Increasing amount of renewables in the medium and low-voltage distribution systems is causing challenges to realize a stable and reliable operation of the grid. Especially, the power production from renewable power sources such as solar and wind may cause reverse power flow and voltage rise in the medium-voltage distribution network. Without proper control measures, due to fluctuating renewable power production, power flows in some parts of the distribution network may exceed the thermal and voltage limits and cause overloading of the network assets. These conditions are called network congestions in this paper. The distribution system operators in coordination with the transmission system operator has to manage the power flow to avoid such congestions. This work addresses two aspects to manage network congestions: (i) observance of the network by means of load/generation forecasting and dynamic state estimation, (ii) optimal control of the flexible network assets. These are linked to the electricity market to utilize the ancillary services offered by the flexible resources in the network. These services could be purchased by the distribution system operator during network congestions. The proposed observability module gets measurements from a minimum set of network nodes and ensures full observability of the network. A coordinated control algorithm is proposed to compute the aggregated flexibility at each MV bus, which can be used by the area controllers at each MV bus to control the individual flexible resources connected to the MV bus or distributed at the LV network. These modules get inputs from the state estimation and forecasting algorithms. The objective of the proposed control is to accommodate maximum power from renewable energy sources and maintain the power flows within the thermal and voltage limits of the feeder. A model predictive control algorithm based on conic programming with quadratic constraints is used to achieve the above objective. The proposed algorithm is applied and tested in a simulation platform using DigSilent Power factory and Matlab software using a model of a real-life Danish 10 kV medium-voltage distribution network. The results obtained are therefore directly applicable to Danish distribution network grids and may be feasible in other countries with similar grid structures.
- Active distribution network
- Load forecasting
- network congestions
- Model Predictive Control (MPC)