The Fair Allocation of Power to Air Conditioners on a Smart Grid

Abstract

Our objective is to allocate power to air conditioners fairly when demand exceeds supply or when the load in regions of the network exceeds the capacity of the links. This is an alternative to the current practice of brown outs or black outs and the proposed pricing mechanisms. We hypothesis a smart grid in which the power company can communicate with the customers, measure the temperature and power consumption of each residence, and remotely set each thermostat. We develop simple techniques to predict changes in power consumption for modest changes in the setting of a thermostat, algorithms to apply two fairness metrics used in communications networks to power allocation, and an algorithm to assign power to different parts of a network on a capacity constrained local distribution network. We also propose a practical procedure to apply these algorithms.