonance techniques and non-nuclear devices have been used for measuring the density of pavement layers (Chang, 1994). Nuclear density gauges represent the current state-of-practice or baseline for QA application and are not discussed or evaluated as part of NCHRP Project 10-65. All discussions and evaluations with regard to density measurements were limited to the newer non-nuclear devices. 49

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report Table 17. Comparison of Selected NDE Methods (ASTM E543) Method Properties Sensed or Measured Typical Discontinuities Detected Advantages Limitations Ultrasonic examination Changes in acoustic impedance. Cracks, voids, porosity, lamination, delaminations and inclusions. Excellent penetration; readily automated; good sensitivity and resolution; requires access to only one side; permanent record, if needed. Requires acoustic coupling to surface; reference standard usually required; highly dependent upon operator skill; relative insensitivity to laminar flaws which are parallel to the sound beam. Sonic examination Changes in acoustic impedance. Disbonds, delaminations, cracks or voids. Simple to implement; readily automated; portable. Geometry sensitive; poor definition. Ultrasonic holography Same as ultrasonic examination. Used primarily for evaluation of discontinuities detected by other members. Produces a viewable image of discontinuities. Cost; limited to small regions of the structure; poor definition compared to radiography. Acoustic emission Stress wave energy generated by growing flaws, areas of high stress, leaks. Cracks, structural anomalies, leaks, also delamination, fiber fracture and matrix failure in composite materials. 100% volumetric examination in real time, complication geometries, very high sensitivity, permanent record, accurate flaw location. Structure must be loaded, sensors must be in contact with structure. X and gamma radiography Changes in density from voids, inclusions, material variations, placement of internal parts. Voids, porosity, inclusions and cracks. Detects internal discontinuities; useful on a wide variety of materials; portable; permanent record. Cost; relative insensitivity to thin or laminar flaws such as fatigue cracks or delaminations which are perpendicular to the radiation beam; health hazard. Neutron radiography Compositional inhomogeneiti es; selectively sensitive to particular atomic nuclei. Presence, absence, or mislocation of components or variations of suitable composition. Good penetration of most structural metals; high sensitivity to favorable materials; permanent record. Cost; relatively portable; health hazard. Strain gages Mechanical strains. Not used for detection or discontinuities. Low cost; reliable. Insensitive to preexisting strains; small area coverage; requires bonding to surface. Brittle coatings Mechanical strains. Not commonly used for detection of discontinuities. Low cost; produces large area map of strain field. Insensitive to preexisting strains. Optical holography Mechanical strains. Disbonds; delaminations; plastic deformation. Extremely sensitive; products map of strain field; permanent record if needed. Cost; complexity; requires considerable skill. Liquid penetrant examination Surface openings. Cracks, porosity, laps and seams. Inexpensive; easy to apply; portable. Discontinuity must be open to an accessible surface; false indications often occur. 50

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report Properties Typical Advantages Limitations Method Sensed or Discontinuities Measured Detected Eddy current examination Changes in electrical and magnetic properties caused by surface and near-surface discontinuities. Cracks, seams, laps, voids, and variations in alloy composition and heat treatment Moderate cost; readily automated; portable; permanent record if needed. Conductive materials only; shallow penetration; geometry sensitive; reference standards often necessary. Microwave examination Anomalies in complex dielectric coefficient; surface anomalies in conductive materials. In dielectrics: disbonds voids, and cracks; in metal surfaces: surface cracks. Noncontacting; readily automated; rapid inspection. No penetration of metals; comparatively poor definition of flaws. Magnetic particle examination Leakage in magnetic field flux caused by surface or near-surface discontinuities. Surface or near-surface cracks, laps, voids, and nonmetallic inclusions. Stable; inexpensive. Ferromagnetic materials only; surface preparation may be required; false indications often occur. Magnetic flux leakage examination Leakage in magnetic field flux caused by surface or near-surface discontinuities. Surface or near-surface cracks, laps, voids, and nonmetallic inclusions. Sensitivity to typical discontinuities; readily automated; moderate depth penetration; permanent record, if needed. Ferromagnetic materials only; proper magnetization of part sometimes difficult when parts do not have uniform cross section. Infrared testing Surface temperature; anomalies in thermal conductivity or surface emissivity, or both. Voids or disbonds in nonmetallics; location of hot or cold spots in thermally active assemblies. Produces a viewable thermal map. Cost; difficult to control surface emissivity; poor definition. Leak detection Pressure changes, bubbles, acoustic hiss, or the passage of a tracer fluid through a pressure boundary. Leaks in closed systems. Good sensitivity; wide range of instrumentation available. Requires internal and external access to system; contaminants may interfere; can be costly. 51

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report The remaining sections of this chapter provide a discussion on the fundamental principles of physics and engineering concepts used by different NDT technologies that have been used for measuring material properties and features of flexible pavements. Table 18 lists the technologies or methods that have been used to measure the properties and features of flexible pavements. The review provides a summary of the measurement techniques, equipment and output, and identifies those technologies that have been and can be used to measure specific properties. Table 18. NDT Technologies and Methods Used to Measure Material Properties and Features of Flexible Pavements Type of Property Volumetric Structural Functional Density-Fluids Content: • Nuclear gauges • Non-nuclear gauges • GPR – ground & air coupled methods Thickness: • GPR • Impact Echo • DCP Profile: • ARAN Van • Profilometer • Profilograph Density: • Pavement Quality Index • PaveTracker gauge • Humboldt nuclear density gauge • Onboard Density Measuring System Modulus: • Seismic (SASW, Impulse Response, Impact Echo) • Deflection (LWD & FWD) Noise: • Noise Trailer Segregation: • Infrared camera • Profile sensor van • R0SAN unit • GPR Modulus/Shear Strength: • DCP Friction: • Skid Trailer 3.4 Impact Devices and Technology for Unbound Materials and Layers Two types of impact testers for measuring the strength of unbound materials are discussed in this section—the DCP and the Clegg Impact Soil Tester. 3.4.1 Dynamic Cone Penetrometer The DCP is a testing device to estimate the in place strength and deformation characteristics of unbound pavement layers. The DCP was developed in South Africa in 1975 (Kleyn, 1975) and has been used in many parts of the world, including Australia, New Zealand, United Kingdom, Central Africa, Israel, Norway, New Zealand, Germany, Canada, Portugal, and several state and federal agencies in the United States. It is used primarily to measure the structural capacity of unbound pavement layers and embankments, and it can provide an assessment on the uniformity of compaction. It is considered a feasible tool for use in QA during construction because it is amenable to many types of evaluations and is easy to handle. 52

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report DCP testing can be performed on unfinished or compacted unbound pavement layers. By ASTM’s definition of NDT, the DCP is a quasi-NDT device, because it changes the structure of the material being tested. It also requires that the HMA surface layer be removed to test the supporting unbound materials and soils. Nonetheless, the DCP is considered an NDT device under NCHRP Project 10-65. The test involves driving a cone shaped probe into the soil or aggregate layer using a dynamic load and measuring the advancement of the device for each applied blow or interval of blows. The factors that have a direct impact on depth of penetration are drop height of the weight, cone size, and cone shape. The strength of the material being tested also has a direct impact on the depth of penetration with each blow. In other words, the resistance to penetration is dependent on the strength of the material. The strength, in turn, is dependent on density, moisture, and material type of the layer evaluated. Principle of Operation, Equipment, and Software The DCP is defined by ASTM 6951-03 as a device used to assess the in place strength of soils or compacted materials. The DCP device consists of a0.62 in. (15.8 mm) - diameter steel rod with a standard cone shaped tip, a 17.6-lb (8-kg) hammer that is dropped by a fixed height of 22.6 in (575 mm), a coupler assembly, and a handle. The cone tip has a diameter of 0.79 in (20 mm) with an angle of 60 degrees to reduce side friction. This is accompanied with a sliding rod, 0.62 in (15.8 mm) in diameter, but shorter in length, and attached parallel to the main rod to measure the penetration of the device. Figure 3 shows the manual DCP device in operation. The entire device is made of stainless steel to protect it from corrosion. However, the cone tip is made of hardened tool steel or a similar material to resist wear and tear. The test is conducted by dropping the weight and measuring the penetration of the cone. The data recorded include the number of blows and the depth of penetration. The rate of penetration is defined as the depth of penetration per blow, and is often referred to as the penetration index or the DCP ratio. The units used are mm/blow or in/blow. The penetration rate is determined as the slope of the curve relating the number of blows to the depth of penetration. The device can be operated manually (see Figure 3) or can be automated by installing it on a trailer, as shown in Figure 4. The automated DCP (ADCP) has a fully developed software tool to determine soil support values. Application to Flexible Pavement Testing The use of DCP in pavement evaluation and QA during construction has gained increased popularity mainly because the equipment is simple and easy to handle. It is amenable to many types of evaluations and several material types. It is also an economical device with minimal operator training needs. The information gathered with regard to base/subbase relative thickness and strength is invaluable compared to the resources and time consumed to perform the test. The DCP penetration rate, PR, in in/blow, has been correlated to several engineering properties of the material. It was initially correlated to the California Bearing Ratio (CBR) of the pavement subgrade (Kleyn, 1975; Livneh and Ishai, 1987; Livneh, 1989). These models 53

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report defined a relationship between the log of CBR and the log of PR. Further studies extended a different relationship for fine and coarse-grained soils (Harrison, 1989). More recently, the U.S. Army Corps of Engineers (Webster et al., 1992) developed a relationship based on a wide range of tests on granular and cohesive materials. This relationship, shown as equation 1, is the most widely used one today. Log(CBR) = 2.465 – 1.12Log(PR) (1) Figure 3. Photo of Manual DCP in Operation (courtesy of Minnesota Road Research Section, Office of Materials, Minnesota DOT) 54

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report ( )2555 CBRM R = Figure 4. Automated DCP Attached to a Trailer (courtesy of Minnesota Road Research Section, Office of Materials, Minnesota DOT) This correlation is included in the MEPDG for all unbound materials and soils. Correlations of the PR to other engineering properties are accomplished based on the relationship these properties have with CBR. For example, equation 2 is the more common relationship that has been used between resilient modulus (MR) and CBR (Heukelom and Klomp, 1962). MR = 1500(CBR) (2) The particular correlation or regression equation between MR and CBR that is included in the MEPDG is shown below, as equation 3. The CBR is measured in accordance with AASHTO T193, and provides a direct tie between use of the DCP for QA application and structural design of flexible pavements. (3) 64.0 Equations 1 and 2 or 3 can be used to relate the PR to MR values for subgrade soils. This type of correlation was validated by studies that verified the computed modulus to the backcalculated modulus from FWD data (Chen, 2001). Several studies have also correlated the PR to the elastic modulus of the subgrade (Chua and Lytton, 1981). In addition, some agencies are developing correlations between resilient modulus and the penetration rate or 55

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report index.4 The correlation included in the MEPDG, by combining equations 1 and 3, is shown as equation 4. ( ) ( )PRLogMLog R 7168.0985.4 −= (4) 3.4.2 Clegg Impact Soil Tester The Clegg Impact Soil Tester is an instrument for monitoring and controlling the quality of placing unbound aggregate materials and embankment soils. Generally, it is not used for testing the unbound layers of existing flexible pavements, unless the bound layers can be sawed and removed without disturbing the layers to be tested. The device is manufactured by SDi of Trowbridge. There are three versions of this device—one that uses a 4.5-kg hammer, another using a 2.25-Kg hammer, and a lightweight version that uses a 0.5-Kg hammer. The equipment consists of a compaction hammer operating within a vertical guide tube, as shown in Figure 5. Figure 5. Photo of the Clegg Impact Soil Tester (courtesy of SDi WebSite) The hammer falls within the guide tube when it is released and contacts the surface of the material or layer being tested. The hammer then decelerates at a rate determined by the 4 As an example, the Colorado, Minnesota, Montana, and Ohio DOTs have been collecting DCP and resilient modulus data on a range of soils and materials to develop similar correlations for their own use in pavement design and/or QA activities. 56

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report ) stiffness of the material within the region of impact. A precision accelerometer mounted on the hammer measures the deceleration and provides an output in units of an Impact Value. The test procedure recommends that five tests or drops of the hammer be used to obtain a reliable Impact Value for each test location. The first two drops take up the surface irregularities, and the final three drops should result in consist readings. The device can be fitted with software for data collection and quality control of the data. The Impact Value (IV) resulting from the tests has been correlated to CBR, much like the penetration rate from the DCP. The relationship that has been developed and included with the literature from SDi is shown as equation 5. (5) [ ]( 2124.0 += IVCBR 3.5 Deflection Measuring Devices and Technology The FWD and heavy weight deflectometer (HWD) are well documented in the literature. These two devices generally are categorized as one device with different maximum load capacities, with the higher-load capacity HWD primarily designed for airfield use. Either device can be used for evaluating the strength and response of pavements using appropriate load levels. The light weight deflectometer (LWD) operates in a similar fashion to the FWD; however, the LWD is small and light enough to be carried and operated by one person and is mainly used on unbound materials, where lighter loads are required. The LWD is less suitable for deflection tests on thick bound layers. On the other hand, the HWD is not well suited for tests on unbound layers. Of the deflection-based methods, the FWD has the greatest potential range of use for new or rehabilitated pavement structures. This method of evaluation essentially involves measuring surface deflections to applied loads of known magnitudes. The measured deflections are the pavement’s structural response, and an indicator of its structural capacity. This evaluation process has been used in rehabilitation designs, pavement management, forensics, and in determining seasonal load restrictions and overload permits. Deflection-based NDT devices can be grouped into three main categories: • Static or slow moving load deflection devices, such as the Benkelman beam, California traveling deflectometer, and the LaCroix deflectometer. • Steady state deflection (vibratory) devices, such as the Road rater and Dynaflect trailer. • Impact load deflection devices, such as the FWD, HWD, and LWD. The nature of loading imparted by a moving truck on the pavement is closest to the weight application simulated by the FWD and HWD impact load devices, making them a natural choice over the other deflection testing devices. The percentage of state agencies using the FWD has increased from 10 percent in 1980 to over 80 percent after the Strategic Highway Research Program (SHRP) was completed (Von Quintus et al., 1996; Von Quintus and 57

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report Killingsworth, 1998). Thus, the review of deflection-based methods focuses on the impact load devices. Two other deflection measuring devices recently have been developed that have great potential for use in evaluating, managing, and measuring the quality of flexible pavements. These two devices are the Rolling Dynamic Deflectometer (RDD), developed at the University of Texas, and the Rolling Wheel Deflectometer (RWD), developed by Applied Research Associates (ARA, 2004). Both of these devices are still considered to be in the research and development stage. Thus, they are mentioned but not included in the review of NDT technologies for measuring layer properties and features of flexible pavements and HMA overlays. 3.5.1 Falling Weight Deflectometer Principle of Operation, Equipment, and Software The FWD consists of an impact loading mechanism and a set of sensors to measure vertical surface displacements at the load location and at specified offsets from the load. The device delivers a transient load to the pavement surface, and the sensors measure the surface deflection at the specified locations. The system is trailer mounted, as shown in Figure 6. The loading device consists of a load plate that can apply an impulse load of different magnitudes ranging from1500 to 27000 lb (6.7 to 120 kN). The load can be applied from three standard drop heights resulting in a load pulse of 0.025 to 0.03 seconds. The load plate is circular and has a standard diameter of 6 inches (150 mm). Figure 6 also shows a close up of the FWD loading plate in contact with the pavement surface. A larger loading plate is used to test unbound aggregate layers and subgrade soils prior to placing the HMA or other bound layers. The vertical deflection response is measured at the surface of the pavement at different sensor locations. The sensor locations are typically chosen to adequately characterize the pavement structure being tested. These deflection measurements are used to characterize the deflection basin of the pavement (see Figure 7). A backcalculation algorithm is used to estimate the modulus of pavement layers based on the measured deflections and the layer thicknesses derived from cores. The HWD works on the same principle but uses a higher level of load. It is designed primarily for airfield use and is not discussed in this study. Test Procedure The developmental work for the FWD was conducted mainly at the Technical University of Denmark between 1969 and 1976, and the device has been used in the United States since 1978. ASTM D4695-03, Standard Guide for General Pavement Deflection Measurements, provides the guidance and procedural information for measuring pavement surface deflections, directly under, or at locations radially outward (offset) from the load. 58

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report Figure 6. Trailer Mounted FWD 59

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report Figure 7. Typical Deflection Basin Measured from FWD (after AASHTO, 1993) 3.5.2 Light Weight Deflectometer or Portable FWD After successful use of the FWD, the LWD, also known as the PFWD, was developed in the late 1970s to perform deflection based evaluations in a much simplified manner. The PFWD also uses plate bearing test equipment, but the deflection is measured at a single point under the loading plate (although some LWDs also measure the deflection at a few points from the loading plate). The measured deflection is used to calculate or estimate the bearing capacity or composite modulus of the pavement layers. Figure 8 shows a picture of two LWDs. More sophisticated versions of the LWD, with advanced software packages to process the load and deflection readings, were developed in Europe in the early 1990s. 3.5.3 Application to Flexible Pavement Testing Deflection-based measurements historically have been conducted to assess the bearing capacity of pavements in need of rehabilitation at the project level or for relative bearing capacity assessments at the network level as part of a complete pavement management system. In recent years, however, interest has grown to use deflection measurements (primarily with the FWD, but also with the LWD) to assess the structural properties of pavement layers that are under construction. Many of the same analysis techniques used for evaluating in-service pavements can also be applied to new construction measuring quality characteristics of QA programs (stiffness, modulus, etc.). The traditional backcalculation techniques have incorporated static-linear analyses to calculate the elastic modulus of the pavement layers. Commonly used programs include BISDEF, ELMOD, ELSDEF, EVERCALC, ISSEM4, MODCOMP, MICHBACK, 60

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report MODULUS, and WESDEF. However, these programs do not result in unique solutions of layer moduli (Von Quintus and Killingsworth, 1998; ASTM, 1994). Figure 8. Two Different LWDs Forward-calculation techniques have been developed and utilized in research studies to obtain unique solutions for layer modulus values for surface courses and the underlying unbound and bound layers (Stubstad, 2002). For forward-calculation of HMA layers, the first three sensors must be placed at 0, 8, and 12 inches (0, 203, and 305 mm), in compliance with LTPP’s protocol for testing flexible pavement layers. These techniques have been used in selected agencies (for example, Mississippi). The FWD test protocol for control or acceptance testing is much the same as the normal test protocols used for pavement rehabilitation design. Test spacing depends on the length of a particular project (for example, 10 meters or 25 feet, per lane), and the applied load has to be adjusted to realistic levels, with lower stress levels applicable to unbound layers. More drops of the FWD weight are needed for testing unbound materials than for bound layers, because of the increased variability associated with unbound material. In any event, a statistically significant sample of test results must be available in order to use PWL or other statistical approaches. 61

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 3.6 Ground Penetrating Radar Devices and Technology GPR is a geophysical nondestructive technique that uses electromagnetic pulses to test, characterize, or detect subsurface materials based on changes in electrical and magnetic properties of the subsurface layers. GPR is also referred to as ground probing radar, georadar, subsurface radar, or earth sounding radar. Literature traces its first use in Austria in 1929. Commercial GPR equipment became available in the 1970s, and wide research was carried out with its application in the pavements area over the next three decades. Both ground coupled and air horned antenna systems are available for multiple applications in the transportation area (see Figure 9.) Limitations of GPR have included the cost and complexity of the equipment, the need for interpretive expertise, and the requirement for office data processing. Recent developments with GPR hardware have yielded systems which are less expensive and easier to operate, which could overcome equipment complication issues. On the data processing side, prototype software for automated on-site processing has been developed (Maser, 2002 and 2003) to overcome some of the complicated processing issues. GPR has been used extensively for measuring pavement layer thickness and more recently has been applied to the measurement of pavement density and air content. ASTM D4748-98 (Standard Test Method for Determining the Thickness of Bound Pavement Layers Using Short-Pulse Radar) is used for estimating layer thickness; there is no ASTM nor AASHTO standard test method for estimating air voids. GPR layer thickness and air content measurements recently have been used for QA of new HMA pavement layers, as described in the following sections. 3.6.1 Operation Principles GPR works using short electromagnetic pulses radiated by an antenna which transmit these pulses and receive reflected returns from the pavement layers, as shown in Figure 10.a. The reflected pulses are received by the antenna and recorded as a waveform, as shown in Figure 10.b. As the equipment travels along the pavement, it generates a sequence of waveforms as shown in Figure 10.c. The layer boundary between the HMA and aggregate base is clearly visible in this sequence of waveforms. These waveforms are digitized and interpreted by computing the amplitude and arrival times from each main reflection. The pavement thickness and dielectric permittivity can be computed from these amplitudes and arrival times according to equations 6 and 7 (Maser and Scullion, 1992): ][ *150 2 * )( GPRa GPRGPRGPR ttVmmThickness ε== (6) Where: 2 ][ ⎥⎦ ⎤⎢⎣ ⎡ + += GPR GPR GPRa AApl AAplε (7) 62

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report Where VGPR is the velocity and calculated from εa[GPR], the dielectric constant of the HMA; tGPR is the time delay between the reflections from the top and bottom of the HMA layer, computed automatically from each waveform; AGPR is the amplitude of the reflection from the top of the HMA, computed from each waveform; and Apl is the amplitude of the reflection from a metal plate, obtained during calibration. GPR Data Acquisition & Storage GPR Antenna GPS Receiver b. Ground-Coupled Antenna Arrays Attached to Survey Vehicle Ground coupled array a. Air-Coupled GPR Antenna Attached to Survey Vehicle Figure 9. GPR Antennas Attached to a Standard Survey Vehicle 63

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report b) GPR Waveform asphalt bottom reflection Horn Antenna 1 3 Air Asphalt Base Volts Time a) Measurement Setup c) Sample Field Data Distance t Figure 10. Principle of GPR Operation for Pavement Layer Thickness Evaluation 3.6.2 Equipment and Software The current GPR technology used in transportation-related applications emerged over 30 years ago through two separate efforts: (a) the development of ground-coupled antenna systems for geological and geotechnical applications and (b) the development of air-coupled horn antennas for mine detection (see Figure 9). The ground-coupled equipment traditionally has been used for maximum depth penetration and where information is more qualitative rather than quantitative. This technology has been used for a variety of subsurface applications, including mapping of groundwater, bedrock, and soil layers, detecting pipes, buried drums, and subsurface contamination, and locating reinforcement. Antennas are available with center frequencies ranging from 80 MHz to 1.5 GHz, providing a wide range of penetration depths and resolutions. The use of higher frequencies (8-12 GHz) with ground-coupled antenna has been investigated at Iowa State (Jaselskis et al., 1998) for measurement of pavement density. The system incorporated high frequency antennas deployed in front of and behind the roller. A prototype was constructed and initial results showed promise, but problems were encountered and further development of this system has not been reported. The 1 GHz horn, air-coupled horn antenna equipment is operated 20 to 50 cm (8 to 20 in) above the pavement surface from a moving vehicle and thus allows data collection at highway speed. This antenna has proven to be suitable for pavement and bridge deck applications, where quantitative results are required at high resolution but for shallow penetration. The antenna is non-contact, and the typical 1 GHz horn antenna produces a clear, one-cycle pulse revealing interfaces in the pavement structure as close as 5 cm. Due to 64

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report the large surface reflection and the high frequencies used, the horn antenna is limited in depth of penetration to about one meter, which is adequate for most QA applications. Data from ground-coupled equipment generally are analyzed from a graphical display on a computer screens. The analysis seeks to relate arrival patterns from different reflectors to depths, horizontal locations, and qualitative descriptions of subsurface conditions (Ulricksen, 1982). Air horn antenna data generally are processed numerically with custom or commercially available software. Fully automated software has been developed for thickness and density measurement in newly constructed pavements (Scullion and Chen, 1999; Maser, 2002). The automation of thickness calculation is possible for new construction, since the nominal HMA layer thickness should be known for QA application. For pavement thickness evaluation on existing pavements, different software packages are available, but the analysis usually requires user intervention to identify the pavement layers to be analyzed and to ensure that these layers are properly tracked (e.g., Infrasense PAVLAYER; Roadscanner Road DOCTOR; GSSI RADACT; and Hyper OpticsTM Pavement Thickness Analysis). Most procedures in use to-date have employed a single antenna in their evaluation method. Electronic Pavement Infrastructure, Inc. (EPIC), however, has a proprietary analysis method that employs multiple antennas. The EPIC proprietary system is referred to under the trade name Hyper OpticsTM. This multiple antenna system includes four major modules that provide an assessment of the pavement: Pavement Thickness Analysis (PTA), Pavement Composition Analysis (PCA), Pavement Voids Analysis (PVA), and Relative Compaction Profile (RCP). The multiple antenna system can provide complete coverage of a lane, while providing multiple volumetric properties. The accuracy of the single or multiple antenna methods for estimating the volumetric properties has not been well established to-date. The accuracy of these methods will be addressed in chapter 4. 3.6.3 Application to Flexible Pavement Testing The most common application of GPR to pavements has been for thickness measurements associated with rehabilitation design and for pavement structure inventory data input to pavement management systems (Maser, 1999).ASTM D 4748-98 (Standard Test Method for Determining the Thickness of Bound pavement Layers Using Short-Pulse Radar) is commonly used for this application. Numerous research evaluations have been conducted to establish and confirm the accuracy of the GPR method for determining the thickness of existing pavements. The capability of GPR to determine HMA layer thickness has been verified for HMA surface layers and bituminous base layers (Roddis et al., 1992). Investigation of GPR for measurement of unbound aggregate base layer thickness has been limited, and the available data are for existing pavements. Studies in Texas and Florida showed that the average per site deviation between GPR and core measurements ranged from 19 to 25 mm, or 10 to 15% (Maser and Scullion, 1992; Fernando et al., 1994). These studies showed that base layer thickness could not be measured for cement treated bases, since there is inadequate dielectric contrast between the stabilized base and the subgrade. Subsequent studies of existing pavements have shown that the ability to detect the bottom of the base varies considerably, 65

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report and cannot be predicted in advance. 5 This unpredictability might be due to the "blurring" of the boundary between unbound base and subgrade caused by migration of fine material from the subgrade over time. The application to QC/QA for new construction is more recent, in part because of stringent accuracy requirements. The ability to detect unbound aggregate base layer thickness in new construction has not been reported. It is likely that the limitations to detect the base/subgrade boundary in older pavements will not be present in new construction. GPR has also been used to some extent for pavement condition diagnosis, such as detection of voids, segregation, and stripping in HMA (Scullion and Rmeili, 1997). As an example, GPR was used in a Georgia DOT study to identify areas of stripping in thick HMA pavements (Hammons et al., 2005 and 2006). GPR was only partially successful and primarily used to create analysis segments with different dielectric properties for taking cores and conducting seismic tests. As noted above, the accuracy of using GPR to detect stripping, segregation, density, and voids has not been established. 3.6.4 Application to HMA Density and Air Void Content Determinations Recent work has shown that air horn antenna GPR equipment can accurately characterize the air void content of a newly placed HMA mat. The air content is determined by correlation with the dielectric permittivity measured directly from the GPR signal (see equation 6). Saarenketo and Roimela (1998) demonstrated a correlation between the GPR measured dielectric and newly placed HMA air void content. The developed relationship that was used, shown as equation 8, led to the implementation of the GPR air horn antenna as a standard test method for QC of HMA density in Finland. ( ) ][GPRsbGPR GPReAAirVoids ε= (8) Where AGPR and bGPR are determined from calibration cores, and ε is the surface dielectric from GPR data. Results of this data correlation are discussed in chapter 4. Work in the U.S. confirmed the Finnish approach and demonstrated the ability to map air content variations on a two-dimensional plan view of the newly constructed pavement (Sebesta and Scullion, 2003; see Figure 11). Utilizing the methods developed in Finland for relating the surface dielectric to in-place air voids, probability distributions are generated for the air void content of each control strip or section by using cores to calibrate the equation 8. Air void predictions are made for approximately 5,000 GPR readings within each lot or test section. Other agencies have also recognized the benefits of applying the GPR technology in their QA program for measuring density or air voids. As an example, Florida DOT is currently sponsoring a project for confirming the use of GPR for evaluating density measurements and 5 Comments and statements are based on the personnel experience of Dr. Kenneth Maser and based on his extensive experience with the use of GPR. 66

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report has already completed some work using the Hyper OpticsTM technology. As noted above, the EPIC system uses multiple antennas and proprietary software to provide an estimate of the HMA mixture volumetric properties for complete coverage of a lane at highway speeds in real time. Details of this procedure and its reported accuracy are discussed in chapter 4. Figure 11. Contours of Air Void Content from GPR Data (Sebesta and Scullion, 2003) 3.7 Infrared Thermography Technology Infrared (IR) thermography is a diagnostic NDE method which relates changes in surface temperature of a material to subsurface or internal flaws. Thermography simply means the mapping of isotherms, i.e. contours of the equal temperature over the surface of a material or component, and is a method of evaluating materials by measurement of their surface temperature. Basically, heat-sensing materials are used to detect irregularities in temperature contours and such thermal irregularities can be related to defects and/or flaws. Test standards of this type include: E 1543-94: Standard Test Methods for Noise Equivalent Temperature Difference of Thermal Imaging Systems E 1213-92: Standard Test Method for Minimum Resolvable Temperature Difference for Thermal Imaging Systems Infrared is a particular implementation of thermography in which an infrared camera is used as the means for making surface temperature measurement. An infrared camera detects the infrared radiation emitted by a material surface. With appropriate calibration for material properties and background radiation, this radiation can be converted into a direct measurement of temperature of the material surface. IR thermography has been used for the detection of segregation in newly placed HMA, as well as stripping and debonding between HMA layers due to discontinuity in the temperature caused by the difference in voids between two areas (see Figure 11). The effectiveness of IR 67

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report thermography in identifying segregated areas was recently evaluated in Texas by taking measurements on new HMA overlays at the time of placement, coring, then identifying relationships between changes in the IR data with changes in the measured volumetric and engineering properties of field cores (Sebesta and Scullion, 2003). Analyses of results showed that changes in IR data were significantly related to changes in HMA properties, such as air void content and gradation. Figure 12. Infrared Camera Attached to Data Collection Vehicle The potential of IR to serve as a QA tool for evaluating HMA materials or flexible pavement construction has been studied in detail only within the past 5 years or so. It is expected that the use of IR cameras or sensors has potential for newly placed HMA lifts located near the surface, but the technology is less reliable for thicker HMA layers to detect anomalies located below the surface. Based upon current Texas DOT specifications, significant changes in the HMA are expected if temperature differentials greater than 25 ºF (13.9 ºC) are measured after placement but before breakdown rolling. Work performed in Texas indicates that results from IR imaging are indeed relatable to changes in HMA constituents, and the proposed acceptability limits being used for temperature differentials are reasonable. Washington DOT has also performed research that uses an IR camera to view the process of placing HMA (Willoughby et al., 2003). Their results show how the temperature changes are 68

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report depicted with the IR camera, and the correlation between temperature, density, and air void. A similar process was used in Georgia to detect stripping and segregation within thick HMA layers. The results from the Georgia study were less encouraging (Hammons et al., 2005). 3.7.1 Operation Principles Infrared thermography involves the use of non-contact surface temperature measurement to diagnose subsurface conditions. The basis of the measurement is that the surface temperature at a defect will differ from the "normal" or background surface temperature. In some applications, the object to be tested is artificially heated to produce the desired temperature differentials. In other applications, the heat input is either from solar radiation or from the natural temperature of the material or structure being tested. In either case the infrared sensor detects the infrared radiation emitted from the object, and converts the radiation measurement into a temperature measurement using the Stefan-Boltzmann Law: Q = σE(T4-T04) (9) Where: Q = Radiation emitted from an object (watts/sq.meter). σ = Stefan-Boltzman constant. E = Emissivity of the object. T = Absolute temperature of the object. T0 = Absolute temperature of the surroundings. Surface temperature is detected using an infrared sensor or an infrared camera that provide the required infrared radiation. Infrared cameras have been used for civil structures because they provide a two-dimensional image of large surfaces and they operate like a conventional video camera. The main difference is that the intensity levels of the infrared image are related to the infrared radiation (i.e., surface temperature) rather than the intensity of light. For example, where the infrared camera is set to a 50 degree temperature range and provides an eight-bit grey scale image, each pixel of that image provides 256 shades of gray representing the 50 degree range (or a resolution of 0.2 degree). Infrared camera operators often use a pseudo-color scale, in which the gray scale is replaced by a color scale. The use of color can highlight temperature changes that are less obvious in the gray scale image. An infrared sensor produces an output voltage proportional to the received infrared radiation at a point. These sensors are less expensive than infrared cameras and generally used more in automated manufacturing and QC operations, where the need is to monitor temperature at a fixed point or group of points. Infrared thermography has been used for the past 15 years as a method for detecting delaminations in bridge decks. The underlying principle is that, with solar radiation, the areas above delaminations will heat up more quickly than the "sound" areas due to the insulating effect of the delamination. These small temperature differentials (about 1 to 2 degrees), or "hot spots," can be observed as bright spots on a high-resolution infrared image. The results, 69

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report produced by mapping these identified areas onto a plan view of the bridge deck, are used for making rehabilitation decisions, and for scoping and estimating repair projects. 3.7.2 Equipment and Software Initial work by Washington State DOT and Texas Transportation Institute (TTI) utilized commercial infrared cameras, producing real time video images, such as shown in Figure 13. These cameras allow the user to adjust to the temperature limits so that the appropriate range is being viewed. For example, the selected range for the image in Figure 13 is 20.0 to 114.2 ºC (68 to 238 ºF). These cameras also allow taking snapshots in addition to continuous video, and they provide a cursor that displays numeric temperature values on the image. The low viewing angle required of the video camera creates some distortion of the temperature measurement, due both to the angle and to the range of distances from the pavement surface to the camera. However, temperature differentials associated with segregation seem to be large enough to overcome this distortion. Using the infrared camera, pavement surface locations with temperature anomalies have to be manually marked on the pavement surface while the image is being viewed, since the camera has no distance scale. Figure 13. Illustration of Variable Density Due to Temperature Differentials (Willoughby et al., 2003) Subsequent work by TTI and by Auburn University (Stroup-Gardiner and Brown, 2000; Stroup-Gardiner, 2003) has favored the use of an array of infrared sensors mounted behind the screed of a paver in a line transverse to the pavement. With this setup, the collection of infrared data is automated and continuous as the paver moves forward. Distance is monitored using a conventional distance encoder (used by TTI) or a global positioning system (GPS) (used by Auburn). The individual lines of temperature data are contoured to produce a continuous two dimensional strip chart thermal image of the pavement. Figure 14 shows an example of this type of equipment layout and the results. The prototype equipment is separated from the screed itself so as not to interfere with the paving process. Eventually, however, the equipment needs to be attached to the screed for routine use. Custom software 70

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report has been designed by TTI and Auburn research groups to automate the generation of output of the type shown in Figure 14(b). A third equipment option is an infrared thermometer, or "infrared gun." This is a hand-held point sensor used to obtain spot temperature measurements. Overall thermal patterns are more difficult to obtain with this equipment, but it is easy to use. 3.7.3 Application to Flexible Pavement Testing The use of infrared thermography for detecting segregation in newly placed HMA was recommended in NCHRP Report 441, Segregation in Hot Mix Asphalts (Stroup-Gardiner and Brown, 2000). The report noted the ability of infrared to detect two types of segregation: (1) temperature segregation (or temperature differences), where the HMA has been unevenly cooled due to uneven exposure to cold surfaces during transport, and (2) gradation segregation, where coarse aggregate segregates, resulting in localized areas with high air voids and thus more rapid cooling. Both of these types of segregation appear to be associated with eventual deficiencies in HMA properties. The infrared measurement, however, cannot distinguish one from the other. The report recommended that the thermal measurement be made prior to the first pass of the roller, since this is where the temperature differentials are greater. The effectiveness of IR in identifying segregated areas has been evaluated by the Texas and Washington DOTs. The Texas work involved taking measurements on new HMA overlays at the time of placement, coring, then identifying relationships between changes in the IR data with changes in the measured volumetric and engineering properties of field cores (Sebesta and Scullion, 2002 and 2003). Analyses of results showed that changes in IR data were significantly related to changes in HMA properties, such as air void content and gradation. The Washington DOT research work also used an IR camera to view the process of placing HMA (Willoughby et al., 2003). Figure 13 showed an example of their work. The figure shows temperature changes depicted with the IR camera, and the correlation between temperature fluctuations and density and air void content. 71

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report Infrared Sensor Bar (a) Infrared sensor bar behind paver screed (b) Sample infrared strip contour plot output showing segregation at 130 foot intervals Figure 14. TTI Continuous Infrared System (courtesy of Tom Scullion, TTI) 72

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 3.8 Ultrasonic/Seismic Devices and Technology Acoustic velocity, attenuation, and related frequency effects are the primary ultrasonic measurements which have been used in other industries to characterize and inspect materials. Basically, the ultrasonic velocity of surface waves is measured with special electromagnetic- acoustic transducer (EMAT) probes and correlated to Young's modulus and shear modulus of the material. Variations from the modulus values of undamaged materials are used to predict creep damage of in service components. The fundamental theory relates to the fact that wave velocity is sensitive to material condition. NDT using stress or seismic waves has been used extensively in the geotechnical and soil mechanics area for some time. Some of the tests based on wave propagation that have been standardized include the following: ASTM D4428 - Test Method for Crosshole Seismic Testing ASTM D4633 - Test Method for Stress Wave Energy Measurement for Dynamic Penetrometer Testing Systems ASTM D4945 - Test Method for High Strain Dynamic Testing of Piles Ultrasonic testing in the pavement area has been in use for many years for void detection by simply dragging a steel chain across the surface of a portland cement concrete (PCC) pavement. However, application of this technology for full-scale pavement evaluation did not begin until the early 1980s. Since then, studies have resulted in the development of sophisticated equipment for use by pavement engineers. Examples include the development of the DOCTOR (Sansalone and Carino, 1986), the application of scanning methods by Olson (Olson et al., 1992), and the development of the Seismic Pavement Analyzer (SPA [Nazarian and Baker, 1993]). Figure 15 is an illustration of the SPA, while Tables 19 and 20 summarize the ultrasonic testing techniques used by the SPA for determining various properties of the pavement structure and distress precursors, respectively. Other selected uses of ultrasonic test methods have included: • The adhesive and compressive strengths of HMA mixtures. • Determination of layer/component thicknesses and delaminations. • QC of compaction of HMA paving mixtures. • Identification and location of moisture damage and stripping in HMA mixtures at various depths in flexible pavements. • Determination of the elastic moduli and viscoelastic properties of materials and structures. • Identification and determination of material degradation due to creep and fatigue and glass-transition temperatures. The basis for NDT with stress waves is propagation of seismic waves in the material of concern. Seismic wave velocity is the speed at which a wave advances in a medium. It is a direct indication of the stiffness of the material—the higher the wave velocity, the higher the stiffness. Seismic wave methods simply consist of measuring time required for those waves to travel a given distance. Once travel time and distance have been measured, wave 73

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report velocities are calculated by dividing distance by travel time. Wave motions created by a disturbance on top of a layered system are described as compression (P-), shear (S-), and surface (Rayleigh, R-) waves. a) Device in Use Receivers (Geophones & Accelerometers Raise-lower mechanism Air Tank Low Frequency Source High Frequency Source PC PC Bus A/D Board I/O Board Amplifiers & Multiplexer A1, G1 A2 A3 G2A4 A5 G3 b) Schematic Figure 15. Illustration of the Seismic Pavement Analyzer (SPA) 74

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report Table 19. Strengths of Testing Techniques Used by the SPA (Nazarian and Baker, 1993) Testing Technique Strengths Ultrasonic Body Wave Young's Modulus of top paving layer Ultrasonic Surface Wave Shear modulus of top paving layer Impulse Response Modulus of subgrade reaction of foundation layers Spectral Analysis of Surface Waves Modulus of each layer Thickness of each layer Variation in modulus within each layer Impact Echo Thickness of paving layer or depth to delaminated layer Table 20. Levels and Nature of Measurements for Each Distress Precursor as Used by the SPA (Nazarian and Baker, 1993) Distress Precursor Test Quantity Measured Pavement Component Evaluated Impulse Response Change in flexibility due to change in moisture content Overall pavement system Moisture in Base Spectral Analysis of Surface Waves (SASW) Change in Young's modulus due to change in moisture content Base, subbase and subgrade Impulse Response Reduction in rigidity of the paving layer due to cracks Overall pavement system Fine Cracking Body Wave Velocity Delay in travel time of compressional wave because of longer travel path and lower rigidity Paving layer Impulse Response Significant increase in flexibility of slab due to lack of support under the slab Supporting layer Voids or Loss of Support Impact Echo Return (resonant frequency) associated with the thickness of slab Upper layer (HMA or concrete) Impulse Response Significant increase in flexibility of overlay due to lack of support under the overlay Overall pavement system Overlay Delamination Impact Echo Return (resonant) frequency associated with the thickness of overlay Overlay Ultrasonic SASW Shear wave velocity of HMA layer HMA layer Aging Body Wave Velocity Poisson's ratio, by measuring compression wave velocity of HMA layer and combining with shear wave velocity HMA layer 75

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report The solution of the equations of motion yields two types of stress waves in an unbonded, isotropic medium. These waves are called body waves, and they propagate with different velocities and generate different particle motions in the medium. The faster wave is called the dilatational, compression, or P-wave. This wave exhibits a push-pull type motion in the direction of wave propagation and is in the same direction of particle motion. The slower wave is called the distortional, shear, or S-wave. The shear wave exhibits rotational motion in which the direction of particle motion is perpendicular to the direction of wave propagation. The shear wave is sometimes subdivided into two special cases. The first is when both directions of particle motion and wave propagation are contained in a vertical plane. In this case, the shear wave is referred to as a vertically polarized shear wave or SD-wave. The other special case is when particle motion and wave propagation are contained in a horizontal plane. The shear wave in this case is referred to as a horizontally polarized shear wave or SH-wave. In an isotropic elastic medium, the magnitude of the SD-wave and SH-wave velocities are the same. "Rayleigh waves" are introduced into the pavement by hitting the pavement's surface with a hammer. The velocity of these waves will depend on the mechanical properties of the pavement's surface and to a lesser extent on the properties of the base, subbase, and subgrade materials. The sensitivity to depth is proportional to their wavelength (Seed et al., 1986). More recently, researchers have used wave propagation theory and seismic methods to evaluate different features of pavements (Lee et al., 1993; Lo et al., 1989; Nazarian, 1986- 2002; Seed et al., 1986; Hammons et al., 2005 and 2006). Some of these are: • Fine cracks in a pavement's surface that are not visible, but they will disrupt the local mechanical properties. A surface wave passing through a cracked area should experience high attenuation and a significant delay in arrival. These characteristics can be rapidly detected with geophones or accelerometers placed on the pavement surface. • Aging of the asphalt results in the hardening of the HMA at the surface layer. This hardening is associated with an increase in elastic modulus. Surface waves with wavelengths less than the pavement thickness will be sensitive to this change in elastic properties. The depth sensitivity noted above will reveal higher velocities for the shorter wavelengths (which are more sensitive to the near surface modulus increase) and lower velocities for the longer wavelengths. • Identification of moisture damage, stripping, and other forms of damage in HMA mixtures at varying depths in flexible pavements and HMA overlays. Seismic methods (specifically, the PSPA) were used with GPR to identify and locate the damaged areas. The surface waves are affected by the reduced stiffness of the damaged areas. Cores were taken and laboratory wheel load tests were used to confirm the damaged areas identified by the field tests. The seismic-GPR 76

