Recent reports by the American Water Works Association and the U.S. Environmental Protection Agency indicate a significant need of investment in the nation’s drinking water infrastructure to maintain desired service levels. Coupled with a continuously decaying asset inventory, most utilities are funding their capital improvement projects by prioritizing the riskiest pipes for renewal. A system-wide risk analysis is generally implemented to identify high-risk assets in the system that have the most significant impact in the event of a failure. Such analyses usually require a wealth of system-related data, such as pipe break history, hydraulic model of the network, and other social, economic, and environmental pipe data. However, many utilities do not have the possibility to evaluate and manage such data, especially small- to mid-sized utilities which often struggle with limited financial and human resources to fully digitize their system. To overcome the limitations of lack of sufficient quantitative data for performing a data-intensive risk analysis, this paper presents a data-driven qualitative risk framework that provides a baseline risk evaluation for a water system using minimally available data. Specifically, using information about pipe material and diameter only, a water network’s baseline risk can be evaluated. For this, two multicriteria models were developed to assess the water pipes’ likelihood of failure (LoF) and consequence of failure (CoF). To capture and model the complex relationships between variables in both the LoF and CoF models, Bayesian network (BN) modeling was adopted. BN is a probabilistic graphic model that captures the interdependencies of the LoF and CoF model variables together with their probabilistic relationship, therefore being an adequate modeling technique to capture the complexity of a water system. The proposed risk model has been applied to a Mid-Atlantic Water utility network. The results show that the proposed risk model can be used as a baseline risk assessment in the preliminary justification of the utility’s decision-making process for next steps in their capital improvement project planning.

Click Here  for the ASCE Library Article

Recent reports by the American Water Works Association and the U.S. Environmental Protection Agency indicate a significant need of investment in the nation’s drinking water infrastructure to maintain desired service levels. Coupled with a continuously decaying asset inventory, most utilities are funding their capital improvement projects by prioritizing the riskiest pipes for renewal. A system-wide risk analysis is generally implemented to identify high-risk assets in the system that have the most significant impact in the event of a failure. Such analyses usually require a wealth of system-related data, such as pipe break history, hydraulic model of the network, and other social, economic, and environmental pipe data. However, many utilities do not have the possibility to evaluate and manage such data, especially small- to mid-sized utilities which often struggle with limited financial and human resources to fully digitize their system. To overcome the limitations of lack of sufficient quantitative data for performing a data-intensive risk analysis, this paper presents a data-driven qualitative risk framework that provides a baseline risk evaluation for a water system using minimally available data. Specifically, using information about pipe material and diameter only, a water network’s baseline risk can be evaluated. For this, two multicriteria models were developed to assess the water pipes’ likelihood of failure (LoF) and consequence of failure (CoF). To capture and model the complex relationships between variables in both the LoF and CoF models, Bayesian network (BN) modeling was adopted. BN is a probabilistic graphic model that captures the interdependencies of the LoF and CoF model variables together with their probabilistic relationship, therefore being an adequate modeling technique to capture the complexity of a water system. The proposed risk model has been applied to a Mid-Atlantic Water utility network. The results show that the proposed risk model can be used as a baseline risk assessment in the preliminary justification of the utility’s decision-making process for next steps in their capital improvement project planning.

Click **[Here](https://ascelibrary.org/doi/10.1061/9780784483602.015)**  for the ASCE Library Article

Risk Analysis in the Absence of Data to Help Plan Capital Improvement Projects

Papers from sessions of Pipelines 2021, held virtually August 3–6, 2021. Sponsored by the Utility Engineering and Surveying Institute of ASCE.

This collection contains 43 peer-reviewed papers on the planning and asset management of pipelines.
Topics include: machine learning and artificial intelligence; sewermain condition assessment; asset management; watermain condition assessment; and case studies.
This collection will be of interest to design and consulting engineers, utility and pipeline owners, and researchers in the pipeline profession.

Paper Proceedings: https://ascelibrary.org/doi/book/10.1061/pipelines2021