TRANSFERRING ALABAMA?S SMOOTHNESS SPECIFICATIONS FROM PI-BASED TO IRI-BASED Except where reference is made to the work of others, the work described in this thesis is my own or was done in collaboration with my advisory committee. This thesis does not include proprietary or classified information. _____________________________________ Wen Huang Certificate of Approval: _______________________ __________________________ David H. Timm Mary Stroup-Gardiner, Chair Assistant Professor Associate Professor Civil Engineering Civil Engineering _______________________ __________________________ Rod E. Turochy Stephen L. McFarland Assistant Professor Acting Dean Civil Engineering Graduate School TRANSFERRING ALABAMA?S SMOOTHNESS SPECIFICATIONS FROM PI-BASED TO IRI-BASED Wen Huang A Thesis Submitted to the Graduate Faculty of Auburn University in Partial Fulfillment of the Requirements for the Degree of Master of Science Auburn, Alabama August 7, 2006 iii TRANSFERRING ALABAMA?S SMOOTHNESS SPECIFICATIONS FROM PI-BASED TO IRI-BASED Wen Huang Permission is granted to Auburn University to make copies of this thesis at its discretion, upon request of individuals or institutions at their expense. The author reserves all publication rights. _______________________ Signature of Author _______________________ Date of Graduation iv THESIS ABSTRACT TRANSFERRING ALABAMA?S SMOOTHNESS SPECIFICATIONS FROM PI-BASED TO IRI-BASED Wen Huang Master of Science, August 7, 2006 (M.S., Tongji University, 2004) (B.S., Chang?an University, 2001) 95 Pages Typed Directed by Mary Stroup-Gardiner Currently, Profilograph Index (PI) is deployed as the pavement smoothness evaluation index in the Alabama DOT?s smoothness specifications. The specifications set the incentive, full and disincentive payment levels to encourage the construction of smoother pavement. The problems of this index are the poor correlation between PI and the driving comfort, and its walking-speed operation, which makes it infeasible to keep track of pavement smoothness condition over time and traffic. With the development of inertial profilers, smoothness specifications based on International Roughness Index (IRI), which can accurately evaluate the driving quality right after construction up to rehabilitation needs, are expected to address these problems. An analysis was conducted on the profile database pooled from a range of Alabama asphalt concrete pavements and Quebec Portland cement concrete pavements. Correlations between PI and IRI were developed by several statistic methods. According v to these relationships, the PI-based smoothness specifications were transferred to IRI-based smoothness specifications. vi ACKNOWLEDGEMENTS The author would like to thank Dr. Mary Stroup-Gardiner for her guidance and strong support in all my study and the analysis and writing portion of this thesis. The author would also like to thank Dr. David Timm and Dr. Rod E. Turochy for their great course teaching and correction of this thesis. vii Style manual used: MLA Handbook for Writers of Research Papers (5th Edition) Computer Software Used: Microsoft Word, Microsoft Excel, and ProVAL 2.5 viii TABLE OF CONTENTS LIST OF TABLES ....................................................................................................... x LIST OF FIGURES .................................................................................................... xi CHAPTER ONE INTRODUCTION ....................................................................................................... 1 1.1 Background..................................................................................................... 1 1.2 Objective......................................................................................................... 4 1.3 Scope............................................................................................................... 4 CHAPTER TWO LITERATURE REVIEW............................................................................................. 5 2.1 Roughness Index............................................................................................. 5 2.1.1 Profilograph Index ................................................................................ 6 2.1.2 International Roughness Index ............................................................11 2.1.3 Comparison of PI with IRI.................................................................. 13 2.1.4 Correlation between PI and IRI .......................................................... 15 2.2 Smoothness Specifications Conversion Methods ......................................... 20 2.3 Effect of Short Interval on Estimating Pavement Smoothness..................... 23 2.4 Smoothness Specification ............................................................................. 24 2.4.1 ALDOT Smoothness Specifications ................................................... 24 2.4.2 Smoothness Specifications of Other DOTs......................................... 26 2.5 Summary....................................................................................................... 29 CHAPTER THREE DATA COLLECTION AND DATABASE DEVELOPMENT.................................. 31 3.1 Data Collection ............................................................................................. 31 3.1.1 Asphalt Pavement Profiles .................................................................. 32 3.1.2 Concrete Pavement Profiles................................................................ 33 ix 3.2 ProVAL ......................................................................................................... 34 CHAPTER FOUR DATA ANALYSIS ..................................................................................................... 38 4.1 Data Quality.................................................................................................. 38 4.1.1 Smoothness Data of Asphalt Concrete (AC) Pavement...................... 38 4.1.2 Smoothness Data of Portland Cement Concrete Pavement ................ 44 4.2 Effect of Different Index on Evaluating Pavement Smoothness .................. 48 4.3 Conversion of PI Specifications to IRI Specifications ................................. 52 4.3.1 Specification Conversion Using Regression Equations...................... 52 4.3.2 Specification Conversion Using Distribution Method........................ 54 4.3.3 Effect of Material Transfer Devices (MTD) on Asphalt Pavement Smoothness .................................................................................................. 58 4.4 IRI-based Specification................................................................................. 60 4.5 Comparison of Converted IRI Specification with Other DOT?s Specifications ............................................................................................................................. 62 CHAPTER FIVE CONCLUSIONS........................................................................................................ 64 5.1 Conclusions................................................................................................... 65 5.2 Limitations .................................................................................................... 66 REFERENCES .......................................................................................................... 68 APPENDICES ........................................................................................................... 72 Appendix A Regression Relationship between IRI and PIx at 0.1 and 0.01 Mile Interval ................................................................................................................ 73 Appendix B Histogram Distribution of PI and IRI values of AC and PCC Pavement............................................................................................................. 76 x LIST OF TABLES Table 2. 1 Summary of Documented PI-IRI Relationships. ...................................... 19 Table 2. 2 Alabama Pavement Smoothness Specifications for PI 0.2 (ALDOT, 2002). ............................................................................................................................ 25 Table 2. 3 Alabama Pavement Smoothness Specifications for PI 0.0 (ALDOT, 2003). ............................................................................................................................ 26 Table 3. 1 Project Descriptions (Alabama Mill and Fill Projects)............................. 33 Table 3. 2 Descriptions of Concrete Pavements. ....................................................... 34 Table 4. 1 Correlation Equations between IRI and PI in this Study and LTPP (Asphalt Overly Pavement)...................................................................................................... 53 Table 4. 2 Converted IRI Specifications for Asphalt Pavement at 0.1 Mile Interval by Regression Equations. ............................................................................................... 54 Table 4. 3 IRI Specifications for PCC Pavement....................................................... 54 Table 4. 4 Converted IRI Specifications for AC Using Distribution Method............ 58 Table 4. 5 Combination of Converted IRI Specifications.......................................... 61 Table 4. 6 IRI Specification at 0.1 mile interval........................................................ 62 Table 5. 1 Transferred IRI based Smoothness Specifications for Asphalt and Concrete Pavement in Alabama................................................................................................ 66 xi LIST OF FIGURES Figure 2. 1 McCracken California Profilograph.......................................................... 7 Figure 2. 2 A Typical California Profilograph with 12 Support Wheels (FAA, 2005). 7 Figure 2. 3 Profilograph Trace (FAA, 2005)................................................................ 9 Figure 2. 4 ProScan (Smith et al., 1997).................................................................... 10 Figure 2. 5 Sensitivity of PI and IRI to Wavelength (Evans et al. 2003)....................11 Figure 2. 6 Quarter Car Model. (Gillespie, T.D., 1992)............................................. 12 Figure 2. 7 Human Body Sensitivity of the Vertical Vibration (Sayers and Karamihas, 1998). ................................................................................................................. 14 Figure 2. 8 Relationship between Simulated PI 0.2 and IRI in ILDOT Bridge Smoothness Study (Rufino et al., 2001). ........................................................... 21 Figure 2. 9 Conversion from Old Smoothness Specification to New One by Distribution Method (Hossain et al., 1995). ...................................................... 22 Figure 2. 10 Comparison of IRI Value at 0.1 mile interval with 0.01 mile interval. . 24 Figure 2. 11 2002 ALDOT Specification for Pavement Roughness.......................... 25 Figure 2. 12 PI 0.0 Specifications for AC Pavement from other DOTs (Pellinen et al., 2003). ................................................................................................................. 27 Figure 2. 13 PI 0.0 Specifications for PCC Pavement from other DOTs (Pellinen et al., 2003). ................................................................................................................. 28 Figure 2. 14 IRI Specifications from other DOTs...................................................... 29 Figure 3. 1 Model 4300 of ARAN Van (Roadware, 2005). ....................................... 32 Figure 3. 2 Main Function of ProVAL 2.5................................................................. 35 Figure 3. 3 Profilograph Simulation Function Tab. ................................................... 36 Figure 3. 4 Ride Statistics Function Tab. ................................................................... 37 Figure 4. 1 Histogram of AC IRI Value Distribution at 0.01 Mile Interval............... 39 Figure 4. 2 Histogram of AC IRI Value Distribution at 0.1 Mile Interval................. 39 Figure 4. 3 Histogram of AC PI 0.2 Value Distribution at 0.01 Mile Interval. ............ 41 xii Figure 4. 4 Histogram of AC PI 0.2 Value at 0.01 Mile Interval after Taking out PI 0.2 Values of 0 in/mile. ............................................................................................ 41 Figure 4. 5 Histogram of AC PI 0.2 Value Distribution at 0.1 Mile Interval. .............. 42 Figure 4. 6 Histogram of AC PI 0.0 Value Distribution at 0.01 mile interval. ............. 43 Figure 4. 7 Histogram of AC PI 0.0 Value Distribution at 0.1 Mile Interval. .............. 43 Figure 4. 8 Histogram of IRI Value Distribution of PCC at 0.01 Mile Interval. ....... 45 Figure 4. 9 Histogram of IRI Value Distribution of PCC at 0.1 Mile Interval. ......... 45 Figure 4. 10 Histogram of PI 0.2 Value Distribution of PCC at 0.01 Mile Interval..... 46 Figure 4. 11 Histogram of PI 0.2 Value Distribution of PCC at 0.1 Mile Interval....... 46 Figure 4. 12 Histogram of PI 0.0 Value Distribution of PCC at 0.01 Mile Interval..... 47 Figure 4. 13 Histogram of PI 0.0 Value Distribution of PCC at 0.1 Mile Interval....... 48 Figure 4. 14 Comparison of PI 0.2 of AC and PCC Pavement at 0.1 Mile Interval..... 49 Figure 4. 15 Comparison of PI 0.0 of AC and PCC Pavement at 0.1 Mile Interval..... 50 Figure 4. 16 Comparison of IRI of AC and PCC Pavement at 0.1 Mile Interval. ..... 51 Figure 4. 17 Pavement Percentages of AC Pavements in Each Pay Level according to PI 0.0 Specifications............................................................................................. 56 Figure 4. 18 Pavement Percentages of AC Pavements in Each Pay Level according to Adjusted PI 0.0 Specifications.............................................................................. 57 Figure 4. 19 Limits of Each Pay Range for IRI ......................................................... 58 Figure 4. 20 Effect of MTD on Pavement Smoothness at 0.1 mile interval.............. 60 Figure 4. 