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

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

12 Construction Quality Management Using Intelligent Compaction Introduction Technological improvement of construction technologies has resulted in the popularity of the IC techniques. Even though the basic concept of IC was developed in the early 1970s (Adam and Pistrol 2016), this technology has been under continuous development and implementation during the past decade. The following sections summarize the current body of research relevant to the performance management methods, including the IC applica- tions. The information gathered from the literature on different methods for estimating the modulus of compacted geomaterials in the field and laboratory, and the various factors that impact the mechanical properties also are summarized. The chapter includes a brief review of the numerical modeling techniques, constitutive and material models, and soil-drum contact mechanics used for simulating roller compaction of geomaterials. It also provides a review of Intelligent Compaction Measurement Values (ICMVs) and current techniques for backcalculation of mechanical properties. Finally, the chapter offers a summary of current specifications for implementing IC technology. Estimation of Modulus of Compacted Geomaterials The stiffness/modulus of the compacted geomaterials can be estimated either from the laboratory or from in-situ tests. The behavior of the unbound materials under repeated loading is quite complex and involves many different factors. Given the complexity and time-consuming nature of the resilient modulus tests, simple methods have been proposed for estimating the modulus of the geomaterials in the laboratory. Strength tests, such as the unconfined compres- sive strength or laboratory California Bearing Ratio (CBR), also are commonly used to estimate the modulus. Correlations have been developed by various studies in the literature to predict the modulus from the soil index parameters; however, most models exhibit poor predictive power when they are tested on soils not used to develop the relationships (Von Quintus and Killingsworth 1998; Yau and Von Quintus 2002; Wolfe and Butalia 2004; Malla and Joshi 2007). A comprehensive review of the concepts and definitions regarding the response of geomaterials is provided in Appendix C. Factors Impacting Modulus of Compacted Geomaterials Puppala (2008) and Tutumluer (2013), among others, synthesized the body of literature regarding the estimation of the modulus of the unbound geomaterials. The following sections summarize the body of literature regarding the factors influencing the modulus/stiffness of the earthwork and unbound geomaterials: C H A P T E R 2

Construction Quality Management Using Intelligent Compaction 13 State of Stress. Several material models have evolved in the decades since the 1960s. Different forms of the stress state have been implemented to explain the stress-dependency of the modulus. The representative modulus of a given geomaterial placed in a pavement section is not a unique value and depends on the underlying and/or overlying layers. The state of the stress of a given geomaterial placed in a pavement section can only be estimated if the moduli of all layers are known. The estimation of the modulus must be carried out iteratively using an analytical layered structural model. Residual Stresses During Compaction. The stresses imposed by compaction equipment during the construction process are usually the largest stress states that the compacted unbound geomaterials will experience during their service lives. The particle interlock that is formed during the compaction process, along with the lateral confining pressure, forms a residual stress within the geomaterial layer that could affect the responses of the pavement layers during the repeated traffic loading. Moisture Content. A review of the literature reveals that significant efforts have been dedicated to studying the impact of the moisture variation in terms of the moisture content or matric suction (Gupta et al. 2007). Like those of Cary and Zapata (2010), most of these studies are based on the concepts of the unsaturated soil mechanics. Wolfe and Butalia (2004) acknowledged the significance of unsaturated soil mechanics in characterizing the behavior of pavement subgrades and the influence of suction or moisture content variation on modulus. The soil may be subjected to variation in stiffness due to interaction with the atmosphere, leading to repeated cycles of infiltration and evaporation, referred to as hydraulic hysteresis, which in turn can lead to a change in soil stiffness (McCartney and Khosravi 2013). As detailed in Appendix C, several research efforts have documented the specific impact of moisture variations and the general impacts of environmental changes on the modulus/ stiffness of compacted earthwork and unbound geomaterials. The lack of a correlation between the modulus of the compacted geomaterials and the field moisture content is discussed in Richter (2006). Von Quintus et al. (2010) and Pacheco and Nazarian (2011) also attempted to address that concern. The importance of the difference between the moisture content at the time of compaction and at the time of testing has been suggested by Khoury and Zaman (2004) and by Nazarian et al. (2014). The Mechanistic-Empirical Design Guide (MEPDG) recommended the following function to consider the effects of the environmental factors, Fenv, on the resilient modulus, MR, at any degree of saturation: F MR MRenv opt (2-1)= where MRopt = the resilient modulus at the optimum moisture content (OMC). That model was further calibrated in terms of degree of saturation using a series of experi- ments under different moisture conditions in the following form:     = + − + ( )β− − MR MR a b a eopt K S Sm opt log 1 (2-2) where S = current degree of saturation (decimal), Sopt = degree of saturation at OMC (decimal), a = minimum of log (MR/MRopt),

