Predicting Flexible Pavement Structural Response Using Falling. Weight Deflectometer Deflections. A thesis presented to.

Predicting Flexible Pavement Structural Response Using Falling Weight Deflectometer Deflections A thesis presented to the faculty of the Russ College of Engineering and Technology of Ohio University In partial fulfillment of the requirements for the degree Master of Science Jianfeng Qin June 2010 2010 Jianfeng Qin. All Rights Reserved.

This thesis titled 2 Predicting Flexible Pavement Structural Response Using Falling Weight Deflectometer Deflections by JIANFENG QIN has been approved for the Department of Civil Engineering and the Russ College of Engineering and Technology by Shad M. Sargand Russ Professor of Civil Engineering Dennis Irwin Dean, Russ College of Engineering and Technology

ABSTRACT 3 QIN, JIANFENG, M.S., June 2010, Civil Engineering Predicting Flexible Pavement Structural Response Using Falling Weight Deflectometer Deflections (94 pp.) Director of Thesis: Shad M. Sargand This thesis presents a model to predict the pavement response using Falling Weight Deflectometer (FWD) deflection data for asphalt concrete (AC) pavement. Evercalc 5.0 and Elmod 6.0 were chosen to conduct the backcalculation of pavement layer moduli using FWD deflections. Everstress 5.0 was used to do the forward calculation using backcalculated layer moduli. Predicted pavement responses (tensile strain at the bottom of the AC layer) were compared to the measured pavement responses from U.S. Route 30 to check the validity and accuracy of the selected prediction model. The predicted results show a good agreement with the measured responses. A comparison between FWD and truck load conditions was also conducted. The results show that FWD can accurately simulate the magnitude and the duration of a moving single wheel load. Approved: Shad M. Sargand Russ Professor of Civil Engineering

ACKNOWLEDGMENTS 4 First of all, I would like to express my sincere appreciation for my advice, Dr. Shad M. Sargand for his support, help and guidance during my graduate study. I would also give my deep appreciation for my committee members Dr. Sang-Soo Kim, Dr. Deborah McAvoy, and Dr. Gaurav Sinha for their valuable time, help and suggestions. I would like to thank Issam Khoury and David Beegle for their help on collecting and analyzing the pavement response data. Next, I would like to extend thanks to CE graduate students David Keatley, David Padilla-Llano, Jose Antonio, Aziz Ahmad Gulistani, Abdalla Alrawashdeh, Farid Momand and Hanxiao. Thanks for their help and accompany during my master s study. I also want to thank all professors and classmates who helped me during my study in Athens. Finally, I would like to thank the friends I made in Athens. Thank you for accompanying me through the good and bad times. I also want to express my deep gratitude to my parents who always loved, supported and encouraged me.

TABLE OF CONTENTS 5 Page Acknowledgments... 4 List of Tables... 8 List of Figures... 9 Chapter 1: Introduction... 12 1.1. Background... 12 1.2. Pavement Types and Design... 12 1.3. Objective of the Study... 14 1.4. Outline... 15 Chapter 2: Literature Review... 17 2.1. Introduction... 17 2.2. Flexible Pavements Structural Response Models... 19 2.2.1. Boussinesq s equations... 19 2.2.2. Two-layer Theory... 22 2.2.3. Multi-Layer Theory... 25 2.2.4. Multi-Layer Computer Programs... 27 2.2.5. Finite Element Method... 28 2.3. Summary... 29 Chapter 3: Falling Weight Deflectometer Testing and Backcalculation... 30 3.1. Introduction... 30 3.2. Falling Weight Deflectometer... 30 3.3. Backcalculation... 32

6 3.4. Backcalculation Programs... 34 3.4.1. Evercalc 5.0... 34 3.4.2. Modulus 6.0... 35 3.4.3 Elmod 6.0... 36 3.5. Summary... 37 Chapter 4: Experimental Site U.S. Route 30 Demonstration Project... 38 4.1. Site Description... 38 4.2. Introduction of Perpetual Pavement... 40 4.3. Instrumentation Plan of AC Sections... 42 Chapter 5: Predict Pavement Response Using FWD Deflections... 48 5.1. Introduction... 48 5.2. FWD Deflection and Pavement Response Data... 49 5.3. Backcalculation of Pavement Layer Moduli... 50 5.4. Forward Calculation... 61 5.5. Conclusion... 64 Chapter 6: Comparison Between FWD and Truck Loading Conditions... 65 6.1. Introduction... 65 6.2. Controlled Load Vehicle Testing... 65 6.3. Comparison of FWD and CLV test... 67 6.3.1. Magnitude of Longitudinal Tensile Strain... 67 6.3.2. Response time of Longitudinal Tensile Strain... 75 6.4. Summary... 76 Chapter 7: Conclusions and Recommendations... 78

7 7.1. Conclusions... 78 7.2. Recommendations... 79 References... 81 Appendix A: Deflection Data Report... 85 Appendix B: Strain Responses in FWD Tests and CLV tests... 88

LIST OF TABLES 8 Page Table 2.1 Boussinesq s equations for a concentrated load (Ullidta, 1998).... 21 Table 2.2 Multi-layer Computer Programs... 28 Table 4.1 Perpetual Pavement Built-Up (Liao, 2007)... 40 Table 4.2 Pavement Data Acquisition Instrumentation... 43 Table 5.1 Typical ranges of resilient modulus and Poisson s ratio.... 52 Table 5.2 Summary of backcalculated moduli from Evercalc 5.0.... 53 Table 5.3 Summary of backcalculated moduli from Elmod 6.0 FEM.... 54 Table 5.4 Summary of backcalculated moduli from Elmod 6.0 LET.... 55 Table 5.5 Summary of backcalculated moduli from Elmod 6.0 MET.... 56 Table 5.6 Predicted and measured strain values at the bottom of asphalt laye.... 63 Table 6.1 Maximum strains at the bottom of FRL layer under Single Axle Truck (Rear Axle)... 71 Table 6.2 Maximum strains at the bottom of FRL layer under Tandem Axle Truck (Front Axle)... 71 Table 6.3 Maximum strains at the bottom of FRL layer under FWD loading... 72 Table 6.4 Ratio of the strains at the bottom of FRL layer under FWD loadings to the maximum strains under controlled load vehicle loadings... 73 Table 6.5 Strains at the bottom of FRL layer caused by the front axle of single axle truck (5950lbs)... 74 Table 6.6 Strains at the bottom of FRL layer caused by the front axle of tandem axle truck (8550lbs).... 74

