Evaluating the Cracking Predicted by the MEPDG using Results from the S.R. 22 Smart Pavement Study

Transcription

1 Evaluating the Cracking Predicted by the MEPDG using Results from the S.R. 22 Smart Pavement Study Final Report Prepared by: Rania E. Asbahan Jennifer K. McCracken Julie M. Vandenbossche University of Pittsburgh Department of Civil and Environmental Engineering Pittsburgh, Pennsylvania Prepared for: Pennsylvania Department of Transportation, April 2008

2 ACKNOWLEDGEMENTS The success of this project would not have been possible without the assistance and expertise of many individuals. First, the authors gratefully acknowledge the financial and technical support provided by the Pennsylvania Department of Transportation (PennDOT). Specifically, the authors would like to thank Ms. Michelle Tarquino of Central Office for her assistance. The authors would also like to extend a sincere appreciation to the PennDOT District 12 Office. The effort and support provided by Mr. Joseph Szczur and Mr. Gary Barber of District 12 are especially appreciated. Finally, the authors would also like to extend their sincere gratitude to Mr. Michael Dufalla (formerly of the PennDOT District 12) who provided vision that was critical in the development of the research objectives and persistence that was essential in turning this research idea into a funded project. The contents of this report reflect the views of the author who is responsible for the facts and accuracy of the data presented herein. The contents do not necessarily reflect the views or policies of the Pennsylvania Department of Transportation.

3 TABLE OF CONTENTS Page Number List of Figures... v List of Tables... vi CHAPTER 1: INTRODUCTION Primary Goals of the Smart Pavement Project Goals Completed in Phase I Goals Completed in Phase II Primary Goals of Contract /WO Project Location and Site Description Pavement Structure and Design Details Layout of Test Sections... 4 CHAPTER 2: STRESS IN THE SMART PAVEMENT Introduction Dynamic Sensor Locations Axle Loads and Configurations used for the Field Testing Stress Corresponding to Measured and Predicted Strains Stress along the Transverse Joint (Group 1) Stress at Midpanel (Group 3) Effect of Environmental and Applied Loads on Stress in the Smart Pavement Effect of Slab Gradient on Stress in the Smart Pavement Effect of Load Magnitude on Stress in the Smart Pavement Combined Effect of Slab Gradient and Applied Load on Stress in the Smart Pavement CHAPTER 3: EVALUTION OF THE FATIGUE CRACKING MODEL OF THE MEPDG i

4 3.1.0 Introduction MEPDG Fatigue Cracking Model MEPDG Inputs General Inputs Environmental and Climatic Inputs PCC Material Properties Inputs ATPB Material Properties Inputs Granular Material and Subgrade Properties Inputs Traffic Inputs Fatigue Damage and Slab Cracking Results CHAPTER 4: CONCLUSIONS AND RECOMMENDATIONS References ii

5 Figures Figure Title Page Number Figure 1.1. Design thicknesses of the pavement layers [1; 2; 3] Figure 1.2. Layout of the Smart Pavement section [1; 2; 3] Figure 2.1. Sensor layout for Cell 1 and Cell Figure 2.2. Typical dimensions of dynamic strain gages and dynamic pressure cell Figure 2.3. Axle configuration and tire spacing for the Class 6 truck Figure 2.4. Axle configuration and tire spacing of the Class 7 truck Figure 2.5. Axle configuration and tire spacing of the Class 10 truck Figure 2.6. Loading conditions evaluated in the stress analysis for the restrained and unrestrained slabs Figure 2.7. Axle configuration and tire spacing for the single axle Figure 2.8. Effect of environmental and loading conditions on stress in the restrained slabs Figure 2.9. Effect of environmental and loading conditions on stress in the unrestrained slabs Figure 3.1. PCC strength characteristics over the 20-year design life Figure 3.2. Axle configuration and load for modified Class 6 truck Figure 3.3. Traffic and number of load application predicted over the 20-year design life Figure 3.4. Slab fatigue damage over the 20-year design life Figure 3.5. Slab cracking over the 20-year design life Figure 3.6. Critical stress causing 50 percent slab cracking iii

6 Tables Table Page Number Table 1.1. Summary of sensors installed in the Smart Pavement Project [1; 2; 3]... 4 Table 2.1 Stress for the gage along the transverse joint (Group 1 sensors) during truck testing Table 2.2. Stress at midpanel (Group 3 sensors) during truck testing Table 2.3. Temperatures in the slab and ATPB during the maximum positive and negative temperature gradients Table 2.4. Effect of slab gradient on critical tensile stress in the restrained slab Table 2.5. Effect of slab gradient on critical tensile stress in the unrestrained slab Table 2.6. Effect of load magnitude on critical tensile stress in the restrained slabs Table 2.7. Effect of load magnitude on critical tensile stress in the unrestrained slabs Table 2.8. Effect of gradient and load magnitude on critical tensile stress in the restrained slabs Table 2.9. Effect of gradient and load magnitude on critical tensile stress in the unrestrained slabs Table 3.1. General MEPDG design inputs Table 3.2. General structure inputs Table 3.3. PCC and ATPB temperatures at 2:00 PM during the month of May Table 3.4. General drainage inputs Table 3.5. Mixture properties for the MEPDG Table 3.6. PCC strength characteristics required for Level Table 3.7. General and thermal PCC properties Table 3.8. Shrinkage-related PCC properties Table 3.9. General asphalt inputs for SR Table Asphalt mix and binder characteristics Table Granular material and subgrade properties Table General traffic inputs Table General traffic inputs Table Base modulus and dynamic k-value estimated by the guide Table Base modulus and k-value used for the stress analysis Table Concrete stress based on the different traffic loading configurations and base material parameters Table Fatigue damage and cracking at the end of the 20-year design life iv

