Identification Of Hydraulic Fracturing Cracks Based On Magnetotelluric Method

Since the 1940 s,hydraulic fracturing technology has been widely used in the development of oil and gas fields in tight formations,and has achieved good results.Correct identification and evaluation of fracture shape parameters(such as size,orientation,etc.)is the basis for predicting reservoir stimulation level.At the same time,it is of great significance for further optimizing well completion strategy and improving resource utilization efficiency.Among the existing crack monitoring methods,electromagnetic monitoring is favored by researchers because it can accurately describe the effective support volume containing proppant distribution.Among them,as a geophysical electromagnetic exploration technology with far-reaching influence in many fields such as mineral resources exploration and groundwater monitoring,magnetotelluric method can not only provide relevant information such as underground resistivity distribution and geological structure,but also has a wide range of detection depth,which can better meet the needs of hydraulic fracturing fracture monitoring.At present,due to the multi-scale problem caused by the special three-dimensional geometric characteristics of hydraulic fractures,this method is rarely studied in this field.Therefore,this paper takes the hydraulic fractures of vertical wells as the research object,takes the principle of magnetotelluric method as the theoretical support,and takes the deep learning model as the identification method.Through the sensitivity analysis of hydraulic fractures and reservoir parameters,a set of magnetotelluric hydraulic fracturing fracture monitoring scheme based on deep learning is proposed.Firstly,this paper briefly reviews the development of magnetotelluric exploration and hydraulic fracturing fracture monitoring technology.Secondly,based on the theory of finite element method,the magnetotelluric response of hydraulic fracturing fracture is forward simulated by using transition boundary conditions,and the correlation characteristics between apparent resistivity,impedance offset and fracture parameters including length,width and azimuth at different frequencies are explored,which provides strong data support for the identification model.Finally,by establishing the fracture shape recognition model based on deep learning,FRN successfully realizes the fracture parameter recognition,and further improves the recognition effect through the introduction of attention mechanism.The experimental results show that the deep learning model proposed in this paper has achieved good application results in the identification of hydraulic fractures in vertical wells.At the same time,the model has certain anti-interference ability,and can still maintain strong recognition ability on the premise of adding background noise.