DHS Fellow and USCG Intern Using Data Science to Study Oil-Rig Blowout in the Chukchi Sea
Department of Homeland Security (DHS) Fellow Richard Garrett stared in amazement at the bedlam of activity in the Incident Command post where the US Coast Guard and other government agencies were directing efforts to clean up a May 19, 2015, oil spill at Refugio State Beach in Santa Barbara County, California.
A rupture in an oil-transport pipe owned by Plains All American Pipeline caused the spill, leaking 101,000 gallons of crude oil into a highway culvert. Of that amount, 21,000 gallons slid into the Pacific Ocean at the ecologically fragile Refugio State Beach.
Garrett arrived soon after the cleanup had begun. In the command center, he witnessed response-team members huddled in small groups, speaking animatedly. Others traced their fingers over maps. Large electronic screens displayed up-to-the-minute information. Men and women in colorful reflective vests whisked in and out of the post, delivering reports from the beach.
It was Garrett’s first inside look at response-team decisions being made to an oil spill in real time. For a doctoral student working as an intern with the Coast Guard to develop a sophisticated mathematical model to assess the effectiveness of planned responses to a hypothetical well blowout in the Arctic’s Chukchi Sea, it was an eye opener.
“I still can’t believe how everything works in reality,” says Garrett. “It seems like organized chaos when you first look at it.”
Student of ‘Response Logistics’
At the time, Garrett, a doctoral student studying Industrial and Systems Engineering at Rensselaer Polytechnic Institute in Troy, NY, was also an intern with the US Coast Guard’s Pacific Area division (PAC-7). His assignment was to research decision support methods for oil-spill-response logistics in the Arctic, which is why the Coast Guard had invited him to witness the Refugio beach response.
The internship was made possible by the Command, Control and Interoperability Center for Advanced Data Analysis (CCICADA), which is using its expertise in data analysis and mathematical modeling to help the Coast Guard with a number of homeland security projects. Rensselaer Polytechnic Institute (RPI) is one of CCICADA’s 17 partner institutions.
Garrett is also a prime example of how CCICADA is fulfilling one of its most important missions as a University Center of Excellence funded by the US Department of Homeland Security, which is to educate and train the next generation of homeland-security research scientists. “Richard Garrett’s US Coast Guard internship is a good example of a high-level implementation of the DHS education mandate,” says CCICADA Director Fred Roberts.
Garrett has been working on the Dynamic Modeling for Arctic Resource Allocation (DMARA) project, a CCICADA/USCG venture that uses data analytics and mathematical modeling to determine the optimal use of scarce equipment to respond to oil spills in the Arctic. The DMARA research project is designed to help the Coast Guard implement its May 2013 Arctic Strategy.
Chukchi Sea Blowout Study
Garrett’s new assignment is to find ways to adapt those modeling efforts to a standalone project: evaluating the likely effectiveness of contingency plans to respond to a hypothetical blowout of an exploratory drilling rig in the Chukchi Sea.
Given potentially harsh Arctic weather; the vast distances between the rig and Alaska’s north shore; and the scarcity of deep-water ports, airplane runways and other infrastructure on the north shore’s ecologically fragile tundra, the complexity of responding to an even moderate blowout beggars the imagination. Understanding that complexity, however, is within the reach of science.
Garrett, under the close supervision of his academic advisors Drs. Thomas Sharkey, Martha Grabowski, and William A. Wallace, and in coordination with members of industry and government, is using sophisticated data science to determine what might—or might not—make an effective response. Specifically, they are seeking an objective understanding of oil spill responses from one hypothetical case study so experts can study how logistical demands affect responses in remote regions.
Garrett has already made a contribution to the effort by developing a software tool to organize, connect, and encode millions of pieces and layers of data related to the Chukchi Sea drilling rig’s Oil Spill Response Plan (OSRP). The data primarily comes from three publicly available sources: the 450-page ORSP itself and two response tactics manuals, one written by the energy company that owns the rig, the other by Alaska Clean Seas, a prominent spill-response organization in the region.
Ultimately, the plan is to feed this and other data, such as the extent and position of the spill, into a computer-driven “optimization” model, which would evaluate the effectiveness of planned movements of manpower and equipment and the scheduling of those logistics. After digesting this information—comprising hundreds of thousands, if not millions, of data points from many sources—the computer would spit out a single number rating the efficacy of a planned response.
Promise of Success
The optimization model would reach this number by analyzing the assumptions made by the current response plan, and then comparing these assumptions to the response tasks laid out in the plan. It would, in essence, provide a reality check of the potential effectiveness of the plan based on scientific data, not guesswork.
Garrett’s project, mind boggling in its complexity, is designed to support the USCG District 17’s Spill of National Significance (SONS) Logistics Assessment. There’s still much to do, and it remains a work in progress.
However, the promise of its success is huge.
Spill-response commanders and their teams currently lack the benefit of science-based data to support the thousands of decisions and snap judgements they must make in highly compressed time frames.