7 CHAPTER 1: INTRODUCTION Primary Goals of the Smart Pavement Project The Smart Pavement research project is aimed at the design and construction of more cost effective concrete pavements. There are four primary objectives for this research: 1. Evaluate the ability of High Performance Paving (HIPERPAV) software to predict strength gain and early-age stress development. 2. Characterize the seasonal response of a Jointed Plain Concrete Pavement (JPCP) to environmental and applied loads. 3. Establish inputs for a pavement constructed in Pennsylvania to use in the Mechanistic-Empirical Pavement Design Guide (MEPDG). 4. Evaluate stress in the pavement by developing and validating/calibrating finite element models using field measurements. The approach taken to accomplish these objectives was to construct an instrumented JPCP section, perform laboratory testing to characterize the material properties of the paving concrete and finally perform seasonal load testing and surface profile measurements on the instrumented pavement. The following sections present a summary of the tasks completed under Phase I and Phase II of the Smart Pavement Project. This report presents the results of the work completed under Contract /WO-003, which includes characterizing the stress in Smart Pavement using the validated finite elements developed under Phase II [1] Goals Completed under Phase I The project consists of two phases. Phase I involved the instrumentation of the Smart Pavement, the evaluation of the early-age (first 28 days) concrete material properties, the evaluation of HIPERPAV and an analysis of the early-age (first 28 days) pavement response characteristics. A summary of these findings can be found in the Phase I report submitted in October 2005 [2] Goals Completed in Phase II The second phase involves characterizing the design inputs for the MEPDG, characterizing longer-term trends in the response of the slab to environmental and applied 1

8 loads, and the development of finite element models to estimate pavement performance. A Phase II Interim Report was published in November 2006 summarizing the results from the load testing and surface profile measurements for the first year after the pavement was constructed as well as the one-year material properties of the concrete [3]. The Phase II Final Report was published in February 2008 summarizing the results from the load testing and surface profile measurements for the first three years after construction as well as the design inputs for the MEPDG and validation of the finite element models [1] Primary Goals of Contract /WO-003 Contract /WO-003 tasks include: 1. Evaluate stress in the Smart Pavement during truck testing using the validated finite element models. 2. Evaluate the effect of environmental conditions and applied loads on stress in the pavement. 3. Evaluate the accuracy of calculated stress in the structural models of the MEPDG. A brief section is first provided that describes the project location, site description pavement cross-section and design details. A general overview of the location of the dynamic, environmental and static sensors embedded in the pavement is also included. Only a brief description of the test section is provided below. A more detailed description can be found in the Phase I Report [2] Project Location and Site Description The location of the Smart Pavement is a 3.4 mile section of U.S. Route 22, along construction Section B01. The majority of this section runs through the municipality of Murrysville in Westmoreland County. Murrysville is located approximately 20 miles east of Pittsburgh. The test section consists of 14 Portland cement concrete (PCC) slabs, which are located in the westbound truck lane of US Route 22 between Tarr Hollow Road and School Road. The test section is located in front of a shopping plaza (Franklin Plaza) on the westbound side of the roadway and a manufacturing facility (Cleveland Brothers Machinery Company) on the eastbound side. 2

9 1.3.0 Pavement Structure and Design Details The pavement is a jointed plain concrete pavement (JPCP) with 15-ft transverse joints and 12-ft wide lanes. This section of roadway is crowned with a 2.0 percent transverse slope. The longitudinal slope along the research section is approximately 2.4 percent. The concrete medians vary in width from 14.4 ft to 2.0 ft with concrete mountable curbs. The Smart Pavement section contains 2.6-ft wide concrete curb-and-gutter shoulders at an 8 percent transverse slope adjacent to the outside lane. The pavement structure is composed of a 12-in thick PCC layer placed over a 4-in thick asphalt treated permeable base. The subbase material consists of slag material and is 5- in thick. Originally, the pavement was to be constructed directly on the subgrade but poor soil conditions required the removal of 24 in of the subgrade material. The 24 in was replaced with a gap-graded soil and aggregate mixture [2]. The cross section of the pavement structure is shown in Figure in 4 in 5 in 24 in Portland Cement Concrete Asphalt Treated Permeable Base PENNDOT 2A Subbase Cut and Fill Subgrade Figure 1.1. Design thicknesses of the pavement layers [1; 2; 3]. The test section consists of both restrained and unrestrained slabs. The restrained slabs contain No. 5 epoxy-coated tie bars placed 2.5 ft apart along the lane/shoulder and centerline joints. Epoxy coated 1.5-in dowel bars were spaced every 12 in along transverse joints. The unrestrained slabs do not contain either dowel or tie bars. 3