21 Comparison of Transferred AL IRI specifications with Specifications from other DOT?s............................................................................................... 63 Figure A. 1 PI 0.2 vs. IRI for AC at 0.1 mile interval. ................................................. 74 Figure A. 2 PI 0.0 vs. IRI for AC at 0.1 mile interval. ................................................. 74 Figure A. 3 PI 0.2 vs. IRI for AC at 0.01 mile interval. ............................................... 75 Figure A. 4 PI 0.0 vs. IRI for AC at 0.01 mile interval. ............................................... 75 1 CHAPTER ONE INTRODUCTION Pavement smoothness, defined as the lack of roughness, is considered as one of the most important indicators of overall construction quality and subsequent riding comfort (Smith et al. 1997). Initially smooth pavement, which is the result of a good construction quality, provides a longer service life than initially rough pavement (Smith et al. 1997). For the driving public, smoothness is the primary means of assessing pavement quality. A rough-riding pavement increases fuel costs, vehicle maintenance and repair costs, slows traffic flow which can increase congestion, and in extreme cases, creates safety issues. Due to the importance of pavement smoothness, smoothness specifications are applied to encourage the construction of good ride quality of the final surface. Good-riding smooth pavements can earn the incentives, while contractors building rough-riding pavement product are only paid a reduced portion of the contract price (i.e., disincentives). 1.1 Background The nationwide application of smoothness specifications has led to the development of a variety of devices to measure pavement profiles, which generate various ride quality statistics as the outputs. The most commonly employed device is the California-type profilograph, used to calculate the profile index (PI) as the index to assess pavement 2 smoothness. The PI represents the total accumulated deviations of the longitudinal profilograph beyond a tolerance zone, which is also referred as a blanking band. Until recently, the Alabama Department of Transportation (ALDOT) deployed the McCracken California-style profilograph as the standard measuring device, and Profile Index with 0.2 inch blanking band as the smoothness index. Contractors received a 5% bonus by providing pavements with a PI of less than 2 inch/mile (ALDOT, 2002). However, an analysis study conducted by ALDOT in 1999 indicated that 0.2 inches blanking band specification raised some concerns (Bowman et al., 2003). The most important one was that the wide blanking band (0.2?) ignores defects (localized bumps) in the surface that are felt by the driving public but not necessarily identified as a penalty to the contractor. In this analysis, more than three-quarters of all 0.1 mile segments were found falling within the bonus range for the contractor without improving the public?s ride comfort. Therefore, Profile Index calculated with a 0.2 blanking band has a limited ability to reflect riding quality of the newly constructed pavement, which results in the failure of the PI to motivate good construction. After 2003, ALDOT decreased the 0.2? blanking band to 0.0? blanking band, which helps to count irregularities hidden by the blanking band. However, PI still represents the physically accumulated pavement deviations, which do not directly connect to the ride quality of the pavement. And besides, since California-type profilograph is hand-propelled and operated in walking speed, it is extremely time-consuming and infeasible for PI to keep track of the pavement smoothness condition during the whole service life because of the required traffic control. 3 With the development of inertial profilers, especially light-weight inertial profilers, the longitudinal profiles of pavement can be collected at highway speeds, even right after paving is finished. These technologies make International Roughness Index (IRI) a universally accepted ride quality statistic. IRI accumulates the response of vehicle to the roughness of the road surface. It can precisely evaluate the riding comfort by simulating the way a reference vehicle would response to the pavement roughness and accumulating the vehicle suspension travel. And also, the inertial profilers are operated at highway speeds, which provides an efficiently fashion to investigate the smoothness of the new pavement and monitor the subsequent pavement condition over traffic and time. All these evident advantages encourage the development of IRI as a portable and repeatable smoothness scale to evaluate both short and long-term pavement ride quality. Although PI is used in the present ALDOT pavement smoothness specification, an urge to employ IRI in specification is claimed by ALDOT because of the advantages of IRI. In order to transfer the current PI based specifications to the corresponding specifications with IRI, the relationships between the PI and IRI indices are needed to connect different smoothness indices. Currently, most agencies including ALDOT measure the pavement smoothness over a 0.1 mile segment during the quality assessment. But as observed in the quality assessment and construction, localized irregularities at the construction joint or caused by discontinuous paving practices can be averaged in the whole 0.1 mile interval without being noticed. In order to mark these bumps and accurately evaluate pavement smoothness, a smaller interval, such as a 0.01 mile segment, has the potential for identifying and quantifying these localized irregularities. 4 1.2 Objective The main objective of this study was to move the current PI-based smoothness specifications to the corresponding IRI specifications. To address this transfer from PI to IRI in the specifications, the correlation between these two indices needed to be established. Based on these connections, the IRI limits, corresponding to PI limits for bonus, full pay, and penalty pay range, can be calculated and determined. 1.3 Scope The 57 sets of longitudinal profiles from a range of Alabama asphalt concrete pavements and Quebec concrete pavements were collected for this study. All asphalt pavement sections are HMA overlay sections located in the same climatic zone (a wet, no-freeze region), while concrete pavement sections come from wet and freeze climate zone. Due to the different climate zone and other different conditions, PCC data from Quebec has the limitation to be applied in Alabama specification, PCC data was only used to primarily compare with AC data, and to present the way for different smoothness indexes to evaluate the pavement roughness. Both IRI and mathematical-simulated PI value were calculated for each profile using the ProVAL Version 2.5 software, for 0.1 mile and 0.01 mile interval. The transferred specifications were only based on 0.1 mile interval. Since 0.01 mile interval is just used in localizing the bumps (WFLHD, 2003), the 0.01 mile interval specification for bump detection needs future development. 5 CHAPTER TWO LITERATURE REVIEW According to the definition of roughness (i.e., lack of smoothness) from ASTM E 867 (1998), traveled surface roughness is the deviations of a pavement surface from a true planar surface with characteristic dimensions that affect vehicle dynamics, ride quality, dynamic loads, and drainage, for example, longitudinal profiles, transverse profile and cross slope. Therefore, pavement roughness can be described by the magnitudes of the profile irregularities and their distribution on the measured surface. 2.1 Roughness Index The primary objective for any ride quality index is to indicate information about a pavement surface that is sufficient to estimate the satisfaction of riding comfort. Mathematically, a pavement profile can be described as a combination of varied sine waves, which includes the long wavelengths like slope of pavement and the short wavelengths like the teeth-jarring waves (Sayers and Karamihas, 1998). Not all waves contribute to the driver?s perception of pavement roughness. Good design of the vehicle suspension system and tire system are used to filter out the effect of some pavement wavelengths. The wavelengths that can not be filtered out with vehicle design and cause the unwanted vehicle vibration are felt as the pavement roughness. Consequently, the roughness index is required to attenuate the unnecessary road features and highlight the 6 driving-discomfort wavelengths. As a matter of fact, different indices use different mechanical filters or mathematical algorithms to collect pavement roughness information. Profile index is the typical representative for mechanical filter based indices; International Roughness Index is for profile based indices. Due to the different filter methods, some wavelength bands may be noticed by one roughness index and ignored by another index. 2.1.1 Profilograph Index Profilograph Index (PI), also called as profile index, is derived from low-speed rolling system, which uses its own geometry to filter the profile. PI is derived from rolling straightedge systems such as California profilograph, which is a 25 ft long truss with a set of wheels at either end that travels over the pavement surface, presented in Figure 2.1 and Figure 2.2 (FAA, 2005). The wheel in the center of the truss is attached to a recording device (e.g. chart recorder), which documents the deviations. This rolling system functions as the mechanical filter. The wheels of truss except the middle wheel establish the average surface, and then the middle wheel records the deviation from this surface. According to this filter method, the long surface wavelength is removed by the average; the high-frequency wavelength is emphasized. 7 Side View Recorder Recording wheel 25' c c c cc c c Top View 2.5' 2.5' 2.5' 8.75' 2.5'2.5' 2.5'8.75' 2.5' Figure 2. 1 McCracken California Profilograph Figure 2. 2 A Typical California Profilograph with 12 Support Wheels (FAA, 2005). The calculation procedure to produce the profilograph recording from the rolling systems can be expressed as in Equation 2.1(FAA, 2005). This equation is also the algorithm for profile software to simulate PI value from pavement profiles collected by the inertial profiler. )())(()( 1 rrii N i i dxPdxPCxR ????= ? = (Equation 2.1) Where, 8 R(x) = the computed profilograph recording at the position x, mm N = the total number of the wheels in the left and right side of the support system P i = the profile on which the ith wheel is traveling, mm Ci = the influence coefficient corresponding to the ith wheel. It is equal to the vertical displacement at the recorder position caused by a unit vertical movement at the ith wheel. From the structure geometry and the definition of the influence coefficients, Ci = 1/16 for the 8 right side wheels and Ci = 1/8 for the 4 left side wheels is used here. D i = the offset distance from the location x for each wheel, mile Items with subscript r refer to those of the recording wheel. After recording the profilograph in the field, the operator needs to return to the office to have the chart paper profiles processed. The analysis starts with the location of a floating blanking band, which is determined by tracing these curve outlines. Figure 2.3 presents one sample of this process. The blanking band is located for allowing as many of irregularities as possible to be covered and blanked out. Since defects within blanking band are considered having no effect on riding quality, these defects are excluded from calculating PI values. In Figure 2.3, the two dash lines indicate the location of the blanking band. Each deviation exceeding the blanking band is called a scallop, with the number of scallops being accumulated to compute PI. PI value has the unit of slope, in/mile or m/km. A length of 0.1 miles (528 ft) is used as the interval over which the number of scallops is considered. 9 Figure 2. 3 Profilograph Trace (FAA, 2005). 0.0, 0.1 and 0.2 inches, are the commonly used blanking band to calculate PI value. From Federal Highway Administration survey results (Smith et al. 1997), 19 states used 0.2 inches blanking band for AC pavement quality assurance; one state used 0.1 inches blanking band and two states used 0.0 inches blanking band. 2000 America Concrete Pavement Association database shows that the different usages of blanking band for PCC pavement are distributed: 0.2? blanking band 77.2%, 0.1? blanking band 13.9%, 0.0? blanking band 11.1%. These different blanking bands can generate the different effect on evaluating the pavement ride quality. The vertical deviations smaller than 0.2 inches are not counted when 0.2? blanking band is applied to compute the PI value. This has raised some concerns because in some cases newly constructed pavements received the riding quality complaints even though they met the smoothness criterion (Bowman et al. 2003). 0.0 inches blanking band can count every irregularities to better assess the pavement roughness. There is a trend to move toward 0.0? blanking band to compute PI value. Two methods are widely used to conduct PI calculation, manual method or automated method (ProScan TM , shown in Figure 2.4). Manual tracing includes the personal judgment about the location of blanking band, which leads to variations in the final result. Alabama uses automatic tracing program, ProScan TM system to process the trace (ALDOT, 2003). In general, the profiler curve is scanned to digitize its tracing. An image enhancement program is then used to prepare the image for analysis. After the 10 enhancement, mathematical filtering is applied to the digitized traces to reduce the noise of the traces and to mimic the process of an operator drawing the outline on the trace. A linear regression analysis is then performed to establish the location of a floating centerline and blanking band, along the outline of the trace (Pellinen et al. 2003). Figure 2. 