14 Evaluating Mechanical Properties of Earth Material During Intelligent Compaction b = maximum of log (MR/MRopt), β = regression parameter = ln (–b/a), and Km = regression parameter. Studies about the impact of the moisture content/degree of saturation/suction variation on the stiffness and ICMV of the geomaterials during and shortly after the compaction process are limited. Thompson and White (2007) used test strips constructed at three different moisture contents to evaluate the impact of the moisture content on the ICMVs. They discussed that the inevitable variation in the moisture content of the compacted geomaterials during construction could affect the quality management of earthwork and unbound geomaterials. Siddagangaiah et al. (2014) performed an extensive field evaluation of IC for the quality control of the base and soils in Texas. Even though some weak correlation between the LWD modulus and the moisture content was observed, a reasonable relationship between ICMV and the moisture content could not be reported. They also confirmed that a certain level of uncertainty was associated with the estimation of the in-situ moisture contents using NDG. White and Vennapusa (2015) empha- sized the need for the utilization of moisture content sensors in the future development of the IC monitoring of the earthwork and soil layers. Density. The study of the variation in modulus of compacted geomaterials with density has been limited in the literature. Some efforts were aimed at investigating the impact of the density combined with the moisture content in terms of the degree of saturation (Cary and Zapata 2011). Other studies, such as those conducted by Mooney and Rinehart (2009), Von Quintus et al. (2010), Pacheco and Nazarian (2011), and Nazarian et al. (2014), could not establish a direct correlation between the modulus and density. Nazarian et al. (2014) indicated that, considering the uncertainties associated with the estimation of the density by NDG, a reason- able correlation could not be found among any of the in-situ moduli and density. Floss et al. (1991) reported several correlations between ICMV and density in terms of the percent compaction. The goodness of the fit for their correlations was less than that obtained with the plate load test results. Bräu et al. (2004) reported correlations between ICMV and spot- test results, including density with significant scatter. Mooney et al. (2003 and 2005) correlated ICMV to several spot-test results, including the dry density. They showed that such correlations improved when the lifts were stiff. Peterson (2005) and White and Thompson (2008) reported poor correlations between the ICMV and dry density. Gradation and Plasticity. Many empirical models can estimate the modulus of a geo- material using index properties such as gradation parameters and Atterberg limits. Navarro et al. (2012) and Nazarian et al. (2014) summarized several such regression models; however, most models are applicable and reasonable only for the specific soil conditions for which the test protocols were developed. Long-Term Versus Short-Term Behaviors of Geomaterials. Significant work has been done to predict the long-term changes in the moisture content/suction and modulus of the compacted geomaterials under the in-service pavement. However, the amount of work related to the short-term behavior of the exposed geomaterials (as related to quality management) has been limited. In a proper field compaction, the geomaterial is placed near the optimum moisture content and the moisture change is due to either evaporation or the introduction of moisture. The moduli obtained from this process can be vastly different from the moduli measured in the laboratory under a constant compaction effort (Khoury and Zaman 2004; Sabnis et al. 2009; Pacheco and Nazarian 2011). During the first few days, a freshly compacted material experiences several phenomena (e.g., thixotropy, moisture loss and equilibrium) that cannot be modeled using most models, which were developed to represent the long-term behavior of the materials in relation to seasonal and other environmental variations.

Construction Quality Management Using Intelligent Compaction 15 Numerical Modeling Techniques of Roller Compaction Experimental data collected with instrumented roller compactors has revealed complex nonlinear roller vibration behaviors, which include the loss of contact between the drum and the soil, as well as the drum and the frame rocking (Adam and Kopf 2004; Anderegg and Kaufmann 2004; Mooney et al. 2006). Various numerical modeling techniques have been attempted to address some of these concerns. Numerical models can be either physical-based models or finite element (FE) models. Physical-based models, which include lumped-parameter, boundary-element (BE), and discrete-element (DE) models, have been studied by van Susante and Mooney (2008), Mooney and Facas (2013), and Buechler et al. (2012), among others. Research on FE models has been conducted by Xia and Pan (2010), Mooney and Facas (2013), Erdmann and Adam (2014), and Keneally et al. (2015), among others. The following summary comments can be made about the implementation of the different models for simulating roller compaction: • Lumped-parameter models are the simplest but the least realistic models. These models can provide results rapidly, but site-specific calibration may be time-consuming. • BE models reduce the dimensionality of the problem, resulting in savings of computation time and resources; however, roller dimensions and operation parameters, as well as the plastic response of geomaterials associated with compaction, must be addressed using iterative processes and indirect means to adjust soil responses. • DE models can provide a wealth of information about the performance of granular materials under the rollers. However, the execution time at realistic scales is prohibitive. • FE models are the most versatile tools for obtaining the responses of geomaterials under rollers. Simple linear elasto-static models (especially 2D models) are rapid to execute. As the problem is extended to 3D with dynamic loading and with plastic and nonlinear geomaterial behavior, the execution time becomes rather time-consuming for routine use. • The best numerical model to use is the one that best balances the execution time, the accuracy of the results, and the amount of time a DOT is willing to spend to obtain the necessary material parameters. Detailed information about the modeling techniques that have been implemented for simulating roller compaction of geomaterials is provided in Appendix D. Constitutive/Material Models of Geomaterials It is essential to incorporate reasonable geomaterial constitutive/material models. A summary of constitutive models of geomaterials is provided in Appendix D. The following statements can be made about them: • Linear elastic models are the most versatile and recognized constitutive models for geo- materials. Because most of the design algorithms are based on linear elastic models, the harmonization of selecting the target modulus/stiffness for field IC compaction with the design parameter selected for each layer becomes straightforward. However, linear elastic material models may only be applicable to stiffer materials at the end of the compaction process. • Nonlinear (resilient modulus) models are more realistic than linear elastic models because the highest state of stress that most layers experience is during compaction. However, these models should be implemented with caution. Based on a survey of DOTs conducted during the NCHRP 10-84 study by Nazarian et al. (2014), one of the top two reasons that would