10 1.4.0 Layout of Test Sections Nearly 400 sensors were installed at various depths throughout the pavement structure. The sensors were installed in groups of slabs known as cells. There are a total of four cells consisting of three slabs each. The cells are labeled 1 through 4, with numbers increasing in the westward direction. Cells 1 and 2 contain sensors for measuring dynamic strains and pressures and Cells 3 and 4 measure both static strains and environmental conditions. The dynamic strain sensors in Cell 1 are a replicate of the dynamic strain sensor layout in cell 2. The same is true for Cells 3 and 4 with the exception that Cell 4 also contains environmental sensors. Figure 1.2 presents the layout of the Smart Pavement section. Although the sensor arrangements in these two sets of cells are repetitive, there is one unique factor that separates them. Cells 2 and 3 are unrestrained by dowel and tie bars while Cells 1 and 4 contain dowels and tie bars. A summary of the types and quantities of the dynamic sensors installed in Cells 1 and 2 and environmental sensors installed in Cell 4 is presented in Table 1.1. Additional information on the instrumentation used for the Smart Pavement project can be obtained from the Phase I Report [2]. Table 1.1. Summary of sensors installed in the Smart Pavement Project [1; 2; 3]. Sensor Type Sensor Name Qty. Measurement Cell Environmental Thermocouple 60 Temperature 4 Environmental Moisture Sensor 24 Relative Humidity 4 Environmental Time Domain Reflectometer 16 Moisture Content 4 Static Load Vibrating Wire Strain Gage 156 Static Strain 3, 4 Static Load Static Pressure Cell 8 Static Pressure 3, 4 Dynamic Load Dynamic Strain Gage 112 Dynamic Strain 1, 2 Dynamic Load Dynamic Pressure Cell 8 Dynamic Pressure 1, 2 Each cell has its own datalogging equipment that collects data from the sensors in the cell. Data from the dynamic sensors is collected manually at the time of dynamic loading. Data from the environmental and static sensors in Cell 3 and 4 are collected automatically. The datalogger in Cells 3 and 4 automatically retrieves data every 15 minutes. Once per day, the data on the datalogger is sent via telephone modems to a database located on a computer housed at the University of Pittsburgh. 4

11 = 210 = 64 m CELL 1 CELL 2 CELL 3 CELL 4 Westbound Traffic Slabs with Dynamic Strain Sensors Cell 1: 3 Restrained Slabs Cell 2: 3 Unrestrained Slabs Slabs with Static Strain and Environmental Sensors Cell 3: 3 Unrestrained Slabs Cell 4: 1 Intermediate Slab Cell 4: 3 Restrained Slabs Datalogger Enclosure with Remote Communications System Datalogger Enclosure without Remote Communications System Power Supply Phone Supply Power Cable Phone Cable Sensor Lead Wires Coaxial/Fiber Optic Cable Figure 1.2. Layout of the Smart Pavement section [1, 2; 3]. 5

12 CHAPTER 2: STRESSES IN THE SMART PAVEMENT Introduction This chapter presents the evaluation of stresses in the pavement using the finite element models developed and presented in the Phase II Final Report [1]. These models were used to: 1. Evaluate stress induced during truck testing 2. Evaluate stress in the pavement for a wider range of temperature, moisture and support conditions and a wider range of axle loads than those represented during testing. This chapter begins with an evaluation of stress corresponding to the measured and calculated strains presented in chapter 7 of the Phase II Final Report [1]. A comparison between the strains calculated using the finite element models and the measured strains was made to validate the finite element models. This validation procedure was performed using data collected when the gradient of the slab was approximately zero so stresses produced purely by the applied loads could be isolated from the residual stresses that result as a function of environmental loads. These calculated stresses, along with the corresponding measured and calculated strains are presented below. Next, the finite element models are used to evaluate stress in the pavement for a wider range of environmental and loading conditions than what was experienced during the data collection outings. The effects of slab gradient and load magnitude on pavement response will be evaluated independently and then the combined effect of both parameters will be considered Dynamic Sensor Layout Figure 2.1 outlines the locations of the dynamic strain gages and dynamic pressure cells in Cells 1 and 2. Longitudinally oriented gages are located in the wheelpath at the center of the slab (Group 3) and in the slab corner along the edge (Group 2). The dynamic strain gages oriented in the transverse direction measure strains in the wheelpath near the transverse joints (Group 1). As shown in Figure 2.1, the sensor layout for the unrestrained cell (Cell 2) is almost identical to that of the restrained cell (Cell 1). Figure 2.2 shows the 6

13 Instrumented Restrained Panels (Dowel and Tie Bars) Slab A Slab B Slab C Slab D CE07, CE08 CE05, CE06 CE03, CE04 CE01, CE02 CE09, CE10 CE11, CE12 CE13 CE14 CE15, CE16 CE25, CE26 CE23, CE24 CE21, CE22 CE19, CE20 CE17, CE18 CE33, CE34 CE51, CE56 CE31, CE32 CE48, CE49 CE29, CE30 CE46, CE47 CE27, CE28 CE44, CE45 CE35, CE36 CE43, CE44 CE37 CE38 CE39 CE40 CE41, CE42 CELL 1 Instrumented Unrestrained Panels (No Dowel and Tie Bars) Slab A Slab B Slab C CE07, CE08 CE05, CE06 CE03, CE04 CE01, CE02 CE09, CE10 CE11, CE12 CE13 CE14 CE15, CE16 CE25, CE26 CE23, CE24 CE21, CE22 CE19, CE20 CE17, CE18 CE33, CE34 CE31, CE32 CE29, CE30 CE27, CE28 CE35, CE36 CE37 CE38 CE39 CE40 CE41, CE42 CELL 2 CE52, CE54 CE48, CE49 CE46, CE47 CE44, CE45 CE43, CE44 Dynamic Strain Gage (CE) Dynamic Pressure Cell (DP) Figure 2.1. Sensor layout for Cell 1 and Cell 2. 7

14 Representative Dimensions Cells 1 and Sensor Clearance from Panel Edge (Typical All Sensors) Sensor Group Sensor Group Sensor Group Dynamic Strain Gage (CE) Dynamic Pressure Cell (DP) Figure 2.2. Typical dimensions of dynamic strain gages, and dynamic pressure cells. 8

