Coordinated voltage control via distributed model predictive control

Over the past decades many voltage collapse incidents have occurred, caused by uncoordinated interactions of local controllers, following a major disturbance in electric power system operating closer and closer to their safety limits. Currently most voltage control schemes are rule-based and only rely on local measurements. However the availability of wide-area phasor measurements (WAMS/PMU) suggests that the risk of voltage collapse could be reduced by coordinating control actions in neighboring control regions of the network. This paper shows how distributed model predictive control (MPC) can be used in order to design such a coordinating controller. Local control agents carry out an on-line optimization by comparing the plant behavior over a finite window in time, for different possible switching sequences of the local tap changing transformer (LTC). The evolution of the plant behavior is obtained by a fast Modelica simulation of the hybrid systems model, using information on the tap switching sequences that neighboring control regions are planning to implement on their LTCs. Through simulation of a 12-bus power system this paper shows that the distributed MPC controllers can prevent, or at least postpone, voltage collapse in circumstances where classical uncoordinated controllers fail. The required communication exchange is very limited, making practical applications feasible. However extensions to larger systems will require consideration of modeling abstractions of neighboring control regions, so as to keep the simulation time for each local decision maker short.