2013 Session: 659

• Bayesian Approach to Updating Markov-Based Pavement Performance Prediction Models
  Abstract: The Markov decision process is one of the most common probabilistic prediction models used in infrastructure management. Where insufficient data exists, expert knowledge is commonly used to derive a Markovian transition probability matrix (TPM). Eventually, every pavement management system will progress to a level where inspection measurements from the network are organized into a database to be used for performance prediction. The best way to use this body of data to improve the initially-developed TPM is to combine prior expert knowledge with new observations. This paper proposes a method to periodically update Markovian transition probabilities as new inspection data becomes available. Bayesian inference is used to accommodate uncertainty in the expert-derived initial probabilities and measurement errors from inspection of the network. A data set of asphalt concrete pavement observations from the MnROAD test facility is used to illustrate the proposed method.
  Authors: Tabatabaee, Nader; Ziyadi, Mojtaba
  Authors: Tabatabaee, Nader; Ziyadi, Mojtaba
  Year: 2013
  Document Type: Paper
  Subject: Design; Pavements
  Session: 659
  Paper Number: 13-0993
• Sigmoid Distress Prediction Models at Project Level for Main Urban Flexible Pavements Based on Historical Data
  Abstract: The development of distress prediction models for overlaid flexible main street pavements in large municipalities cross Saudi Arabia is described. The sigmoid of models represented quantitatively predict distress density versus pavement age and are based on pavement condition data maintained by the General Directorate of Operation and Maintenance (GDOM) at the Municipality of Riyadh city, Jeddah city, Makkah holy city, Madinah city, and Damam city. Different model forms of the sigmoid family were examined in an attempt to identify the most appropriate one for fitting the data. Models are available for the following distress types in flexible main street pavements: Block Cracks, Longitudinal and Transverse Cracking, Patching, Potholes, Depressions, Weathering and Raveling, and Cracking (due to patching). The seven models have been developed using more than 11 years survey data for overlaid sections on the road network of Riyadh, Jeddah, Makkah, Madinah, and Damam. In all prediction models, age is by far the most significant predictor of deterioration. The traffic volume in terms of Annual Daily Traffic (ADT) and the drainage play only a secondary role in forecasting prediction of distress propagation. In general, the developed models provided an
  Authors: Mubaraki, Muhammad; Thom, Nick
  Authors: Mubaraki, Muhammad; Thom, Nick
  Year: 2013
  Document Type: Paper
  Subject: Design; Pavements
  Session: 659
  Paper Number: 13-1803
• Probabilistically Characterizing Uncertain Inputs in Life-Cycle Cost Analysis of Pavements
  Abstract: Life Cycle Cost Analysis (LCCA) is an important tool to evaluate the merits of alternative investments. Inevitably, inputs for an LCCA are subject to a high level of uncertainty both in the short-term and long-term. Under its current implementation in the field, however, LCCA inputs are treated as static, deterministic values. Conducting such an analysis, although computationally simpler, hides the underlying uncertainty of the inputs by only considering a few possible permutations, which could potentially lead a decision-maker to the incorrect pavement selection. One methodology that has gained traction the past decade is to describe uncertain parameters probabilistically, allowing the analysis to consider a range of possible outcomes. Although this methodology is recommended by the Federal Highway Administration (FHWA), practitioners still generally conduct deterministic LCCAs. One of the major reasons is that further research must be conducted to statistically characterize uncertainty for input parameters. This research attempts to build upon previous work by probabilistically characterizing several input parameters with and without readily available empirical data. Having characterized uncertainty and variability in the LCCA of pavements, the methodology presented is applied to a case study to determine whether the pavement selection would change by treating parameters probabilistically. The results of the analysis showed the pavement selection could change by conducting a probabilistic LCCA and are dependent upon the risk-perspective of the decision-maker.
