EFFECT OF TEMPORAL FWD AND PROFILE MEASUREMENTS ON DERIVED PAVEMENT PARAMETERS By Hamad Bin Muslim A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Civil Engineering - Master of Science 2020 ABSTRACT EFFECT OF TEMPORAL FWD AND PROFILE MEASUREMENTS ON DERIVED PAVEMENT PARAMETERS By Hamad Bin Muslim Evaluating a analyzing deflections measured by Falling Weight Deflectometer (FWD) while assessing surface roughness, estimated from longitudinal profile measurements , helps determine its functional performance . However, s easonal and diurnal changes (temperature and moisture) influence such measurements . E v aluati ng temporal variations and quan tifying their impact on these measurements may aid in a better understanding of pavement parameters derived from these measurements. Long - Term Pavement Performance (LTPP) Seasonal Monitoring Program (SMP) study i s designed to understand environmental fact ors and their effects on pavement structural and functional performance. Analyzing data from the SMP study show s that FWD and profile measurement season and time of the day have a significant impact on parameters derived from these measurements. Relating the observed effects with recorded ambient temperatures helped developing general guidelines for FWD and profile measurements in different climatic regions. The recommended temperature range for FWD testing on rigid and flexible pavements in freeze climates is 55 to 70 F ; 65 to 75 F and 60 to 75 F in the non - freeze climates for flexible and rigid pavements, respectively . The study recommends b efore - noon FWD te sting for rigid , while no time limit within a day for the flexible pavements. Also, the research suggests a temperature range between 50 to 75 F for flexible pavement profiles with no time limitation. For rigid pavements, profile measurement s in the aftern oon are recommended with temperature ranges of 50 to 65 F, and 50 to 70 F in freeze and non - freeze climates, respectively . iii This thesis is dedicated to my family, my fellow graduate students, and all my mentors who supported me throughout my journey. iv ACKNOWLEDGMENTS I would like to thank my advisor, Dr. Syed Waqar Haider, for his continued guidance and support throughout the study . Dr. Haider has been a fantastic mentor who has taught me valuable lessons concerning the research process. He was always approachable and willing to help. This time that he lent me throughout the study and while writing my thesis. I would also like to acknowledge the committee members: Dr. Karim Cha tti , Dr. Neeraj Buch, and Dr. M . E min Kutay, who taught me valuable lessons during my graduate studies. I found each of them very helpful and inspirational. I would like to thank my family and friends in the department and back home , who always supported a nd motivated me. Also, I would like to appreciate the help and support from all those with whom I crossed my paths during this journey. v TABLE OF CONTENTS LIST OF TABLES ................................ ................................ ................................ ........................ vii LIST OF FIGURES ................................ ................................ ................................ ......................... x CHAPTER 1 INTRODUCTION ................................ ................................ ................................ ..... 1 1.1 Problem Statement ................................ ................................ ................................ ................. 1 1.2 Objectives ................................ ................................ ................................ .............................. 3 1.3 Outline of the Thesis ................................ ................................ ................................ .............. 3 CHAPTER 2 LITERATURE REVIEW ................................ ................................ .......................... 4 2.1 Evaluating Structural Capacity of an Existing Pavement ................................ ...................... 4 2.2 Environmental Effects on FWD Measurements ................................ ................................ .... 4 2.2.1 Temperature Effects on AC Materials ................................ ................................ .......... 5 2.2.2 Environmental Effects on Unbound Materials ................................ .............................. 8 2.2.3 Environmental Effects on PCC Slab ................................ ................................ ........... 12 2.3 Recommended FWD Guidelines ................................ ................................ ......................... 14 2.4 Evaluating Functional Performance of an Existing Pavement ................................ ............ 17 2.5 Environmental Effects on IRI Measurements ................................ ................................ ...... 17 2. 6 Summary ................................ ................................ ................................ .............................. 21 CHAPTER 3 DATA EVALUATION ................................ ................................ ........................... 22 3.1 Task 1: Identification of Data Elements ................................ ................................ .............. 22 3.2 Task 2: Data Availability and Extents ................................ ................................ ................. 27 3.2.1 Pavement Cross - Sections ................................ ................................ ............................ 28 3.2.2 FWD Based Parameters ................................ ................................ .............................. 29 3.2.3 Sub - surface Temperature Data ................................ ................................ .................... 37 3.2.4 Longitudinal Profile Measurements ................................ ................................ ............ 38 3.3 Summary ................................ ................................ ................................ .............................. 39 CHAPTER 4 DATA ANALYSIS ................................ ................................ ................................ . 41 4.1 Analysis for Flexible Pavements (by climatic region) ................................ ......................... 41 4.1.1 FWD based Pavement Parameters ................................ ................................ .............. 41 HMA Layer Moduli ................................ ................................ ................................ ..... 43 Base Layer Moduli ................................ ................................ ................................ ...... 55 Subgrade Layer Moduli ................................ ................................ .............................. 64 Discussion on ANOVA Results (HMA, Base and Subgrade Moduli) .......................... 72 4.1.2 Longitudinal Profile Measurements (IRI) Flexible Pavements ............................... 79 4.2 Analysis for R igid Pavements (by climatic region) ................................ ............................. 93 4.2.1 FWD based Pavement Parameters ................................ ................................ .............. 94 PCC Layer Moduli ................................ ................................ ................................ ...... 96 Modulus of Subgrade Reaction (k - value) ................................ ................................ . 103 Discussion on ANOVA Results (PCC modulus and k - value s) ................................ .. 109 vi Load Transfer Efficiency (LTE) ................................ ................................ ................ 115 4.2.2 Longitudinal Profile Measurements (IRI) JPCP Pavements ................................ .. 125 4.3 Summary ................................ ................................ ................................ ............................ 137 CHAPTER 5 CONCLUSIONS AND RECOMMENDATIONS ................................ ................ 139 5.1 Conclusions ................................ ................................ ................................ ........................ 139 5.2 Recommendations for FWD and Profile Measurements ................................ ................... 141 5.3 Recommended Future Work ................................ ................................ .............................. 142 APPENDIX ................................ ................................ ................................ ................................ .. 144 REFERENCES ................................ ................................ ................................ ............................ 152 vii LIST OF TABLES Table 1 Devices used for measurements specific to SMP test sections [44] ................................ 24 Table 2 LTPP database tables ................................ ................................ ................................ ....... 26 Table 3 Original SMP experiment design [43] ................................ ................................ ............. 27 Table 4 Available pavement sections in the LTPP SMP database ................................ ............... 28 Table 5 Layer thicknesses of flexible pavement sections located in different climates ............... 29 Table 6 Layer thicknesses of JPCP pav ement sections located in different climates ................... 30 Table 7 Available backcalculated pavement structural parameters by FWD pass SMP flexible sections ................................ ................................ ................................ ................................ .......... 31 Table 8 Maintenance work categorization details ................................ ................................ ........ 31 Table 9 Available backcalculated pavement structural parameters by FWD pass and maintenance category in different climatic regions SMP flexible pavements ................................ ................ 32 Table 10 Monthly data distribution of backcalculated pavement parameters - SMP AC sections 34 Table 11 Available backcalculated pav ement structural parameters (E and k - values) by FWD pass in different climatic regions SMP JPCP sections ................................ ............................... 35 Table 12 Available backcalculated pavement structural parameters by FWD pass and maintenance category in different climatic regions SMP JPCP section s ................................ .. 35 Table 13 Monthly data distribution of backcalculated parameters - SMP JPCP sections ............ 36 Table 14 Hourly data distribution of backcalculated parameters - SMP JPCP sections .............. 37 Table 15 Available HMA layer mid - depth temperatures on FWD measurement days SM P sections ................................ ................................ ................................ ................................ .......... 38 Table 16 Available temperature gradients on FWD measurement days matched with LTE values SMP JPCP sections ................................ ................................ ................................ ......... 38 Table 17 Available temperature gradients on FWD measurement days matched with backcalculated moduli values SMP JPCP sections ................................ ................................ .... 38 Tabl e 18 Available profile measurements in different climatic regions SMP experiment ........ 39 viii Table 19 Descriptive statistics HMA layer moduli SMP flexible pavements ........................... 45 Table 20 ANOVA results for HMA moduli values DF climatic region ................................ .... 47 Table 21 ANOVA results for HMA moduli values DNF climatic region ................................ . 50 Table 22 ANOVA results for HMA moduli values WF climatic region ................................ ... 51 Table 23 ANOVA results for HMA moduli values - WNF climatic region ................................ . 53 Table 24 Descriptive statistics for aggregate base layer modul i data SMP flexible pavements 57 Table 25 ANOVA results for base layer moduli values DF climatic region ............................. 58 Table 26 ANOVA results for base layer moduli values DNF climatic region .......................... 60 Table 27 ANOVA results for base layer moduli values WF climatic regi on ............................ 62 Table 28 ANOVA results for base layer moduli values WNF climatic region ......................... 63 Table 29 Descriptive statistics subgrade layer moduli SMP flexible pavements ...................... 66 Table 30 ANOVA results for subgrade layer moduli values DF climatic region ...................... 66 Table 31 ANOVA results for subgrade layer moduli values DNF climatic region ................... 67 Table 32 ANOVA results for subgrade layer moduli values WF climatic region ..................... 70 Table 33 ANOVA results for subgrade layer moduli values WNF climatic region .................. 70 Table 34 Descriptive st atistics IRI values SMP flexible pavement sections ............................. 82 Table 35 ANOVA results for IRI data for flexible pavements DF climatic region ................... 82 Table 36 ANOVA results for IRI data for flexible pavements DNF climatic region ................ 85 Table 37 ANOVA results for IRI data for flexible pavements WF climatic region .................. 89 Table 38 ANOVA results for IRI data for flexible pavements WNF climatic region ............... 91 Table 39 Descriptive statistics PCC moduli SMP JPCP pavements ................................ ......... 97 Table 40 ANOVA results for PCC moduli values DF climatic region ................................ ...... 97 Table 41 ANOVA results for PCC moduli values DNF climatic region ................................ ... 99 ix Table 42 ANOVA results for PCC moduli values WF climatic region ................................ ... 100 Table 43 ANOVA results for PCC moduli values WNF climatic region ................................ 101 Table 44 Descriptive statistics k - values SMP JPCP pavements ................................ ............. 105 Table 45 ANOVA results for k - values DF climatic region ................................ ..................... 105 Table 46 ANOVA results for k - values DNF climatic region ................................ ................. 106 Table 47 ANOVA results for k - values WF Climatic region ................................ ................... 107 Table 48 ANOVA results for k - values WNF Climatic region ................................ ................. 108 Table 49 Descriptive statistics LTE values ................................ ................................ ................. 115 Table 50 ANOVA results for LTE values DF climatic region ................................ ................ 118 Table 51 ANOVA results for LTE values DNF climatic region ................................ ............. 120 Table 52 ANOVA results for LTE values WF cl imatic region ................................ .............. 120 Table 53 ANOVA results for LTE values WNF climatic region ................................ ............ 122 Table 54 Descriptive statistics IRI values SMP JPCP pavement sections ............................... 127 Table 55 ANOVA results for IRI values DF climatic region ................................ .................. 130 Table 56 ANOVA results for IRI values DNF climatic region ................................ ............... 131 Table 57 ANOVA results for IRI values WF climatic region ................................ ................. 134 Table 58 ANOVA results for IRI values WNF climatic region ................................ .............. 135 Table 59 Hourly data distribution of backcalculated parameters - SMP AC sections ................ 145 Table 60 Monthly distribution of IRI visit data - SMP AC sections ................................ .......... 147 Table 61 Hourly distribution of IRI visit data - SMP AC sections ................................ ............. 148 Table 62 Monthly distribution of IRI visit data - SMP PCC sections ................................ ........ 150 Table 63 Hou rly distribution of IRI visit data - SMP PCC sections ................................ ........... 151 x LIST OF FIGURES Figure 1 Measured FWD deflection time histories [17] ................................ ................................ . 7 Figure 2 Backcalculated deflection time histories [17] ................................ ................................ .. 7 Figure 3 Backcalculated dynamic modulus master - curves at high and lo w temperatures [17] ...... 7 Figure 4 Monthly comparison of combined (MODULUS) and AASHTO subgrade moduli for Station 1 on US - 160 [18] ................................ ................................ ................................ ................ 9 Figure 5 Monthly average moisture (top) and temperature (bottom) in a class 5 (base) layer [20] ................................ ................................ ................................ ................................ ....................... 10 Figure 6 Seasonal average m oisture variation in UAB and subgrade layers [20] ........................ 11 Figure 7 Seasonal average temperature variation in UAB and subgrade layers [20] ................... 11 Figure 8 Measured curling on August 7, 2003 [25] ................................ ................................ ...... 13 Figure 9 Tensile stress location under temperature - , moisture - , and load - related curvatures [33] ................................ ................................ ................................ ................................ ....................... 14 Figure 10 Diurnal curvature analysis [40] ................................ ................................ .................... 20 Figure 11 Joint functionality analysis showing diurnal effects [41] ................................ ............. 20 Figure 12 Pavement IRI plotted against curvature index for LTPP section 04 - 0215 (Based on profiles collected in 1997) [42] ................................ ................................ ................................ ..... 21 Figure 13 Seasonal and diurnal variations in HMA layer moduli ................................ ................ 44 Figure 14 Visualizing available HMA layer moduli data SMP flexible pavements .................. 45 Figure 15 Evaluating normality of the HMA moduli data - DF climatic region .......................... 46 Figure 16 Interaction means plot - DF climatic region ................................ ................................ . 47 Figure 17 Multiple mean comparisons with 95% confidence intervals - DF climatic region ...... 48 Figure 18 Interaction means plot - DNF climatic region ................................ .............................. 50 Figure 19 Mean HMA moduli difference for season and time interaction DNF climatic region ................................ ................................ ................................ ................................ ....................... 51 xi Figure 20 Factorial plots ANOVA for HMA layer moduli data WF climatic region ................ 52 Figure 21 ANOVA plots - WNF climatic region ................................ ................................ .......... 54 Figure 22 Seasonal and diurnal variations in base layer moduli ................................ .................. 56 Figure 23 Visualizing available base layer moduli data SMP flexible pavements .................... 57 Figure 24 Evaluating normality of the aggregate base moduli data - DF climatic region ............ 58 Figure 25 ANOVA plots for base moduli data - DF climatic region ................................ ............ 59 Figure 26 Factorial plots ANO VA for base layer moduli data DNF climatic region ................ 61 Figure 27 Main effects plot WF climatic region ................................ ................................ ........ 62 Figure 28 Main effects plot WNF climatic region ................................ ................................ ..... 63 Figure 29 Seasonal and diurnal variations in subgrade layer moduli ................................ ........... 65 Figure 30 Visualising avai lable subgrade layer moduli data SMP flexible pavements ............. 66 Figure 31 ANOVA plots for subgrade moduli data - DF climatic region ................................ .... 68 Figure 32 Interaction means plot for subgrade moduli data DNF climatic region .................... 69 Figure 33 ANOVA plots for subgrade moduli d ata - WF climatic region ................................ ... 71 Figure 34 ANOVA plots WNF climatic region ................................ ................................ ......... 72 Figure 35 HMA layer moduli temperature correction using the Asphalt Institute equation ........ 73 Figure 36 Pavement surface and ambient temperatures during FWD measurements on flexible pav ements - DF climatic region ................................ ................................ ................................ .... 74 Figure 37 Pavement surface and ambient temperatures during FWD measurements on flexible pavements - DNF climatic region ................................ ................................ ................................ . 75 Figure 38 Pavement surface and ambient temperatures during FWD measurements on flexible pavements - WF climatic region ................................ ................................ ................................ ... 76 Figure 39 Pavement surface and ambient temperatures during FWD measurements on flexible pavements - WNF climatic region ................................ ................................ ................................ 77 Figure 40 Monthly variation in layer moduli values - SMP flexible pavement sections ............. 78 Figure 41 As sessing available IRI data SMP flexible pavement sections ................................ . 81 xii Figure 42 An example of ensuring normality of the IRI data - WF climatic region .................... 83 Figure 43 Interaction means plot for IRI data - DF climatic region ................................ ............. 84 Figure 44 Mean IRI multiple comparison plots DF climatic region ................................ .......... 84 Figure 45 IRI measurement temperatures with 95% confidence intervals - DF climatic region . 85 Figure 46 Interaction means plot for IRI data - DNF climatic region ................................ .......... 86 Figure 47 Mean IRI multiple comparison plots DNF climatic region ................................ ....... 87 Figure 48 IRI measurement temperatures with 95% confidence intervals - DNF climatic region 88 Figure 49 ANOVA results for IRI data of flexible pavements WF climatic region .................. 90 Figure 50 ANOVA results for IRI of flexible pavements WNF Climatic region ...................... 91 Figure 51 Monthly IRI variation - SMP flexible pavements ................................ ....................... 93 Figure 52 Assessing available PCC moduli data SMP JPCP pavement sections ...................... 96 Figure 53 Ensuring normality of the PCC moduli data - DF climatic region ............................... 98 Figure 54 ANOVA plots for PCC moduli data - DF climatic region ................................ ........... 99 Figure 55 Main effects plot for PCC moduli data DNF climatic region ................................ . 100 Figure 56 ANOVA plots for PCC moduli data - WF climatic region ................................ ........ 101 Figure 57 ANOVA plots for PCC moduli data - WNF climatic region ................................ ..... 102 Fig ure 58 Assessing available k - values ................................ ................................ ...................... 104 Figure 59 ANOVA plots for k - values - DF climatic region ................................ ....................... 106 Figure 60 ANOVA plots for k - values - DNF climatic region ................................ ................... 107 Figure 61 ANOVA plots for k - values - WF climatic region ................................ ...................... 108 Figure 62 ANOVA plots for k - values - WNF climatic region ................................ ................... 109 Figure 63 Pavement surface and ambient temperatures DF climatic region ........................... 110 Figure 64 Pavement surface and ambient temperatures DNF climatic region ........................ 111 xiii Figure 65 Pavement surface and ambient temperatures WF climatic region .......................... 112 Figure 66 Pavement surface and ambient temperatures WNF climatic region ....................... 112 Figure 67 Histogram of the available temperature gradients data for E pcc and k - values SMP JPCP sections ................................ ................................ ................................ .............................. 113 Figure 68 Variation in PCC layer moduli and k - va lues with temperature gradient within the PCC layer ................................ ................................ ................................ ................................ ............. 114 Figure 69 Assessing available LTE data ................................ ................................ ..................... 116 Figure 70 Evaluating normality of the LTE data - DNF climatic region ................................ .... 117 Figure 71 ANOVA results - DF climatic region ................................ ................................ ......... 119 Figure 72 ANOVA results - DNF clima tic region ................................ ................................ ...... 121 Figure 73 ANOVA results WF and WNF climatic regions ................................ ..................... 123 Figure 74 Seasonal and diurnal pavement surface and air temperatures ................................ .... 124 Figure 75 Histogram of the available temperature gradients data for LTE values SMP JPCP sections ................................ ................................ ................................ ................................ ........ 124 Figure 76 Variation in LTE with temperature gradient - JPCP sections ................................ .... 125 Figure 77 Assessing available IRI data SMP JPCP pavement sections ................................ ... 128 Figure 78 Evaluating normality of the IRI data - DF climatic region ................................ ........ 129 Figure 79 ANOVA plots for IRI data - DF climatic region ................................ ........................ 131 Figure 80 Pavement surface and air temperatures during profile measurements - DF climatic region ................................ ................................ ................................ ................................ .......... 132 Figure 81 ANOVA results for IRI - DNF climatic region ................................ .......................... 133 Figure 82 ANOVA results for IRI data - WF climatic region ................................ .................... 135 Figure 83 ANOVA results for IRI data - WNF climatic region ................................ ................. 136 Figure 84 Variation in IRI with temperature gradients JPCP sections ................................ .... 