Optimal Control Actions in an Electrical Grid with Variable Renewable Energy Sources: Evaluation of the ANM4L Control Algorithms

Abstract: The increasing amount of renewable energy in the electrical grid comes with several challenges, of which overvoltage and congestion (overcurrents) are the subjects of this work. One flexible way of handling these problems is active management by some control system. There is a trade-off between simplicity when implementing the control system and the performance. In the project Active Network Management for All (ANM4L), an algorithm using PI controllers for management of voltage has been developed. It can control active power and reactive power. To make it easy to implement, PI controllers at different buses cannot communicate. This work is a comparison between control actions acquired from optimization in the program GAMS and algorithm control actions to see how far from optimal the latter ones are. In a five-bus low voltage test network with a 30 kW photovoltaic generator at each bus, the least curtailment of active power compared to the optimal actions was obtained when the voltage management algorithm controlled curtailment of active power and decided the reactive power based on a constant power factor of 0.8. This simulation represented a case of today. When the generation was increased to 50 kW per bus, representing a distant future scenario, the least curtailment of active power compared to the optimal actions was obtained when the active and reactive power were independently controlled. Control of active and reactive power independently could, in this respect, be regarded as the best choice in the long term. However, the consumption of reactive power is not entirely independent as the inverter of a photovoltaic generator poses a limit on the power flow from and to that generator. Thus, if the consumption of reactive power cannot be increased further because of this limit, active power has to be curtailed. If a fair distribution of the control actions is prioritized, using a constant power factor of 0.8 is best in both the case of today and the future scenario. There is also an ANM4L algorithm for congestion management. Congestion management demands some communication between buses but the aim has, analogous to the approach in the voltage management, been to make it easy to implement. In a medium voltage test network, the CIGRE medium voltage distribution network benchmark, the congestion management algorithm came very close to the optimal actions when being set on prioritizing equal sharing between the buses.