|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
1 DEVELOPMENT OF A CONSTRUCTION Q UALITY INDEX AND AN INTEGRATED CONSTRUCTION QUALITY INDEX TO EVALUATE THE QUALITY OF PAVEMENT CONSTRUCTION PROJECTS By JUNYONG AHN A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2008
2 2008 Junyong Ahn
3 To my Wife
4 ACKNOWLEDGMENTS It is m y great pleasure to express my tha nks to the teachers and friends who supported me to complete this dissertation. First and foremost I would like to thank my graduate supervisory committee, all of whom have offered invaluab le support and encouragement throughout this process, but special thanks and gratitude must go to Dr. Ralph D. Ellis, my advisor and committee chair. He has guided, challenged, and enc ouraged me to complete the project. I wish to express gracious thanks to Dr. Zohar J. Herbsman, who demonstrated excellence in scholarship and provided construc tive ideas. I thank Dr. Michael I. Hammons for his advice on research design and opportunities he gave me to work with industries for the research. I also wish to express my gratitude to Dr. Robert C. Stroh for his invaluable comments and warm encouragement. My immense gratitude goes esp ecially to Dr. R. Edward Minchin, who opened my collaborative world. I appreciate his support throughout my graduate experience. I also wish to extend my gratit ude to Dr. Charles R. Glagola for his extra time and effort devoted to read my dissertation and constructive comments. I al so thank Dr. Gary Long for the opportunity to work with as his teaching assistant. I would like to thank my collea gues at Applied Research Asso ciates and personnel at State Material Office in Gainesville for the support they provided to access and retrieve the research data. Without their support and help, th is research could not be completed. To all my friends at UF, I thank them for en riching my experience in Gainesville. I thank them for sharing my happiness with them and thei r happiness with me. To my extended friends and family, who are too numerous to name he re, I thank them for calling or writing with supportive words at the moment when I need them most. I would like to express my most sincere a ppreciation to my parents for their love, encouragement, and support. Without their s upport, this achievement would not have been
5 possible. I wish to thank my children, Kyoungbin, Jaemin, and Jaeyon, who have enabled me to see the worlds differently. They have made me laugh everyday. And, finally, but not last, I dedicate this dissertati on to my wife, Heeja Lee, with love and respect.
6 TABLE OF CONTENTS page ACKNOWLEDGMENT..................................................................................................................4 LIST OF TABLES................................................................................................................. ..........8 LIST OF FIGURES.........................................................................................................................9 LIST OF ABBREVIATIONS........................................................................................................ 10 ABSTRACT...................................................................................................................................11 CHAP TER 1 INTRODUCTION..................................................................................................................13 1.1 Background.......................................................................................................................13 1.2 Objective...........................................................................................................................14 1.3 Scope.................................................................................................................................17 2 LITERATURE REVIEW.......................................................................................................19 2.1 Contractor Evaluation Systems of SHAs.......................................................................... 19 2.2 Performance-Related Specifications.................................................................................23 2.3 Analytic Hierarchy Process.............................................................................................. 28 2.4 Percent with in Lim its...................................................................................................... ..32 2.5 Acceptance Quality Characteristics.................................................................................. 33 2.6 Mechanistic-Empirical Pavement Design Guide (MEPDG)............................................ 33 2.7 Conclusions.......................................................................................................................35 3 METHODOLOGY................................................................................................................. 36 3.1 Data Collection.................................................................................................................36 3.1.1 Project List for Data Collection..............................................................................37 3.1.2 Expert Panel Surveys..............................................................................................39 18.104.22.168 Expert panel surveys for CQI.......................................................................39 22.214.171.124 Expert panel surveys for ICQI.....................................................................41 3.2 Model Formulation...........................................................................................................43 3.2.1 Construction Quality Index (CQI) Model.............................................................. 43 126.96.36.199 Model concepts............................................................................................ 43 188.8.131.52 Model weighting factors...............................................................................45 184.108.40.206 Adaptation of the model for more than one asphalt mix.............................. 48 220.127.116.11 Additional data required o ther than test results............................................ 49 18.104.22.168 Model implementation................................................................................. 50 3.2.2 Integrated Construction Quality Index (ICQI) Model ............................................ 50 22.214.171.124 Mechanistic-empirical pavement design guide (MEPDG)..........................50 126.96.36.199 Model concepts............................................................................................ 51
7 4 DATA ANALYSIS................................................................................................................ 71 4.1 Overview...........................................................................................................................71 4.2 Data Deduction............................................................................................................. ....76 4.3 Model Validation........................................................................................................... ...83 4.3.1 Construction Quality Index (CQI) Model.............................................................. 83 188.8.131.52 Validation process........................................................................................83 184.108.40.206 The CQI model validation through al l projects from every district............. 84 220.127.116.11 The CQI model validation by district........................................................... 88 4.3.2 Integrated Construction Quality Index (ICQI) Model ............................................ 89 18.104.22.168 Validation process........................................................................................90 22.214.171.124 The ICQI model validation by district......................................................... 93 5 CONCLUSIONS AND RECOMME NDATIONS................................................................. 97 5.1 Conclusions.......................................................................................................................97 5.2 Recommendations.............................................................................................................98 APPENDIX A FDOT PAVEMENT ACCEPTANCE QUALITY CHARACTERISTICS ..........................100 B PROJECT LIST....................................................................................................................104 C EXPERT PANEL MEETING FORMS................................................................................111 D TABULATION OF RESULTS FROM EXPERT PANEL MEETINGS............................. 114 E PWL TABLES..................................................................................................................... .118 F TARGET VALUE REPORT SAMPLE............................................................................... 121 LIST OF REFERENCES.............................................................................................................123 BIOGRAPHICAL SKETCH.......................................................................................................126
8 LIST OF TABLES Table page 2-1 Current project rating system s around the U.S.A..............................................................19 2-2 Predicted and actual market shares.................................................................................... 30 3-1 Flexible pavement weighting factors................................................................................. 46 3-2 Sample calculation of revised layer weighting factors ...................................................... 48 3-3 Sample calculation for multiple Superpave mixes............................................................. 49 3-4 Project classification by FDOT.......................................................................................... 53 3-5 Project re-classifica tion by the researcher ......................................................................... 53 3-6 Sample of an expected distress in tabular format.............................................................. 61 3-7 Sample of a reliability summary 1..................................................................................... 63 3-8 Sample of a reliability summary 2..................................................................................... 64 3-9 Conversion of predicted dist resses to overall reliability ....................................................67 3-10 Conversion example of the overall relia bility to ratio to target value ............................... 68 4-1 All provided projects for validation from FDOT............................................................... 71 4-2 Eliminated projects from CQI analysis.............................................................................. 78 4-3 Selected projects for CQI analysis..................................................................................... 80 4-4 Summary of the CQI projects by FDOT ratings................................................................ 87 4-5 Mean value of the CQ I projects by districts ...................................................................... 89 4-6 Regression on friction course ICQI ................................................................................... 91 4-7 ANOVA statistics for regre ssion on friction course ICQI .................................................91 4-8 Regression on Superpave ICQI.......................................................................................... 92 4-9 ANOVA statistics for re gression on Superpave ICQI ....................................................... 93 4-10 Performance of the ICQI model.........................................................................................94 4-11 Mean value of the ICQI projects by districts ..................................................................... 95
9 LIST OF FIGURES Figure page 2-1 Overall goal: market share of competitor group................................................................ 29 3-1 Schematic of a pavement structure with n layers...............................................................44 3-2 Sample asphalt (Superp ave) report from LIMS.................................................................55 3-3 Layout of the MEPDG program for input and output........................................................ 56 3-4 Climate input Gainesville, FL......................................................................................... 57 3-5 Color-coded inputs......................................................................................................... ....58 3-6 Asphalt material properties asphalt mix......................................................................... 58 3-7 Asphalt material properties asphalt general.................................................................... 59 3-8 Sample of an expected distress in graphical format........................................................... 62 3-9 SuperDecisions hierarchical m odel for weighting distresses............................................. 65 3-10 Matrix pair-wise comparison screen..................................................................................66 3-11 Results of the pair-wise comparisons................................................................................ 66 4-1 Distribution of CQ Is for good projects ..............................................................................85 4-2 Distribution of CQIs for average projects .......................................................................... 86 4-3 Distribution of CQ Is for poor projects ............................................................................... 86 4-4 Overall CQIs of good and poor projects............................................................................ 88
10 LIST OF ABBREVIATIONS AHP Analytic hierarchy process. ANP Analytic network process. AQC Acceptable quality characteristics. CQI Construction quality index. ERS End-result specifications. HMA Hot mixed asphalt. ICQI Integrated cons truction quality index. LBR Lime bearing ratio. LCC Life-cycle cost. LIMS Laboratory information management system. LSL Lower specification limit. LTPP Long-term pavement performance. MEPDG Mechanistic-empirical pavement design guide. PD Percent defective. PMS Pavement management system. PRS Performance-related specifications. PWL Percent within limits QA Quality assurance. QC Quality control. RAP Recycled asphalt pavement. RQL Rejectable quality level. SHA State highway agency. USL Upper specification limit.
11 Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy DEVELOPMENT OF A CONSTRUCTION Q UALITY INDEX AND AN INTEGRATED CONSTRUCTION QUALITY INDEX TO EVALUATE THE QUALITY OF PAVEMENT CONSTRUCTION PROJECTS By Junyong Ahn August 2008 Chair: Ralph D. Ellis Major: Civil Engineering Using some type of evaluation process to rate contractors is definite ly the current national trend. In the past, several actions by state highway agencies have been taken to develop some kind of rational system of rating the quality of highway constructi on projects and to utilize these ratings for contractor qualific ation or bidding purposes. Howeve r, setting contractors bidding limits is the most substantial use of quality ratings while some stat es do not use them for anything, and some states do not even rate the qual ity of their projects. Th e subjectivity is one of the main complaints against current systems. The Construction Quality Index (CQI) and the Integrated Construction Quality Index (ICQI) ar e ratings of the quality of materials and workmanship on highway projects that, unlike current quality rating models used by state highway agencies, is totally objective. The CQI/ICQI model tended to assign quality index values consistent with the owners level of satisfaction with the limited data procured through th e LIMS database. FDOT was asked to provide 105 projects to the research team along with an associated rating for each project. The projects submitted by FDOT were to be ones that had data entered into the LIMS database.
12 For the projects that met the criteria for incl usion, the CQI model showed positive results, rendering higher CQI for projects rated good by FDOT. The results were clearer when the projects were put into two categories; good and poor. Because the rating by FDOT is a criterion of the model, having a consistent and precise ratin g is essential for successful application of the CQI model. There is a tendency that bette r rated projects by FDOT pr oduce higher ICQI. The mean value of ICQI for projects rated good by FDOT construction and materials personnel was higher than the other ICQI mean values for pr ojects rated average or poor. The same relationship can be seen be tween average and poor.
13 CHAPTER 1 INTRODUCTION 1.1 Background In 1992, construction w as one of the largest i ndustries in the United States, constituting approximately 7.9 percent of the Gross Nati onal Product (GNP) (MacAuley et al. 1993). Comparing the post-World War II peak of 11.9 per cent of GNP attained in 1966, this 7.9 percent seems low even though the construction industr y tends to fluctuate widely according to variations in the economy or the cycle of the s easons. However, this measure does not illustrate the importance of construction in the economy be cause the above data only represents new construction. Several types of construction activ ity that are not included in new construction data, for example, maintenance and repair, commercial/industrial renovation, and hazardous waste cleanup, are also a big part of the U.S. indus try. With the start of the new century, various construction agencies have already begun sear ching for cost-effective and labor-effective management practices. State Highway Agencies (SHAs) have been seeking to develop mechanisms that elevate the quality of the transportation infrastructure an d reduce project cost and agency personnel. As one of the alternative ways to shave project costs and reduce agency personnel, policy changes concerning the use of contractor-conducted testin g in quality assurance decisions have been made by SHAs, which has reinforced the importan ce of quality-driven cont ractors. These policy changes forced contractors to implement more requirements in handling quality management in their field operations. As more contractors administered the qua lity control (QC) service, the SHAs do not do QC any more, thus a quality assura nce (QA) remains as their role. Within this quality assurance process, a need arose for co mprehensive evaluation methods for contractors
14 quality control processes, which led to a need for researching quality performance measurement techniques and approaches. All SHAs accept their projects in accordance w ith one or more construction specifications that contain guidance and minimu m requirements that contractors should follow. Theoretically, when contractors comply with the specifications the final product must provide the expected level of service for the e xpected periods of time. In many cases, low-quality work and high-quality work are treated the same as long as they meet the requirements of the specifications. The current approach for evaluating the quality of pavement projects is generally based on engi neering judgment. Variou s material properties, such as the percentage of laboratory densit y, the limerock-bearing ra tio, the asphalt binder content, and gradation, are used for evaluating pavement construc tion. It is generally believed that these properties should be related to the qua lity of the pavement construction; however, no fundamental engineering propertie s are used to connect the components of the pavement system to evaluate the pavement system performance. Many researchers recently have emphasized the development of a Construction Quality Index (CQI). The CQI supplies a rational m eans for measuring the overall quality of a constructed structure by determining the quality of the individual components and combining them as a system to obtain a composite quality index. The CQI value can be used either to determine the contractors compensation according to the quality level or to determine the contractors qualification status as a contractor rating system. 1.2 Objective The prim ary purpose of this research was to develop two practical and objective models to evaluate the quality of pavement construction. One model generates CQI, and the other model produces integrated constructi on quality index (ICQI), includ ing aspects of mechanistic-
15 empirical analysis. The CQI/ ICQI were implemented withou t substantial modification to FDOTs current test and measurement system. The CQI/ICQI include ma terial, structural, and pavement smoothness characteristic s and are applicable for both new and rehabilitation projects. Soils, bound and unbound granular base materials, and asphalt were considered. Concrete materials for concrete pavement projects were not considered in validating the models because of the limited number of concrete pavement project s and significant problems in the procuring of the data. The goal of the developed CQI/ICQI was to be able to use it as an objective tool for evaluating the quality of pavement construction. The formulation and data set used to develop the CQI/ICQI model must be objective; that is, it must be based upon quality characteristics that are explicitly addressed in the construction specifi cations and directly within the control of the contractor. In order to be transparent a nd stick to the specifications, the CQI model was applied using the percent within limits (PWL) concept, which already has been used by FDOT and is familiar to the contractor. In general terms, the PWL de termines whether the CQI for a project is good or bad. The PWL for each quality characteristic were multiplied by weighting factors. In the CQI model, the analytic hierarchy process (AHP) wa s applied to obtain weighting factors for each factor (acceptance quality characteristics) and each layer of the pavement system. Acceptance quality characteristics such as #200 sieve passin g rate and asphalt content from specifications were considered as factors of the model, but interactions between the factors were not considered. In the ICQI model, the mechanistic-empirical model and statistical (regression) analysis were used to generate ICQI. The same factors (acceptance quality char acteristics) in the CQI
16 model were initially considered for developm ent of the ICQI model, but the mechanisticempirical pavement design guide had different a cceptance quality characteristics from the CQI model to predict future pavement performance. Data for the different acceptance quality characteristics of the mechanistic-empirical pavement design guide was stored in FDOTs database system, and the acceptance quality char acteristics from the pavement design guide were used to generate the ICQI model. The factors (acceptance quality characteristics from the pavement design guide) were considered as independent variables, and a m odified output from the mechanistic-empirical model was used as a dependent variable of the regression model in developing the ICQI model. Interactions between the f actors in each layer of the pavement system were analyzed in the ICQI model; however, weighting factors for each layer of the pavement system were obtained from the AHP. For validation purposes, personnel from asphalt pa ving construction projec ts were asked to provide FDOT Construction and Mate rials officials with their ratings for the projects. Once the CQI and ICQI were developed and generated the sc ores for the projects, they were compared to the FDOT ratings under the assumption that FDOT ratings were the correct evaluations for the quality of the projects. Even though the mechanistic-empirical model can produce a variety of output, such as expected distre sses according to pavement syst em ages, the generated output cannot represent a project but, rather, only represen t one sample in this research. Therefore, when comparing a projects quality rating, the mechanistic-empirical model was not used. It was used to develop the ICQI. To the greatest extent possi ble, the CQI/ICQI should use data from the Laboratory Information Management System (LIMS) because it is FDOTs enterprise database system,
17 which, theoretically, contains all construction qu ality data and is ava ilable to general LIMS users. 1.3 Scope In order to achieve the objective of this res earch, the detailed scope of this study was as f ollows: To collect data related to major com ponents of asphalt pavement construction To identify the factors that in fluence the performance of as phalt pavement construction To identify interactions between the factor s that influence the performance of asphalt pavement construction To measure and analyze the effects of constr uction quality characteristics on pavement project performance To validate CQI and ICQI with FDOT ratings In keeping with a straightfo rward approach, the CQI/ICQI only focuses on quality factors for the major components of asphalt pavement cons truction such as Superpave (asphalt), base course, subgrade, and embankment. Other aspect s of contractor performance typical in the current rating system (e.g., financial resources, ownership of equipment or ability to lease equipment, adherence to schedule, job safety, and past performance) are not included in the CQI/ICQI formulation. The CQI/ICQI models were formulated in a mo dular fashion. The flexible models enable them to be scaled to all pavement constr uction projects, from routine mill and overlay rehabilitation to major new highway pavement construction. This means that the CQI/ICQI models should be able to generate layer-lev el CQI/ICQI for each layer of the pavement system, and the sum of the layer-level CQI/ICQI times a weighti ng factor for each layer makes the project-level CQI/ICQI. By allowing the mode l to have layer-level CQI/ICQI, any type of layer combination for the pavement system shoul d not be a problem. According to the layer
18 combination, revised weighting factors for each layer are calculated by weighting their respective contributi on to the project. This modular approach leads the ICQI model to consider interactions between factors only within the same layer because th e layer-level ICQI should have its own value (score), which is not combined with other factors in other layers. Because there are not enough concrete paveme nt projects, and therefore not enough data, to validate the model, concrete pavement has been excluded from the CQI/ICQI. Finally, the ICQI was developed with m echanistic-empirical (M-E) analysis. The procedures of ICQI development and the genera ting of ICQI scores from the model were as follows: Collect data for acceptance quality characteristics from LIMS Generate output (such as expected distress es) from the Mechanistic-Empirical Pavement Design Guide (MEPDG) for each sample Determine how to translate expected distresses to pavement performance in MEPDG. Develop regression model for each layer usi ng the collected data and the generated output Enter the collected data to the layer-level IC QI model to have sample-level ICQI for each sample Compute ICQIlayer from sample-level ICQI Calculate the project-level ICQI by the sum of the layer-level ICQI times a weighting factor for each layer Even though the MEPDG allows researchers to connect fundamental material quality measures to facility performance, it still n eeds evaluation, revision, va lidation, and calibration. For evaluation and calibration, the MEPDG was used to develop ICQI. The result of the ICQI model was compared to the CQI model and FDOT ratings.
