U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
2023664000
Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations
This report is an archived publication and may contain dated technical, contact, and link information 

Publication Number: FHWARD02057 Date: OCTOBER 2002 
PDF files can be viewed with the Acrobat® Reader®
A key factor in the longterm performance of both asphalt and portland cement concrete pavements is initial pavement smoothness. In general, the smoother a pavement is built, the smoother it stays over time, resulting in lower maintenance costs and more comfort and safety for the traveling public. State highway agencies recognized in the 1960s the importance of controlling initial pavement smoothness, and began developing and implementing smoothness specifications. As the technology and equipment for measuring pavement smoothness advanced, two predominant methods emerged.
The profilograph is widely used to measure and control initial smoothness by producing profile traces, which can be evaluated to identify severe bumps and to establish an easily understood, overall measure of smoothness, the profile index (PI). However, concerns about the accuracy of the profilograph have grown significantly in the last decade. The more recently developed inertial profiler is used to quickly and accurately monitor inservice pavements, and produces a more definitive profile of a pavement from which the widely accepted International Roughness Index (IRI) can be computed. Use of inertial profilers has remained limited in initial construction acceptance testing due to their higher cost and constraints on timeliness of testing. Thus, in many agencies, initial pavement smoothness has been measured one way (profilograph PI) and smoothness over time has been measured another way (inertial profiler IRI).
Despite efforts to make adjustments for more accuracy in the computation of PI, it is evident that IRI will become the statistic of choice in future smoothness specifications. So how do agencies make the switch from their current PIbased specifications to IRI specifications? This study attempts to provide answers through the analysis of comprehensive time history smoothness data collected by highspeed inertial profilers under the LongTerm Pavement Performance (LTPP) program. Using advanced computer simulation algorithms, it is possible to compute PI values from surface profile data, thereby allowing detailed comparisons between IRI and PI.
T. Paul Teng, P.E.
Director, Office of Infrastructure Research and Development
Notice
This document is disseminated under the sponsorship of the Department of Transportation in the interest of information exchange. The U.S. Government assumes no liability for its contents or use thereof. This report does not constitute a standard, specification, or regulation.
The
U.S. Government does not endorse products or manufacturers. Trademarks or manufacturers' names appear in this report only because they are considered essential to the objective of the document.
Technical Report Documentation Page
1. Report No.
FHWARD02057 
2. Government Accession No.  3 Recipient's Catalog No.  
4. Title and Subtitle Pavement Smoothness Index Relationships, Final Report 
5. Report Date OCTOBER 2002 

6. Performing Organization Code C6B 

7. Author(s)
Kelly L. Smith, Leslie TitusGlover, Lynn D. Evans 
8. Performing Organization Report No.


9. Performing Organization Name and Address ERES Division of Applied Research Associates, Inc., 505 W. University Avenue, Champaign, IL 61820. 
10. Work Unit No. (TRAIS) 

11. Contract or Grant No. GS10F0298K 

12. Sponsoring Agency Name and Address
Office of Pavement Technology, Federal Highway Administration, 400 Seventh St., S.W., Washington, D.C. 20590 
13. Type of Report and Period Covered
Final Report, February  November 2001 

14. Sponsoring Agency Code


15. Supplementary Notes FHWA Contracting Officer's Technical Representative (COTR): Mark Swanlund, HIPT1 

16. Abstract
Nearly all State highway agencies use smoothness specifications to ensure that hotmix asphalt (HMA) and Portland cement concrete (PCC) pavements are built to high levels of smoothness. Not only is an initially smooth pavement generally indicative of quality workmanship, but it has been shown to last longer than a pavement built rougher. About half of all current State smoothness specifications for HMA and more than threefourths of all current PCC smoothness specifications are centered around the Profile Index (PI), as often measured using a profilograph. The vast majority of these specifications utilize a 5mm (0.2inch) blanking band in computing PI (i.e., PI_{5mm}). Unfortunately, because of the technical limitations of the profilograph equipment and PI computation procedures, the adequacy of PI_{5mm} in characterizing roughness and having it relate to user response has come into question. The International Roughness Index (IRI) or the Profile Index using a 0.0mm blanking band (PI_{0.0}) seem to provide better measures of smoothness and better correlation with user response. However, one barrier to more widespread implementation of these new smoothness standards is the lack of objective, verifiable correlation methods for use in establishing specification limits using the IRI or PI_{0.0}. Assistance in selecting appropriate IRI and PI_{0.0} specification limits is needed to provide a basis for modifying current specifications to these more reproducible and portable smoothness indices. This research effort has developed a series of relationships between IRI and PI that can assist States in transitioning to an IRI or PI_{0.0} smoothness specification for HMA and PCC pavements. 

17. Key Words
LTPP, pavement smoothness, profilograph, inertial profiler, international roughness index, profile index, smoothness limits. 
18. Distribution Statement
No restrictions. This document is available to the public through the National Technical Information Service, Springfield, VA 22161. 

19. Security Classification Unclassified 
20. Security Classification Unclassified 
21. No. of Pages 116 
22. Price 
Form DOT F 1700.7  Reproduction of completed page authorized 
Technical Report Documentation Page
Approximate Conversions to SI Units
Length:
inches (in) multiply by 25.4 to get millimeters (mm)
feet (ft) multiply by 0.305 to get meters (m)
yards (yd) multiply by 0.914 to get meters (m)
miles (mi) multiply by 1.61 to get kilometers (km)
Area:
square inches (in^{2}) multiply by 645.2 to get square millimeters (mm^{2})
square feet (ft^{2}) multiply by 0.093 to get square meters (m^{2})
square yard (yd^{2}) multiply by 0.836 to get square meters (m^{2})
acres (ac) multiply by 0.405 to get hectares (ha)
square miles (mi^{2}) multiply by 2.59 to get square kilometers (km^{2})
Volume:
fluid ounces (fl oz) multiply by 29.57 to get milliliters (mL)
gallons (gal) multiply by 3.785 to get liters (L)
cubic feet (ft^{3}) multiply by 0.028 to get cubic meters (m^{3})
cubic yards (yd^{3}) multiply by 0.765 to get cubic meters (m^{3})
NOTE: volumes greater than 1000 L shall be shown in m^{3}
Mass:
ounces (oz) multiply by 28.35 to get grams (g)
pounds (lb) multiply by 0.454 to get kilograms (kg)
short tons  2000 lb (T) multiply by 0.907 to get megagrams or "metric ton" (Mg or "t")
Temperature (exact degrees):
Fahrenheit (°F) multiply by 5 (F32)/9 or (F32)/1.8 to get Celsius (°C)
Illumination:
footcandles (fc) multiply by 10.76 to get lux (lx)
footLamberts (fl) multiply by 3.426 to get candela/m^{2} (cd/m^{2})
Force and Pressure or Stress:
poundforce (lbf) multiply by 4.45 to get newtons (N)
poundforce per square inch (lbf/in^{2}) multiply by 6.89 to get kilopascals (kPa)
Approximate Conversions From SI Units
Length:
millimeters (mm) multiply by 0.039 to get inches (in)
meters (m) multiply by 3.28 to get feet (ft)
meters (m) multiply by 1.09 to get yards (yd)
kilometers (km) multiply by 0.621 to get miles (mi)
Area:
square millimeters (mm^{2}) multiply by 0.0016 to get square inches (in^{2})
square meters (m^{2}) multiply by 10.764 to get square feet (ft^{2})
square meters (m^{2}) multiply by 1.195 to get square yards (yd^{2})
hectares (ha) multiply by 2.47 to get acres (ac)
square kilometers (km^{2}) multiply by 0.386 to get square miles (mi^{2})
Volume:
milliliters (mL) multiply by 0.034 to get fluid ounces (fl oz)
liters (L) multiply by 0.264 to get gallons (gal)
cubic meters (m^{3}) multiply by 35.314 to get cubic feet (ft^{3})
cubic meters (m^{3}) multiply by 1.307 to get cubic yards (yd^{3})
Mass:
grams (g) multiply by 0.035 to get ounces (oz)
kilograms (kg) multiply by 2.202 to get pounds (lb)
megagrams or "metric ton" (Mg or "t") multiply by 1.103 to get short tons  2000 lb (T)
Temperature (exact degrees):
Celsius (°C) multiply by 1.8C+32 to get Fahrenheit (°F)
Illumination:
lux (lx) multiply by 0.0929 to get footcandles (fc)
candela/m^{2} (cd/m^{2}) multiply by 0.2919 to get footLamberts (fl)
Force and Pressure or Stress:
newtons (N) multiply by 0.225 to get poundforce (lbf)
kilopascals (kPa) multiply by 0.145 to get poundforce per square inch (lbf/in^{2})
*SI is the symbol for the International System of Units. Appropriate rounding should be made to comply with Section 4 of ASTM E380.
(Revised March 2002)
BackgroundChapter 2. Literature Review
Problem Statement
Study Objectives
IntroductionChapter 3. LTPP Data Collection and Project Database Development
Past Studies on PIIRI Relationships
Summary
IntroductionChapter 4. Development of LTPPBased Smoothness Index Relationships
Collection of LTPP Profile Data
Conversion of Profile Data to Simulated PI Values
Populating the Project Database
IntroductionChapter 5. Adaptation of LTPPBased Models to Current State Smoothness Specifications
Step 1  Preliminary Evaluation
Step 2  Selection of Appropriate Model Form
Step 3  Group Data into Sets with Similar Smoothness Relations
Steps 4 and 5  Develop Tentative Models and Assess Models for Reasonableness
Step 6  Select Final Models
IntroductionChapter 6. Conclusions and Recommendations
Overview of State Smoothness Specifications
Development of Recommended Initial IRI and PI_{0.0} Levels
ConclusionsAppendix A: IRI and PI Relationships for AC Pavements
Recommendations
Appendix B: IRI and PI Relationships for PCC Pavements
1. Sensitivity of simulated profilograph to spatial frequency.
2. Relationship between IRI and manually generated PI in PTI profilograph calibration study.
3. Relationship between IRI and computergenerated PI in PTI profilograph calibration study.
4. Correlation of IRI and PI in Arizona pavement smoothness study.
5. IRIPI_{5mm} (PI_{0.2inch}) correlations established in Florida's ride quality equipment study.
7. Relationship between IRI and computersimulated PI values in TTI equipment comparison study.
11. Graphical comparison of documented PI_{5mm}IRI smoothness relationships.
12. Graphical comparison of documented PI_{2.5mm}IRI smoothness relationships.
13. Graphical comparison of documented PI_{0.0}IRI smoothness relationships.
14. Flow chart for developing pavement smoothness models.
15. Histogram showing the distribution of IRI data used in model development (all AC pavements).
16. Histogram showing the distribution of IRI data used in model development (all PCC pavements).
19. IRI vs. PI_{0.0} for all AC pavements and climates.
20. PI_{0.0} vs. PI_{5mm} for all PCC pavements and climates.
21. IRI vs. PI_{0.0} by AC pavement type for all climates.
22. PI_{0.0} vs. PI_{5mm} by climate for all PCC pavement types.
23. Graphical comparison of PI_{5mm}IRI smoothness relationships for AC pavements.
24. Graphical comparison of PI_{5mm}IRI smoothness relationships for PCC pavements.
25. Graphical comparison of PI_{2.5mm}IRI smoothness relationships for AC pavements.
26. Graphical comparison of PI_{0.0}IRI smoothness relationships for AC pavements.
27. Graphical comparison of PI_{0.0}IRI smoothness relationships for PCC pavements.
28. Conceptual plot showing relationships of smoothness indices within and between cells.
1. Summary of documented PIIRI relationships.
2. Breakdown of test sections by LTPP experiment.
3. Breakdown of test sections by State.
4. Summary of basic statistics of data used in model development.
5. Matrix of scatter plots created for model development.
6. Factorial of cells used for model development.
7. ANOVA results on the effect of pavement type and climate on PIIRI relationship for AC pavements.
9. Summary of groupings (merged cells) used for model development.
10. PItoIRI index conversion equations and variability indices for AC pavements.
11. PItoPI index conversion equations and variability indices for AC pavements.
12. PItoIRI index conversion equations and variability indices for PCC pavements.
13. PItoPI index conversion equations and variability indices for PCC pavements.
14. State agency smoothness specifications for asphalt pavements.
15. State agency smoothness specifications for concrete pavements.
Background
Initial pavement smoothness is a key factor in the performance and economics of a pavement facility. All other things being equal, the smoother a pavement is built, the smoother it will stay over time. The smoother it stays over time, the longer it will serve the traveling public, thereby benefiting the public in terms of investment (initial construction and upkeep) and vehicular wear costs, as well as comfort and safety.
As a means of controlling initial pavement smoothness, several highway agencies began developing and implementing smoothness specifications in the late 1950s and 1960s. These specifications generally included straightedge testing and a form of ride quality testing using responsetype mechanical equipment, such as the Bureau of Public Roads (BPR) Roughometer, the Mays Ridemeter, and the Portland Cement Association (PCA) Ridemeter, or simple profiling devices, such as the Chloe profilometer and the profilograph.
Between the late 1960s and the 1980s, the profilograph emerged as the clear choice among highway agencies for measuring and controlling initial smoothness, particularly for concrete pavements. This 7.6meters (25feet) rolling reference system is capable of producing profile traces, which can be evaluated to identify severe bumps and to establish an overall measure of smoothness (i.e., the profile index [PI]).
During this same period of time, more complex profiling systems were being developed and marketed, which provided a much quicker assessment and more accurate representation of pavement smoothness. Inertial profilometers or profilers consist of an integrated set of vertical displacement sensors, vertical accelerometers, and analog computer equipment mounted in a fullsized vehicle (usually a van or large automobile) equipped with a distancemeasuring instrument (DMI). These pieces of equipment, which can be operated at highway speed, are capable of producing a more definitive profile of a pavement, from which the universally accepted International Roughness Index (IRI) can be computed.
Inertial profilers' first major role in the pavements realm involved longterm condition monitoring of inservice pavements. The reliability and repeatability of these devices greatly enhanced the quality of the pavement management data used by highway agencies in programming maintenance and rehabilitation (M&R) activities. Although the use of inertial profilers in condition monitoring increased substantially in the 1980s and early 1990s, their application in construction acceptance testing remained limited due to their high cost and constraints on the timeliness of testing (i.e., tests on rigid pavements could not be performed until after a few days of curing). Thus, in many agencies, initial pavement smoothness has been measured one way (profilograph PI) and smoothness over time has been measured another way (inertial profiler IRI).
In recent years, the technology of inertial profiling systems present on fullsized vehicles has been adopted on smaller motorized vehicles, such as the John Deere and Kawasaki utility carts and fourwheel allterrain vehicles (ATVs). These lightweight profilers, which are currently being evaluated by several agencies and have been approved for use by a few, enable testing personnel to obtain timely and highly definitive measurements of surface profiles at rates of speed significantly higher than profilographs (24 kilometers/hour [15 miles/hour] versus 5 km/hr [3 mi/hr]). The profilers are capable of producing IRI and other indices (e.g., simulated PI and Mays output, ride number [RN]) commonly used in controlling and monitoring pavement smoothness.
Problem Statement
Although the profilograph has served the highway community fairly well as an easily understood index of initial pavement smoothness, concerns about its accuracy and relationship with user response (fair to poor) have grown significantly in the last decade. For instance, because the device measures only wavelengths within the range of 0.3 to 23 m (1 to 75 ft) and because it amplifies wavelengths that are factors of its length (i.e., 7.6 m [25 ft]), the profile it produces is biased from a pavement's true profile. This can be seen in figure 1, where a true profile would be represented by a gain of 1.0. Coupled with the fact that a 2.5 or 5millimeter (0.1 or 0.2inch) blanking band is often applied when computing PI, thereby masking some roughness, it is understandable how correlation with user response is generally deemed inadequate.
Over the last 6 years, a handful of State agencies have moved toward using a zero blanking band PI (PI_{0.0}) statistic for construction acceptance testing. This has reportedly improved the ability to control initial smoothness and bettered the relationship between profilograph PI and user response. However, the fact that the same biased profiles are being used to compute PI_{0.0} does not fully alleviate the major concerns with the profilograph. Among many agencies, the belief persists that inertial profilers are the best means for specifying and evaluating initial smoothness.
Additional support for using inertial profilers in construction acceptance testing comes from the desire for a "cradletograve" smoothness index. Since it has been shown that future smoothness is a function of initial smoothness, use of one index for tracking smoothness over the entire life of a pavement would significantly benefit pavement managers and designers through improved performance prediction modeling.
Recent surveys of State highway agencies indicate that about 10 percent (4 of 34 respondents) use IRI to control initial smoothness (Baus and Hong, 1999), while about 84 percent (31 of 37 respondents) use IRI to monitor pavement smoothness over time (Ksaibati et al., 1999).). It is quite evident that IRI will become the statistic of choice in future smoothness specifications, given that: many agencies are investigating lightweight inertial profilers, and that the proposed 2002 Design Guide under development by the National Cooperative Highway Research Program (NCHRP) will include IRI prediction models that are a function of initial IRI (IRI_{0}).
So, how do agencies make the switch from their current PIbased specifications to IRI specifications? What levels of IRI should be specified which would be comparable or equivalent to the PI values currently stipulated? How confident can an agency be that newly established IRI levels reflect the levels of ride quality previously specified? These are all questions that must be properly addressed in light of the fact that several past pavement smoothness studies show poor correlation between PI values produced by a profilograph and IRI values generated by inertial profilers.
This study attempts to provide answers to the above questions through the analysis of comprehensive time history smoothness data collected by highspeed inertial profilers under the LongTerm Pavement Performance (LTPP) program. These smoothness data include archived surface profile data and corresponding computed IRI values for many General Pavement Studies (GPS) and Specific Pavement Studies (SPS) test pavements located throughout the United States. Using advanced computer simulation algorithms, it is possible to compute PI values from the surface profile data, thereby allowing detailed comparisons between IRI and PI.
Study Objectives
The specific objectives of the study include the following:
Introduction
To begin the investigation of the relationship between IRI and PI, a fairly extensive literature search was performed focusing on national and Statesponsored pavement smoothness studies conducted in the last 15 years. This search resulted in the collection of many reports, papers, and articles on the topic of smoothness, but only a handful dealing specifically with the correlation of IRI and PI.
Presented in this section is a synopsis of seven documented studies and the PItoIRI correlations developed in those efforts. Most of the correlations involve PI readings from actual profilograph equipment; however, a few are based on computersimulated PI values produced from surface profiles measured by inertial profilers.
Past Studies on PIIRI Relationships
Pennsylvania Transportation Institute Profilograph Calibration Study
As part of a major effort to develop calibration procedures for profilographs and evaluate equipment for measuring the smoothness of new pavement surfaces, the Pennsylvania Transportation Institute (PTI) conducted a fullscale fieldtesting program on behalf of the Federal Highway Administration (FHWA) (Kulakowski and Wambold, 1989). Concrete and asphalt pavements at five different locations throughout Pennsylvania were selected for the experiment; each pavement was new or newly surfaced. Multiple 0.16km (0.1mi) long pavement sections were established at each location, resulting in 26 individual test sections over which 2 different types of profilographs (California and Rainhart), a Mays Meter, and an inertial profiler were operated. The resulting smoothness measurements were evaluated for correlation.
Figure 2 shows the relationship between the inertial profiler IRI and the PI_{5mm} (PI_{0.2inch}) determined manually from the Californiatype profilograph. As can be seen, the resulting linear regression equation had a coefficient of determination (R^{2}) of 0.57. Figure 3 shows the relationship between the inertial profiler IRI and the computergenerated PI_{5mm} (PI_{0.2inch}) from the Californiatype profilograph. Although the resulting linear regression equation had a similar coefficient of determination (R^{2} = 0.58), its slope was considerably flatter. For any given IRI, the data show a wide range of PI_{5mm} (PI_{0.2inch}).
Although both of these relationships were based on measurements from both concrete and asphalt pavement sections, neither one is considerably different from regressions based solely on data from the concrete sections.
Arizona DOT Initial Smoothness Study
In 1992, the Arizona Department of Transportation (AZDOT) initiated a study to determine the feasibility of including their K.J. Law 690 DNC Profilometer (opticalbased inertial profiler) as one of the principal smoothness measuring devices for measuring initial pavement smoothness on PCC pavements (Kombe and Kalevela, 1993). At the time, the AZDOT used a Cox Californiatype profilograph to test newly constructed PCC pavements for compliance with construction smoothness standards.
To examine the correlative strength of the Profilometer (IRI) and profilograph (PI) outputs, a group of twelve 0.16km (0.1mi) pavement sections around the Phoenix area were selected for testing. The smoothness levels of the sections spanned a range that is typical of newly built concrete pavementPI_{5mm} (PI_{0.2inch}) between 0 and 0.24 m/km (15 inches per mile). A total of three smoothness measurements were made with the Profilometer over each wheelpath of each selected section, whereas a total of five measurements were made by the profilograph over each wheelpath of each section. The mean values of each set of three or five measurements were then used to correlate the IRI and PI_{5mm} (PI_{0.2inch}) values.
Simple linear regression analyses performed between the left wheelpath, right wheelpath, and both wheelpath sets of values indicated generally good correlation between the two indexes. Figure 4 shows the scatter plots of each group, as well as the regression line associated with the both wheelpath data group. As can be seen, the R^{2} for the both wheelpath regression line was very high (0.93).
University of Texas Smoothness Specification Study
In the course of developing new smoothness specifications for rigid and flexible pavements in Texas, researchers at the University of Texas conducted a detailed field investigation comparing the McCracken Californiatype profilograph and the Face Dipstick, a manual Class I profile measurement device (Scofield, 1993). The two devices were used to collect smoothness measurements on 18 sections of roadway consisting of both asphalt and concrete pavements. For both devices, only one test per wheelpath was performed.
Results of linear regression analysis showed a strong correlation (R^{2} = 0.92) between the IRI and PI_{5mm} (PI_{0.2inch}) values. The resulting linear regression equation had a higher intercept value than those obtained in the PTI and AZDOT studies, while the slope of the equation was more in line with the slopes generated in the PTI study.
Florida DOT Ride Quality Equipment Comparison Study
Looking to upgrade its smoothness testing and acceptance process for flexible pavements, the Florida DOT (FLDOT) undertook a study designed to compare its current testing method (rolling straightedge) with other available methods, including the California profilograph and the highspeed inertial profiler (FLDOT, 1997). A total of twelve 0.81km (0.5mi) long pavement sections located on various Florida State highways were chosen for testing. All but one of the sections represented newly constructed or resurfaced asphalt pavements.
The left and right wheelpaths of each test section were measured for smoothness by each piece of equipment. The resulting smoothness values associated with each wheelpath were then averaged, yielding the values to be used for comparing the different pieces of equipment. The inertial profiler used in the study was a model manufactured by the International Cybernetics Corporation (ICC). Because one of the objectives of the study was to evaluate different technologies, the ICC inertial profiler was equipped with both laser and ultrasonic sensors. Separate runs were made with each sensor type, producing two sets of IRI data for comparison.
Figure 5 shows the relationships developed between the profilograph PI_{5mm} (PI_{0.2inch}) and the IRI values respectively derived from the laser and ultrasonic sensors. As can be seen, both correlations were fairly strong (R^{2} values of 0.88 and 0.67), and the linear regression equations were somewhat similar in terms of slope. As is often the case, however, the ultrasonicbased smoothness measurements were consistently higher than the laserbased measurements, due to the added sensitivity to items such as surface texture, cracking, and temperature. This resulted in a higher yintercept for the ultrasonicbased system.
Figure 6 shows the correlations developed between IRI and PI_{2.5mm} (PI_{0.1inch}) and IRI and PI_{0.0}. It is quite clear from this and the previous figure that the application of smaller blanking bands results in higher PI values, since additional components of roughness are considered. More significant, however, is the fact that both the slopes and the yintercept values in the resulting linear regression equations decrease with smaller blanking bands. This is, again, the result of additional profile roughness being considered.
It is reasonable to surmise from these observations that, if the PI_{0.0} was computed from a more accurate pavement profile than the one generated by a profilograph, the yintercept would be much closer to zero. This is because the roughness associated with long wavelengths (e.g., long dips or humps) is automatically filtered out as a result of the short baselength of profilographs.Texas Transportation Institute Smoothness Testing Equipment Comparison Study
As part of a multistaged effort to transition from a profilographbased smoothness specification to a profilebased specification, the Texas Transportation Institute (TTI) was commissioned by the Texas DOT (TXDOT) in 1996 to evaluate the relationship between IRI and profilograph PI (Fernando, 2000). The study entailed obtaining longitudinal surface profiles (generated by one of the Department's highspeed inertial profiler) from 48 newly AC resurfaced pavement sections throughout Texas, generating computersimulated profilograph traces from those profiles using a fieldverified kinematic simulation model, and computing PI_{5mm} (PI_{0.2inch}) and PI_{0.0} values using the ProScan computer software.
A total of three simulated runs per wheelpath per section were performed, from which an average PI value for each section was computed. The resulting section PI values were then compared with the corresponding section IRI values, which had been computed by the inertial profiling system at the time the longitudinal surface profiles were produced in the field. Since both the PI and IRI values were based on the same longitudinal profiles, potential errors due to differences in wheelpath tracking were eliminated.
Illustrated in figure 7 are the relationships between the IRI and the simulated PI response parameters. As can be seen, a much stronger trend was found to exist between IRI and PI_{0.0} than between IRI and PI_{5mm} (PI_{0.2inch}). Again, this is not unexpected since the application of a blanking band has the natural effect of masking certain components of roughness. In comparison with the other IRIPI_{5mm} (IRIPI_{0.2inch}) correlations previously presented, the one developed in this study is quite typical. The linear regression equation includes a slightly higher slope but a comparable yintercept value.
Kansas DOT Lightweight Profilometer Performance Study
The major objective of this 1999/2000 study was to compare asconstructed smoothness measurements of concrete pavements taken by the Kansas DOT's (KDOT) manual Californiatype profilograph, four lightweight inertial profilers (Ames Lightweight Inertial Surface Analyzer [LISA], K.J. Law T6400, ICC Lightweight, and Surface Systems Inc. [SSI] Lightweight), and two fullsized inertial profilers (Kansas DOT South Dakotatype profiler, K.J. Law T6600) (Hossain et al., 2000). The simulated PI_{0.0} values produced by the various lightweight systems were statistically compared with the Californiatype profilograph PI_{0.0} readings to determine the acceptability of using lightweight systems to control initial pavement smoothness. In addition, IRI values generated by the lightweight systems were statistically compared with those generated by the fullsized, highspeed profilers to investigate whether the IRI statistic can be used as a "cradletograve" statistic for road roughness.
The field evaluation was performed at eight sites along I70 west of Topeka. Each lane (driving and passing) at each site was tested with the KDOT's profilograph and fullsized profiler, while the remaining profilers tested at only some of the eight sites. At a given site, one run of each wheelpath was made with the profilograph, and the average of the two runs was determined and reported. For the lightweight and fullsized profilers, three and five runs were made, respectively, with both wheelpaths measured and averaged during each run.
Statistical analysis of the data indicated that the lightweight systems tended to produce statistically similar PI_{0.0} values when compared to the KDOT manual profilograph. It also showed similarities in IRI between the KDOT fullsized profiler and three of the four lightweight profilers, giving some credence to the "cradletograve" roughness concept.
The study included correlation analysis between the PIs from the manual profilograph and those from the lightweight systems. It also included correlation analysis between the simulated PI and IRI values produced by each inertial profiler. Plots of these data are provided in figure 8, which also shows the linear IRIPI_{0.0} relationship that results when data from all profiling devices are considered.
No correlations were made in the KDOT study between profilograph PI_{0.0} and inertial profiler IRI. However, using data from the report, several such trends have now been developed and are illustrated in figure in 9. Each data point in this figure represents the mean smoothness (profilograph PI_{0.0} and profiler IRI) of one lane at one test site. As can be seen, only the IRI data taken by two of the lightweight profilers (Ames LISA and ICC) and the KDOT fullsized profiler are represented. The other three profilers collected data from only two of the eight sites, which resulted in very limited data sets.
Illinois DOT Bridge Smoothness Specification Development Study
As part of an effort to develop a preliminary bridge smoothness specification for the Illinois DOT (ILDOT), the University of Illinois coordinated a series of bridge smoothness tests in 1999 using the K.J. Law T6400 lightweight inertial profiler (Rufino et al., 2001). A total of 20 bridges in the Springfield, Illinois area were chosen and tested, with each bridge measured for IRI and PI_{5mm} (PI_{0.2inch}). At least one run per wheelpath of the driving lane was made, and each run extended from the front approach pavement across the bridge deck to the rear approach pavement.
A correlation analysis of the IRI and simulated PI_{5mm} (PI_{0.2inch}) values produced by the lightweight profiler was performed in the study, which resulted in the graph and linear relationship given in figure 10. Unlike other relationships presented earlier in this chapter, this relationship covers a larger spectrum of PI values  PI_{5mm} (PI_{0.2inch}) values largely in the range of 0.4 to 1.0 m/km (25 to 63 inches per mile)  due to the fact that bridges are often much rougher than pavements.
Summary
Table 1 summarizes the various regression equations found in the literature relating IRI from an inertial profiling system with PI statistics (PI_{5mm}, PI_{2.5mm}, and PI_{0.0}) generated by Californiatype profilographs or simulated by inertial profilers. How these various relationships compare visually with one another can be seen in figures 11 through 13. Generally speaking, there is considerable disparity in the vertical positioning of each trend, but the slopes are rather similar. The fact that different pavement types, different roughness ranges, and different pieces of testing equipment are represented by the various trends is believed to account in large part for the disparities observed.
Introduction
As mentioned previously, the main thrust of this study involves the comprehensive analysis of LTPP smoothness data. Since the time the LTPP program was initiated in 1989, several hundred test pavements throughout the country have been tested for smoothness on an annual or biennial basis using fullsized, highspeed inertial profilers. In each test, the longitudinal surface profile of each wheelpatch was measured and recorded, and from those profiles the IRI of each wheelpath was computed and recorded for inclusion in the LTPP Information Management System (IMS) database. The sections below describe in detail the collection of LTPP data and the development of the project database used to examine the relationship between IRI and PI.
Collection of LTPP Profile Data
To retrieve the profile and smoothness data required for this study, a data request was submitted to the LTPP IMS database manager. All 1996  2001 archived profile data contained in the Ancillary Information Management System (AIMS) and IRI data contained in the IMS were requested, covering all LTPP test sections. Data for this time period only were requested, as they represented data collected by a specific model of profiling equipment  the 1995 version of the K.J. Law T6600 inertial profiler. Four such profilers were purchased by LTPP in 1996 for use by each LTPP Regional Contracting Office (North Central, North Atlantic, Southern, Western).
The 1995 T6600 profiler is considered a class I accelerometerestablished inertial profiling reference based on American Society of Testing and Materials (ASTM) E95098. It is a vanmounted system containing two infrared sensors spaced 1,676 mm (66 inches) apart. The system collects longitudinal profile data at 25.4mm (1inch) intervals, and these data are processed through a movingaverage smoothing filter to generate 152mm (6inch) profile data, which are subsequently downloaded and stored in the IMS database. Also stored in the IMS database are the individual wheelpath IRI values computed from the 152mm (6inch) profile data.
The original 25.4mm (1inch) profile data are also archived, but they are done so in the AIMS databases managed by each Regional Contracting Office. Because current automated profilographs record profile traces on 32mm (1.25inch) intervals, the AIMS profile data represent a closer match of the profile traces than the 152mm (6inch) IMS profile data. Hence, in addition to requesting IRI and relevant test section data (e.g., State ID, SHRP ID, experiment number, pavement type, climatic information) contained in the IMS database, all available 25.4mm (1inch) profile data were solicited.
Conversion of Profile Data to Simulated PI Values
To model profilograph traces and generate simulated PI values from the AIMS profile data, a calibrated software modeling system was used. In 1995, K.J. Law developed software to model Californiatype profilograph traces and output PI values. This software is now used with their lightweight profilers to compute PI and IRI. K.J. Law's lightweight profilers use the same vertical elevation sensors that are mounted on the T6600 profiler, which again has been the device used to collect profiles for the LTPP program. Although there are several good lightweight profilers and PI modeling systems available, the K.J. Law modeling software was selected for this study to provide the most compatibility with the available LTPP profile data.
Using the modeling and index computation software currently installed on their commercial lightweight profilers, K.J. Law developed interface for analysis of the LTPP data. Named "Indexer," the software computes PI, IRI, and ride number (RN) values using University of Michigan Transportation Research Institute (UMTRI) Engineering Research Department (ERD) format input files. The operator can set the blanking band, as well as several other parameters, such as the type of smoothing filter (moving average or thirdorder Butterworth) and the type of scallop filter (height, length, rounding).
In this study, the 25.4mm (1inch) AIMS profile data were processed into 0.0, 2.5, and 5mm (0.0, 0.1, and 0.2inch) blanking band PI values (herein designated as PI_{0.0}, PI_{2.5mm}, and PI_{5mm}) for each profile data set using the K.J. Law Indexer 3.0 software. These simulated PI values were computed using a 0.76m (2.5ft) movingaverage filter, along with minimum height, maximum height, and rounding scallop filters settings of 0.9, 0.6, and 0.25 mm (0.035, 0.024, and 0.01 inches), respectively.
During the conversion of profile data into simulated PI values, the issue of subsectioning of SPS profile data was addressed. Unlike GPS test sites, which serve as individual 152.5m (500ft) test sections, each SPS test site contains between 3 and 20 test sections comprised of different designs, materials, and construction practices. Profile data for each SPS site are collected in one pass, and the data are subsectioned only after conversion to 152mm (6inch) intervals.
To extract 25.4mm (1inch) profile data for each SPS test section, a special subsectioning program was developed and applied to each continuous SPS test site profile. Each subsectioned profile was then processed for IRI, PI_{0.0}, PI_{2.5mm}, and PI_{5mm} using the Indexer program. As a data quality control measure, each IRI value computed by Indexer was compared with the IRI value computed in the field and subsequently reported in the IMS database. All profiler runs that showed more than 0.0075 m/km (0.475 inches per mile) difference between the Indexercomputed IRI and the IMS database IRI were excluded from the project database.
Populating the Project Database
IRI and relevant test section data obtained from the IMS database were downloaded into Microsoft® Access, a database management system that provides easy extraction of data into spreadsheets and statistical analysis input files. The IRI data consist of right and left wheelpath IRI values generated from individual profiler runs conducted on GPS and SPS sites between 1996 and 2001.
Simulated PI_{0.0}, PI_{2.5mm}, and PI_{5mm} values derived from the 25.4mm (1inch) profile data were also added to the project database. Moreover, to successfully carry out the data analyses for the project, mean IRI and mean simulated PI values were computed from each pair of left and rightwheelpath smoothness values. The resulting means were then added to the project database.
A total of 1,793 LTPP test sections located in 47 States and 8 Canadian Provinces formed the basis for this evaluation. The sections represent a variety of pavement types, including original and restored AC and PCC pavements, asphalt overlays of both AC and PCC pavements, and concrete overlays of PCC pavements. They also span all four climatic zonesdry freeze, dry nonfreeze, wet freeze, wet nonfreeze  as defined by mean annual precipitation (wet being greater than 508 mm [20 inches] of precipitation per year) and mean annual freezing index (FI) (freeze being more than 66Â°Cdays [150°Fdays] per year).
Each test section in the database includes IRI and simulated PI values corresponding to individual profiler runs made between 1996 and 2001. Breakdowns of the test sections by LTPP experiment and by State are provided in tables 2 and 3, respectively.
GPS Experiment No.  No. of GPS LTPP Sections  SPS Experiment No.  No. of SPS LTPP Sections 

