Unraveling fluid-fluid interfaces through molecular simulations and machine learning

• Grant number: 21/00953-4
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: March 01, 2021
• End date: November 30, 2021
• Field of knowledge: Physical Sciences and Mathematics - Physics - Condensed Matter Physics
• Agreement: BG E&P Brasil (Shell Group)
• Principal Investigator: Caetano Rodrigues Miranda
• Grantee: Dan Ni Lin
• Host Institution: Instituto de Física (IF). Universidade de São Paulo (USP). São Paulo , SP, Brazil
• Associated research grant: 14/50279-4 - Brasil Research Centre for Gas Innovation, AP.PCPE

Abstract

The control of fluid-fluid interfaces is of significant importance in the oil and gas industry. We have the reduction of the interfacial tension oil-brine that favors the emulsification of hydrocarbons, considered one of the mechanisms associated with improved oil extraction. Previous studies in the proponent's group showed correlations between the molecular composition of fluids (brine and oil) with the interfacial tension, providing the injected solution's ionic adjustment. Here we will continue the theme with a systematic investigation of the parameters that influence interfacial phenomena to understand the underlying molecular mechanisms that occur at the oil/brine interface. Configurations obtained by previous and new molecular dynamics simulations will be used to create an interfacial database. We will examine quantities that characterize the fluid-fluid interfaces, such as the interface width, density profile, and order parameters. Besides, hydrogen water-water and water-oil bonds will be analyzed through graph theory. These and other quantities will compose a database used in machine learning to describe the interfacial tension as a function of the interface's molecular characteristics. Concomitantly, these data will be analyzed using the sonification technique, converting data into sounds, as an alternative in representing the variations of properties and phenomena emerging from these interfaces' molecular dynamics. The proposal aims to introduce machine learning methods and deepen the knowledge in molecular simulations within the perspective of how basic science can improve the understanding of interfaces and advances in enhanced oil recovery processes. (AU)