Consensus Based Distributed Optimization For Energy Management Of Active Distribution Network

The grid technology has undergone through an enormous revolution over the past few years.This is due to the increased participation of distributed generation,renewable energy resources(i.e.solar,wind,biomass etc.)and energy storages to accommodate the load demands in the power grid.These all factors contribute to offer a more reliable,sustainable and clean power.However,these factors also create many challenges in energy management and made it more complex.The tradition power grid is based on a centralized approach in which all the information is gathered a single point with the help of a central controller and then optimization is performed,however,this approach increases the computational burden and makes it more prone to a single-point-failure.With the revolution of distributed generation,the future power grid will be more distributed and requires distributed optimization techniques to solve the energy management problems.Following this trend,a distributed consensus protocol is used in this thesis to solve the energy management optimization problem for Active Distribution Network(ADN).A 33-bus system distribution system equipped with distributed generators(DG),photovoltaic(PV),wind and battery storage are used as a base model for case study.The incremental price of electricity is used as a consensus variable,the designed distributed algorithm tries to find a single value of cost for all agents.The computed cost is then used to find the optimal power for each agent.In order to meet the total load demand in the network,local power mismatch at each bus is shared with only neighboring agents.This mismatch information acts as feedback variable and helps the agents to adjust their powers to meet the load demand economically.The designed model is as a social welfare maximization optimization problem and consensus protocol is used to find the best optimal solution of the problem.The designed algorithm is fully distributed and only incremental cost and local power mismatch is shared with the adjacent neighbors to satisfy the privacy concerns of the future power grid.To validate the effectiveness of the algorithm a full day profile of PVs,Winds and loads is used to solve optimization problem.Convergence analysis is performed for the peak load condition under normal operation.A very important feature i.e.plug-and-play of future grid is then opted to demonstrate the convergence and adaptiveness of the proposed algorithm.Furthermore,the load demand is varied during the optimization process to test the performance under varying load demand.A convergence speed analysis is also presented in the end to further demonstrate the effects of convergence factor on the convergence speed of the algorithm.