Network theorists usually assume blackouts spread the same way diseases do – by close contact – but for power grids “the relevant network is not the physical network” of lines and towers and transfer stations, says SFI Professor Cris Moore.

The cascade of failures in a blackout can jump from one power line to another line hundreds of miles away as the flow of power in the network rearranges itself, says Moore, who hosted a workshop at SFI September 9-12, 2013, to explore ways to adapt network science to better model critical infrastructure networks such as power grids.

Rethinking such critical infrastructure networks as “virtual or effective networks of dynamic connections” is a promising approach, he says, with possible insights to come from studying other highly dynamic networks such as food webs.

“A lot of what we were focusing on in the workshop,” says workshop co-organizer Matthew Koehler, MITRE’s Complexity Sciences Area Lead and a member of SFI’s Business Network, “was trying to figure out what is the next step in the development of the tools and techniques of network science to get it past its current limitations.”

Participants, including co-organizer Paul Hines of the University of Vermont, agreed that the next step was overcoming four challenges: getting more data to better understand the complexity and heterogeneity of critical infrastructure systems, understanding tradeoffs between detailed but slow physics-based models and abstract and quick network-based models of such systems, validating models once they are identified, and translating the output of validated models into information that is useful to decision-makers.

This was the second recent SFI workshop on the topic organized by Moore. He says he is pleased with this workshop’s results but recognizes the problems of protecting critical infrastructure as long-term challenges. He adds that he is hopeful that the coming year will bring sabbatical visitors to SFI “so we’ll have critical mass in house to make some major progress.”