Geospatial Inference and Management of Utility Infrastructure Networks

Abstract

Modern cities consist of spatially and temporally complex networks that connect urban infrastructure assets to the buildings they service. Critical infrastructure networks include transport, electricity, water supply, waste water and gas, all of which play a key role in the functioning of modern cities. Understanding network spatial connectivity, resource flow, dependencies and interdependencies is essential for infrastructure planning, management, and assessment of system robustness and resilience. However, there is a sparsity of fine spatial scale data from which such understanding can be derived or inferred. Often data is held within commercially sensitive organisations and may be incomplete topologically and/or spatially. Thus, there is an urgent need to develop new approaches to the integrated inference, management and analysis of the complex utility infrastructure networks. Such approaches should allow the highly granular representation of utility network connectivity to be represented in a spatially explicit manner, employing methods of data and information management to ensure they are scalable and generic. This thesis presents the development of such an approach, one that employs a geospatial ontology to formally define the key entities, attributes and relationships of fine spatial scale utility infrastructure networks. This ontology is used as the conceptual framework for the development of a suite of algorithms that allow the heuristic inference of the spatial layout of utility infrastructure networks for any urban conurbation within the UK. This is demonstrated via several case studies where the electricity feeder network between substations and buildings is generated for several different cities within the UK. Validation against the known network for the city of Newcastle upon Tyne indicates that the network can be inferred to high levels of accuracy (about 90%). Moreover, the algorithm is shown to be a transferable to the inference and integration of other utility infrastructure networks (gas, water supply, waste water, and new road layouts). ii The representation, management and analysis of such spatially complex and large utility networks is, however, a major challenge. The efficient storage, management and analysis of such spatial networks is explored via a comparison of a traditional RDMS approach (PgRouting within Postgres), spatial database (PostGIS) and a NoSQL graph-database (Neo4j), as well as a bespoke hybrid spatial-graph framework (combination of PostGIS and Neo4j). A suite of comparison tests of data writing, data reading and complex network analysis demonstrated that significant performance benefits in the use of the NoSQL graph database approach for data read (around 210% faster) and network analysis (between 420 and 1170 % faster). However, this was at the expenses of data writing which was found to be between 135 and 150% slower.