16 Evaluating Mechanical Properties of Earth Material During Intelligent Compaction impede the implementation of the modulus-based quality acceptance was the incorporation of the resilient modulus tests in the specifications. If nonlinear models are to be adopted, an indirect means of estimating the nonlinear model parameters should be considered. In addition, the uncertainty of estimating these parameters should be weighed against the benefits gained in terms of the more realistic results. • Plasticity and hypo-plasticity models have limitations similar to those of resilient modulus models. Though these models offer more realistic responses than the linear elastic models, implementation of plasticity and hypo-plasticity models would require incorporation of laboratory tests such as shear, triaxial, and consolidations tests into the specifications. Inclusion of estimation methods must be developed for parameters used by these models, and their uncertainty must be assessed. Soil-Drum Contact Mechanism Several alternative models, from the classical Hertzian models to the more modern and sophisticated BE models, have been used for addressing the soil-drum contact mechanism. The following statements can be made about the reviewed contact models: • Hertzian models are easy to implement but are overly simplified, as they overlook the nonlinear response of the geomaterial. Their implementation would require the inclusion of methods to adjust the calculated contact widths in order to address the nonlinear behavior of soils. The use of Hertzian models may, nonetheless, be applicable to stiffer granular materials or at the end of the compaction process. • BE models better address soil-drum contact width and stress distribution than do Hertzian models. The implementation of BE models would require an iterative approach to adjust the contact width, which is greatly expedited due to faster execution times than common FE approaches. • DE models have been found to be more realistic in addressing contact widths than Hertzian models, and DE models have been found to better address cohesive soils than do other methods. Further research to investigate the stress fields is still needed, however, and the dynamic analysis required by DE models makes them time prohibitive. • FE models generally include different contact models depending on the program used. These models address the stress fields and contact widths reasonably well, given the FE software’s capacity to consider the soil’s nonlinearity, but they require the implementation of dynamic loading, which leads to time-consuming execution analyses that would be unsuitable for routine applications. More realistic contact widths and stress fields will depend on the chosen numerical model and constitutive material model that better addresses execution time suitable for routine use. Intelligent Compaction Measurement Value (ICMV) Mooney et al. (2010) described the roller measurement values in detail. The various data measurement values used for compaction control are listed in Table 2-1. The ratio between the amplitude of the first harmonic and the amplitude of the excitation frequency was first correlated to the stiffness of the soil as measured by dynamic plate load tests (Thurner and Forsblad 1978). Thurner and Sandström (1980) introduced the compaction meter value (CMV). Since then, various measuring systems have been implemented by the roller manu- facturers. In 1982, Bomag introduced the OMEGA value and the Terrameter measuring system, and these systems were followed by the vibration modulus Evib, a measure of dynamic soil stiffness (Ferris 1985; Floss et al. 2001; Kröber et al. 2001). In 1999, Ammann introduced the

