WO2024131160A1 - 储层分类方法及装置 - Google Patents

储层分类方法及装置 Download PDF

Info

Publication number
WO2024131160A1
WO2024131160A1 PCT/CN2023/118717 CN2023118717W WO2024131160A1 WO 2024131160 A1 WO2024131160 A1 WO 2024131160A1 CN 2023118717 W CN2023118717 W CN 2023118717W WO 2024131160 A1 WO2024131160 A1 WO 2024131160A1
Authority
WO
WIPO (PCT)
Prior art keywords
logging
parameters
reservoir
principal component
parameter set
Prior art date
Application number
PCT/CN2023/118717
Other languages
English (en)
French (fr)
Inventor
刘建宇
王国栋
仇庭聪
张寒
潘树新
曲永强
许多年
马永平
王彦君
郭娟娟
马德龙
关新
杨超
Original Assignee
中国石油天然气股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 中国石油天然气股份有限公司 filed Critical 中国石油天然气股份有限公司
Publication of WO2024131160A1 publication Critical patent/WO2024131160A1/zh

Links

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
    • G01V11/00—Prospecting or detecting by methods combining techniques covered by two or more of main groups G01V1/00 - G01V9/00
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/10—Pre-processing; Data cleansing
    • G06F18/15—Statistical pre-processing, e.g. techniques for normalisation or restoring missing data
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
    • G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques

Definitions

  • the present invention relates to the technical field of petroleum exploration, and in particular to a reservoir classification method and device.
  • Reservoir fine classification is of great significance to reservoir classification evaluation, sweet spot selection and production capacity prediction.
  • reservoir classification methods mainly include core reservoir classification method, well logging reservoir classification method and geophysical reservoir classification method.
  • well logging reservoir classification method has the advantages of rich data sources, high vertical resolution and continuous well section processing, and is an important method for reservoir fine classification.
  • well logging reservoir classification can be summarized into four categories: (1) Semi-quantitative reservoir classification method based on intersection charts. This method is simple to operate and has a wide range of applications, but its applicability is not strong; (2) Well logging reservoir classification method based on core calibration logging. This method is mainly based on core physical property data and has clear geological significance, but it depends on core data; (3) Well logging reservoir classification method based on multivariate statistics and machine learning algorithms can effectively avoid interference from human factors, is fast and has diverse methods, but the classification results are unclear; (4) Reservoir classification method based on new well logging technology methods. This method carries a large amount of geological information and can evaluate reservoirs more effectively and accurately, but the new well logging technology methods are relatively expensive and have limited application scope.
  • the logging reservoir classification method based on core calibration logging effectively integrates core analysis data and logging data.
  • the conventional logging reservoir classification method based on core calibration logging only relies on simple cutoff or intersection of single or double parameters, which is difficult to meet the needs of fine classification of complex reservoirs.
  • the flow unit method only considers physical property data such as core porosity and permeability
  • the pore structure method only considers core mercury injection data or core nuclear magnetic resonance data.
  • the above conventional logging reservoir classification method based on core calibration logging can achieve good application results in conventional sandstone or carbonate reservoirs, but it is difficult to achieve good application results in complex unconventional reservoirs such as tight sandstone, conglomerate, igneous rock and shale, because such complex unconventional reservoirs often have complex mineral composition, mud content, particle size distribution and pore structure.
  • the purpose of the present invention is to provide a reservoir classification method and device to solve the technical problem that the current reservoir classification method is difficult to achieve good application effect in complex unconventional reservoirs such as tight sandstone, conglomerate, igneous rock and shale.
  • the present invention provides a reservoir classification method, comprising the following steps: obtaining characteristic parameter sets of different depth points of multiple reservoir sections of a well section to be classified; the characteristic parameter sets include a well logging characteristic parameter set and a core characteristic parameter set, the well logging characteristic parameter set includes lithology well logging characteristic parameters, physical well logging characteristic parameters and electrical well logging characteristic parameters, and the core characteristic parameter set includes physical property characteristic parameters, mercury injection characteristic parameters, nuclear magnetic resonance characteristic parameters and particle size characteristic parameters; analyzing the characteristic parameter sets of multiple depth points by principal component analysis, thereby extracting at least one principal component parameter of each depth point; clustering analysis is performed on the principal component parameters of multiple depth points, and the multiple depth points are classified into multiple types; and the reservoir type of the well section to be classified is classified according to the various types of well logging parameters.
  • the invention also provides a reservoir classification device, comprising: a feature parameter set acquisition module, used to acquire feature parameter sets of different depth points of multiple reservoir sections of a well section to be classified; the feature parameter sets include a well logging feature parameter set and a core feature parameter set, the well logging feature parameter set includes lithology well logging feature parameters, physical property well logging feature parameters and electrical well logging feature parameters, and the core feature parameter set includes physical property feature parameters, mercury injection feature parameters, nuclear magnetic resonance feature parameters and particle size feature parameters; a principal component analysis module, used to analyze the feature parameter sets of multiple depth points by principal component analysis, so as to extract at least one principal component parameter of each depth point; a depth point classification module, used to perform cluster analysis on the principal component parameters of multiple depth points, and classify the multiple depth points into multiple types; a reservoir type classification module, used to classify the reservoir type of the well section to be classified according to the various types of well logging parameters.
  • a feature parameter set acquisition module used to acquire feature parameter sets of different depth points of multiple reservoir sections of a
  • the reservoir classification method and device of the present invention fully explores the lithology logging characteristic parameters, physical logging characteristic parameters and electrical logging characteristic parameters at different depths of multiple reservoir sections; as well as physical property characteristic parameters, mercury injection characteristic parameters, nuclear magnetic resonance characteristic parameters and particle size characteristic parameters; and then uses the principal component analysis method to extract at least one principal component parameter, and then clusters the principal component parameters of each core, so that the cores of multiple reservoir sections are divided into multiple types, and then the reservoir types of the well sections to be classified are classified according to the types of the multiple cores, so that the fine classification of complex unconventional reservoirs can be achieved, and a reference for the comprehensive evaluation of complex unconventional reservoirs and the formulation of exploration and development plans is provided, which solves the problem that the logging reservoir classification method based on core calibration logging in the prior art only relies on a single parameter.
  • the technical problem is that simple cutoff or intersection of single or double parameters cannot achieve fine classification of complex reservoirs.
  • FIG1 is a flow chart of a reservoir classification method of the present invention.
  • FIG2 is a cross-plot of core porosity and permeability at multiple depth points in this embodiment
  • FIG3 is a set of high-pressure mercury injection original curves of multiple reservoir sections in this embodiment