137 1 CHAPTER 1 INTRODUCTION 1.1 Problem Statement measurements and deflection analysis. On the other hand, assessing functional performance requires longitudinal profile measurements to estimate pavement surface roughn ess. These measurements are of vital importance in evaluating the structural and functional performance of pavements both at the network and project level. However, temporal variations ( i.e., seasonal temperature and moisture and diurnal temperature change s) influence these measurements. Consideration of this influence over the FWD and longitudinal profile measurements is necessary for accurate assessment of the pavement condition and limiting chances of under - or over - estimating pavement structural and fun ctional performance. Besides, the evaluation of these temporal variations and quantifying their effects on the measurements will assist in a better understanding of the derived pavement parameters: International Roughness Index (IRI) and backcalculated lay er moduli based on the measured longitudinal profiles and FWD deflections, respectively. The Long - Term Pavement Performance (LTPP) initiated the Seasonal Monitoring Program (SMP) study to better understand the environmental factors and their effects on pav ement structural and functional performance. The two major environmental factors are temperature and moisture. The design of the SMP study aimed to evaluate the influence of temporal changes in the pavement structural characteristics due to the diurnal and seasonal variations of these factors. The primary measure of a change in the pavement structural characteristics is the FWD deflection data. T he LTPP SMP sites were instrumented to measure the local weather, subsurface 2 moisture, and temperature t o explain the observed temporal changes in the FWD measurements. Therefore, the data can be analyzed to seek the full potential of the FWD deflection data for investigating the corresponding pavement condition for both flexible and rigid pavements. In addition to F WD deflection measurements, the LTPP also collected daily and seasonal data on longitudinal surface profiles on the SMP test sections. The results and findings presented in published studies on LTPP data analysis showed that on some LTPP Jointed Plain Conc rete Pavement (JPCP) sections, diurnal temperature variations could have a significant influence on the IRI computed from longitudinal pavement profile measurements [1 - 4] . Nearly all states collect roughness data on their interstate and primary highway net work either annually or biennially. This collected data at the network level is used to (a) assess the current pavement condition, (b) forecast the future condition. Pavement agencies use s uch information to define policies, allocate and justify budget req uests, and prioritize pavement rehabilitation works [2] . Variations in seasonal and diurnal IRI measurement s can influence the current practices used by highway agencies to measure smoothness for payment management purposes, construction quality control, a nd determining construction - related pay factors. Based on such information, the LTPP implemented a program to measure diurnal changes in longitudinal profile on JPCP test sections. However, there is a need to perform statistical data analyses to quantify t he impact of such measurements on the roughness indices such as IRI. Subsequently, based on the findings of such a study, the current procedures and guidelines on how to properly account for temporal variations in measured pavement p arameters should be ref ined. While the LTPP SMP was initiated to obtain data on the influence of temporal changes on pavement surface deflections and roughness, seasonal and diurnal data contained in the LTPP database, have not been analyzed to improve the deflection and roughne ss measurement 3 practices. Besides, there is a need to evaluate the current availability of the data along with its extent to quantify the effect of temporal FWD and longitudinal profile measurements on derived pavement parameters. 1.2 Objectives The objectives of this thesis are to ; a. E valuate the availability and extent of current data in the LTPP SMP study database for both flexible and rigid (JPCP) pavement sections. b. A ssess the effects of temporal (seasonal and diurnal) FWD and longitudinal profi le measurements on the derived pavement parameters. c. P ropose general recommendations for the improved use of FWD and profile measurements . 1.3 Outline of the Thesis This thesis has five chapters . C hapter 1 documents the problem statement and the objectives of this thesis. Chapter 2 presents a literature review of the environmental effects on FWD and longitudinal profile measurements. Chapter 3 describes the data elements identified for the accomplishment of the objectives of this study, details of the available data, and the data extents. Chapter 4 contains the data analysis to quantify the temporal effects on each data element, an explanation of the observed effects along with some demonstrative examples to illustrate these effects . Chapter 5 summarizes the res ults along with general recommendations for undertaking the FWD and longitudinal profile measurements. 4 CHAPTER 2 LITERATURE REVIEW 2.1 Evaluating Structural Capacity of an Existing Pavement As observed in early work on pavement deflections and fatigue failures, there is a strong correlation between pavement deflections and their ability to carry traffic loads at a prescribed minimum level of service [5] . Therefore, there is extensive use of d eflection - based techniques for evaluation of the structural capacity of existing pavements and the estimation of the in - situ elastic moduli of the pavement systems. Accurate assessment of the pavement structural condition and characterization of the materi als in different layers is necessary to determine cost - effective pavement treatment types. It also helps highway agencies in the allocation of funds and resources to maintain and rehabilitate the deteriorating infrastructure. Any pavement management system , responsible for making suitable preventive and corrective decisions, largely depends on the assessment of the pavement's existing structural condition. The key to success is the proper assessment of the present condition of the pavement structure and an accurate prediction of its future performance. In either case, the characterization of the pavement properties plays a critical role [6] . The need for accurate characterization of the structural condition of existing pavements has also increased manifolds because of the technology development s , ongoing improvements, and implementation of the Mechanistic - Empirical Pavement Design Guide (MEPDG) which require several accurately defined material inputs [7] . 2.2 Environmental Effects on FWD Measurements Variations i n environmental factors such as diurnal temperature, and seasonal temperature and moisture typically leads to change in the pavement material properties. Properties that get 5 influenced by the changes in the environmental factors include the moduli values o f asphalt bound materials or unbound materials such as base and subgrade. 2.2.1 Temperature Effects on AC Materials The seasonal and diurnal variation of asphalt pavement temperature is related to environmental factors such as air and pavement surface temperatur e. The stiffness of an AC layer is directly related to its temperature. The temperature dependent property of AC materials results in diurnal and seasonal variations in their structural performance. I nternal temperature changes within the AC layers caused by fluctuations in air temperature and solar radiation are the primary cause of the changes in their moduli and thus, the structural performance of AC materials. Besides, a part of the changes in stiffness can also be related to binder properties that chan ge over time because of age hardening and micro - cracking [8] . Pavement surface temperature, therefore, is a critical factor that requires due consideration as part of any pavement deflection testing program. A significant amount of work is available on th e temperature correction of the FWD data. T emperature correction s have been applied to the measured FWD data on raw deflections [9 - 12] , deflection basin parameters [8, 11, 13] , or on the backcalculated modulus [14] . These temperature corrections make use o f non - linear correction functions of temperature empirically. However, a work by Fernando et al. pointed out that temperature correction of the FWD deflections may not be suitable for project - level analysis because such correction methods alter the actual shape of the deflection basin. Instead, it recommends temperature correction of the backcalculated modulus [14] . Other literature also showed that correcting the raw FWD deflections or deflection basin parameters could be problematic because of the highly empirical, non - linear functions used for temperature correction [12] . 6 Since AC materials, by nature, are viscoelastic (VE), their modulus values depend on temperature and loading time (or frequency). According to the well - established theory of viscoelastic ity, the effect of time and temperature on the behavior of a linearly viscoelastic material can be addressed using the Time - Temperature Superposition Principle (TTSP). Although the TTSP may offer a more mechanistic means for temperature correction of the A C modulus, the use of this approach is not in practice due to the challenges associated with back - calculating the time - or frequency - dependent dynamic modulus (|E*|) from FWD time history data. Recent developments in dynamic/ VE back - calculation methodologies have shown potential for back - calc ulating the time - or frequency - dependent E(t)/|E*(f)| of asphalt concrete from FWD load and deflection time histories. Contrary to static solutions, dynamic back - calculation methodologies involve wave propagation - based theories that can model the stress wa ve propagation within a pavement structure m ore realistically. Also, the elastic - viscoelastic correspondence principle can be used to consider the viscoelastic nature of the asphalt material s [15] . ViscoWave, a new finite layer solution recently introduced , can model the dynamics of a variety of flexible pavement structures, including elastic/ VE layers, with or without a stiff layer, and with or without free vibrations [16, 17] . Besides, the preliminary back - calculation performed (see Figure 1 and Figure 2 ), using ViscoWave has also shown the potential for back - calculating the VE |E*| and, the master curve of the asphalt from the FWD time histories (see Figure 3 ) [17] . 7 Figure 1 Measured FWD deflection time histories [17] Figure 2 Backcalculated deflection time histories [17] Figure 3 Backcalculated dynamic modulus master - curves at high and low temperatures [17] 8 2.2.2 Environmental Effects on Unbound Materials R esilient modulus (MR ) characterizes unbound materials used as a base, subbase, and subgrade layers in pavement structures. MR of the material is a function of its density, the applied stress, and moisture content and is generally assumed to be temperature independent property. The density of the unbound material may vary with time (related to rutting). However, it is usually consider ed constant in pavement design. Therefore, stress level and moisture variations are of primary interest with regards to changes in unbound material resilient moduli values. A Kansas study investigated the seasonal variation of pavement material properties and behavior due to the change s in temperature and moisture. Temperature, moisture con tent, and FWD data were collected every month on four asphalt pavement sections for a year. The subgrade moduli were backcalculated using elastic layer theory using two different calculation schemes; MODULUS and AASHTO equations. Back - calculation of the su bgrade moduli involved dividing the subgrade layer into two sub - layers; a compacted subgrade layer and a natural soil subgrade layer. The compacted subgrade layer resulted in being more sensitive to seasonal variations at all the sites. Although the season al temperature variations are expected not to affect the subgrade moduli values ; however, the study observed that increasing temperature had a significant effect on the subgrade moduli values. The subgrade moduli value decreased with increasing temperature . The observed behavior was due to the temperature effects on the stiffness of the AC layer in a pavement structure, which in turn influences the deviatoric or bulk stresses in the subgrade layer. The increasing temperature softens the AC layer, thus incre ase the bulk stress. The increase in bulk stress resulted in a decrease of subgrade moduli values, as the subgrade soils were cohesive (silty clay soils) . The study also concluded that the subgrade moduli calculated from FWD measurements taken at unusually higher pavement surface 9 temperatures resulted in lower backcalculated subgrade moduli irrespective of the backcalculation scheme used. A possible reason is a violation of the linearity assumption resulting from variable deviatoric stresses on the subgrade . The monthly backcalculated subgrade moduli using AASHTO and MODULUS (combined), as seen in Figure 4 varied ; however, the effective roadbed soil resilient moduli computed with AASHTO algorithm yielded similar results irrespective of the calculation scheme used. Based on this finding, the study recommended three FWD measurements (i.e., at 4 - month interval) to capture the seasonal subgrade response [18] . Figure 4 Monthly comparison of combined (MODULUS) and AASHTO subgrade moduli for Station 1 on US - 160 [ 18] The behavior of the unbound aggregate base (UAB) and subgrade layers are significantly affected by seasonal moisture and temperature fluctuations. These fluctuations eventually influence the overall performance of the pavements by chang ing the load - bea ring capacity of these layers. High moisture content tends to decrease the stiffness of granular and subgrade layers, increase deflections, and, ultimately, reduce the life of the pavement systems [19] . Therefore, all pavement design procedures, including AASHTO and the MEPDG, rely on the use of M R to characterize unbound granular and subgrade layers. A study carried out on five 10 pavement sections in Minnesota observed the effects of seasonal moisture and temperature variations on UAB and subgrade layer. The study involved four different base/subbase aggregate material types commonly used in Minnesota. The study concluded that overall, subsurface material exhibited similar monthly moisture and temperature trends throughout the year. The moisture content incre ased as the temperature increased from spring to summer months; low moisture content was observed in the fall and winter months (see Figure 5 ). The study also observed s easonal variations in moisture content and temperature in different classes of base and subgrade material, with the highest and lowest moisture readings occurred in summer and winter, respectively (see Figure 6 ). Figure 7 shows the average seasonal temperature variations within the different classes of the base/subgrade layers [20] . Figure 5 Monthly average moisture (top) and temperature (bottom) in a class 5 (base) layer [20] 11 Figure 6 Seasonal average moisture variation in UAB and subgrade layers [20] Figure 7 Seasonal average temperature variation in UAB and subgrade layers [20] M R is a crucial design parameter in rigid pavement design procedures as well, which is converted to the modulus of subgrade reaction ( k ) using empirical relationship in the design proc ess. As an alternative to determining k - value by the static plate load test, the use of FWD is a 12 popular choice for determining it based on measurements from the field [21] . However, seasonal temperature variations and freeze - thaw cycles affect the FWD bas ed k - value. A two - year study, observing five rigid pavement sections in Iowa, concluded that mid - slab peak deflections (D 0 ) measured directly under the load in frozen conditions were 45% lower and were about the same once measured before freezing and durin g the thaw period, respectively. After thawing, D 0 values recovered to the same levels as before freezing and remained constant throughout summer. The corresponding k - values during the freezing period were twice as compared to pre - freeze conditions, droppe d to and remained constant during the thaw period and summer, respectively [22] . Thus, seasonal variations in temperature and moisture conditions affect pavement deflection response of both flexible and rigid pavements due to the effect on the response of underlying layers. Typically, deflections are higher in the spring because of wet conditions and reduced pavement support and are lower in the winter when the underlying layers and subgrade are frozen. However, rigid pavements are less affected by seasonal variations in support conditions comparatively. 2.2.3 Environmental Effects on PCC Sla b The environmental effects on a PCC slab of a rigid pavement can unfold in two ways. One, deterioration of the concrete itself, which is generally related to concrete mixture design and construction (which is not the focus here). Two, volumetric changes t hat cause the concrete slab to change size in either the horizontal or vertical direction. The later is called curling, which develops due to differential volume changes across the slab thickness. The design of a JPCP involves e ither AASHTO [23] or PCA [24 ] methods . Temperature gradients (i.e., the difference between the top and bottom of the PCC slab at a particular time of the day) cause the slab to curl 13 up or down during the night (negative gradient) and daytime (positive gradient), respectively (see Fig ure 9 ). Both upward and downward curling increases as the temperature gradients increase [25]. This curling results in loss of support at the corners or center coupled with the self - weight of the slab , and vehicle loads contribute to early - age slab crackin g or even slab failure [25 - 27]. The distribution of the thermal stresses had been considered linear across the slab thickness historically [28]. However, now it is well known to be non - linear [25, 27, 29]. A drawback of considering non - linear thermal gradi ents result in under - and over - estimating the stresses during day and night time, respectively [26]. The tensile stresses on an edge and a corner position can be as high as 5.5 and 8.8 times compared to stresses induced by a standard axle; both once negati ve gradients occur. Such was a conclusion in a study on the effects of thermal gradients on JPCP slab behavior [30]. Figure 8 Measured curling on August 7, 2003 [25] . In addition to the curling of the concrete slabs due to temperature gradient, the differential moisture gradient can also induce volume changes and cause upward and downward warping in JPCP slabs [31 - 33]. Generally, in most cases, the top surface of the PC C slab is partially saturated while the bottom; close to saturation. Such moisture conditions produce an upward 14 warping due to the presence of a negative moisture gradient for almost the entire duration of the day [34]. Besides warping, other moisture - rela ted behaviors in JPCP slabs are the built - in curling and early age shrinkage, which cause the slab to curl upward within a few months after paving and remain curled up therein [27, 35, 36]. Such curling results in an increase in top - down fatigue cracking a s opposed to the traditionally assumed bottom - up cracking [35]. A temperature model is proposed in a study to quantify the built - in curling in JPCP constructed in Pennsylvania. The state was divided into three climatic regions and using the developed model , monthly built - in temperature gradients in the C/cm units were calculated [36]. Figure 9 Tensile stress location under temperature - , moisture - , and load - related curvatures [33] 2.3 Recommended FWD Guidelines As discussed earlier, pavement deflections under the application of load by an FWD device vary with the seasonal and diurnal variations of pavement temperature and moisture. The variability in 15 the measured deflections influences the material properties obtained from these deflections using ba ck - calculat ion analyses. The existing literature has some guidelines on how to conduct FWD measurements in the field that can help minimize the effects of temporal changes of temperature and moisture on the backcalculated pavement parameters . The current recommended FWD deflection measurement guidelines include [37] : a. Temperature measurements should be collected during FWD testing on flexible pavements. Since HMA is a temperature - dependent material, the equivalent modulus obtained during backcalculation represents the material's temperature at the time of testing. Having accurate temperature data helps to determine the correction factor to apply to the backcalculated HMA modulus to obtain a value at a standard temperature (typically 21°C (70°F)) for use in design. b. FWD t esting on PCC pavements must consider the temperature at the time of the testing. Ideally, testing should be performed at a time (typically night or early morning hours) when the slab is flat (i.e., when the slab edges or center are not lifted off the base ). However, this may be impractical for an agency that must test many kilometers (miles) of pavement every day. In general, deflection testing on PCC pavements should be conducted when the ambient temperature is below 27 °C (80 °F). While the backcalculati on procedures for PCC pavements do not currently incorporate temperature corrections, such measurements are useful in evaluating backcalculation results for PCC pavements. This information is useful to determine the potential for slab curing severity that may be affecting the results. Also , knowledge of the temperature conditions at the time of testing assists in evaluating LTE data. 16 c. Air and surface temperature should be recorded at each test location, and most FWD equipment has temperature sensors for obta ining such data. The daily average temperatures for five days preceding testing should also be obtained , main ly if the air and surface temperatures will be used to predict the mean pavement temperature. d. For rigid pavements, it is essential to obtain the temperature gradients. Such data can be obtained by drilling holes at various depths and measuring temperatures with thermometers. A minimum of three temperature readings, roughly correlated with the beginning, middle, and end of testing, should be obtained for smaller projects with shorter testing times. A recent Federal Highway Administration (FHWA) study [6] reg arding FWD data collection to conduct viscoelastic and dynamic analyses of deflection data recommended: a. The temperature of the asphalt concrete (AC) layer needs to be collected during the FWD testing , preferably at every 2 inches of depth of the AC layer. b. Either a single FWD run on an AC layer with a large temperature gradient or FWD runs at different temperatures can be sufficient to compute the relaxation modulus, E ( t ) and the dynamic modulus, | E* | master curve of asphalt pavements. c. Tests should be conducted at a minimum of two different temperatures, preferably 18 °F or more apart. FWD data collected at a set of temperatures between 68 and 104 °F will maximize the accuracy of backcalculated E(t) or | E* | master curve up to less than 1 0 - percent error. d. For backcalculation, using a single FWD test dataset at a known AC temperature profile, the FWD test should be conducted under a temperature gradient of preferably ±9°F or more. 17 e. Either temperature variation with depth needs to be measure d (and included in the analysis) or the FWD test (with multiple pulses) needs to be run at different pavement temperatures (e.g., different times of the day) to obtain the time - temperature shift factor coefficients. The above recommended FWD measurement g uidelines most ly cover the seasonal or diurnal variations of pavement temperature and moisture so that the influence of these variations on material properties may be quantified by using backcalculation analyses. 2.4 Evaluating Functional Performance of an Ex isting Pavement One of the essential elements of any pavement management system is a means to measure the performance of the pavement system in terms of surface roughness, distresses, and other properties. The use of devices that measure the longitudinal profile of the pa vements to assess its surface roughness is common in most of the pavement management systems. When longitudinal profile measurements are used to determine the functional condition of the road surface , these are always summarized by an index that reduces th e thousands of elevation values into a single value. The IRI was developed in research sponsored by NCHRP and the World Bank and is the most broadly used index [38, 39] . However, no matter which index is calculated from a longitudinal profile, the quality of the information is only as good as the profile measurement [2]. 2.5 Environmental Effects on IRI Measurements Similar to surface deflections measured by FWD, diurnal and seasonal temperature and moisture variations also influence the longitudinal profile me asurements. As compared to flexible pavements, this effect can be more pronounced for rigid pavements due to curling and warping 18 of slabs. Concerns for measuring longitudinal profiles related to the environmental variables (i.e., temperature and moisture), as documented in an NCHRP report, include several aspects of the pavement surface shape that confound profile measurements, such as [2] : a. Transverse, daily, and seasonal variations in profile all combine to make an individual measurement a mere sample of t he road shape. b. The lateral position of the measurement has a strong influence on the longitudinal profile because the pavement surface shape changes across the lane. c. In PCC pavements, roughness variations of 10 percent are common over a 24 - hour cycle. d. T hin asphalt pavements over a granular base are subject to large temporary increases in roughness in winter caused by frost heave. A n FHWA study examined the roughness and roughness progression of 2 1 rigid pavement test sections on the LTPP Specific Pavemen t Studies - 2 (SPS - 2) experiment site in Arizona over the first 16 years of the experiment. The site included 12 test sections from the standard experiment and nine supplemental test sections selected by the Arizona Department of Transportation [3]. The find ings of the investigation showed that: a. Traditional profile analyses revealed roughness is caused by transverse and longitudinal cracking on some test sections and some localized roughness caused by built - in defects. b. Detailed profile analyses showed that curl and warp contributed significantly to the roughness of many of the test sections. c. Surface roughness did not increase steadily with time because of diurnal and seasonal changes in slab curl and warp. To better understand the effects of curling and war ping, the study documented an objective profile analysis method known as the Second Generation Curvature Index (2GCI) analysis 19 procedure for quantifying the level of curl and warp on each section. Automated algorithms estimated the gross strain gradient ne eded to deform each slab into the shape present in the measured profile in terms of a pseudo strain gradient (PSG) value. The average PSG values summarized the levels of curl and warp within each profile. For the JPCP test sections, variations in average P SG over time explained many of the changes in roughness over time. Such changes include diurnal variations in slab curl. The overall curl and warp level s increased throughout the life of the pavements with corresponding increases in the roughness [3, 40, 4 1] . According to the study, the roughness on a JPCP has two components; (a) a curvature - related component (i.e., due to curling/ warping), (b) a non - curvature - related component (i.e., due to pavement distresses). Based on the procedure developed, the diurn al changes in slab curvatures were analyzed in terms of PSG and plotted in a global curvature plot, as shown in Figure 10 , which demonstrates the effect of diurnal measurements on the shape of the slab, hence its roughness. According to the study, the diu rnal impacts of slab curling on a Half - car Roughness Index can be as high as 0.63 m/km averaging around 0.16 m/km. This finding makes it essential to emphasize the time of roughness measurement in specifications, primari ly where agencies use incentive - disi ncentive specifications. The diurnal impacts on roughness can also affect maintenance programming as it is likely that the estimated pavement functional condition may vary significantly, depending on survey timings and pavement curling characteristics. The study also reports s ignificant diurnal variations in joint edge geometry (see Figure 11 ) [40] . In a recent study, a new method of separating the curvature - and non - curvature - related IRI have been proposed using only a single profile measurement. The new a pproach builds on the existing 2GCI analysis procedure; however, it eliminates the need for establishing linear regression between IRI and PSG, which requires multiple surveys of the same pavement section 20 with slabs subjected to different thermal gradients . The study also showed the diurnal and seasonal variation in IRI for a JPCP section, as shown in Figure 12 , where the diurnal variation in IRI measured on a particular day is as significant as the seasonal variation assessed from the dataset [42] . Figure 10 Diurnal curvature analysis [40] Figure 11 Joint functionality analysis showing diurnal effects [41] Another investigation on flexible pavement longitudinal profiles in Michigan showed the effects of seasonal variations on profile measurements, especially frost action [1]. According to the study, pavements subject to frost action are rougher in the winters as opposed to summers. On the other hand, a pavement not exposed to frost action behaves oppositely; it tends to be 21 rougher in summer than in winter. Moreover, the study results illustrated that using a longitudinal profile has far greater utility than simply determining pavement surface roughnes s . The winter and summer longitudinal profiles of three flexible and composite pavements were used to determine the cause of deterioration, appropriate maintenance, rehabilitation, and reconstruction treatments to improve ride quality. Figure 12 Pavement IRI plotted against curvature index for LTPP section 04 - 0215 (Based on profiles collected in 1997) [42] 2.6 Summary Literature shows that there is an essential influence of the seasonal and diurnal FWD and longitudinal profile measurements, which ultimately translates into the parameters obtained from these measurements: IRI and the different pavement layer s moduli. Consideration of such effects is essential to know the pavement condition accurately. Also, the correct interpretation of the IRI will help agencies manage the pavements systems more effectively. Thus, there is a need to evaluate and quantify these effects. Also, based on the analysis, formulation of the general guidelines for the FWD and profile measurements are needed, which could help agencies in better interpretation of the obtained deflections and profiles. 