Tales Pulinho Ramos

Parallel computing applied to the stochastic dynamic programming for long term operation planning of hydrothermal power systems

In this paper, parallel processing techniques are employed to improve the performance of the stochastic dynamic programming applied to the long term operation planning of electrical power system. The hydroelectric plants are grouped into energy equivalent reservoirs and the expected cost functions are modeled by a piecewise linear approximation, by means of the Convex Hull algorithm. In order to validate the proposed methodology, data from the Brazilian electrical power system is utilized.

Stochastic Dynamic Programming Applied to Hydrothermal Power Systems Operation Planning Based on the Convex Hull Algorithm

This paper presents a new approach for the expected cost-to-go functions modeling used in the stochastic dynamic programming (SDP) algorithm. The SDP technique is applied to the long-term operation planning of electrical power systems. Using state space discretization, the Convex Hull algorithm is used for constructing a series of hyperplanes that composes a convex set. These planes represent a piecewise linear approximation for the expected cost-to-go functions. The mean operational costs for using the proposed methodology were compared with those from the deterministic dual dynamic problem in a case study, considering a single inflow scenario. This sensitivity analysis shows the convergence of both methods and is used to determine the minimum discretization level. Additionally, the applicability of the proposed methodology for two hydroplants in a cascade is demonstrated. With proper adaptations, this work can be extended to a complete hydrothermal system.

Using stochastic dual dynamic programming and a periodic autoregressive model for wind-hydrothermal long-term planning

In this study, we use a stochastic representation of wind for medium/long-term planning problems that are associated with the operation of hydro-thermal systems. The stochastic dual dynamic programming (SDDP) technique is used in this study. Synthetic wind and hydrological scenarios are generated using a periodic autoregressive model (PAR (p)). This algorithm has wide applicability in countries with a predominantly hydroelectric energy matrix that is associated with high penetration of thermal and wind generation, as in the Brazilian power system. The nonlinearities of the hydraulic production function also was been taking into account. The developed technique can be applied due to the global expansion of power generation over the last two decades with the increasing integration of alternative sources.

Modelo para o despacho de usinas individualizadas no planejamento hidrotérmico de médio prazo baseado em PNL

This paper presents a method to the hydrothermal dispatch using optimization techniques based on non linear programming techniques. To do so, the expected cost-to-go functions from a long term operation plannning strategic decision model are used. This decision model is based on stochastic dual dynamic programming and energy equivalent reservoirs. The proposed method considers a set of historical water inflow scenarios to the hydroelectric reservoirs. Those scenarios are used to simulate the long term operation planning to a given horizon. The results obtained from this disaggregation model (MIUH) are compared with those from the model officially adopted in the Brazilian power system, SUISHI-O. The latter is based on operation heuristics aiming at operating the reservoir maintaining the water storag e in similar levels, that is, trying to operate them in parallel.

Incorporation of rationing effects to hydrothermal dispatch optimization

In Brazil, power supply safety criteria is established by the sector regulation, the power supply insufficiency risk must not exceed 5% limit. Thus, the tools that enable agents to model the rationing more accurately became necessary, estimating the impact caused by such kind of event and the future consequences. The official model used by Brazilian electric sector to carry out the expansion planning and medium/long-term operation is based on Stochastic Dual Dynamic Programming technique (SDDP). The main results that this model presents are the monthly evolution projection of Short Run Marginal Cost (SRMC) and hydrothermal dispatch, considering various hydrological scenarios. However, such model does not contain the representation of conjuncture changes occurred during a rationing, such as, for example, demand retraction and thermal dispatch out of merit order. This paper presents a new alternative to estimate rationing effects on power operation planning.

Stochastic dynamic programming with discretization of energy interchange between hydrothermal systems in the operation planning problem

This work presents an alternative strategy to solve hydrothermal systems operation planning by stochastic dynamic programming. Under the presented approach, the hydroelectric plants are grouped into equivalent subsystems of energy and the expected cost functions are modeled by a piecewise linear approximation, by means of the convex hull algorithm. Also, under this methodology, the problem is solved independently for each subsystem such that the state variables to be considered are the energy storage and energy net interchange of the subsystem. The presented results have shown that this subsystems separation technique reduces significantly the computation time when compared with the traditional techniques of stochastic dynamic programming, demonstrating the effectiveness of the proposed methodology.

Generation of multivariate monthly synthetic water streamflows through multiplicative parma model

This paper describes the methodology applied to develop an object oriented computer model for multivariate monthly synthetic water streamflows generation through multiplicative PARMA model (1.0)times(1.0)12. This model assumes that the streamflow of a certain month depends explicitly on the streamflow of the prior month, the streamflow of the same month in the previous year and in the preceding year, as well as on a random noise. The stochastic model tries to guarantee that the historical serial correlations of unitary lag from the annual aggregated streamflows are properly reproduced, as well as the other annual, periodic and statistical parameters.