GPS1(Conventional AC pavement)  140  SPS1(Structural factors for flexible pavement)  207 
GPS2(Fulldepth AC pavement)  77  SPS2(Structural factors for rigid pavement)  166 
GPS3(JPC pavement)  120  SPS3(Preventive maintenance of AC pavement)  259 
GPS4(JRC pavement)  50  SPS4(Preventive maintenance of PCC pavement)  72 
GPS5(CRC pavement)  64  SPS5(AC overlays on AC pavement)  152 
GPS6(AC overlays on AC pavement)  169  SPS6(CPR and AC overlays on PCC pavement)  103 
GPS7(AC overlays on PCC pavement)  55  SPS7(Bonded PCC overlays on PCC pavement)  27 
GPS9(Unbonded PCC overlays)  24  SPS8(Environmental effects on AC, PCC pavements)  47 
SPS9(SHRP AC mix designs)  80 
State/Province  No. of LTPP Sections  State/Province  No. of LTPP Sections  State/Province  No. of LTPP Sections 

Alabama  58  Michigan  62  South Dakota  27 
Arizona  81  Minnesota  68  Tennessee  25 
Arkansas  57  Mississippi  38  Texas  156 
California  78  Missouri  79  Utah  31 
Colorado  42  Montana  36  Vermont  4 
Connecticut  7  Nebraska  40  Virginia  29 
Delaware  28  Nevada  37  Washington  41 
Florida  46  New Hampshire  1  West Virginia  5 
Georgia  28  New Jersey  23  Wisconsin  53 
Idaho  22  New Mexico  33  Wyoming  21 
Illinois  36  New York  7  Alberta  16 
Indiana  35  North Carolina  39  British Columbia  4 
Iowa  60  North Dakota  16  Manitoba  20 
Kansas  56  Ohio  41  New Brunswick  3 
Kentucky  15  Oklahoma  55  Newfoundland  2 
Louisiana  13  Pennsylvania  30  Ontario  12 
Maine  17  Rhode Island  1  Quebec  11 
Maryland  21  South Carolina  7  Saskatchewan  17 
Massachusetts  3 
Introduction
Based on a comprehensive review of past model development research, the following procedure was utilized in developing LTPPbased PItoPI_{0.0} and PItoIRI relationships:
The steps outlined for model development are summarized in the flow chart shown in figure 14 and are explained in greater detail in the sections that follow. This approach has been used in previous research studies and has been improved to provide practical and accurate models.
Step 1  Preliminary EvaluationIn step 1 of model development, the assembled database was examined to determine its general properties and to identify possible data anomalies (i.e., outliers, missing or erroneous data). The data were "cleaned" as appropriate and then sorted to allow for the development of various PI PI_{0.0} and PIIRI scatter plots for use in model development.
Data Quality Evaluation
Basic statistics, such as the mean and range of data, were used to identify possible gaps in the data and to determine whether the database was representative of the expected inference space. Of specific interest in this process were the following:
Figures 15 through 18 present histograms showing the distribution of IRI and PI_{5mm} for all AC and PCCsurfaced pavements. A detailed summary of the information depicted in the plots (categorized by pavement type and climatic region) is provided in table 4. It is clear from these exhibits that the data used for analysis (i.e., the cleaned data), and for developing PIPI_{0.0} and PIIRI relationships, fully cover the ranges of smoothness typical of new construction and AC overlays (i.e., IRI between 800 and 2,000 mm/km [50 and 125 inches per mile], PI_{5mm} between 0 and 235 mm/km [0 and 15 inches per mile]).
Development of Scatter Plots
As summarized in table 5, 111 scatter plots of IRI versus PI_{5mm}, PI_{2.5mm}, and PI_{0.0}, and 111 scatter plots of PI_{0.0} versus PI_{5mm} and PI_{2.5mm} were produced to aid the model development process. The scatter plots represent various combinations of climatic zone and pavement type. Complete sets of the scatter plots developed in the study are provided in appendixes A and B.
At the broadest level, over 14,000 asphalt pavement smoothness data points (representing the average roughness of right and left wheelpaths) and over 8,000 concrete pavement data points representing all four climatic zones were available for plotting and model development. Figure 19 shows the PI_{0.0}IRI scatter plot for all AC pavements, and figure 20 shows the PI_{5mm}PI_{0.0} scatter plot for all PCC pavements. These plots, which are typical of most of the scatter plots, show reasonably strong (R^{2} > 0.75) and virtually linear relationships between the smoothness indices. They also, however, illustrate the considerable amount of variation due in large part to the inherent differences in the way the smoothness indices process different surface wavelengths.
Examples of the effects of pavement type and climatic zone on the smoothness relationships can be seen in figures 21 and 22. In the case of the PI_{0.0}IRI trends for different AC pavements (figure 21), the differences are almost negligible. Slightly more distinct differences, however, are discernible among the PI_{5mm}PI_{0.0} trends representing different climatic zones (figure 22).
Pavement Type  Climate  N  IRI Mean  IRI Std. Dev.  PI_{0.0} Mean  PI_{0.0} Std. Dev.  PI_{2.5mm} Mean  PI_{2.5mm} Std. Dev.  PI_{5mm} Mean  PI_{5mm} Std. Dev. 

AC  DF  2,720  1,252.5  671.5  373.8  216.9  190.3  187.8  91.0  137.6 
AC  DNF  1,740  1,031.4  518.7  297.7  179.1  142.0  139.4  63.8  89.9 
AC  WF  6,502  1,330.3  661.2  450.8  238.7  254.2  215.1  131.8  163.4 
AC  WNF  4,046  1,186.0  492.6  370.9  181.9  193.5  165.0  92.5  122.4 
AC/AC  DF  1,856  1,397.6  629.0  412.3  218.2  220.4  187.3  106.4  130.8 
AC/AC  DNF  1,502  1,011.1  422.2  274.6  141.9  106.5  103.0  40.1  63.4 
AC/AC  WF  3,832  1,135.8  423.9  342.2  159.6  174.4  140.5  79.5  98.3 
AC/AC  WNF  1,426  1,125.6  491.2  346.6  192.3  181.9  158.8  91.0  110.4 
AC/PC  DF  90  1,072.4  196.9  336.4  88.2  152.8  115.3  63.2  72.7 
AC/PC  DNF  0                 
AC/PC  WF  3,774  1,208.5  444.5  387.1  167.7  181.0  138.2  74.3  89.6 
AC/PC  WNF  376  1,280.6  461.8  358.4  140.3  163.0  130.1  68.4  100.5 
JPC^{a}  DF  2,154  1,536.8  545.2  475.2  210.6  226.4  195.8  99.9  139.5 
JPC^{a}  DNF  1,270  1,464.9  500.1  394.8  175.1  162.1  168.3  68.9  117.8 
JPC^{a}  WF  6,542  1,639.8  703.4  572.3  294.8  334.2  286.5  180.4  236.1 
JPC^{a}  WNF  2,196  1,737.3  612.7  594.1  228.3  337.7  216.0  161.5  174.1 
JRC^{a}  DF  0                 
JRC^{a}  DNF  0                 
JRC^{a}  WF  1,950  1,955.7  508.0  780.0  233.1  531.2  231.2  321.9  197.4 
JRC^{a}  WNF  349  2,053.8  349.0  785.4  144.4  546.3  142.8  328.9  129.2 
CRC^{a}  DF  39  1,330.3  54.0  508.8  25.8  336.3  55.3  160.8  29.0 
CRC^{a}  DNF  120  1,275.2  397.6  395.0  207.2  180.3  200.7  98.0  142.2 
CRC^{a}  WF  722  1,575.6  478.0  541.7  202.2  332.9  204.0  181.7  166.7 
CRC^{a}  WNF  358  1,620.7  457.5  563.4  193.3  343.6  188.6  189.2  153.7 
Climatic Zone  Model  All AC  AC  AC/AC  AC/PCC  All PCC  JPC  JRC  CRC 