15 typical dimensions of the sensor layout. The top sensor is 0.5 in below the surface of the slab and the bottom sensor is located directly on the bottom of the slab, therefore each location measures strain at the top and bottom of the slab Axle Loads and Configurations used for Field Testing The truck load testing conducted over the dynamic strain gages used three different axle configurations, as shown in Figure 2.3 through Figure 2.5. The three trucks consisted of the following: 1. 6-axle semi (FHWA Class 10); 2. 4-axle dump truck with a triple axle in the rear (FHWA Class 7) and 3. 3-axle dump truck with a tandem axle in the rear (FHWA Class 6). Each truck was loaded with three different loads representing an average, high and overload condition. Additionally, for each axle and load configuration, the truck made two passes over the test section. One pass was with the outside wheels passing directly adjacent to the lane/shoulder edge. The other pass was in the wheelpath, approximately two feet from the lane/shoulder. Each truck pass was completed at creep speed. The axle that produced the maximum strain was identified for each truck included in the field study. This critical axle is highlighted in gray in Figure 2.3 through Figure 2.5 for each truck classification. A finite element analysis was performed to determine the stress along the transverse joint where the Group 1 sensors are located and at midpanel where the Group 3 sensors are located. The result from this analysis is presented in the next section. 14 and tire width is Figure 2.3. Axle configuration and tire spacing for the Class 6 truck. 9

16 Figure 2.4. Axle configuration and tire spacing of the Class 7 truck Figure 2.5. Axle configuration and tire spacing of the Class 10 truck Stress Corresponding to Measured and Predicted Strains The Smart Pavement was modeled using finite element. The inputs for the models were established based on FWD deflection data and material property testing [1; 3]. The models were then validated using field strain measurements. The stress induced during truck testing for the critical axle was determined using these validated models. The predicted stress corresponding to the strain measured using the second sensor from the longitudinal joint within sensor Group 1 (adjacent to transverse joint) was determined. The predicted stress corresponding to the first sensor for the unrestrained slabs and third sensor for the restrained slabs from the longitudinal joint within sensor Group 3 (midpanel) was also determined. 10

17 The location of the stress calculated for the Group 1 sensors is 4 in from the transverse joint in the longitudinal direction and 36 in from the lane/shoulder joint in the transverse direction. The location of the calculated stress for Group 3 sensors is 90 in from the transverse joint in the longitudinal direction and 10 in from the lane/shoulder joint in the transverse direction for the unrestrained slabs and 22 in from the lane/shoulder joint in the restrained slabs. Additionally, stress was determined for each truck classification (Class 6, 7, and 10). The results of this analysis are provided below and begin with a discussion of stress calculated for the Group 1 sensor Stress along the Transverse Joint (Group 1 Sensor Location) Table 2.1 presents the results for the maximum tensile stress calculated at the transverse joint in the restrained and unrestrained slabs for each truck classification (Class 6, 7, and 10). Again, the stress was determined at location of the strain gage when the critical axle is directly on top of the gage. for truck loads applied to the Smart Pavement when the effective gradient in the slab is close to zero. Stress at the transverse joint of the restrained slab ranged between 20 and 22 psi for the range of axle configurations and axle loads considered. The corresponding measured strains varied between 4 and 6 microstrain. Stress in the unrestrained slab ranged between 19 and 35 psi, while the measured strain ranged between 4 and 8 microstrain. The stress for both slabs is very small since the stress is the result of just an applied load and not a combination of both an applied and an environmental load. Stress in the unrestrained slabs was approximately 37 percent larger than stress in the restrained slabs, while strain was only 26 percent larger. The fact that the unrestrained slab is thinner (approximately 2.5 inches) than the restrained slab contributes to the lower stress. The reduced potential for stress transfer across the joints also results in larger stresses in the unrestrained slabs. It also must be remembered that comparisons can not be made between trucks because not all of the strains were measured at the same time of the year. 11

18 Table 2.1. Stress for the gage along the transverse joint (Group 1 sensors) during truck testing. Truck Type GAGE IN SENSOR GROUP 1 RESTRAINED SLABS Load per Measured Predicted Axle Microstrain Microstrain (lbs) Stress (psi) Class 6 15, Class 7 18, Class 10 18, UNRESTRAINED SLABS Class 6 15, Class 7 18, Class 10 18, Stress at Midpanel (Group 3 Sensor Location) Table 2.2 provides the stress determined at midpanel for both the restrained and unrestrained slabs during truck testing. Again, the stress was determined at location of the strain gage when the critical axle is directly on top of the gage for truck loads applied to the Smart Pavement when the effective gradient in the slab is close to zero. The strains in table 2.2 for the unrestrained slab were recorded by the first sensor away from the lane/shoulder joint. The strains for the restrained slab represent strains measured by the third sensor from the lane/shoulder joint. Stress at midpanel ranged between 36 and 52 psi for the restrained slabs while the measured strain varied between 3 and 10 microstrain for the range of axle loads and vehicle axles considered. Stress in the unrestrained slabs ranged between 40 and 43 psi while corresponding measured strain ranged between 5 and 8 microstrain. As the strain increases, the predicted strains more closely approximate the measured strains. This reflects the difficulty in measuring such small strains with the strain gages and data collection equipment used. Again, the stresses corresponding to the measured strains are quite small because the testing was performed when the effective gradient in the slabs was close to zero. Table 2.2. Stress at midpanel (Group 3 sensors) during truck testing. 12