LIST OF FIGURES 9 Page Figure 2.1: Critical locations in flexible pavement (Muench, et al., 2003).... 18 Figure 2.2: Notation for Boussinesq s equations in polar coordinates (Ullidta, 1987).... 20 Figure 2.3: Burmister s two layer system (Burmister, 1943).... 23 Figure 2.4: Boundary and continuity conditions of Burmister s two layer system (Burmister, 1943).... 24 Figure 2.5: Notation for multi-layer elastic model in cylindrical coordinates (Huang, 2004).... 27 Figure 3.1: Dynatest Model 8000 Falling Weight Deflectometer (Dynatest).... 31 Figure 3.2: Schematic of FWD load and deflection measurement.... 32 Figure 3.3: Backcalculation Flowchart (Lytton, 1989).... 34 Figure 4.1: Plan view of the U.S. Route 30 perpetual pavement project (Source: Google map).... 39 Figure 4.2: Perpetual pavement design concept (HMA = hot-mix asphalt). (Newcomb, 2001)... 41 Figure 4.3: Instrumentation Plan of AC Section 876A.... 44 Figure 4.4: Instrumentation Plan of AC Section 876B.... 45 Figure 4.5: Dynatest PAST II AC strain gauges... 46 Figure 5.1: The process of predicting pavement response based on FWD deflections.... 49 Figure 5.2: Typical pavement surface deflection basins for different FWD load levels.. 50 Figure 5.3: Comparison of estimated moduli for AC Surface Layer from different backcalculation methods.... 57

10 Figure 5.4: Comparison of estimated moduli for AC Intermediate Layer from different backcalculation methods.... 58 Figure 5.5: Comparison of estimated moduli for AC Base Layer from different backcalculation methods.... 59 Figure 5.6: Comparison of estimated moduli for Aggregate Layer from different backcalculation methods.... 60 Figure 5.7: Comparison of estimated moduli for Subgrade from different backcalculation methods.... 61 Figure 6.1: Tire loads and axle characteristics of Single Axle truck for CLV test.... 66 Figure 6.2: Tire loads and axle characteristics of Tandem Axle truck for CLV test.... 66 Figure 6.3: Typical longitudinal tensile strain at the bottom of FRL layer caused by single axle truck in CLV tests.... 68 Figure 6.4: Typical longitudinal tensile strain at the bottom of FRL layer caused by tandem axle truck in CLV tests.... 69 Figure 6.5: Typical longitudinal tensile strain at the bottom of FRL layer in FWD tests. 69 Figure 6.6: Comparison of strain responses under FWD loading and truck loading at the same level of loads.... 75 Figure 6.7: Average duration of the impulses under different loading conditions.... 76 Figure B.1: Strain responses under FWD loading (Strain gauge 5).... 88 Figure B.2: Strain responses under FWD loading (Strain gauge 6).... 88 Figure B.3: Strain responses under FWD loading (Strain gauge 8).... 89 Figure B.4: Strain responses under FWD loading (Strain gauge 9).... 89 Figure B.5: Strain responses under FWD loading (Strain gauge 13).... 90

11 Figure B.6: Strain responses under FWD loading (Strain gauge 16).... 90 Figure B.7: Strain responses under Single-Axle truck loading (speed at 5 mph).... 91 Figure B.8: Strain responses under Single-Axle truck loading (speed at 25 mph).... 91 Figure B.9: Strain responses under Single-Axle truck loading (speed at 45 mph).... 92 Figure B.10: Strain responses under Single-Axle truck loading (speed at 55 mph).... 92 Figure B.11: Strain responses under Tandem-Axle truck loading (speed at 5 mph).... 93 Figure B.12: Strain responses under Tandem-Axle truck loading (speed at 25 mph).... 93 Figure B.13: Strain responses under Tandem-Axle truck loading (speed at 45 mph).... 94 Figure B.14: Strain responses under Tandem-Axle truck loading (speed at 55 mph).... 94

CHAPTER 1: INTRODUCTION 12 1.1. Background The highway system serves as an important factor to a country s economic development and defense. In United States, the national highway system includes approximately 162,000 miles of roadway, which carry more than seventy-five percent of heavy truck traffic and ninety percent of tourist traffic (AASHO, 2007). Due to the tremendous increases of the highway traffic volumes in the last twenty years, highway pavement failures occurred earlier than expected. Rehabilitation or reconstruction is important and necessary to maintain a good condition of highway system. Every year, the state and federal government spend billions of dollars on the construction and maintenance of highway pavements. During the period of 1995 to 2004, about 900 billion dollars were spent on the highway construction and maintenance (FHWA, 2004). With the aging of the highway infrastructure, a large portion of the highway system built during the 1950s and 1960s need major rehabilitation or reconstruction which will cost billions of dollars. Efforts continue to be made to improve pavement analysis and design methods will result in cost-effective improvement in pavement construction and rehabilitation (Tayabji, et al, 2000). 1.2. Pavement Types and Design Basically, pavements can be divided into three types: flexible pavements, rigid pavements and composite pavements. A flexible pavement typically consists of a hot-mix asphalt (HMA) wearing course, an intermediate asphalt course and one or more base and