4 ProScan (Smith et al., 1997). Figure 2.5 shows the sensitivity of PI to the wavelengths, where a gain equals 1 for the true profile (Smith et al. 2002). If the gain value corresponding to one wavelength is larger than 1, this wavelength is considered as having an important effect on the discomfort riding and would be amplified in PI calculation. On the other hand, if the gain value is less than 1, the wavelength is recognized to have an insensitive effect on riding quality and would be attenuated in PI calculation. According to Figure 2.5, it is indicated that PI addresses the wavelengths from 0.3 to 23 m (1 to 75 ft), especially wavelengths from 0.3 to 1 m. The filtering function of the rolling system is limited by its own geometry, which minimizes the impact of wavelengths shorter than 0.3 m or longer than 11 23 m. Figure 2. 5 Sensitivity of PI and IRI to Wavelength (Evans et al. 2003). 2.1.2 International Roughness Index International Roughness Index (IRI) is the ride quality statistic deriving from the response-type road roughness measuring systems (RTRRMS). In RTRRMS, the devices (also called as roadmeters) accumulate the suspension motion of a passenger car running over a pavement surface at a given speed. IRI mathematically standardizes the PTRRMS and duplicates the vibrations level of the vehicles. IRI is based on the response of a generic passenger car (known as the quarter-car model) to the roughness of a pavement surface. This simple dynamic model is a sprung mass resting on a suspension system with stiffness and damping (Figure 2.6). The wheel contacts the road through a tire-like spring. Road inputs to the car flex the tire, stroke the suspension and cause the sprung and unsprung masses to vibrate in the vertical direction (Shahin, 1994). The vertical velocity difference between sprung mass body and unsprung mass body produces the stroke of the suspension system, which is perceived by human body as the roughness of pavement. Equation 2.2 illustrates the algorithm used by IRI to 12 record these acceleration differences (Sayers, 1995). About 70% of vertical vibration of a passenger experiences can be described by the response of quarter-car to pavement. The IRI is the accumulated vehicle vibration divided by the distance traveled to give a ride quality statistic with units of slope (in/mile, m/km). Figure 2. 6 Quarter Car Model. (Gillespie, T.D., 1992) dtzz L IRI V L ts ? ?? ?= 0 1 (Equation 2.2) Where, IRI = International Roughness Index, in/ft; L = the distance quarter-car travels over, ft; V = the velocity of quarter-car, ft/s, ? s z = the vertical velocity of sprung mass, ft/s 2 ? t z = the vertical velocity of unsprung mass, ft/s 2 13 As viewed in Figure 2.5, IRI has sensitive gain value larger than 1 for wavelengths from 2.2 to 16.1 m (7.1 to 52.5 ft), which means this wavelength range are sensitive to the pavement ride quality based on IRI algorithm. This wavelength range is within the sensitivity band range for the PI statistic (i.e., between 0.3 to 23 m). However, it is also evident that PI focuses more on the smaller wavelengths around 1 meter, whereas IRI amplifies the larger band wavelengths from 3 to 11 m. 2.1.3 Comparison of PI with IRI Automotive engineers measure accelerations on the seat of the car to evaluate the suspension performance and the riding comfort of passengers. From numerous studies of the human body sensitivity to vibration in a sitting position, a vertical frequency of around 5 Hz is critical to the riding comfort. It is generally recognized that the human body has minimum tolerance to vertical vibration when the vibration frequency is about 5 Hz due to resonance of the abdominal cavity (Sayers and Karamihas, 1998). For example, Figure 2.7 shows that in the SAE J6A research, human body only can endure about 0.13 g acceleration when the vibration frequency equals to 5 Hz, while when the vibration is decreased to 1 Hz or increased to 20 Hz, the tolerable vertical acceleration for human body can be about 0.8g. Therefore, the pavement wavelengths raising the critical frequency should be emphasized by the pavement roughness index. In other words, the roughness index needs to have gain value larger than 1 to this wave band and be sensitive to these wavelengths. 14 Figure 2. 7 Human Body Sensitivity of the Vertical Vibration (Sayers and Karamihas, 1998). Assume the average speed of the vehicle ranges from 25 mi/hr to 80 mi/hr, the pavement surface wavelengths from 2.2 m to 9.4 meters can cause the vertical vibration of 5 Hz, which is mostly uncomfortable to passengers. A good ride quality index that accurately reflects user discomfort is required to make these wavelengths pronounced in evaluating the pavement roughness. Based on the former discussion about the sensitivity range of IRI and PI, it can be concluded that IRI well covers this critical wavelengths from 2.2 m to 9.4 m responsible for creating vertical vibration. As for PI, it not only covers this critical wavelength but also emphasizes other wavelengths unnecessary to producing vibration. This means that IRI can more accurately assess the ride quality through focusing on these uncomfortable pavement features. The quarter-car model uses the suspending system and pneumatic tire damping to isolate the effect of some speed-related vertical frequencies, and records the accelerations of the passenger seat. Its rational algorithm makes this model more related to the vertical 15 acceleration of vehicle than the hand-operated California profilograph rolling systems. Therefore, IRI can better represent the driving comfort than PI. 2.1.4 Correlation between PI and IRI Since PI and IRI statistic amplifies or attenuates profile features occurring at different wavelengths range, it is difficult to find an exact correlation between these two indices. Nevertheless, Figure 2.5 also presents that both of indices amplify the wavelengths from 2 to 10, even though at different degree. It makes the possibility to develop the connection between these two indices. Some previous research has presented that there is a relatively good statistic relationship between PI and IRI. In 1989, Pennsylvania Transportation Institute (PTI) conducted a full-scale field-testing program on behalf of Federal Highway Administration (FHWA) to develop calibration procedures for profilograph and evaluate equipments for measuring the smoothness of new pavement surfaces (Kulakowski and Wambold, 1989). In this project, 26 individual 0.1 mile long sections were selected from five different locations around Pennsylvania, including new or newly surfaced concrete pavements and asphalt pavements. Pavement roughness was recorded by profilograph and laser-type inertial profiler. Table 2.1 shows the relationship developed in this correlation. Solely based on the data from this research, the regression was not considerably different between concrete sections and asphalt sections. The manually calculated PI 0.2 had a correlation equation with IRI different from the correlation equation between computer-generated PI 0.2 and IRI. Slope from the regression equation from computer-generated PI 0.2 was considerably flatter. 16 1992 saw Arizona Department of Transportation (AZDOT) initiated a study to determine the feasibility of their inertial profiler (K.J.Law 690 DNC profilometer) on measuring the initial PCC pavement smoothness (Kombe and Kalevela, 1993). To examine the correlation between the profiler (IRI) and profilograph (PI) output, twelve typical newly-constructed 0.1-mile PCC pavement sections were selected to measure the smoothness by both devices. Simple linear regression (presented in Table 2.1) were performed between IRI and PI 0.2 values and indicated that generally good correlation existed between these two indexes with high R 2 of 0.93. During developing the new smoothness specifications for rigid and flexible pavements in Texas, University of Texas operated an investigation between McCracken California-type profilograph and the Face Dipstick, a manual profile measurement device in 1993 (Scofield, 1993). After collecting smoothness of 18 pavement sections including both asphalt and concrete pavements using these two devices, linear regression analysis (presented in Table 2.1) showed a strong collection (R 2 =0.92) between IRI and PI 0.2 . In order to compare its current rolling straightedge with other available measurement devices such as inertial profilers, Florida DOT conducted a study in 1997 (FLDOT, 1997). Twelve 0.5-mile sections from newly-constructed or resurfaced asphalt pavement were chosen for testing. Two type sensors were equipped in the inertial profilers, laser sensor and ultrasonic sensors. The linear relationships between IRI and PI 0.0 were developed respectively for each kind of inertial profiles. Both correlations (presented in Table 2.1) were fairly strong, with R 2 value of 0.88 and 0.67. Since the ultrasonic-based measurement adds the smoothness sensitivity to surface texture, cracking and temperature, the measurements deriving from ultrasonic profiler were higher than 17 laser-based and resulted in higher intercept in the regression equation. In 1996, as part of research on transfer a profilograph-based smoothness specification to a profile-based specification, Texas Transportation Institute (TTI) was involved to evaluate the relationship between PI and IRI (Fernando, 2000). Longitudinal surface profiles from 48 newly resurfaced AC pavement sections throughout Texas were measured to produce PI and IRI values. PI values were simulated by using ProScan software, IRI was automatically created from the inertial profiling system. In the relationship evaluation, a much stronger trend (presented in Table 2.1) was found between IRI and PI 0.0 than between IRI and PI 0.2. Since the application of blanking band mask the effect of certain component of the roughness, PI 0.2 was found to have a poorer relationship with IRI than PI 0.0 . In developing a series of relationships between IRI and PI that can assist states in transitioning to IRI or PI 0.0 smoothness specifications for AC and PCC pavement, research project using the Long Term Pavement Performance (LTPP) DataPave database to establish the relationships was sponsored and conducted by FHWA in 2002, hereafter referred as 2002 LTPP. A total of 1,793 LTPP test sections located in 47 states and 8 Canadian Provinces, which span all four climatic zone (dry freeze, dry nonfreeze, wet freeze and wet nonfreeze), formed the database for this evaluation (Smith.K.L et al. 2002). All these archived profile were measured with K.J. Law T-6600 inertial profiler from 1996 to 2001.PI and IRI values were generated from ?Indexer?, a profiler software developed by K.J. Law in 1995. Finally, the linear regression models were developed between IRI and PI 0.0 , PI 0.1 , PI 0.2 . Different pavement type and climate zone were found to have a significant effect on the regression model. The models in wet nonfreeze climate 18 zone, where Alabama belongs, are presented in Table 2.1. The regression equations from all these research are summarized in Table 2.1. Since the blanking band covers some components of pavement roughness, the correlation between IRI and PI 0.0 was found generally stronger than correlation between IRI and PI 0.2 . Table 2.1 shows that both the slope and intercept of the regression equations are dependent on the blanking band selected for calculating the PI ride quality statistic. When a 0.2 blanking band is used, the average slope is 3.7, and the average intercept is 64.6 in/mi. The values are various among different studies. When a 0.0 blanking band is used, both the slope and intercept decrease. The average of slope is 2.2, and the intercept is 18.2 in/mi. The values are more consistent between different studies than values in 0.2 blanking band. Data from PTI, ADOT, University of Texas and FLDOT research were developed by calculating one statistic for each of two independently obtained profiles. It is extremely difficult to track the identical profile with two different devices which can have a large influence in the quality of the correlations obtained. The data of 2002 LTPP and TTI were developed using a single source of raw profile data, then calculating both the IRI and PI from the same profile. One single source raw profile data eliminates the variation between two profilers used to respectively calculate IRI and PI value. The correlations would be sensitive only to the choice of blanking band and not of changes in profile characteristics. Table 2. 1 Summary of Documented PI-IRI Relationships. Study (Year) Pavement Types No. of Test Sections Remarks Linear Regression Equation, m/km Linear Regression Equation, in/mi R 2 PTI (1989) AC and PCC 26 Manual profilograph PI Laser-type inertial profiler IRI = 4.02* PI 0.2 + 1.11 IRI = 4.02* PI 0.2 + 70.13 0.57 PTI (1989) AC and PCC 26 Computerized profilograph PI Laser-type inertial profiler IRI = 2.46* PI 0.2 + 1.04 IRI = 2.46* PI 0.2 + 66.22 0.58 Arizona DOT (1992) PCC 12 Computerized profilograph PI Laser-type inertial profiler IRI = 6.10 * PI 0.2 + 0.83 IRI = 6.10* PI 0.2 + 52.90 0.93 University of Texas (1992) AC and PCC 18 Computerized profilograph PI Manually computed IRI (Dipstick) IRI = 2.83* PI 0.2 + 1.16 IRI = 2.83* PI 0.2 + 73.70 0.92 Texas Transportation Institute(1996) AC overlays 48 Computer-simulated PI Laser-type inertial profiler IRI = 4.08* PI 0.2 + 0.84 IRI = 4.08 * PI 0.2 + 52.74 0.56 LTPP (2002) AC Overlay (wet nonfreeze) 5126 LTPP Measurement data IRI = 3.43*PI 0.2 + 0.88 IRI = 3.43*PI 0.2 + 55.54 0.63 LTPP (2002) PCC (wet nonfreeze) 2888 LTPP Measurement data IRI= 2.87*PI 0.2 + 1.23 IRI= 2.87*PI 0.2 + 77.89 0.74 Florida DOT (1996) AC 12 Computerized profilograph PI Laser-type inertial profiler IRI = 2.19* PI 0.0 + 0.22 IRI = 2.19* PI 0.0 + 13.75 0.90 Florida DOT (1996) AC 12 Computerized profilograph PI Ultrasonic-type inertial profiler IRI = 2.20* PI 0.0 + 0.31 IRI = 2.20* PI 0.0 + 19.36 0.88 Texas Transportation Institute(1996) AC overlays 48 Computer-simulated PI Laser-type inertial profiler IRI = 2.14* PI 0.0 + 0.31 IRI = 2.14* PI 0.0 + 19.33 0.85 LTPP (2002) AC Overlay (wet nonfreeze) 5126 LTPP Measurement data IRI =2.42*PI 0.0 + 0.30 IRI =2.42*PI 0.0 + 19.12 0.84 LTPP (2002) PCC (wet nonfreeze) 2888 LTPP Measurement data IRI= 2.36* PI 0.0 + 0.32 IRI= 2.36* PI 0.0 + 20.09 0.84 19 20 2.2 Smoothness Specifications Conversion Methods With the update of the pavement roughness measurement devices or evaluation method, some states already had the experience on moving their former smoothness specifications to the new specifications. There are several methods widely used for making this conversion. The first method is based on engineering judgment without performing any comparative measurements. Indiana DOT and Missouri DOT selected their new reasonable IRI specifications from the practical knowledge and field experience of old specifications (Pellinen et al., 2003). The second method is to build the regressed correlation equations between old smoothness index and the new IRI index. Through the regress equations, the old smoothness index based specifications are transferred to specifications based on the new smoothness index. Illinois DOT established the regressed relationship between IRI and PI from an available database, such as LTPP (Rufino et al., 2001). The bonus and penalty range for the new index IRI, corresponding to the old PI index range, were determined by the correlations, shown in figure 2.8. 