Investigation of Relationship Between Pavement Condition Index and International Roughness Index of Asphalt Roads Based on the Long-Term Pavement Performance Data
S. Madeh Piryonesi, University of TorontoShow Abstract View Presentation
Tamer El-Diraby, University of Toronto
Two of the most popular pavement performance indicators are the International Roughness Index (IRI) and the Pavement Condition Index (PCI). The Long-Term Pavement Performance (LTPP) database does not include the latter. Therefore, limited research is available on the correlation between the PCI and IRI based on the LTPP roads. This study aims to determine the correlation between these two performance indicators using LTPP data. To this end, 3,954 records of IRI and PCI were collated to determine the correlation. The aggregate correlation was not substantial (R 2 =0.31) as the data was collected over 61 different states and provinces and in a 28-years timeline. So, in the next step the data was clustered into more meaningful groups based on location (province/state) and functional class in the hope of finding higher correlations. It was observed that the R 2 within each group was substantially higher than the aggregate data, with some reaching above 0.70. Preparing an unprecedentedly large dataset gave us the freedom of segmenting the data into smaller and less noisy subsets, which can reveal the more significant correlations. Moreover, another dataset collected by Ontario Ministry of Transportation (MTO) was studied and the results were contrasted against that of the LTPP. It was observed that the MTO data is more cohesive, and the correlation between the IRI and the PCI was more significant. Finally, this study investigated the variations not explained by regression models, i.e. reasons that road sections can have an excellent PCI and poor IRI and vice versa.
3D Visualization of Airport Pavements Quality Based on BIM and WebGL Integration
Zhen Liu, Southeast UniversityShow Abstract View Presentation
Xingyu Gu, Southeast University
Qiao Dong, Southeast University
Shanshan Tu, Southeast University
Shuwei Li, Southeast University
Impact of New-Generation, Wide-Base Tires on Fuel Consumption and Roughness
Izak Said, University of Illinois, Urbana ChampaignShow Abstract
Egemen Okte, University of Illinois, Urbana Champaign
Jaime Hernandez, Marquette University
Imad Al-Qadi, University of Illinois, Urbana Champaign
Multiple approaches were combined to evaluate the structural and economic impact of using new-generation wide-base tires (NG-WBT) in New Brunswick, Canada. A three-dimensional finite element model of a typical pavement structure was used to predict critical pavement responses. The model included measured tire-pavement contact forces among other variables overlooked in conventional flexible pavement analysis approaches. Using the model output, regression analysis may be used to predict the responses under various loading without the need to perform finite element analysis. Eight-year weight-in-motion data along with the critical pavement responses were used in transfer functions to predict pavement damage and the corresponding progression of international roughness index (IRI) over an analysis period of 60 years. Most pavement responses from NG-WBT compared to dual tire assembly (DTA) were between 20 and 30% higher. The smallest difference was the vertical strain on top of subgrade. The life-cycle cost analysis considered NG-WBT’s benefits such as fuel savings and higher hauling capacity. Two scenarios were analyzed: (1) Case A, where maintenance was performed periodically and independent of IRI values; and (2) Case B, where IRI threshold triggered maintenance. Reduction in fuel cost were significant in both cases. M aintaining a low pavement IRI would increase savings in truck fuel costs.
Assessment of Pavement Condition on the Interstate Highway System
Amy Simpson, Wood Technical Consulting SolutionsShow Abstract View Presentation
Sareh Kouchaki, Wood Environment & Infrastructure Solutions, Inc.
Pedro Serigos, Wood Environment & Infrastructure Solutions, Inc.
Gonzalo Rada, Wood Environment & Infrastructure Solutions, Inc.
Jonathan Groeger, Wood Environment & Infrastructure Solutions, Inc.
The MAP-21 and FAST ACT surface transportation authorization bills have required that pavement performance measures be established for the Interstate Highway System (IHS). As a result of their enactment, the Federal Highway Administration (FHWA) issued a rule defining pavement performance measures to assess the condition of the pavements on the IHS. Because the measures rely on pavement condition data stored in the Highway Performance Monitoring System (HPMS), FHWA undertook a study in 2015 to: (1) collect an unbiased baseline condition of a statistically significant sample of the entire IHS and produce a report indicating the pavement condition on the IHS nationally and in each State where data were collected, (2) determine if HPMS is an unbiased representation of the pavement condition of the IHS, and (3) recommend improvements to HPMS data collection and reporting necessary to make HPMS unbiased or improve its precision. The results provided the outcomes needed at the time. Two years later, FHWA pursued a follow-up study addressing the same objectives. The 2016 HPMS data were reviewed for data completeness and various stratification factors and subsequently used to define a 7,500-mile route for data collection along the IHS. This paper summarizes the work performed under the project and the results of the analysis.
