How to measure an oil spill

EMERGENCY RESPONSE: Mechanical engineering professor Ömer Savaş helped federal agencies estimate the official flow rate during last year’s Deepwater Horizon oil spill in the Gulf of Mexico, the largest offshore oil spill in U.S. history. PEG SKORPINSKIEMERGENCY RESPONSE: Mechanical engineering professor Ömer Savaş helped federal agencies estimate the official flow rate during last year’s Deepwater Horizon oil spill in the Gulf of Mexico, the largest offshore oil spill in U.S. history. (Photo by Peg Skorpinski.)When oil was flowing from BP’s broken well at the bottom of the Gulf of Mexico last spring, the company estimated the flow rate at about 1,000 barrels a day. But news outlets wanted an independent estimate.

Could Ömer Sava?, an expert in fluid mechanics and turbulent flows, help? At first, the Berkeley mechanical engineering professor declined, later explaining, “I didn’t want to contribute to the sensationalism.”

But soon afterward Sava? received an e-mail addressed to the membership of the American Physical Society (APS)’s Division of Fluid Dynamics, to which he belongs, asking for estimates.

He could no longer ignore the request. He watched videos of the leaking well and sent off an initial estimation, one that was much higher than 1,000 barrels a day. Thus, Sava? became involved in a national effort to establish the “official” flow rate, a number that would dictate not only the level of resources assigned to the cleanup but also its legal ramifications once the emergency had passed.

Soon afterward, he received a request from Bill Lehr, senior scientist at the National Oceanic and Atmospheric Administration Office of Response and Restoration. Lehr, who had seen Sava?’s APS estimate, was assembling a team of academic researchers and federal scientists charged with estimating the flow rate.

Sava? joined Lehr’s Plume Team and spent hours reviewing BP’s high-definition videos of the leaking well. He was soon exchanging hundreds of e-mails with other team members from around the country, debating how best to approach the problem.

The team settled on a method called particle image velocimetry (PIV), in which a fluid is seeded with tracer particles whose velocities are extracted from successive images using elaborate image-processing techniques.

Since PIV couldn’t be performed in a controlled manner at the bottom of the gulf, instead scientists had to be content to base their calculations on video streams coming from the gulf. Their initial estimates put the flow rate at 12,000 to 18,000 barrels a day, which they later revised to 24,000 to 30,000 barrels.

After running his own calculations using PIV, Sava? concluded that the technique posed serious difficulties in this case.

“We’re looking at oil, this dark fluid, and we can only look at the face of it, what is called turbulent flow interface,” he explains. “There was a fine scale of turbulence that was changing too rapidly to be captured by the video streams, and that’s why I chose not to employ PIV for flow rate estimation.”

AGAINST THE FLOW: Using a probability density function graph, the National Institute of Standards and Technology summarized some of the expert estimates of the rate of oil flowing from the damaged well. Savaş’s estimate came closest to the mean and median values and very close to the direct flow estimate before the well was capped. COURTESY ÖMER SAVAŞAGAINST THE FLOW: Using a probability density function graph, the National Institute of Standards and Technology summarized some of the expert estimates of the rate of oil flowing from the damaged well. Savaş’s estimate came closest to the mean and median values and very close to the direct flow estimate before the well was capped. (Photo by ÖMER SAVAŞ)Instead of focusing on minute structures, Sava? selected larger features—the big billows coming out of the leak—and ran calculations based on the celerity of those “coherent structures,” as they’re known in fluid dynamics.

In e-mails, teleconferences and meetings, his colleagues questioned his approach, which consistently produced higher numbers, but Sava? didn’t back down.

After spending many weeks on his analysis, he finally concluded that the flow rate was not lower than 44,600 barrels a day, maybe even 10 or 20 percent higher. His results, along with everyone else’s on the team, were statistically analyzed by the National Institute of Standards and Technology.

Lehr’s Plume Team was joined by a Department of Energy team that had done its own calculations, producing results even higher than Sava?’s. On June 15, the federal government announced the official flow rate: between 35,000 and 60,000 barrels a day.

Right before BP engineers finally capped the well in mid-July, they inserted a pressure device to gauge the flow and measured 53,000 barrels a day, plus or minus 5 percent.

“Ömer had a different approach than the others, and that was a good thing,” Lehr says. “We were working under emergency, not experimental, conditions and under extremely tight timeframes, so there was a lot we simply didn’t know. None of our estimates was exact, but Ömer’s came closest to the direct pressure reading. It was a great contribution to the country, and Berkeley should be proud.”

For his contributions, along with his team members, Sava? received a U.S. Geological Survey Director’s Award in December.

That same month, BP formally challenged the official rate, calling it “flawed” and arguing that it should be 20 to 50 percent lower (which could lower BP’s fine by several billion dollars). In response, the Department of Justice is mounting a defense and has requested Sava?’s computer files related to his analysis.

Whatever the outcome, Sava? stands by his contribution, which he calls “the power of the pencil.”

“To think that some simple calculations done here at Berkeley had such a huge impact, that’s an awesome and at the same time a sobering feeling.”