By Thea Singer

(Released 4 November 2015) The 1999 Odisha Cyclone struck the eastern coast of India, knocking out whole swaths of the Indian Rail­ways Net­work, bringing the eastern IRN system to a halt. Cyclones Hudhud and Phailin caused sim­ilar mayhem in 2014 and 2013, while in 2012 power black­outs in northern and eastern Indialed 300 inter­city pas­senger trains and com­muter lines. Closer to home, severe winter storms that hit Boston in 2014–2015 brought the MBTA mass- transit system to its knees.

Here and abroad, there is an urgent need for sys­tem­atic strate­gies for recov­ering crit­ical life­lines once dis­as­ters strike. Thanks to North­eastern researchers, that need is being met.

A tool for recovery First- year grad­uate stu­dent Udit Bhatia, under the direc­tion of Auroop R. Gan­guly, asso­ciate pro­fessor in the Depart­ment of Civil and Envi­ron­mental Engi­neering, has drawn on net­work sci­ence to develop a com­put­er­ized tool for guiding stake­holders in the recovery of large-scale infra­struc­ture sys­tems. In addi­tion to the IRN and MBTA, the method can be extended to water-distribution sys­tems, power grids, com­mu­ni­ca­tion net­works, and even nat­ural eco­log­ical systems.

This unique tool, which has been filed for inven­tion pro­tec­tion through North­eastern University’s Center for Research Inno­va­tion, also informs devel­op­ment of pre­ven­ta­tive mea­sures for lim­iting damage in the face of a disaster.

The study—which Bhatia and Gan­guly coau­thored with Devashish Kumar, PhD’16, and Evan Kodra, PhD’14—appears in the Nov. 4 issue of the journal PLOS ONE.

“The tool, based on a quan­ti­ta­tive frame­work,en­ti­fies the order in which the sta­tions need to be restored after full or par­tial destruc­tions,” says Bhatia, PhD’18, who is a stu­dent in Northeastern’s Sus­tain­ability and Data Sci­ence Lab­o­ra­tory, directed by Gan­guly. “We found that, gen­er­ally, the sta­tions between two impor­tant stops were most crit­ical,” he says, alluding to the net­work sci­ence con­cept of “cen­trality mea­sures,” whichen­tify sta­tions that enable a large number of station- pairs to be con­nected to one another.

A new top- down approach Bhatia credits Northeastern’s inter­dis­ci­pli­nary engi­neering grad­uate pro­gram with opening his mind to the pos­si­bility of con­structing the model.

Through the pro­gram, he took courses with experts in a variety of fields. They include: “Crit­ical Infra­struc­tures Resilience,” co- taught by Gan­guly, an expert in cli­mate, hydrology, and applied data sci­ences, and Stephen Flynn, a pro­fessor of polit­ical sci­ence and director of the Center for Resilience Studies and co- director of the George J. Kostas Research Insti­tute for Home­land Secu­rity, and “Com­plex Net­works,” taught by Albert- László Barabási, Robert Gray Dodge Pro­fessor of Net­work Sci­ence. Insights from Jerome F. Hajjar, CDM Smith Pro­fessor and CEE Chair and an expert in struc­tural engi­neering, also helped shape the model.

“Struc­tural engi­neers have typ­i­cally focused on rebuilding large infra­struc­tures from the bottom up,en­ti­fying indi­vidual com­po­nents or small- scale infra­struc­ture sys­tems,” says Bhatia. For IRN, this might meant tar­geting the busiest sta­tion to begin repairs.

Bhatia’s paper—based on a mix of real- world met­rics, resilience, civil engi­neering prin­ci­ples, and net­work science- based algorithms—provides what Gan­guly calls “a generic and quan­ti­ta­tive top- down approach.”

A com­pre­hen­sive strategy requires a blend of bottom- up and top- down approaches, says Gan­guly. “If these nodes of the system go down, here is a timely, resource- efficient, and overall effec­tive way to speed recovery.”

“Auroop and Udit are devel­oping a system frame­work, which is a new approach for solving com­plex system prob­lems,” says Jalal Mapar, Director of the Resilient Sys­tems Divi­sion, Depart­ment of Home­land Secu­rity, Sci­ence & Tech­nology Direc­torate. “This new approach is very impor­tant and answers many of the com­plex ques­tions that we will be facing in the next 5 to 50 years. It will help us under­stand the inter­de­pen­den­cies and cas­cading effects of our crit­ical infra­struc­ture, and help us as a nation to be better pre­pared because we know what we are dealing with.”

Mining datasets, con­structing a network For the study, Bhatia mined open- source datasets on ticket- reservation web­sites to track the ori­gins and des­ti­na­tions of trains run­ning on the IRN—the world’s most trav­eled railway in terms of pas­senger kilo­me­ters per day. He then con­structed a com­plex net­work, with the sta­tions as nodes and the lines con­necting those nodes as the “edges,” or links, between them, and over­laid it on a geo­graph­ical map of the country. Next he applied nat­ural and man- made dis­as­ters to the system, knocking out sta­tions using net­work science- derived algorithms.

“We con­sid­ered real- life events that have brought down this net­work,” says Bhatia, ticking off the 2004 Indian Ocean Tsunami and the 2012 North Indian blackout due to a power grid failure, as well as a sim­u­lated cyber- physical attack, par­tially mod­eled after the November 2008 Mumbai terror attack. “We asked: Should this recovery be based on the number of trains each sta­tion han­dles, the number of con­nec­tions each sta­tion has, the impor­tance of the con­nec­tions, where that sta­tion is located in the net­work, or some­thing else?”

The researchers devel­oped addi­tional algo­rithms “to assign pri­ority to each sta­tion,” Bhatia says, indi­cating when it should be brought back online to pro­duce the fastest recovery of the entire system.

Preparing for the worst In the IRN study, “between­ness cen­trality” often came to the fore. Bhatia cau­tions, how­ever, that a single metric or strategy does not apply in all cir­cum­stances; for example, if just part of a net­work is dis­rupted, a par­tic­ular sta­tion with an out­size number of con­nec­tions might take prece­dence as a starting point over a sta­tion sit­u­ated between two impor­tant stops.

“This model gives you the ability to say, ‘These are the most crit­ical nodes in the net­work, which if they failed, would cause a domino effect in the case of a disruption—meaning a cas­cading failure when there’s a major shock,’” says Flynn, who recently tes­ti­fied before the U.S. House of Rep­re­sen­ta­tives on the pre­ven­tion of and response to the arrival of a dirty bomb at a U.S. port.  “‘So that’s obvi­ously where we should go first.’”

If the Boston MBTA had this tool during last winter’s his­toric snow­fall, he says, they would have known where to start to get the transit system back up and running.

More­over, Flynn says, the model gives decision-makers—urban plan­ners, emer­gency man­agers, oper­a­tions per­sonnel who run the system day-to-day—insight into how to design the most secure system upfront. “And then,” he says, “it enables them to pri­or­i­tize where to put mit­i­ga­tion measures—resources, such as backup power, and other safe­guards, including computer- security mea­sures, to make the overall system better with­stand the risk of disruption.”