Abstract
The residential sector represents approximately 30% of global electricity consumption, but the
underlying drivers are still poorly understood. The drivers are many, varied, and complex,
including local climate, household demographics, household behaviour, building stock and the
type and number of appliances. There is considerable variation across households and, until
recently, often a lack of good data.
This thesis draws upon a detailed household dataset from the Australian Smart Grid Smart City
project to build a residential demand modelling tool set. This data covers a part of greater
Sydney. Two statistical models for household annual electricity demand and half-hourly peak
electricity demand were established and tested for both individual households and regional
aggregations of households. The model showed only reasonable performance in forecasting the
consumption of individual households, highlighting the influence of factors beyond those
surveyed. However, the model demonstrated good performance for aggregated household
consumption: 3.9% MAPE for annual electricity consumption forecast and 4.57% MAPE for peak
demand forecast. Models such as this would be highly useful for a range of stakeholders,
including individual households, trying to understand the potential implications of different
choices and utilities looking to better forecast the impact of different possible residential trends.
The model would also be very helpful to grid operators seeking better reliability while avoiding
augmentation and to policy makers seeking to improve householder’s energy efficiency through
targeted policies and programs. Based on the developed tool set, models were built to simulate
various strategies for annual and peak demand reduction, and socio-economic evaluations were
calculated and compared between different reduction options. Results showed that
behavioural and demand response interventions were found to provide the most cost effective
peak reduction. The results were scaled up to the Sydney geographical region to provide realistic
recommendations for policy makers, utility operators and other stakeholders. In addition,
annual demand reduction intervention using feedback systems were investigated. Results
showed that feedback interventions have different effectiveness on households with different
characteristics. The statistically significant findings directly support the fact that demand
reduction intervention should be tailored to match specific household types to achieve optimum
and cost effective outcomes.