Parallelized Bi-level optimization model with continuous search domain for selection of run-of-river hydropower projects

A comprehensive optimization model was developed to identify the optimal set of non-storage based small hydropower projects (SHP) along a given river reach. The model was based on a greedy genetic algorithm that maximizes net annual benefit. For each project, the model determines intake location, penstock diameter and length, and turbine number and capacity. The novel features in the model include: 1) bi-level optimization that produces robust results, 2) improved search for SHP intake locations, 3) inclusion of turbine number as an explicit optimization variable, and 4) nested optimization modules and parallelized execution. Model performance was assessed through application to the Mamquam River in Canada and the Guder River in Ethiopia. For the Mamquam River, the model identified seven feasible projects. The model provided suggestions for further improvement of one of the two existing SHPs on the river through shifting the intake 1.2 km upstream of the existing intake to provide a greater net benefit. No SHPs have been constructed on the Guder River yet; however, the model suggested 19 potential SHPs that could provide 62% higher annual net benefit compared to the results of a previous model recently applied to the same study area. (c) 2020 Elsevier Ltd. All rights reserved.

Automated summary

The optimization model developed in this study can identify the optimal set of non-storage based small hydropower projects along a given river reach. The model, with novel features, was successfully applied and proven effective through case studies.