19 CHAPTER 2 LITERATURE REVIEW 2.1 Contractor Evaluation Systems of SHAs A lite rature review related to contractor ev aluation systems of SHAs, performance-related specifications, analytical hierarchical process, percent within limits, and acceptance quality characteristics was conducted for this research proposal. When low-quality construction work is tr eated the same as high-quality work in competitive-bid construction projects, both owners and contractors acknowledge that it is a problem. Hybert (1996) examined quality problems in owners that use a contracting process to provide customized, large-scale sy stems or products, which can be extended to quality problems on many construction projects. Too often, he f ound that contractors we re winning contracts by underbidding, exaggerating delivery capabilities, underestimating the project risks, or undersolving the technical problems just to get a lowe r price than their competitors. They then earn extra fees by change orders for their ability to argue specification interpretation issues. SHAs have expressed this same concern. Sin ce each SHA has its ow n political and cultural circumstances in dealing with contractor quali ty, every states contra ctor evaluation system should not be the same. However, a survey wa s conducted concerning thei r policies to find out whether the methods used by them can be impl emented in FDOT. Survey results concerning SHAs plans or policies of contractor eval uation systems can be seen in Table 2-1. Table 2-1. Current project ra ting systems around the U.S.A. State State Wide Rating System Description Alabama N N/A Alaska N N/A Arizona N N/A
20 Table 2-1. Continued State State Wide Rating System Description Arkansas N N/A California N Some districts informally rate some projects. Delaware N N/A Georgia Y A subjective form is used to affect bidding capacity. Hawaii N N/A Idaho N N/A Illinois Y A subjective form is used to create a performance factor that affects the pre qualification amount. Indiana Y A subjective form is used to create a performance factor that affects the pre qualification amount. Iowa Y A subjective form is used to create a performance factor that affects the pre qualification amount. Kansas Y For work completed by the prime only, a subjective form is used to form a factor that affects bidding capacity. Kentucky Y A subjective form is used to designate a performance level which affects bidding capacity. Louisiana N N/A Maine Y A subjective form is used to create a performance rating that affects the pre qualification amount. Maryland Y A subjective form is used, and based upon its score the retainage amount is adjuste d. Currently revising their form to make it less subjective. Massachusetts Y A subjective form is used by the prequalification committee in their decision making process. A contractor must have an 80% passing rate. Michigan Y A subjective form is used to affect the prequalification amount. If the contractors' scores are too low, they are given three chances to impr ove prior to disciplinary action. Minnesota N Trying to become a prequalific ation state; currently rate particular projects base d upon quality vs. cost. Mississippi N N/A
21 Table 2-1. Continued State State Wide Rating System Description Missouri Y A subjective form is used to rate the contractor. If their average score becomes to low, they can be placed on probation or suspended. Montana N N/A Nebraska Y A subjective form is used to create a rating factor that affects bidding capacity. Nevada Y A subjective form is used to affect prequalification amount. New Hampshire Y A subjective form is used to rate the contractor. This form is then used by the prequalification board along with other information to qualify the contractor. New Jersey Y A subjective form is used to create a factor that affects prequalification amount. North Carolina N Have a safety and environmental index but no performance evaluation. North Dakota N N/A Ohio Y A subjective form is used to affect the financial capacity portion of prequalification. Oklahoma Y A subjective form is filled out for each project; currently the forms are filed away and not used. Oregon Y A subjective form is used for all projects; once the average score drops below a threshold, a discipline process begins. Pennsylvania Y A subjective form is used to create an ability factor that increases or decreases thei r bidding capacity. If the contractor is sub par in particular types of project, they can be barred from bidding. Rhode Island N N/A South Carolina Y A subjective form is used to create a score for each project. If the contractor's quarterly average drops below the minimum score, they are brought in for a meeting; if no improvement occurs, they are disqualified from bidding. The state is trying to implement more strict standards for important projects.
22 Table 2-1. Continued State State Wide Rating System Description South Dakota N The state used to have a standard form; due to lack of use only some districts still use the form. Tennessee N Trying to begin an evaluation system by 2007. Texas N N/A Utah Y A subjective form is used to let contractors know where they stand. On design build jobs the contractor with the highest score gets the job. Vermont Y A subjective form is used indirectly by the prequalification board as another source to qualify the contractor. Virginia Y In the process of revamping system, currently the subjective performance rating affects 70% of prequalification amount. Washington Y A subjective form is used to affect bidding capacity. West Virginia Y A subjective form is used to affect bidding capacity. Wisconsin Y A subjective form is used to increase or decrease the prequalification amount. Wyoming Y A subjective form is used to create a rating factor that affects the prequalification amount. Washington DC N N/A Puerto Rico N Wants to implement some evaluation process; however currently can not legally. Comparing the nationwide interview results of NCHRP 10-54 (Minchin and Smith 2001), the expectation that more states would rate cont ractors and even several states would facilitate the results of the ratings with a qualification process was not met. Instead, a few states that once rated contractor quality have abandoned their systems. Currently, 28 states have an evaluation process, and four states are in the final stages of attempting to implement one. Nineteen of the st ate DOTs use their subjective forms to evaluate contractors affecting either the contractors bidding capacity or prequalification amount. Other
23 states uses for the evaluation scores include, but are not limited to, br iefing requirements when the scores drops too low, adjust ing the contractors retainage, being used by a prequalification committee in their decision-making process, an d being placed on probation. Again, no states used objective forms to rate contractors. As a way to be objective, several states evaluation forms now have questions that can be answered only with yes or no instead of the traditional scaled responses. However, not a single DOT me ntioned trying to include any type of material and workmanship test results. While all forms incl ude a section that contains either workmanship or materials, all are based on whether the work manship meets specificati ons, not on numeric test results. 2.2 Performance-Related Specifications Specifications have been changed by tran sportation agencies from quality assurance specifications that define end-product quality to performance-related specifications (PRS) that specify quality in terms of desired long-term performance. PRS help transportation agencies forecast future performance, maintenance requir ements, and life-cycle costs. PRS describes the desired levels of key construction quality characteristics that have been shown to correlate to engineering properties and applies mathematical models to predict future pavement performance. The benefits of using PRS in clude identification of a direct relati onship between quality characteristics and product performance, identification of an optim um level of quality, a rational basis upon which to set the appropri ate level of penalty/bonus for inferior/superior quality, and a critical link between construction and engi neering management systems (FHWA 2001). Buttlar and Harrell (1998) reported the SHAs efforts to develop and implement end-result specifications (ERS) and performance-related spec ifications (PRS) in Il linois. Authors of the paper stated that PRS provided the ultimate method of compensation for a delivered product even though such a system could be challenging to develop. They also suggested development
24 and implementation of a specification that combin ed elements of ERS and PRS considering the existing technology level, available materials, and test equipment. As key steps for developing the combination specification, the authors presented the following. 1. Make an initial move to statistical QC/QA. 2. Develop a comprehensive ERS to consider all relevant quality characteristics. 3. Monitor and foster devel opment of primary and secondar y prediction relationships. 4. Develop performance-related pay factors. 5. Compare performance-related pay factors with ERS pay factors, which were developed based upon experience. 6. Periodically repeat steps 3, 4, and 5 to move from ERS to PRS. An approach was presented to estimate the deviation from pavement performance life due to any deviation in the as-bui lt characteristics from the as-designed characteristics. The deviations can be used to set up the basis for measuring the rational pay adjustment. To get the estimation, key quality control aspe cts in asphalt pavement construction such as asphalt content, aggregate characteristics, pave ment layer thickness and their degree of compaction, and initial pavement smoothness are quantified using a partial derivatives appr oach (Noureldin 1997). The authors claim that current QC/QA protocols cannot be used to evaluate the overall construction quality of a project and that it is hard to compare the construction quality of different projects or different co ntractors. An overall CQI is deve loped using material, thickness, and structural data so that it can evaluate the overall construction quality and facilitate the comparison between different pavement sections and projects (Beckh eet et al. 2004). Because of the need for a rational method to rela te as-built quality to expected performance and ultimate value as the basis for reliable and defensible pay schedules, a method of developing pay schedules was proposed. The pay factor in th e method can be expressed as a monetary value
25 rather than as a percentage of the bid price of the pavement. This method is believed to more appropriately reflect the true va lue of departures from the design level of quality since the actions upon which the pay reduction is based are not a function of the thickness of the pavement layer or bid price (Weed 1998). In his research, analytical data and survey data were used to develop mathematical models for predicting pa vement performance. The models were then combined with other models, which relate expect ed life to present value to obtain rational and practical pay schedules. Weed (2000) presented a method for combining the effects of multiple deficiencies. Air voids and thickness of hot mixed asphalt (HMA) pave ment are factors used to decide whether a large amount of HMA pavement is rejectable. In the current New Jersey Department of Transportation specifications for HMA pavement, th e rejectable quality level (RQL) for both air voids and thickness is 75, which me ans that if any one RQL of th e two characteristics is more than 75, then the agency reserves the right to order removal a nd replacement of the deficient pavement. This might not consistently distinguish poor quality pavement from acceptable quality pavement because a pavement job with two items rated as having poor quality levels but each barely within the acceptable range may be a wo rse case than the other pavement job with an excellent quality level for one characteristic but a quality level above the RQL in another characteristic. To determine an appropriate method for assessing the combined effects of deficiencies in air voids and th ickness, survey data were used Based on the performance model with combined effects, several pay equations were presented. A rational and feasible method for quantitativel y formulating pay factors was described for asphalt concrete construction. Performance models were developed for fa tigue and rutting based
26 on the analysis of accelerated pavement tests from the Caltrans Heavy Vehicle Simulator (HVS) and the WesTrack accelerated pavement performance test program. The development of pay factors in the rese arch considers the economic impact to the highway agency. The amount of penalty/bonus wa s sought under the assumption that the penalty should be the extra cost to th e agency, and the bonus should not be greater than the added savings to the agency. For new construction, these costs/ savings to the agency are related mainly to prospective pavement rehabilitation. Inferior construction increases the presen t worth of future rehabilitation costs; in contrast, superior cons truction decreases the present worth of the costs. Differences in the present worth of future rehabilitation costs between as-built and as-designed can be applied in setting the appropriate leve l of penalty/bonus for inferior/s uperior pavement construction quality. However, the authors admit that pena lties/bonuses might be too low because only the first rehabilitation cycle was consid ered in their performance model. The performance-based approach highlights th e importance of uniformity in both materials and placement and the importance of adhering to th e design target value (Monismith et al. 2004). Killingsworth argued that of 13 factors analyzed, five were proven to have a significant influence on the overall performance of HMA pa vement and should be included in performancerelated HMA construction specifications. The selected factors are segr egation, initial ride quality, in-place pavement density, density at longitudinal joints, and permeability. These quality characteristics of as-produced and as-con structed hot mix asphalt directly affect asdesigned performance quality and life. Practical test methods for measuring these five quality characteristics, specification criteria, and threshold values are suggested for PRS (Killingsworth 2004).
27 Many statistical construction specifications and acceptance procedures define an end result, and according to the end result, bonuses are awarded by many agencies. A significant amount of research was found to improve these pr ocedures and the develo pment of better PRS. An assumption that legitimate mathematical relationships have been established between characteristics measured at the job site and the expected performance of the construction activity is required in order to determine an appropriate amount of pay reduction/addition as penalties/bonuses of inferior/superior construction quality. For mo st factors, however, there are no such convenient or simple relationships; therefore, a method for developing the required relationships is necessary. Weed and Tabrizi (2005) explained the developm ent of a statistical acceptance procedure for hot-mix asphalt pavement smoothness usi ng the international roughness index. As procedural steps, Weed and Tabr izi (2005) suggested the following. Select a quality characteristic that relates to performance. Select a statistical quality measure upon which acceptance will be based. Select an appropriate mathematical form for the performance model. Obtain data to calibrate the performance model. Apply life-cycle cost analysis to determine appropriate pay levels. Convert this information into an appropriate pay schedule. Define lot size and sample size. Finalize the prototype specification. A comprehensive approach for the developmen t of performance models for network-level Pavement Management System (PMS) using Long-Term Pavement Performance (LTPP) data was presented by Bekheet et al (2005). Conventionally, historic al performance and inventory data have been used for developing these pave ment performance models; however, historical data may not be appropriate to use because fiel d data collection equipmen t has been continually improved, and inventory records may be incomplete.
28 As an alternative reliable source of data fo r developing pavement performance models, the LTTP was used. Once base pavement performan ce models have been developed, they can be adapted to agency-specific experience and data to render agency-specific m odels (Bekheet et al. 2005). Whiteley et al. (2005) devel oped a method for obtaining pay factors based on pavement life-cycle cost (LCC) by establishing the relations hip between design life and LCC, as well as between LCC and pay factors. The follo wing are the results of the research. Overlay thickness increases result in th e increase of pavement service life. More than 80% of the contribu tion to the variance in paveme nt service life predictions is made by overlay thickness, whereas less than 20% of the variance is contributed by combined variables of accumulated ESALs af ter eight years and to tal prior cracking. Regardless of overlay thickness distribution t ype, the resulting life-cycle costs show a normal distribution. The pay factor values presente d in the research shows that disincentives for inferior performance are greater than incen tives for superior performance. 2.3 Analytic Hierarchy Process The Analytic Network Process (ANP) is st ructured for the analysis of societal, governm ental, and corporate decisions for decision makers. The ANP allows for inclusion of all the factors and criteria that are relevant for making the best deci sion whether they are tangible or intangible. Both interaction and feedback are allowed to be applied in the ANP. Such feedback is known to take into account the co mplex effects of interplay in human society, especially when risk and uncertainty are involved (Saaty 1996). The ANP derives ratio scale prio rities for the distribution of influence among the factors and groups of factors in the de cision by providing a way to input judgments and measurements. The ANP can be applied to allocate resources accor ding to their ratio scale priorities because the process is based on deriving ratio scale measurements.
29 An example of applying ANP to the problem of predicting the market share for the biggest three companies in the hamburger fast food i ndustry is shown in Figure 2-1 (Saaty 1999). Clusters of elements form the ANP model, which are connected by their dependence on one another; therefore, decision makers should consider grouping elements that share a set of attributes when identifying clusters. For example, the marketing mix is a cluster whose elements are price, product, promotion, and location. When identifying clusters and their elements, the elements should be similar. Figure 2-1 below from the research shows the structure of the m odel described by its clusters and elements and by the connection betw een them. The flow of influence between the elements is indicated by these connections. The prevailing understanding of the marketplace is diagrammed in the ANP model through the pro cess of analyzing dependencies. A cluster is connected to another cluster when at least one element in it has infl uence on at least one element Figure 2-1. Overall goal: market share of competitor group (Saaty 1999)
30 in another cluster. The elements, themselves, ar e not displayed in this figure. Except for the customer group cluster, which doe snt have a curving arrow in the figure, inner dependence exists for all other clusters. The connections between elements for the other clusters exist in the same cluster. With the ANP, the research showed the comp arison between the predicted market shares and the actual market shares. The results can be seen in Table 2-2 below. The predicted market shares were close to the actual market shares. Table 2-2. Predicted and actua l market shares (Saaty 1999) Market Share Company Predicted % Actual % McDonalds 62.9% 58.2% Burger King 23.9% 28.6% Wendys 13.2% 13.2% Applications of the ANP also can be found in construction performance measurement (Isik et al. 2007). In this paper, Isik et al. presen t a conceptual performance measurement framework that considers company-level fact ors (objectives, strategies, and resources), as well as projectlevel (risks and opportun ities) and market-level factors (competition and demand). In order to use the conceptual framework for measuring perf ormance, the ANP was used because authors of the paper regarded the ANP as the best-suite d methodology that consider s both quantitative and qualitative factors and the interrelations between them. The ANP was the effective tool in those cases in which the interactions among the elem ents of a system form ed a network structure. The well-known decision theory, the Analytic Hierarchy Process (AHP), is a special case of the ANP. Both the AHP and the ANP derive ratio scale priorities for elements and clusters of elements and make paired comparisons of elements on a common property or criterion. The
31 difference in the AHP is that the elements are or ganized in a hierarchic decision structure, while the ANPs structure has one or more flat networks of clusters that contain the elements. As mentioned earlier, the AHP is a special case of the ANP and the decision-making technique developed by Saaty (1980 ). He claims that AHP is a comprehensive, logical, and structured decision-making process to help deci sion makers set priorities and make the best decision when both qualitative and quantitative asp ects of a decision need to be considered. The AHP is designed to consider a variety of subj ective and objective evalua tion measures, providing a useful mechanism for inspecting the consistency of the evaluation measures and alternatives. Although there are critics of the method, the general consensus is that the AHP is useful both technically and practically (McCaffrey 2005). The AHP relies on three fundamental assumptions. Preferences for different alternatives depe nd on separate criteria that can be reasoned about independently and given numerical scores. The score for a given criteria can be calculated from sub-criteria; that is, the criteria can be arranged in a hierarchy, and the score at ea ch level of the hierar chy can be calculated as a weighted sum of the lower-level scores. The model can be as many levels deep as necessary to represent th e information appropriately. At a given level, suitable scores can be ca lculated from only pair-wise comparisons. Various components, such as social, political technical, and economic factors, can be involved in the decision-making process. The fi rst step in using th e AHP involving complex decisions is to break down the goal into its com ponent parts, progressing from the general to the specific. This hierarchical stru cture involves not only goals but also criteria and alternatives. Each set of alternatives can be divided into as many further levels as nece ssary. The next step is to make a series of one-on-one comparisons of al l criteria. The results are then synthesized to determine the priorities of the alternatives with respect to each criterion and the weight of each criterion with respect to the goal. For each pairing, the score is decided on a relative basis, not an
32 absolute basis, comparing one criterion to anot her. For the final step, the mathematical model computes the relative scores for each pairing, gene rating a relative weight for each criterion. The sum of the weights is normalized to 1, or 100%. Applications of the AHP can be found in transportation engineering and construction. Smith and Tighe (2006) used the AHP as a tool fo r infrastructure management. The uses of the AHP by Smith and Tighe range from comparing repa ir products of fast-tract concrete based on priorities set by an agency to comparing ma intenance, rehabilitation, and reconstruction strategies for asphalt pavements. Smith et al (1995) used the AHP to characterize bridge material selection decisions of st akeholders, specifically as it rela tes to using timber as a bridge material. 2.4 Percent within Limits There are several qu ality measures that can be used in an acceptance plan. In early QA specifications in the late 1960s and 1970s, the mean or the average deviation from a target value was often used as the quality measure. However, the use of the av erage alone provides no measure of variability, a nd it is now recognized that variability is often an important predictor of performance (Burati, et al. 2004). Several quality measures, includ ing percent defective (PD) a nd PWL, have been preferred in recent years because they simultaneously meas ure the average level an d the variability in a statistically efficient way. PD and PWL are, in reality, the same quality measure since they are directly related by the simp le relationship PWL = 100 PD (Burati, et al. 2004). The PWL concept is applied to calculate pa y factors in Section 334: Superpave Asphalt Concrete in the FDOT specification, 2007 Standard Specifications for Road and Bridge Construction. Under Section 334 in the spec ifications, the quality char acteristics to measure and their upper or lower specification limits are presente d to determine the quality of the constructed
33 material. Based upon the quality of the material, a pay adjustment to the bid price of the material is applied. 2.5 Acceptance Quality Characteristics Federal Highway Adm inistration (FHWA 1999) has defined the acceptance quality characteristics (AQC) as an inherent measurable pa vement characteristic that significantly affects pavement performance, is under the direct control of the contractor, and is measurable at or near the time of construction. The AQC for this research will be identic ally selected to those in the FDOT specification, 2007 Standard Specifications for Road and Bridge Construction which is currently used by FDOT for acceptance of pavement materials at the mine, plant, or roadway. The selected AQC will be considered as fact ors of the CQI model. The AQC are listed in Appendix A. For developing and validating the ICQI model, mo st of the AQCs selected as factors of the CQI model will be also considered as factors of the ICQI model as long as the MEPDG program can accept them as input variables. In this rese arch, #8 sieve aggregate passing rate and asphalt sample density will be excluded from the ICQI model factors. The MEPDG program utilizes #4 sieve aggregate passing rate for an input variab le instead of #8 aggreg ate sieve passing rate. Because the ranges of the AQC are not required fo r developing the ICQI model, the AQC of the ICQI are not listed in Appendix. 2.6 Mechanistic-Empirical Paveme nt Design Guide (MEPDG) Methodology for pavement design and evaluatio n of paving m aterials is improved by the proposed MEPDG procedure. The new procedur e depends on the characterization of the fundamental engineering properties of paving mate rials. In order to use the MEPDG to design flexible pavements, many states put forth an ex tensive effort to characterize asphalt mixes used in their states.
