Pavement management with dynamic traffic and artificial neural network: a case study of Montreal

Amin, Md. Shohel Reza; Amador-Jiménez, Luis E. · 2016 · Crossref

DOI: 10.1139/cjce-2015-0299

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This study addresses the deterioration of Montreal’s aging road network, which has suffered from inadequate funding and a lack of comprehensive pavement management systems (PMS). Traditional PMS tools often rely on static traffic predictions and deterministic or stochastic models that fail to account for dynamic traffic loads and modeling uncertainties. To resolve these limitations, the authors developed a linear programming optimization framework for Montreal’s road network that integrates simulated dynamic traffic over a 50-year period and utilizes Artificial Neural Networks (ANN) to minimize uncertainty in pavement performance modeling. The methodology comprised three main steps: traffic simulation, pavement performance modeling, and lifecycle optimization. First, a travel demand model simulated Annual Average Daily Traffic (AADT) and Equivalent Single Axle Loads (ESALs) for each road segment every five years from 2009 to 2058. This model accounted for demographic factors, land use, and travel behavior using a discrete choice model and a deterministic User Equilibrium assignment. Second, a Backpropagation Neural Network (BPN) with a Generalized Delta Rule learning algorithm was employed to predict pavement deterioration. The BPN used input variables including AADT, ESALs, pavement age, structural characteristics, and the difference in Pavement Condition Index (PCI) between years to model performance for four road categories: arterial-flexible, arterial-rigid, local-flexible, and local-rigid. Third, linear programming was applied to optimize maintenance and rehabilitation (M&R) strategies, minimizing costs while maximizing network condition subject to budget constraints. The results indicated that dynamic traffic simulation revealed significant increases in traffic loads compared to simple growth rates, with arterial-flexible roads seeing a 26.23% increase in simulated traffic. The BPN analysis determined that PCI values were predominantly influenced by the change in PCI from the previous year and pavement age, rather than structural characteristics like slab thickness or structural number, likely due to data aggregation. The optimization model identified that a minimum annual budget of CAD 150 million is required to maintain most arterial and local roads in "good" condition (PCI ≥ 70). Budgets below this threshold, such as CAD 125 million, led to rapid deterioration after the 31st year, while budgets above CAD 150 million did not significantly improve overall condition but shifted roads from "good" to "excellent." The significance of this work lies in its dual improvement over conventional PMS: it incorporates dynamic traffic predictions based on travel demand rather than static historical data, and it addresses modeling uncertainties through ANN. This approach allows transportation authorities to make more accurate, cost-effective long-term investment decisions. The study demonstrates that integrating dynamic traffic and robust performance modeling is essential for managing pavement assets efficiently, particularly in cities with aging infrastructure and complex traffic patterns.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• signage environment