The transportation infrastructure systems in the United States were built between the 50's and 80's, with 20 years design life. As most of them already exceeded their original life expectancy, state transportation agencies (STAs) are now under increased needs to rebuild deteriorated transportation networks. For major highway maintenance projects, a federal rule enforces to perform a life-cycle cost analysis (LCCA). The lack of analytical methods for LCCA creates many challenges of STAs to comply with the rule. To address these critical issues, this study aims at developing a new methodology for quantifying the future maintenance cost to assist STAs in performing a LCCA. The major objectives of this research are twofold: 1) identify the critical factors that affect pavement performances; 2) develop a stochastic model that predicts future maintenance costs of flexible-type pavement in Texas. The study data were gathered through the Pavement Management Information System (PMIS) containing more than 190,000 highway sections in Texas. These data were then grouped by critical performance-driven factor which was identified by K-means cluster analysis. Many factors were evaluated to identify the most critical factors that affect pavement maintenance need. With these data, a series of regression analyses were carried out to develop predictive models. Lastly, a validation study with PRESS statistics was conducted to evaluate reliability of the model. The research results reveal that three factors, annual average temperature, annual precipitation, and pavement age, were the most critical factors under very low traffic volume conditions. This research effort was the first of its kind undertaken in this subject. The maintenance cost lookup tables and stochastic model will assist STAs in carrying out a LCCA, with the reliable estimation of maintenance costs. This research also provides the research community with the first view and systematic estimation method that STAs can use to determine long-term maintenance costs in estimating life-cycle costs. It will reduce the agency's expenses in the time and effort required for conducting a LCCA. Estimating long-term maintenance cost is a core component of the LCCA. Therefore, methods developed from this project have the great potential to improve the accuracy of LCCA.

Ma thèse de doctorat doit contribuer à la diminution du gaspillage des ressources naturelles (en particuliers les sables naturels) en étendant leur acceptabilité dans les enrobés bitumineux. La compréhension de l’effet des particules nocives, notamment la fraction argileuse contenue dans les sables, sur le comportement des mélanges bitumineux est visé. Pour évaluer le niveau de nocivité des particules fines, le test d'adsorption du bleu de méthylène noté MB (EN13043, NF EN 933-9) est appliqué, mais un tel test a été modifié à quatre reprises au cours des 22 dernières années tandis que la valeur de bleu limite, utilisée pour déterminer la conformité ou non du sable, n’a pas été modifiée. Le projet de thèse vise à modifier cette valeur limite pour étendre la gamme de sable utilisable. Cet objectif sera atteint grâce à l’identification de la composition minéralogique des sables, une recherche sur la meilleure manière de quantifier les phases qui composent les sables et une meilleure compréhension du rôle de la fraction fine sur la durabilité et les pathologies des mélanges bitumineux. Le travail de thèse est organisé en différentes tâches : 1. Tester le protocole de mesure de la valeur de bleu. Nous voulons comprendre l'effet de différents facteurs (tels que l'échantillonnage, la cinétique des additions de bleu, la température de séchage du sable avant l’essai ...) sur la mesure de la valeur de bleu. Le développement d'un appareil automatique ou d'une nouvelle méthode (qui sera validée par comparaison avec les résultats obtenus à partir de la méthode standardisée) est prévu.2. Établir une corrélation entre la valeur de bleu (ou d'autres paramètres à trouver) et la quantité de phases minérales présentes dans les sables. Un jeu de données a d’ores et déjà été établi au cours de mon stage de master. Pour atteindre cet objectif, les différentes méthodes de quantification (utilisant en particuliers la diffraction de rayons X) doivent être testées et comparées.3. Corréler la valeur de bleu couplée à la nature minéralogique des particules fines avec le comportement mécanique des mélanges sable/bitume. Nous allons étudier la dégradation en présence d’eau des propriétés du matériau bitumineux (par exemple l'adhésion du liant bitumineux sur granulat) et le rôle de la couche de particules fines qui entourent les grains de sable sur les propriétés mécaniques de ce matériau. Nous espérons comprendre en particulier l'effet de la présence d’argile gonflante dans les matériaux constitutifs de la chaussée bitumineuse.4. Améliorer le comportement des granulats non conformes par l’application d’un traitement. Après l'identification de l'origine des pathologies (nous faisons l'hypothèse que les argiles gonflantes jouent un rôle majeur), nous proposons d'appliquer un traitement en ajoutant de la chaux (ou de traiter en ajoutant des déchets qui contiendraient des substances actives comme la chaux), et de tester également l’utilisation de polymères ou du greffage avec des composés organiques des argiles comme traitement / Siliceous fines (clays) in fine aggregates used for AC pavement stimulate the moisture entering the bitumen-aggregates interface in AC mixture and create channels for water penetration. MB adsorption for qualifying fine aggregates is in accordance with the layer charge and the accessibility of consisted mineralogy in fine aggregates. However, the correlation between MB qualification for fine