34 The Virginia Department of Transportation sponsored a project whose objective was to perform a full HMA characterization in accordan ce with the procedure established by the proposed MEPDG to support its implementation in Virginia (Flintsch et al. 2007). To achieve this objective, samples of surface, intermedia te, and base mixes were tested. The dynamic modulus, the main HMA material property requ ired by the MEPDG, and creep compliance and tensile strength were studied. The Virginia Department of Transportation re search revealed that the dynamic modulus test is quite effective for determining the mechan ical behavior of HMA at different temperatures and loading frequencies. The test results de monstrated that the dynamic modulus can be sensitively changed as the mix constituents (aggregate type, asphalt content, percentage of recycled asphalt pavement, etc.) change. Even mixes of the same type used had different measured dynamic modulus values when the mixes did not have the same constituents. The Montana Department of Transportati on (MDT) conducted a similar research project as the Virginia investigation above. The research of the MDT was to develop performance characteristics of flexible pavements in Mont ana and to use them in the implementation of distress prediction models (VonQuintus and Moulthrop 2007). Utilizing reliable distress prediction models, the MDT intended to use Mech anistic-Empirical (ME) based principles for flexible pavement design and manage the Montana highway network. This study focused on developing local calibration factor s for Montana climate, structur es, and materials for flexible pavements by using the MEPDG software. Chehab and Daniel (2006) studied the sensitiv ity of the predicted performance of recycled asphalt pavement (RAP) mixtures to the assu med binder grade using MEPDG software. In the research, MEPDG software was used to predict the performance of a specific flexible pavement
35 structure with a RAP-modified hot-mix asphalt surf ace layer. To minimize va riations of the other conditions, all the pavement pr operties and conditions were held constant except for the properties of the surface layer while di fferent design runs were conducted. Comparison of the predicted performance of th e various runs was presented by Chehab and Daniel (2006). The importance of determining the effective binder grade of RAP mixtures was emphasized from the results of the study. 2.7 Conclusions Through the literature review relating to the existing construction quality evaluation study, the res earcher could consider several unanswered questions that might be answered by adopting the CQI/ICQI. How can the differences in constructi on quality be quantified objectively? How can quality indicators required by the spec ifications and stored in FDOTs LIMS be linked rationally to formulat e quality discriminators? What acceptance quality characteristics are mo st important in determining contractor quality? Are there any interactions among the construc tion quality characteristics in determining construction project performance? What is the relationship between contractor quality and performance of constructed facilities? How can the various components of a paveme nt construction project be combined to develop an overall indicator of construction quality? Can concepts from performance-related speci fications be used to assess construction quality? What is the role of mechanistic-empirical procedures in assessing construction quality? Can the mechanistic-empirical procedures be combined with a mechanistic, statistical, or empirical model?
36 CHAPTER 3 METHODOLOGY 3.1 Data Collection In this rese arch, two models for pavement construction quality measurement will be developed. The CQI model will be develope d using the PWL concept, not considering interactions between factors. The ICQI mode l will be developed with consideration of interactions and mechanistic-empirical analysis. These models then will be compared to FDOT ratings, assuming their evaluation is correct. In order to develop and validate the CQI/ICQI model, actual test re sults of the acceptance quality characteristics from various pavement construction projects were required. Therefore, the project list for data collection was asked to be provided to FDOT personnel in order to retrieve actual test results of th e acceptance quality characteristics from LIMS. In the process of development and validation of the CQI model, the re trieved data was used only for validation purposes. The CQI model wa s formulated using the percent within limit concept from Florida specifications and the analy tic hierarchy process fo r weighting factors of the CQI model. In order to acquire the weightin g factors of the CQI mo del, an expert panel survey was held. Once the CQI model was develope d, the retrieved actual test data was used to obtain CQI scores of the projec ts; then, the CQI scores were compared with FDOT ratings. Unlike the process of CQI model development, th e actual test data from LIMS was used to formulate the ICQI model. The ICQI model required output from the MEPDG program to use for regression analysis. When a set of actual test data is entered into the MEPDG program, the program produces its output in the form of severa l predicted distresses, such as roughness, topdown cracking, bottom-up cracking, thermal cracking, and rutting; therefore, conversion of the predicted distresses to the exp ected performance of pavement construction projects should be
37 executed. In order to convert the output from the MEPDG program, information about the relationship between the predicte d distresses and the performance of pavement construction projects was essential. To obtain their relations hip, the expert panel survey and SuperDecisions program based on analytic hierarchy process were used. 3.1.1 Project List for Data Collection When developing a CQI m odel, one of the c onsiderations is making an objective model using objective data. As shown in Table 2-1, ever y state having a contractor rating system uses subjective forms. Development of a CQI that is as objective as possible may lower or eliminate the concern for construction owners. In an effort to maintain objectivity, two CQI models that rely completely on material and workmanship test results were developed. In addition, the data for the CQI must be the actual test results of the AQC and easily accessible from the database of the FDOT without significantly changing the current test properties or methods. Therefore, LIMS was used to retrieve data. In order to retrieve data from LIMS, FDOT Construction and Mate rials officials were asked to provide project number s to the study. The projects were to be recent enough to be relevant (having used current methods, such as Superpave) and use LIMS as the data management system (data in previous data mana gement system may not be retrievable) but old enough so that sufficient post-construction test ing would have been performed. The most important requirement was for the projects to ha ve their relevant data stored in the LIMS database. It was also requested th at FDOT provide a level of satis faction, or rating, of each project provided in order to use for later eval uation purposes. As a result of the request for project numbers to study, FDOT supplied a total of 105 project numbers from all districts including th e turnpike district w ith FDOTs level of satisfaction. The number of projects provided varied from 1 project in District 6 to 28 projects in
38 District 2. Even though there were a total of 105 projects, not all of them could be used after close investigation because some seemed as if large portions of important data had not been entered into the database or they did not have any data in LIMS at all. Project details are summarized in Appendix B, including project nu mber, location, managing district, type of construction (resurfacing or rec onstruction), FDOT rating, maximu m number of possible layers according to the type of construction, number of layers containing data, and number of samples for each layer. Each pavement system consists of several layers with different functions and materials. Possible flexible pavement layers include embankment, subgrade, base, Superpave, and friction courses. The reason that maximum number of possibl e layers, number of layers containing data, and number of samples are incl uded in Appendix B is to eliminate unsuitable projects from analysis. For example, when a pr oject should have five layers according to its project type (i.e., new constructio n), if the project has data from only two layers, it may not be acceptable to claim that the data from the two la yers represent the quality of the project. The number of samples for each layer was also invest igated. When the number of samples for a layer was too small to represent the layer, the laye r or the project was excluded from analysis. In the early stage of this research, there were significant problems in the procuring of the data. The first problem encountered was gain ing access to LIMS, which took months. Once the data was available, it became apparent that large portions of important data had not been saved in the LIMS when new construction projects in cluded adding lanes that did not have an embankment, subgrade, or base data. In contrast, when resurf acing construction projects had foundation (embankment, subgrade, or base) da ta, inquiries for more detailed project descriptions to FDOT project ma nagers were required in order to reveal whether the foundation data was actually related to the road construction.
39 For Superpave and friction course layers, having several mixes with different target values for the certain AQCs is not uncommon. In such ca ses, the tonnage report should be run through LIMS to obtain revised weighting factors for th e mixes; however, for some reason, the tonnage report worked for only some proj ect numbers. Target values for different design mixes are required to calculate CQI even though the target va lues are not saved in LIMS. The target values can be retrieved from FDOTs intranet. Some target values for design mixes were missing, and CQI for the design mix are excluded from analysis. As seen in Appendix A, all AQC except material mixing depth are used as factors affecting the CQI. Mixing depth is not stored in the LIMS. 3.1.2 Expert Panel Surveys 126.96.36.199 Expert panel surveys for CQI The weighting factors for both CQI and ICQI m odels were obtained using the analytic hierarchy process (AHP) concept. In the CQI mo del, weighting factors for parameter level as well as layer level weighting factors were requir ed. The layer level weight ing factors define the relationship among layers of pavement systems such as friction course, Superpave, base, subgrade, and embankment. The parameter leve l weighting factors define the relationship between the expected performance of pavement construction projects and the parameters, which are acceptance quality characteri stics such as sieve passin g rate, #4 sieve passing rate, #8 sieve passing rate, #200 sieve pass ing rate, asphalt content of mixe s, air void ratio, density, lime bearing ratio (LBR), and ride number. To obt ain experts opinions concerning the factors affecting pavement construction performan ce, expert panel meetings were held. Approximately 30 expert panel members from FDOT, the constructi on industry, academia, and consultants were selected a nd requested to attend the expert panel meeting, which was held three times in Gainesville, Orlando, and Ta llahassee between June 2006 and August 2006. The
40 three meetings were identical, so the panel me mbers were requested to attend the one most convenient for them. As a result of the meetings, the research t eam was able to obtain opinions from 18 people: 7 members from contractors or contractor advocacy groups, 8 members from FDOT personnel, 2 members from academia, and 1 member who did not sign in. Initially, there was strong confrontation from mostly the contractors side concerning the purpose of the research and anticipated consequences after completing the research. The contractors concern focused on w hy quality rating of construction projects is needed when the projects already at least meet, and in some cases exceed, the requirements of specifications to be accepted by the owners. They were also concerned about how the owners would use the research results. The contractors argued th at if the owners want to get higher quality construction products than current projects, which already have passed the requirements according to specifications, then the owners can simply raise the specificatio n requirements. The contractors seemed to be disturbed because there was a possibility that new regulatio ns would be applied without increasing their fees. The research team explained that the purpose of the CQI research was to develop practical and objective models and evaluate the quality of pavement c onstruction. Not only meeting the requirements of the specifications by the use of the average values of the construction quality characteristics but also keeping PWL in certain ranges has been preferred in recent years because they simultaneously measure the average level and the variability in a statistically efficient way. Therefore, the research teams perspective of the construction quality with CQI was that variability might be an important predictor of performance. Howe ver, the research team could not answer the contractors concern about the usage of the CQI because it belonged to the owners authority.
41 The panel members were requested to fill out forms prepared by the rules of AHP. The forms used for flexible pavements are shown in Appendix C. The instructions for panel members answering questions in the form include: 1) Each response only re presents your opinion concerning the relative importance of the pair of items on a single line, 2) Fill out all portions of the form for which you feel qualified to have an opinion, and 3) Fill out the forms without discussion with your neighbor. The incorporation of all relevant decision criteria and their pair-wise comparison allows the decision maker to determine the trade-offs among objectives. This procedure recognizes and incorporates the knowledge and ex pertise of the participants by making use of their subjective judgments. The results of the survey are rearranged in Appendix D for CQI flexible pavements. The average values were entered into SuperDecisions software to determine the weighting factors for the CQI formulation. These flexible pavement weighting factors for the CQI are presented in Table 3-1. At any given pavement layer, the su m of the weighting factors is always 1. The first part of the forms shown in Appe ndix C concerning flexible pavement system components asks which layers have the greater in fluence on quality. In other words, this question was used to determine weighting factors for each layer of the flexible pavement system. These weighting factors for each layer will be identi cally applied to the ICQI model. However, weighting factors for the parameter level in th e ICQI model will be determined through MEPDG results and statistical analysis For this purpose, the expert panel surveys for ICQI were conducted and described in the next chapter. 188.8.131.52 Expert panel surveys for ICQI When FDOT evaluates a pavem ent, they consider cracking, roughness, rutting, raveling, patching, and friction (FDOT 2003).
42 There are five distresses in the MEPDG m odel applicable for HMA: rutting, fatigue cracking (bottom-up), fatigue cracking (top-down), therma l cracking, and roughness. The MEPDG does not directly address friction, patc hing, and raveling. The researcher dismissed thermal cracking as negligible for Florida cond itions. Therefore, roughness, rutting, and the fatigue cracking (both top-down and bottom-up) we re used to predict pavement construction performance. Once the MEPDG program is run, all the results of the expected distresses are produced. Then, a way to rationally combine the results into a metric that can be used to compare to the ICQI needs to be develope d. In order to correlate the results of the expected distresses and expected construction performance, the expe rt panel surveys for ICQI were conducted. At this time, approximately 15 FDOT personnel specializing in asphalt material, most from State Material Office in Gainesville, were asked to fill out the survey forms in Appendix C. Just as the previous CQI survey forms were prepared by the rules of AHP, the ICQI survey forms followed the rules of AHP. From the experience of the previous CQI expert panel meeting, the researcher decided to do the surveys individually. The rules of AHP require attendants to fill out the forms without discussion with others because strong opinion leaders can change other panel members opinions. The survey forms with instruc tions were sent and then returned by e-mail. The instructions for panel members answering ques tions on the ICQI form were same as the CQI survey forms. The results of the survey are presented in Appendix D for ICQI flexible pavements. The average values were entered into SuperDecisions software to determine the relationship between the expected distresses and pavement perf ormance. Once the relationship was found, the distresses were calculated to get the expected reli ability of a sample. This reliability value served as the independent variable in the regression analysis.
43 3.2 Model Formulation All FDOT projects are accep ted in accordance with one o r more construction specifications. The final product must pass certain crit eria to show that it meets the expectations of the designer to protect public safety and provid e the expected level of service. These criteria are enforced through testing of materials and work manship, and the results of these tests are what the CQI/ICQI model uses to calcul ate a CQI/ICQI for each project. In this research, two models are developed. In the CQI model, the CQI values for layers and AQC are obtained applying the percent within limits (PWL) concept, which is embedded in FDOT specifications. In the ICQI model, coeffici ents of the model are computed by statistical analysis (regression analysis) using the mechan istic-empirical pavement design guide (MEPDG). When the statistical analysis is executed, intera ctions between factors ar e also considered and analyzed. In addition, weighting factors for each layer for both CQI and ICQI models are computed using AHP. Details will be explained in later chapters. 3.2.1 Construction Quality Index (CQI) Model 184.108.40.206 Model concepts The CQI for mulation is based on the modular approach, allowing it to be scaled to all pavement construction projects. Figure 3-1 shows a conceptual pavement system consisting of a series of n layers. All existing layers in the pavement system do not have to be a part of the construction project. For example, a typical flex ible pavement resurfac ing project may involve rehabilitation of the friction cour se (Layer 1) and a portion of the structural course (Layer 2). In this case, the CQI model considers only the two layers, and according to the change, weighting factors for Layer 1 and Layer 2 are revised, w ith no other layers being considered in the calculations.
44 Layer 1 Layer 2 Layer n. Figure 3-1. Schematic of a pavement structure with n layers The general form of the CQI for a layered pavement system is given by layer layerCQIWCQIlayers (Equation 1) where Wlayer = weighting factor for layer i CQIlayer = construction quality index for layer i For each layer, the CQI is based on the sum of the Acceptance Quality Characteristics (AQC) for each layer times a weighting factor: AQC AQC AQC layercqiw CQI (Equation 2) where wAQC = weighting factor for AQC i cqiAQC = construction quality index for AQC i Finally, the construction quality index for each AQC is given by
45AQC AQCPWL cqi )()( (Equation 3) where (PWL)AQC is the percent within limits. (PWL)AQC is calculated based on statistical principles assuming that random samples are ta ken from a normally distributed population using the procedures outlined in Evaluation Procedures for Qual ity Assurance Specifications (Burati, et al. 2004). This procedure re quires the use of a PWL table, shown in Appendix E. The PWL concept is also applied to FDOT specifications. A Q value is determined from the difference between the sample mean ( X ) and the lower or upper specification limit (LSL/USL) divided by the samples standard deviation (s): s LSLX QL and s XUSL QU (Equation 4) A Q value is a quality index for its specification limit. For one-sided limits, the appropriate Q value is calculated and cross-referenced in F DOTs PWL table to find the PWL of that sample. Should the Q value be negative, the absolute value can be located in the tables, and the resulting PWL is subtracted from 100 to find the actual PW L. Two-sided limits require both Q values to be calculated and cross-referenced in the table. The two-sided percent within limits is then given by the difference between the sum of those two values and 100: 100 L U TPWL PWL PWL (Equation 5) For all PWL values that fall between rows in th e PWL table, linear in terpolation is used. 220.127.116.11 Model weighting factors The weighting factors are obtaine d u sing the analytic hierarchy process (AHP) concept that is a multi-criteria assessment tool for decision structuring and analysis (Isik et al. 2007). As mentioned earlier, the AHP is a special case of the ANP. The difference of the AHP from the ANP is that interconnections between decision f actors at the same level cannot be considered
46 with the AHP because the decision-making fram ework in an AHP model assumes a one-way hierarchical relationship among d ecision levels. AHP may not be an effective or correct decisionmaking tool to apply for some cases where ther e are interactions between decision variables. However, for the purpose of this research, AHP is a powerful and flexible decision-making technique that helps decision make rs to set priorities and choose the best alternative because interactions between AQCs are not considered in 2007 Standard Specifications for Road and Bridge Construction, which is the FDOT specification from which the AQCs are selected. Therefore, the AHP is selected to ob tain weighting factors for CQI model. Expert panel members from FDOT, the constr uction industry, academia, and consultants were requested to fill out forms prepared by the rules of AHP. As mentioned earlier, the results of the survey were obtained in Appendix D, and the average values for each pair-wise comparison were entered into SuperDecisions softwa re to determine the weighting factors for the CQI formulation. These flexible pavement weighting factors for the CQI are presented in Table 3-1. At any given pavement layer, the sum of the weighting factors is always 1. Table 3-1. Flexible pave ment weighting factors Pavement Component Weighting Factor, Wlayer Embankment 0.046 Stabilized Subgrade 0.074 Base Course 0.175 Superpave 0.400 Friction Course 0.305 Embankment Weighting Factor, wi Density 1.000 Stabilized Subgrade Weighting Factor, wi Density 0.617 LBR 0.383
47 Table 3-1. Continued Base Weighting Factor, wi Density 1.000 Superpave Weighting Factor, wi Passing #200 0.089 Passing #8 0.089 Air Voids 0.269 Asphalt Content 0.237 Density 0.316 FC5 Weighting Factor, wi Passing #8 0.096 Passing #4 0.107 Passing 3/8" 0.151 Asphalt Content 0.333 Ride Number 0.313 FC9.5 Weighting Factor, wi Passing #200 0.073 Passing #8 0.073 Air Voids 0.241 Asphalt Content 0.200 Density 0.198 Ride Number 0.215 FC12.5 Weighting Factor, wi Passing #200 0.073 Passing #8 0.073 Air Voids 0.241 Asphalt Content 0.200 Density 0.198 Ride Number 0.215
48 As mentioned earlier, typically fo r rehabilitation projec ts, one or more layers of the system will not be a part of the construction project. For example, in a typical rehabilitation job that includes milling and overlaying, only the structural Superpave and friction course layers may be constructed in the project while all other layers remain undistur bed from previous construction projects. In such cases, revised layer weigh ting factors are recalcul ated by weighting their respective contributions to the pr oject as shown in Table 3-2. Table 3-2. Sample calculation of revised layer weighting factors Layer Layer Weighting Factor Calculation of Revised Layer Weighting Factor Friction Course WFC = 0.305 WFC revised = 0.305/0.705 = 0.433 Superpave WSP = 0.400 WSP revised = 0.400/0.705 = 0.567 Total 0.705 1.000 18.104.22.168 Adaptation of the model for more than one asphalt mix For construction projects using Superpave a nd friction course layers, it is not uncomm on for several mixes with different target values for the certain AQC s to be involved in a project. For example, the CQI for the Superpave layer is adapted as follows: i mix i SPCQIt CQI (Equation 6) where ti is a tonnage weighti ng factor given by Superpave of tonstotal imix of tonsit (Equation 7) For example, suppose a construction project us ed three Superpave mi xes, designated by SP1, SP2, and SP3. A total of 20,000 tons was pl aced on the example project: 4,000 tons of SP1, 10,000 tons of SP2, and 6,000 tons of SP3. Table 33 explains how the Superpave layer CQI is calculated for this example. If that friction course has multiple mixes, the same process can be applied to adapt the CQI to the friction course layer.
