Productivity Model Of Fractured Horizontal Wells In Tight Reservoirs And Application Study

Due to bad physical properties and low permeability in tight reservoirs,there is no effective productivity by vertical well,therefore,multiple-fractured horizontal wells(MFHWs)are usually used for the development of tight reservoirs.The seepage flow characteristics and development methods of tight reservoirs are quite different from those of conventional reservoirs,which makes it difficult to accurately predict the productivity of tight reservoirs.Therefore,it is of great significance to study the productivity prediction of MFHWs in tight reservoirs.On the basis of research on productivity prediction of tight reservoirs,aiming at the shortcomings of research,the following research has been carried out and some knowledge has been obtained.Based on Green's function and superposition principle,a productivity model of MFHW in tight reservoir is established,which takes into account the irregular shape of artificial fractures,stress sensitivity of artificial fractures and fracture network,and wellbore friction,etc.Then,the model is solved by using Matlab programming,and the model is verified.Furthermore,by using the proposed productivity model,the distribution characteristics of confluence flow rate along fracture,main control factors of productivity and reasonable production pressure difference are analyzed,which provides a basis for the development of tight reservoirs.In order to improve the field applicability of the model,based on the stress-sensitive theory of fracture,the Eclipse Plugin(a productivity prediction plug-in based on Eclipse platform)is programmed and implemented,and the productivity numerical model that can describe the whole production process of MFHW is established.Then,the case calculation of JHW023 well is carried out.Firstly,the grid size is optimized and designed.Then,the history of actual production data is fitted to obtain the corresponding stress sensitive parameters of the reservoir.Finally,the results of production performance,pressure and permeability in the next three years are predicted.The numerical model can be applied to predict productivity in oil fields,which increases the applicability of the model.Based on field data,combined with machine learning and artificial intelligence algorithm,researches on the main control factors and the establishment and improvement of productivity model are carried out.Principal Component Analysis(PCA)combined with K-means clustering method is applied to distinguish production characteristics of production wells.Then,by the analysis of Biplot diagram,the main factors affecting high production of oil wells are determined intuitively,such as number of stages,horizontal well length,fracturing fluid volume,proppant volume,nozzle diameter.Furthermore,support vector machine(SVM)and artificial neural network(ANN)are used predict productivity in tight reservoirs.From the prediction results,the SVR method is more accurate than the ANN method,but it still needs to be improved.Finally,the accuracy of the prediction model is improved by feature selection.The research results of this paper provide a reference for productivity prediction of MFHWs in tight reservoirs,analysis of the main control factors,determination of reasonable production system and analysis and application of field data.