19 Truck Type Load Level (lbs) SENSOR GROUP 3 RESTRAINED SLABS Measured Microstrain Predicted Microstrain Stress (psi) Class 6 15, Class 7 18, Class 10 18, UNRESTRAINED SLABS Class 6 15, Class 7 18, Class 10 18, Effect of Environmental and Applied Loads on Stress in the Smart Pavement The stresses determined during the validation of the finite element models in the previous section were quite small. This is because the stress is the result of only an applied load and not the combination of both an applied and environmental load. This section evaluates the effects of both environmental and applied loads on the stress state in both the restrained and unrestrained slabs. The first analysis will evaluate the effect of temperature gradients in the Smart Pavement on stress. Three temperature gradient conditions (maximum positive, maximum negative, and zero) will be investigated for two loading configurations. These loading configurations are provided in Figure 2.6. The first loading configuration consists of placing a single axle in the wheelpath (approximately 24 inches from slab edge) and at midpanel. This is the critical load condition for bottom-up transverse cracking. The axle configuration and tire spacing of the single axle is provided in Figure 2.7. A tire pressure of 120 psi was used for all analyses. The second load configuration evaluates stress when a single axle is placed adjacent to each transverse joint. This represents an axle spacing that is approximately equal to the joint spacing, which is the critical load condition for top-down transverse cracking. Next, the effect of the magnitude of the applied load on stress in the Smart Pavement is evaluated. This analysis will compare stress in the restrained and unrestrained slabs when there is no gradient for each loading condition, and for load magnitudes of 12,000, 18,000 and 24,000 lbs. Finally, the effect of combined environmental and applied loads on stress in the Smart Pavement will be evaluated. 13

20 Load Condition 1 Load Condition 2 Figure 2.6. Loading configurations evaluated in the stress analysis for the restrained and unrestrained slabs Figure 2.7. Axle configuration and tire spacing for the single axle Effect of Slab Gradient on Stress in the Smart Pavement The effect of the temperature gradient in the slab on stress in the restrained and unrestrained slabs is evaluated first. The stress is determined for each load configuration in the presence of a positive, negative and zero temperature gradient condition. An 18,000 lb axle load was applied. The largest positive and negative temperature gradient to develop in the Smart Pavement during the first three years after construction, occurred during the spring of 2007 and was F/in and F/in, respectively. Table 2.3 provides the temperature conditions in both the restrained and unrestrained slabs during these gradients. 14

21 The temperature throughout the slab and the asphalt treated permeable base (ATPB) during the zero gradient condition were 104 F and 90 F, respectively, since these were the temperatures at the time the concrete set. A detailed description of the finite elements models can be found in the Phase II Final Report [1]. Several other parameters remained constant throughout this analysis and they include: modulus of subgrade reaction (k-value), 212 pci elastic modulus of the base, 336,000 psi load transfer efficiency of the joints - Restrained 87% - Unrestrained 67% Table 2.3. Temperatures in the slab and ATPB during the maximum positive and negative temperature gradients. Temperature in Slab, F Location Restrained Unrestrained in Slab Positive Negative Positive Negative Top Middepth Bottom ATPB Table 2.4 and Table 2.5 summarize the stresses determined using the finite element models for each gradient for both the restrained and unrestrained slabs. The maximum stress for the first load configuration (when the single axle was placed at midslab) occurred when a positive gradient was present. The maximum tensile stress at the bottom of the pavement was 501 psi in the restrained slabs and 362 psi in the unrestrained slabs. These are very high stresses. Very few load applications could be applied for this combination of temperature gradient and applied load. Again, this is the peak positive gradient that developed in the slab throughout the first three years after paving. The maximum tensile stress to develop at the top of the slab occurred when the second load configuration was applied along with the maximum negative gradient. The maximum tensile stress for this loading condition was 517 psi for the restrained slabs and 207 psi for the unrestrained slabs. This is a substantial difference in stress that results from 15

22 the different restraining conditions indicating that a large portion of the stress that is generated in the restrained slab is probably induced by the restraint the dowel and tie bars provide to curling. Evidence of this can be found by comparing the substantially higher stresses determined for the restrained slabs (Table 2.4) compared to the unrestrained slabs (Table 2.5) for both positive and negative gradients. The stress varied between 70 and 139 in the restrained slabs and between 57 and 92 in the unrestrained slabs when no gradient was present. Table 2.4. Effect of slab gradient on critical tensile stress in the restrained slab. CRITICAL TENSILE STRESS, PSI Restrained Slab Loading Condition Positive Gradient Negative Gradient Zero Gradient Table 2.5. Effect of slab gradient on critical tensile stress in the unrestrained slab. CRITICAL TENSILE STRESS, PSI Unrestrained Slab Loading Condition Positive Gradient Negative Gradient Zero Gradient Effect of Load Magnitude on Stress in the Smart Pavement The effect of load magnitude on stress in the restrained and unrestrained slabs is presented in this section. This analysis evaluates stress when no temperature gradient was present for both load conditions at three different load magnitudes (12,000, 18,000 and 24,000 lbs). The critical stress for each combination of variables is provided in Table 2.8 and Table 2.9 for the restrained and unrestrained slabs, respectively. Stress in the restrained slabs ranged between 113 and 162 psi for the loading Condition 1 and between 65 and 70 psi for the loading Condition 2. In the unrestrained slabs, stress ranged between 63 and 118 psi for loading Condition 1 and between 38 and 74 psi for loading Condition 2. 16