  • FIG4 is a collection of NMR T2 spectra of multiple reservoir sections in this embodiment
  • FIG5 is a set of probability distribution curves of core grain sizes of multiple reservoir sections in this embodiment.
  • FIG6 is a cross-plot of first principal component parameters and second principal component parameters at multiple depth points in this embodiment
  • FIG7 is a flow chart of extracting principal component parameters according to the present invention.
  • FIG8 is a flow chart of the present invention for continuously classifying the reservoir type of the well section to be measured
  • FIG. 9 is a diagram showing the reservoir classification effect of a well section in a typical well in this embodiment.
  • the present invention provides a reservoir classification method, comprising the following steps:
  • Step 101 obtaining characteristic parameter sets of different depth points of multiple reservoir sections of the well section to be classified;
  • the characteristic parameter sets include a well logging characteristic parameter set and a core characteristic parameter set
  • the well logging characteristic parameter set includes lithology well logging characteristic parameters, physical well logging characteristic parameters and electrical well logging characteristic parameters
  • the core characteristic parameter set includes physical property characteristic parameters, mercury injection characteristic parameters, nuclear magnetic resonance characteristic parameters and particle size characteristic parameters.
  • the logging curve statistics are used to obtain the logging characteristic parameter sets at different depth points.
  • the lithology logging curve is used to calculate the lithology logging characteristic parameters at each depth point.
  • the lithology logging characteristic parameters include natural gamma logging values and natural potential logging values.
  • the physical property logging curve is used to calculate the physical property logging characteristic parameters at each depth point.
  • the physical property logging characteristic parameters include density logging value, compensated neutron logging value and sonic time difference logging value.
  • the electrical logging characteristic parameters of each depth point are statistically obtained through the electrical logging curve.
  • the electrical logging characteristic parameters include deep resistivity logging value, medium resistivity logging value, and shallow resistivity logging value.
  • the acquisition of the core characteristic parameter set includes: drilling rock samples at different depths of multiple reservoir sections and pre-processing the rock samples into cores that meet the requirements.
  • the pre-processing of the rock samples includes: cutting and grinding the rock samples into standard plunger-shaped cores of 3cm-4cm and 2.5cm in diameter; drying the cores at a temperature of 90 degrees Celsius-110 degrees Celsius for 12 hours-24 hours to remove moisture in the cores; and conducting experiments on each core to obtain the core characteristic parameter set at different depths.
  • the physical property characteristic parameters of the core are obtained by conducting physical property parameter experiments.
  • the physical property characteristic parameters include but are not limited to core porosity and core permeability.
  • the core porosity is measured by injecting helium using the Boyle single chamber method.
  • the core permeability is measured by gas measurement. In this embodiment, the intersection of the core porosity and the core permeability at each depth point is shown in Figure 2.
  • the characteristic parameters of mercury injection at the depth point are obtained.
  • the characteristic parameters of mercury injection include but are not limited to median pressure, median radius, displacement pressure, maximum pore throat radius, mercury withdrawal efficiency, average pore throat radius and maximum mercury injection saturation.
  • the original curves of core high-pressure mercury injection at each depth point are shown in Figure 3.
  • the nuclear magnetic resonance characteristic parameters of the depth points are obtained.
  • the nuclear magnetic resonance characteristic parameters include but are not limited to magnetic porosity, geometric mean, arithmetic mean, mud bound water saturation, capillary bound water saturation and movable water saturation.
  • the core nuclear magnetic resonance T2 (transverse relaxation time) spectrum set at each depth point is shown in Figure 4.
  • the particle size characteristic parameters of the core are obtained.
  • the particle size characteristic parameters include mud content, C value, median particle size, peak value, skewness and sorting coefficient.
  • the probability distribution of core particle size at each depth point is shown in Figure 5.
  • Step 102 Analyze the characteristic parameter sets of multiple depth points by principal component analysis, so as to extract at least one principal component parameter of each core.
  • the multiple principal component parameters can be obtained, and by calculating the contribution rate of each principal component parameter, the multiple principal component parameters are arranged from large to small according to their contribution rate as the first principal component parameter, the second principal component parameter, ..., the nth principal component parameter, where n is equal to the number of data in the characteristic parameter set.
  • the first two principal component parameters i.e., the first principal component parameter and the second principal component parameter, are extracted for each depth point.
  • the first principal component parameter can be extracted, and the third principal component parameter or more principal component parameters can also be extracted.
  • Step 103 Perform cluster analysis on the principal component parameters of the multiple depth points to classify the multiple depth points into multiple types.
  • the K-means clustering method can be used to classify the principal component parameters of multiple depth points, and the multiple depth points can be divided into multiple types.
  • Other clustering analyses in the prior art can also be used for classification.
  • multiple depth points are divided into four types by clustering the first principal component parameters and the second principal component parameters of each depth point.
  • Step 104 Classify the reservoir types of the well sections to be classified according to multiple types.
  • the reservoir classification method of the present invention fully explores the lithology logging characteristic parameters, physical logging characteristic parameters and electrical logging characteristic parameters of different depth points of multiple reservoir sections; as well as physical characteristic parameters, mercury injection characteristic parameters, nuclear magnetic resonance characteristic parameters and particle size characteristic parameters; and then uses the principal component analysis method to extract at least one principal component parameter, and then performs cluster analysis on the principal component parameters of each depth point, so as to classify the depth points of multiple reservoir sections into multiple types, and then classifies the reservoir types of the well sections to be classified according to the types of the multiple depth points, so as to achieve fine classification of complex unconventional reservoirs, provide reference for comprehensive evaluation of complex unconventional reservoirs and formulation of exploration and development plans, and solve the technical problem that the logging reservoir classification method based on core calibration logging in the prior art cannot achieve fine classification of complex reservoirs by relying only on simple cutoff or intersection of single or double parameters.
