Performance analysis of a distributed algorithm for dynamic reactive power compensation

We consider the problem of dynamic reactive power compensation in power distribution networks populated by a large number of microgeneration devices. We model the power losses minimization problem in the case of stochastic, time-varying reactive power demands. For this control problem, we propose a randomized iterative optimization algorithm, and we define a suitable metric for evaluating this algorithm´s (and other algorithms´) performance. For this performance index, we derive an upper that makes the optimal tuning of the algorithm parameters possible. Numerical simulations confirm that, via proper design of the algorithm, it is possible to achieve remarkable performance with limited but strategical communication among the agents.