23 It was determined that stress in the restrained slabs was approximately 28 percent larger than stress in the unrestrained slabs when averaged for all the runs. However, when looking at the difference in stress between the restrained and unrestrained slabs for each load level, the difference between the two decreases as the load magnitude increases. For example, during loading Condition 2, stress in the restrained slabs is 42 percent larger then the unrestrained slabs when a 12,000 lb load is applied. However, when a 24,000 lb load is applied, the difference in stress between the two slabs types is only 4 percent. Table 2.6. Effect of load magnitude on critical tensile stress in the restrained slabs. Loading Condition Restrained Slabs Load Stress (psi) 1 12, , , , , , Table 2.7. Effect of load magnitude on critical tensile stress in the unrestrained slabs. Unrestrained Slabs Loading Condition Load Stress (psi) 1 12, , , , , , Combined Effect of Slab Gradient and Applied Load on Stress in the Smart Pavement The combined effect of slab gradient and load magnitude on stress in the restrained and unrestrained slabs is presented herein. This analysis evaluates stress for the critical loading condition in the presence of the maximum positive, maximum negative, and zero 17

24 gradient as the load magnitude is varied between 12,000 and 24,000 lbs. The results for the restrained slabs are summarized in Table 2.8. The variation of stress in the restrained slabs can be seen in Figure 2.8. The range of the stress in the restrained slabs was 82 psi for loads between 12,000 and 24,000 lbs applied using loading Condition 1 when a o F/in was present. The maximum stress developed at the bottom of the slab at midpanel for this combination of loads. For a 1.72 o F/in gradient and applying loads between 12,000 and 24,000 lbs using load Condition 2, the stress varied by only 2 psi. The stress in the slab was close to 500 psi for the restrained slabs when a gradient is present. At such a high stress level, very few applications can be applied before failure. Fortunately, gradients this high rarely occur. With no gradient, the stress in the slab varied 49 psi when load Condition 1 was applied and only 12 psi for load Condition 2. Table 2.9 provides the variation of stress in the unrestrained slabs due to the combined effect of environmental and applied loads, while Figure 2.9 shows this variation graphically. A o F/in gradient and load Condition 1 produced a change in stress of 46 psi as the load magnitude varied between 12,000 and 24,000 lbs for the unrestrained slab. Load Condition 2 in combination with a gradient results in a change in stress of 36 psi. The magnitude of the stress in the unrestrained slabs is substantially lower than the restrained slabs. This emphasizes the fact that even though dowels help to reduce the deflections at the joint; they also can contribute to increases in stress in the slab. 18

25 Table 2.8. Effect of gradient and load magnitude on critical tensile stress in the restrained slabs. Restrained Slabs Loading Condition Load (lb) Stress (psi) 1 12, , , , , , Stress (psi) Effect of Environmental and Loading Conditons on Stress (Restrained Slabs) Load Condition 2 LC Load Condition 1 Load Condition 1 Load Condition 2 0 5,000 10,000 15,000 20,000 25,000 30,000 Positive (2.2 F/in) Load (lbs) Negative (-1.7 F/in) Zero (Loading Cond. 1) Zero (Loading Cond. 2) Figure 2.8. Effect of environmental and loading conditions on stress in the restrained slabs. 19

26 Table 2.9. Effect of gradient and load magnitude on critical tensile stress in the unrestrained slabs. Unrestrained Slabs Loading Condition Load (lb) Stress (psi) 1 12, , , , , , Stress (psi) Effect of Environmental and Loading Conditons on Stress (Unrestrained Slabs) Load Condition 1 Load Condition 2 Load Condition 1 Load Condition 2 0 5,000 10,000 15,000 20,000 25,000 30,000 Positive (2.2 F/in) Load (lbs) Negative (-1.7 F/in) Zero ( Loading Cond. 1) Zero (Loading Cond. 2) Figure 2.9. Effect of environmental and loading conditions on stress in the unrestrained slabs. 20

27 CHAPTER 3: EVALUATION OF THE FATIGUE CRACKING MODEL OF THE MEPDG INTRODUCTION This chapter presents the evaluation of the accuracy of the calculated stress in the fatigue cracking model of the MEPDG. Fatigue damage is accumulated over the entire pavement life due to the combined effects of environmental and traffic loads. To evaluate the fatigue cracking model, the MEPDG is used to predict damage for an unrestrained 12-in thick jointed plain concrete pavement over a 20-year design life. The model is based on the pavement structure properties of the Smart Pavement, which were determined in the Phase II Final Report [1]. Since fatigue takes into account the combined effects of environmental and traffic loads, and to facilitate the evaluation of this model, constant climatic conditions are used in this model. This will help isolate the effect of traffic loading alone on a concrete slab with a constant temperature and moisture profile through it. In addition, to simplify the analysis of the cracking model and the evaluation of concrete stress, one type of traffic loading is applied on the pavement during the 20-year design life. Moreover, the same pavement structure, subjected to the same temperature and moisture conditions and the same traffic loading is then analyzed using the finite element program, Illislab, to determine the resulting stresses in the concrete. The stress is evaluated based on the base properties and stiffness of the underlying layers that are estimated by the MEPDG and those estimated based on the results of the calibrated finite element models presented in the Phase II Final Report [1]. The resulting fatigue damage and slab cracking based on both sets of data are compared and the critical stress to cause slab failure is estimated. This chapter begins with an overview of the fatigue cracking model used in the MEPDG. Then the inputs used to model the pavement structure in the MEPDG are presented, and finally, the results of the fatigue damage and slab cracking are presented MEPDG FATIGUE CRACKING MODEL The fatigue damage due to the combined effect of environmental and traffic loads is accumulated according to Miner's damage hypothesis by summing the damage over the 21