21 Figure 2. 8 Relationship between Simulated PI 0.2 and IRI in ILDOT Bridge Smoothness Study (Rufino et al., 2001). The third method is to statistically examine the surface smoothness data, and plot the probability or distribution curve for both old and new index. The new index limits for incentive/disincentive pay ranges correspond to the limits of old index by having the same amount of sections in each smooth level. Kansas DOT, Minnesota DOT and Wisconsin DOT applied this histogram method to establish new index specifications (Pellinen et al., 2003). Figure 2.9 shows an example how this approach is used. Simulated PI 0.2 , in/mile 22 Figure 2. 9 Conversion from Old Smoothness Specification to New One by Distribution Method (Hossain et al., 1995). From the distribution plot of PI 0.2 , it can be calculated that based on PI 0.2 specification from Kansas, 10% segments having PI 0.2 value less than 2 in/mile were qualified to the incentive, 80% segment would be full paid, 10% segments with PI 0.2 PI 0.0 , in/mile PI 0.2 , in/mile 23 value larger than 10 in/mile located in the penalty range. Therefore, based on distribution method, in order to allow 10% segments still could achieve bonus, the lower limit for PI 0.0 full pay range needed to be set at 10 in/mile. For having 80% segments in full pay range, the upper limit for PI 0.0 full pay range would be 26 in/mile. Consequently, the specifications based on the new roughness index were determined after setting those limits. 2.3 Effect of Short Interval on Estimating Pavement Smoothness Some bumps or localized irregularities are not detected by the average IRI values over long distance. Figure 2.10 shows a continuous plot of average IRI values over 0.1 mile interval and 0.01 mile interval of one pavement section. Assume the upper limit of average roughness considered barely acceptable without correction is 95 in/mi (WFLHD, 2003), 17% segments at 0.1 mile interval are recognized as bumps with needed correction, while 23% segments at 0.01 mile interval are detected as irregularities. By examining the IRI values at short interval, some of the segments requiring correction can be readily identified as the result of poorly constructed joints. Compared to long interval spacing, short interval spacing more accurately locates and quantifies localized ride quality problems. 24 0 50 100 150 200 0 5000 10000 15000 20000 Distance, ft I R I , in /m il e 0.01 Mile interval 0.1 Mile interval Figure 2. 10 Comparison of IRI Value at 0.1 mile interval with 0.01 mile interval. 2.4 Smoothness Specification 2.4.1 ALDOT Smoothness Specifications As of 2002, ALDOT had different pavement smoothness specifications for asphalt and concrete pavements (Table 2.2 and Figure 2.11). Both of the specifications were PI-based using a 0.2 inches blanking band. The smoothness values were required to be measured as soon as practical after paving and compaction. The measurement interval in quality assessment was 0.1 mile. The specifications for asphalt pavement combined continuous and step function pay factors for different smoothness levels. Concrete pavement had the step function pay factors for each smoothness level. Pay factors for concrete pavement were higher than asphalt, either in bonus range or penalty range. 25 Table 2. 2 Alabama Pavement Smoothness Specifications for PI 0.2 (ALDOT, 2002). Price Adjustments Pavement Type Equipment Section Length Blanking Band Profile Index, in/mile Contract Price Adjustment of pavement unit bid price, % Under 2 105 - (profile index/4) 2 to 4 100 4 to 10 100 - (profile index-4)/0.3 Asphalt Pavement California profilograph 0.1 mile 0.2 inches Over 10 Unacceptable Under 3 105 3 to 6 100 6 to 8 95 8 to 10 90 Concrete Pavement California profilograph 0.1 mile 0.2 inches Over 10 Unacceptable 70 75 80 85 90 95 100 105 110 024681012 PI0.2, in/mile P er cen t P ay , % Asphalt Pavement Concrete Pavement Figure 2. 11 2002 ALDOT Specification for Pavement Roughness. The ALDOT smoothness specifications were changed in 2003 so that ride quality would be evaluated using a 0.0 blanking band (PI 0.0 ). At the same time, the separate specifications for asphalt concrete and Portland cement concrete pavements were eliminated. There is only one specification for ride quality, regardless of the type of pavement. Pavement products are paid only by the ride service they can provide, concrete pavements are required to reach the same comfort level as asphalt pavement to earn the same pay. The pay functions of concrete were also changed from step functions to the 26 combination of step and continuous functions. Table 2.3 states the current PI 0.0 ALDOT smoothness specifications. Table 2. 3 Alabama Pavement Smoothness Specifications for PI 0.0 (ALDOT, 2003). Profile Index In/mi/Section Contract Price Adjustment Percent of Pavement Unit Price Under 10.0 105 ? (PI/2) 10.0 to less than 20.0 100 20.0 thru 50.0 100 - (PI -20)/1.5 Over 50.0 Unacceptable While the current ALDOT specification is based on PI using the 0.0 blanking band, the analyses in the following chapters will include the evaluation of PI calculated with both blanking bands and the IRI. The PI 0.2 is included because a number of states still use this value; there is also a substantial amount of previous research based on this value. 2.4.2 Smoothness Specifications of Other DOTs After changing its smoothness specification toward 0.0 inches blanking band, it is still necessary for ALDOT to track the implement of this new specification. In this study, smoothness specifications based on PI 0.0 from other states were collected to compare and evaluate the current ALDOT smoothness specification. The specifications from five states (plotted in Figure 2.12 and Figure 2.13) state that these states employ different smoothness specifications for AC pavements and PCC pavements. Figure 2.12 records the PI 0.0 specifications for AC pavement. In this figure, these five states all deploy the step functions to pay the pavements at each smoothness level. And also, the incentive and disincentive ranges are divided into several steps to have an accurate pay for each riding quality level. The lower limits for PI 0.0 full pay range are averaged around 17 in/mile; the average of upper limits is 27 in/mile. Compared to 27 current ALDOT PI 0.0 specifications, ALDOT uses continuous function for paying different smoothens level. And also, ALDOT conducts considerable stricter specifications, where PI 0.0 full pay range from 10 in/mile to 20 in/mile, than these states. Figure 2. 12 PI 0.0 Specifications for AC Pavement from other DOTs (Pellinen et al., 2003). Figure 2.13 plots the PI 0.0 specifications for PCC pavement from other DOTs. All these state have the more lenient specifications for PCC pavement than AC pavement. For example, Kansas DOT pays more incentive for smooth PCC pavement than smooth AC pavement, and has the same penalty for pavement generating PI 0.0 values larger than 40 in/mile for both pavement types. The step functions are still used for paying concrete pavement. The incentive and disincentive ranges are also separated into several steps to have an accurate pay for each riding quality level. The lower limits of full pay range are averaged around 26 in/mile. States like Indiana and Pennsylvania have no penalty to the PCC pavement. The upper limits of full pay range are around 42 in/mile averaged from Kansas and Wisconsin specifications. 28 Figure 2. 13 PI 0.0 Specifications for PCC Pavement from other DOTs (Pellinen et al., 2003). Currently, IRI is already applied in quality assessment of some states. Since the main objective of this study is to transfer PI based specification to IRI based, the IRI specifications from other states are plotted in Figure 2.14 to provide a reference for establishing ALDOT IRI specifications. Within the seven states in Figure 2.14, Maine and Virginia have the same specifications for AC pavement and PCC pavement. In other states, like Connecticut, South Dakota, Vermont and Washington, IRI-based specifications are just for evaluating flexible pavement; PI-based specifications are still used to investigate rigid pavement. Except Maine, other states use the step function to pay the pavement at different smoothness levels. The lower limits of full pay range from these seven states have the average of 58.5 in/mile; the average of the upper limits is around 73 in/mile. 29 72 74 76 78 80 82 84 86 88 90 92 94 96 98 100 102 104 106 108 40 50 60 70 80 90 100 110 120 IRI, in/mile P e r c e n t of U n it P e r c e n t, % Indiana Maine Vermo n t Connecticut* South Dakota Virginia Washington *: For Connecticut, South Dakota, Virginia and Washington, there is no detailed pay factor value available. 105% and 95% was assumed as the bonus and penalty pay factor. Figure 2. 14 IRI Specifications from other DOTs. 2.5 Summary From the literature reviews discussed in this chapter, several conclusions can be drawn as follows: z PI is the physical accumulation of pavement deviations. The geometry of the rolling system limits PI sensitive to pavement wavelength from 0.3 to 23 m, especially from 0.3 to 1 m. However, IRI is the accumulated vertical vibration simulated by Quarter-car model. This index is sensitive to the wavelengths spanning from 2 to 16 m. IRI wavelength range well covers the waves responsible for 5Hz critical frequency vibration, which ranges from 2 m to 10 m and human body has the 30 minimum tolerance to. Therefore, IRI better represents the pavement riding quality. z The short interval spacing makes the localized bumps pronounced. Some of the bumps averaged in the 0.1 mile interval can be detected in the 0.01 mile interval. Short interval localizes and quantifies the local riding problems. z ALDOT moved its smoothness specifications from PI 0.2 to PI 0.0 in 2003. In PI 0.0 specifications, AC pavement and PCC pavement have the same pay standard. The specifications provide full pay for pavement smoothness ranged from 10 in/mile to 20 in/mile. Compared to ALDOT, most of other states have different specifications for each pavement type. Either for AC pavement or PCC pavement, the specifications from several other states are more lenient than ALDOT specifications. Base on the specifications from five states plotted in Figure 2.12 and Figure 2.13, the average lower limit of full pay for AC pavement is 17 in/mile, upper limit is 27 in/mile. The average lower limit of full pay for PCC pavement is 26 in/mile, upper limit is 46 in/mile. z As the specifications of seven states using IRI plotted in Figure 2.14, step functions are used to pay the different smooth level pavement. The lower limits of full pay range from these seven states have the average of 58.5 in/mile; the average of the upper limits is around 73 in/mile. 31 CHAPTER THREE DATA COLLECTION AND DATABASE DEVELOPMENT 3.1 Data Collection The Roadware ARAN (automated pavement analyzer) vehicle was used to collect pavement longitudinal profiles in this study. This vehicle has several subsystems, which can collect the raw profile data in each wheel path for calculating ride quality statistics, such as IRI and PI. Other pavement condition information collected includes rut depth estimates (both wheel paths) and pavement macro texture in the right wheel path only. Auburn University has an ARAN van of model 4300, which uses the South Dakota Profiler (SDP) inertial profiling system sensor set-up. This is a laser-accelerometer combination system to measure the longitudinal profile. This system measures the pavement profile at intervals as short as 100 mm (4 in) at variable speeds up to 100 km/h (60 mph) (Roadware, 2005). An automated standard moving-average filter from ARAN translates the digital sensor data into a representation of the relative surface profiles. Therefore, the output profiles from ARAN system are considered pre-filtered before any further analysis is conducted. Pavement longitudinal profile measured by laser inertial profiler, like ARAN van, covers a slice of pavement. With the variation between different driver and the variation of start point, it is hard to repeat the exact same profile measure. But the former research has indicated that inertial profiler has high repeatability. In 2000 and 2001, Highway 32 Research Center in Auburn University operated the repeatability estimates for inertial profiler in National Center for Asphalt Technology (NCAT) test track. The research showed that IRI had the coefficient of variance (COV) around 9% between different repeat measures. For rough and high ESALs pavement, COV value increased to 15% (Stroup Gardiner, 2004). It was suggested that the one-time measurement of profile from ARAN was sufficient. In this study, the ARAN Van was driven over a range of asphalt pavement and concrete pavement projects to collect longitudinal profiles (total 20 sections) in both right and left wheelpath. The longitudinal profiles of all sections were measured at least twice (i.e., 2 replicates). When possible, the profiles were measured three times for one section, ending up with a total of 57 pavement profiles. Figure 3. 1 ARAN Van Model 4300 (Roadware, 2005). 3.1.1 Asphalt Pavement Profiles Longitudinal profiles of asphalt pavement were collected from four Alabama paving projects using ARAN inertial profiler. These projects are briefly described in Table 3.1. Projects were HMA overlays after an initial mill only, or a mill and chip seal preparation. 