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report combination was found to be excellent based on the visual observations from the cores and laboratory test results. • Depending on the soil type, moisture in the subgrade will change the shear wave velocity in the soil. The velocity of surface waves of sufficient wavelength will thus be affected by this moisture. A delaminated or debonded HMA overlay will change the mechanical structure of the pavement, and therefore, change the surface wave velocity. The presence of a subsurface void has a similar effect. Direct compression or shear waves (in the sonic range) involve propagation of P- or S-waves in the top pavement layer (Kolsky, 1963; Manning, 1985). The waves are initiated with a hammer blow and measured with an array of geophones or accelerometers. Anomalous decreases in wave velocity measured in this fashion will indicate some reduction in the average moduli of the pavement, and are used as an indicator of cracking and other deterioration in the material or layer (e.g., stripping, freeze/thaw damage). Most ultrasonic analytical methods used in determining material properties from seismic tests make the assumption that the material (or pavement layer) is elastic, isotropic, and homogeneous. This assumption is reasonable for PCC and some granular materials. For HMA and selected fine-grained soils, however, that assumption is inappropriate, especially for low frequencies and/or high temperatures. Corrections or adjustments to load frequency and temperature are required in most cases, when evaluating nonlinear and visco-elastic materials. The mathematical relationships of wave propagation theory that have been utilized in developing the different methods for determining pavement material properties from various seismic tests are given below. By employing elastic theory, the compression wave velocity, Vp, can be defined as: 5.0 2 ⎟⎠ ⎞⎜⎝ ⎛ += ρλ GVP (10) Where λ, G, and ρ are the Lame's constant, shear modulus, and mass density, respectively. The shear wave velocity, Vs, is equal to: 5.0 ⎟⎟⎠ ⎞⎜⎜⎝ ⎛= ρ GVS (11) Compression (P-) and shear (S-) wave velocities are interrelated by Poisson's ratio, ν. The ratio of the compressive to shear wave velocities is expressed in terms of Poisson's Ratio as: ( ) 5.0 5.0 1 ⎟⎠ ⎞⎜⎝ ⎛ − −= υ υ S P V V (12) 77

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report Rearranging the above equation, results in an expression for Poisson's Ratio. 1 15.0 2 2 −⎟⎟⎠ ⎞ ⎜⎜⎝ ⎛ −⎟⎟⎠ ⎞ ⎜⎜⎝ ⎛ = S P S P V V V V υ (13) For a layer having constant properties, surface (Rayleigh or R-) wave and shear (S-) wave velocities are related by Poisson's ratio as well. Although the ratio of R-wave velocity to S- wave velocity increases as Poisson's ratio increases, its change is insignificant. The velocity of Rayleigh waves (VR) in a material is slightly less than that of the shear waves, and the ratio of the two waves can be expressed by: ( )υ υ + += 1 12.187.0 s R V V (14) Wave propagation velocities have limited use in engineering applications, but the calculation of elastic moduli from them has direct application. The velocities of body waves are directly related to the constrained (M) and shear (G) moduli of an isotropic material by: (15) ( )2PVM ρ= (16) ( )2SVG ρ= Young's and shear moduli are related by equation 17. ( )υ+= 12GE (17) 3.8.1 Spectral Analysis of Surface Waves Test Method—SASW The Spectral-Analysis-of-Surface-Waves (SASW) test is a non-intrusive seismic test method that relies on the measurement of Rayleigh type surface waves. For non-intrusive seismic methods, all instruments are placed on the ground surface. The key point in the SASW method is the measurement of the dispersive nature of the surface waves, which are used to determine the shear wave velocity of the pavement, the base, and the subgrade. SASW is a powerful NDT method which indicates material modulus (stiffness) versus depth while measuring from one surface and without any coring or other material intrusion required (see Tables 19 and 20). This method for evaluating pavements was largely developed by Nazarian and Stokoe (1983, 1985, and 1986) and significantly improved by Nazarian and Baker (1993). It has been used by the Department of Energy, U.S. Air Force, and Department of Transportation in various capacities to evaluate the strength of subsurface 78

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report soils and to locate subsurface features in geophysical explorations. It has been used by a few agencies to determine the depth and thickness of soft (loose) fill, depth to rigid layers (limestone), and identification of granular layers that are saturated. SASW is the primary measurement technique employed in the SPA (see Table 19 and Figure 15). The SASW method is based upon measuring surface waves propagating in layered elastic media. The generation and detection of surface waves are controlled by an impact source and two receivers (or accelerometers) placed on the pavement surface. The two vibration transducers are located at known distances from the source. Typically, one of the distances is kept equal to two times the shorter distance. At the surface, the direction of particles in motion forms a retro-grade ellipse in a uniform material. The amplitude of Rayleigh-wave motion decays with depth and is less than about 10% of the surface amplitude at a depth equal to about 1.5 times the wave length. Surface wave velocity varies with frequency in a layered system with velocity contrasts, and this frequency dependence of velocity is termed dispersion. A plot of surface wave velocity versus wavelength is called a dispersion curve. The SASW tests and analyses are performed in three phases: 1) collection of data, 2) construction of an experimental dispersion curve from the field data, and 3) inversion (forward modeling) of the theoretical dispersion curve to match the experimental curve and provide the shear wave velocity versus depth profiles. The ratio of surface wave velocity to shear wave velocity varies slightly with Poisson's ratio (as stated above), but is usually assumed to be equal to 0.90 with an error of less than 5 percent for most materials. Measurement of the surface wave velocity with the SASW method similarly allows calculation of compression wave velocity for use in Impact-Echo (IE) test analysis, discussed in the next subsection. Knowledge of the seismic wave velocities (surface and compression) and mass density of the material layers allows calculation of shear and Young's moduli for low strain amplitudes (Heisey et al., 1982; Kolsky, 1963; Nazarian et al., 1987; Nazarian and Stokoe, 1983; Nazarian and Baker, 1994). The elastic modulus values determined by the SASW method need to be modified to account for higher strain levels when used as inputs to models that include nonlinear constitutive relationships for the material response. The SASW field tests typically consist of impacting the test surface to generate surface wave energy at various frequencies that are transmitted through the material. For single point stationary SASW, two accelerometer receivers are evenly spaced on the surface in line with the impact point to monitor the passage of the surface wave energy. To obtain increasingly deeper data, several tests with different receiver spacings can be performed by simply doubling the distance between the receivers about the imaginary centerline between the receivers. A PC-based data acquisition system digitizes the analog receiver outputs and records the signals for spectral (frequency) analyses to determine the phase information of the cross power spectrum between the two receivers for each frequency. The dispersion curve is developed by knowing the phase shift (Ф) in degrees at a given frequency (f) and then calculating the travel time (t) between receivers of that frequency/wavelength using equation 18. 79

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report ( )ft 360 φ= (18) Surface wave velocity (VR) is obtained by dividing the receiver spacing (X) by the travel time at a specific frequency (f) in accordance with equation 19. t XVR = (19) The wavelength (Lr) of the corresponding surface wave is related to the phase velocity and frequency by equation 20. f VL Rr = (20) By repeating the above procedure for any given frequency, the surface wave velocity corresponding to the given wavelength is evaluated, and the dispersion curve is determined. The phase velocity at wavelengths shorter than the thickness of the pavement layer is indicative of the quality of the material of the surface layer. Changes in the stiffness of the surface layer are manifested by different phase velocities. To obtain the material properties for layers, a forward modeling process to match the experimental dispersion curve is performed. Forward modeling is the process of determining the "true" shear wave velocity profile from the "apparent" velocity of the dispersion curve. The forward modeling inversion process is iterative and involves assuming a shear wave velocity profile and constructing a theoretical dispersion curve. The experimental (field) and theoretical curves are compared, and the assumed theoretical shear wave velocity profile adjusted until the two curves match. The SASW method and an interactive computer algorithm for both 2-dimensional and 3-dimensional analyses have been developed to compute a theoretical dispersion curve based upon an assumed shear wave velocity and layer thickness profile (Roesset et al., 1990; Stokoe and Hoar, 1978). The advantage of the SASW method over the IE method (discussed in the next paragraph) is that the thickness and the shear modulus of all layers in a pavement system can be determined with the SASW method, as opposed to only the top layer with the IE method. 3.8.2 Impact-Echo Test Method The IE test method is an acoustic wave-based method that has long been recognized as a powerful tool for the nondestructive testing of PCC, wood, and other materials. It has the advantage of requiring access to only one side of a test member and is capable of determining thickness as well as flaw depth when used on structures with simple geometries. It is one of the more successful NDE methods for evaluating the internal condition of structural concrete (floors, on-grade walls and decks), and was largely developed and improved at the National Institute of Standards and Technology (NIST) and Cornell University in the mid-1980s 80

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report (Sansalone and Carino, 1986; Sansalone et al., 1987). The method has also been used for evaluating HMA overlays on concrete bridge decks and PCC pavements. Nazarian adopted the IE test method in the SPA for measuring the thickness of the surface layer (see Tables 19 and 20). In an IE test, the surface is impacted with a small instrumented impulse hammer or impactor. From the mechanical impact, a transient stress pulse is generated and introduced into the material (Sansalone, 1993). This pulse travels as pressure waves with spherical wave-fronts, which are reflected by internal cracks, voids, and/or interfaces. The response, reflected wave energy, is measured by an accelerometer or displacement transducer placed on the surface a few inches away from the impact point. Use of a displacement transducer is the preferred response measuring device. The wave form is dominated by the compression or P-wave, as a result of this sensor placement relative to the impact point. Thus, the method relies on reflected compression wave energy from interfaces and/or discontinuities that are present in a pavement system. These pressure waves are reflected at the free surface back into the material, to be reflected again by the internal interfaces, and so on. Therefore, a transient resonance condition is set up by multiple reflections of pressure waves between the free surface and the internal defects. Resonant conditions are also set up by the plate-like vibrations of thin concrete sections which are delaminated from the total deck structure. These resonant conditions are identified in the accelerometer signal by using the Fast Fourier Transform (FFT) algorithm. Resonant responses are associated with the normal deck dimensions (HMA and PCC thickness) as well as with delaminations. The frequency of these resonances can be analyzed to uniquely distinguish "normal" thickness-related resonant conditions associated with the deck structure, from "abnormal" resonant conditions associated with delamination and concrete deterioration. In a typical PCC slab supported by soil or crushed stone base material, almost all of the energy is reflected because of the large difference in stiffness between PCC and unbound materials. Thus, the greater the difference in impedance between two adjacent layers, the larger the amplitude of the return resonant frequency. Conversely, if the two materials have similar impedances, little to no energy will be reflected at the interface—diminishing the resonant return frequency from being measured. Separation of the energy related to surface waves from the reflected energy is difficult, especially for thin slabs or low modulus materials, such as HMA in the summer months. This separation process is made simpler when the signal is Fourier transformed into the frequency domain to obtain the amplitude spectrum. The amplitude of the peak is directly proportional to the difference in impedance between materials on either side of an interface. Thus, the time domain signals are FFT transformed to the frequency domain and a transfer function is calculated. The transfer function between the accelerometer and load is used to determine the resonant frequency. Mathematically, the transfer function is defined as: ( )( )fX fYTF = (21) 81

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report Y(f) is the FFT transform of the receiver output in displacement units and X(f) is the FFT transform of the hammer input. Dominant peaks in the transfer function plot are resonances resulting from discontinuities such as the interface between the pavement layer and the base layer, delamination of an overlay, or cracks present in the pavement system. The depth of such reflectors is easily calculated from the following equation: ( )r P f VD 2 = (22) Where D is the depth of the reflector, VP is the velocity of compression waves and fr is the resonance frequency. However, good contact between the sensor and test surface is essential for obtaining accurate and reliable results. The IE method can operate effectively on bare concrete and with an HMA overlay. Testing on an HMA overlay is possible because the high frequency impact excites ultrasonic resonant behavior in the PCC below the HMA surface. Traditional sounding methods do not have this capability. Software is available to automatically interpret the IE data, and the system (hardware and software) has been tested in the field and checked against chain drag results when the HMA was removed (Sansalone, 1993). The resulting correlations of delamination locations based on impact-echo and on chain drag were described as excellent. Olson Engineering has developed test equipment for measuring the thickness of stiff or harder (aged) HMA surfaces. 6 However, using the IE test method on thicker HMA lifts at elevated temperatures after compaction, when the HMA is relatively soft, has yet to be validated for measuring layer thickness. 3.8.3 Impulse Response Test Method The Impulse Response method (also known as transient dynamic response and mechanical impedance method) is similar to the IE method and historically has been used to evaluate subgrade support conditions. IR involves striking or shaking an object or structure and determining its response versus frequency. The frequency response reflects the material properties of the structure, including modulus, layer thickness, etc. Resonant frequencies are often simple indicators of anomalous structural conditions. The response of a pavement with moisture in the subbase material will be different than for the same pavement without moisture. For the IR test, the surface of the pavement is impacted to produce a low frequency wave in the surface layer. A portion of the energy is reflected back to the surface at the interface between the surface and base layers and the remainder is transmitted to the bottom layers. The response of the pavement's surface and impacted energy are measured with a geophone and load cell, respectively. The geophone measures particle velocity which is numerically 6 Personal communication with Olson Engineering (Dr. Larry Olson), and observing a demonstration of the equipment for testing HMA surfaces. 82

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report converted to displacement. Both the force and displacement signals are captured, digitized and processed to develop a flexibility spectrum. The flexibility spectrum is the ratio of the displacement and load as a function of frequency. The load and displacement time-histories are transformed to the frequency domain using a FFT algorithm. A velocity transducer is used, as compared to an accelerometer in the IE method, to measure the response of a pavement system subjected to an impact source. The method is efficient in evaluating the support conditions under the slab. A smooth transfer function (called a mobility function) with a high dynamic stiffness are expected for good support conditions. A transfer function with peaks manifested at low frequencies and a low dynamic stiffness are indicative of a loss of support/existence of voids. Based on studies by Reddy (Reddy, 1992), the stiffness obtained by the IR method is also a good representation of the subgrade modulus provided the pavement is rigid with surface layer thicknesses greater than 3 inches. In these cases, the effect of the surface layer on the stiffness values is minimal. Preliminary research has shown that IR scanning should also be possible with the basic scanner approach. Thus, the combination of IE, SASW, and IR scanning with PC-based systems offers the promise for innovative, continuous measurement of pavement system conditions. 3.8.4 Ultrasonic Surface Wave Test Method—USW The Ultrasonic Surface Wave (USW) method is a variation of the SASW method. With the USW method, the properties of the top layer can be easily and directly determined without the use of a complex inversion algorithm. The USW method was used by Nazarian in developing the SPA (see Table 19) and a PSPA referred to as the "Lunch Box." Figure 16 shows the PSPA in operation and carriage case that was recently developed to facilitate its use in data collection during construction. To perform the test, a disturbance is applied to the surface to generate stress waves that propagate mostly as surface waves of various wavelengths. The waves are monitored and captured with a data acquisition system (through the receivers). Signal and spectral analyses are then used to determine the phase information of the transfer function (phase spectrum) and the coherence function between the two receivers. This information is used to develop a dispersion curve. A dispersion curve depicts the variation in the velocity of propagation with wavelength. To obtain the dispersion curve, the velocity of wave propagation, VR, and wavelength, Lr, are determined from the phase spectrum, Ν, at any frequency, f, (equations 19 and 20). In a theoretical dispersion curve for a two-layer system, two distinct branches are obvious. First, up to a wavelength approximately equal to the thickness of the uppermost layer, the velocity of propagation is independent of the wavelength. For wavelengths greater than the thickness of the surface layer, the dispersive characteristic of surface waves (i.e., variation of velocity with wavelength) is normally clearly evident. Therefore, if one simply generates high-frequency (short-wavelength) waves and assumes that the properties of the uppermost 83

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report ) layer are uniform, the shear modulus of the top layer, G, can be determined using equation 23. (23) ( )( 2RVkG γ= Where: υ16.013.1 −=k (24) γ = Density of surface layer. Carriage case recently developed for facilitating the use of the PSPA & DSPA in data collection. Figure 16. PSPA in Operation for Testing HMA Layers, While the DSPA is Used for Testing Unbound Layers 84

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report An estimate of the thickness of the surface layer can be made by determining the wavelength above which the surface wave velocity is constant. The methodology can be further simplified by assuming that the stiffness of the top layer is constant. With that assumption, equation 18 can be written as: mff V X R =⎟⎟⎠ ⎞ ⎜⎜⎝ ⎛= 360φ (25) Equation 25 represents a linear relationship between the phase of the transfer function and frequency, provided the phase velocity is constant. Thus, VR can be easily determined by performing a least-squares linear regression over the high-frequency region of the cross power spectrum and obtaining the slope of the best-fit line. This method uses the time signals measured with two accelerometers. These two signals are Fourier-transformed, and the ratio of one to the other is calculated in the form of a transfer function. Since this is a complex-valued function, each point can be represented by its magnitude and phase. With this method, only the phase of the transfer function is used. To ensure high quality results, the source should impart surface wave energy over a wide range of frequencies (in the range of 2 KHz to 40 KHz), and the sensors should be in intimate contact with the pavement. There are two main advantages to using this method (relative to determining the "bulk" surface wave travel time using the time record). First, the variation in velocity with depth (via a dispersion curve) can be determined so that the extent of pavement damage or surface deterioration can be estimated. Second, as the velocity is averaged over the thickness of the HMA (or PCC), the near-surface deterioration does not significantly affect thickness determination, as is the case with the time-domain method. 3.8.5 Ultrasonic Body Wave Method—UBW According to equation 22, to effectively use the IE method, the compression wave velocity through the HMA (or PCC) must be known. Otherwise, as proven by Sansalone and Carino (1986), large errors will result. For simplicity of the equipment according to Nazarian, the compression wave velocity can be determined using the ultrasonic body wave method. The Ultrasonic Body Wave (UBW) method was also incorporated into developing the SPA and the PSPA (refer to Tables 19 and 20, and Figures 15 and 16). As indicated before, compression waves travel faster than any other types of seismic wave and are detected first on seismic records. An automated technique for determining the arrival of compression waves, as suggested by Willis and Toksoz (1983), has been implemented in the SPA. With this method, the detection of P-wave arrival is done in two steps: 1) event detection and 2) fine adjustment. Event detection is carried out by triggering on the first amplitude that falls within a time window satisfying a predetermined amplitude threshold. The threshold value depends on the background noise and anticipated amplitude of the compression wave. This value is typically set as the average of the voltage levels of the background noise and half the anticipated 85

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report maximum amplitude of the P-wave. The user defines a window within which the compression wave velocity energy is most likely to be concentrated. This is done by defining the most likely value and possible range for the modulus of HMA or PCC. The first point within the window that has a voltage above the threshold is considered a candidate for the arrival of the P-wave. In the second step, to verify that the arrival of the wave is correctly detected, a semblance correlation is carried out between the peaks selected for different records. The semblance discriminates between amplitude differences and the shape of the signal corresponding to the compression wave energy. This two-step procedure not only yields a robust procedure for detecting the compression wave velocity but also determines if the arrival times were falsely selected because of uncorrelated changes in the background noise level. To clearly detect the arrival of P-waves, the records must be greatly amplified. The compression wave velocity is calculated from the distance between receivers and the difference in travel time. Practically speaking, an intimate contact between the receiver and surface, a strong source, and extremely low-noise amplifiers are needed to obtain repeatable, interpretable, and reliable signals. To minimize the contamination of the signal with the so-called near-source energy, the source should be able to generate very short-duration impulses. Minor near- surface imperfections and cracks can adversely affect the results (in terms of detecting the compression wave energy or accurately determining the velocity of the material). Even though this sensitivity to minor defects is undesirable for estimating thickness, it can be effectively used to evaluate the quality of the layer or material. Although the determination of the compressive wave velocity can be carried out with one receiver (through so called "direct measurement"), numerous studies have shown that more reliable results are obtained when the difference in travel time between two receivers (so called "interval measurement") is used. Many problems (such as interval delays) in the system can be avoided when two sensors are used. When using the PSPA at a given test location and initiating the testing sequence through use of the computer, the high-frequency source is activated. The source is fired at least seven times. For its last three impacts, the output voltages of the receivers are saved and averaged (stacked) in the frequency domain. The other impacts are stacked to determine the arrival of compression waves. The gains are set so that the output of the sensors is optimized. The collection and reduction of data at one point take less than 15 sec when an i-386-20 MHz IBM-PC compatible is used. The data acquisition system for the PSPA can acquire data at a rate of 500 K-sample per channel. 3.8.6 Scanning Test Systems The ultrasonic testing rate of traditional point-by-point limits the utility of the stress-wave based methods. Thus, IE and SASW scanning systems have been developed and evaluated to fill the need for rapid collection, processing, and display of NDT data from stress-wave based methods. 86

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report IE Scanner System. The IE scanner system developed by Olson Engineering is capable of measuring the thickness of a wall or floor slab with an accuracy of typically 5% or better, depending on the accuracy of the velocity used for computing member thickness. The processed data displayed by the system are plots of either frequency peak(s) or member thickness versus position, along with signal energy or amplitude versus position. The IE measured member thickness profile can then be compared to the design thickness for verification of as-built conditions. The IE scanner also shows the locations of defects, such as cracks and delaminations. To allow rapid scanning with the IE method, the fixed accelerometer or displacement transducer was replaced with a rolling velocity transducer assembly. The hand-held hammer or other manually operated impactor was replaced with an electrically driven solenoid. The impactor drive and the rolling transducer signals are controlled by a digital controller which measures the distance traveled and tests at pre-set distance increments, regardless of speed of motion. The current IE scanning system is capable of testing either 1 or 4 simultaneous lines of points at incremental point spacings of about 3/4 inch (1.9 cm) for 1 channel scanning, or about 3-4 inches (7.6-10 cm) per channel for 4 channel scanning. Testing rates on smooth, open floor slabs have been typically less than 5 minutes for a 50 lineal feet (15 m) scan, during which about 600-800 IE tests will be performed by the 4 scanners. The system has been found to be sensitive to slab or wall thickness changes of as little as 0.1 inch (2.5 mm), with the absolute accuracy dependent on the accuracy of the measured material velocity. However, the IE scanner has yet to be validated and used on a production basis for testing HMA with rougher surfaces and lower modulus materials. SASW Scanning System. The IE scanning system was modified at Olson Engineering to allow the performance of SASW scanning. The SASW scanner hardware is based on the IE scanner hardware, modified to allow 2-channel data acquisition from a moving source- receiver-receiver line. The SASW scanner allows near continuous acquisition of data at fixed increments as the unit is rolled across the test surface. A distance wheel controls the impact source that generates the surface wave energy, and tracks the scanner position. The SASW accelerometers are replaced by two rolling displacement transducers, and the impact source consists of an electrically operated solenoid impactor. The SASW method gives information as to the material properties versus depth, based on the measurement of the surface wave velocity versus wavelength of propagating surface waves. The SASW scanning technology is new and allows rapid acquisition of SASW data, but has yet to become commercially available. Faster data processing software is in the development stage to allow full utilization of this new testing system. Most of the initial data with the use of this scanner has been collected on relatively smooth PCC slabs. The roughness of the surface is also a problem for this method to be applicable on a production basis for flexible pavement testing. 87

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 3.8.7 Acoustic Emission Test Method Acoustic methods are based on monitoring and identifying variations in sound waves emitted by the test specimen to locate flaws in the material. Acoustic emission (AE) qualifies as a passive technique because it involves monitoring or listening to sound waves generated by stresses within a material. AE sensors are piezoelectric transducers that detect out-of-plane displacements created by sound waves. The devices are highly mechanized and automated with sophisticated computerized measurements of resulting voltages. Most NDT methods, such as ultrasonics, radiography, or infrared thermography, rely on the application of some form of energy to the material and/or structure under tests. The difference between the applied energy and that detected at some later time or distance give an indication of the state of the material. AE, however, relies upon the detection of energy released by the material and/or structure in itself. Although AE has been used to test materials for as long as other ultrasonic tests, it is a relatively new field that only recently is being used as an important research tool for the study of materials. AE is simply a form of skilled listening to structures and/or materials as changes within the materials occur. It is applicable to a wide variety of materials, including metals, ceramics, polymers, HMA, composites, and wood. The use of AE has been largely confined to a laboratory environment to monitor the failure or damage of specimens under full-scale testing or when specimens are being tested by other methods. Liu and Li (1989) used the total energy of AE to measure the degree of the damage caused by plastic deformation during tensile tests of selected steels. The same techniques have been used in rock mechanics to identify failure plans (or plans of weakness) in testing rock samples for foundations. AE inspection has several advantages over conventional NDT techniques in that it can access the dynamic response of a flaw to imposed stresses. The following lists some of the standards which are used to test and inspect a variety of materials: E 976-94: Standard Guide for Determining the Reproducibility of Acoustic Emission Sensor Response E 750-88: Standard Practice for Characterizing Acoustic Emission Instrumentation E 1106-92: Standard Method for Primary Calibration of Acoustic Emission Sensors A typical AE test uses a piezoelectric sensor mounted on a test piece with a couplant (often some type of gel or grease). Because the output signal from the sensor is small, a preamplifier is used to amplify the signal, eliminate some noise, and provide impedance matching with the analysis system. The analysis system usually includes further amplification and the means to measure and record data from the signal. Conventional AE systems use digital counters to measure parameters of the analog signal, but some of the newer systems use microprocessors to digitize and record the signals. Recent developments and high-speed microcomputers and microprocessor chips have made it feasible to digitize and record the entire AE wave form. This development has led to new analysis techniques and better identification of the sources. Analysis methods have often focused on locating sources of AE and then trying to identify those sources. However, 88

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report sources are often further inspected using other NDE techniques to determine their size and severity. The AE technique was employed within Strategic Highway Research Program (SHRP) for monitoring the micro-fracture in the asphalt phase and/or between asphalt and aggregate when the HMA was subjected to mechanical loading (Pearson, 1994). Together with mechanical results, these give a comprehensive indication of the internal structure change of HMA during the loading process. Based on the preliminary test results included in the SHRP, the micro-damage process in HMA can be detected using AE methods. The results from both single-cycle and multi-cycle tests indicate that the deformation of a specimen relates closely to the AE event which is the measurement of damage within the specimen. Other uses of AE tests included in the literature have included: • QC of HMA mixtures, as related to laboratory compaction. • Determining the fracture energy (toughness) of PCC mixtures, the fracture-failure planes of rock, the cohesion of asphalt, and the adhesive strength in HMA mixtures. • Prediction of rock hardness, drillability, and fracture toughness. • Investigation and monitoring of avalanches, soil stability, retaining walls, earth-dam stability, and foundations. Currently there is no commercial equipment that can be used to collect, analysis, and interpret AE data for evaluating the in-place condition and quality of HMA mixtures and other pavement layers. The AE method is considered to be in the initial research and development stage. 3.8.8 Laser-Induced Ultrasonic Test Method One of the problems associated with the conventional use of piezoelectric transducers is that they must be acoustically coupled to the test material. A number of techniques for the non- contact generation and detection of ultrasound are available. Two of these include Electro Magnetic Acoustic Transducer (EMAT) and laser-based interrogation systems. EMAT generally are limited to metals and must be close to the test specimen. Laser-based ultrasonic interrogation systems are in use to a limited extent for inspections during the manufacturing process. These systems are mainly used to detect the existence of piping and the quality of cast steel. The ultrasonic pulses are produced by focusing a series of light impulses from a laser on the surface of the test specimen. The laser sends out a series of short high-energy light impulses. These impulses are 20 nanoseconds in length and are encountered by thermo-mechanical effects into sound impulses at a frequency of 1MHz to 100 MHz. The advantage of this technique is that no mechanical coupling is required and the acquisition of test results is rapid. The disadvantage is that the sensitivity of the system is lower than that for conventional ultrasonic pulse-echo testing systems (piezoelectric techniques). Thus, these methods are not considered applicable for evaluating the in-place condition and quality of HMA mixtures and flexible pavements. 89

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 3.9 Steady-State Vibratory Devices and Technology The Humboldt Stiffness Gauge (referred to as the GeoGauge and manufactured by Humbolt Manufacturing Company) is a hand-held portable instrument that provides a simple and rapid means of measuring the elastic modulus of compacted soils and aggregate base layers (Humboldt, 2002). Figure 17 shows the GeoGauge in operation. The GeoGauge was developed under a project co-sponsored by the FHWA and the Advanced Research Programs Administration in the late 1990s for potential use as a QA tool in evaluating the quality of unbound aggregate base materials, embankment soils, and HMA mixtures. The GeoGauge has been beta-tested by FHWA and multiple state highway agencies through a pooled fund study that was initiated in the early 2000s. Figure 17. GeoGauge Used to Estimate the Stiffness and Modulus of Unbound Layers and Soils in Flexible Pavements Outside of the initial pooled fund study, the GeoGauge has had limited use in measuring the stiffness and modulus of flexible pavement layers. FHWA used the GeoGauge to determine the repeatability of the measurements and to evaluate the unbound materials that had been placed at their accelerated loading facility at Turner-Fairbank Station (FHWA, 2003). A few state agencies have also used the GeoGauge on a limited basis, and there are mixed reports of success. Some of the reports found good repeatability, while other verbal reports found it to be highly variable. As a result of the beta-testing under the pooled fund study, revisions were made to the gauge and calibration procedure. Some of the revisions included minor mechanical design changes to reduce variability, enclosing the shaker to eliminate the effect of electromagnetic fields on 90

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report the motion sensor, and fixing errors identified in the signal generation and processing circuit boards. More recent use of the gauge by the FHWA has shown that the variability has been reduced and that the modulus values reported with the gauge are more correlated to the results from other NDT devices (for example, backcalculated modulus values from the deflection basins measured with the FWD). 3.9.1 Principle of Operation Wang et al. (2003) provides a detailed discussion on the principle of operation and measurements of the GeoGauge. In summary, the GeoGauge measures the impedance at the surface of an unbound layer. It imposes a stress to the surface of a layer and the resulting surface velocity is measured as a function of time. The GeoGauge imparts very small displacements to the soil (< 1.27 x 10-6 m or < 0.00005") at 25 steady state frequencies between 100 and 196 Hz. The stiffness value is determined for each frequency, and the average value is displayed on the surface of the gauge. The entire testing process takes about 1.5 minutes. At the low frequencies, the impedance at the surface is stiffness controlled and is proportional to the shear modulus of the soil. With a Poisson's ratio and the GeoGauge's foot dimensions, shear and Young's modulus are derived. The GeoGauge weighs about 22 lb (10 kg), is 11 in (28 cm) in diameter, 10 in (25.4 cm) tall, and rests on the soil surface through a circular foot. The foot bears directly on the soil and supports the weight of the GeoGauge through several rubber pads, or what Humbolt calls “isolators.” The shaker that drives the gauge is also attached to this foot. The sensors that measure the force and displacement-time history of the foot are also attached to the foot. The connection between the shaker and force sensor of the gauge is manufactured as a rigid column. It is powered by six disposable D-cell batteries. The GeoGauge is placed on the soil or aggregate base layer. A sand cushion or thin layer is usually placed to ensure that the gauge and surface of the layer are adequately coupled. A slight push or rotation of the GeoGauge is applied to ensure at least 60 percent contact area between the foot and soil. The GeoGauge displays and logs the data in memory. These data can be downloaded to a laptop or PC for more detailed data analysis. 3.9.2 Application to Flexible Pavement Testing As noted above, outside of the pooled fund study there has been minimal use of the GeoGauge related to flexible pavement evaluation or QA. However, the device has direct application for judging the quality of unbound pavement layers and results in an estimate of the elastic or resilient modulus of the unbound layer. 3.10 Magnetic Imaging or Tomography Technology Magnetic imaging tools function by emitting and detecting magnetic field in the localized area being scanned for the presence of any magnet-attracting materials. This technology has been applied in pavement evaluation in European countries and more recently in the U.S. 91

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report The technology is being used to locate the alignment of dowel bars and tie bars in PCC pavements and for determining the thickness of both HMA and PCC. The device being used in the U.S. for monitoring dowel bars is called the MIT-Scan-2, and for thickness measurements the MIT-Scan-T. Figure 18 shows the MIT-Scan-T being used to measure pavement thickness. Figure 18. MIT-Scan-T to Measure Flexible Pavement Thickness 3.10.1 Principle of Operation Magnetic imaging tomography enables users to determine the thickness of flexible or rigid pavements nondestructively, in an accurate and cost-effective manner using data collected from the magnetic field. The response signals of metallic reflectors at the bottom of the layer to be measured induced by short magnetic pulses are recorded and evaluated using magnetic tomography. Consequently, any metallic objects within the proximity of the scan unit will influence the measurements. To obtain reliable results, the surface of the joint to be scanned must be free of any metallic objects (e.g., coins, keys). A standard test procedure has been developed in Germany for this test and is identified as Specification TPD StB99. 3.10.2 Measuring Procedure The device may be operated in three modes: the service mode, the search mode, and the measuring mode. 1. In the service mode, different data for identification of the measuring place and the type of reflector may be put in. 92

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 2. In the search mode, the device is moved meander-like in a distance of 2 to 4 in (5 to 10 cm) above the ground. The unknown location of the reflector is indicated by a four-line bar diagram and, if desired, by an acoustic signal. 3. In the measuring mode, the reflector is crossed by the device. The crossing starts before the reflector and ends about 3.3 ft (1 meter) behind it. The constant given measuring time of 4 seconds prevents subjective measuring errors. The exact localization of the center of the reflector is not necessary, but must be met only within about 7.9 in (20 cm). During measurement, all necessary environmental conditions are recorded, so no calibrations of the device are required. The magnetic device requires certain conditions and precautions during use: • Minimum distance between neighbored reflectors: 1.6 ft (0.5 m), from edge to edge • Distance to guardrails: 3.3 ft (1 m) • Distance to parking vehicles: 6.6 ft (2 m) • Measuring temperature: 23 to 112 °F (-5 to 50 °C) Metallic parts on the pavement must be removed. Safety shoes with metallic caps induce perturbations too. However, no errors are induced by wet pavements, weakly conducting or magnetic additives within the pavement. Reflectors in the device, which are essentially circular sheets of galvanized sheet steel, are 0.04 in (0.6 mm) thick for both PCC and HMA, and have a diameter of 11.8 in (30 cm) and 2.8 in (7 cm) for layer thicknesses up to 18 and 8 in (45 and 20 cm), respectively. The device has a measuring accuracy of 0.04 in (0.1 cm). 3.11 Non-Nuclear Density Estimating Devices and Technology The acceptance of flexible pavement layers typically is based on in-place density of individual layers. Cores are used by most agencies for acceptance, while nuclear density gauges are used by most contractors for controlling the materials and compaction process. Non-nuclear density gauges are being investigated for use by many agencies. For example, Indiana, Kentucky, Louisiana, Maryland, Nebraska, New York, North Carolina, Ohio, Oklahoma, Pennsylvania, Tennessee, Texas, Utah, Washington, and Wisconsin have obtained different devices and are comparing them to the densities measured by traditional nuclear gauges and cores. The Pavement Quality Indicator (PQI) was used in field studies of NCHRP Project 9-15 and have been used at the National Center for Asphalt Technology (NCAT) test track (Killingsworth, 2002). Non-nuclear methods have been under development for HMA density testing for many years, but none have proved very successful. The systems can be categorized into two basic types—compactor or roller-mounted and non-roller mounted devices. The early non-roller- mounted systems created interest but failed to become commercially viable. These early systems included the “Density on the Run” system (Seamon, 1988) and the “Rolling Dynamic Deflectometer” (Bay et al., 1995). These systems used some form of acoustic or vibratory waves to estimate density. The focus later changed to electrical current for the non- roller-mounted systems, and systems like the PQI, PaveTracker, and the Electrical Density Gauge (EDG) show promise today. 93

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report The roller-mounted density measuring methods are also not a new phenomenon. Since the mid-1970s, attaching a reliable density estimating device to a roller has been tried. The technologies used to achieve the density measurement have included piezo-electric acceleration measurement (the most popular), gamma rays, dielectric probes, GPR, and microwaves. The four most recent attempts have been in the areas of microwaves and piezo- electric acceleration measurement. Jaselskis et al. (1998 and 2003) used microwaves to measure density with the “Roller- Mountable Asphalt Pavement Quality Indicator.” Minchin, Thomas, and Swanson (1999, 2001, and 2003) used piezo-electric acceleration measurement to measure HMA density with the “Onboard Density Measuring System,” and BOMAG and Geodynamic have similar systems. BOMAG’s system is currently on the American market, while Geodynamic is attempting to translate their success in measuring the density of embankment soils to HMA. Many of these systems do not actually measure density, but rather measure the response of the roller’s vibration and translate that response into a density or material stiffness. The status of GPR and laser technologies in the area of densities were covered in an earlier section dealing specifically with those technologies. 3.11.1 Roller-Mounted Density/Stiffness Systems Some of the roller-mounted systems are being referred to as Intelligent Compaction (IC). IC is an emerging technology that actually monitors layer stiffness during compaction by instruments attached to the roller to measure the reaction of the material being compacted. Many of these rollers are not true intelligent rollers, because they only measure the response of the roller as the material densifies. In other words, the compaction effort is not varied during the rolling process; the roller operator is responsible for discontinuing the rolling operation, once the maximum stiffness has been achieved. Briaud and Seo (2003) provided a thorough overview of the different roller-mounted compaction control processes, including IC rollers. IC and other roller-mounted systems give the contractor the opportunity to continuously monitor or test and document layer stiffness at the time of compaction, producing more uniformly-compacted material layers and allowing real-time compaction modifications based on response outputs. The output from this technology also provides documentation for owner and contractor management regarding material quality of all pavement layers. Thus, the technology has the potential to improve density and quality, having a positive impact on pavement construction and performance. IC technology has been in existence for several years and provides real-time, in-place material stiffness data that can be used by roller operators to make better decisions. The use of IC technology as a viable construction quality measure has increased over the past decade. Over 15 state and federal agencies have sponsored demonstration projects or case studies to date. Multiple manufacturers (Ammann, BOMAG, Caterpillar, Geodynamic, and Sakai) now build compaction monitoring or IC equipment with varying outputs and controls (see Figure 19). These manufacturers have fully equipped rollers, but also have instrumentation kits that can be attached to existing vibratory rollers. As noted above, however, not all of these rollers 94

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report can vary the compaction effort during rolling to account for increasing stiffness of the layer being compacted. a. BOMAG Asphalt Manager IC Roller c. Caterpillar IC Roller d. Vibratory Roller Instrumented by TTI for Use on Research Projects b. AMMANN IC Roller Figure 19. Fully Equipped Rollers Measuring the Stiffness of the Material Being Compacted Most of the instrumented rollers used to control or monitor the compaction process in real- time can be grouped into three categories: relative compaction monitoring equipment, absolute compaction monitoring equipment, and intelligent compaction equipment. Three of the roller-mounted systems that fall within these categories include: the Onboard Density Measuring System (ODMS) patented by Pennsylvania State University—the absolute compaction monitoring equipment; the Continuous Compaction Control (CCC) system marketed by Geodynamik—the relative compaction monitoring equipment; and the Asphalt Manager manufactured by BOMAG—the IC monitoring equipment. These systems and others offer real-time pavement quality measurement tools and use accelerometers to measure parameters of the compactor’s vibratory signature. Other sensors are also used to gain information about the pavement during the compaction process. 95

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report Information from all sensors is then used to evaluate pavement density or stiffness and quality. The true test of the “intelligent compaction” system, however, is whether it actually saves time (fewer passes), improves uniformity of the mat, and provides accurate, consistent readings. Minnesota DOT, FHWA, and NCAT all recently sponsored demonstrations and workshops on the use of IC rollers. Other agencies where case studies have been completed include Alabama, Florida, Iowa, Maine, Oklahoma, Virginia, Texas, and Wisconsin, to name a few. In addition, FHWA has developed a strategic plan (Horan and Ferragut, 2005) to implement IC technology within the U.S., which includes a systematic procedure to encourage state agencies and industry to expedite the implementation process. Several agencies, including Minnesota, Virginia, North Carolina, Louisiana, Iowa, New Jersey, and Wisconsin, are conducting or planning to conduct field and laboratory based studies to evaluate IC technology and to develop specifications for pilot projects. The Minnesota DOT demonstration at the Minnesota Road Research (MnROAD) test track used the Bomag CCC system and other NDT devices to independently measure soil properties at each test point. The other NDT devices used in the demonstration included the DCP, the Geogauge, and the LWD. In general, it was found that the moisture content in the soil greatly influenced the compaction process and the modulus measurement. The study suggests future demonstration projects on “real-world” construction projects. NCHRP Project 21-09 was initiated in 2006 to determine the reliability of IC systems and to develop QA specifications for the application of IC in the compaction of unbound materials. The study is ongoing and has included an evaluation of three IC systems for soils using field data—Ammann, BOMAG, and Caterpillar. Based on current progress of this study, the findings suggest smooth drum rollers are more reliable than sheet-foot rollers (Mooney et al., 2007). In addition, a change in displacement amplitude of the roller proved to have a larger effect on soft clays than on granular materials. Soils properties such as modulus and the DCP penetration index are correlated to roller output for both subgrade clays and aggregate materials. The field study was comprehensive, and future tasks to be undertaken will result in the development of construction specifications for embankments and granular bases. The following paragraphs summarize the operational characteristics of three of the systems currently in existence. Onboard Density Measuring System The Onboard Density Measuring System (ODMS), a model patented by Pennsylvania State University, offers density measurements in real time at a rate of one per second during the compaction process, thereby affording the contractor and roller operator the opportunity to recognize and correct compaction problems immediately, while maintaining a permanent record of the entire compaction process. The ODMS was developed from the rather simple idea that, the denser the material that the vibratory roller is rolling over, the more excited the vibratory response of the roller. This response is measured by an onboard computer connected to a machine milled accelerometer 96

