Gabriel, Brubaker developing game theory water market models for river users

news story image

[Click on photo for larger image]

Above: The Anacostia River is an 8.7-mile tributary of the Potomac River that flows through Maryland and Washington, D.C. It is an urban river with a long history of industrial development and pollution; recently the efforts of governments and businesses have led to better water quality.
Below: The Duck River (loosely approximated here) is a 284-mile rural river that winds through the lower middle of Tennessee. Most of the river is free-flowing; it is the most biologically diverse river in the U.S.

With funding from the National Science Foundation, two University of Maryland Clark School of Engineering researchers are building game theory-based water market equilibrium models to help improve water use in the watersheds of the Anacostia River in Maryland and D.C., and the Duck River in Tennessee.

Game Theoretic Modeling for Improved Management of Water and Wastewater Resources Using Equilibrium Programming and Feedback Mechanisms is a three-year, $553,407 grant from NSF’s Civil, Mechanical and Manufacturing Innovation division and Civil Infrastructure Systems program.

ISR-affiliated Professor Steven Gabriel (ME), the principal investigator, is joined by Associate Professor Kaye Brubaker (CEE), director of the Maryland Water Resources Research Center as co-PI. The research team also includes ME doctoral student Nathan Boyd, industrial advisors from the District of Columbia Water and Sewer Authority (Matt Ries), the global engineering company Ramboll (George Rest, Tom Dumm), and the Tennessee Duck River Development Agency (Doug Murphy).

“To better manage the Anacostia and Duck Rivers, we will be developing game theory models of the users in their networks, such as water utilities, municipalities, residential consumers and dam operators,” says Professor Gabriel. “We anticipate the result will be improved water market designs that will incentivize better use of water resources.”

New management approaches are needed to improve cooperation among independent water resources users and stakeholders. This is because cooperation is not naturally incentivized—actions beneficial to upstream users often negatively impact users downstream. This makes it difficult to foster cooperation over water withdrawal rights, water quality responsibilities, and risks associated with flooding.

Historically, cooperative agreements among independent entities have required static legal agreements that create barriers to adaptation or policy improvement. This research will explore novel management approaches, such as market-based mechanisms, to overcome these cooperation challenges, based on game theoretic models they will develop and use. The researchers anticipate that the approaches’ efficiency and equity gains will benefit municipal, industrial, and agricultural water users; treatment plant and network operators; and the natural environment.

The Anacostia River testbed will focus on urban river restoration, while the Duck River testbed will emphasize economic development and ecological preservation.

Professors Gabriel and Brubaker will use rigorous mathematical techniques to model deterministic and stochastic water infrastructure systems from a one-level and two-level equilibrium problem perspective based on non-cooperative game theory. The developed models will combine engineering, water policy, machine learning, risk analysis, resilience planning and economic elements.

The project will account for risk and benefits in a systematic, unified, and endogenous manner across all entities and their interactions and will allow the system operator/regulator to effectively balance risk and cost under uncertain and/or changing conditions.

The researchers also anticipate the work will lead to algorithmic advances in decomposition methods for water equilibrium problems as well as stochastic equilibrium models for this general class of infrastructure equilibrium problems. A rolling-horizon, stochastic mathematical program with equilibrium constraints will develop strategic learning algorithms for water stakeholders to improve their decision-making over time.

Published April 28, 2021