  • the characteristic parameter set of multiple depth points is analyzed by principal component analysis to extract at least one principal component parameter of each depth point, including the following steps:
  • Step 201 Combine characteristic parameter sets of multiple depth points into a first matrix X 1 , wherein the number of rows n of the first matrix X 1 is equal to the number of data in the core characteristic parameter set, and the number of columns m of the first matrix is equal to the number of depth points.
  • Step 202 Subtract the mean value of each row of data in the first matrix X1 from the data in the row to obtain a second matrix X2 ;
  • Step 203 Calculate the covariance matrix C according to the second matrix X 2 .
  • the relationship between the covariance matrix C and the second matrix X 2 is:
  • Step 204 using singular value decomposition method to obtain multiple eigenvalues and corresponding eigenvectors of the covariance matrix C, and sorting the multiple eigenvectors from large to small according to their corresponding eigenvalues, and respectively using them as row vectors from top to bottom to form a third matrix X 3 ;
  • Step 205 Take the first multiple rows of data in the third matrix X3 to form a fourth matrix X4 ; wherein the number of rows of the fourth matrix X4 is equal to the number of principal component parameters to be extracted for each depth point.
  • the first principal component parameter and the second principal component parameter are selected, so the first row of data and the second row of data in the third matrix X3 are selected.
  • the fourth matrix X 4 is formed.
  • Step 206 Extract the principal component parameters of each depth point according to the fourth matrix X4 and the first matrix X1 .
  • the reservoir type of the well section to be classified is classified according to various types of logging parameters, including the following steps:
  • Step 301 Calculate various types of logging parameters using core calibration logging curves.
  • the logging parameters include one or more of the porosity calculated by logging, the permeability calculated by logging, natural gamma, natural potential, acoustic wave, density, compensated neutron and resistivity.
  • the logging parameters include six parameters, namely, the porosity calculated by logging, the permeability calculated by logging, density (DEN), acoustic wave (AC), compensated neutron (CNL), natural gamma ( ⁇ GR) and natural potential ( ⁇ SP).
  • the statistical table of the four types of logging parameters is as follows:
  • Step 302 Establish various types of discriminant models based on Fisher discriminant analysis method according to various types of logging parameters.
  • F is the discriminant standard of each type; among them, A 1 , A 2 , A 3 , A 4 , A 5 are respectively the unknown coefficients, D 1 , D 2 , D 3 , D 4 are all logging parameters.
  • D 1 , D 2 , D 3 , D 4 are preferably logging sensitive parameters, that is, the logging parameters that are most accurate in distinguishing types.
  • DEN density
  • AC acoustic wave
  • ⁇ GR natural gamma
  • ⁇ SP natural potential
  • the discriminant model of type one is:
  • the discrimination standard F 1 of type 1 is obtained.
  • 16 depth points are type 1, and the discrimination model is used to discriminate the types of the 16 depth points with an accuracy rate of 87.5%.
  • the discriminant model of type 2 is:
  • the discrimination standard F 2 of type 2 is obtained.
  • 38 depth points are type 2
  • the discrimination model is used to discriminate the 38 depth points, with an accuracy rate of 89.4%.
  • the discriminant model of type three is:
  • the discrimination standard F 3 of type three is obtained.
  • 25 depth points are type two, and the discrimination model is used to discriminate the types of the 20 depth points, and the accuracy is The rate is 84.0%.
  • the discriminant model of type 4 is:
  • the discrimination standard F 4 of type 4 is obtained.
  • 10 depth points are type 4, and the discrimination model is used to discriminate the types of the 10 depth points, with an accuracy rate of 80.0%.
  • Step 303 extract the logging sensitivity curve of the well section to be classified, and substitute the sensitive logging curve value of the logging sensitivity curve into various types of discrimination models, so as to realize the continuous classification of the reservoir type of the well section to be classified.
  • the well section to be classified is the well section whose mud content and the porosity calculated by logging meet the effective reservoir standard.
  • the reservoir type discrimination and classification are not performed.
  • the reservoir type at the depth position is the type corresponding to the discrimination model, so that in this way, the reservoir type at each depth position of the logging sensitivity curve can be determined, and the continuous classification of the reservoir type of the well section can be achieved.
  • the well section to be classified is a well section with a depth of about 4090m to 4160m in a typical well.
  • the present invention also provides a reservoir classification device, which can be implemented with reference to the above-mentioned reservoir classification method, and will not be described in detail here.
  • the reservoir classification device of the present invention includes: a feature parameter set acquisition module, which is used to acquire feature parameter sets of different depth points of multiple reservoir sections of the well section to be classified; the feature parameters include a well logging feature parameter set and a core feature parameter set, the well logging feature parameters include lithology logging feature parameters, physical property logging feature parameters and electrical property logging feature parameters, and the core feature parameter set includes physical property feature parameters, mercury injection feature parameters, nuclear magnetic resonance feature parameters and particle size feature parameters.
  • a principal component analysis module is used to analyze the feature parameter sets of multiple depth points by principal component analysis, so as to extract at least one principal component parameter of each depth point;
  • a core classification module is used to perform cluster analysis on the principal component parameters of multiple depth points and classify the multiple depth points into multiple types;
  • a reservoir type classification module is used to According to various types of logging curves, the reservoir types of the well sections to be classified are classified.
  • the reservoir type classification module includes: a logging parameter statistics unit, which is used to use core calibration logging curves to count various types of logging parameters; a discrimination model establishment unit, which is used to establish various types of discrimination models based on the Fisher discriminant analysis method according to various types of logging parameters; a continuous classification unit, which is used to extract the logging sensitivity curve of the well section to be classified, and substitute the sensitive logging curve value of the logging sensitivity curve into various types of discrimination models, thereby realizing continuous classification of the reservoir type of the well section to be classified.