22 CHAPTER 3 DATA EVALUATION The d ata to be used for this work should have multiple FWD and longitudinal profile measurements, both within a day to evaluate the diurnal effects and over a year for the evaluation of seasonal effects. For example, the needed data should include multiple FWD and profile measurements for the same sections during a day to capture morning (i.e., before noon) and afternoon temperature variations and the impact of such change s on slab curling for rigid pavements . Therefore, data from the LTPP SMP study were used to accomplish the objectives of this thesis. The d ata were evaluated to (a) identify the required data elements, (b) quantify the various attributes to determine its suitability and appropriateness. The three major tasks set up to accomplish the objectives are summarized below. a. Task 1 : Identify the data elements required for accounting for the effect of temporal variations on FWD and longitudinal profile measurements. b. Task 2 : Review the contents of the LTPP SMP study database and discuss the availability of the data elements identified in Task 1. c. Task 3 : Analyze the data, determine the findings, and recommend guidelines related to FWD and longitudinal profile measurements. 3.1 Tas k 1: Identification of Data Elements Temperature, moisture , and freeze/thaw related changes in pavement layers within a day and over a year can have a significant impact on the structural characteristics of pavement layers. Such variations in temperature and moisture can influence the material properties of pavement layers and , therefore, affect the response of the pavement structure under traffi c loads, and ultimately the life of the pavement. However, the magnitude and relationship of these effects are 23 not fully understood, making them difficult to address with any degree of confidence, in pavement design and evaluation. The LTPP SMP study can o vercome this limitation . The primary objective of the SMP study is to provide data needed to attain a fundamental understanding of the magnitude and impact of temporal variations in pavement response and material properties due to the separate and combined effects of temperature, moisture and frost/thaw variations. The SMP experiment data and subsequent analysis can provide [43] : a. "The means to link pavement response data obtained at random points in time to critical design conditions; b. The means to validate models for relationships between environmental conditions (e.g., temperature and precipitation) and in situ structural properties of pavement materials; and c. Expanded knowledge of the magnitude and impact of the changes involved." Parallel to the above go als, the SMP data may also assist in developing guidelines for FWD and longitudinal profile measurements to account for seasonal and diurnal temperature and moisture variations. The LTPP SMP test sections were instrumented and monitored temperature and mo isture at a higher rate than the regular measurement intervals for deflections, distresses, and longitudinal profiles. Table 1 shows the list of devices that were used to instrument the SMP sections. The table also indicate s the data item that was measured by each device for sections in the SMP study [44] . The critical elements of the monitoring plans for the SMP study include [43] : a. Deflection basin testing to evaluate temporal variations in structural properties b. Load transfer testing on joints and cracks i n rigid pavements for monitoring load transfer conditions 24 c. Joint faulting and joint opening measurements, for determining the effects of temperature variations on joint condition d. Surface elevation measurements for evaluating the effects of frost heave and swelling soil e. Transverse and longitudinal profile measurements for characterizing pavement rutting and roughness f. Distress surveys for monitoring the progression of pavement distress over time. g. In - situ a mbient temperature and precipitati on measurements over time h. Subsurface temperature s and moisture conten ts with depth over time i. Frost and thaw depth measurements, where applicable, for defining changes in support conditions over time Table 1 Devices used for measurements specific to SMP tes t sections [44] Measurement Device Measured Data Item Time - Domain Reflectometry Subsurface moisture changes Thermistor Probes Subsurface temperature changes Electrical Resistivity Frost/thaw depth Piezometer Groundwater table determination Air Temperature Probes Ambient temperature Tipping - Bucket Rain Gauge Precipitation For evaluating the temporal (i.e., diurnal temperature and seasonal temperature/ moisture) effects of FWD and longitudinal profile measurements on derived pavement parameters, e.g., pavement layer moduli or IRI, the first step was to identify the required data elements . Based on the discussion presented earlier in this section, several data elements, needed to accomplish the objective of this thesis, were identified in the LTPP SMP study database. Of particular interest were the pavement structure data, pavement site - specific data, over time monitored performance data, and pavement deflection data along with derived parameters. Additionally, the longitudinal profile d ata measured overtime for all the flexible and JPCP sections included in the LTPP SMP study. The LTPP Standard Data Release (SDR) 33 (the most up - to - date version of the available 25 data) was examined and used in this study. The identified data elements were extracted from the LTPP InfoPave , imported into a Microsoft Access ® database, and stored in different data tables. These tables were set up as a relational database so that these can be manipulated and linked together for various analysis pur poses. These data elements are briefly listed below and summarized in Table 2 , along with the used LTPP database tables and description of the contained data within the tables: a. Pavement structure details: cross - sections, age, and material types for all pav ement layers b. Pavement site - specific data: ambient temperatures, sub - surface temperatures, sub - surface moisture, and precipitation data , and the climatic regions c. Measured longitudinal profile data, testing dates, and timings d. Monitored FWD deflection data, w ith air and surface temperatures, temperature gradients, testing dates, and timings e. Monitored performance data over time (IRI, LTE) f. FWD testing based derived pavement parameters ( backcalculated layer moduli and k - values) with testing dates and timings 26 T able 2 LTPP database tables Data Elements LTPP Table Description Layer no., type, representative thickness, and material types SECTION_LAYER_STRUCTURE The table contains a consolidated set of pavement layer structure information for all LTPP test sections. State code, SHRP ID, experiment name, no., assign date , const no, const. no. change reason, and de - assign date EXPERIMENT_SECTION It is the master control table for all test sections and project sites included in the LTPP database. Precipitatio n and Climatic Regions TRF_ESAL_INPUTS_SUMMARY The contents of this table include average annual precipitation and freeze index, LTPP experimental climate region, and the source for this classification. Subsurface moisture content SMP_TDR_AUTO_MOISTURE This table contains the volumetric and gravimetric moisture contents from TDR. SMP_TDR_DEPTHS_LENGTHS This table contains information on the physical characteristics of the TDR probes, including the depth, the length of the probe, and its installation date. Subsurface temperature SMP_MRCTEMP_AUTO_HOUR This table contains most of the subsurface average hourly temperature data at a series of depths. SMP_MRCTEMP_DEPTH This table contains the installation depths for each tempe rature probe at an SMP sectio n and the date of installation. Mean IRI, visit date, visit no., run no., start time, cloud conditions, air temperature, average speed MON_HSS_PROFILE_SECTION High - speed profile computed parameters and statistics based on a 150 mm interval. MON_HSS_RUN_NO Identification of each high - speed survey run during each visit Test date, time, deflection unit identifier, point location, drop height, load, peak deflection sensor, layer temperature gradient, and depths MON_DEFL_DROP_DATA Peak deflection, peak load, and other drop - specific FWD measurements MON_DEFL_TEMP_DEPTHS It contains the l ocation and depth at which temperature gradient measurements wer e performed during FWD t esting. MON_DEFL_TEMP_VALUES In - pavement temperature gradient measurements obtained during FWD testing L oad T ransfer E fficiency (LTE) MON_DEFL_LTE This table contains the load transfer efficiency (LTE) computed parameter. LTE is computed from FWD measurements at transverse joints and cracks on PCC pavements Test date, FWD pass, average modulus, backcalculated layer number, average k - value, std. k - va lue BAKCAL_MODULUS_SECTION_LAYER Backcalculated modulus values averaged for each FWD pass. BAKCAL_BEST_FIT_SECTION_MASTER Best - fit back - calculation quality measures and other non - layer specific information for each FWD pass for PCC surfaced sections only 27 3.2 Task 2: Data Availability and Extents The purpose of initiating the LTPP SMP study was to measure the effects of diurnal and seasonal temperature and moisture variations on pavement structures and its response to loads. The original SMP design included 64 LTPP test sections arranged in a facto rial design that covered different pavement types, subgrade types, and the climate regions. Table 3 shows the design setup under the SMP study. Planning included 48 flexible and 16 rigid sections to be part of this study. The flexible pavements were divide d into two categories based on HMA thickness; thick HMA pavements with a layer thickness of greater than five inches (125 millimeters) and thin with fewer than five inches HMA thickness. Three flexible pavement test sections each were planned for every com bination of the pavement and subgrade soil type (i.e., fine and coarse) in each of the four climate regions (i.e., dry, freeze (DF), dry, no - freeze (DNF), wet, freeze (WF) and wet, no - freeze (WNF)). Moreover, one rigid pavement test section for the two pav ement types (i.e., jointed reinforced concrete pavement (JRCP) and JPCP) was planned for each combination of subgrade soil type and climate region [43] . Table 3 Original SMP experiment design [43] Pavement Type Subgrade Soil No Freeze Freeze Dry Wet Dry Wet Flexible Thick AC (> 5 - inch surface) Fine 3 3 3 3 Coarse 3 3 3 3 Flexible Thin AC (< 5 - inch surface) Fine 3 3 3 3 Coarse 3 3 3 3 JPCP Fine 1 1 1 1 Coarse 1 1 1 1 JRCP Fine 1 1 1 1 Coarse 1 1 1 1 Note: Number in a cell indicates the desired number of pavement sections. Task 2 had two parts; one was to evaluate the availability of the data in terms of the sections currently present in the LTPP database as part of the SMP study, while the second was to evaluate the extent of the extracted data for the critical data element identified in Task 1. 28 As per SDR 33, the SMP study data in the LTPP database has 67 flexible and 22 rigid pavement sections, including six JRCP sections. Table 4 shows the number and distribution of the flexible pavement sections by HMA thickness, subgrade type, and climatic zone. The table also shows the number and distribution of JPCP pavement sections within each climatic zone. T he table excludes the six JRCP pavement sections as the analysis undertak en in this thesis excluded them . Also, a significantly fewer number of sections are available in the dry climatic regions as opposed to wet climates. The same is true while comparing the available number of thin HMA flexible pavement sections with thick HM A sections. Table 4 Available pavement sections in the LTPP SMP database Pavement Type Subgrade DF DNF WF WNF Total Flexible - Thick (HMA > 5 inches) Fine 4 - 3 4 11 Coarse 4 6 26 6 42 Subtotal 8 6 29 10 53 Flexible Thin (HMA < 5 inches) Fine - - 2 3 5 Coarse 1 1 2 5 9 Subtotal 1 1 4 8 14 Total Flexible 9 7 33 18 67 JPCP 2 2 6 6 16 After the availability of the pavement sections in the LTPP SMP study database was evaluated , the next part of Task 2 was to assess the extent of the available data . The later part of this section presents the assessment of the extents of the data for each data element identified in Task. It is noteworthy that the precipitation data confounds with in th e climatic region s. 3.2.1 Pavement Cross - Sections Pavement cross - section is one of the vital property which has a significant bearing on its performance. Table s 5 and 6 show the descriptive statistics regarding the pavement structures within each climatic region for the 67 flexible and 16 JPCP pavement sections currently available within the LTPP SMP study, respectively. 29 Table 5 Layer thicknesses of flexible pavement sections located in different climates Layer Type Thickness Statistics, in DF DNF WF WNF Available Sections 9 7 33 18 HMA Average 5.69 6.43 7.18 5.59 Std* 2.46 2.05 2.41 3.00 Minimum 2.80 4.40 1.40 1.00 Maximum 10.70 11.00 14.30 11.30 Base Average 10.40 14.39 13.36 10.66 Std 5.41 7.48 6.38 4.64 Minimum 5.40 6.30 4.00 4.80 Maximum 22.80 24.00 25.80 24.00 Subbase Average 21.82 24.00 24.19 20.78 Std 6.13 0.00 5.78 10.31 Minimum 2.50 24.00 7.80 1.60 Maximum 24.00 24.00 66.00 39.00 * Standard deviation. 3.2.2 FWD Based Parameters The primary objective of the data mining was to obtain the availability of FWD deflection based parameters during a given day and in different seasons for all the SMP pavement sections. T his section presents the data extents of all such critical pavement p arameters identified in Task 1 . These parameters include the backcalculated HMA, base, and subgrade layer moduli for the SMP flexible pavement sections and PCC layer moduli, k - value, and the LTE for SMP JPCP sections. The BAKCAL tables of the LTPP database were used to summarize the available number of backcalculated moduli values for each pavement layer within the pavement structure. Table 7 shows the available FWD backcalculated moduli values for flexible pavement sections by the number of passes on the same day. The data includes FWD tests performed between the years 1989 to 201 2 . An FWD pass serves to distinguish multiple runs of the same lane number on the same day [44] . The results show that several flexible pavement sections contain multiple F WD passes on the same day. Most pavement sections have at least three passes per day. A few test sections have more than five FWD passes on the same day. 30 Table 6 Layer thicknesses of JPCP pavement sections located in different climates Layer Type Thicknes s Statistics, in DF DNF WF WNF Available Sections 2 2 6 6 PCC Average 10.70 10.16 9.79 9.34 Std* 0.74 1.07 1.19 0.97 Minimum 10.20 8.80 8.40 8.00 Maximum 11.80 11.00 11.80 11.20 Base Average 4.69 5.62 7.22 6.68 Std 1.02 0.87 4.39 2.47 Minimum 4.00 4.50 2.50 1.50 Maximum 6.20 6.30 14.40 9.30 Subbase Average 20.50 5.80 9.36 6.13 Std 0.00 0.00 4.80 1.95 Minimum 20.50 5.80 5.90 3.80 Maximum 20.50 5.80 16.00 8.00 * Standard deviation To further explain the table, consider the DF climatic zone as an example. There are 3 40 FWD test d ays with only one pass per day , i.e., a total of 3 40 FWD test runs involving all the nine pavement sections (number in parentheses). Out of those 3 40 test da ys , there were 2 67 test da ys where deflections on the section were measured for a second time, i.e., a second F WD pass on the same day; 2 30 test da ys where a third FWD pass measured deflections on the same day; 1 04 test da ys involving a fourth FWD pass on the same day. On the other hand, th ere were only 1 4 test da ys, which in clud ed only four of the nine sections where deflections measurement took place for a fifth time on the same day. Pavement age and maintenance history of the pavement structure are critical factor s that could affect pave ment deflection response or roughness. Th e construction number assigned to a pavement section in the LTPP database reflect s the later one. The construction number identifies changes in the pavement structure caused by the application of maintenance or reha bilitation treatments. When a test section first enters the LTPP program, it is assigned a construction number of 1. The construction number is then incremented by 1 for each subsequent maintenance or rehabilitation event regardless of its impact on the pa vement structure. For 31 example, crack sealing causes the generation of a new construction event, even though it does not c reat e a significant change in the experiment assignment or pavement structure [44] . It is , therefore , essential to consider the constru ction number of the pavement section at the time of the FWD test and the corresponding measured deflections used for back - calculating pavement structural parameters. Table 7 Available backcalculated pavement structural parameters by FWD pass SMP flexible sections FWD Passes DF DNF WF WNF 1 340 (9) 207 (7) 779 (33) 599 (18) 2 267 (9) 160 (5) 445 (23) 458 (17) 3 230 (9) 139 (5) 349 (14) 318 (17) 4 104 (9) 59 (5) 150 (13) 114 (17) 5 14 (4) 17 (5) 4 (3) 23 (8) Note: Values in parenthesis are the number of sections. Based on the discussion above, the available data presented in Table 7 were subdivided based on the construction number assigned to each flexible pavement section in the LTPP database. The maintenance and rehabilitation activities that cause a change in the construction number assigned to the pavement sections have been divided into four ca tegories (see Table 8 ). Table 8 Maintenance work categorization details Maintenance Category Description 0 No maintenance at all 1 Maintenance work with no significant effect on FWD deflections or profile measurements 2 Localized patchwork 3 Maintenance works with a potential effect on pavement structure or response, i.e., deflections or IRI Table 9 shows the distribution of the available backcalculated structural parameters of the SMP flexible pavements by the maintenance category based on the assigned construction number for the different climate regions. For example, referring to Table 8 , there are 340 backcalculated pavement structural parameters in the DF region for a single pass of the FWD test. Among these 340 measurements, 233 took pl ace while the pavement sections had a construction number of 1 32 (newly constructed or maintenance category 0) 8 1 FWD measurements were undertaken after some minor maintenance works. (category 1; i.e., crack seal, fog seal, etc.) . 18 measurements took place after localized maintenance/ rehabilitation works (category 2; i.e., patchworks) while, the last eight happened after a major maintenance/rehabilitation work (category 3; i.e., overlays or full - depth patchworks, etc.). Table 9 Available backcalculated pavement structural parameters by FWD pass and maintenance category in different climatic regions SMP flexible pavements Maintenance category FWD Pass DF DNF WF WNF 0 1 233 (8)* 117 (4) 382 (28) 381 (17) 2 192 (7) 98 (4) 284 (11) 284 (14) 3 167 (7) 80 (4) 228 (11) 208 (13) 4 77 (7) 36 (4) 97 (11) 77 (13) 5 8 (3) 15 (4) 3 (2) 12 (5) 1 1 81 (5)* 53 (4) 54 (5) 130 (6) 2 61 (4) 48 (4) 42 (3) 119 (6) 3 54 (4) 47 (4) 36 (3) 82 (6) 4 25 (4) 15 (3) 17 (3) 31 (5) 5 6 (1) - 1 (1) 10 (2) 2 1 18 (3)* 17 (1) 92 (6) 21 (4) 2 14 (3) 14 (1) 53 (3) 13 (3) 3 9 (2) 12 (1) 44 (3) 1 (1) 4 2 (1) 8 (1) 18 (3) 5 - 2 (1) - - 3 1 8 (4)* 20 (3) 251 (29) 67 (8) 2 - - 66 (14) 42 (3) 3 - - 41 (5) 27 (3) 4 - - 18 (4) 6 (1) 5 - - - 1 (1) Note: Values in parenthesis are the number of sections. *233 + 81 + 18 + 8 = 340 Another aspect that required consideration while data mining was the distribution of the data over different months, necessary for the investigation of the seasonal effects. Similarly, the daily distribution of the data with time was essential to look at the diurnal effects. T he number of the available backcalculated moduli within each month of the year was determined f or each pavement se ction. 33 Table 10 shows the monthly distribution of the backcalculated layer moduli data for the flexible pavement sections. The data distribution was irrespective of the FWD pass number, maintenance category, and the year of measurement. Table 10 also shows that most of the sections have multiple FWD measurements within eve ry month of the year. Similarly, the section - wise hourly distribution of data was also determined (see A ppendix). The majority of the available FWD measurement s are between 8:00 am and 4:00 pm. Preparing such data distribution helped in identifying differe nt factors and the number of levels within each factor, which will be discussed later in the text. Similar to flexible pavement sections, FWD deflections based parameters were summarized to isolate the effect of diurnal and seasonal measurements on rigid p avement sections (i.e., E and k - values). Table 11 shows up - to - date available data for rigid pavement sections of the LTPP SMP study that are extracted from the BAKCAL tables of the LTPP monitoring module database. The data includes FWD tests performed duri ng the years 1989 to 2012. The table shows the available FWD measurements by different passes on the same day. As mentioned earlier, the FWD pass serves to distinguish multiple runs of the same lane on the same day [44] . The numbers in parentheses indicate the number of sections involved. The tables show that several rigid pavement sections contain multiple FWD passes on the same day, with the majority having three passes a day. Note that the number of LTE values available for the same number of SMP rigid p avement sections is double of the values shown in Table 11 since each LTE measurement requires two deflection measurements (i.e., J4 and J5). 34 Table 10 Monthly data distribution of backcalculated pavement parameters - SMP AC sections Climate region State State code & section ID Month Total 1 2 3 4 5 6 7 8 9 10 11 12 DF Colorado 8_1053 12 8 23 13 10 6 3 6 7 7 12 11 118 DF Idaho 16_1010 4 9 16 15 8 8 4 5 5 6 10 6 96 DF Montana 30_0114 9 13 13 1 25 18 12 91 DF Montana 30_8129 2 3 23 28 13 7 7 9 8 13 9 4 126 DF Nevada 32_0101 14 7 19 6 15 7 9 10 15 3 15 7 127 DF Saskatchewan 90_6405 6 14 11 16 8 7 7 10 8 87 DF South Dakota 46_0804 17 26 21 11 16 12 17 5 12 137 DF South Dakota 46_9187 3 4 7 12 6 8 8 4 8 7 5 2 74 DF Wyoming 56_1007 11 9 16 9 8 6 3 12 5 4 11 7 101 DNF Arizona 4_0113 6 22 9 19 9 15 9 20 4 12 9 14 148 DNF Arizona 4_0114 6 22 8 17 8 15 7 18 17 10 15 143 DNF Arizona 4_1017 3 1 2 2 1 1 10 DNF Arizona 4_1018 1 1 1 2 1 6 DNF Arizona 4_1024 8 7 9 7 7 6 8 10 4 3 5 7 81 DNF New Mexico 35_1112 3 11 7 11 11 9 13 9 6 6 9 8 103 DNF Utah 49_1001 6 11 12 20 3 3 6 4 6 9 11 91 WF Connecticut 9_1803 9 4 23 17 13 9 5 5 5 9 8 6 113 WF Maine 23_1026 3 7 15 15 11 7 10 12 7 10 7 104 WF Manitoba 83_1801 4 28 23 8 21 10 14 12 13 133 WF Manitoba 83_3802 1 1 2 WF Massachusetts 25_1002 7 6 20 18 11 17 7 3 8 2 5 6 110 WF Minnesota 27_1018 10 11 16 10 10 13 8 8 9 95 WF Minnesota 27_1028 5 9 8 13 8 8 5 5 8 10 3 82 WF Minnesota 27_6251 14 28 30 9 19 11 26 13 16 3 169 WF Nebraska 31_0114 13 8 3 13 1 16 11 9 10 10 3 97 WF New Hampshire 33_1001 3 11 11 14 14 8 6 9 10 6 3 95 WF New Jersey 34_0501 1 2 1 1 1 1 2 3 2 14 WF New Jersey 34_0502 1 2 2 1 1 1 3 3 14 WF New Jersey 34_0503 1 2 3 1 2 1 3 3 2 18 WF New Jersey 34_0504 1 2 3 2 1 1 2 3 1 16 WF New Jersey 34_0505 1 2 3 2 1 1 2 2 1 15 WF New Jersey 34_0506 1 2 4 1 2 1 2 3 2 18 WF New Jersey 34_0507 1 2 4 1 2 1 3 3 2 19 WF New Jersey 34_0508 1 2 3 1 2 1 2 3 2 17 WF New Jersey 34_0509 1 2 3 1 2 1 2 3 2 17 WF New Jersey 34_0559 1 1 1 2 1 1 3 2 1 13 WF New Jersey 34_0560 1 2 2 1 1 1 3 3 1 15 WF New Jersey 34_0901 2 1 1 2 2 2 10 WF New Jersey 34_0902 2 2 1 2 1 2 10 WF New Jersey 34_0903 2 1 1 2 1 2 9 WF New Jersey 34_0960 2 1 1 2 1 2 9 WF New Jersey 34_0961 2 2 1 2 1 1 2 11 WF New Jersey 34_0962 2 3 1 2 1 1 2 12 WF New York 36_0801 4 16 21 10 19 5 15 15 13 10 11 9 148 WF Ohio 39_0901 16 16 10 21 5 10 5 32 11 6 6 138 WF Ontario 87_1622 4 16 11 8 8 4 9 6 1 67 WF Quebec 89_3015 2 1 1 4 WF Vermont 50_1002 2 3 15 25 11 22 6 14 13 11 17 5 144 WNF Alabama 1_0101 7 5 10 8 9 8 6 6 6 11 9 6 91 WNF Alabama 1_0102 4 4 10 8 9 6 4 6 5 11 8 7 82 WNF Delaware 10_0102 4 6 1 8 2 3 5 6 2 13 4 4 58 WNF Georgia 13_1005 6 6 6 4 5 4 7 8 14 10 6 76 WNF Georgia 13_1031 8 12 3 11 10 5 3 12 12 15 8 9 108 WNF Maryland 24_1634 7 10 2 30 7 1 3 1 5 7 3 7 83 WNF Mississippi 28_1016 1 2 4 4 4 5 5 5 12 4 4 50 WNF Mississippi 28_1802 4 6 4 6 6 6 7 4 8 13 10 5 79 WNF North Carolina 37_1028 10 7 9 9 9 7 4 7 9 5 4 80 WNF Oklahoma 40_4165 3 5 12 6 16 6 5 6 2 2 6 3 72 WNF Texas 48_1060 11 9 9 7 10 10 7 7 6 4 7 8 95 WNF Texas 48_1068 10 6 9 8 9 10 6 9 8 1 10 10 96 WNF Texas 48_1077 10 9 10 8 8 10 7 7 9 11 7 10 106 WNF Texas 48_1122 11 12 10 10 10 16 13 11 9 7 14 13 136 WNF Texas 48_3739 7 6 14 12 9 7 11 8 7 8 6 14 109 WNF Virginia 51_0113 5 7 11 9 5 9 7 4 5 12 9 7 90 WNF Virginia 51_0114 12 9 16 7 7 9 12 3 14 14 7 11 121 WNF Washington 53_3813 1 1 35 Table 11 Available backcalculated pavement structural parameters (E and k - values) by FWD pass in different climatic regions SMP JPCP section s FWD passes DF DNF WF WNF 1 54 (2) 73 (2) 153 (6) 131 (6) 2 43 (2) 59 (2) 105 (5) 85 (4) 3 14 (2) 23 (2) 37 (5) 18 (3) 4 2 (1) 1 (1) 2 (2) 2 (2) Note: Values in parenthesis are the number of sections. As for flexible pavement sections, the available data for JPCP pavement sections were also summarized based on the assigned construction numbers and the maintenance activities that these pavement sections have undergone. Table 12 shows the distribution of the av ailable backcalculated structural parameters in the LTPP database by the FWD pass and maintenance category for the different climate regions. For instance, referring to Table 11 , there are 54 backcalculated pavement structural parameters in the dry freeze climate region for a single pass of the FWD test. Table 12 shows that among the 54 measurements : 22 measure ments happened while the pavement sections did not undergo any maintenance activity , 18 took place after some minor maintenance (i.e., crack seal) , w hile 14 took place after significant maintenance/ rehabilitation work (i.e., PCC slab replacement and full - depth patchwork). Table 12 Available backcalculated pavement structural parameters by FWD pass and maintenance category in different climatic region s SMP JPCP section s Maintenance category FWD pass DF DNF WF WNF 0 1 22 (1)* 61 (2) 81 (6) 130 (6) 2 19 (1) 51 (2) 54 (4) 85 (4) 3 8 (1) 23 (2) 16 (3) 18 (4) 4 - 1 (1) - 2 (2) 1 1 18 (2)* 12 (1) 16 (1) - 2 12 (1) 8 (1) 13 (1) - 3 2 (1) - 11 (1) - 4 - - 1 (1) - 3 1 14 (1)* - 56 (4) 1 (1) 2 12 (1) - 38 (3) - 3 4 (1) - 10 (3) - 4 2 (1) - 1 (1) - Note: Values in parenthesis are the number of sections. * 22 + 18 + 14 = 54 36 The data shown in Table 11 was also stratified in a way to know its section - wise monthly and hourly distribution, as was done for flexible pavement sections. Table s 13 and 14 presents the discussed data distribution for the JPCP sections of the LTPP SMP database irrespective of the maintenance categor y, age, and measurement year. It can be observed from Table 13 that 12 out of the available 16 JPCP pavement sections have multiple FWD measurements (and the corresponding backcalculated parameters) in each month. Table 14 shows that the majority of the FW D measurements lie between 8:00 am and 2:00 pm. As mentioned earlier, such data distribution tables helped in deciding the number of levels of factors used in the analysis, which will be discussed later in the text. Table 13 Monthly data distribution of ba ckcalculated parameters - SMP J PC P sections Climate region State State code & section ID Month Total 1 2 3 4 5 6 7 8 9 10 11 12 DF Nevada 32_0204 4 3 5 4 2 4 2 2 2 3 1 3 35 DF Utah 49_3011 7 5 14 10 5 5 5 5 6 4 10 2 78 DNF Arizona 4_0215 3 10 10 12 6 10 5 14 3 12 6 13 104 DNF California 6_3042 3 4 5 6 5 4 6 5 2 2 6 4 52 WF Indiana 18_3002 5 3 2 5 3 1 3 2 8 4 3 1 40 WF Manitoba 83_3802 4 6 7 9 3 5 5 7 2 48 WF Nebraska 31_3018 3 8 11 3 11 10 11 8 8 14 1 88 WF Ohio 39_0204 3 4 4 7 5 4 5 3 2 2 39 WF Quebec 89_3015 1 2 10 11 6 11 8 4 10 2 10 3 78 WF South Dakota 46_3010 2 1 1 4 WNF Georgia 13_3019 6 6 8 10 8 5 2 9 10 9 6 2 81 WNF North Carolina 37_0201 6 5 6 7 15 2 15 4 7 10 7 7 91 WNF North Carolina 37_0205 2 1 2 5 WNF North Carolina 37_0208 1 2 1 1 2 7 WNF North Carolina 37_0212 2 2 2 1 2 9 WNF Washington 53_3813 4 2 2 4 2 2 7 4 2 4 4 6 43 37 Table 14 Hourly data distribution of backcalculated parameters - SMP J PC P sections Climate region State State code & section ID Hour of the day Total 7 8 9 10 11 12 13 14 15 16 17 19 DF Nevada 32_0204 1 1 8 8 4 8 4 1 35 DF Utah 49_3011 1 7 17 13 6 17 5 5 6 1 78 DNF Arizona 4_0215 2 16 20 20 20 13 7 4 2 104 DNF California 6_3042 1 6 13 9 8 9 5 1 52 WF Indiana 18_3002 7 11 7 5 6 3 1 40 WF Manitoba 83_3802 15 8 4 6 10 4 1 48 WF Nebraska 31_3018 2 10 20 20 13 16 4 3 88 WF Ohio 39_0204 3 2 9 9 4 6 4 1 1 39 WF Quebec 89_3015 1 5 15 16 15 12 11 2 1 78 WF South Dakota 46_3010 2 2 4 WNF Georgia 13_3019 4 10 11 15 4 17 6 5 6 3 81 WNF North Carolina 37_0201 2 23 15 14 28 6 2 1 91 WNF North Carolina 37_0205 1 2 1 1 5 WNF North Carolina 37_0208 2 1 1 2 1 7 WNF North Carolina 37_0212 1 1 1 1 3 1 1 9 WNF Washington 53_3813 1 1 7 12 8 9 4 1 43 3.2.3 Sub - surface Temperature Data The sub - surface temperature data from the SMP module were obtained for both the flexible and rigid pavement sections. For flexible pavements, the obtained data helped determine the HMA layer mid - depth temperature used to correct the moduli values. Table 15 shows the available data by the FWD pass that was measure d within 30 minutes of the FWD test. Again, more data in the wet regions are due to a larger number of sections in these regions. For rigid pavements, the temperature data were essential to calculate the temperature gradient (i.e., top minus bottom) for t he PCC slab. Temperature gradients were calculated within ±1.25 inches of the top or bottom surface of the PCC layer. The available gradients data were matched for each FWD based parameter according to the location of the FWD test (i.e., J1, J4/J5, etc.). Table s 16 and 17 show the details of the available temperature gradients data with the FWD pass of the day. 38 Table 15 Available HMA layer mid - depth temperature s on FWD measurement days SMP sections FWD Passes DF DNF WF WNF 1 539 (9) 301 (7) 1303 (30) 1149 (17) 2 389 (9) 229 (5) 557 (19) 819 (17) 3 323 (9) 200 (5) 445 (13) 566 (17) 4 132 (9) 98 (5) 193 (13) 211 (17) 5 15 (4) 28 (5) 7 (3) 48 (8) Note: Values in parenthesis are the number of sections. Table 16 Available temperature gradients on FWD measurement days matched with LTE values SMP JPCP sections FWD Passes DF DNF WF WNF 1 37 (1) 79 (2) 113 (6) 187 (6) 2 32 (1) 57 (2) 73 (5) 121 (3) 3 8 (1) 19 (2) 21 (2) 30 (3) 4 2 (1) 1 (1) 4 (2) Note: Values in parenthesis are the number of sections. Table 17 Available temperature gradients on FWD measurement days matched with backcalculated moduli values SMP JPCP sections FWD Passes DF DNF WF WNF 1 37 (1) 85 (2) 104 (6) 164 (6) 2 35 (1) 68 (2) 74 (5) 133 (3) 3 9 (1) 23 (2) 23 (4) 30 (3) 4 1 (1) 1 (1) 2 (1) Note: Values in parenthesis are the number of sections. 