Metal culvert renewal prioritization framework development: A study for Saskatchewan Highways

Saskatchewan comprises over 250,000 km (160,000 mi) of roads, the highest length of road surface compared to any other Canadian province. Along these roads, over 26,500 culverts have been installed for passaging water flow. Some of which have been installed over a century ago with no recorded installation dates. The failure of culverts, caused by a wide range of factors such as environmental, structural, or material failures, can result in sudden and catastrophic consequences, leading to injuries or loss of life. Additionally, damaged culverts can negatively impact water quality by causing erosion and scour, as well as impeding the passage of aquatic species between their habitats upstream and downstream. Thus, timely replacement of culverts can play a crucial role in minimizing such damages. This study evaluates three methods, namely, ordinal logistic regression, artificial neural network, and Fuzzy Inference System, for the prioritization of culverts for renewal. Specifically, the approach presented in this research is developed for metal pipe materials, which are extensively used in culvert installations. Using the condition of 1,000 metal culverts located along Saskatchewan highways, the three methods were employed to rank their condition and determine their renewal time. The evaluation of model performance was conducted using a range of established metrics including the area under the ROC (Receiver Operating characteristic) curve, percentage of correct predictions (PCP), confusion matrix, accuracy, precision, recall, and F1 score. The results of the study indicate that the artificial neural network optimized by genetic algorithm outperforms the other two methods, providing the most effective approach for culvert renewal prioritization. Keywords: Culvert renewal; Ordinal Logistic Regression; Artificial Neural Network; Fuzzy Inference System.

Name thesis:degree_name

Master of Applied Science (MASc)

Level thesis:degree_level

Master's

Discipline thesis:degree_discipline

Engineering - Industrial Systems

Grantor dc:publisher

Faculty of Graduate Studies and Research, University of Regina

Year dc:date.issued

2023

Author dc:creator

• Marzbani, Pooneh

Advisor dc:contributor.advisor

• Kabir, Golam

Committee members dc:contributor.committeemember

• Khan, Sharfuddin
• Khondoker, Mohammad

Language dc:language.iso

en

DOI

https://doi.org/10.82465/4701

OAI identifier oai:identifier

oai:uregina.scholaris.ca:10294/16497