< Back to Session View 

Data-Driven Approach to Identify Maintained Pavement Segments and Estimate Maintenance Type for Local Road

Date and Time: Monday, July 24: 3:30 PM - 5:00 PM
Location: Grand Ballroom C

 

Lead Presenter: Abdallah Al-Hamdan

Graduate Research Assistant, Iowa State University

Speaker Biography

Abdallah Al-Hamdan is an enthusiastic graduate research assistant and PhD student at the Department of Civil, Construction, and Environmental Engineering at Iowa State University. Co-majoring in transportation engineering and intelligent infrastructure engineering, he focuses on areas such as transportation engineering, transportation asset management, pavement performance modeling, GIS, spatial analysis, and navigation applications. With a passion for making a positive impact on transportation systems, Abdallah's research aims to optimize pavement assets and accurately predict their conditions. He strives to develop robust tools and models for efficient pavement asset management practices. His work is driven by a passion for making a positive impact on transportation systems, with a vision to create smarter and more sustainable solutions for the industry, benefiting communities worldwide.

Co-Authors 

Inya Nlenanya, Research Scientist, Iowa State University; Omar Smadi, Associate Professor | CCEE Department, Iowa State University

Presentation Description/Paper Summary

Missing maintenance records is one of the challenges that face the pavement management process for local transportation agencies. Some studies investigated possible solutions to overcome this issue. Part of them suggested using the rate of deterioration concept to generate performance models when the age of pavements is unknown. However, this approach did not give the ability to detect maintenance activities on the road network. Other studies suggested using either probabilistic or deep learning techniques for maintenance detection. In this study, a data-driven approach was proposed and utilized to detect probable maintenance activities on the network. Pavement condition data for municipal roads in Iowa was used to generate performance models for flexible, rigid, and composite pavements, which were used to predict the condition for the year 2021. Predictions were compared to actual data, and the difference between the actual and predicted values was calculated for each road segment. The data clusters were obtained using the mini-batch k-means clustering algorithm. This process was done separately for flexible, rigid, and composite pavement at low-volume and high-volume traffic levels. It was found that the resulting clusters could be used to roughly determine the maintenance type. On the other hand, clusters might be better used to indicate the probability of a segment being maintained based on the value of observed condition improvement for that segment.

Presentation File

Poster