DryFreeze  IRI vs. PI_{0.0}  Yes  Yes  Yes  Yes  Yes  Yes    Yes 
DryFreeze  IRI vs. PI_{2.5mm}  Yes  Yes  Yes  Yes  Yes  Yes    Yes 
DryFreeze  IRI vs. PI_{5mm}  Yes  Yes  Yes  Yes  Yes  Yes    Yes 
DryNonfreeze  IRI vs. PI_{0.0}  Yes  Yes  Yes    Yes  Yes    Yes 
DryNonfreeze  IRI vs. PI_{2.5mm}  Yes  Yes  Yes    Yes  Yes    Yes 
DryNonfreeze  IRI vs. PI_{5mm}  Yes  Yes  Yes    Yes  Yes    Yes 
WetFreeze  IRI vs. PI_{0.0}  Yes  Yes  Yes  Yes  Yes  Yes  Yes  Yes 
WetFreeze  IRI vs. PI_{2.5mm}  Yes  Yes  Yes  Yes  Yes  Yes  Yes  Yes 
WetFreeze  IRI vs. PI_{5mm}  Yes  Yes  Yes  Yes  Yes  Yes  Yes  Yes 
WetNonfreeze  IRI vs. PI_{0.0}  Yes  Yes  Yes  Yes  Yes  Yes  Yes  Yes 
WetNonfreeze  IRI vs. PI_{2.5mm}  Yes  Yes  Yes  Yes  Yes  Yes  Yes  Yes 
WetNonfreeze  IRI vs. PI_{5mm}  Yes  Yes  Yes  Yes  Yes  Yes  Yes  Yes 
All  IRI vs. PI_{0.0}  Yes  Yes  Yes  Yes  Yes  Yes  Yes  Yes 
All  IRI vs. PI_{2.5mm}  Yes  Yes  Yes  Yes  Yes  Yes  Yes  Yes 
All  IRI vs. PI_{5mm}  Yes  Yes  Yes  Yes  Yes  Yes  Yes  Yes 
DryFreeze  PI_{0.0} vs. PI_{5mm}  Yes  Yes  Yes  Yes  Yes  Yes    Yes 
DryFreeze  PI_{0.0} vs. PI_{2.5mm}  Yes  Yes  Yes  Yes  Yes  Yes    Yes 
DryFreeze  PI_{2.5mm} vs. PI_{5mm}  Yes  Yes  Yes  Yes  Yes  Yes    Yes 
DryNonfreeze  PI_{0.0} vs. PI_{5mm}  Yes  Yes  Yes    Yes  Yes    Yes 
DryNonfreeze  PI_{0.0} vs. PI_{2.5mm}  Yes  Yes  Yes    Yes  Yes    Yes 
DryNonfreeze  PI_{2.5mm} vs. PI_{5mm}  Yes  Yes  Yes    Yes  Yes    Yes 
WetFreeze  PI_{0.0} vs. PI_{5mm}  Yes  Yes  Yes  Yes  Yes  Yes  Yes  Yes 
WetFreeze  PI_{0.0} vs. PI_{2.5mm}  Yes  Yes  Yes  Yes  Yes  Yes  Yes  Yes 
WetFreeze  PI_{2.5mm} vs. PI_{5mm}  Yes  Yes  Yes  Yes  Yes  Yes  Yes  Yes 
WetNonfreeze  PI_{0.0} vs. PI_{5mm}  Yes  Yes  Yes  Yes  Yes  Yes  Yes  Yes 
WetNonfreeze  PI_{0.0} vs. PI_{2.5mm}  Yes  Yes  Yes  Yes  Yes  Yes  Yes  Yes 
WetNonfreeze  PI_{2.5mm} vs. PI_{5mm}  Yes  Yes  Yes  Yes  Yes  Yes  Yes  Yes 
All  PI_{0.0} vs. PI_{5mm}  Yes  Yes  Yes  Yes  Yes  Yes  Yes  Yes 
All  PI_{0.0} vs. PI_{2.5mm}  Yes  Yes  Yes  Yes  Yes  Yes  Yes  Yes 
All  PI_{2.5mm} vs. PI_{5mm}  Yes  Yes  Yes  Yes  Yes  Yes  Yes  Yes 
As can be seen, in most instances, the LTPP regression envelope covers the individual PIIRI relationships. In the case of the PI_{5mm}IRI relationships for asphalt (figure 23), the PTI relationship and one of the Florida relationships (IRI using ultrasonic profiler) extended outside the LTPP envelope. Equipment is likely a contributing factor with respect to the Florida relationship, as ultrasonic sensors were used as opposed to the infrared sensors used in the LTPP program). And, as noted in the figure, the PTI relationship was based on measurements for both AC and PCC pavements.
In the PI_{5mm}IRI relationships for concrete (figure 23), the Arizona relationship contrasted sharply with the LTPP relationship. The fact that the Arizona relationship was based on measurements from only 12 concrete pavement sections may help explain this departure. However, other factors, such as sensor type (Arizona used optical sensors), are likely to have also contributed to this phenomenon.
Step 2  Selection of Appropriate Model Form
Model development began with the selection of the most suitable functional form that best describes the relationship between IRI and PI. As indicated by the scatter plots presented in figures A1 through A37 and B1 to B37 in appendixes A and B, the PIIRI relationship is virtually linear and, thus, a linear function with IRI as the dependent variable and PI as the independent variable was adopted.
A similar functional form was selected for developing the PI_{0.0} versus PI models, as indicated by the corresponding scatter plots in appendixes A and B. The linear relationship as shown in the figures was true for both AC and PCC surface pavements. The magnitude of the slope, however, varied according to pavement type (AC vs. PCC, or Jointed Plain Concrete (JPC) vs. Continuosly Reinforced Concrete (CRC)) and the climatic region in which the pavement was located. The model form selected is shown as equation 1.
IRI = alpha + beta*PI_{X}, Eq. 1
where:
IRI = International roughness index, mm/km.
PI_{X} = Profile index for blanking band X (X = 0.0, 2.5, or 5.0 mm).
alpha, beta = regression constants.
Step 3  Group Data into Sets with Similar Smoothness Relations
Ideally, a single model could be developed to relate the various smoothness indices (e.g., IRI versus PI_{0.0} for all pavement types, climatic regions). However, unless the influences of different climatic zones and pavement types were statistically insignificant, this would result in the development of models with low prediction capabilities and the introduction of significant levels of error in predicted indices.
On the other hand, developing models for all the different combinations of pavement types (e.g., AC, AC/AC, JPC, CRC) would result in the development of a minimum of 144 models, as illustrated in table 6. So many models is not only impractical from a user's point of view, but could not be developed with the level of accuracy required, due to the lack of sufficient amounts of data in some of the cells in table 6.
For this study, it was deemed important to merge cells within the two main blocks (AC and PCCsurfaced pavements) in table 6 with similar relationships between the different smoothness indices. Models were developed for a total of six combinations of smoothness indices as follows:
Block  Pavement Type  Climatic Region  IRI vs PI (0.0)  IRI vs PI (2.5mm)  IRI vs PI (5mm)  PI (0.0) vs PI (2.5mm)  PI (0.0) vs PI (5mm)  PI (2.5mm) vs PI (5mm) 

Block 1(ACsurfaced pavements)  AC  DF  1  2  3  4  5  6 
Block 1(ACsurfaced pavements)  AC  DNF  7  8  9  10  11  12 
Block 1(ACsurfaced pavements)  AC  WF  13  14  15  16  17  18 
Block 1(ACsurfaced pavements)  AC  WNF  19  20  21  22  23  24 
Block 1(ACsurfaced pavements)  AC/AC  DF  25  26  27  28  29  30 
Block 1(ACsurfaced pavements)  AC/AC  DNF  31  32  33  34  35  36 
Block 1(ACsurfaced pavements)  AC/AC  WF  37  38  39  40  41  42 
Block 1(ACsurfaced pavements)  AC/AC  WNF  43  44  45  46  47  48 
Block 1(ACsurfaced pavements)  AC/PC  DF  49  50  51  52  53  54 
Block 1(ACsurfaced pavements)  AC/PC  DNF  55  56  57  58  59  60 
Block 1(ACsurfaced pavements)  AC/PC  WF  61  62  63  64  65  66 
Block 1(ACsurfaced pavements)  AC/PC  WNF  67  68  69  70  71  72 
Block 2(PCCsurfaced pavements)  JPC  DF  73  74  75  76  77  78 
Block 2(PCCsurfaced pavements)  JPC  DNF  79  80  81  82  83  84 
Block 2(PCCsurfaced pavements)  JPC  WF  85  86  87  88  89  90 
Block 2(PCCsurfaced pavements)  JPC  WNF  91  92  93  94  95  96 
Block 2(PCCsurfaced pavements)  JRC  DF  97  98  99  100  101  102 
Block 2(PCCsurfaced pavements)  JRC  DNF  103  104  105  106  107  108 
Block 2(PCCsurfaced pavements)  JRC  WF  109  110  111  112  113  114 
Block 2(PCCsurfaced pavements)  JRC  WNF  115  116  117  118  119  120 
Block 2(PCCsurfaced pavements)  CRC  DF  121  122  123  124  125  126 
Block 2(PCCsurfaced pavements)  CRC  DNF  127  128  129  130  131  132 
Block 2(PCCsurfaced pavements)  CRC  WF  133  134  135  136  137  138 
Block 2(PCCsurfaced pavements)  CRC  WNF  139  140  141  142  143  144 
Similar relationships between the smoothness indices listed above was defined as cells with the same surface type with statistically insignificant differences in mean slope or gradient of a linear model developed relating the two given indices. This is shown conceptually in figure 28.
Thus, cells with similar PIIRI or PIPI relationships were merged for model development, so as to limit the number of models. As shown in table 6, cells were defined according to pavement type (e.g., AC, AC/PCC, Jointed Reinforced Concrete (JRC)) and climatic region. For PCC pavements, the categories of surface type were limited to JPC, JRC, and CRC because there were an insufficient number of PCC overlays (e.g., JPC/JPC) to perform a detailed and thorough analysis.
The procedures used to compute mean slopes for the smoothness index relationships for each cell in table 6 are as follows:
Figures 29 through 32 show, for both AC and PCCsurfaced pavements, examples of the distribution of slopes for IRI versus PI_{0.0} and PI_{0.0} versus PI_{5mm}.
The next step involved testing for similarities or differences in the mean slopes (beta) among cells. This analysis was limited to cells within each pavement category, as it was assumed that there were differences in slopes between AC and PCC pavements. To determine potential similarities or differences among cells, analysis of variance (ANOVA) was performed at the following two levels:
Level 1 Analysis
Level 1 analysis involved the following tasks:
The data used in level 1 analysis were as follows:
The basic ANOVA type I statistical model was used in analysis. Like other basic regression models, it was a linear statistical relation between the independent variables and the dependent variable. The model is presented as follows:
Beta = gamma(1) + gamma(2)*CLIMATE + gamma(3)*PVMT, Eq. 2
where:
Beta = Slope of PIIRI or PIPI linear model.
CLIMATE = Test pavement climate location.
PVMT = Pavement type.
gamma(1), gamma(2), gamma(3) = Regression constants.
The following hypothesis was tested under the level 1 analysis:
Acceptance or rejection of the null hypothesis was accomplished by computing the level of significance (pvalue) for each of the independent classification variables in equation 2 and comparing it to a predetermined level of significance. For this study, a 95 percent level of significance (pvalue = 5 percent) was used. Thus, a computed pvalue of 0.05 or less would cause the null hypothesis to be rejected, whereas a pvalue greater than 0.05 would confirm the null hypothesis.
The results of the ANOVA are presented in tables 7 and 8 for AC and PCC pavements, respectively. The results show that both pavement type and climate had a significant effect on the PIIRI and PIPI relationships. That is, the ANOVA Ftest results indicated that one or more of the mean slopes for the different cells in the matrix presented in table 6 were significantly different.
Dependent Variable^{a}  Grouping Variable^{b}  N  FStatistic Value  Probability > F (pvalue) 

Slope of PI_{0.0}IRI linear relationship  Pavement Type  2,395  2.24  0.1061^{c} 
Slope of PI_{0.0}IRI linear relationship  Climate  2,395  7.61  0.0001^{d} 
Slope of PI_{2.5mm}IRI linear relationship  Pavement Type  2,395  5.91  0.0028^{d} 
Slope of PI_{2.5mm}IRI linear relationship  Climate  2,395  3.87  0.0089^{d} 
Slope of PI_{5mm}IRI linear relationship  Pavement Type  2,395  0.72  0.4870^{e} 
Slope of PI_{5mm}IRI linear relationship  Climate  2,395  0.95  0.4177^{e} 
Notes:
^{a} Computed for each wheelpath within a given pavement section within a uniform construction period.
^{b} Pavement type considered  AC, AC/AC, and AC/PCC and climate types  DF, DNF, WF, and WNF.
^{c} Borderline significance at the 10 percent significance level.
^{d} Significant at the 5 percent significance level.
^{e} Not significant.
Dependent Variable^{a}  Grouping Variable^{b}  N  FStatistic Value  Probability > F (pvalue) 

Slope of PI_{0.0}IRI linear relationship  Pavement Type  1,123  3.44  0.0630^{d} 
Slope of PI_{0.0}IRI linear relationship  Climate  1,123  4.82  0.0024^{d} 
Slope of PI_{2.5mm}IRI linear relationship  Pavement Type  1,123  5.51  0.0190^{d} 
Slope of PI_{2.5mm}IRI linear relationship  Climate  1,123  13.68  0.0001^{d} 
Slope of PI_{5mm}IRI linear relationship  Pavement Type  1,119  2.96  0.0850^{c} 
Slope of PI_{5mm}IRI linear relationship  Climate  1,119  12.46  0.0001^{4} 
Notes:
^{a} Computed for each wheelpath within a given pavement section within a uniform construction period.
^{b} Pavement type considered  AC, AC/AC, and AC/PCC and climate types  DF, DNF, WF, and WNF.
^{c} Borderline significance at the 10 percent significance level.
^{d} Significant at the 5 percent significance level.
^{e} Not significant.
Level 2 Analysis
Although the ANOVA Ftest results listed in tables 7 and 8 indicate significant differences in mean slope for the various cells evaluated, they do not show which cells were similar or how the cells differed from each other. This information is required in order to merge cells that have similar slopes or trends in their PIIRI and PIPI relationships, so as to optimize and reduce the number of models to be developed.
Duncan's multiple comparison method in ANOVA was used to group cells with similarities among their mean slopes at a 95 percent significance level. Table 9 provides a summary of the grouping based on the Duncan's multiple comparison tests. The final groupings were based not only on the results of the statistical analysis, but also on the practicality of the groupings and engineering judgment.
Steps 4 and 5  Develop Tentative Models and Assess Models for Reasonableness
Linear regression models for all of the groupings (merged cells) in table 9 were developed and are presented in tables 10 through 13. Each model was verified for accuracy and reasonableness by evaluating diagnostic statistics, such as the standard estimate of the error (SEE), coefficient of determination (R^{2}), and the number of data points used in model development.
In general, the models appeared to be reasonable. For ACsurfaced pavement models, R^{2} was typically greater than 70 percent, with only 3 out of 33 models having reported R^{2} values less than 70 percent. SEE ranged from 178 to 308 mm/km (11.2 to 19.5 inches per mile) for IRI and 21 to 79 mm/km (1.3 to 5.0 inches per mile) for PI. These models contained the largest number of data points to date for modeling the PIIRI relationships, ranging from 1,800 to 14,170 data points per model.
Block  Pavement Type  Climatic Region  IRI vs PI_{0.0}  IRI vs PI_{2.5mm}  IRI vs PI_{5mm}  PI_{0.0} vs PI_{2.5mm}  PI_{0.0} vs PI_{5mm}  PI_{2.5mm} vs PI_{5mm} 

Block 1 (ACsurfaced pavements)  AC  DF  1  2  3  4  5  6 
Block 1 (ACsurfaced pavements)  AC  DNF  1  2  3  7  8  9 
Block 1 (ACsurfaced pavements)  AC  WF  1  2  3  4  5  6 
Block 1 (ACsurfaced pavements)  AC  WNF  1  2  3  7  8  9 
Block 1 (ACsurfaced pavements)  AC/AC  DF  10  11  12  13  14  15 
Block 1 (ACsurfaced pavements)  AC/AC  DNF  16  17  18  19  20  21 
Block 1 (ACsurfaced pavements)  AC/AC  WF  22  23  24  25  26  27 
Block 1 (ACsurfaced pavements)  AC/AC  WNF  22  23  24  25  26  27 
Block 1 (ACsurfaced pavements)  AC/PC  DF  28  29  30  31  32  33 
Block 1 (ACsurfaced pavements)  AC/PC  DNF  28  29  30  31  32  33 
Block 1 (ACsurfaced pavements)  AC/PC  WF  28  29  30  31  32  33 
Block 1 (ACsurfaced pavements)  AC/PC  WNF  28  29  30  31  32  33 
Block 2 (PCCsurfaced pavements)  JPC  DF  34  35  36  37  38  39 
Block 2 (PCCsurfaced pavements)  JPC  DNF  40  41  42  43  44  45 
Block 2 (PCCsurfaced pavements)  JPC  WF  34  35  36  46  47  48 
Block 2 (PCCsurfaced pavements)  JPC  WNF  49  50  51  52  53  54 
Block 2 (PCCsurfaced pavements)  JRC  DF  34  35  36  37  38  39 
Block 2 (PCCsurfaced pavements)  JRC  DNF  40  41  42  43  44  45 
Block 2 (PCCsurfaced pavements)  JRC  WF  34  35  36  46  47  48 
Block 2 (PCCsurfaced pavements)  JRC  WNF  49  50  51  52  53  54 
Block 2 (PCCsurfaced pavements)  CRC  DF  34  35  36  37  38  39 
Block 2 (PCCsurfaced pavements)  CRC  DNF  40  41  42  43  44  45 
Block 2 (PCCsurfaced pavements)  CRC  WF  34  35  36  46  47  48 
Block 2 (PCCsurfaced pavements)  CRC  WNF  49  50  51  52  53  54 
Note: Cells with the same numbers share the same model.
Pavement Type  Climate^{a}  Blanking Band (mm)  Correlation Equation (IRI = mm/km, PI = mm/km)  N  SEE  R^{2} 