33 Longitudinal profiles were collected as soon as practical after the paving and rolling was completed. Table 3. 1 Project Descriptions (Alabama Mill and Fill Projects). Project Location Layer Mix Design Max. Agg. Size, in Traffic Level Preparation Binder Superpave 424 1 ESAL E 1 Milling and chip seal 1 US 280 Wearing Surface Superpave 424 0.75 ESAL C/D 1 Patching and chip seal Binder Superpave 424 1 ESAL E Milling 2 Selma Wearing Surface Superpave 424 0.75 ESAL C/D None Binder Superpave 424 1 ESAL E Milling 3 US 82 Wearing Surface Superpave 424 0.75 ESAL E None 4 Opelika Binder SMA 423 1 ESAL E Milling 1 ESAL C/D range: 1.0X10 6 ? ESALs < 1.0X10 7 E range: 1.0X10 7 ? ESALs < 3.0X10 7 (ALDOT, 2002) Project 1, 2 and 3 had Superpave bituminous concrete binder and wearing surface layers constructed according to Section 424 of ALDOT 2002 Specification. Project 4 had SMA 423 as binder concrete according to Section 395 of ALDOT 2002 Specification. These detailed gradation information about these mixtures were presented in somewhere else (Williams, 2003). 3.1.2 Concrete Pavement Profiles Concrete pavement data were surveyed in Montr?al, Quebec. The description of the four concrete sections is stated in Table 3.2 (Carter, 2005). All of these sections are new concrete pavement, except project 1 with short slabs. While project 1 was not new, the concrete pavement was still in its good shape and condition. 34 Table 3. 2 Descriptions of Concrete Pavements. Project Slabs Texture Year of Construction Length of Section, Km 1 Short Slabs Skid abrader 1997 1.5 2 Continuous Slabs Transverse Tinning 2004 1.5 3 Continuous Slabs Transverse Tinning 2004 0.5 4 Continuous Slabs Longitudinal Tinning 2004 0.5 Since the different climate conditions between Alabama and Quebec, this concrete profile database has the limitation to be used in Alabama smoothness specification development. However, these four concrete pavement projects located in the same urban highway system, had the structure of the typical 9? thick slab, and were built by the same contractor in the recent years. The data based on them can be considered as a homogenous database deriving from newly-constructed concrete pavements. Moreover, Alabama has very few new concrete pavement constructions, which make it difficult to build a new concrete pavement database. Therefore, concrete pavement database from Qu?bec were just used to compare the way different roughness indices evaluate pavement riding quality. Only asphalt pavement data were used to transfer ALDOT specifications. 3.2 ProVAL ProVAL was performed as the analysis tool in this study. ProVAL (Profile Viewing and Analysis), published by Federal Highway Administration in 2005, is an engineering software application that allows users to view and analyze pavement profiles in many different ways (ProVAL, 2005). This software can perform various filters to pavement profiles, provide power spectral density information of profiles, simulate profilograph trace and operate the smoothness statistic analysis. Also, ProVAL can complete these 35 analyses with two unit systems: Metric and USCS. Finally, an analysis report can be created automatically. Figure 3.2 shows the major function tab of this software. Figure 3. 2 Main Function of ProVAL 2.5. Profilograph simulation is designed to emulate profilograph traces, like California Profilograph, for the profiles collected using inertial profilers. The default wheel offsets is the geometry of the California rolling system. The algorithm here to calculate the deviations of pavement similarly follows Equation 2.1. The elevation of the referred surface can be computed by averaging the elevations of wheel groups. The deviation of the recording wheel can be calculated from the disparity of its elevation from the surface. The location of the blanking band is determined by the least squares linear fit, which makes the centerline of the blanking band pass through the middle of the profile. Therefore, the blanking band can cover as many of irregularities as possible. In this software, after setting the input value of blanking band, minimum scallop width, minimum scallop height and scallop rounding increment, the button of Run Filter is pressed to perform the Profilograph simulation filter. As a result, the California 36 Profilograph trace appears on the screen with the default interval set as 0.1 mile (528 feet). If smaller, larger segment or part of the profiles is interested to be analyzed, the Segments button allows adding and deleting segments, even changing the desired analysis interval. After the input of all parameters, the Analyze button is pressed to run and compute the California Profilograph Index. Consequently, the simulated PI values are calculated for each segment of profiles. In this study, ProVAL2.5 was used to model the California profilograph trace and calculate the PI values in different blanking band (0.0and 0.2 inches) and different segment intervals (0.01 and 0.1 miles). Figure 3. 3 Profilograph Simulation Function Tab. The second main function of ProVAL is to compute ride statistics, such as International Roughness Index and Half-car Roughness Index, which is the IRI algorithm applied to average of two wheelpath profiles. In ProVAL, the algorithm of quarter-car model is used for calculating IRI value. The raw profile provides the height information 37 of the unsprung and sprung mass body. With parameters of the suspension system and the tire system in quarter-car, the vertical acceleration difference between these two body parts can be computed with integration method. In ProVAL, the default values of vehicle velocity and segment length are 80 km/h and 528 feet, respectively. If the input profiles are not pre-filtered, the required 250 mm moving average filter or other desired filters can be performed on the raw profiles before further analysis. After that, the Analyze button starts to run the analysis. As a result, IRI value of each segment of each wheelpath appears on the screen. Figure 3.4 presents one ride statistics analysis example. This study applied ProVAL to calculate IRI for each segment of each section tested, with 0.01 mile and 0.1 mile segment interval. Figure 3. 4 Ride Statistics Function Tab. 38 CHAPTER FOUR DATA ANALYSIS 4.1 Data Quality After profiles were processed using the ProVAL2.5 software, database consisting of IRI and simulated PI values were developed for the further analysis. During the data collection using ARAN Van, sometimes optical triggers were placed on the pavement before the segment collections to indicate the start of another segment. The triggers produced evident peaks on the profiles. To eliminate the effect of those peaks, the remainder of the database were evaluated and deleted as outliers, which were defined as values beyond plus and minus three standard deviations of the average. After removing these abnormal values, the data were plotted in Figure 4.1 to 4.13 to evaluate the quality. 4.1.1 Smoothness Data of Asphalt Concrete (AC) Pavement Figures 4.1 to 4.7 present the range and distribution of IRI and PI values at each of two intervals (0.1 m, 0.01 mi) for asphalt concrete pavement. These figures demonstrate that PI and IRI values fully cover the range of typically reported smoothness values of new construction and AC overlays (i.e., IRI between 50 to 125 in/mi, PI 0.2 between 0 and 15 in/mi) (Smith et al. 2002). Therefore, the assembled AC overlay database can be considered as a representative of asphalt overlay pavement projects in Alabama. 39 0 20 40 60 80 100 120 140 160 180 15 30 45 60 75 90 105 120 135 150 165 180 195 210 IRI, in/mile Fr e q u e n c y 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% C u m u la tive F r e q u e nc y Figure 4. 1 Histogram of AC IRI Value Distribution at 0.01 Mile Interval. 0 5 10 15 20 25 30 15 30 45 60 75 90 105120135150165180195210 IRI, in/mile Fr e q u e n c y 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% C u mu la tiv e F r e q u e n c y Figure 4. 2 Histogram of AC IRI Value Distribution at 0.1 Mile Interval. Figures 4.1 and 4.2 indicate that distributions of IRI values at 0.01 mile interval and 0.1 mile interval are similar. The 50 th percentile of 0.1 mile interval values is 72 in/mile, 40 0.01 mile interval values has close 50 th percentile of 68 in/mile. As it was expected, the distribution of the IRI values calculated at 0.01 mile interval has a flatter distribution with more data spreading into both tails than 0.1 mile interval. The 0.1 mile interval averages the bumps, and therefore smoothes out the tail in the longer distance to gain standard deviation of 38 in/mile. However, the smaller interval, accounting for the shorter areas with localized irregularities, spreads data to wider tails and has larger standard deviation of 49 in/mile. As seen in Figure 4.1 and Figure 4.2, a transformation of the database may be helpful in order to obtain a more normally distributed distribution of IRI data. However, as already presented in smoothness specifications, pavements are sorted into four population by its smoothness according to the practical experiences and engineer judgments: very smooth pavement which is the product of excellent construction and is qualified for the incentive, smooth pavement which is the result of qualified construction and would earn the full pay, the rough pavement which is created by the unqualified construction and only achieves parts of the bid price, the very rough pavement which is produced by the poor construction and could not be accepted without correction. Therefore, there would actually be several populations represented by the data, but the separation of different population is not readily evident. There is no sufficient data in these particular projects to provide project-specific information, which is needed to sort each data base into independent databases of low, med, and high roughness. Compared to the IRI values distribution, PI 0.2 values have completely different trends, either for 0.01 mile interval or for 0.1 mile interval. For the 0.01 mile interval, Figure 4.3 shows that 64% segments have PI 0.2 value of 0 in/mi. These data initially appear to have a 41 very limited distribution. However, this appearance is a function of the high frequency of values at 0 in/mi. If 0 in/mile values were taken out, the remaining data present other populations (Figure 4.4). 0 1000 2000 3000 4000 5000 6000 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 PI0.2, in/mile Fr e q u e n c y 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Cum u la tiv e F r e q ue n c y Figure 4. 3 Histogram of AC PI 0.2 Value Distribution at 0.01 Mile Interval. 0 50 100 150 200 250 300 4 9 14 19 24 29 34 39 44 49 54 59 64 69 PI0.2, in/mile Fre q u e n c y 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Cu m u la ti v e F r e q u e n c y Figure 4. 4 Histogram of AC PI 0.2 Value at 0.01 Mile Interval after Taking out PI 0.2 Values of 0 in/mile. 42 The data distribution for 0.1 mile interval shows that PI 0.2 values calculated using at 0.1 mile interval comprise 18.5% of the segments having a value less than 2 in/mi. These segments would qualify for a 5% bonus by the pre-2003 specifications. Based on pre-2003 specifications, 13% segments associated with PI 0.2 from 2 to 4 in/mile can receive full pay; 32% segments would have deducted pay; 36.5% segments are unacceptable without correction. 0 10 20 30 40 50 60 70 80 90 100 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 PI0.2, in/mile Fre q u e n c y 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Cum u la tiv e F r e q ue n c y Figure 4. 5 Histogram of AC PI 0.2 Value Distribution at 0.1 Mile Interval. Figure 4.6 and figure 4.7 present distributions of PI 0.0 values, like IRI distributions are skewed to the left. The 50 th percentile is associated with a PI 0.0 of 32 in/mi when using an interval of 0.01 miles. The current specified interval of 0.1 mile shows 50 th percentile of PI 0.0 value is 27 in/mile. 0.1 mile interval also has smaller standard deviation of 18 in/mile than 0.01 mile standard deviation of 28 in/mile. Smaller interval moves more data to the tails of the distribution and creates higher standard deviation. 43 0 50 100 150 200 250 300 6 111621263136414651566166717681869196 PI0.0, in/mile Fre q u e n c y 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Cum u la tive F r e q u e nc y Figure 4. 6 Histogram of AC PI 0.0 Value Distribution at 0.01 mile interval. 0 10 20 30 40 50 60 6 111621263136414651566166717681869196 PI0.0, in/mile Fre q u e n c y 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Cum u la tive F r e q u e nc y Figure 4. 7 Histogram of AC PI 0.0 Value Distribution at 0.1 Mile Interval. 44 4.1.2 Smoothness Data of Portland Cement Concrete Pavement Figure 4.8 to Figure 4.13 demonstrate the range of PI and IRI values for a range of differently textured PCC pavements at 0.1 mile and 0.01 mile intervals. According to these figures, it can be seen that PI and IRI values fully cover the range of typical smoothness values of newly constructed PCC pavement (i.e., IRI between 50 to 150 in/mi, PI 0.2 between 0 and 25 in/mi) (Smith et al. 2002). Therefore, this new PCC pavement database can be considered as one representative of new PCC pavement. As seen from figure 4.8 and figure 4.9, IRI values of concrete pavement at both 0.1 mile interval and 0.01 mile interval have slightly skewed distributions, with 50 th percentile around 95 in/mi. Like IRI value distributions of AC pavement, the 0.01 mile interval has a larger standard deviation than the 0.1 mile interval, which flattens the distribution curve and brings more segments into the right side tails. IRI values using 0.01 mile interval have a standard deviation of 40 in/mile; the 0.1 mile interval has a standard deviation of 20 in/mile. 45 0 5 10 15 20 25 30 35 40 34 50 66 82 98 114 130 146 162 178 194 210 IRI, in/mile Fr e q u e n c y 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% C u m u la tive F r e q u e nc y Figure 4. 8 Histogram of IRI Value Distribution of PCC at 0.01 Mile Interval. 