Complex Sciences. First International Conference, Complex 2009, Shanghai, China, February 23-25, 2009, Revised Papers, Part 2

Research Article

Modeling Failure Propagation in Large-Scale Engineering Networks

Download
559 downloads
  • @INPROCEEDINGS{10.1007/978-3-642-02469-6_89,
        author={Markus Schl\aa{}pfer and Jonathan Shapiro},
        title={Modeling Failure Propagation in Large-Scale Engineering Networks},
        proceedings={Complex Sciences. First International Conference, Complex 2009, Shanghai, China, February 23-25, 2009, Revised Papers, Part 2},
        proceedings_a={COMPLEX PART 2},
        year={2012},
        month={5},
        keywords={cascading events failure dynamics decaying networks mean field approximation large-scale engineering systems},
        doi={10.1007/978-3-642-02469-6_89}
    }
    
  • Markus Schläpfer
    Jonathan Shapiro
    Year: 2012
    Modeling Failure Propagation in Large-Scale Engineering Networks
    COMPLEX PART 2
    Springer
    DOI: 10.1007/978-3-642-02469-6_89
Markus Schläpfer1,*, Jonathan Shapiro2,*
  • 1: ETH Zurich
  • 2: University of Manchester
*Contact email: schlaepfer@mavt.ethz.ch, jls@cs.man.ac.uk

Abstract

The simultaneous unavailability of several technical components within large-scale engineering systems can lead to high stress, rendering them prone to cascading events. In order to gain qualitative insights into the failure propagation mechanisms resulting from independent outages, we adopt a minimalistic model representing the components and their interdependencies by an undirected, unweighted network. The failure dynamics are modeled by an anticipated accelerated “wearout” process being dependent on the initial degree of a node and on the number of failed nearest neighbors. The results of the stochastic simulations imply that the influence of the network topology on the speed of the cascade highly depends on how the number of failed nearest neighbors shortens the life expectancy of a node. As a formal description of the decaying networks we propose a continuous-time mean field approximation, estimating the average failure rate of the nearest neighbors of a node based on the degree-degree distribution.