A Fuzzy Multidimensional Risk Assessment Method for Sewer Asset Management

Abstract

This paper presents a flexible framework that aims at estimating the risk of structural failure in sewer pipes by utilizing limited or imperfect data. To this end, classical risk analysis is enhanced by incorporating fuzzy logic and multi-criteria decision making. To account for the multi-dimensionality of collapse risk at the pipe level as a decision parameter, its distinct impacts on the environment, traffic and road condition, and quality of life are taken into account. The proposed method is applied to the sewer network of Tehran, the capital of Iran. Results show how the integration of different risk indexes can influence the criticality of pipelines for the selection of rehabilitation activities. While using the first individual risk index, only considering the risk posed to the natural environment by a collapsing pipe in terms of contamination, approximately half of the pipe lengths are classified as extremely critical by the clustering algorithm. However, when the integrated risk is calculated, this cluster encompasses only approximately 30% of the total pipe length. With a database that contains various levels of uncertainty (from 10 to 60%), the predictive reproducibility for the exact same risk cluster is above 20% and above 50% for the same or only marginally better or worse. Furthermore, pipelines that are predicted to have a better risk class than the situation without considering uncertainty, thereby underestimating the likelihood of failures or consequences, are below 15%, showing a measure of quite good robustness. Considering the budget constraints of utilities, the proposed method can be applied to any urban, aiding in the identification of high-risk sections. Nevertheless, incorporating physical validation might be beneficial for further improving the analysis.

Bridging Geo-Data and Natural Gas Pipeline Design Standards: A Systematic Review of BIM-GIS Integration for Natural Gas Pipeline Asset Management

“…In order to carry out activities that will prevent these risks on time and reduce maintenance costs, accurate data must be collected in order to maintain fault detection and maintenance practices effectively. Using these data, decision support systems can be developed with various data analysis methods such as machine learning, fuzzy logic, and regression [108][109][110][111].…”