0 1 2 3 4 5 6 7 8 9 34 50 66 82 98 114 130 146 162 178 194 210 IRI, in/mile Fr e q u e n c y 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Cum u la tiv e F r e q ue n c y Figure 4. 9 Histogram of IRI Value Distribution of PCC at 0.1 Mile Interval. The PI 0.2 data distribution of concrete pavement is also similar to asphalt pavement. When the interval changes from 0.1 mile to 0.01 mile, 50% segments focus on the PI 0.2 of zero. This emphasizes 0.2 inches blanking band is unable to record small roughness and 46 produces a large percent of segments reaching the bonus. 0 100 200 300 400 500 600 700 800 900 0 6 12 18 24 30 36 42 48 PI0.2, in/mile Fr e q u e n c y 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Cum u la tiv e F r e q ue n c y Figure 4. 10 Histogram of PI 0.2 Value Distribution of PCC at 0.01 Mile Interval. 0 2 4 6 8 10 12 14 16 18 20 0 6 12 18 24 30 36 42 48 PI0.2, in/mile Fr e q u e n c y 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% C u m u la tiv e F r e q u e nc y Figure 4. 11 Histogram of PI 0.2 Value Distribution of PCC at 0.1 Mile Interval. As for PI 0.0 data of 0.1 mile or 0.01 mile interval, concrete pavement also has slightly skewed distributions. PI 0.0 values have almost same shape of distribution curves with IRI. 47 0.1 mile interval generates the PI 0.0 value of concrete pavement with 50 th percentile of 41 in/mile, with a standard deviation of 13 in/mile. The 0.01 mile interval creates larger average of 44 in/mile and larger standard deviation of 23 in/mile. Unlike the PI 0.2 , the different intervals present dissimilar distribution patterns; IRI and PI 0.0 have a similar pattern either for 0.1 mile interval or 0.01 mile interval. 0 20 40 60 80 100 120 10 18 26 34 42 50 58 66 74 82 90 98 PI0.0, in/mile Fr e q u e n c y 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Cum u la tiv e F r e q ue n c y Figure 4. 12 Histogram of PI 0.0 Value Distribution of PCC at 0.01 Mile Interval. 48 0 5 10 15 20 25 30 10 18 26 34 42 50 58 66 74 82 90 98 PI0.0, in/mile Fr e q u e n c y 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% C u m u la tiv e F r e q ue nc y Figure 4. 13 Histogram of PI 0.0 Value Distribution of PCC at 0.1 Mile Interval. 4.2 Effect of blanking band on Evaluating Pavement Smoothness Numbers of states still use PI 0.2 in the quality assessment, especially for concrete pavement. As known in literature review, 0.2? blanking band covers some components of pavement roughness. And also, the same specifications were recommended for both AC and PCC pavements (Smith et al. 1997), so it is meaningful to see whether this blanking band has the same influence on the AC pavement and PCC pavement. Since the database in this study came from limited projects, there are limitations for these data to represent the roughness feature of the whole new pavements. Therefore, the emphasis here focuses on the comparison of the effects of different roughness indexes, not the comparison of the roughness of different pavement type. From PI 0.2 distributions of AC and PCC pavements in Figure 4.14, it can be seen that these two groups of AC and PCC pavements have close roughness condition based on 49 PI 0.2 . If paid by Alabama Pre-2003 specification, contractors from both industries can achieve similar degree of pay for providing the PI 0.2 - based ride quality. There would be approximately 31% asphalt segment (PI 0.2 between 0 and 4 in/mi) and 22% concrete segment (PI 0.2 between 0 and 6 in/mi) receiving full pay or bonus. 32% asphalt segments and 30% concrete segments would get penalty price. 34% asphalt segments and 44% concrete segments (PI 0.2 larger than 10 in/mile) need to be corrected. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0 1020304050 PI0.2, in/mile Cu m u la tiv e F r e q u e n c y AC Pavement PCC Pavement Figure 4. 14 Comparison of PI 0.2 of AC and PCC Pavement at 0.1 Mile Interval. However, the PI 0.0 cumulative frequency curves display a large disparity between these two groups of asphalt pavements and concrete pavements (Figure 4.15). When using the old PI 0.2 specification, similar pay for asphalt and concrete pavements could be obtained. But for the same pavement profile database, the current PI 0.0 specification highlights the rougher service provided by these concrete pavements compared to asphalt pavements in this study. Following the current Alabama PI 0.0 specification, 20% of the AC projects would receive full pay while 0% of the PCC projects would receive full pay. 50 Concrete pavements have 85% segments get disincentive pay and 15% segments need extra correction. Based on PI 0.0 values, contractors of those concrete pavements would need a large improvement in construction procedures to achieve the same ride quality and earn the same pay as those AC pavement contractors. Although smoothness specifications need to provide fair competition between asphalt pavement and concrete pavement industry, there is no reason to accept worse ride quality with the same pay. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0 2040608010 PI0.0, in/mile C u m u la tive F r e q ue n c y AC Pavement PCC Pavement Figure 4. 15 Comparison of PI 0.0 of AC and PCC Pavement at 0.1 Mile Interval. Figure 4.16 also shows an evident difference between the IRI value distributions of asphalt pavement and concrete pavements. Assumed that full-pay upper limits of IRI is set on 75 in/mile, only 12% of the concrete segments could achieve the full pay, while 62% asphalt segments would be qualified for 100% pay. 51 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 35 55 75 95 115 135 155 175 IRI, in/mile Cum u la tiv e F r e q u e nc y AC Pavement PCC Pavement Figure 4. 16 Comparison of IRI of AC and PCC Pavement at 0.1 Mile Interval. These two comparisons further prove that PI with 0.2 inches blanking band makes the small roughness unnoticeable and moves segments to ?smooth? level. As PI 0.0 and IRI are more sensitive to smaller vertical displacements, those segments defined as ?smooth? by PI 0.2 would be considered as rough segments by both the PI 0.0 and IRI. Although 0.2? blanking band covers the small defects for both pavements, 0.2? blanking band has different effect on evaluating the smoothness of asphalt and concrete pavements in this study. Based on the database developed in this study, it shows that more amounts of irregularities from those concrete pavements are concealed by 0.2? blanking band than these AC pavements. 0.2? blanking band allows the worse-quality PCC pavement to earn the same pay as AC pavement. Accordingly, those concrete pavements contractors need more rapid improvement on the roughness measurement and smoothness specification. Deleting the blanking band would promote smoother concrete pavements. 52 4.3 Conversion of PI Specifications to IRI Specifications 4.3.1 Specification Conversion Using Regression Equations As indicated by the scatter plot in Appendix A and the previous research about the correlations between PI X and IRI, the simple linear relation model was chosen to describe the relationship between PI X and IRI. The model is shown in equation 4.1. IRI = ? 0 + ? 1 * PI X (Equation 4.1) Where, IRI = International Roughness Index, in/mile PI X = Simulated Profile Index for blanking band x (x= 0.0, 0.1 or 0.2 inches), in/mile ? 0 , ? 1 = Regression parameter In the 2002 LTPP study, regression equations from different climate zones have significant differences between each other. Asphalt pavement data used in this study was collected in Alabama. This corresponds with the LTPP population of asphalt pavement in the wet no-freeze (WNF) climate zone. The equations based on this database should be applicable to the profiles obtained for this study. The concrete database used in this study was gathered at Quebec, Canada, which is located in wet-freeze (WF) climate zone. Due to the climate limitation and other construction or material difference between Quebec PCC pavement and Alabama PCC pavement, the regression equations developed on this database could not adapt to Alabama. So the correlation model of WNF zone PCC pavement in 2002 LTPP was applied here to transfer Alabama specifications. By following models of equation 4.1, the regression equations for asphalt pavement 53 were developed and shown in Table 4.1. Compared with the regression equations from 2002 LTPP study (asphalt pavement at 0.1 mile interval), the equations developed in this study and those for the LTPP study have similar intercepts: 55 in/mile for the 0.2? blanking band and 18 in/mi for 0.0? blanking bands. The slopes between IRI and PI x from the 2002 LTPP equations are slightly higher than those found in this study. Equations for 0.01 mile interval are distinct from 0.1 mile interval equations, with a slightly higher slope and a noticeably higher intercept. The short interval creates the database with a higher variation than 0.1 mile interval, contributing to the smaller R 2 . Table 4. 1 Correlation Equations between IRI and PI in this Study and LTPP (Asphalt Overlay Pavement). Correlation Equation (IRI,PI=in/mile) Climate Zone Number of segments Interval (mile) PI 0.2 PI 0.0 8332 0.01 IRI=1.9295*PI 0.2 +62.82, R 2 =0.70 IRI=1.5699*PI 0.0 +19.91, R 2 =0.79 This Study WNF a 869 0.1 IRI=2.3688*PI 0.2 +54.10, R 2 =0.91 IRI=2.0708*PI 0.0 +17.84, R 2 =0.92 LTPP(2002) WNF 5126 0.1 IRI=3.4267*PI 0.2 +55.54, R 2 =0.63 IRI=2.4230*PI 0.0 +19.12, R 2 =0.84 a WNF: Wet-Nonfreeze climate zone Table 4.1 shows that regression equations for asphalt pavement in this study have high significance of regression with R 2 values consistently above 0.9. Even for 0.01 mile interval, regression models still have a good R 2 (around 0.75). In other words, 75% change of IRI can be explained by the linear change of PI. Using these developed equations, the current Alabama asphalt pavement PI based specification could be reasonably transferred to IRI based specification. Table 4.2 presents the converted IRI-based specification results. The continuous pay factor functions were retained through these regression models. 54 Table 4. 2 Converted IRI Specifications for Asphalt Pavement at 0.1 Mile Interval by Regression Equations. Price Adjustment of Pavement Unit Bid Price by PI 0.2 PI 0.0 , in/mi IRI, in/mi Price Adjustment of Pavement Unit Bid Price by IRI 105-(PI/20) Under 10 Under 38 109.3 - 0.24*IRI 100 10 to 20 38 to 60 100 100-(PI-20)/1.5 20 to 50 60 to 121 119.1 - 0.322*IRI Unacceptable Over 50 Over 121 Unacceptable Owning to the absence of Alabama rigid pavement data, the linear regression model of WNF zone PCC pavement from 2002 LTPP study (Table 2.1) were used for concrete pavement smoothness specification transfer. According to regression equations established for WNF climate zone from 2002 LTPP study, the current ALDOT concrete pavement PI 0.0 specifications were changed to IRI base specification, shown in table 4.3. Table 4. 3 IRI Specifications for PCC Pavement. Price Adjustment of pavement Unit Bid Price by PI 0.2 PI 0.0 , in/mi IRI, in/mi (LTPP) Price Adjustment of pavement Unit Bid Price by IRI 105-(PI/20) Under 10 Under 44 112 - 0.24* IRI 100 10 to 20 44 to 67 100 100-(PI-20)/1.5 20 to 50 67 to 138 119 ? 0.282* IRI Unacceptable Over 50 Over 138 Unacceptable 4.3.2 Specification Conversion Using Distribution Method Pavements with different smoothness levels can be paid for different percentages of the initial bid: bonus pay, full pay or penalty pay. For contractor, if an existing smoothness specification is converted to new specification based on another index, the same pavement product is expected to receive the same pay either based on former smoothness index or new one. But for the agency and the public, the transfer of smoothness index is for more accurately evaluating the pavement roughness and promoting the good construction. If the pavement product does not improve the driving comfort but is paid the incentive by the former index, its payment needs to be adjusted in 55 the new index. The distribution method is to transfer specification limits between different indices by using the concept that each index will have the same number of segments in the same payment level. The percentages of bonus, full or penalty pay pavement determined by the former specifications are used as the reference to start a new specification. This conversion makes the change of evaluation system comfortable for contractors, but it also makes the public having the risk to receive the worse paving product with paying the same amount. Therefore, the result from the distribution method is just a first step to establish the new specifications. With the application of this primary result, the further adjustments are needed to decide the reasonable percentage of pavement having incentive/disincentive. Herein, the distribution method provided a primary result for moving PI-based specifications to IRI-based; the further adjustment is out of the range of this study. The distribution curves of PI 0.0 were employed to determine the number of segments at different pay levels: the incentive range, full pay range, disincentive range and the unacceptable range, respectively. Consequently, the limits of IRI specifications can be determined by having the same number of segments in each roughness level as PI 0.0 specifications. Due to the current ALDOT PI 0.0 specification, 0% segment of asphalt pavement projects used in this study would reach the bonus pay; PI 0.0 range for full payment is from 0 th to 20 th percentile. It should be pointed out that the projects in this study were all mill and overlay over existing distressed HMA pavements, which is a contributing factor to the contractors? ability to restore a new pavement ride. Figure 4.17 also indicates that 56 71% of asphalt segments would have a penalty pay, while 9% would be unacceptable without correction. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0 10203040506070809010 PI0.0, in/mile Cum u la tive F r e q u e nc y Figure 4. 