  Authors: Swei, Omar; Gregory, Jeremy; Kirchain, Randolph
  Authors: Swei, Omar; Gregory, Jeremy; Kirchain, Randolph
  Year: 2013
  Document Type: Paper
  Subject: Design; Pavements
  Session: 659
  Paper Number: 13-2858
• Development of Roughness Models Using Alberta Transportation’s Pavement Management System
  Abstract: Alberta Transportation (AT) conducts annual automated roughness measurements on Alberta’s core highway network. The measures are used to classify the physical conditions of each pavement segment in the Pavement Management System (PMS) and to establish rehabilitation and preservation strategies for the year. Currently, no prediction model is available to correlate climatic, mechanistic and structural characteristics of the pavement to roughness development for Alberta’s network. Existing models incorporated in the Mechanistic Empirical Pavement Design Guide (MEPDG) and Highway Development and Management Tool (HDM) comprise comprehensive and detailed distress records. Such data record is not currently available for Alberta, so existing models cannot be calibrated for local conditions. The present study focuses on identifying significant variables to roughness development for Alberta’s highway network. In this study, the data available in the PMS is used to develop two new prediction models for the International Roughness Index (IRI): one for new, and the other for straight overlaid asphalt concrete (AC) sections with a granular base. Regression analysis conducted in the study revealed that variables such as age, traffic, percent fines in the subgrade, rut depth and cracking are most significant to IRI for the new AC sections with GBS, and that variables such as age, freezing index (FI), base and overlay thickness and rut depth are most significant to IRI for overlaid sections. The model for new AC sections was validated using data from the General Pavement Section (GPS)-1 in the Long Term Pavement Performance (LTPP) database located in the provinces of Alberta, Manitoba and Saskatchewan in Western Canada.
  Authors: Nassiri, Somayeh; Shafiee, Mohammad Hussein; Bayat, Alireza
  Authors: Nassiri, Somayeh; Shafiee, Mohammad Hussein; Bayat, Alireza
  Year: 2013
  Document Type: Paper
  Subject: Design; Pavements
  Session: 659
  Paper Number: 13-2897
• Roughness Performance Model for Double Surface Treatment Highways
  Abstract: This paper presents the development of a roughness performance prediction model for double surface treatment highways. It is based on a factorial experiment design developed with data from the Department of Transport Infrastructure of the State of Bahia, Brazil, and also from data collected in 2009, when approximately 3,000 km of highways were travelled and 650 km of roughness survey were performed. The factors considered were pavement age, traffic volume and climate, the last one mainly in terms of rainfall. An Analysis of Variance (ANOVA) was performed to assess the significance of the factors and to define the parameters of the performance model. The obtained model was compared to other roughness prediction models and showed better correlation between observed and predicted values, indicating the validity of its use in pavement management analysis of double surface treatment road networks.
  Authors: Soncim, Sergio Pacifico; Fernandes, Jose Leomar
  Authors: Soncim, Sergio Pacifico; Fernandes, Jose Leomar
  Year: 2013
  Document Type: Paper
  Subject: Design; Pavements
  Session: 659
  Paper Number: 13-3191
• Probabilistic Simulation of Cracking in Asphalt Pavements: Chilean Case
  Abstract: Deterioration models allow predicting pavement condition and the development of maintenance programs. Normally, when evaluating pavement performance through model deterioration, the forecast given is a determinist value. However, pavement engineering projects, as any engineering project, have a degree of uncertainty. This implies that an adequate performance of the engineering solution cannot be absolutely guaranteed.The aim of this research is to incorporate probability in the output of a structural cracking model. To achieve this objective the model of crack initiation and progression of HDM-4 was used under several scenarios defined based on geographic location, type of traffic and structural capacity of 86 roads located in Chile. The input data for each scenario were obtained from the Ministry of Public Works of Chile and calibration studies of deterioration models to local conditions. To incorporate probability in the structural cracking models, a simulation model that reproduced the deterioration due to cracking for a lifecycle of 25 years was developed, based on random input data sets. With the set of outputs of the simulation, probability density functions that represent the probabilistic response of the deterioration model were developed.The main output of this research is a set of probability density functions of cracking initiation and cracking progression of all structural cracks and wide cracks of 14 groups of Chilean roads. Although the research was carried out using Chilean data, the methodology presented could be applied to other states or countries.