Construction Quality Management Using Intelligent Compaction 17 ACE (Ammann Compaction Expert) that calculated the soil stiffness parameter ks (also called kB) (Anderegg and Kaufmann 2004; Anderegg 1997). Mechanistic ICMVs The introduction of Evib and ks signaled an important evolution toward the measurement of more mechanistic, performance-related soil properties (e.g., soil stiffness/modulus). These two ICMVs are determined from the force-displacement hysteresis loops. The hysteresis loops are interpreted from the drum acceleration time histories collected by the IC rollers. The force-displacement loops are created by plotting the time-varying contact force, Fc, versus time-varying drum displacement, zd, where contact force is calculated from the vertical response of the drum. The Ammann ACE system calculates the secant soil stiffness, ks, from the gradient of the line passing through the point of zero dynamic displacement (i.e., displacement due to the static weight of the roller) to the point representing the maximum dynamic drum displacement, as shown in Figure 2-1 (Anderegg and Kaufmann 2004; Mooney et al. 2010). To determine these parameters, the system takes advantage of the lumped-parameter model. That model uses a roller and a 2 degrees of freedom (2 DOF) model representing the vertical kinematics of the drum-frame system. The drum/soil contact force, Fs, consists of the machine weight, the eccentric force, and the drum and frame inertias, and is described in Appendix C. The Ammann Measurement Value Manufacturers Parameters Used Relations Used Compaction Meter Value (CMV) Dynapac, Caterpillar, Hamm, Volvo Ratio of vertical drum acceleration amplitudes at fundamental vibration frequency and its first harmonic. 2ACMV c A where c is constant around 300, A2Ω is the amplitude of second harmonic, and AΩ is amplitude of fundamental frequency. Compaction Control Value (CCV) Sakai Algebraic relationship of multiple vertical drum vibration amplitudes, including fundamental frequency, and multiple harmonics and sub harmonics. 1 3 4 5 6 1 2 100A A A A ACCV A A where Ai are amplitudes at the excitation frequencies. Stiffness, ks (kb) Ammann Vertical drum displacement, drum-soil contact force. 2 0 0 cos s d d m ek m z where md is drum mass, m0e0 is eccentric mass moment, is phase angle, zd is drum displacement, and Ω is frequency. Vibration Modulus, Evib Bomag Vertical drum displacement, drum-soil contact force. 22 1 1.8864 lnsd vib F Lz E L b where Fs is drum soil interaction force, L is the drum length, b is contact width, is Poisson ratio, and zd is drum displacement. Machine Drive Power (MDP) Caterpillar Difference of gross power and the power associated with sloping grade and machine loss. sing aMDP P WV mV b g where Pg is gross power, W is roller weight, a is acceleration, g is acceleration due to gravity, θ is slope angle, V is roller velocity, and m and b are internal loss coefficients. Table 2-1. Commercially available roller measurement values (from Mooney et al. 2010).

18 Evaluating Mechanical Properties of Earth Material During Intelligent Compaction system determines the drum inertia and the eccentric force by measuring the vertical drum acceleration and eccentric mass position, whereas the frame inertia is neglected. The vertical drum displacement amplitude, zd, is determined by the spectral decomposition and integration of the measured peak drum accelerations (Anderegg and Kaufmann 2004). Secant stiffness, ks, is calculated from k m m e z s d d cos , (2-3)2 0 0= Ω + φ    where md = the drum mass, m0e0 = the eccentric mass moment, W = the excitation frequency, and φ = the phase lag between the eccentric mass and the drum displacement. Like the Ammann ACE system, the Bomag Variocontrol system makes use of the force- displacement hysteresis curves to determine the tangent stiffness (defined as the slope to the force-displacement loop at locations of 80% and 20% of the difference between the maximum and minimum contact forces in calculating a “vibration modulus,” Evib [Mooney et al. 2010]). Correlation Analysis Studies Several studies have evaluated the roller measurement values for the compaction quality management of different pavement layers and embankment soils. Research has also been carried out to correlate the roller measurement values with the in-situ point test measurements. Even though different manufacturers recommend different ICMVs, the vertical, longitudinal, and transverse heterogeneity of the underlying soil strata is the most important factor influ- encing ICMVs and the modulus-based spot-test results. The correlations developed with ICMVs and the spot tests change whenever there is a change in the underlying condition. The hetero- geneity stems from the changes in material type, compaction effort, and moisture contents at the time of compaction and testing (Nazarian et al. 2014). The depth of influence for a regular (11 ton to 15 ton) roller is reported to vary between 2.5 ft. to 4 ft. (Mooney et al. 2010). Hence, the ICMVs measured will reflect the composite stiffness of the geomaterials up to a depth of 2.5 ft. to 4 ft.; however, the spot tests typically reflect the material property up to a depth of 0.5 ft. to 1 ft. (Mooney et al. 2010). Fo rc e Displacement kt ks Figure 2-1. Calculation of secant stiffness, ks, and tangent stiffness, kt.