13 subbase courses. It is called flexible pavement because the structure of pavement will flex under the load (Yoder et al., 1975). A rigid pavement, also known as Portland Cement Concrete (PCC) pavement is composed of a PCC slab surface, a base course directly under the PCC layer and a subbase course below the base layer. Since the PCC material has a high elastic modulus, this type of pavement is much stiffer than the flexible pavement. A composite pavement is basically a rigid pavement overlain with an asphalt concrete layer. Because of the high construction cost, composite pavements are rarely constructed initially (Huang, 2004). Most of composite pavements are rehabilitated pavements with asphalt overlays on PCC pavement (Huang, 2004). In United States, about 82.2 percent of paved roads are flexible pavements, 11.3 percent are composite pavements and only 6.5 percent of the paved roads are rigid pavements (Muench, et al., 2003). Since most of the paved roads are flexible pavements, this research will mainly focus on flexible pavements. The purpose of pavement design is to select a series of material layers to comprise a pavement structure which can carry an estimated volume of traffic for a specified design life. There are two principal pavement design methods: Empirical methods and mechanistic methods. The empirical methods are used in many pavement structural design procedures. They are developed based on empirical studies of pavement materials and structures. The 1993 AASHTO Guide for Design of Pavement Structures used an empirical approach and it is still widely used by most of the states. The mechanistic methods are based on physical principles. Pavement response models are the backbone of the mechanistic approach. By using pavement response models, the stresses, strains, and deflections in the pavement structure can be calculated.

14 The Mechanistic-Empirical design procedure is the most advanced pavement design method. It combines the advantages of both mechanistic approach and empirical approach. After calculate the stresses, strains and deflections using the mechanistic approach, engineers use empirical elements to determine what value of calculated stresses and strains can cause pavement failure. In 2004, the National Cooperative Highway Research Program (NVHRP) released a new pavement design guide, the Mechanistic- Empirical Design Guide (MEPDG) for New and Rehabilitated Pavement Structures. By using this guide, more reliable pavement structures can be designed by pavement engineers. 1.3. Objective of the Study In flexible pavements, fatigue cracking is one of the major distresses caused by repeated traffic loading. Extensive researches have been conducted on fatigue cracking and many models have been developed using empirical and mechanistic-empirical approaches to predict the fatigue failure for asphalt concrete. In most of these models, fatigue failure is correlated to the critical tensile strain at the bottom of HMA layer and the elastic modulus of the asphalt concrete layer. So engineers can predict fatigue failure in asphalt concrete layer using the horizontal tensile strain at the bottom of HMA layer. The value of the horizontal tensile strain can be determined using the pavement response models. The major objective of this study is following: 1. Conduct a review of available pavement structural response models and pavement analysis softwares.

15 2. Select suitable structural analysis softwares to predict the pavement response (horizontal tensile strain at the bottom of the asphalt layer) based on the deflection data from Falling Weight Deflectometer (FWD) Tests. Compare the predicted results with available measured data and check the validity and accuracy of the selected structural analysis programs 3. Compared the loading condition between the FWD loading and truck loading based on the test results from Controlled Load Vehicle (CLV) tests and FWD tests. 1.4. Outline The report is structured in the following format to effectively present information and data clearly. In Chapter 2, a briefly review of the major theoretical structural response models and structural analysis programs for flexible pavements is presented. The most commonly used structural response model is multi-layer elastic model. Chapter 3 presents a briefly introduction about the perpetual pavement and the U.S. 30 perpetual pavement project along with its design and instrumentation plans for asphalt concrete sections. Falling Weight Deflectometer deflection data and pavement response data collected from AC section of U.S 30 perpetual pavement were used in this research. Chapter 4 introduces Falling Weight Deflectometer (FWD) and the backcalculation method to calculate the elastic moduli of pavement layers using pavement surface deflections. Several commonly used backcalculation programs are also described in this chapter.

16 In Chapter 5, the process to predict the pavement response (tensile strain at the bottom of HMA layer) based on FWD deflections is introduced. Comparisons between the predicted and measured values of pavement response are also presented in this chapter. Chapter 6 compares the loading condition between FWD test and controlled load vehicle (CLV) test. The FWD tests and CLV tests were conducted at the same location on the same day. So the environment effects can be ignored. Finally, the conclusion and recommendations are presented in Chapter 7.

17 CHAPTER 2: LITERATURE REVIEW 2.1. Introduction In flexible pavement analysis, there are three elements should be considered: Theoretical structural response model, Material properties, External conditions (traffic loading, environment condition). Theoretical structural response models are developed to examine the response (deflection, strain, stress) of the pavement under the traffic loads based on a continuum mechanics approach. Responses at some critical locations are often used in pavement analysis. For instance, the horizontal tensile strain at the bottom of the asphalt layer can be used to predict the fatigue failure in the asphalt. The compression strain at the top of intermediate layer is used to predict rutting failure (Muench, et al., 2003). The critical locations in a pavement structure are shown in figure 2.1.

18 Figure 2.1: Critical locations in flexible pavement (Muench, et al., 2003). Material properties include three aspects: the strain and stress relationship of the material, the degree to recover strain after stress removal, and the time and temperature dependency of strain. Based on these three aspects, materials can be categorized into linear or nonlinear, elastic or plastic, viscous or nonviscous.

19 External conditions include pavement traffic loading and environmental conditions. They are two main distress mechanisms which affect the performance of pavement. In pavement structural analysis, temperature and moisture content are the most important environmental parameters. For example, thermal cracking is a result of temperature effect on pavement. It usually occurs at extremely low temperatures during the winter. Stripping is a result of the moisture damage. It is caused by the interaction between the moisture and asphalt binder-aggregate adhesion (Muench, et al., 2003). In this study, author will focus on the theoretical structural response models for pavement analysis. The environmental conditions will not be considered and the properties of the materials are assumed to be linear elastic. In the following sections, literature is reviewed on the available structural response models for flexible pavement. 2.2. Flexible Pavements Structural Response Models 2.2.1. Boussinesq s equations The first pavement response model was developed by J. Boussinesq (1885). He examined the pavement s response to a load and proposed a series of equations called Boussinesq s equations. These equations can be used to calculate stresses, strains, and deflections subjected to a concentrated load. In this model, Boussinesq assumed the pavement layer is a homogeneous, isotropic, linear elastic half space. Figure 2.2 presents the notation in polar coordinates for Boussinesq s equations. z is the depth and r is the horizontal distance between the load P and the point where the responses are desired. Table 2.1 lists some of the Boussinesq s equations for a point load P.

