Prediction method for methane solubility in high-temperature and high-pressure aqueous solutions in ultra-deep drilling

Abstract

During ultra-deep oil and gas exploration and development, overflow can easily occur when drilling in high-temperature (≤473.15 K) and high-pressure (≤150 MPa) reservoirs. The invaded methane gas is more soluble in the drilling fluid under high temperature and pressure, and the concealment is enhanced. Low temperature and pressure at the wellhead cause rapid expansion and precipitation of the dissolved gas, thereby increasing the risk of blowout. Therefore, the safe development of the ultra-deep oil and gas drilling process is critical to accurately determine the solubility of natural gas under alternating high- and low-temperature and high-pressure conditions in ultra-deep drilling wellbores. In this study, considering the presence and absence of hydrates at low temperature, and the conditions of high temperature and high pressure, the phase equilibrium prediction fugacity–fugacity model of methane and the aqueous solution is established. At the same time, the solubility experimental data of existing CH4 in aqueous solution are counted. The prediction model is established according to the particle swarm optimization support vector machine (PSO–SVM) algorithm. The new thermodynamic fugacity–fugacity model can be applied to predict the methane solubility and water content in the gas phase under the following ultra-deep drilling conditions: temperature from 273.15 to 473.15 K, pressure up to 150 MPa, and salinity from 0 to 5.4 mol kg−1. The prediction errors for the presence and absence of hydrate at 273.15–293.15 K and 273.15–473.15 K are within 8%. The calculation error from the PSO–SVM machine learning model for the interactive solubility of methane and water is maintained within 10%, but the prediction error at atmospheric pressure exceeds 10%. The machine learning model strongly depends on abundant statistical experimental data. The prediction of gas dissolution in ultra-deep oil and gas development should be based on thermodynamic models.