No abstract

The CO 2 −brine interfacial tension (IFT) is a key parameter affecting the CO 2 storage capacity in saline aquifers and therefore should be accurately characterized to ensure the optimal design of CO 2 sequestration projects. This paper proposed the use of the extreme gradient boosting (XGBoost) trees for the fast and accurate modeling of the CO 2 −brine IFT. Results show that the novel model is capable of not only estimating the IFT but also reproducing the underlying correlation between the IFT and each input variable with remarkably high accuracies. Statistical matrices and point-wise error analyses demonstrate that the new model outperforms previous machine learning (ML) methods significantly. The estimation model was then applied for determining the optimum CO 2 sequestration depth in saline aquifers, which reveals that higher pressure and/or lower geothermal gradients result in a significant increase in the maximum structural trapping capacity that occurs at noticeably shallower formations.

Gas flooding has proven to be a promising method of enhanced oil recovery (EOR) for mature water-flooding reservoirs. The determination of optimal well control parameters is an essential step for proper and economic development of underground hydrocarbon resources using gas injection. Generally, the optimization of well control parameters in gas flooding requires the use of compositional numerical simulation for forecasting the production dynamics, which is computationally expensive and time-consuming. This paper proposes the use of a deep long-short-term memory neural network (Deep-LSTM) as a proxy model for a compositional numerical simulator in order to accelerate the optimization speed. The Deep-LSTM model was integrated with the classical covariance matrix adaptive evolutionary (CMA-ES) algorithm to conduct well injection and production optimization in gas flooding. The proposed method was applied in the Baoshaceng reservoir of the Tarim oilfield, and shows comparable accuracy (with an error of less than 3%) but significantly improved efficiency (reduced computational duration of ~90%) against the conventional numerical simulation method.

No abstract