Font Size: a A A

Study On Morphology Evolution And Neural Network Prediction Model Of Hydraulic Fracture Network In Shale With Natural Fracture

Posted on:2022-01-27Degree:MasterType:Thesis
Country:ChinaCandidate:M J FengFull Text:PDF
GTID:2481306536465424Subject:Safety science and engineering
Abstract/Summary:PDF Full Text Request
With the problem of insufficient energy supply around the world,shale gas has been vigorously developed as a new clean energy source.However,due to the complex geological environment,hydraulic fracturing technology may cause a series of environmental problems,such as induced microearthquakes,groundwater pollution in shale gas production,among which the prapagation of hydraulic fracture network is the key to these problems.Natural fractures and layered discontinuities in shale reservoirs are well developed,leading to complex fracture propagation laws.Moreover,it is difficult to establish an efficient prediction model for fracture morphology due to the numerous and coupling factors affecting fracture propagation.Therefore,in this paper,the morphologic evolution and intelligent prediction model of hydraulic fracture network in shale with natural fractures are studied.The interaction mechanism between hydraulic fractures and natural fractures is firstly revealed through theoretical analysis,and a mathematical model of fracture mode at the intersection point of hydraulic fractures and natural fractures considering the plastic zone at the tip of hydraulic fractures is established.Then,based on Well Jiaoye 194-x of Fuling shale gas field,a series of numerical studies on fracture network propagation law of shale reservoir embedded with natural fractures(DFN)and interfaces under the influence of geological and operation parameters at field scale are conducted.Finally,the neural network prediction model of fracture network morphology under the influence of multiple factors is established,and the ranking of the influence importance of each factor is obtained by using the forward selection method.The results of the study are as following:(1)The failure mode is mainly affected by approaching angle,stress difference and strength difference of rock matrix and discontinuities when hydraulic fractures meet natural fractures.When the stress difference and approaching angle are small,tensile failure on natural fractures is more likely to occur.With the increase of stress difference and approaching angle,it is gradually transformed into shear failure on natural fractures,and then into hydraulic fractures crossing natural fractures directly.The smaller the strength difference is,the smaller the critical stress difference and approaching angle for fracture mode transformation are,and vice versa.(2)Natural fractures and interfaces are the prerequisites for the formation of complex network.The fracture complexity will increase when the dip angle of natural fractures is greater than 20°and when the dip angle of interfaces is greater than 30°.The greater the density of natural fractures is,the greater the complexity of fracture network is.The failure of natural fractures is mainly caused by shear failure,and its mechanical strength mainly affects the proportion of failure modes,but little effect on fracture morphology and complexity.The complexity of fracture network of X-type fractures is larger than that of orthogonal type.Larger horizontal stress ratio is beneficial to the formation of fracture networks with higher complexity.High viscosity and high flow rate will reduce the leak-off and form fracture network with wider range and greater complexity.(3)A 5-layer hidden layer prediction model(including 6 weight matrices and 6deviation matrices)with 9 input parameters(inclination of natural fracture to the direction of maximum horizontal principal stress,density of natural fracture group,etc.)and 9 output parametersis(fracture propagation range,fracture area,fracture volume,control volume,etc.)is established.The R~2 range between the prediction results of fracture network morphology by using the established model and the real value is0.81?1.The important degree of the influence factors is the permeability of matrix,fluid viscosity,the density of natural fracture,the dip angle of natural fracture,fluid rate,the type of natural fracture,horizontal stress ratio,the cohesion of natural fracture,the tension of natural fracture.
Keywords/Search Tags:Shale, Hydraulic fracturing, Fracture network propagation, Numerical study, Neural network model
PDF Full Text Request
Related items