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report (MMA) attached to the frame of the roller just inside the roller’s damping mechanism. The computer receives the signal from the MMA and uses a FFT to transfer the information into the frequency domain, producing a power spectrum such as the one seen in Figure 20. The computer then integrates the power spectrum into an algorithm that calculates the HMA density at one-second intervals and transports them via radio signal to interested project personnel at a remote computer. Two major aspects of the ODMS separate it from other acoustics-based density gauges. 1. It is the only acoustics-based density gauge that takes physical parameters other than the vibratory response of the roller into account. 2. It does not give a relative density reading, but a direct density reading in pounds-per- cubic-foot. Principle of Operation of ODMS The relationships between established vibration parameters are both unique and convenient. The displacement is written as ∆(t). The velocity of the vibration (V) is the derivative of the displacement (∆) as a function of time; therefore, velocity can be written as: ( ) dt tdtV Δ=)( (26) Acceleration is defined as the rate of change in velocity at a given point in time. Acceleration is the derivative of velocity; therefore, the acceleration can be written as: 2 2 )( dt d dt dVta Δ== (27) Thus, these relationships can be derived, as shown in the following equations: ( ) ( )( )tAtntDisplaceme ωsin=Δ= (28) ( ) ( )( twA dt dtVVelocity ωcos=Δ== ) (29) ( ) ( )( tA dt d dt dVtaonAccelerati ωω sin22 2 =Δ=== ) (30) Where: A = amplitude t = time ω = circular frequency (in radians / sec.) 97

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report STATE ROAD 61 POWER SPECTRUM -80 -70 -60 -50 -40 -30 -20 -10 0 0 36.25 72.5 108.75 145 181.25 FREQUENCY (HERTZ) A C C EL ER A TI O N (d B - ar bi tr ar y) Figure 20. Typical Power Spectrum Produced by the ODMS As the HMA mat is compacted with each successive pass of the vibratory roller, the density level rises, and the effective stiffness increases. The theory that the system is based upon is that the acceleration level of the vibratory response is affected in a repeatable manner. The fundamental relationship between the displacement, the velocity, and the acceleration of a mechanical system is demonstrated above. These established, fundamental relationships allow the restatement of the theory. As shown in Figure 21, the concept of mat stiffness and the presence of damping induced by the mat were introduced. The assumption that the mat contains these properties is symbolized by the traditional spring, shown as stiffness, K, and dashpot, shown as damping, C. The acceleration, a(t), of the compacting vibrations for this study depends on mat stiffness, K, and the damping, C. At very low frequencies, stiffness K is the dominant physical factor affecting a(t). This is shown by equation 31: Fdamping = CV = CAω sinωt (31) Where: V = Velocity of the vibration 98

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report Mechanical Model f(t) M a(t) Revised Mechanical Model C M a(t) Ke f(t) Kp Ku Figure 21. Illustration of the Mechanical Model on which the ODMS was Developed Equation 31 shows how dependent the damping force is on the frequency. Since the frequencies of the vibratory components studied are in a sufficiently low range, the damping, C, can be initially ignored. This theory is a continuation of the original hypothesis, discussed earlier. Mathematically, it can be stated that: Acceleration, a(t) = f(Stiffness, K) As material under the roller becomes denser, the reduction of air voids adds structure to the material, resulting in higher densities and mat stiffness. The theoretical relationship for stiffness can be expressed as a function of HMA density, so acceleration can be expressed as a function of density. Acceleration a(t) = f(Density) During the developmental phase, the system was tested on twelve highway construction projects in four states. In the validation phase, it was tested on two interstate highway construction projects—one in Florida, and one in New York. Both tests were conducted with the system mounted on an Ingersoll-Rand DD-110 vibratory asphalt roller. The results definitely demonstrated the potential use of these roller-mounted systems. The parameters that have a large impact on the accuracy of the system include the temperature and thickness of the HMA mat, the vibratory frequency and amplitude of the compactor, and the strength of the underlying pavement structure. BOMAG Asphalt Manager BOMAG’s system, called the Asphalt Manager, was introduced in 2001 and combines VARIOMATIC technology with a new method for HMA compaction and testing to provide 99

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report an assessment during the compaction process (see Figure 22). BOMAG’s system first calculates the stiffness of the HMA and then ties the stiffness to a density, producing a reading of how much the density of the mat has increased in pounds-per-cubic-foot (BOMAG, 2003, Kloubert, 2002). Actually, Asphalt Manager is a total HMA compaction management system. The density measurement is only one component of that, and it is a descendent of the original Hypac Terrameter. Figure 22. The Asphalt Manager from BOMAG BOMAG reports and hypothesizes that the dynamic stiffness value calculated by the Asphalt Manager, termed EVIB [MN/m²], can be used as a measure for the level of compaction under uniform subgrade stiffness and under consideration of the HMA temperature. EVIB and the Marshall density have been shown to be related to one another. The contact force between the HMA and roller drum together with the vibration path is determined by acceleration measurements taken on the vibrating roller drum. When calculating the contact force over the vibration path of the drum each rotation of the eccentric produces a loading and unloading curve in which the enveloped area defines the compaction work done. As with the plate bearing test used in soils, the dynamic stiffness of the HMA is calculated using the load curve. The cylindrical shape of the drum and the changing contact area of drum and HMA is thereby taken into account. The physical measurement value of HMA stiffness is the vibration modulus, EVIB (BOMAG, 2003). 100

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report The roller operator reads the EVIB value on an analogue display. Since HMA stiffness is temperature-related, the surface temperature is sensed by an infrared measuring unit underneath the cab and also displayed on an analogue gauge. Site experience shows that temperature sensitivity of the EVIB value is between 100 and 150oC (212 and 302oF) and is therefore within reasonable limits. The effect of increasing compaction is very distinct and provides an assessment of the compaction progress. Compaction measurements using a nuclear density gauge show a direct correlation between the EVIB vibration modulus and density—given a uniform and stable base under the HMA layer and taking the mat temperature into account. Continuous Compaction Control Geodynamik has traditionally focused on embankment density measurements but has initiated operations to service the HMA construction community. Currently, its embankment density measuring system produces a compaction meter value that is a dimensionless unit that measures the compaction state of a material, and its absolute value varies with the material’s rigidity. At the end of compaction, a documentation report of both the compaction process and the compaction results is available. The system consists of an accelerometer, a processor, a compaction meter value (CMV), and a frequency meter. Both meters are fed information by the accelerometer. The accelerometer detects the drum's vertical vibrations and transforms them into electric signals. The signals are amplified, filtered, and then sent to the processor via a cable. In the processor, the signals are analyzed and the CMV is calculated. This CMV signal is then sent to the CMV meter and to the compaction documentation system (CDS). CMV is a measure of how much the vibration signal differs from a pure sinusoidal signal. The reason the vibration signal differs from a sinusoidal signal is because the drum hits the ground. If the ground is hard, the impact will be very short in duration and very powerful and, consequently, the distortion of the signal from the A-sensor will be very large. This method of measuring compaction can be considered to be a continuous loading test of the material (i.e., while the drum rolls on, a load test for every impact into the ground). In principle, about 25 to 40 load tests per second are obtained. In order to level out the CMV variations from impact to impact, the processor builds up a moving average that is valid over a given period of time. The overall effect is that the CMV values that the processor sends at a given time is the average of the load test results for the last half second. The system also has an oscillometer. The oscillometer is a patented roller-integrated compaction meter for oscillating rollers. It can be mounted on all types of oscillating rollers of all fabrications. The oscillometer consists of an accelerometer, a processor, an I-sensor (a proximity transducer that produces an electric pulse whenever a metallic object passes by), an OMV (oscillo-meter-value) meter, a roll-speed meter, and an oscillation frequency meter. The operation of the oscillometer is based on the indirect measurement of the reaction force in the horizontal direction brought about as a result of the drum's contact with the ground. This reaction force accelerates the whole roller horizontally. An A-sensor registers this 101

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report horizontal acceleration and transforms it into an electrical signal. This signal is then filtered, amplified, and sent to a processor unit via a cable. In the processor, the signal is analyzed and the OMV is computed and sent to an OMV-meter or to a CDS-system. The analysis includes the computation of the maximum reaction force into the material during the oscillation of the drum. This reaction force increases with increasing rigidity of the ground in a specific way, provided all the other factors affecting the whole system are kept constant. The friction between the material and the oscillating drum is not big enough to keep the drum and the material in contact during the whole oscillation. Instead, there is always some gliding between the drum and the ground. The processor takes this into consideration and uses only that part of the signal where the gliding between the ground and the drum does not occur. The values produced by this process correspond to the reaction force that would have existed if the friction were big enough to prevent gliding between the drum and the ground. This method of measurement can be considered as equivalent to a continuous dynamic loading test of the ground while the drum rolls. The compacto-meter analyses loading tests with vertically directed loads, as the oscillometer uses horizontally directed oscillating loads. In principle, even here, 25 to 40 load tests per second are obtained. In order to level out the variations between cycles, the processor builds up a moving average. The overall effect is that the CMV values that the processor sends at a given time is the average of the load test results for the last half second. The density measurement is a function of the CMV and OMV values. Before using the continuous compaction control device, it must be calibrated using a traditional density measuring system. Calibration instructions for the system are very detailed and the calibration procedure continues until no considerable rise in compaction occurs or when a double jump occurs for the first time. After the compaction for purposes of calibration is finished, there is a six-step procedure to complete the calibration process. 3.11.2 Non-Roller-Mounted, Non-Nuclear Electric Devices The four non-roller-mounted systems that show the most promise at this point in time are the PQI, manufactured by TransTech; the PaveTracker manufactured by Troxler; the EDG; and the Purdue Time Domain Reflectometry (TDR) method. The PQI and PaveTracker devices are used for measuring the in place density of HMA, while the EDG and TDR devices are used to measure the in-place density and water content of unbound aggregate base materials and embankment soils. The use of the non-nuclear density gauges has received much attention over the past 5 years because of the increased regulations and safety issues with the use of nuclear density gauges. Studies conducted by Hausman and Buttlar (2000), Henault (2001), and Romero (2002) found poor correlations between the HMA density devices and the results from cores and nuclear density gauges. Most studies have found that the results from these gauges are related to changes in the material density, but an absolute density value is not reported by any of the 102

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report gauges. More importantly, the gauges need to be calibrated for the same material being placed. In addition, most studies have recommended that the devices can be used for controlling the materials placement, but that they should not be used for acceptance and establishing payment. A more recent study, (Schmitt et al., 2006), that used significantly upgraded and improved non-nuclear density devices, recommended its use for routine QA and established a procedure for calibrating the device. The recommendations were, however, based on companion nuclear gauge readings, a baseline selected to reflect the agency’s current practice. Unbound Materials and Soils Electrical Density Gauge—EDG The EDG was developed for measuring the density and moisture content of soils and other unbound materials using capacitance. This device has not been used to-date for testing HMA layers. The EDG is relatively new and has not been used by many agencies. Nonetheless, those agencies that used the device (for example, Nevada DOT) were satisfied with its performance and recommended it for continued use.7 The biggest advantage of using the EDG and similar devices is its safety and accompanying lack of regulation and required licenses, as compared to nuclear density gauges. EDG, LLC has 10 EDG beta units and have made these units available to various highway agencies and consultants. A user’s manual is available for this device which describes its operation and calibration.8 In addition, a draft test standard has been prepared and submitted to ASTM D-18.08 (Soils and Rocks for Engineering Purposes) for balloting purposes. The unit weighs 11 pounds (5 kg) and is 13.5x12x6 inches (343x305x152 mm) in size. It is placed at the location of interest by the technician. While in operation, the unit emits far less energy than does a cell phone, since the EDG uses radio frequency measurements to measure the density of the material. The current beta units include “soil darts” that are driven into the material and act as probes as well as controlling the sample size. The EDG and all its accessories are shown in Figure 23 (EDG, 2003). A benchmark or calibration is required for each soil and material being compacted to estimate the density and moisture content of the compacted material. This calibration is a moisture-density relationship prepared in accordance with ASTM D698 or ASTM D1557 or other similar methods. The EDG is then used to determine the electrical conductivity through the different conditions of moisture and density. A relationship is developed between the electrical conductivity and actual volumetric values of the material or soil. The calibration can also be completed during construction by varying the compaction effort (number of roller passes) and measuring the density with the sand-cone method or other devices within each area with different compaction levels and then preparing a relationship between the EDG readings and the actual density and moisture content of the material over the expected range of values encountered in the field. 7 Density data on a sand/clay soil measured with the Electrical Density Gauge provided by Ali Regimand with Instro Tec, Inc. and Dennis Anderson with Anderson Resource Associates, Inc. 8 User’s Manual, P/N 9093927; Electrical Density Gauge for Compacted Soil, Electrical Density Gauge, LLC; Carson City, Nevada. 103

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report Figure 23. Photo of the Electronic Density Gauge Purdue TDR Device The Purdue TDR method measures both the density and water content of unbound embankments and other fill materials. The system consists of a TDR device, a coaxial cable, a coaxial head, either a coaxial cylinder or multiple probe arrangement, and a portable computer interface. Figure 24 shows the parts of this system and the gauge in operation. The procedure is currently being marketed by Durham Geo. This system is similar to the EDG, is relatively new, and has yet to be used or evaluated by many agencies. Other Devices for Estimating Moisture Content There are other devices where commercial equipment is available for measuring the water content of unbound materials and soils, such as the Field Moisture Oven (FMO 200) that is manufactured by Kessler Soils Engineering Products and the Speedy Moisture Testing Kit manufactured by Humbolt Manufacturing. The Field Moisture Oven measures the water content in accordance with ASTM D 4959 (Determination of Water (Moisture) Content of Soil by Direct Heating Method), while the Speedy Moisture Testing kit measures water content in accordance with ASTM D 4944 (Field Determination of Water (Moisture) Content of Soil by the Calcium Carbide Gas Pressure Tester Method). Although these devices can be used in the field, they do not actually measure the water content of the in-place soil. Soil must be sampled and removed from the layer, unlike the EDG and Purdue TDR devices that make the measurements in place without physically sampling the soil. Thus, these devices do not meet the definition of NDT devices, as used within this project. 104

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report Figure 24. Photo of the Purdue TDR Method (courtesy of Durham Geo Website) HMA Mixtures PaveTracker Device Troxler has long been the leader in nuclear density gauge technology. More recently, Troxler has developed a non-nuclear field density measuring device called the PaveTracker (Model 2701). This particular instrument, unlike Troxler’s nuclear gauges and unlike the other non- nuclear gauges covered (PQI and EDG), is not made to be lifted by the operator and placed on the location of interest. The PaveTracker is shown in Figure 25 (Troxler, 2003). This unit is the lightest of the three models reviewed, at 2 lb (0.9 kg). It gives a reading every second and displays density in pounds-per-cubic-foot. Unlike some non-nuclear, non- mounted gauges, this model needs no moisture or temperature corrections, and the 3-foot (0.9 m) telescoping handle allows the operator to slide the gauge into position and reduces bending. Presently, Troxler recommends this instrument as a QC tool, but in time, it may advance to the point of being considered for acceptance purposes. 105

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report Figure 25. Photo of the Troxler PaveTracker A unique feature of the PaveTracker is its very small size, with dimensions of 3.5x4.5x2.25 inches (89x114x57 mm). Since it is so small, the PaveTracker can be used in the laboratory for calibration by placing it top of a 6-inch (150-mm) gyratory compacted specimen. The instrument is able to probe to depths of 1.75 inches (44.5 mm) and demonstrates a repeatability of +/- 0.5 units. Pavement Quality Indicator Device—PQI The PQI is a non-nuclear density gauge for instantaneous, in-situ measurement of asphalt pavement density, invented and manufactured by TransTech Systems, Inc. PQI is a lightweight device (under 16 lb [7.3 kg]), easy to use, no special licensing requirement, and can provide density data in several seconds (see Figure 26). PQI measures pavement density by measuring the electrical impedance of the material (see Figure 27). A toroidal electrical sensing field is established in the paving material and measured via a flat sensing plate. The measured electrical impedance is a function of the composite dielectric constant, which is further related to the density of the paving material. An embedded computer is used to determine the density of the paving material, and perform calibration and correction functions (TransTech, 2003). 106

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report Figure 26. Photo of PQI Device by TransTech Figure 27. Operational Theory Schematic of the PQI 107

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report TransTech started the work on the PQI in 1995 under an agreement with the New York State Energy Research and Development Authority, and later got support from FHWA and AASHTO, delivered through the NCHRP under the IDEA program. Now the PQI Model 300 has been widely evaluated; the latest PQI Model 301, with the ability to compensate for surface water, has been commercially available (TransTech, 2003). By January 2003, about 400 units of PQI have been sold to more than 10 countries. Many evaluations have been made on the use of PQI. But the conclusions are not consistent. Several research projects reported accurate and reliable results and suggested the use of PQI for QC on asphalt paving (Allen et al., 2003; Rogge et al., 1999; Sully-Miller, 2000). An FHWA five-state pooled fund study provided similar results (TransTech, 2003). But Henault (2001) reported a poor correlation between PQI density and core density, conditioned by the presence of moisture introduced into the asphalt during rolling operations. One of the most thorough studies performed on the PQI was done by the Kentucky Transportation Center (KTC) at the University of Kentucky. The study took two PQI units and one Troxler thin-lift nuclear density gauge out to an ongoing construction project and let the contractor operate one PQI (HHR PQI), the KTC operate the other PQI (KTC PQI), and the DOT inspector operated the nuclear density gauge (TMTL) and took cores—all on the same project. Results of this study are summarized in Table 21. Table 21. Summary of Density Comparisons Density Measurement HHR PQI KTC PQI TMTL Cores Number of samples 735 740 453 149 Average (pcf) 144.4 143.4 142.3 144.1 Standard Deviation (pcf) 4.91 3.52 4.15 2.72 The following are quotes from the study’s reported conclusions: • The standard deviations of the density readings of all the gauges were greater than the standard deviations of the density readings of all the cores. This indicates that there was more scatter in the data from each of the gauges than from the cores. • There was no significant difference between the mean density of the HHR PQI and the mean density of the cores. However, there was a statistically significant difference between the mean density of the TMTL and KTC PQI density gauges and the cores. • The density distribution of the HHR PQI gauge most closely matched the distribution of the cores with an 88% overlap in the distributions. The TMTL gauge and the KTC PQI gauge had overlaps in their density distribution functions with the density distribution function of the cores of 83% and 78%, respectively. This information indicates that the HHR PQI gauge, not only in the mean density, but also in the overall distribution of readings, most closely approximated the results of the cores. 108

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report • If pay factors were determined from gauge densities, then using the densities provided by the TMTL gauge would have resulted in a five-percent reduction in overall pay for lane densities. One hundred percent overall pay would have resulted from using the two non-nuclear density gauges. The report went on to recommend that the PQI gauge be approved for use as a QC device. This is because the ease of operation of the two gauges (Troxler Model 4640-B and PQI) are similar, and because the gauge that most closely approximates the data from the cores (both by comparing the means and the distributions) was the HHR PQI non-nuclear gauge (Allen et al., 2003). 3.12 Surface Condition Measuring Systems and Devices The quality of the surface condition of the pavement includes the measurement of ride quality or surface profile, surface texture, noise, and friction or skid. Each surface property or characteristic is briefly discussed below in relation to QA applications. 3.12.1 Surface Texture Several nondestructive methods have been developed to measure pavement texture. The pavement’s surface texture is a function of and can be defined by surface wavelengths. The surface wavelengths can be separated into two groups, microtexture and macrotexture. The microtexture wavelengths are defined in a range of 1μm to 0.5 mm, while the macrotexture is defined by wavelengths from 0.5 mm to 50 mm. Microtexture provides a gritty surface to penetrate thin water films and produce good frictional resistance between the tire and the pavement. Macrotexture provides drainage channels for water expulsion between the tire and the pavement, thus allowing better tire contact with the pavement to improve frictional resistance and prevent hydroplaning. Measurements of macrotexture may also indicate pavement uniformity or lack of segregation. This potential use of identifying segregation in HMA mixtures would be applicable to QA use. Currently there is no system capable of measuring microtexture profiles at highway speeds. Therefore, microtexture can be evaluated by using pavement friction at low speeds as a surrogate. The classic measure of pavement macrotexture is a volumetric method, typically referred to as the “sand patch” method, ASTM E965 (ASTM, 1999). The sand patch method, while historically used, is time intensive and operator dependent (Henry, 2000). Thus, it will not be discussed further. On the other hand, with the significant advances that have been made in laser technology and data processing, systems are now available to measure macrotexture at traffic speeds. There are three types of laser based systems to measure pavement macrotexture. • MGPS commercial version of the Road Surface Analyzer (ROSAN). • Circular track texture meter (CT Meter). • Various proprietary systems using inertial profilers with laser based sensors. 109

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report All three systems use laser range finding technology. As shown in Figure 28, a photosensitive diode measures reflections from a pulsating laser diode source. The MGPS and inertial profiler systems are linked with precision electronic distance measuring systems (McGhee and Flintsch, 2003). Both systems also use accelerometers to account for vehicle body movement. The systems are capable of continuous or semi-continuous measurements at high speeds. Pulsing Laser Light Source Photosensitive diode array Photosensitive Diode Array Road surface Road Surface Pulsing laser light source Figure 28. Schematic of Laser Sensor (Stroup-Gardiner and Law, 2000) The MGPS system is an outgrowth of FHWA’s ROSAN project. The MGPS high frequency laser is focused to a smaller diameter, making it more suitable for texture measurements (McGhee et al., 2003). The use of the MGPS system to detect segregation was reported in NCHRP Report 441 (Stroup-Gardiner and Brown, 2000). The MGPS can be mounted on the bumper of most vehicles. The MGPS system measures mean profile depth (MPD) according to ASTM E1845. McGhee and Flintsch (2003) found a good correlation between the MPD measured by the MGPS system and the MTD measured by the sand patch test. Inertial profilers typically are used for measuring pavement smoothness. However, several companies (ARAN, Australian Road Research Board, Dynatest, Greenwood Engineering, International Cybernetics, and WDM) produce vehicle-mounted inertial profilers with high frequency lasers and software for estimating pavement texture. The CT Meter is a stationary device with a laser displacement sensor mounted on a rotating arm. The arm rotates in a circular path with a diameter of 284 mm and takes measurements every 0.9 mm (ASTM, 2005). The device calculates a MPD according to ASTM E2157. Several studies have shown good correlation between MPD measurements with the CT Meter 110

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report and MTD measurements made with the sand patch test (Abe et al., 2000; McGhee and Flintsch, 2003). Segregation is a major problem in the placement of HMA layers. NCHRP 441 (Stroup- Gardiner and Brown, 2000) defined segregation as: “a lack of homogeneity in the HMA constituents of the in-place mat of such a magnitude that there is reasonable expectation of accelerated pavement distress.” NCHRP 441 identified both gradation and temperature related segregation. Historically segregation has been identified visually, assuming that segregation is confined to the surface. Visually identified areas could then be cored and the extracted gradations compared with the job mix formula or control areas. NCHRP 441 proposed using texture ratios determined from the MGPS system to identify segregation (Stroup-Gardiner and Brown, 2000). The ratio of the area in question to a non- segregated area defines the texture ratio. In lieu of identifying a non-segregated area, the texture depth may be estimated from equation 32 (Stroup-Gardiner and Brown, 2000; McGhee et al., 2003): )(004861.)(1038.0 )75.4.(%004984.0)..(max01980.0 uc CC mmpasssizeaggETD ++ −= (32) Where: ETD = Estimated texture depth in mm. Max. Agg. Size = smallest sieve size with 100 percent passing. % pass. 4.75 mm = Percent passing the 4.75 mm sieve. Cc = Coefficient of curvature = (D30)2 / (D10 D60). Cu = Coefficient of uniformity = D60/D1.. D10 = Sieve size, in mm, with 10 percent passing. D30 = Sieve size, in mm, with 30 percent passing. D60 = Sieve size, in mm, with 60 percent passing. Included with NCHRP Report 441 was a draft AASHTO test method that includes the proposed texture ratios shown in Table 22. Research by McGhee et al. (2003) suggests that these limits may not be applicable to large stone and gap graded mixes, such as Stone Matrix Asphalt (SMA) or Open-Graded Friction Course (OGFC). McGhee et al. (2003) examined two equations for predicting the ideal texture based on the job mix formula, acceptance bands based on the AASHTO Implementation Manual for Quality Assurance and empirically established target standard deviation levels. The authors felt that this last approach could hold some promise for future use in QA application. However, this system and equipment cannot identify segregation at the bottom of an HMA lift. 3.12.2 Noise and Skid The measurement of noise and skid along a pavement’s surface are becoming important regarding the performance of flexible pavements. NCAT recently developed equipment and test procedures to measure both of these parameters. However, these test methods and parameters have yet to be used in any QA format. 111

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report Table 22. Proposed Texture Ratios Corresponding to Various Levels of Segregation (Stroup-Gardiner and Brown, 2000) Level Upper Limit Lower Limit Non-segregated ETD*1.15 ETD*0.7 Low Segregation ETD*1.56 ETD*1.16 Medium Segregation ETD*2.02 ETD*1.57 High Segregation >ETD*2.02 NA 3.12.3 Ride Quality Pavement ride quality is the key to user satisfaction. Achieving an appropriate level of smoothness is therefore a strategic goal for highway agencies. Studies have further validated that pavements built smooth remain smooth longer and generally exhibit better performance and serviceability. ASTM defines pavement roughness as the deviations of a pavement surface from a true planar surface with characteristic dimensions that affect vehicle dynamics, ride quality, dynamic loads, and drainage. Several NDT techniques have been developed to measure the pavement profile to determine pavement roughness and texture. The profile-measuring devices can be classified as: 1. Response-type road roughness measuring (RTRRM) devices that measure the movement of an axle with respect to the vehicle or trailer frame. 2. Inertial profilers that employ an accelerometer and vertical measuring device to measure the “true” profile of the pavement surface, for a range of wavelengths, at highway speeds. 3. Profilographs that record deviations of the pavement surface from the plane of a rolling straightedge. 4. Inclinometer and manual devices that measure the slope from one point to the next or the elevation of each point as the unit is moved along the pavement. During the last decade, FHWA developed the ROSAN, briefly described earlier. This is a laser-based profiler and is capable of measuring longitudinal texture and pavement profiles at highway speeds. More importantly, the inertial profilers, profilometers, and profilographs are already included in many agencies QA programs for acceptance. In addition, none of these devices result in an estimate of the structural or volumetric properties of HMA. 112

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 113 3.13 Summary of NDT Technologies for HMA Pavement Evaluation Table 23 summarizes those NDT technologies and methods that have been used to measure different properties and features of flexible pavements. Chapter 4 of this report details the evaluation of these methods in regard to their potential for application to QA. As tabulated, the GPR has been used for estimating many more properties than any other NDT technology for the volumetric properties, while the seismic technologies have been used more extensively for estimating the structural properties. Table 23. Summary of NDT Methods Used to Measure Properties and Features of Flexible Pavements In Place NDT Technologies and Methods Type of Property or Feature HMA Layers Unbound Aggregate Base and Soil Layers Density GPR, PQI, PaveTracker, ODMS GPR, EDG, Purdue TDR Air Voids or Percent Compaction GPR, Infrared, Acoustic Emissions, Roller-Mounted Density Devices GPR, Roller-Mounted Density Devices Fluids Content GPR GPR, EDG, Purdue TDR Gradation; Segregation GPR, Infrared, ROSAN NA Volumetric Voids in Mineral Aggregate GPR (Proprietary Method) NA Thickness GPR, Impact Echo, SPA, SASW, Magnetic Tomography GPR, SASW, SPA Modulus; Dynamic or Resilient PSPA, FWD, LWD, SASW, Asphalt Manager & Other Roller-Mounted Response Systems DCP, Clegg Hammer, DSPA, SPA, SASW, FWD, LWD, GeoGauge, Roller- Mounted Response Systems Structural Bond/Adhesion Between Lifts SASW, Infrared, Impulse Response NA Profile; IRI Profilometer, RTRRM, Inertial Profilers NA Noise Noise Trailers NA Functional Friction CT Meter, ROSAN NA

Page intentionally left blank.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report CHAPTER 4 NDT TECHNOLOGIES FOR APPLICATION TO QUALITY ASSURANCE Multiple NDT technologies and devices have been used to varying extents by agencies in the U.S. for pavement evaluation, forensic studies, and construction evaluation. Several of these agencies were contacted to ask about their specific use of these NDT devices for potential application within their acceptance plans and the test methods associated with them. These contacts were considered essential to evaluate the various NDT technologies and determine their practical application to QA practices. Typically, the emergence of a new technology will initiate a research and evaluation effort by an agency before the process is accepted for use as a good engineering tool. This chapter presents an evaluation of the NDT technologies and provides the reasons and justification for selecting specific NDT methods with potential use for judging the quality of flexible pavement construction. 4.1 Evaluation Factors and Topics As noted in chapter 3, a number of NDT technologies and inspection systems provide information on the quality of the material without altering or damaging the materials being tested. Utility analyses have been completed regarding the potential use of NDT methods and devices for measuring critical pavement properties and features (Von Quintus et al., 1995; Saeed et al., 2001). These analyses were used to rank selected NDT methods and devices for measuring important properties and features of pavements. The factors used in those analyses were considered in this study. However, this evaluation was focused towards specific requirements related to QA programs referred to in the Research Problem Statement, specifically: This will lead to increased measurement of layer moduli by owner agencies, an activity that is not at present a typical component in the acceptance of a completed project. The results will identify NDT technologies ready and appropriate for implementation in routine, practical quality control and acceptance operations. Thus, those NDT technologies and methods were evaluated for their ability to accurately measure QA properties that are strongly correlated to performance, including estimates of layer moduli, and can be used in routine, practical QA operations by contractor and agency personnel. Those layer properties that have been found to be strongly correlated to performance and included as inputs to the MEPDG are dynamic and resilient modulus, density, air voids, fluids content, and gradation. Field permeability, although an important characteristic of HMA mixtures, was excluded from this evaluation, as explained in chapter 2 115

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report of this report. The following lists the factors used to evaluate specific NDT devices for inclusion into a QA program. 1. Accuracy and precision of the test equipment and protocols in measuring a specific material property—one of the difficulties of this category is defining the target value of some properties for nonlinear and viscoelastic materials. The accuracy and precision of the technology also are tied to the data interpretation procedures. 2. Data collection guidelines and interpretation procedures—this category includes whether there are generalized guidelines and procedures available for performing the tests and analyzing the data to estimate the material properties and/or features. 3. Availability of standardized test procedures (test protocols)—this category includes whether there is a test standard available for use in collecting NDT data to estimate the required material properties and features. 4. Data collection—production rate of the NDT equipment in collecting the data. 5. Data interpretation—time and ancillary equipment/software required to analyze and interpret the data for estimating the specific layer property (see Tables 15 and 23). 6. Cost of the equipment—this category includes the initial cost of the test equipment, additional software and hardware requirements necessary to perform the test, and the operational and maintenance costs, including calibration. 7. Complexity of the equipment or personnel training requirements. 8. Ability of the test method and procedure to quantify the material properties needed for QA, mixture design, and structural design (see Figure 2). In other words, is the NDT test result applicable to mixture and structural design? 9. Relationship between the test result and other traditional and advanced tests used in mixture design and structural design. Calibration of field and laboratory equipment is important to reduce the error and variability in the test results. However, calibration of the system is often overlooked or confused with equipment calibration in many QA projects. Figure 2 identified some of the calibration steps in the systems approach to ensure that the flexible pavement will meet the design expectations—reducing fracture, distortion, and disintegration distress over the design period. Stated simply, there can be a bias between the structural and mixture design tests, and those tests used for QA. For example, a good modulus (seismic design modulus or dynamic modulus) for one project might be an inferior value for another project under different conditions. Calibration should eliminate or account for this bias for project success and a reduction of contractor disputes. 116

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report The advantages and limitations of the equipment and data processing techniques relative to QA application, as well as for measuring material properties needed for forensic studies and rehabilitation design, are included in this chapter. For example, the Florida, Texas, and Washington DOTs routinely use various NDT methods for evaluating flexible pavement construction, but not necessarily for accepting flexible pavement construction. 4.2 Data Sources for Evaluation Several highway agencies were contacted to collect information on their practices and their experiences with using NDT devices. Research reports of several agencies were also reviewed. These agencies include Arizona, California, Connecticut, Florida, Georgia, Illinois, Maryland, New Hampshire, Minnesota, Mississippi, Missouri, Nevada, Ohio, Oklahoma, Pennsylvania, Texas, Virginia, Washington, and Wisconsin DOTs, the FHWA, FHWA Eastern Federal Lands Division and Central Federal Lands Division, U.S. Air Force, U.S. Army Corps of Engineers Engineer Research and Development Center, Loughborogh University, Nottingham Trent University, TRRL, University of Illinois, University of Mississippi, Louisiana State University, Worcester Polytechnic Institute, and Texas Transportation Institute. Appendix A lists the specific topics and items discussed with these agencies. The following questions were also asked to address important issues and barriers related to the use of NDT technologies for QA. These questions are in addition to the issues already noted in chapter 2: • What types of NDT methods and devices have been used? • For what purpose of application have these methods been used? • How long have these NDT technologies and methods been used? • What are some of the advantages and disadvantages of the NDT technology? • What is the operational cost of the equipment? • Are there test protocols for the specific NDT technologies that have been used? • What is the repeatability and variability in the measured response? • What criteria have been used for interpretation of the data collected? • What properties and features have been used for acceptance? • What properties have been used for process control? Some of the equipment manufacturers and suppliers were also contacted to obtain specific information and data on the items listed above. The manufacturers contacted include Olson Engineering, Blackhawk, GSSI, Transtech Systems Inc., and others. In addition, the utility analyses completed by Von Quintus et al. (1995) and Saeed et al. (2001) were reviewed for use in this evaluation. The agency/contractor contacts and literature reviews were used in the evaluation of those NDT methods that have been used for testing pavements. The information obtained from the surveys and relevant literature for each topic was synthesized and is presented in the remainder of this chapter for each NDT device. 117

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 4.3 Deflection-Based Methods—FWD and LWD The deflection-based device in widespread use in North America is the FWD (see Figure 6 in chapter 3). This device has been used in the U.S. since 1978. Accordingly, it is believed that the FWD is well suited to QA applications, even though its use for QA has only recently begun to emerge. Most of the 150+ FWDs in service today are devoted to pavement rehabilitation design or pavement management purposes. The other deflection measuring devices discussed in chapter 3, the RDD and RWD, are not suited for QA applications at this point in time, because they are considered to be in the research and development stage. These deflection-based devices are not considered state-of- the art. The LWD is considered to be a manual or portable form of the FWD and was considered for use as a supplemental device to the FWD (see Figure 8 in chapter 3). Some of the earlier and most recent efforts for using the FWD for QA have used backcalculated and forward-calculated layer moduli because these values can be directly tied to the inputs required by the MEPDG and other M-E based methods for unbound and HMA layers. To demonstrate the practical use and effective application of deflection data for use in control or acceptance plans, the deflections measured with an FWD on a select granular fill material from a construction project in Oklahoma were used. Deflections were measured along a project at random locations along three lines—parallel to the centerline. For this example, a control chart was prepared using the forward-calculated layer modulus to determine whether the construction process to place and compact the select fill is in control or out of control. Figure 29 illustrates the control chart prepared using the calculated elastic modulus values from the measured deflections. As shown, the elastic modulus is changing from one end of the project to the other, with the stiffer material occurring in a localized area. The design elastic modulus for this project for the select fill was 12,000 psi (82.7 MPa). The control chart would indicate that the agency received a higher quality or modulus-material than assumed in design. Research conducted under NCHRP Project 10-48 by North Carolina State University (Kim et al., 2000) presented a method for assessing pavement layer condition on the basis of surface deflection data obtained from the FWD. The uniqueness of this study is that it attempted to correlate “along project” variability in deflection basin characteristics (e.g., shapes of basins, magnitudes of deflections) directly with pavement distress parameters (e.g., rutting, cracking). Kim et al. did not use backcalculation programs because of their inherent assumptions and biases, adding subjectivity and complexity to the correlations. Data from the LTPP program and those collected by state DOTs were used for database development. A sophisticated, commercial finite-element program was used to develop the required pool of theoretical pavement responses. Damage indicators such as the deflection basin parameter (DBP), effective modulus, base curvature index (BCI), subgrade compressive strain, and subgrade stress ratio (SSR) were developed. These indicators, derived from raw deflection data, were deemed to correlate strongly with the pavement layer 118

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report performance (e.g., subgrade rutting potential is strongly correlated with subgrade effective modulus, BCI, subgrade compressive strain, and SSR). Models were developed to relate the pavement condition indicators created with surface deflection information available using both regression analysis and artificial neural networks. The entire methodology including processing of raw deflection data, pavement condition indicator calculations, and condition assessment criteria was coded into a software program designated as APLCAP (Asphalt Pavement Layer Condition Analysis Program) for ease of implementation. Figure 29. Example of a Control Chart for the Elastic Modulus Calculated from Deflection Basins on an Unbound Granular Embankment (Select Fill) Material Statistical Control Chart for Elastic Modulus 0 10000 20000 30000 40000 0 20 40 60 Lot Identification El as tic M od ul us , p si Series1 To date, however, the NCHRP Project 10-48 methodology has demonstrated limited success for HMA pavements with granular bases and full-depth HMA pavements. It could not be extended to HMA pavements with cement-treated bases or to HMA overlays of PCC pavements. Even for cases where the methodology was successful, it was only able to establish a relative assessment of condition by comparing the section under evaluation with an “ideal” or “good” section. The methodology, in its present form, does not relate the difference between current and expected performance to specific distress types that may be prevalent in the pavement layer being examined or other reasons. Agency Use or Adoption Most states surveyed responded that they use the FWD as an engineering tool for research, forensic studies, and rehabilitation designs. The Mississippi DOT recently sponsored a study to correlate subgrade moduli calculated from FWD deflection basins measured on the subgrade to moduli measured with LTPP TP46 resilient modulus test protocol and DCP tests. 119

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report The subgrade moduli were calculated using forward-calculation techniques and correlated to the DCP and laboratory measured values (George et al., 2003). The moduli calculated from FWD deflections were found to be related to moduli estimated from the DCP penetration rate and resilient modulus values measured in the laboratory. The correlation, however, was dependent on soil type. The Mississippi DOT has yet to adopt or include this method into their acceptance plan or QA procedure. The Florida DOT is tracking subgrade moduli using their own in-house forward-calculation method and the current AASHTO method using FWD tests. These methods use the deflection basins rather than the deflection measured at a single sensor offset from the loading plate. The AASHTO and Florida DOT methods, as well as backcalculation methods, over-estimate laboratory derived resilient modulus values. AASHTO recommends that the subgrade modulus values calculated from deflection basins be adjusted by a C-factor of 0.33 (AASHTO, 1993). Von Quintus and Killingsworth (1998) confirmed the use of the C-factor in relating elastic moduli calculated from deflection basins to those measured in the laboratory at comparable stress states for many of the LTPP test sections. However, they found that the C-factor was not constant, but pavement and layer dependent. NCHRP Project 20-50(09) (Stubstad, 2002) investigated the efficacy of utilizing FWD tests measured during construction of the Special Pavement Study (SPS)-1 projects within the LTPP program. Deflection basins were measured on the base (unbound or bound) and surface layers during construction. The study showed the potential for QA application using the FWD and forward-calculation techniques (see Figure 29). These results are also applicable to PWL approaches, because so many tests can be conducted within an individual lot. More recently, the Danish Road Directorate (Road Institute) conducted a study (Hildebrand et al., 2003) evaluating three versions of the LWD and comparing the LWD results with those from the FWD and Static Plate Load (SPL) tests conducted on a cohesive fine-grained (clay) subgrade. Traditional density tests (sand cone and nuclear gauge densities along with moisture-density relationships for compaction control) were also conducted on the subgrade soil. The main goal of this research was to determine whether SPL test equipment and analysis procedure could be replaced by the FWD or LWD. Based on the composite modulus, EO, as calculated from the center deflection, the results supported the hypothesis that there is a good correlation between the different types of equipment. This single test section found that the FWD produces EO-values that were almost identical to those determined through SPL tests, while the three LWDs produced similar results. The test results indicate that the LWD from Keros Technology and, to a lesser degree, the LWD from Loadman, have a reasonable relationship to the FWD and SPL test results conducted on the subgrade soil. The EO-values from the Zorn LWD were lower than those derived from any other piece of equipment included in the test program. The study concluded that the FWD and LWD can play an active role in QA procedures along new construction projects. The United Kingdom has conducted substantial research in recent years in their move towards developing PRS for foundations materials (Fleming et al., 2002; Frost et al., 2001). 120

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report The research evaluated several in-place testing devices for a direct measurement of stiffness (or resilient modulus) during construction. The devices included in the research studies were the FWD, the German Dynamic Plate (GDP), similar to an LWD, the TRL Foundation Tester (TFT), and the Soil Stiffness Gauge (SSG; now referred to as the GeoGauge). The research found that the stiffness values determined from the various test devices were significantly different, but showed similar trends in the data. A reason for the disparity is the stress dependency of nonlinear, unbound materials. The different devices apply different load levels and load pulse-durations, and use different transducers and mounting devices. Miller (2006) also reported similar results in his master thesis work at the Colorado School of Mines. The GDP device included in the United Kingdom study consistently predicted lower stiffness than the other devices. The modulus values from each test was verified and correlated to FWD results. Interrelationships between the indices measured from each device were found to be site-specific. TFT test results were comparable to the FWD, being within 20 percent of the FWD results. The data also indicated that the stiffness modulus from the SSG was about 1.3 times that from the GDP, but with more scatter in the data. The data showed less scatter for sections with thick subbase. It was recommended that any specification must account for the expected variability in the stiffness modulus from one point to another. Test Protocol and Data Collection Guidelines The FWD test protocol for control or acceptance testing is much the same as the normal test protocols used for pavement rehabilitation design. Test spacing depends on the length of a particular project (for example, 10 meters or 25 feet, per lane), and the applied load has to be adjusted to realistic levels, with lower stress levels applicable to unbound layers. In addition, setting drops must be used prior to collecting the deflection basin data for data analyses. More importantly, a statistically significant sample of test results must be available in order to use PWL or other statistical approaches. The test protocols recommended for initial use in QA operations are those developed in support of the FHWA-LTPP program and standardized through ASTM. For forward-calculation of HMA layers, the first three sensors must be placed at LTPP’s protocol positions of 0, 8, and 12 inches, respectively. In addition, more drops of the FWD weight should be carried out for unbound materials than for bound layers, because of the increased drop-to-drop variability associated with FWD tests on unbound materials. Interpretation of Test Data and Determination of Material/Mixture Property FWD load-deflection test data are processed through automated spreadsheet equations or macros. The end result is a set of spatially variable pavement layer moduli, using forward- calculation techniques that are not subject to the “art” of backcalculation procedures. Variables that are subject to input errors are the quality of the FWD data itself (with more testing problems associated with unbound layer tests) and the layer thickness of the bound pavement layer when that layer is under investigation. For HMA pavements and overlays, the mid-depth temperature of the bound layer is also an important factor. 121