Incorporating Pavement Smoothness Benefits to Enhance the Iowa Department of Transportation’s Pavement Type Determination Process
MAX GROGG, Applied Pavement Technology, Inc.Show Abstract View Presentation
Kelly Smith, Applied Pavement Technology, Inc.
Christopher Williges, HDR
Scott Schram, Iowa Department of Transportation
The FHWA’s Pavement Policy as codified in 23 CFR 626 states, “ Pavement shall be designed to accommodate current and predicted traffic needs in a safe, durable, and cost effective manner ” to be eligible for federal highway funding. To meet this requirement, state highway agencies have developed pavement type determination (PTD) policies, also known as pavement type selection, and implemented pavement management. The Iowa Department of Transportation’s (DOT’s) previous PTD had been in place for many years. In 2018, the Iowa DOT looked at enhancing their PTD process to address gaps between past practice and best practice. Among the enhancements, user benefit as defined by pavement smoothness was utilized when net present value (NPV) alone could not definitively distinguish a preferred alternative. The smoothness benefit would become the divisor in a cost-benefit ratio used to determine the preferred alternate for the PTD. The cost portion of the ratio would remain the NPV of agency costs for the construction and projected rehabilitations during the analysis period. After a literature review and interviews of comparable state DOT’s, several further modifications to Iowa DOT’s PTD and the cost-benefit ratio were analyzed and adopted. The modifications range from accepted practice changes, such as the use of a longer analysis period (50 years), to unconventional techniques, such as the consideration of smoothness. Iowa DOT believes these changes provide a more robust PTD. Iowa DOT is also considering additional improvements based upon additional research and policy making.
A Bayesian Approach to Enhance Accuracy of Time-Series Analyses and Prediction of Pavement Condition Data
Tim Blumenfeld, Technische Universitaet DarmstadtShow Abstract View Presentation
An accurate description of road deterioration is one of the most challenging aspects within the scope of PMS. When deriving probabilistic deterioration models from empirical condition data, one often compares section-based condition states of two measuring campaigns. However, research has shown a high amount of road sections with measured condition improvements. The reasons for these improvements are often unknown due to restricted data quality of recorded maintenance works. A consequent exclusion of measured condition improvements leads to a systematic violation of type one and type two errors from a statistical point of view: Road sections whose condition truly got worse (or sometimes got even better) are excluded due to a measured condition improvement. Even though the mentioned problem is already partially known, suitable methods to handle this issue are hardly existing. This study aims at taking a closer look at Bayesian approaches, which allow combining a series of condition measurements with the statistical noise of the measuring method into one model. By applying this method, it is possible to estimate the most likely condition development over time, including uncertainty in prediction. The described method was applied to condition data of a long-term observation section in Germany, which is measured twice a year. The results show that it is possible to accurately estimate the condition deterioration and the associated uncertainties using a probabilistic forecast. Within the scope of PMS, this approach helps to give an improved basis for decision making and finance investments.
Non-Parametric Deterioration Curves Using Difference Data on Condition Measurements: Methodology and Initial Empirical Evidence
Craig Richmond, University of PittsburghShow Abstract View Presentation
Tariq Saeed, Purdue University
One of the central questions of pavement deterioration modeling is the selection of an adequate functional form. The selection is often made a priori leaving only the parameter values open for adjustment by the data, but not the curves general shape. Non-parametric methods reduce or eliminate this restriction and allow the data to determine both the shape of the deterioration curve and the parameter values. Here, the specific non-parametric method involves calculating local rates of deterioration and then summing the expected time it takes to cross a set of predetermined condition state intervals. This yields a connected curve whose shape is unconstrained. The method is carried out on a set of data from Switzerland. The results lead to a set of general observations about deterioration within the data set. It is found that the key differentiating factor for longevity occurs after the roads have begun the process of minor preventive maintenance. In that phase, the harshness of the climate plays a key role. This has important implications for deterioration modeling in the context of PMS applications.
