Final Results of the 2007 International Contest on LTPP Data Analysis

University students and professors were invited to participate in the 2007 International Contest on LTPP Data Analysis, co-sponsored by the American Society of Civil Engineers (ASCE) and the Federal Highway Administration (FHWA). The contest received 12 papers. It was designed to encourage students and professors from around the world to get involved in using the LTPP database. The contest created an opportunity for students and professors to use this significant database for research, class projects, Master's and Doctoral theses, and practical fieldwork to resolve common engineering problems. This year, ASCE/LTPP received papers in two categories.

Contest Award Ceremony

Winners of this years contest received their awards on Sunday January, 13, 2008 during the LTPP State Coordinators Meeting.

group photo-The people photographed are, top row from left to right, Gary Henderson (FHWA), Jared Richard Wiesman (Penn State), Brian Wolfgang (Penn State), Kam K. Movassaghi (ASCE), Orlando Núñez (VTech), and Oscar Gonzalez (VTech). Bottom row from left to right Vanessa I. Rosales-Herrera (UT Austin), Feng Hong, PhD (UT Austin), and Soly Alvarez Claro (VTech).
The people photographed are, top row from left to right, Gary Henderson (FHWA), Jared Richard Wiesman (Penn State), Brian Wolfgang (Penn State), Kam K. Movassaghi (ASCE), Orlando Núñez (VTech), and Oscar González (VTech). Bottom row from left to right Vanessa I. Rosales-Herrera (UT Austin), Feng Hong, PhD (UT Austin), and Soly Alvarez Claro (VTech).

Contest Categories and Prizes

The papers were evaluated using the following criteria:

Usefulness of product and potential benefit to end users.
Originality of concept, demonstrated use of the LTPP database.
Organization of paper, and presentation (including clarity and style).

This Year's Winners

UNDERGRADUATE CATEGORY:

Prize	Name	Faculty Advisor
1st Place	Jared Weisman Brian Wolfgang	Shelley M. Stoffels, D.E.

GRADUATE CATEGORY:

Prize	Name	Faculty Advisor
1st Place	Orlando Nunez Soly Alvarez Oscar González	Gerardo W. Flintsch
2nd Place	Chen Chen	Gerardo W. Flintsch
3rd Place	Feng Hong, PhD Vanessa I. Rosales-Herrera	Jorge A. Prozzi, PhD

Abstracts of the Winning Papers

OBSERVATIONAL STUDY OF UNBONDED CONCRETE OVERLAYS IN THE LTPP DATABASE
Jared Weisman
Brian Wolfgang
First Place, UNDERGRADUATE CATEGORY

This paper analyzes the relationship between structural characteristics of unbonded concrete overlays and various forms of distress. Due to the complex nature of concrete overlays and the inability to accurately model their behavior, this is an investigation for the sake of identifying significant trends in pavements that have been in use for many years. This analysis is purely an observational study. In this study, 26 sections of pavement with concrete overlays were analyzed using the Long-Term Pavement Performance database. The structural and physical characteristics that were investigated include interlayer thickness, overlay thickness, overlay to underlayer ratio, and age of the overlay. Each of these components was compared with various forms of pavement distress, including cracking, spalling, corner breaks, IRI (International Roughness Index), and faulting.

The findings of this investigation confirm some previously accepted notions regarding concrete overlay performance. This study also allowed for more specific recommendations. When relating interlayer thickness to pavement distress, a thicker interlayer yielded less distress than thinner ones. Overall, a thickness greater than 2 inches performed significantly better than smaller interlayers. In terms of actual overlay thickness, once again a thicker layer outperformed thinner overlays. A thickness of at least 8.75 inches showed fewer signs of distress. In addition, pavements with an overlay to underlayer thickness ratio of 0.80 or less are more prone to higher frequency and severity of distress. Lastly, age was found to significantly play a role in faulting, corner breaks, and spalling, but that it did not have a dominant effect on the various types of cracking.