aggregates and fine aggregates triggering moisture susceptibility of AC mixture has not ever been addressed; moreover, the questions from MB adsorption still remain to be solved. In order to evaluate the stripping and to clarify MB adsorption on fine aggregates from a mineralogical perspective, the most common clays in natural aggregates, kaolinite, illite and montmorillonite, were extracted from their clay rocks, and used to prepare thin clay film, artificial fine aggregates and asphalt concrete mixtures consist of artificial fine aggregates. Moreover, the 5 blinded fine aggregates received from quarries without any given information were applied as the blinded experiments to confirm the practicability with mineralogical diversity. The mineralogical analysis was identifying and quantifying the mineral phases in extracted clay and fine aggregates by using X-Ray diffraction, and the quantitative results were judged by complementary test. MB adsorption was studied using drop method, UV-Photometer method and cation exchange capacity from clays and aggregates. Water-bitumen-clay interaction was studied using the sessile drop and the Oliensis spot tests on those thin clay films. The water intrusion routes in AC mixture was investigated immersing AC mixture in solution with chemical probe, whereas Duriez tests allowed measuring the moisture susceptibility of AC mixtures containing varied clay mineralogy in fine aggregates. The results show that clay mineralogy is in relation to water-bitumen-clay interaction, and it is agreed by moisture susceptibility of AC mixture indicated by Duriez test. With the dispersive nature bitumen used in this study, cohesive failure is in the light of kaolinite-AC mixture in the presence of water, whereas swelling property is responsible for the stripping in illite- and montmorillonite- AC mixture. In order to exclude the physical interferences on MB adsorption, the vale of 100% of MB adsorption on fine aggregates referred to CEC value is necessary to apply, and MB value measured from drop method efficiently excludes the significant impacts from layer charge, pH value and exchangeable cations. Although MB adsorption is in relation to clay mineralogy, this test does not sufficiently indicate the stripping of AC mixture occurred by clay in fine aggregates. There are still rooms to use the non-qualified fine aggregates for the use of AC pavement. As Rietveld method from X-Ray diffraction derives the most reasonable quantitative results, the X-Ray powder diffraction and the X-Ray orientated EG treated K-saturated clay diffraction are proposed for analyzing clay mineralogy and its swelling property for judging the use of non-qualified fine aggregates for AC mixture.MB2 and Duriez0.8 surfaces can be established on clay mineralogical map since we discovered MB adsorption and moisture susceptibility are related to clay mineralogy in fine aggregates. The fine aggregates which is not qualified by MB adsorption is going to confirm the possibility for the use of AC pavement by the mineralogical analysis through X-Ray diffraction. By adopting the quantitative results of clay on mineralogical map with Duriez0.8 surface and the swelling property of clays in fine aggregates, the use of fine aggregates for AC pavement can be properly judged according to water-bitumen-aggregates interaction and stripping which has been proved in this study

Previous studies show that pavement condition properties are significant factors to enhance road safety and riding experience, and pavements with low quality might have inadequate performance in terms of safety and riding experience. Pavement Management System (PMS) databases include pavement properties for each segment of the road collected by the agencies. Understanding the impact of road characteristics on crash frequency is a key step to prevent crashes. Whereas other studies analyzed the effect of different characteristics such as International Roughness Index (IRI), Rutting Depth (RD), Annual Average Daily Traffic (AADT), this thesis analyzed the effect of Critical Condition Index (CCI) on crash frequency, in addition to the other factors identified in previous studies. Other characteristics such as Percentage of Heavy Vehicles, Road Surface Condition, Road Lighting Condition, and Driver Conditions are taken into the consideration. The scope of the study is the interstate highway system in Fairfax County, Virginia. Negative Binomial, Least Square and Nominal Logistic Models were developed, showing that the CCI value is a significant factor to predict the number of crashes, and that it has different effect for different values of AADT. The result of this study is a substantial step towards developing an integrated transportation control and infrastructure management framework. / Master of Science / Many factors cause crashes in the roads. Although there is a common sense that road characteristics such as asphalt quality are important in terms of road safety, there are few studies that scientifically prove that statement. In addition, asphalt maintenance decisions making process is mainly based on cost benefit optimization, and traffic safety is not considered at the process. The purpose of this study is to analyze crashes and road characteristics related to each crash to understand the effect of those characteristics on crash frequency, and eventually, to build a model to predict the number of crashes at each part of the road. The model can help transportation agencies to have a better understanding in terms of safety consequences of their infrastructure management plans. The scope of this study is the highway interstate system in Northern Virginia. Results suggest that pavement condition has a significant impact on crash frequency.