The main objective of the operation planning in a hydrothermal system is to compute a so-called operation policy, that estimates the water values in the reservoirs and allows to determine monthly generation targets for each plant of the system which meets the energy demand and at the same time minimizes the expected operation costs throughout the planned period, also considering a risk aversion criterion. This cost is composed by the variable fuel cost of thermoelectric plants and the cost assigned to power supply deficits, represented by an energy deficit penalization function. This creates a link between current operation decisions and their future consequences. As it is impossible to accurately predict the inflows, the problem is essentially stochastic. The existence of multiple interconnected reservoirs, the transmission restrictions and the need to perform a multi-period optimization turns the hydrothermal planning problem a very complex task. For this reason, the solution is obtained in stages, when models with different detail levels are used to represent the system, covering periods of study with different horizons, namely medium-term, short-term and daily operation scheduling.
In this line of research, Cepel develops methodologies and software programs for:
(i) Long and medium-term operation planning –
In long-term horizons (10-year expansion planning) and medium-term (5 years ahead)there is a monthly discretization and hydropower plants are usually represented individually during the first years and in Equivalent Energy Reservoirs (REEs) in the remaining years. The optimization technique used is Stochastic Dynamic Programming in a Benders decomposition scheme. It supplies information to the short-term model through cost-to-go functions. Given the large size of the optimization problem to be solved, parallel/distributed processing techniques are also employed. Besides determining the dispatch of the power generation plants, probabilistic indicators of system performance are also obtained, as well as empirical distribution of probabilities of several magnitudes, including the system marginal costs.
(ii) Simulation of power plant operation through interconnected hydrothermal systems – In interconnected hydrothermal systems, as is the case of Brazil, several studies to support decision making in both, operation and expansion planning require detailed monthly simulation of the individual power plants. The inflows to the reservoirs can be either represented by historical series or by multivariate synthetic series generated by stochastic models, and alternative scenarios for multiples uses of water can also be considered. System operation can be simulated in a static way, where a fixed hydrothermal configuration is considered, or the problem can dynamically vary along time. The firm energy of the hydropower system - which is the largest energy market that the system may meet without energy deficits considering the occurrence of the inflow historical series - can be calculated, as well as the assured energy of an interconnected hydrothermal system to a given risk level. The simulation of hydrothermal system operation, for each month and hydrologic scenario, is divided into two stages: hydrothermal balance optimization between subsystems and simulation of the operation of individual hydroelectric plants. In the first stage the following values are decided solving an optimization problem: controllable hydraulic generation, thermal generation and energy interchanges for each subsystem which minimize the sum of current cost plus the currently expected value for operational future cost, the latter obtained from the model for long and medium-term energy operation planning. In the second stage, the final values of turbined outflow and storage of each power plant are obtained through heuristic rules so as to meet the hydraulic generation targets calculated in the first stage. Given the size of the optimization problem to be solved, parallel / distributed processing techniques are employed.
Contact the responsible area via email: