Well Testing And Production Forecast Of Tight Oil And Gas Reservoirs

In the face of the increasingly strong domestic demand for oil and gas resources,as a type of typical unconventional oil and gas resources with good economics,tight oil and gas reservoirs are important supplements to conventional oil and gas resources.However,unconventional oil and gas not only follow different flow mechanisms,but also generally be mined with complex structure wells or large-scale fracturing methods.So the well test analysis and production forecast methods are different from conven-tional oil and gas reservoirs.In this paper,relevant research is carried out based on the characteristics of unconventional oil and gas reservoirs,and the main achievements are as follows:Multi-branched horizontal wells are complex structural wells constructed by hor-izontal well technology and are currently used in the mining of coalbed methane and shale gas.According to the characteristics of the multi-branched horizontal wells,the formula of ground pressure distribution is deduced by Newman method and discussed with the bottom hole pressure curve.The performance problems in the calculation of ground pressure distribution are gradually optimized,and the GPU based general-purpose computing technology based on CUDA framework is introduced,which greatly improves the calculation efficiency.Hydraulic fracturing technology is an important way to improve the production of tight reservoirs.In addition to the popular horizontal well multi-stage fracturing technology,fracturing of traditional vertical wells can also significantly increase the capacity of tight reservoirs.Based on the research methods of multi-layer reservoir well testing and fractured well testing,the well testing model of multi-layer partially fractured vertical well is deduced based on the no-crossover hypothesis and infinite diversion model of fracture.A piece of existing well test data were compared with the introduced model,and the parameters such as the half length of fractures in each layer were successfully explained,which reflected the characteristics of the multi-layer partially fractured vertical well.The traditional production data analysis method of oil and gas reservoirs is a curve analysis method based on parameter regression.In the face of tight oil and gas reservoirs as a type of unconventional oil and gas resources,it is easy to reflect its limitations.Based on the full investigation,it's introduced a statistical machine learning algorithm based on support vector machine in shale gas production forecast.Regression fitting of bottom hole pressure of a single well was performed on the actual production data and also the bottom hole pressure curve of a single well predicted by the multi-well learning model.Among them,the correlation coefficient of regression fitting of a single well is good,while the multi-well model forecast has achieved good results in the data intensive interval.