Published in 06/09/2023 at 14:35:50
Six Petrobras supercomputers are among the most eco-efficient machines in the world, according to the Green 500 global ranking – a list that features equipment that combines high power with greater energy efficiency. Petrobras' supercomputers are in the top 6 among all Latin American machines. One of the most recent computers, Gaia, began its operation in August at the company's Development and Innovation Research Center (Cenpes). With a processing capacity of 7.7 Petaflops FP64, equivalent to 1.5 million cell phones or 40,000 laptops, the supercomputer, exclusively used for research purposes, has an energy consumption of 574 KW, equivalent to that of a city of 2,400 people, such as Superior, in Wisconsin.
Gaia is used to develop and improve Geophysics technologies, which can be applied both in pre-salt fields and in other areas. The idea is to improve seismic images processing tools - a technique that produces 3D images of the Earth’s interior - to obtain very high definition reproductions of rock layers in the subsurface.
“More than recognition, being in the list of the most eco-efficient supercomputers in the world proves that we are on the right track towards energy efficiency, combining operational performance with low energy consumption. We are proud of that and channeling our efforts to moving forward in incorporating high-performance technology into our production processes, generating increasingly reliable and efficient data”, said Carlos Travassos, Director of Engineering, Technology and Innovation.
Gemini, which also began its operation in August, has its 3.9 Petaflops FP64 and is used for geophysical processing and the development of new algorithms.
Another supercomputer whose production has recently begun is Albacora, which was named in celebration of the 35th anniversary of production at Petrobras' first giant field. Together with Cazarin - in memory of geologist Caroline Cazarin, a reference in the company -, it has, in total, around 500 servers, 30,000 processing cores and a RAM memory of 130 terabytes. Both will be used for geophysical processing and for the construction of flow and geomechanical simulation models for the reservoirs.
Petrobrás atop high-performance computing
The supercomputers now in operation Pegaso, Dragão, Fênix and Atlas have made it to Green500. Pégaso is considered the largest in Latin America in processing capacity and eco-efficiency. The result endorses the company's leading position in high-performance computing (HPC). The machine has a processing power equivalent to the sum of 6,000 cell phones or 150,000 modern laptops.
Dragão went into operation in 2021 with a RAM (Random Access Memory) of 200 terabytes, 100 gigabits per second, and millions of mathematical processors. The supercomputer's processing capacity is equivalent to that of 4 million smartphones or one 100,000 modern laptops. It is an important ally of the company to further increase the performance of geophysical data processing, reducing geological and operational risks, with an eye to supporting the company's strategic projects.
Lastly, Fenix and Atlas, the oldest ones in operation, together have a processing capacity equivalent to 2.5 million smartphones or 67,000 modern laptops. The two, which are ranked 5th and 6th, respectively, among the most eco-efficient supercomputers in Latin America, are responsible for processing geophysical data generated during exploration activities and development of oil and gas production.
A great deal of processing power is needed to create representative images of the ocean floor, where the geological layers are and where oil reservoirs are found. The seismic images, fundamental for oil and gas discoveries, are spread over hundreds of square kilometers and thousands of meters down deep into the sea.
Therefore, the algorithms that process them involve complex mathematical equations, with a huge volume of data, generating images that geologists and geophysicists can interpret. The volume of data for a single seismic project can reach tens of terabytes, more than the capacity of the HDs of a desktop computer used today.
Use of this material is authorized for editorial purposes only.