A Risk-Based Framework to Evaluate Infrastructure Investment Options for a Water Supply System

Abstract

Increasing water demand due to socio-economic development often requires structure-level interventions, e.g., infrastructure expansion of water supply systems to bridge the gap between demand and water supply. Evaluating different expansion options is critical to making informed decisions. This study focuses on evaluating infrastructure investment options from the reliability perspectives of a water supply system. The evaluation framework is comprised of probabilistic demand projections, stochastic streamflow generation, a mixed-integer programming, and system performance evaluation. Different “industry understood” performance metrics, e.g., annual reliability, maximum delivery capacity, the probability of unsuccessful status, and the magnitude of potential water shortage, can be employed to evaluate the performance of water supply systems. This framework has been applied to Tampa Bay Water, a regional supply agency on the west coast of Florida, United States. The probabilistic demand projections for the future (2021–2040) consider socio-economic projections and climate conditions based on the statistics of historical observations. The Latin Hypercube Sampling algorithm is used to capture the bivariate distribution of supply and demand. This mixed-integer programming is a daily system optimization that minimizes operational cost and considers operational constraints and preferences, e.g., water withdrawal permits, facility production capacity, and water production preferences. Current infrastructure capacity and operational conditions are defined as the baseline scenario. Two additional planning scenarios of expanding the treatment capacity of a surface water treatment plant by 75,708 m3/day [20 million gallons per day (mgd)] and 113,562 m3/day (30 mgd) in the year 2028 are investigated. Results reveal that annual reliability would increase with the expansion of surface water treatment plant capacity. The spatio-temporal supply and demand variability determines that the firm yield from the surface water treatment is less than its expanded production capacity. Except for low demand realizations in the future, the duration and magnitude of potential water shortage could be reduced in both scenarios with engineering intervention. Although this framework is demonstrated for a regional water utility in Florida, it can be applied to other water supply systems around the world.