3.2.4 Longitudinal Profile Measurements The longitudinal profile is measured five times as an LTPP standard practice on every visit [44] at a pavement section . MON_HSS_RUN_NO is the LTPP table contains the calculated left wheel path, right wheel path, and the mean IRI value s for each run of a visit. Table 18 shows the overall availability of longitudinal profile measurements for all the SMP pavement sections (83 sections), excluding the six JRCP sections, where a single measurement is the mean IRI value of the five runs per visit. Similar to FWD deflection ba sed parameter availability, more longitudinal profile measurement data are available for pavement sections located in wet climates, mainly because of a higher number of pavement sections. Also, the data has been classified based on the maintenance category . 39 Table 18 shows that a ll 16 SMP JPCP sections (highlighted in bold) were subjected at some point in time to multiple profile measurements during the same day; on the other hand, only five flexible sections (highlighted in bold) had multiple profile measur ements during the same day. A higher number of multiple profile measurements on JPCP sections is logical since diurnal profile variations are significant only in rigid pavements (upward versus downward curling). While the number of multiple measurements du ring a day varies for each section, more data per day are available for rigid pavements. Table 18 Available profile measurements in different climatic regions SMP experiment Pavement type Cate - gory DF DNF WF WNF Visit number Visit number Visit number Visit number 1 2 3 1 2 3 1 2 3 1 2 3 AC 0 134 (9) 72 (4) 245 (29) 4 (1) 206 (16) 2 (2) 1 60 (5) 4 (2) 51 (6) 42 (5) 63 (6) 2 11 (3) 7 (1) 83 (7) 14 (4) 3 25 (6) 38 (4) 313 (26) 65 (8) Total by visit 230 (9) 4 ( 2 ) 168 (7) 68 3 (30) 4 ( 1 ) 3 48 (17) 2 ( 2 ) AC Total 234 (9) 168 (7) 687 (30) 350 (17) JPCP 0 11 (1) 1 (1) 1 (1) 51 (2) 24 (2) 1 (1) 71 (6) 16 (4) 5 (5) 142 (6) 39 (6) 8 (3) 1 22 (2) 6 (2) 15 (1) 5 (1) 1 (1) 13 (1) 12 (1) 2 1 (1) 3 10 (2) 3 (1) 1 (1) 72 (4) 15 (4) 5 (3) 6 (2) Total by visit 43 ( 2 ) 10 ( 2 ) 1 ( 1 ) 67 ( 2 ) 29 ( 2 ) 2 ( 2 ) 156 (6) 43 ( 6 ) 10 ( 6 ) 1 49 ( 6 ) 39 ( 6 ) 8 ( 3 ) JPCP Total 54 (2) 98 (2) 209 (6) 196 (6) Note: Values in parenthesis are the number of sections. 3.3 Summary To investigate the effects of seasonal and diurnal FWD deflections and profile measurements on the derived pavement parameters, the determination of the required data elements along with the 40 d ata extents, was critical. The desired data needed to have multiple FWD and profile measurements to achieve the objectives of this study. The LTPP SMP study database has the required data for each element identified with multiple FWD deflections and profil e measurements. This chapter presented a brief account of all the identified data elements, the LTPP data tables that contained the required data, and the extents of the availability of data for each of the data elements used in this study. More data are a vailable for flexible pavements (64 sections) as compared to JPCP (16 sections), primarily due to the number of sections for each pavement type in the database. 41 CHAPTER 4 DATA ANALYSIS The data analysis to accomplish the objectives of the thesis required looking into the effects of seasonal and diurnal FWD and profile measurements on the different deflection - and profile - based parameters. For each pavement type and data element, a relational database was prepared for the data analysis. The data an alysis and findings for each pavement type are presented in this chapter . 4.1 Analysis for Flexible Pavements (by climatic region) FWD deflection - based parameters for flexible pavements include the backcalculated layer moduli values. The backcalculated moduli for each pavement layer were obtained from the LTPP SMP database. Each of these parameters was analyzed separately within all climatic region s . Dividing the data and analysis by climatic regions helped to und erstand better the effects of seasonal and diurnal measurements on these parameters. Also, it aided in explaining the variance in the parameters . As compared to including all the climates together, it also helped reduce the data required while improving th e power of the analysis. Similarly, using available IRI data from the LTPP SMP database, the seasonal and diurnal effects on pavement roughness were investigated. The IRI data extracted from the database was analyzed similarly, dividing the data by climati c region, to gain the benefits as mentioned earlier. 4.1.1 FWD based Pavement Parameters HMA, base , and subgrade layer moduli are backcalculated using deflections measured with an FWD device. To investigate the effects of seasonal and diurnal FWD measurements on these moduli values, the main factors included (a) month and, (b) time of FWD measurement. HMA layer thickness is also used as a factor since it is a part of the SMP original design. The 42 maintenance category of the pavement sections is used as a blocking factor. Blocking is a technique that increases the precision in an experiment by reducing the experimental error variance. It is achieved by considering factors believed to affect the response; however, not considered to be of primary importance in the ana lysis. An analysis for investigating the seasonal and diurnal effects on each of the mentioned parameters is presented next. The data analysis for each data element involved the following steps: 1. The pavement structure details, layer moduli values, FWD meas urement dates and timings, maintenance history, pavement surface temperature, air temperature, and temperature gradient measurements are variables identified to perform the analysis. 2. The investigation of the problem at hand required an analysis involving a group of sections within each climatic region . 3. The flexible pavements sections in the SMP LTPP database were identified in each of the four climatic regions. Note that climatic regions are defined in terms of temperature (i.e., freeze/no freeze ), and moisture conditions (i.e., wet/dry). 4. The structural details, including layer types and thicknesses, were also obtained. 5. The backcalculated layer moduli, along with the date and time of FWD measurements, the air and pavement surface temperatures meas ured at the time of FWD testing, and the temperatures measured at different depths of the structure were obtained. Also, the maintenance history details for all the sections were also extracted from the database (i.e., construction no. and type of treatmen t) and categorized. 6. The data was arranged in a relational database. 43 7. The data were evaluated by using a histogram and boxplot to identify outliers. 8. The factors used in the analysis include; (a) measurement month discretized into four seasons (i.e., levels) to look at the seasonal effects, (b) measurement time with two levels (before noon and afternoon) for the diurnal effects, (c) maintenance category, (d) HMA layer thickness with two levels; thick (>5 inches) and thin (<5 inches) based on the original SMP e xperiment [43]. 9. Analysis of Variance (ANOVA) was conducted for flexible pavement sections within each climatic region to investigate the temporal effects on each of the layer moduli values. A level of significance of 5% ( = 0.05) is used in the analysis. H MA Layer Moduli To evaluate the effects of seasonal and diurnal FWD measurements, observing the general trends of the backcalculated HMA layer moduli is useful. Figure 13 shows the diurnal and seasonal HMA layer variations by FWD pass and months, respectiv ely. Figure 13 (a) indicates that the HMA layer moduli decrease with FWD pass (i.e., from morning to afternoon) for both thin and thick HMA layers. However, the data show somewhat mixed trends between different climatic regions, which could be because of va rying HMA mixtures and field aging on the pavement sections [see Figure 13 (b) and (c)]. As expected, the HMA layer moduli display significant variations with higher values in winter months and lower values in the summer months [ Figure 13 ( d )]. Also, there i s less overall variation in the backcalculated moduli values within the summer season (i.e., May, June, and July) than winters (December, January, and February). 44 (a) Overall by FWD pass (d) Overall by month (b) Thick HMA Layer by FWD pass (e) Thick HMA Layer by month (c) Thin HMA Layer by FWD pass (f) Thin HMA Layer by month Figure 13 Seasonal and diurnal variations in HMA layer moduli 45 Table 19 shows the descriptive statistics of the available HMA layer moduli values with the number (N) of measurements within each climatic region. The HMA moduli value considerably varies within each climatic region, especially in the DF, WF, and WNF regions. The variation can be explained by looking at the histogram and box - plot of the available data in Figure 14 . HMA moduli values over 3000 ksi exist in the database for the three climatic regions mentioned earlier, which is potentially a reason for higher standa rd deviations in these climatic regions. Table 19 Descriptive statistics HMA layer moduli SMP flexible pavements Climatic region N Mean St d. Minimum Q1 1 Median Q3 1 Maximum DF 957 1219.8 1009.9 113.1 613.9 940.9 1581.0 7817.9 DNF 582 1085.6 577.8 102.3 639.2 1015.2 1474.9 2805.2 WF 1739 1113.2 809.1 105.2 606.2 959.5 1387.6 7880.9 WNF 1533 1268.5 1259.5 109.9 438.6 789.8 1605.0 7634.9 Note: HMA moduli values shown are in ksi units. 1 1 st and 3 rd quartiles. (a) Histogram - Available HMA moduli data (b) Box - plot - Available HMA moduli data Figure 14 Visualizing available HMA layer moduli data SMP flexible pavements Figure 15 demonstrates the normality of the data for the DNF climatic region after a suitable transformation used in the analysis. This study excludes HMA modulus values higher than 2000 ksi , as these can potentially mask the findings. The satisfaction of the norma lity assumptions is essential to draw meaningful conclusions from the ANOVA analysis. Similarly, 46 the data were transformed adequately for the rest of the climatic regions to ensure the satisfaction of the normality assumptions. Figure 15 Evaluating norma lity of the HMA moduli data - D N F climatic region Table 20 demonstrates the results of the ANOVA for the DF climatic region . FWD measurement season, time, AC thickness, and the maintenance category (blocking factor) are significant factors affecting the HM A layer modulus values based on a type - = 0.05). However, the interaction (highlighted in bold font) between the FWD measurement season, time, and AC thickness also has a significant influence on the HMA layer moduli values. Thus, one should look at the interaction means plot rather than the main effects plot for interpreting the results for the analysis. 47 Table 20 ANOVA results for HMA moduli values DF climatic region Source DoF 1 Seq SS 2 Contribution Adj SS 3 Adj MS 4 F - v alue p - v alue Season 3 76.161 26.27% 73.624 24.5414 141.20 0.000 Time 1 12.279 4.23% 9.170 9.1697 52.76 0.000 AC_th 1 7.251 2.50% 1.358 1.3583 7.82 0.005 Maint. Cat. 3 28.197 9.72% 21.716 7.2386 41.65 0.000 Season*Time 3 2.122 0.73% 2.401 0.8005 4.61 0.003 Season*AC_th 5 3 19.957 6.88% 20.589 6.8630 39.49 0.000 Time*AC_th 1 0.781 0.27% 0.781 0.7814 4.50 0.034 Error 824 143.211 49.39% 143.211 0.1738 Lack - of - Fit 31 9.819 3.39% 9.819 0.3167 1.88 0.003 Pure Error 793 133.393 46.00% 133.393 0.1682 Total 839 289.960 100.00% 1 Degrees of freedom. 2 Sequential sum of squares. 3 Adjusted sum of squares. 4 Adjusted mean squares. 5 AC thickness. Figure 16 Interaction means plot - DF climatic region 48 (a) Mean HMA moduli difference for season and time interaction (b) Mean HMA moduli difference for the season and AC thickness interaction (c) Mean HMA moduli difference for time and AC thickness interaction Figur e 17 Multiple mean comparisons with 95% confidence intervals - DF climatic region 49 Diurnally, the HMA moduli values backcalculated using FWD measurements conducted before noon are higher than the afternoon measurements (see Figure 16 ). Such a difference can be explained due to the absorption of heat throughout the day by the pavement. There is a significant mean difference (around 20 0 ksi) between HMA backcalculated moduli values obtained from FWD measurements taken before - noon and in the afternoon between spring and summer seasons [see Figure 17 (a)]. Also, the time has a pronounced effect on HMA moduli values backcalculated from FWD measurements obtained over pavements with thick and thin HMA layers. Generally, t hin HMA layers tend to display higher moduli values as compared to thick HMA layers. Such a trend can be due to the back - calculation process that estimates higher stiffness va lues for thin HMA layers than thicker ones. Figure 17 (c) shows that the mean moduli difference is higher (260 ksi) for thick HMA layers than pavements with a thin HMA layer (170 ksi). Seasonally, HMA moduli values are generally highest for winters, as expected, followed by spring and fall seasons, while the summer season has the lowest moduli values (see Figure 16 ). A noticeable mean difference (around 400 ksi) exists between moduli values obtained in winter and summer seasons over pavements with different HMA layers. In winters, thicker HMA layers are stiff er than thin layers due to low temperatures , while in summers, the opposite is true. However, higher moduli values for thin HMA layers can also be due to the back - calculation process, which tends to calculate higher moduli values for thinner layers. A simi lar analysis for the DNF climatic region shows that the interaction of season and time influences the HMA moduli values. The AC layer thickness, on the other hand, does not have a significant effect (see Table 21 ). The maintenance category being the blocki ng factor also appears to contribute towards the variation of HMA moduli values in the DNF climatic region. 50 Winter season FWD measurements result in higher HMA moduli values as compared to other seasons. Lowest values are obtained once FWD deflections meas ured in the summer season are used to back - calculate the HMA layer moduli (see Figure 18 ). Also, the mean difference between the HMA layer moduli values based on before - noon and afternoon FWD deflections is significant (>170 ksi) among all seasons except f or fall (see Figure 19 ) in the DNF climatic region. Table 21 ANOVA results for HMA moduli values D N F climatic region Source DoF Seq SS Contribution Adj SS Adj MS F - v alue p - v alue Season 3 13479.0 40.55% 13289.5 4429.84 188.72 0.000 Time 1 809.4 2.44% 904.6 904.64 38.54 0.000 AC_th 1 307.5 0.93% 6.9 6.91 0.29 0.588 Maint. Cat. 3 6214.6 18.70% 6005.0 2001.68 85.28 0.000 Season*Time 3 224.8 0.68% 227.2 75.72 3.23 0.022 Season*AC_th 3 129.2 0.39% 128.2 42.73 1.82 0.142 Time*AC_th 1 7.4 0.02% 7.4 7.37 0.31 0.575 Error 514 12064.8 36.30% 12064.8 23.47 Lack - of - Fit 29 1117.1 3.36% 1117.1 38.52 1.71 0.013 Pure Error 485 10947.7 32.94% 10947.7 22.57 Total 529 33236.7 100.00% Figure 18 Interaction means plot - D N F climatic region 51 Figure 19 Mean HMA moduli difference for season and time interaction DNF climatic region Table 22 ANOVA results for HMA moduli values W F climatic region Source DoF Seq SS Contribution Adj SS Adj MS F - v alue p - v alue Season 3 34775.9 37.50% 13091.8 4363.93 132.47 0.000 Time 1 2467.8 2.66% 1322.7 1322.72 40.15 0.000 AC_th 1 251.2 0.27% 19.2 19.21 0.58 0.445 Maint. Cat. 3 1591.4 1.72% 2140.8 713.61 21.66 0.000 Season*Time 3 236.6 0.26% 162.6 54.19 1.64 0.177 Season*AC_th 3 1665.9 1.80% 1633.0 544.32 16.52 0.000 Time*AC_th 1 22.5 0.02% 22.5 22.53 0.68 0.408 Error 1570 51719.4 55.77% 51719.4 32.94 Lack - of - Fit 36 2648.1 2.86% 2648.1 73.56 2.30 0.000 Pure Error 1534 49071.3 52.92% 49071.3 31.99 Total 1585 92730.7 100.00% ANOVA for the WF climatic region shows that the interaction of season with HMA layer thickness has a significant effect on its stiffness (see Table 22 ). Time also plays a role in defining the variation of HMA moduli values. Figure 20 presents the interaction means plot for the season and HMA thickness interaction and the main effects plot for time. Figure 20 (a) shows that FWD deflections measured before - noon results in higher backcalculated HMA layer moduli values as compared to those measured in the afternoon. FWD deflections measured in the winter season show higher HMA moduli values as compared to the winter season FWD tests, with noticeable differences among HMA thick and thin layer pave ment sections. On the other hand, there is no 52 difference between tests conducted in the fall and spring seasons irrespective of HMA layer thickness [see Figure 20 (b)]. (a) Main effects plot (b) Interaction means plot Figure 20 Factorial plots ANOVA for HMA layer moduli data WF climatic region 53 The interactions between season and time and season and HMA layer thickness appeared to influence the HMA moduli values in the WNF climatic regions (see Table 23 ). Generally, winter seas on FWD measurements give the highest HMA moduli values, while summers provide the lowest. Fall and spring season FWD tests bear similar backcalculated HMA moduli values [see Figure 21 (a)]. Also, before - noon tests show higher HMA modulus values than afterno on ones. The interaction means plot shows a noticeable difference (> 130 ksi) between mean HMA moduli values for all seasons backcalculated from FWD tests conducted before - noon and afternoon except winter season [see Figure 21 (b)]. Table 23 ANOVA result s for HMA moduli values - WNF climatic region Source DoF Seq SS Contribution Adj SS Adj MS F - v alue p - v alue Season 3 126.701 21.61% 132.221 44.0738 142.42 0.000 Time 1 16.195 2.76% 17.220 17.2197 55.64 0.000 AC_th 1 8.013 1.37% 0.966 0.9665 3.12 0.077 Maint. Cat. 3 45.048 7.68% 43.764 14.5879 47.14 0.000 Season*Time 3 3.081 0.53% 3.135 1.0451 3.38 0.018 Season*AC_th 3 3.878 0.66% 3.933 1.3110 4.24 0.005 Time*AC_th 1 0.996 0.17% 0.996 0.9956 3.22 0.073 Error 1236 382.489 65.23% 382.489 0.3095 Lack - of - Fit 38 39.671 6.77% 39.671 1.0440 3.65 0.000 Pure Error 1198 342.818 58.46% 342.818 0.2862 Total 1251 586.400 100.00% Figure 21 (c) shows the mean comparison plot for the season and HMA layer thickness interaction. There is no significant difference between the HMA moduli values with a different layer thickness (i.e., thick or thin) in all three seasons except summer season. FWD de flections measured in the summer season result in a mean difference of about 100 ksi between backcalculated HMA moduli values once conducted on pavements with thick (>5inches) HMA layers as opposed to thin layers (<5 inches). 54 (a) Interaction means plot (b) Mean difference with 95% confidence intervals for season and time interaction (c) Mean difference with 95% confidence intervals for season and HMA thickness interaction Figure 21 ANOVA plots - WN F climatic region 55 Base Layer Moduli Figure 22 (a) shows the overall variations in the granular base moduli with different FWD passes between SMP thick (>5 inches) and thin (<5 inches) HMA layer pavements. As expected, there is no significant difference in the base layer moduli values with FWD pass num ber (i.e., from morning to evening) with different HMA layer thicknesses and different climatic regions. Overall, the base layer moduli for pavements with thin or thick HMA layers show moduli values ranging between 20 - 40 ksi [see Figure 22 (b) and (c)]. How ever, base moduli for pavements in the WNF region with thin HMA layers are significantly high er [see Figure 22 (c)]. A ge neral expectation is that granular base moduli values may show a significant influence of seasons (months). As expected, substantial dif ferences are observed in the aggregate base modul i for flexible pavements with thin and thick HMA layers in different months [see Figure 22 (d)]. Back - calculation results in higher base layer moduli values in winter months, probably due to freezing conditio ns, as compared to summer months irrespective of HMA layer thickness in the freeze climates. However, the aggregate base layers moduli values for flexible pavement sections having thin HMA layers are highest in the summer months within the WNF climatic reg ion [see Figure 22 (f)]. Table 24 shows the descriptive statistics for the granular base moduli data along with the number (N) within each climatic region. The WNF climatic region has a higher variability with a standard deviation of 46.5 ksi. The DNF clima tic region has the lowest variability; however, it may be due to fewer data in this region as compared to the others. Histogram and box - plot for the available data reveal values beyond the range of 10 - 50 ksi. For the analysis, it is beneficial to exclude a ny such values to prevent masking of the ANOVA results (see Figure 23 ). 56 (a) Overall by FWD pass (d) Overall by month (b) Thick HMA Layer by FWD pass (e) Thick HMA Layer by month (c) Thin HMA Layer by FWD pass (f) Thin HMA Layer by month Figure 22 Seasonal and diurnal variation s in base layer moduli 57 Table 24 Descriptive statistics for aggregate base layer moduli data SMP flexible pavements Climatic region N Mean St d. Minimum Q1 Median Q3 Maximum DF 957 25.6 27.6 5.1 13.7 19.8 26.2 195.4 DNF 582 27 18.4 6.2 12.8 21.8 34.3 95.8 WF 1458 31.6 25.5 5.2 18.8 24.7 34.4 195.8 WNF 1219 44.3 46.5 5.1 15.3 23.7 52.7 195.4 Note: Base moduli values shown are in ksi units. (a) Histogram - Available base moduli data (b) Box - plot - Available base moduli data Figure 23 Visualizing available base layer moduli data SMP flexible pavements To investigate the effects of diurnal and seasonal FWD deflection measurements on base layer moduli, a similar ANOVA was performed as for HMA moduli values. Factors used in the analyses included season, time, HMA layer thickness, and maintenance category ( as a blocking factor). Ensuring the normality of the data used was critical. The satisfaction of the normality assumptions was achieved by suitably transforming the data (see Figure 24 ). ANOVA results for the DF climatic region show that the interaction of season and HMA layer thickness is an essential factor that can explain the variations i n base layer moduli due to temporal (seasonal and diurnal) changes affecting FWD based deflections (see Table 25 ) . Thus, looking at the means plot for the significan t i nteracti o n will help understand the effects on base 58 layer moduli. Also, the maintenance category shows to be a significant contributing factor in the base layer moduli variations; however, this factor is used as a blocking factor only. Figure 24 Evaluat ing normality of the aggregate base moduli data - DF climatic region Table 25 ANOVA results for base layer moduli values DF climatic region Source DoF Seq SS Contribution Adj SS Adj MS F - v alue p - v alue Season 3 0.07817 6.50% 0.096508 0.032169 24.93 0.000 Time 1 0.00352 0.29% 0.002027 0.002027 1.57 0.211 AC_th 1 0.08118 6.76% 0.035506 0.035506 27.51 0.000 Maint. Cat. 3 0.04372 3.64% 0.052592 0.017531 13.58 0.000 Season*Time 3 0.00488 0.41% 0.005118 0.001706 1.32 0.266 Season*AC_th 3 0.03073 2.56% 0.031172 0.010391 8.05 0.000 Time*AC_th 1 0.00066 0.06% 0.000663 0.000663 0.51 0.474 Error 743 0.95885 79.79% 0.958846 0.001291 Lack - of - Fit 30 0.03892 3.24% 0.038918 0.001297 1.01 0.460 Pure Error 713 0.91993 76.55% 0.919928 0.001290 Total 758 1.20171 100.00% 59 (a) Interaction means plot (b) Mean difference with 95% confidence intervals for season and HMA thickness interaction Figure 25 ANOVA plots for base moduli data - DF climatic region The interaction means plot, and the multiple comparison plot shows that the base moduli values are significantly different for all seasons except winters between HMA layers with different thicknesses (thick or thin) [see Figure 25 (a) and (b)]. The d ifference in base layer moduli is highest for the fall season (> 5ksi), while the summer season shows the lowest (< 3ksi). The differences are also practical to alter pavement performance. Another noteworthy finding 60 demonstrated by the interaction means pl ot is that in the winter season where the thick HMA layers showed higher AC moduli corresponds to a lower base layer moduli and vice versa for the rest of the seasons. Low or sub - freezing temperatures are an explanation of the observed trends; where a stif fer HMA layers lead to underestimation of the modulus for the underlying base layers in the winter season. Table 26 shows the ANOVA results for the DNF climatic region. HMA layer thickness and the interaction of FWD deflections measurement season and time appears to be significantly contributing to the seasonal and diurnal effects on base moduli. The main effects plot shows that the difference between base moduli values backcalculated using FWD deflections on pavements with thick and thin HMA layers differ by 3 ksi, enough to cause a change in the pavement performance [see Figure 26 (a)]. Interaction means plot for the season, and time interaction shows that, generally, base layer moduli are higher in the winter season and those resulting from FWD deflections measured before - noon; however, the differences are not practically significant [see Figure 26 (b)]. Table 26 ANOVA results for base layer moduli values D N F climatic region Source DoF Seq SS Contribution Adj SS Adj MS F - v alue p - v alue Season 3 4.1379 4.76% 4.0706 1.35686 8.32 0.000 Time 1 0.2737 0.32% 0.3089 0.30890 1.89 0.169 AC_th 1 2.3093 2.66% 1.6189 1.61887 9.93 0.002 Maint. Cat. 3 6.9675 8.02% 7.1733 2.39110 14.67 0.000 Season*Time 3 1.3266 1.53% 1.3697 0.45658 2.80 0.040 Season*AC_th 3 1.2617 1.45% 1.2494 0.41648 2.55 0.055 Time*AC_th 1 0.0016 0.00% 0.0016 0.00155 0.01 0.922 Error 433 70.5893 81.26% 70.5893 0.16302 Lack - of - Fit 28 14.1451 16.28% 14.1451 0.50518 3.62 0.000 Pure Error 405 56.4442 64.98% 56.4442 0.13937 Total 448 86.8674 100.00% ANOVA of the base layer moduli data available for the WF climatic region shows that FWD deflections measured in different seasons have an imprint on these values (see Table 27 ). 