Stochastic community domestic energy modelling

Amin, Amin 2022. Stochastic community domestic energy modelling. PhD Thesis, Cardiff University. Item availability restricted.

	PDF (PhD Thesis) - Accepted Post-Print Version Restricted to Repository staff only until 2 May 2024 due to copyright restrictions. Download (28MB)
	PDF (Cardiff University Electronic Publication Form) - Supplemental Material Restricted to Repository staff only Download (336kB)

Abstract

The residential sector is a stochastic energy consumer due to the diversity and complexity of individual household energy demand, which is responsible for around 40% of global electricity consumption. An accurate and robust domestic model for estimating and evaluating energy loads, environmental impacts, and savings potential is crucial for supporting decision-making and formulation in many energy applications during their lifecycle stages. However, inherent energy modelling uncertainties are due to socio-technical attributions, local microclimate, and the usage of oversimplified methods. This research aims at developing and evaluating a comprehensive hybrid framework for estimating daily electricity demand in UK dwellings by considering key factors influencing household power loads, including local urban microclimate and dynamic and stochastic socio-technical factors. The framework is built on top of a combination of a machine learning model for generating a representative local weather dataset for energy simulations, two statistical-based models that behave as key components for stochastic and detailed forecasting of household behaviour patterns and electric load schedules, and a physics-based model to estimate the impacts of the thermal behaviours of electric appliances on the overall dwelling thermal performance and total electricity demand. The research provides several original contributions to knowledge that reside in the integration of geospatial actual observation interpolation and reanalysis data down�scaling approaches to provide local measures for twelve key variables required in energy simulations; the incorporation of random occupancy behaviour modelling into a statistical model to determine the electrical power usage within a household; the integration of three different model types to overcome energy modelling limitations and to provide robustness and simplicity in forecasting the household electricity loads; and the validation and calibration of the developed framework on multiple annual, daily, and sub-hourly basis. iii The implementation of the developed framework illustrates the framework’s effect�iveness in reflecting the daily temporal fluctuations of electrical power demand and peak loads with an accuracy of up to 70% and 90% on a sub-hourly and daily basis, respectively. The framework facilitates more reliable results with slighter demand peaks of up to 49% when compared to conventional energy simulation practises and around 15% and 10% less of the overall daily and annual electricity demand, respectively. The research findings demonstrate the capability of the developed framework for synthesising simulation scenarios and conditions that allows estimating a "baseline" electricity demand for the domestic sector, scaling the energy demand estimations from individual devices and dwellings to a national level, and assessing the techno�economic and environmental potential impacts of the implementation of renewable energy sources to reduce energy costs, mitigate peak power loads, and reduce greenhouse gas emissions.

Item Type:	Thesis (PhD)
Date Type:	Completion
Status:	Unpublished
Schools:	Engineering
Uncontrolled Keywords:	1). Domestic energy 2). Energy forecasting 3). Community electricity 4). Weather localisation 5)._Occupation pattern 6). Household electricity
Date of First Compliant Deposit:	2 May 2023
Last Modified:	03 May 2023 11:24
URI:	https://orca.cardiff.ac.uk/id/eprint/159133

Actions (repository staff only)

Edit Item