Landscapes

  • Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • General Engineering & Computer Science (AREA)
  • Geophysics (AREA)
  • General Life Sciences & Earth Sciences (AREA)
  • Probability & Statistics with Applications (AREA)
  • Geophysics And Detection Of Objects (AREA)

Abstract

本发明公开了一种储层分类方法及装置，所述储层分类方法包括以下步骤：获取待分类井段的多个储层段的不同深度点的特征参数集；特征参数集包括测井特征参数集以及岩心特征参数集；通过主成分分析法对多个深度点的特征参数集进行分析，从而提取出各个深度点的至少一主成分参数；对多个深度点的主成分参数进行聚类分析，将多个深度点分为多种类型；根据各种类型的测井参数，对待分类井段的储层类型进行分类。本发明可以实现复杂非常规储层的精细分类，对复杂非常规储层的综合评价及勘探开发方案制定提供参考，解决了目前基于岩心刻度测井的测井储层分类方法仅依靠单参数或双参数的简单截止或交会而无法实现复杂储层精细分类的技术问题。

Description

储层分类方法及装置
相关申请
本申请要求专利申请号为202211632985.0、申请日为2022年12月19日、发明名称为“一种储层分类方法及装置”的中国发明专利的优先权。
技术领域
本发明涉及石油勘探技术领域，特别地，有关于一种储层分类方法及装置。
背景技术
本部分旨在为权利要求书中陈述的本发明实施例提供背景或上下文。此处的描述不因为包括在本部分中就承认是现有技术。
储层精细分类对储层分级评价、甜点优选及产能预测等方面具有重要意义。按照输入数据类型的差异，储层分类方法主要包括岩心储层分类方法、测井储层分类方法及物探储层分类方法，其中测井储层分类方法具有数据来源丰富、纵向分辨率高，可连续井段处理等优势，是储层精细分类的重要方法。
目前，测井储层分类可归纳为四大类：(1)基于交会图版的半定量储层分类方法，此方法操作简单、应用范围广，但适用性不强；(2)基于岩心刻度测井的测井储层分类方法，该方法主要基于岩心物性数据，具有明确的地质意义，但依赖于岩心数据；(3)基于多元统计法及机器学习算法的测井储层分类方法，可有效避免人为因素的干扰，速度快且方法多样，但分类结果意义不明确；(4)基于测井新技术方法的储层分类方法，该方法携带了大量的地质信息，可以更有效准确的评价储层，但测井新技术方法较为昂贵，应用范围有局限。基于岩心刻度测井的测井储层分类方法有效融合了岩心分析化验数据和测井数据，但常规基于岩心刻度测井的测井储层分类方法仅依靠单参数或双参数的简单截止或交会，难以实现复杂储层精细分类的需要，如流动单元法仅考虑了岩心孔隙度和渗透率等物性数据，孔隙结构法仅考虑了岩心压汞数据或岩心核磁数据。
上述常规基于岩心刻度测井的测井储层分类方法在常规砂岩或碳酸盐岩储层中可以获得较好的应用效果，但在致密砂岩、砂砾岩、火成岩及页岩等复杂非常规储层中难以取得较好的应用效果，因为此类复杂非常规储层常具有复杂的矿物组分、泥质含量、粒度分布及孔隙结构。
发明内容
本发明的目的是提供一种储层分类方法及装置，以解决目前的储层分类方法在致密砂岩、砂砾岩、火成岩及页岩等复杂非常规储层中难以取得较好的应用效果的技术问题。
本发明的上述目的可采用下列技术方案来实现：
本发明提供一种储层分类方法，包括以下步骤：获取待分类井段的多个储层段的不同深度点的特征参数集；所述特征参数集包括测井特征参数集以及岩心特征参数集，所述测井特征参数集包括岩性测井特征参数、物性测井特征参数及电性测井特征参数，所述岩心特征参数集包括物性特征参数、压汞特征参数、核磁共振特征参数以及粒度特征参数；通过主成分分析法对多个所述深度点的所述特征参数集进行分析，从而提取出各个所述深度点的至少一主成分参数；对多个所述深度点的所述主成分参数进行聚类分析，将多个所述深度点分为多种类型；根据各种所述类型的测井参数，对所述待分类井段的储层类型进行分类。
发明还提供一种储层分类装置，包括：特征参数集获取模块，用于获取待分类井段的多个储层段的不同深度点的特征参数集；所述特征参数集包括测井特征参数集以及岩心特征参数集，所述测井特征参数集包括岩性测井特征参数、物性测井特征参数及电性测井特征参数，所述岩心特征参数集包括物性特征参数、压汞特征参数、核磁共振特征参数以及粒度特征参数；主成分分析模块，用于通过主成分分析法对多个所述深度点的所述特征参数集进行分析，从而提取出各个所述深度点的至少一主成分参数；深度点分类模块，用于对多个所述深度点的所述主成分参数进行聚类分析，将多个所述深度点分为多种类型；储层类型分类模块，用于根据各种所述类型的测井参数，对所述待分类井段的储层类型进行分类。
本发明的特点及优点是：
本发明的储层分类方法及装置，通过充分挖掘多个储层段的不同深度点的岩性测井特征参数、物性测井特征参数及电性测井特征参数；以及物性特征参数、压汞特征参数、核磁共振特征参数以及粒度特征参数；进而采用主成分分析法提取出至少一主成分参数，进而通过对各个岩心的主成分参数进行聚类分析，从而将多个储层段的岩心分为多种类型，进而根据多种岩心的类型对待分类井段的储层类型进行分类，从而可以实现复杂非常规储层的精细分类，对复杂非常规储层的综合评价及勘探开发方案制定提供参考，解决了现有技术中基于岩心刻度测井的测井储层分类方法仅依靠单参 数或双参数的简单截止或交会而无法实现复杂储层精细分类的技术问题。
附图说明
为了更清楚地说明本发明实施例中的技术方案，下面将对实施例描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本发明的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1为本发明的储层分类方法的流程图；
图2为本实施例中多个深度点的岩心孔隙度与渗透率交会图；
图3为本实施例中多个储层段的高压压汞原始曲线集；
图4为本实施例中多个储层段的核磁共振T2谱集；
图5为本实施例中多个储层段的岩心粒度概率分布曲线集；
图6为本实施例中多个深度点的第一主成分参数及第二主成分参数的交会图；
图7为本发明的提取主成分参数的流程图；
图8为本发明的对待测井段的储层类型进行连续分类的流程图；
图9为本实施例中一口典型井中一井段的储层分类效果图。
具体实施方式
