Students develop hurricane response plans on Cambridge roads, gaining practical experience in computational science

Imagine a powerful hurricane has wreaked havoc on the city of Cambridge, Mass. Thousands of residents are injured, but debris blocks roads everywhere, preventing medical workers from reaching the victims.

Crews are mobilizing to clear paths between the victims and two medical centers, Mount Auburn Hospital and Harvard University Health Services. Which roads should they open first, in order to quickly reach the largest number of victims? How many of those roads can they actually clear each day with the equipment available?

This was the problem posed to tech-savvy students participating in the IACS Computational Challenge in January. The competition was part of ComputeFest, a 2-week program hosted by the recently created Institute for Applied Computational Science (IACS) within the Harvard School of Engineering and Applied Sciences (SEAS).

“The amount of debris created by regularly occurring disasters is huge,” said Özlem Ergun, Visiting Associate Professor of Applied Mathematics at SEAS. In her usual post, Ergun is co-director of the Center for Health and Humanitarian Logistics at Georgia Institute of Technology, where she helps emergency management officials plan their response to disasters.

“The first problem,” she said, “is really to figure out in what order to open the streets so that you create connectivity between the population and the critical infrastructure.”

The Cambridge debris data was generated by Georgia Tech graduate students using the Federal Emergency Management Agency (FEMA) Hazus software, which visually models the human and environmental impacts of earthquakes, hurricanes, and floods.

In the Challenge scenario, 2,478 disaster victims were distributed unevenly across 443 Cambridge locations served by two hospitals (one large, one small), connected by 604 road segments (blocked by varying amounts of debris), and accessed via a fleet of bulldozers that roughly doubled in size over a 9-day cleanup period. A penalty was imposed to simulate the real-life pressure of time—the chance of people losing their lives if help took too long to arrive.

In short, the number of data points and constraints was huge…. [more]