Construction Quality Management Using Intelligent Compaction 19 Current Backcalculation Techniques Backcalculation (also called system identification or inversion) is an optimization process performed to inverse map a known relation established by discrete or continuous data points. The most commonly known backcalculation process in pavement engineering is related to the interpretation of the results from the FWD. In FWD backcalculation, the measured deflections are “matched” with the calculated deflections from a numerical algorithm. Usually, the matching process between the measured and calculated responses is performed by an iterative process, in which the responses are calculated using different sets of assumed mechanical properties. Göktepe et al. (2006) provided a thorough comparison of the various backcalculation tech- niques in terms of modeling precision, computational expense, calculation details, and data requirements. Figure 2-2 presents an overview of these backcalculation methods. Their imple- mentation has been possible due to the tremendous advances in computational power, which has significantly minimized the computation time required for the backcalculation processes. In the context of this study, the backcalculation methods can generally be categorized as static, dynamic, or adaptive (Göktepe et al. 2006). Static and dynamic methods are classified by their loading types and utilize the conventional pavement response models. Adaptive methods, such as neural networks and neuro-fuzzy systems, do not directly use a response model; instead, they simulate inverse mapping by learning the target behavior via known input-output data patterns. A forward model and an inverse algorithm are used in the backcalculation process. In the forward process, the responses are computed based on the loading and pavement structure, typically using a linear elastic procedure. The inverse process can be performed using various optimization (error minimization) processes. Optimization can be performed using a param- eter identification algorithm (PIA) such as nonlinear least-squares, database search algorithms (DSA), or genetic algorithms (GAs). GAs are used in genetic programming (GP), which offers a model-free artificial intelligence (AI)-based optimization technique that mimics the theory of evolution. Source: Göktepe et al. (2006) Figure 2-2. Overview of backcalculation methods.

20 Evaluating Mechanical Properties of Earth Material During Intelligent Compaction In the inverse process, the calculated responses are compared to the measured responses so that the new mechanical properties are determined by a parameter identification routine. Optimization is achieved through an iterative process until the differences between the calculated and measured deflections stay under a certain error criterion. Backcalculation Methods in IC Technology Mooney and Facas (2013) evaluated different backcalculation processes for determining layer moduli with a forward process that used a static BE model simulating roller compaction. The forward process predicted the stiffness over a wide range of two-layer pavement structures with different layer moduli and top-layer thicknesses, as shown in Figure 2-3. Based on a sensitivity analysis, the authors suggested that the simple minimization algorithms could be used without the need of more complex techniques. Mooney and Facas found their approach time intensive, as each inversion required 5 to 15 iterations, and each iteration required forward modeling. To increase efficiency in the backcalculation process, they used direct inverse models created through regression analyses to substitute for the simulations in the forward model. The authors found that a local tri-cubic (LTC) interpolation, a ninth-order polynomial fit regression model, and an artificial neural network (ANN) model were able to simulate the responses with acceptable error. Approaches to Include IC in Specifications This section summarizes the approaches to include IC in construction specifications and reviews of the current U.S. and European IC specifications. Mooney et al. (2010) proposed several options to include IC in construction specifications. In Option 1, the IC results are used to identify the weakest areas and test them with the conventional spot-testing devices for acceptance. In Option 2, the acceptance is based on limiting pass-to-pass percentage change in (b) (a) Source: Mooney and Facas (2013) Figure 2-3. Comparison of simulated stiffness, k, values from BE analysis for a two-layer system with (a) bottom-layer modulus, E2, versus top-layer modulus, E1, and top-layer thickness, h1 = 30 cm; and (b) k versus top-layer modulus, E1, for variable top-layer thickness, h1.

Construction Quality Management Using Intelligent Compaction 21 either mean ICMV (Option 2a) or spatial percentage change in ICMV (Option 2b). In Option 3, the acceptance is based on whether the ICMV has met the requirements of target ICMVs, which can be determined either by relating ICMV to spot-tests (Option 3a), by determining the plateau value of ICMV compaction curve (Option 3b), or by relating ICMV to laboratory tests (Option 3c). Mooney et al. (2010) summarized the challenges to widespread utilization of the IC as a quality acceptance tool in the following manner: • Heterogeneity in underlying layer support condition; • High moisture content variation and adjustment of ICMV according to moisture condition; • Narrow range of spot test measurements and ICMV; • Machine operation setting variation (e.g., amplitude, frequency, speed) and roller double- jumping; • Nonuniform drum/soil contact conditions; • Uncertainty in spatial paring of spot test measurements and ICMV; • Limited number of spot test measurements; • Lack of other construction and materials information to help interpreting the IC results; and • Intrinsic measurement errors associated with ICMV and spot test measurements. Uniformity is recognized as a key parameter of compaction that relates to performance. Uniformity can be gauged by univariate statistics (e.g., variance) or spatial geostatistics (e.g., semi-variograms), but areas with the same variance may vary in spatial statistics. Another approach is to use the roller coverage as a method-based acceptance process. Most U.S. IC speci- fications have been using such an approach (e.g., 70% of compacted areas must have the target number of line passes or more). This method does not apply for this project, however, because of the goal of determining mechanical properties of the geomaterials. Review of Current Specifications FHWA and AASHTO IC Specifications The current FHWA specifications provide generic guidelines to various state and local high- way agencies for the implementation of the IC technology. These guidelines can be modified by the agencies to meet the state’s specifications and specific project requirements. The current provision, AASHTO PP 81-14, “Standard Practice for Intelligent Compaction Technology for Embankment and Asphalt Pavement Applications,” is a combined specification for both soils and hot-mix asphalt applications. The key features of the two specifications are compared in Table 2-2. U.S. States’ Specifications Figure 2-4 shows the distribution of the current IC specifications across the United States. The “Quality Management Special—Intelligent Compaction (IC) Method” (Minnesota DOT) specifies all the requirements listed by AASHTO PP 81-14. The “Intelligent Compaction for Soils” (Georgia DOT), “Intelligent Compaction Mapping of Subbase and Aggregate Base” (Michigan DOT), and “Intelligent Compaction for Subbase and Reclaimed Stabilized Base (RSB) Applications” (Vermont Agency of Transportation [VTrans]) use a RTK-GPS with a 1.6-in. verification tolerance. The IC rollers to be used on Georgia DOT projects require approval from the DOT and need to be listed in the agency’s “Approved Vendor List.” The Michigan DOT waives the pre-mapping requirement, and the VTrans waives the training requirement. The Indiana DOT’s “Quality Control/Quality Assurance, QC/QA, Soil Embankment” specifi- cation follows the general requirements of the FHWA specification using RTK-GPS with a 6-in. verification tolerance. However, the specification does not require the contractor to perform pre-mapping prior to construction.

