| With the breakthrough development of unconventional oil and gas development technology,unconventional oil and gas have become important strategic replacement resources,and tight oil is one of the most valuable resources for exploration and development of unconventional oil and gas resources.At present,multi-stage horizontal fracturing technology is usually used for depletion development of tight reservoirs.However,there are problems such as rapid decline of single well production,short stable production period and low recovery factor.Refracturing technology is an effective measure to restore single well production and prolong stable production period.However,the optimal design of refracturing in tight reservoirs is still to be further studied.The key problems of refracturing technology are the selection of refracturing Wells,the determination of refracturing timing,and the optimization of refracturing methods and construction parameters.Firstly,considering the particularity of tight oil reservoir and the imbibition,oil change and energy enhancement effects of fracturing fluid,a numerical simulation model for the refracturing of horizontal wells in tight oil was established.and the relationships among geological,completion and dynamic production data and the oil increase by refracturing were analyzed.On this basis,three comprehensive indices of geology,engineering and production were constructed.The influence weights of each comprehensive index on oil increase of refracturing were quantified by XGBoost decision tree model,and the decision indices of well selection for refracturing were constructed comprehensively.Secondly,a learning sample library with geological and engineering parameters and their corresponding refracturing timing as characteristic variables was constructed by combining numerical simulation with field statistics.Through the performance comparison and analysis of artificial neural network,support vector machine and XGBoost algorithm,the algorithm with higher prediction accuracy was selected as the base model,and the model stacking method of ensemble learning was used to carry out the fusion research on the base model,and the prediction method of refracturing timing based on ensemble learning algorithm was established.Finally,numerical simulation method was used to analyze the stimulation potential of the target well with refracturing,and the favorable area for refracturing was identified.Through the analysis of the development effect and the comparison of the stimulation amplitude,the optimal refracturing method was optimized.The sensitivity analysis and orthogonal design test method were used to determine the optimal combination of fracturing parameters under the optimal refracturing method.The results show that the numerical simulation model established is in good agreement with the actual production well data,and the effect of refracturing depends on many factors such as reservoir geology,fracturing completion and dynamic production.There is a good linear relationship between the composite index and the oil increment of refracturing.The weight calculation results show that the main controlling factors affecting the stimulation effect of refracturing are: fracturing completion,dynamic production and reservoir physical properties.In the research of refracturing timing prediction,the support vector machine and XGBoost regression algorithm have better generalization than the artificial neural network algorithm,and the established prediction method can replace the reservoir numerical simulation process,and then predict the refracturing timing quickly and accurately.In the research of refracturing optimization design,the areas far from fractures in the longitudinal direction of the target well and the transverse areas between fractures are the favorable areas for refracturing stimulation.The best refracturing method of target oil well is to increase energy of old fractures and supplement fracturing new fractures.The combination optimization results of refracturing parameters show that the fracturing design method of “close cutting + medium and low conductivity” can effectively increase the production of single well and obtain higher economic benefits. |