49 Table 3-3 Sample calculation for multiple Superpave mixes Mix Tons Produced Mix CQI Calculation of ti Calculation of CQISP SP1 4,000 0.958 tSP1 = 4,000/20,000 = 0.200CQISP1 = 0.200.958 = 0.192 SP2 10,000 0.923 tSP2 = 10,000/20,000 = 0.500CQISP2 = 0.500.923 = 0.462 SP3 6,000 0.976 tSP3 = 6,000/20,000 = 0.300CQISP3 = 0.300.976 = 0.293 Total 20,000 CQISP = 0.947 22.214.171.124 Additional data required other than test results In order to calculate CQI, som e additional da ta other than AQC test results are required. These data included target values of each AQC a nd asphalt tonnage constructed for each asphalt mix. Target values are used to get the PWL of th e quality characteristics and eventually acquire the CQI of the same equality ch aracteristics. Target value repor ts were posted in the FDOT intranet. With permission from FDOT, the research er could gain access to the reports. A sample of the target value report is presented in Appendix F. From the target value reports in Appendix F, general information and specific target values for the mixes can be obtained. General informati on concerning asphalt mixes includes the asphalt production contractor, asphalt mix number, and type of the mix identifying whether the mix is for Superpave or friction course and whether dens ity of the mix is fine or coarse. For target values of the mixes, the aggregate passing percen tages of various sieves can be found in the front page of the report. In the second page, optimum asphalt content and target air void ratio can be located. The target air void ratio for the asphalt mi xes in the project list that the research team has was always 4%. In order to determine constructed asphalt we ight in a project where there are multiple asphalt mixes to determine revised weighting f actors for each mix, the tonnage report should be run through LIMS. For some reas on, the tonnage report worked fo r some project numbers but
50 not for others. For example, some projects have asphalt mix numbers and test data for AQCs, but there is no asphalt amount tested according to the tonnage report. Even though the reliability level of the tonnage report was not as good as was expected, it was the only report that contained the asphalt tonnage according to FDOT asphalt materi als personnel. Therefore, the research team used the tonnage report for the purpose of we ighting factors revision. However, when some tonnage values for design mixes were missing or too big to be realistic, results of the tonnage reports for the design mix were excluded from anal ysis. In such cases, the number of constructed lots for each design mix was used to determ ine the amount of constr ucted asphalt under the assumption that weight variation of the cons tructed asphalt for different design mixes is insignificant. 126.96.36.199 Model implementation The CQI model uses the Microsoft Windows operating system and is a stand-alone application called CQI Calculator. The application runs from one window and displays several screens to simplify and organize data entry. Data can be easil y imported or exported from text files, and reports in HTML format can be produced from the input data. At the current time, the application cannot read input files directly from LIMS. 3.2.2 Integrated Construction Quality Index (ICQI) Model 188.8.131.52 Mechanistic-empirical pave ment design guide (MEPDG) The Am erican Association of State Highway and Transportation Officials (AASHTO) Guide for Design of Pavement Structures is currently the primary document for designing highway pavements. The AASHTO Guide is em pirically based on performance equations developed using 1950s American Association of State Highway Official s (AASHO) Road Test data. Since the AASHO Road Test, the develo pment of revisions of the AASHTO Guide has been made. Recently, the limitations of the AASHT O Guide have been apparent, and the need
51 for developing an improved Design Guide has b een recognized. AASHTO decided to develop a design guide based as fully as possible on mechanistic principles. This MEPDG is the result of the decision. MEPDG is also known as NCHRP 1-37A Guide, 2002 Design Guide, New Design Guide, or Guide for M-E Design. It was developed to overcome limitations of design procedures based on the AASHO Road Test, which include consid ering only one climate condition, limited axle loads (2 million) in traffic, use of outdated vehicles and materials, and designing only new construction. The benefits of MEPDG are its wide range of pavement structure (new and rehabilitation) and direct consider ation of major factors such as tr affic, climate, materials, and support. The most notable advantage of this a pproach over the 1993 AASHTO Guide is that it is based upon a rigorous analytical and mechanistic approach usi ng the best available technology. MEPDG is designed to calculate the pavement response and predict the pavement life by entering more than 35 kinds of input for flex ible pavement. Even though it has not been calibrated for local conditions, the MEPDG is approved and under review for implementation by FHWA. The MEPDG software is based on ME desi gn concepts. This means that the design procedure considers pavement responses such as st resses, strains, and deflections, as well as the incremental damage over time. Damage over time is related to pavement distresses according to ME design concepts. 184.108.40.206 Model concepts In the ICQI developm ent, a combination of empirical methods and mechanistic methods are executed. Weighting factors for layers (Wlayer) obtained from AHP are used in the model, while the ICQI for layers (ICQIlayer) are obtained using MEPDG and regression analysis.
52 The general form of the ICQI model for a laye red pavement system is the same as CQI as shown in Equation 8. layer layers layerICQI W ICQI (Equation 8) where Wlayer = weighting factor for layer i (obtained from AHP) ICQIlayer = integrated construction quality inde x for layer i (obtai ned from MEPDG and regression analysis) For each layer, the ICQI will be acquired util izing a regression model. Regression analyses will be performed to determine the statistical relationship between a response and the variables. The following steps explain the ICQI developmen t procedures and how to get ICQI scores from the model. STEP 1 : Collect data for acceptance qua lity characteristics from LIMS. In order to retrieve data fr om LIMS, FDOT Construction and Materials officials from all districts including Turnpike district were asked to provide proj ect numbers to the study with their ratings. Even though the resear cher asked them to rate the projects in three levels, the rating categories from each district were different. As sh own in Table 3-4, some classified their projects in two categories as higher qual ity and lower quality, and some divided their projects ratings into good to excellent, averag e, and poor, while the others classified them as good, fair, and poor or excellent, good, fair, and average. In Table 3-4, the numbers in parentheses are the number of projects FDOT personnel provided regardless of whether the projects cont ained enough data for analysis. From District 2, 30 projects with ratings were supplied separately. The first 13 projects were provided for validation purposes for the CQI research projec t from FDOT, and the second 17 projects were requested and provided later for ICQI model development and validation.
53 Table 3-4. Project cl assification by FDOT District Project Ratings 1 & 7 Good to Excellent (12), Average (8), Poor (5) 2 (First Supply) Higher Quality (6), Lower Quality (7) 2 (Second Supply) Good (4), Average(12), Poor (1) 3 Good (9), Fair (5), Poor (3) 4 & 6 Excellent (1), Good (9 ), Fair (4), Average (1) 5 Good to Excellent (5), Average (3), Poor 2() Turnpike Excellent (3), Good (2) Poor (2), Very Poor (1) The researcher re-classified project ratings by FDOT personnel into three categories: good, average, and poor. Original FDOT project ratin gs of good to excellent, excellent, and higher quality are assigned to the good category. The average category contains fair and average. Poor, very poor, and lower quali ty comprise the poor category. This is shown in Table 3-5. Table 3-5 Project re-classification by the researcher District Good Average Poor 1 & 7 Good to Excellent Average Poor 2 Higher Quality, Good Average Lower Quality 3 Good Fair Poor 4 & 6 Excellent, Good Fair, Average 5 Good to Excellent Average Poor Turnpike Excellent, Good Poor, Very Poor The projects should be recent enough to use Superpave since the model is based on current FDOT specifications which embedded Superpave. The projects should be also recent enough to use LIMS as the data management system since retrieving data from previous data management system is quite impractical. The most important requirement was for the projects to have their relevant data stored in the LIMS database.
54 Once the project numbers were given, data retrieving from LIMS was executed. The retrieved data using Crystal Report from LIMS wa s saved to Excel spread sheets as shown in Figure 3-2 below, which has a sample Superpav e report. Among many test characteristics, five AQC are highlighted. However, the #8 sieve aggr egate passing rate and asphalt sample density were not required for ICQI model factors among the highlighted AQCs. The MEPDG program uses the #4 sieve aggregate passing rate for an input variable instead of the #8 aggregate sieve passing rate for Superpave. For developing and validating the ICQI mode l, the required quality characteristics of the CQI model are in. aggregate passing rate, 3/8 in. aggregate passing rate, #4 sieve aggregate passing rate, #200 sieve aggreg ate passing rate, asphalt content, and air void percentage because the MEPDG program can accept them as input variables. While each column represents a sample, all of the samples sometimes could not be used for the ICQI model without data correction. Note that the second and third samples do not have test values for density (Gmm). Even though the density (Gmm) is not entered into MEPDG software for analysis, the missing value can occur in other quality characteristics, for example asphalt content or air void percentage. Sometimes, asphalt reports retrieved inaccurate data as shown in Figure 3-2. The second, third, and fifth samples retrieved wrong values from LIMS. For example, the #8 sieve aggregate passing rate is 940%, where it must be between 0% and 100%, so the researcher had to search LIMS individually, and the correct value of the te st was 49.14%. In many cases, the extra effort to retrieve data individually was required. STEP 2 : Run MEPDG to get output for each sample. In the CQI, target values for AQC are requi red to obtain the CQI for the mix because the model was applied using the PWL concept. Howe ver, when the MEPDG calculates pavement
55 performance, it only requires test results of a mix regardless of the target range of the mix to predict pavement performance. In the ICQI model, layer-level ICQI equa tions for each layer will be developed using regression analysis. To develop the regression model (layer-level ICQI), sample data for a dependant variable is required. The MEPDG output using test results of the construction quality characteristics in asphalt reports from LIMS will be used as sample data for dependant variables. 93.85 94.52 93.89 Average Core % Gmm Core 5 Gmb 2.296 2.285 2.304 2.326 Core 4 Gmb 2.317 Core 3 Gmb 2.254 2.324 2.326 2.312 Core 2 Gmb 2.341 Core 1 Gmb 2.304 2.67 2.68 2.670 2.663 2.670 Gse 0.94 Dust/AC 0.94 0.86 0.94 0.84 72.46 70.13 74.54 70.01 70.93 % VFA 15.1 % VMA 15.20 15.07 15.40 15.07 4.15 4.51 3.88 4.52 4.48 % Air Voids 117 Hgt @ Nd 117.4 117.3 117.4 117.3 123.7 122.9 123.7 123.3 123.5 Hgt @ Ni 2.352 Gmb 2.353 2.347 2.344 2.353 2.455 2.463 2.448 2.458 2.454 Gmm 5.49 % A C 5 75 2 65 5 25 5 1 4.01 4.36 4.69 4 4.51 % Passing No.200 1280.9 % Passing No.100 11.96 1658.1 11.28 1655.6 26.87 27.43 28.37 26.54 28.49 % Passing No.50 948.2 % Passing No.30 36.96 34.82 37.36 34.66 40.4 849.1 43.27 40.5 43.52 % Passing No.16 714.2 % Passing No.8 52.91 940 52.87 940.7 66.1 64.69 70.38 649.8 69.2 % Passing No.4 88.94 % Passing 3/8in 92.39 86.05 89.7 88.81 95.52 96.2 % Passing 1/2in 97.74 93.46 95.86 100.00 100 98.83 % Passing 3/4in 100.00 98.91 2C005V 2 / 1 0700021086 E61001381-000 2C006Q 2 / 2 0700014546 E61001381-000 2C007Q 2 / 3 0700016752 E61001381-000 Project No.: 41551315201 Pay Item: 2 334 1 13 Design Mix: SP 05-4423B Mix Type: SP125C Contractor: RANGER Date Sample Taken: Sample L / S : LIMS ID : TIN : 2C004V 1 / 4 0700016451 E61001381-000 2C005Q 2 / 1 0700013846 E61001381-000 2/1/2007 2/2/2007 2/5/2007 1/31/2007 Figure 3-2. Sample asphalt (S uperpave) report from LIMS
56 To minimize variance of the model, two things are taken into consideration. First, when data is entered into the MEPDG program, all in put except asphalt quality characteristics for the structure category need to be invariable. Even though the MEP DG has a wide range of input requirements (traffic, climate, etc.), the ICQI only needs pavement performance as an output of the MEPDG using as many AQCs in CQI as possi ble. Of course, AQC values are entered according to the test results. Then, samples that dont have inaccurate test results for AQC will be used. For example, in Figure 3-2, the second, th ird, or fifth sample will not be used unless the incorrect data is corrected. To run the MEPDG software, input was need ed. The input for the software included general project information, traffic, climate, an d structure layering as shown in Figure 3-3. As mentioned earlier, traffic and c limate input remained the same, using default values. Only the test results were put into the struct ure category to run the MEPDG program. Figure 3-3. Layout of the MEPDG program for input and output (VonQuintus and Moulthrop 2007)
57 For the default value of the climate input, Gain esville climate data wa s used for all projects analyzed to eliminate variation except for cons truction quality characte ristics even though the projects are widely separated acro ss Florida. Figure 3-4 shows th at Gainesville data was entered into the climate input. Figure 3-4. Climate input Gainesville, FL Status of all input for the program can be di stinguished by colors as shown in Figure 3-5. Red means that data entry is required for the de sign process. The yellow input screens means that default values that are not yet verified and accep ted will be used for the design. Input that has been verified and accepted by the user is coded in green. To run the program, all input screens must be either yellow or green. After verifying traffic and climat e input, the data for the struct ure input was entered. In this research, six quality characteristics were filled wi th actual test results from LIMS. As shown in Figure 3-6, the required asphalt gradation for the MEPDG program includes cumulative % retained inch sieve, cumulative % retained 3/8 inch sieve, cumulative % retained #4 sieve, and
58 Figure 3-5. Color-coded inputs (V onQuintus and Moulthrop 2007) Figure 3-6. Asphalt material properties asphalt mix
59 % passing #200 sieve. Except for the #200 sieve passi ng rate, the retained pe rcentage instead of passing rate was required for the ot her properties. Therefore, the pa ssing rate for the properties in the asphalt report as shown in Figur e 3-2 must be converted to the retained rate for the sieves before entering the data into the MEPDG program. Effective binder content and air voids are the ot her two quality characte ristics to be entered into the MEPDG program. They can be found under the Asphalt Genera l tab in asphalt material properties as shown in Figure 3-7. Exce pt for the two quality characteristics, effective binder content and air voids, all the other cate gories will be filled with default values. Figure 3-7. Asphalt material properties asphalt general As seen on asphalt material properties in Figure 3-6 and Figure 3-7, analysis by this program is for one design or one sample. Once all input is entered into the program, the analysis of the sample is ready to begin. When the run is completed, the program creates a summary of all
60 input and output of the design. The summary of output is the distress an d performance prediction in both tabular and graphical formats. The output report is created in Microsoft Excel files. STEP 3 : Determine how to translate expected distresses to pavement performance in MEPDG. Once the MEPDG program run is complete, it generates a Microsoft Excel file with the expected pavement distresses a nd the International Roughness Inde x (IRI) as an output report. Table 3-6 is an example of an expected distress in tabular form at. The table shows that design life of the project is 15 years.
61Table 3-6. Sample of an expect ed distress in tabular format Fatigue Cracking: Proj ect 19384825201-fc4376B-01 Top Down at Surface Top Down at 0.5" Bottom Up at hac Reliability Pavement age mo yr Month Maximum Damage (%) Maximum Cracking (ft/mi) Location (in) Maximum Damage (%) Maximum Cracking (ft/mi) Location (in) Maximum Damage (%) Maximum Cracking (%) Location (in) Top Down Cracking (ft/mi) Bottom Up Cracking (%) 1 0.08 June 0.212 0.91 0 0. 00205 0 0 0.00305 0 0 474.63 1.45 2 0.17 July 0.857 7.61 0 0. 0035 0 0 0.00916 0 0 937.33 1.45 3 0.25 August 1.46 17 0 0.00476 0 0 0.0153 0.01 0 1240.89 1.45 4 0.33 September 1.61 19.8 0 0.00678 0 0 0.0178 0.01 0 1304.38 1.46 5 0.42 October 1.65 20.5 0 0. 00892 0.01 0 0.0188 0.01 0 1320.59 1.46 6 0.5 November 1.66 20.8 0 0. 0111 0.01 0 0.0189 0.01 0 1324.72 1.46 7 0.58 December 1.67 21 0 0.0125 0.01 0 0.0189 0.01 0 1328.74 1.46 8 0.67 January 1.68 21.1 0 0. 0139 0.01 0 0.0189 0.01 0 1332.64 1.46 9 0.75 February 1.69 21.4 0 0. 0158 0.02 0 0.019 0.01 0 1336.72 1.46 10 0.83 March 1.71 21.7 0 0.0179 0.02 0 0.0193 0.01 0 1344.54 1.46 11 0.92 April 1.74 22.3 0 0.0198 0.02 0 0.0203 0.01 0 1356.3 1.46 12 1 May 1.96 26.7 0 0.0212 0. 03 0 0.0243 0.01 0 1438.33 1.46 13 1.08 June 2.49 38.3 0 0.0224 0.03 0 0.0312 0.01 0 1611.28 1.46 169 14.1 June 44.5 2390 0 0.231 1.04 0 0.843 0.52 0 5369.74 1.97 170 14.2 July 45.5 2450 0 0.233 1.05 0 0.861 0.53 0 5434.03 1.98 171 14.3 August 46.7 2520 0 0.234 1.06 0 0.881 0.54 0 5508.95 2 172 14.3 September 47 2540 0 0. 235 1.07 0 0.891 0.55 0 5530.15 2 173 14.4 October 47 2550 0 0.237 1.08 0 0.893 0.55 0 5540.15 2 174 14.5 November 47 2550 0 0. 24 1.1 0 0.894 0.55 0 5540.15 2 175 14.6 December 47.1 2550 0 0.241 1.11 0 0.894 0.55 0 5540.55 2.01 176 14.7 January 47.1 2550 0 0.243 1.12 0 0.894 0.55 0 5540.55 2.01 177 14.8 February 47.1 2550 0 0. 245 1.13 0 0.895 0.55 0 5540.55 2.01 178 14.8 March 47.1 2550 0 0.247 1.15 0 0.896 0.55 0 5540.55 2.01 179 14.9 April 47.3 2560 0 0.248 1.16 0 0.903 0.56 0 5551.33 2.01 180 15 May 47.7 2590 0 0.25 1. 17 0 0.913 0.56 0 5582.89 2.02
62 Figure 3-8 is an example of an expected di stress in graphical format. In this graph, expected surface cracking at the e nd of 180 months (blue line) is much larger than design limits (red line). If this sample is for the real design of a pavement system, the expected cracking must not exceed the design limits, so a redesign might be required in this case. However, the fact that expected cracking is above design limits is not important because all of the other design criteria, such as traffic, climate, and structure properties, in the other layers of the pavement system are not properly designed for ICQI model developmen t. For the ICQI model development, only the relative ratio to target values of the mix will be required. Surface Down Cracking Longitudinal0 600 1200 1800 2400 3000 3600 4200 4800 5400 6000 01836547290108126144162180198Pavement Age (month) LongitudinalCr acking(ft/mi) Surface Depth = 0.5" Surface at Reliability Design Limit Figure 3-8. Sample of an expect ed distress in graphical format
63 As shown in Table 3-7, the MEPDG program generates a reliability summary for the expected pavement distresses and the IRI. Th e expected pavement distresses include surface down cracking, known as top-down cracking, bottom-up cracking, thermal cracking, and permanent cracking, known as rutting. Table 3-7. Sample of a reliability summary 1 Performance Criteria Distress Target Reliability Target Distress Predicted Reliability Predicted Acceptable Terminal IRI (in/mi) 172 90 118.2 94.99 Pass AC Surface Down Cracking (Long. Cracking) (ft/mile): 1000 90 2590 24.8 Fail AC Bottom Up Cracking (Alligator Cracking) (%): 100 90 0.6 99.999 Pass AC Thermal Fracture (Transverse Cracking) (ft/mi): 100 90 1 93.61 Pass Permanent Deformation (AC Only) (in): 0.25 90 0.83 0.25 Fail Permanent Deformation (Total Pavement) (in): 0.75 90 0.88 26.52 Fail Table 3-7 shows the output obtained from the MEPDG program after running a sample from a friction course mix in a District 1 pr oject. From the same mix in the same project, another sample was chosen for running the program and comparing the resu lts. Table 3-8 is the result of the run. Comparing these two tables fo r a reliability summary, D istress Predicted and Reliability Predicted have different values from each other; which means that the expected pavement performance will be different, so the way in which the predicted distress is related to pavement performance needs to be revealed. When permanent deformation for total pavement from the above two tables are compared, values for distress predicted are the same at 0.88 each; however, values for reliability predicted are different at 26.52 and 27.05, respectively. The reason is that the values for distress predicted are rounded, and values for reliability predicte d are more sensitive. In addition, values for
64 reliability predicted range from 0 to 100, which means a scale adjustment between distress categories is not required. Table 3-8. Sample of a reliability summary 2 Performance Criteria Distress Target Reliability Target Distress Predicted Reliability Predicted Acceptable Terminal IRI (in/mi) 172 90 117.8 95.18 Pass AC Surface Down Cracking (Long. Cracking) (ft/mile): 1000 90 229 66.27 Fail AC Bottom Up Cracking (Alligator Cracking) (%): 100 90 0.1 99.999 Pass AC Thermal Fracture (Transverse Cracking) (ft/mi): 100 90 1 93.61 Pass Permanent Deformation (AC Only) (in): 0.25 90 0.83 0.26 Fail Permanent Deformation (Total Pavement) (in): 0.75 90 0.88 27.05 Fail For these reasons, predicted reliability was us ed to find the relationship between distresses and pavement performance. The definition of target design reliability is th e probability that the pavement will not exceed performance criterion limits over the design period. The relationship between distresses and pavement performance were sought using the expert panel survey prepared by the rules of AHP. As mentioned in chapter 220.127.116.11, the results of the survey were presented in Appendix D for ICQI flexible pavements. The average values were entered into SuperDecisions software to dete rmine the relationship between the expected distresses and the pavement performance. Supe rDecisions hierarchical model for weighting distresses to convert the summary of distresses to pavement perf ormance is shown in Figure 3-9. Once the hierarchical model was set up, Figur e 3-10 shows pair-wise comparisons with respect to the goal when mean values in Appe ndix D for ICQI flexible pavement are entered. Numbers in the matrix are the dominance judgment th at is derived from the expert panel survey. The blue number in the matrix indicates the element listed at the left is more important than the
65 Figure 3-9. SuperDecisions hierarchi cal model for weighting distresses element listed at the top. The red number means that the element listed at the top is dominant. A judgment of 1.0 means that they are equally important, and a j udgment of 5.0 means strongly or five times as much if the categories are meas urable. The maximum number of judgment is 9. Judgments greater than 9 may be entered, but it is suggested that they be avoided. In those cases, the hierarchical structure should be re-organized so that such a comparison is not required. After entering average values from the survey results of ICQI flexible pavement, the results of the pair-wise comparis ons are shown in Figure 3-11; the inconsistency is shown in the
66 top part of the properties screen. At 0.07, it is less than 0.10, so no correction of judgment is needed. Figure 3-10. Matrix pair -wise comparison screen Figure 3-11. Results of the pair-wise comparisons Once the relationship is found, the distresses ar e calculated to get the expected reliability of a sample. For example, Table 3-7 and Table 38 show reliability summaries after running the
67 MEPDG program. Each table represents expected distresses of one sample. Table 3-9 shows how to convert predicted distresse s to overall performance reliabil ity of the samples. When the overall reliability is obtained from a sample (here, 42.39 and 55.25 are the overall reliability of each sample), it is ready to go to the next step, which is developing a regression model for each layer. Table 3-9. Conversion of predicted distresses to overall reliability Sample in Table 3-8 Sample in Table 3-9 Weight Reliability Predicted Revised Reliability Reliability Predicted Revised Reliability Top-Down Cracking 0.3039 24.80 7.54 66.27 20.14 Bottom-Up Cracking 0.1804 99.999 18.04 99.999 18.04 IRI (Roughness) 0.0458 94.99 4.35 95.18 4.36 Rutting 0.4699 26.52 12.46 27.05 12.71 SUM 42.39 55.25 STEP 4 : Develop regression model for each layer. The MEPDG program has no target values for input, which means that the MEPDG produces its output according to th e input value without considera tion of specification limits or target values. Therefore, to compare sample qualities when each sample has different target values (or specification limits), the response should be th e ratio between MEPDG output using test (sample) data and MEP DG output using target value. For example, overall reliabilities of 42.39 and 55.25 in Table 3-10 do not have any meaning in themselves. The overall reliability shoul d be compared to target values of the same mixes. Table 3-10 explains why the overall reliabili ty should be compared to target values of the same asphalt mixes to be used in developing a regression model.