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report Accuracy, Repeatability, and Reproducibility Apart from tests on unbound materials, the accuracy, repeatability, and reproducibility of the FWD are well documented in the literature. For tests on unbound materials, the accuracy and precision are somewhat lower, depending on the plate configuration (solid, split, segmented), the drop load, the smoothness or evenness of the tested surface, and the nature of the unbound material (cohesive vs. non-cohesive, etc.). Calibration Requirements FWD calibration should be carried out in accordance with the FHWA-LTPP protocols or other equivalent methods and procedures. It is also important, since the entire deflection basin is utilized, to ensure that the sensors are positioned properly for subsequent data analysis. The SLIC program (developed under LTPP) can be used for this purpose. For HMA pavements and overlays it is critical that the infrared sensor mounted on the FWD is calibrated and providing accurate data to produce values that are consistent with the use of surface temperatures determined through the BELLS procedure. Production Rate The FWD production rate is between 1 to 2 minutes per test point, plus set-up time. This time is dependent on the pavement being tested, whether the layer is bound or unbound, and the number of drops that are used. Some states report about 5 minutes per test location when accounting for the time to move from one test point location to the next. Initial Cost, Maintenance, and Complexity of the Equipment and Data Interpretation Procedure – Operator Technical Requirements The initial cost of an FWD is between $100,000 and $150,000 for a new machine, excluding the tow vehicle. Maintenance can be considered to be between 10 and 20 percent of the initial cost, per year, with increasing costs as the equipment ages. Most states have to arrange for an out-of-state calibration. The equipment is well known and is not complex to operate (although it is somewhat complex to maintain). Data interpretation procedures are very easy, by virtue of newly developed forward-calculation techniques. Results from these forward-calculation techniques, however, have yet to be compared to the values measured in the laboratory, at least for a diverse range of soil types. Results from the back-calculation methods have been correlated to laboratory measured resilient modulus values (Von Quintus and Killingsworth, 1998). Advantages The advantages of using the FWD for QA purposes include widespread use, availability, and support from equipment manufacturers. The FWD is also easy to use in a production environment, allowing good coverage for the PWL method to be used for acceptance. In addition, the FWD can be used to test the final flexible pavement structure immediately after construction, rather than just individual layers during construction. It also provides loads that are compatible with the expected range of loads applied by trucks. Conversely, the LWD uses much lighter loads and needs to be adjusted to laboratory values. The forward-calculation methods developed for QA when using the FWD are easy to use and result in unique layer moduli. Backcalculation procedures traditionally used for pavement 122

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report evaluation are less likely to be used in acceptance plans because it would be difficult to defend non-unique layer moduli in disputes with the contractor. Disadvantages and Limitations Most agencies reported that the analysis of the deflection data can be challenging. The use of forward or backcalculation methods results in composite layer modulus values that are affected by the thickness variations of the layer being tested, as well as non-uniform support conditions. Like any other testing method, data interpretation must be accurate to obtain reliable results, especially under QA operations. Care must also be taken to avoid errors resulting from data normalized to 70 °F or some other standard temperature for HMA mixtures. A major limitation for QA purposes is the fact that the FWD is not recommended for estimating the modulus of thin layers (bound or unbound layers). The thin layers usually are combined with thicker layers of similar materials, resulting in composite layer modulus values. In addition, the FWD is not widely used for QA to-date, probably because of the thin layer limitation, and universal threshold values and PWL protocols need to be developed. 4.4 Impact Method—DCP The use of DCP in controlling and accepting unbound layers has gained increased popularity because the equipment is simple and easy to handle (see Figure 3 in chapter 3). It is also an economical device, with minimal operator training needs and little to no equipment maintenance. The information gathered with regard to base/subbase relative thickness and strength is invaluable compared to the resources and time consumed to perform the test. Figure 30 shows the results from the DCP for a set of tests from a site with select fill. These data are for the same site used for the deflection-based example (see Figure 29). 123

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report DCP penetration curves. Site 1 - Phase I. 0 10 20 30 40 0 10 20 30 40 50 60 70 80 90 100 Blow no. D ep th , i n Series1 Series2 Series3 -5+00 Series5 Series6 Series7 Series8 Series9 Series10 Figure 30. Graphical Presentation of the DCP Test Results on a Granular Select Fill Material Test data from the DCP were also used to prepare a similar statistical control chart that was prepared for the deflection-based method; see Figure 31. In general, the lower strength material was found where the higher deflections were measured and where the lower elastic modulus values were calculated. The DCP, however, did not show the high elastic modulus values resulting from the deflection basins. Relationships were provided in chapter 3 between the penetration rate or index and resilient modulus. Thus, these values can be tied back to the structural design values used in the MEPDG and other M-E based design methods. 124

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report DCP Penetration Index vs. station. Site 1. 0.0 0.5 1.0 1.5 -5 +0 0 -4 +5 0 -3 +5 0 -3 +0 0 -2 +5 0 -2 +0 0 -1 +5 0 -1 +0 0 -0 +5 0 +0 +5 0 1+ 00 1+ 50 2+ 00 2+ 50 3+ 00 3+ 50 4+ 00 4+ 50 5+ 00 5+ 50 6+ 00 6+ 50 7+ 00 7+ 50 8+ 00 8+ 50 9+ 00 9+ 50 Station Pe ne tra tio n pe r b lo w , i n Mean M 6-24 Figure 31. Illustration of the Statistical Control Values for the DCP Being Used to Measure the Strength of the Unbound Aggregate Select Fill Material Used in the Example for the Deflection-Based Method Agency Use or Adoption The DCP is used extensively by various agencies for evaluating unbound layers prior for rehabilitation designs. Those agencies with extensive experience include Illinois, Indiana, Louisiana, Minnesota, Mississippi, Oklahoma, Texas, Pennsylvania Turnpike, and the Corp of Engineers. Many other agencies have the DCP but use it on a limited basis. Still others are in the process of evaluating the DCP for use in rehabilitation design and new pavement design through existing and on-going research studies (for example, Montana DOT). These agencies have realized the benefit of using the DCP to provide input data to the new MEPDG, as well as other M-E based pavement design procedures. CSIR, South Africa has developed a software program, WinDCP 5.0 (http://asphalt.csir.co.za/DCP/index.htm visited in February 2004), with a user-friendly interface and post-processors. WinDCP5.0 automatically gives average DCP penetration rates in mm/blow. DCP results have also been calibrated with unconfined compressive strength and CBR values (provided in chapter 3). The program estimates the structural capacity of granular and weakly stabilized pavements base materials and classifies the pavement in various categories based on the structural capacity estimations. Test Protocol and Data Collection Guidelines ASTM recently standardized a procedure for general use of the DCP, ASTM D 6951. Minnesota, Mississippi, Oklahoma, and other agencies all have test protocols and data 125

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report collection guidelines for using the DCP for rehabilitation design and forensic studies. No agency contacted uses the DCP as an acceptance tool or device for subgrades to-date. Minnesota DOT, however, uses the DCP in the QA process of aggregate base layers. Testing must be performed within 24 hours of placement of the base layer. A minimum of two tests are conducted within each 1,000 cubic yard (800 cubic meters) volume. The Minnesota DOT specification requires that the compacted base layer has a penetration index less than 10mm/blow. Any layer that exceeds this requirement has to be re-compacted and retested until that rate of penetration is not exceeded. A compacted layer is defined as one with a compacted thickness of at least 3 inches (76 mm) but no more than 6 inches (152 mm) for each lift. Interpretation of Test Data and Determination of Material/Mixture Property DCP is easy to interpret, especially for relative comparison of material strengths along a project. The penetration rate or index has been used for design and evaluation. The units used are penetration depth per blow. The more difficult issue is relating the penetration rate or index to the elastic modulus of the material. Regression equations have been developed relating the penetration rate to the CBR value of the material (see chapter 3), but these are believed to be material specific. In addition, any regression equation relating DCP test results to resilient modulus will depend on the pavement type and thickness. It is expected, however, that criteria can be developed for specific type of materials for use in QA programs. The criteria should be tied back to the assumptions used for the elastic modulus of the material during structural design. The accuracy of the relationships used to develop the criteria noted above is dependent on the number of soil and aggregate types used to develop these regression relationships. The test method can be highly variable, simply because the materials being tested are variable. The variability of the test results is greater for embankment and fill materials, and decreases for processed materials. Production Rate The DCP production rate is dependent on the type of material, strength, and thickness of the layer being tested. In general, most DCP tests at a specific location can be completed within 5 to 10 minutes. This time, however, excludes the coring of the HMA surface for testing unbound aggregate base layers, embankments, and subgrade soils. Coring of HMA should not be needed for QA purposes, unless dispute resolution is required after an HMA layer has been placed. Initial Cost, Maintenance, and Complexity of the Equipment The DCP is easy to operate and requires minimal training, as compared to most of the other NDT technologies used for testing pavement materials. Maintenance of the equipment is also minimal, even for the automated device. The initial cost of the equipment is less than $30,000 for the automated device and less than $15,000 for the manual device. Advantages The advantages of using the DCP for QA include ease of use for testing unbound aggregate and soil layers, the results can be easily understood by field technicians, and the results are 126

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report related to the modulus of the layer/material being tested. In addition, layers greater than 1 foot thick can be tested, and the results are layer specific, unlike some of the other NDT devices that are influenced by the underlying layers (composite modulus values). The cost and technical support for the equipment are low compared to the cost and support requirements for most of the other NDT technologies. Disadvantages and Limitations The major disadvantages of the DCP are the inability to test HMA layers, and that coring is required to remove any HMA layer prior to testing unbound aggregate materials and soils. In addition, no threshold values have been developed that can be used immediately within a QA program for determining the quality of the unbound layers—other than the value being used by the Minnesota DOT for aggregate base layers. Another limitation of the DCP is trying to test embankments with boulders or larger aggregate particles that can result in refusal or low penetration rates at specific points. 4.5 Ground Penetrating Radar Methods—GPR GPR has been used as part of research, forensics, or evaluation studies. Agencies and organizations that have used GPR include Florida, Illinois, Missouri, Minnesota, Nevada, Oklahoma, Texas, Virginia, Washington, and Wisconsin DOTs. The Corp of Engineers, FHWA, Federal Aviation Administration (FAA), and U.S. Air Force under the Department of Defense have also used GPR for measuring layer thickness and identification of subsurface features for evaluating pavement structures and rehabilitation designs. Some agencies (such as the Georgia and Texas DOTs) are also considering the use of GPR technology in support of their pavement management programs. GPR was used within the FHWA-LTPP program for measuring the layer thickness within the test sections. It was also used for measuring layer thickness and volumetric properties (density and/or air voids) of the HMA layers placed at the WesTrack, MnROAD, and NCAT test tracks. To demonstrate the practical and effective use of the GPR, the data measured along the WesTrack test sections were used to estimate two quality characteristics: HMA thickness and air voids. The equations and data analyses presented in chapter 3 were used to calculate both properties in accordance with the Finland/Texas DOT correlations. In summary, Figure 32 shows the distribution of the HMA layer thickness, while Figure 33 shows the distribution of air voids. As shown, the HMA layer thickness data set has a normal distribution, while the air void data set has a skewed distribution. The thickness distribution is typical of data measured from other projects where both GPR and a sufficient number of cores were recovered to accurately determine the distributions. The mean thickness from the GPR data at WesTrack closely matched the thickness values measured from cores, but the mean GPR-estimated air voids did not always match the average air voids measured on cores recovered from the test sections. 127

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report WesTrack Thickness Distribution 0% 5% 10% 15% 20% 25% 127 141 155 169 182 Thickness (mm) Fr eq ue nc y (% ) mean = 152 σ = 7.5 Figure 32. Frequency Distribution of the Variation in HMA Layer Thickness Estimated with GPR Technology at WesTrack Asphalt Air Voids from Dielectric Data 0% 5% 10% 15% 20% 25% 2 3 4 6 7 8 10 11 12 13 15 16 17 19 20 21 23 Air Void (%) Fr eq ue nc y (% ) mean = 8.24% σ = 3.56% Figure 33. Frequency Distribution of the Variation in HMA Air Voids Estimated with GPR Technology at WesTrack The distribution of air voids after construction (see Figure 33) also typically has a normal distribution. The skewed distribution resulting from the GPR measurements made at WesTrack could be related to the non-compliance values (high and low air voids) built into some of the test sections along the test track. 128

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report The data from each lot at WesTrack are shown in Figure 34, which represents the average air voids between the right wheel path, left wheel path, and centerline using 20 sublots within each lot or test section. As shown, a statistical control chart was prepared for the average air voids that vary down the roadway. A similar statistical chart could also be prepared using the range of values to determine whether the variation in air voids of the population is in control or out of control. Statistical Control Chart for Air Voids 0 2 4 6 8 10 12 0 10 20 30 40 50 Lot Identification Air Voids, % Series1 Figure 34. Example of a Statistical Control Chart for the Average Air Voids Estimated Using GPR Technology The arrows shown in Figure 34 represent typical target air voids for the HMA base layer, and the upper and lower control limits for that specific contractor. As shown, the placement of the HMA layer may be out-of-control in some of the lots. More importantly, there appears to be a drift or uniform decrease in air voids along the project, especially for the higher lot numbers. Although this decrease in air voids was designed into the experimental plan at WesTrack and the lower air voids of some section were not identified in the GPR data, it shows the power of the GPR technology in QA application. The amount of data collected with the GPR is more than sufficient to properly determine the calculated air voids for that population or mat, whereas that would be highly unlikely using the traditional QA sampling and testing programs. Proper calibration of the GPR with at least some cores is essential. Cores and bulk HMA samples are not normally used to determine the type of distribution because there is an insufficient amount of time to take the required number of samples to ensure that the distribution has been properly described. Normality is assumed for most 129

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report cases, and a small number of samples are taken to estimate the properties of the population. The GPR technology can take a large number of tests in a short time period to clearly identify the type of distribution and the characteristics of that distribution for the population—rather than for a sample from the population. As noted in chapter 3, EPIC is promoting the use of a proprietary system that uses multiple antennas and produces different volumetric properties for HMA mixtures, as well as HMA layer thickness. These properties include relative compaction (density) and asphalt content, and a composition analysis of the HMA mixture. The accuracy and reliability of this system has been investigated by an independent organization (Greene, 2006), while some agencies (such as the Florida DOT) are expanding previous investigations of this system. The accuracies that have been reported are provided in the following pertinent paragraphs. Agency Use or Adoption Some agencies (for example, the California, Florida, Minnesota, New Hampshire, and Texas DOTs) have their own GPR systems that are used for various purposes. Minnesota and Texas use their devices for forensic analyses and investigations, while Florida uses their system as a rehabilitation design tool. These and other agencies are also considering the GPR technology as a QA tool for supplementing their current acceptance plans for flexible pavement construction. Since 1999, the New Hampshire DOT has been using the GSSI ground-coupled 1.5 GHz GPR for QA and pay factor calculations for PCC cover over the top rebar on new bridge decks.9 The method collects data along survey lines parallel to the centerline of each lane. The system uses software to calculate the arrival time of the reflection from each rebar in the scan. The relationship between time and depth is obtained through calibration holes. The Finnish Road Administration has adopted the GPR technology as a method for assessment of penalties for HMA pavement whose air void content is found to be outside of the specifications.10 The HMA dielectric constant is calculated from the GPR surface reflection, and calibration cores are used to calibrate the relationship between dielectric constant and air void content. A similar method for using GPR as a tool for QA for new flexible pavement has been developed by TTI, and this method currently is under evaluation by the Texas DOT (Sebesta and Scullion, 2002). The protocol for this method is as follows: 1. Collect a series of parallel GPR survey lines at 2-foot (0.6-m) lateral offsets (5 lines per 12-foot [3.7-m] lane). 2. Process the data to compute the HMA dielectric constant from the GPR surface reflection. 3. Calculate the mean dielectric constant. 4. Plot the GPR data, and identify areas where the dielectric constant is less than the mean by more than 0.8 (for coarse-graded mixtures) or 0.4 (for fine-graded, dense 9 NHDOT GPR Specification for Concrete Cover; Special Provision—Amendment to Section 520 – Portland Cement Concrete; Subsection 3.1.7 on Quality Assurance. 10 Finish Road Administration Specification—PANK 4122: Air Void Content of Asphalt Pavement, Ground Penetrating Radar Method. 130

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report mixtures). These are areas where the mix properties deviate significantly and where problems are likely to occur. 5. Calibrate the relationship between dielectric constant and air void content using 3 cores and equation 8 (see chapter 3). The result of the correlation is shown in Figure 35. 6. Plot the air void content, and locate areas where the air void content is outside of the specified values. The plot can be linear or a surface contour plot, as shown in Figure 11 in chapter 3. The same GPR data used for evaluation air void content, listed above, can be used to calculate the variability in pavement thickness, and to identify locations where pavement thickness is outside the specified limits. Figure 35. Correlation of Dielectric Values with HMA Air Voids In 2003, the California DOT (Caltrans) completed an investigation of GPR methods for thickness in their QA program. The objective was to use the method as a basis for pay factors. The Caltrans study showed that GPR using a non-contact, air horn antenna was able to determine the average pavement thickness on a newly constructed segment to within 0.1 inch (2.5 mm) of the value obtained from a sample of 20 cores on that segment (Maser et al., 2002). A protocol for this thickness evaluation was prepared as part of a Caltrans project (Maser, 2003).11 Tests were conducted under the Caltrans project on six newly constructed pavement segments, each 1,000 feet (305 m) long representing full depth HMA construction and HMA overlays over existing flexible and rigid pavements. The GPR method was able to cover the 11 CalTrans Proposed Test Method—Horn Antenna GPR Method for Asphalt Thickness. 131

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report pavement section in a small fraction of the time than what was required to take cores. The GPR survey produced thousands of thickness data points, as compared to the limited number of core data points. A recent study by the Virginia DOT using similar equipment produced similar results (Al Qadi et al., 2003). The capability of GPR to determine HMA layer thickness has been verified for HMA surface and base layers (e.g., Roddis et. al, 1992). Investigation of GPR for measurement of unbound aggregate base layer thickness has been limited. Most of the available data are for existing pavements. Studies in Texas and Florida showed that the average per site deviation between GPR and core measurements ranged from 0.75 to 1 inch (19 to 25 mm), or 10 to 15 percent (Maser and Scullion, 1992; Fernando et al., 1994). These studies showed that base layer thickness could not be measured for cement treated bases, since there was inadequate dielectric contrast between the stabilized base and the subgrade soil. Subsequent studies of existing flexible pavements (Maser, personal experience) have shown that the ability or accuracy to detect the bottom of the base varies considerably. This unpredictability might be due to the "blurring" of the boundary between unbound base and subgrade caused by migration of fine material from the subgrade over time. The ability to detect unbound aggregate base layer thickness in new construction has not been reported. It is very likely that the limitations to detection of the base/subgrade boundary in existing pavements may not apply to new construction. Other than the applications described above, the adaptation of GPR by most state agencies in the U.S. has been focused towards forensics or pavement evaluation for rehabilitation purposes. The Texas DOT, with support from TTI, has adapted GPR as a standard tool for project-level pavement evaluation and diagnostics. The Florida DOT acquired a GPR system for layer thickness inventory data in 1996 (Fernando and Maser, 1997). In addition, the New Jersey DOT conducted an evaluation on 1200 lane miles of pavement in 2002-2003 as part of their equipment evaluation process prior to adapting the GPR technology to day-to-day practices. Accuracy and Repeatability Air Void Content and Other Volumetric Properties The Finnish Road Administration reported that the measuring accuracy of the GPR surface reflection technique for estimating air voids is + 0.9 percent. This statistical analysis result has been achieved through comparison of core sample results and GPR measurements conducted as static shots over each individual measurement points (R=0.9223). Greene completed an independent precision and bias study of the Hyper OpticsTM technology for EPIC in 2006 and 2007 (Greene, 2007; Greene and Hammons, 2006). The study found good results when the system was calibrated with properties measured on cores recovered from multiple locations along the projects. In summary, Greene reported a 95 percent precision tolerance of + 2.32 percent for air voids when outliers were removed from the statistical analyses. Similarly, Greene and Hammons reported a 95 percent precision tolerance of + 0.36 for asphalt content; + 0.050 for bulk specific gravity; and + 2.69 percent for VMA. For the relative compaction module included 132

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report in Hyper OpticsTM technology, the percent compaction was within AASHTO T-166 criteria for 61 percent of the time. Although EPIC does not routinely estimate the maximum specific gravity for HMA mixtures, values have been reported for some mixtures or projects. The 95 percent precision tolerance reported by Greene is + 0.022. Layer Thickness The Caltrans study noted above (Maser, 2003) involved six newly constructed test sections, two sections each of the following measurement types: (a) full depth (multiple lift) HMA; (b) HMA overlay on PCC; and (c) single and multiple lift HMA over HMA. The results showed the following accuracy measures based on comparisons to core thickness: Mean GPR thickness for a section: std. error = 2.1 mm = 1.6% Thickness variability for a section: std. error = 4.3 mm = 3.3% (as measured by the thickness standard deviation) GPR thickness at a point: std. error = 8.6 mm = 9.4% (R=0.933) The local thickness error is reduced when sites representing a new HMA overlay over an old flexible pavement are removed: GPR thickness at a point: std. error = 6.5 mm = 7.0% (HMA overlay on HMA excluded) The Virginia DOT study (Al-Qadi et al., 2003) involved measurements at one full depth HMA construction site at which thickness measurements were made after each lift of HMA was placed. The reported point by point results based on comparison to cores using this cumulative thickness method were as follows: 100 mm HMA Base Layer (at a point) std. error = 4% 180 mm (HMA Base Layer + 1st HMA Layer) std. error = 1.7% 250 mm (HMA Base Layer + 2 HMA Layer s) std. error = 2.2% The accuracy results suggest that the accuracy estimate is somewhat sensitive to the type of construction (overlay vs. full depth) and the timing of the measurement, similar to the Caltrans study. Measurements after each lift will provide greater accuracy when compared to one measurement after the paving is complete. Greene reported similar values for the Hyper OpticsTM technology with a 95 percent precision tolerance of + 0.5 inches using field core thickness measurements. Unbound Layer Thickness Previous studies have shown that the error of GPR thickness measurements of unbound base layers below HMA ranges from 10 to 15 percent. In summary, properly calibrated equipment produces repeatable data, even though there are minimal published repeatability statistics. Repeatability studies carried out by Maser and Scullion (1992) identified the relationship between antenna height calibration and the 133

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report achievement of repeatable data using identical equipment. This study also showed that using different equipment on the same pavement can produce different results for weak layer boundaries (e.g., base/subgrade boundary) due to differences in the equipment's ability to resolve these boundaries. Calibration The GPR equipment should be calibrated each survey day. The daily calibrations consist of static metal plate test and direct wave (air wave) test. Static plate tests should be conducted at the beginning of the survey day after 20 minutes of equipment warm-up, and at the end of the survey day before equipment breakdown. The direct wave test, which requires partial breakdown of the equipment, should be conducted after the second metal plate test at the end of the survey day. The antenna height calibration function (to compensate for vehicle bounce and setup height) should be evaluated periodically, and before any major projects. System time calibrations should be carried out according to ASTM D 4748-9b at the beginning of each survey. The distance-measuring instrument (DMI) should be calibrated at the beginning of each survey and at the end of each week of ongoing surveys. Production Rate GPR data collection consists of conducting a series of parallel survey lines with vehicle- mounted equipment driving at normal speeds. Assuming that setup and calibrations are already completed, it is estimated that data collection on 1 mile of new pavement would take less than 30 minutes. Initial Cost A single antenna GPR system, including data acquisition and control system, cables, antenna, and mounting, would cost approximately $50,000. An example of the equipment would be a GSSI SIR-20 data acquisition and control system, with a model 4108 1 GHz horn antenna and cable. The survey vehicle would have to be equipped with an electronic DMI so that the data in the survey lines could be coordinated with actual locations on the test pavement. The system would also have to include automated software for calculation of dielectrics and layer thickness. Prototype software of this type has been developed, but none is commercially available. Most of the software that has been developed is proprietary. Advantages The advantage of GPR is that it can acquire thousands of measurement points quickly, and by doing so it can provide a complete representation of the variation of thickness and air void content in each layer for the entire population. Some firms that have developed proprietary software report that density (or relative compaction), thickness, asphalt content, and VMA can be extracted from the GPR measurements using multiple antennas. The tests can be conducted shortly after construction or placement of each layer and lot. Though cores are required for calibration, the number of cores needed is less—and the resulting information far greater—than for any other system. 134

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report Disadvantages and Limitations The required GPR equipment is not simple to operate and generally is operated by well- trained individuals. The software to automate the data analysis has been shown to be feasible for this application, but it is not commercially available and has not been verified to date. As noted above, some of the available software packages are proprietary. In addition, the analysis to review the data and resulting charts takes more time than normally available for QA procedures. This limitation can be overcome with continued use and verification of the results. As an example, EPIC has developed software that can be used on a real-time basis; the limitation of the Hyper OpticsTM technology is that the software is proprietary. GPR Status with the FCC – A Potential Limitation of the Equipment In February 2002, the Federal Communications Commission (FCC) issued a first report and order which severely limited the use of GPR. The order stated that GPR operation would be "restricted to law enforcement, fire and rescue organizations, to scientific research institutions, to commercial mining companies, and to construction companies." The order also stated that “GPR equipment must be operated below 960 MHz or in the frequency band of 3.1 to 10.6 GHz” and it specified strict limits for radiation emitted above 960 MHz. After considering some objections raised by the GPR industry, and the fact that no document occurrence of GPR-related interference had ever been reported, the FCC made some amendments. In July 2002, the FCC ruled that existing GPR operators can continue to operate once they register their operation with the FCC, and that existing equipment used by these operators would be given a blanket waiver. In February 2003, the FCC rules were amended to allow for GPR operation over 960 MHz, and to allow operations "related to" construction (interpreted to mean all highway-related applications). The one remaining issue with the FCC's rules is that the radiation limits above 960 MHz cannot be met by the current manufactured air horn antennas. The GPR industry is working to obtain a waiver for this type of antenna, recognizing its value to highway engineers and its relatively limited use compared to the universe of communications equipment. 4.6 Infrared Tomography Methods and Technology Infrared tomography has been used by relatively few agencies in their research, forensics, and evaluation studies. The agencies and organizations that have used this technology include the California, Connecticut, Florida, Nevada, Minnesota, Missouri, Texas, Virginia, and Washington DOTs, and the U.S. Army Corps of Engineers, FHWA, and U.S. Air Force through the Department of Defense. The Connecticut, Florida, and Texas DOTs have used infrared cameras to demonstrate temperature anomalies (cold spots) in HMA mats prior to compaction and their effect on mat density. The infrared cameras have not been used extensively in QC operations to date. Washington DOT is the only known agency that uses the infrared cameras (see Figure 13) in their acceptance plan based on density. Agency Use and Adoption Washington DOT was an early adapter of the infrared camera as a tool for QC/QA. Their work with infrared began in 1995, and their study and use of infrared to determine variability 135

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report in density has continued. Washington DOT has concluded that significant density differentials occur when the HMA transport vehicle dumps its load into the paver, leaving a concentrated area of lower temperature HMA with every truckload (see Figures 13 and 14.b in chapter 3). Because of this cyclic nature of density differential, traditional statistically based random sampling for QA of field density does not have the ability to characterize this problem (Willoughby et al., 2003). By investigating the relationship between temperature differentials and density, the Washington DOT has shown that temperature differentials greater than 14ºC (25ºF) correspond to changes in air void content greater than 2 percent. Washington DOT has implemented a density specification that locates potential areas of low density using the greater than 14ºC (25ºF) temperature differential criterion. These areas are tested for density and must meet a specified minimum. Washington DOT has incorporated these temperature measurements into their nuclear density method specification, prepared special data sheets, and prepared a "Cyclic Density Special Provision" in which the infrared based density results are incorporated as a pay item. At present, Washington DOT has four infrared cameras—three in use by district engineers and one in use by the central office for continued studies (Willoughby, 2004). The University of Washington, in conjunction with Washington DOT, has set up an infrared image database that has incorporated documented infrared pavement images from states participating in a pooled fund study (Connecticut, Minnesota, Texas, California, and Washington State). A sample entry in this database is shown in Figure 36. The Texas DOT has implemented specifications using the greater than 14ºC (25ºF) temperature differential as an indication of significant problems in the HMA (Scullion, 2003). These differentials are measured after placement but before breakdown rolling. Figure 36. Sample Entry Form in Pooled Fund Study Database 136

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report The use of infrared tomography for QC appears to have provided valuable feedback related to problems in the paving process. Both Washington and Texas DOTs report that the infrared data has been extremely valuable in the early stages of the construction process, where inadequate material handling and remixing has led to temperature segregation. Once this information is available, the problem can be isolated and corrected, and the temperature variations no longer appear (Scullion, 2004). Test Protocol and Data Collection Guidelines A test protocol for the use of an infrared camera or a hand-held infrared spot thermometer is included in the Washington DOT specification. The Texas DOT system provides automated temperature contour charts, but there is no known evaluation protocol associated with the use of these charts. Interpretation of Test Data The infrared data are used as a qualitative QC method for identifying problems in the construction process. For QA, the Washington DOT uses the infrared data to determine the locations of conventional density measurements. They do not use the infrared data as a direct measure of any pavement property. The Texas DOT uses the temperature differential as a direct measure of a pavement deficiency based on previous correlation between temperature differentials and density. However, the Texas DOT does not use the infrared data as a direct measure of a mixture or mat property. Accuracy, Repeatability, and Reproducibility of the Infrared Measurement No known documented studies have been conducted dealing with this subject related to measuring temperature differences in an HMA mat and corresponding density measurements. Calibration Requirements Standard calibration methods are used for ensuring the proper temperature reading from an infrared camera or sensor. These involve determining the emissivity value of the pavement surface based on either the IR reading from a material of know emissivity (e.g., black tape) or using a surface thermocouple. HMA emissivity is typically between 0.90 and 0.98 (Sebesta and Scullion, 2002). Production Rate Both the infrared camera and sensor bar approaches can operate directly behind the paver and have a production rate equivalent to the rate of paving. Cost, Maintenance, Complexity, Interpretation, Operator Technical Requirements Infrared Cameras Infrared cameras sell for anywhere from $20,000 to $50,000. The cameras are reasonably robust, but they need the same care and treatment as a video camera. No regular maintenance is required. After an initial training session (1-2 days), the cameras are easy to operate. The results are directly apparent in the video or still images. A field technician with experience with electronic equipment should be capable of operating an infrared camera. 137

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report Infrared Sensor Bar The infrared sensor bar is a custom built item, so its cost has to be estimated. The primary purchased components are 10 infrared sensors at $250-$350 each, a laptop computer (or dedicated processor) equipped with a 10-channel data acquisition system, a mounting bar for the sensors with an attachment to the screed, an electronic distance encoder, and software to produce the real time temperature contour plots. If manufactured, one might estimate a sales price of $15,000-$20,000. Advantages The advantage of infrared tomography is that it provides immediate information on the location of cold spots within the paving process which can be readily understood by all field personnel. This immediate feedback can be used to correct deficiencies in the paving process, as well as to identify locations for density measurements. In some respects, infrared tomography provides information that is already known, such as cyclic temperature differentials (and therefore, density) resulting from transport of HMA and loading of the paver. These temperature differentials can be minimized by using material transfer devices (MTV). The infrared camera or sensor bar can simply confirm whether such a problem is occurring, and lead to the implementation of a solution. The simple presence of the infrared equipment appears to motivate the contractor to deal proactively with potential material transfer problems (Willoughby, 2004). In summary, the infrared cameras do provide valuable forensic data to a contractor on determining the cause for not being able to obtain adequate densities, and to the agency for selecting bias locations for density tests. Disadvantages and Limitations TTI researchers found the camera to be cumbersome and difficult to use on a routine basis, and they went to the automated sensor bar approach. In addition, Connecticut DOT (one of the agencies participating in the pooled fund study) found "no significant correlation between temperature differentials and changes in density" (Sebesta and Scullion, 2002). Thus, the relationship between temperature differentials and density found in Washington and Texas may require more work to improve on the location of the density tests from the temperature differential measured with the cameras. More importantly, the infrared data alone have not been related to fundamental properties of the HMA mixtures that can be used in traditional acceptance programs, and these data do not provide a direct or indirect tie to performance or an input to the MEPDG. Another area of concern for some agencies is the identification of bias sample locations in a statistically-based acceptance program. 4.7 Ultrasonic/Seismic Methods—PSPA and DSPA A procedure based on seismic techniques to measure the modulus layer-by-layer shortly after placement was developed for the Texas DOT. The procedure measures layer modulus of pavement materials with four inter-related seismic devices. Two of these are laboratory devices: the free-free resonant column device for testing base and subgrade and the ultrasonic device for testing HMA cores and laboratory prepared specimens. The other two are field 138

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report devices: the PSPA for testing HMA layers and a version of it that works on the base and prepared subgrade layers (called DSPA, for Dirt Seismic Pavement Analyzer). The proposed QA procedure consists of several steps. The first step consists of selecting the most suitable material or mixture for a given project. The second step is dedicated to determining the variation in modulus with the primary parameter of interest and determining the desired modulus. For base and subgrade materials, this step consists of developing a moisture (water)-modulus curve (similar to an M-D curve). For HMA materials, this step consists of developing voids in total mix (VTM)-modulus curve. In the third step, the variation in modulus with environmental factors is considered. For example, the change in modulus with changes in water content of a base layer can be determined in the laboratory. In the case of HMA, the change in modulus with varying temperature and asphalt content is important. The fourth step consists of determining the desirable or target (design) modulus for the material. The final step is to compare the field modulus with the acceptable laboratory measured modulus. These procedures allow rapid data collection and interpretation. Thus, any problem during the construction process can be identified and adjusted. Performing the simplified laboratory and field tests along with more traditional tests may result in a database that can be used to confirm the assumptions used for pavement structural design—integrating structural design and acceptance of flexible pavement materials (refer to Figure 2 in chapter 2). The method has shown promise as a practical tool and is being implemented on a trial basis by the Texas DOT. The key to this procedure is that the NDT is calibrated to the specific mixture in the laboratory. The simplified laboratory tests can also be used to develop the ranges of acceptable properties for a given material. NDT field tests are performed to determine whether the contractor has achieved the minimum specified stiffness that can be related back to the value used for structural design. The seismic scanners that were discussed in chapter 3 have not been used extensively, with the exception by highly trained consultants that have been involved in the development of the equipment and university personnel for research and forensic purposes. However, these devices have high potential for use in the future because of the extensive coverage of the area tested within a minimal amount of time. These devices are not suggested for use in QA operations until their use become routine in pavement engineering applications. In addition, there are relatively few of the SPA devices or trailers available for testing and use. The PSPA and DSPA devices can be made easily available for future use and application to QA operations. Thus, the PSPA and DSPA were considered for use in the field evaluation study, while the SPA was not. Agency Use or Adoption No agency currently uses seismic technology in routine QA operations. However, Florida, Texas, FHWA, U.S. Army Corp of Engineers, and the U.S. Air Force through the Department of Defense have used this technology for forensic studies of pavement structures. 139

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report A few agencies have on-going research projects to develop these methods for use in QA applications. Accuracy and Repeatability There is little information available on the accuracy and repeatability of these test methods. The University of Texas at El Paso, through the Texas DOT, has data from selected projects. These laboratory test methods are repeatable considering the variability between test specimens of HMA mixtures and unbound materials and soils. The coefficient of variation measured on similar HMA mixtures tested in the laboratory is less than 5 percent when the test specimens are compacted to the same air void level in a highly controlled manner. Calibration Calibration is important to obtain reliable results using seismic test methods. The test methods and test results need to be calibrated for each project or material used to ensure that the results can be related back to the modulus value used in structural design. Production Rate Although the SPA and PSPA have not been used in routine pavement evaluation projects, it is estimated that data can be collected on a basis of about 1 test point in 2 minutes for the SPA (trailer mounted device; see Figure 15 in chapter 3), while 1 test in less than a minute can be completed for the manual PSPA (see Figure 16 in chapter 3). Advantages The major advantage of seismic methods is that similar results are anticipated from the field and laboratory tests as long as the material is tested under comparable conditions. This unique feature of seismic methods in material characterization is particularly significant in QA operations. In the procedure recommended for use in Texas, simplified field and laboratory tests are suggested that can be performed and interpreted rapidly so that noncompliant materials can be identified during construction. The field and laboratory methods are incorporated in a manner in which the results can be reconciled without any scaling or simplifying assumptions. A major advantage of the DSPA is that it can be used to develop modulus-growth curves to optimize the compaction of unbound materials and soils. Disadvantages and Limitations One of the limitations of using seismic technology for QA application is that the mixture modulus value does not represent the stress levels that occur under truck loadings. The modulus values have to be adjusted to account for the design loading frequency and temperature. One of the disadvantages of the equipment is that the HMA mixture must be allowed to cool down to a temperature less than about 160 °F. When the device is placed on HMA at elevated temperatures, the rubber pads of the response detectors begin to melt or become easily damaged. In addition, the wave form usually is not well defined at elevated temperatures. Other disadvantages of the equipment are that it takes highly trained personnel to collect and interpret the data and there is relatively little information available to determine the precision and bias of test output. Training will be an important step in the implementation process for QA application. 140

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 4.8 Steady-State Vibratory Response Method—GeoGauge The GeoGauge is a relatively new device, but it has been recently updated, as discussed in chapter 3. Although the device has significant potential in QA application, it has been used minimally in flexible pavement diagnostic and forensic studies. Agency Use or Adoption No agency currently uses the GeoGauge in routine QA operations, nor do any contractors use it in their QC operations. FHWA and other agencies have on-going research projects to investigate the accuracy of the gauge for QA applications. Accuracy and Repeatability Information on the accuracy and repeatability of the GeoGauge was obtained during the initial development study and from the pooled fund study that was recently completed. Some of the projects used in those studies used the older gauges. The data collected to-date would suggest that the coefficient of variation for a single operator using a single gauge is between 2 to 5 percent, while for multiple operators and gauges the coefficient of variation increases to a value of 5 to 8 percent. Calibration Calibration is important to obtain reliable results using the GeoGauge. It is recommended by the manufacturer that a local calibration be completed at the beginning of each day’s use. The GeoGauge comes with a relative calibration procedure and mass verification equipment. Production Rate Although the GeoGauge has not been used in routine construction or evaluation projects, it is estimated that data can be collected on a basis of about 1 test in 2 minutes. One test would include clustered tests at a single point, similar to the DSPA. Advantages A major advantage of the GeoGauge is that it can be used to develop density and modulus growth curves as the unbound materials and soils are being compacted by the rollers. Another major advantage of GeoGauge is that results from the field tests have been found to be similar to those measured in the laboratory with proper calibration. This unique feature in material characterization is particularly significant in QA operations. In addition, the coefficient of variation is reasonable for clustered testing, and those tests can be performed in a short period of time. The gauge is also easy to use and requires minimal training. Disadvantages and Limitations One of the limitations of the GeoGauge is that the material’s modulus value does not represent the stress levels that occur under truck loadings. The modulus values have to be adjusted to account for the design loads. Another limitation of the GeoGauge is that the underlying or supporting materials can influence the results of the upper layer when trying to test relatively thin unbound layers (less than 6 inches in depth). Conversely, it cannot be used to accurately measure the modulus of thick unbound layers (greater than 12 inches in thickness). 141

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 4.9 Non-Nuclear, Electrical Sensing Methods—PQI, PaveTracker, EDG, and Purdue TDR Most agencies and contractors consider density as a quality characteristic and include this material property in their acceptance and control plans. The most recent survey conducted to determine the prevalence of NDT methods for HMA density acceptance was the 2002 Binder Payment Method and Compaction Specification Survey conducted by the Colorado DOT on behalf of the AASHTO Subcommittee on Materials. The survey was fairly complete in that 41 states and the FHWA participated. The survey showed that 28 states use cores for HMA acceptance, while 22 use the nuclear density gauge for acceptance. While three states were counted as using something other than cores or the nuclear density gauge, these methods were simply a version of the core method or a method specification. As part of NCHRP Project 10-65, nine states not covered in the Colorado survey were contacted. None of these states use anything other than cores or nuclear density gauges. Since the 2002 survey, however, many agencies are now investigating the non-nuclear electrical sensing NDT devices for both HMA and unbound layers. Some agencies that have on-going research studies of these devices include Alabama, Colorado, Florida, Nevada, Ohio, Texas, and Wisconsin. These methods being investigated include the non-roller- mounted, as well as roller-mounted devices or systems. To date, none of these devices is being used for acceptance, but some contractors have incorporated the non-nuclear density measuring devices (PQI or PaveTracker) into their QC plans for HMA.12 The use of non-nuclear density measuring devices for unbound layers has been much more limited. Most agencies and contractors use sand-cone tests and nuclear density gauges for QA. The Colorado, Nevada, and Texas DOTs have on-going research studies that include the use of or are evaluating the EDG device. The following provides supplemental information to that included in chapter 3 on the non-nuclear density gauges for use in QA of flexible pavement construction. • No state has adopted this technology for acceptance of HMA and unbound materials/soils. Selected contractors, however, are using this technology for controlling the compaction of HMA mixtures. Both the PQI and PaveTracker are being used. The Nevada and Texas DOT studies on the EDG show promising results for monitoring the density and water content of unbound layers.13 • Calibration is important to obtain accurate results in terms of material density. Adjustments can be programmed or entered into the HMA devices for measuring density. However, these adjustments need to be periodically checked against cores, similar to the nuclear gauges. The gauges used for measuring density and water content of unbound materials require the development of a soil model between the electrical readings and the density and water content of the in-place soil. The soil 12 Agencies contacted provided the names of contractors that use these devices for QC operations. These states were considered a high priority for the field evaluation projects (Alabama, Michigan, Minnesota, Texas, and Wisconsin). 13 Information obtained from the Nevada DOT and University of Texas at Austin (conducting the Texas DOT study) during the survey of NDT devices for use in QA and other applications. 142

