Research On Production Forecast Method Of Gas Well In Tight Gas Reservoir Based On Big Data Analysis

During development process of tight gas reservoirs,accurate prediction of gas well production and rational allocation of production are of great significance for the prediction and evaluation of gas field development trends.In the actual production process,gas well production is affected by various factors.Limited by the mathematical model,the traditional production decline method considers the influencing factors to be limited,and the gas well production prediction results have poor applicability.With the development of intelligent gas field and the development of computer big data analysis technology,the method of big data analysis for gas well production prediction has a good prospect.This paper based on the analysis of the influencing factors of gas production in the west of Sichuan,using the big data analysis method established a gas production prediction model.The model was used to predict production and achieved good prediction results.In this paper,the works accomplished are:(1)The classification of dynamic characteristics of gas production in Xinchang Shaximiao gas reservoir was carried out.Classification of dynamic characteristics of gas production by dynamic and static combination method.The characteristics and distribution characteristics of different types of gas wells were determined.(2)The factors affecting the production of gas were analyzed.Using statistical methods,the influence of static geological parameters of Xinchang Shaximiao gas reservoir on gas productivity is analyzed,which provides a basis for the setting of neural network input parameters.(3)Using reservoir static parameters as static constraints,wellhead oil pressure,water-gas ratio and remaining recoverable reserves as dynamic constraints,the gas production prediction model was constructed using recurrent neural network and long-short time memory neural network.The applicable conditions of the two production prediction models are analyzed,and the internal parameters of the recurrent neural network and the long-short-time memory neural network are optimized.(4)Taking the gas production as the dynamic constraint condition and the reservoir static parameter as the static constraint condition,the support vector machine method is used to establish the single well controlled reserve calculation model.The single well production data and geological parameters are input,and the relationship between the well control and the single well control reserves is established by the support vector machine,and the well control reserves of the gas well can be calculated.(5)Based on the gas production prediction model and the well control reserve calculation model,the production predictions of three different types of gas wells in the Xinchang gas field in western Sichuan are compared with the traditional production decline analysis model,and good prediction results are obtained.This paper introduces the big data analysis method into the gas well productivity prediction research in the petroleum engineering field,and has achieved good results for the tight gas reservoir production forecast in western Sichuan,which can be further promoted and applied.