< Back to Session View 

Predicting Iowa Concrete Overlay Performance and Remaining Service Life using Statistics and Deep Learning Techniques

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

 

Lead Presenter: Nazik Citir Razavi

, 

Speaker Biography

Nazik Citir Razavi is a Ph.D. candidate specializing in Intelligent Infrastructure Engineering and Structural Engineering at Iowa State University. She actively contributes as a research member of the Program for Sustainable Pavement Engineering and Research (PROSPER). With a focus on pavement engineering, she has delved into the fascinating world of infrastructure asset management by evaluating pavement performance and structural capacities.

Her research background and interests cover a wide range of areas. She is deeply involved in developing performance predictive modeling, leveraging artificial intelligence and statistical methods. Her involvement extends to evaluating pavement structural capacities, functional and structural performance, and exploring innovative ways to develop user-friendly automation tools that estimate pavement performance and remaining service life. Her work aims to contribute to more effective pavement management decision-making processes. She is also interested in using non-destructive techniques to assess pavement condition, promoting sustainable and cost-efficient solutions.

Beyond her academic pursuits, she enjoys engaging with industry professionals and peers. Through active knowledge exchange, she strives to foster creativity in addressing challenges within the intelligent infrastructure field.

Co-Authors 

Halil Ceylan, Ph.D., Dist.M.ASCE, Pitt-Des Moines, Inc. Endowed Professor in CCEE, Iowa State University;
Orhan Kaya, Ph.D., Assistant Prof., Adana Alparslan Türkeş Science and Technology University, Adana, Turkey;
Sunghwan Kim, Ph.D., P.E., Research Scientist, Iowa State University;
Danny R. Waid, Executive Director, Iowa County Engineers Association (ICEA) Service Bureau

Presentation Description/Paper Summary

This study focuses on developing a pavement performance analysis tool for Iowa County Portland-cement concrete (PCC) overlay roads and evaluates statistical and deep learning techniques using an artificial neural network (ANN) for predicting the international roughness index (IRI) and resulting in forecasting pavement remaining service lives (RSLs). In terms of statistical technique, a sigmoid pavement deterioration curve-based approach was utilized for IRI calculations for county PCC overlaid pavement sections. For developing ANN models, input parameters consisted of material and construction design properties of pavement, traffic, and condition measurements. This input dataset was created by combining a historical database provided by the Iowa Concrete Paving Association, a condition database by the Iowa Pavement Management Program, and traffic data from the Iowa DOT Roadway Asset Management System/open data online. Overall, the statistics- and ANN-based IRI models produced high accuracy in model development and independent test. These models were incorporated into a macro-enabled Microsoft Excel and Visual Basic for Applications (VBA)-based automation tool (Iowa Pavement Analysis Techniques [IPAT]) developed for analysis of project-level pavement performance, prediction of future pavement performance, and estimation of RSL developments for any given road section.

Presentation File

Poster