Heterogeneity Evaluation Of Oil Shales Containing Mud

Logging method to evaluate organic carbon content and reservoir heterogeneity of an economic, efficient, high vertical resolution characteristics, value and promote the potential is enormous. To take advantage of shale mud logging data to evaluate the first to grasp the corresponding geophysical logging features, and understand with conventional sandstone reservoirs are essentially different reservoir space and reservoir characteristics. In this paper, a detailed analysis of shale reservoir characteristics is provided, summed up its salient features different from conventional reservoirs: ① rich in kerogen shale reservoirs and other organic matter; ② shale oil and gas reservoirs have a complex mineral composition and strong non-flat quality; ③ shale reservoir permeability is extremely low, poor relationship between porosity and permeability; ④ shale reservoirs have diverse types of reservoir space; ⑤ shale rock types, vertical heterogeneity obvious; ⑥ fluid occurrence way diverse. This article describes the use of logging data to calculate various organic and inorganic carbon content heterogeneity evaluation of organic, evaluation method include the wave amplitude difference method and new method based on artificial neural network; inorganic heterogeneity evaluation methods are empirical approach, the whole volume of the model law and multi-mineral volume model law, the text of the principles and procedures of each method are introduced, and simple application. In this paper, the Δlog R methods and artificial neural network is improved for Bonan Sag practical application, analysis and comparison of the applicability and the advantages and disadvantages of each method. Select a representative sample of 162 measuring points, where 149 points are used to model RBF neural network training and Δlog R law established, with the remaining 13 points and Δlog R RBF neural network model to evaluate the effects of validation. Comparative results show, RBF neural network and Δlog R method of single well logging evaluation process modeling and verification of TOC have better results, when the linear relationship between the sonic or resistivity curves and TOC is not obvious, RBF neural network is better than Δlog R law; RBF neural network of logging evaluation for S1 has a good effect, which can meet the requirements of the application of extrapolation, this is important for shale oil evaluation. This article also utilize inorganic Luo 69 well data modeling, and in the longitudinal direction of the mineral content to continue calculations, the effect is more ideal for the subsequent admissibility pave the way for evaluations.