Please use this identifier to cite or link to this item:
Title: Integrated topological representation of multi-scale utility resource networks
Authors: Gilbert, Thomas Tavernor
Issue Date: 2022
Publisher: Newcastle University
Abstract: The growth of urban areas and their resource consumption presents a significant global challenge. Existing utility resource supply systems are unresponsive, unreliable and costly. There is a need to improve the configuration and management of the infrastructure networks that carry these resources from source to consumer and this is best performed through analysis of multi-scale, integrated digital representations. However, the real-world networks are represented across different datasets that are underpinned by different data standards, practices and assumptions, and are thus challenging to integrate. Existing integration methods focus predominantly on achieving maximum information retention through complex schema mappings and the development of new data standards, and there is strong emphasis on reconciling differences in geometries. However, network topology is of greatest importance for the analysis of utility networks and simulation of utility resource flows because it is a representation of functional connectivity, and the derivation of this topology does not require the preservation of full information detail. The most pressing challenge is asserting the connectivity between the datasets that each represent subnetworks of the entire end-to-end network system. This project presents an approach to integration that makes use of abstracted digital representations of electricity and water networks to infer inter-dataset network connectivity, exploring what can be achieved by exploiting commonalities between existing datasets and data standards to overcome their otherwise inhibiting disparities. The developed methods rely on the use of graph representations, heuristics and spatial inference, and the results are assessed using surveying techniques and statistical analysis of uncertainties. An algorithm developed for water networks was able to correctly infer a building connection that was absent from source datasets. The thesis concludes that several of the key use cases for integrated topological representation of utility networks are partially satisfied through the methods presented, but that some differences in data standardisation and best practice in the GIS and BIM domains prevent full automation. The common and unique identification of real-world objects, agreement on a shared concept vocabulary for the built environment, more accurate positioning of distribution assets, consistent use of (and improved best practice for) georeferencing of BIM models and a standardised numerical expression of data uncertainties are identified as points of development.
Description: PhD Thesis
Appears in Collections:School of Engineering

Files in This Item:
File Description SizeFormat 
Gilbert T 2022.pdf6.43 MBAdobe PDFView/Open
dspacelicence.pdf43.82 kBAdobe PDFView/Open

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.