Using mixed integer programming for the volt/var optimization in distribution feeders

One of the functions that characterize modern management systems of electric power distribution networks is the periodical short-term optimization of the operating conditions. Such a function is typically designated as volt/var optimization (VVO). The usual objective is the minimization of network loss or demand. The main constraints are the maximum current values in lines or transformers and a few percentage point deviation of bus voltages from the rated value. VVO exploits the availability of two-way communication and the possibility to control transformer load tap changers, switchable capacitor banks, and reactive outputs of specific embedded generators (being active outputs often fixed by market decisions or by energy resource characteristics). The use of mixed integer linear programming (MILP) appears to have been less explored than other approaches for the solution of VVO problems. This paper presents a MILP model that includes the approximate representation of the main characteristics and constraints of short-term distribution system operation. The quality of the results obtained for different test feeders in various operating conditions and the corresponding performances of the solver appear promising for online applications.