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report model can be developed in the laboratory but should be periodically checked during construction. • Data on the accuracy of these gauges is limited, but many agencies have indicated that the accuracy and repeatability of these gauges is equal to or better than that of the nuclear gauges. • The advantages of the HMA non-nuclear gauges are the speed and safety of the devices, as compared to nuclear density gauges. The disadvantage is that the manufacturers recommend that the gauges be used immediately after compaction— during the same day of paving. Surface water and other contaminates can affect the readings. • The advantage of the unbound non-nuclear gauges is safety. The disadvantage is that these gauges have had limited use to demonstrate their immediate and practical use in QA programs. In summary, the material density is an important quality characteristic that deserves additional consideration for further evaluation of NDT techniques that measure this property. 4.10 Non-Nuclear-Roller-Mounted Density/Stiffness Methods—IC Rollers Although rollers with the IC devices are commercially available, few contractors have purchased these rollers. Some agencies have and are sponsoring demonstration and research projects on the use of this technology for controlling and accepting flexible pavement construction of individual layers. Few projects have included the use of this technology for constructing the entire flexible pavement or all HMA lifts placed on a rehabilitation project. Thus, chapter 3 provided a summary of the information available for review on the roller- mounted systems. It should be noted, however, that NCHRP, Colorado, Ohio, Michigan, Virginia, and Wisconsin are sponsoring or participating in research studies evaluating these systems. Results from many of these studies will be available within the new 2 to 4 years. 4.11 Surface Condition Characteristics The surface condition factors consist of longitudinal and transverse profiles for determining the ride quality or smoothness of the pavement surface, the texture for determining the noise abatement features of surface layer, and the friction of the surface layer for determining skid resistance. 4.11.1 Smoothness Most agencies use some measure of smoothness (typically an IRI value) in their specifications. The equipment used to measure smoothness is the profilometer, lightweight profilometer, and profilograph. The lightweight profilometer and profilograph are used by most agencies in their acceptance plan. These devices are readily available for use, but they only provide a measure of the smoothness of the surface layer. In addition, there are numerous manufacturers or suppliers of the lightweight profilometer that can be used. The purpose of the field evaluation study was not to compare and evaluate different lightweight 143

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report profilometers. Thus, smoothness was not considered in the field study, and the different devices not included. 4.11.2 Noise Noise is also becoming a significant factor in pavement surface type or rehabilitation strategy selection. However, most of the work completed to-date has been with the NCAT noise trailer. The number of trailers available is limited. In addition, those agencies contacted (for example, Arizona, Michigan, Ohio, Texas) have no plans to incorporate noise into their acceptance procedures.14 Noise could be included as a factor in an acceptance plan, but only after the tie between mixture properties and noise is developed and verified. In other words, there are no criteria available that can be applied during the HMA mixture design stage. Thus, noise was excluded as a factor in the field study. 4.11.3 Surface Texture Although macrotexture is an important component in wet weather skid resistance and noise generation (Henry, 2000), agencies have not specified macrotexture levels in relation to skid resistance and identifying surface defects, such as segregation. Surface macrotexture measurements cannot be used to identify segregation beneath the surface. In addition, there is no known HMA mixture design or structural design procedures that use macrotexture. More importantly, none of the DOTs contacted have plans to incorporate surface texture into their acceptance plans. Thus, surface texture was excluded from the field study. 4.11.4 Skid Resistance Skid resistance testing and its use in acceptance is a concern of many agencies because of a potential liability problem. Excluding Virginia and the FHWA, most state agencies conduct skid resistance testing on an as-needed or requested basis and as required by FHWA policies on federal routes, but have no plans for including friction into their acceptance plans. Due to its limited use, skid resistance was not considered in the field study. 4.12 Summary of Evaluation Table 24 shows a summary of technologies evaluated or used by different state agencies. Tables 25 through 28 summarize some of the critical points of each NDT technology as they relate to routine QA operations and to inputs needed for the MEPDG. [Note: Tables 24 through 28 are included at the end of this chapter.] The ones selected include those with high to moderate applicability to QA and low to moderate risk for implementation of the technology. Those technologies not selected were those with a low applicability to QA or a high risk for implementation of the technology. 14 From personal contacts and correspondence: Agencies are considering the use of noise as a factor in their rehabilitation and surface layer selection policy, but do not believe that it should be a part of the acceptance plan. The reasoning for this position is that there is little information that can be used to design “quiet” HMA mixtures. 144

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report NDT Devices Included in the Field Evaluation The following lists the NDT technologies and devices that were selected for use in the field study, in no particular order: 1. Deflection Based Technologies—The FWD and LWD were selected for use because of the large number of devices that are being used in the U.S. and the large database that has been created under the FHWA-LTPP program. The LWD will be used to evaluate individual layers, especially unbound layers, while the FWD will be used to evaluate the entire pavement structure at completion to ensure that the flexible pavement structure or HMA overlay have met the overall strength requirements used in the structural design process. Deflection measuring devices are generally readily available within most agencies for their immediate use in QA. 2. Dynamic Cone Penetrometer—The DCP was selected for use because of its current use in QA operations in selected agencies and ability to measure the in-place strength of unbound layers and materials. In addition, the DCP does not require extensive support software for evaluating the test results. DCP equipment is being manufactured and marketed by various organizations, so its availability is not a problem. 3. Ground Penetrating Radar—GPR was selected for study because of its current use in pavement forensic and evaluation studies for rehabilitation design and for estimating both the thickness and air voids of pavement layers. If proven successful, this will be one of the more important devices used for acceptance of the final product by agencies, assuming that the interpretation of the data can become readily available on a commercial basis. The GPR air-coupled antenna was successfully used within the FHWA-LTPP program to measure the layer thickness within many of the 500- foot test sections. 4. Seismic Pavement Analyzer—Both the PSPA and DSPA were selected for use because it provides a measure of the layer modulus and can be used to test thin, as well as thick layers shortly after placement. This technology can also be used in the laboratory to test both HMA and unbound materials compacted to various conditions—different fluids content for unbound materials and soils or temperatures for HMA to evaluate the effect of fluids and temperature on materials. 5. GeoGauge—The GeoGauge has had mixed results from its use to test unbound pavement layers. It was selected for use because it is simple to use and provides a measure of the resilient modulus of unbound pavement layers and embankment soils and can be used to test typical lift thicknesses. 6. Non-Nuclear Electric Gauges; Non-Roller-Mounted Devices—Non-nuclear density gauges have a definite advantage over the nuclear ones simply from a safety standpoint. These gauges have been used on many projects but with varying results. They were selected for use because many agencies are allowing their use by contractors on a QC basis, and agencies are beginning to use the contractors QC 145

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report results for acceptance. They also represent the baseline comparison to the results from the nuclear gauges for measuring density for use in acceptance procedures. Thus, the location specific non-nuclear density gauges were selected. The gauges selected for initial use were the PQI and PaveTracker for HMA mixtures, while the EDG was selected for unbound materials. NDT Devices Excluded from the Field Evaluation The following lists some of the basic reasons for excluding specific NDT technologies and devices from the field evaluation study. • Roller-Mounted-Density/Stiffness Devices—Non-nuclear density and stiffness monitoring devices attached to the rollers (Bomag Variocontrol and Onboard Measuring System) were excluded because these devices have not been extensively used for QC, no agency has immediate plans to implement them in their project requirements for future use, and there are a limited number of these rollers available for contractor use. Although the roller-mounted devices were excluded from the experimental plan for the field evaluation study, the roller manufacturers were contacted to determine their availability and use on selected projects. Thus, they were not totally excluded from the field evaluation. • Surface Condition Systems—None of the surface condition measuring systems or devices was recommended for further evaluation under NCHRP Project 10-65. Although the initial IRI is an input to the MEPDG, the smoothness measuring devices used for acceptance of the wearing surface are already included in many agencies QA programs. In addition, none of the devices provide an estimate of the volumetric and structural properties of the wearing surface. • Noise and Friction Methods—Noise and friction measuring devices were excluded from further consideration, because these properties are not needed in the MEPDG. Or any other structural design procedure and no agency is considering their use for acceptance. • Infrared Tomography—The infrared cameras and sensors were excluded from the field evaluation because their output only provides supplemental information to current acceptance plans. In other words, the devices are used to identify “cold spots” or temperature anomalies and other test methods are still used to determine whether the contractor has met the density specification. This statement does not imply that this technology should be abandoned or not used—the infrared cameras and sensors do provide good information and data on the consistency of the HMA being placed by the contractor. • Other Ultrasonic Test Methods—The IE and impulse response methods, as well as the ultrasonic scanners, were excluded because they are perceived to have a high risk of implementation into practical and effective QA operations. 146

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report Table 24. Technologies Used by State Agencies as a Research, Forensic, or Evaluation Tool Based on Information Collected During the Project Survey15 NDT Technology State Highway Agencies Using Technology FWD Arizona, Alabama, California, Connecticut, Florida, Georgia, Illinois, Minnesota, Mississippi, Missouri, Nevada, Ohio, Oklahoma, Texas, Utah, Virginia, Washington, Wisconsin DCP California, Colorado, Florida, Illinois, Iowa, Kansas, Kentucky, Maryland, Michigan, Minnesota, Mississippi, Missouri, Montana, Nevada, New Jersey, North Carolina, North Dakota, Ohio, Oklahoma, Pennsylvania, South Carolina, Texas, Utah, Virginia, Wisconsin GPR Florida, Georgia, Illinois, Minnesota, Missouri, Oklahoma, Texas, Nevada, North Dakota, Oklahoma, Virginia, Washington, Wisconsin PSPA & DSPA Colorado, Florida, Texas GeoGauge FHWA (other agencies have used the device, but only during the pooled fund studies sponsored by FHWA) IR Connecticut, Minnesota, Missouri, Nevada, Texas, Washington Non-nuclear density Alabama, Connecticut, Florida, Illinois, Minnesota, Mississippi, Missouri, Nevada, New York, Oklahoma, Texas, Washington, Wisconsin Smoothness Arizona, Alabama, Colorado, California, Connecticut, Florida, Illinois, Minnesota, Mississippi, Missouri, Nevada, Ohio, Oklahoma, Virginia, Washington, Wisconsin Skid and Texture Arizona, Alabama, Colorado, California, Florida, Illinois, Minnesota, Mississippi, Missouri, Nevada, Ohio, Oklahoma, Virginia, Washington, Wisconsin 15 This summary does not reflect the status of all State DOTs in the U.S., because not all 50 States were consulted in this survey. Detailed information about the DCP was obtained from the Minnesota Road Research Section, Office of Materials, MnDOT, as well as from the Oklahoma and Montana DOTs and consulting organizations. 147

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report Table 25. Summary Evaluation Results for DCP, Deflection-Based Methods, GPR, and Infrared Devices for Use in QA Deflection Evaluation Factor DCP FWD LWD GPR Infrared Acceptance No No No No No Test Method Used for QA Process Control No No No No No HMA Mixes NA Layer Modulus Composite Modulus Thickness; Density/Voids NA Material Properties Estimated Unbound Layers Strength Thickness Layer Modulus Composite Modulus Thickness; Density NA Modulus Estimated by: Regression equations Back- calculation Forward- calculation NA NA Structural Design Yes; Strength & Thickness Yes; Back- Calculated E No Yes; Density, Thickness NA Output Applicable To: Mixture Design NA No No No NA Status of Use State-of-Practice State-of- Practice State-of-Art State-of-Art State-of-Art Test Protocol Used by Agencies Yes, ASTM Standardized Yes, ASTM Standardized No Yes Yes, ASTM Standardized Acceptance Yes Yes, final structure Yes Yes, ASTM Standardized No Test Method Applicable To: Process Control Yes No Yes No Yes Test Local Point Specific Area Specific Point Specific Continuous Continuous Identification of Localized Defects Yes, but requires more testing No No Yes, low density, segregation Yes; cold spots Applicability to Test Thin Layers No No Yes Yes Yes Production Rate 1 test/10 min. 1 test/5 min. 1 test/2 min. High High Analysis Effort Easy Difficult Easy Difficult Easy Cost of Equipment $15,000 $125,000 $20,000 $40,000 $12,000 Applicability of Device High Moderate Moderate High Low for QC; NA for QA Practical & Effective QA Operations Risk of Implementation Low Moderate Moderate Low Low 148

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report Table 26. Summary Evaluation Results for Magnetic Imaging, Density, and Smoothness Devices for Use in QA Density Evaluation Factor Magnetic Imaging Non-Nuclear Humbolt, Nuclear Roller- Mounted Gauges Smoothness, Profilometer Acceptance No No No No Yes Test Method Currently Used for QA Process Control No Yes Yes No No HMA Mixes Thickness Density Density Density & Stiffness IRI, Surface Material Properties Estimated Unbound Layers NA Density Density Density & Stiffness NA Modulus Estimated by: NA NA NA NA NA Structural Design Yes, mat thickness No No No Yes, initial IRI Output Applicable To: Mixture Design No No No No NA Status of Use Research & Forensics Research Research Development State-of- Practice Test Protocol Used by Agencies None Supplier Supplier None Yes; AASHTO Standardized Acceptance Yes; mat thickness Yes, density Yes, density No Yes Test Method Applicable To: Process Control No Yes Yes Yes Yes Test Local Area Specific Point Specific Point Specific Continuous Continuous Identification of Localized Defects No Yes, but more testing is needed Yes, but more testing is needed No Yes, bumps Applicability to Test Thin Layers Yes Yes Yes Yes Yes, surface Production Rate Based on Location of metallic washers 1 test/2 min. 1 test/ 2 min. High High Analysis Effort Limited Easy Easy Limited Extensive Cost of Equipment $15,000 $12,000 Van-$100,000 Light-Weight - $35,000 Applicability of Device Low Moderate Low Moderate High Practical & Effective QA Operations Risk of Implementation High Moderate Moderate High Low 149

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report Table 27. Summary Evaluation Results for Seismic and Steady State Vibratory Devices for Use in QA Seismic Evaluation Factor SPA; PSPA & DSPA Impact-Echo Scanners; I-E, SASW GeoGauge Acceptance No No No No Test Method Currently Used for QA Process Control No No No No HMA Mixes Thickness, Modulus Thickness Thickness, Modulus No Material Properties Estimated Unbound Layers Thickness, Modulus NA NA Modulus Modulus Estimated by: Calibration NA Calibration Direct Reading Structural Design Thickness, Modulus Thickness Thickness Modulus Output Applicable To: Mixture Design Modulus NA NA NA Status of Use Limited R&D R&D Limited Test Protocol Used by Agencies Forensics Forensics No No Acceptance Yes Yes Yes Yes Test Method Applicable To: Process Control Yes Yes Yes Yes Test Local Point Specific Point Specific Continuous Point Specific Identification of Localized Defects Yes, but more testing is needed Yes, but more testing is needed Yes Yes, but more testing is needed Applicability to Test Thin Layers Yes Yes Yes Yes, to some degree Production Rate 1 test/2 min. 1 test/5 min. High 1 test/5 min. Analysis Effort Extensive Extensive Extensive Minimal Cost of Equipment $25,000 $20,000 $35,000 $25,000 Applicability of Device High Moderate High High Practical & Effective QA Operations Risk of Implementation Moderate High High Moderate 150

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 151 Table 28. Summary Evaluation Results for Noise and Skid Devices for Use in QA Evaluation Factor Noise Skid Acceptance No No Test Method Currently Used for QA Process Control No No HMA Mixes NA Friction, surface Material Properties Estimated Unbound Layers NA NA Modulus Estimated by: NA NA Structural Design NA NA Output Applicable To: Mixture Design No No Status of Use R&D State-of-Art Test Protocol Used by Agencies Limited Yes Acceptance Yes Yes Test Method Applicable To: Process Control No No Test Local Continuous Continuous Identification of Localized Defects No Yes, surface Applicability to Test Thin Layers Yes Yes Production Rate High High Analysis Effort Extensive Moderate Cost of Equipment Applicability of Device Low Low Practical & Effective QA Operations Risk of Implementation High High

Page intentionally left blank.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 153 CHAPTER 5 FIELD EVALUATION OF NDT DEVICES As noted in chapter 1, the field evaluation was divided into two parts, referred to as Part A and Part B. Part A was to confirm the applicability of NDT technologies that were judged to be ready and appropriate for implementation into routine, practical, and effective QA programs for flexible pavement construction and HMA overlays. Part A of the field evaluation also included selecting those NDT technologies and devices that can consistently and accurately identify construction anomalies. Part B of the field study was to use those NDT technologies and devices selected from Part A and refine the test protocols and data interpretation procedures for judging the quality of flexible pavement construction. Part B also included identifying limitations and boundary conditions of selected NDT test methods. This chapter summarizes all testing completed within the field evaluation. The interpretation and analyses of the data are included in Part III of the research report. 5.1 Projects and Materials Included in Field Evaluation Table 29 summarizes the project and materials included in the field evaluation for Parts A and B. Appendix B provides a discussion of all projects and materials included in the field evaluation. The anomalies included within specific segments or lots are identified and discussed in the following sections of this chapter. 5.2 Field Testing Plan The NDT technologies and devices recommended for use in the field evaluation were identified in chapter 4. Table 30 lists the specific devices that were used for each project listed in Table 29, while Figure 37 shows the general layout of test points for each section or lot within a project under Part A. Areas with anomalies were included in the test plan to confirm that the NDT devices can estimate at least one quality characteristic and identify areas with anomalous features. The testing plan for the segments included in Part B was similar to Figure 37, but did not purposely include any localized anomalies. The remainder of this chapter presents the NDT data measured on the projects included within Part A of the field evaluation for identifying the localized anomalies within each project. The NDT data collected on the Part B projects are discussed in more detail in chapters 7 and 8. 5.3 NDT Test Results of Unbound Materials and Soils This section presents the NDT responses measured on the unbound materials at each project listed in Table 29. It also provides a brief evaluation of the materials based on those measured responses and compares the responses measured by different NDT devices on the same material.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 154 Table 29. Listing of Projects and Material Types Included in the Field Evaluation Part Project Identification & Location Layer/Material Evaluated A 1 MnRoad Demonstration (Note 1) Embankment Low Plasticity, Fine-Grained Soil HMA Dense-Graded Base Mixture Granular Base Class 6, Crushed Aggregate A 2 TH-23 Reconstruction Project; Wilmar/Spicer Minnesota Class 5 Embankment Low Plasticity, Improved Soil with Gravel & Large Aggregate Particles A 3 I-85 Overlay Project; Auburn, Alabama HMA 12.5mm Stone Matrix Asphalt Mix; PG76-22 HMA Coarse-Graded Base Mixture; PG67-22 Granular Base Crushed Limestone Base A 4 US-280 Reconstruction Project; Opelika, Alabama Embankment Improved Soil; Aggregate-Soil Mix A 5 I-85 Ramp Construction Project; Auburn, Alabama Embankment Low Plasticity, Fine-Grained Soil HMA Coarse-Graded 19mm Base Mixture; PG64-22 A 6 SH-130 New Construction Project; Georgetown, Texas Embankment Coarse-Grained Aggregate/Soil; Improved Soil A 7 SH-21 Widening Project; Caldwell, Texas Subgrade High Plasticity Fine-Grained Soil with Gravel HMA Coarse-Graded Base Mixture HMA Fine-Graded Wearing Surface B 8 US-47 Widening Project; St. Clair, Missouri Granular Base Crushed Aggregate; In place material HMA Coarse-Graded Base Mixture B 9 US-47 Reconstruction Project; Union, Missouri Granular Base Crushed Aggregate Base HMA Dense-Graded Binder Mixture; Type 3C B 10 I-75 Rehabilitation Project, Rubblization; Saginaw, Michigan HMA Fine-Graded Wearing Surface HMA Coarse-Graded Base Mix; PG58-28 Granular Base Crushed Gravel with Surface Treatment; Class 5 B 11 US-2 New Construction; North Dakota Embankment Soil-Aggregate Mixture HMA Coarse-Graded Binder Mixture B 12 US-53 New Construction; Toledo, Ohio Granular Base Crushed Aggregate; Type 304 HMA Coarse-Graded Mixture; CMHB B 13 I-20 Overlay; Odessa, Texas Granular Base Crushed Stone B 14 County Road 103; Pecos, Texas Granular Base Caliche, Aggregate Base NCAT; Alabama Overlay, Section E-5, Opelika, Alabama HMA Wearing Surface with 45% RAP; PG67, no modifiers used. NCAT; Alabama Overlay, Section E-6, Opelika, Alabama HMA Wearing Surface with 45% RAP; PG76 with SBS. B 15 NCAT; Alabama Overlay, Section E-7, Opelika, Alabama HMA Wearing Surface with 45% RAP; PG76 with Sasobit. HMA PMA Mixture with SBS; PG76 HMA Neat Asphalt Binder Mix; PG67 B 16 NCAT; Florida; Structural Test Sections N-1 & N-2 Granular Base Limerock Base HMA Polymer Modified Asphalt Mix; PG76 (SBS) HMA Neat Asphalt Binder Mix; PG64 B 17 NCAT; Missouri; Structural Test Section N-10 Granular Base Crushed Limestone B 18 NCAT; Oklahoma; Structural Test Sections N-8 & N-9 Subgrade Soil High Plasticity Clay with Chert Aggregate HMA Coarse-Graded Base Mix; PG67; Limestone B 19 NCAT; South Carolina; Structural Test Section S-11 Granular Base Crushed Granite Base CMHB – Coarse Matrix, High Binder Content (mixture type term used by the Texas DOT specifications) PG – Performance Grade PMA – Polymer Modified Asphalt RAP – Recycled Asphalt Pavement NOTE: Shaded cells with italic text in table were excluded from the field evaluation for different reasons; see explanation in Appendix B.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 155 Table 30. NDT Devices Used at Each of the Field Evaluation Projects NDT Technologies Part Project ID Material Impact (DCP) Deflect. Seismic GeoGauge GPR Non- Nuclear Density Roller Mounted Devices A MnRoad Demonstration Embankment √ √ --- --- --- --- √ HMA NA --- √ NA √ √ --- Crushed Stone √ √ √ √ √ √ √ A TH-23 Project, MN Embankment √ √ √ √ √ √ --- A I-85 Overlay, AL SMA NA √ √ NA √ √ √ HMA NA √ √ NA √ √ √ A US-280, AL Crushed Stone √ √ √ √ √ √ --- A I-85 Ramp, AL Embankment √ √ √ √ √ √ √ HMA NA √ √ NA √ √ --- A SH-130, TX Embankment √ √ √ √ √ √ --- A SH-21, TX Subgrade Soil √ √ √ √ --- √ √ B US-47, MO HMA NA --- √ NA --- √ --- B I-75, MI HMA NA --- √ NA --- √ --- HMA NA --- √ NA --- √ --- Crushed Stone √ --- √ √ --- --- --- B US-2; ND Embankment √ --- √ √ --- --- --- HMA NA --- √ NA --- √ --- B US-53, OH Crushed Stone √ --- √ √ --- --- --- B I-20, TX HMA NA --- √ NA --- √ --- B Co. Rd. 103, TX Caliche Base √ --- --- √ --- --- --- B NCAT, Alabama HMA NA --- √ NA √ √ √ HMA NA --- √ NA √ √ --- B NCAT, Florida Limerock Base √ --- √ √ --- --- --- HMA NA --- √ NA √ √ --- B NCAT, Missouri Crushed Stone √ --- √ √ --- --- --- B NCAT, Oklahoma Subgrade Soil √ --- √ √ --- --- --- HMA NA --- √ NA √ √ --- B NCAT, South Carolina Crushed Stone √ --- √ √ --- --- √

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 156 50025050 350 X X X X X X X X X X X X X X X 150 450 NDT Technology (Unbound Layers) Frequency of Test Segments or Lots; Localized Anomalies • Non-Roller-Mounted Density Devices (EDG) • Steady-State Vibratory (GeoGauge) • Seismic (DSPA) • Deflections (LWD, FWD) • Impact/Penetration Devices (DCP) • GPR, Air Horned Antenna • Triplicate Tests • Clustered Tests • Varying density between & within a lot • Varying water content between & within a lot • Material types NDT Technology (HMA Layers) Frequency of Test Segments or Lots; Localized Anomalies • Non-Roller-Mounted Density Devices (PQI, PaveTracker) • Seismic (PSPA) • Deflections (FWD) • GPR, Air Horned Antenna • Triplicate Tests • Clustered Tests • Varying density between & within lots • Varying asphalt content between lots • Mixture types Figure 37. General Layout of Test Points and Testing Sequence for Each Section or Lot Included within a Project in Part A of the Field Evaluation The initial testing under Part A of the field evaluation was to confirm that the NDT technologies can accurately identify differences in construction quality of unbound pavement layers. The specific hypothesis used for this part of the field evaluation was that the NDT technology and device can detect changes in the physical condition of the materials. Table 31 summarizes the anomalies between the unbound materials placed along each project. During nondestructive testing, the NDT operators were not advised of these anomalies. Conversely, no anomalies were planned for the Part B field evaluation projects. 5.3.1 Impact Penetration Test—DCP The manual DCP was used to estimate the in-place strength of the unbound materials in accordance with ASTM D 6951 (see Figure 3 in chapter 3). However, the sequence of drops and penetration readings were modified based on the layers and thicknesses being evaluated at each project site.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 157 Table 31. Description of the Local Anomalies in the Unbound Materials and Soils Placed Along Each Project Included in Part A Project Identification Unbound Sections Description of Differences Along Project Area 2, No IC Rolling No planned difference between the points tested. SH-21 Subgrade, High Plasticity Clay; Caldwell, Texas Area 1, With IC Rolling With IC rolling, the average density should increase; lane C received more roller passes. Lane A of Sections 1 & 2 Prior to IC rolling, Lane A (which is further from I-85) had thicker lifts & a lower density. I-85 Embankment, Low Plasticity Clay; Auburn, Alabama All sections tested After IC rolling, the average density should increase & the variability of density measurements should decrease. South Section – Lane C Construction equipment had disturbed this area. In addition, QA records indicate that this area has a lower density. TH-23 Embankment, Silt-Sand-Gravel Mix; Spicer, Minnesota North Section – Lane A The area with the higher density and lower moisture content – a stronger area. SH-130, Improved Embankment, Granular; Georgetown, Texas All sections tested No planned differences between the areas tested. Section 2 (middle section) – Lane C Curb and gutter section; lane C was wetter than the other two lanes because of trapped water along the curb from previous rains. The water extended into the underlying layers. TH-23, Crushed Aggregate Base; Spicer, Minnesota Section 1 (south section) – Lane A Area with a higher density and lower moisture content; a stronger area. US-280, Crushed Stone Base; Opelika, Alabama Section 4 Records indicate that this area was placed with higher moisture contents and is less dense. It is also in an area where water (from previous rains) can accumulate over time. For each point, the test was begun by using one seating drop from full height. The penetration was recorded for the seating drop. The penetration was then recorded after each drop or five successive drops (depending on its strength) throughout the layer thickness. One DCP test was performed at each test point. At a few test locations, however, refusal of the DCP occurred when large aggregates were encountered (see Figure 38). When refusal occurred, the DCP was moved slightly and the test repeated. The penetration rate has been correlated to resilient modulus, as presented in the chapter 3. Equation 33 was used to calculate the resilient modulus for each test point. Table 32 lists the average resilient modulus values for each area tested within Part A, while Table 33 lists the average resilient modulus values for the projects included in Part B. The DCP test or penetration of the device was continued into the supporting layer. All incremental

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 158 penetration rates are provided in Appendix C. The average penetration rates through the test material were used to calculate the average elastic modulus at each test point. Figure 38. Photo of the DCP Test and Large Aggregate Particles Encountered at Some of the Projects Resulting in Refusal of the Test ( ) 64.0 12.1 2926.17 ⎟⎟⎠ ⎞ ⎜⎜⎝ ⎛= DPI ER (33) Where: ER = Resilient modulus, MPa. DPI = Penetration rate or index, mm/blow. Figure 39 compares the standard deviation to the mean elastic modulus calculated from the DCP penetration rate for both fine and coarse-grained materials. As shown, the standard deviation increases with material strength or increasing elastic modulus. In addition, the coarse-grained materials were found to be consistently stronger than fine-grained soils, as expected. Large aggregate particles in the embankment soil caused refusal of the DCP in localized areas. These particles found near the surface also had an impact on the DSPA and GeoGauge readings.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 159 Table 32. Summary of Resilient Modulus Values Calculated from DCP Test Results Measured Within Specific Sections of the Projects Included in Part A, ksi Location or Designated Area Project Identification A B C D Mean, ksi 5.41 6.71 6.04 5.47 I-85 Low Plasticity Clay; Section 1; Before IC Rolling COV, % 13.7 48.4 29.4 24.3 Mean, ksi 4.98 5.32 5.44 4.73 I-85 Low Plasticity Clay; Section 2; Before IC Rolling COV, % 11.9 25.2 22.7 25.2 Mean, ksi 6.66 7.74 7.10 6.23 I-85 Low Plasticity Clay; Section 1; After IC Rolling COV, % 19.5 38.0 24.7 26.6 Mean, ksi 6.07 6.20 6.54 6.01 I-85 Low Plasticity Clay; Section 2; After IC Rolling COV 18.5 15.3 12.5 26.5 Mean, ksi --- --- --- 11.9 SH-21, High Plasticity Clay; Area 2, No IC Rolling COV, % --- --- --- 16.6 Mean, ksi 9.1 8.3 9.9 --- SH-21, High Plasticity Clay; Area 1, With IC Rolling COV, % 40.2 16.8 19.1 --- Mean, ksi 14.77 15.55 11.47 --- TH-23 Embankment, Silt-Sand- Gravel Mix; South Section COV, % 4.8 14.8 22.3 --- Mean, ksi 18.52 20.22 17.80 --- TH-23 Embankment, Silt-Sand Gravel Mix; North Section COV, % 21.5 26.2 28.3 --- Mean, ksi 20.50 18.65 24.18 --- SH-130 Granular, Improved Embankment; Section 1 COV, % 14.0 25.0 24.0 --- Mean, ksi 21.31 20.32 18.85 --- SH-130 Granular, Improved Embankment; Section 2 COV, % 43.4 36.8 10.4 --- Mean, ksi 22.99 23.87 19.18 --- SH-130 Granular, Improved Embankment; Section 3 COV, % 37.5 58.9 40.6 --- Mean, ksi 42.25 33.07 18.55 --- TH-23 Crushed Aggregate; Middle Section COV, % 46.6 38.3 20.0 --- Mean, ksi 48.23 44.66 24.11 --- TH-23 Crushed Aggregate; South Section COV, % 50.5 20.6 16.6 --- Mean, ksi 53.79 --- US-280, Crushed Stone; Section 1 COV, % 23.8 --- Mean, ksi 45.90 --- US-280, Crushed Stone; Section 2 COV, % 21.8 --- Mean, ksi 51.19 --- US-280, Crushed Stone; Section 3 COV, % 8.9 --- Mean, ksi 34.31 --- US-280, Crushed Stone; Section 4 COV, % 11.9 --- Note: The shaded cells designate those areas with anomalies (refer to table 31); black cells denote weaker areas, while the gray cells denote stronger areas for a specific project.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 160 Table 33. Summary of Resilient Modulus Values Calculated from DCP Test Results Measured within Projects Included in Part B, ksi Project Identification Section and Material Mean Modulus, ksi Coefficient of Variation, % Standard Deviation of Means, ksi Crushed Gravel; No prime coat 28.7 2.9 0.095 Crushed Gravel; Prime Coat 31.7 14.01 4.44 US-2, ND Embankment, Soil-Aggr. 22.2 15.6 3.45 US-53, OH Crushed Aggregate 31.9 23.2 7.4 CR-103, TX Caliche Base 22.5 6.5 2.81 Limerock Base, Sect. N1 43.1 22.8 9.82 NCAT, Florida Limerock Base, Sect. N2 49.3 20.2 9.97 NCAT, Missouri Crushed Limestone, Sect. N10 --- --- --- NCAT, SC Crushed Granite --- --- --- High Plasticity Clay, Sect. N8 8.15 27.4 2.23 NCAT, Oklahoma High Plasticity Clay, Sect. N9 9.02 21.1 1.91 NOTE: The DCP was excluded from the Missouri and South Carolina test sections because of problems encountered in compacting these base materials that delayed the final completion of these test sections. 0 5 10 15 20 25 30 0 10 20 30 40 50 60 Mean Elastic Modulus, DCP, ksi S ta nd ar d D ev ia tio n, k si Fine-Grained Coarse-Grained Poly. (Fine-Grained) Power (Coarse-Grained) Figure 39. Relationship Between the Standard Deviation and Mean of the Elastic Modulus Values of Unbound Materials Calculated from the DCP Penetration Rate The cells in Table 32 that correspond to those conditions listed in Table 31 have been shaded. The following list summarizes the results of the DCP tests in accordance with those anomalies identified in Table 31:

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 161 • I-85 Low Plasticity Soil Embankment—The DCP found both outside lanes (lanes A and D) to be weaker than the two inside lanes, both before and after IC rolling. The DCP results also indicate a consistent increase in the embankment’s strength after IC rolling, but not a reduction in variability of strength. • SH-21 High Plasticity Clay Soil—The DCP found area 1, with IC rolling and testing, to be weaker than area 2. This observation is inconsistent with construction records. However, area 2 was found to have some gravel mixed in with the high plasticity clay near the surface (top 6 to 8 inches) during the sampling process. This could explain the higher strengths in area 2. • TH-23 Gravelly, Silty Clay (Silt-Sand Gravel Mix) Embankment—The DCP correctly found lane C of the south section to be the weaker of the areas tested, and found the entire north section to be significantly stronger than the south section. Lane A was not stronger than the other two lanes tested in the north section, which is inconsistent with construction records. • SH-130 Improved Granular Embankment—The DCP found no significant difference between the areas tested, which was planned. • TH-23 Crushed Aggregate Base—The DCP found lane C in the middle section to be the weaker and lane A in the south section to be stronger. The paving schedule prevented the north section from being tested with the DCP. • US-280 Crushed Stone Base—The DCP found area 4 to be softer of the four areas tested. However, its strength is still high and consistent with adequately compacted crushed stone. 5.3.2 Deflection Testing—FWD and LWD Two types of deflection measuring equipment were used on some of the projects: the trailer- mounted FWD and the portable FWD or LWD. Falling Weight Deflectometer Deflection basins were measured with the FWD in accordance with the test protocol being used in the LTPP program (see Figure 6 in chapter 3). The procedure was to use two seating drops, followed by two drops at each drop height. Three drop heights were used at each test point. The deflection basins were recorded for each drop, including the seating drops. After the first set of tests, the FWD was moved forward (where the loading plate would be in contact with a different area) and the test sequence repeated. This sequence of drops and replicate testing was used at each test point. The 18-inch-diameter loading plate was used for all unbound materials testing, and the deflections were measured at seven sensors at the spacing recommended for use in LTPP. The deflection basins were used to forward-calculate the elastic modulus of the layer being evaluated using the procedure developed by Stubstad et al. (2003). The calculated elastic modulus values are summarized in Table 34 for the US-280 project. Elastic moduli were also backcalculated using other traditional methods and more sophisticated pattern recognition methods. The forward-calculation method resulted in the least variation of elastic moduli within a specific area.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 162 Light Weight Deflectometer Deflections were also measured with different LWD devices in accordance with the manufacturer’s recommendations. One to three LWD devices were used on the Part A projects. These devices are defined as the Loadman, Dynatest Prima 100, and Carl Bro. The Loadman and Dynatest Prima 100 were used to measure the deflection at the center of the loading plate, while the Carl Bro device was used to measure the deflections under the loading plate and at two additional sensors spaced at 8 and 12 inches from the loading plate. Figure 40 shows the LWDs that were used on selected projects. Table 34. Summary of the Calculated Resilient Modulus Values from the FWD Deflection Basins, ksi Section of Project Project Identification Area A B C D 1 18.1 15.7 15.7 --- 2 8.10 6.3 7.2 --- 3 16.7 17.9 20.2 --- 4 27.5 26.8 25.9 --- 5 32.3 35.1 38.3 --- Mean, ksi 20.794 --- Std. Dev., ksi 9.979 --- US-280; Crushed Stone; Section 1 COV, % 48.0 --- 1 16.6 13.8 --- --- 2 11.9 8.6 9.2 --- 3 15.5 18.2 14.9 --- 4 26.4 32.1 30.0 --- 5 --- 31.4 28.7 --- Mean, ksi 19.798 --- Std. Dev., ksi 8.686 --- US-280; Crushed Stone; Section 2 COV, % 43.9 --- 1 32.3 31.7 26.9 --- 2 14.2 11.7 10.6 --- 3 7.8 8.2 9.2 --- 4 22.3 18.5 20.3 --- 5 20.3 18.7 19.6 --- Mean, ksi 18.166 --- Std. Dev., ksi 7.969 --- US-280; Crushed Stone; Section 3 COV, % 43.9 --- 1 5.5 5.0 5.4 --- 2 5.7 5.4 5.7 --- 3 7.3 7.2 7.5 --- 4 7.5 6.6 7.7 --- 5 6.2 6.8 5.7 --- Mean, ksi 6.352 --- Std. Dev., ksi 0.9196 --- US-280; Crushed Stone; Section 4 COV, % 14.5 --- Note: The shaded cells designate those areas with anomalies (refer to table 31); black cells denote weaker areas.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 163 (a) Loadman LWD used on selected projects. (b) Loading plate for the LWD. (c) Prima 100 LWD. Figure 40. LWDs Used for Testing the Unbound Materials and Soils

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 164 A seating drop was used to begin the test for each device, and both load and deflection were recorded. The seating drop was followed by five successive drops. Elastic modulus values were calculated from the measured load and deflections for each drop, in accordance with the procedures recommended by the individual manufacturers. The average elastic modulus values, excluding the seating drop, are provided in Table 35 for the Carl Bro device. Table 36 lists the average elastic modulus values calculated from the loads and deflections measured with the other LWD devices (Dynatest Prima 100 and Loadman). Table 35. Summary of the Elastic Modulus Values Calculated from the Deflections Measured with the CarlBro LWD Device, ksi Section of Project Project Identification Area A B C D Mean, ksi --- --- --- --- I-85 Low Plasticity Clay; Section 1; Before IC Rolling COV, % --- --- --- --- Mean, ksi --- --- --- --- I-85 Low Plasticity Clay; Section 2; Before IC Rolling COV, % --- --- --- --- Mean, ksi 9.767 8.989 13.06 8.145 I-85 Low Plasticity Clay; Section 1; After IC Rolling COV, % 20.5 31.6 6.5 84.0 Mean, ksi 11.78 I-85 Low Plasticity Clay; Section 2; After IC Rolling COV, % 47.1 Mean, ksi --- --- --- --- SH-21 High Plasticity Clay, Area 2; No IC Rolling COV, % --- --- --- --- Mean, ksi 8.7 7.3 12.9 --- SH-21 High Plasticity Clay, Area 1; With IC Rolling COV, % 27.9 36.3 45.8 --- Mean, ksi 6.082 5.264 5.552 --- TH-23 Embankment, Silt-Sand- Gravel Mix; South Section COV, % 14.0 27.6 14.9 --- Mean, ksi 4.685 4.618 4.800 --- TH-23 Embankment, Silt-Sand- Gravel Mix; North Section COV, % 13.9 23.6 27.9 --- Mean, ksi 27.8 23.6 21.7 --- SH-130 Granular, Improved Embankment, Section 1 COV, % 51.2 60.3 22.4 --- Mean, ksi 23.6 29.7 21.3 SH-130 Granular, Improved Embankment, Section 2 COV, % 42.7 26.2 28.2 Mean, ksi 21.4 30.2 20.7 SH-130 Granular, Improved Embankment, Section 3 COV, % 65.4 80.5 19.3 Mean, ksi 15.45 12.80 7.95 --- TH-23; Crushed Aggregate, Middle Section COV, % 53.6 42.8 9.0 --- Mean, ksi 17.66 21.10 8.67 --- TH-23 Crushed Aggregate, South Section COV, % 61.1 42.0 22.5 --- Mean, ksi 51.23 --- US-280 Crushed Stone; Section 1 COV, % 56.1 --- Mean, ksi 37.82 --- US-280 Crushed Stone; Section 2 COV, % 44.0 --- Mean, ksi 50.334 --- US-280 Crushed Stone; Section 3 COV, % 42.2 --- Mean, ksi 18.53 --- US-280 Crushed Stone; Section 4 COV, % 16.8 --- Note: The shaded cells designate those areas with anomalies (refer to table 31); black cells denote weaker areas, while the gray cells denote stronger areas within a specific project.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 165 Table 36. Summary of the Elastic Modulus Values Calculated from the Other LWD Test Results, ksi Loadman LWD Device Dynatest Prima 100 Project Identification Area A B C A B C Mean, ksi 3.085 3.029 1.036 --- --- --- TH-23 Embankment; South Section COV, % 19.9 64.7 51.6 --- --- --- Mean, ksi 5.200 4.488 3.000 --- --- --- TH-23 Embankment; North Section COV, % 47.1 44.2 81.8 --- --- --- Mean, ksi 25.922 44.704 16.026 --- --- --- TH-23 Crushed Aggregate; Middle Section COV, % 46.0 74.4 31.5 --- --- --- Mean, ksi 35.22 36.14 20.90 --- --- --- TH-23 Crushed Aggregate; South Section COV, % 77.3 56.6 18.2 --- --- --- Mean, ksi --- --- --- 32.87 US-280 Crushed Stone; Section 1 COV, % --- --- --- 41.0 Mean, ksi --- --- --- 14.20 US-280 Crushed Stone; Section 2 COV, % --- --- --- 37.2 Mean, ksi --- --- --- 26.21 US-280 Crushed Stone; Section 3 COV, % --- --- --- 21.5 Mean, ksi --- --- --- 9.64 US-280 Crushed Stone; Section 4 COV, % --- --- --- 20.3 Note: The shaded cells designate those areas with anomalies (refer to table 31); black cells denote weaker areas, while the gray cells denote stronger areas tested with a specific project. Comparison of Test Results from Deflection Based Devices Figure 41 compares the average elastic modulus calculated from the loads and deflections measured with various deflection measuring devices. As shown, the Carl Bro device consistently measured higher elastic modulus values than the Dynatest Prima 100 and FWD. The elastic modulus values from the Loadman are more diverse for the weaker layers and much higher for the stronger layers (Figure 41.a). Figure 42 presents a cumulative frequency diagram of the standard deviation or repeatability of the deflection based methods. The standard deviations in Figure 42 represent the variability between the five successive drops at the same test point. The repeatability of the LWD devices (excluding the Loadman device) is considered good, with a mean standard deviation less than 0.5 ksi. Figure 43 compares the standard deviation of the measurements made within an area to the mean elastic modulus calculated for that area. As shown, the standard deviation continues to increase with increasing elastic modulus. Similar to the DCP, the LWD measured consistently higher elastic modulus values for coarse-grained materials than for fine-grained materials, as expected. Figure 44 includes a comparison of the coefficient of variation in elastic modulus values determined with each of the deflection measuring devices to normalize differences caused by changes in material strength. The Carl Bro and Dynatest Prima 100 devices measured similar variability, while the Loadman device and FWD consistently measured higher variability.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 166 Thus, test results from the Carl Bro and Dynatest devices were used in comparison to the other NDT technologies. 0.1 1 10 100 0.1 1 10 100 Elastic Modululs, CarlBro Device, ksi El as tic M od ul us , O th er L W D D ev ic es , ks i LoadMan, Soil 1 Loadman, Soil 2 LoadMan, Base 1 LoadMan, Base 2 Prima100, Area 1 Prima100, Area2 Prima100, Area 3 Prima100, Area 4 Line of Equality (a) Other LWD devices, as compared to the CarlBro LWD device. 0 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 80 90 Elastic Modulus, CarlBro Device, ksi El as tic M od ul us , O th er D ef le ct io n- B as ed D ev ic es , k si FWD, Crushed Stone Line of Equality Pima-100, Crushed Stone (b) FWD and Prima 100 devices, as compared to the CarlBro LWD device. Figure 41. Comparison of the Average Elastic Modulus Calculated from Deflections and Loads Measured with Different LWD Devices