28 entire design period. When the estimated value of accumulated damage is small, the pavement structure is not expected to have physical distresses. When the accumulated damage is large, physical distresses can be expected [4]. Several key factors are taken into account in the fatigue cracking model. They include the following: Traffic load and number of applications, Slab curvature at the time of loading, which is affected by the climatic conditions, PCC material properties, and Base material and subgrade soil properties. The critical traffic loading condition varies depending on the slab curvature. When the slab curvature is downward (positive gradient), the critical traffic loading at midpanel results in high tensile stress at the slab bottom. When the slab curvature is upward (negative gradient), the critical traffic loading at the slab edges results in high tensile stress at the slab top. The PCC material properties influence the strength of the concrete at the time of loading. The base material and subgrade soil properties are used to characterize the subgrade k-value needed for the design analysis. The subgrade k-value is obtained through a conversion process, which transforms the actual pavement structure into an equivalent structure that consists of the PCC slab, base, and an effective dynamic k-value. The E-to-k conversion is performed internally in the MEPDG software as a part of input processing. The effective k-value used in this Guide is a dynamic k-value, which should be distinguished from the traditional static k-values used in previous design procedures. The dynamic k-value is typically considered to be approximately twice that of the static k-value. The general expression for fatigue damage accumulation is presented in equation 3-1. ni FD = N, j, k, l, m, n i, j, k, l, m, n (Equation 3-1) Where: FD = Total fatigue damage (top-down or bottom-up) n i,j,k,l,m,n = Applied number of load applications at condition i, j, k, l, m, n. N i,j,k,l,m,n = Allowable number of load applications at condition i, j, k, l, m, n. 22

29 i = Age (accounts for change in PCC modulus of rupture, layer bond condition, deterioration of shoulder LTE) j = Month (accounts for change in base and effective dynamic modulus of subgrade reaction) k = Axle type (single, tandem, and tridem for bottom-up cracking; short, medium, and long wheelbase for top-down cracking) l = Load level (incremental load for each axle type) m = Temperature difference n = Traffic path Each load application (n i,j,k,l,m,n ) is identified by the axle type k at load level l that passed through traffic path n at a specific age i during month j. A temperature difference m is present at the time the load is applied. The allowable number of load applications is the number of load cycles at which fatigue failure is expected (corresponding to 50 percent slab cracking) and is a function of the applied stress and PCC strength. The allowable number of load applications is determined using the field calibrated fatigue model presented in equation 3-2. MRi c2 log( Ni j, k, l, m, n) = C1.( ) σ, + i, j, k, l, m, n (Equation 3-2) Where: N i,j,k,l,m,n = Allowable number of load applications at condition i, j, k, l, m, n MR i = PCC modulus of rupture at age i, psi σ i,j,k,l,m,n. = Applied stress at condition i, j, k, l, m, n C 1 = Calibration constant = 2.0 C 2 = Calibration constant = 1.22 The final calibrated model, which shows percentage of slabs with transverse cracks of all severities in a given traffic lane, provides the amount of transverse cracking. This model is represented by equation 3-3. The model is used for predicting both bottom-up and topdown cracking. The total amount of cracking is determined using equation CRK = (Equation 3-3) FD 23

30 Where: CRK = Predicted amount of bottom-up or top-down cracking (fraction). FD = Fatigue damage calculated using equation 3-1. TCRACK=(CRK Bottom-up +CRK Top-down -CRK Bottom-Up.CRK Top-down ).100% (Equation 3-4) Where: TCRACK = Total cracking (percent). CRK Bottom-up = Predicted amount of bottom-up cracking (fraction). CRK Top-down = Predicted amount of top-down cracking (fraction) MEPDG INPUTS The MEPDG was used to model the Smart Pavement to determine the amount of fatigue damage and slab cracking sustained under a repeated load. For this, the pavement design inputs were established using Level 1, 2 and 3 data. Details on how these inputs were defined are provided in the Phase II Final Report [1]. Whenever available, Level 1 inputs are used in the analysis since they provide the highest level of accuracy. The inputs used in the analysis are summarized below GENERAL INPUTS The general design inputs define the analysis period, type of design, pavement construction month and traffic opening month, as presented in Table Table General MEPDG design inputs. Input Parameter Value Design life 20 years Construction month August-04 Traffic opening month September-04 Type of Design JPCP The pavement consists of an undoweled 12-in PCC slab resting on the following layers: a 4-in asphalt treated permeable base (ATPB), a 5-in PennDOT 2A subbase and 24 in of fill. The general structural design inputs are provided in Table In this analysis, the built-in construction gradient in the slab is established as zero. 24

31 Table General structure inputs. Input Parameter Value Permanent curl/warp 0 F Joint spacing 15 feet Sealant type Liquid Base type Asphalt treated Erodobility index very erosion resistant (2) PCC-base interface full friction contact Loss of full friction 229 months ENVIRONMENTAL AND CLIMATIC INPUTS The environmental factors significantly affect performance of JPCP. Factors such as precipitation, temperature, and moisture determine the shape and critical stresses of a concrete slab, which affects performance. In the MEPDG, the environmental analysis is performed by the Enhanced Integrated Climatic Model (EICM) which simulates changes in the pavement and subgrade materials that are caused by seasonal changes in environmental conditions [4]. The EICM predicts temperature, resilient modulus adjustment factors, pore water pressure, water content, frost and thaw depth, frost heave, and drainage throughout the entire pavement structure. The output from the EICM is used by the structural response models and performance prediction models of the MEPDG to evaluate the performance of the trial design pavement over the design life. When the MEPDG uses the damage accumulation model, the design analysis period is divided into monthly time increments to analyze the proposed pavement structure. Each month is then subdivided into 2-hour periods to establish the temperature profiles in the slab. For each time increment, the equivalent linear temperature difference through the concrete slab is accounted for in increments of 2 F for both positive (daytime) and negative (nighttime) top-to-bottom temperature differences. In addition, all other factors that affect pavement response and damage are held constant within each time increment; they include: concrete strength and modulus, base modulus, subgrade modulus and joint load transfer across transverse and longitudinal joints. For each time increment, 25

32 critical stresses, strains and deflections are determined along with damage accumulated during that time increment. The fatigue damage due to the combined effect of environmental and traffic loads is accumulated according to Miner's damage hypothesis by summing the damage over the entire design period. When the estimated value of accumulated damage is small, the pavement structure is not expected to have physical distresses. When the accumulated damage is large, physical distresses can be expected [4]. The fatigue damage model is discussed in more detail in section In this project, the climatic data is held constant throughout the design life. This will help isolate the effect of climatic changes on the development of stress in the concrete slab and provides a means for evaluating the fatigue damage and slab cracking model. A sitespecific weather station with the following constant conditions was created for this project using the Integrated Climatic Model (ICM) [5]: Hourly air temperature: 80 F Hourly precipitation: 0 in Hourly wind speed: 0 mph Hourly percentage sunshine: 1 percent Hourly relative humidity: 95 percent The above values were selected to minimize daily and seasonal temperature and moisture changes in the slab. The temperature of 80 F was also selected as the zero-stress temperature in the concrete and the reference temperature in the asphalt base to help reduce environmental-related stress. The precipitation was set at zero and the hourly relative humidity was set at 95 percent to minimize variations in the concrete moisture gradient. The MEPDG assumes a constant relative humidity of 95 percent throughout the depth of the slab accept for the top 2 in of the slab. The top 2 in fluctuates as a function of the ambient relative humidity. By establishing a constant ambient relative humidity of 95 percent, the upper 2 in of the concrete will maintain a constant relative humidity of 95 percent throughout the depth of the slab. This allows the pavement structure to be loaded while the slab is not warped. The hourly sunshine was set to 1 percent to avoid daily and seasonal changes in the temperature gradient that are caused by exposure to the sun. However, the ICM automatically calculates the solar radiation factor based on the project location. As a result, 26

33 the pavement structure was found to undergo some temperature changes throughout the 20- year design life. To make sure the temperature distribution in the slab was kept constant throughout the design life, the pavement was only loaded during the same one-hour period everyday for one month. (Note: The MEPDG reduces the analysis period to less than onemonth increments during spring-thaw. Since the ambient temperature was kept constant at 80 F, the subgrade never froze. Therefore, it can be safely assumed that the analysis period was one month for every month of the year.) The pavement was modeled using a zero temperature gradient in the slab and a positive gradient in the slab. It was found that the MEPDG predicts no damage or cracking for the slab when the loads are applied when a zero temperature gradient is present in the slab. As a result, the traffic was applied when a positive temperature gradient is present in the slab. A positive temperature difference of 13.8 F is selected for this study. This is equivalent to a temperature gradient of F/in across the slab. This gradient was selected because it is suitable for the estimation of slab stresses using Illislab. It has been shown that when using a large positive gradient, Illislab tends to overestimates slab curvature when compared to curvature measured using a Dipstick. This results in an overestimation of the stress [6]. This gradient was found to occur daily at 2:00 PM during the month of May. And the corresponding temperatures in the slab and underlying base layer are presented in Table Table PCC and ATPB temperatures at 2:00 PM during the month of May. Location Temperature ( F) PCC slab: - Top Middepth Bottom ATPB Groundwater-related inputs also play a significant role in the overall accuracy of the foundation/pavement moisture contents. The depth to the water table for this project was identified from the results of soil borings performed near the test section. According to the 27

34 boring log, the water table for the test section at stations and is 9 ft below the surface of the soil. Other drainage inputs affecting the infiltration of water into the pavement and the drainage of the pavement that were used are discussed in the Phase II Final Report [3] and are summarized in Table Table General drainage inputs. Input Parameter Value Type of infiltration Minor Cross slope 2.0 % Longitudinal slope 2.4 % Lane width 12 feet Edge drain trench width 6 in Cross-section geometry Crowned PCC MATERIAL PROPERTIES INPUTS PCC material properties play a significant role in the performance of slabs in response to environmental and applied loads and are very important inputs in the distress prediction models of the MEPDG. PCC properties can be classified under three major conceptual groups: Strength/mechanical behaviour: Modulus of elasticity, Poisson s ratio, modulus of rupture, indirect tensile strength, compressive strength, PCC unit weight and coefficient of thermal expansion. Shrinkage: Ultimate shrinkage, reversible shrinkage, time to reach 50 percent ultimate shrinkage. Thermal behaviour: Thermal conductivity, specific heat and surface short wave absorptivity. Most of the input variables within the first group vary with PCC age in the short and long term. Due to the incremental nature of the distress prediction models used in the MEPDG, a time dependent variation of these properties is considered throughout the design life. The PCC material property inputs required for the MEPDG are presented in this section. 28