Evaluation Of Macroscopic Heterogeneity And Development Potential Of Coalbed Methane Reservoir Based On Spatial Stratified Heterogeneity

Coalbed Methane(CBM)is a clean energy source that promotes low-carbon and sustainable energy development through its extraction and utilization.However,there are still many challenges facing CBM development,particularly in high-rank coal reservoirs where macroscopic heterogeneity has a significant impact on the selection and potential of CBM development.Therefore,the evaluation of macroscopic heterogeneity is of great importance for the assessment of coalbed gas reservoirs.Geographical Information System(GIS)has strong data processing and spatial analysis capabilities and,when combined with the theory of spatial stratified heterogeneity(SSH),can be used to describe the distribution characteristics and spatial variability of reservoir heterogeneity,enabling quantitative evaluation of macroscopic heterogeneity in CBM reservoirs.Thus,this thesis evaluates the macroscopic heterogeneity of CBM reservoirs based on GIS and SSH and explores its influence on the potential of CBM development.The main research outcomes are as follows:1)Based on the geological conditions of coal-bed methane reservoirs,the spatial characteristics of reservoir macro inhomogeneity and the influencing factors were analyzed.By collecting and organizing the coal reservoir data in the study area,a dataset of 12 reservoir evaluation parameters including gas content,coal thickness,and reservoir pressure was obtained.At the same time,the reservoir was partitioned according to the hydrodynamic conditions in the study area,and the Geodetector Model(GDM)was used to calculate the reservoir spatial differentiation characteristics of different hydrodynamic partitioning units,and analyze the influence of individual reservoir parameters and reservoir parameters stacked by two on the reservoir spatial differentiation characteristics.Finally,the relationship between different reservoir parameters and CBM production capacity is verified by using the Maximum Information Coefficient(MIC)method,and the main influencing factors of CBM production capacity in different hydrodynamic zoning units are derived.2)A SMOTEENN-XGBoost machine learning model was constructed for predicting the productivity.Through comparison and selection of methods,the SMOTEENN-XGBoost model was developed by combining XGBoost with SMOTEENN under sampling and oversampling techniques.The results showed that the model had an accuracy of 79% and 73% for identifying high-and low-yield wells,respectively,effectively resolving the problem of imbalanced training data.Based on this,coalbed methane productivity was predicted using the machine learning model,expanding the dataset for evaluating reservoir heterogeneity and utilizing the fast and efficient features of machine learning to address issues of missing or imbalanced data,laying a data foundation for fuzzy evaluation of reservoir development potential.3)A coalbed methane development potential evaluation model was developed based on the macroscopic heterogeneity and spatial differentiation characteristics of the coalbed methane reservoir using a data-driven approach.Starting from the macroscopic heterogeneity of the reservoir,the GDM-MIC model was used to quantitatively analyze the spatial differentiation characteristics of the reservoir’s macroscopic heterogeneity.Combining the game theory-based weighted combination model with the fuzzy comprehensive evaluation methodology,the fuzzy evaluation of the reservoir’s development potential was carried out based on the results of machine learning-based productivity prediction.By revealing the mechanism behind the production differences caused by the spatial differentiation of reservoir heterogeneity through a data-driven research approach,this study provides a reference for coalbed methane exploration and development selection.28 figures,29 tables and 126 references are included in this dissertation.