Real-time Measurement and Control of Urban Stormwater Systems

Abstract

Urban watersheds are being stressed beyond their capacity as storms are becoming more frequent and intense. Flash flooding is the leading cause of natural disaster deaths in the United States. Simultaneously, population pressures are changing landscapes and impairing water quality by altering the composition of urban stormwater runoff. Presently, the only solution to combat these challenges relies on the construction of larger infrastructure, which is cost prohibitive for most cities and communities.

Advances in technology and autonomous systems promise to usher in a new generation of “smart” stormwater systems, which will use city-scale sensing and control to instantly “redesign” themselves in response to changing inputs. By dynamically controlling pumps, valves and gates throughout the entire city this paradigm promises to push the performance of existing assets without requiring the construction of new infrastructure. This will allow for entire urban watersheds to be dynamically controlled to meet a variety of desired outcomes.

Despite technological advances and an established fundamental knowledge of water systems, it is presently entirely unclear how “smart” stormwater systems can actually be built. This dissertation conducts a review of existing “static” solutions and provides an assessment of a number of limited, but highly promising, real-world control studies. An analysis of sensor network scalability is then carried out, focusing on how large water sensor networks can be enabled by leveraging wireless connectivity and web-services. A study of urban water quality follows, which shows how real-time data improve our watershed-scale understanding of pollutant loads during storm events. In turn, through an unprecedented real-world study, it is illustrated how this improved understanding can be used to control flows across a watershed. A feedback control-based approach is then introduced to enable the control of urban watersheds. Through extensive simulation, this framework is applied to identify which control assets have the highest potential to improve watershed performance and to determine how many sites must be retrofitted to achieve desired outcomes. Finally, an analysis of input uncertainty is carried out, which quantifies the importance of weather forecasts in improving control performance across the scale of urban headwater catchments. The dissertation closes by laying out future directions in the emerging field of “smart” stormwater research.