61 Besides, HMA layer thickness also contributes to the variations in base moduli due to seasonal and diurnal FWD tests. Maintenance category and either of the interactions, on the other hand, does not appear to be contributing factors. (a) Main effects plot (b) Interaction means plot Figure 26 Factorial plots ANOVA for base layer moduli data DNF climatic region The main effects plot shows that base layer moduli are highest for the winter season (see Figure 27 ). Also, there is a clear difference between the base layer moduli values between the 62 winter and spring (2.5 ksi) and winter and fall (3.5 ksi) seasons. The plot also shows that base moduli values obtained from FWD deflections measured on pavements with thick HMA layers are noticeably higher than those measured over thin HMA layers with a difference greater than 3 ksi. Table 27 ANOVA results for base layer moduli values W F climatic region Source DoF Seq SS Contribution Adj SS Adj MS F - v alue p - v alue Season 3 2.218 1.43% 1.845 0.61501 5.06 0.002 Time 1 0.008 0.00% 0.001 0.00107 0.01 0.925 AC_th 1 2.076 1.34% 2.458 2.45803 20.21 0.000 Maint. Cat. 3 0.620 0.40% 0.640 0.21338 1.75 0.154 Season*Time 3 0.567 0.37% 0.567 0.18893 1.55 0.199 Season*AC_th 1 0.000 0.00% 0.000 0.00004 0.00 0.986 Time*AC_th 1227 149.236 96.45% 149.236 0.12163 Error 37 12.949 8.37% 12.949 0.34998 3.06 0.000 Lack - of - Fit 1190 136.287 88.08% 136.287 0.11453 Pure Error 1239 154.725 100.00% Total 3 2.218 1.43% 1.845 0.61501 5.06 0.002 Figure 27 Main effects plot for base layer moduli WF climatic region ANOVA for available base moduli data within the WNF climatic region resulted in similar findings except that higher base moduli values are observed within fall as compared to other 63 seasons (see Table 28 and Figure 28 ). Also, base moduli obtained from FWD deflections measured in the spring season are the lowest. Moisture variations can be a possible explanation of such effects. Also, there is a clear mean difference of base moduli between values o btained from FWD tests conducted in spring and winter (3 ksi) and fall and spring seasons (4 ksi). Table 28 ANOVA results for base layer moduli values W N F climatic region Source DoF Seq SS Contribution Adj SS Adj MS F - v alue p - v alue Season 3 0.05553 3.66% 0.05380 0.017933 9.46 0.000 Time 1 0.00407 0.27% 0.00435 0.004346 2.29 0.130 AC_th 1 0.06641 4.37% 0.05749 0.057486 30.32 0.000 Maint. Cat. 3 0.00501 0.33% 0.00482 0.001608 0.85 0.468 Season*Time 3 0.00433 0.29% 0.00436 0.001454 0.77 0.513 Season*AC_th 3 0.01292 0.85% 0.01293 0.004310 2.27 0.079 Time*AC_th 1 0.00004 0.00% 0.00004 0.000041 0.02 0.883 Error 723 1.37058 90.24% 1.37058 0.001896 Lack - of - Fit 39 0.18683 12.30% 0.18683 0.004791 2.77 0.000 Pure Error 684 1.18375 77.93% 1.18375 0.001731 Total 738 1.51890 100.00% Figure 28 Main effects plot W N F climatic region 64 Subgrade Layer Moduli The subgrade layer moduli values found in the LTPP SMP study database are evaluated similar to the HMA and base layers for analyzing the diurnal (i.e., by FWD pass) and seasonal (i.e., monthly) variation in the moduli values. Generally, the unbound layer m oduli are expected not to demonstrate noticeable changes within a day. Figure 29 (a) shows the overall variations in the subgrade layer moduli with different FWD passes between SMP thick (>5 inches) and thin (<5 inches) HMA layer pavements. No significant v ariation exists in the moduli values with FWD pass number (i.e., morning to afternoon) with different HMA layer thicknesses and different climatic regions. Overall, pavements with thick HMA layers show higher subgrade moduli values [see Figure 29 (b)]. Also , for pavements with thick HMA layers, higher subgrade moduli values are observed in the dry climates. On the other hand, higher moduli values are seen in the freeze regions for pavements with thin HMA layers [see Figure 29 (c)]. It is likely to see a signi ficant influence of seasons (months) in subgrade layer moduli values. Overall, not many variations are observed in the subgrade layer moduli for flexible pavements with thin and thick HMA layers [see Figure 29 (d)] within different months. However, somewhat mixed trends occur in the subgrade moduli values within different climatic regions irrespective of HMA layer thickness [see Figure 29 (e) and (f)]. Table 29 shows the descriptive statistics for the available subgrade moduli data in the LTPP SMP database. T he data reveal a higher variability in the backcalculated subgrade moduli values in the DF climatic region, which also has the highest mean subgrade modulus. The absence of moisture in the DF climatic region might be the reason for the highest mean subgrad e modulus value. 65 (a) Overall by FWD pass (d) Overall by month (b) Thick HMA Layer by FWD pass (e) Thick HMA Layer by month (c) Thin HMA Layer by FWD pass (f) Thin HMA Layer by month Figure 29 Seasonal and diurnal variations in subgrade layer moduli 66 Table 29 Descriptive statistics subgrade layer moduli SMP flexible pavements Climatic region N Mean St d. Minimum Q1 Median Q3 Maximum DF 957 40.3 34.2 5.1 16.7 26.6 55 148.9 DNF 582 34.8 22.6 8.1 17.1 27.9 47 146.6 WF 1739 35 25.3 5.7 15.7 28.3 46 146.6 WNF 1533 30.7 25.1 5.3 15 23.2 34.5 146.6 Note: Subgrade moduli values shown are in ksi units. (a) Histogram - Available subgrade moduli data (b) Box - plot - Available subgrade moduli data Figure 30 Visualising available subgrade layer moduli data SMP flexible pavements Figure 30 demonstrates the backcalculated subgrade layer moduli data obtained from the SMP database. The data shows backcalculated subgrade moduli values higher than 40 ksi. Such values are better to exclude from the analysis to prevent masking of the ANOVA results. Table 30 ANOVA results for subgrade layer moduli values D F climatic region Source DoF Seq SS Contribution Adj SS Adj MS F - v alue p - v alue Season 3 5.707 4.28% 8.4809 2.8270 19.40 0.000 Time 1 0.356 0.27% 0.4684 0.4684 3.21 0.073 AC_th 1 23.619 17.71% 14.8824 14.8824 102.13 0.000 Maint. Cat. 3 11.590 8.69% 11.5644 3.8548 26.45 0.000 Season*Time 3 0.517 0.39% 0.5880 0.1960 1.35 0.259 Season*AC_th 3 0.036 0.03% 0.0287 0.0096 0.07 0.978 Time*AC_th 1 0.440 0.33% 0.4398 0.4398 3.02 0.083 Error 625 91.075 68.30% 91.0749 0.1457 Lack - of - Fit 31 5.769 4.33% 5.7694 0.1861 1.30 0.133 Pure Error 594 85.306 63.98% 85.3055 0.1436 Total 640 133.341 100.00% 67 The available subgrade data were transformed to ensure the normality assumptions that are critical for drawing meaningful conclusions from an ANOVA. Table 30 demonstrates the ANOVA results for the subgrade layer moduli values available for the DF climatic region . The results reveal that the FWD measurement season and HMA layer thickness are critical factors that can influence the subgrade moduli values in the DF climatic region. Figure 31 (a) illustrates the main effects of these factors. Subgrade moduli obt ained from FWD tests conducted in the summer season result in higher values while spring season tests bear the lowest. The absence of moisture in the summer season can be a possible explanation of such effects. There is a sizeable mean subgrade moduli diff erence between backcalculated values obtained from deflections measured in different seasons [see Figure 31 (b)]. Also, subgrade moduli obtained from FWD tests over thick HMA layers are lower than the opposite [see Figure 31 (a)]. ANOVA results for DNF clima tic region show that the interaction of the FWD measurement season and HMA layer thickness is contributing to the variations in subgrade moduli (see Table 31 ). Explanation of the interaction effects warrants looking into the interaction means plot. Table 31 ANOVA results for subgrade layer moduli values D N F climatic region Source DoF Seq SS Contribution Adj SS Adj MS F - v alue p - v alue Season 3 3.8321 7.25% 4.7716 1.59052 13.11 0.000 Time 1 0.0020 0.00% 0.0511 0.05109 0.42 0.517 AC_th 1 0.0782 0.15% 0.0055 0.00551 0.05 0.831 Maint. Cat. 3 3.8304 7.24% 3.6687 1.22290 10.08 0.000 Season*Time 3 0.4036 0.76% 0.2903 0.09676 0.80 0.496 Season*AC_th 3 1.1690 2.21% 1.2046 0.40152 3.31 0.020 Time*AC_th 1 0.3742 0.71% 0.3742 0.37422 3.08 0.080 Error 356 43.1967 81.68% 43.1967 0.12134 Lack - of - Fit 27 7.6753 14.51% 7.6753 0.28427 2.63 0.000 Pure Error 329 35.5214 67.17% 35.5214 0.10797 Total 371 52.8863 100.00% 68 (a) Main effects plot (b) Seasonal mean difference with 95% confidence intervals Figure 31 ANOVA plots for subgrade moduli data - DF climatic region 69 Figure 32 Interaction means plot for subgrade moduli data DNF climatic region Observing the interaction between season and HMA layer thickness in Figure 32 shows that there is no noticeable difference between the subgrade moduli values in the winter and fall seasons irrespective of the HMA layer thickness. However, there is a clear distinction between subgrade moduli values in spring season backcalculated fo r pavements with thin and thick HMA layers (around 5 ksi). Also, a noticeable difference exists between subgrade moduli values in the fall season as well for values obtained from testing on pavements with different HMA layer thickness. Overall, moduli valu es related to summers are the highest, while those related to the fall season are the lowest. Results from a similar ANOVA for the WF region also show that the interaction between season and HMA layer thickness influences the subgrade moduli of the pavemen ts. Table 32 displays the ANOVA results for the WF climatic region. Figure 33 shows the interaction means plot along with the mean comparisons for the two interacting factors. There is a clear difference between the moduli values obtained in different seas ons from pavements with different HMA layer thicknesses. A minimum mean difference of 4 ksi (for spring season) exists between 70 subgrade moduli values obtained for a thick and a thin HMA layered pavement in all seasons. The maximum mean (21 ksi) difference exists between mean subgrade moduli measured over thick and thin HMA layer pavements in the winter season; a possible explanation can be extremely low temperatures. Table 32 ANOVA results for subgrade layer moduli values W F climatic region Source DoF Seq SS Contribution Adj SS Adj MS F - v alue p - v alue Season 3 2.829 1.16% 3.404 1.1348 5.97 0.000 Time 1 0.020 0.01% 0.209 0.2093 1.10 0.294 AC_th 1 5.737 2.36% 16.808 16.8082 88.40 0.000 Maint. Cat. 3 13.447 5.53% 11.792 3.9306 20.67 0.000 Season*Time 3 0.160 0.07% 0.183 0.0610 0.32 0.810 Season*AC_th 3 4.853 2.00% 4.960 1.6532 8.70 0.000 Time*AC_th 1 0.140 0.06% 0.140 0.1395 0.73 0.392 Error 1135 215.795 88.81% 215.795 0.1901 Lack - of - Fit 36 18.451 7.59% 18.451 0.5125 2.85 0.000 Pure Error 1099 197.343 81.22% 197.343 0.1796 Total 1150 242.981 100.00% Table 33 ANOVA results for subgrade layer moduli values W N F climatic region Source DoF Seq SS Contribution Adj SS Adj MS F - v alue p - v alue Season 3 3.231 1.58% 3.493 1.16436 7.30 0.000 Time 1 0.044 0.02% 0.091 0.09104 0.57 0.450 AC_th 1 0.527 0.26% 0.117 0.11658 0.73 0.393 Maint. Cat. 3 9.983 4.87% 10.030 3.34322 20.97 0.000 Season*Time 3 0.358 0.17% 0.397 0.13219 0.83 0.478 Season*AC_th 3 0.573 0.28% 0.574 0.19127 1.20 0.309 Time*AC_th 1 0.001 0.00% 0.001 0.00063 0.00 0.950 Error 1193 190.202 92.82% 190.202 0.15943 Lack - of - Fit 45 11.801 5.76% 11.801 0.26225 1.69 0.003 Pure Error 1148 178.400 87.06% 178.400 0.15540 Total 1208 204.919 100.00% A similar ANOVA analysis for the WNF climatic region reveals that the season of FWD testing has an impact on the subgrade modulus backcalculated from the measured deflections (see Table 33 ). The mean difference plot within different seasons shows that the difference of subgrade moduli values between summer and winter (about 2 ksi), fall and winter (about 2.5 ksi), 71 performance (see Figure 34 ). (a) Interaction means plot (b) Mean difference with 95% confidence intervals for season and HMA layer thickness Figure 33 ANOVA plots for subgrade moduli data - W F climatic region 72 (a) Main effects plot (b) The seasonal mean difference with 95% confidence interval Figure 34 ANOVA plots WNF climatic region Discussion on ANOVA Results (HMA, Base and Subgrade Moduli) Different techniques are used in practice to correct the HMA layer moduli values to a standard temperature of 77 F. The LTPP SMP database contains the uncorrecte d backcalculated HMA moduli values. This study presents the correction of these values based on the Asphalt Institute equation: Where; E 0 = corrected HMA layer modulus in psi; E = backcalculated uncorrected HMA layer modulus in psi; t = test temperature in degree Fahrenheit; and t 0 = reference temperature = 77 F. The calculation of the mid - depth HMA layer temperatures for each pavement section involved using the data from the MON_DEFL_TEMP_DEPTH table of the LTPP monitoring module. 73 (a) HMA layer moduli correction - DF climatic region (b) HMA layer moduli correction - DNF climatic region (c) HMA layer moduli correction WF climatic region (d) HMA layer moduli correction WNF climatic region Figure 35 HMA layer moduli temperature correction using the Asphalt Institute equation 74 Figure 35 shows the corrected HMA moduli values for each climatic region using the Asphalt Institute equation. The equation performed well in correcting the available HM A moduli values, as shown in the figure. The data showed the highest variation in HMA layer moduli values in the WNF climatic region; hence, the high variation in the corrected values can also be seen. Thus, it may not be critical at what temperature or we ather conditions the FWD deflections are measured on flexible pavements for the HMA layer. Since a single FWD measurement gives the moduli for all the layers (i.e., surface, base, and subgrade), the underlying unbound layers become critical. Therefore, the general guidelines should consider the variation of moduli values for the unbound layers together for all layers. (a) Temperature ranges for season and time interaction (b) Seasonal temperature ranges Figure 36 Pavement surface and ambient temperatures during FWD measurements on flexible pavements - DF climatic region Figure 36 presents the temperature ranges during FWD measurements in the DF climatic region. ANOVA results showed that the interaction of season and time and season and HMA thickness were significant factors for HMA and base layer moduli in the DF climatic region, respectively. Subgrade moduli analysis showed the main effects of the season as a prominent feature in defining change in the moduli. The ambient temperature ranges occurring in the spring 75 and fall seasons appear to have the least variation of the moduli v alues in the HMA layer. Thus, the subgrade and the base moduli backcalculated from using FWD deflection in these temperature ranges would result in values close to the actual. Consequently, the suggested ambient temperature range for conducting FWD in the DF climatic region is between 55 - 70 F. Figure 37 Pavement surface and ambient temperatures during FWD measurements on flexible pavements - DNF climatic region Figure 37 shows the temperature ranges recorded during FWD measurements in the DNF climatic region. In this climate, the season and time interaction showed noticeable effects on HMA and base layer moduli values according to ANOVA results. For subgrade, season and HMA layer thickness interaction was deemed critical. Based on the ANOVA results pres ented earlier and the observed effects of each of the layer moduli, the spring and fall season ambient temperatures appear to have minimal variation in the base and HMA layer moduli, respectively. The unbound layer moduli did not show considerable differen ce in the spring and fall seasons as well. Therefore, the ambient temperatures between 65 80 F appear to result in the unbound layer moduli values that may be closest to the actual. 76 In the WF climate, the interaction of season and time appears significant for effects in HMA layer moduli according to ANOVA results. Season and season and HMA layer thickness interaction showed to have a noticeable impact on base and subgrade layer moduli , respectively. Figure 38 shows the temperature ranges for season and time interaction, and the seasonal temperature ranges in the WF climatic region. HMA moduli showed the least variation between the spring and fall seasons. Base moduli showed no change b etween spring and summer seasons, while subgrade moduli remained similar in the spring season. Based on the observed effects, the ambient temperatures within the spring season can help in determining the actual unbound layer moduli in the WF climatic regio n. Consequently, the ambient temperature range suggested for FWD deflection measurements is between 55 65 F. (a) Temperature ranges for season and time interactions (b) Seasonal temperature ranges Figure 38 Pavement surface and ambient temperatures during F WD measurements on flexible pavements - W F climatic region The season and time interaction was termed significant as per the ANOVA results for WNF climatic region concerning HMA layer moduli. The interaction means plot showed minimal variation in the HMA l ayer moduli values between fall and spring seasons. ANOVA results for the base and subgrade layers revealed the FWD measurement season as an essential 77 factor affecting their moduli. Based on the effects seen on the unbound layers, the FWD deflections measu red within the ambient temperatures occurring in the spring season can result in moduli values closest to the actual. Figure 39 shows the ambient temperature in the spring season to be 65 75 F; thus, suggested to conduct FWD deflection measurements in th e WNF climatic region. (a) Temperature ranges for season and time interactions (b) Seasonal temperature ranges Figure 39 Pavement surface and ambient temperatures during FWD measurements on flexible pavements - W N F climatic region Consequently, based on the discussion presented in this section, the suggested ambient temperature deemed appropriate to result in the unbound layer moduli, which represents their actual condition, is between 55 75 F. Also, since the time of the day did not appear to affect th e unbound layers, as expected, FWD deflections can be measured at any time of the day with the ambient temperature in the suggested range. 78 (a) HMA moduli for South Dakota section ID 46 - 9187 DF climatic region (b) HMA moduli for Maine section ID 23 - 1026 WF climatic region (c) Base moduli for South Dakota section ID 46 - 9187 DF climatic region (d) Base moduli for Maine section ID 23 - 1026 WF climatic region (e) Subgrade moduli for South Dakota section ID 46 - 9187 DF climatic region (f) Subgrade moduli for Maine section ID 23 - 1026 WF climatic region Figure 40 Monthly variation in layer moduli values - SMP flexible pavement sections 79 Figure 40 shows the monthly variation in layer moduli values for two sections, each in the DF and WF climatic region. The plots show that the HMA layer moduli decrease in the summer months while these increase as the temperatures get colder (i.e., winter and fall m onths). The corresponding base and subgrade moduli values also show monthly (i.e., seasonal ) variation , which is in agreement with the ANOVA results for these climatic regions. 4.1.2 Longitudinal Profile Measurements (IRI) Flexible Pavements The smoothness of a pavement surface determines its functional performance, evaluated using longitudinal profile measurements. The longitudinal profile measurements are commonly summarized by the International Roughness Index (IRI) that reduces the thousands of elevation va lues into a single value [38, 39] . However, no matter which index is calculated from a longitudinal profile, the quality of the information is only as good as the profile measurement [2] . But seasonal and diurnal (temperature/moisture) changes influence th ese profile measurements. Th erefore , there is a need to evaluate the impacts of temporal (seasonal temperature/moisture and daily temperature) variations on longitudinal profile measurements . The steps involved in the analysis are enumerated as follows: 1. Th e pavement structure details, IRI values, profile measurement dates and timings, maintenance history, pavement surface temperature, and air temperature were some of the variables identified to perform the analysis. 2. The investigation required an analysis in volving the available flexible pavement sections in the LTPP SMP database. 3. The available flexible pavement sections were identified in the SMP LTPP database with their climatic region. Note that the c limatic region s are defined based on temperature (i.e., freeze/no freeze), and moisture (i.e., wet/dry). 80 4. The structural details of the identified sections (i.e., layer types and thicknesses) were obtained. 5. T he IRI values, along with the date and time of profile measurement s, the air and pavement surface temperatures were obtained. Also, the maintenance history details for the section were extracted from the database (i.e., construction no. and type of treatment) and categorized. 6. Five profile measurements (runs) per visit are the LTPP stan dard [44] ; the analysis used each of the calculated IRI values . 7. All the required data elements were arranged in a relational database. 8. Inspected the data using a histogram and boxplot to identify outliers. 9. Two factors were used in this analysis; (a) measu rement month discretized into four seasons (i.e., levels) to look at the seasonal effects, (b) measurement time with two levels (before noon and afternoon) for the diurnal effects. Besides, the maintenance category was used as a blocking factor. Additional ly, an additional factor used was the HMA layer thickness with two levels (i.e., thick and thin) based on the original SMP study design [43] . 10. Analysis of Variance (ANOVA) was conducted for JPCP sections within each climatic region to investigate the tempo ral effects on the joint LTE values. 81 (a) IRI by hour (b) IRI by month (c) Histogram Available IRI data (d) Box - plot Available IRI data Figure 41 Assessing available IRI data SMP flexible pavement sections Figure 41 shows the overall variations in the mean IRI values by the hour and month of profile measurements within different climatic regions. The IRI values show an important hourly influence, i.e., the IRI values are different in each hour among the different cli matic regions [see Figure 41 (a)]. Also, significant variations can be seen in the IRI values among different months. [see Figure 41 (b)]. Figure 41 (c) and (d) show the histogram and box - plots, while Table 34 displays the descriptive statistics of the data w ithin each climatic region. There is generally higher variability in the IRI values within different climates, with the highest in the wet climates. 82 The variation in these climates may be due to IRI values over 170 inch/mile. Such higher values can potenti ally mask the results of ANOVA. The IRI analysis for the flexible pavement section presented in this study uses values between 30 - 170 inch/mile. Table 34 Descriptive statistics IRI values SMP flexible pavement sections Climatic region N Mean St d . Minimum Q1 Median Q3 Maximum DF 1163 65.9 20.8 35.8 51.2 59.2 77.8 149.9 DNF 840 63 17.9 29.2 51.5 59.8 75.2 141.4 WF 3387 81.1 35.8 32.5 54.6 70.5 95.8 231.8 WNF 1727 81.1 35.7 29.1 59.1 69.4 93.6 261.1 Note: IRI values shown are in inch/mile units. Table 35 ANOVA results for IRI data for flexible pavements DF climatic region Source DoF Seq SS Contribution Adj SS Adj MS F - v alue p - v alue Season 3 0.006833 1.95% 0.003750 0.001250 7.97 0.000 Time 1 0.000001 0.00% 0.000265 0.000265 1.69 0.194 AC_th 1 0.089827 25.63% 0.045984 0.045984 293.25 0.000 Maint. Cat. 3 0.064755 18.48% 0.057951 0.019317 123.19 0.000 Season*Time 3 0.003519 1.00% 0.004306 0.001435 9.15 0.000 Season*AC_th 3 0.005018 1.43% 0.004505 0.001502 9.58 0.000 Time*AC_th 1 0.000635 0.18% 0.000635 0.000635 4.05 0.044 Error 1147 0.179856 51.32% 0.179856 0.000157 Lack - of - Fit 31 0.029292 8.36% 0.029292 0.000945 7.00 0.000 Pure Error 1116 0.150564 42.96% 0.150564 0.000135 Total 1162 0.350443 100.00% ANOVA results for DF climatic region show that the interaction of season and time influences the IRI of the flexible pavement sections. Also, the interactions of season and time Table 35 ). T he data was transformed to satisfy the normality assumptions to draw meaningful conclusions from the ANOVA (see Figure 42 ). 83 Figure 42 An example of ensur ing normality of the IRI data - WF climatic region Figure 43 shows the interaction means plot from the ANOVA for the IRI data of the DF climatic region. There is no considerable difference between the IRI values obtained in the morning (i.e., before - noon) and the afternoon with seasons. The fall season shows a variation of the morning and afternoon IRI of around 8 inches/mile, which is not practically relevant. The IRI changes considerably between thick (> 5inches) and thin (< 5inches) HMA layered pavements with different seasons and times of the day. The thin HMA layered pavements exhibited higher IRI v alues overall. The IRI differs around 15 20 inches/mile in all seasons except fall between thick and thin HMA layered pavements [see Figure 44 (a)]. A similar IRI difference exists between pavements with thick and thin HMA layers across different times (i .e., before noon and afternoon) of the day [see Figure 44 (b)]. 84 Figure 43 Interaction means plot for IRI data - DF climatic region (a) Mean differences with 95% confidence intervals for season and HMA layer thickness interaction (b) Mean differences with 95% confidence intervals for time and HMA layer thickness interaction Figure 44 Mean IRI multiple comparison plots DF climatic region 85 Figure 45 IRI measurement temperatures with 95% confidence intervals - DF climatic region Figure 45 displays the temperatures at the time of profile measurements. A comparison of Figures 43 and 45 infers that the ambient temperatures related to the fall season prod uce IRI with less variation irrespective of the season, time of the day, and HMA layer thickness. The ambient temperatures in the fall season range between 50 - 65 F. Such temperatures are suggested for profile measurements in DF climates on flexible pavem ents. Besides, at such temperatures where possible, the pavement profile should be measured in the morning times (i.e., before noon). Table 36 ANOVA results for IRI data for flexible pavements D N F climatic region Source DoF Seq SS Contribution Adj SS Adj MS F - v alue p - v alue Season 3 2.2935 3.55% 0.1371 0.04570 1.11 0.343 Time 1 0.0044 0.01% 0.4303 0.43033 10.49 0.001 AC_th 1 16.0635 24.83% 9.3492 9.34916 227.79 0.000 Maint. Cat. 3 9.2876 14.36% 8.2945 2.76482 67.36 0.000 Season*Time 3 2.5753 3.98% 2.4898 0.82994 20.22 0.000 Season*AC_th 3 0.4597 0.71% 0.3531 0.11770 2.87 0.036 Time*AC_th 1 0.3952 0.61% 0.3952 0.39516 9.63 0.002 Error 819 33.6137 51.96% 33.6137 0.04104 Lack - of - Fit 24 7.2782 11.25% 7.2782 0.30326 9.15 0.000 Pure Error 795 26.3355 40.71% 26.3355 0.03313 Total 834 64.6929 100.00% 86 Figure 46 Interaction means plot for IRI data - D N F climatic region Table 36 displays the results from ANOVA on IRI data for DNF climatic region. The three interactions between profile measurement season, time, and the HMA layer thickness appears significant. To dissect these interactions, one needs to look at the interaction mean s plots for each significant interaction. Figure 46 shows the interaction means plot for the ANOVA for DNF climatic region. Noticeable differences exist between IRI obtained from profile measurements undertaken in different seasons. Similar to the DF clima te, pavements with thick HMA layer has lower IRI, with little variation, as compared to thin HMA layered pavements across different seasons in the DNF climatic region. Also, pavements with thick HMA layers display low IRI without any changes across the day as compared to pavements with thin HMA layer thickness; it has higher IRI and varies with time in the day. 87 (a) Mean difference with 95% confidence interval for season and time interaction (b) Mean difference with 95% confidence interval for season and HMA layer thickness (c) Mean difference with 95% confidence interval for time and HMA layer thickness Figure 47 Mean IRI multiple comparison plots D N F climatic region 88 Figure 47 shows the multiple comparison plots for the significant interactions for the DNF climatic region. There is an IRI difference of less than 10 inches/mile between morning (i.e., before noon) and afternoon within every season [see Figure 47 (a)]. A considerab le difference in mean IRI values also exists between thick and thin HMA layered pavements within every season. The lowest difference (< 15 inches/mile) occurs in the summer season in IRI values determined with different HMA layer thickness; 22 25 inches/ mile difference exists among the rest of the seasons [see Figure 47 (b)]. Pavements with different HMA layer thickness also show considerable variation (> 16 inches/mile) in IRI during the day from morning to afternoon [see Figure 47 (c)]. Figure 48 shows th e profile measurement temperatures for the DNF climatic region. Based on the discussion and observing the temperatures, a range between 50 - 75 F can be suggested for IRI measurements in the DNF climatic region Figure 48 IRI measurement temperatures with 95% confidence intervals - D N F climatic region For the WF climatic region, ANOVA declares HMA layer thickness and interaction between profile measurement season and time to be driving the IRI variations (see Table 37 ). The main effects plot illustrates th at, although statistically significant, there is no practical difference between IRI of pavements with different HMA layer thickness [see Figure 49 (a)]. 