22 Evaluating Mechanical Properties of Earth Material During Intelligent Compaction Reference FHWA Soils AASHTO GPS verbiage HPPS 1 RTK-GPS GPS verification tolerance 12 inch 6 inch Temperature verification tolerance (for asphalt) NO 5°F Require alignment files NO YES 2 Departmental approval of rollers NO YES Roller vendors listed NO NO Test strip required YES YES Pre-construction mapping YES 3 NO Veta software required YES YES IC training 4-8 hours 4 Required 5 IC training includes Veta YES NO 6 Data submittal Daily Daily IC-based acceptance NO 7 NO Basis of payment Lump sum Lump sum and partial 8 1HPPS is intended to include all positioning technologies including GPS, laser, cellular signals, and so forth. 2AASHTO requires agencies to provide alignment files. 3FHWA only allows pre-mapping on soils and granular subbase. 4FHWA recommends project-specific just-in-time training. 5AASHTO requires yearly certification. 6AASHTO assumes that agencies are trained to use Veta. 7FHWA includes “IC Construction Operations criteria” that are based on pass count and ICMV coverage. 8AASHTO includes partial payments: 10% based on certification, 10% based on providing IC equipment, and 80% based on roller pass counts coverage. Table 2-2. Comparison of FHWA and AASHTO IC specifications. Figure 2-4. Current IC specifications for soils application.

Construction Quality Management Using Intelligent Compaction 23 The “Special Note for Intelligent Compaction of Aggregate Bases and Soils” of the Kentucky Transportation Cabinet (KYTC) follows the FHWA guidelines and specifies the use of RTK-GPS. In addition, the KYTC specification includes an approval process for the IC rollers. The Iowa DOT, North Carolina DOT, Pennsylvania DOT, and Texas DOT use the RTK-GPS but do not require any verification of GPS, and do not require pre-mapping to be performed prior to construction. The North Carolina DOT requires the roller vendor to have prior IC experience (three completed projects). The Texas DOT requires approved rollers from the vendor list only. The Texas DOT also waives the training requirement. European Specifications Specifications for roller-integrated measurement systems for quality control and quality assurance have been developed and implemented in Europe as national compaction standards for more than two decades under the term Continuous Compaction Control (CCC). Introduced in Austria in 1990 with revisions in 1993 and 1999, these revisions were adopted in Germany in 1994 with revisions in 1997 and 2009 (see ZTVA-StB 97); in Finland in 1994; in Sweden in 1994 with revisions in 2004 (see VVR VÄG 2009); and in Switzerland in 2006. The International Society for Soil Mechanics and Geotechnical Engineering (ISSMGE) developed specifications based primarily on the Austrian specifications (ISSMGE 2005; Adam 2007). Mooney et al. (2010) classified the European specifications under two options: • Option 1: Spot Testing in Roller-Identified Weak Areas. Following these specifications, the CCC roller performs a vibratory proof roll over the earthwork area under evaluation. The quality acceptance inspector uses the roller MV data map to identify the weakest area(s) for spot testing. If the roller-identified weakest areas meet the spot-test requirements, the entire evaluation area meets the threshold for acceptance. In principle, the validity of this option is tied to the existence of a positive correlation between the spot-test measurements and the ICMVs (i.e., high ICMVs correspond to high density and vice versa). Such positive correlations should be verified before quality acceptance testing. • Option 2: Calibration of ICMVs to Spot Test Values. Specifications classified under this option consist of (1) An on-site correlation assessment to relate the ICMV to the selected (or contract specified) spot test measurement; (2) identification of an ICMV target value (ICMVTarget) consistent with the required spot-test value based on the regression performed in Step 1; and (3) acceptance testing by comparing the ICMV data from the evaluation area with the ICMVTarget. Given its relative simplicity, Option 1 is the more common of the two options, and it is the only option permitted in Sweden. The principal components of the various specifications and planned revisions are described in Ninfa (2013). Spot-testing requirements vary by country. Acceptance of earthwork materials in Europe is based primarily on use of the plate load test (PLT) or the lightweight deflectometer (LWD), whereas in the United States acceptance is primarily based on the dry density measure- ments using NDG. Common Elements of IC Specifications Common elements of the federal and state IC specifications can be categorized using four sets of characteristics: • Pre-approval of IC system requirements and quality control plan, • Field operation requirements, • Data requirements and submission, and • Measures and payment.