AC  1,2,3,4  0.0  IRI = 2.66543*PI_{0.0} + 213.01  14,170  200.17  0.89 
AC  1,2,3,4  2.5  IRI = 2.97059*PI_{2.5mm} + 638.74  14,160  231.69  0.86 
AC  1,2,3,4  5.0  IRI = 3.78601*PI_{5mm} + 887.51  13,775  292.26  0.77 
AC/AC  1  0.0  IRI = 2.74599*PI_{0.0} + 265.42  1,854  191.97  0.91 
AC/AC  2  0.0  IRI = 2.68169*PI_{0.0} + 274.67  1,494  184.64  0.81 
AC/AC  3,4  0.0  IRI = 2.42295*PI_{0.0} + 301.90  5,126  178.81  0.84 
AC/AC  1  2.5  IRI = 3.12622*PI_{2.5mm} + 708.56  1,854  230.03  0.87 
AC/AC  2  2.5  IRI = 3.33564*PI_{2.5mm} + 655.67  1,494  246.64  0.66 
AC/AC  3,4  2.5  IRI = 2.68324*PI_{2.5mm} + 660.34  5,126  216.98  0.76 
AC/AC  1  5.0  IRI = 4.25316*PI_{5mm} + 957.80  1,824  288.17  0.79 
AC/AC  2  5.0  IRI = 4.39478*PI_{5mm} + 883.20  1,345  308.23  0.45 
AC/AC  3,4  5.0  IRI = 3.42671*PI_{5mm} + 876.80  4,906  265.85  0.63 
AC/PCC  1,2,3,4  0.0  IRI = 2.40300*PI_{0.0} + 292.93  4,156  205.58  0.79 
AC/PCC  1,2,3,4  2.5  IRI = 2.78217*PI_{2.5mm} + 716.87  4,156  229.68  0.73 
AC/PCC  1,2,3,4  5.0  IRI = 3.94665*PI_{5mm} + 939.22  4,052  259.58  0.65 
^{a} Climatic zones: 1=DF, 2=DNF, 3=WF, 4=WNF.
Pavement Type  Climate^{a}  Correlation Equation (PI = mm/km)  N  SEE  R^{2} 

AC  1,3  PI_{0.0} = 1.08722*PI_{2.5mm} + 174.42  5,744  47.73  0.96 
AC  1,3  PI_{0.0} = 1.35776*PI_{5mm} + 275.48  5,684  83.58  0.88 
AC  1,3  PI_{2.5mm} = 1.28213*PI_{5mm} + 87.79  5,684  46.62  0.95 
AC  2,4  PI_{0.0} = 1.12338*PI_{2.5mm} + 152.84  8,418  45.23  0.95 
AC  2,4  PI_{0.0} = 1.46417*PI_{5mm} + 240.09  8,093  71.73  0.86 
AC  2,4  PI_{2.5mm} = 1.34055*PI_{5mm} + 73.13  8,093  38.64  0.95 
AC/AC  1  PI_{0.0} = 1.14153*PI_{2.5mm} + 160.70  1,856  43.41  0.96 
AC/AC  1  PI_{0.0} = 1.56038*PI_{5mm} + 250.89  1,826  73.74  0.88 
AC/AC  1  PI_{2.5mm} = 1.39462*PI_{5mm} + 75.55  1,826  40.47  0.95 
AC/AC  2  PI_{0.0} = 1.28067*PI_{2.5mm} + 138.15  1,496  52.26  0.86 
AC/AC  2  PI_{0.0} = 1.75837*PI_{5mm} + 222.84  1,347  79.32  0.66 
AC/AC  2  PI_{2.5mm} = 1.52523*PI_{5mm} + 56.60  1,347  34.14  0.89 
AC/AC  3,4  PI_{0.0} = 1.11926*PI_{2.5mm} + 145.85  5,128  44.86  0.93 
AC/AC  3,4  PI_{0.0} = 1.45876*PI_{5mm} + 233.59  4,908  71.53  0.81 
AC/AC  3,4  PI_{2.5mm} = 1.36739*PI_{5mm} + 71.17  4,908  38.12  0.93 
AC/PCC  1,2,3,4  PI_{0.0} = 1.15412*PI_{2.5mm} + 177.08  4,158  44.46  0.93 
AC/PCC  1,2,3,4  PI_{0.0} = 1.61123*PI_{5mm} + 271.11  4,054  71.07  0.81 
AC/PCC  1,2,3,4  PI_{2.5mm} = 1.44895*PI_{5mm} + 76.83  4,054  36.99  0.93 
^{a} Climatic zones: 1=DF, 2=DNF, 3=WF, 4=WNF.
Pavement Type  Climate^{a}  Blanking Band (mm)  Correlation Equation (IRI = mm/km, PI = mm/km)  N  SEE  R^{2} 

PCC  1,3  0.0  IRI = 2.12173*PI_{0.0} + 439.76  12,039  259.63  0.84 
PCC  2  0.0  IRI = 2.58454*PI_{0.0} + 423.09  1,448  176.54  0.88 
PCC  4  0.0  IRI = 2.3582*PI_{0.0} + 317.19  2,888  236.51  0.84 
PCC  1,3  2.5  IRI = 2.15316*PI_{2.5mm} + 947.05  12,039  278.69  0.81 
PCC  2  2.5  IRI = 2.5921*PI_{2.5mm} + 1024.73  1,448  226.53  0.80 
PCC  4  2.5  IRI = 2.40731*PI_{2.5mm} + 888.10  2,888  264.46  0.79 
PCC  1,3  5.0  IRI = 2.62558*PI_{5mm} + 1205.73  11,946  305.96  0.77 
PCC  2  5.0  IRI = 3.51673*PI_{5mm} + 1226.35  1,364  268.70  0.72 
PCC  4  5.0  IRI = 2.87407*PI_{5mm} + 1229.63  2,885  297.37  0.74 
^{a} Climatic zones: 1=DF, 2=DNF, 3=WF, 4=WNF.
Pavement Type  Climate^{a}  Correlation Equation (PI = mm/km)  N  SEE  R^{2} 

PCC  1  PI_{0.0} = 1.39512*PI_{5mm} + 343.08  2,182  71.19  0.87 
PCC  2  PI_{0.0} = 1.36715*PI_{5mm} + 313.25  1,366  66.42  0.86 
PCC  3  PI_{0.0} = 1.20723*PI_{5mm} + 367.91  9,764  86.73  0.91 
PCC  4  PI_{0.0} = 1.19909*PI_{5mm} + 390.49  2,885  85.19  0.85 
PCC  1  PI_{0.0} = 1.04364*PI_{2.5mm} + 238.13  2,237  46.91  0.95 
PCC  2  PI_{0.0} = 1.02028*PI_{2.5mm} + 229.78  1,448  44.34  0.94 
PCC  3  PI_{0.0} = 1.01255*PI_{2.5mm} + 238.65  9,800  49.98  0.97 
PCC  4  PI_{0.0} = 1.01320*PI_{2.5mm} + 244.81  2,888  56.94  0.94 
PCC  1  PI_{2.5mm} = 1.36458*PI_{5mm} + 96.46  2,180  43.27  0.95 
PCC  2  PI_{2.5mm} = 1.38376*PI_{5mm} + 74.90  1,364  39.84  0.95 
PCC  3  PI_{2.5mm} = 1.20990*PI_{5mm} + 123.95  9,764  53.62  0.96 
PCC  4  PI_{2.5mm} = 1.212677*PI_{5mm} + 138.43  2,885  42.99  0.96 
^{a} Climatic zones: 1=DF, 2=DNF, 3=WF, 4=WNF.
All of the PCCsurfaced pavement models had R^{2} greater than 70 percent. SEE ranged from 177 to 306 mm/km (11.2 to 19.4 inches per mile) for IRI and 21 to 79 mm/km (1.3 to 5.0 inches per mile) for PI. These models contain the largest number of data points to date for modeling the PIIRI relationships, ranging from 1,366 to 12,039 data points per model.
Step 6  Select Final Models
Fifteen models were developed for the PIIRI relationships and 18 models were developed for the PIPI relationships for ACsurfaced pavements. For PCCsurfaced pavements, 9 and 12 models were developed for PIIRI and PIPI relationships, respectively. The models were developed using a database that represented a reasonable inference space (IRI ranged from 300 to 4,000 mm/km [19 to 253 inches per mile] and PI ranged from 0 to 1,700 mm/km [0 to 108 inches per mile] for all blanking bands). The number of data points used in model development ranged from 1,300 to 14,000.
In general, the models developed were adequate and predicted IRI and PI well. An evaluation of diagnostic statistics, such as SEE and R^{2}, showed that there was a good correlation between the measured and predicted smoothness indices from the models (R^{2} was typically > 70 percent) with a reasonable level of error (ranged from 34 to 86.7 mm/km [2.1 to 5.5 inches per mile] for PI and 177 to 308 [11.2 to 19.5 inches per mile] for IRI).
The models presented in tables 10 through 13 predict the mean smoothness index (IRI or PI) for the sample LTPP data used in model development. In this case, the sample means are probably a reasonable estimate of means of the population of pavements within the limits of the reference data. However, they do not necessarily indicate the range of values within which the true population means lies.
The range of values within which the true population mean lies can be obtained by computing a confidence interval around the predicted sample mean. The confidence interval for the mean provides a range of values around the mean where one can expect the "true" (population) mean to be located (with a given level of certainty). Confidence interval can be computed using the following equation:
CI = mean ± t_{alpha/2}sigma, Eq. 3
where:
CI = Confidence interval.
mean = Predicted smoothness index.
t = Value of tstatistic at a given significance level.
alpha = Significance level (usually 90 or 95 percent).
sigma = Model standard error of estimate (SEE).
For example, if the predicted mean IRI (computed using models based on the LTPP data sample) is 1,000 mm/km (63.4 inches per mile), and the lower and upper limits at a significance level of 95 percent are 900 and 1,100 mm/km (57.0 and 69.7 inches per mile) respectively, then it can be concluded that there is a 95 percent probability that the population mean is between 900 and 1,100 mm/km (57.0 and 69.7 inches per mile). If the significance level is set to a smaller value (say 99 percent), then the interval would become wider thereby increasing the certainty of the estimate, and vice versa.
In essence, the larger the sample size, the more reliable will be its mean, and the larger the variation (SEE), the less reliable will be the mean. Sample size used for development of both the LTPP PIIRI and PIPI models ranged from 1,347 to 14,170 data points. These numbers are greater than the generally required minimum of 100 and should provide reliable results.
The SEE values associated with the PIIRI models in tables 10 and 12 ranged from 179 to 292 mm/km (11.3 to 18 inches per mile). The SEE values associated with the PIPI models in tables 11 and 13 ranged from 25 to 58 mm/km (1.6 to 3.7 inches per mile). These SEE values are reasonable, considering the inherent differences in the way surface wavelengths are processed for IRI and PI.
Introduction
The vast majority of U.S. highway agencies use smoothness specifications to ensure an adequate level of initial smoothness for newly constructed and resurfaced pavements. Smoothness specifications typically define the type of equipment and testing procedures to be used to measure initial smoothness, the method of identifying significant bumps to be removed, the type of smoothness statistics to be computed and reported, and the levels of smoothness required for full pay, bonuses, penalties, and corrective work.
As discussed in chapter 1, most specifications are based on the PI smoothness statistic, as measured using a profilograph. Although these specifications differ primarily in terms of PI limits for acceptable smoothness and pay adjustment provisions, there are also differences in testing procedures and equipment. For instance, the length, location, and timeframe specified for testing may be different, as might the responsibility (i.e., contractor vs. agency) for testing. Also, there are various makes and models of profilographs (Ames, Cox, and McCracken Californiatype profilographs, Rainharttype profilograph, and manual or computerized trace reduction), and different filters (3^{rd} order Butterworth, 1^{st} order Cox, and moving average), and blanking band sizes (0, 2.5, and 5 mm [0, 0.1, 0.2 inches]) that can be applied to compute PI.
This chapter provides a summary of States' current AC and PCC smoothness specifications and presents the results of an effort to develop recommended IRI smoothness limits that correspond to existing specified PI limits. The recommended IRI limits were derived using the PIIRI conversion models developed and reported in chapter 4.
Overview of State Smoothness Specifications
In the last 10 years, at least five different national surveys have been conducted to show the status of State smoothness specifications. In each of these surveys, about half of the responding agencies use a California or Rainharttype profilograph for testing new AC pavements, whereas slightly more than threefourths of the agencies use profilographs for new PCC.
Usage of responsetype testing devices (e.g., Mays meter) on AC pavements declined slightly during this time, from about 15 percent in the mid 1990s to about 10 percent now. In contrast, the use of inertial profilers on AC pavements increased appreciably, from about 6 percent in the early 1990s to about 24 percent now. For testing of PCC pavements, the use of responsetype systems stayed the same (about 2 percent), while the use of inertial profilers increased from about 6 percent in 1992 to about 10 percent now.
Tables 14 and 15 list some of the key aspects of current State smoothness specifications, including the type of equipment and smoothness index used, the testing interval, and the smoothness ranges specified for acceptance, correction, bonus, and penalty. The information contained in these tables is based largely on data compiled by the FHWA in 2000 (Rizzo) and on inquiries made to selected State agencies.
As can be seen in table 14, 26 of the 50 States and Puerto Rico have a PIbased smoothness specification for asphalt pavements. Of these 26 agencies, 21 use the 5mm (0.2inch) blanking band, 1 uses the 2.5mm (0.1inch) blanking band, and 4 use the zero blanking band. Collectively, the ranges for full pay are as follows:
For concrete pavements (table 15), 42 of the 50 States and Puerto Rico have a PIbased smoothness specification. Of these 42 agencies, 1 uses a 7.5mm (0.3inch) blanking band, 31 use the 5mm (0.2inch) blanking band, 4 use the 2.5mm (0.1inch) blanking band, and 6 use the zero blanking band. The collective ranges for full pay are as follows:
Development of Recommended Initial IRI and PI_{0.0} Levels
To assist agencies in transitioning from their existing PItype specification to a PI_{0.0} or IRI specification, the LTPPbased correlation models developed and presented in chapter 4 were applied to the fullpay PI limits given in tables 14 and 15. For each State with a PI specification, the respective correlation model was used to develop best estimates of the fullpay PI_{0.0} and IRI limits for new AC, new PCC, AC overlays on AC, and AC overlays on PCC.
The results of this effort are summarized in tables 16 through 19. Each table lists, for a given State, its currently reported PI_{5mm}, PI_{2.5mm}, or PI_{0.0} fullpay smoothness limits, its different climate types, and the estimated equivalent PI_{0.0} and IRI values, computed using the PIIRI model reflective of the State's predominant climate (highlighted in column 3). These estimated equivalent PI_{0.0} and IRI values can be used as a starting point for developing specifications based on one of these two indices.
Because the IRI and PI indices are not exactly correlated, tables 16 through 19 include a 90 percent standard error of the estimate range for the projected specification limit. This error rating should assist specification writers in defining their limits. It also can be used as a basis for refining the specification on an ongoing basis.
State  Testing Device  Index  Testing Interval  Bonus Range  Full Pay Range  Penalty Range  Correction Range 

AL  Californiatype profilograph  PI_{5mm}  0.16 km^{a} (0.1 mi)  <32 mm/km (<2 inches per mile)  32  63 mm/km (2  3.9 inches per mile)  64  160 mm/km (4  10 inches per mile)  >160 mm/km (<10 inches per mile) 
AK               
AZ  GMtype profiler  MRN  0.16 km^{a} (0.1 mi)  <520 mm/km^{a} (<33 inches per mile)  520  710 mm/km^{a} (33  45 inches per mile)  711  1578 mm/km^{a} (46  100 inches per mile)  <1578 mm/km^{a} (>100 inches per mile) 
AR  Californiatype profilograph, lightweight profiler  PI_{5mm}  0.2 km (0.1 mi)  </= 45 mm/km (</= 3 inches per mile)  46  75 mm/km (3.1  5 inches per mile)  76  110 mm/km (5.1  7 inches per mile)  >110 mm/km (< 7 inches per mile) 
CA  Californiatype profilograph  PI_{5mm}  0.16 km (0.1 mi)^{a}    </= 80 mm/km (</= 5 inches per mile)^{a}    >80 mm/km (>5 inches per mile)^{a} 
CO  Californiatype profilograph  PI_{2.5mm}  0.15 km (0.095 mi)  </= 222 mm/km (</= 14 inches per mile)  222.1  252 mm/km (14.1  16 inches per mile)  252.1  378 mm/km (16.1  24 inches per mile)  >378 mm/km (>24 inches per mile) 
CT  ARAN inertial profiler  IRI  0.16 km^{a} (0.1 mi)  >950 mm/km^{a} (<60 inches per mile)  950  1260 mm/km^{a} (60  80 inches per mile)  1261  1894 mm/km^{a} (80.1  120 inches per mile)  >1894 mm/km^{a} (>120 inches per mile) 
DE  Rolling straightedge             
FL  Rolling straightedge             
GA  Inertial profiler  IRI  1.6 km(1.0 mi)    </= 750 mm/km (</= 47.5 inches per mile)^{a}    >750 mm/km (>47.5 inches per mile)^{a} 
HI               
ID  Californiatype profilograph  PI_{5mm}  0.1 km (0.1 mi)    </= 8 mm/0.1km (</= 0.5 in/0.1mi)    >8 mm/0.1km (>0.5 in/0.1 mi) 
IL  Californiatype profilograph  PI_{5mm}  0.16 km (0.1 mi)  </= 8 mm/km (</= 0.5 inches per mile)^{b}  9  160 mm/km (0.6  10 inches per mile)  161  235 mm/km (10.1  15 inches per mile)  >235 mm/km (>15 inches per mile) 
IN  Californiatype profilograph  PI_{5mm}  0.16 km (0.1 mi)    </= 30 mm/0.16 km (</= 1.2 in/0.1 mi)  31  38 mm/0.16 km (1.21  1.5 in/0.1 mi)  >38 mm/0.16 km (>1.5 in/0.1 mi) 
IA  Californiatype profilograph  PI_{5mm}  0.16 km (0.1 mi)  </= 48 mm/km (</= 3 inches per mile)  49  110 mm/km (3.1  7 inches per mile)  111  160 mm/km (7.1  10 inches per mile)  >160 mm/km (>10 inches per mile) 
KS  Californiatype profilograph  PI_{0.0}  0.1 km (0.1 mi)  </= 160 mm/km (</= 10 inches per mile)  161  475 mm/km (10.1  30 inches per mile)  476  630 mm/km (30.1  40 inches per mile)^{c}  >630 mm/km (>40 inches per mile) 
KY  Inertial profiler  RI  1.6 km^{a} (1.0 mi)  RI >/= 4.05  3.70 </= RI < 4.05  3.45 </= RI < 3.70  RI < 3.45 
LA  Californiatype profilograph  PI_{5mm}  Lot    </= 47 mm/km (</= 3 inches per mile)  48  95 mm/km (3.1  6 inches per mile)  >95 mm/km (>6 inches per mile) 
ME  Rolling dipstick profiler  IRI  0.2 km (0.12 mi)  </= 945 mm/km^{a} (</= 60 inches per mile)  946  1105 mm/km^{a} (60.1  70 inches per mile)  1106  1260 mm/km^{a} (70.1  80 inches per mile)  >1260 mm/km^{a} (>80 inches per mile) 
MD  Californiatype profilograph  PI_{5mm}  0.16 km^{a} (0.1 mi)  </= 63 mm/km^{a} (</= 4.0 inches per mile)  64  110 mm/km^{a} (4.1  7 inches per mile)  111  190 mm/km^{a} (7.1  12 inches per mile)  >191 mm/km^{a} (>12 inches per mile) 
MA  Inertial Profiler  IRI  0.2 km (0.12 mi)^{a}  *  *  *  * 
MI  Californiatype profilograph or GMtype inertial profiler  PI_{5mm}RQI^{d}  0.16 km^{a} (0.1 mi)  </= 63 mm/km^{a} (</= 4 inches per mile)or RQI <45  64  158 mm/km^{a} (4.1  10 inches per mile) or 45 </= RQI </= 53    >158 mm/km^{a} (>10 inches per mile)or RQI > 53 
MN  Californiatype profilograph  PI_{5mm}  0.1 km (0.1 mi)  </= 38.7 mm/km (</= 2.4 inches per mile)  38.8  78.9 mm/km (2.5  5 inches per mile)  79  118.3 m/km (5.1  7.5 inches per mile)  >118.3 mm/km (>7.5 inches per mile) 
MS  Californiatype profilograph  PI_{5mm}  0.16 km^{a} (0.1 mi)  </= 79 mm/km^{a} (</= 5 inches per mile)  80  110 mm/km^{a} (5.1  7 inches per mile)  111  158 m/km^{a} (7.1  10 inches per mile)  >158 mm/km^{a} (>10 inches per mile) 
MO  Californiatype profilograph  PI_{0.0}  0.1 km (0.1 mi)  </= 284 mm/km (</= 18 inches per mile)  285  395 mm/km (18.1  25 inches per mile)  396  711 m/km (25.1  45 inches per mile)  >712 mm/km (>45 inches per mile) 
MT               
NE  Californiatype profilograph  PI_{5mm}  0.2 km (0.1 mi)  </= 75 mm/km (</= 5 inches per mile)  76  110 mm/km (5.1  7 inches per mile)  111  155 mm/km (7.1  10 inches per mile)  >155 mm/km (>10 inches per mile) 
NV  Californiatype profilograph  PI_{5mm}  0.1 km (0.1 mi)    </= 80 mm/km (</= 5 inches per mile)    >80 mm/km (>5 inches per mile) 
NH  GMtypeinertial profiler  RN  0.16 km^{a} (0.1 mi)  **  **  **  ** 
NJ  Rolling straightedge             
NM  Californiatype profilograph  PI_{5mm}  0.1 km (0.1 mi)  </= 65 mm/km (</= 4 inches per mile)  66  80 mm/km (4.1  5 inches per mile)  81  160 m/km (5.1  10 inches per mile)  >160 mm/km (>10 inches per mile) 
NC  Hearne straightedge  CSI  0.76 km (0.47 mi)  CSI=10,20  CSI=30,40  CSI=11,21,31,41,50,51,60,61   
ND               
OH  Californiatype profilograph  PI_{5mm}  0.16 km^{a} (0.1 mi)  </= 63 mm/km^{a} (</= 4 inches per mile)  64  110 mm/km^{a} (4.1  7 inches per mile)  111  190 m/km^{a} (7.1  12 inches per mile)  >190 mm/km^{a} (>12 inches per mile) 
OK  Californiatype profilograph  PI_{5mm}  0.16 km^{a} (0.1 mi)  </= 79 mm/km^{a} (</= 5 inches per mile)  80  110 mm/km^{a} (5.1  7 inches per mile)  111  190 m/km^{a} (7.1  12 inches per mile)  >190 mm/km^{a} (>12 inches per mile) 
OR  Californiatype profilograph  PI_{5mm}  016 km^{a} (0.1 mi)  </= 80 mm/km^{a} (</= 5 inches per mile)  81  110 mm/km^{a} (5.1  7 inches per mile)  111  155 mm/km^{a} (7.1  10 inches per mile)  >155 mm/km^{a} (>10 inches per mile) 
PA  Californiatype profilograph  PI_{0.0}  0.16 km^{a} (0.1 mi)  </= 442 mm/km^{a} (</= 28 inches per mile)  443  536 mm/km^{a} (28.1  34 inches per mile)  537  726 mm/km^{a} (34.1  46 inches per mile)  >726 mm/km^{a} (>46 inches per mile) 
PR  Californiatype profilograph  PI_{5mm}  0.16 km^{a} (0.1 mi)  </= 110 mm/km^{a} (</= 7 inches per mile)  111  205 mm/km^{a} (7.1  13 inches per mile)    >205 mm/km^{a} (>13 inches per mile) 
RI               
SC  Maysmeter  MRN  1.6 km^{a} (1.0 mi)  </= 552 mm/km^{a} (</= 35 inches per mile)  553  630 mm/km^{a} (35.1  40 inches per mile)  631  868 mm/km^{a} (40.1  55 inches per mile)  >868 mm/km^{a} (>55 inches per mile) 
SD  Inertial profiler  IRI  0.16 km^{a} (0.1 mi)  </= 868 mm/km^{a} (</= 55 inches per mile)  869  1105 mm/km^{a} (55.1  70 inches per mile)  1106  1262 mm/km^{a} (70.1  80 inches per mile)  >1262 mm/km^{a} (>80 inches per mile) 
TN  Maysmeter  MRN  1.6 km^{a} (1.0 mi)  </= 315 mm/km a (</= 20 inches per mile)  316  475 mm/km^{a} (20.1  30 inches per mile)  476  950 mm/km^{a} (30.1  60 inches per mile)  >950 mm/km^{a} (>60 inches per mile) 
TX  Californiatype profilograph  PI_{0.0}  0.16 km^{a} (0.1 mi)  </= 237 mm/km^{a} (</= 15 inches per mile)  238  315 mm/km^{a} (15.1  20 inches per mile)  316  630 m/km^{a} (20.1  40 inches per mile)  >630 mm/km^{a} (>40 inches per mile) 
UT  Californiatype profilograph  PI_{5mm}  0.2 km (0.12 mi) a    </= 110 mm/km (</=7 inches per mile) a    >110 mm/km (>7 inches per mile)^{a} 
VT  Maysmeter  IRI  0.32 km^{a} (0.2 mi)  <950 mm/km^{a} (<60 inches per mile)  950  1090 mm/km^{a} (60  69 inches per mile)  1091  1500 mm/km^{a} (70  95 inches per mile)  >1500 mm/km^{a} (>95 inches per mile) 
VA  South Dakotatype profiler  IRI  0.16 km^{a} (0.1 mi)  </= 868 mm/km^{a} (</= 55 inches per mile)  869  1105 mm/km^{a} (55.1  70 inches per mile)  1106  1578 km^{a} (70.1  100 inches per mile)  >1578 mm/km^{a} (>100 inches per mile) 
WA^{e}  Lightweight inertial profiler  IRI  0.1 km (0.1 mi)^{a}  </= 946 mm/km^{a} (</= 60 inches per mile)  947  1500 mm/km^{a} (60.1  95 inches per mile)  1501  1815 mm/km^{a} (95.1  115 inches per mile)  >1815 mm/km^{a} (>115 inches per mile) 
WV  Maysmeter or inertial profiler  MRN  0.16 km (0.1 mi)    </= 1000 mm/km (</= 65 inches per mile)  1001  1500 mm/km (66  97.5 inches per mile)  >1500 mm/km (>97.5 inches per mile) 
WI  Californiatype profilograph  PI_{5mm}  0.16 km^{a} (0.1 mi)    </= 158 mm/km^{a} (</= 10 inches per mile)  159  237 m/km^{a} (10.1  15 inches per mile)  >237 mm/km^{a} (>15 inches per mile) 
WY  Inertial profiler  IRI  0.16 km^{a} (0.1 mi)  ***  ***  ***  *** 
*Percent Within Limits Specification: Upper Spec Limit = 1500 m/km (95 inches per mile)
**Percent Within Limits Specification: Lower Spec Limit = RN = 4.1
***Statistical Based Specification: Full Pay approximately equal to 8681105 mm/km (5570 inches per mile)
^{a} Limits are a direct EnglishMetric conversion from counterpart limits. Actual limits given by the Agency were not available.
^{b} Based on average profile index for entire project.
^{c} For PI between 476 mm/km (30.1 inches per mile) and 630 mm/km (40 inches per mile), must also grind to 475 mm/km (30 inches per mile) or below.
^{d} RQI: Ride quality index.
^{e} Draft specification.
State  Testing Device  Index  Testing Interval  Bonus Range  Full Pay Range  Penalty Range  Correction Range 

