Multi-model Predictive System for Solubility of CH4 and H2S in Water at Shale Gas Sites

Technological advancements in horizontal drilling and hydraulic fracturing make extraction of natural gas from shale formations economically feasible. However, those activities may induce environmental risks associated with regional water quality due to migration of gases like CH4 and H2S through fractures and accidental spills. Thus, predicting the solubility of these gases in different aqueous conditions relevant to typical subsurface conditions is important. Nine models, including equations of state and empirical models, for predicting CH4 solubility in aqueous phases and six models for H2S are considered and evaluated. The goal of this study was to develop an integrated predictive system for each of these gases for a range of ionic strengths varying from freshwater, saline water and brine conditions over a temperature range of 298 - 483 K and a pressure range of 1 - 350 bars. The predictive accuracy of each model varies with different aqueous conditions. A variance - based weighted model is developed to predict the solubility of the two gases under different subsurface conditions (i.e., temperature, pressure and salt concentration (T-P-X)), and the performance of the weighted model is compared to the best fitting individual model in each case. Predicted and observed values are compared using a 5 - fold cross validation. Cases for which the weighted model outperforms the best predictive model for each of the two gases are identified and discussed. The modeling approach increases the predictive accuracy of CH4 and H2S solubility across the subsurface T-P-X conditions likely to be encountered at shale gas extraction sites.