AI-Driven Decision Support Using CIRx for Smart Urban Water Infrastructure

Abstract

Smart urban water infrastructure spans drinking water distribution, stormwater conveyance, and wastewater collection systems, each generating growing volumes of sensor, inspection, and operational data as utilities adopt SCADA telemetry, smart metering, and IoT-based condition monitoring. Translating this data into timely, defensible capital and operational decisions remains difficult, particularly for resource-constrained municipal utilities that must prioritize a limited rehabilitation budget across a large and heterogeneous asset base. This paper presents an artificial intelligence (AI) driven decision support framework built around the Condition-Index Rx (CIRx) methodology, a composite scoring approach that fuses structural condition, hydraulic or hydraulic-equivalent performance, consequence of failure, and machine-learning-forecast deterioration risk into a single, prescriptive rehabilitation recommendation. We describe how CIRx extends naturally across the three principal urban water subsystems, and we present a decision-support architecture that couples AI/ML predictive models with an interactive prioritization layer intended for use by utility engineers and capital planners. The paper discusses model selection considerations for heterogeneous urban water asset types, human-in-the-loop decision workflows, integration with geographic information systems and asset management platforms, and the governance considerations relevant to using AI-generated prescriptions to inform public infrastructure spending. We conclude with a discussion of adoption barriers among small and mid-sized utilities and directions for future research, including transferable deterioration models and explainability requirements for public-sector AI decision support.