下面将结合本发明实施例中的附图，对本发明实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本发明一部分实施例，而不是全部的实施例。基于本发明中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都属于本发明保护的范围。
实施方式一
如图1所示，本发明提供一种储层分类方法，包括以下步骤：
步骤101、获取待分类井段的多个储层段的不同深度点的特征参数集；特征参数集包括测井特征参数集以及岩心特征参数集，测井特征参数集包括岩性测井特征参数、物性测井特征参数及电性测井特征参数，岩心特征参数集包括物性特征参数、压汞特征参数、核磁共振特征参数以及粒度特征参数。
其中，利用测井曲线统计获取不同深度点的测井特征参数集。
其中，利用岩性测井曲线统计出各个深度点的岩性测井特征参数。具体的，岩性测井特征参数包括自然伽马测井值及自然电位测井值。
其中，利用物性测井曲线统计出各个深度点的物性测井特征参数。具体的，物性测井特征参数包括密度测井值、补偿中子测井值及声波时差测井值。
其中，通过电性测井曲线统计出各个深度点的电性测井特征参数。具体的，电性测井特征参数包括深电阻率测井值、中电阻率测井值、浅电阻率测井值。
其中，岩心特征参数集的获取，包括：通过钻取多个储层段的不同深度位置的岩样并将岩样预处理成符合要求的岩心，岩样的预处理包括：将岩样切割并打磨成3cm-4cm，直径为2.5cm的标准柱塞状的岩心；将岩心在90摄氏度-110摄氏度的温度下烘干12小时-24小时，以除去岩心中的水分；通过对各个岩心开展实验，从而获取不同深度点的岩心特征参数集。其中，通过开展物性参数实验，获取岩心的物性特征参数。具体的，物性特征参数包括但不限于岩心孔隙度以及岩心渗透率。采用波义耳单室法注入氦气测得岩心孔隙度。采用气测法测得岩心渗透率。本实施例中，各个深度点的岩心孔隙度与岩心渗透率的交会如图2所示。
其中，通过开展岩心高压压汞实验，获取深度点的压汞特征参数。具体的，压汞特征参数包括但不限于中值压力、中值半径、排驱压力、最大孔喉半径、退汞效率、平均孔喉半径及最大进汞饱和度。本实施例中，各个深度点的岩心高压压汞原始曲线如图3所示。
其中，通过开展岩心核磁共振实验，获取深度点的核磁共振特征参数。具体的，核磁共振特征参数包括但不限于磁孔隙度、几何平均值、算数平均值、泥质束缚水饱和度、毛管束缚水饱和度及可动水饱和度。本实施例中，各个深度点的岩心核磁共振T2(横向弛豫时间)谱集如图4所示。
其中，通过开展岩心粒度分析实验，获取岩心的粒度特征参数。具体的，粒度特征参数包括泥质含量、C值、粒度中值、峰值、偏度及分选系数。本实施例中，各个深度点的岩心粒度概率分布如图5所示。
步骤102、通过主成分分析法对多个深度点的特征参数集进行分析，从而提取出各个岩心的至少一主成分参数。
其中，通过主成分分析法对多个深度点的特征参数集进行分析可以获得多个主成分参数，通过计算各个主成分参数的贡献率，将多个主成分参数根据其贡献率由大至小地排列为第一主成分参数、第二主成分参数，…，第n主成分参数，其中n等于特征参数集中数据的数量。本实施例中，每个深度点提取出前两个主成分参数，即第一主成分参数和第二主成分参数。当然也可以仅提取出第一主成分参数，还可以提取出第三主成分参数或者更多个主成分参数。
步骤103、对多个深度点的主成分参数进行聚类分析，将多个深度点分为多种类型。
其中，可以采用K均值聚类法将多个深度点的主成分参数进行分类，将多个深度点分为多种类型。也可以采用现有技术中其他聚类分析进行分类。如图6所示，本实施例中，通过对各个深度点的第一主成分参数和第二主成分参数进行聚类分析，将多个深度点分为四种类型。当然，对于不同的井段所获取的多个储层段的深度点，通过上述步骤进行分类，也有可能分为两种、三种、五种、六种，甚至更多种。
步骤104、根据多种类型，对待分类井段的储层类型进行分类。
本发明的储层分类方法，通过充分挖掘多个储层段的不同深度点的岩性测井特征参数、物性测井特征参数及电性测井特征参数；以及物性特征参数、压汞特征参数、核磁共振特征参数以及粒度特征参数；进而采用主成分分析法提取出至少一主成分参数，进而通过对各个深度点的主成分参数进行聚类分析，从而将多个储层段的深度点分为多种类型，进而根据多种深度点的类型对待分类井段的储层类型进行分类，从而可以实现复杂非常规储层的精细分类，对复杂非常规储层的综合评价及勘探开发方案制定提供参考，解决了现有技术中基于岩心刻度测井的测井储层分类方法仅依靠单参数或双参数的简单截止或交会而无法实现复杂储层精细分类的技术问题。
如图7所示，本发明的实施方式中，通过主成分分析法对多个深度点的特征参数集进行分析，从而提取出各个深度点的至少一主成分参数，包括以下步骤：
步骤201、将多个深度点的特征参数集组合为第一矩阵X1。其中，第一矩阵X1的行数n等于岩心特征参数集中数据的数量，第一矩阵的列数m等于深度点的数量。
步骤202、将第一矩阵X1中每行的数据都减去该行数据的均值，得到第二矩阵X2；
步骤203、根据第二矩阵X2，计算出协方差矩阵C。具体的，协方差矩阵C与第二矩阵X2之间的关系为：
步骤204、利用奇异值分解法求取协方差矩阵C的多个特征值及对应的特征向量，并将多个特征向量按其对应的特征值从大到小地排序，并分别作为行向量由上至下地组成第三矩阵X3；
步骤205、取第三矩阵X3中的前面多行的数据组成第四矩阵X4；其中，第四矩阵X4的行数等于每个深度点所需提取的主成分参数的数量。本实施例中，所需选取为第一主成分参数和第二主成分参数，因此选取第三矩阵X3中的第一行数据和第二行数据 组成第四矩阵X4。
步骤206、根据第四矩阵X4和第一矩阵X1，提取出各个深度点的主成分参数。具体的，第四矩阵X4与第一矩阵X1之间的关系为：Y＝X4X1，Y为各个深度点所提取出的主成分参数所构成的矩阵。
下面列举一简单的例子对上述步骤进行说明：假设深度点的数量为5，5个深度点的特征参数集分别为{-1 -2}、{-1 0}、{0 1}、{2 1}以及{0 1}，则第一矩阵X1为由于每行均值为0，因此第二矩阵X2与第一矩阵X1相同，则协方差矩阵进而计算出协方差矩阵C的特征值λ1＝2，对应的特征向量则第三矩阵X3为假设仅提取第一主成分参数，则第四矩阵X4为因此，5个岩心的第一主成分参数所构成的矩阵为：
如图8所示，本发明的实施方式中，根据各种类型的测井参数，对待分类井段的储层类型进行分类，包括以下步骤：
步骤301、利用岩心标定测井曲线统计各种类型的测井参数。
其中，测井参数包括测井计算孔隙度、测井计算渗透率、自然伽马、自然电位、声波、密度、补偿中子以及电阻率中的一种或多种。本实施例中，测井参数包括六种参数，分别为测井计算的孔隙度、测井计算的渗透率、密度(DEN)、声波(AC)、补偿中子(CNL)、自然伽马(ΔGR)以及自然电位(ΔSP)。四种类型的测井参数的统计表如下所示：