17 Pavement Percentages of AC Pavements in Each Pay Level according to PI 0.0 Specifications. Since these data profiles of asphalt pavement come from overlay projects, overlay pavements are possibly rougher than totally new-constructed pavements. That is one of the reasons that just small amount of the asphalt pavement segments in this database are reached bonus or full pay limits. Another reason is the strict requirement of PI 0.0 full pay limits in current ALDOT specification (PI 0.0 value from 10 in/mile to 20 in/mile), which results in small number of full-pay segments and bonus-pay segments. Figure 2.12 shows that the lower limit of full pay from other DOTs is 17 in/mile; the upper limit of full pay from other DOTs is 27 in/mile. This means that currently ALDOT specifications are stricter than most other DOT ride quality specifications, which suggests that a little lenient range in limits could be more reasonable. 91% 20% 0% 57 If the ALDOT specification is adjusted to the average limits of full pay range from other DOT (PI 0.0 value from 17 in/mile to 27 in/mile), 11% asphalt segments would achieve a bonus, 40% segments would earn full pay, 40% segments would receive penalty payment and 9% segments would need to be corrected. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0 10203040506070809010 PI0.0, in/mile Cum u la tive F r e q u e nc y Figure 4. 18 Pavement Percentages of AC Pavements in Each Pay Level according to Adjusted PI 0.0 Specifications. According to the percentage ranges calculated after adjusting the PI 0.0 limits for different pay levels, the limits of IRI-based specifications were determined for having the same number of segments for each pay level. In order to have 11% segments receiving the incentive, the lower limit of full pay range for IRI equals to 52 in/mile based on the cumulative frequency curve. The upper limit of full pay range for IRI is 72 in/mile for having 40% full-pay segments. The upper limit of penalty range is 128 in/mile to make 9% segments unacceptable. Figure 4.19 and Table 4.4 presents the result of the limits of each pay range for IRI at 0.1 mile interval. 91% 51% 11% 17 27 50 58 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 35 55 75 95 115 135 155 175 IRI, in/mile C u m u la tv ie F r e q u e nc y Figure 4. 19 Limits of Each Pay Range for IRI Table 4. 4 Converted IRI Specifications for AC Using Distribution Method. Price Adjustment of pavement Unit Bid Price Current PI 0.0 at 0.1 mile, in/mile Adjusted PI 0.0 at 0.1 mile, in/mile Percent of segments in different pay level based on adjusted PI 0.0 PI 0.0 at 0.01 mile, in/mile IRI at 0.1 mile, in/mile IRI at 0.01 mile, in/mile 105 ? (PI/20) Under 10 Under 17 11.5% Under 17 Under 51 Under 42 100 10 to 20 17 to 27 40% 17 to 32 51 to 72 42 to 68 100-(PI-20)/1.5 20 to 50 27 to 50 40% 32 to 74 72 to 128 68 to 140 Unacceptable Over 50 Over 50 8.5% Over 74 Over 128 Over 140 Moreover, the statistical relationships between 0.1 mile and 0.01 mile smoothness indices were developed during the analysis for possible use of the smaller interval for localized bump detection in further studies. 4.3.3 Effect of Material Transfer Devices (MTD) on Asphalt Pavement Smoothness One of the important purposes of smoothness specification is to encourage contractors provide better products and pursue higher payment by employing new 51 72 128 91% 51% 11% 59 technologies. Hence the payment level should be set to motivate contractors to use these technologies. During the asphalt pavement paving projects, material transfer devices, also called remixers, are proven to play an important role on decreasing the material segregation and yielding smooth pavement (Roberts et al., 1996). MTD is used between the paver and the loading truck in the construction. Because of it, the paver can process the paving at a more uniform speed with less stop. MTD also remixes the material before supplying them to the paver and decreases the segregation of the materials. In this study, the pavement smoothness data were collected from paving projects using MTD and projects without MTD. Figure 4.20 plots the distributions of pavement smoothness data at 0.1 mile interval with and without using MTD. The figure shows that MTD has a strong affect on the distribution of segments having IRI value less than 70 in/mile. Paving projects with using MTD provide 26% segments having IRI value less than 55 in/mile, but only 5% segments in paving projects without MTD have IRI value less than 55 in/mile. IRI value of 55 in/mile reveals the biggest disparity between projects with MTD and without MTD. Consequently, IRI value of 55 in/mile is a good value as incentive limit to encourage contractors to pursuit the incentive with using MTD. Projects constructed without MTD or with MTD but not using best paving practices would both have penalties assessed when the IRI is greater than 70 in/mi. Given that the cost of purchasing, using, and maintaining a MTD is high; it is to the contractors? advantage to make sure that the equipment is used properly. Alternatively, lower traffic volume roadways can have a higher initial IRI and still be considered acceptable. It is also difficult to use some of the MTD equipment in single lane paving operations, as is 60 common on two-lane roadway paving. In this case, projects that would be acceptable with an IRI of 70 in/mi would not use an MTD, which would result in a lower bid for the agency and both less capital cost and maintenance for the contractor. IRI value of 70 in/mile is a good value as the upper limits of full pay. With the proper paving practices, contractor can provide the IRI less than 70 in/mile and achieve the 100% pay, either using MTD or not. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 35 45 55 65 75 85 95 105 115 125 135 145 155 165 IRI, in/mile C u m u l at i v e F r eq u e n c y With MTD Without MTD Figure 4. 20 Effect of MTD on Pavement Smoothness at 0.1 mile interval. 4.4 IRI-based Specification The converted IRI specifications based on the above methods and analysis provides the reasonable references to determine the final IRI specifications recommendation. Currently, most of states still use 0.1 mile as the test interval, and 0.01 mile interval is just employed to further detect localized bump for some states (WFLHD, 2003). Therefore, this study recommends the IRI based specification at 0.1 mile interval; leaving 61 the 0.01 mile interval specification for bump detection for future development. From the preceding analysis, Table 4.5 provides the combined analysis results from regression method, distribution method, literature review, and effectiveness of a material transfer device. Table 4. 5 Combination of Converted IRI Specifications. Price Adjustment of Pavement Unit Bid Price by PI 0.2 at 0.1 mile interval Current PI 0.0 at 0.1 mile, in/mile IRI at 0.1mile from AC Regression, in/mi IRI at 0.1mile from PCC Regression, in/mi IRI at 0.1 mile from AC Distribution Method, in/mile IRI at 0.1 mile suggested by MTD application, in/mile 105 ? (PI/20) Under 10 Under 38 Under 44 Under 51 Under 55 100 10 to 20 38 to 60 44 to 67 51 to 72 55 to 70 100-(PI-20)/1.5 20 to 50 60 to 121 67 to 138 72 to 128 ------ Unacceptable Over 50 Over 121 Over 138 Over 128 ------ Table 4.5 indicates that transferred IRI specifications developed from both methods reach similar conclusions. Asphalt pavement and concrete pavement also have close smoothness limits after conversion. In addition, the analysis result of MTD effects, for asphalt pavement at 0.1 mile interval, that IRI of 55in/mile is suitable for incentive limit and IRI of 70 in/mile is for 100% pay upper limit, also closely follow the converted IRI specifications by other methods. Since the continuous specification is more accurate to evaluate the pavement smoothness than stepped pay specifications, continuous functions were also considered in the recommendations for an IRI-based specification. Balancing the final recommendation to account for these limitations, the final IRI specifications for asphalt pavement at 0.1 mile interval were determined in Table 4.6. 62 Table 4. 6 IRI Specification at 0.1 mile interval. Price Adjustment of Pavement Unit Bid Price by PI 0.2 at 0.1 mile interval Current ALDOT PI 0.0 at 0.1 mile, in/mile Suggested IRI at 0.1 mile interval, in/mile Price Adjustment of Pavement Unit Bid Price by IRI 105 ? (PI/20) Under 10 Under 55 112 -0.22*IRI 100 10 to 20 55 to 70 100 100-(PI-20)/1.5 20 to 50 70 to 110 121-0.37*IRI Unacceptable Over 50 Over 110 Unacceptable 4.5 Comparison of Converted IRI Specification with Other DOT?s Specifications Since some other DOTs have applied IRI in evaluating pavement roughness, current specifications from other DOT were plotted together to verify the feasibility of transferred IRI smoothness specification for Alabama. All of the DOT specifications included for comparisons in Figures 4.21 use a 0.1 mile segment interval to test pavement smoothness. Figure 4.21 shows that most of IRI full pay ranges are from 55 to 85 in/mile. The full pay range of transferred Alabama IRI-based specification is from 55 to 70 in/mile, which belongs the typical full pay range. It is also seen in Figure 4.21 that compared with other states expect Maine, the transferred Alabama IRI specifications pay less bonus for the smooth pavement with IRI value less than 55 in/mile, and make a higher penalty for the pavement roughness higher than IRI of 85 in/mile. Therefore, the transferred Alabama IRI-based specifications are within the typical pay factor function trend as other DOT?s specifications, and slightly stricter in the incentive and disincentive range. 63 72 74 76 78 80 82 84 86 88 90 92 94 96 98 100 102 104 106 108 40 50 60 70 80 90 100 110 120 IRI, in/mile P a y fa c t or, % Indiana Maine Vermont Connecticut* South Dakota Virginia Washington Alabama (Converted) Figure 4. 21 Comparison of Transferred AL IRI specifications with Specifications from other DOT?s. 64 CHAPTER FIVE CONCLUSIONS Over recent years, both the inertial profiler electronic technology and mathematical algorithms for evaluating the user?s perception of ride quality have developed rapidly. Inertial profilers can record pavement profiles at highway speed, encouraging IRI to become widely used as both an initial smoothness acceptance assessment and an evaluation of ride quality changes with time and traffic. The IRI ride quality statistic accumulates the vertical movement response of a vehicle running over a pavement surface at a given speed. This method of profiling better highlights the wavelengths that reflect the riding comfort than other smoothness indices calculating the physical deviation of pavement surface beyond certain tolerance band, such as the PI obtained from the California-style profilograph. And also, IRI is an index estimating the pavement condition from immediately after construction up to rehabilitation needs, which makes pavement management more efficient and economical. For these reasons, the Alabama Department of Transportation is considering moving from a PI 0.0 based specification to an IRI based specification. The reasonable and practical relationships between PI and IRI needed for a specification conversion were developed in this study. The current specifications use 0.1 mile segment as the test interval, which averages the roughness and makes the localized irregularities unnoticed. To address this problem, the shorter interval, 0.01 mile was utilized to analyze the pavement roughness. 65 5.1 Conclusions This study is based on 57 pavements longitudinal profiles from four Alabama asphalt pavement projects and four Quebec concrete pavement projects, measured with the ARAN inertial profiler. The raw profiles were analyzed using the ProVAL 2.5 software, which conducted the calculation of both the PI and IRI values for the each obtained raw profile. The asphalt pavement database fully covered the typical smoothness value range of newly surfaced AC pavement, and was considered as the representative of Alabama (wet and no-freeze climate zone) asphalt pavement smoothness database. Since concrete pavements examined in this study were located in Quebec (wet and freeze climate zone), this database was just used to compare the blanking band effect on AC pavement and PCC pavement, and not used in Alabama specification development. Through the analysis of the database, the following conclusions can be drawn from this research: z 0.2? blanking band hides the irregularities of pavements and has the limitation to evaluate pavement roughness, especially in short interval like 0.01 mile interval. It also shows that 0.2? blanking band has much more influence on evaluating PCC pavements than AC pavements in this study. According to the database herein, 0.2? blanking band covers much more amount of defects of rigid pavements than flexible pavements in this study, which allows rougher-driving concrete pavement to earn the same payment as asphalt pavement. z Good linear regressions (R 2 > 0.7) between PI and IRI were developed. According to the correlation analysis, the current Alabama pavement smoothness specifications were moved to the single IRI-based smoothness specifications. The IRI based 66 specifications were decided at 0.1 mile interval, presented in table 5.1. Table 5. 1 Transferred IRI based Smoothness Specifications for Asphalt and Concrete Pavement in Alabama. Price Adjustments Pavement Type Equipment Section Length IRI, in/mile Contract Price Adjustment of pavement unit bid price, % Under 55 112 -0.22*IRI 55 to 70 100 70 to 110 121-0.37*IRI AC and PCC Pavement Inertial Profiler 0.1 mile Over 110 Unacceptable z In addition, the statistical relationships between 0.1 mile and 0.01 mile smoothness indices were established in Table 4.4 for possible use of the smaller interval for investigating localized bumps in further studies. 5.2 Limitations z Smoothness data from Alabama concrete pavement need to be verified if or when new concrete pavements are constructed in Alabama. Due to dataset of concrete pavement used herein collected from Montr?al, Qu?bec, even the database falling in the typical concrete pavement smoothness value range, the different climate zone still has an obvious impact on the relationship linking IRI and PI. In order to accurately move the PI based specifications to IRI based in Alabama, correlations developed by the smoothness data from Alabama are required. z Since there is limited database in this study, in order to examine the effect of blanking band on evaluating all type of pavement smoothness, the database qualified for representing all new pavements need to be collected. z Transferring the specifications from 0.1 mile interval to 0.01 mile interval is based on distribution method, which ensures contractor earning the same pay level for 67 either index limits. The primary reason for introducing 0.01 mile interval is to motivate fewer localized bumps in an otherwise good quality pavement. Specific recommendations for use of this approach for minimizing localized bumps are beyond the scope of this current research project. 