68 In Table 3-10, samples in Table 3-7 and Ta ble 3-8 are taken as examples. The overall reliability of sample A is 42.39, whereas the over all reliability of sample B is 55.25. This does not necessarily mean that asphalt pavement was laid better in sample A than in sample B. Even though overall reliability of sample B is larger th an that of sample A, the ratio between the sample and target value for sample A is larger than the ratio for sample B. This means that in the good sample, construction was executed close to the design target, or even exceeds the design target quality. Overall re liability of sample A, 42.39, can be regarded as in the good category when overall reliability of th e target value is 41.05. As seen in Table 3-10, unlike the CQI model, sample quality can be better than target quality; accordingly, the ratio can exceed 1, or 100%. Table 3-10. Conversion example of the overall reliability to ratio to target value Sample A Sample B Overall Reliability of a Sample (1) 42.39 55.25 Overall Reliability of Target Value (2) 41.05 56.20 Ratio (1)/(2) 1.033 0.983 In addition, the variables are construction qua lity characteristics such as asphalt content and air void ratio. By using the ratio between test (sample) MEPDG output and target MEPDG output, pavement performance from MEPDG ca n be properly estimated under the assumption that pavement performance from MEPDG is 100% when target values for the mix are entered. In the Equation 9 below, y is the same value as the ratio in Table 3-10. Several assumptions are required for the re gression model, which is expressed by ICQIlayer = y = ( )arg( )( ett sampley y) =0 + 1 x1 + 2 x2 + + p x1 x2 + + (Equation 9)
69 where y(sample) is the pavement performance from MEP DG when test results from samples are entered, and y(target) is the pavement performance from MEPDG when target values for the samples are entered. The assumptions of the model are (Rawlings et al. 1998): y is regarded as the response that corresponds to the levels of the explanatory variables x1, x2, xp. 0, 1, p are assumed to be the coefficients in the linear relationship. If there is a single factor (p = 1) for the equation, 0 will be the intercept, and 1 will be the slope of the straight line defined. 1, 2 n are assumed to be errors that make a scatter pattern around the linear relationship at each of the n observations. These errors are mutually independent, are normally distributed, and have a zero mean and variance, under the assumption of the regression model. STEP 5 : Generate sample-level ICQI for each sample. The developed regression model in step 4 is an ICQI model which was used to calculate layer-level ICQI. By entering th e construction quality characteri stics values to the model, sample-level ICQI can be calculated for the laye r. For example, if there are 30 samples in a specific layer, 30 ICQI ar e obtained for the layer. STEP 6 : Compute ICQIlayer from sample-level ICQI. In the CQI model development, the PWL con cept was used. The basic assumption of the concept is that variability of quality can affect performance. However, in the ICQI model, mechanistic-empirical and mathematical (reg ression) approaches are used, so the ICQIlayer can be computed to the average value of the calculat ed sample-level ICQI for the layer from step 5. When there are several mixes in a layer, the same concept used in chapter 18.104.22.168. is applied. STEP 7 : Multiply weighting factors for each layer and ICQIlayer to get the project-level ICQI.
70 As shown in Equation 8 earlier, the genera l form of the ICQI model for a layered pavement system is as follows: layer layers layerICQI W ICQI (Equation 8) where Wlayer = weighting factor for layer i ICQIlayer = integrated construction qual ity index for layer i (Step 5)
71 CHAPTER 4 DATA ANALYSIS 4.1 Overview In order to validate the model, the research team asked FDOT to provide projects to study. The m ost important requirement was that the proj ects have their relevant data stored in the LIMS database. It was also requested that FDOT provide a level of satisfaction, or rating of good or poor, for each project provided. FDOT supplied 105 flexible paving projects from all districts for re view. The projects consisted of 12 new or reconstruction projects that require earth wo rk and 93 resurfacing projects that require only asphalt work. Details of these projects such as constructed district, project number, construction type, and FDOT rating, can be seen in Table 4-1. Table 4-1. All provided projects for validation from FDOT District Project Number Cons truction Type FDOT Rating 1 404201-1-52-01 Resurfacing Average 1 193848-2-52-01 Resurfacing Good 1 194437-2-52-01 Resurfacing Average 1 201015-2-52-01 Resurfacing Good 1 194100-2-52-01 Resurfacing Good 1 194172-2-52-01 Resurfacing Poor 1 195736-1-52-01 Add Lanes & Rehabilitation Poor 1 197308-2-52-01 Resurfacing Average 1 197291-2-52-01 Resurfacing Good 1 197388-2-52-01 Resurfacing Average 1 197309-2-52-01 Resurfacing Average 1 197679-1-52-01 Add Lanes & Reconstruction Poor
72 Table 4-1. Continued District Project Number Cons truction Type FDOT Rating 1 411862-1-52-01 Resurfacing Average 1 196960-3-52-01 Resurfacing Good 1 196960-2-52-01 Resurfacing Good 1 197007-2-52-01 Resurfacing Good 2 207545-2-52-01 Resurfacing Average 2 208085-2-52-01 Resurfacing Good 2 208363-1-52-01 Add Lanes & Reconstruction Average 2 208366-2-52-01 Resurfacing Good 2 209301-1-52-01 New Road Construction Average 2 209543-2-52-01 Resurfacing Average 2 209648-3-52-01 Miscellaneous Construction Poor 2 209949-2-52-01 Resurfacing Average 2 209970-2-52-01 Resurfacing Average 2 210221-2-52-01 Resurfacing Average 2 210273-2-52-01 Resurfacing Average 2 210374-2-52-01 Resurfacing Good 2 210384-2-52-01 Widening/Resurfacing Average 2 210432-2-52-01 Resurfacing Average 2 210889-3-52-01 Resurfacing Average 2 210949-2-52-01 Resurfacing Good 2 213520-2-52-01 Resurfacing Average 2 207947-2-52-01 Resurfacing Good 2 207956-2-52-01 Resurfacing Poor
73 Table 4-1. Continued District Project Number Cons truction Type FDOT Rating 2 207956-3-52-01 No Data 2 208200-2-52-01 Resurfacing Poor 2 208226-4-52-01 Resurfacing Good 2 209166-3-52-01 Resurfacing Poor 2 209692-2-52-01 Resurfacing Poor 2 209999-1-52-01 Add Lanes & Reconstruction Poor 2 210004-1-52-01 No Data 2 210253-1-52-01 Add Lanes & Reconstruction Poor 2 213251-2-52-01 Resurfacing Good 3 217947-1-52-01 Add Lanes & Reconstruction Good 3 217948-1-52-01 Add Lanes & Reconstruction Poor 3 409022-1-52-01 Resurfacing Average 3 413426-1-52-01 Resurfacing Good 3 411389-1-52-01 Resurfacing Average 3 408878-1-52-01 Resurfacing Good 3 409017-1-52-01 Resurfacing Good 3 411391-1-52-01 Resurfacing Good 3 411395-1-52-01 Resurfacing Good 3 218539-1-52-01 Add Lanes & Reconstruction Poor 3 409006-1-52-01 Resurfacing Poor 3 403930-1-52-01 Resurfacing Good 3 411697-1-52-01 Resurfacing Good 3 406326-1-52-01 Resurfacing Average
74 Table 4-1. Continued District Project Number Cons truction Type FDOT Rating 3 411396-1-52-01 Resurfacing Good 3 411397-1-52-01 Resurfacing Average 3 413442-1-52-01 Resurfacing Average 4 411321-1-52-01 Resurfacing Fair 4 411322-1-52-02 Resurfacing Good 4 231726-1-52-01 Resurfacing Average 4 231735-1-52-02 Resurfacing Good 4 415397-1-52-01 Resurfacing No rating 4 231737-1-52-01 Resurfacing Good 4 411323-1-52-01 Resurfacing Fair 4 227861-1-52-01 Resurfacing Fair 4 228110-1-52-01 Resurfacing Fair 4 228188-1-52-01 Resurfacing Excellent 4 228615-1-52-01 Resurfacing Good 4 413801-1-52-01 Resurfacing No rating 4 411441-1-52-01 Resurfacing Good 4 228135-1-52-01 Intersection Good 4 403605-1-52-01 Resurfacing Good 4 403619-1-52-01 Resurfacing Good 5 415513-1-52-01 Resurfacing Average 5 419993-1-52-01 No Data 5 411603-1-52-01 Resurfacing Good 5 413585-1-52-01 Resurfacing Average
75 Table 4-1. Continued District Project Number Cons truction Type FDOT Rating 5 415514-1-52-01 Resurfacing Average 5 415512-1-52-01 Resurfacing Good 5 239725-1-52-01 Add Lanes & Reconstruction Poor 5 417155-1-52-01 Resurfacing Good 5 415526-1-52-01 Resurfacing Good 5 413583-1-52-01 Resurfacing Good 6 250548-2-52-01 Reconstruction Good 7 411332-1-52-01 Resurfacing Good 7 411266-1-52-01 Resurfacing Good 7 406560-1-52-01 Resurfacing Good 7 257076-1-52-01 Resurfacing Poor 7 411277-1-52-01 Resurfacing Average 7 408913-1-52-01 Resurfacing Average 7 413413-1-52-01 Resurfacing Good TPK 232352-1-52-01 No Data TPK 406092-1-52-01 Add Lanes & Reconstruction Good TPK 411533-1-52-01 Resurfacing Good TPK 411532-1-52-01 Resurfacing Excellent TPK 413623-2-52-01 Resurfacing Excellent TPK 413623-1-52-01 Resurfacing Excellent TPK 413669-1-52-01 Resurfacing Poor TPK 406102-1-52c-01 Interchange Poor
76 The researcher had significant problems in the procuring of data from LIMS. LIMS seemed to be an efficient tool that different le vels of users could employ to enter various test results; however, it was a difficult system from wh ich to retrieve data. When LIMS specialists were asked to show how to retrieve particular da ta, only a few of them were able to do so. Even after the data was procured, it became apparent that large portions of important data had not been entered into the database. Another problem in getting correct data wa s that the LIMS report sometimes showed incorrect data without specific reasons. For example, the researcher needed to know the #8 sieve passing rate, which must be between 0 and 100. The LIMS report sometimes retrieved #8 sieve passing weight instead of passing ra te, showing hundreds or thousands for passing percentage as a result. Therefore, once that type of incorrect da ta was procured, the researcher had to check and re-retrieve data individually. 4.2 Data Deduction Each pavement system consists of several laye rs with different func tions and materials. Embankment, subgrade, base, Superpave, and friction courses are among possible flexible pavement layers. As shown in Table 4-1, constr uction type of the flexible pavement can be divided into two categories: resurfaci ng construction or reconstruction. Resurfacing construction generally requires milling of top layers of the flexible pavement system and does not require earth work such as embankment, subgrade, or base. Therefore, the maximum number of possible layers for resurfacing construction is two: Superpave and friction course. However, it was impossible to know whether the actual number of layers for a particular resurfacing construction project was one or two because the am ount of information available through LIMS was limited.
77 Reconstruction and new construction require earth work that includes embankment, subgrade, or base according to design. Therefore, the maximum number of possible layers for reconstruction or new construction is five: embankment, subgrade, base, Superpave, and friction course. The actual number of layers for rec onstruction or new construction also depended upon the design of a particular project. For more reliable analysis, some projects were excluded from the pr oject list provided by FDOT. The first factor to consider was the numbe r of layers constructed. When the number of constructed layers is over 50 pe rcent of the possible number of layers, then CQI analysis was done. As mentioned earlier, some projects, such as straight asphalt resurfacing projects, have only two layers; therefore, anal ysis was done to determine the CQI for any project for which data was available for either Superpave or friction course layers constructed. For new construction or reconstruction proj ects, at least three constructe d layers containing data were required to determine the CQI because three layers was the minimum requirement for over 50 percent of the possible number of layers. Howeve r, if a projects three constructed layers were all earth work, such as embankment, subgrade, and base, then the project was excluded from the CQI analysis because the sum of earth work laye rs CQIs could represent only slightly less than 30% of the projects CQI. Another factor to consider was the number of samples that were tested and those test results entered into the LIMS database. Extremely low sample counts greatly reduce the reliability of any model result; therefore, any project that had less than 30 samples was excluded from the CQI analysis.
78 Table 4-2 below shows all the projects that we re excluded from the CQI analysis and the reasons for the exclusions. As shown in Table 4-2, some projects were eliminated from the analysis because no data were found in the LIMS or no rating from FDOT was provided. Table 4-2. Eliminated projects from CQI analysis District Project Number Max. Possible Layers No. Layers of Data No. of Samples Reason for Elimination 1 195736-1-52-01 5 2 11 Not enough number of layers 1 403890-1-52-01 No Data 1 196960-3-52-01 2 2 15 Not enough number of samples 1 196960-2-52-01 2 2 15 Not enough number of samples 1 197007-2-52-01 2 2 30 No target value 1 201015-2-52-01 2 1 23 Not enough number of samples 2 208363-1-52-01 5 2 232 Not enough number of layers 2 210374-2-52-01 2 2 13 Not enough number of samples 2 210949-2-52-01 2 2 25 Not enough number of samples 2 207956-3-52-01 No Data 2 208226-4-52-01 2 1 2 Not enough number of samples 2 209999-1-52-01 5 2 104 Not enough number of layers 2 210004-1-52-01 No Data 2 210253-1-52-01 5 1 34 Not enough number of layers 3 409022-1-52-01 2 2 22 Not enough number of samples 3 411391-1-52-01 2 2 16 Not enough number of samples
79 Table 4-2. Continued District Project Number Max. Possible Layers No. Layers of Data No. of Samples Reason for Elimination 3 411395-1-52-01 2 2 9 Not enough number of samples 3 218539-1-52-01 5 2 74 Not enough number of layers 3 413442-1-52-01 2 2 24 Not enough number of samples 4 415397-1-52-01 No rating from FDOT 4 413801-1-52-01 No rating from FDOT 4 411441-1-52-01 2 2 22 Not enough number of samples 4 228135-1-52-01 No Data 4 403605-1-52-01 2 2 18 Not enough number of samples 5 419993-1-52-01 No Data 5 415514-1-52-01 2 1 16 Not enough number of samples 5 239725-1-52-01 5 2 27 Not enough number of layers 5 417155-1-52-01 2 2 24 Not enough number of samples 5 413583-1-52-01 2 1 25 Not enough number of samples 7 408913-1-52-01 2 2 31 No tonnage for mixes 7 411266-1-52-01 2 2 61 No tonnage for mixes TPK 232352-1-52-01 No Data After 32 projects were eliminated from CQI an alysis for various reasons, 73 projects from all 8 districts were selected for CQI analysis. Table 4-3 below shows the list of the projects, including district in which the projects were constructed, projec t number, construction type, and ratings from FDOT personnel.