89 Considering the ANOVA termed statistically significant season and time interaction, the interaction mea ns plot does not show any practical IRI differences within a day between various seasons [see Figure 49 (b)]. No difference exists between IRI values determined during the day, irrespective of the time, in spring and summer seasons. The preceding discussion , along with the temperatures shown in Figure 49 (c), suggests that it might be better to determine IRI using profile measurements conducted within the ambient temperatures ranging between 50 75 F. Table 37 ANOVA results for IRI data for flexible pavemen ts W F climatic region Source DoF Seq SS Contribution Adj SS Adj MS F - v alue p - v alue Season 3 0.04613 2.14% 0.01814 0.006048 13.66 0.000 Time 1 0.00051 0.02% 0.00034 0.000340 0.77 0.381 AC_th 1 0.04056 1.88% 0.00318 0.003181 7.18 0.007 Maint. Cat. 3 0.63157 29.24% 0.61529 0.205098 463.07 0.000 Season*Time 3 0.00519 0.24% 0.00556 0.001853 4.18 0.006 Season*AC_th 3 0.00260 0.12% 0.00283 0.000945 2.13 0.094 Time*AC_th 1 0.00058 0.03% 0.00058 0.000577 1.30 0.254 Error 3235 1.43280 66.33% 1.43280 0.000443 Lack - of - Fit 43 0.26613 12.32% 0.26613 0.006189 16.93 0.000 Pure Error 3192 1.16668 54.01% 1.16668 0.000365 Total 3250 2.15995 100.00% Table 38 displays the ANOVA results for the WNF climatic region. It shows that the profile measurement season, time of the day, and the HMA layer thickness has an essential practically significant difference (about 22 inches/mile) between IRI in pavements with different HMA layer thickness [see Figure 50 (a)]. No practical difference between the pavement IRI exists within seasons or different times of the day. 90 (a) Main effects plot (b) Interaction means plot (c) Profile measurement temperatures with 95% confidence intervals Figure 49 ANOVA results for IRI data of flexible pavements WF climatic region 91 Table 38 ANOVA results for IRI data for flexible pavements W N F clima tic region Source DoF Seq SS Contribution Adj SS Adj MS F - v alue p - v alue Season 3 0.00542 0.38% 0.005599 0.001866 3.15 0.024 Time 1 0.03279 2.28% 0.005389 0.005389 9.08 0.003 AC_th 1 0.31419 21.84% 0.202213 0.202213 340.85 0.000 Maint. Cat. 3 0.09186 6.39% 0.090810 0.030270 51.02 0.000 Season*Time 3 0.00252 0.18% 0.002422 0.000807 1.36 0.253 Season*AC_th 3 0.00325 0.23% 0.003165 0.001055 1.78 0.149 Time*AC_th 1 0.00051 0.04% 0.000505 0.000505 0.85 0.356 Error 1665 0.98778 68.68% 0.987783 0.000593 Lack - of - Fit 41 0.14180 9.86% 0.141802 0.003459 6.64 0.000 Pure Error 1624 0.84598 58.82% 0.845981 0.000521 Total 1680 1.43832 100.00% (a) Main effects plot (b) Seasonal temperature ranges (c) Diurnal temperature ranges Figure 50 ANOVA results for IRI of flexible pavements WNF Climatic region 92 Figure 50 (b) and (c) show the temperature ranges during pavement profile measurements used to determine their IRI. These plots show that temperatures in the WNF climatic region vary between 50 - 85 F. As discussed earlier, there is no considerable difference in IRI between different seasons as well as within different times of the day; the temperature plot suggests that ambient temperature range between 55 - 75 F may produce IRI with less variability irrespective of season and time. Generally speaking, the IRI of f lexible pavements do not show a definite trend at a group or climatic region level. Such effects may result from the variation of the initial pavement IRI values determined just after construction. Based on the discussion for individual climatic regions, t he ambient temperature range between 50 - 75 F appears to show less difference in IRI values across all climates with no limitation on time of the day. Figure 51 presents four flexible pavement sections, each from a climatic region. No particular trend ex ists for IRI values determined from profile measurements in different months. It is in agreement with ANOVA results, which showed statistically significant seasonal differences; however, these were not practically important. 93 (a) IRI for Montana section ID 30 - 0114 DF climatic region (b) IRI for Arizona section ID 4 - 0113 DNF climatic region (c) IRI for Vermont section ID 50 - 1002 WF climatic region (d) IRI for V irginia section ID 5 1 - 0 1 14 W N F climatic region Figure 51 Monthly IRI variation - SMP flexible pavements 4.2 Analysis for Rigid Pavements (by climatic region) FWD deflection - based parameters for rigid pavements include the PCC layer modul us, the modulus of subgrade reaction ( k - value), load transfer efficiency (LTE) at joints/cracks, estimation of edge support, and determination of void potential under the slab . Estimation of PCC layer modulus and the k - value uses mid - slab deflections measured by an FWD device. The evaluation of joint or crack LTE uses deflections mea sured at these locations. Corner and edge 94 deflections help determine the edge support and void potential under the PCC slab, respectively. This study used the available PCC layer moduli, k - values, and LTE values to assess the effects of seasonal and diurna l measurements on these parameters. A relational database was prepared for the obtained data from the LTPP SMP database. Each of these parameters was analyzed separately for each climatic region ; it aided to exclude a factor, i.e., the climatic region from the analysis. The removal of one factor helped reduce the data required while improving the power of the analysis. Similarly, using the available IRI data from the LTPP SMP database, the seasonal and diurnal effects on pavement roughness were investigated for the available JPCP sections of the SMP study . The IRI data extracted from the database was analyzed similarly, dividing the data by climatic region, to gain the benefits as mentioned earlier. 4.2.1 FWD based Pavement Parameters Elastic modulus of the PCC slab and modulus of the subgrade reaction ( k - value) are calculated using deflections measured with an FWD device at the center of the slab . The main factors t o investigate the effects of seasonal and diurnal FWD measurements on t hese moduli values include (a) FWD testing month and (b) time of FWD measurement. The maintenance category of the pavement sections is used as a blocking factor. An analysis for investigating the seasonal and diurnal effects on each of the mentioned param eters is presented next. The data analysis for each data element involved the following steps: 1. The pavement structure details, layer moduli values, FWD measurement dates and timings, maintenance history, pavement surface temperature, air temperature, and 95 temperature gradient measurements were some of the variables identified to perform the analysis. 2. The investigation required an analysis involving a group of sections within each climatic region . 3. The JPCP pavement sections in the SMP LTPP database were identified in each of the four climatic regions. Note that climatic regions are defined in terms of temperature (i.e., freeze/no freeze), and moisture (i.e., wet/dry). 4. The structural details, including layer types and thicknesses, were also obtained. 5. The backcalculated layer moduli values, along with the date and time of FWD measurements, the air and pav ement surface temperatures measured at the time of FWD testing, and the temperatures measured at different depths of the structure were obtained. Also, the maintenance history details for all the sections were also extracted from the database (i.e., constr uction no. and type of treatment) and categorized. 6. The data was arranged in a relational database. 7. The data were inspected using a histogram and boxplot to identify outliers. 8. Two factors were used in this analysis; (a) measurement month discretized into fo ur seasons (i.e., levels) to look at the seasonal effects, (b) measurement time with two levels (before noon and afternoon) for the looking into the diurnal effects. 9. Analysis of Variance (ANOVA) was conducted for JPCP sections within each climatic region t o investigate the temporal effects on each data element using a level of significance of 95% (i.e., = 0.05). 96 PCC Layer Moduli Figure 52 (a) shows the overall variations in the PCC moduli with different FWD passes in various climatic regions. The PCC moduli decrease with increasing pass number (i.e., morning to afternoon) which may be due to the curling down of the PCC slab with the rise in temperature as the day progresses . However, somewhat mixed trends were observed within months between different climates [see Figure 52 (b)]. Lower moduli values in the summer months may be due to the curl down of the PCC slabs. (a) FWD pass (b) FWD measuremen t month (c) Histogram available PCC moduli data (d) Box - plot available PCC moduli data Figure 52 Assessing available PCC moduli data SMP JPCP pavement sections 97 Table 39 Descriptive statistics PCC moduli SMP JPCP pavements Climatic region N Mean St d. Minimum Q1 Median Q3 Maximum DF 113 7653 2462 3513 5699 7197 9788 13612 DNF 156 6953 2374 3606 5200 5942 8717 14801 WF 297 6506 1064 3268 586 6397 7042 10978 WNF 236 6945 3567 3146 5208 6485 7489 31574 Note: PCC moduli values shown are in ksi units. Table 39 shows the descriptive statistics for PCC layer moduli values for the available JPCP sections along with the available number (N) of the FWD measurements within each climatic zone. The DF climatic re gion has the highest variability in the PCC layer moduli values found in the database. Figure 52 also shows the histogram and box - plot for the backcalculated PCC moduli data. It appears that each climatic region has some PCC moduli values higher than 10,00 0 ksi. ANOVA presented in the study excluded such high values as these can influence the findings from the analysis. Table 40 ANOVA results for PCC moduli values DF climatic region Source DoF Seq SS Contribution Adj SS Adj MS F - v alue p - v alue Season 3 0.74327 11.37% 0.78846 0.26282 10.59 0.000 Time 1 0.12278 1.88% 0.42028 0.42028 16.93 0.000 Maint. Cat. 2 3.70207 56.65% 3.69586 1.84793 74.45 0.000 Season*Time 3 0.00595 0.09% 0.00595 0.00198 0.08 0.971 Error 79 1.96099 30.01% 1.96099 0.02482 Lack - of - Fit 13 0.34869 5.34% 0.34869 0.02682 1.10 0.377 Pure Error 66 1.61230 24.67% 1.61230 0.02443 Total 88 6.53506 100.00% Table 40 displays the ANOVA results for the available PCC layer moduli data within the DF climatic region. The data was transformed to satisfy the normality assumptions, which is essential for drawing meaningful conclusions from ANOVA (see Figure 53 ). A similar da ta transformation was undertaken for the rest of the climatic regions as well. Results show that FWD measurement season and time of the day impacts the PCC layer backcalculated moduli 98 values. The maintenance category, although used as a blocking factor, al so influences the PCC layer moduli. The main effects plot in Figure 54 (a) shows that there is a significant difference in mean PCC modulus within the summer season compared to the rest of the seasons. High temperatures causing the curling down of the slab rendering it unsupported at the center could be a possible explanation for the lowest PCC moduli values in the summer season. The opposite is true for the winter season, where curling up causes a full contact between the slab center and the underlying froz en layers. The minimum mean difference between PCC moduli obtained from FWD tests conducted in summer and any other season is more than 850 ksi [see Figure 54 (b)]. The highest mean PCC modulus difference (more than 1600 ksi) exists between summer and winte r seasons. Figure 53 E nsur ing normality of the PCC moduli data - DF climatic region 99 (a) Main effects plot (b) Seasonal mean differences with 95% confidence intervals Figure 54 ANOVA plots for PCC moduli data - DF climatic region There is a noticeable difference (more than 1000 ksi) between the PCC moduli obtained using FWD deflections measured before - noon and afternoon within a day. Such effects are possible due to the temperat ure difference between the two times of the day, causing the slab to transition from curled up in the morning to curled down in the afternoon. This transition in the moduli values from morning (i.e., before noon) to the afternoon. Table 41 ANOVA results for PCC moduli values D N F climatic region Source DoF Seq SS Contribution Adj SS Adj MS F - v alue p - v alue Season 3 4937928 1.34% 1989360 663120 0.33 0.801 Time 1 1374659 0.37% 923473 923473 0.46 0.497 Maint. Cat. 1 100613678 27.38% 92756115 92756115 46.59 0.000 Season*Time 3 9746626 2.65% 9746626 3248875 1.63 0.185 Error 126 250842641 68.25% 250842641 1990815 Lack - of - Fit 4 8903891 2.42% 8903891 2225973 1.12 0.349 Pure Error 122 241938750 65.83% 241938750 1983105 Total 134 367515533 100.00% ANOVA results for the DNF climatic region, as shown in Table 41 , displays that only the maintenance category, used as a blocking factor, is significant having an influence on the PCC 100 modulus values in this climatic region. Neither the FWD measurement season nor the time of the day has any essential effects on the PCC layer moduli (see Figure 55 ). Insufficient data could be a reason for such results as only two JPCP sections are available within the SMP database in this climatic region. Figure 55 Main effects plot for PCC moduli data DNF climatic region Table 42 ANOVA results for PCC moduli values W F climatic region Source DoF Seq SS Contribution Adj SS Adj MS F - v alue p - v alue Season 3 0.25083 3.54% 0.14709 0.049030 2.57 0.055 Time 1 0.48332 6.83% 0.28108 0.281082 14.74 0.000 Maint. Cat. 2 0.88365 12.48% 0.88449 0.442246 23.18 0.000 Season*Time 3 0.02572 0.36% 0.02572 0.008572 0.45 0.718 Error 285 5.43628 76.79% 5.43628 0.019075 Lack - of - Fit 14 0.46040 6.50% 0.46040 0.032885 1.79 0.040 Pure Error 271 4.97589 70.28% 4.97589 0.018361 Total 294 7.07980 100.00% The ANOVA results for the WF climatic region shows that FWD measurement time has a noticeable bearing on the PCC modulus (see Table 42 ). FWD testing season, on the other hand, appears an insignificant factor with no influence on the PCC modulus. Figure 56 ( a) shows the main effects plot for the WF climatic region. There is a considerable difference (500 ksi) 101 between the PCC moduli obtained from FWD deflection measured before - noon and in the afternoon. The higher PCC moduli in the before - noon can be related to the close contact between the slab and the layer beneath due to curling (i.e., curl up in the morning while curl down in the afternoon). Figure 56 (b) shows the seasonal differences between PCC moduli values for different seasons. Although statistically insignificant at the type - I error rate of 5%, it can be termed significant practically (or at a type - I er (greater than 400 ksi) between some of the seasons (winter versus spring and summer). (a) Main effects plot (b) Seasonal mean differences with 95% confidence intervals Figure 56 ANOVA plots for PCC moduli data - W F climatic region Table 43 ANOVA results for PCC moduli values W N F climatic region Source DoF Seq SS Contribution Adj SS Adj MS F - v alue p - v alue Season 3 17449689 3.93% 13337363 4445788 2.40 0.069 Time 1 25240885 5.69% 25853245 25853245 13.95 0.000 Maint. Cat. 1 4217980 0.95% 4045433 4045433 2.18 0.141 Season*Time 3 131875 0.03% 131875 43958 0.02 0.995 Error 214 396734753 89.40% 396734753 1853901 Total 222 443775183 100.00% The ANOVA results for the WNF climatic region are similar to those of the WF region. Table 43 displays the ANOVA results for WNF climatic region; deflections measurement time 102 appears to influence the PCC layer moduli values while the FWD measurement season has no bearing the PCC layer moduli. Figure 57 (a) displays the main effects plot for the ANOVA on the PCC moduli data within WNF climatic region. The mean PCC moduli values obtained from deflections measured in the morning (i.e., before - noon) are signific antly higher (> 700 ksi) than those obtained from the afternoon FWD tests. appears to have a practical impact and can be termed significant a type - I error rate of 10% ( 0.10). Figure 57 (b) shows that considerable differences (higher than 500 ksi) between PCC moduli exist within different seasons (i.e., winter versus spring and summer, fall versus spring, and summer). Wider confidence intervals for the seasonal mean PC C moduli differences observed in the WF and the WNF climatic regions may be due to insufficient data to explain the seasonal and diurnal effects on these moduli [see Figure 56 (b) and Figure 57 (b)]. (a) Main effects plot (b) Seasonal mean differences with 95% confidence intervals Figure 57 ANOVA plots for PCC moduli data - W N F climatic region 103 Modulus of Subgrade Reaction (k - value) Similar to the PCC layer modulus, FWD mid - slab deflections are also used to calculate the modulus of subgrade reaction ( k - value). This study uses the available k - values in the LTPP SMP database to investigate the effects of seasonal and diurnal on this pavement parameter. Figure 58 (a) shows the overall variations in the k - values with different FWD passes in different climati c regions. The k - values decrease with increasing pass number (i.e., morning to afternoon), which may be related to the curling down of the PCC slab with the rise in temperature as the day passes. However, there are somewhat mixed trends observed in the k - v alues with months between different climates [see Figure 58 (b)]. 104 (a) FWD Pass (b) FWD Measurement Month (c) Histogram available k - values (d) Box - plot available k - values Figure 58 Assessing available k - values Figure 58 (c) and (d) displays the distribution of the k - values within different climatic regions, and Table 44 presents its descriptive statistics. The wet climatic regions show higher variability in the backcalculated k - values. The data also shows very high k - values within each climate. Including such values in the ANOVA can mask the findings. Thus, k - values higher than 250 pci were not used in the analysis. 105 Table 44 Descriptive statistics k - values SMP JPCP pavements Climatic region N Mean St d. Minimum Q1 Median Q3 Maximum DF 113 245.1 50.1 137 209.5 243 276 394 DNF 156 223 51.2 110 187 212.5 265.7 346 WF 297 178.5 77.6 67 121 155 238.5 661 WNF 236 175.9 60.2 66 142 163 204.7 355 Note: k - values shown are in pci units. Table 45 ANOVA results for k - values DF climatic region Source DoF Seq SS Contribution Adj SS Adj MS F - v alue p - v alue Season 3 1635 3.16% 163.8 54.61 0.10 0.957 Time 1 6286 12.14% 554.2 554.25 1.06 0.308 Maint. Cat. 2 10252 19.79% 10333.4 5166.71 9.90 0.000 Season*Time 3 7530 14.54% 7529.7 2509.89 4.81 0.005 Error 50 26093 50.38% 26093.3 521.87 Lack - of - Fit 11 5080 9.81% 5079.8 461.80 0.86 0.587 Pure Error 39 21013 40.57% 21013.5 538.81 Total 59 51796 100.00% ANOVA results revealed that interaction between the deflections measurement season and time of the day influences the k - values rather than the individual factors (see Table 45 ). Interaction being of significance is interesting because one would expect the season to be more relevant as compared to the time of the day for unbound layers. Figure 59 (a) shows the interaction means plot for season and time interaction effects on k - va lues. Before - noon FWD deflection measurements result in higher values, which is consistent with the trend seen for PCC moduli. During this time of the day, the slab is in close contact with the layers beneath it due to curling down during the winter season and near - flat slab condition in the spring and summer seasons. Curling down of the PCC slab is a possible explanation for the lower values in the afternoon for all the seasons except fall. The fall season is an exception where the k - values are lower in th e morning (i.e., before - noon) than in the afternoon. The mean differences between all seasons except winter are practically noticeable (higher than 20 pci) , but broader confidence 106 intervals (due to insufficient data) are displaying the differences to be ot herwise [see Figure 59 (b)]. (a) Interaction means plot (b) Seasonal mean differences with 95% confidence intervals Figure 59 ANOVA plots for k - values - DF climatic region Table 46 ANOVA results for k - values D N F climatic region Source DoF Seq SS Contribution Adj SS Adj MS F - v alue p - v alue Season 3 1171133182 9.68% 1266565893 422188631 4.40 0.006 Time 1 920437865 7.61% 1009560297 1009560297 10.52 0.002 Maint. Cat. 1 145359751 1.20% 146136486 146136486 1.52 0.220 Season*Time 3 270840478 2.24% 270840478 90280159 0.94 0.424 Error 100 9592745466 79.28% 9592745466 95927455 Lack - of - Fit 5 386114724 3.19% 386114724 77222945 0.80 0.555 Pure Error 95 9206630742 76.08% 9206630742 96911903 Total 108 12100516743 100.00% For the DNF climatic region, ANOVA reveals the deflection measurement season and time of the day as significant factors having an impact on the k - values (see Table 46 ). Curling of the PCC slab may be the reason for the effect of time on the k - values. Figur e 60 (a) shows the main effects plot from the ANOVA analysis. It shows that k - values are highest in the winter season and have a mean difference of 15 - 20 pci as compared to the values in spring and summer seasons [see Figure 60 (b)]. The figure also shows a noticeable difference of around 20 pci between k - 107 values obtained from deflections measured in the morning and the afternoon. The reason for the diurnal change, as mentioned earlier, is linked to the curling of the PCC slab. (a) Main effects plot (b) Seasonal mean differences with 95% confidence intervals Figure 60 ANOVA plots for k - values - D N F climatic region Looking at the seasonal and diurnal effects on the k - values in the WF climatic region, ANOVA shows that the deflections measurement season ha s a noticeable impact (see Table 47 ). The main effects plot and the seasonal mean difference plot, in Figure 61 , show that the difference of k - values between fall season as compared to spring and summers is not significant (about 10 pci). However, the k - va lues in winters are considerably different (difference higher than 20 pci) then the rest of the seasons. Table 47 ANOVA results for k - values WF Climatic region Source DoF Seq SS Contribution Adj SS Adj MS F - v alue p - v alue Season 3 0.000080 9.10% 0.000071 0.000024 7.37 0.000 Time 1 0.000022 2.47% 0.000006 0.000006 1.83 0.177 Maint. Cat. 2 0.000048 5.43% 0.000046 0.000023 7.24 0.001 Season*Time 3 0.000005 0.57% 0.000005 0.000002 0.52 0.668 Error 226 0.000721 82.44% 0.000721 0.000003 Lack - of - Fit 14 0.000079 9.03% 0.000079 0.000006 1.86 0.032 Pure Error 212 0.000642 73.41% 0.000642 0.000003 Total 235 0.000875 100.00% 108 (a) Main effects plot (b) Seasonal mean differences with 95% confidence intervals Figure 61 ANOVA plots for k - values - W F climatic region Table 48 ANOVA results for k - values W N F Climatic region Source DoF Seq SS Contribution Adj SS Adj MS F - v alue p - v alue Season 3 26.754 4.98% 28.308 9.436 4.05 0.008 Time 1 32.703 6.09% 29.396 29.396 12.61 0.000 Maint. Cat. 1 2.473 0.46% 1.877 1.877 0.81 0.371 Season*Time 3 27.581 5.14% 27.581 9.194 3.94 0.009 Error 192 447.504 83.33% 447.504 2.331 Total 200 537.016 100.00% The ANOVA results using the available k - values in the WNF climatic region shows that the interaction between the deflections measurement season and time of the day has a bearing on these values (see Table 48 ). This warrants looking at the interaction means plot to understand the temporal (seasonal and diurnal) effects on the obtained k - values. Figure 62 (a) shows that there is no noticeable difference between k - values obtained at different times of the day in the winter and fall seasons. However, a considerable difference exists between k - values obtained using FWD deflections measured in the spring and summer seasons. The mean difference in k - values is about 30 pci in spring, while in the summer season, the mea n difference reaches up to 60 pci from morning to the afternoon [see Figure 62 (b)]. 109 (a) Interaction means plot (b) Seasonal mean differences with 95% confidence intervals Figure 62 ANOVA plots for k - values - WN F climatic region Discussion on ANOVA Results (PCC modulus and k - values) The same FWD deflection measurements conducted on the mid - slab of a rigid pavement are used to obtain the elastic modulus of the PCC slab and the modulus of subgrade reaction ( k - value). Ambient and surface temperature val ues accompany each of the FWD drops. A better insight into the effects discussed earlier on the PCC moduli and k - values is possible by considering the measured temperatures. This section presents a discussion on the temperatures obtained from the LTPP SMP database while linking them to the observed effects. Figure 63 shows the pavement surface and ambient temperatures recorded during FWD measurements in the DF climatic region. Figure 63 (a) and (b) are useful as the main effects of season and time were signi ficant factors according to ANOVA on the PCC moduli values in the DF climate. Figure 63 (c) is essential as the interaction of the two factors (i.e., season and time) turned out to be significant while looking into the effects on k - values. 