Smart grids are seen as the solution to deal with future challenges in the eld of electrical generation, transport and supply. Environmental, economic and social issues have encouraged distributed generation and new control methods should be applied to ensure grid frequency deviation to be low enough in order to prevent potential damage to devices connected to the electrical grid. At the same time, supply cost and power should be controlled to avoid excess. The fact that the classical generation structure based on large power plants as the main supply sources for urban areas is changing to a set of smaller generation distributed points. To adapt to this change in terms of generation model, a distributed control is proposed in this study, where all areas have an in uence on neighbour areas. To do so, a hierarchical cooperative model predictive control is presented in this thesis. The idea of the study is creating a hierarchical system based by means of overlapping the original subsystem areas to create several control layers. Lower layers are controlled through load frequency control (LFC) for each layer and transmit the results to following layers beneath each one. In addition, the upper level symbolises a centralised optimization unit computing the reference values based on cost optimization. The fact that the control is distributed also deals with eventual communication blackout, due to the fact that other areas apply the control to the momentarily disconnected area. All these studies have been checked implementing a simulation program using Matlab/Simulink.