Figure 2.2: Notation for Boussinesq s equations in polar coordinates (Ullidta, 1987). 20

Table 2.1 Boussinesq s equations for a concentrated load (Ullidta, 1998). Normal Stresses 21 2 3 1 2 1 1 2 2 1 1 3 2 Normal Strains 1 2 3 3 2 1 2 1 1 1 2 2 1 1 2 3 2 Shear Stress 3 2 Shear Strain 1 1 2 Displacements 1 2 1 1 2 2 1 Boussinesq s signal layer model is probably the simplest model of a pavement structure. It is developed originally for a static concentrated load. Later, equations for a uniformly distributed load were derived by integration. Although the assumptions in this model seem to be hypothetical and unrealistic, many researches have shown that there is

22 a good correlation between computed deflections by Boussinesq s equations and the measured deflections (Yang, 1972). The biggest advantage of this model is its simplicity and it provides the basis for several pavement structural models which are currently being used. 2.2.2. Two-layer Theory A Typical flexible pavement is usually composed of several layers. Therefore, Boussineq s signal layer model cannot accurately simulate the flexible pavement structure, a better model is need for flexible pavement analysis. In 1943, Burmister (1943) developed solutions for a two layer flexible pavement. In his model, certain essential assumptions were made in order to compute the stresses, strains, and deflection. The hypotheses made by Burmister in two layer elastic theory include (Burmister 1943): The material in each layer is homogeneous, isotropic, and linear elastic with an elastic modulus E and Poisson s ratio. The lateral direction of the surface layer is infinite in extent while the vertical direction has a finite depth. Both the horizontal and vertical directions of the bottom layer are infinite in extent. All layers have a uniform thickness. The material at each layer is weightless. The layers are in continuous contact. Stresses and strains are continuous across boundaries. Shearing forces are not present in the surface. Loads applied on the pavement are represented by the uniformly distributed pressure.

23 The dynamic and thermal effects are not considered. Figure 2.3: Burmister s two layer system (Burmister, 1943).

24 Figure 2.4: Boundary and continuity conditions of Burmister s two layer system (Burmister, 1943).

25 In Burmister s theory, stresses and deflections depend on the modular ratio (E 1 /E 2 ), where E 1 is the modular of the surface layer and E 2 is the modular of the subgrade layer. Burmister (1958) developed a chart for computer surface deflections in a two-layer system. The equations to computer the deflections are following (Burmister, 1958): Deflection under a flexible plate: Deflection under a rigid plate: 1.5 E 1.18 E F F Where p is the united load on circular plate, a is the radius of plate, E 2 is the elastic modulus of lower layer, F 2 is the deflection factor depend on the values of E 1 /E 2 and a/h 1, h 1 is the thickness of the surface layer. 2.2.3. Multi-Layer Theory In order to build a better model for flexible pavements, Burmister (1945) extended his two layer theory to a three layer system. Later in 1951, Acum and Fox developed the exact solutions of normal and radial stresses in a three layer system based on Burmister s theory. The hypotheses in Burmister s theory were also used in their models. Later, Acum and Fox s solutions were extended by Jones (1962) and Peattie (1962) to a much wider range of solution parameters. Jones (1962) developed solutions for horizontal stresses in three-layer systems. Peattie (1962) presented graphical solutions for vertical stresses. A poisson s ratio of 0.5 was used for all layers in these researches. In the same year, Schiffman presented a solution of stresses and displacements in a multi-layer elastic system. This was considered to be a significant breakthrough in

26 flexible pavement analysis. In previous researches, the load types are limited to a point load or a uniformly distributed load, the responses due to the Non-uniform loads and tangential loads were not considered. Using Schiffman s theory, the stresses and displacement under different kinds of load such as non-uniform loads, tangential loads, rigid loads were able to be computed. Figure 2.5 shows the notation for Multi-layer elastic model in cylindrical coordinates. Each layer has its elastic modular (E), Poisson s ratio ( ) and thickness.

27 Figure 2.5: Notation for multi-layer elastic model in cylindrical coordinates (Huang, 2004). 2.2.4. Multi-Layer Computer Programs With the development of computer technology, many computer programs were developed for pavement analysis based on the multi-layer elastic theory. The available computer programs which can be used in pavement analysis and design include: BISAR, CHEVRON, KENLAYER, ELSYM5, Everstress and WESLEA. Typically these softwares are able to compute the stresses, strains, and deflections under a circular

28 surface loads. The inputs of these softwares include: material properties (modulus and poisson s ratio), layer thickness, and load conditions (magnitude of load, radius, or contact pressure). The outputs include stresses, strains, and deflections. Table 2.2 Multi-layer Computer Programs Programs Developer Description BISAR Shell Oil Co. Developed based on linear elastic theory. BISAR 3.0 can be used to calculate comprehensive stress and strain profiles, deflections, and slip between the pavement layers via a shear spring compliance at the interface. CHEVRON KENLAYER ELSYM5 Everstress WESLEA Chevron research company Yang H. Huang FHWA University of Washington U.S. Army Corps of Engineers Developed based on linear elastic theory. The program can accept more than 10 layers and up to 10 wheel loads. Developed based on Burmister s elastic layered theory. It can be used to compute the responses for maximum of 19 layers with an output of 190 points. Developed based on linear elastic theory. The program can analyze a pavement structure containing up to five layers, 20 multiple wheel loads. The program can be used to determine the stresses, strains, and deflections in a layered elastic system (semi-infinite) under circular surface loads. The program is able to analyze up to five layers, 20 loads and 50 evaluation points. The current version can analyze more than 10 layers with more than 10 loads. 2.2.5. Finite Element Method The finite element method is a numerical method which can be used for analysis of stress, strain and displacement in a pavement structure. ILLI-PAVE, developed by Raad and Fifueroa (1980), is a 2-D finite element program commonly used for analysis of

29 flexible pavements. The advantage of this 2-D finite element program is that it allows the use nonlinear constitutive relationships which can describe nonlinear elastic, visco-elastic, or plastic behavior. Another finite element program called 3D-Move was developed by Siddharthan et al. (1998) based on continuum mechanics. The advantage of finite element method is that it can evaluate the dynamic response of flexible pavements. Many other finite element programs (such as CAPA-3D, CESAR-LCPC, Mich-PAVE, and FeBack) were developed for analysis of flexible pavement. Due to the complicated nature of finite element method, the finite element programs are only suited for forward analysis of pavement structures. The backcalculation of elastic modulus based on surface deflections is unable to accomplish by using these programs. 2.3. Summary Currently, the multi-layer elastic theory is the most commonly used pavement structural model in pavement design and analysis. In this research, the multi-layer elastic theory was used both in the moduli backcalculation process and the forward calculation process. Detailed information about backcalculation and forward calculation will be introduced in Chapter 3.

