The electric grid is undertaking a change of paradigm. New requirements such as the high power loads of Electric Vehicles’ charging are putting pressure on grids, and the popularisation of the figure of the prosumers, are creating new needs on the management of the system. Although the investment on infrastructure is one of the ways of alleviating the arrival of this new loads, the existence of smart controls, actuators and smart appliances opens also the door for demand modification. The so called Demand Response Events are strategies to reallocate the loads of the consumers, so the power demand is shifted towards times less critical for the grid. Some smart appliances can be considered as shiftable. The operation of them can be rescheduled to other times and still provide their services. On this paper we suggest an optimisation methodology consisting on a parallelisation method together with a generic algorithm that would allow an overlooking controller (such as aggregator or DSO) reduce the power peaks even with a large number of devices to control. The research here presented has looked for the smallest group size at which the optimisation is more effectible done in parallel groups rather than on a single large set. With this, we eliminated the exponential growth of the computational effort that the optimisation algorithm would need to do organising all devices in one go. The method achieves a computational time reduction of 80%, and it is only worsening the reduction of peak by 2.8% compared to an optimisation with a complete set.