Forecast-informed hydropower optimization at long and short-time scales for a multiple dam network

This study presents a scheme for co-optimizing the long-term (seasonal) reservoir operating objectives with the short-term (daily) objectives for multi-dam networks to maximize hydropower generation. Long-term optimal reservoir storage provides temporal space to optimize operation of the dams at short-term based on forecasted reservoir inflow. This study asks if there is an added benefit of co-optimization of operations at long- and short-term scales for hydropower generation. The multi-objective optimization problem at both the temporal scales was simultaneously solved considering Pareto optimality between conflicting objectives. Constraints pertaining to flood control, dam safety, and environmental flow were imposed. The proposed scheme was implemented over a network of Blue Mesa, Morrow Point, and Crystal dams in the Upper Colorado Basin. Ensemble forecast forcings from publicly available numerical weather prediction and climate models were used as inputs for the daily and monthly scale inflow forecasts. The results show improvements of 41%, 27%, and 15% in hydropower generation using the co-optimization during wet, moderate, and dry years, respectively, against a benchmark that neglects inflow forecasts. This study demonstrates added benefit of co-optimizing the operations for hydropower generation based on short- and long-term forecasted reservoir inflow. Given that most dams today operate as a network in a river basin, we recommend moving away from single dam and single time scale optimization to a multiple-dam with long- and short-term scale co-optimization-based operations to make renewable energy generation more efficient.