Advances Towards a Comprehensive Model of Laser-Rock Interaction

Abstract

Laser applications provide unique advantages to applications in discovery, recovery, and production of hydrocarbons. A comprehensive numerical model would enable prediction, optimization, increase efficiency, enhance control, and further innovation. This work reviews the modeling methods, discusses key variables and physics, presents results, and introduces innovative solutions that make use of machine learning and artificial intelligence to solve an inherently multi-scale and multi-physics problem. Two possible methods have been explored to model laser-rock interaction: mechanistic and statistical — the former uses as a set of coupled partial differential equations that adequately describe the physics involved. The statistical method uses advanced statistical analysis and supervised-learning to elucidate relations between the experimental settings and observations. The full-physics or mechanistic model was developed using finite-element and finite-difference methods; it incorporates coupled solvers for electromagnetic, thermodynamics, and geomechanics. The statistical model uses advanced statistical analysis and machine learning to characterize the dynamics and build a prediction algorithm. A numerical model of laser-rock interaction must comprise physical dynamics that span over different time and spatial scales. When a laser beam impinges on a rock, a portion of the incident energy is absorbed as thermal energy, and a thermal gradient is created. The result is a distribution of physical and chemical changes such as spallation, melting, dissociation, calcination, or vaporization. The full-physical model can adequately capture the transient process; however, it requires a functional characterization of the dynamic rock properties, environmental conditions, and laser parameters. The statistical approach provides a prediction of the overall outcome of the process departing as a function of known input parameters, yet its precision depends on the availability of experimental data (outcomes and conditions). Key parameters are identified using statistical analysis. The modeling results agreed with experimental tests. Further, they evince that thermal properties and geomechanical stresses configuration have a significant impact on the process’ outcome. These methods can optimize and predict the interaction for multiple applications, ranging from heat treatment to stimulation. Subsurface laser operations could provide the next generation of stimulation and workover tools for Oil and Gas. Numerical models of laser-rock interaction are essential to predict, optimize, adapt, and evaluate subsurface laser applications during development, test, and operation. This work provides a basis for the development of future numerical models and enables the next generation of subsurface photonic tools.

Intelligent element: Coupling Green function approach and artificial intelligence to reduce discretization effort

“…[42][43][44][45][46][47][48] In these works, neural networks are used to build constitutive models based on laboratory tests. 49,50 Only one research work was found that proposes a numerical method based on machine learning. 51 In this work, a model reduction is proposed to predict displacements and stresses combining "proper orthogonal decomposition" (POD) and "multi-output support vector machine" (MSVM).…”

“…By using this combination, the search found papers that study the training of neural networks to perform inverse analysis, 39–41 which uses in loco measurements to build meta models for the massive 42–48 . In these works, neural networks are used to build constitutive models based on laboratory tests 49,50 . Only one research work was found that proposes a numerical method based on machine learning 51 .…”