Stochastic Security—Constrained Generation Scheduling with Wind Power Generation based on Dynamic Line Rating

Abstract

The penetration of renewable energy resources is strongly on the rise worldwide. However, difficulty in accurately forecasting the power output, which is an unavoidable and inherent nature of renewable energy resources such as wind power, is expected to cause short-term scarcity events in the operation of power system. Consistently increased loads and limited transmission line resources also impose issues to power system operation, such as transmission congestion, re-dispatch of generators, and wind power spillage. Accordingly, operating reserve requirements must be revised in order to maintain a desired system reliability level in an uncertain operating environment, and the effective employment of network resources must be developed. This dissertation proposes day-ahead generation scheduling that utilizes demand-side resources (DSR) and wind power generation (WPG) itself as a reserve for managing increased uncertainty. When optimizing the scheduling process, dynamic line rating (DLR) is employed as a power flow limit for the transmission line rather than static line rating (SLR). However, when incorporating DLR into the generation scheduling problem, it should not be overlooked that new uncertain factors arise with regard to the determined DLR. Hence, a stochastic problem is suggested to properly manage the newly introduced uncertainty that originates from DLR. The uncertainty in load is expressed using a time series model after analyzing real data in-depth. In order to reflect the tendency of time sequential change in uncertainty, an autoregressive integrated moving average (ARIMA) model is adopted. After the diagnostic check, i.e., Q-Q plot, autocorrelation function (ACF), and partial autocorrelation function (PACF) of residuals, it is confirmed that the model fits the real data well enough. Wind speed forecast error is also designed as a time series model. Because this dissertation aims for security-constrained generation scheduling, equipment failures in power systems are also considered, from which the probability distribution model is determined based on a Markov chain. All the constraints and objective functions of the generation scheduling problem with DLR are converted into mixed integer linear forms in order to solve the optimization problem by applying mixed integer linear programming (MILP) generally adopted in practical system operation. Heat balance equations related to DLR are expressed as a linear form with reasonably acceptable assumptions. On the other hand, because the common simplified decoupled power flow model cannot be used for various reasons, in this study, relatively less approximated power flow equations are proposed for the generation scheduling problem. The optimization problem is designed by a two-stage decision model where the objective function is to minimize the expected operating cost that comprises the generation, reserve, DSR, and load-shedding costs. The problem is constrained by individual generation constraints (e.g., ramping up and down, minimum up and down time, and power generation limits) and by system constraints (e.g., system active/reactive power balance, reserve requirement, and transmission flow limit). The expected energy not supplied at each bus is also calculated to assess reliability. The performance of the proposed generation scheduling was verified using a six-bus system and modified IEEE 118-bus system integrated with wind power units. The simulation results clearly demonstrate that with DLR, a system operator (SO) can significantly reduce the expected operating cost. This advantage stems from the sufficient utilization of the existing transmission line, which leads to an optimal commitment of the generating units and dispatched volume, and efficient use of harvested wind energy. The expected energy that is not served also drastically decreases by means of incorporating DLR. In other words, the proposed method is more reliable in terms of reduced load-shedding amount. The simulation results also show that additional cost savings can be achieved with demand-side participation and deloaded control of the wind power generator. It can be interpreted that DSR and the reserve from WPG are clearly capable of managing the increased uncertainty due to wind power integration with the desired reliability. The effects are remarkable, especially at higher wind power penetration levels. The computational time, a crucial factor for short-term generation-scheduling tools, increases with multiple scenarios and newly inserted constraints, but it is not excessive owing to the reduced non-linearity of the mathematical formulation.