110 (a) Seasonal temperatures ranges (b) Diurnal temperature ranges (c) Temperature ranges for season and time interaction Figure 63 Pavement surface and ambient temperatures DF climatic region PCC slab modulus (E pcc ) and the corresponding k - values are lowest in the summer in DF climatic region. The high ranges of the ambient temperatures (75 90 F) in the summer season are the reason for such effects that causes the slab to curl down, rendering the slab unsupported at the center. Avoiding FWD tests at such high temperatures is beneficial as the obtained values (E pcc and k ) will not be the true representative of the actual pavement condition. Moderate ambient temperatures (55 70 F) recorded in the spring season in the DF climatic region are probably be tter to obtain these values. Diurnally, Figure 63 (b) suggests that before - noon 111 deflection measurements will produce results close to the in - field pavement conditions. However, avoiding early morning times may prove useful. (a) Seasonal temperature ranges (b) Diurnal temperature ranges Figure 64 Pavement surface and ambient temperatures D N F climatic region Based on the ANOVA results, none of the factors, i.e., season and time, significantly affect the PCC slab modulus in the DNF climatic region. However, a probable reason for such effects is insufficient data for the JPCP sections in the SMP database. On the other hand, looking at k - values, ANOVA showed significant season and time influence. Figure 64 (a) suggests the best temperature ranges to conduct FWD in the DNF climatic region are between 60 70 F shown within the spring season. Such temperatures will r esult in PCC moduli values that are near to the mean (see Figure 55 ). Thus, the obtained k - value will represent the in - field modulus accurately. As far as a better time of the day is concerned, where possible, before - noon measurements could be beneficial. Figure 64 (b), however, shows that ambient temperatures recorded in the morning Figure 65 shows the seasonal and diurnal temperature ranges recorded during FWD measurements in the WF climatic region. ANOV and time of the day have a bearing on the PCC layer modulus. On the other hand, only the season 112 impacts k - values as per ANOVA results in the WF climate. Figure 65 (a) suggests the ambient temperature range of 55 - 65 F occurring in the fall seasons to be a better candidate for FWD measurements in the WF climate that can result in representative E pcc and k - values closest to the real conditions. Also, the suggested ambient temperature ranges (55 - 65 F) advocates before - noon FWD measurements [see Figure 65 (b)]. (a) Seasonal temperature ranges (b) Diurnal temperature ranges Figure 65 Pavement surface and ambient temperatures W F climatic region (a) Seasonal temperature ranges (b) Diurnal temperature ranges Figure 66 Pavement surface and ambient temperatures W N F climatic region 113 ANOVA results showed that the time of FWD measurement within a day is an influencing factor for the PCC layer modulus in the WNF climatic region. On the other hand, season and time i nteraction is essential to explain variations of k - values in the WNF climatic region, as shown by the ANOVA results. Figure 66 suggests that the ambient temperature range between 60 - 75 F occurring in the fall season is ideal in the WNF climatic region th at will result in realistic E pcc and k - value. Diurnally, such a temperature range also suggests before - noon FWD measurements. Figure 67 shows the histogram of the available temperature gradients data available in the SMP database for the JPCP sections (for E pcc and k - values). Zero or near - zero temperature gradients would be ideal for measuring deflections resulting in E pcc and k - values that are close to the in - field actual values. Around 50 FWD measurements used to obtain these pavement parameters had a tem perature gradient between - 1 to 1. All these measurements occurred from 8 am to about noon. Thus, suggesting that diurnally before noon measurements are better, based on the available data, to find a zero gradient temperature condition. Figure 67 Histogram of the available temperature gradients data for E pcc and k - values SMP JPCP sections 114 (a) PCC moduli for Arizona section ID 4 - 0215 DF climatic region (b) PCC moduli for Nebraska section ID 31 - 3108 WF climatic region (c) k - Values for Arizona section ID 4 - 0215 DF climatic region (d) k - Values for Nebraska section ID 31 - 3108 WF climatic region Figure 68 Variation in PCC layer moduli and k - values with temperature gradient within the PCC layer Figure 68 shows examples of two JPCP sections, each from the DF and the WF climatic regions. The plots show PCC moduli and k - values determined from FWD measurements on different days in multiple years. The plots display a decreasing trend in the PCC slab modulus and the corresponding k - values as the tempera ture gradient shifts from negative to positive (i.e., before noon to afternoon). The higher PCC layer moduli and k - values when the temperature gradient is negative are due to the PCC slab curling up with the bottom of the mid - slab in 115 complete contact with the underlying layers. The opposite happens as the temperature gradient becomes positive, causing the PCC slab to curl down; when the mid - slab is unsupported. Also, it is noteworthy that both the rigid pavement parameters (E pcc and k - value) follow a simila r trend while the temperature gradients change from negative to positive. Thus, PCC moduli are more critical in formulating general guidelines for FWD testing on rigid pavements than k - values. Load Transfer Efficiency (LTE) Figure 69 illustrates the effects of diurnal and seasonal FWD deflection - based LTE values in different climatic zones for all JPCP pavement sections in the SMP experiment. It also presents the distribution of the available data within each climatic region. It shows that LTE values are affected by diurnal and seasonal measurements. Higher LTE values are expected at higher temperatures due to slab expansion and joint locking while lower LTE value correspond s to slab contraction in lower temperatures. Table 49 shows th e descriptive statistics for LTE values for the available JPCP sections along with the available number (N) of the LTE measurements within each climatic zone. A higher variation evident from the higher standard deviation (Std . ) in the freeze regions. There fore, it is essential to consider the diurnal and seasonal temperatures for such measurements for rigid pavements. Table 49 Descriptive statistics LTE values Climatic region N Mean Std. Minimum Q1 Median Q3 Maximum DF 234 58.9 27.5 15.2 30.7 66.2 85.3 96.8 DNF 326 69.2 17 25 53.8 71.6 84.2 96.9 WF 610 70.6 26 16 44.3 85 91.5 98.2 WNF 478 75.2 19 18.4 57.8 86.700 90.3 97 Note: LTE values shown as a percentage (%). 116 (a) LTE(%) by FWD pass (b) LTE(%) by FWD measurement month (c) Histogram - available LTE (%) values (d) Box - plot - available LTE (%) values Figure 69 Assessing available LTE data Figure 70 demonstrates the normality of the data for the DNF climatic region that was used in the analysis. As mentioned earlier, the satisfaction of the normality assumptions is essential to draw meaningful conclusions from the ANOVA analysis. Similarly, the data were transformed adequately for the rest of the climatic regions to ensure the satisfaction of the normality assumptions. 117 Figure 70 Evaluating normality of the LTE data - DNF climatic region Table 50 presents the results of the ANOVA analysis for the DF climatic region. The results show that the FWD measurement season significantly affects the LTE values based on a type - significant effect on the LTE values at the same error rate. The results also show that the interaction between the two factors, i.e., season and time, also contributes significantly to the variations in the LTE values. Thus, looking at the interaction effects rather than the main effects is essential. Figure 71 (a) shows the mean LTE value in the summer season varies from 68% for before - noon measurement to 85% in the afternoon. Also, the LTE values change in the spring and fall seasons in between before - noon and afternoon measurements; however, the difference is not as significant as in summer. Figure 71 (b) also shows that the difference of mean LTE values between any pair of season and time interaction involving the summer season is substantial. Thus, these results suggest that summer season LTE values are sig nificantly higher from other 118 seasons irrespective of the time of the day with a more significant difference in the afternoon measurements. Table 50 ANOVA results for LTE values DF climatic region Source DoF Seq SS Contribution Adj SS Adj MS F - v alue p - v alue Season 3 20540 12.94% 20158 6719.5 25.52 0.000 Time 1 1010 0.64% 1031 1031.3 3.92 0.049 Maint. Cat. 2 77063 48.56% 73316 36658.2 139.22 0.000 Season*Time 3 3743 2.36% 3743 1247.5 4.74 0.003 Error 214 56350 35.51% 56350 263.3 Lack - of - Fit 14 16188 10.20% 16188 1156.3 5.76 0.000 Pure Error 200 40163 25.31% 40163 200.8 Total 223 158707 100.00% To further explain the summer LTE values, Figure 71 (c) shows the 95% confidence interval of the pavement surface and air temperatures at the time of FWD measurements. There is a clear difference between the temperatures recorded between the before - noon and afternoon times while conducting FWD measurements. Thus, suggesting that elevated temperatures beyond 80°F should be avoided while conducting FWD testing on joints. A similar ANOVA analysis for the DNF climatic region also reveals that the interaction of the two factors, that is, the FWD measurement season and time has an essential influence on the LTE values rather tha n the individual factors (see Table 51 ). The maintenance category is also a contributing factor influencing the LTE values; however, it is being used as a blocking factor. The interaction means do not show a significant difference between LTE values measur ed before noon and in the afternoon [see Figure 72 (a)]. Also, the overall mean LTE values are higher than for DF climatic region. A possible explanation can be the higher range of measurement temperatures, i.e., around and over 75°F, which should be avoide d while measuring LTE at joints [see Figure 72 (b)]. 119 (a) Interaction means plot (b) Mean difference of LTE values with 95% confidence interval for season and time interaction (c) Pavement surface and air temperatures during FWD measurements Figure 71 ANOVA results - DF climatic region 120 Table 51 ANOVA results for LTE values DNF climatic region Source DoF Seq SS Contribution Adj SS Adj MS F - v alue p - v alue Season 3 547985885 32.70% 520306982 173435661 56.42 0.000 Time 1 33623209 2.01% 17249536 17249536 5.61 0.018 Maint. Cat. 1 87039916 5.19% 94121166 94121166 30.62 0.000 Season*Time 3 32552731 1.94% 32552731 10850910 3.53 0.015 Error 317 974519223 58.16% 974519223 3074193 Lack - of - Fit 6 179477449 10.71% 179477449 29912908 11.70 0.000 Pure Error 311 795041774 47.44% 795041774 2556404 Total 325 1675720963 100.00% Table 52 ANOVA results for LTE values WF climatic region Source DoF Seq SS Contribution Adj SS Adj MS F - v alue p - v alue Season 3 756157660 13.11% 680113939 226704646 34.88 0.000 Time 1 154530676 2.68% 43189635 43189635 6.65 0.010 Maint. Cat. 2 1058360121 18.35% 1062049115 531024558 81.71 0.000 Season*Time 3 4293469 0.07% 4293469 1431156 0.22 0.882 Error 584 3795286065 65.79% 3795286065 6498778 Lack - of - Fit 14 414978033 7.19% 414978033 29641288 5.00 0.000 Pure Error 570 3380308032 58.60% 3380308032 5930365 Total 593 5768627991 100.00% The ANOVA analysis for wet regions shows that the main effects of the two factors, namely season and time, are significant contributors to the variations in LTE while their interaction is not (see Table 52 and Table 53 ). The maintenance category, used as a blocking factor, is also contributing to the variation of LTE values in the WF region. However, it is not a significant factor in the WNF region. The main effects plot in Figure 73 (a) shows that the mean LTE value in summer is significantly higher than other seasons, while lowest in winter season in the WF climate region [also see Figure 73 (c)]. While for the WNF climatic region, the LTE values are statistically different [see Figur e 73 (b) and Figure 73 (d)] between all the seasons. However, the difference between winter and spring season LTE values (>15%) and summer and spring season LTE values (>20%) can be termed as practically different only. 121 (a) Interaction means plot (b) Pavement surface and air temperatures during FWD measurements Figure 72 ANOVA results - DNF climatic region As discussed earlier, the higher LTE values in the summer season in the WF climatic region is explained based on the temperature range during the F WD measurements. The temperature range in summer for the WF climatic region is between 80 - 84°F, which causes an increase in LTE values [see Figure 74 (a)]. Similarly, the difference between the seasonal LTE values in the WNF climate is also evident from the temperature ranges during FWD testing [see Figure 74 (b)]. 122 Table 53 ANOVA results for LTE values WNF climatic region Source DoF Seq SS Contribution Adj SS Adj MS F - v alue p - v alue Season 3 47986 33.88% 49449.3 16483.1 86.60 0.000 Time 1 3256 2.30% 3217.5 3217.5 16.91 0.000 Maint. Cat. 1 352 0.25% 342.5 342.5 1.80 0.180 Season*Time 3 1161 0.82% 1160.7 386.9 2.03 0.108 Error 467 88882 62.75% 88882.4 190.3 Total 475 141637 100.00% Observing the diurnal trends in both the WF and WNF climates, in Figure 73 (a) and Figure 73 (b), it is found that before noon measurements result in a lower LTE value as compared to an afternoon measurement. Figure 74 (c) and Figure 74 (d) shows the diurnal t emperature ranges between the wet climates. In the WF region, the before - noon air temperatures are not very different from the afternoon temperatures; however, there is a significant difference between the corresponding pavement surface temperature ranges. Such temperature difference of the pavement surface causes the LTE values measured before - noon to be lower than the afternoon LTE values. The difference in the diurnal temperature ranges, in the WNF climatic region, explains the lower before - noon LTE valu es as opposed to afternoon LTE values. Generally, FWD measurements conducted before noon result in lower LTE values as compared to those measured in the afternoon, irrespective of the climatic region. Therefore, it is beneficial to perform deflection measu rements on joints for LTE determination before noon when slabs are flatter. Also, Figure 75 shows the histogram of the temperature gradients data available in the SMP database for the JPCP sections. Ideally, a zero or near - zero temperature gradient would b e ideal for measuring deflections at joints and, ultimately LTE, as such LTE values will depict the actual load transfer capacity of the joint. Around 50 measurements were found in the data where the temperature gradient was between - 1 to 1. The time range of these measurements was found to be ranging from 8 am to around 1 pm. Hence, the available data also 123 suggests that diurnally before noon measurements are ideal to see a zero gradient temperature condition, which can help to identify joints condition acc urately. (a) Main effects plot for LTE - WF climatic region (b) Main effects plot for LTE - WNF climatic region (c) The seasonal mean difference of LTE values with 95% confidence WF climatic region (d) The seasonal mean difference of LTE values with 95% confidence WF climatic region Figure 73 ANOVA results WF and WNF climatic regions 124 (a) WF climatic region (b) WNF climatic region (c) WF climatic region (d) WNF climatic region Figure 74 Seasonal and diurnal pavement surface and air temperatures Figure 75 Histogram of the available temperature gradients data for LTE values SMP JPCP sections 125 (a) Arizona section ID 4 - 0215 DF climatic region (b) Nebraska section ID 31 - 3018 WF climatic region Figure 76 Variation in LTE with temperature gradient - JPCP sections Figure 76 displays examples of LTE for two JPCP sections, each in the DF and WF climatic regions. The plots display the increasing trend of LTE as the temperature gradient within the PCC layer shi fts from negative to positive within a day (i.e., before noon to afternoon). The displayed LTE values were determined on different days in multiple years. The shift of the PCC slab shape from curled up to curled down causes the LTE to increase as the day p asses. 4.2.2 Longitudinal Profile Measurements (IRI) JPCP Pavements The functional performance of new or existing pavements is assessed in terms of smoothness, which is estimated using longitudinal profile measurements. However, the longitudinal profiles of J ointed Concrete Pavements (JCP) are affected significantly by temporal (seasonal and diurnal) variations in temperature and moisture. These temporal variations translate into influencing the curling and warping of the concrete slabs of JPCP. Therefore, it is essential to consider such effects for accurately assessing the pavement condition. 126 The longitudinal profile measurements are commonly summarized by the International Roughness Index (IRI) that reduces the thousands of elevation values into a single va lue [38, 39] . However, no matter which index is calculated from a longitudinal profile, the quality of the information is only as good as the profile measurement [2] . Thus, there is a need to evaluate the impacts of temporal (seasonal temperature/moisture and daily temperature) variations on longitudinal profile measurements, especially for JCPs. The steps involved in the analysis are enumerated as follows: 1. The pavement structure details, IRI values, profile measurement dates and timings, maintenance histor y, pavement surface temperature, and air temperature were some of the variables identified to perform the analysis. 2. The investigation required an analysis involving the available JPCP sections in the LTPP SMP database. 3. The available JPCP pavement sections were identified in the SMP LTPP database with their climatic region. Note that the climatic regions are defined based on temperature (i.e., freeze/no freeze), and moisture (i.e., wet/dry). 4. Obtained slab thickness es for each JPCP section s. 5. The IRI values, along with the date and time of profile measurements, the air and pavement surface temperatures were obtained. Also, the maintenance history details for the section were extracted from the database (i.e., construction no. and type of treatment) and categorized. 6. Five profile measurements (runs) per visit are the LTPP standard [44] ; the analysis used each of the calculated IRI values. 7. All the required data elements were arranged in a relational database. 127 8. The data were inspected using a histogram and boxplot to identify outliers. 9. Two factors were used in this analysis; (a) measurement month discretized into four seasons (i.e., levels) to look at the seasonal effects, (b) measurement time with two levels (before noon and afternoon) for the diurnal effec ts. Besides, the maintenance category was used as a blocking factor. 10. Analysis of Variance (ANOVA) was conducted for JPCP sections within each climatic region to investigate the temporal effects on the joint LTE values. Figure 77 shows the overall variation s in the mean IRI values by the hour and month of profile measurements within different climatic regions. The IRI values show a significant hourly influence, i.e., the IRI values are different in each hour among the different climatic regions [see Figure 7 7 (a)]. Also, significant variations can be seen in the IRI values among different months, with higher values in the freeze regions [see Figure 77 (b)]. Figure 77 (c) and (d) show the histogram and box - plot, while Table 54 displays the descriptive statistics of the data within each climatic region. There is generally higher variability in the IRI values within different climates, with the highest in the WF climatic region. The variation in the WF climatic region may be due to IRI values over 250 inch/mile. Suc h higher values can potentially mask the results of ANOVA. The IRI analysis presented in this study uses values between 70 - 170 inch/mile. Table 54 Descriptive statistics IRI values SMP JPCP pavement sections Climatic region N Mean St d. Minimum Q1 Median Q3 Maximum DF 258 122.1 26.1 54.8 110.9 124.4 136.9 185.3 DNF 474 95.7 27.1 47.1 65.6 98.5 119.2 138.5 WF 1038 127.2 53.2 49.1 96.4 121.8 151.1 274 WNF 969 102.4 23 45.6 86 102.3 117.8 163 Note: IRI values shown are in inch/mile units. 128 (a) IRI by hour (b) IRI by month (c) Histogram - Available IRI data (d) Box - plot - Available IRI data Figure 77 Assessing available I RI data SMP JPCP pavement sections Figure 78 shows the normality of the data used for the IRI analysis of the DF climatic region after a suitable transformation. As mentioned earlier, the satisfaction of the normality assumptions is critical to draw a meaningful conclusion from an ANOVA. Similar to the DF climatic region, data for all other climatic regions were transformed to ensure the satisfaction of the normality assumptions. 129 Figure 78 Evaluating normality of the IRI data - DF climatic region Table 55 shows the ANOVA results for IRI analysi s of the DF climatic region. Results reveal that profile measurement season and time influence the IRI values in the DF climatic region. The main effects plot in Figure 79 (a) shows a noticeable difference in IRI values between the different seasons. Multip le mean comparisons plot reveals that there is no significant mean difference between fall and winter and summer and spring seasons [see Figure 79 (b)]. It also shows that a mean difference higher than 12 inches/mile exists between the other season combinat ions (i.e., spring and winter, summer and winter, fall and spring, and fall and summer). Although statistically significant, the difference between IRI obtained from profile measurements before - noon and afternoon is not practically noticeable (< 6 inches/m ile). For an explanation of the effects seen in the DF climatic region, it is essential to know the ambient and surface PCC temperature at the time of the profile measurements. Figure 80 (a) shows the ambient temperature range between different seasons duri ng profile measurements. 130 Pavement surface temperature data were available for the spring season only. Low temperatures (below 40 F) explain higher IRI observed in the winter season. The increase in IRI could be due at such temperatures. Whereas, downward curling of the JPCP slab at elevated temperatures (i.e., around 80 F) results in low IRI. The drop in IRI in the spring season can be related to the corresponding high pavement surface temperature causing the slab to curl down. On the other hand, the practically insignificant difference between before - noon and afternoon IRI is because the temperature ranges do not vary between these timings [see Figure 80 (b)]. Based on the discussion and Figure 80 , profile measuremen ts at an ambient temperature range of 55 65 F recorded in the spring and fall seasons are suggested in the DF climatic region. Also, considering the diurnal trends, the pavement profiles are better to measure in the afternoon. Table 55 ANOVA results for IRI values D F climatic region Source DoF Seq SS Contribution Adj SS Adj MS F - v alue p - v alue Season 3 10364 12.27% 10687 3562.5 20.30 0.000 Time 1 2360 2.79% 1472 1472.4 8.39 0.004 Maint. Cat. 2 31102 36.82% 29127 14563.3 82.97 0.000 Season*Time 3 1325 1.57% 1325 441.7 2.52 0.059 Error 224 39318 46.55% 39318 175.5 Lack - of - Fit 12 13392 15.85% 13392 1116.0 9.13 0.000 Pure Error 212 25926 30.69% 25926 122.3 Total 233 84469 100.00% 131 (a) Main effects plot (b) Seasonal mean IRI differences Figure 79 ANOVA plots for IRI data - DF climatic region Table 56 ANOVA results for IRI values D N F climatic region Source DoF Seq SS Contribution Adj SS Adj MS F - v alue p - v alue Season 3 2091.2 2.43% 8148 2715.8 15.93 0.000 Time 1 37.5 0.04% 3639 3638.9 21.35 0.000 Maint. Cat. 1 25864.0 29.99% 24336 24336.4 142.76 0.000 Season*Time 3 4535.3 5.26% 4535 1511.8 8.87 0.000 Error 315 53699.8 62.28% 53700 170.5 Lack - of - Fit 1 1012.6 1.17% 1013 1012.6 6.03 0.015 Pure Error 314 52687.2 61.10% 52687 167.8 Total 323 86227.7 100.00% 132 (a) Seasonal temperature ranges (b) Diurnal temperature ranges Figure 80 Pavement surface and air temperatures during profile measurements - DF climatic region ANOVA results for the DNF climatic region shows that the interaction between profile measurement season and time is a pivotal contributor to the variations in IRI value s (see Table 56 ). The interaction means plot illustrates a significant difference between IRI obtained from profile measurements conducted before - noon and the afternoon in spring and summer seasons [see Figure 81 (a)]. Also, before - noon IRI values are highe r than the afternoon ones for all seasons except winter, which displays almost similar IRI values irrespective of the time. Figure 81 (b) shows that there is no significant difference between before - noon and afternoon IRI values in winter and fall seasons. However, there is a considerable difference (> 25 inches/mile) between the IRI values obtained before - noon and afternoon in the summer season. 133 (a) Interaction means plot (b) Mean IRI difference with 95% confidence interval for season and time interaction (c) Pavement surface and air temperatures Figure 81 ANOVA results for IRI - DNF climatic region 134 The pavement surface temperature ranges for before - noon and afternoon timings in the winter season does not vary; hence, the similar IRI values [see Figure 81 (c)]. The driving factor for the substantial IRI difference in the summer season is the clear distinction between the ambient temperatures in the season (pavement surface temperatures are unavailable in the data). Such a significant temperature d ifference could affect the level of the slab curling; thus, changes the IRI substantially between the two times (i.e., before - noon and the afternoon). Also, the IRI difference observed during the spring season (around 10 inches/mile) seen in Figure 81 (b) i s also explained by the noticeable difference between the ambient and pavement surface temperatures. Basing on the discussion and Figure 81 (c) suggests profile measurements at ambient temperatures of 60 70 F occurring in the winter and fall seasons in th e DNF climatic region. Additionally, the afternoon pavement profile appears beneficial, considering the overall diurnal trends. Table 57 ANOVA results for IRI values W F climatic region Source DoF Seq SS Contribution Adj SS Adj MS F - v alue p - v alue Season 3 120876738 0.58% 36801493 12267164 0.59 0.624 Time 1 623684726 2.99% 599793853 599793853 28.67 0.000 Maint. Cat. 2 6130412326 29.37% 6129409617 3064704809 146.51 0.000 Season*Time 3 48800265 0.23% 48800265 16266755 0.78 0.507 Error 667 13952259180 66.83% 13952259180 20917930 Lack - of - Fit 14 1293556633 6.20% 1293556633 92396902 4.77 0.000 Pure Error 653 12658702547 60.64% 12658702547 19385456 Total 676 20876033235 100.00% Table 57 shows that the time of the profile measurement within a day influences the IRI of pavement sections in the WF climatic region. However, the statistically significant difference (less than 8 inches/mile) is not practical [see Figure 82 (a)]. Although there is a definite difference in ambient temperatures between before - noon and afternoon times, however, the pavement surface temperatures are not very different (larger error bars are due to insufficient availability of pavement surface temperatures data). Thus , the corresponding difference in IRI 135 values is not practically significant. The discussion on the ANOVA results along with temperature ranges shown in Figure 82 (b) suggests profile measurements within the ambient temperature range of 50 - 65 F; conducted in the afternoon in the WF climatic region (a) Main effects plot (b) Pavement surface and air temperatures Figure 82 ANOVA results for IRI data - WF climatic region ANOVA results for the WNF climatic region are almost similar to the WF climatic region, with the only difference that the profile measurement season is the significant factor influencing Table 58 ). However, the mean difference betwe en any two seasons is not practically noteworthy (i.e., less than 6 inches/mile), which is explained by the similar ambient and pavement surface temperature ranges observed during the profile measurements [see Figure 83 (a) and (b)]. Table 58 ANOVA results for IRI values W N F climatic region Source DF Seq SS Contribution Adj SS Adj MS F - v alue p - v alue Season 3 0.4160 1.29% 0.5137 0.17125 4.85 0.002 Time 1 0.0244 0.08% 0.0393 0.03926 1.11 0.292 Maint. Cat. 1 0.4832 1.49% 0.5154 0.51542 14.58 0.000 Season*Time 3 0.1949 0.60% 0.1949 0.06498 1.84 0.139 Error 883 31.2090 96.54% 31.2090 0.03534 Total 891 32.3276 100.00% 136 An interesting finding from the ANOVA results is that in the WNF climatic region, IRI values in the winter season are the lowest while those in the spring are the highest. Such effects are possibly due to relatively moderate winter temperatures, while the pavement surface temperatures corresponding to the spring season are much high. The high pavement temperatu res coupled with moisture could be a possible explanation of the IRI effects observed in the spring season. The main effects plot shows that the mean IRI values do not differ between the different seasons within the WNF climatic region. Thus, suggesting th at the ambient temperature ranges within these seasons could be better for profile measurements. Observing the temperature ranges in Figure 83 (b) indicates that the ambient temperature range between 50 - 65 F could be better to measure profile that could g enerate near actual IRI values. Additionally, suggested time for profile measurements is in the morning (i.e., before noon) within this climatic region. (a) Main effects plot (b) Pavement surface and air temperatures Figure 83 ANOVA results for IRI data - W N F climatic region 137 (a) Arizona section ID 4 - 0215 DF climatic region (b) Nebraska section ID 31 - 3018 WF climatic region Figure 84 Variation in IRI with temperature gradients JPCP sections Figure 84 shows an example of two JPCP sections in the DF and WF climatic regions. These examples display the effect of a temperature gradient within the PCC layer on the e., from morning to afternoon) within a day, the IRI of the pavement decreases displaying the diurnal effects. Also, the figure displays the growth of IRI with age. 4.3 Summary This section presented the ANOVA analysis of the available data arranged in a relational database for each of the identified elements. The factors used in the analysis mainly consist of FWD and profile measurement season and time of the day. For flexible pavement sections, the HMA layer thickness classified into two levels based on the original SMP experiment was also an additional factor. Besides, the maintenance category was used as a blocking factor to increase precision in the analysis by reducing the experimental error. The analysis revealed an essential impact of the season and time of the day on the HMA layer moduli. However, it is less critical owing to the temperature correction procedures for AC 138 layer moduli. The Asphalt Institute temperature correction equation performed well in correcting the HMA layer moduli values. Never theless, since a single FWD deflection measurement helps estimate the base and subgrade layer moduli, the seasonal variation effects become more critical for the unbound layers. The ANOVA results displayed the impact of the FWD measurement season to be sig nificant on the unbound layer moduli. The analysis of the available data for JPCP sections also showed that FWD measurement season and time of the day influence the PCC slab, thus changes the estimated moduli and k - values. The temperature differential wit hin a day between the top and bottom of the PCC slab displayed an essential impact on the load transfer capacity of the JPCP joints as well. However, some of the results from the ANOVA for the JPCP pavements were not very clear due to the data limitation w ithin the SMP experiment database. The season and time of the profile measurements also showed an influence on the IRI of both pavement types based on the ANOVA results. While for the flexible pavement sections, the interaction between the season and time of profile measurements had a pronounced influence, their main effects were dominant over the IRI for the rigid pavements. Also, the shift of temperature gradient from negative to positive displayed a decrease in the IRI of JPCP pavements. The section also related the measurement temperatures of the FWD and profile measurements with the parameters determined from them. Ambient temperature ranges are suggested for each climatic region to conduct FWD and profile measurements in anticipation of getting the mod uli and IRI values that accurately represent the actual pavement conditions. 