24 Evaluating Mechanical Properties of Earth Material During Intelligent Compaction Pre-Approval of IC System Requirements and Quality Control Rollers Types, Vendors and Department Approval. Agency specifications typically require that contractors submit documentation of the roller supplier, make, and roller model, along with the number of IC rollers to be provided for the project. Application-based instru- mented roller types are specified, along with the requirements for their accuracy, GPS, rolling speed, frequency, and amplitude. Agency approval of the instrumented rollers may be required, as may the demonstration of the rollers at agency-approved locations. Some specifications include lists of approved or recommended IC roller brands and models. One key requirement is an “accelerometer-based” measurement system that provides a measure of the vibration amplitude of the roller, preferably in conjunction with measurement of the applied load, which enables an assessment of the compaction quality (generically as an ICMV). The preferred param- eters that meet this requirement include kb (Ammann/Case) and Evib (BOMAG and Dynapac). Alternative systems that might be acceptable include CCV (Sakai), CMV (Caterpillar, Trimble, and TOPCON), EDV (VOLVO), and HMV (Hamm). GPS Requirements. The GPS radio and receiver units are required to be mounted on each IC roller to track the roller passes. A handheld GPS also is required for measuring the locations of the spot tests with the density/modulus-based devices for correlation with the ICMV data. General specifications related to the GPS, including the definitions, devices, and networks, are included. GPS is often a loose definition to cover all global navigation satellite systems (GNSS), including GPS from the United States and GLANOSS from Russia. The precision requirements often specify real-time kinematic (RTK) or survey-grade precision. A high precision positioning system (HPPS) is used to cover all positioning technologies such as the GNSS, laser-based, and cellular-based systems. Contractor Quality Control Plan. The contractor is required to prepare and submit a written Quality Control Plan (QCP) for the project. In addition to meeting the requirements of a general construction QCP, the plan should list the person(s) responsible for operating the IC roller(s) and attached IC equipment, and include the training documentation for the roller operator(s). Field Operation Requirements IC Training. Appropriate IC training is critical to implement the IC procedure successfully because most contractors and DOT staff are not familiar with this technology. The IC vendors’ technical support normally provides the IC training. GPS Field Validation. The requirements to check the proper setup and to verify the accuracy of the GPS on the IC rollers against a hand-held rover are common and critical. Without such validation, the GPS offsets may render the IC data useless. The tolerances for the difference between the measurements have been specified as 6 in. to 12 in. The daily field GPS validation is often specified but not always enforced. Alignment Files. An alignment file may be loaded onto the onboard documentation system of the instrumented roller and into the cloud-computing mapping software when used. A requirement for the agency to provide the relevant alignment files is usually included. Pre-Paving Mapping. Performing pre-mapping with an IC roller on the existing support geomaterials is recommended to identify the weak areas and to evaluate the condition of the underlying materials such as soils subgrade, aggregate bases, or similar materials. Test Strip. A test strip to test section construction in order to successfully establish target compaction pass counts and target values for the strength of the materials is specified. Test strips are not always enforced at actual projects, especially for embankment work.

