Understanding, Prediction, Mitigation of Cascading Failures in Interdisciplinary Complex Systems

In our society, critical services are underpinned by increasingly complex networks, some of them highly vulnerable to natural hazards and malevolent attackers. Reducing the probability and consequences of blackouts and the cascading effects is critical to improving systemic resiliences. However, the paucity of data in this area calls for a different approach where research expertise from other fields can be drawn upon and transferred to meet our energy security needs. This paper analyses cascading models that have a wide applications in many different disciplines (complex networks) based mainly on the interdependent network theory. It gives an overview of the current techniques, challenges and emerging methods of analysis and simulation. The nature of the models discussed indicates the potential applicability of large integrated financial, biological, social and transport systems. Such examples are the electrical power markets and grids, urban regional transportation systems, gene regulatory networks, infectious diseases, financial networks. Through this research, we created a forum where researchers can engage together in examining the challenges of reliability faced by operational systems in various fields, and on various scales. We identified analogies between different disciplines, mainly power engineering, medicine, biology, finance. Cascading failures may happen in various types of systems. Examples of such systems are: power transmission - when one of the elements fails and shifts its load to nearby elements in the system, medicine - when heart’s pumping function is not working adequately leads to a cascade of a number of symptoms and poor overall body function, finance - when the failure of one financial institution impacts on others causing a domino effect, biology - where a small reaction can impact on the overall system; transport - when the failure of one segment has an impact on the traffic within the broader transport network , stock market, etc. In any of these systems, to minimize disruptions, it is necessary to effectively understand, forecast and control the response of complex systems to outside perturbations. This requires tackling two major technical challenges: to control abnormal behaviour in local sub-systems when these have the potential to disrupt the whole system and to offer security, stability and reliability. As such events are fortunately infrequent, the shortage of data available makes the technical challenges even more difficult to solve. Complex systems are often very difficult to predict and control. Due to the wide number of possible interactions, the analysis of complex systems challenge our ways of thinking. Often, our perception of them is oversimplified or biased. What we proposed is to investigate methodologies and approaches applied to different complex systems, reviewing the statistics and dynamics of cascading failures in various individual networks and interconnected networks. A failure in one part of a system can make the entire system to collapse and that happens simply because of the dynamics of the network. By questioning existing approaches and putting forward new ways of examining existing problems, the findings should ultimately foster a radically new approach to modelling blackouts and cascading events following them.

Understanding, Prediction, Mitigation of Cascading Failures in Interdisciplinary Complex Systems. Presented at: European Conference on Complex Systems Barcelona, Spain.

Spataru, C. (2013)

The full text of this article is not available through UCL Discovery.