Researchers at the Moscow Institute of Technical Physics have developed an innovative molecular model based on machine learning that can determine the properties of oil during the extraction process with an accuracy of up to 90%.
This technology aims to significantly increase the efficiency of oil extraction by accurately calculating the surface tension between fluids inside porous rocks, according to the Russian news agency Novosti, citing the institute's press service.
Nikolay Kondratyuk, executive director of the Center for Computational Physics at the Moscow Institute of Physics and Technology, said: "In order to extract oil efficiently, we usually need a long time to determine the optimal salinity of the water and take into account the gas composition in a particular oil field, while our model greatly accelerates this process."
He pointed out that previous models gave a margin of error of up to 40%, while laboratory tests took months and required exorbitant costs.
The new multi-component model, developed by researchers in collaboration with scientists from the Institute of High Thermal Physics of the Russian Academy of Sciences and the Tyumen Oil Scientific Center, uses a supercomputer. It takes into account the oil's composition, temperature, and pressure, as well as dissolved gases and salts, enabling the use of machine learning algorithms to achieve a prediction accuracy of up to 90%.
This technology is particularly important for methods of injecting carbon dioxide into rock formations. Ilya Kopanichuk, a senior researcher at the Moscow Institute of Physics and Technology, noted that the prototypes have successfully passed testing, adding that scientists plan to adapt the system for heavy oil and study the behavior of fluids in nanostructures.
It is worth noting that scientists at the Moscow Institute of Physics and Technology had previously developed a new neural network based on "liquid light." Alexander Dyukov, head of Gazprom Neft, also stated earlier that Russia has the potential to increase oil production using modern technologies.
