含油气泥页岩层非均质性评价方法研究

发布时间：2018-02-03 13:01

本文关键词： 泥页岩储层有机质非均质性 ΔlogR方法人工神经网络(RBF)　出处：《东北石油大学》2015年硕士论文　论文类型：学位论文

【摘要】：测井方法评价有机碳含量及储层非均质性具有经济、快捷、纵向分辨率高的特点,应用价值和推广潜力极大。要利用测井资料对泥页岩层进行评价首先要掌握其地球物理测井相应特征,并了解其与常规砂岩储层有本质区别的储层空间及储集特性。本文对页岩储层特征进行细致分析,总结出其区别于常规储层的显著特征:①页岩储层中富含干酪根等有机质;②页岩油气储层中矿物组分复杂,非均质性强;③页岩储层渗透率极低,孔渗关系差;④页岩储层储集空间类型多样;⑤泥页岩岩石类型多样,纵向非均质性明显;⑥流体赋存方式多样。本文介绍了多种利用测井资料计算有机碳含量及无机非均质性评价的方法,有机非均质性评价方法包括声波电阻率幅度差法及新方法人工神经网络法;无机非均质性评价方法有经验统计法、全体积模型法及多矿物体积模型法,文中对各方法的原理及步骤进行介绍,并进行简单的应用。本文利用改进的Δlog R方法和人工神经网络法对渤南洼陷进行实际应用,分析对比各方法的适用性与优缺点。在实测点中选取162个代表性样本,其中149个点用于RBF神经网络训练和Δlog R法的模型建立,用其余13个点对RBF神经网络和Δlog R模型评价效果进行验证。对比研究结果表明,RBF神经网络法和Δlog R法在对单井进行测井评价TOC的建模和验证过程中均有较好的效果,当声波或电阻率曲线与TOC的线性关系不明显时,RBF神经网络法优于Δlog R法;RBF神经网络法对于S1的测井评价具有较好效果,基本可以达到外推应用的要求,对于泥页岩含油性的评价具有重要意义。本文还利用罗69井数据进行无机建模,并在纵向上对矿物含量进行连续计算,效果较理想,为后续可采性的评价工作做铺垫。
[Abstract]:The evaluation of organic carbon content and reservoir heterogeneity by logging method has the characteristics of economy, rapidity and high vertical resolution. The application value and popularization potential are great. In order to evaluate shale formation with logging data, it is necessary to master the corresponding geophysical logging characteristics. And understand the reservoir space and reservoir characteristics which are essentially different from the conventional sandstone reservoir. This paper makes a detailed analysis of the shale reservoir characteristics. It is concluded that there are abundant organic matter such as kerogen in the Wei 1 shale reservoir, which is different from the conventional reservoir. (2) the mineral composition in shale oil and gas reservoir is complex and heterogeneity is strong; (3) the permeability of shale reservoir is extremely low and the relationship between pore and permeability is poor; (4) the reservoir space types of shale reservoir are various; (5) the types of shale rocks are various and the longitudinal heterogeneity is obvious; (6) there are many ways to store fluids. This paper introduces a variety of methods for evaluating organic carbon content and inorganic heterogeneity by using logging data. The evaluation methods of organic heterogeneity include acoustic resistivity amplitude difference method and artificial neural network method. The evaluation methods of inorganic heterogeneity include empirical statistical method, full-volume model method and multi-mineral volume model method. The principles and steps of each method are introduced in this paper. In this paper, the modified 螖 log R method and artificial neural network method are applied to Bonan sag. 162 representative samples were selected from the measured points, 149 of which were used for RBF neural network training and 螖 log R modeling. The evaluation results of RBF neural network and 螖 log R model were verified with the other 13 points. RBF neural network method and 螖 log R method have good results in the modeling and verification of TOC for single well logging evaluation, when the linear relationship between acoustic or resistivity curves and TOC is not obvious. RBF neural network method is superior to 螖 log R method. RBF neural network method has a good effect on the logging evaluation of S1, and can basically meet the requirements of extrapolation. It is of great significance to evaluate the oil content of shale. This paper also uses the data of Luo 69 well to build inorganic model and calculate the mineral content continuously in longitudinal. The result is satisfactory. For the follow-up evaluation of admissibility to do the groundwork.
【学位授予单位】：东北石油大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：P618.13

【参考文献】