Prediction of Pavement Performance Using Non-Homogeneous Markov Models: Incorporating the Impact of Preventive Maintenance
Mohamed S. Yamany, Purdue UniversityShow Abstract View Presentation
Dulcy M. Abraham, Purdue University
Highway agencies strive to establish effective pavement management systems that contribute to accurate prediction of pavement condition and optimal decision-making of maintenance and rehabilitation (M&R). Hence, various modeling methodologies have been developed to probabilistically model pavement performance. However, the impact of preventive maintenance has not been considered in the probabilistic pavement performance models owing to the lack of preventive maintenance data. This paper introduces a research methodology to estimate the times and types of preventive maintenance treatments most likely to be implemented to pavements and integrate them into pavement condition data. The times and types of preventive maintenance treatments were identified by investigating and analyzing the corresponding state of practice and pavement performance curves. Then, they were integrated into pavement condition data using a newly developed optimization algorithm. Furthermore, the non-homogeneous transition probabilities of pavement condition were estimated using the ordered-probit method to predict pavement performance stochastically. The introduced methodology and developed models were validated and found to be robust in estimating and predicting the probabilistic pavement condition while incorporating the impact of preventive maintenance. The results emphasize the significance of considering the effect of preventive maintenance in probabilistic pavement performance models. The contribution of this research can help State Transportation Agencies (STAs) to predict pavement condition accurately, make M&R decisions optimally, and eventually improve their pavement management systems.
Road Load–Based Model for Vehicle Repair and Maintenance Cost Estimation
Muluneh Sime, Nevada Automotive Test CenterShow Abstract View Presentation
Gary Bailey, Childrens Heart Center Nevada
Elie Hajj, University of Nevada, Reno
Rami Chkaiban, University of Nevada, Reno
Reid Pulley, Nevada Automotive Test Center (NATC)
A techno-economic model is developed based on road load simulation results expressed in terms of slip energy (SE) at the tire–pavement interface and the repair and maintenance (R&M) cost obtained from published sources and data from state agencies. R&M costs were estimated for various vehicle categories and accumulated vehicle mileage. The approach is based on relating the probability density functions (PDFs) of slip energy and R&M costs. Asymptotic series expansion for an incomplete gamma function was used to approximate the gamma functions and to determine the gamma ratio function that is used as the coefficient to SE to estimate R&M costs. The average R&M cost per mile results from the model compared well with the arithmetic mean R&M cost data from fleet operators and published data. The model can serve as a method of predicting repair and maintenance cost as a function of road load to vehicle fleet.
Development of Machine Learning Models for Predicting Rutting Classification in Asphalt Pavement Using Network-level Data
Seyedamin Banihashemrad, Texas A&M University, College StationShow Abstract View Presentation
Charles Gurganus, Texas A&M University, College Station
Maryam Sakhaeifar, Texas A&M University, College Station
Automated rut measurements are used as indicators of maintenance and rehabilitation needs. A common practice by many agencies is to use the average rut depth of the left wheel-path rutting and right wheel-path. However, there are significant variations of rut depth between left and right wheel-paths. If the roadways are paved with the same material, equipment, and workforce concurrently, the only difference between the left and right wheel path is the roadside features and the drainage quality. This paper describes the development of machine learning model for predicting rutting classification based on roadside features, readily available soil data, and pavement condition data. Several machine learning methods including logistic regression, k-nearest neighborhood (KNN), support vector machines (SVM), decision tree, random forest trees (RF), and gradient boosting trees (GB) are tested. A framework is implemented to obtain classification models with optimum parameters. The models were evaluated using an independent test dataset. Moreover, the critical factors affecting rutting classification were derived from the relative importance concept available in the tree models. Finally, the marginal effect of these attributes on the output was analyzed and visualized using partial dependent plots. The results suggest that a minimum roadside ditch depth of 2 feet is required to guard against rutting in the adjacent pavement. Keywords: LiDAR data, Machine Learning, Rutting, Network-Level Pavement Management
Large-Scale Evaluation of Pavement Performance Models Utilizing Automated Pavement Condition Survey Data
Xiang Shu, California Department of Transportation (CALTRANS)Show Abstract View Presentation
Zhongren Wang, California Department of Transportation (CALTRANS)
Imad Basheer, California Department of Transportation (CALTRANS)
Pavement performance models form an essential component of a pavement management system (PMS) and have a direct impact on future pavement condition prediction, selection of future pavement maintenance and rehabilitation (M&R) methods, and budget planning and allocations. Therefore, it is critical to develop and maintain pavement performance models as accurate as possible. The California Department of Transportation (Caltrans) has implemented a modern pavement management system called PaveM. To maintain the intended functions of PaveM, regular updates of its databases and key components are necessary, including pavement performance models. This paper aims to evaluate the network-level pavement performance models in PaveM by utilizing the concept of deterioration rate and the most recent automated pavement condition survey (APCS) data. First, the concept of pavement deterioration rate was defined. The actual deterioration rates were calculated using the latest two cycles of APCS data and then compred to the predicted deterioration rates obtained using APCS data and the current configurations of PaveM. Two typical pavement distresses (International Roughenss Index and the Caltrans’ asphalt pavement Alligator B cracking) for one selected pavement treatment (thin overlay) were predicted using PaveM and compared to the actual APCS measurements. The results from this study show that the simple concept of deterioration rate was effective in evaluating the overall quality of performance models in PaveM as well as identifying the time spans in which the models over- or under-predict pavement performance.