AL  Californiatype profilograph  PI_{5mm}  0.16 km (0.1 mi)  <45 mm/km (<3 inches per mile)  45  94 mm/km (3  5.9 inches per mile)  95  160 mm/km (6  10 inches per mile)  >160 mm/km (>10 inches per mile) 
AK               
AZ  Californiatype profilograph  PI_{5mm}  0.16 km^{a}(0.1 mi)  <110 mm/km^{a}(<7 inches per mile)  110  142 mm/km^{a}(7  9 inches per mile)    >142 mm/km^{a}(>9 inches per mile) 
AR  Californiatypeprofilograph, lightweight profiler  PI_{5mm}  0.2 km (0.1 mi)  </= 90 mm/km (</= 6 inches per mile)  91  110 mm/km (6.1  7 inches per mile)    >110 mm/km (>7 inches per mile) 
CA  Californiatype profilograph  PI_{5mm}  0.1 km (0.06 mi)^{a}    </= 110 mm/km (</= 7 inches per mile)^{a}    >110 mm/km (>7 inches per mile)^{a} 
CO  Californiatype profilograph  PI_{2.5mm}  0.15 km (0.095 mi)  </= 222 mm/km^{a} (</= 14 inches per mile)  222.1  252 mm/km^{a} (14.1  16 inches per mile)  252.1  378 mm/km^{a} (16.1  24 inches per mile)  >378 mm/km^{a} (>24 inches per mile) 
CT  Californiatype profilograph  PI_{5mm}  0.15 km (0.1 mi)^{a}  </=160 mm/km (10 inches per mile)^{a}  161  190 mm/km (10.1  12 inches per mile)^{a}  191  315 mm/km (12.1  20 inches per mile)^{a}  >315 mm/km (>20 inches per mile)^{a} 
DE  CA profilograph or rolling straightedge  PI_{5mm}  0.16 km^{a}(0.1 mi)  <50 mm/km^{a}(<3.2 inches per mile)  50  200 mm/km^{a}(3.2  12.7 inches per mile)    >200 mm/km^{a}(>12.7 inches per mile) 
FL  Californiatype profilograph  PI_{5mm}  0.1 km (0.1 mi)  </= 80 mm/km (</= 5 inches per mile)  81  95 mm/km (5.1  6 inches per mile)  96  110 mm/km (6.1  7 inches per mile)  >110 mm/km (>7 inches per mile) 
GA  Rainhart profilograph  PI_{2.5mm}  0.4 km^{a}(0.25 mi)    </= 110 mm/km^{a}(</= 7 inches per mile)    >110 mm/km^{a}(>7 inches per mile) 
HI  Californiatype profilograph  PI_{5mm}  0.16 km^{a}(0.1 mi)    </= 157 mm/km^{a}(</= 10 inches per mile)  158  236 mm/km^{a}(10.1  15 inches per mile)  >236 mm/km^{a}(>15 inches per mile) 
ID  Californiatype profilograph  PI_{5mm}  0.1 km (0.1 mi)    </= 8 mm/0.1 km (</= 0.5 in/0.1mi)    >8 mm/0.1 km (>0.5 in/0.1mi) 
IL  Californiatype profilograph  PI_{5mm}  0.16 km (0.1 mi)  </= 67 mm/km (</= 4.25 inches per mile) b  68  160 mm/km (4.26  10 inches per mile)  161  235 mm/km (10.01  15 inches per mile)  >235 mm/km (>15 inches per mile) 
IN  Californiatype profilograph  PI_{5mm}  0.16 km (0.1 mi)  </= 23mm/0.16 km (</= 0.9 in/0.1mi)  23  25 mm/0.16km (0.9  1.0 in/0.1 mi)    >25 mm/0.16 km (>1.0 in/0.1 mi) 
IA  Californiatype profilograph  PI_{5mm}  0.16 km (0.1 mi)  </= 48 mm/km (Â£3 inches per mile)  49  110 mm/km (3.1  7 inches per mile)  111  160 mm/km (7.1  10 inches per mile)  >160 mm/km (>10 inches per mile) 
KS  Californiatype profilograph  PI_{0.0}  0.1 km (0.1 mi)  </= 285 mm/km (</= 18 inches per mile)  286  475 mm/km (18.1  30 inches per mile)  476  630 mm/km (30.1  40 inches per mile)^{c}  >630 mm/km (>40 inches per mile) 
KY  Rainhart profilographand inertial profiler  PI_{2.5mm}RI  0.3 km^{a}(0.19 mi)  RI >/= 4.05  </=125 mm/km^{a} (</= 8 inches per mile)  126  190 mm/km^{a}(8.1  12 inches per mile)  >190 mm/km^{a}(>12 inches per mile) 
LA  Californiatype profilograph  PI_{5mm}  Lot    Â£94 mm/km (</= 6 inches per mile)  95  126 mm/km (6.1  8 inches per mile)  >126 mm/km (>8 inches per mile) 
ME               
MD  Californiatype profilograph  PI_{5mm}  0.16 km^{a}(0.1 mi)  </= 63 mm/km^{a} (</= 4.0 inches per mile)  64  110 mm/km^{a}(4.1  7 inches per mile)  111  190 mm/km^{a}(7.1  12 inches per mile)  >191 mm/km^{a}(>12 inches per mile) 
MA               
MI  Californiatype profilograph or GMtype inertial profiler  PI_{5mm}RQI d  0.16 km^{a}(0.1 mi)  </= 63 mm/km^{a} (</= 4 inches per mile)or RQI < 45  64  158 mm/km^{a} (4.1  10 inches per mile)or 45 </=RQI </= 53    >158 mm/km^{a}(>10 inches per mile)or RQI > 53 
MN  Californiatype profilograph  PI_{5mm}  0.16 km (0.1 mi)  </= 63 mm/km (</= 4 inches per mile)  64  94 mm/km (4.1  6 inches per mile)  95  126 m/km (6.1  8 inches per mile)  >126 mm/km (>8 inches per mile) 
MS  Californiatype profilograph  PI_{5mm}  0.16 km^{a}(0.1 mi)    </= 110 mm/km^{a}(</= 7 inches per mile)  111  190 m/km^{a}(7.1  12 inches per mile)  >190 mm/km^{a}(>12 inches per mile) 
MO  Californiatype profilograph  PI_{0.0}  0.1 km (0.1 mi)  </= 284 mm/km (</= 18 inches per mile)  285  395 mm/km (18.1  25 inches per mile)  396  711 m/km (25.1  45 inches per mile)  >712 mm/km (>45 inches per mile) 
MT  Californiatype profilograph  PI_{5mm}  0.16 km^{a}(0.1 mi)  </= 94 mm/km^{a}(Â£6 inches per mile)  95  158 mm/km^{a}(6.1  10 inches per mile)  159  237 m/km^{a} (10.1  15 inches per mile)  >237 mm/km^{a}(>15 inches per mile) 
NE  Californiatype profilograph  PI_{5mm}  0.2 km (0.1 mi)  </= 75 mm/km (</= 5 inches per mile)  76  155 mm/km (5.1  10 inches per mile)  156  230 mm/km (10.1  15 inches per mile)  >230 mm/km (>15 inches per mile) 
NV  Californiatype profilograph  PI_{5mm}  0.1 km (0.1 mi)    </= 80 mm/km (</= 5 inches per mile)    >80 mm/km (>5 inches per mile) 
NH               
NJ  Rolling straightedge             
NM  Californiatype profilograph  PI_{5mm}  0.1 km (0.1 mi)  </= 80 mm/km (</= 5 inches per mile)  81  110 mm/km (5.1  7 inches per mile)  111  190 m/km (7.1  12 inches per mile)  >190 mm/km (>12 inches per mile) 
NY  Californiatype profilograph  PI_{5mm}  0.16 km (0.1 mi)^{a}  </= 79 mm/km^{a}(</= 5 inches per mile)  80  190 mm/km^{a}(5.1  12 inches per mile)    >190/km^{a}(>12 inches per mile) 
NC  Rainhart profilograph  PI_{5mm}  0.18 km^{a}(0.11 mi)    </= 63 mm/km^{a}(</= 4 inches per mile)    >63 mm/km^{a}(>4 inches per mile) 
ND  Californiatype profilograph  PI_{5mm}  0.16 km^{a}(0.1 mi)  <8mm/0.16 km^{a}(<0.3 in/0.1mi)  8  13 mm/0.16 km^{a}(0.3  0.5 in/0.1mi)  14  23 mm/0.16 km^{a}(0.51  0.9 in/0.1mi)  >23 mm/0.16 km^{a}(>0.9 in/0.1mi) 
OH  Californiatype profilograph  PI_{5mm}  0.16 km^{a}(0.1 mi)  </= 78 mm/km^{a}(</= 5 inches per mile)  79  110 mm/km^{a}(5.1  7 inches per mile)  111  190 m/km^{a}(7.1  12 inches per mile)  >190 mm/km^{a}(>12 inches per mile) 
OK  Californiatype profilograph  PI_{5mm}  0.16 km^{a}(0.1 mi)  </= 79 mm/km^{a}(</= 5 inches per mile)  80  110 mm/km^{a}(5.1  7 inches per mile)  111  190 m/km^{a}(7.1  12 inches per mile)  >190 mm/km^{a}(>12 inches per mile) 
OR  Californiatype profilograph  PI_{5mm}  0.2 km (0.1 mi)^{a}  </= 80 mm/km (</= 5 inches per mile)^{a}  81  110 mm/km (5.1  7 inches per mile)^{a}    >110 mm/km (>7 inches per mile)^{a} 
PA  Californiatype profilograph  PI_{0.0}  0.16^{a}(0.1 mi)  </= 568 mm/km^{a}(</= 36 inches per mile)      >568 mm/km^{a}(>36 inches per mile) 
PR  Californiatype profilograph  PI_{5mm}  0.16 km^{a}(0.1 mi)  </= 110 mm/km^{a}(</= 7 inches per mile)  111  205 mm/km^{a}(7.1  13 inches per mile)    >205 mm/km^{a}(>13 inches per mile) 
RI               
SC  Rainhart profilograph  PI_{5mm}  0.4 km^{a}(0.25 mi)    </= 158 mm/km^{a}(</= 10 inches per mile)    >158 mm/km^{a}(>10 inches per mile) 
SD  Californiatype profilograph  PI_{0.0}  0.1 km (0.1 mi)  </= 395 mm/km (</= 25 inches per mile)  396  550 mm/km (25.1  35 inches per mile)  551  630 mm/km (35.1  40 inches per mile)  >630 mm/km (>40 inches per mile) 
TN  Rainhartprofilograph  PI_{2.5mm}  0.1 km (0.1 mi)    </= 160 mm/km (</= 10 inches per mile)  161  235 mm/km (10.1  15 inches per mile)  >235 mm/km (>15 inches per mile) 
TX  Californiatype profilograph  PI_{0.0}  0.16 km^{a}(0.1 mi)  237 mm/km^{a}(</= 15 inches per mile)  238  315 mm/km^{a}(15.1  20 inches per mile)  316  630 m/km^{a}(20.1  40 inches per mile)  >630 mm/km^{a}(>40 inches per mile) 
UT  Californiatype profilograph  PI_{5mm}  0.2 km (0.12 mi)^{a}    </= 110 mm/km (</= 7 inches per mile)^{a}    >110 mm/km (>7 inches per mile) a 
VT               
VA  South Dakotatype profiler  IRI  0.16 km^{a} (0.1 mi)  </= 946 mm/km^{a} (</= 60 inches per mile)  947  1262 mm/km^{a} (60.1  80 inches per mile)  1263  1578 km a(80.1  100 inches per mile)  >1578 mm/km^{a} (>100 inches per mile) 
WA  Californiatype profilograph  PI_{7.5mm}  0.1 km (0.1 mi) a  </= 60 mm/km (</= 3.8 inches per mile)^{a}  61  100 mm/km (3.9  6.3 inches per mile) a  >100 mm/km (>6.3 inches per mile) a,e   
WV  Maysmeter or inertial profiler  MRN  0.16 km (0.1 mi)    </= 1000 mm/km (</= 65 inches per mile)  1001  1500 mm/km (66  97.5 inches per mile)  >1500 mm/km (>97.5 inches per mile) 
WI  Californiatype profilograph  PI_{01inch}  0.16 km^{a} (0.1 mi)  </= 400 mm/km^{a} (</= 25.3 inches per mile)  401  700 mm/km^{a} (25.4  44.3 inches per mile)  701  800 m/km^{a} (44.4  50.7 inches per mile) f  >800 mm/km^{a} (>50.7 inches per mile) 
WY  Californiatype profilograph  PI_{5mm}  *  *  *  *  * 
* Perf. Related Spec (PCC thickness, strength, smoothness) >80 mm/km (>5.0 inches per mile).
^{a} Limits are a direct EnglishMetric conversion from counterpart limits. Actual limits given by the agency were not available.
^{b} Based on average profile index for entire project.
^{c} For PI between 476 mm/km (30.1 inches per mile) and 630 mm/km (40 inches per mile), must also grind to 475 mm/km (30 inches per mile) or below.
^{d} RQI: Ride quality index.
^{e} For PI greater than 100 mm/km (6.3 inches per mile), must also grind to 100 mm/km (6.3 inches per mile) or less.
^{f} For PI greater than 700 mm/km (44.3 inches per mile), must also grind to 700 mm/km (44.3 inches per mile) or less.
PI_{5mm} Specification:
Agency  Existing FullPay Range, mm/km  Climatic Zone  Estimated PI_{0.0} FullPay Range, mm/km  SEE, mm/km^{a} Estimated PI_{0.0}  Estimated IRI FullPay Range, mm/km  Estimated IRI SEE, mm/km^{a} 

AL  32  63  WNF  287  332  72  1009  1126  292 
AR  46  75  WF,WNF  307  350  72  1062  1171  232 
CA  </= 80  DNF,WF,WNF  </= 357  72  </= 1190  292 
ID  </= 80^{b}  DF,WF  </= 384  84  </= 1190  292 
IL  9  160  WF  288  493  84  922  1493  292 
IN  </= 187^{c}  WF  </= 529  84  </= 1595  292 
IA  49  110  WF  342  425  84  1073  1304  292 
LA  </= 47  WNF  </= 309  72  </= 1065  292 
MD  64  110  WF  362  425  84  1130  1304  292 
MI  64  158  WF  362  490  84  1130  1486  292 
MN  38.8  78.9  WF  328  383  84  1035  1187  292 
MS  80  110  WNF  357  401  72  1190  1304  292 
NE  76  110  DF,WF  379  425  84  1175  1304  292 
NV  </= 80  DF  < 384  84  < 1190  292 
NM  66  80  DF,DNF  337  357  84  1137  1190  292 
OH  64  110  WF  362  425  84  1130  1304  292 
OK  80  110  DF,WF,WNF  357  401  72  1190  1304  292 
OR  81 110  DF,WNF  385  425  84  1194  1304  292 
PR  111  205  WNF  403  540  72  1308  1664  292 
UT  </= 110  DF,WF  < 425  84  < 1304  292 
WI  </= 158  WF  < 490  84  < 1486  292 
PI_{2.5mm} Specification:
Agency  Existing FullPay Range, mm/km  Climatic Zone  Estimated PI_{0.0} FullPay Range, mm/km  SEE, mm/km^{a} Estimated PI_{0.0}  Estimated IRI FullPay Range, mm/km  Estimated IRI SEE, mm/km^{a} 

CO  222.1  252  DF,WF  415  446  48  1295  1381  232 
PI_{0.0} Specification:
Agency  Existing FullPay Range, mm/km  Climatic Zone  Estimated PI_{0.0} FullPay Range, mm/km  SEE, mm/km^{a} Estimated PI_{0.0}  Estimated IRI FullPay Range, mm/km  Estimated IRI SEE, mm/km^{a} 

KS  161  475  DF,WF  642  1479  200  
MO  285  395  WF  973  1266  200  
PA  443  536  WF  1394  1642  200  
TX  238  315  DF,DNF,WF,WNF  847  1053  200 
^{a} SEE = Standard error of the estimate. Range of values with 90 percent confidence.
^{b} Extrapolated from actual specification, which calls for PI </= 8 mm per 0.1 km.
^{c} Extrapolated from actual specification, which calls for PI </= 30 mm per 0.16 km.
PI_{5mm} Specification:
Agency  Existing FullPay Range, mm/km  Climatic Zone  Estimated PI_{0.0} FullPay Range, mm/km  SEE, mm/km^{a} Estimated PI_{0.0}  Estimated IRI FullPay Range, mm/km  Estimated IRI SEE, mm/km^{a} 

