Get the picture

09.01.2006
With gasoline prices up and equipment to drill new wells backlogged, oil producers want to squeeze every drop out of existing wells. But which sites have more oil to give up? To revisit old drilling data for possible clues, Halliburton Co.'s energy services group turned to a new-breed data visualization tool, and so far the analysis has led to more contracts for the Houston firm.

By creating a "picture" of the data -- a process that entails rendering multiple variables into a graphical presentation -- Halliburton analysts have found ways to see deep into the earth, spotting patterns, trends and anomalies in the data.

"We wanted sites that had very high production at first but then decreased rapidly," says Judy Tiffin, U.S. business and marketing analyst at Halliburton, noting that such reservoirs might have gotten clogged somehow and could be kick-started into production by one of Halliburton's services. "We can go in and stimulate or treat the well differently and maybe get the production up higher," she says.

However, wading through data to find those sleeper sites is a tricky process that involves comparing many variables. "It's usually not just one thing -- there's so much [to consider]," Tiffin says. "We don't know what's going on underneath the ground."

To juggle the millions of pieces of this multidimensional or multivariant data, Halliburton's production optimization group tried Spotfire Inc.'s DecisionSite software. Until then, the five-member production optimization team had plodded through internal and public well data and seismic readings one Microsoft Excel data plot at a time.

Halliburton, Dunn & Bradstreet Corp., Cullen/Frost Bankers Inc. and Microsoft Corp. are among the companies embracing such analysis software, which devours data either from existing spreadsheets and database products or from data warehouses and business intelligence systems, and transforms it into visual representations. The tools -- some of which cost as little as $999 -- enable users to slice and dice data on the fly multiple times and to spit out the findings in multicolor bar and pie charts, scatterplots and diagrams.