30 CHAPTER 3: FALLING WEIGHT DEFLECTOMETER TESTING AND BACKCALCULATION 3.1. Introduction Pavement surface deflection measurement plays an indispensable role in evaluating a flexible pavement structure. It can be used to monitor the pavement performance, calculate the pavement layer moduli and the subgrade resilient modulus, and identify potential problem areas in the pavement. Many nondestructive deflection testing equipments are available for pavement engineers. These equipments can be divided into three categories: static deflections (Benkelman Beam), steady state deflections (Dynaflect and Road Rater), impact load deflections (Falling Weight Deflectometer). This chapter will introduce the Falling Weight Deflectometer (FWD) and the backcalculation method to calculate the elastic moduli of pavement layers based on surface deflections. 3.2. Falling Weight Deflectometer Falling Weight Deflectometer (FWD) is a nondestructive test device widely used in pavement engineering. It plays an important role in evaluating the physical properties and performance of a pavement. The main components of a FWD system include: control system, hydraulic system, loading weight and plate, load cell and deflection sensors.

31 Figure 3.1: Dynatest Model 8000 Falling Weight Deflectometer (Dynatest). During the test, the FWD applies a load to the pavement surface by dropping a large weight onto a load plate positioned on the pavement surface. This load simulates the magnitude and duration of a moving wheel load. The pavement response (surface deflection) due to the load is then measured by a series of deflection sensors mounted at various distances from the loading point (one sensor is located directly over the loading point). Usually, the deflections are measured at 0 inch, 8 inches, 12 inches, 18 inches, 24 inches, 36 inches and 60 inches away from the center of the loading plate. The measured deflections at each sensor called deflection basin. Figure 3.2 shows a schematic of FWD load and deflection measurement.

32 Figure 3.2: Schematic of FWD load and deflection measurement. The advantages of FWD test include: it is accurately simulated the traffic load, it is quicker (can test up to 60 points in an hour) and can be operated by one person. The loading range of a FWD varies from 1,500 to 27,000 lbf (Dynatest, 2009). 3.3. Backcalculation Backcalculation is the process of computing pavement layer moduli and the subgrade resilient modulus based on pavement deflection basins generated by Falling Weight Deflectometer (Muench, et al., 2003). In order to conduct a backcalculation, the initial moduli of pavement layers should be first assumed, the values are usually estimated base on engineer s experience or equations. After assuming the initial layer moduli, pavement surface deflections can be calculated using pavement response models. The calculated deflections are then compared to the measured values. By adjusting the pavement layer moduli, a good match (within some tolerable error) between the measured and theoretical deflections can be reached. The process of backcalculation is

33 usually iterative. Many programs were developed for backcalculation such as Modulus 6.0, Elmod 6.0, and Evercalc 5.0. Figure 3.3 presents a basic flowchart of backcalculation program. The main components in a backcalculation process include (Lytton, 1989): Layer thicknesses and loads: Thickness of each pavement layer and load levels applied on the pavement surface. Measured deflections: Surface deflections measured during FWD tests. Seed moduli: Initial modulus used to compute theoretical surface deflections. Deflection calculation: Use pavement response models to calculate theoretical surface deflections. Error check: Compare the calculated and measured deflections. Search for new moduli: Iteratively search for the new modui of pavement layers until the calculated and measured deflection are matched (within acceptable error). Controls on the range of moduli: The backcalculation programs usually can define a range of modulus for each pavement layer to prevent unreasonable pavement layer moduli.

34 Figure 3.3: Backcalculation Flowchart (Lytton, 1989). 3.4. Backcalculation Programs Evercalc 5.0, Elmod 6.0, and Modulus 6.0 are three most commonly used backcalculation programs. In this research, these three backcalculation programs were used to estimate the pavement layer moduli. A comparison of the results from Evercalc 5.0, Elmod 6.0, and Modulus 6.0 were conducted by author. Strains at the bottom of the AC layer can then be calculated based on the results of backcalculation. Software Everstress 5.0 was used to compute the strains based on the backcalculated layer moduli. 3.4.1. Evercalc 5.0 Evercalc 5.0 is a pavement analysis program developed by the University of Washington. It can be used to estimate the elastic moduli of pavement layers, and determine the stresses and strains at various locations. Evercalc 5.0 uses WESLEA

35 program (a multi-layer computer program developed by the U.S. Army Corps of Engineers) as a subroutine to calculate the theoretical deflections. It also uses an inverse solution technique to determine the set of layers moduli from FWD deflection data (Everseries User s Guide, 2005). Before running the program, the user can define the deflection tolerance, moduli tolerance and the maximum number of iterations. When one of the conditions is satisfied, the program will terminate. In Evercalc 5.0, deflection tolerance is given by: RMS (%) = 100% Where, Root Mean Square (RMS) is the primary measure of convergence used for error check. n is the number of deflection sensors used in FWD test, d ci is the calculated pavement deflection at sensor i, and d mi is the measured pavement deflection at sensor i. Generally a RMS of 1to 2 percent would be acceptable. Moduli Tolerance is expressed by the following equation: ε E E 100 E Where, E and E are respectively the i-th layer moduli at the (K+1)-th and K-th iteration (Everseries User s Guide, 2005). 3.4.2. Modulus 6.0 Modulus 6.0 was developed by Texas Transportation Institute. It is the newest version of the Modulus program and can be used to process Falling Weight Deflectometer data and flexible pavement design. Modulus 6.0 is based on the multilayer linear elastic theory. It uses WESLEA (developed by the U.S. Army Corps of Engineers) as a subroutine for forward calculation (William, 1999). With assumed seed

