Browse by author
Lookup NU author(s): Dr Gaihua Fu, David Alderson, Professor Richard DawsonORCiD, Professor Stuart Barr
Full text for this publication is not currently held within this repository. Alternative links are provided below where available.
Critical infrastructure systems, such as communication and energy systems, are important to our society. Their reliability and integrity is vital for ensuring national security, public health and economic growth. Previous research has largely focused on analysing single, isolated systems and has tended to ignore the fact that modern infrastructure systems have evolved into a network of networks (or a system of systems) [2, 4]. Furthermore, analysis has traditionally tended to focus on relatively simple, spatially constrained representations of systems and their interdependencies [1]. This results in network models, analytical methods and modelling outputs being often far removed from those required for real-world spatially complex interdependent infrastructure systems. With regards to interdependencies in particular, it is widely recognised that as their complexity increases so does the level of risk to infrastructure systems [5]. If we wish to understand the risk faced by complex interdependent infrastructure systems and develop appropriate mitigation and adoption policies, we require significantly improved analysis and modelling of interdependent networks that are spatially explicit and represented at the regional to national scale. In order to address the challenge above, the Resilient Futures project (http://r-futures.ecs.soton.ac.uk) and the Infrastructure Transitions Research Consortium (http://www.itrc.org.uk) have developed a suite of tools to represent and analyse infrastructure networks and their associated interdependencies. The aim of this research is to facilitate system-scale understanding of infrastructure networks and the impact of interdependencies upon them. This understanding could be used to improve the design of infrastructure that is more resilient and adaptable to future socio-economic and climate change. In this paper, we report two aspects of the infrastructure network modelling work conducted to date. 1. Using open source, spatially-enabled relational database software, PostgreSQL with PostGIS, we have constructed a comprehensive data set of national infrastructure assets in the UK. For the first time, we have established and connected data layers relating to the energy, water, waste water, solid waste and transport networks. The construction of the database using publicly available data sources and project partner data contributions has resulted in both the location and varied attribution of infrastructure network components being recorded. We engage with our stakeholder community to contribute data directly to the database, as this offers has, and will lead to the most detailed inventory of national infrastructure assets in the UK. The ability to model and subsequently analyse these networks and their interdependencies is underpinned by the availability of adequate data. To facilitate any analysis, we have also developed a bespoke database schema within PostgreSQL for network and interdependency storage. The schema offers the ability to store data that can be read in to our network analysis package. 2. We have developed a modelling framework for exploring cascading failure of interdependent networks [3]. Here, we first establish a number of isolated networks, each representing an infrastructure system exhibiting some topology. Interdependencies between networks are represented by a number of edges, each connecting a node in one network with a node in another. We model an attack on one network that disables some proportion of the network nodes directly and indirectly brings about additional node failures as a consequence of compromised interdependencies. We show that the post-attack performance of both networks is mediated by (i) network topology (e.g., lattice, small world, centralised, decentralised), (ii) the nature and extent of network interdependency (e.g., directed versus undirected dependencies, their density and correlation structure), (iii) the type of attack (i.e., random, targeted, spreading, or spatial), and (iv) the rule for establishing post-attack viability (e.g., a network node may only be viable if it is part of the network’s largest component, or, alternatively, an entire network component might be viable only if it is above some critical size, or if it is sufficiently connected to a viable component of the other interdependent network, etc.). We demonstrate that disruption in interdependent networks can be disproportionate to attack size, and that the magnitude of cascading failure can be significantly increased when interdependencies are sub-optimal. The approach described here has value for infrastructure stakeholders, providing hitherto unavailable analysis of how to: 1) maximise the reliability of interconnected infrastructures subject to disruption; 2) adapt existing infrastructure systems to meet the challenges imposed by natural and malicious threats and hazards. Ongoing work is extending this analysis to consider issues around capacity, lag and latency in network connections, and systems with more than two networks. A case study using the UK electricity transmission grid and railway network is also underway.
Author(s): Fu G, Khoury M, Alderson D, Dawson R, Bullock S, Barr S
Publication type: Conference Proceedings (inc. Abstract)
Publication status: Unknown
Conference Name: Network of Networks: Systemic Risk and Infrastructural Interdependencies
Year of Conference: 2012
URL: https://sites.google.com/site/netonets2012/
Notes: https://sites.google.com/site/netonets2012/home/presentations/NON2-Fu.pdf?attredirects=0&d=1