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 167 0 10 20 30 40 50 60 70 80 90 100 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 Standard Deviation of Elastic Modulus, LWD, ksi C um ul at iv e Fr eq ue nc y, % Low Plasticity Soil Crushed Aggregate Dense Crushed Stone Figure 42. Cumulative Frequency of the Standard Deviation from the Deflection-Based Test Methods 0 5 10 15 20 25 30 35 0 10 20 30 40 50 60 Mean Elastic Modulus, LWD Devices, ksi St an da rd D ev ia tio n, k si Fine-Grained Coarse-Grained Log. (Fine-Grained) Log. (Coarse-Grained) Figure 43. Relationship Between the Standard Deviation and Mean of the Elastic Modulus Values of Unbound Layers Calculated from Deflections

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 168 0 20 40 60 80 100 0 20 40 60 80 100 Coefficient of Variation, CarlBro Device, % C oe ffi ci en t o f V ar ia tio n, O th er D ef le ct io n D ev ic es , % Loadman Prima 100 FWD Line of Equality Figure 44. Comparison of the Coefficient of Variations for Calculated Elastic Modulus from Different Deflection Measuring Devices The cells in Tables 34, 35, and 36 that correspond to those conditions listed in Table 31 have been shaded. The following list summarizes the results of the deflection-based methods in accordance with those conditions listed in Table 31: • I-85 Low Plasticity Soil Embankment – The deflection-based methods were not used to test the embankment prior to IC rolling. No significant difference in stiffness was found between the areas tested after IC rolling, as planned. • SH-21 High Plasticity Clay – The deflection-based methods found lane C of area 1 to be stronger than lanes A and B, which is inconsistent with construction records. • TH-23 Gravelly, Silty Clay Embankment – The deflection-based tests found no significant difference in stiffness between the areas tested in the south section, and found the north section to be weaker than the south section, with the exception of the Loadman device. This finding is inconsistent with QA records and other tests. It is expected that the calculated modulus values are being influenced by the underlying foundation. The Loadman device resulted in low modulus values for the TH-23 embankment that are extremely variable. This result (low modulus values) is questionable based on visual observations of construction traffic using this area. Heavy construction equipment did not cause any visible deformation of the surface. • SH-130 Improved Granular Soil – The deflection-based methods found no consistent difference between the three areas tested, which was planned. • TH-23 Crushed Aggregate Base – The deflection-based methods found lane C to be the weakest of all areas tested, similar to the results from the DCP. The paving schedule prevented the north section from being tested with the deflection-based

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 169 devices. The deflection-based methods also found the south section to be stronger than the middle section. • US-280 Crushed Stone Base – The deflection-based methods found area 4 to be the weakest of the four areas tested, similar to the DCP results. However, the modulus values calculated from deflections for area 4 are inconsistent with a good quality crushed stone. It is expected that the calculated modulus values in this area are being influenced (lowered) by the underlying layers. 5.3.3 Ultrasonic Test—DSPA The DSPA was used to measure the seismic modulus of the unbound materials in accordance with the procedure developed by Nazarian et al. (2002). Triplicate tests were performed at each test point. One test or measurement was taken with the device parallel to the direction of compaction, the second measurement with the device 90 degrees to the first measurement (perpendicular to the direction of compaction), and the final measurement taken 180 degrees to the first measurement. Figure 16 in chapter 3 shows the DSPA in operation, while Tables 37 and 38 provide the average seismic modulus values measured along the Parts A and B projects, respectively. The test results from the DSPA will be discussed with the GeoGauge results in subsection 5.3.5. 5.3.4 Steady-State Vibratory Test—GeoGauge The GeoGauge was used to measure the resilient modulus of the unbound materials in accordance with the procedure recommended by the manufacturer—with one exception (see Figure 17 in chapter 3). The test was performed with and without a sand cushion below the plate on selected projects (SH-12, TH-23, and US-280), because one of the agencies that hosted a project had been using the gauge without a sand cushion. Triplicate tests were performed at each test point. The gauge was placed and seated on the surface by applying a slight pressure and rotation to ensure uniform contact—making sure that the surface and gauge were coupled. The gauge was then lifted and this sequence repeated. The sand cushion did make a difference in the measured values for some materials. The resilient modulus values were found to be greater when using the sand cushion on rough surfaces, similar to a crushed aggregate or granular base. A ratio of approximately 2.2 was determined between the two conditions. This ratio or difference (modulus measured with and without a sand cushion) decreased on fine-grained surfaces. In fact, no systematic difference (ratio equal to 1.0) was detected on the SH-21 project with high plasticity clay soil without surface shrinkage cracks. For consistency, however, the sand cushion was used in all testing. Table 39 summarizes the average resilient modulus values measured with the GeoGauge within each section for the Part A field evaluation projects, while Table 40 lists the average values measured on the Part B projects. The results from these projects are discussed in the next subsection, along with the DSPA test results.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 170 Table 37. Summary of the Seismic Modulus Measured with the DSPA within Specific Sections of the Projects Included in Part A, ksi Project Identification Area A B C D Mean, ksi 26.2 31.8 27.9 34.2 I-85 Low Plasticity Clay; Section 1, Before IC Rolling COV, % 28.4 7.7 14.1 20.9 Mean, ksi 24.1 27.2 38.3 44.4 I-85 Low Plasticity Clay; Section 2, Before IC Rolling COV, % 22.9 23.9 9.6 21.7 Mean, ksi 42.5 38.7 37.0 39.5 I-85 Low Plasticity Clay; Section 1, After IC Rolling COV, % 5.9 22.4 20.0 21.8 Mean, ksi 33.2 39.7 45.1 43.7 I-85 Low Plasticity Clay; Section 2, After IC Rolling COV, % 12.8 27.6 10.7 25.8 Mean, ksi --- --- --- 23.6 SH-21 High Plasticity Clay; Area 2, No IC Rolling COV, % --- --- --- 7.6 Mean, ksi 25.8 25.0 30.4 --- SH-21 High Plasticity Clay; Area 1, With IC Rolling COV, % 18.1 11.3 11.5 --- Mean, ksi 42.00 45.13 31.12 --- TH-23 Embankment, Silt-Sand- Gravel Mix; South Section COV, % 14.5 20.8 43.9 --- Mean, ksi 51.66 40.20 31.13 --- TH-23 Embankment, Silt-Sand- Gravel Mix; North Section COV, % 23.2 23.4 29.7 --- Mean, ksi 38.4 39.0 34.4 --- SH-130 Granular, Improved Embankment; Section 1 COV, % 9.0 23.0 22.1 --- Mean, ksi 33.5 38.5 35.3 --- SH-130 Granular, Improved Embankment; Section 2 COV, % 33.1 27.5 18.8 --- Mean, ksi 29.9 26.7 30.1 --- SH-130 Granular, Improved Embankment; Section 3 COV, % 15.8 21.1 6.6 --- Mean, ksi 71.87 119.9 61.4 --- TH-23 Crushed Aggregate; North Section COV, % 41.2 40.4 43.0 --- Mean, ksi 89.47 69.67 28.0 --- TH-23 Crushed Aggregate; Middle Section COV, % 79.7 48.6 37.2 --- Mean, ksi 112.8 108.6 62.8 --- TH-23 Crushed Aggregate; South Section COV, % 71.4 41.1 53.2 --- Mean, ksi 233.5 --- US-280 Crushed Stone, Section 1 COV, % 13.8 --- Mean, ksi 189.0 --- US-280 Crushed Stone; Section 2 COV, % 22.0 --- Mean, ksi 173.2 --- US-280 Crushed Stone; Section 3 COV, % 16.2 --- Mean, ksi 117.4 --- US-280 Crushed Stone; Section 4 COV, % 12.8 --- Note: The shaded cells designate those areas with anomalies (refer to table 31); black cells denote weaker areas, while the gray cells denote stronger areas tested within a specific project.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 171 Table 38. Summary of Resilient Modulus Values Estimated from the DSPA within Projects Included in Part B, ksi Project Identification Section Mean Modulus, ksi Coefficient of Variation, % Standard Error, ksi Crushed Gravel; No Prime Coat 52.0 35.3 18.35 Crushed Gravel; with Prime 81.9 22.9 18.79 US-2, ND Embankment, Soil-Aggr. 33.1 21.4 7.08 US-53, OH Crushed Aggregate 61.3 26.9 16.49 CR-103, TX Caliche Base --- --- --- Limerock Base, Sect. N8 80.3 21.2 16.96 NCAT, Florida Limerock Base, Sect. N9 96.5 24.1 23.22 NCAT, Missouri Crushed Limestone, Sect. N10 97.1 22.9 22.20 NCAT, SC Crushed Granite, Sect. S11 91.5 22.6 20.7 High Plasticity Clay, Sect. N8 43.0 14.1 6.05 NCAT, Oklahoma High Plasticity Clay, Sect. N9 40.2 17.6 7.08 5.3.5 Comparison of Test Results from the DSPA and GeoGauge Figure 45 compares the seismic and resilient modulus values measured with this technology. As shown, the seismic modulus values measured with the DSPA are greater than the resilient modulus values measured with the GeoGauge. The difference between the two values increases with stiffer and coarser materials. There is correspondence between the two seismic devices, but that difference is material dependent. Figure 46 presents a cumulative frequency diagram of the standard deviation or repeatability of the DSPA and GeoGauge. The standard deviations in Figure 46 represent the triplicate measurements taken at the same test point. The variability of measurements made with the DSPA (Figure 46.a) is higher than for the GeoGauge (Figure 46.b), especially for the stiffer and coarse-grained materials. The repeatability of the GeoGauge devices is considered good but is material dependent. The mean standard deviation of the GeoGauge varies from about 0.5 ksi for the weaker soils to 3.5 ksi for dense base materials. The DSPA has a mean standard deviation varying from about 1.5 ksi to over 21 ksi. A reason for the higher variability of the DSPA was the rotation of the sensor bar relative to the roller direction. The GeoGauge was not rotated between repeat readings because of the circular loading plate. Another reason for the higher variability is that the DSPA measures the stiffness of the layer evaluated, while the GeoGauge and other NDT devices (excluding the DCP) can be influenced by the supporting layers. More importantly, the moisture gradient is much greater nearer the surface which has a greater influence on those devices that measure material responses closer to the surface—the DSPA. Thus, the mean seismic modulus values and variance of those values should be higher. The measurement and effect of this resilient modulus gradient is discussed in more detail in chapter 7 and at the end of this chapter.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 172 Table 39. Summary of the Resilient Modulus Values Measured with the GeoGauge within Specific Sections of the Part A Projects, ksi Project Identification Area A B C D Mean, ksi 14.5 16.3 14.9 15.7 I-85 Low Plasticity Clay; Section 1, Before IC Rolling COV, % 20.7 7.4 19.2 12.2 Mean, ksi 10.6 15.9 17.1 18.1 I-85 Low Plasticity Clay; Section 2, Before IC Rolling COV, % 26.9 15.7 7.7 20.8 Mean, ksi 17.43 16.35 16.633 17.85 I-85 Low Plasticity Clay; Section 1, After IC Rolling COV, % 11.7 --- 27.5 --- Mean, ksi 18.42 18.50 19.64 19.4 I-85 Low Plasticity Clay; Section 2, After IC Rolling COV, % 7.6 --- 0.4 --- Mean, ksi --- --- --- 19.6 SH-21 High Plasticity Clay; Area 2, No IC Rolling COV, % --- --- --- 6.3 Mean, ksi 24.0 24.7 20.1 --- SH-21 High Plasticity Clay; Area 1, After IC Rolling COV, % 15.5 24.8 11.5 --- Mean, ksi 10.07 10.86 7.537 --- TH-23 Embankment, Silt-Sand- Gravel Mix; South Section COV, % 10.2 11.0 9.4 --- Mean, ksi 12.568 10.00 10.31 --- TH-23 Embankment, Silt-Sand- Gravel Mix; North Section COV, % 15.6 4.5 22.0 --- Mean, ksi 28.74 26.82 27.72 --- SH-130 Granular, Improved Embankment; Section 1 COV, % 14.2 15.3 9.0 --- Mean, ksi 22.92 26.71 25.21 --- SH-130 Granular, Improved Embankment; Section 2 COV, % 17.5 14.4 21.2 --- Mean, ksi 24.62 22.97 19.21 --- SH-130 Granular, Improved Embankment; Section 3 COV, % 7.7 1.5 17.2 --- Mean, ksi 13.64 15.16 12.374 --- TH-23 Crushed Aggregate; North Section COV, % 11.1 10.2 9.1 --- Mean, ksi 12.97 12.55 9.838 --- TH-23 Crushed Aggregate; Middle Section COV, % 25.0 15.8 17.6 --- Mean, ksi 15.64 14.37 11.718 --- TH-23 Crushed Aggregate; South Section COV, % 24.3 14.5 16.2 --- Mean, ksi 48.84 --- US-280 Crushed Stone; Section 1 COV, % 7.9 --- Mean, ksi 49.98 --- US-280 Crushed Stone; Section 2 COV, % 5.0 --- Mean, ksi 44.96 --- US-280 Crushed Stone; Section 3 COV, % 9.9 --- Mean, ksi 35.12 --- US-280 Crushed Stone; Section 4 COV, % 4.6 --- Note: The shaded cells designate those areas with anomalies (refer to table 31); black cells denote weaker areas, while the gray cells denote stronger areas tested within a specific project.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 173 Table 40. Summary of Resilient Modulus Values Estimated with the GeoGauge Within Projects Included in Part B, ksi Project Identification Section Mean Modulus, ksi Coefficient of Variation, % Standard Error, ksi Crushed Gravel, No Prime Coat 17.2 8.61 1.93 Crushed Gravel, without Prime 26.7 16.9 3.54 US-2, ND Embankment, Soil-Aggr. 13.1 20.1 2.75 US-53, OH Crushed Aggregate 23.5 7.01 3.29 CR-103, TX Caliche Base 26.6 11.9 2.81 Limerock Base, Sect. N1 49.7 11.6 5.95 NCAT, Florida Limerock Base, Sect. N2 49.8 13.1 12.86 NCAT, Missouri Crushed Limestone, Sect. N10 25.7 8.1 5.28 NCAT, SC Crushed Granite, Sect. S11 15.1 8.31 2.09 High Plasticity Clay, Sect. N8 25.0 8.32 4.36 NCAT, Oklahoma High Plasticity Clay, Sect. N9 26.8 7.22 3.62 0 50 100 150 200 250 300 0 10 20 30 40 50 60 Resilient Modulus, GeoGauge, ksi Se is m ic M od ul us , D SP A , ks i I-85 Soil TH-23 Soil TH-23 Base US-280 Base Line of Equality SH-130 Improved Soil Figure 45. Comparison of the Seismic Modulus Values Measured with the DSPA and Resilient Modulus Values Measured with the GeoGauge

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 174 0 10 20 30 40 50 60 70 80 90 100 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 Standard Deviation of Elastic Modulus, DSPA, ksi C um ul at iv e Fr eq ue nc y, % Low Plasticity Soil High Plasticity Soil Crushed Aggregate Dense Crushed Stone (a) Standard deviation or repeatability for the DSPA. 0 10 20 30 40 50 60 70 80 90 100 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 Standard Deviation of Elastic Modulus, GeoGauge, ksi Cu m ul at iv e Fr eq ue nc y, % Low Plasticity Soil High Plasticity Soil Crushed Aggregate Dense Crushed Stone (b) Standard deviation or repeatability for the GeoGauge. Figure 46. Cumulative Frequency of the Standard Deviation from the DSPA and GeoGauge Figure 47 compares the standard deviation to the mean of the elastic modulus values determined from the DSPA and GeoGauge for different unbound materials. The standard deviation of the DSPA (Figure 47.a) and GeoGauge (Figure 47.b) slightly increases with increasing elastic modulus values. Figure 47.a does not show all of the DSPA data—it only shows the mean elastic modulus values less than 200 ksi for visual comparison to the other NDT devices.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 175 0 10 20 30 40 50 60 70 80 0 50 100 150 200 Mean Elastic Modulus, DSPA, ksi St an da rd D ev ia tio n, k si Fine-Grained Coarse-Grained Log. (Fine-Grained) Log. (Coarse-Grained) (a) Standard deviations from the DSPA for all projects; mean elastic modulus values greater than 200 ksi are not shown in the graph. 0 1 2 3 4 5 6 7 0 10 20 30 40 50 60 Mean Elastic Modulus, GeoGauge, ksi St an da rd D ev ia tio n, k si Fine-Grained Coarse-Grained Log. (Fine-Grained) Log. (Coarse-Grained) (b) Standard deviations from the GeoGauge for all projects. Figure 47. Standard Deviations of the Elastic Modulus Values Resulting from the DSPA and GeoGauge for Testing Unbound Layers

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 176 Contrary to the findings from the DCP and deflection-based methods, both the DSPA and GeoGauge found that the elastic modulus values of fine and coarse-grained materials were within the same range for many of the test sections. This difference between the different technologies will be discussed in greater detail in chapter 7 of Part III of the research report. Figure 48 compares the coefficient of variations determined in different areas of a project with each device. In general, the GeoGauge was found to have the lower variability in modulus values. The coefficient of variation in the TH-23 crushed aggregate base modulus values from the DSPA tests was found to be high. The reason for the high variation is unknown. However, the moisture gradient could be higher along this project because of rains that occurred prior to NDT. 0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80 Coefficient of Variation, GeoGauge, % C oe ffi ci en t o f V ar ia tio n, D SP A , % I-85 Soil TH-23 Soil TH-23 Base US-280 Base Line of Equality SH-130 Soil Figure 48. Comparison of the Coefficient of Variations of the Modulus Values Measured with the DSPA and GeoGauge The cells in Tables 37 and 39 that correspond to those conditions listed in Table 31 have been shaded. The following list summarizes the results of the DSPA and GeoGauge tests in accordance with those conditions listed in Table 31: • I-85 Low Plasticity Soil Embankment – Both devices found lane A of section 2 to be the weakest, prior to IC rolling. This is the area where thicker lifts had been placed. The results from both devices also indicate an increase in the embankment’s strength after IC rolling, but not a reduction in stiffness variability. • SH-21 High Plasticity Clay – Both devices found section 2 to be slightly weaker than section 1, which is consistent with construction records. Both devices showed a slight benefit when using the IC roller for testing and compaction. • TH-23 Gravelly, Silty Clay Embankment – Both devices correctly found lane C of the south section to be the softer (less stiff) of the areas tested, and found lane A of the north section to be stronger.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 177 • SH-130 Improved Granular Embankment – The GeoGauge did not detect any difference between the three areas tested, while the DSPA found section 3 to be consistently weaker, which was not planned. • TH-23 Crushed Aggregate Base – Both devices found that lane C of the middle section was weaker, and lane A of the south section was the stronger of the areas tested, which is consistent with construction records. • US-280 Crushed Stone Base – Both devices found area 4 to be weaker of the four areas tested. However, its strength is still high and consistent with adequately compacted crushed stone, similar to the findings with the DCP. 5.3.6 Ground Penetrating Radar Testing—Air Horn Antenna A GPR, single air-coupled antenna was used to take dielectric measurements of the unbound materials in accordance with ASTM and the procedure outlined by Maser and others (Maser., 2003; see also Figure 9 in chapter 3). Triplicate runs were made for each line of points within a section. Table 42 summarizes the average dielectric values measured at each test point for the other NDT devices for comparison purposes. One of the key advantages of the GPR is that a continuous profile of the dielectric values can be measured—in contrast to point-based devices. Contours of the dielectric measurements were prepared and used to determine the values at specific points where other tests were performed. Obviously, the increased sampling error between repeat runs will increase the overall variability of the GPR point measurements. Where the measurement lanes were well defined, the coefficient of variation (COV) of the dielectric values was significantly less than for the wider areas. As an example, the COV for the I-85 embankment area was as high as 50 percent, while the COV along the narrow US-280 test lane never exceeded 12 percent (refer to Table 42). Density contours and profiles were also prepared for each layer. Figures 49 and 50 present examples of contours that were prepared from the dielectric readings. Wet densities were calculated from these dielectric values, assuming a water content for the unbound materials in a specific area. Figure 51 shows an example of the density profile for the TH-23 crushed aggregate base material. The wet densities were found to be highly variable and generally did not coincide with the actual densities measured from the sand cones and nuclear density gauge readings. As an example, Table 41 lists the average total unit weights (pcf) that were estimated from the GPR data for the crushed aggregate base material placed along the TH-23 project in Minnesota. Table 41. Density Estimated by GPR on TH-23 Project, pcf. Lane A B C Comment North Section --- 129.2 142.4 Middle Section --- 130.8 150.6 Lane C had the less dense base. South Section --- 131.0 145.8 Lane A & B had the denser base.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 178 Conversely, all other NDT devices found lanes A and B to be stronger than lane C (refer to Tables 32, 34, 35, 37, and 39). In general, the GPR did not adequately identify those areas with anomalies. As noted above, a reason for this observation is that the water content for a particular area was assumed to be constant to identify changes in density, and vise-versa for water content. Another reason is that the anomaly may have been caused by variations in gradation and other physical properties that would be difficult, at best, to identify with the GPR. Table 42. Summary of the Dielectric Values Measured with the GPR on the Unbound Layers Project Identification Area A B C D Mean 15.38 15.79 14.29 15.19 I-85 Embankment, Silty Clay; Section 1, Before Rolling COV, % 17.8 23.3 53.6 25.7 Mean 13.91 17.47 16.82 16.38 I-85 Embankment, Silty Clay; Section 2, Before IC Rolling COV, % 29.0 20.5 30.7 24.1 Mean 20.37 21.23 21.61 23.23 I-85 Embankment, Silty Clay; Section 1, After IC Rolling COV, % 15.8 10.6 15.0 12.6 Mean 19.13 23.75 23.77 25.36 I-85 Embankment, Silty Clay; Section 2; After IC Rolling COV 10.2 10.7 17.6 8.4 Mean 23.004 13.468 19.334 --- TH-23 Embankment, Silt-Sand- Gravel Mix; South Section COV, % 11.3 7.0 14.4 --- Mean 20.324 34.438 23.882 --- TH-23 Embankment, Silt-Sand- Gravel Mix; North Section COV, % 22.2 32.7 22.7 --- Mean 9.225 10.00 7.65 --- SH-130 Improved Embankment; Section 1 COV 33.1 42.3 42.9 --- Mean 12.875 8.875 9.825 --- SH-130 Improved Embankment; Section 2 COV 90.3 47.4 20.1 Mean 8.775 9.025 11.85 SH-130 Improved Embankment; Section 3 COV, % 51.5 50.8 48.7 --- Mean --- 8.796 10.042 --- TH-23 Crushed Aggregate; North Section COV, % --- 1.6 5.4 --- Mean --- 8.950 10.87 --- TH-23 Crushed Aggregate; Middle Section COV, % --- 6.1 10.9 --- Mean --- 9.792 10.378 --- TH-23 Crushed Aggregate; South Section COV, % --- 8.2 4.3 --- Mean 11.723 US-280 Crushed Stone; Section 1 COV, % 8.3 Mean 12.222 US-280 Crushed Stone; Section 2 COV, % 11.4 Mean 11.919 US-280 Crushed Stone; Section 3 COV, % 7.3 Mean 11.569 US-280 Crushed Stone; Section 4 COV, % 7.0 Notes: • The shaded cells designate those areas with anomalies (refer to table 31); the black cells denote the weaker areas, while the gray cells denote the stronger areas tested within a specific project. • Due to construction sequencing, lane A of the TH-23 crushed aggregate base sections could not be tested with the GPR after it arrived on site.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 179 0 20 40 60 80 100 120 140 160 180 200 0 5 10 15 C en te r L in e O ffs et (f t) 8241+13.0 8240+13.7 200 220 240 260 280 300 320 340 360 380 400 Distance from Station 8241+13.0 (ft) 0 5 10 15 C en te r L in e O ffs et (f t) 8239+12.4 8238+11.3 8237+12.2 12 14 16 18 20 22 24 26 Subgrade Section 1 - Dielectric INFRASENSE, Inc. TH23 - Spicer, MN Arlington, MA 02476 Sheet: 1 of 2 Analyzed by: GLM Date: 10/21/04 Checked by: KRM Date: 10/22/04 Dielectric Scale (a) Section 1 of the Embankment. 0 20 40 60 80 100 120 140 160 180 200 0 5 10 15 20 C en te r L in e O ffs et (f t) 8229+98.3 8228+97.4 8227+98.9 200 220 240 260 280 300 320 340 360 380 400 Distance from Station 8241+13.0 (ft) 0 5 10 15 20 C en te r L in e O ffs et (f t) 8226+98 8226+00.3 14 18 22 26 30 34 38 42 46 50 Subgrade Section 2 - Dielectric INFRASENSE, Inc. TH23 - Spicer, MN Arlington, MA 02476 Sheet: 2 of 2 Analyzed by: GLM Date: 10/21/04 Checked by: KRM Date: 10/22/04 Dielectric Scale (b) Section 2 of the Embankment. Figure 49. Dielectric Contours Generated from the GPR Test Results for the Gravelly- Silty Clay Embankment Placed Along the TH-23 Reconstruction Project

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 180 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 -15 -10 -5 D is ta nc e fro m E dg e (ft ) 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290 300 -15 -10 -5 D is ta nc e fro m E dg e (ft ) 310 320 330 340 350 360 370 380 390 400 410 420 430 440 450 Distance from Station 1 (meters) -15 -10 -5 D is ta nc e fro m E dg e (ft ) 5 9 13 17 21 25 29 33 Precompaction - Subgrade Dielectric INFRASENSE, Inc. AL I-85 Ramp Subgrade Test Arlington, MA 02476 Sheet: 1 of 2 Analyzed by: GLM Date: 12/21/04 Checked by: KRM Date: 01/03/05 Dielectric (a) Pre-IC Compaction of the Low Plasticity Clay Embankment. 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 -15 -10 -5 D is ta nc e fro m E dg e (ft ) 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290 300 -15 -10 -5 D is ta nc e fro m E dg e (ft ) 310 320 330 340 350 360 370 380 390 400 410 420 430 440 450 Distance from Station 1 -15 -10 -5 D is ta nc e fro m E dg e (ft ) 5 9 13 17 21 25 29 33 Postcompaction - Subgrade Dielectric INFRASENSE, Inc. Arlington, MA 02476 Sheet: 2 of 2 Analyzed by: GLM Date: 12/21/04 Checked by: KRM Date: 01/03/05 Dielectric AL I-85 Ramp Subgrade Test (b) Post-IC Compaction of the Low Plasticity Clay Embankment. Figure 50. Dielectric Contours Generated from the GPR Test Results for the Low Plasticity Soil Embankment Placed on the I-85 Exit Ramp Reconstruction Project

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 181 200.0 300.0 400.0 500.0 600.0110 120 130 140 150 160 170 D en si ty (p cf ) 7956+00.4 7957+00 7958+00 7959+00 7959+99.4 1800.0 1900.0 2000.0 2100.0 Distance from S Catch Basin (ft.) 110 120 130 140 150 160 170 D ep th (i n. ) Layer 1 Density, Li ne C Layer 1 Density, Li ne B GPS Stations 7971+23.3 7972+23.4 7973+22.3 7974+21.9 7975+21.9 Base Sections - Base Density INFRASENSE, Inc. TH23 - Spicer, MN Arlington, MA 02476 Sheet: 1 of 1 Analyzed by: GLM Date: 10/21/04 Checked by: KRM Date: 10/22/04 800.0 900.0 1000.0 1100.0 1200.0110 120 130 140 150 160 170 D en si ty (p cf ) 7962+00.5 7963+02.1 7964+02.5 7965+04 7966+04.2 Moisture Content Assumed Constant at 4.12% Figure 51. Density Profiles Generated from the GPR Test Results for the Crushed Aggregate Base Layer Placed Along the TH-23 Reconstruction Project Layer thickness was also determined from the GRP test results, and it concurred with the thickness reported during construction. The thicknesses resulting from the GPR are provided in the appendices. Figure 52 presents an example of the thickness profiles (crushed aggregate base layer for the TH-23 reconstruction project) that were prepared from the dielectric readings. In general, thicknesses estimated with the GPR were within an acceptable error of the thicknesses reported from QC tests and other destructive sampling methods. 5.3.7 Non-Roller-Mounted Density Testing, Non-Nuclear—EDG The EDG was used to measure the density and water content of the unbound materials placed along each project. Figure 53 shows the EDG and its setup for measuring the density and water content in unbound materials. The test was performed in accordance with the manufacturer’s recommendations.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 182 800.0 900.0 1000.0 1100.0 1200.0 6 8 10 12 14 D ep th (i n. ) 7962+00.5 7963+02.1 7964+02.5 7965+04 7966+04.2 200.0 300.0 400.0 500.0 600.06 8 10 12 14 D ep th (i n. ) 7956+00.4 7957+00 7958+00 7959+00 7959+99.4 1800.0 1900.0 2000.0 2100.0 Distance from S Catch Basin (ft.) 6 8 10 12 14 D ep th (i n. ) Layer 1 Dielectric, Line A Layer 1 Dielectric, Line C GPS Stations 7971+23.3 7972+23.4 7973+22.3 7974+21.9 7975+21.9 Base Sections - Base Dielectric INFRASENSE, Inc. TH23 - Spicer, MN Arlington, MA 02476 Sheet: 1 of 1 Analyzed by: GLM Date: 10/21/04 Checked by: KRM Date: 10/22/04 Figure 52. Thickness Profiles Generated from GPR Test Results for the Crushed Aggregate Base Layer Placed Along the TH-23 Reconstruction Project Triplicate readings were made at each test point without moving the 6-inch probes. The EDG measurements made at each test point were adjusted based on calibration densities obtained from sand cones or nuclear density readings that were suppose to cover the range of values expected for the project. A soil model was developed for each unbound material, and that model was used to determine the actual densities and water contents from the EDG readings. The accuracy of the EDG, as for the GPR, is heavily dependent on the calibration values obtained from other test results. Any error in density or water content from these other tests is included in the EGD values. Table 43 summarizes the average dry density, while Table 44 provides the average water content measured within each section where the EDG was used. The amount of deviation in the test results was found to be small. The COV of the density readings were generally less than 1 percent, and it was less than 5 percent for water content readings. The water contents listed in Table 44 are generally below the optimum values obtained from construction records and measured in the laboratory. Based on observations at each site, it is expected that the water content of the upper layer materials are less than the optimum values for most of the areas tested, with the exception of the I-85 embankment material. Table 45 summarizes the maximum dry densities and optimum water contents recovered from construction records in comparison to the average values measured along the project. In some cases, multiple M-D relationships exist for a single layer within the same project.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 183 The values included in Table 45 represent the M-D curve or relationship for the material nearest the location of the specific test section. Figure 53. EDG Used to Measure the Density and Water Content of the Unbound Layers The cells in Tables 43 and 44 that correspond to those conditions listed in Table 44 have been shaded. The following list summarizes the results of the EDG tests in accordance with those conditions listed in Table 44: • I-85 Low Plasticity Soil Embankment –No difference in water content was detected by the EDG between all areas tested. The EDG found both outside lanes to be less dense prior to and after IC rolling, similar to the DCP test results. The variation in dry density and water content was found to be low. • TH-23 Gravelly, Silty Clay Embankment – Higher water contents and lower dry densities were measured in the south section, but not along lane C. Lane C had the greater variability in water content. The variability of the dry density was found to be low. • SH-130 Improved Granular Embankment – The EDG found no significant difference in density and water content between all areas tested, which was planned. • TH-23 Crushed Aggregate Base – The EDG found no significant difference in density and water content between all areas tested, which is inconsistent with construction records.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 184 • US-280 Crushed Stone Base – The EDG found no significant difference in density and water content between all areas tested, also inconsistent with construction records. Table 43. Summary of the Dry Densities Measured with the EDG, pcf Project Identification Area A B C D Mean, pcf 107.92 108.9 108.6 107.7 I-85 Embankment, Silty Clay; Section 1, Before IC Rolling COV, % 1.3 0.5 1.1 1.7 Mean, pcf 107.2 107.5 108.9 107.2 I-85 Embankment, Silty Clay; Section 2; Before IC Rolling COV, % 0.8 0.8 1.1 1.9 Mean, pcf 108.1 108.2 108.5 108.4 I-85 Embankment, Silty Clay; Section 1, After IC Rolling COV, % 1.0 0.5 0.7 0.3 Mean, pcf 107.4 107.7 108.0 107.6 I-85 Embankment, Silty Clay; Section 2, After IC Rolling COV, % 0.5 0.5 0.8 1.3 Mean, pcf 123.9 123.7 124.4 --- TH-23 Embankment, Silt-Sand- Gravel Mix; North Section COV, % 0.4 0.1 1.0 --- Mean, pcf 122.5 122.9 122.9 --- TH-23 Embankment, Silt-Sand- Gravel Mix; South Section COV, % 1.8 1.8 0.8 --- Mean, pcf 123.7 123.7 124.9 --- SH-130 Improved Embankment; Section 1 COV, % 0.3 0.1 0.6 --- Mean, pcf 122.6 123.1 122.7 --- SH-130 Improved Embankment; Section 2 COV, % 2.0 2.0 0.8 --- Mean, pcf 123.3 122.3 123.7 SH-130 Improved Embankment; Section 3 COV, % 1.4 0.1 0.2 Mean, pcf 129.9 129.8 129.8 --- TH-23 Crushed Aggregate; North Section COV, % 0 0 0 --- Mean, pcf 129.8 129.8 129.8 --- TH-23 Crushed Aggregate; Middle Section COV, % 0 0 0 --- Mean, pcf 129.8 129.9 129.8 --- TH-23 Crushed Aggregate; South Section COV, % 0.1 0.1 0 --- Mean, pcf 147.4 US-280 Crushed Stone; Section 1 COV, % 0.7 Mean, pcf 148.8 US-280 Crushed Stone; Section 2 COV, % 0.3 Mean, pcf 145.9 US-280 Crushed Stone; Section 3 COV, % 0.5 Mean, pcf 148.2 US-280 Crushed Stone; Section 4 COV, % 0.3 Note: The shaded cells designate those areas with anomalies (refer to table 31); the black cells denote the weaker areas, while the gray cells denote the stronger areas tested within a specific project.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 185 Table 44. Summary of the Water Contents Measured with the EDG, percent Project Identification Area A B C D Mean, % 16.9 16.8 16.9 16.9 I-85 Embankment, Silty Clay; Section 1, Before IC Rolling COV, % 0.8 0.3 0.3 1.0 Mean, % 16.9 16.9 16.8 17.0 I-85 Embankment, Silty Clay; Section 2; Before IC Rolling COV, % 0.7 0.3 0.3 1.5 Mean, % 16.9 16.9 16.9 16.9 I-85 Embankment, Silty Clay; Section 1, After IC Rolling COV, % 0.5 0.3 0.4 0 Mean, % 17.0 16.9 16.9 16.9 I-85 Embankment, Silty Clay; Section 2, After IC Rolling COV, % 0.5 0.3 0 0.7 Mean, % 8.0 8.0 7.6 TH-23 Embankment, Silt-Sand- Gravel Mix; North Section COV, % 5.1 1.1 11.9 Mean, % 9.8 8.7 7.6 TH-23 Embankment, Silt-Sand- Gravel Mix; South Section COV, % 7.5 7.3 15.8 Mean, % 8.1 8.05 7.23 SH-130 Improved Embankment; Section 1 COV, % 4.4 1.2 6.8 Mean, % 8.85 8.43 8.7 SH-130 Improved Embankment; Section 2 COV, % 19.8 21.6 8.4 Mean, % 8.35 9.1 8.05 SH-130 Improved Embankment; Section 3 COV, % 14.4 1.6 0.9 Mean, % 4.26 4.28 4.34 TH-23 Crushed Aggregate; North Section COV, % 1.3 1.0 2.1 Mean, % 4.24 4.28 4.30 TH-23 Crushed Aggregate; Middle Section COV, % 1.3 2.0 1.6 Mean, % 4.18 4.18 4.38 TH-23 Crushed Aggregate; South Section COV, % 3.9 3.9 1.0 Mean, % 3.92 US-280 Crushed Stone; Section 1 COV, % 3.1 Mean, % 4.18 US-280 Crushed Stone; Section 2 COV, % 2.9 Mean, % 3.77 US-280 Crushed Stone; Section 3 COV, % 2.9 Mean, % 4.06 US-280 Crushed Stone; Section 4 COV, % 2.6 Note: The shaded cells designate those areas with anomalies (refer to table 31); the black cells denote weaker areas, while the gray cells denote the stronger areas tested within a specific project.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 186 Table 45. Listing of the Maximum Dry Density and Optimum Water Content for the Unbound Materials and Soils, as Compared to the Average Test Results from the EDG Project Material Maximum Dry Unit Weight, pcf Optimum Water Content, % Average Dry Density, pcf Average Water Content, % NCAT, Oklahoma High Plasticity Clay 99.9 21.8 96.7 21.3 SH-21, TX High Plasticity Clay 108.0 21.9 107.3 18.4 Low Plasticity Soil; Pre-IC 107.98 16.9 I-85, AL Low Plasticity Soil; Post-IC 112.7 13.1 107.98 16.9 SH-130, TX Improved Granular Embankment 122.0 9 123.3 8.32 Silt-Sand-Gravel Mix – South Area 122.77 8.69 TH-23, MN Silt-Sand-Gravel Mix – North Area 122.6 12 123.80 7.87 US-2, ND Soil-Aggregate, Embankment 128.0 9.0 123.1 12.1 NCAT, FL Limerock Base 116.1 12.5 110.5 13.4 CR-103 Caliche Base 127.5 10.0 125.0 9.5 NCAT, MO Crushed Limestone 130.0 10.0 124.4 9.0 TH-23, MN Crushed Aggregate Base 135.3 7.8 129.82 4.3 US-53, OH Crushed Aggregate Base 134.1 8.5 136.0 9.1 NCAT, SC Crushed Granite Base 138.1 5.0 130.0 4.7 US-2, ND Crushed Gravel Base 141.1 6.0 134.4 5.9 US-280, AL Crushed Stone Base 148.5 6.2 147.58 3.9 NOTE: The maximum dry density and optimum water content for most of the materials and layers were determined using AASHTO T 180. The exception is the high plasticity clay from the Texas project and the North Dakota embankment material. 5.3.8 Roller-Mounted Density and Stiffness Devices TH-23 Base Material The Caterpillar IC roller was used to test the Class 6 crushed aggregate base materials on the TH-23 project in Spicer, Minnesota. The IC roller (shown in Figure 19 in Chapter 3) was set in low amplitude so that the roller would not decompact or damage the existing base material. Figures 54 and 55 show example print outs that were obtained from the IC roller’s instrumentation. The stiffness responses recorded by the IC roller were about the same between both areas tested. Based on the interpretation of the readings by the operator, the IC roller suggests that

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 187 the crushed aggregate base material is as dense as it can be along these lanes. Further compaction could damage or decompact the aggregate base layer. Table 46 tabulates the results from the stiffness measurements made with the IC roller, which are explained and discussed in the bullets that follow. Table 46. Stiffness Responses Recorded by the IC Roller, Tabulation of Results Area Lanes Tested A B C Mean 35.00 24.22 41.80 Standard Deviation 9.25 9.38 6.78 1 – South Section COV, % 26.4 38.7 16.2 Mean 38.40 31.78 31.30 Standard Deviation 13.01 8.70 5.79 2 – Middle Section COV, % 33.9 27.4 18.5 • South Section, Area 1 (Figure 54) – Lanes A and C were found to be the stiffer based on the measured responses by the IC roller. The lowest stiffness readings were recorded in the northern part of lane B. Conversely, the other NDT devices found lane C to be weaker (refer to Tables 32, 35, 37, and 39). • Middle Section, Area 2 (Figure 55) – The IC roller found no consistent difference between the three lanes. The weaker area identified by the DSPA, GeoGauge, DCP, and LWD was found to be along lane C (refer to Tables 32, 35, 37, and 39). Lane C has the lower densities and higher moisture contents. The IC roller may have bridged the less dense area along lane C making it difficult to detect the lower strengths. I-85 Exit Ramp 51 – Embankment Material Nondestructive testing was performed on the embankment material prior to final compaction. The Ammann IC roller was used to complete the compaction of the two embankment sections along the I-85 reconstruction of the Exit ramp 51 (refer to Figure 19 in chapter 3). After IC rolling, selected NDT devices were used to re-test each area. The results from this testing were provided in the respective tables for each NDT device, previously discussed in this chapter. Figure 56 compares the modulus values before and after IC rolling, as measured by the GeoGauge, DSPA, and DCP. As shown, the modulus values consistently increased after IC rolling, with the exception of the DCP device. In general, the test results from those NDT devices suggest increases in density of the embankment. The GRP test results also show a benefit (increased density) of the additional compaction (refer to Table 42). Conversely, the EDG did not show any increase in the embankment density (refer to Table 43). Figure 57 compares the coefficient of variation of those average modulus values before and after IC rolling. The variability in the modulus values did not decrease. In other words, the uniformity of the stiffness of the embankment did not significantly increase. The GPR test results, however, did show a significant reduction in variability of the dielectric values.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 188 Figure 54. Printouts from the IC Roller Used to Test the Crushed Aggregate Base in Area 1 of the TH-23 Reconstruction Project in Spicer, Minnesota Questionable Response Data Questionable Response Data

