Study On The Technique Of Log Interpretation And Productivity Prediction For Coalbed Methane Reservoir

Based on the coal bed logging data, the experimental data and the productivity data of a study area in Qinshui Basin, and taking the technique of productivity prediction for coalbed methane reservoir, the log evalutaion, the division of the productivity level, the influencing factors on coalbed methane reservoir productivity, the productivity log forecasting for coalbed methane reservoir and the prediction of the productivity rate dynamic variation of the coalbed methane reservoir are systematically studied.Based on the coal bed logging data and experimental data, the coal rock industrial components are calculated with the methods of the multiple regression analysis model and the genetic optimization model; The coal bed gas content are calculated with the methods of the multiple regression analysis model and the isothermal adsorption model; The coal fracture porosity is calculated with the methods of the dual laterolog iteration model, the dual laterolog numerical simulation model and the simplified archie model. Based on the crack width and the fracture porosity calculated by the dual laterolog, fracture permeability of the coal is obtained by the mesh model. The prediction accuracy of the reservoir parameters of the logging interpretation model are all well interrelated to the experimental test results.With the analysis of the average daily gas productivity and the shape of accumulative gas production curve in the study block, the productivity level of coalbed methane reservoir can be divided into three levels: high(>1000m3/d), middle(300~1000m3/d) and low(<300m3/d). The influence of the parameters obtained by the logging method on the productivity are analyzed And the effect weights of the different parameters on productivity are sorted by the grey correlation analysis method. The weight order from high to low is: gas content, ash content, carbon content, fracture porosity, fracture permeability, thickness and depth.The polynomial index model, the fuzzy comprehensive evaluation model and the particle swarm optimization support vector machine(SVM) model are established to predict the coalbed methane reservoir productivity. Polynomial index model can reflect the productivity in whole by combining the different factors influencing the productivity. The fuzzy comprehensive evaluation model can balance all the influence factors and reflect the differences of all the factors with the method of weighting and scoring. The particle swarm optimization support vector machine model can reflect the complicated relationship between the factors influencing the productivity and the coalbed methane reservoir productivity. The above-mentioned productivity prediction models are all well interrelated to the experimental test results.The Weng Shi model, the grey model and the Elman neural network model are established to predict the productivity rate dynamic variation of the coalbed methane reservoir. And the applicability of the different models is also analyzed. For the coalbed methane reservoir with the characteristics of the obvious gas productivity peak, the low gas productivity fluctuation and the long stable gas productivity, the Weng Shi model can accurately predict the dynamic variation. The grey model can accurately predict the dynamic variation in the decline stage with the characteristics of the obvious gas productivity peak, the low gas productivity fluctuation and the long stable gas productivity. For the coalbed methane reservoir with the different productivity characteristics, the Elman neural network model can accurately predict the productivity rate dynamic variation.