The formation of marine gas hydrates is controlled by gas migration and accumulation from lower sediments and by the conditions of the hydrate stability zone. Permeability and porosity are important factors to evaluate the gas migration capacity and reservoir sealing capacity, and to determine the distribution of hydrates in the stable region. Based on currently available geological data from field measurements in the Shenhu area of Baiyun Sag in the northern South China Sea, numerical simulations were conducted to estimate the influence of heterogeneities in porosity and permeability on the processes of hydrate formation and accumulation. The simulation results show that: (1) The heterogeneity of the hydrate stability zone will affect the methane migration within it and influence the formation and accumulation of hydrates. This is one of the reasons for the formation of heterogeneous hydrates. (2) When the reservoir is layered heterogeneously, stratified differences in gas lateral migration and hydrate formation will occur in the sediment, and the horizontal distribution range of the hydrate in a high porosity and permeability reservoir is wider. (3) To determine the dominant enrichment area of hydrate in a reservoir, we should consider both vertical and lateral conditions of the sedimentary layer, and the spatial coupling configuration relationships among the hydrate stability region, reservoir space and gas migration and drainage conditions should be considered comprehensively. The results are helpful to further understand the rules of hydrate accumulation in the Shenhu area on the northern slope of the South China Sea, and provide some references for future hydrate exploration and the estimation of reserves.

To further understand the interactions of CO₂-brine-rock at geological time scales, in this study, a 1D reactive transport model of CO₂ intrusion into sandstone of the Longtan Formation (P₂l) in the Huangqiao area, China, was constructed based on site-specific data. The simulation time is consistent with the retention time of CO₂ in the Longtan sandstone Formation and is set to 20 Ma. The reactive transport model is calibrated and revised using the measured data for sandstone samples from Well X3 (i.e., the natural analogue). By comparing the simulation results with measured data for the natural analogue, the long-term geochemical reactions are investigated. The simulation results indicate that the brine-rock interactions induced by CO₂ can be roughly divided into two stages. First, susceptible minerals (e.g., chlorite, ankerite, calcite, and feldspar minerals) dissolve rapidly under acidic conditions formed by the dissolution of CO₂. The precipitation of siderite is facilitated by the dissolution of ankerite and chlorite. Smectite-Ca and dawsonite precipitate due to the dissolution of anorthite and albite, respectively. Dawsonite begins to convert into smectite-Na when albite is completely dissolved. As the reactions continue, intermediate products (i.e., illite, smectite-Na, and smectite-Ca) generated in the first stage become the reactants and subsequently react with CO₂ and brine. These three clay minerals are not stable under acidic conditions and transform into kaolinite and paragenetic quartz in the later stage of reaction. Comparing the simulation results of the Base Case with the measured data for the natural analogue and inspired by previous studies, the scour of kaolinite is supposed to have occurred in this region and is considered in the revised model by introducing a coefficient of the scour of kaolinite (i.e., Case 2). The simulation results of Case 2 fit well with the measured data on mineral assemblage, and the trend of the sandstone porosity growth caused by the CO₂-brine-rock reaction is captured by our simulation results. The combination of numerical simulation and natural analogue study indicates that the joint effects of long-term CO₂-brine-rock reactions and scour of kaolinite increase the pore space of the host rock and result in an increase in quartz content in the sandstone.

Thermodynamic models are usually employed to predict formation condition of hydrates. However, these thermodynamic models usually require a large amount of calculations to approach phase equilibrium. Additionally, parameters included in the thermodynamic model need to be calibrated based on the experimental data, which leads to high uncertainties in the predicted results. With the rapid development of artificial intelligence (AI), machine learning as one of sub-discipline has been developed and been widely applied in various research area. In this work, machine learning was innovatively employed to predict the formation condition of natural gas hydrates to overcome the high computation cost and low accuracy. Three data-driven models, Random Forest (RF), Naive Bayes (NB), Support Vector Regression (SVR) were tentatively used to determine the formation condition of hydrate formed by pure and mixed gases. Experimental data reported in previous work were taken to train and test the machine learning models. As a representative thermodynamic model the Chen–Guo (C-G) model was used to analyze the computational efficiency and accuracy of machine learning models. The comparison of results predicted by C-G model and machine learning models with the experimental data indicated that the RF model performed better than the NB and SVR models on both computation speed and accuracy. According to the experimental data, the average AADP calculated by the C-G model is 7.62 times that calculated by the RF model. Meanwhile, the average time costed by the C-G model is 75.65 times that by the RF model. Compared with the other two machine learning models, the RF model is expected to be used in predicting the formation condition of natural gas hydrate under field conditions.