VERIFICATION OF THE PROPOSED MEPDG ASPHALT PAVEMENT REHABILITATION IRI MODEL USING THE LTPP DATABASE
Orlando Nunez
Soly Alvarez
Oscar González
First Place, GRADUATE CATEGORY

A verification of the MEPDG IRI model for flexible rehabilitated pavement sections for the state of Virginia was performed. The IRI model verification was based on data available from the Long-Term Pavement Performance (LTPP) program. The values entered to the MEPDG software (version 0.910) were obtained from the LTPP website, DataPave Online; the data used included material characterization, traffic, climate, structure description and characterization, performance monitoring, and rehabilitation information. The IRI model prediction for eight sections was compared to the obtained IRI values from the LTPP database. Statistical verification analyses which included a confidence-interval, a correlated inspection, a probability calculation, and graphical approach were performed on the data obtained.

The various results showed that the MEPDG IRI model values are not statistically similar to the field obtained ones at a 90% confidence level. However, from visual inspections of the different plots, it can be observed that the model tends to follow a similar trend to the field IRI measurements. A correlation of R2 = 0.7788 (coefficient of determination) from an equality plot was calculated in addition to a visual analysis. This result suggests that the MEPDG IRI model for flexible rehabilitated pavements needs to be calibrated to match local conditions.

CALIBRATING FUZZY-LOGIC-BASED PAVEMENT REHABILITATION DECISION MODELS USING THE LTPP DATABASE
Chen Chen
Second Place, GRADUATE CATEGORY

This paper establishes a systematic method to calibrate a fuzzy-logic-based rehabilitation decision model using real cases extracted from the Long Term Pavement Performance (LTPP) database. The fuzzy system was developed to conduct life cycle costs analysis (LCCA) transportation infrastructure assets. Fuzzy logic systems are efficient for combining expert knowledge and numeric data. This feature makes this technology ideal for establishing a decision-support model based on expert knowledge at the initial phase and tuning the model with numeric data as further data is collected. The following tasks have to be accomplished to develop the proposed method: (1) extract representative rehabilitation events and related pavement information from the database; (2) identify proper input areas for engineering knowledge and numeric data; and (3) simultaneously tune two fuzzy logic systems with shared membership functions for input variables.

A total of eight tables in the LTPP database were used from which to extract pavement and rehabilitation information. The investigation started with 62 rehabilitation cases but only six overlay rehabilitations with thicknesses between 1.5 and 2.5 inches had all the required information and were thus selected to calibrate the decision model. To make the dataset unbiased, six do-nothing cases were created based on the rehabilitation cases. The steepest descent method and back-propagation learning were used to tune the model based on the selected rehabilitation events. By reinterpreting the model in the form of neural fuzzy system, the calibration algorithm successfully tuned the decision model to distinguish between rehabilitation cases and do-nothing cases with an error rate of zero for the calibration dataset.

Development of Transverse Crack Initiation Models in Asphalt Pavements by Applying Generalized Linear Models to Long Term Pavement Performance (LTPP) Data
Feng Hong, PhD
Vanessa I. Rosales-Herrera
Third Place, GRADUATE CATEGORY
Pavement distress models are vital for infrastructure management. Accurate models that predict the initiation of pavement distress can provide a crucial insight into the expected condition of pavement infrastructure along its service life, which in turn is particularly useful in planning and budget allocation for maintenance and rehabilitation activities at both the project and network levels. This paper focuses on developing a distress initiation model for transverse cracking. Data from the real world pavements, from the Long Term Pavement Performance (LTPP) program is used to develop a survival model which accounts for censoring bias due to unobserved events such as limited observation duration. A statistical technique - maximum likelihood estimation is applied to estimate parameters for the model. The model estimation results reflect pavement transverse cracking initiation time from a probabilistic viewpoint and also are consistent with engineering judgment.