36 layer moduli, the program uses WESLEA to calculate a deflection basin. The calculated basin is then compared with the measured basin. After several iterations, a set of layer moduli that produce an acceptable error between the calculated and measured basins can be determined. The Modulus 6.0 program is able to analyze up to four unknown layers, 7 deflection sensors. It also has a database which can assign the modulus range and Poisson s ratio by selecting the material type (William, 1999). 3.4.3 Elmod 6.0 Elmod 6.0 was developed by Dynatest International A/S. It is used to evaluate the pavement layer moduli and overlay design based on FWD deflection data. There are three backcalculation options available in this program: Linear Elastic Theory (LET), Finite Element Method (FEM), and Method of Equivalent Thickness (MET). Basically, different forward analysis methods are used in these three options. The LET method uses WESLEA for forward analysis, the FEM uses an axial symmetric finite element program to calculate the theoretical deflections, while the MET makes use of method of equivalent thickness with improved adjustment factors. The Elmod 6.0 program can directly read the Dynatesr-FWD files. By selecting the analysis option, the program is able to automatic fit the calculated and measured deflection basins either for all points or point by point in the FWD files (Elmod 6 Quick Start Manual, 2009). For Elmod FEM option, the program treats all pavement layers as non-linear elastic. This may take a longer processing time. For Elmod MET option, the program treats subgrade as non-linear. (Elmod 6 Quick Start Manual, 2009).

3.5. Summary 37 The pavement surface deflections from the FWD tests are used to predict the pavement response (tensile strain at the bottom of the HMA layer). By using the backcalculation method, the pavement layer moduli can be calculated base on FWD deflections, further, the pavement response can be calculation using the backcalculated pavement moduli.

38 CHAPTER 4: EXPERIMENTAL SITE U.S. ROUTE 30 DEMONSTRATION PROJECT 4.1. Site Description In 2002, the Ohio Department of Transportation (ODOT) and Flexible Pavement of Ohio decided to build a demonstration project for perpetual pavements. They chose the section of U.S Route 30 in Wayne County, the Wooster Bypass. The project is a four-lane divided rural freeway, begins on the west by State Route 83, and extends on the east by Kansas Road near State Route 57. The asphalt perpetual pavement was constructed in the westbound lanes and the long lasting Portland cement pavement was built in the eastbound. The total length of this perpetual pavement is approximately 8 miles. The road was open to traffic in December, 2005. Figure 4.1 displays the plan view of the U.S. Route 30 perpetual pavement project from Google map.

39 Figure 4.1: Plan view of the U.S. Route 30 perpetual pavement project (Source: Google map). The design of this perpetual pavement was based on the mechanistic analyses conducted by a research team led by Dr. Sang-Soo Kim (Ohio Asphalt, 2004). These analyses were performed using a larger design load (1.2 times of the legal load), with limiting the strain less than 70 microstrains at the bottom of the HMA layer. Due to these limits, the thickness of the HMA layer was determined using the layer elastic analysis and a HMA thickness of 16.25 inches was used in this project. The HMA layer is composed of four courses: 1.5 inches wearing course, 1.75 inches intermediate course, 9 inches asphalt base and 4 inches fatigue resistance course. The dimensions and materials for each layer are listed below in Table 4.1.

Table 4.1 Perpetual Pavement Built-Up (Liao, 2007) Course Thickness ODOT (inches) Item No. Description ODOT 12.5 mm stone mastic asphalt with Surface 1.5 856 a PG 76-22M polymer modified binder. 93% - 97% target density. Intermediate 1.75 442 ODOT 19 mm Superpave, Type A, with a PG 76-22M polymer modified binder. 93% - 97% target density. 40 Asphalt Base 9 302 ODOT s large stone mix, PG 64-22 asphalt binder. 93% - 96% target density. Fatigue resistant layer 4 Special 302 ODOT s large stone mix, PG 64-22 asphalt binder with 3% air void, 94% - 97% target density Aggregate Base 6 304 Highly crushed densely graded granular base with under drain 4.2. Introduction of Perpetual Pavement A perpetual pavement, or long-lasting asphalt pavement, is defined by the Asphalt Pavement Alliance (APA) as a hot mix asphalt pavement designed and constructed to last longer than 50 years without requiring major structural rehabilitation or reconstruction. It only needs periodic surface renewal in response to distresses confined to the top of the pavement (APA, 2002). The concept of perpetual pavement is not new. Actually, the design and construction of long-lasting hot-mix asphalt pavement has been in progress since the 1960s (APA, 2002).

41 A perpetual pavement usually consists of three layers of asphalt with different mix formats and a strong foundation to produce a safe, smooth, and long-lasting road. Figure 3.1 shows a design concept of a perpetual pavement. The design begins with a strong and stable foundation at the bottom of the pavement to preclude distresses (APA, 2002). The hot-mix asphalt base layer (bottom layer) is designed to resist fatigue cracking. The strong intermediate layer is designed to carry most of the traffic load, and the wearing surface is the top layer designed specifically to resist top-down cracking and rutting (TRB, 2001). The surface layer is intended to be periodically overlaid with more hot-mix asphalt to restore condition. Ideally, with scheduled surface restoration, perpetual pavements can be maintained and cost-effectively without removing the road structure for reconstruction (Kuennen, 2004). Figure 4.2: Perpetual pavement design concept (HMA = hot-mix asphalt). (Newcomb, 2001)

