| Petroleum is an important non-renewable strategic resource with the reputation of"Blood of Industrial".China is a major oil consumer and importer.At present,the proven recoverable quantity of domestic crude oil is limited.Most of the main oilfields have entered the late stage of stable production,and it is difficult to start production in the new area.However,the rapid growth of oil consumption has led to prominent conflicts between supply and demand.In 2019,China's oil consumption was nearly 700 million tons,and its external dependence was as high as 72%.For many years,major oil exporting countries such as the United States and Saudi Arabia have held the right to speak about oil prices,and the international shipping lanes have been frequently deterred.Therefore,it is of great practical significance to increase China's own oil supply and reduce its external dependence.In this context,tertiary oil recovery technology has received widespread attention.Tertiary oil recovery is a type of technology that uses chemical flooding,gas flooding,thermal flooding or microbial oil recovery to improve crude oil recovery.For the low-permeability and ultra-low-permeability oil fields that prevail in China,gas flooding is a method of tertiary recovery commonly used to increase crude oil recovery and extend the life of the oil field.According to different displacement principles,gas flooding is divided into two types:miscible and immiscible.The theory and experiments show that the miscible displacement efficiency is much higher than the immiscible displacement.The National Oil and Gas Resources Exploration and Production Report issued in 2019 states that China's investment in crude oil exploration and production has continued to increase in recent years,and gas injection miscible flooding enhanced recovery technologies will support and lead the efficient development of oil fields.The guidance outlines the important status of miscible gas flooding.Among the flooding agents for gas flooding,CO2 has a lower critical temperature and pressure,and it is easy to form a miscible state under the flooding conditions.It is easy to obtain and recyclable,and its permanent storage in the crust can also reduce the impact of the greenhouse effect.At present,CO2 enhanced oil recovery(CO2-EOR)technology has become a widely used tertiary oil recovery technology in the industry.Minimum Miscible Pressure(MMP)is a key parameter to measure the miscibility of CO2 flooding.Miscible flooding is possible when the displacement pressure is higher than the minimum miscible pressure of the system.Its value is affected by displacement temperature,composition of crude oil and CO2 injection gas.Obtaining the minimum miscible pressure of CO2 and crude oil accurately is very important to improve displacement efficiency,reduce operating costs,and produce social and ecological benefits.This study focuses on the prediction of the minimum miscibility pressure,and the following work has done:To begin with,four types of machine learning-based models were used to simulate and predict the minimum miscibility pressure.Practically,the composition of crude oil varies in different regions,the composition of CO2 injection gas is also different,so the parameters of the minimum miscible pressure are numerous.Among the various methods for determining the minimum miscible pressure,the experimental measurement method is not only complicated in operation,time-consuming,but expensive,the theoretical calculation method is often simpler,faster and more economical.To explore their principles and prediction capabilities of four commonly used machine learning models(neural network analysis,genetic function approximation,multiple linear regression,and partial least squares).In this study,nine minimum miscible pressure parameters such as displacement temperature were selected,and 147 sets of raw data were selected from a large number of literatures.After analysis of outliers,four machine learning-based models were trained and predicted,and cross-validation was adopted to avoid overfitting or fall into local optimum,and then the simulation results were compared with the literature model.The research carried out a detailed analysis of the algorithm principles of the four models,and found that the four models all have good prediction capabilities.Neural network analysis and genetic function approximation models have better prediction results than linear models,and the results obtained by models with similar algorithms are not much different.Finally,the results of the neural network model with the highest accuracy were applied to the sensitivity analysis,and the influence trends and degrees of the influence parameters on the minimum miscible pressure were obtained.Secondly,a novel minimum miscible pressure prediction model using molecular dynamics method was proposed.After exploring four prediction models based on machine learning,this paper innovatively applied the molecular dynamics method to the prediction of the minimum miscible pressure,and developed a statistical model for predicting the minimum miscible pressure of CO2 and crude oil at the molecular scale.The study first established multiple miscible systems with obvious phase interfaces,then proposed the concept of the initial miscibility time,and used a variety of analysis methods:first-order variance,solvent accessible surface area,root mean square deviation,interaction energy,and local magnification for validation.The research further uses the proportion of atoms passing through the initial interface to characterize the miscible state and obtain the minimum miscible pressure of different systems.Finally,the difference between the minimum miscible pressure predicted from CO2 and crude oil,and the relationship between the oilfield reservoir temperature and the minimum miscible pressure were discussed.The comparison of the research results with various empirical models shows that the new model proposed in this study has good prediction ability,and the prediction results are consistent with theoretical analysis and literature conclusions.Finally,the microscopic mechanism of using alcohols to reduce the minimum miscibility pressure was explored.When the actual displacement pressure is insufficient to reach the minimum miscible pressure,it is a feasible way to reduce the minimum miscible pressure,which can be achieved by adding an entrainer to the CO2 injection gas.Yang et al.(2019)found through experiments that alcohols can significantly reduce the minimum miscible pressure of CO2 and crude oil,but the mechanism of action at the molecular level is not clear.In this study,a molecular dynamics method is used,with ethanol as the entrainer.Under the conditions of constant temperature and pressure,multiple simulation systems were established,and which distinguished based on whether ethanol is added or not.The effects of alcohols on CO2 injection gas,the miscible process of CO2 and different crude oil components,and then the minimum miscible pressure were successively explored.The selected crude oil component models include aromatic,cycloalkanes and straight-chain alkanes.Studies have found that ethanol aggregates supercritical CO2 molecules through interactions such as van der Waals and hydrogen bonding,on the one hand,it can increase the viscosity of the CO2 injection gas,making it a stronger driving force for crude oil.On the other hand,it can also reduce the viscosity of crude oil,increasing its mobility and making it easier to extraction.When CO2 comes into contact with crude oil,the volume of crude oil will be further expanded by the addition of ethanol,the displacement front will be enlarged,and the mass transfer efficiency will be increased accordingly,which will increase the solubility and dissolution rate of crude oil to CO2.Therefore,the minimum misible pressure of the system can be reduced.This article aims to provide a deeper understanding of the minimum miscible pressure,and try to serve the application and promotion of CO2 enhanced crude oil recovery technologies better. |
PDF Full Text Request