The challenge of reservoir management is to manage the reservoir outflow such that various objectives are met as good as possible. Some of these objectives are in line with each other, while others are conflicting. For most reservoirs these objectives include:

• water supply for domestic, industrial and agricultural use,
• hydro power production,
• flood control,
• recreation,
• environmental obligations,
• minimum flow in the downstream river for cargo ship navigation, to support downstream water extractions and/or to ensure a certain level water quality.

Hydrological conditions and water demand vary over time, making operational reservoir management a complex task. This becomes even more complex if a system consists of multiple reservoirs, for example in a cascade.
 

Mathematical optimization is a powerful approach to support operational reservoir management: with the help of optimization techniques it is possible to adapt the management to forecasted events like floods and droughts, and future water demand.
For operational reservoir management it is often important to account for

• minimum and maximum operational limits, e. g. maximum discharge at a certain location, minimum operational water levels to account for flood protection or environmental obligations;
• target values, e. g. a target forebay elevation in a reservoir on a specific day, or a specific hydropower production in order to obtain load balance;
• economic objectives like maximum hydropower output under given operational goals.

Optimization can also be a powerful tool for long-term strategic planning, for feasibility studies, design studies, climate change adaptation studies, and climate change stress tests. How large must a reservoir be in order to bridge the largest expected drought period, how resilient is the reservoir system with respect to climate change, and how are costs and potential benefits balanced?
Strategic planning studies often aim to optimize for

• performance indicators that address supply security for water supply, power, and environmental issues;
• cumulated annual revenue for hydro power production or water sales.

RTC-Tools supports a very flexible set-up of the optimization problem to account for all the different aspects of reservoir management on different time scales.
 

The increased production of electric energy from wind and solar sources results in higher fluctuation of power supply over time. Complementary energy sources are necessary to balance the energy demand and supply for the sake of a stable electric power grid operation. Hydropower production from reservoirs can play an important role as complementary source. RTC-Tools can support the reservoir operators to optimally respond to anticipated fluctuations in electricity supply, under consideration of other water use practices.

Pump storage operations can be included as well. Essential for hydro power operational management with RTC-Tools is that the model is supplied with hydrological forecasts of water availability and forecast of water and energy demands from the system. For this reason, RTC-Tools is often embedded in operational systems based on Delft-FEWS.
 

RTC-Tools is able to use ensemble forecasts for robust predictions. Weather services often provide meteorological forecasts as ensembles, with each ensemble member having the same probability. With hydrological models, these ensembles can be translated to ensemble streamflow prediction for inflows into the reservoir system. With RTC-Tools, the optimal reservoir management for each ensemble member can be computed individually, such that results can be used to objectively assess prediction uncertainties.

The implementation of the scenario tree reduction methodology is an ongoing development. In order to provide effective decision support for reservoir operations with ensemble forecast, ensemble members with redundant information will be aggregated. Branching points are introduced at time steps when information becomes available in the forecast horizon, for example, when a forecasted precipitation event is observed. With the tree reduction ensembles can not only be used to quantify uncertainty, but, due to lower computation time, to effectively support decisions about operational choices.