Gas Transport Mechanism And Evaluation Of Production From Shale Reservoirs With Multiscale Pores And Fractures

Due to the increased demand in energy recently,knowledge of mechanics has been predominantly used in providing alternative solutions to the depleting energy resources through providing efficient means of exploiting shale resources,especially the analysis of the solid-fluid interaction.Shale gas reservoirs(SGRs)have intricate fractures which connect with the hydraulic fractures(HFs)to create a channel for the gas stored in the kerogen to migrate to the production well(PW).As the gas migrates through the generated channel,it passes through many scales including the nano scale(kerogen),the microscale(inorganic matrix(IM)),mesoscale(natural fractures(NFs))and finally the macro/reservoir scale(HFs and PW)thus completing a quadruple-domain-transport.Therefore,how to capture the transfer of gas across the various domains of the reservoir and account for the deformation of the shale formation as a result of the transport of gas,is not only essential for the exploitation of SGRs but also fundamental for mechanics,which relates the key scientific problems about rock fracturing,fluid transport,and fluid-solid interactions.Herein,a comprehensive analysis of the key features for mechanics,storage,exchange,and forecast of gas recovery is conducted through numerical simulation using finite element method(FEM),where a set of partial differential equations(PDEs)are implemented in Comsol Multiphysics,using the time-dependent study since most of the variables in this work change over time.The following are the major contributions of this work.1.The storage of gas in the layers of the kerogen was comprehensively investigated.In the first work,the contribution of dissolved gas to transport was revealed through analysis of excess sorption and absolute sorption based on Langmuir and Brunauer-Emmett-Teller(BET)isotherms.The equations were solved using the modules of Darcy law and coefficient form PDE to describe the mechanisms and gas flux,respectively.In the numerical model,the step function for pressure was defined around the PW and the flux terms were treated as the source and sink to describe the transformation from dissolved gas to adsorbed gas and finally to free gas.Simulation results match with experimental results and indicate that the amount of adsorbed gas increases drastically with the increase in pressure,then reduces to a transition stage beyond which any increase in pressure would not influence the quantity of adsorbed gas.This study revealed that pressure,layers,and volume are the major determinants of the storage and transport behavior.2.The matrix-fracture transport of gas was deeply studied through a dual-domain model.To capture the transport behavior of the free gas,herein,a matrix-fracture model(dual-domain)with crisscrossing NFs embedded in the matrix,was developed.The Langmuir isotherm was adopted to describe the behavior of the storage of gas because the pressure and temperature considered were below the supercritical condition,thus,a single layer was suitable for this study.Unlike random fractures,crisscrossing NFs improve the overall exchange and transport of gas by increasing the rate at which the gas migrates from the matrix through the fracture to the PW.Fracture and matrix properties directly influence the recovery of gas with the following order of precedence,fracture length,matrix porosity,matrix permeability,adsorption volume,and number of NFs while the increase in fracture spacing dramatically reduces the recovery of gas.Through this model,the study revealed the strength of analyzing transport at the micro and macro scales in the realm of mechanics.3.The dual-domain model was extended to capture the transport of gas across many scales.Herein,the matrix was divided into the kerogen and IM and the fracture was divided into the NFs and HFs,yielding a fully coupled multiscale model.Using the principle of normally distributed random functions,the NFs were developed in Matlab and exported to the numerical model for coupling.In the framework of FEM,the modules of Darcy law and coefficient form PDE were used to define each of the domains and implement the set of PDEs.During the extraction process,the four domains are interrelated and support the exchange of gas from one domain to another domain,which implies that a quadruple-domain is essential in describing the transport of gas across the different scales of the subsurface.Overall,matrix and fracture properties,especially,the geometry of the fractures,porosity of IM,and porosity of kerogen,improve gas recovery while deformation properties adversely impact recovery.This multi-domain modeling approach is essential in understanding the exchange of gas from the nanoscale to the macroscale and provides efficient scientific and engineering solutions.4.The quadruple-domain model was modified to include the unstimulated region(USR)and capture the bifurcating tendency of the HFs and stress-dependent transport regimes.In the numerical model,the kerogen,IM,and NFs were captured in both the USR and stimulated region(SR).The SR also had bifurcating hydraulic fractures(BHFs)which were generated in Matlab using the Lindenmayer system(L-system).Based on the framework of the L-system,a set of production rules and strings were defined to capture the geometry of the growing fracture,which was later exported to Comsol Multiphysics for further processing.The geometry of the generated BHFs conforms to that of the microseismic events(MSE),experiments,and performs better than the planar HF.To effectively describe the interaction of the matrix and fracture,the geometry of the NFs and BHFs are essential and are useful to estimate the amount of gas migrating at a time.This modified quadruple-domain model provides a foundation for understanding the principle of discrete fractures and stress-dependent transport regimes,which are critical scientific issues pertinent to mechanics.5.The modified quadruple-domain model was used to generate input dataset for forecasting of gas recovery using deep learning technique.As the model increased in complexity through adding many domains,the time for predicting an output from a single well increased because of the prolonged the simulation time.To overcome this challenge,a fast and accurate way of predicting gas recovery using an algorithm of long short-term memory(LSTM)was established.A parametric sweep on four parameters(porosity,permeability,adsorption volume and the pressure of the well)was conducted in Matlab with Comsol to generate the dataset to be used as input.The gas rate predicted by numerical simulation was used as the target output and the adaptive moment estimation(ADAM)was used as the optimizer.The results of forecast by the LSTM algorithm match the results from the numerical simulation and efficiently solve the time-consuming problem of most simulation models.This deep learning algorithm saves extra costs to be spent on looking for expensive data from the field and provides an interface for directly handling data and visualizing the outcome.In summary,all the multiphysics numerical models are tested against either the experiment data or field data and the verified models are applied to various sensitivity investigations.This study introduces an in-depth investigation of the multiscale transport and recovery evaluation of gas from fractured shale reservoirs subjected to deformation due to the transport of gas.Insights gained from this work provide scientific guidelines for exploiting shale reservoirs and extend the knowledge of mechanics to modeling of complex domains.