Causal inference approach to appraise long-term effects of maintenance policy on functional performance of asphalt pavements

TL;DR

This paper introduces a causal inference framework to evaluate the long-term impact of maintenance policies on asphalt pavement performance, accounting for environmental uncertainties and treatment effects over five years.

Contribution

It develops a novel causal inference method combining structural and potential outcome models to assess maintenance effects on pavement cracking.

Findings

Accurately evaluates the effects of preventive maintenance treatments.

Assists policymakers in selecting effective pavement maintenance strategies.

Provides a framework for long-term pavement performance assessment.

Abstract

Asphalt pavements as the most prevalent transportation infrastructure, are prone to serious traffic safety problems due to functional or structural damage caused by stresses or strains imposed through repeated traffic loads and continuous climatic cycles. The good quality or high serviceability of infrastructure networks is vital to the urbanization and industrial development of nations. In order to maintain good functional pavement performance and extend the service life of asphalt pavements, the long-term performance of pavements under maintenance policies needs to be evaluated and favorable options selected based on the condition of the pavement. A major challenge in evaluating maintenance policies is to produce valid treatments for the outcome assessment under the control of uncertainty of vehicle loads and the disturbance of freeze-thaw cycles in the climatic environment. In this study, a novel causal inference approach combining a classical causal structural model and a potential outcome model framework is proposed to appraise the long-term effects of four preventive maintenance treatments for longitudinal cracking over a 5-year period of upkeep. Three fundamental issues were brought to our attention: 1) detection of causal relationships prior to variables under environmental loading (identification of causal structure); 2) obtaining direct causal effects of treatment on outcomes excluding covariates (identification of causal effects); and 3) sensitivity analysis of causal relationships. The results show that the method can accurately evaluate the effect of preventive maintenance treatments and assess the maintenance time to cater well for the functional performance of different preventive maintenance approaches. This framework could help policymakers to develop appropriate maintenance strategies for pavements.