Optimization of Hydrologically Linked Hydropower Systems with Multiple Owners through Water Payments

Open Access
Beevers, David Bryan
Graduate Program:
Mechanical Engineering
Doctor of Philosophy
Document Type:
Date of Defense:
July 29, 2015
Committee Members:
  • Horacio Perez Blanco, Dissertation Advisor
  • Horacio Perez Blanco, Committee Chair
  • Hosam Kadry Fathy, Committee Member
  • Anil Kamalakant Kulkarni, Committee Member
  • Vinayak V Shanbhag, Committee Member
  • Seth Adam Blumsack, Committee Member
  • hydropower
  • optimization
  • day-ahead
  • operations
  • scheduling
  • dynamic programming
Before the advent of deregulated electricity markets in the 1990’s, hydroelectric power production schedules were typically coordinated with the schedules of other generating technologies by utility companies to meet expected regional demand. However, the growing adoption of deregulated markets requires the development of new methods for determining optimal utilization of hydroelectric resources. Much work has been performed regarding the optimization of cascaded hydropower systems owned and operated by single entities. However, these works fail to address the complexity involved in optimizing a cascaded system with owners along the same river. In those instances, each plant is limited to operate with whatever water flow the upstream plants choose to make available. In this situation, system optimal operation results in higher net system revenue than individual optimization of each plant with whatever water flow is available. However, such an optimal system solution must also result in lower net revenue for some owners compared to the value they would obtain through individual optimization. Hence, a novel methodology is proposed in which individual optimization naturally leads to system optimization through the use of iteratively determined water payments to direct the release of water toward a more optimal utilization. The proposed method alternates between dynamic programming for individual optimization and gradient analysis to determine and update hourly, location-specific water values for modified release. The sequence is repeated until there are no further additions or modifications to the payment schedule that results in increased total revenue. The water values are based upon local and downstream revenue gradients with respect to the flow rate released by the upstream facility during each time period. The method is applied in this thesis to a hydropower system consisting of three connected hydropower facilities, successfully increasing the total gross revenue of each installation. Although the algorithm it is not yet able to achieve the globally optimal solution, it comes quite close and offers unique insights into the value of water.