Dynamic Multimachine Modeling and Optimal Tuning of Automatic Generation Control

Abstract

This article describes the control scheme of an actual SCADA/EMS AGC (Automatic Generation Control) function and proposes a methodology for tuning the parameters of both generation units and AGC control systems. AGC tuning is split into two stages: unit control logic parameters and AGC system. In order to determine the optimal parameters, the application of a heuristic optimization algorithm, named MVMO (Mean-Variance Mapping Optimization), is proposed to solve a defined optimization problem. The proposed tuning methodology is validated using a realistic simulation environment named Operator Training Simulator (OTS). In addition, AGC system parameters are determined from real SCADA records and primary frequency response (PFR) obtained from frequency events recorded by the Ecuadorian WAMS (Wide Area Monitoring System). Finally, the AGC system is modeled in PowerFactory with the aim of achieving a realistic AGC model, considering a multimachine power system. For this aim, AGC control is implemented considering the realistic four-seconds sampling period related to a SCADA/EMS and the implementation of a filter that measures the rate of change of frequency (ROCOF) for allowing the primary frequency control to be previously performed. This model is then validated by simulations of generation outage events using a reduced model of the Ecuador-Colombia interconnected power system in PowerFactory.