42 4.3. Instrumentation Plan of AC Sections Ohio University was granted three associated research projects to evaluate the performance of the perpetual pavement. The comprehensive instrumentation plans were developed by Ohio Research Institute for Transportation and the Environment (ORITE). Three test sections were constructed with instrumentation. One test section constructed at Station 664+00 was named as Section 664, the other two test sections constructed at Station 876+60 were named as Section 876A and Section 876B. Falling Weight Deflectometer (FWD) Testing and Controlled Load Vehicle (CLV) Testing was performed periodically by research teams from Ohio University and ODOT. Data acquisition instruments such as strain gauges, pressure cells, linear variable differential transformers (LVDTs) and thermocouples were installed in different pavement layers in section 664, section 876A and 876B during the construction. Table 3.2 shows the pavement data acquisition instruments used in this project. These instruments are used to measure pavement loads, strains, deflections, monitor the environmental parameters include temperature, moisture, frost depth, and ground water table levels. A self weather station was also built to monitor air temperature, wind speed and direction, relative humidity and solar radiation. In this research, the author only used the pavement response data collected at section 876A and section 876B. The detailed instrumentation plans for these two sections are shown in Figure 4.3 and 4.4.

Table 4.2 Pavement Data Acquisition Instrumentation 43 Measurement Sensor Manufacturer Displacement GDP 121-500 LVDT GDP 121-250 LVDT Lucas Schaevita Inc. Strain Dynatest PAST II - AC SG Dynatest Consulting Inc. Pressure Geokon 3500 PC Geokon Inc. Temperature MRC Thermistor Measurement Corporation Research Moisture FHWA TDR Probe Cambell Scientific Inc. Frost Depth CRREL Resistivity Probe Cold Regions Research & Engineering Laboratory Groundwater Table Piezometers -----

Figure 4.3: Instrumentation Plan of AC Section 876A. 44

45 Figure 4.4: Instrumentation Plan of AC Section 876B. As shown in Figure 4.3 and 4.4, strain gauges were placed at the bottom of the top layer, AC base layer and FRL layer. Dynatest PAST II AC strain gauges were used in this project. This type of strain gauge is designed for the measurement of strains in asphalt concrete pavements. It is an H shaped precision transducer and can measure either longitudinal or transverse strains in asphalt pavements (Dynatest, 2009). Strain

46 gauges in the bottom FRL layer are used to monitor fatigue resistance (Muench, et al., 2003). Only longitudinal strain was measured at the bottom of the FRL layer. Strain gauges are very delicate and vulnerable. Some strain gauges failed during compaction of an HMA layer. Figure 4.5: Dynatest PAST II AC strain gauges Two pressure cells were installed per test section on the top of the subgrade. These pressure cells measure the vertical stress under the dynamic loading which are can be utilized to evaluate the potential of rutting (Muench, et al., 2003). Linear variable differential transformers (LVDTs) are used to measure the displacement of the pavement. Four LVDTs were installed in each test section. As shown in Figure 4.3 and 4.4, two shallow referenced LVDTs measure the displacement above the subgrade while the other two deep referenced LVDTs measure the total displacement of the pavement.

47 Controlled Load Vehicle (CLV) Testing and Falling Weight Deflectometer (FWD) Testing are performed periodically by research teams from Ohio University and ODOT. The CLV testing is used to record the structural response of the pavement under controlled truck loading. The FWD testing is designed to apply a dynamic loading to the pavement surface that simulates the load of a signal moving wheel load. Based on the data from these tests, engineers can calculate the layer stiffness of the pavement and evaluate the physical properties of the pavement.

48 CHAPTER 5: PREDICT PAVEMENT RESPONSE USING FWD DEFLECTIONS 5.1. Introduction As introduced in Chapter 4, the Falling Weight Deflectometer (FWD) is a nondestructive test device which has been widely used to evaluate the physical properties and performance of a pavement. In U.S. 30 perpetual pavement project, the FWD tests were conducted during construction in order to detect weak areas and to evaluate construction quality (Liao, 2007). After construction, the FWD tests were also performed by Ohio University and ODOT periodically to assess the performance of the pavement (Sargand, 2008). This chapter is dedicated to predict pavement response based on the FWD deflection data. The process of predicting pavement response is shown in Figure 5.1. Backcalculation process was used in this research in order to calculate the pavement layer moduli. Thus, the pavement response can be predicted using backcalculated pavement layer moduli based on the pavement structural response models. The predicted strain values were compared to the measured values from U.S. Route 30 to check the validity and accuracy of this model.

49 FWD Deflection Data Backcalculation Process Pavement Layer Moduli Forward Calculation Pavement Response Figure 5.1: The process of predicting pavement response based on FWD deflections. 5.2. FWD Deflection and Pavement Response Data The Falling Weight Deflectometer (FWD) tests were conducted at section 876A and 876B on July 18, 2006. Three levels of load were used during the FWD tests. They are 6000lb, 9000lb, and 12000lb. The pavement surface deflection data were collected by FWD. Typical surface deflections under different levels of FWD load are displayed in Figure 5.2. The detailed FWD load and deflection data are listed in Appendix A.

50 Figure 5.2: Typical pavement surface deflection basins for different FWD load levels. Pavement responses (strains and deflections in the pavement layer) corresponding to different levels of load (6000lb, 9000lb, and 12000lb) during the FWD test were recorded by strain gages and LVDTs. Pressure cells were installed on the top of the subgrade to measure the vertical stress, however, these pressure cells were not working during the FWD tests and controlled load vehicle (CLV) tests. 5.3. Backcalculation of Pavement Layer Moduli In this research, Evercalc 5.0 and Elmod 6.0 were selected for backcalculation process. As introduced in Chapter 4, Evercalc 5.0 and Elmod 6.0 are two commonly used backcalculation programs which can analyze up to five unknown layers. Evercalc 5.0 program is based on multi-layer elastic forward calculation subroutines, while Elmod 6.0

51 has three backcalculation options based on three different structural response models. These three options include: Linear Elastic Theory (LET), Finite Element Method (FEM), and Method of Equivalent Thickness (MET). So a total of four methods were used in the backcalculation process. These four methods are designated as Evercalc, Elmod LET, Elmod FEM and Elmod MET. Since Evercalc 5.0 and Elmod 6.0 can only analyze up to five unknown layers, a five-layer pavement model was used in backcalculation process. This five-layer backcalculation model was composed of 1.5 inches surface layer, 1.75 inches intermediate layer, 13 inches asphalt base layer, 6 inches aggregate base layer and subgrade layer. The detailed information of each layer including layer material type, poisson s ratio and modulus range are shown in Table 5.1. Poisson s ratio and modulus range for different pavement layers were selected based on previous researches and past testing of similar materials. Initial modulus for each layer was estimated according to the material properties and the layer temperature during the FWD test.