80 Table 4-3. Selected projects for CQI analysis District Project Number Cons truction Type FDOT Rating 1 404201-1-52-01 Resurfacing Average 1 193848-2-52-01 Resurfacing Good 1 194437-2-52-01 Resurfacing Average 1 194100-2-52-01 Resurfacing Good 1 194172-2-52-01 Resurfacing Poor 1 197308-2-52-01 Resurfacing Average 1 197291-2-52-01 Resurfacing Good 1 197388-2-52-01 Resurfacing Average 1 197309-2-52-01 Resurfacing Average 1 197679-1-52-01 Add Lanes & Reconstruction Poor 1 197368-2-52-01 Resurfacing Good 1 411862-1-52-01 Resurfacing Average 2 207545-2-52-01 Resurfacing Average 2 208085-2-52-01 Resurfacing Good 2 208366-2-52-01 Resurfacing Good 2 209301-1-52-01 New Road Construction Average 2 209543-2-52-01 Resurfacing Average 2 209648-3-52-01 Miscellaneous Construction Poor 2 209949-2-52-01 Resurfacing Average 2 209970-2-52-01 Resurfacing Average 2 210221-2-52-01 Resurfacing Average 2 210273-2-52-01 Resurfacing Average 2 210384-2-52-01 Widening/Resurfacing Average
81 Table 4-3. Continued District Project Number Cons truction Type FDOT Rating 2 210432-2-52-01 Resurfacing Average 2 210889-3-52-01 Resurfacing Average 2 213520-2-52-01 Resurfacing Average 2 207947-2-52-01 Resurfacing Good 2 207956-2-52-01 Resurfacing Poor 2 208200-2-52-01 Resurfacing Poor 2 209166-3-52-01 Resurfacing Poor 2 209692-2-52-01 Resurfacing Poor 2 213251-2-52-01 Resurfacing Good 3 217947-1-52-01 Add Lanes & Reconstruction Good 3 217948-1-52-01 Add Lanes & Reconstruction Poor 3 413426-1-52-01 Resurfacing Good 3 411389-1-52-01 Resurfacing Average 3 408878-1-52-01 Resurfacing Good 3 409017-1-52-01 Resurfacing Good 3 409006-1-52-01 Resurfacing Poor 3 403930-1-52-01 Resurfacing Good 3 411697-1-52-01 Resurfacing Good 3 406326-1-52-01 Resurfacing Average 3 411396-1-52-01 Resurfacing Good 3 411397-1-52-01 Resurfacing Average 4 411321-1-52-01 Resurfacing Fair 4 411322-1-52-02 Resurfacing Good
82 Table 4-3. Continued District Project Number Cons truction Type FDOT Rating 4 231726-1-52-01 Resurfacing Average 4 231735-1-52-02 Resurfacing Good 4 231737-1-52-01 Resurfacing Good 4 411323-1-52-01 Resurfacing Fair 4 227861-1-52-01 Resurfacing Fair 4 228110-1-52-01 Resurfacing Fair 4 228188-1-52-01 Resurfacing Excellent 4 228615-1-52-01 Resurfacing Good 4 403619-1-52-01 Resurfacing Good 5 415513-1-52-01 Resurfacing Average 5 411603-1-52-01 Resurfacing Good 5 413585-1-52-01 Resurfacing Average 5 415512-1-52-01 Resurfacing Good 5 415526-1-52-01 Resurfacing Good 6 250548-2-52-01 Reconstruction Good 7 411332-1-52-01 Resurfacing Good 7 406560-1-52-01 Resurfacing Good 7 257076-1-52-01 Resurfacing Poor 7 411277-1-52-01 Resurfacing Average 7 413413-1-52-01 Resurfacing Good TPK 406092-1-52-01 Add Lanes & Reconstruction Good TPK 411533-1-52-01 Resurfacing Good TPK 411532-1-52-01 Resurfacing Excellent
83 Table 4-3. Continued District Project Number Cons truction Type FDOT Rating TPK 413623-2-52-01 Resurfacing Excellent TPK 413623-1-52-01 Resurfacing Excellent TPK 413669-1-52-01 Resurfacing Poor TPK 406102-1-52c-01 Interchange Poor 4.3 Model Validation Once projects to be analyzed were id entifie d, the available data for each project were entered into the CQI and ICQI models. Then, calculated CQI and ICQI was grouped by district or score to compare to FDOT ratings. 4.3.1 Construction Quality Index (CQI) Model 22.214.171.124 Validation process As explained earlier, a p rojects CQI is the sum of each layers CQI of the project. Unfortunately, for several projects, it is clear that some data we re missing or irretrievable from LIMS. In addition, of course, not every project should have data for every possible layer. In resurfacing construction, usually only the top two layers of the pavement systemfriction course, and Superpaveare cons tructed; therefore, when ther e are missing layers, a weight correction of the layers with data should be considered. As explained in Table 3-2 in chapter 126.96.36.199, in order to make the sum of the remaining layers weights 100 percent, or 1.00, the revise d layer weighting factors are rearranged by weighting their respective contri butions to the projects. A wei ght correction of parameters followed the same rule. For example, if a density and ride number are the only two parameters for a friction course layer out of six parameters, then the revi sed weighting factors of
84 density and ride number should be recalculated to make the sum of the factor s 1, or 100 percent. Then, the revised CQIs for density and ride num ber represent the CQI for the entire friction course. This arrangement was not ideal, so the researcher avoided analyzing these projects. When the researcher asked for the project num bers to study, a total of 105 projects were initially provided. Among the 105 projects, 12 were categorized as reconstruction or new construction that had embankment, subgrade, or base in their pavement system. After data deduction, seven reconstruction or new construction projects were eliminated from analysis, so five reconstruction projects were included in the 73 projects to be analyzed. Because of the limited number of reconstruction projects, comp arison between reconstr uction and resurfacing construction could not be done. 188.8.131.52 The CQI model validation through all projects from every district All projects were com pared by FDOT rating cat egories. Figure 4-1, Figure 4-2, and Figure 4-3 show the CQI results of projects classified as good, average, and poor by FDOT, respectively. In these figures, the CQI results are displaye d by three columns representing overall CQI, friction course CQI, and Superpav e CQI, respectively. Of course, so me projects have layers of embankment, subgrade, or base in their pavement constructio n. However, most projects are resurfacing projects that only have layers of friction course and Superpave; in addition, the sum of the layer weighting factors of the friction course and Superpave is 0.705. This means the combined CQIs of the friction course and Superpave represent an overall CQI for resurfacing projects and 70% of the CQI for reconstruction projects. As seen in Figure 4-1, Figure 4-2, and Figure 4-3, some projects CQIs are affected severely by one layers extremely low CQI. Fo r most of these cases, thats because one particular parameters CQI is very low. For example, when all other parameters CQIs are
85 marked normally, ranging from 0.7 to 1.0, one parameters CQI was 0.05. When this low parameter CQI is included in a project, the CQI for the project usually marks much lower than expected and may not represen t the projects performance. In Figure 4-1, CQIs of 32 projects from ever y district are arranged in the order of CQI score. Projects with lower CQI among this grou p show a tendency to have an extremely low layer CQI of either friction course or Superpav e rather than having laye r CQIs of both friction course and Superpave in the normal range. The nu mber of projects whose CQI is greater than 0.9 is 15 out of 33, while the number of projects whose CQI is less than 0.8 is four. CQIs for Good Projects0.4 0.5 0.6 0.7 0.8 0.9 12281 88 1 231735 1 4116 97 1 4116 03 1 4 1 1 3 96 1 2080 85 2 2179471 4136 23 2 4088 78 1 2 1 3251 2 4136 23 1 4113321 1938 48 2 4090 1 71 2 0 8366 2 1973 68 2 4134131 Project NumbersCQI Overall CQI Friction Course Superpave Figure 4-1. Distribution of CQIs for good projects In Figure 4-2, 28 projects from every district are arranged in the order of CQI score. Projects with lower CQI among this group showed the same tendency as the group of good projects, having an extremely low layer CQI of either friction course or Superpave. Upon close investigation, it was revealed that four projects out of last six projects had extremely low ride
86 number (roughness) CQIs. The number of projects whose CQI is greater than 0.9 is five out of 28, while the number of projects w hose CQI is less than 0.8 is six. Figure 4-3 shows CQIs of 12 pr ojects arranged in the order of CQI score. Unlike the good CQIs for Average Projects0.4 0.5 0.6 0.7 0.8 0.9 12278 61 1 2103842 2104 3 22 1944 3 72 2095 43 2 2102 21 2 2099 4 92 2108 8 93 4118 62 1 2099 70 2 4042 01 1 4113 2 31 4063 26 1 4113 21 1Project NumbersCQI Overall CQI Friction Course Superpave Figure 4-2. Distribution of CQIs for average projects CQIs for Poor Projects0.4 0.5 0.6 0.7 0.8 0.9 12 1 7948-1 2 5 7076-1 409 0 061 2 07 9 5 6 2 2 09 6 9 2 2 2 0 8200-2 4 1 3669-1 2 09 1 6 6 3 2 09 6 4 8 3 1 97 6 7 9 1 4 0 6102-1 1 9 4172-2Project NumbersCQI Overall CQI Friction Course Superpave Figure 4-3. Distribution of CQIs for poor projects
87 and average groups, projects with lower CQI among this group ha ve low layer CQIs of both friction course and Superpave. That means lo w CQIs of the projects are not caused by a particular parameter performance but, rather, by each parameters unsatisfactory performance. The number of projects whose CQI is greater than 0.9 is three out of 12, while the number of projects whose CQI is less than 0.8 is three. Table 4-4 shows mean value, number of pr ojects whose CQI is greater than 0.9, and number of projects whose CQI is less than 0.8. Good-rated proj ects have a higher mean value and a higher number of projects whose CQI is greater than 0.9 than the other two groups; however, the average group and th e poor group show almost the same results. The mean CQI has a difference of 0.003, and the proportion of nu mber of projects whose CQI is greater than 0.9 or less than 0.8 is very close. Table 4-4. Summary of the CQI projects by FDOT ratings FDOT Raring Good Average Poor Mean CQI 0.874 0.844 0.847 Number of total projects 33 28 12 Number of projects (CQI>0.9) 15 5 3 Number of projects (CQI<0.8) 4 6 3 One possible reason is, first, the FDOT ra ting from one personnel member cannot be the same as another members rating. For example, Di stricts 4 and 6 personnel rated their levels of satisfaction as good, fair, and average, while most other districts classified projects as good, average, and poor. In this case, it is doubtful that the average in Districts 4 and 6 have the same meaning (or quality) as the other districts averag e; therefore, in order to remove the ambiguity, comparison of only the good and poor groups might be a reasonable approach.
88 Figure 4-4 shows CQIs of good projects and poo r projects. Even though there are several projects with very low CQI in the good proj ect group, generally CQIs of good projects are higher than the poor group. Even though not all FDOT-rated good projec ts have a higher CQI than any FDOT-rated poor projects, as seen in Table 4-4 and Figure 4-4, good projects had a better mean CQI value than poor projects. Figure 4-4. Overall CQIs of good and poor projects 184.108.40.206 The CQI model validation by district As m entioned in the previous chapter, FDOT ratings were done by different persons in different districts; thus, they may not be consistent. In this chapter, then, CQI comparison by district is discussed. Table 4-5 shows the compar ison of CQI means by district. Districts 4 and 6 and District 5 did not have poor projects to analyze; in fact, Di stricts 4 and 6 did not have a Overall CQI 0.4 0.5 0.6 0.7 0.8 0.9 1228188-1 231735-1 411697-1 411603-1 411396-1 208085-2 217947-1 413623-2 408878-1 213251-2 413623-1 411332-1 193848-2 409017-1 208366-2 197368-2 413413-1 257076-1 207956-2 208200-2 209166-3 197679-1 194172-2Project NumbersCQI Overall CQI Good Projects Poor Projects
89 poor category at all. District 5 had one project classified as poor in the list they provided, but the poor project was excluded from the analysis because of a lack of data. Except for Districts 1 and 7 and District 3, good project groups of other districts have higher mean values of CQI than poor project grou ps. In Districts 1 and 7 and District 3, the means of the poor projects are higher than thos e of the good projects. However, it should be noted that because the project rating groups are divided by six districts, the number of samples of each category are limited. In particular, the number of poor projects by districts is very small because total number of poor projects to be anal yzed was originally only 12. Districts 1 and 7 had three poor projects. Even though two of them had low CQ I, the average CQI was much higher because the other project had th e second highest CQI in the district. Table 4-5. Mean value of th e CQI projects by districts District FDOT Rating 1 & 7 2 3 4 & 6 5 Turn Pike Good 0.783 0.887 0.891 0.925 0.900 0.880 Average 0.815 0.854 0.825 0.856 0.894 Poor 0.799 0.863 0.916 0.809 4.3.2 Integrated Construction Quality Index (ICQI) Model ICQI m odels for only friction course and S uperpave layers were developed in this research. The number of projects that had embankment, subgrade, and base layers were too small to develop a model. In the development of the CQI model, actual test data was not required because the CQI model obtained its weig hting factors from AH P without test data. However, not having a high enough number of projects that incl ude construction of embankment, subgrade, and base layers is a prob lem because actual test data is required in developing ICQI.
90 220.127.116.11 Validation process A total of 1,334 data sets were used to develop a regression m odel for the ICQI. Among those data sets, 487 were used for friction cour se and 847 for Superpave. T he data sets were derived from LIMS and the MEPDG output report af ter running the program. From the data sets, all six construction quality char acteristics used in the MEPDG program and 15 combinations of two characteristics, a total of 21 f actors, were chosen for the regres sion analysis at the first stage. Combinations of two characteristics whose form at is multiplication of the two characteristics were selected to consider interactions between ch aracteristics; however, a factor (3/4 inch sieve passing rate) was removed from the first regression equation because most of the values were either 100 or very close to 100. The regression analysis disclo sed quite interesting results comparing the CQI model. Air void ratio, which is the most im portant factor in the CQI mode l for friction course 9.5 and 12.5 as seen in Table 3-1, is also the most affecting factor in the ICQI model for both friction course and Superpave. Some construction quality charac teristics such as ride number, #8 sieve passing rate, and density were not used as factors of the ICQI model. Among the unused construction quality characteristics as inputs of the MEPDG, the ride number, representing roughness, was rather used as one of the performance indicators of the ICQI model. The #200 sieve passing rate, which is the leas t important factor in the CQI model for friction course 9.5 and 12.5 as seen in Table 3-1, is the second most affecting factor, next to air void ratio, in the ICQI model for bot h friction course and Superpave. Friction course regression model: The results of the regression analysis revealed that at the 95% confidence level, ten of the 21 factors were statistically significant. Tables 4-6 and 4-7 show the regression and the ANOVA statistics fo r the final equation. From Table 4-6, it is interesting to note that asphalt content, it self, is not a significant factor,
91 while some combination factors including the asphalt content factor have a significant relationship with the ICQI. When the regression analysis was ex ecuted without consideration of correlations between the factors, asphalt content was the significant factor. In Table 4-7, the P-value indicates that the regression as a whole is very significant for a significance level of less than 1%. The coefficient of determination, R2, indicates that over 80% of the variation in the variables is explaine d by the regression model, shown in Equation 10. Table 4-6. Regression on friction course ICQI Variables Coefficients Standard Error Coefficients t-ratio P-value Constant 1.12 0.12 9.41 0.000 3/8 in. passing rate 0.00828 0.0010 7.99 0.000 No. 4 passing rate 0.00882 0.0019 4.61 0.000 No. 200 passing rate -0.0527 0.0175 -3.01 0.003 Air void ratio -0.226 0.0270 -8.38 0.000 3/8 No.4 -0.000049 0.000013 -3.81 0.000 3/8 No.200 0.00151 0.00018 8.22 0.000 3/8 Asphalt content -0.00190 0.00014 -13.25 0.000 No. 4 Air void -0.00107 0.00041 -2.6 0.010 No. 200 Air void -0.0173 0.0028 -6.13 0.000 Asphalt content Air void 0.0425 0.0030 14.03 0.000 Table 4-7. ANOVA statistics for re gression on friction course ICQI Source Degree of Freedom Sum of Squares Mean Square F P Regression 10 3.04338 0.30434 250.71 0.000 Residual Error 476 0.57783 0.00121 Total 486 3.62121 R2 84.0 % ICQIFC =1.12 + 0.00828 3/8in + 0.00882 No.4 0.0527 No.200 0.226 AV 0.000049 3/8*No. 4 + 0.00151 3/8*No.200 0.00190 3/8*AC 0.00107 No.4*AV 0.0173 No.200*AV + 0.0425 AC*AV (Equation 10)
92 In the above equation 10, ICQIFC is ICQI at friction course layers. Superpave regression model: The results for the regression an alysis revealed that at the 95% confidence level, eight of th e 21 factors were statistically significant in the Superpave layer regression analysis. Tables 4-8 and 4-9 show the regression and the ANOVA statistics fo r the final equation. From Table 4-8, even though the 3/8 inch sieve passing rate itself is not the significant factor, some combination of factors, including the 3/8 inch sieve passing ra te, have a significant relationship with the ICQI. Table 4-8. Regression on Superpave ICQI Variables Coefficients Standard E rror Coefficients t-ratio P-value Constant 2.09 0.12 18.07 0.000 No. 4 passing rate 0.0172 0.0046 3.78 0.000 No. 200 passing rate -0.236 0.0625 -3.77 0.000 Asphalt content -0.150 0.0146 -10.28 0.003 Air void ratio -0.254 0.0281 -9.02 0.000 3/8 No.4 -0.00020 0.000049 -4.05 0.000 3/8 No.200 0.00345 0.00070 4.88 0.000 No. 200 Air void -0.0188 0.0036 -5.21 0.000 Asphalt content Air void 0.0367 0.0038 9.57 0.000 In Table 4-9, the P-value indicates that the regression as a whole is very significant for a significance level of less than 1%. The coefficient of determination, R2, indicates that over 70% of the variation in the variables is explaine d by the regression model, shown in Equation 11. ICQISP =2.09 + 0.0172 No.4 0.236 No.200 0.150 AC 0.254 AV 0.000200 3/8*No. 4 + 0.00345 3/8*No.200 0.0188 No.200*AV + 0.0367 AC*AV (Equation 11) In the equation 11, ICQISP is ICQI at Superpave layers.
93 Table 4-9. ANOVA statistics fo r regression on Superpave ICQI Source Degree of Freedom Sum of Squares Mean Square F P Regression 8 9.9702 1.2463 275.29 0.000 Residual Error 838 3.7938 0.0045 Total 846 13.7639 R2 72.4 % 18.104.22.168 The ICQI model validation by district When developing the ICQI regression m odel fo r the friction course and Superpave layer, the project list was divided into two parts. One part was compos ed of projects from Districts 1 and 7, District 2, and District 3 and was used to develop the ICQI regression model, and another part included Districts 4 and 6, Di strict 5, and the Turnpike distri ct and was used to analyze and validate the regression model. Table 4-10 shows the results of the analysis of projects in Districts 4 and 6, District 5, and the Turnpike district. As me ntioned earlier, it is unreasonabl e to expect that all FDOT construction and materials personnel from diff erent districts would have the same rating standard. In other words, fair projects in one district might be good projects in another district. For example, Districts 4 and 6 personnel originally rate d 17 projects as excellent, good, fair, and average when they were asked to provide the project list with three-level ratings. In this research, excel lent and good are classified as good. Fair and average are categorized as average. For this reason, comparison of ICQI model pe rformance within the same group should be a rational method of analysis. In Table 4-10, pr ojects rated by the same FDOT personnel are grouped together. ICQI does not ha ve a wide range, so it is ha rd to tell whether the model performs correctly as expected. The highest ICQI was 1.041 from project number 41551215201
94 in District 5, and the lowest ICQI was 0.9835 from project number 41551315201 also in District 5. Therefore, the maximum ICQI difference pos sible from Table 4-10 is 0.0575. In order to compare these projects easily, Table 4-11 s hows the mean values for each category. Table 4-10. Performance of the ICQI model District Project Number Overall ICQI FDOT Rating 4 228188-1 1.0157 Good 4 228615-1 1.0307 Good 4 231735-1 1.0055 Good 4 231737-1 1.0305 Good 4 403619-1 1.0015 Good 4 411322-1 0.9839 Good 6 250548-2 1.0210 Good 4 228110-1 1.0221 Average 4 411321-1 1.0255 Average 4 411323-1 0.9894 Average 4 231726-1 0.9989 Average 5 411603-1 1.0020 Good 5 415512-1 1.0410 Good 5 415526-1 0.9993 Good 5 413585-1 1.0264 Average 5 415513-1 0.9835 Average TPK 411532-1 1.0181 Good TPK 413623-1 1.0064 Good TPK 413623-2 1.0198 Good TPK 406092-1 1.0144 Good TPK 411533-1 1.0157 Good TPK 413669-1 0.9847 Poor Table 4-11 shows that the model has rendered ICQI values that are in line with the considered opinion of the owner regarding the quality of the project even though differences of
95 ICQI values between rating categories range only from 0.004 to 0.0302. However, considering the ICQI ranges are narrow, having the correct or der by the FDOT ratings shows that there is a tendency for better rated projects by FDOT to produce higher ICQI. The most important result is that no mean value of ICQI for projects ra ted good by FDOT construction and materials personnel was lower than any other ICQI mean va lues for projects rated average or poor. The same relationship can be seen between average and poor. Table 4-11. Mean value of the ICQI projects by districts FDOT Rating District 4 & 6 District 5 Turnpike Good 1.0127 1.0141 1.0149 Average 1.0087 1.0050 Poor 0.9847 In order to determine if there is a signif icant difference between good projects and average projects from District 4 and 6 and Distri ct 5, t-tests were executed. The t-test results showed that there is a signifi cant difference at the 74% level of confidence between the ICQI mean of good projects and average projects from District 4 and 6. Also, there is a significant difference at the 72% level of confidence betw een the ICQI mean of good projects and average projects from District 5. Because the nu mber of projects rated poor in the Turnpike district is only 1, t-test for the Tu rnpike district could not be executed. When all the good projects from District 4 a nd 6, District 5, and th e Turnpike district and average projects from District 4 and 6 and Di strict 5 are combined for t-test, the results showed that there is a signifi cant difference at the 44% level of confidence between the ICQI mean of the good projects and average projects This 44% level of confidence for combined districts t-test results is much lower than 74% and 72% level of confiden ce for each districts ttest results.
