Comprehensive Sweet Spot Evaluation Of M Volcanic Rock Reservoir Based On Gradient Boosting Regression Algorithm

In recent years,China has made great progress in the exploration and development of volcanic rock reservoirs,and the efficient exploitation of volcanic rock reservoirs has long-term significance.The geological reserves of the M volcanic reservoir are considerable,but there are some problems in the study area,such as low development degree,unclear control factors of reservoir sweet spots,imperfect evaluation method of reservoir sweet spots,and lack of systematic analysis,which lead to high risk of large-scale horizontal well development.From the perspective of data analysis,this paper predicts the distribution of integrated sweet spots in M volcanic reservoir by using machine learning method and comprehensive index evaluation method.Firstly,geological and engineering data of M volcanic reservoir were collected,and various useful data were integrated and cleaned by software.Then,the correlation analysis model between multiple parameters and production were established by using Gradient Boosting Decision Tree(GBDT)algorithm to quantitatively analyze the main controlling factors affecting reservoir production,and the results were taken as the main controlling factors of reservoir sweet spots.By referring to the comprehensive evaluation method of shale gas compressibility,an evaluation model M volcanic oil comprehensive sweet spot index is obtained,which can comprehensively consider reservoir geological sweet spot index and engineering sweet spot index,and further realize the prediction of M volcanic reservoir sweet spot.The result shows that:(1)The correlation analysis model of multi-characteristic parameters and production established by GBDT algorithm can quantitatively evaluate and rank the influencing factors of productivity,and determine the main controlling factors affecting reservoir sweet spot.Geological sweet spot indexes are oil saturation,fracture density and porosity,and engineering sweet spot indexes are brittleness index and horizontal stress difference.(2)The comprehensive sweet spot index model was established according to the weight coefficient of each index parameter,and it was found that the sweet spot index had a good positive correlation with the output of both vertical and horizontal Wells in the working area,and when Fn>0.38,the productivity of the working area was higher.When 0.34<Fn<0.38,the productivity of horizontal Wells in the working area is greatly affected by fracturing operation scale.When Fn<0.34,the well productivity in the working area is low.(3)The sweet spot index Fn>0.38 is used as the basis for the division of the reservoir sweet spots area,and the distribution map of sweet spots in M volcanic reservoir is drawn in layers,and the prediction results can be used as a reference basis for the horizontal well scale development of M volcanic reservoir.