Table 5.1 Typical ranges of resilient modulus and Poisson s ratio. Layer Type Material Type Thickness Poisson s Modulus Range (inches) Ratio (ksi) Surface Asphalt Concrete 1.5 0.35 200 2500 Intermediate Asphalt Concrete 1.75 0.35 200 2500 Asphalt Base Asphalt Concrete 13 0.35 200 2500 52 Sub-Base Aggregate 6 0.35 10-100 Subgrade Soil - 0.4 3-30 A total of 24 locations were tested for section 876A and 876B. Resilient moduli of pavement layers for each test location in section 876A and 876B were backcalculated using Evercalc 5.0, Elmod FEM, Elmod LET and Elmod MET. The estimated moduli for each location were averaged and summarized in Table 5.2, 5.3, 5.4 and 5.5.

Table 5.2 Summary of backcalculated moduli from Evercalc 5.0. Estimated Resilient Modulus (ksi) Station AC AC Asphalt Number Aggregate Subgrade Surface Intermediate Base RMS Error (%) 53 1 1153.4 269.4 544.8 3 86.7 1.13 2 622.9 316.1 673.3 3 84.8 0.8 3 270.2 382.7 1005.7 3 86.8 0.47 4 262.6 408.7 837.6 3.6 73.9 0.46 5 270.5 486.4 662.6 4.2 68.7 0.47 6 260.2 407.6 982.2 3.3 81.7 0.53 7 303.1 379.2 942.3 3.2 87.5 0.34 8 276.1 445.2 639.6 4.5 69.8 0.47 9 268.5 427.9 939.4 3.8 74.4 0.66 10 332.6 450.3 998.9 3.1 88.3 0.46 11 281.9 431.5 1098.2 3.2 86.3 0.58 12 240.6 500.3 905.4 3.4 86.1 0.34 13 412.3 392.6 888.9 3.5 84.8 0.38 14 388.3 456.2 1095.5 3 88.6 0.54 15 381.3 417.5 840.4 3.7 70 0.54 16 329.7 467.7 797.6 3.6 73.6 0.31 17 276.9 454.8 871.2 4 70 0.59 18 562.8 354.5 1006.2 3.1 75.9 0.23 19 246.3 386.5 1100 4.8 58.4 0.6 20 305.6 383.7 1099.3 5.1 52 0.35 21 224.1 559 897 4.4 48.8 0.66 22 230.2 723.9 685.9 3 56.3 0.9 23 210.9 582.3 789.6 3 61 0.64 24 314.8 429.7 903.1 3 64.9 0.58

Table 5.3 Summary of backcalculated moduli from Elmod 6.0 FEM. 54 Station Number AC Surface Estimated Resilient Modulus (ksi) AC Asphalt Base Aggregate Subgrade Intermediate 1 658.7 336.2 466.2 4 51.1 2 513.5 347.4 565.3 4.5 46.2 3 234.5 479.8 675.4 4 50.1 4 246.9 420.4 786.8 4.3 48 5 326.5 382.2 880.4 3.9 51.9 6 234.5 431.9 805 4.1 50.1 7 293.8 365.6 846.9 4.6 46.2 8 305.8 348.4 858.3 4.7 46.3 9 239.9 444.8 844.5 4.1 50.7 10 271.2 527.4 788.3 4.2 50.7 11 310.2 386.8 1003.1 4.4 47.8 12 263.7 444.8 929 3.7 57.6 13 475.5 359.2 863.3 4.1 55.5 14 389.9 436.2 1064.6 3.8 54.5 15 336.9 469.4 743.4 3.7 50.6 16 309.3 459.9 747.4 4.5 48.7 17 278.3 404.1 1035 3.9 54.2 18 442.3 404.4 826.2 3.9 49.2 19 265.3 344.9 1159.5 4.8 43.8 20 366.2 339.2 1226.7 4 46.8 21 221 506.5 1089.5 3.6 40.7 22 239.9 538.2 852.9 3 41.9 23 254.1 492.7 814.9 3.4 42.2 24 227.9 513.8 786.8 3.5 43.5

Table 5.4 Summary of backcalculated moduli from Elmod 6.0 LET. 55 Station Estimated Resilient Modulus (ksi) Number AC Asphalt AC Intermediate Surface Base Aggregate Subgrade 1 719.5 327.5 407.9 4.1 65.8 2 442.3 360.3 554.1 4.5 61.2 3 238 504.6 615.8 4.4 61.6 4 267.7 392.2 810.3 4.4 60.2 5 327.7 413.3 718.6 4.2 64.2 6 263.1 434.8 702.1 4.1 67.4 7 283 396.3 718.3 4.8 61.2 8 283 396.3 718.3 4.8 61.2 9 314.5 391.3 829.5 4.1 67.1 10 253.2 570 714.2 4.2 67.4 11 289.7 421.3 929 4.4 64.4 12 276.5 421.3 931.4 5 61.4 13 447.4 382.9 773.3 4.4 66.3 14 303.2 508.5 917.2 3.9 69.2 15 340.4 508.5 649.9 3.9 64.7 16 292.6 476.2 737.8 4.4 63 17 264.6 435.9 894.3 4.2 63.7 18 359.1 478.2 741.2 4.2 62.2 19 225.3 405.7 1175.4 4.2 57.9 20 320.1 365.6 1039.9 5.6 49 21 251.9 452.3 1335.7 3.2 54.2 22 247.6 587.3 792.3 3.3 51.7 23 219.9 548.6 784.4 3.5 52.3 24 239.2 459.9 955.9 3.4 56.5