96 Because the FDOT rating is either not tangib le or does not have concrete guidance, it might be a better idea to have just two rating categories. Even though th e number of projects analyzed is insufficient, the Turnpike shows th e largest difference of mean ICQI between the good and poor categories.
97 CHAPTER 5 CONCLUSIONS AND RECOMMENDATIONS 5.1 Conclusions A practical and effective pavem ent CQI/ICQI ha s been developed that only uses the most objective data and test results. The CQI formul ation employed the PWL concept that is already used by FDOT and familiar to the pavement c ontractor. The ICQI formulation used the MEPDG program to develop the ICQI model base d on regression analysis. Both the CQI and ICQI used data from the LIMS, which serves as the departments enterprise database system for all construction quality data. The CQI/ICQI applies material and structural characteristics. The CQI uses pavement smoothness characteristics as input, whereas the ICQI recognizes pavement smoothness characteristics as a future performan ce indicator in terms of distress. The CQI is applicable for both new and rehab ilitation projects; howev er, the ICQI is only developed for rehabilitation because of the limited number of projects that have stored test data into LIMS. Because a pavement system consists of a series of multiple layers, the CQI/ICQI formulation is based upon the modular approach, allowing a summation of the CQI/ICQI of each individual layer multiplied by a weighting factor. The CQI of each layer is similarly determined by adding the products of the PWL of each AQC multiplied by a matching weighting factor. However in the layer level, ICQI is determined by a regression model that uses construction quality characteristics in MEP DG as dependant variables. Weighting factors were calcula ted from information gathered from expert panel surveys, which was prepared by rules of AHP. These were applied to both the CQI and ICQI models.
98 The CQI/ICQI model showed positive results with the limited data procured through the LIMS database. FDOT was asked to provide 105 projects to the research team along with an associated rating for each project. The projects submitted by FDOT were to be ones that had data entered into the LIMS database. For the projects that met the criteria for incl usion, the CQI model showed positive results, rendering higher CQI for projects rated good by FDOT. The result was clearer when the projects were put into two categories; good and poor. Because the rating by FDOT is a criterion of the model, having a consistent and precise ratin g is essential for successful application of the CQI model. There is a tendency for betterrated projects by FDOT to produce a higher ICQI. No mean value of ICQI for projects rated good by FDOT construction and materials personnel was lower than any other ICQI mean values for pr ojects rated average or poor. The same relationship can be seen be tween average and poor. 5.2 Recommendations The research team found that a great deal of da ta had either not been entered into LIMS or had been entered into the wrong place. For inst ance, for some asphalt paving projects, several key asphalt data were not found when the job st atus was listed as Construction Completed. Even when asphalt data were present, there were cases in which no asphalt data were found under some asphalt design mix numbers that had been listed for the project. The ultimate goal is to have the data available to the model so that the model can obtain it directly and automatically calculate the CQI/IC QI. Obviously, the kinds of database problems experienced by the research team on this project will doom these plans to failure. It is the observation of the research team that problems of this sort are not limited to Florida. In fact, Florida is far ahead of many states in its ability to store accurate construc tion data electronically.
99 It is therefore recommended that all states upgrade their electr onic construction testing results database if using something such as CQI is of interest. In addition, further research should be conduc ted on this model. Because there was no actual performance data for the projects, the F DOT rating was the only applicable measure for project performance. This res earch used the most objective data (test data) excluding all other aspects of contractor performance (e.g., financial re sources, ownership of e quipment or ability to lease equipment, adherence to schedule, job safety, past performance). Even though the CQI/ICQI model was developed based on the ob jective data, the final decision regarding whether the model performs properly was determin ed by the FDOT rating, which is subjective. Therefore, it is recommended that the CQI/ICQI results be compared with performance data.
100 APPENDIX A: FDOT PAVEMENT ACCEPTANCE QUALITY CHARACTERISTICS
101Table A-1. Acceptance Quality Charact eristics for Flexible Pavements Specification Layer AQC Units Upper Range Target Lower Range Section 120: Excavation and Embankment Embankment Density Percent Standard Proctor Maximum Density None 100 0 Bearing Value LBR (soaked) None 40 5 Bearing Value LBR (soaked) None 35 4 Bearing Value LBR (soaked) None < 30 2.5 Bearing Value LBR = 40 (unsoaked) None 43 0 Mixing Depth inches 2 Per plans 0 Section 160: Stabilizing Stabilized Subgrade Density Percent Modified Proctor Density None 98 0 Section 200: Rock Base Base Course Density Percent Modified Proctor Density None 98 0 Section 204: Graded Aggregate Base Base Course Density Percent Modified Proctor Density None 98 0 Passing No. 8 Sieve Percent 3.1 Per plans 3.1 Passing No. 200 Sieve Percent 1.0 Per plans 1.0 Asphalt Content Percent 0.40 Per plans 0.40 Air Voids (Coarse Mix) Percent 1.40 4.00 1.40 Air Voids (Fine Mix) Percent 1.20 4.00 1.20 Density (Coarse) Percent Gmm 1.30 94.50 1.30 Section 234: Superpave Asphalt Base Base Course Density (Fine) Percent Gmm 2.00 93.00 1.20
102 Table A-1. Acceptance Quality Characteristi cs for Flexible Pavements (Continued) Specification Layer AQC Units Upper Range Target Lower Range Section 283: Reclaimed Asphalt Base Base Course Density Percent Modified Proctor Density None 95 0 Passing No. 8 Sieve Percent 3.1 Per plans 3.1 Passing No. 200 Sieve Percent 1.0 Per plans 1.0 Asphalt Content Percent 0.40 Per plans 0.40 Air Voids (Coarse Mix) Percent 1.40 4.00 1.40 Air Voids (Fine Mix) Percent 1.20 4.00 1.20 Density (Coarse) Percent Gmm 1.30 94.50 1.30 Section 334: Superpave Asphalt Concrete Structural Course Density (Fine) Percent Gmm 2.00 93.00 1.20
103Table A-1. Acceptance Quality Characteristi cs for Flexible Pavements (Concluded) Specification Layer AQC Units Upper Range Target Lower Range Asphalt Binder Content Percent 0.45 Per plans 0.45 Passing 3/8 in Seive Percent 6.00 Per plans 6.00 Passing No. 4 Sieve Percent 4.50 Per plans 4.50 FC-5 Passing No. 8 Seive Percent 2.50 Per plans 2.50 Passing No. 8 Sieve Percent 3.10 Per plans 3.10 Passing No. 200 Sieve Percent 1.00 Per plans 1.00 Asphalt Content Percent 0.40 Per plans 0.40 Air Voids (Coarse Mix) Percent 1.40 4.00 1.40 Air Voids (Fine Mix) Percent 1.20 4.00 1.20 Density (Coarse) Percent Gmm 1.30 94.50 1.30 FC-9.5 Density (Fine) Percent Gmm 2.00 93.00 1.20 Passing No. 8 Sieve Percent 3.10 Per plans 3.10 Passing No. 200 Sieve Percent 1.00 Per plans 1.00 Asphalt Content Percent 0.40 Per plans 0.40 Air Voids (Coarse Mix) Percent 1.40 4.00 1.40 Air Voids (Fine Mix) Percent 1.20 4.00 1.20 Density (Coarse) Percent Gmm 1.30 94.50 1.30 Section 337: Asphalt Concrete Friction Courses FC-12.5 Density (Fine) Percent Gmm 2.00 93.00 1.20 Ride Number Friction Course Ride Number None 5 1
104 APPENDIX B PROJECT LIST
111 APPENDIX C EXPERT PANEL MEETING FORMS Table C-1. Survey Form of CQI Sheet 1 of 2 Name: Location: Date: Affliation: Concerning: Factor98765432 1 23456789Factor Embankment Stablized Subgrade Embankment Base Embankment Superpave Embankment Friction Course Stablized Subgrade Base Stablized Subgrade Superpave Stablized Subgrade Friction Course Base Superpave Base Friction Course Superpave Friction Course Concerning: Factor98765432 1 23456789Factor Density LBR Density Thickness LBR Thickness Concerning: Factor98765432 1 23456789 Factor Density Thickness Concerning: Factor98765432 1 23456789Factor Air Voids Passing #200 Air Voids Asphalt Content Air Voids Thickness Air Voids Roadway Density Passing #200 Asphalt Content Passing #200 Thickness Passing #200 Roadway Density Asphalt Content Thickness Asphalt Content Roadway Density Roadway Density ThicknessCONTINUED ON NEXT PAGEFlexible Pavement System Components Which factor has the greater influence on quality? Stablized Subgrade Which factor has the greater influence on quality? Base Which factor has the greater influence on quality? SuperPave Which factor has the greater influence on quality? Consultant Academia Other FDOT CONSTRUCTION QUALITY INDEX EXPERT PANEL RATING SHEETFlorida Department of Transportation Construction IndustryFLEXIBLE PAVEMENT
112 Table C-1. Continued Sheet 2 of 2 Concerning: Factor98765432 1 23456789Factor Binder Content Passing 3/8 in. Binder Content Passing #4 Binder Content Passing #8 Binder Content Ride Number Passing 3/8-in. Passing #4 Passing 3/8-in. Passing #8 Passing 3/8-in. Ride Number Passing #4 Passing #8 Passing #4 Ride Number Passing #8 Ride Number Concerning: Factor98765432 1 23456789Factor Air Voids Passing #200 Air Voids Asphalt Content Air Voids Thickness Air Voids Roadway Density Air Voids Ride Number Passing #200 Asphalt Content Passing #200 Thickness Passing #200 Roadway Density Passing #200 Ride Number Asphalt Content Thickness Asphalt Content Roadway Density Asphalt Content Ride Number Ride Number Roadway Density Ride Number Thickness Roadway Density Thickness FC-9.5/FC-12.5 Which factor has the greater influence on quality?FLEXIBLE PAVEMENTFC-5 Which factor has the greater influence on quality?FDOT CONSTRUCTION QUALITY INDEX EXPERT PANEL RATING SHEET
113 Table C-2. Survey Form of ICQI Name: Date: *** Instructions for panel members answering questions in the form below *** 1) Each response only represents your opinion concerning the relative importance of the pair of items on a single line. 2) Fill out all portions of the form for which you feel qualified to have an opinion. 3) Fill out the forms without discussion with your neighbor. Example: On the first line below, if you think IRI (Roughness) and Cracking (Top-Down) are equally important to the flexible pavement performance, then mark on "0". If you think Cracking (Top-Down) is a stronger indicator than IRI (Roughness), mark on between "1" and "8" on the right side of the line. Marking on right side "8" means that almost only Cracking (Top-Down) of the 2 factors affects to the pavement performance. Factor87654321012345678Factor IRI (Roughness) Cracking (Top-Down) IRI (Roughness) Cracking (Bottom-Up) IRI (Roughness) Rutting Cracking (Top-Down) Cracking (Bottom-Up) Cracking (Top-Down) Rutting Cracking (Bottom-Up) Rutting Which factor is the stronger indicator of flexible pavement performance?FDOT INTEGRATED CONSTRUCTION QUALITY INDEX EXPERT PANEL RATING SHEETFLEXIBLE PAVEMENT
114 APPENDIX D TABULATION OF RESULTS FROM EXPERT PANEL MEETINGS
115Table D-1. Results of the CQI Survey FDOT CONSTRUCTION QUALITY INDEDX FLEXIBLE PAVEMENT Comparison Construction Industry FDOT Academia Other + 1 2 3 4 5 6 7 AVG 8 9 10 11 12 13 14 15 AVG 16 17 AVG 18 GRAND MEAN Flexible Pavement System Components Embankment vs. Stabilized Subgrade 1 2 3 3 3 0 0 1.71 5 1 2 2 7 4 0 2 2.88 7 2 4.50 -4 2.22 Embankment vs. Base 2 2 6 8 1 1 0 2.86 6 2 2 6 8 6 0 6 4.50 7 6 6.50 4 4.06 Embankment vs. Superpave 3 4 8 6 8 (1) 0 4.00 6 3 2 8 8 7 (3) 7 4.75 3 3.00 4 4.29 Embankment vs. Friction Course 0 4 8 6 8 (2) 0 3.43 6 4 6 8 8 8 (4) 5 5.13 4 0 2.00 4 4.06 Stabilized Subgrade vs. Base 1 2 5 8 8 0 0 3.43 4 1 2 3 6 1 0 2 2.38 1 5 3.00 4 2.94 Stabilized Subgrade vs. Superpave 2 2 8 6 8 (1) 0 3.57 4 2 2 8 7 3 (3) 5 3.50 4 4.00 4 3.59 Stabilized Subgrade vs. Friction Course 0 2 8 6 8 (2) 0 3.14 3 3 6 8 7 3 (4) 6 4.00 4 1 2.50 4 3.50 Base vs. Superpave 1 2 8 (4) 8 (2) 0 1.86 4 2 2 6 5 0 (3) 5 2.63 0 0.00 4 2.24 Base vs. Friction Course 0 2 8 (4) 8 (2) 0 1.71 3 3 6 6 6 0 (7) 3 2.50 1 (6) (2.50) 0 1.50 Superpave vs. Friction Course (1) (1) (6) (4) 0 (3) 0 (2.14) (3) 4 0 2 2 4 (4) 2 0.88 (3) (3.00) 0 (0.65) Stabilized Subgrade Density vs. LBR 2 1 (5) 4 (1) 0 0 0.14 (7) 0 0 (1) 0 (7) 2 (1.86) 6 0 3.00 0 (0.35) Density vs. Thickness 1 0 (2) (4) (1) 2 2 (0.29) (1) 0 (2) 4 (5) (7) (1) (1.71) 6 0 3.00 -4 (0.71) LBR vs. Thickness 0 0 2 (4) 1 1 2 0.29 8 0 (2) 5 (4) 7 (1) 1.86 0 3 1.50 4 1.29 Base Density vs. Thickness 0 0 4 (8) 2 1 (0.17) (3) 0 0 2 (3) (6) 2 (1.14) 5 (2) 1.50 -4 (0.63) Superpave Air Voids vs. Passing #200 (4) 0 0 0 (8) 0 1 (1.57) 1 (3) (2) (5) (8) (3) 0 (5) (3.13) 0 3 1.50 -4 (2.06) Air Voids vs. Asphalt Content 2 0 0 0 (2) 0 1 0.14 2 0 (2) (2) (3) (1) 0 1 (0.63) 0 3 1.50 -1 (0.11) Air Voids vs. Thickness 4 1 1 3 (8) 2 1 0.57 0 1 0 0 (2) (2) 0 (2) (0.63) (3) 2 (0.50) 0 (0.11) Air Voids vs. Roadway Density 1 1.00 (1) (1.00) 0.00 Passing #200 vs. Asphalt Content 4 0 2 (2) 5 1 (1) 1.29 2 2 0 0 8 5 3 2 2.75 0 0 0.00 1 1.78 Passing #200 vs. Thickness 8 1 (2) 3 (6) 2 (1) 0.71 (1) 3 2 2 8 3 3 3 2.88 0 1 0.50 -2 1.50 Passing #200 vs. Roadway Density (1) (1.00) 4 4.00 1.50 Asphalt Content vs. Thickness 8 1 (1) 3 (8) 2 0 0.71 (1) 0 2 1 0 0 0 2 0.50 1 1.00 -3 0.41 Asphalt Content vs. Roadway Density 0 0.00 2 2.00 1.00 Roadway Density vs. Thickness 0 0.00 1 1.00 0.50
116Table D-1. Continued Comparison Construction Industry FDOT Academia Other + 1 2 3 4 5 6 7 AVG 8 9 10 11 12 13 14 15 AVG 16 17 AVG 18 GRAND MEAN FC-5 Binder Content vs. Passing 3/8 in. (3) (2) (1) 2 (4) (1) 0 (1.29) (6) 0 0 (2) (5) 0 (4) (4) (2.63) (1) (2) (1.50) -4 (2.06 ) Binder Content vs. Passing #4 (6) (2) (1) 2 (4) (1) 0 (1.71) (6) (1) 0 (3) (5) 0 (4) (5) (3.00) (1) (2) (1.50) -4 (2.39) Binder Content vs. Passing #8 (5) (2) (1) 2 (1) (1) 0 (1.14) (2) (3) 0 (4) (6) 0 (4) (6) (3.13) (1) (2) (1.50) -4 (2.22) Binder Content vs. Ride Number 0 (2) (4) 8 (4) 0 0 (0.29) (2) 0 1 0 (2) (2) 0 (2) (0.88) 5 0 2.50 0 (0.22) Passing 3/8 in. vs. Passing #4 (2) 0 1 2 0 0 0 0.14 0 (3) 0 (2) (1) (2) (4) (4) (2.00) 0 0.00 4 (0.65) Passing 3/8 in. vs. Passing #8 (2) 1 1 4 0 0 0.67 5 (4) 0 (1) (4) (2) (4) (6) (2.00) 0 0.00 4 (0.50) Passing 3/8 in. vs. Ride Number 3 0 (4) 8 0 1 0 1.14 4 0 1 0 3 (1) 0 2 1.13 5 2 3.50 4 1.56 Passing #4 vs. Passing #8 0 1 1 2 4 0 0 1.14 5 (3) 0 0 (4) (2) (4) (3) (1.38) 0 0.00 0 (0.18) Passing #4 vs. Ride Number 3 0 (4) 8 (3) 1 0 0.71 4 2 1 0 3 (1) 4 2 1.88 5 2 3.50 -3 1.33 Passing #8 vs. Ride Number 3 0 (4) 8 (3) 1 0 0.71 (1) 5 1 0 7 0 4 3 2.38 5 2 3.50 -3 1.56 FC-9.5 and FC-12.5 Air Voids vs. Passing #200 (3) 0 0 0 (6) 0 0 (1.29) 1 (3) 0 (4) (8) (3) 0 (5) (2.75) (2) 2 0.00 -3 (1.89) Air Voids vs. Asphalt Content 0 0 0 2 (2) 1 0 0.14 2 0 0 (1) (3) (2) 0 1 (0.38) 0 2 1.00 -1 (0.06) Air Voids vs. Thickness 3 1 1 5 (6) 1 0 0.71 2 1 1 0 (2) (2) 0 (2) (0.25) 0 2 1.00 -3 0.11 Air Voids vs. Roadway Density 0 0.00 (2) (2.00) (1.00) Air Voids vs. Ride Number 0 (2) (4) 8 (6) 1 0 (0.43) 4 1 1 0 (3) (2) 0 (2) (0.13) 7 2 4.50 -3 0.11 Passing #200 vs. Asphalt Content 2 0 1 1 3 1 0 1.14 3 2 0 0 8 2 2 2 2.38 0 0 0.00 2 1.61 Passing #200 vs. Thickness 4 1 (1) 5 (2) 1 0 1.14 3 3 1 0 8 2 0 2 2.38 7 0 3.50 -1 1.83 Passing #200 vs. Roadway Density 0 0.00 2 2.00 1.00 Passing #200 vs. Ride Number 4 (1) (4) 8 (1) 1 0 1.00 5 4 1 0 8 1 2 2 2.88 7 2 4.50 -1 2.11 Asphalt Content vs. Thickness 0 1 (2) 5 (6) 0 0 (0.29) (1) 0 0 0 0 0 0 (2) (0.38) 0 0 0.00 -2 (0.39) Asphalt Content vs. Roadway Density 0 0.00 2 2.00 1.00 Asphalt Content vs. Ride Number 1 (1) (4) 8 (6) 1 0 (0.14) (2) 0 1 0 0 0 0 1 0.00 7 0 3.50 -1 0.28 Ride Number vs. Roadway Density 0 0.00 1 1.00 0.50 Ride Number vs. Thickness 2 1 4 (8) (1) 0 0 (0.29) (2) 0 (1) 0 0 1 0 (1) (0.38) (7) 2 (2.50) -4 (0.78) Roadway Density vs. Thickness 0 0.00 (1) (1.00) (0.50)
