Genetic Algorithm-based Smart Appliance Scheduling for Energy-cost and Peak-demand Reduction: A Five-household Simulation Case Study
Gafar Abiola Adepoju
*
Department of Electronic and Electrical Engineering, Faculty of Engineering and Technology, Ladoke Akintola University of Technology, Ogbomoso, Nigeria.
Olatoun Oluwatoyin Subair
Department of Electronic and Electrical Engineering, Faculty of Engineering and Technology, Ladoke Akintola University of Technology, Ogbomoso, Nigeria.
Samuel Okeolu Omogoye
Department of Electrical and Electronic Engineering, Lagos State University of Science and Technology, Ikorodu, Nigeria.
*Author to whom correspondence should be addressed.
Abstract
This study presents a Genetic Algorithm (GA)-based multi-objective framework for scheduling residential appliances to reduce electricity cost, peak-to-average ratio (PAR), and modelled user waiting time. The framework was implemented in MATLAB R2022b and evaluated in two simulation cases: a single household with 16 representative appliances and an aggregate case comprising five households and 50 appliances associated with the Monatan 11-kV feeder area of the Ibadan Electricity Distribution Company, Nigeria. A weighted objective function combined normalised electricity billing, PAR, and waiting-time ratio, subject to appliance operating windows, required durations, continuity requirements, and an aggregate power limit. Because an operational Nigerian time-of-use tariff was not available for the study, a representative peak, shoulder, and off-peak tariff structure was used; the economic outcomes therefore describe the adopted simulation scenario rather than verified customer savings. In the single-household case, daily cost decreased from USD 44.74 to USD 42.13 (5.83%), while PAR decreased from 2.00 to 1.60 (20.00%). In the five-household aggregate, daily cost decreased from USD 265.39 to USD 255.20 (3.84%), peak demand decreased from 26 to 10 kW (61.54%), PAR decreased from 6.55 to 2.58 (60.61%), and the reported average waiting time decreased from 4.5 to 1.2 h. At an exchange rate of USD 1 = NGN 1,628, the aggregate daily saving corresponds to approximately USD 917.10 (NGN 1,493,039) over 90 days. The results indicate that appliance rescheduling can flatten the simulated aggregate load profile while preserving stated operating constraints. However, the limited household sample, assumed tariff, single reported optimisation run, incomplete raw-data disclosure, and absence of feeder power-flow or field validation restrict generalisation. The framework should therefore be interpreted as a case-study simulation that supports further benchmarked and pilot-scale evaluation.
Keywords: Appliance scheduling, demand-side management, genetic algorithm, home energy management, peak-to-average ratio, residential demand response