AL  32  63  WNF  280  325  72  986  1093  266 
AR  46  75  WF,WNF  301  343  72  1034  1134  217 
CA  </= 80  DNF,WF,WNF  < 364  79  < 1235  308 
ID  </= 80^{b}  DF,WF  < 376  74  < 1298  288 
IL  9  160  WF  247  467  72  908  1425  266 
IN  </= 187^{c}  WF  < 506  72  < 1518  266 
IA  49  110  WF  305  394  72  1045  1254  266 
LA  </= 47  WNF  < 302  72  < 1038  266 
MD  64  110  WF  327  394  72  1096  1254  266 
MI  64  158  WF  327  464  72  1096  1418  266 
MN  38.8  78.9  WF  290  349  72  1010  1148  266 
MS  80  110  WNF  350  394  72  1151  1254  266 
NE  76  110  DF,WF  369  423  74  1281  1426  288 
NV  </= 80  DF  < 376  74  < 1298  288 
NM  66  80  DF,DNF  339  364  79  1173  1235  308 
OH  64  110  WF  327  394  72  1096  1254  266 
OK  80  110  DF,WF,WNF  350  394  72  1151  1254  266 
OR  81 110  DF,WNF  377  423  74  1302  1426  288 
PR  111  205  WNF  396  533  72  1257  1579  266 
UT  </= 110  DF,WF  < 423  74  < 1426  288 
WI  </= 158  WF  < 464  72  < 1418  266 
PI_{2.5mm} Specification:
Agency  Existing FullPay Range, mm/km  Climatic Zone  Estimated PI_{0.0} FullPay Range, mm/km  SEE, mm/km^{a} Estimated PI_{0.0}  Estimated IRI FullPay Range, mm/km  Estimated IRI SEE, mm/km^{a} 

CO  222.1  252  DF,WF  413  446  43  1399  1490  230 
PI_{0.0} Specification:
Agency  Existing FullPay Range, mm/km  Climatic Zone  Estimated PI_{0.0} FullPay Range, mm/km  SEE, mm/km^{a} Estimated PI_{0.0}  Estimated IRI FullPay Range, mm/km  Estimated IRI SEE, mm/km^{a} 

KS  161  475  DF,WF  708  1570  191  
MO  285  395  WF  992  1259  179  
PA  443  536  WF  1375  1601  179  
TX  238  315  DF,DNF,WF,WNF  913  1119  185 
^{a} SEE = Standard error of the estimate. Range of values with 90 percent confidence.
^{b} Extrapolated from actual specification, which calls for PI </= 8 mm per 0.1 km.
^{c} Extrapolated from actual specification, which calls for PI </= 30 mm per 0.16 km.
PI_{5mm} Specification:
Agency  Existing FullPay Range, mm/km  Climatic Zone  Estimated PI_{0.0} FullPay Range, mm/km  SEE, mm/km^{a} Estimated PI_{0.0}  Estimated IRI FullPay Range, mm/km  Estimated IRI SEE, mm/km^{a} 

AL  32  63  WNF  323  373  71  1066  1188  260 
AR  46  75  WF,WNF  345  392  71  1121  1235  260 
CA  </= 80  DNF,WF,WNF  < 400  71  < 1255  260 
ID  </= 80^{b}  DF,WF  < 400  71  1255  260 
IL  9  160  WF  286  529  71  975  1571  260 
IN  </= 187^{c}  WF  < 572  71  < 1677  260 
IA  49  110  WF  350  448  71  1133  1373  260 
LA  </= 47  WNF  < 347  71  < 1125  260 
MD  64  110  WF  374  448  71  1192  1373  260 
MI  64  158  WF  374  526  71  1192  1563  260 
MN  38.8  78.9  WF  334  398  71  1093  1251  260 
MS  80  110  WNF  400  448  71  1255  1373  260 
NE  76  110  DF,WF  394  448  71  1239  1373  260 
NV  </= 80  DF  < 400  71  < 1255  260 
NM  66  80  DF,DNF  377  400  71  1200  1255  260 
OH  64  110  WF  374  448  71  1192  1373  260 
OK  80  110  DF,WF,WNF  400  448  71  1255  1373  260 
OR  81 110  DF,WNF  402  448  71  1259  1373  260 
PR  111  205  WNF  450  601  71  1377  1748  260 
UT  </= 110  DF,WF  < 448  71  < 1373  260 
WI  </= 158  WF  < 526  71  < 1563  260 
PI_{2.5mm} Specification:
Agency  Existing FullPay Range, mm/km  Climatic Zone  Estimated PI_{0.0} FullPay Range, mm/km  SEE, mm/km^{a} Estimated PI_{0.0}  Estimated IRI FullPay Range, mm/km  Estimated IRI SEE, mm/km^{a} 

CO  222.1  252  DF,WF  432  466  45  1332  1412  230 
PI_{0.0} Specification:
Agency  Existing FullPay Range, mm/km  Climatic Zone  Estimated PI_{0.0} FullPay Range, mm/km  SEE, mm/km^{a} Estimated PI_{0.0}  Estimated IRI FullPay Range, mm/km  Estimated IRI SEE, mm/km^{a} 

KS  161  475  DF,WF  680  1434  206  
MO  285  395  WF  978  1242  206  
PA  443  536  WF  1357  1581  206  
TX  238  315  DF,DNF,WF,WNF  865  1050  206 
^{a} SEE = Standard error of the estimate. Range of values with 90 percent confidence.
^{b} Extrapolated from actual specification, which calls for PI </= 8 mm per 0.1 km.
^{c} Extrapolated from actual specification, which calls for PI </= 30 mm per 0.16 km.
PI_{5mm} Specification:
Agency  Existing FullPay Range, mm/km  Climatic Zone  Estimated PI_{0.0} FullPay Range, mm/km  SEE, mm/km^{a} Estimated PI_{0.0}  Estimated IRI FullPay Range, mm/km  Estimated IRI SEE, mm/km^{a} 

AL  45  94  WNF  444  503  85  1359  1500  297 
AZ  110  142  DF,DNF  464  507  66  1613  1726  269 
AR  91  110  WF, WNF  500  522  85  1491  1546  297 
CA  </= 110  DNF,WF,WNF  </= 464  66  </= 1613  269 
CT  161  190  WF  562  597  87  1628  1705  306 
DE  50  200  WF  428  609  87  1337  1731  306 
FL  81  95  WNF  382  587  66  1402  1930  269 
HI  </= 157  WNF  </= 579  85  </= 1681  297 
ID  </= 80 b  DF,WF  </= 445  71  </= 1416  306 
IL  68  160  WF  450  561  87  1384  1626  306 
IN  144  156 c  WF  542  556  87  1584  1615  306 
IA  49  110  WF  427  501  87  1334  1495  306 
LA  </= 94  WNF  </= 503  85  </= 1500  297 
MD  64  110  WF  445  501  87  1374  1495  306 
MI  64  158  WF  445  559  87  1374  1621  306 
MN  64  94  WF  445  481  87  1374  1453  306 
MS  </= 110  WNF  </= 522  85  </= 1546  297 
MT  95  158  WF  483  559  87  1455  1621  306 
NE  76  155  DF,WF  449  559  71  1405  1613  306 
NV  </= 80  DF  </= 455  71  </= 1416  306 
NM  81  110  DF,DNF  424  464  66  1511  1613  269 
NY  80  190  WF  464  597  87  1416  1705  306 
NC  </= 63  WF,WNF  </= 466  85  </= 1411  297 
ND  50  81 d  DF,WF  413  456  71  1337  1418  306 
OH  79  110  WF  463  501  87  1413  1495  306 
OK  80  110  DNF,WF,WNF  486  522  85  1460  1546  297 
OR  81 110  DF,WNF  456  497  71  1418  1495  306 
PR  111  205  WNF  524  636  85  1549  1819  297 
SC  </= 158  WNF  </= 580  85  </= 1684  297 
UT  </= 110  DF,WF  </= 497  71  </= 1495  306 
WY  </= 80  DF  </= 455  71  </= 1416  306 
PI_{2.5mm} Specification:
Agency  Existing FullPay Range, mm/km  Climatic Zone  Estimated PI_{0.0} FullPay Range, mm/km  SEE, mm/km^{a} Estimated PI_{0.0}  Estimated IRI FullPay Range, mm/km  Estimated IRI SEE, mm/km^{a} 

CO  222.1  252  DF,WF  469  499  47  1423  1485  279 
GA  </= 110  WF,WNF  </= 469  57  </= 1420  264 
KY  </= 125  WF  </= 365  50  </= 1216  279 
TN  </= 160  WF,WNF  </= 407  57  </= 1273  317 
PI_{0.0} Specification:
Agency  Existing FullPay Range, mm/km  Climatic Zone  Estimated PI_{0.0} FullPay Range, mm/km  SEE, mm/km^{a} Estimated PI_{0.0}  Estimated IRI FullPay Range, mm/km  Estimated IRI SEE, mm/km^{a} 

KS  286  475  DF,WF  1047  1448  306  
MO  285  395  WF  1044  1278  306  
PA  568 e  WF  </= 1645  306  
TX  396  550  DF,DNF,WF,WNF  1038  1237  269  
WI  401  700  WF  1291  1925  306 
^{a} SEE = Standard error of the estimate. Range of values with 90 percent confidence. ^{b} Extrapolated from actual specification, which calls for PI </= 8 mm per 0.1 km. ^{c} Extrapolated from actual specification, which calls for 23 </= PI </= 25 mm per 0.16 km. ^{d} Extrapolated from actual specification, which calls for 8 </= PI </= 13 mm per 0.16 km. ^{e} Actual specification calls for incentives for PI</=568 mm/km (36 inches per mile) and correction for PI>568 inches per mile (36 inches per mile).
For example, if Maryland (a wetfreeze climatic zone state) entertained thoughts of switching from PI_{5mm} to PI_{0.0}, it could refer to table 16 to identify the comparable PI_{0.0} range for the current PI_{5mm} fullpay range 64 to 110 mm/km (4 to 7 inches per mile). This range is estimated to be 321 to 403 mm/km (20.3 to 25.5 inches per mile). With a standard error of the estimate (SEE) of 72 mm/km (4.6 inches per mile) for this relationship, the specification writer can assume that variability within the relationship results in a reasonable range for comparable PI_{0.0} values of 321 to 403 mm/km (20.3 to 25.5 inches per mile).
If, on the other hand, the agency desired to transition to an IRI specification, it could use the comparable IRI range of 1,096 to 1,254 mm/km [70 to 79.5 inches per mile]). The SEE for this relationship is 266 mm/km (16.9 inches per mile).
Direct StatetoState comparisons of derived specification limits may not be appropriate due to individual agencies' implementation practices. Factors that may affect the specification limits for a specific agency include segment length, whether an agency aggregates segments, scope of application (new pavements or overlays, and type facilities), and method of index computation (halfcar roughness index, individual wheelpath IRI, or average IRI).
Conclusions
In the search for reasonable, practical relationships that link IRI with PI_{5mm}, PI_{2.5mm}, and PI_{0.0}, and PI_{0.0} with PI_{5mm} and PI_{2.5mm}, a comprehensive evaluation was made of trends documented in past pavement smoothness studies, as well as trends developed in this study from vast amounts of LTPP profile and smoothness data. The background and results of these studies were presented and discussed in previous chapters of this report.
Although past documented PIIRI relationships were rather limited (particularly with respect to PI_{2.5mm}IRI relationships) and showed varying degrees of disparity, factors such as pavement type, equipment characteristics, and filtering methods contributed significantly to these disparities.
A much broader and more controlled evaluation using over 43,000 LTPP smoothness data points showed generally similar PIIRI trends as the past study trends. The data points consisted of IRI and simulated PI values computed from the same longitudinal profiles measured multiple times for 1,793 LTPP pavement test sections.
Detailed statistical analyses of IRI and simulated PI data indicated a reasonable correlation between IRI and PI (PI_{5mm}, PI_{2.5mm}, and PI_{0.0}) and between PI_{0.0} and PI (PI_{5mm} and PI_{2.5mm}). However, it was determined that pavement type (i.e., AC, JPC, AC/PCC) and climatic conditions (i.e., dryfreeze, wetnonfreeze) are significant factors in the relationship between IRI and PI.
The effects of these variables were taken into consideration in the development of PItoIRI and PItoPI conversion models. A total of 15 PItoIRI models and 18 PItoPI models covering all three PI blanking band sizes (5, 2.5, and 0 mm [0.2, 0.1, and 0 in]) and all four climatic zones (dryfreeze, drynonfreeze, wetfreeze, and wetnonfreeze) were developed for ACsurfaced pavements. Similarly, for PCCsurfaced pavements, 9 PItoIRI models and 12 PItoPI models were developed.
The equations, estimated standard errors, and other relevant statistics for all 54 models are summarized in tables 10 through 13 in chapter 4. These equations can be used to assist highway agency personnel in transitioning smoothness specification limits from PI to IRI or to PI with a tighter blanking band. Chapter 5 of this report illustrated the results of applying these conversion equations to existing State smoothness specifications. Each State's PI_{5mm}, PI_{2.5mm}, or PI_{0.0} fullpay range was converted to an estimated equivalent IRI range, and each State's PI_{5mm} or PI_{2.5mm} fullpay range was converted to an estimated equivalent PI_{0.0}. Results of this exercise are summarized in tables 16 through 19 in chapter 5.
Recommendations
The major goal of this research was to develop a practical tool to assist in the transition from PI to PI_{0.0} or IRI specifications. Correlation and error estimates have been provided to allow agencies to estimate the level of IRI and PI smoothness that is associated with their current specifications. To make this research useful, agencies are asked to:
Baus, R., and W. Hong. 1999. "Investigation and Evaluation of Roadway Rideability Equipment and Specifications," Federal Highway Administration/South Carolina Department of Transportation Report FHWASC9905, Columbia, South Carolina.
Fernando, E.G. 2000. "Evaluation of Relationship Between Profilograph and ProfileBased Roughness Indices," Paper Prepared for the 79th Annual Meeting of the Transportation Research Board, Washington, D.C.
Florida Department of Transportation (FLDOT). 1997. "Comparison of Ride Quality Monitoring Equipment Used for New Construction," Draft Report, Pavement Systems Evaluation Study 972, Florida Department of Transportation State Materials Office, Tallahassee, Florida.
Hossain, M., M. Akhter, J. Hancock, and J. Boyer. 2000. "Evaluation of Performance of Lightweight Profilometers," Final Report prepared for Kansas Department of Transportation, Kansas State University, Manhatten, Kansas.
Kombe, E.M. and S.A. Kalevela. 1993. "Evaluation of Initial Pavement Smoothness for the Development of PCCP Construction Specifications," Phase I Final Report, Arizona Department of Transportation, Phoenix, Arizona.
Ksaibati , K., R. McNamera, W. Miley, and J. Armaghani. 1999. "Pavement Roughness Data Collection and Utilization," Transportation Research Record 1655, Transportation Research Board, Washington, D.C.
Kulakowski, B.T. and J.C. Wambold. 1989. "Development of Procedures for the Calibration of Profilographs." Final Report, Pennsylvania Transportation Institute (PTI), Pennsylvania State University, University Park, Pennsylvania.
Rizzo, R. 2000 "Survey of State Smoothness Specifications," Federal Highway Administration, Michigan Division, Lansing, Michigan.
Rufino, D., K. Baraka, and M.I. Darter. 2000. "Development of a Bridge Smoothness Specification for Illinois DOT," Interim Report prepared for Illinois Department of Transportation, Report No. FHWAILUI279, University of Illinois, UrbanaChampaign, Illinois.
Scofield, L. 1993. "Profilograph Limitations, Correlations, and Calibration Criteria for Effective PerformanceBased Specifications," NCHRP Project 207 Task 53 Draft Report, Transportation Research Board, Washington, D.C.