Benchmarking optimal utilisation of residential distributed generation with load control

In this paper we develop a mathematical model of the power usage of a community of 7 typical New Zealand houses with realistic demand profiles. Using mixed-integer programming (MIP) we apply load shifting and energy management techniques to minimise the overall cost of power. The loads are controlled based on their energy and state of charge, and a meaningful representation of energy has been ascribed to loads that do not intuitively have `energy'. It is proposed that modeling the loads in this fashion can provide guarantees to the consumer that their consumption habits will be uninfluenced by load control. Within these energy bounds, the MIP finds the optimal allocation of a set of loads for two scenarios: maximising utilization of a local wind resource in a fixed electricity price environment, and in an hourly time-of-use (ToU) price environment. The MIP provides two key insights useful in the development of real-time load control methodologies: firstly it highlights the optimal utilization of wind for certain time periods in a given day; and secondly it quantifies the maximum value that can actually be extracted by time-shifting all of the loads within a specific system. The MIP henceforth serves as a benchmark against which near-optimal load control algorithms can be compared in simulation and validated in practice.