  Authors: Rodríguez, Mario Alberto Moreno; Thenoux, Guillermo; Gonzalez, Alvaro Andres
  Authors: Rodríguez, Mario Alberto Moreno; Thenoux, Guillermo; Gonzalez, Alvaro Andres
  Year: 2013
  Document Type: Paper
  Subject: Design; Pavements
  Session: 659
  Paper Number: 13-3251
• Pavement Condition States Before and After Treatments
  Abstract: When a pavement section is subjected to preservation or rehabilitation treatments, its surface conditions transfer from one set of condition states before treatment to another set of condition states after treatment. This transformation is a function of the pavement condition states before treatment, the treatment types and timing, the project boundaries, and the quality of construction. For most pavement sections, the before and after treatment surface conditions vary substantially along the section and over time. In this study, which is sponsored by the Federal Highway Administration (FHWA), the distributions of the pavement surface conditions and distress before and after treatment along several flexible pavement projects in the State of Louisiana were analyzed. It is shown that the current practice regarding the selection of pavement treatment type, treatment time, and project boundaries are independent of the pavement surface conditions and distress before treatment.For all pavement projects that received certain treatment types in the past, knowledge of the relationships between the before and the after treatment distributions of the pavement surface conditions and distress is crucial to the establishment of future cost-effective pavement treatment strategy. This paper shows that such relationships can be expressed by the probabilities of transforming the before treatment condition states to the after treatment condition states. These probabilities could be housed in one matrix format called the “Treatment Transition Matrix (T2M)”. For each treatment type, the T2M presents a snap shot of the state-of-the-practice.
  Authors: Baladi, Gilbert Y.; Dean, Christopher M.
  Authors: Baladi, Gilbert Y.; Dean, Christopher M.
  Year: 2013
  Document Type: Paper
  Subject: Design; Pavements
  Session: 659
  Paper Number: 13-3599
• Sprinkle Treatment on Thin Durable Asphalt Overlays
  Abstract: This paper presents techniques to improve asphalt overlays through the use of innovative surfacing technologies with durable aggregates spread on a cost-effective base wearing surface mixture using local aggregates. The sprinkle treatment was previously used in a few states in order to improve skid resistance of pavement surfaces on top of typical asphalt pavements. Over past decades, numerous approaches have been introduced to improve the performance of asphalt pavement. These improvements often add cost because of using high quality aggregates and modified binder. Since sustainable and cost-effective pavements are being emphasized, innovative application of the sprinkle treatment has been considered. A fine dense-graded asphalt wearing surface mixture, developed using the Bailey method, provides a promising aggregate structure that makes it possible to ensure compactibility using relatively thinner asphalt mixture layers. The mix design of the base mix was developed using less expensive local aggregates, and its performance was evaluated and compared with two control mixtures for durability, rut resistance, moisture susceptibility, fracture, and complex modulus in the laboratory. In addition, test pavements of various thicknesses were constructed in northern Illinois during October and November 2010 to evaluate the performance of sprinkle treatment pavements under real traffic loading. On-site performance tests that include noise, friction, rut depth and texture profiles were conducted at four-month intervals following construction. The study concluded that it is feasible to use a cost effective alternative wearing surface that performs well.
  Authors: Son, Songsu; Al-Qadi, Imad L.; Lippert, David Leslie; Zehr, Thomas G.
  Authors: Son, Songsu; Al-Qadi, Imad L.; Lippert, David Leslie; Zehr, Thomas G.
  Year: 2013
  Document Type: Paper
  Subject: Design; Pavements
  Session: 659
  Paper Number: 13-3612
• International Roughness Index Models for Hot-Mix Asphalt Overlay Treatment of Flexible and Composite Pavements for Louisiana
  Abstract: Timely rehabilitation and preservation of pavement systems are imperative to maximize benefits in terms of driver’s comfort and safety. However, the effectiveness of any treatment largely depends on the time of treatment and triggers governed by treatment performance models. This paper presents the development of International Roughness Index (IRI) models for overlay treatment of composite and flexible pavements in the state of Louisiana. Various factors affecting the IRI of overlay treatment were identified. New climatic factors were developed and regression analysis was conducted and IRI prediction models were generated. The models provided a good agreement between the measured and predicted IRI values with majority of data within 5% of prediction error. The models could be used as a good pavement management tool for pavement maintenance and rehabilitation actions.
  Authors: Khattak, Mohammad Jamal; Nur, Mohammad Abdullah; Bhuyan, Mohammad Reza-ul Karim; Gaspard, Kevin
  Authors: Khattak, Mohammad Jamal; Nur, Mohammad Abdullah; Bhuyan, Mohammad Reza-ul Karim; Gaspard, Kevin
  Year: 2013
  Document Type: Paper
  Subject: Design; Pavements
  Session: 659
  Paper Number: 13-4396
• Calibration and Application of Treatment Performance Models in Pavement Management System
  Abstract: To utilize the pavement management system (PMS) for maintenance strategy decision making, it is necessary to use locally calibrated performance models and other related parameters. In this study, the performance family models of typical asphalt resurfacing treatments used in Tennessee were calibrated for the PMS by investigating historical maintenance projects. Multiple regression method was employed to analyze the influence of pre-treatment PSI, traffic level, overlay thickness and milling depth on the post-treatment performance curves. The specific designs of asphalt resurfacings and performance classes were determined based on the regression analysis. Then, the performance curves for each identified treatment methods at different performance classes were established and calibrated for the PMS. Rehabilitation trigger values, typical treatments and decisions trees for pavement with different functional classes were determined and designed by investigating the historical maintenance projects. A network optimization using calibrated models and determined parameters was conducted by the PMS.Investigation on the treatment performance models indicated that pavement with lower pre-treatment roughness, thick overlay and deep milling deteriorated slower. Pavement with high traffic level tended to have lower post-treatment roughness. The distress condition decreases much faster than smoothness. Network optimization analysis of the PMS indicated the most cost-effective maintenance strategy can be determined by conducting a network optimization analysis with a sufficient high budget limit, in which the highest annual cost is lower than the budget limit. Several suggestions on how to use the budgets according to the HPMA analysis include combining cost-effective and worst-first, rolling over budget and redistributing budget among the four regions.
  Authors: Dong, Qiao; Huang, Baoshan; Richards, Stephen H.
  Authors: Dong, Qiao; Huang, Baoshan; Richards, Stephen H.
  Year: 2013
  Document Type: Paper
  Subject: Design; Pavements
  Session: 659
  Paper Number: 13-4745
• Factors Influencing Life-cycle Performance of Pavement Utilizing Decision Tree Analysis
  Abstract: Accurate prediction of pavement performance during its lifecycle plays an important role in selecting the most cost-effective solution during pavement type selection process. However, State Highway Agencies (SHAs) have reported lack of consensus between asphalt and concrete industries on life-cycle cost analysis (LCCA) models utilized for pavement type selection. This is despite the fact that the Oklahoma Department of Transportation (ODOT) along with other state DOTs stores a large amount of highway project data throughout the life-cycle of highway projects. This paper applies the decision tree analysis in data mining to the historical pavement treatment data set available in the ODOT in order to identify the timing of maintenance and rehabilitation (M&R) activities. The data is divided into training and validation data sets. Therefore the models are created utilizing the training data set then their complexities are optimized to minimize the classification errors utilizing the validation data set. The results of the classification trees indicate that categorizing performance of pavement sections based on the type of material and level of traffic create errors in predicting the performance of pavements. Utilizing the historical pavement treatment projects, new criteria are identified in this paper that predicts the timing of M&R strategies adopted by ODOT. Since decision tree models developed in this paper are based on the actual treatment strategies performed by ODOT, it is expected that classifying pavement sections based on these factors help in predicting the future performance of pavement more accurately.
  Authors: Abdollahipour, Saeed; Jeong, David Hyung Seok
  Authors: Abdollahipour, Saeed; Jeong, David Hyung Seok
  Year: 2013
  Document Type: Paper
  Subject: Design; Pavements
  Session: 659
  Paper Number: 13-4983
• Probabilistically Characterizing Uncertain Inputs in Life-Cycle Cost Analysis of Pavements
  Authors: Swei, Omar
  Authors: Swei, Omar
  Year: 2013
  Document Type: Presentation; Poster
  Subject: Design; Pavements
  Session: 659
  Paper Number: 13-2858
• Calibration and Application of Treatment Performance Models in Pavement Management System
  Authors: Dong, Qiao
  Authors: Dong, Qiao
  Year: 2013
  Document Type: Presentation; Poster
  Subject: Design; Pavements
  Session: 659
  Paper Number: 13-4745
• Factors Influencing Life-cycle Performance of Pavement Utilizing Decision Tree Analysis
  Authors: Abdollahi Pour, Saeed
  Authors: Abdollahi Pour, Saeed
  Year: 2013
  Document Type: Presentation; Poster
  Subject: Design; Pavements
  Session: 659
  Paper Number: 13-4983
• Pavement Condition States Before and After Treatments
  Authors: Baladi, Gilbert
  Authors: Baladi, Gilbert
  Year: 2013
  Document Type: Presentation; Poster
  Subject: Design; Pavements
  Session: 659
  Paper Number: 13-3599