【作者】 周宇；

【导师】 魏国齐； 郭和坤；

【作者基本信息】 中国科学院研究生院（渗流流体力学研究所） ， 流体力学， 2011， 博士

【摘要】 岩心分析实验是认识油气层地质特征的必要手段。近年来,以火山岩、砂砾岩、泥岩和碳酸盐岩等特殊岩性为主要储层的油气勘探不断取得新突破,特殊岩性储层的岩心分析工作变得越来越重要。这些特殊岩性的化学成分复杂,物理状态多样,导致常规分析方法在分析特殊岩性岩心时有很大的局限性。因此,为了获取准确的实验数据,正确认识特殊岩性储层的渗流规律,本文通过应用CT和核磁共振这两种现代化岩心分析手段,结合人工智能方法与计算机三维可视化技术,开发了一套先进的特殊岩性岩心实验分析方法。并应用这套方法编写了相应的软件,完成了大量的岩心分析工作,取得了良好的效果。通过应用CT技术对岩心内部结构进行无损检测、建立特殊岩性非均质性特征CT分析评价指标、进行岩石CT图像可视化研究和相应的软件开发,实现了对岩样内孔洞和裂缝等的发育特征、发育程度以及岩石非均质性特征的分析。CT分析评价指标包括定性和定量评价指标,指标中给出了相应的评判标准和计算公式,为特殊岩性CT分析提供了依据。CT图像可视化研究采用了三维体素模型、Freeman链码、改进的Marching Cubes算法和视觉增强等多种图形学算法,并在此基础上开发了专业软件,为特殊岩性岩CT分析提供了必要的手段。应用上述方法对211块特殊岩性岩心进行CT实验分析的结果表明：1)砾岩中普遍发育有砾缘缝,其的发育程度与砾岩的砾级成正比,与压实程度成反比；砾岩中的孔洞主要是砾间溶孔和颗粒溶孔,其发育程度与胶结物的成分和压实作用有关。随着砾级的降低,孔洞和裂缝的数量下降。2)在多数火山岩内均有较多的孔洞发育,且存在较大的洞和较多的孔；有个别岩心内可见少量微裂缝,极少数岩心内可见裂缝或较多微裂缝。不同类别的火山岩孔洞和裂缝发育程度相当,但是类型略有不同。针对特殊岩性非均质性强,孔隙空间结构复杂的特点,提出了特殊岩性物性参数核磁共振检测方法,并开发了相应的岩心分析专家系统。方法包括：全直径岩心核磁共振束缚水饱和度测量、人工智能核磁孔隙度校准和遗传算法与神经网络相结合的渗透率预测模型。这些新方法实现了特殊岩性全直径岩心物性参数的检测,解决了常规分析方法测试周期长、样本难以获取以及可能产生黏性指进等问题。人工智能技术的应用在一定程度上降低了特殊岩性复杂物理化学性质对核磁共振测量的影响,将原有核磁孔隙度实验误差降低了50%以上,渗透率预测精度提升了3倍,并且具有更高的自动化程度。岩心分析专家系统采用最先进的B/S架构,通过在服务器上建立适用于岩心分析的决策树、神经网络和遗传算法等几种人工智能算法和多种核磁共振反演算法,达到了对实验结果进行校准和对储层参数的进行预测和评价的目的。这套系统填补了国内核磁共振岩样分析数据解释专家系统的空白,成为目前特殊岩性储层岩心的物性参数测量分析最有效的工具。本文通过以上研究,开创了新的实验手段和实验数据分析方法,提高了特殊岩性岩心分析水平。从而提升了人们对特殊岩性油气储层的科学认识,对推动油气田勘探开发具有重要意义。更多还原

【Abstract】 Core analysis is one of the most necessary means to recognize the geological characteristics of reservoirs. Recently, the oil and gas exploration of particular lithology such as volcanic, conglomerate, mud and carbonate rocks continued to make new breakthroughs. As a result, core analysis of particular lithology reservoir rock is increasingly significant. However, conventional methods of core analyzing have great limitations on particular lithology due to the complex chemical components and variable physical status. In order to obtain accurate experimental data and understand the seepage law of particular lithology, advanced core analysis methods are developed through the application of two modern core analysis tool, computed tomography (CT) and nuclear magnetic resonance (NMR), and two data analysis methods, artificial intelligent (AI) and computer visualization (CV). Corresponding software are developed and applied to core analysis works bases on these methods, the application has archived good results.Through the application of nondestructive CT testing for inner construction of core, the establishing of CT evaluation criterion for the heterogeneity of particular lithology, the research of CT visualization technology for core, and the development of CT analyzing software, the goals of analyzing the features and growth of holes and fractures in core and heterogeneity of core are archived. The evaluation criterion is composed by qualitative evaluation and quantitative evaluation, and gives the corresponding evaluation method and formula. The evaluation criterion provides the basis for CT analysis of particular lithology. The research of CT visualization technology for core used 3D cube model, Freeman chain, improved Marching Cubes algorithms, visual enhancements and many other graphics algorithms. Software developed based on these algorithms provides necessary means for CT analysis of particular lithology.The application of CT analysis for 211 particular lithology cores shows:1) Most conglomerates develops gravel-edge fractures, growth of the fracture is proportional to the level of gravel and inverse proportional to the degree of compaction. Holes in conglomerate are mainly corroding holes between gravel and corrode gravel holes. The growth of holes is related to the composition of cement and compaction. With the lower gravel level, the number of holes and fractures decline.2) Most volcanic rocks develop holes, some of them develop big holes or lots of holes. Several volcanic samples contain micro-fractures. Very few contain big fractures or many micro-fractures. The growth of holes and fractures in different volcanic rock categories are equivalent, but the type of holes and fractures are different.For the strong heterogeneity and complex pore structure of particular Iithology, methods using NMR for physical parameter testing are proposed, and a corresponding expert system is developed. These methods contain irreducible water saturation measurement for whole core, the NMR porosity measurement refined by AI algorithms, and the NMR permeability estimation model using artificial neural network based on genetic algorithm. These new methods achieved the goal of physical parameters detection for full-diameter and special Iithology core, and solved the defacts of conventional methods, such as size and shape restricts, long test cycles and viscous fingering. The use of AI algorithms reduced the influence of complex chemical components and variable physical status to NMR testing. With higher degree of automation, the new methods successfully reduced the average relative error of NMR porosity by 50%, and increased the accuracy of NMR permeability by 3 times. The expert system uses B/S structure, and achieved on experimental results calibration and reservoir parameters prediction and evaluation. This system filled a blank of NMR data analysis expert system of core samples in China, and has becomes the most efficient tool for physical parameter measurement of particular Iithology.This research has opened new experimental means and new data analysis methods for particular Iithology cores. By increasing the skills of core analysis, this research has increased people’s knowledge of particular Iithology reservoir, and means a lot to oil and gas exploration and development.更多还原