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 189 Figure 55. Printouts from the IC Roller Used to Test the Crushed Aggregate Base in Area 2 of the TH-23 Reconstruction Project in Spicer, Minnesota

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 190 0 10 20 30 40 50 60 0 10 20 30 40 50 60 Layer Modulus Before IC Rolling, ksi La ye r M od ul us A fte r I C R ol lin g, k si GeoGauge DSPA DCP Line of Equality Figure 56. Comparison of Modulus Values Measured with Different NDT Devices Before and After IC Rolling of the I-85 Low Plasticity, Fine-Grained Soil Embankment 0 10 20 30 40 50 0 10 20 30 40 50 Coefficient of Variation, Before IC Rolling, % C oe ffi ci en t o f V ar ia tio n, A fte r I C R ol lin g, % GeoGauge DSPA DCP Line of Equality GPR Figure 57. Uniformity of the Embankment Along I-85 Exit 51 Before and After IC Rolling, as Determined Through Modulus Measurements from Different NDT Devices

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 191 5.4 NDT Test Results of HMA Mixtures This section presents the NDT responses measured on HMA mixtures at each project listed above and discussed in Appendix B and identified in Tables 29 and 30. It also provides a brief evaluation of the mixtures based on those measured responses and compares the responses measured by different NDT devices on the same material. As noted in chapter 1, the initial testing under Part A was to confirm that the NDT technologies can identify differences in construction quality of HMA mixtures. Table 47 summarizes the anomalies between the HMA materials placed along each project. During nondestructive testing, none of the NDT operators were advised of those anomalies or differences. Table 47. Description of the Different Physical Conditions (Localized Anomalies) of the HMA Mixtures Placed Along Projects within Part A Project Identification HMA Sections Description of Differences Along the Project TH-23 HMA Base; Spicer, Minnesota Section 2, Middle or Northeast Section QA records indicate lower asphalt content in this area – asphalt content was still within the specifications. Section 2, Middle; All lanes QA records indicate higher asphalt content in this area, but it was still within the specifications. I-85 SMA Overlay; Auburn, Alabama Lane C, All Sections This part or lane was the last area rolled using the rolling pattern set by the contractor, and was adjacent to the traffic lane. Densities lower within this area. Initial Test Sections, defined as A; Section 2, All Lanes Segregation identified in localized areas. In addition, QA records indicate lower asphalt content in this area of the project. Densities lower within this area. Supplemental Test Sections near crushed stone base sections, defined as B. Segregation observed in limited areas. US-280 HMA Base Mixture; Opelika, Alabama IC Roller Compaction Effort Section, Defined as C. Higher compaction effort was used along Lane C. SH-130 HMA Base Mixture; Georgetown, Texas All Sections No differences between the different sections tested.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 192 5.4.1 Seismic Testing—PSPA The PSPA was used to measure the seismic modulus of the HMA materials in accordance with the procedure and software developed by Dr. Nazarian for the Texas DOT (see Figure 16 in chapter 3). Triplicate tests were performed at each test point—similar to what was done for the unbound materials. Table 48 provides the average seismic modulus values measured within each area of the Part A projects, while Table 49 provides the average seismic values measured along the Part B projects. The cells in Table 48 that are shaded correspond to those conditions listed in Table 47. The sensor bar for the PSPA was rotated relative to roller direction for the repeat readings, similar to the test procedure for unbound materials. Figure 58 compares the differences between the measurements taken parallel and perpendicular to roller direction. No significant difference was found between the measurement directions. Figure 59 presents the cumulative frequency of the standard deviation or repeatability of the PSPA, while Figure 60 compares the standard deviations to the mean seismic modulus. The standard deviations in Figures 59 and 60 represent the triplicate measurements taken at the same test point. The repeatability of the PSPA is considered good. The mean standard deviation varies from about 10 to 50 ksi and appears to be independent of the mean seismic modulus. The standard deviation, however, generally decreases with increasing mean seismic modulus for the I-35/SH-130 HMA base, while it increases with increasing mean seismic modulus for the US-280 HMA base. The reason for this disparity between the different projects is unknown. Figure 58.a does show a consistent difference between the seismic modulus measured parallel and perpendicular to the rolling direction for the US-280 project. Figure 58.b also shows more diversity between these two measurements made at a point for the interior of the HMA lane or mat. The following bullets summarize the results from the PSPA seismic tests in accordance with those conditions listed in Table 47. • TH-23 HMA Base – The PSPA found that section 2 (northeast section) had the lower seismic modulus values, which could be consistent with a lower asphalt content. The coefficient of variation for all areas tested suggests low variability or uniform construction of the HMA mixture within each area. • I-85 SMA Overlay – The PSPA found higher seismic modulus in the area with the higher asphalt content; lanes A and B in the middle section. The lowest seismic value measured with the PSPA was in lane C of section 1. The seismic values measured in lane C of the two other areas were similar to the other lanes tested. This observation suggests that the delay in compaction along this lane may have been discontinued with the placement-compaction operations. After section 1, the rollers were able to keep up with the paver.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 193 Table 48. Summary of the Seismic Modulus Measured with the PSPA within the Projects Included in Part A, ksi Location or Designated Area Project ID A B C Mean, ksi 475.6 505.4 481.8 TH-23 HMA Base, Section 1 – South COV, % 3.9 2.9 8.9 Mean, ksi 454.7 447.2 461.4 TH-23 HMA Base, Section 2 – Northeast COV, % 9.6 8.8 6.6 Mean, ksi 481.0 501.4 493.6 TH-23 HMA Base, Section 3 – Northwest COV, % 4.0 6.6 10.1 Mean, ksi 472.6 504.4 521.4 TH-23 HMA Base, Section 4, North – One Day After Paving COV, % 3.8 2.3 5.6 Mean, ksi 272.4 240.6 215.4 I-85 SMA Overlay, Section 1 – North COV, % 32.2 14.1 9.7 Mean, ksi 290.8 279.2 278.8 I-85 SMA Overlay, Section 2 – Middle COV, % 9.1 7.9 16.6 Mean, ksi 269.4 265.4 298.2 I-85 SMA Overlay, Section 3 – South COV, % 7.6 6.6 14.9 Mean, ksi 490.0 535.4 474.2 US-280 HMA Base, Section 1, Initial Testing COV, % 6.9 11.6 10.8 Mean, ksi 465.8 432.2 373.8 US-280 HMA Base, Section 2, Initial Testing COV, % 1.6 6.5 9.7 Mean, ksi 305.8 US-280 HMA Base, Joint Measurements, Initial Testing COV, % 29.4 Mean, ksi 372.0 NA 287.8 US-280 HMA Base, Segregated Areas, Initial Testing COV, % 3.7 NA 26.0 Mean, ksi 550.4 554.4 574.6 US-280 HMA Base, Section 1, Supplemental Tests COV, % 15.2 12.8 6.1 Mean, ksi 537.2 559.8 553.5 US-280 HMA Base, Section 2, Supplemental Tests COV, % 6.8 11.7 6.6 Mean, ksi 596.0 US-280 HMA Base, Joint Measurements, Supplemental Tests, Section 1 COV, % 9.7 Mean, ksi 391.3 US-280 HMA Base, Segregated Areas, Supplemental Tests, Section 2 COV, % 12.9 Mean, ksi --- 238.6 268.8 US-280 HMA Base, Supplemental Testing, IC Roller, One Day After Paving COV, % --- 15.0 17.5 Mean, ksi 354.3 427.0 373.4 I-35/SH-130 HMA Base, Section 1 COV, % 12.1 10.0 13.4 Mean, ksi 260.4 317.4 300.0 I-35/SH-130 HMA Base, Section 2 COV, % 3.6 11.6 17.1 Mean, ksi 441.8 475.9 467.3 I-35/SH-130 HMA Base, Section 3, One Day After Paving COV, % 12.4 6.5 4.5 Mean, ksi --- --- 297.5 I-35/SH-130 HMA Base, Joints, Section 1 COV, % --- --- 15.3 Note: The shaded cells designate those areas with anomalies (refer to table 47).

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 194 Table 49. Summary of the Seismic Modulus Measured with the PSPA within Projects Included in Part B, ksi Project Identification Section and Material Mean Modulus, ksi Coefficient of Variation, % Standard Deviation of Means, ksi Coarse-Graded Base 605.3 14.2 85.58 US-47, MO Fine-Graded Wearing Surface 457.6 18.5 84.7 I-75, MI Dense-Graded Binder Mix; Type 3-C 676.3 8.9 69.35 US-2, ND Coarse-Graded Base Mix; PG58-28 344.3 10.6 36.58 US-53, OH Coarse-Graded Binder Mix 666.7 6.0 40.26 I-20, TX Coarse-Graded Binder Mix, CMHB 435.5 10.2 44.54 Surface Mix; 45% RAP, PG67; E-5 510.7 6.5 33.38 Surface Mixture; 45% RAP, PG76, SBS ; E-6 473.4 12.2 57.79 NCAT, Alabama Surface Mixture; 45% RAP, PG76, Sasobit ; E-7 444.3 9.1 40.62 Coarse-Graded Base Mix; PG67; N-1 447.1 10.6 47.57 NCAT, Florida Coarse-Graded Base Mix; PG76; PMA; N-2 475.8 11.5 54.87 NCAT, Missouri Dense-Graded Surface Mix; PG76; N-10 528.7 15.1 136.18 NCAT, SC Coarse-Graded Base Mix; PG67; S-11 495.2 6.6 32.68 • US-280 HMA Base – The PSPA found that section 2 of the project used for the initial testing has lower seismic modulus values than section 1, as expected. Similarly, the PSPA measured lower modulus in the area with lower compaction, as compared to the area compacted with the IC roller for the supplemental sections. The seismic modulus values from the supplemental sections are higher than for the initial sections along this project. The reason for higher values in the supplemental sections is unknown, other than some change in the mixture occurred between the two testing periods. This issue will be discussed in more detail under section 5.5. • US-280 HMA Base – The seismic modulus values measured at the locations with some minor segregation were found to be lower than for the areas without segregation for both the initial and supplemental sections. Similarly, the seismic modulus values measured along the longitudinal joints were consistently less than those modulus values measured within the interior of the sections. • I-35/SH-130 HMA Base – The PSPA found section 2 to be weaker or less stiff than the other two sections. This difference was not planned, and the specific reason for the less stiff mixture is unknown.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 195 0 100 200 300 400 500 600 700 800 0 100 200 300 400 500 600 700 800 Seismic Modulus, Parallel to Roller, ksi S ei sm ic M od ul us , Pe rp en di cu la r to R ol le r, k si SH-130 I-85, SMA US-280 TH-23 Line of Equality (b) Seismic modulus measured within interior of the mat. 0 100 200 300 400 500 600 700 800 0 100 200 300 400 500 600 700 800 Seismic Modulus, Parallel to Roller, ksi Se is m ic M od ul us , Pe rp en di cu la r to R ol le r, ks i SH-130 US-280 Line of Equality (a) Seismic modulus measured along a longitudinal joint. Figure 58. Comparison of the PSPA Seismic Modulus Values Measured Parallel and Perpendicular to Roller Direction

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 196 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 110 120 Modulus Standard Deviation, DSPA, ksi Cu m ul at iv e Fr eq ue nc y, pe rc en t TH-23, HMA US-280, HMA I-85, SMA SH-130, HMA Figure 59. Cumulative Frequency of the Standard Deviation from the PSPA 0 20 40 60 80 100 0 100 200 300 400 500 600 700 Average Seismic Modulus, PSPA, ksi S ta nd ar d D ev ia tio n, k si SMA US-280 SH130 TH23 Log. (US-280) Log. (SH130) Figure 60. Relationship Between the Standard Deviation and Mean Seismic Modulus Values from the PSPA

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 197 NDT devices must be able to identify the quality of joint construction and segregation. Table 50 provides a summary of the seismic modulus measured in areas with different features— along longitudinal joints and within segregated areas. In most areas, the PSPA measured lower seismic modulus values along the longitudinal joints than within the interior of the HMA lane, as expected. The ratios between the joint seismic modulus and interior value are listed in Table 50 and range from 0.54 to 1.05. Table 50. Summary of the Seismic Modulus Measured with the PSPA Along Longitudinal Joints or in Areas with Limited Segregation Project Location/Area Lane Interior Area with Feature Ratio LONGITUDINAL JOINTS Mean, ksi 499.0 305.8 US-280 HMA Base; Initial Sections Multiple Days After Paving COV, % 10.8 29.4 0.613 Mean, ksi 565.2 596.0 US-280 HMA Base; Supplemental Sections Section 1, During Paving COV, % 6.2 9.7 1.054 Mean, ksi 373.4 297.5 Multiple Days After Paving COV, % 13.4 25.3 0.797 Mean, ksi 208.0 111.2 I-35/SH-130 HMA Base During Paving COV, % 32.9 6.4 0.535 SEGREGATED AREAS Mean, ksi 490.0 372.0 Section 1, Lanes A,B COV, % 6.9 3.7 0.759 Mean, ksi 474.2 287.8 US-280 HMA Base; Initial Sections Section 1, Lane C COV, % 10.8 26.0 0.607 Mean, ksi 557.2 391.2 US-280 HMA Base; Supplemental Sections Section 2 COV, % 11.5 12.9 0.702 NOTE: The ratio listed above is the mean seismic modulus measured at the feature divided by the seismic modulus measured within the interior of the lane or area. Similarly, lower seismic modulus values were measured in those areas with limited segregation. The ratio of the seismic modulus measured within the segregated area to non- segregated areas ranged from 0.61 to 0.76. The segregation found in these areas is truck to truck segregation and was not considered severe. The PSPA, however, did detect these areas. Seismic modulus values were also measured during different times after paving. Table 51 lists the seismic modulus for three different times, when available. These time periods included: (1) during paving or immediately after compaction, (2) the day following placement, and (3) multiple days after placement. As shown, time and temperature of the mixture during testing have a significant effect on the seismic modulus, as expected. In most cases, one-day after placement resulted in similar mean seismic modulus values to those measured multiple days after placement, adjusting for temperature differences. This was not the case for the US-280 HMA base mixture. More importantly, those areas tested

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 198 with the PSPA resulting in high variability (COV greater than 15 percent) were tested immediately after compaction or during the same day of placement. The different mat temperatures in combination with other changes in the volumetric properties between test points exaggerate the mixture’s variability, as measured by the PSPA. Table 51. Summary of the Seismic Modulus Measured with the PSPA for Different Times After Paving, Excluding Sections or Areas with Anomalies Project Section Time of Readings Location Mean, ksi COV, % During Paving --- --- A 472.6 3.8 B 504.4 2.3 Section 4 One-Day After Paving C 521.4 5.6 A 478.6 3.8 B 503.9 4.8 TH-23 HMA Base Sections 1,3 Multiple Days After Paving C 487.7 9.0 A 270.9 22.1 B 253.0 11.3 During Paving C 256.8 21.3 I-85 SMA Overlay Sections 1,3 Multiple Days After Paving --- --- A 165.0 12.2 B 185.0 6.8 Section 3 During Paving C 188.0 7.5 A 490.0 6.9 B 535.4 11.6 US-280 HMA Base; Initial Sections Section 1 Multiple Days After Paving C 474.2 10.8 Section 3 During Paving B 171.0 8.3 B 238.6 15.0 IC Roller Section One Day After Paving C 268.8 17.5 A 543.8 11.2 B 557.2 11.5 US-280 HMA Base; Supplemental Sections Sections 1,2 Multiple Days After Paving C 565.2 6.2 A 238.5 28.5 B 179.0 54.0 Section 3 During Paving C 208.0 32.9 A 441.8 12.4 B 475.9 6.5 Section 3 One Day After Paving C 467.3 4.5 A 354.3 12.1 B 427.0 10.0 I-35/SH-130 HMA Base Section 1 Multiple Days After Paving C 373.4 13.4

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 199 5.4.2 Deflection Testing—FWD Deflection basins were measured with the FWD in accordance with the test protocol being used in the LTPP program. The procedure was to use two seating drops at a low drop height, followed by two drops at each drop height. Three drop heights were used at each test point. The deflection basins were recorded for each drop, including the seating drops. After the first set of drops, the FWD was moved forward and the test sequence repeated. This sequence of replicate testing was used at each project. The deflection basins were used to forward-calculate the elastic modulus of the layer being evaluated using the procedure developed by Stubstad and used for evaluating the unbound materials. The calculated elastic modulus values are summarized in Table 52 for those projects where the FWD was used. Elastic modulus values were also backcalculated using other traditional methods. The forward-calculation method resulted in the least variation of elastic modulus values within a specific area, as found for the unbound materials. Table 52. Summary of the Elastic Modulus Calculated from the Deflection Basins Measured with the FWD Location or Designated Area Project Identification A B C Mean, ksi 363.9 437.2 --- I-85 SMA Overlay, Section 1 – North COV, % 14.1 27.9 --- Mean, ksi 562.8 575.0 --- I-85 SMA Overlay, Section 2 – Middle COV, % 14.4 9.6 --- Mean, ksi 343.3 477.0 --- I-85 SMA Overlay, Section 3 – South COV, % 24.7 1.9 --- Mean, ksi 195.8 231.2 183.0 US-280 HMA Base, Section 1, Initial Testing COV, % 10.6 18.6 27.4 Mean, ksi 138 143 96.6 US-280 HMA Base, Section 2, Initial Testing COV, % 6.0 6.3 23.0 Mean, ksi 82.8 87.0 73.8 US-280 HMA Base, Section 3, Just After Paving, Initial Testing COV, % 13.9 15.5 8.3 Mean, ksi 125.2 US-280 HMA Base, Joint Measurements, Initial Testing COV, % 23.9 Mean, ksi 149 --- 140 US-280 HMA Base, Segregated Areas, Initial Testing COV, % 16.6 --- 49.1 Mean, ksi 630 568 509 US-280 HMA Base, Section 1, Supplemental Tests COV, % 12.5 10.8 15.8 Mean, ksi 1,324 1,172 1,060 US-280 HMA Base, Section 2, Supplemental Tests COV, % 17.6 22.1 28.8 Mean, ksi 379 US-280 HMA Base, Joint Measurements, Supplemental Tests COV, % 53.9 Mean, ksi 707 US-280 HMA Base, Segregated Areas, Supplemental Tests COV, % 28.2 Note: The shaded cells designate those areas with anomalies (refer to table 47).

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 200 Figure 61 presents a cumulative frequency diagram of the standard deviation or repeatability of the FWD, while Figure 62 compares the standard deviations to the mean elastic modulus calculated from the deflection basins. The standard deviations in Figures 61 and 62 represent the triplicate measurements taken at the same test point. The repeatability of the FWD and forward-calculation procedure is considered poor for new construction. The reason for this increased variability is the effect or influence from the supporting layers. More importantly, the standard deviation appears to be more heavily dependent on the elastic modulus— increasing standard deviation with increasing elastic modulus values (refer to Figure 62). The cells in Table 52 that correspond to those conditions listed in Table 47 have been shaded. The following list summarizes the results from the FWD and forward-calculation procedure in accordance with those conditions listed in Table 47: • I-85 SMA Overlay – Deflections were not measured along lane C, adjacent to existing traffic, for safety reasons. The FWD found higher elastic modulus in the area with the higher asphalt content, which is consistent with the seismic test results. However, the COV values from the FWD and forward-calculation procedure suggests more construction variability than estimated with the PSPA. The traditional QA test results found the SMA to be within specification. The reason for the higher variability is related to the thickness variations (a constant value was used in the forward- calculation processing of the data) and the variation of the supporting HMA pavement (a constant pavement thickness was also assumed for the underlying structure for overlay placement). • US-280 HMA Base – The FWD found that section 2 of the project used for the initial testing has lower elastic modulus values, excluding the tests completed immediately after compaction. In addition, the FWD found the HMA base mixture in the supplemental sections to be stiffer than in the initial sections. These observations are consistent with the seismic test results. However, the elastic modulus values calculated from the FWD deflection basins are low and representative of an inferior mixture. These elastic modulus values are believed to be influenced by the supporting layers and are not representative of the in-place material. Similar to the findings from testing the I-85 SMA overlay, the FWD estimate of construction variability is greater than estimated with the PSPA. • US-280 HMA Base – The elastic modulus values calculated at the locations with some minor segregation were higher than some of the other areas tested without segregation. This finding is contrary to the PSPA results and previous experience. A potential reason for this discrepancy between the two devices is that the FWD measures response from a much larger area than for the PSPA. Placing the FWD loading plate over a small area with segregation would bridge the coarse aggregate particles having little to no effect on the measured deflection basin. The PSPA measures responses that are more localized. • US-280 HMA Base – The elastic modulus calculated from the FWD deflection basins measured along the longitudinal joints is lower than within the interior of the lanes. This observation is consistent with the findings from the PSPA. The difference between the two results, however, is the magnitude of the differences. The PSPA

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 201 found the joint to interior seismic modulus ratios to be in the range of 0.54 to 1.054, while the FWD found ratios to average 0.62 for the supplemental sections and 0.43 for the initial sections. It is believed that the FWD is measuring a more structural response along the longitudinal joints, while the PSPA is more of a mixture response in localized areas. 0 10 20 30 40 50 60 70 80 90 100 0 50 100 150 200 250 300 350 400 Modulus Standard Deviation, FWD, ksi Cu m ul at iv e Fr eq ue nc y, % SMA Overlay HMA Base, Reconstruction Figure 61. Cumulative Frequency of the Standard Deviation from the Deflection-Based Test Methods, FWD 5.4.3 Ground Penetrating Radar Testing—Air Horn Antenna A GPR, single air-coupled antenna was used to take dielectric measurements of the HMA materials in accordance with ASTM and the procedure outlined by Maser (2003). Replicate or triplicate runs were made for each line of points within a section of a project. The data were analyzed by software written by Dr. Maser for use in QA applications. Table 53 summarizes the average air voids calculated from the dielectric values measured at each test point for the other NDT devices and technologies for comparative purposes. The cells in Table 53 that correspond to those conditions listed in Table 47 have been shaded. As noted in chapter 3, one of the advantages of the GPR is that a continuous profile can be measured for different properties. Contours of the dielectric measurements were prepared and used to determine the values at specific points where other point-based nondestructive tests were performed. Figure 63 presents the contours of the dielectric values measured within section 1 of the TH-23 HMA base project. In addition, contour plots of the HMA air voids and base thickness were also prepared by the software. Examples are provided in

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 202 Figures 64 and 65. These contour plots can be beneficial for evaluating in more detail areas that appear to be deficient, strategically locating areas with different dielectric properties (a key benefit of GPR). 0 50 100 150 200 250 300 350 0 200 400 600 800 1000 1200 1400 Average Elastic Modulus, FWD, ksi St an da rd D ev ia tio n, k si SMA HMA Poly. (HMA) Figure 62. Relationship Between the Standard Deviation and Mean Elastic Modulus Calculated from Deflection Basins

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 203 Table 53. Summary of the Air Voids Calculated from the Dielectric Values Measured with the GPR Location or Designated Area Project ID A B C Mean, % 5.80 6.58 6.62 TH-23 HMA Base, Section 1 – South COV, % 5.1 1.3 3.9 Mean, % 8.34 7.24 5.52 TH-23 HMA Base, Section 2 – Northeast COV, % 3.8 3.6 3.5 Mean, % 7.02 6.90 6.90 TH-23 HMA Base, Section 3 – Northwest COV, % 10.7 9.3 7.0 Mean, % 6.72 10.00 8.62 I-85 SMA Overlay, Section 1 – North COV, % 15.6 6.6 4.1 Mean, % 5.71 6.64 8.76 I-85 SMA Overlay, Section 2 – Middle COV, % 16.9 5.7 24.3 Mean, % 13.11 10.74 8.25 I-85 SMA Overlay, Section 3 – South COV, % 12.5 13.0 15.3 Mean, % 7.02 6.80 7.27 US-280 HMA Base, Section 1, Initial Testing COV, % 7.1 8.2 13.8 Mean, % 7.12 6.59 6.72 US-280 HMA Base, Section 2, Initial Testing COV, % 6.3 10.3 11.8 Mean, % 7.70 US-280 HMA Base, Joint Measurements, Initial Testing COV, % 11.1 Mean, % 7.51 NA 7.04 US-280 HMA Base, Segregated Areas, Initial Testing COV, % 8.8 NA 9.3 Mean, % 5.76 5.48 5.42 US-280 HMA Base, Section 1, Supplemental Tests COV, % 2.7 2.4 2.0 Mean, % 5.36 5.44 5.54 US-280 HMA Base, Section 2, Supplemental Tests COV, % 3.7 3.3 3.3 Mean, % 5.78 US-280 HMA Base, Joint Measurements, Supplemental Tests COV, % 2.2 Mean, % 5.64 US-280 HMA Base, Segregated Areas, Supplemental Tests COV, % 2.2 Mean, % 5.60 6.19 6.05 I-35/SH-130 HMA Base, Section 1 COV, % 1.1 3.4 1.1 Mean, % 5.13 5.70 6.01 I-35/SH-130 HMA Base, Section 2 COV, % --- 1.3 2.7 Mean, % --- --- 5.19 I-35/SH-130 HMA Base, Joints, Section 1 COV, % --- --- 1.6 Mean, % --- --- 4.97 COV, % --- --- --- I-35/SH-130 HMA Base, Joints, Section 2 Note: The shaded cells designate those areas with anomalies (refer to table 47).

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 204 AC Section 1 - Base Dielectric INFRASENSE, Inc. TH23 - Spicer, MN Arlington, MA 02476 Sheet: 1 of 3 Analyzed by: GLM Date: 10/21/04 Checked by: KRM Date: 10/22/04 600 620 640 660 680 700 720 740 760 780 800 Distance (ft) from Station 7917+07 -10 -5 0 C L O ffs et (f t) 7925+00 7 7.4 7.8 8.2 8.6 9 9.4 9.8 10.2 10.6 11 Base Dielectric 400 420 440 460 480 500 520 540 560 580 600 -10 -5 0 C L O ffs et (f t) 7923+01 200 220 240 260 280 300 320 340 360 380 400 -10 -5 0 C L O ffs et (f t) 7921+03 0 20 40 60 80 100 120 140 160 180 200 -10 -5 0 C L O ffs et (f t) 7917+07 7919+05 Figure 63. Contours of the Dielectric Values Measured with GPR; Section 1, TH-23 HMA Base Layer The repeatability of the GPR was found to be higher when testing HMA mixtures than when testing unbound materials. Cumulative frequency plots were not prepared because the variation between repeat runs was low for both air voids and thickness estimates. One reason for these lower standard deviations, in comparison to unbound materials, is that the driving lines and test points within a section were always well defined for the HMA surfaces, which was not always the case for unbound layers. Table 54 summarizes the GPR repeatability for testing HMA mixtures. Table 54. Summary of the GPR Repeatability for Testing HMA Mixtures Standard Deviation in Air Voids, percent Standard Deviation in Layer Thickness, inches Mixture Mean Median Range Mean Median Range US-180, HMA 0.54 0.38 0.0 to 2.10 0.057 0.04 0.0 to 0.28 I-35/SH-130, HMA 0.087 0.08 0.0 to 0.33 0.045 0.03 0.0 to 0.14 I-85 SMA 1.37 1.29 0.05 to 1.37 0.036 0.03 0.0 to 0.34

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 205 As shown, the lowest repeatability or higher standard deviations was found for the SMA mixture along I-85 in Auburn, Alabama. The reason for this higher variation between repeat measurements is unknown. However, the SMA did have the thinner layer tested, as well as a higher fluids content. Most of the COV values for estimating air voids within an area were less than 10 percent. Overall, the repeatability for the GPR is considered very good for both air voids and layer thickness. AC Section 1 - Air Void (uncalibrated) INFRASENSE, Inc. TH23 - Spicer, MN Arlington, MA 02476 Sheet: 1 of 3 Analyzed by: GLM Date: 10/21/04 Checked by: KRM Date: 10/22/04 600 620 640 660 680 700 720 740 760 780 800 Distance (ft) from Station 7917+07 -10 -5 0 C L O ffs et (f t) 7925+00 3.5 4.5 5.5 6.5 7.5 8.5 Air Void (%) 400 420 440 460 480 500 520 540 560 580 600 -10 -5 0 C L O ffs et (f t) 7923+01 200 220 240 260 280 300 320 340 360 380 400 -10 -5 0 C L O ffs et (f t) 7921+03 0 20 40 60 80 100 120 140 160 180 200 -10 -5 0 C L O ffs et (f t) 7917+07 7919+05 Figure 64. Contours of Air Voids Calculated from the Dielectric Values Measured Along Section 1 of TH-23 for the HMA Base Layer

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 206 0 20 40 60 80 100 120 140 160 180 200 -10 -5 0 C L O ffs et (f t) 7917+07 7919+05 200 220 240 260 280 300 320 340 360 380 400 -10 -5 0 C L O ffs et (f t) 7921+03 400 420 440 460 480 500 520 540 560 580 600 -10 -5 0 C L O ffs et (f t) 7923+01 600 620 640 660 680 700 720 740 760 780 800 Distance (ft) from Station 7917+07 -10 -5 0 C L O ffs et (f t) 7925+00 1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4 2.5 Thickness Scale (in) AC Section 1 - AC Thickness INFRASENSE, Inc. TH23 - Spicer, MN Arlington, MA 02476 Sheet: 1 of 3 Analyzed by: GLM Date: 10/21/04 Checked by: KRM Date: 10/22/04 Figure 65. Contours of the Layer Thickness Determined from Dielectric Values Measured Along Section 1 of TH-23 HMA Base Layer The following list summarizes the results from the GPR measurements in accordance with those conditions listed in Table 47: • TH-23 HMA Base – The GPR found that section 2 (northeast section) had the higher air voids, which would be consistent with a lower asphalt content. This observation is also similar to that from the PSPA test results. The coefficient of variation for all areas tested suggests low variability of the HMA mixture within each area, which is also consistent with the PSPA test results. • I-85 SMA Overlay – The GPR found lower air voids in the area with the higher asphalt content, which is consistent with the PSPA and FWD results. The mean air voids measured in lane C were consistent between all sections, and lower than the air voids calculated for lane B in sections 1 and 3 and lane A in section 3. Section 3 was found to have high air voids with the mean exceeding 10 percent in lanes A and B. In fact, the highest air voids were measured in lane A of section 3, which suggests inadequate compaction of the SMA mixture within this area. This observation is not believed to be the case, based on the test results from field cores and other nondestructive tests.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 207 • US-280 HMA Base – The GPR found no consistent difference in air voids between the initial sections tested a couple of days after placement. This finding is inconsistent with construction records and results from the PSPA and FWD. In addition, the GPR did not distinguish any difference between the initial and supplemental sections, while both the PSPA and FWD found the supplemental sections to have higher stiffness. It should be noted that the differences picked up by the PSPA and FWD were not planned. This issue will be addressed in more detail in chapter 7 of Part III. • US-280 HMA Base – No consistent difference in air voids was measured between the longitudinal joints and interior of the lanes. More importantly, the GPR air voids measured at the locations with some minor segregation were found to be the same or less than the other areas without segregation, which is inconsistent with previous experience and the results from the PSPA. • I-35/SH-130 HMA Base – The GPR found no difference in air voids between the sections tested. The lower air voids were measured within section 2. Conversely, the PSPA found the less stiff mixture to be in section 2. More importantly, the GPR found the air voids along the longitudinal joints to be lower than within the interior of the area tested—directly in contrast to the results from the other NDT devices. • Lift Thickness for all Projects – The GPR estimated the thickness for the first lift of the HMA base placed along the TH-23 project in Minnesota to be about 2.2 inches (see Figure 65)—the average thickness of the limited cores recovered for QA purposes. The GPR also provided an accurate estimate of the lift thickness in comparison to the cores recovered from the I-85 SMA overlay (1.7 inches) and SH- 130 base mixture (2.9 inches). The GPR results for the US-280 HMA base mixture, however, did have a negative bias of about 0.5 inches. This project was the only one included in the field evaluation where a 4-inch permeable asphalt treated base (PATB) layer was used. The HMA base mixture included in the project was placed directly above the PATB layer. All cores removed from the 3.5-inch HMA base lift for volumetric property determination and QA purposes did not include the PATB layer. It is possible that the PATB may have been less than the design thickness or the high air voids and moisture in that layer could have caused the bias in the HMA base layer placed above the PATB. 5.4.4 Non-Roller-Mounted Density Testing, Non-Nuclear—PQI and PaveTracker Two non-nuclear density gauges were used to measure the density of the HMA mixtures: the PQI and the PaveTracker. Figure 66 shows both devices used on the project. The tests were performed in accordance with the manufacturer’s recommendation. Five readings were made at each point. Density readings were made directly over the marked test point, and four additional reading were made at 90 degree deviations around the test point with the edge of the gauge’s base adjacent to the test point. Repeatability with the non-nuclear gauges was found to be very good. As an example, in Part B tests, the COV of cluster readings taken at each point was 1.32% and 99 percent of the test points had a COV of less than 5%.. This is in agreement with the results from a study sponsored by Wisconsin DOT on the evaluation of nonnuclear gauges (Schmitt et al., 2006, Rao et al., 2007).

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 208 Tables 55 and 56 summarize the average densities for each area tested in the Part A evaluation with the PQI and PaveTracker, respectively. During the testing sequence, it was noticed that the PaveTracker device started going out of calibration and malfunctioned during the testing along the US-280 project. It was returned to the manufacturer, and thus, fewer tests were performed with the PaveTracker device. Tables 57 and 58 summarize the average densities measured along the projects included in Part B. Figure 67 presents a cumulative frequency diagram of the standard deviation or repeatability of the non-nuclear density gauges. The standard deviations represent the five readings taken at each point. The repeatability for both gauges is considered very good. Figure 68 shows the cumulative frequency diagram of the standard deviation in density measured with a nuclear density gauge. With the exception of the readings taken on the HMA base mixture placed along US-280, both devices resulted in comparable, if not lower standard deviations than the nuclear density gauge. Figure 69 shows a comparison of the density readings made with each device at the same test points. As shown, there is a significant bias or difference between the two non-nuclear density measuring devices. The PaveTracker measured significantly lower densities. These low values, however, are not believed to be representative of the in-place mixture. (a) PQI device (b) Pavetracker device Figure 66. Non-Nuclear Density Measuring Devices for HMA Mixtures One reason for this observation is that the PaveTracker device was not calibrated to local materials at the beginning of the test program, while the PQI device was calibrated to the specific mixture prior to testing. Conversely, Figure 70 shows the results from the Wisconsin study referred (Schmitt et al., 2006, Rao et al., 2007). On that project, the PaveTracker device was calibrated to local mixtures, and the PQI device was used as received. As shown, the PQI resulted in significantly lower densities, which was not representative of the in-place HMA mixture.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 209 Table 55. Summary of the Densities Measured with the Non-Nuclear PQI Along the Projects Included in Part A Location or Designated Area Project Identification A B C Mean, pcf 148.2 147.6 147.0 TH-23 HMA Base, Section 1 – South COV, % 0.5 1.0 0.6 Mean, pcf 146.2 145.2 144.1 TH-23 HMA Base, Section 2 – Northeast COV, % 0.5 0.8 0.6 Mean, pcf 145.5 145.6 145.9 TH-23 HMA Base, Section 3 – Northwest COV, % 0.7 0.3 0.6 Mean, pcf 143.4 144.7 143.6 TH-23 HMA Base, Section 4, North – Just after paving and compaction COV, % 0.7 0.6 0.7 Mean, pcf 144.4 146.9 140.3 I-85 SMA Overlay, Section 1 – North COV, % 1.9 1.4 1.1 Mean, pcf 148.3 151.5 140.9 I-85 SMA Overlay, Section 2 – Middle COV, % 2.0 1.9 1.4 Mean, pcf 144.8 150.3 142.3 I-85 SMA Overlay, Section 3 – South COV, % 2.1 2.4 3.2 Mean, pcf 148.3 149.4 146.3 US-280 HMA Base, Section 1, Initial Testing COV, % 0.4 1.0 0.9 Mean, pcf 156.0 157.0 150.4 US-280 HMA Base, Section 2, Initial Testing COV, % 1.0 2.3 2.7 Mean, pcf 141.5 141.3 140.9 US-280 HMA Base, Section 3, Just After Paving, Initial Testing COV, % 1.5 1.1 1.8 Mean, pcf 145.7 US-280 HMA Base, Joint Measurements, Initial Testing COV, % 2.9 Mean, pcf 147.6 NA 146.6 US-280 HMA Base, Segregated Areas, Initial Testing COV, % 4.8 NA 2.6 Mean, pcf 141.6 140.2 139.3 US-280 HMA Base, Section 1, Supplemental Tests COV, % 1.8 1.2 1.4 Mean, pcf 139.5 140.6 141.4 US-280 HMA Base, Section 2, Supplemental Tests COV, % 1.6 1.1 2.6 Mean, pcf 142.9 US-280 HMA Base, Section 3, Supplemental Tests, Just after paving COV, % 1.2 Mean, pcf 135.8 US-280 HMA Base, Joint Measurements, Supplemental Tests COV, % 4.3 Mean, pcf 136.6 US-280 HMA Base, Segregated Areas, Supplemental Tests COV, % 1.2 Mean, pcf 140.8 141.2 US-280 HMA Base, IC Roller COV, % 1.4 1.5 Mean, pcf 127.2 126.9 125.4 I-35/SH-130 HMA Base, Section 1 COV, % 0.6 0.5 1.1 Mean, pcf 123.4 124.1 124.6 I-35/SH-130 HMA Base, Section 2 COV, % 1.1 1.1 1.6 Mean, pcf 124.8 125.3 125.2 I-35/SH-130 HMA Base, Section 3 COV, % 0.9 0.6 1.0 Mean, pcf --- --- 118.8 I-35/SH-130 HMA Base, Joints, Section 1 COV, % --- --- 1.4 Mean, pcf --- --- 120.1 I-35/SH-130 HMA Base, Joints, Section 3 COV, % --- --- 1.2 Note: The shaded cells designate those areas with anomalies (refer to table 47).

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 210 Table 56. Summary of the Densities Measured with the Non-Nuclear PaveTracker Along Projects Included in Part A Location or Designated Area Project ID A B C Mean, pcf 132.6 132.6 132.7 TH-23 HMA Base, Section 1 - South COV, % 0.4 0.6 1.5 Mean, pcf 130.2 131.6 131.0 TH-23 HMA Base, Section 2 – Northeast COV, % 0.8 0.6 0.4 Mean, pcf 130.6 131.8 132.4 TH-23 HMA Base, Section 3 – Northwest COV, % 0.3 0.4 0.7 Mean, pcf 131.6 132.1 131.9 TH-23 HMA Base, Section 4, North – After Paving COV, % 0.6 0.7 1.0 Mean, pcf 134.6 137.1 133.4 I-85 SMA Overlay, Section 1 - North COV, % 1.8 2.2 2.0 Mean, pcf 137.82 140.2 131.9 I-85 SMA Overlay, Section 2 - Middle COV, % 2.4 2.3 2.8 Mean, pcf 132.8 136.6 132.1 I-85 SMA Overlay, Section 3 - South COV, % 1.7 2.0 2.7 Mean, pcf 127.2 131.6 128.8 US-280 HMA Base, Section 1, Initial Testing COV, % 2.8 1.4 0.7 Mean, pcf NA NA NA US-280 HMA Base, Section 2, Initial Testing COV, % NA NA NA Mean, pcf NA NA NA US-280 HMA Base, Section 3, Just After Paving, Initial Testing COV, % NA NA NA Mean, pcf 126.5 US-280 HMA Base, Joint Measurements, Initial Testing COV, % NA Mean, pcf NA NA NA US-280 HMA Base, Segregated Areas, Initial Testing COV, % NA NA NA Mean, pcf 142.4 142.0 139.9 I-35/SH-130 HMA Base, Section 1 COV, % 1.2 3.0 1.4 Mean, pcf --- --- --- I-35/SH-130 HMA Base, Section 2 COV, % --- --- --- Note: The shaded cells designate those areas with anomalies (refer to table 47).