Construction Quality Management Using Intelligent Compaction 25 Conventional Spot Testing. IC technology is currently used for quality control, but the acceptance is still based on the geomaterial density and sometimes on moisture measurements made with the NDG. A need exists for identification of the standard testing device(s) and frequency for measuring the in-place density and moisture content of the soil. Data Requirements and Submission IC Data Requirement. An IC roller is both a construction machine and a data collection system. The IC data is the key to leveraging the benefits of the IC technologies. The data header block and data blocks are often required, with the most common data elements including the roller type and size information, GPS system setting, and IC measurements (including the roller passes, vibration amplitudes and frequencies, and ICMV). Veta Compatibility. Currently, most states require that the IC data be compatible with Veta software. Veta (formerly known as Veda) can import data from various IC systems to perform standardized viewing and analysis. Veta is required as a standardized tool in the FHWA and AASHTO IC soils specifications. IC Data Submission. The timely submission of the IC data is normally required. The IC data may include all passes, but including the data from the final coverage is mandatory. Measures and Payment IC-Based Acceptance. The IC construction operations criteria does not generally affect the standard agencies’ acceptance processes for the materials or construction operations since IC is mainly used for quality control. Basis of Payment. The incorporation of the IC process in a project is currently based on a lump-sum price in most DOTs. This item includes all costs related to providing the IC roller(s) including the fuel, roller operator, GPS system, or any other equipment required for the IC process. All quality control procedures, including the IC rollers and GPS systems representatives’ support, on-site training, and testing facility, are included in the contract lump-sum price. It is becoming increasingly popular to include payment breakdown to cover the IC equipment provided, training conducted, IC data submitted, and IC coverage (e.g., minimum number of line passes within a certain project boundary). Issues with Implementing Current IC Specifications Given that the intention of the research team was to ensure the practicality and robustness of the specification, it was desirable to consider the current issues related to the implementation of the IC. Based on the experience of the research team, the important issues included: • Uniformity Across the Country. Most state IC specifications are primarily modeled after the FHWA’s generic IC specifications, but variations remain. (For more information, see Appendix C in the downloadable “Appendices.pdf” available from the NCHRP Research Report 933 web page at www.trb.org.) State IC specifications range from 3 pages in length (e.g., the KYTC specifications) to 19 pages (e.g., the Minnesota DOT specifications). The IC vendors often report difficulties communicating with the DOTs and meeting their specifica- tions, though vendors sell the same machines and systems across the country. • Lack of Field Qualification/Certification Process for IC Rollers and Operators. The IC technology is primarily used for quality control. The inclusion of pay items that tie to the IC data is gaining popularity in specifications. As a result, an increasing need is to have a field procedure to qualify and certify the IC systems to ensure valid data is collected for calculating pay items. Currently, only the AASHTO IC specification includes “certification” verbiage in

26 Evaluating Mechanical Properties of Earth Material During Intelligent Compaction the appendices regarding contractor personnel and provides a checklist to approve IC rollers. Like the pavement profiler certification, this certification is not trivial, and it may be costly to establish and operate. • Qualification of On-Site Training. The IC training is one of the critical requirements to ensure the success of any IC project. Though most state IC specifications require an on-site or “just-in-time” training, it is difficult to provide qualified trainers to do so. A need may exist to include additional training-related language in the specifications to clarify who would provide the training and who would receive the training. • Difficulty of Conducting Daily GPS Validation and Complications of Not Performing Such Tests. Most of the current problems with smooth implementation of IC technology are GPS-related issues. Most of the issues resulted from failing to perform the GPS validation prior to the fieldwork. Most contractors still do not understand that implementing the IC and GPS is not a turnkey solution, but rather requires rigorous steps to ensure that all elements of the IC technologies, with the GPS as the core, function as they are supposed to function. Among states that do require daily GPS validation, enforcement of this requirement may be neglected because it is time-consuming to perform the necessary tasks. • Controversy of Pre-Mapping Requirement. Pre-mapping can be used to identify less stiff areas and perform corrective actions. Due to the limitations set by the unknown depth of the identified less stiff layer(s) using the current ICMV technologies, the question, “How weak is too weak?” does not yet have a definitive answer. In addition, the roller manufacturers do not recommend pre-mapping of hard surfaces due to concerns with the potential damage to the rollers. • Practicality of Conducting Test Strips. For various reasons, it may not be practical to conduct a test strip for every project. Besides being time consuming, it can be impractical to conduct test strips when variations in moisture content or soil type affect support conditions throughout a project. • Difficulty of Determining Target Values from Test Strip Data. Several DOT IC specifications include the requirements to determine target roller passes and target ICMV from test strip data. The material, production, equipment, and climatic variabilities encountered during construction make it difficult to determine and set target values, not to mention the differ- ences in test mechanism and testing footprint and measurement depths. Therefore, one of the main goals of this research was to provide proven, practical field procedures to determine layer-specific target values for compaction acceptance. • Issues with Data and Report Submission (Data Management). Currently, the IC systems allow either manual data storage or wireless data transmission. Manual data storage makes use of USB drives and is prone to errors or file loss with ill-trained personnel. Wireless transmis- sion relies on a cellular connection for automated submission, and the connection is not always reliable given potential losses of cellular coverage or incorrect project setup. Pitfalls related to both data submission methods are still unresolved. Also, it is still a steep learning curve for the contractors to learn the vendors’ software to export the IC data. Although several DOTs have started requiring their contractors to submit both the IC data and the IC analysis reports, the contractors and IC vendors are still struggling to meet such requirements due to a lack of training. Therefore, data management is currently one of the main concerns in the IC implementation.