步骤302、根据各种类型的测井参数，基于Fisher判别分析法建立各种类型的判别模型。
其中，判别模型为：F＝A1*D1+A2*D2+A3*D3+A4*D4+A5，F为各类型的判别标准；其中，A1、A2、A3、A4、A5分别为待定系数，D1、D2、D3、D4均为测井参数，为提高判别模型的准确度，D1、D2、D3、D4优选测井敏感参数，即区分类型最准确的测井参数。
本实施例中，选取密度(DEN)为D1，声波(AC)为D2，选取自然伽马(ΔGR)为D3，选取自然电位(ΔSP)为D4，从而获得各个类型的判别模型，具体的：
类型一的判别模型为：
F1＝1282.83*DEN+21.23*AC-207.14*ΔGR+66.75*ΔSP-2292.35；通过将上述统计表中类型一对应的测井参数值代入该判别模型中，得到类型一的判别标准F1。本实施例中，16个深度点为类型一，采用该判别模型对16个深度点的类型进行判别，准确率为87.5％。
类型二的判别模型为：
F2＝1323.38*DEN+21.52*AC-212.04*ΔGR+74.55*ΔSP-2414.28；通过将上述统计表中类型二的对应的测井参数值代入该判别模型中，得到类型二的判别标准F2。本实施例中，38个深度点为类型二，采用该判别模型对38个该深度点进行判别，准确率为89.4％。
类型三的判别模型为：
F3＝1300.87*DEN+20.97*AC-207.03*ΔGR+69.81*ΔSP-2315.76；通过将上述统计表中类型三的对应的测井参数值代入该判别模型中，得到类型三的判别标准F3。本实施例中，25个深度点为类型二，采用该判别模型对20个该深度点的类型进行判别，准确 率为84.0％。
类型四的判别模型为：
F4＝1313.57*DEN+20.7*AC-198.78*ΔGR+68.78*ΔSP-2337.74；通过将上述统计表中类型四对应的测井参数值代入该判别模型中，得到类型四的判别标准F4。本实施例中，10个深度点为类型四，采用该判别模型和对10个该深度点的类型进行判别，准确率为80.0％。
步骤303、提取待分类井段的测井敏感曲线，并将测井敏感曲线的敏感测井曲线值代入各种类型的判别模型中，从而实现待分类井段的储层类型的连续分类。其中，待分类井段为泥质含量及测井计算的孔隙度达到有效储层标准的井段，对于泥质含量或测井计算的孔隙度未达到有效储层标准的井段不进行储层类型的判别及分类。
具体的，当测井敏感曲线中某深度位置对应的敏感测井曲线值，也即与判别模型中对应的各个测井参数的值，代入判别模型中得到的结果符合该判别模型对应的判别标准，则说明该深度位置的储层类型为该判别模型对应的类型，从而以此方式，便能确定测井敏感曲线各深度位置的储层类型，实现该井段的储层类型的连续分类。例如，将待测井段的测井敏感曲线一深度位置的密度(DEN)、声波(AC)、自然伽马(ΔGR)以及自然电位(ΔSP)的值分别代入上述四种类型的判别模型后分别得到四个结果，只有当代入类型二的判别模型获得的结果与类型二的判别标准F2相符，则说明该深度位置的储层分类为类型二。本实施例中，待分类井段为一口典型井中深度约为4090m至4160m的井段。利用上述四种判别模型和该待分类井段的敏感测井曲线，将该井段进行连续分类的效果图如图9所示。
实施方式二
本发明还提供一种储层分类装置，其可以参照上述储层分类方法进行实施，在此不再赘述。
本发明的储层分类装置包括：特征参数集获取模块，用于获取待分类井段的多个储层段的不同深度点的特征参数集；特征参数包括测井特征参数集以及岩心特征参数集，测井特征参数包括岩性测井特征参数、物性测井特征参数及电性测井特征参数，岩心特征参数集包括物性特征参数、压汞特征参数、核磁共振特征参数以及粒度特征参数。主成分分析模块，用于通过主成分分析法对多个深度点的特征参数集进行分析，从而提取出各个深度点的至少一主成分参数；岩心分类模块，用于对多个深度点的主成分参数进行聚类分析，将多个深度点分为多种类型；储层类型分类模块，用于 根据各种类型的测井曲线，对待分类井段的储层类型进行分类。
本发明的实施方式中，储层类型分类模块包括：测井参数统计单元，用于利用岩心标定测井曲线统计各种类型的测井参数；判别模型建立单元，用于根据各种类型的测井参数，基于Fisher判别分析法建立各种类型的判别模型；连续分类单元，用于提取待分类井段的测井敏感曲线，并将测井敏感曲线的敏感测井曲线值代入各种类型的判别模型中，从而实现待分类井段的储层类型的连续分类。
以上所述仅为本发明的几个实施例，本领域的技术人员依据申请文件公开的内容可以对本发明实施例进行各种改动或变型而不脱离本发明的精神和范围。