Deterioration Modeling of Flexible Pavements Based on As-Produced and As-Constructed Properties
Arash Mohammad Hosseini, Temple UniversityShow Abstract View Presentation
Ahmed Faheem, Temple University
Hani Titi, University of Wisconsin, Milwaukee
The goal of this study is to improve the understanding of the long-term performance of flexible pavements with regards to construction and production quality and to promote the sustainability of this critical infrastructure. New advancements in data analytics allow for utilization of historical data to evaluate the effectiveness of the Quality Control Programs(QCP). The data is geo-referenced to establish a connection between QCP and long-term performance. The developed relational database is used to associate the influence of production and construction quality on measured performance-related indicators. The data used in this paper is sampled from the Wisconsin Department of Transportation. In this paper, data is filtered to include pavement sections of the same traffic load and environmental conditions to avoid potential bias. Information of 42 highways with the total length of 240 miles were collected and analyzed for this study. Pavement deterioration metamodels were developed using machine learning techniques. Three ML techniques including Decision Tree Regression (DTR), Random Forest (RF), and Gene-Expression Programming (GEP) were utilized using coded subroutines. The outcomes of DTR, RF, and GEP showed promising results in the prediction of pavement performance by considering the contribution of mix production quality indicators factors such as individual lots Voids in Mineral Aggregates (VMA), Air Voids of the mixture (VA), In-place density of asphalt concrete, asphalt content, surface thickness, and the age. This approach provides the basis of adaptive pavement management system that is continuously reflecting the network performance.
Statistics and Artificial Intelligence–Based Pavement Performance and Remaining Service Life-Prediction Models for Iowa Flexible and Composite Pavement Systems
Orhan Kaya, Iowa State UniversityShow Abstract
Halil Ceylan, Iowa State University
Sunghwan Kim, Iowa State University
Danny Waid, Iowa County Engineers Association Service Bureau
Brian Moore, Iowa County Engineer Association
In their pavement management decision-making processes, state highway agencies (SHAs) are required to develop performance-based approaches based on The Moving Ahead for Progress in the 21st Century (MAP-21) Federal Transportation Legislation. One of the performance-based approaches to facilitate pavement management decision-making process is use of remaining service life (RSL) models. In this study, a detailed step-by-step methodology for the development of pavement performance and RSL prediction models for Iowa flexible and composite (Asphalt concrete (AC) over Jointed Plain Concrete Pavement (JPCP)) pavement systems is described. To develop such RSL models, pavement performance models based on statistical and artificial intelligence (AI) techniques were initially developed. While statistically (or mathematically) defined pavement performance models were found to be accurate in predicting pavement performance at project level, AI based pavement performance models were found to be successful in predicting pavement performance in network level analysis. Network level pavement performance models using both statistical and AI based approaches were also developed to evaluate the relative success of these two models for network-level pavement-performance modeling. As part of this study, in development of pavement RSL prediction models, automation tools for future pavement performance predictions were developed and used along with Federal Highway Administration (FHWA)-specified threshold limits for various pavement performance indicators. These RSL models will help engineers in both network and project level decision-making processes and for different types of pavement-management business decisions.