Smart residential energy scheduling utilizing two stage Mixed Integer Linear Programming

In this paper, we design and evaluate the feasibility of a system which minimizes residential electricity cost of individual homes by shifting demand over a daily forecast price cycle. Ideally, our system will accept use-time preferences from consumers and optimize their appliances' operation around those given patterns. However, using the system to recommend optimum use-times to consumers is also possible by accepting ideal use preferences from an external load manager and computing the cost savings of these preferences relative to the cost of the consumer's current use patterns. We implement the optimization problem in two stages by using Mixed Integer Linear Programming (MILP). In the first stage, we obtain the optimum scheduling for appliances to be connected to the outlet; the output of this stage only shows the hours that we can use the appliance and does not reflect the actual consumption hours of each appliance. In the second stage, we model the random behavior of users via Monte Carlo simulation by running the appliances within the times they receive power, specified in the first stage. Finally, we evaluate the effectiveness of the proposed model in terms of cost savings by considering three appliances and four pricing schemes.