68 REFERENCES Alabama Department of Transportation, (2003). ?Measuring Profile Index of a Paved Surface?, Alabama Department of Transportation Bureau of Materials and Tests, ALDOT-335-83 Alabama Department of Transportation, (2002). ?2002 Standard Specifications for Highway Construction?, Alabama Department of Transportation, Section 410.05 American Concrete Pavement Association, (2000). http://www.pavement.com/practices/practice-a.html American Society for Testing and Materials E867, (1998). ?Terminology Relating to Vehicle-Pavement Systems?, Annual Book of ASTM Standards, Vol.04.03 Bowman, B., Ellen, B.P., III and Stroup Gardiner, M. (2003). ?Evaluation of Pavement Smoothness and Pay Factor Determination for the Alabama Department of Transportation?, ASTM STP 1433, West Conshohocken, PA Carter, A. and Stroup Gardiner, M. (2005). ?Field Noise and Texture Evaluation of Different Pavement Surface Draft, Final Report for Phase I?, Auburn University Evans, L.D., Smith, K.L., Swanlund, M.E. and et al. (2003). ?Smoothness Index Relationships for HMA Pavements?, ASTM STP 1433, West Conshohocken, PA Federal Aviation Administration, (2005). ?Computation of Profile Index from CA Profilograph Model for ProFAA?, www.airporttech.tc.faa.gov/NAPTF/Download/ 69 Fernando, E.G. (2000). ?Evaluation of Relationship between Profilograph and Profile-Based Roughness Indices?, Transportation Research Record, Transportation Research Board, National Research Council, Washington D.C. Florida Department of Transportation, (1997). ?Comparison of Ride Quality Monitoring Equipment Used for New Construction Draft Report?, Pavement Systems Evaluation Study 97-2, Florida Department of Transportation State Materials Office, Tallahassee, FL Gillespie, T.D. (1992). ?Everything You Always Wanted to Know about the IRI, but Were Afraid to Ask?, Presented in Road Profile Users Group Meeting, Lincoln, NE Hossain, M. and Parcells, W. (1995). ?Smoothness Control in Asphalt Pavement Construction: Development of Specifications, Implementation, and Results?, Transportation Research Record 1491, Transportation Research Board, National Research Council, Washington D.C., pp.40-45 Komve,E.M. and Kalevela.S.A. (1993). ?Evaluation of Initial Pavement Smoothness for the Development of PCCP Construction Specifications? Phase I Final Report, Arizona Department of Transportation, Phoenix, AZ Kulakowski, B.T. and Wambold, J.C. (1989). ?Development of Procedures of the Calibration of Profilographs? Final Report, Pennsylvania Transportation Institute, Pennsylvania State University, University Park, Pennsylvania Pellinen, K.T. and Chou, S.F., (2003). ?Evaluation of INDOT Construction Smoothness Specifications? Final Report, FHWA/IN/JTRP-2003/8 Roadware, (2005). http://www.roadware.com/products_services/aran/ ProVAL, (2005). http://www.roadprofile.com/ 70 Roberts, F.L., Kandhal, P.S., Brown, E.R. and et al. (1996). ?Hot Mix Asphalt Materials, Mixture Design, and Construction?, National Asphalt Pavement Association Research and Education Foundation, Maryland Rufino, D., Baraka, K. and Darter, M.I. (2001). ?Development of a Bridge Smoothness Specification for Illinois DOT?, FHWA Interim Report Sayers, M.W. (1995). ?On the Calculation of International Roughness Index from Longitudinal Road Profile? Transportation Research Record 1501, Transportation Research Board, National Research Council, Washington D.C., pp.1-12 Sayers, M.W. and Karamihas, S.M., (1998). ?The little Book of Profiling?, Transportation Research Institute, University of Michigan Shahin, M.Y., (1994). ?Pavement Management for Airport, Roads, and Parking Lots?, Chapman and Hall, New York Smith, K.L., Smith, K.D, Evans, L.D., Hoerner, T.E., Darter, M.I., and Woodestrom, J.H., (1997). ?Smoothness Specifications for Pavements?, NCHRP Final Report 1-31, Transportation Research Record, National Research Council, Washington, D.C. Smith, K.L., Titus-Glover, L. and Evans, L.D. (2002). ?Pavement Smoothness Index Relationships? Final Report, FHWA-RD-02-057 Stroup Gardiner, M., Carter, A. Das, P. and et al. (2005). ?An initial Ride Quality Using an Inertial Profiler?, Transportation Research Record 1907, Transportation Research Board, National Research Council, Washington D.C., pp 105-112 Western Federal Lands Highway Division, (2005). ?FP Specification Change Pavement Smoothness/Roughness?, www.wfl.fha.dot.gov/other/IRI_Spec.pdf Williams, M.B. (2003). ?The Effect of Material Transfer Devices on the Texture and 71 Temperature Differentials of Hot Mix Asphalt Concrete?, Thesis. Auburn University, AL 72 APPENDICES 73 Appendix A Regression Relationship between IRI and PIx at 0.1 and 0.01 Mile Interval 74 y = 2.3688x + 54.098 R 2 = 0.9138 0 100 200 300 400 500 600 0 20 40 60 80 100 120 Simulated PI 0.2 , in/mile I R I, i n /m i l e Figure A. 1 PI 0.2 vs. IRI for AC at 0.1 mile interval. y = 2.0708x + 17.842 R 2 = 0.9192 0 100 200 300 400 500 600 0 20 40 60 80 100 120 140 Simulated PI 0.0 , in/mile IR I, i n /m i l e Figure A. 2 PI 0.0 vs. IRI for AC at 0.1 mile interval. 75 y = 1.9295x + 62.822 R 2 = 0.7014 0 100 200 300 400 500 600 0 50 100 150 200 250 Simulated PI 0.2 , in/mile IR I , i n /m i l e y = 1.5699x + 19.909 R 2 = 0.7925 0 100 200 300 400 500 600 0 50 100 150 200 250 300 350 Simulated PI 0.0 , in/mile IRI, in /mile Figure A. 3 PI0.2 vs. IRI for AC at 0.01 mile interval. Figure A. 4 PI0.0 vs. IRI for AC at 0.01 mile interval. 76 Appendix B Histogram Distribution of PI and IRI values of AC and PCC Pavement 77 Table B. 1 Histogram Distribution of PI 0.2 Value of AC Pavement at 0.1 mile interval. Bin Frequency Cumulative Bin Frequency Cumulative 0 20 2.30% 26 15 89.87% 2 140 18.41% 28 13 91.37% 4 110 31.07% 30 3 91.71% 6 120 44.88% 32 9 92.75% 8 89 55.12% 34 7 93.56% 10 72 63.41% 36 1 93.67% 12 39 67.89% 38 4 94.13% 14 40 72.50% 40 3 94.48% 16 41 77.22% 42 4 94.94% 18 31 80.78% 44 3 95.28% 20 23 83.43% 46 4 95.74% 22 27 86.54% 48 0 95.74% 24 14 88.15% And more 37 100.00% Table B. 2 Histogram Distribution of PI 0.0 Value of AC Pavement at 0.1 mile interval. Bin Frequency Cumulative Bin Frequency Cumulative 6 1 0.11% 69 7 95.88% 10 0 0.11% 73 2 96.11% 14 23 2.75% 77 1 96.22% 18 92 13.27% 81 5 96.80% 22 126 27.69% 85 2 97.03% 26 152 45.08% 89 7 97.83% 30 158 63.16% 93 1 97.94% 34 92 73.68% 97 4 98.40% 38 49 79.29% 101 1 98.51% 42 48 84.78% 105 0 98.51% 46 37 89.02% 109 3 98.86% 50 23 91.65% 113 1 98.97% 54 8 92.56% 117 5 99.54% 58 8 93.48% 121 0 99.54% 62 10 94.62% 125 1 99.66% 65 4 95.08% And more 3 100.00% Table B. 3 Histogram Distribution of IRI Value of AC Pavement at 0.1 mile interval. Bin Frequency Cumulative Bin Frequency Cumulative 25 1 0.11% 177 1 96.22% 35 2 0.34% 187 7 97.02% 45 32 4.01% 197 6 97.71% 55 111 16.74% 207 2 97.94% 65 163 35.44% 217 2 98.17% 75 187 56.88% 227 3 98.51% 85 109 69.38% 238 5 99.08% 78 96 70 77.41% 248 0 99.08% 106 58 84.06% 258 3 99.43% 116 32 87.73% 268 1 99.54% 126 27 90.83% 278 2 99.77% 136 14 92.43% 288 0 99.77% 146 10 93.58% 299 0 99.77% 156 11 94.84% 309 1 99.89% 167 11 96.10% And more 1 100.00% Table B. 4 Histogram Distribution of PI 0.2 Value of PCC Pavement at 0.1 mile interval. Bin Frequency Cumulative Bin Frequency Cumulative 0 3 1.66% 26 5 93.37% 2 7 5.52% 28 2 94.48% 4 4 7.73% 30 0 94.48% 6 25 21.55% 32 1 95.03% 8 35 40.88% 34 1 95.58% 10 28 56.35% 36 2 96.69% 12 9 61.33% 38 2 97.79% 14 13 68.51% 40 2 98.90% 16 14 76.24% 42 0 98.90% 18 8 80.66% 44 0 98.90% 20 10 86.19% 46 1 99.45% 22 5 88.95% 48 1 100.00% 24 3 90.61% And more 0 100.00% Table B. 5 Histogram Distribution of PI 0.0 Value of PCC Pavement at 0.1 mile interval. Bin Frequency Cumulative Bin Frequency Cumulative 20 0 0.00% 72 2 97.80% 22 1 0.55% 74 0 97.80% 24 2 1.65% 76 1 98.35% 26 3 3.30% 78 0 98.35% 28 5 6.04% 80 1 98.90% 30 3 7.69% 82 1 99.45% 32 9 12.64% 84 0 99.45% 34 17 21.98% 86 0 99.45% 36 16 30.77% 88 0 99.45% 38 20 41.76% 90 0 99.45% 40 11 47.80% 92 0 99.45% 42 24 60.99% 94 0 99.45% 44 13 68.13% 96 0 99.45% 46 10 73.63% 98 0 99.45% 48 12 80.22% 100 0 99.45% 50 4 82.42% 102 0 99.45% 52 6 85.71% 104 0 99.45% 54 6 89.01% 106 0 99.45% 79 56 2 90.11% 108 0 99.45% 58 5 92.86% 110 0 99.45% 60 0 92.86% 112 0 99.45% 62 4 95.05% 114 0 99.45% 64 2 96.15% 116 0 99.45% 66 0 96.15% 118 0 99.45% 68 1 96.70% 120 0 99.45% 70 0 96.70% And more 1 100.00% Table B. 6 Histogram Distribution of IRI Value of PCC Pavement at 0.1 mile interval. Bin Frequency Cumulative 25 0 0.00% 35 0 0.00% 45 0 0.00% 55 0 0.00% 65 8 4.40% 75 14 12.09% 85 45 36.81% 96 39 58.24% 106 29 74.18% 116 19 84.62% 126 16 93.41% 136 4 95.60% 146 2 96.70% 156 3 98.35% 167 1 98.90% 177 2 100.00% And more 0 100.00% Table B. 7 Histogram Distribution of PI 0.2 Value of AC Pavement at 0.01 mile interval. Bin Frequency Cumulative Bin Frequency Cumulative 0 5300 63.61% 51 15 95.31% 1 0 63.61% 52 16 95.50% 2 0 63.61% 53 12 95.64% 3 0 63.61% 54 14 95.81% 4 278 66.95% 55 9 95.92% 5 201 69.36% 56 8 96.02% 6 124 70.85% 57 14 96.18% 7 154 72.70% 58 6 96.26% 8 102 73.92% 59 9 96.36% 9 126 75.43% 60 8 96.46% 10 93 76.55% 61 12 96.60% 11 92 77.65% 62 10 96.72% 12 77 78.58% 63 10 96.84% 80 13 89 79.64% 64 9 96.95% 14 71 80.50% 65 8 97.05% 15 80 81.46% 66 5 97.11% 16 63 82.21% 67 9 97.22% 17 53 82.85% 68 4 97.26% 18 61 83.58% 69 7 97.35% 19 62 84.33% 70 10 97.47% 20 54 84.97% 71 3 97.50% 21 51 85.59% 72 6 97.58% 22 48 86.16% 73 6 97.65% 23 50 86.76% 74 5 97.71% 24 44 87.29% 75 3 97.74% 25 40 87.77% 76 4 97.79% 26 43 88.29% 77 3 97.83% 27 30 88.65% 78 5 97.89% 28 40 89.13% 79 6 97.96% 29 38 89.58% 80 4 98.01% 30 31 89.95% 81 7 98.09% 31 33 90.35% 82 4 98.14% 32 25 90.65% 83 5 98.20% 33 27 90.97% 84 3 98.24% 34 26 91.29% 85 3 98.27% 35 33 91.68% 86 2 98.30% 36 29 92.03% 87 9 98.40% 37 21 92.28% 88 3 98.44% 38 28 92.62% 89 2 98.46% 39 18 92.83% 90 3 98.50% 40 16 93.03% 91 3 98.54% 41 18 93.24% 92 4 98.58% 42 17 93.45% 93 3 98.62% 43 19 93.67% 94 1 98.63% 44 12 93.82% 95 7 98.72% 45 16 94.01% 96 6 98.79% 46 24 94.30% 97 4 98.84% 47 22 94.56% 98 0 98.84% 48 19 94.79% 99 1 98.85% 49 13 94.95% 100 1 98.86% 50 15 95.13% And more 95 100.00% Table B. 8 Histogram Distribution of PI 0.0 Value of AC Pavement at 0.01 mile interval. Bin Frequency Cumulative Bin Frequency Cumulative 0 48 0.58% 130 15 98.28% 4 18 0.79% 133 11 98.42% 7 63 1.55% 137 14 98.58% 11 201 3.96% 140 6 98.66% 81 14 316 7.75% 144 11 98.79% 18 521 14.01% 147 2 98.81% 21 650 21.81% 151 8 98.91% 25 820 31.65% 154 4 98.96% 28 808 41.35% 158 8 99.05% 32 705 49.81% 161 5 99.11% 35 665 57.79% 165 3 99.15% 39 538 64.25% 169 9 99.26% 42 484 70.06% 172 5 99.32% 46 377 74.58% 176 7 99.40% 49 293 78.10% 179 6 99.47% 53 251 81.11% 183 5 99.53% 56 192 83.41% 186 4 99.58% 60 175 85.51% 190 5 99.64% 63 155 87.37% 193 4 99.69% 67 135 88.99% 197 2 99.71% 70 110 90.31% 200 1 99.72% 74 95 91.45% 204 3 99.76% 77 81 92.43% 207 3 99.80% 81 51 93.04% 211 0 99.80% 84 67 93.84% 214 1 99.81% 88 56 94.52% 218 3 99.84% 91 50 95.12% 221 3 99.88% 95 47 95.68% 225 1 99.89% 98 30 96.04% 228 0 99.89% 102 29 96.39% 232 0 99.89% 105 33 96.78% 235 0 99.89% 109 17 96.99% 239 0 99.89% 112 29 97.34% 242 2 99.92% 116 17 97.54% 246 0 99.92% 119 18 97.76% 249 2 99.94% 123 16 97.95% 253 1 99.95% 126 13 98.10% And more 4 100.00% Table B. 9 Histogram Distribution of IRI Value of AC Pavement at 0.01 mile interval. Bin Frequency Cumulative Bin Frequency Cumulative 25 38 0.46% 175 58 95.40% 35 325 4.36% 185 48 95.98% 45 944 15.68% 195 47 96.54% 55 1290 31.17% 205 50 97.14% 65 1271 46.42% 215 38 97.60% 75 1074 59.31% 225 34 98.01% 85 827 69.23% 235 18 98.22% 95 585 76.25% 245 13 98.38% 105 393 80.97% 255 13 98.54% 115 281 84.34% 265 10 98.66% 82 125 271 87.59% 275 9 98.76% 135 199 89.98% 285 18 98.98% 145 196 92.33% 295 16 99.17% 155 118 93.75% 305 11 99.30% 165 80 94.71% And more 58 100.00% Table B. 10 Histogram Distribution of PI 0.2 Value of PCC Pavement at 0.01 mile interval. Bin Frequency Cumulative Bin Frequency Cumulative 0 842 47.84% 26 7 87.27% 1 0 47.84% 27 8 87.73% 2 6 48.18% 28 2 87.84% 3 9 48.69% 29 6 88.18% 4 63 52.27% 30 14 88.98% 5 69 56.19% 31 10 89.55% 6 56 59.38% 32 5 89.83% 7 40 61.65% 33 16 90.74% 8 46 64.26% 34 11 91.36% 9 42 66.65% 35 8 91.82% 10 25 68.07% 36 9 92.33% 11 32 69.89% 37 9 92.84% 12 37 71.99% 38 7 93.24% 13 27 73.52% 39 4 93.47% 14 23 74.83% 40 10 94.03% 15 24 76.19% 41 5 94.32% 16 29 77.84% 42 7 94.72% 17 19 78.92% 43 4 94.94% 18 24 80.28% 44 2 95.06% 19 13 81.02% 45 6 95.40% 20 19 82.10% 46 2 95.51% 21 22 83.35% 47 3 95.68% 22 15 84.20% 48 3 95.85% 23 16 85.11% 49 3 96.02% 24 16 86.02% 50 1 96.08% 25 15 86.88% And more 69 100.00% Table B. 11 Histogram Distribution of PI 0.0 Value of PCC Pavement at 0.01 mile interval. Bin Frequency Cumulative Bin Frequency Cumulative 10 2 0.11% 58 49 76.14% 12 2 0.23% 60 39 78.35% 14 2 0.34% 62 41 80.68% 16 11 0.97% 64 40 82.95% 83 18 15 1.82% 66 29 84.60% 20 31 3.58% 68 30 86.31% 22 24 4.94% 70 25 87.73% 24 45 7.50% 72 22 88.98% 26 48 10.23% 74 28 90.57% 28 62 13.75% 76 17 91.53% 30 68 17.61% 78 18 92.56% 32 74 21.82% 80 15 93.41% 34 74 26.02% 82 12 94.09% 36 76 30.34% 84 15 94.94% 38 90 35.45% 86 11 95.57% 40 95 40.85% 88 6 95.91% 42 65 44.55% 90 6 96.25% 44 84 49.32% 92 4 96.48% 46 99 54.94% 94 4 96.70% 48 77 59.32% 96 5 96.99% 50 66 63.07% 98 4 97.22% 52 72 67.16% 100 2 97.33% 54 56 70.34% And more 47 100.00% 56 53 73.35% Table B. 12 Histogram Distribution of IRI Value of PCC Pavement at 0.01 mile interval. Bin Frequency Cumulative Bin Frequency Cumulative 34 1 0.06% 137 78 89.77% 45 24 1.42% 149 63 93.35% 57 134 9.03% 160 27 94.89% 68 224 21.76% 172 18 95.91% 80 298 38.69% 183 20 97.05% 91 298 55.63% 195 13 97.78% 103 244 69.49% 207 13 98.52% 114 161 78.64% And more 26 100.00% 126 118 85.34%