Use of Artificial Intelligence to Quantify Geomechanical Uncertainties Induced by Fluid Injection in Subsurface Systems
Injecting fluid underground for energy production (hydrocarbon, geothermal) or storage purpose (CO2, gas, waste) can impact heavily the stress equilibrium of the underground, resulting in geomechanical phenomena like induced fracturing or fault reactivation. When they are not controlled such phenomena might jeopardy the energy or storage process. Unfortunately the computational cost to model and simulate those phenomena is large, making a Monte Carlo type quantification of uncertainty intractable.
The PhD aims at using Artificial Intelligence (AI) techniques to rapidly quantify geomechanical risks and uncertainties. By embedding some of the system’s physics into the AI algorithm, the predictions should more realistically reproduce the essential mechanics of fracture development than a traditional AI approach would do. Besides It should also reduce the computational cost. Last such workflow might enable a more robust and more interpretable quantification of geomechanical uncertainty so as to enhance the decision making process.
Dr Helen Lewis
Professor Vasily Demyanov
Dr Dan Arnold