117Table D-2. Results of the ICQI Survey +12345678910111213 Flexible Pavement System Distresses IRI (Roughness) vs. Cracking (Top-Down)54864563843685.38 IRI (Roughness) vs. Cracking (Bottom-Up)6286056386(3)784.77 IRI (Roughness) vs. Rutting7688(2)665847786.00 Cracking (Top-Down) vs. Cracking (Bottom-Up)(2)(5)(8)(3)(4)0(5)004(3)40(1.69) Cracking (Top-Down) vs. Rutting3(1)00(4)2(6)2415581.46 Cracking (Bottom-Up ) vs. Rutting (7)10622(6)2406081.38FDOT INTEGRATED CONSTRUCTION QUALITY INDEDXFLEXIBLE PAVEMENTExpert Panel MembersMEAN
118 APPENDIX E PWL TABLES Table E-1. P WL Table (n=3-11) PWL n = 3 n = 4 n = 5 n = 6 n = 7 n = 8 n = 9 n = 10 to 11 100 1.16 1.50 1.79 2.03 2.23 2.39 2.53 2.65 99 1.47 1.67 1.80 1.89 1.95 2.00 2.04 98 1.15 1.44 1.60 1.70 1.76 1.81 1.84 1.86 97 1.41 1.54 1.62 1.67 1.70 1.72 1.74 96 1.14 1.38 1.49 1.55 1.59 1.61 1.63 1.65 95 1.35 1.44 1.49 1.52 1.54 1.55 1.56 94 1.13 1.32 1.39 1.43 1.46 1.47 1.48 1.49 93 1.29 1.35 1.38 1.40 1.41 1.42 1.43 92 1.12 1.26 1.31 1.33 1.35 1.36 1.36 1.37 91 1.11 1.23 1.27 1.29 1.30 1.30 1.31 1.31 90 1.10 1.20 1.23 1.24 1.25 1.25 1.26 1.26 89 1.09 1.17 1.19 1.20 1.20 1.21 1.21 1.21 88 1.07 1.14 1.15 1.16 1.16 1.16 1.16 1.17 87 1.06 1.11 1.12 1.12 1.12 1.12 1.12 1.12 86 1.04 1.08 1.08 1.08 1.08 1.08 1.08 1.08 85 1.03 1.05 1.05 1.04 1.04 1.04 1.04 1.04 84 1.01 1.02 1.01 1.01 1.00 1.00 1.00 1.00 83 1.00 0.99 0.98 0.97 0.97 0.96 0.96 0.96 82 0.97 0.96 0.95 0.94 0.93 0.93 0.93 0.92 81 0.96 0.93 0.91 0.90 0.90 0.89 0.89 0.89 80 0.93 0.90 0.88 0.87 0.86 0.86 0.86 0.85 79 0.91 0.87 0.85 0.84 0.83 0.82 0.82 0.82 78 0.89 0.84 0.82 0.80 0.80 0.79 0.79 0.79 77 0.87 0.81 0.78 0.77 0.76 0.76 0.76 0.75 76 0.84 0.78 0.75 0.74 0.73 0.73 0.72 0.72 75 0.82 0.75 0.72 0.71 0.70 0.70 0.69 0.69 74 0.79 0.72 0.69 0.68 0.67 0.66 0.66 0.66 73 0.76 0.69 0.66 0.65 0.64 0.63 0.63 0.63 72 0.74 0.66 0.63 0.62 0.61 0.60 0.60 0.60 71 0.71 0.63 0.60 0.59 0.58 0.57 0.57 0.57 70 0.68 0.60 0.57 0.56 0.55 0.55 0.54 0.54 69 0.65 0.57 0.54 0.53 0.52 0.52 0.51 0.51 68 0.62 0.54 0.51 0.50 0.49 0.49 0.48 0.48 67 0.59 0.51 0.47 0.47 0.46 0.46 0.46 0.45 66 0.56 0.48 0.45 0.44 0.44 0.43 0.43 0.43 65 0.52 0.45 0.43 0.41 0.41 0.40 0.40 0.40 64 0.49 0.42 0.40 0.39 0.38 0.38 0.37 0.37 63 0.46 0.39 0.37 0.36 0.35 0.35 0.35 0.34 62 0.43 0.36 0.34 0.33 0.32 0.32 0.32 0.32 61 0.39 0.33 0.31 0.30 0.30 0.29 0.29 0.29 60 0.36 0.30 0.28 0.27 0.27 0.27 0.26 0.26 59 0.32 0.27 0.25 0.25 0.24 0.24 0.24 0.24 58 0.29 0.24 0.23 0.22 0.21 0.21 0.21 0.21 57 0.25 0.21 0.20 0.19 0.19 0.19 0.18 0.18 56 0.22 0.18 0.17 0.16 0.16 0.16 0.16 0.16 55 0.18 0.15 0.14 0.14 0.13 0.13 0.13 0.13 54 0.14 0.12 0.11 0.11 0.11 0.11 0.10 0.10
119 Table E-1. Continued PWL n = 3 n = 4 n = 5 n = 6 n = 7 n = 8 n = 9 n = 10 to 11 53 0.11 0.09 0.08 0.08 0.08 0.08 0.08 0.08 52 0.07 0.06 0.06 0.05 0.05 0.05 0.05 0.05 51 0.04 0.03 0.03 0.03 0.03 0.03 0.03 0.03 50 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
120 Table E-2. Pwl Table (n=12) PWL n = 12 to 14 n = 15 to 18 n = 19 to 25 n = 26 to 37 n = 38 to 69 n = 70 to 200 n = 201 to 100 2.83 3.03 3.20 3.38 3.54 3.70 3.83 99 2.09 2.14 2.18 2.22 2.26 2.29 2.31 98 1.91 1.93 1.96 1.99 2.01 2.03 2.05 97 1.77 1.79 1.81 1.83 1.85 1.86 1.87 96 1.67 1.68 1.70 1.71 1.73 1.74 1.75 95 1.58 1.59 1.61 1.62 1.63 1.63 1.64 94 1.50 1.51 1.52 1.53 1.54 1.55 1.55 93 1.44 1.44 1.45 1.46 1.46 1.47 1.47 92 1.37 1.38 1.39 1.39 1.40 1.40 1.40 91 1.32 1.32 1.33 1.33 1.33 1.34 1.34 90 1.26 1.27 1.27 1.27 1.28 1.28 1.28 89 1.21 1.22 1.22 1.22 1.22 1.22 1.23 88 1.17 1.17 1.17 1.17 1.17 1.17 1.17 87 1.12 1.12 1.12 1.12 1.12 1.13 1.13 86 1.08 1.08 1.08 1.08 1.08 1.08 1.08 85 1.04 1.04 1.04 1.04 1.04 1.04 1.04 84 1.00 1.00 1.00 1.00 0.99 0.99 0.99 83 0.96 0.96 0.96 0.96 0.95 0.95 0.95 82 0.92 0.92 0.92 0.92 0.92 0.92 0.92 81 0.89 0.88 0.88 0.88 0.88 0.88 0.88 80 0.85 0.85 0.85 0.84 0.84 0.84 0.84 79 0.82 0.81 0.81 0.81 0.81 0.81 0.81 78 0.78 0.78 0.78 0.78 0.77 0.77 0.77 77 0.75 0.75 0.75 0.74 0.74 0.74 0.74 76 0.72 0.71 0.71 0.71 0.71 0.71 0.71 75 0.69 0.68 0.68 0.68 0.68 0.68 0.67 74 0.66 0.65 0.65 0.65 0.65 0.64 0.64 73 0.62 0.62 0.62 0.62 0.62 0.61 0.61 72 0.59 0.59 0.59 0.59 0.59 0.58 0.58 71 0.57 0.56 0.56 0.56 0.56 0.55 0.55 70 0.54 0.53 0.53 0.53 0.53 0.53 0.52 69 0.51 0.50 0.50 0.50 0.50 0.50 0.50 68 0.48 0.48 0.47 0.47 0.47 0.47 0.47 67 0.45 0.45 0.45 0.44 0.44 0.44 0.44 66 0.42 0.42 0.42 0.42 0.41 0.41 0.41 65 0.40 0.39 0.39 0.39 0.39 0.39 0.39 64 0.37 0.36 0.36 0.36 0.36 0.36 0.36 63 0.34 0.34 0.34 0.34 0.33 0.33 0.33 62 0.31 0.31 0.31 0.31 0.31 0.31 0.31 61 0.29 0.29 0.28 0.28 0.28 0.28 0.28 60 0.26 0.26 0.26 0.26 0.26 0.25 0.25 59 0.23 0.23 0.23 0.23 0.23 0.23 0.23 58 0.21 0.21 0.20 0.20 0.20 0.20 0.20 57 0.18 0.18 0.18 0.18 0.18 0.18 0.18 56 0.16 0.15 0.15 0.15 0.15 0.15 0.15 55 0.13 0.13 0.13 0.13 0.13 0.13 0.13 54 0.10 0.10 0.10 0.10 0.10 0.10 0.10 53 0.08 0.08 0.08 0.08 0.08 0.08 0.08 52 0.05 0.05 0.05 0.05 0.05 0.05 0.05 51 0.03 0.03 0.03 0.03 0.03 0.03 0.02 50 0.00 0.00 0.00 0.00 0.00 0.00 0.00
121 APPENDIX F TARGET VALUE REPORT SAMPLE Address Fax No. E-mail Type Mix C 7 5 F.D.O.T. CODE PIT NO. Stockpile 1. A0672-1 2. 87-339 3. 87-339 4. 87-33 9 5. 6. 30%10%30%30% JOB MIX 123456FORMULA3/4" 19.0mm100100100100 100 100 E1/2" 12.5mm9884100100 9890-100 Z3/8" 9.5mm954289100 89 -90 INo. 4 4.75mm76538100 65 SNo. 8 2.36mm604986 4728-5839.1-39.1No. 16 1.18mm533555 34 25.6-31.6ENo. 30 600m473433 26 19.1-23.1VNo. 50 300m352317 17 ENo. 100 150m18226 8 INo. 200 75m6.42.02.02.5 5.02-10 SGSB2.5542.4072.4122.508 2.481 SP 03-2314B (TL-C) Revised to reflect change in JMF and Optimum Asphalt. StructuralThe mix properties of the Job Mix Formula have been conditionally verified, pending successful final verification during produc tion at the assigned plant, the mix design is approved subject to F.D.O.T. specifications. Screenings 51 01 25 2000 01 25 2000 White Rock Quarries 01 25 2000 White Rock Quarries 41 Ed McCarthySP-12.5 RecycleIntended Use of Mix TYPE MATERIALPRODUCERDATE SAMPLED No. 200 reflects aggregate changes expected during production. STATE OF FLORIDA DEPARTMENT OF TRANSPORTATION STATEMENT OF SOURCE OF MATERIALS AND JOB MIX FORMULA FOR BITUMINOUS CONCRETE SUBMIT TO THE STATE MATERIALS ENGINEER, CENTRAL BITUMINOUS LABORATORY, 2006 NORTHEAST WALDO ROAD., GAINESVILLE, FLA. 32609Contractor Pavex Corporation 18300 N.W. 122nd Avenue, Miami, FL 33016 20 01 25 2000 Phone No. Design Traffic Level Submitted By Gyrations @ N des (305) 828-6659 (305) 828-9464 Fi ne Blend Number CONTROL POINTS RESTRICTED ZONE Pavex Corporation A0672 White Rock Quarries PERCENTAGE BY WEIGHT TOTAL AGGREGATE PASSING SIEVES Crushed R.A.P. S-1-A Stone S-1-B Stone
122 VaVMAVFA %Gmm @ Nmax5.314.563 95.8 4.014.472 97.1 2.714.381 98.5 5.7% 45% F149C 141.0 Lbs/Ft32259 Kg/m396.0 F149C 14.4% -0.10% % =5.7% =1.6% =4.1% 2.354 2.338 86.8 88.0 89.2 1.1 1.0 2.244 2.259 2.275 5.5 6.0 6.5 Gmm300 2.370 PbGmb @ NdesPbe Arr-Maz Ad-here LOF 65-00 (D140)4.7 5.2 1.2 4.2 P0.075 / PbeAntistrip 0.5 HOT MIX DESIGN DATA SHEET SP 03-2314B (TL-C) NCAT Oven Calibration Factor VMA Optimum Asphalt FAA %Gmm @ Nini PG 64-22 to be added Optimum Asphalt Compaction Temperature300 %Gmm @ NdesLab. Density(+To Be Added)/(-To Be Subtracted) Mixing Temperature Additives Asphalt using 30% Milled Material @ 5.3% 94.6 95.2 95.7 96.3 96.8 97.4 5.05.56.06.57.0 % Asphalt%Gmm @ Ndes 14.1 14.2 14.3 14.5 14.6 14.7 5 05 56 06 57 0 % Asphalt% VMA 62 66 70 74 78 82 5.05.56.06.57.0 % Asphalt% VFA
123 LIST OF REFERENCES Bekheet, W ., Helali, K., Halim, A. E., and Spring er, J. (2005). A Comprehensive Approach for the Development of Performance Models for Network Level PMS using LTPP Data, Transportation Research Board 84th Annual Meeting. Burati, J. L, Weed, R. M., Hughes, C. S. and. Hill, H. S. (2004). Evaluation of Procedures for Quality Assurance Specifications, FHWA-HRT-04-046, Federal Highway Administration, McLean, VA. Buttlar, W. G. and Harrell, M. (1998). Devel opment of End-Result and Performance-Related Specifications for Asphalt Pavement Construction in Illinois Transportation Conference Proceedings. pp. 195-202. Chehab, G. R. and Daniel, J. S. (2006). Evalu ating Recycled Asphalt Pavement Mixtures with Mechanistic-Empirical Pavement Design Guide Level 3 Analysis, Transportation Research Record 1962, Paper No. 11, pp. 90-100. FDOT State Material Office. (2003). Fl exible Pavement Condition Survey Handbook http://www.dot.state.fl.us/statem aterialsoffice/administration/resources/library/publicatio ns/researchreports/pavement/2003flexhandbook.pdf FHWA. (1999). Guide to Developing Performance Related Specifications for Concrete Pavements, Volume 1: Practical Field Guid e, Final Report, and Appendix A, Publication No. FHWA-RD-98-155, Federa l Highway Administ ration, McLean, VA Flintsch, G. W., Loulizi, A., Diedenderfer, A. D ., Galal, K. A., and Diefenderfer, B. K. (2007). Asphalt Materials Characterization in S upport of Implementation of the Proposed Mechanistic-Empirical Pavement Design Guide, Virginia Transportation Research Council. Hirsch, Jr. E. D., Kett, J. F., and Trefil, J. (2002). The New Dictionary of Cultural Literacy, Third Edition. Houghton Mifflin Company. Hybert, P. (1996). Five Ways to Improve the Contracting Process. American Society of Quality. Quality Progress, Vol. 29-2, pp. 65-70 Isik, Z., Dikmen, I. and Bir gonul, M. T. (2007). Using Analytic Network Process (ANP) for Performance Measurement in Construction, The Construction and Building Research Conference of the Royal Inst itution of Chartered Surveyors, Georgia Tech, Atlanta, USA, 6-7 September 2007. Killingsworth, B. M. (2004). Quality Characteris tics and Test Methods for Use in PerformanceRelated Specifications of Hot Mix Asphalt Pavements NCHRP Project 9-15.
124 MacAuley, P., Cosslett, P. D., and Randol ph, Jr. J.N. (1993). Construction Industry Overview US Industrial Outlook. US Department of Commerce's International Trade Administration. McCaffrey, J. (2005). Test Run: The Analytic Hierarchy Process, MSDN Magazine. Minchin, R.E. and Smith, G.R. (2001). Quality-b ased Performance Rating of Contractors for Prequalification and Bidding Purposes, Final Report for NCHRP Project 10-54, Published as NCHRP Web Document 38, Washington, DC. Monismith, C. L., Popescu, L., and Harvey, J. (2004). Performance-Based Pay Factors for Asphalt Concrete Construction; Comparison wi th a Curruntly Used Experienced-Based Approach Journal of Association of Asphalt Paving Technologists. Volume 73. pp. 147194. Noureldin, A. S. (1997). Quantifying the Effect s of Pavement Performance Characteristics Using Partial Derivatives, Transportation Research Record 1575. Paper No. 970142. pp. 75-84. Rawlings, J. O., Pantula, S. G., and Dickey, D. A. (1998). Applied Regression Analysis, Second Edition, Springer Texts in Statistics. Saaty, T. L. (1980). The Analytic Hierarchy Process, New York, NY, McGraw Hill. Saaty, T. L. (1996). Decision Making with Dependence and Feedback: The Analytic Network Process, RWS Publications, 4922 Ellswo rth Avenue, Pittsburgh, PA. 15213. Saaty, T. L. (1999). Fundamentals of the Analytic Network Process, International Symposium on the Analytic Hierarchy Process (ISAHP), Kobe, Japan, August 12-14, 1999. Saaty, T. L. and Vargas, L. G. (2006). Decision Making with the A nalytic Network Process Economic, Political, Social and Technological Applications w ith Benefits, Opportunities, Costs and Risks, 1st Edition, ISBN: 978-0-387-33859-0, Springer. Smith, R. L., R. J. Bush, and D.L. Schmoldt. (1995). A Hierarchial Model and Analysis of Factors Affecting the Adoption of Timber as a Bridge Material, Wood and Fiber Science, 27(3), pp. 225-238. Smith, J. T. and S. L. Tighe (2006). The An alytical Hierarchy Process as a Tool for Infrastructure Management, pres ented at the 85th Annual Meeting, Transportation Research Board, Washington, DC. VonQuintus, H. L. and Moulthrop, J. S. ( 2007). Mechanistic-Empirical Pavement Design Guide Flexible Pavement Performance Prediction Models for Montana, Research Programs Montana Department of Transportation.
125 Weed, R. M. (1998). A Rational Method for Relating As-Built Quality to Pavement Performance and Value Transportation Research Board 77th Annual Meeting. Weed, R. M. (2000). Method to Model Performance Relationships and Pay Schedules Transportation Research Board 79th Annual Meeting. Weed, R. M. and Tabrizi, K. (2005). Con ceptual Framework for Pavement Smoothness Specification Transportation Research Board 84th Annual Meeting. Whiteley, L., Tighe, S., and Zhang, Z. (2005). Incorporating Variabil ity into Pavement Performance, Life Cycle Cost Analys is and Specification Pay Factors Transportation Research Board 84th Annual Meeting.
126 BIOGRAPHICAL SKETCH Junyong Ahn, a native of Korea, earned his BS in the Civil Engineering Departm ent at Yonsei University in Seoul, Korea in 1991. Th en, he worked at Dongbu Corporation in Korea for eight and half years. The company enga ged in construction and engineering, and he involved various projects such as Kwangyang roadway pavement and drainage construction project, Seoul Metropolitan water supply sy stem project, Seoul Metropolitan subway construction (Line 6) project, a nd Hwoengsung dam construction projec t. He also worked at the head office of the company as duties included cost estimation and project budget control. He earned his MSCE at the Universi ty of Washington, where he st udied construction engineering and management in Civil Engineering Department with Dr. Phillip Dunston. In August 2002, he started his Ph. D. studies at th e University of Florida. He worked with Applied Research Associates during his Ph.D. track for his doctoral dissertation research. He received his Ph.D. degree under the guidance of Dr. Ralph D. Ellis in civil engineering at the University of Florida in 2008. His research interests include cons truction productivity, contracting methods, and construction quality ratings.