Supercomputer Delivers for the Energy Sector

The global energy company Total draws on the power of supercomputers for advanced 4D modeling to locate and simulate the behavior of oil reserves under the surface. 4D seismic consists of repeating 3D seismic surveys over time across the same area. Total’s geophysicists and reservoir engineers develop models based on complex physics by working with advanced algorithms that require a great deal of computational power.

Arnaud Rödde, the manager of Total’s Petroleum Computing Centre in Pau, France, explains that the company has been using the power of computing as part of its research and exploration strategy for a decade. The company installed one of the first supercomputers in Pau in the 1980s and has upgraded every few years to faster and more powerful computers.

When it came time to replace the supercomputer it had been using in Pau in 2012, the company that won the bid was SGI because it offered the Pangea supercomputer, which was very efficient in terms of cost and energy efficiency. That was installed in 2013 and then upgraded at the beginning of 2016.

The reasons for the upgrade are to improve the accuracy of subsurface imaging, optimize the development and production of Total’s oil fields, and to save time by cutting down computation from weeks to days. Rödde explains, “Technology evolves very quickly, particularly in that field.” For that reason, staying on top of it requires an upgrade every two to three years.

Supercomputer Triples Previous Computation Power

Total’s new Pangea supercomputer tripled the company’s previous computation power from 2.3 to 6.7 petaflops (a petaflop does a quadrillion floating point operations per second), which amounts to the combined computing power of more than 80,000 laptops. Its storage capacity has been increased to 26 petabytes, the equivalent of 6 million DVDs. That makes this Pangea computer the top performing one within the industry and one of the top 10 in the world, according to the TOP500 ranking.

The two main applications for the computation are seismic imaging and reservoir simulation. Seismic imaging produces images of what can’t be seen directly, like what is under a salt dome. With more data used for the more complex algorithms, they are able to rely less on approximation to produce the imaging. As a result, they obtain better topography of subsurface to reveal what couldn’t be seen before.

Complex calculations are also applied to reservoir simulation. That entails simulating how the liquid would move within the reservoir. They have to run many scenarios in parallel to optimize the development of the oil field.

Supercomputer’s Results Are Better and Faster

With the more advanced supercomputer, the results are not only better but faster. What had previously taken three weeks for the computer to churn through now takes only three days, Rödde explains.

Time really is money when the company is working within the timeframe set by the host country for it to work on the designated block. If they gain a month or two as a result of the accelerated computation, that could make a big difference in how much they can gain from the block.

Total’s coders bank on that ability to advance when they develop their algorithms in anticipation of the technology that will emerge in the next three to four years. The supercomputers are built to meet those needs by completing the calculations within a reasonable timeframe.

The cost of such advanced technology is significant. The recent investment in Pangea amounted to 60 million euros for the first stage and an additional 35 million euros for the upgrade this year.

However, the amount can be recouped by saving the cost of a single oil well. As a result of the improved simulation, the company can plan more efficiently in order to extract as much oil from 14 wells as it could from 15. That saves the company the millions it would have had to spend on the 15th well.

“Pangea helps optimize development and so optimizes our costs because we’re able to spare a well here or do a more optimized development,” Rödde says. “All that factors into our cost efficiency.”