Risk CDT - Assessing resilience of smart critical infrastructures to deal with emerging risks and threats

Infrastructures, systems and networks are critical components of our moder society. In order to safeguard the future prosperity of our society it is necessesary to design such systems with an ability to anticipate, adapt and respond to changes, natural or man-made, short or long term, local or global.

Resilience can be defined as the ability of a system to quickly recover from unexpected hazardous situations and learn from the past to minimize the effect of similar events in the future. It is an important attribute for modern critical infrastructures but unfortunately still is a poorly defined concept and new research is needed to  “measure resilience”  and achieve better resilience against threats in an integrated manner including natural and human threats/events (e.g. due to human errors or terrorist/criminal attacks). An example of interconnected networks are systems in which electric power, heat, water and transportation optimally interact at various levels and achieve a high resilience level. To do so, robust structures have to be designed and optimized and better-informed operations and management to be considered. The networks need to have the ability to learn from the past, thus, a certain degree of intelligence has to be shown. The system should be capable of mining the available data and extracting the needed information. These features can be achieved using modern Artificial Intelligence and Machine Learning techniques.

However, to properly analyse the resilience of multi-energy-networks many issues have to be tackled including the capacity to effiiciently model very large system size, complex topological structure and highly non-linear behaviours. Moreover, lack of knowledge due to increasing share of environmental-friendly technologies (e.g. renewable heat, renewable electricity, electric transportation), stochastic and variable external conditions (e.g. extreme weather scenarios) and inconsistencies in the data and measurements are relevant sources of uncertainty and, if not accounted, have the potential to undermine the overall system resilience.

The overall goal of this research project is to improve current approaches by providing an innovative and holistic methodology for assessing resilience of critical infrastructure. For instance, the project will define a comprehensive metric to score the system resilience based on quantitative analysis. Its definition is itself a challenging issue with a non-straightforward answer. Resilience metrics should be able to capture relevant resilience features, for instance, the system restoration speed, its structural robustness, its reliability performance and the system capacity to learn from the past accidents. Furthermore, resilience metrics have to be robust and account for the inevitable uncertainties affecting the system assessment. Traditional stochastic, but also advanced probabilistic approaches, can provide a robust mathematical foundation which is necessary to quantify relevant sources of uncertainty. Once a resilience metric for multi-energy-networks is defined, resilience enhancement strategies can be investigated which will likely result in superior solutions when compared to analysis of systems neglecting interactions with other networks. Indeed, an economic dimension is one of the most relevant in realistic applications and the proposed methods can be used to provide support to answer cost-resilience questions. For instance, consider investment in renewable heat sources, renewable electric generators and electric vehicles charging stations (either at regional level or city neighbourhood level). A combined analysis of power-heat-transportation networks will provide better investment solutions (in terms of cost minimization and resilience maximisation) compared to the analysis of for instance power network on its own. To tackle issues related to the networks sizes and structure, efficient graph-theory approaches can be exploited and combined to innovative reliability analysis techniques (e.g. system survival signature).

This project will require the capability to work with colleagues from different disciplines (e.g. Engineering, Mathematics, Electrical Engineering and Computer Science). In addition, a solid mathematical background and good scientific programming skills (e.g. Python, Matlab) are required.

The student will be based at the EPSRC ESRC Centre for Doctoral Training for the Quantification of Risk and Uncertainty in Complex Systems and Environments. First year (fully funded) will be spent by the student doing an MRes in “decision making under uncertainty” to acquire grounded knowledge about existing methods to deal with risk and uncertainty.



Centre for Doctoral Training in Quantification and Management of Risk & Uncertainty in Complex Systems & Environments is funded by EPSRC and ESRC.