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 211 Table 57. Summary of the Densities Measured with the Non-Nuclear PaveTracker Along Projects Included in Part B Project Identification Section and Material Mean Density, pcf Coefficient of Variation, % Standard Deviation of Means, pcf Coarse-Graded Base 131.3 3.2 4.22 US-47, MO Fine-Graded Wearing Surface 136.3 2.0 2.77 Dense-Graded Surface Mix; Type 3-C 156.0 2.3 3.60 I-75, MI Dense-Graded Base Mix 145.5 2.5 3.66 US-2, ND Coarse-Graded Base Mix; PG58-28 132.8 1.1 1.44 US-53, OH Coarse-Graded Binder Mix 158.0 1.3 2.04 I-20, TX Coarse-Graded Binder Mix, CMHB 122.7 1.6 2.01 Surface Mix; 45% RAP, PG67; E-5 124.5 0.9 1.13 Surface Mixture; 45% RAP, PG76, SBS ; E-6 125.0 1.3 1.46 NCAT, Alabama Surface Mixture; 45% RAP, PG76, Sasobit ; E-7 124.9 1.0 1.25 Coarse-Graded Base Mix; PG67; N-1 130.9 1.3 1.78 NCAT, Florida Coarse-Graded Base Mix; PG76; PMA; N-2 131.5 1.0 1.28 NCAT, Missouri Dense-Graded Surface Mix; PG76; N-10 139.9 1.4 1.97 NCAT, SC Coarse-Graded Base Mix; PG67; S-11 134.1 0.7 0.98 Figure 71 shows a comparison of the measured densities with different devices when proper calibration procedures are followed closely. As shown, both non-nuclear devices resulted in densities similar to those measured with a nuclear density gauge. These test results note the importance of calibration to local materials for both devices to be considered for use in a QA program. Figure 72 shows the distribution of the standard deviation between the PaveTracker and PQI. Both result in similar variability when testing the same mixtures. Results from the PQI device will be used in comparison to the other NDT test results, because of this calibration and more extensive use of the device on the projects where the other NDT devices were used. Table 58 summarizes the densities in areas with anomalies or features—along longitudinal joints and in areas with segregation. As shown, the PQI generally measured lower densities along the longitudinal joints, as expected. However, the densities measured in the segregated areas were generally similar to those measured in those areas without segregation. A confounding factor related to the initial testing along US-280 is that wet weather in the form of a light drizzle occurred about halfway through the test program. Without question, some of the surface voids contained water below the surface when the test program was resumed. In fact, the operator made note that the densities began to consistently increase along the roadway after the wet weather began.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 212 0 10 20 30 40 50 60 70 80 90 100 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 7 7.5 Density Standard Deviation, PQI, pcf Cu m ul at iv e Fr eq ue nc y, % US-280, HMA I-85, SMA SH-130, HMA (a) PQI Gauge. 0 10 20 30 40 50 60 70 80 90 100 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 7 7.5 Density Standard Deviation, PaveTracker, pcf C um ul at iv e Fr eq ue nc y, % TH-23, HMA US-280, HMA I-85, SMA SH-130, HMA (b) PaveTracker Gauge. Figure 67. Cumulative Frequency of the Standard Deviation for the Non-Nuclear Density Gauges

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 213 0 10 20 30 40 50 60 70 80 90 100 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 7 7.5 Density Standard Deviation, Nuclear Density Gauge, pcf Cu m ul at iv e Fr eq ue nc y, % SMA - Thin Lift HMA - Thicker Lift Figure 68. Cumulative Frequency of the Standard Deviation for the Nuclear Density Gauges 120 125 130 135 140 145 150 155 160 120 125 130 135 140 145 150 155 160 HMA Density, Nuclear Gauge, pcf H M A D en st iy , N on -N uc le ar D ev is e, p cf PQI - SMA PaveTracker - SMA PQI - HMA Base PaveTracker - HMA Base Line of Equality Figure 69. Comparison of Densities Measured with the PQI and PaveTracker (without Proper Calibration) Devices to Those Measured with a Nuclear Density Gauge

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 214 125.0 130.0 135.0 140.0 145.0 150.0 155.0 142.0 144.0 146.0 148.0 150.0 152.0 154.0 Measured MC3 Density, pcf M ea su re d D en si ty , O th er D ev ic es , A vg ., pc f PaveTracker PQI-301 PQI-300 Line of Equality Figure 70. Comparison of Densities Measured with the PQI (without Proper Calibration) and PaveTracker Devices to Those Values Measured with a Nuclear Density Gauge (courtesy of the Wisconsin DOT) 140.0 142.0 144.0 146.0 148.0 150.0 152.0 154.0 156.0 158.0 140 142 144 146 148 150 152 154 156 158 Predicted MC3 Density, adjustment factor based on 10 points, pcf M ea su re d M C 3 D en si ty , p cf PaveTracker PQI-300 PQI-301 Line of Equality for 43, 19 mm, 6-02 Figure 71. Comparison of Densities Measured with a Properly Calibrated PQI and PaveTracker to Those Values Measured with a Nuclear Density Gauge (courtesy of the Wisconsin DOT)

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 215 0 10 20 30 40 50 60 0-0.5 0.5-1 1-1.5 1.5-2 2-2.5 2.5-3 3-3.5 3.5-4 Standard Deviation, Density Interval, pcf Fr eq ue nc y, p er ce nt PaveTracker PQI-300 Figure 72. Comparison of the Standard Deviations for Each of the Non-Nuclear Density Gauges Resulting from the Wisconsin Project (courtesy Wisconsin DOT) Table 58. Summary of the Densities Measured with the PQI Along Longitudinal Joints and in Areas with Limited Segregation Project Location/Area Lane Interior Area with Feature Ratio LONGITUDINAL JOINTS Mean, pcf 148.0 145.7 US-280 HMA Base; Initial Sections Multiple Days After Paving COV, % 1.20 2.9 0.984 Mean, pcf 139.6 135.8 US-280 HMA Base; Supplemental Sections Section 1, During Paving COV, % 1.42 4.3 0.973 Mean, pcf 126.4 118.8 Multiple Days After Paving COV, % 0.93 1.4 0.940 Mean, pcf 125.1 120.1 I-35/SH-130 HMA Base During Paving COV, % 0.80 1.2 0.960 SEGREGATED AREAS Mean, pcf 148.9 147.6 Section 1, Lanes A,B COV, % 0.83 4.8 0.991 Mean, pcf 146.3 146.6 US-280 HMA Base; Initial Sections Section 1, Lane C COV, % 0.95 2.6 1.002 Mean, pcf 139.5 136.6 US-280 HMA Base; Supplemental Sections Section 2 COV, % 2.78 1.2 0.979 NOTE: The ratio listed above is the mean density measured at the feature divided by the density measured within the interior of the lane or area.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 216 The cells in Tables 55 and 56 that correspond to those conditions listed in Table 47 have been shaded. The following list summarizes the results of the non-nuclear density gauges in accordance with those conditions listed in Table 47: • TH-23 HMA Base – The non-nuclear density gauges found that the higher densities were from section 1, while sections 2 and 3 had similar densities. The lower density was measured along lane C within section 2. Lower densities would be expected in areas with the lower asphalt content. However, construction records only indicate that section 2 had the lower asphalt content. • I-85 SMA Overlay – The PQI definitely measured higher densities in lanes A and B of section 2 with the higher asphalt content. In addition, the PQI found the lower densities consistently along lane C of all areas tested. This finding is consistent with the visual observations made during paving. • US-280 HMA Base – The PQI found that section 2 of the project used for the initial testing had the highest density. This finding is consistent with the GPR test results, but inconsistent the seismic results. This finding or observation is discussed in greater detail in the next paragraph. • US-280 HMA Base – The PQI found that the density in the segregated areas was slightly lower in most cases than in the areas without segregation. This difference was much less than expected based on previous experience. Conversely, the PQI measured large differences between the density along the longitudinal joints and interior of the lane tested. However, the densities measured along I-35/SH-130 project are low for both the interior and joint readings. It is believed that the measured densities are not representative of the in place values. Low densities were measured for both multiple days after placement as well as for immediately after compaction. The reason for this finding is unknown. As noted above, lower densities are expected in areas with lower asphalt content. However, the PQI found a much higher density in the area with lower asphalt content. The reason for these higher densities and lower air voids is believed to be related to moisture, as noted above. During the test program, the HMA surface became wet from earlier rains in the area. Water may have filled some of the permeable voids, especially in some of the segregated areas. Figure 69 showed the PQI density in comparison to those measured with a nuclear gauge. As shown, the density measurements made on the SMA are related and linear. However, there is extensive scatter for the HMA base mixture (the US-280 project). This scatter is believed to be related to moisture in the mixture from previous rains. More importantly, two groups of PQI HMA base density data are shown in the figure: one from the initial sections and the other from the supplemental sections. The initial sections have the higher densities, as measured by the PQI. Table 59 shows the average densities, seismic modulus values, and air voids measured on the mixture placed between these two time periods. The density was found to be much lower on the mixture placed after initial testing. Conversely, the air voids were found to be lower and the seismic modulus higher. The density and seismic modulus are different enough to suggest a change in material, even

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 217 though the results contradict one another. Thus, dynamic modulus tests were performed on samples from both testing areas (presented in section 5.5). Table 59. Summary of HMA Properties Measured on Mixtures Placed During Different Time Periods HMA Mixture Placed: NDT Device Mixture Property Initial Area Used for Testing Supplemental Area; After Initial Testing PQI Density, pcf 149.7 139.3 PSPA Seismic Modulus, ksi 415 540 GPR Air Voids, % 6.2 5.6 5.4.5 Roller-Mounted-Density and Stiffness Measuring Devices The Bomag Asphalt Manager IC roller was scheduled to be used to compact the SMA overlay in the areas where other NDT tests were planned as a demonstration. The IC roller was transported to the I-85 project (refer to Figure 22 in chapter 3), but the demonstration was postponed because of a problem that occurred with one of the roller’s components. The IC roller was used to compact the HMA base along US-280, north of Opelika, Alabama, after the component had been repaired. In summary, a nuclear density gauge and the PQI were used to measure the HMA density after each pass of the IC roller. These data were used to prepare densification curves and evaluate the changes in the measured responses of the IC roller to increases in density of the HMA layer. The density measurements made with each device were compared to the Evib readings from the IC roller. Figures 73 and 74 compare the density and Evib readings, while Figure 89 is an example of one of the densification curves that were prepared during this demonstration. The IC roller was able to adequately compact the HMA base mixture to a density of 153 pcf using the nuclear density gauge (Figure 74) or a value of 144 pcf using the PQI. The densities measured with the nuclear density gauge in this area, but without use of the IC roller, averaged about 151 pcf, while those measured with the PQI averaged about 142 pcf. Both gauges showed an increase in density of about 2 pcf with just the IC roller (no other roller was used). The contractor’s compaction operation included the use of two rollers to achieve the same density level. Figure 75 shows the benefit of using the Evib to determine when the correct number of passes has been used to reach density. The density gauges and the IC roller, through monitoring the Evib value, showed no additional increase in density of the mixture with continued passes of the IC roller. This additional information during lift compaction would be a benefit to most roller operators to ensure that an adequate density or stiffness of the HMA mixture had been reached.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 218 U.S. 280 19.0 mm NMAS y = 0.0819x + 131.94 R2 = 0.47 130 135 140 145 150 155 160 0 50 100 150 200 250 300 Evib * 100, psi N uc le ar D en si ty , l b/ ft^ 3 Figure 73. Comparison of the Nuclear Density Gauge Readings to the Evib Values Measured with the IC Roller As shown in Figures 73 and 74, both the nuclear and PQI gauges resulted in a large amount of data scatter. It should be noted that only 1 density measurement was taken at each of the 12 points along the compaction control strip, whereas 5 density readings were taken at each test point for the other areas tested. It is expected that the fewer density readings caused some of this scatter in the data. The reason for taking one density reading was not to delay the IC roller in compacting the HMA lift, because of mixture cooling with time. Although there is scatter in the data, this demonstration clearly showed the relationship between the density gauge readings and the Evib value measured with the IC roller. The PSPA was also used to measure the seismic modulus of the HMA mixture at the same points used to prepare the densification curves. The average values from this area are included in Table 48. The average seismic modulus measured in this area (the control strip) was 269 ksi, as compared to an average value of 239 ksi without the use of the IC roller. The seismic tests suggest a definite increase in the stiffness of the mixture, in addition to the increase in density. This demonstration clearly identified the potential benefit of using the IC roller for compacting HMA mixtures. The one major issue that has yet to be resolved is correctly taking into account the effect of decreases in temperature on the increase in Evib during compaction of the HMA lift.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 219 U.S. 280 19.0 mm NMAS y = 0.0328x + 134.31 R2 = 0.24 120 125 130 135 140 145 150 0 50 100 150 200 250 300 Evib * 100, psi PQ I D en si ty , l b/ ft^ 3 Figure 74. Comparison of the PQI Density Readings to the Evib Values Measured with the IC Roller 5.5 Laboratory Measured Modulus Dynamic modulus tests were conducted on all HMA mixtures, while repeated load resilient modulus tests were conduced on all unbound granular base materials and subgrade soils included in the field evaluation. The test results from these laboratory tests are included in Appendix C and discussed further in the following paragraphs for each material. 5.5.1 Unbound Aggregate Materials/Embankments and Subgrade Soils—Resilient Modulus Values Laboratory repeated load resilient modulus tests were completed for all of the unbound materials at the average in-place densities and moisture contents. The resilient modulus tests were performed in accordance with the provisional test procedure that resulted from NCHRP 1-28A. Twelve resilient modulus values were measured for each test specimen and are provided in Appendix B. However, only one stress state was used for consistency in comparing the field estimated elastic modulus values from each NDT device to values measured in the laboratory. Table 60 summarizes the resilient modulus values measured in the laboratory at a low stress state for the unbound aggregate base materials and embankment soils included in the field evaluation (Parts A and B).

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 220 Location 3 130 135 140 145 150 155 160 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 Roller Passes D en si ty , l b/ ft^ 3 0 50 100 150 200 250 300 Density Site 6 Density Site 7 Evib Site 6 Evib Site 7 Figure 75. Example of a Density Growth Curve Prepared from the IC Roller Demonstration and NDT Results As noted above, the test specimens were compacted to the average dry density and moisture content reported from the construction records. The dry density, moisture content, and percent compaction that apply to each area tested are also summarized in Table 60. Table 45 listed the optimum moisture contents and maximum dry densities resulting from M-D relationships for each material. In general, the resilient modulus values measured in the laboratory increase with the quality of the material. The dense crushed stone base material placed along US-280 has the highest resilient modulus, while the I-85 low plasticity silty-clay soil embankment prior to IC rolling had the lowest resilient modulus. The high plasticity clay subgrade along SH-21 has a much higher resilient modulus than expected based on previous testing experience with this soil (Von Quintus, 1980 to 1996). Two potential reasons for the larger values are that the average water content at testing (18.4 percent) was below the optimum water content (21.9 percent) and some gravel was mixed in with the upper subgrade soil. Water contents below the optimum value and approaching the plastic limit of the soil can significantly increase the resilient modulus. Similarly, the improved granular embankment placed along SH-130 was found to have a larger value than

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 221 expected. It is believed that the higher in-place densities (above the maximum dry unit weight) account for the higher modulus values. In addition, lime was used on this project to stabilize some of the subgrade soils. Visual observations of the material suggest that lime may have also been used in the improved embankment layer. Conversely, the South Carolina crushed granite base placed at the NCAT test track (test section S-11) had a much lower resilient modulus (14.3 ksi) than expected. One possible reason for this low value is that the density during NDT was well below the maximum dry density, in addition to the water content also being below the optimum value for this material (AASHTO T 180). Table 60. Summary of Average Repeated Load Resilient Modulus Values Measured in the Laboratory at a Specific Stress State Project & Materials Area Dry Density, pcf Moisture Content, % Percent Maximum Density, % Laboratory Resilient Modulus, ksi Before IC Rolling Section 1, Lanes B,C,D 103.0 21.6 0.91 2.5 I-85 Low Plasticity Clay Embankment After IC Rolling Section 1, Lanes B,C,D 108.0 16.9 0.96 4.0 NCAT; Oklahoma High Plasticity Clay 96.7 21.3 0.97 6.9 NCAT; South Caroline Crushed Granite Base 130.0 4.7 0.94 14.3 South Section Lanes A,B 121.0 8.2 0.98 16.0 TH-23 Embankment, Silt-Sand- Gravel Mix North Section Lane B,C 122.4 9.1 1.00 16.4 US-2 Embankment; Soil-Aggregate Mix 123.1 12.1 0.96 19.0 NCAT; Missouri Crushed Limestone Base 124.4 9.0 0.96 19.2 SH-21 High Plasticity Clay Area 1, with IC rolling Lanes A,B 107.3 18.4 0.99 26.8 Middle Area Lane B 139.4 4.3 1.04 24.0 TH-23 Crushed Aggregate Base South Area All Lanes 141.1 4.2 1.03 24.6 US-53 Crushed Aggregate Base, Type 304 136.0 9.1 1.01 27.5 NCAT; Florida Limerock Base 110.5 13.4 0.95 28.6 US-2 Class 5 Crushed Aggregate Base 134.4 5.9 0.95 32.4 SH-130 Improved Granular Sections 2, 3 Lanes A,B 128.7 9.1 1.05 35.3 US-280 Crushed Stone Areas 1,2,3 150.6 3.2 1.01 48.4 NOTES: • Resilient modulus values for the fine-grained soils and embankments are for a low confining pressure (2 psi) and repeated stress of 4 psi, while a confining pressure of 6 psi and repeated stress of 6 psi was used for the granular base materials. These low stress conditions are not based on any theoretical analysis. One stress state for the embankment soils and one for aggregate base layers were selected for consistency in comparing the field estimated elastic modulus values from each NDT device to values measured in the laboratory, which were considered the target values. • Percent maximum density is based on the maximum dry unit weight or density from the moisture-density relationship (the maximum dry densities were included in table 45 for each material tested).

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 222 5.5.2 HMA Mixtures—Dynamic Modulus Values Laboratory dynamic modulus tests were performed on all HMA mixtures sampled during construction in accordance with the provisional standard recommended for the MEPDG. The dynamic modulus was measured without confinement for five test temperatures (14, 40, 70, 100, and 140 F) and six loading frequencies (0.1, 0.5, 1.0 5.0, 10.0, and 25.0 Hz.). All dynamic modulus data measured on test specimens compacted to the in-place air voids and density are provided in Appendix C. Table 61 summarizes the dynamic modulus measured at temperatures of 70 and 130 F and a loading frequency of 5.0 Hz. These values were used for relative comparison purposes between the mixtures. Samples were recovered from the US-280 HMA base mixture for the two testing periods along the US-280 reconstruction project. As listed in Table 61, the dynamic modulus measured on bulk mixture recovered from both time periods is significantly different. The results from the laboratory dynamic modulus tests confirmed the NDT results that identified significant differences between the mixtures placed for the initial and supplemental sections. Thus, these two areas were treated as separate projects or mixtures in the analyses. The reason for this difference is unknown. Table 61. Summary of Dynamic Modulus Values Measured in the Laboratory Dynamic Modulus, ksi Part Project Identification Layer/Mixture 130 °F & 5 Hz 70 °F & 5 Hz B I-75, Michigan Dense-Graded Binder; Type 3-C 190 611 B NCAT, Florida Base, Mix, Sect. N-1; PG67 203 1,163 B NCAT, S. Carolina Base Mixture; Sect. S-11; PG67 214 1,228 B I-75, Michigan Fine-Graded Surface; Type E10 255 780 A I-85, Alabama Wearing Surface; SMA Mixture 230 1,485 B NCAT, Alabama Surface; 45% RAP; Sect. E-5, PG67 250 1,414 B US-47, Missouri Fine-Graded Surface 276 770 A TH-23, Minnesota HMA Base Mixture 319 1,848 A US-280, Alabama HMA Base; Initial Area 330 1,950 B US-47, Missouri Coarse-Graded Base; Shoulder 344 1,076 B US-2, N. Dakota Coarse-Graded Base; PG58-28 356 1,052 B NCAT, Florida Base Mix, SBS, Sect. N-2, PG76 366 1,614 B NCAT, Alabama Surface; 45% RAP, Sect. E-7; PG76 (Sasobit) 421 1,813 B NCAT, Alabama Surface; 45% RAP, Sect. E-6; PG76 (SBS) 427 1,836 B US-53, Ohio Coarse-graded Binder Mix 479 1,053 B I-20, Texas HMA Base Mixture, Type CMHB 520 1,600 A US-280, Alabama HMA Base; Supplemental Area 613 2,668 A SH-130, Texas HMA Base 965 4,271

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 223 5.5.3 Comparison of Laboratory Measured Modulus to NDT Measured Values Resilient Modulus of Unbound Materials Table 62 summarizes the laboratory resilient modulus and average modulus values estimated from the different NDT devices of each area tested within Part A. As shown, the GeoGauge and DSPA provided a reasonable ranking of each area tested, followed by the DCP. The deflection measuring devices did a poor job. This correspondence between the NDT and laboratory determined values would not change if a different stress state was used. Table 62. Elastic Modulus Values Estimated from the NDT Technologies and Devices, Without Adjustments, in Comparison to Resilient Modulus Values Measured in the Laboratory Modulus, ksi Project Material Area Lab.* GeoGage DSPA DCP LWD Section 2, Lane A 2.2 10.6 24.1 5.0 --- Section 1, All Lanes 2.5 15.4 30.0 5.9 --- I-85 Embankment Before IC Rolling Low Plasticity Clay Section 2, Lanes B, C, D 2.5 17.0 36.6 5.2 --- Section 1 4.0 16.8 30.4 6.9 9.99 I-85 Embankment After IC Rolling Low Plasticity Clay Section 2 4.5 19.0 40.4 6.2 11.78 So. Section, Lane C 15.0 13.2 31.1 11.5 5.6 So. Sect., Lanes A,B 16.0 18.3 43.6 15.2 5.7 No. Sect., Lanes B,C 16.4 17.8 35.7 19.0 4.7 TH-23 Embankment Silt-Sand- Gravel Mix No. Sect., Lane A 17.0 22.0 51.7 18.5 4.7 No IC Rolling 22.0 19.6 23.6 11.9 --- SH-21 Subgrade High Plastic Clay After IC Rolling 26.8 22.9 27.1 8.8 9.6 Middle Sect., Lane C 19.5 21.6 28.0 18.6 8.0 North Section, All Lanes; Middle Section Lanes A, B 24.6 28.2 79.3 33.1 12.3 TH-23 Base Crushed Aggregate Base South Section, Lanes A, B 26.0 33.0 110.7 46.4 19.4 Section 3 34.5 19.4 33.3 20.7 24.1 SH-130 Improved Embankment Granular Sections 1, 2 35.3 26.4 34.3 21.3 24.6 Area 4 40.0 35.1 117.4 34.3 18.5 US-280 Base Crushed Stone Areas 1, 2, 3 48.4 47.9 198.6 50.3 46.5 NOTES: * - The repeated load resilient modulus values measured in the laboratory, but corrected to the actual dry density and moisture content measured for the specific section, in accordance with the LTPP procedure and regression equations.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 224 Figure 76 compares the NDT results and those values measured in the laboratory. As shown, the GeoGauge provided a reasonable estimate of the laboratory values across all materials included in the Part A field study (fine-grained soils to crushed stone), with the exception of the I-85 fine-grained, low-plasticity embankment. The DCP also provided a reasonable estimate of the laboratory resilient modulus values. The elastic modulus values estimated from both the DSPA and LWD devices increase with increasing values measured in the laboratory but have a significant bias. The DSPA has a positive bias or over-estimates the laboratory values, while the LWD has a negative bias or under-estimates those values and has the greater dispersion. Dynamic Modulus of HMA Mixtures Unlike for unbound materials, the modulus of HMA mixtures is affected significantly by temperature and frequency. The PSPA values can be adjusted to a temperature and load frequency selected for design. The internal adjustments are included in the software and initially based on global default values. However, these default adjustments can be determined in the laboratory by measuring the seismic modulus on test specimens prepared during the mixture design process, on bulk mixture compacted to the density or air void level expected during construction, or on field cores recovered during construction. Global default values initially were used to calculate the seismic modulus at a load frequency of 5 Hz for the field evaluation projects. Table 63 compares the seismic and deflection-based moduli to values measured in the laboratory. The in-place temperatures for the laboratory values included in Table 63 were measured with multiple devices during NDT. An average mix temperature was used to estimate the laboratory dynamic moduli listed in Table 63. The FWD deflection-based moduli were calculated from the deflection basins using a forward- calculation program. Figure 77 compares the PSPA and laboratory measured HMA modulus values. As shown, the PSPA moduli appear to correlate to the laboratory measured values. The FWD moduli are significantly different than those measured in the laboratory. Two factors, however, have a significant effect on the use of global adjustments. The first is that asphalt binders have different temperature susceptibilities, so the use of a constant global adjustment can result in a significant error. The second is that laboratory compacted test specimens for the dynamic modulus testing do not have checking and mat tears, while field tests are subjected to these fractures which can have a significant effect on the NDT measurements. These issues will be evaluated and discussed in greater detail in chapter 7 of Part III of the research report. 5.6 Summary of Field Projects A diverse range of HMA mixtures and unbound materials/soils were included in the field evaluation of NDT devices (see Table 29). Appendix B provides a summary of each project. Tables 31 and 47, within this chapter, listed the anomalies that exist along those projects included in the Part A field evaluation. Although no anomalies were planned for the Part B projects, construction defects were observed on some of the projects. At the end of this chapter is a listing of those defects and material issues that should have an impact on the

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 225 quality characteristics measured by the QA tests. The effectiveness of the NDT devices in measuring these defects is discussed in Part III of the research report. 0 50 100 150 200 0 10 20 30 40 50 60 Laboratory Resilient Modulus, ksi El as tic M od ul us fr om N DT D ev ic es , k si DSPA, Fine-Grained DSPA, Coarse-Grained Line of Equality Geo., Fine-Grained Geo., Coarse-Grained (a) DSPA and the GeoGauge. 0 10 20 30 40 50 60 0 10 20 30 40 50 60 Laboratory Resilient Modulus, ksi E la st ic M od ul us fr om N DT D ev ic es , k si LWD, Fine-Grained LWD, Coarse-Grained DCP, Fine-Grained DCP, Coarse-Grained Line of Equality (b) Deflection-Based and DCP methods. Figure 76. Comparison of Laboratory Resilient Modulus and the Elastic Modulus Values Estimated with Different NDT Technologies and Devices

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 226 Table 63. Elastic Modulus Values Estimated from NDT Devices, Without Any Adjustments, in Comparison to Dynamic Modulus Values Measured in the Laboratory Laboratory Values, ksi NDT Values, ksi Part Project Identification Layer/Mixture 130 °F & 5 Hz In Place Temp. & 5 Hz PSPA FWD B I-75, Michigan Dense-Graded; Type 3-C 190 400 435.2 --- B NCAT, Florida Base, Mix; PG67 203 390 447.1 --- B NCAT, S. Carolina Base Mix; PG67 214 410 495.2 --- B I-75, Michigan Fine-Graded Surface; Type E10 255 590 676.3 --- A I-85, Alabama SMA Mixture 230 250 237 450 B NCAT, Alabama 45% RAP; Sect. E-5, PG67 250 450 510.7 --- B US-47, Missouri Fine-Graded Surface 276 530 457.6 --- A TH-23, Minnesota HMA Base Mixture 319 810 480 --- A US-280, Alabama HMA Base; Initial Area 330 650 462 165 B US-47, Missouri Coarse-Graded Base 344 420 605.3 --- B US-2, N. Dakota Coarse-Graded Base; PG58-28 356 510 344.3 --- B NCAT, Florida Base Mix, SBS, PG76 366 590 475.8 --- B NCAT, Alabama 45% RAP, Sect. E-7; PG76 (Sasobit) 421 610 444.3 --- B NCAT, Alabama 45% RAP, Sect. E-6; PG76 (SBS) 427 640 473.4 --- B US-53, Ohio Coarse-graded Binder Mix 479 850 666.7 --- B I-20, Texas HMA Base, CMHB 520 340 435.5 --- A US-280, Alabama HMA Base; Supplemental Area 613 780 558 310 A SH-130, Texas HMA Base 965 1,750 342 725

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 227 0 500 1000 1500 2000 0 500 1000 1500 2000 Laboratory Measured Dynamic Modulus (In Place Temperature and 5 Hz.), ksi N D T Es tim at ed M od ul us , ks i Line of Equality FWD Modulus PSPA Modulus - Part A PSPA Modulus - Part B (a) Entire data set. 100 200 300 400 500 600 700 800 900 100 200 300 400 500 600 700 800 900 Laboratory Measured Dynamic Modulus (In Place Temperature and 5 Hz.), ksi N D T Es tim at ed M od ul us , ks i Line of Equality FWD Modulus PSPA Modulus - Part A PSPA Modulus - Part B (b) Excludes data point for very stiff HMA mixture placed along SH-130. Figure 77. Comparison of Laboratory Dynamic Modulus and the Elastic Modulus Values Estimated with Different NDT Technologies and Devices

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 228 Unbound Materials and Embankments: No construction defect was observed in any of the Parts A and B projects. As listed in Table 44, however, there were differences in the condition of the base materials and embankments that were planned to ensure that the NDT devices would identify those differences. HMA Mixtures: • US-280 HMA Base Truck-to-truck segregation observed in some areas. Cores were taken in these areas, but some of the cores disintegrated during the wet coring process. In addition, a significant difference in dynamic modulus was found between the initial and supplemental sections included in the test program. The supplemental section was found to have much higher dynamic modulus values. This difference was not planned. • I-85 SMA Overlay No defects noted. • TH-23 HMA Base No defects noted. • SH-130 HMA Base No defects noted during the time of testing, but there was controversy on the mixture because it had been exhibiting checking during the compaction process. Changes were made to the mixture during production. The change made and the time that the change was made were unclear relative to the time that the NDT evaluation. • US-47 HMA Base The mixture was tender; and shoved under the rollers. • US-47 Wearing Surface Portions of this mixture were rejected by the agency in other areas of the project. • I-75 HMA Base, Type 3-C No defects noted, but mixture was tender – placed along the shoulder. • I-75 HMA, Type E3 & E10 No defects noted, but portions of this mixture were rejected by the agency in other areas of the project. • US-2 HMA Base Checking and mat tears observed under the rollers. • US-53 HMA Base No defects noted. • I-20 HMA CHMB Base No defects noted. • NCAT – Alabama HMA RAP; with & without modifiers No defects noted on any of the test sections. • NCAT – South Carolina HMA Base No defects noted. • NCAT – Missouri HMA Base No defects noted. • NCAT Florida – PMA Base No defects noted. • NCAT Florida – HMA Base, no modification Checking and mat tears observed under the rollers.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 229 Sponsor Agency Missouri DOT Project Location St. Clair & Union, Missouri Project Identification Two-lane widening project – shoulder reconstruction and HMA Overlay across entire width of roadway. The HMA base was 4 inches in thickness, while the HMA overlay was 1.75 inches in thickness. Materials Tested HMA Base and HMA Wearing Surface Special Features Tender HMA base mixture placed along the shoulder. Issues Rain occurred during the shoulder placement or construction. The wet weather did not affect placement of the HMA mixture. Positive Aspects 1. The contractor rolled the HMA mixture placed along the shoulder. The confined edge of the HMA was being rolled using the cold-side pitch method. It was observed that the HMA mixture was being pushed away from the cold joint. The PaveTracker was used to measure the density along the confined joint. The densities were low. The contractor was encouraged to change the rolling pattern – roll from the hot side of the joint. Densities were measured with both the PaveTracker and nuclear density gauge along the confined joint. The densities increased by about 5 to 8 pcf between the two rolling patterns. The contractor changed the rolling pattern to increase the joint density. 2. Another positive aspect is that the PSPA did identify the soft HMA mixture after placement through problems with obtaining a smooth waveform from the PSPA. It was originally believed that the PSPA had been damaged during transport; however, the PSPA was identifying the mixture to be tender. 3. The initial wearing surface/overlay was found to have low air voids and was rejected by the DOT. The PSPA and PaveTracker did identify these differences during construction and placement. The test results for the new mixture placed were found to be statistically different. Negative Aspects 1. Rains resulted in delays and scheduling conflicts. The rain caused the contractor to move off of the job and return weeks later. Thus, the test equipment was not left with the contractor nor agency personnel. However, both the contractor and agency personnel did use the equipment on site during initial testing when the shoulder was being placed with positive results and comments. 2. The HMA mixture being placed along the shoulders was a soft mixture. In fact, the mixture was so soft that indentations could be observed from light loads on the mix after it had cooled down to 160F. The PSPA was used to test the HMA mixture being placed along the shoulder. This point could also be a positive aspect of the project. 3. The unbound aggregate base course was planned to be tested along the shoulder areas after the surface material had been removed, the base material scarified and compacted or removed and replaced. However, the unbound aggregate base along the shoulder area was found to be in excellent structural condition and was able to support the construction equipment. Thus, the unbound aggregate base layer was left in place without any additional compaction and work. Use of the GeoGauge, DSPA, and DCP were excluded from the field testing plan. Sponsor Agency Missouri DOT Project Location NCAT Test Track Project Identification New construction of two structural sections placed at the NCAT test track facility. Both test sections were instrumented by NCAT. Materials Tested Crushed limestone base layer and an HMA binder layer Special Features None. However, the pavement structure did include a high binder content base mix or crack resistant layer. This layer was not tested under the NCHRP Project 10-65.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 230 Issues None. However, the crushed limestone layer was compacted at a water content that was below the optimum water content. This required many more passes or coverages of the roller than expected or planned. Positive Aspects 1. Density and modulus growth curves were measured using both the GeoGauge and DSPA devices. Two DCPs were used to measure the in place strength of the base material. 2. Density growth curves were also measured using the PaveTracker for the HMA binder mixture. Negative Aspects None. Sponsor Agency Michigan DOT Project Location Saginaw, Michigan; I-75 Project Identification I-75 rehabilitation included PCC rubblization in the northbound lanes and milling and overlaying the existing HMA surface in the southbound direction with 7 inches of HMA, consisting of a 3 inch HMA base, 2 inch HMA Binder layer, 2 inch Wearing surface. The portion included in NCHRP Project 10-65 was confined to the southbound lanes. Materials Tested HMA 4-C base and HMA 3-C binder layers. Included both Superpave designed mixtures along the main lanes and Marshall designed mix placed along the shoulders. Special Features None. Issues Rain occurred during the testing period that delayed the paving operation, but it is believed that the rain had no impact on the HMA mixtures being placed. Positive Aspects Density growth curves were developed for the HMA base and binder layers. Negative Aspects During the testing operation, the DOT rejected about 2 miles of HMA that had been previously placed. The contractor ceased paving operations until the cause of the rejected material could be determined. The equipment was not left with the agency and contractor personnel because of this problem and dispute of test results in the DOT’s day-to-day acceptance plan. In addition, the wearing surface was not tested as part of NCHRP Project 10-65. Sponsor Agency North Dakota DOT Project Location Williston, North Dakota Project Identification Realignment and new construction of US-2 between Minot and Williston. Materials Tested HMA base layer (PG58-28); Crushed Gravel – Class 5 Base; Fine-grained embankment (soil-aggregate mix) Special Features The crushed aggregate base layer was tested in two conditions; the first area had been placed over a year ago, while the second area had been placed a couple of weeks prior to arrival at the project site. The surface of the crushed stone base that had been placed the previous construction season received a prime coat to protect it from construction traffic. The test equipment was left with the DOT and contractor personnel for use in testing other areas of the project as the paving materials were placed. Issues Rain and tornados occurred during the week of testing. The unbound materials were tested prior to the rainfall and the HMA layer was tested prior to and after the storm. Positive Aspects Both the DOT and contractor personnel used all gauges during and after the testing under NCHRP Project 10-65. Negative Aspects The HMA base mixture checked and tore during one day’s production. The checking and mat tears occurred under the finish roller – after the density had been obtained by the contractor using the breakdown and intermediate rollers. The PaveTracker and

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 231 PSPA were used to test the area with checking. The checking and tears were found to be severe in localized areas. This is also considered to be a positive aspect of this project, in terms of using the NDT devices. Sponsor Agency Ohio DOT Project Location Freemont, Ohio Project Identification Realignment and widening of SR-53, near Freemont, Ohio; between Toledo and Columbus, Ohio. Materials Tested HMA 19 mm base mixture and crushed stone base layer (304 base material). Special Features None. Issues Rain and wet weather occurred during the week of testing. The contractor also had problems with the plant which delayed another project in the area. Thus, the contractor did not move the paving equipment and rollers back on the project until the testing under NCHRP Project 10-65 had been completed. The test equipment, however, was left with the DOT personnel for use and continued testing for the following weeks. The DOT retained the test equipment for more than two weeks. Positive Aspects 1. Water from the recent rains had accumulated in areas with insufficient drainage to remove the rainfall from the pavement area. The DCP, GeoGauge and DSPA measured low modulus values in the areas where water had been standing and allowed to penetrate the base layer. 2. The PSPA and PaveTracker were used to test the HMA base mixture in all areas where the DOT had taken cores for acceptance testing. Negative Aspects No HMA paving and placement of the unbound aggregate base material was completed during the initial week of testing. Thus, density and modulus growth curves could be obtained from this project for the HMA base mixture and aggregate base layer. Sponsor Agency Alabama DOT Project Location NCAT Test Track Project Identification Mill and HMA overlay of three test sections along the test track. Materials Tested HMA wearing surface with 45 percent RAP and different asphalt binders. The asphalt used in the three sections included a PG 67-22, a PG76-22 with Sasobit, and a PG-76- 22 with SBS. Special Features High amount of RAP in the HMA overlay and different asphalt binders. Issues None. Positive Aspects None, with the exception of comparing the density and seismic modulus for mixtures with high amounts of RAP and varying asphalt grades, as compared to mixtures without RAP. Negative Aspects None; however, the HMA overlay was placed a week prior to the testing under NCHRP Project 10-65. Thus, density growth curves were not obtained. Sponsor Agency Florida DOT Project Location NCAT Test Track Project Identification New construction; two structural sections that were constructed side by side; one with a neat asphalt mix and the other with a polymer modified asphalt mix. Both of these test sections were instrumented by NCAT. Materials Tested Limerock base material; a high binder content HMA base mix considered a crack resistant layer; HMA binder layer with a neat asphalt; an HMA binder layer with a

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 232 modified asphalt. Special Features Pavement cross section included a 3-inch high binder base layer to resist fatigue crack initiation at the bottom of the pavement. This layer or mixture was tested. Two other HMA mixtures were tested – Florida’s standard neat asphalt type mixture and another with a polymer modified asphalt. Issues 1. The temperature of the HMA neat asphalt mix was low at the time of placement. The low temperature made getting density difficult and caused the mixture to check and tear under the rollers. 2. The HMA mixture with the neat asphalt checked during compaction in localized areas. The checking was considered severe in a localized area. The PMA and high binder content base layer did not check or team under the rollers, or at least the checking and tears were not observed during placement. Positive Aspects A comparison of a neat HMA mix to that of a PMA mix. The HMA neat mix did check while the PMA mix did not check. Two DCPs were used to test the limerock base layer. Negative Aspects None. However, the contractor had a lot difficulty in getting the required density for both the HMA neat asphalt mix and the PMA mix. A rubber tired roller was used to continue the compaction operation for many hours. The density did finally reach the required value. PaveTrack and PSPA tests were completed on the mix with the low densities, as well as with the required or specific density. Sponsor Agency Oklahoma DOT Project Location NCAT Test Track Project Identification New construction; two structural test sections that were built side-by-site. It was originally designed that these two sections would be full-depth HMA pavements placed over a high plasticity subgrade soil imported from Oklahoma. Both of these sections were instrumented by NCAT. Materials Tested High plasticity clay soil was in a relatively dry condition (with extensive and wide shrinkage cracks), and high plasticity clay soil compacted to the optimum dry density (without the shrinkage cracks that could be observed at the surface); HMA binder layer. Special Features Wide shrinkage cracks existed in the high plasticity clay soil at the time of initial testing for NCHRP Project 10-65. Issues None. Positive Aspects 1. The effects of wide shrinkage cracks in a high plasticity clay soil can be assessed in terms of their effect on the test results from the GeoGauge and DSPA. 2. Two DCPs were also used to test the subgrade soil in different conditions. 3. Density and modulus growth curves were measured during the original compaction of the clay soil. Negative Aspects The surface of the high plasticity clay soil was removed, the lower soil scarified, reworked, and re-compacted, and a 6-inch layer of local chert aggregate was placed. A misunderstanding of the cross section for these structural sections had occurred. The Oklahoma DOT wanted an intermediate layer of aggregate placed between the HMA base and high plasticity clay. Thus, the NCHRP Project 10-65 tests on the high plasticity clay were performed twice. Sponsor Agency South Carolina DOT Project Location NCAT Test Track Project Identification New construction of a structural section at the NCAT test facility. This section was instrumented by NCAT.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 233 Materials Tested Crushed granite base layer; HMA base mixture and HMA binder layer. Special Features None Issues Contractor tried to use a smaller roller for compacting the crushed granite layer. The water content of the crushed granite base layer was about half of the water moisture content. Contractor could not get density. Tried using the BOMAG Asphalt Manager – which was a problem by disturbing (decompacted) the surface of that base layer. Positive Aspects Density and modulus growth curves were measured for the crushed granite base material. Negative Aspects The water content of the crushed granite base was about half of the optimum water content, and the roller that was available could not densify this material past a specific density. A heavier roller had to be brought to the test section to get the required density. The DSPA and GeoGauge did detect the lower density levels. Sponsor Agency Texas DOT Project Location Odessa, Texas Project Identification Reconstruction of I-20 main lanes, due to construction of overpass, and reconstruction of frontage roads. HMA was placed in two 2-inch lifts. Materials Tested HMA Coarse Matrix High Binder Content Base Layer (CMHB) under new DOT specification; the crushed stone base course material was not tested. A surface treatment had already been placed on top of the crushed stone base layer at the time of testing. Special Features None Issues None Positive Aspects 1. Contractor and DOT were already using the PaveTrack for setting the rolling pattern and DOT was already using the PSPA for acceptance confirmation. Density growth curves were measured by both the contractor’s and NCHRP Project 10-65 PaveTracker devices. Contractor was positive towards using the non-nuclear density gauges and did use the PSPA. Results from the PSPA demonstrated that the HMA mixture was meeting all minimum requirements of the mixture. 2. Multiple PSPAs were used on this project; the one being used under NCHRP 10-65 and by the Odessa district office. The Texas DOT had already used the PSPA for use as a forensic tool in evaluating the failure, prior to the contractor finishing the paving, on a 7-mile section of I-20 through Odessa. The DOT and UTEP agreed to provide that data for use on NCHRP 10-65. Negative Aspects 1. Plant breakdown that significantly delayed paving operation during the week scheduled for the testing under NCHRP Project 10-65. 2. High winds and sand storm occurred during paving that resulted in contractor ceasing paving operations during the week selected for testing under NCHRP Project 10-65. 3. The crushed stone base layer with typical aggregate in west Texas (similar to a caliche) was planned for testing. However, crushed stone base materials had already been covered with a surface treatment prior to NCHRP 10-65 testing. Thus, the DCP, DSPA, and GeoGauge were not used on this project. Sponsor Agency Texas DOT Project Location Odessa, Texas Project Identification Mill and overlay main lanes along Loop 338. Materials Tested --- Special Features HMA overlay was a modified asphalt mixture with rubber. Issues Project was cancelled, as noted below.

NCHRP Project 10-65—Volume 2: Research Report June 2008 Part II—Summary of Findings Final Report 234 Positive Aspects --- Negative Aspects Contractor was delayed from another project and plant breakdown further delayed the paving operation. Contractor’s new schedule was to place the HMA modified asphalt mix with rubber after Thanksgiving. Thus, project was cancelled relative to NCHRP Project 10-65. Sponsor Agency Pecos Research and Test Center Project Location Pecos, Texas Project Identification New construction of the entrance roadway to a private facility located near Pecos, Texas. Materials Tested Caliche base typically used for county roads in west Texas. Special Features Salcido Sand and Gravel Company was placing a caliche base without time restrictions. Material was used to measure the increase in material strength with successive passes of a static steel drum roller. Issues None. Positive Aspects Modulus growth curves were developed using two devices; the DCP and GeoGauge. Negative Aspects None.