Garrett and others in his field are hopeful that the case study of one hypothetical oil-rig blowout in the Arctic will lead to another research project—developing computer models that can be used in the field to give response commanders real-time data to support their decisions.
Garrett says the current project, in his estimation, “is the core, the seed, the foundation for that to happen.”
The Refugio beach oil-spill response that Garrett observed could have benefited from such an approach.
California Spill No Blip
Compared with the nation’s top three oil-spill disasters—the Santa Barbara oil-rig blowout in 1969, the Exxon Valdez disaster in 1989, and the BP Deepwater oil-rig explosion in 2010—each of which spewed millions of gallons of crude oil into the ocean, the Refugio incident might appear to be a blip in the nation’s oil-spill history.
In reality—and certainly to Garrett—this was no blip.
In his report on the incident, he says he obtained “a unique and invaluable understanding of OSR (oil spill response) in practice.” This gave him ideas for future research and allowed him to cultivate a widened network of experts in his field of expertise.
Garrett observed the cleanup efforts from within the response team’s Unified Command, led by the US Coast Guard and National Oceanic and Atmospheric Administration (NOAA).
He was given access to daily command-level briefings, as well as to NOAA’s Emergency Response Management Application and the USCG large-scale, rapid-response, online-information-exchange system.
Garrett observed each level of the decision-making process, which revealed “the deltas between planned courses of action,” response-team communication gaps, and how the response structure and tactics changed to manage “a dynamic, idiosyncratic event.”
The experience left him with only the highest admiration and respect for oil-spill responders and their field commanders. “There’s layer upon layer of decision making and complexity,” Garrett said in his report, “and very often they don’t have data-driven, decision-support tools to provide the feedback they might need.”
Arctic Oil Rush
Melting ice has exposed extensive arctic continental shelves believed to contain the world’s largest undiscovered reserves of oil (an estimated 90 billion barrels) and natural gas (an estimated 1.7 trillion cubic feet). These potential reserves are in addition to the 240 billion barrels of petroleum already discovered in the region.
This has stirred tremendous interest and activity in exploratory drilling among oil producers and the five nations bordering the Arctic: Russia, Denmark, Norway, Canada and the United States.
The Arctic oil rush is on—and with it the prospect of gigantic spills in a harsh environment with a fragile ecology.
Protecting this region presents an enormous challenge for the US Coast Guard, which is also responsible for responding to oil spills there.
Coast Guard Commandant Paul Zukunft recently made it clear that being fully prepared for an effective response to a well blowout in the Chukchi Sea is a high priority for his agency.
“There is zero room for failure, and by that I mean an oil spill in the Arctic,” Zukunft said at a July 2015 Arctic symposium. “We’ve worked very closely with the Department of Interior and the Bureau of Safety and Environmental Enforcement…when it comes to drilling in the Chukchi Sea.”
Exxon Valdez Lessons
The 1989 Exxon Valdez disaster underscored the logistical nightmare of rushing mountains of equipment and armies of manpower to respond to such a spill.
Under the leadership of a captain who was below decks sleeping off a drinking binge, the Exxon Valdez tanker hit a reef, gashing the hull and releasing 11 million gallons of crude oil into Prince William Sound, two hours by boat from the town of Valdez on Alaska’s southern shore.
The spill quickly covered 1,000 square miles of ocean. Before the cleanup was over, Exxon alone had deployed more than 1,800 response personnel, and contracted 25 aircraft and 248 boats to deliver equipment and responders. By the fourth day, 274 tons of booms, skimmers, dispersant and related equipment—flown in from around the world—had been delivered, the EPA said in a Report to the President.
And that was just one part of a largely ineffective response to a spill whose environmental impacts are still being felt more than 25 years later.
“More than 1,000 miles of coastline were fouled, and hundreds of thousands of animals perished…though the oil has mostly disappeared from view, many Alaskan beaches remain polluted to this day, crude oil buried just inches below the surface,” The Atlantic reported in a 25th anniversary article about the spill.
Soon after the spill, Garrett’s main adviser on the Chukchi Sea project, RPI Professor Wallace, led a National Science Foundation quick response team to the Exxon Valdez site. In their 2001 report to the NSF, Wallace and his co-authors explained why the response was ineffective.
“Decision making during the operation was proactive rather than reactive,” the authors said. “The planners did not anticipate the limitations of the technology available and the amount of equipment that would be required, and they did not devise adequate organizational plans and operational strategies. These deficiencies in the plans were not fully revealed until the incident occurred because the response scenarios were not fully developed or exercised.”
They proposed a solution: “The development of useful decision-support models and support systems to aid managers coping with rare, catastrophic events deserves research support. Simulation methodologies could be used to test and preserve the integrity of system safety and to prepare crisis managers for the intense, hostile and complex decision environment they must face.”
And that is precisely the solution—26 years later—that is being worked on in the Chukchi Sea oil-spill-response-plan project.