139 CHAPTER 5 CONCLUSIONS AND RECOMMENDATIONS 5.1 Conclusions Evaluating the structural capacity of a pavement involves measuring and assessing deflections obtained from FWD testing. Longitudinal profile measurements are used to determine the functional performance of pavement in terms of smoothness. However, tempora l changes influence these measurements. Assessing and quantifying the effects of temporal variations can assist in a better understanding of the pavement parameters obtained using these measurements, such as backcalculated layer moduli and IRI. The data f rom the LTPP SMP database was extracted, arranged in different data tables, and analyzed using ANOVA. Data analysis for each element identified involved ANOVA on the climatic region basis. The effects observed and the ambient temperatures recorded at the t ime of FWD and profile measurements were related to formulating general guidelines for these measurements. Analyses of the backcalculated layer moduli for flexible pavements revealed an essential effect of seasonal and diurnal changes. The HMA layer moduli showed minimal variation in the spring and fall seasons in every climatic region. Also, the HMA layer moduli were consistently higher, involving deflections measured before noon. However, temperature correction, generally applied to HMA layer moduli, rend ers the season, time, and temperature of the FWD measurements irrelevant. However, the base and subgrade layer moduli are backcalculated from the same measured deflections using the FWD test conducted on the surface. This practice makes the FWD measurement season, time, and temperatures critical in terms of the moduli values 140 obtained for the unbound layers. Any guidelines concerning FWD testing should, therefore, take into account this fact. Generally speaking, the base and subgrade moduli showed minimal va riation within the temperature ranges occurring in the spring and fall seasons, which lead to concluding that such temperature within each climate region may potentially result in these layer moduli values that are close to their in - field condition. Also, the unbound layer showed no effect of time within a day, suggesting no limits for FWD measurements with regards to a better time for conducting these tests. The PCC layer moduli, k - values, and LTE also exhibit meaningful effects of both the FWD measurement season and time. Interestingly, the temperature ranges within spring and fall seasons appear to have less variation within the PCC layer moduli and LTE values, which are more critical in the case of rigid pavements concerning seasonal and diurnal variatio ns. Therefore, the suggested ambient temperatures for FWD measurements over rigid pavements are those recorded in the spring and fall seasons. Moreover, FWD testing conducted before noon has a better chance of finding a zero or near - zero temperature gradie nts, hence recommended. Profile measurements on flexible pavements do not show a particular trend while dealing with pavement sections on the climatic region basis. However, in general, the effects show that temperature ranges occurring in the fall and spr ing seasons could result in less variable IRI values . While in the WF climatic region, summer season ambient temperatures also displayed little to no variation in determined IRI of the pavement sections. Diurnally, results showed no specific trend in the I RI values within different seasons. Profile measurements on rigid pavements illustrated various trends based on the ANOVA within different climatic regions. The DF region IRI values showed less variation advocating 141 temperature ranges within spring and fall seasons. The results for the DNF climatic region showed lower variability in the determined IRI values within the winter and fall season s . In contrast, the IRI values in the wet regions did not show any particular trend seasonally. As far as diurnal patte rns, afternoon time is better for profile measurements suggested based on the results for all seasons except the WNF climatic region . 5.2 Recommendations for FWD and Profile Measurements The temperature correction of HMA layer moduli makes the backcalculated u nbound layer moduli critical in defining the general guidelines for FWD testing on flexible pavements. The recommended general guidelines for FWD deflection testing based on the data analysis presented in this study are as follow: The recommended ambient t emperature range for FWD testing on flexible pavements in the freeze (i.e., DF and WF) climatic regions is between 55 70 F. In the non - freeze (i.e., DNF and WNF) climates, the recommended ambient temperature range to conduct FWD testing on flexible pavem ents is between 65 75 F; however, in the DNF climatic region, the upper limit might be around 80 F. The study recommends no restriction for the FWD testing time during a day . However, the spring and fall seasons are preferable for FWD deflection measurem ents on flexible pavements. For the rigid pavements, the recommended general guidelines for conducting FWD deflection measurements are as follows: The recommended ambient temperature range for FWD testing on rigid pavements in the freeze (i.e., both the DF and WF) climatic regions is between 55 70 F. 142 In the non - freeze (i.e., DNF and WNF) climates, the recommended ambient temperature range to conduct FWD testing on rigid pavements is between 60 75 F. The recommended time for FWD testing on rigid pavement s is before - noon, preferably between 8 am to noon in the spring and fall seasons . The general recommended guidelines for the profile measurements based on the available IRI data are as follows: The recommended temperature range suitable to measure flexible pavement profile is between 50 - 75 F irrespective of the climatic region , season, and time of the day. The recommended temperature range for profile measurement on rigid pavements in freeze (i.e., DF and WF) regions is between 50 - 65 F; while for the non - freeze (i.e., DNF and WNF) climatic regions, profile measurement is suitable between 50 - 70 F. The study recommends spring and fall seasons for DF, winter and fall seasons for DNF, and no season limitation for the WF and WNF climatic regions for profi le measurements on rigid pavements. This study recommends afternoon time to be better for profile measurements on rigid pavements except for the WNF climate region where before noon measurements are beneficial. 5.3 Recommend ed Future Work Agencies use LTE to determine the load transfer of the joints on a JPCP pavement. However, given the shortcomings of the LTE approach, the us e of differential deflection s instead of LTE can reveal the actual joint condition s better . Therefore, ANOVA on dif ferential deflection s can be performed in a future study. 143 Utilization of the a vailable moisture data within the unbound layers and the correlation between the number of wet days and subsurface moisture before the measurements in the LTPP database are recom mended to be incorporated in the data analysis to enhance the moisture - related impacts on the unbound layers moduli. 144 APPENDIX 145 Table 59 Hourly data distribution of backcalculated parameters - SMP AC sections Climate region State State code & section ID Hour of the day Total 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 DF Colorado 8_1053 11 21 22 22 15 14 9 3 1 118 DF Idaho 16_1010 10 26 11 22 6 9 8 4 96 DF Montana 30_0114 2 17 14 11 17 22 6 2 91 DF Montana 30_8129 10 23 24 26 15 16 8 4 126 DF Nevada 32_0101 1 9 28 19 25 22 18 5 127 DF Saskatchewan 90_6405 4 21 15 13 13 12 5 4 87 DF South Dakota 46_0804 5 20 29 29 28 15 10 1 137 DF South Dakota 46_9187 4 12 12 12 12 10 9 2 1 74 DF Wyoming 56_1007 1 13 20 19 21 10 9 4 4 101 DNF Arizona 4_0113 6 26 22 32 30 21 7 3 1 148 DNF Arizona 4_0114 2 24 26 28 24 21 12 5 1 143 DNF Arizona 4_1017 1 3 2 2 2 10 DNF Arizona 4_1018 1 2 2 1 6 DNF Arizona 4_1024 1 6 15 20 12 14 8 4 1 81 DNF New Mexico 35_1112 14 10 18 10 18 10 18 5 103 DNF Utah 49_1001 2 11 13 18 20 10 8 5 3 1 91 WF Connecticut 9_1803 1 9 21 23 26 15 14 4 113 WF Maine 23_1026 2 7 22 19 19 14 18 3 104 WF Manitoba 83_1801 3 24 29 29 18 16 7 7 133 WF Manitoba 83_3802 1 1 2 WF Massachusetts 25_1002 1 24 19 23 21 18 4 110 WF Minnesota 27_1018 1 3 16 14 18 10 15 14 4 95 WF Minnesota 27_1028 4 13 19 14 12 10 8 2 82 WF Minnesota 27_6251 2 9 32 33 30 28 17 17 1 169 WF Nebraska 31_0114 4 14 17 23 21 14 3 1 97 WF New Hampshire 33_1001 17 18 23 17 20 95 WF New Jersey 34_0501 1 2 1 2 1 1 4 1 1 14 WF New Jersey 34_0502 1 4 4 3 1 1 14 WF New Jersey 34_0503 1 2 3 2 3 4 1 1 1 18 WF New Jersey 34_0504 1 1 2 2 1 2 1 2 2 1 1 16 WF New Jersey 34_0505 2 1 3 1 3 1 2 1 1 15 WF New Jersey 34_0506 3 2 3 3 1 2 2 1 1 18 WF New Jersey 34_0507 4 4 3 3 1 1 1 1 1 19 WF New Jersey 34_0508 1 4 4 3 3 1 1 17 WF New Jersey 34_0509 1 2 3 2 4 3 1 1 17 WF New Jersey 34_0559 1 1 1 1 3 3 2 1 13 WF New Jersey 34_0560 1 1 1 3 4 2 2 1 15 WF New Jersey 34_0901 1 3 1 3 1 1 10 WF New Jersey 34_0902 1 2 1 3 1 1 1 10 146 Climate region State State code & section ID Hour of the day Total 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 WF New Jersey 34_0903 1 2 2 1 2 1 9 WF New Jersey 34_0960 1 2 2 2 1 1 9 WF New Jersey 34_0961 1 2 2 3 1 2 11 WF New Jersey 34_0962 1 1 2 1 5 2 12 WF New York 36_0801 1 2 11 27 31 36 28 11 1 148 WF Ohio 39_0901 1 1 1 1 1 1 1 1 6 17 25 30 22 12 7 4 1 1 1 1 1 1 1 138 WF Ontario 87_1622 7 18 12 10 11 7 2 67 WF Pennsylvania 42_1606 1 1 WF Quebec 89_3015 1 3 4 WF Vermont 50_1002 2 5 41 29 34 29 4 144 WNF Alabama 1_0101 2 8 8 12 13 11 8 15 6 6 2 91 WNF Alabama 1_0102 3 8 11 12 7 15 5 7 8 3 2 1 82 WNF Delaware 10_0102 13 8 11 8 12 5 1 58 WNF Georgia 13_1005 13 15 3 13 10 5 10 3 3 1 76 WNF Georgia 13_1031 1 8 23 12 11 17 11 11 6 7 107 WNF Maryland 24_1634 1 8 4 7 17 26 6 10 1 80 WNF Mississippi 28_1016 4 8 3 1 11 2 4 5 8 1 3 50 WNF Mississippi 28_1802 3 8 15 9 11 9 6 9 4 4 1 79 WNF North Carolina 37_1028 2 15 13 21 15 11 3 80 WNF Oklahoma 40_4165 1 13 16 11 11 7 9 4 72 WNF Texas 48_1060 9 8 14 11 14 11 8 10 7 3 95 WNF Texas 48_1068 11 8 19 10 16 10 4 8 8 2 96 WNF Texas 48_1077 8 11 13 17 7 16 4 11 8 8 3 106 WNF Texas 48_1122 10 9 19 19 16 15 17 12 14 2 3 136 WNF Texas 48_3739 8 8 12 19 15 14 13 8 11 1 109 WNF Virginia 51_0113 17 15 20 17 10 7 2 1 1 90 WNF Virginia 51_0114 1 14 26 24 26 20 8 1 1 121 WNF Washington 53_3813 1 1 147 Table 60 Monthly distribution of IRI visit data - SMP AC sections Climate region State State code & section ID Month Total 1 2 3 4 5 6 7 8 9 10 11 12 DF Colorado 8_1053 1 1 1 4 3 3 3 3 6 3 1 29 DF Idaho 16_1010 1 1 2 2 4 3 2 2 17 DF Montana 30_0114 3 2 1 5 1 2 9 6 4 3 2 2 40 DF Montana 30_8129 1 1 1 4 1 5 4 3 3 4 3 1 31 DF Nevada 32_0101 2 2 3 4 5 1 4 2 3 2 3 31 DF Saskatchewan 90_6405 2 1 3 2 4 4 2 1 3 1 23 DF South Dakota 46_0804 1 4 1 5 3 2 2 6 3 3 30 DF South Dakota 46_9187 1 1 1 2 2 1 1 1 1 11 DF Wyoming 56_1007 1 1 1 2 1 3 1 2 8 1 1 22 DNF Arizona 4_0113 4 2 6 4 2 2 1 2 3 6 32 DNF Arizona 4_0114 4 2 6 4 2 2 1 2 3 6 32 DNF Arizona 4_1017 2 3 2 5 7 19 DNF Arizona 4_1018 1 3 2 4 7 17 DNF Arizona 4_1024 3 4 3 2 2 1 1 1 1 1 3 22 DNF New Mexico 35_1112 3 4 2 1 3 2 1 6 1 23 DNF Utah 49_1001 1 1 1 3 3 3 3 1 3 3 1 23 WF Connecticut 9_1803 3 3 4 4 5 5 5 4 2 35 WF Maine 23_1026 2 3 3 2 1 2 5 2 3 1 24 WF Manitoba 83_1801 2 1 4 3 3 3 4 3 4 1 1 4 33 WF Massachusetts 25_1002 3 4 4 1 4 3 2 5 3 1 30 WF Minnesota 27_1018 2 2 1 3 4 2 1 2 1 18 WF Minnesota 27_1028 2 1 3 1 1 3 1 2 2 2 18 WF Minnesota 27_6251 2 1 4 5 3 3 4 3 4 2 2 3 36 WF Nebraska 31_0114 1 2 2 3 4 1 1 2 1 3 2 22 WF New Hampshire 33_1001 2 4 6 2 2 3 3 2 4 1 29 WF New Jersey 34_0501 2 1 1 1 4 1 1 4 2 17 WF New Jersey 34_0502 2 1 2 1 1 6 2 1 4 2 22 WF New Jersey 34_0503 2 1 2 1 1 5 2 1 4 2 21 WF New Jersey 34_0504 2 1 2 1 1 5 2 1 4 2 21 WF New Jersey 34_0505 2 1 2 1 1 5 2 1 4 2 21 WF New Jersey 34_0506 2 1 2 1 1 3 2 1 4 2 19 WF New Jersey 34_0507 2 1 2 1 1 5 2 1 4 2 21 WF New Jersey 34_0508 2 1 2 1 1 6 2 1 4 2 22 WF New Jersey 34_0509 2 1 2 1 1 5 2 1 4 2 21 WF New Jersey 34_0559 2 1 2 1 1 3 2 1 4 2 19 WF New Jersey 34_0560 2 1 2 1 1 3 2 1 4 2 19 WF New Jersey 34_0901 1 1 2 1 1 1 1 3 2 13 WF New Jersey 34_0902 1 1 2 1 1 1 1 3 2 13 WF New Jersey 34_0903 1 1 2 1 1 1 1 3 2 13 WF New Jersey 34_0960 1 1 2 1 1 1 1 3 2 13 WF New Jersey 34_0961 1 1 2 1 1 1 1 3 2 13 WF New Jersey 34_0962 1 1 2 1 1 1 1 3 2 13 WF New York 36_0801 4 2 3 3 5 3 2 2 3 3 1 31 WF Ohio 39_0901 1 6 1 4 6 2 5 2 3 2 5 37 WF Ontario 87_1622 1 3 1 3 4 1 5 4 4 5 2 33 WF Vermont 50_1002 4 5 2 3 5 5 5 2 2 4 3 40 WNF Alabama 1_0101 3 2 3 1 1 2 1 3 1 17 WNF Alabama 1_0102 3 2 3 1 1 3 1 3 1 18 WNF Delaware 10_0102 1 2 1 1 2 2 2 1 2 2 16 WNF Georgia 13_1005 3 2 3 2 2 3 1 1 17 WNF Georgia 13_1031 4 1 1 3 4 1 2 5 3 1 25 WNF Maryland 24_1634 1 3 3 2 1 7 1 1 2 1 4 26 WNF Mississippi 28_1016 3 1 1 1 3 4 1 14 WNF Mississippi 28_1802 3 2 2 1 1 1 5 4 1 20 WNF North Carolina 37_1028 3 1 3 3 1 2 1 1 2 2 6 25 WNF Oklahoma 40_4165 3 1 2 3 2 1 2 1 3 2 20 WNF Texas 48_1060 3 5 1 2 1 2 1 15 WNF Texas 48_1068 2 1 5 2 1 1 2 3 3 1 21 WNF Texas 48_1077 3 4 1 1 1 1 2 1 14 WNF Texas 48_1122 2 2 1 6 2 1 4 1 2 3 3 27 WNF Texas 48_3739 2 2 3 5 2 1 3 2 1 3 1 1 26 WNF Virginia 51_0113 2 1 1 1 2 3 2 1 2 15 WNF Virginia 51_0114 4 1 2 1 6 1 4 2 1 4 4 4 34 148 Table 61 Hourly distribution of IRI visit data - SMP AC sections Climate region State State code & section ID Hour of the day Total 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 DF Colorado 8_1053 1 1 5 5 1 2 6 3 1 2 2 29 DF Idaho 16_1010 1 1 1 1 1 1 2 4 1 2 2 17 DF Montana 30_0114 1 1 1 2 6 8 2 4 5 2 2 2 2 2 40 DF Montana 30_8129 1 1 1 4 1 5 5 1 3 1 6 1 1 31 DF Nevada 32_0101 1 1 4 1 3 4 6 2 2 6 1 31 DF Saskatchewan 90_6405 2 1 3 5 2 1 3 3 1 1 1 23 DF South Dakota 46_0804 1 4 4 4 1 2 3 3 4 2 1 1 30 DF South Dakota 46_9187 2 1 3 2 1 1 1 11 DF Wyoming 56_1007 2 2 3 2 2 2 2 4 1 2 22 DNF Arizona 4_0113 1 1 3 2 2 5 2 5 5 3 1 1 1 32 DNF Arizona 4_0114 1 3 2 3 6 3 3 5 4 1 1 32 DNF Arizona 4_1017 1 2 5 2 2 1 2 1 1 2 19 DNF Arizona 4_1018 1 1 1 3 2 2 1 1 3 2 17 DNF Arizona 4_1024 1 1 3 1 3 3 4 3 1 1 1 22 DNF New Mexico 35_1112 1 3 2 4 2 3 2 2 2 2 23 DNF Utah 49_1001 1 2 2 1 4 1 1 1 2 6 1 1 23 WF Connecticut 9_1803 5 3 4 1 4 4 2 8 3 1 35 WF Maine 23_1026 3 1 5 3 5 1 3 1 2 24 WF Manitoba 83_1801 1 1 1 3 2 6 3 4 5 3 3 1 33 WF Massachusetts 25_1002 2 7 7 6 1 2 1 1 1 1 1 30 WF Minnesota 27_1018 2 2 2 1 3 1 4 1 2 18 WF Minnesota 27_1028 1 1 1 2 2 2 3 1 1 2 2 18 WF Minnesota 27_6251 1 3 1 1 5 4 2 3 6 7 2 1 36 WF Nebraska 31_0114 1 3 5 1 3 1 1 3 2 1 1 22 WF New Hampshire 33_1001 3 4 5 2 4 2 4 2 1 1 1 29 WF New Jersey 34_0501 1 1 2 4 3 2 1 2 1 17 WF New Jersey 34_0502 1 2 3 2 5 3 2 2 1 1 22 WF New Jersey 34_0503 1 1 3 2 4 3 3 2 1 1 21 WF New Jersey 34_0504 1 1 3 3 4 3 2 2 1 1 21 WF New Jersey 34_0505 1 1 3 3 4 3 3 1 1 1 21 WF New Jersey 34_0506 1 1 1 2 4 4 2 2 1 1 19 WF New Jersey 34_0507 1 2 1 2 6 2 3 2 1 1 21 WF New Jersey 34_0508 1 2 3 3 4 3 2 2 1 1 22 WF New Jersey 34_0509 1 1 2 4 4 3 3 1 1 1 21 WF New Jersey 34_0559 1 1 2 2 4 2 3 2 1 1 19 WF New Jersey 34_0560 1 1 2 1 4 2 5 1 1 1 19 WF New Jersey 34_0901 1 4 2 1 2 2 1 13 WF New Jersey 34_0902 1 1 3 2 1 1 3 1 13 WF New Jersey 34_0903 1 3 3 1 1 3 1 13 WF New Jersey 34_0960 1 1 3 2 1 1 3 1 13 WF New Jersey 34_0961 1 1 3 2 1 1 3 1 13 149 Climate region State State code & section ID Hour of the day Total 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 WF New Jersey 34_0962 1 1 3 2 1 1 3 1 13 WF New York 36_0801 1 2 4 2 2 1 2 9 5 2 1 31 WF Ohio 39_0901 1 2 2 5 2 5 2 2 3 5 2 1 3 1 1 37 WF Ontario 87_1622 3 1 2 2 2 4 4 2 4 4 4 1 33 WF Vermont 50_1002 1 2 2 5 6 2 5 8 3 2 1 3 40 WNF Alabama 1_0101 2 1 2 1 2 4 2 3 17 WNF Alabama 1_0102 1 1 4 2 3 1 1 2 2 1 18 WNF Delaware 10_0102 1 2 1 1 2 2 3 1 2 1 16 WNF Georgia 13_1005 1 1 4 1 1 2 1 3 1 2 17 WNF Georgia 13_1031 1 2 3 2 2 2 1 4 6 2 25 WNF Maryland 24_1634 1 1 2 3 2 3 4 4 3 1 1 1 26 WNF Mississippi 28_1016 1 4 1 1 2 1 1 2 1 14 WNF Mississippi 28_1802 1 2 3 3 2 1 1 3 1 1 1 1 20 WNF North Carolina 37_1028 2 2 2 3 2 2 3 4 4 1 25 WNF Oklahoma 40_4165 2 4 1 3 4 2 1 1 1 1 20 WNF Texas 48_1060 1 4 2 2 1 2 1 1 1 15 WNF Texas 48_1068 1 2 1 3 3 2 2 4 2 1 21 WNF Texas 48_1077 1 3 1 1 1 2 1 1 1 1 1 14 WNF Texas 48_1122 1 1 2 2 3 6 5 3 2 1 1 27 WNF Texas 48_3739 1 1 5 4 3 2 3 1 1 2 2 1 26 WNF Virginia 51_0113 1 2 4 1 2 4 1 15 WNF Virginia 51_0114 1 3 1 7 3 5 6 1 1 4 1 1 34 150 Table 62 Monthly distribution of IRI visit data - SMP PCC sections Climate region State State code & section ID Month Total 1 2 3 4 5 6 7 8 9 10 11 12 DF Nevada 32_0204 2 1 4 1 3 2 1 1 3 18 DF Utah 49_3011 2 1 2 3 5 2 3 5 5 4 4 36 DNF Arizona 4_0215 7 5 5 4 3 6 3 5 3 5 17 63 DNF California 6_3042 4 1 8 3 6 1 2 2 2 4 2 35 WF Indiana 18_3002 1 2 1 5 3 3 4 2 6 2 2 31 WF Kansas 20_4054 2 3 5 7 4 2 2 3 4 32 WF Manitoba 83_3802 3 1 1 6 3 5 5 2 3 1 1 31 WF Minnesota 27_4040 3 1 7 4 4 1 5 3 1 29 WF Nebraska 31_3018 3 2 9 2 6 4 4 7 5 8 50 WF New York 36_4018 2 3 1 10 1 3 4 4 1 29 WF Ohio 39_0204 1 5 1 3 4 5 1 3 3 26 WF Pennsylvania 42_1606 6 4 4 3 9 2 6 3 5 14 1 1 58 WF Quebec 89_3015 2 6 3 3 7 6 5 5 7 5 3 52 WF South Dakota 46_3010 4 1 3 5 1 4 1 19 WNF Georgia 13_3019 5 3 3 5 5 1 8 6 2 2 40 WNF North Carolina 37_0201 10 5 3 1 6 2 6 2 2 5 4 2 48 WNF North Carolina 37_0205 2 2 2 4 1 2 1 3 3 20 WNF North Carolina 37_0208 2 2 3 2 3 6 2 1 2 6 1 30 WNF North Carolina 37_0212 2 2 4 2 5 6 2 1 2 6 1 33 WNF Texas 48_4142 4 2 1 5 1 1 4 4 3 25 WNF Texas 48_4143 3 1 6 1 1 4 4 1 3 24 WNF Washington 53_3813 4 2 2 7 3 2 2 3 25 151 Table 63 Hourly distribution of IRI visit data - SMP PCC sections Climate region State State code & section ID Hour of the day Total 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 DF Nevada 32_0204 1 1 1 3 1 3 6 2 18 DF Utah 49_3011 1 3 3 3 1 2 6 1 7 3 2 2 1 1 36 DNF Arizona 4_0215 1 1 3 2 1 3 10 5 4 7 4 9 6 1 1 1 2 2 63 DNF California 6_3042 1 1 1 6 2 3 2 4 4 2 3 2 2 2 35 WF Indiana 18_3002 1 3 4 1 4 3 2 4 3 3 2 1 31 WF Kansas 20_4054 1 4 2 4 1 4 2 3 2 2 5 1 1 32 WF Manitoba 83_3802 1 1 4 2 1 4 3 4 1 3 3 1 2 1 31 WF Minnesota 27_4040 1 2 1 2 3 1 4 5 1 1 8 29 WF Nebraska 31_3018 1 1 7 7 3 3 2 1 4 6 4 10 1 50 WF New York 36_4018 1 1 5 3 6 1 1 1 5 1 2 1 1 29 WF Ohio 39_0204 1 5 1 2 2 2 1 3 1 3 5 26 WF Pennsylvania 42_1606 1 8 6 3 7 3 5 6 7 5 4 1 2 58 WF Quebec 89_3015 2 2 8 2 2 7 3 9 9 4 3 1 52 WF South Dakota 46_3010 1 4 1 4 1 1 3 2 2 19 WNF Georgia 13_3019 1 3 3 6 5 3 5 8 1 1 3 1 40 WNF North Carolina 37_0201 1 1 6 7 5 4 2 4 1 7 4 2 1 1 2 48 WNF North Carolina 37_0205 1 1 1 1 2 2 2 3 1 1 1 3 1 20 WNF North Carolina 37_0208 1 2 1 3 1 1 2 2 5 2 3 3 2 1 1 30 WNF North Carolina 37_0212 1 2 2 3 2 3 2 2 5 1 3 4 1 1 1 33 WNF Texas 48_4142 3 4 2 1 5 4 2 1 2 1 25 WNF Texas 48_4143 6 1 2 2 4 2 3 2 1 1 24 WNF Washington 53_3813 1 6 2 1 3 1 3 4 1 2 1 25 152 REFERENCES 153 REFERENCES 1. Novak Jr, E.C. and L.E. DeFrain Jr, Seasonal Changes in the Longitudinal Profile of Pavements Subject to Frost Action. Transportation Research Record 1362, 1992: p. 95 - 100. 2. Karamihas, S., et al., Guidelines for Longitudinal Pavement Profile Measurement , in TRB, National Research Council, Washington, DC . 1999. 3. Karamihas, S.M. and K. Senn, Curl and Warp A nalysis of the LTPP SPS - 2 Site in Arizona . 2012. 4. Wiser, L., Curl and Warp Analysis of the LTPP SPS - 2 Site in Arizona: TechBrief . 2013, United States. Federal Highway Administration. 5. Hveem, F.N., Pavement deflections and fatigue failures . 1955. 6. Cha tti, K., et al., Enhanced Analysis of Falling Weight Deflectometer Data for Use With Mechanistic - Empirical Flexible Pavement Design and Analysis and Recommendations for Improvements to Falling Weight Deflectometers . 2017. 7. AASHTO, A., Mechanistic - empiric al pavement design guide: A manual of practice. AAoSHaT Officials, Editor, 2008. 8. Lukanen, E.O., R. Stubstad, and R. Briggs, Temperature predictions and adjustment factors for asphalt pavement . 2000. 9. Chen, D. - H., et al., Temperature correction on fall ing weight deflectometer measurements. Transportation Research Record: Journal of the Transportation Research Board, 2000(1716): p. 30 - 39. 10. Kim, Y.R., B.O. Hibbs, and Y. - C. Lee, Temperature correction of deflections and backcalculated asphalt concrete m oduli. Transportation Research Record, 1995(1473). 11. Mun Park, H., Y. Richard Kim, and S. Park, Temperature correction of multiload - level falling weight deflectometer deflections. Transportation Research Record: Journal of the Transportation Research Boa rd, 2002(1806): p. 3 - 8. 12. Schmalzer, P.N., Temperature Correction of Asphalt Moduli. Presented at the 27th FWD User Group Meeting. . 2018. 13. Zheng, Y., P. Zhang, and H. Liu, Correlation between pavement temperature and deflection basin form factors of asphalt pavement. International Journal of Pavement Engineering, 2019. 20 (8): p. 874 - 883. 154 14. Fernando, E.G., W. Liu, and D. Ryu, Development of a procedure for temperature correction of backcalculated AC modulus . 2001, Texas Transportation Institute, Texa s A & M University System. 15. Kim, Y.R., Modeling of asphalt concrete . 2008. 16. Lee, H.S., Viscowave a new solution for viscoelastic wave propagation of layered structures subjected to an impact load. International Journal of Pavement Engineering, 2014. 15 (6): p. 542 - 557. 17. Lee, H.S., Development of a New Solution for Viscoelastic Wave Propagation of Pavement Structures and Its Use in Dynamic Backcalculation . 2013: Michigan State University. Civil Engineering. 18. Long, B., M. Hossain, and A.J. Gisi, Se asonal Variation of Backcalculated Subgrade Moduli. Transportation Research Record, 1997. 1577 (1): p. 70 - 80. 19. Monismith, C.L., Analytically based asphalt pavement design and rehabilitation: Theory to practice, 1962 - 1992 . 1992. 20. Zegeye Teshale, E., D. Shongtao, and L.F. Walubita, Evaluation of Unbound Aggregate Base Layers using Moisture Monitoring Data. Transportation Research Record, 2019. 2673 (3): p. 399 - 409. 21. Puppala, A., Estimating stiffness of subgrade and unbound materials for pavement design . NCHRP Synthesis 382. Transportation Research Board, Washington, DC, 2008. 22. Zhang, Y., et al., Seasonal variations and in situ assessment of concrete pavement foundation mechanistic properties. International Journal of Pavement Research and Technology, 2018. 11 (4): p. 363 - 373. 23. Highway, A.A.o.S. and T. Officials, AASHTO guide for design of pavement structures . 1993. 24. Packard, R.G., Thickness design for concrete highway and street pavements. 1984. 25. Siddique, Z.Q., M. Hossain, and D. Meggers. Temperature and curling measurements on concrete pavement . in Proceedings of the 2005 Mid - Continent Transportation Research Symposium . 2005. Citeseer. 26. Choubane, B. and M. Tia, Analysis and verification of thermal - gradient effects on concrete pavement. Journal of Transportation Engineering, 1995. 121 (1): p. 75 - 81. 27. Jeong, J. - H. and D.G. Zollinger, Environmental Effects on the Behavior of Jointed Plain Concrete Pavements. Journal of Transpo rtation Engineering, 2005. 131 (2): p. 140 - 148. 28. Westergaard, H . M . Computation of stresses in concrete roads . 155 29. Choubane, B. and M. Tia, Nonlinear temperature gradient effect on maximum warping stresses in rigid pavements. Transportation Research Recor d, 1992. 1370 (1): p. 11. 30. TRUJILLO, P.B. and M.A.S. GUERRERO, Effect of temperature gradients on the behaviour of jointed plain concrete pavements. Revista IBRACON de Estruturas e Materiais, 2019. 12 : p. 398 - 407. 31. Belshe, M., et al., Temperature grad ient and curling stresses in concrete pavement with and without open - graded friction course. Journal of Transportation Engineering, 2011. 137 (10): p. 723 - 729. 32. Harr, M.E., Warping stresses and deflections in concrete slabs. 1958. 33. Mohamed, A.R. and W . Hansen, Effect of nonlinear temperature gradient on curling stress in concrete pavements. Transportation Research Record, 1997. 1568 (1): p. 65 - 71. 34. Ceylan, H.K., Sunghwan; Turner, Dennis J.; Rasmussen, Robert Otto; Chang, George K.; Grove, James; and Gopalakrishnan, Kasthurirangan, Impact of Curling, Warping, and Other Early - Age Behavior on Concrete Pavement Smoothness: Early, Frequent, and Detailed (EFD) Study , in Trans Project Reports. 49. 2007. 35. Asbahan, R.E. and J.M. Vandenbossche, Effects of Te mperature and Moisture Gradients on Slab Deformation for Jointed Plain Concrete Pavements. Journal of Transportation Engineering, 2011. 137 (8): p. 563 - 570. 36. Nassiri, S. and J. Vandenbossche, Establishing Built - in Temperature Gradient for Jointed Plain C oncrete Pavements in Pennsylvania. International Journal of Pavement Research and Technology, 2011. 55 : p. 245 - 256. 37. Pierce, L.M., et al., Using Falling Weight Deflectometer Data with Mechanistic - Empirical Design and Analysis, Volume III: Guidelines for Deflection Testing, Analysis, and Interpretation . 2017. 38. Sayers, M.W., On the calculation of international roughness index from longitudinal road profile. Transportation Research Record, 1995(1501). 39. Sayers, M.W., Guidelines for conducting and calib rating road roughness measurements . 1986, University of Michigan, Ann Arbor, Transportation Research Institute. 40. Chang, G.K., et al. Quantifying the impact of jointed concrete pavement curling and warping on pavement unevenness . in 6th symposium on pave ment surface characteristics (SURF), Potoroz, Slovenia . 2008. 41. Chang, G., et al., Impact of temperature curling and moisture warping on jointed concrete pavement performance . 2010. 42. Lee, H.S., et al., Effect PCC Slab Curling and Warping on Pavement R oughness. 156 43. Rada, G., et al., Seasonal Monitoring Program: Instrumentation Installation and Data Collection Guidelines . 1994, Report No. FHWA - RD - 94 - 110. FHWA, US Department of Transportation. 44. Elkins, E.E., et al., Long - Term Pavement Performance Infor mation Management System User Guide . 2018, Office of Infrastructure Research and Development, Federal Highway Administration,.