基于概率统计反演的储层定量表征方法

发布时间：2018-11-02 08:57

【摘要】：本文主要研究基于概率统计的储层表征及其不确定性定量评价方法与储层建模等相关内容。研究内容涉及储层物性参数反演、储层岩相识别及其不确定性定量评价分析与储层地质统计学建模等方面。本文针对储层表征及建模过程中遇到的相关问题,提出了若干新策略,并结合模型数据与实际资料验证方法的有效性。储层物性信息是储层评价的重要依据,其通常由储层弹性信息通过岩石物理关系反演获得。由于岩石物理模型在数学关系上的复杂性,其反演目标函数通常具有较强的非线性,影响储层物性参数反演精度;此外当弹性参数反演精度较低时,储层含水饱和度信息预测难度通常较大。为此,方法结合蒙特卡洛模拟与智能优化算法,通过在储层物性参数空间大量随机抽样开展岩石物理正演与弹性参数对比分析实现样本优选,采用高斯模型统计储层物性参数后验概率信息。针对储层弹性参数反演精度低导致含水饱和度预测难度较大的问题,通过统计分析储层含水饱和度与孔隙度、泥质含量之间的统计关系,在孔隙度和泥质含量反演结果基础之上利用上述统计关系近一步获取储层含水饱和度信息。实际工区资料应用验证了方法的有效性。方法在储层物性参数反演的基础上开展了储层岩相识别。针对通过地震资料预测储层岩相分布信息不确定性较大的问题,本文采用基于概率统计的多步骤反演方法,在地震资料反演、井震尺度匹配、岩石物理建模以及测井岩相定义等环节建立输入与输出参数之间的概率统计关系,融合各环节概率信息获取储层岩相概率以表征地震岩相识别的不确定性。方法采用属性交绘特征约束反演参数空间以降低概率矩阵的规模,从而提高算法运算效率并降低反演的不确定性。引入熵函数实现地震岩相识别不确定性的定量评价分析,通过求取各环节条件信息约束下的岩相概率及其熵值信息,方法量化分析其不确定性的传递规律与构成特征。针对储层建模预测井间油藏属性不确定性较大的问题,本文将地震岩相识别获取的储层岩相概率作为储层岩相建模的约束信息,采用Tau模型融合测井与地震岩相概率等信息结合地质统计学中的序贯指示模拟方法开展储层岩相建模。通过在建模中引入地震岩相概率信息,储层岩相建模结果的精度和稳定性都获得一定程度的提高,降低了建模的不确定性,为储层精细表征提供了重要参考信息。
[Abstract]:In this paper, the reservoir characterization based on probability statistics and the quantitative evaluation method of uncertainty and reservoir modeling are studied. The research involves inversion of reservoir physical parameters, reservoir lithofacies identification, quantitative evaluation and analysis of uncertainty, and reservoir geostatistical modeling. In view of the problems encountered in reservoir characterization and modeling, some new strategies are proposed in this paper, and the validity of the verification method based on the model data and the actual data is presented. Reservoir physical property information is an important basis for reservoir evaluation, which is usually obtained by inversion of reservoir elastic information through rock physical relations. Because of the complexity of mathematical relation in rock physical model, the inversion objective function usually has strong nonlinearity, which affects the inversion accuracy of reservoir physical parameters. In addition, when the inversion accuracy of elastic parameters is low, it is usually difficult to predict reservoir water saturation information. For this reason, combining Monte Carlo simulation and intelligent optimization algorithm, a large number of random sampling in reservoir physical parameter space is carried out to carry out rock physics forward modeling and elastic parameter contrast analysis to realize sample optimal selection. Gao Si model is used to calculate the posterior probability information of reservoir physical parameters. In view of the difficulty of prediction of water saturation due to the low inversion accuracy of reservoir elastic parameters, the statistical relationship between water saturation and porosity, muddy content of reservoir is analyzed statistically. Based on the inversion results of porosity and muddy content, the information of reservoir water saturation is obtained by using the above statistical relation. The effectiveness of the method is verified by the application of practical area data. Methods based on the inversion of reservoir physical parameters, reservoir lithofacies identification was carried out. In order to solve the problem of uncertainty in predicting reservoir lithofacies distribution information by seismic data, a multi-step inversion method based on probability and statistics is used in this paper. The relationship between the input and output parameters is established in the links of rock physical modeling and log lithofacies definition. The reservoir lithofacies probability is obtained by combining the probabilistic information of each link to characterize the uncertainty of seismic lithofacies identification. In order to reduce the scale of probability matrix, attribute mapping feature constrained inversion parameter space is used to improve the efficiency of the algorithm and reduce the uncertainty of inversion. The entropy function is introduced to quantitatively evaluate and analyze the uncertainty of seismic lithofacies identification. By obtaining the probability and entropy information of lithofacies under the constraint of each link condition information, the transmission law and composition characteristics of uncertainty are quantitatively analyzed. In this paper, the probability of reservoir lithofacies obtained by seismic lithofacies identification is regarded as the constraint information of reservoir lithofacies modeling. Reservoir lithofacies modeling is carried out by using Tau model fusion logging and seismic lithofacies probability information combined with sequential indicator simulation method in geostatistics. By introducing seismic lithofacies probability information into the modeling, the accuracy and stability of reservoir lithofacies modeling results are improved to a certain extent, and the uncertainty of modeling is reduced, which provides important reference information for reservoir fine characterization.
【学位授予单位】：中国石油大学(北京)
【学位级别】：博士
【学位授予年份】：2016
【分类号】：P618.13

【相似文献】