by C. Poulin and M. Kane
Abstract:
When an infrastructure system responds to a disruption, its adaptive behavior can be described by the interactions between elements. These interdependencies are heterogeneous, multi-domain, and span temporal scales. Coupled with infrequent and limited historical data, it is challenging to identify and model key interdependencies. To address this, we propose a novel modeling methodology and iterative simulation approach. Starting with discrete event simulation and simulation graph models, we introduce dependency networks (DN), counterfactual event graphs (CEG), and multiverse simulation. Dependency networks link discrete events and model states to specific infrastructure elements. Counterfactual event graphs expand discrete events to multiple possible outcomes. Multiverse simulation considers each event outcome by duplicating and diverging the simulation into multiple timelines. Through iterative multiverse simulation and focused validation by stakeholders, the model is revised to identify and incorporate interdependencies that impact the system's post-disruption performance. This approach provides a framework for new modeling techniques and implementation in future case studies.
Reference:
C. Poulin and M. Kane, "Identifying Heterogeneous Infrastructure Interdependencies through Multiverse Simulation", In 2019 Resilience Week (RWS), pp. 123-131, 2019.
Bibtex Entry:
@inproceedings{Poulin2020ResWeek,
  author={C. {Poulin} and M. {Kane}},
  booktitle={2019 Resilience Week (RWS)}, 
  title={Identifying Heterogeneous Infrastructure Interdependencies through Multiverse Simulation}, 
  year={2019},
  volume={1},
  number={},
  pages={123-131},
  abstract={When an infrastructure system responds to a disruption, its adaptive behavior can be described by the interactions between elements. These interdependencies are heterogeneous, multi-domain, and span temporal scales. Coupled with infrequent and limited historical data, it is challenging to identify and model key interdependencies. To address this, we propose a novel modeling methodology and iterative simulation approach. Starting with discrete event simulation and simulation graph models, we introduce dependency networks (DN), counterfactual event graphs (CEG), and multiverse simulation. Dependency networks link discrete events and model states to specific infrastructure elements. Counterfactual event graphs expand discrete events to multiple possible outcomes. Multiverse simulation considers each event outcome by duplicating and diverging the simulation into multiple timelines. Through iterative multiverse simulation and focused validation by stakeholders, the model is revised to identify and incorporate interdependencies that impact the system's post-disruption performance. This approach provides a framework for new modeling techniques and implementation in future case studies.},
  keywords={critical infrastructures;discrete event simulation;graph theory;discrete event simulation;simulation graph models;counterfactual event graphs;dependency networks;discrete events;model states;specific infrastructure elements;multiple possible outcomes;event outcome;iterative multiverse simulation;modeling techniques;infrastructure system;adaptive behavior;span temporal scales;infrequent data;limited historical data;model key interdependencies;modeling methodology;iterative simulation approach;heterogeneous infrastructure interdependencies;modeling;simulation;applications;resilience;critical infrastructure;discrete event simulation;failure analysis;network theory (graphs);interdependencies},
  doi={10.1109/RWS47064.2019.8971826},
  ISSN={},
  month={Nov},
  pdf = {http://files.thisismikekane.com/pubs/2019_Identifying_RW.pdf}  
}