Claims (15)

  1. 一种储层分类方法，其中，包括以下步骤：
    获取待分类井段的多个储层段的不同深度点的特征参数集；所述特征参数集包括测井特征参数集以及岩心特征参数集，所述测井特征参数集包括岩性测井特征参数、物性测井特征参数及电性测井特征参数，所述岩心特征参数集包括物性特征参数、压汞特征参数、核磁共振特征参数以及粒度特征参数；
    通过主成分分析法对多个所述深度点的所述特征参数集进行分析，从而提取出各个所述深度点的至少一主成分参数；
    对多个所述深度点的所述主成分参数进行聚类分析，将多个所述深度点分为多种类型；
    根据各种所述类型的测井参数，对所述待分类井段的储层类型进行分类。
  2. 如权利要求1所述的储层分类方法，其中，
    所述测井特征参数集中的所述岩性测井特征参数包括自然伽马测井值及自然电位测井值。
  3. 如权利要求1所述的储层分类方法，其中，
    所述测井特征参数集中的所述物性测井特征参数包括密度测井值、补偿中子测井值及声波时差测井值。
  4. 如权利要求1所述的储层分类方法，其中，
    所述测井特征参数集中的所述电性测井特征参数包括深电阻率测井值、中电阻率测井值、浅电阻率测井值。
  5. 如权利要求1所述的储层分类方法，其中，
    所述岩心特征参数集中的所述物性特征参数包括岩心孔隙度以及岩心渗透率。
  6. 如权利要求1所述的储层分类方法，其中，
    所述岩心特征参数集中的所述压汞特征参数包括中值压力、中值半径、排驱压力、最大孔喉半径、退汞效率、平均孔喉半径及最大进汞饱和度。
  7. 如权利要求1所述的储层分类方法，其中，
    所述岩心特征参数集中的所述核磁共振特征参数包括磁孔隙度、几何平均值、算数平均值、泥质束缚水饱和度、毛管束缚水饱和度及可动水饱和度。
  8. 如权利要求1所述的储层分类方法，其中，
    所述岩心特征参数集中的所述粒度特征参数包括泥质含量、C值、粒度中值、峰值、偏度及分选系数。
  9. 如权利要求1所述的储层分类方法，其中，所述通过主成分分析法对多个所述深度点的所述特征参数集进行分析，从而提取出各个所述深度点的至少一主成分参数，包括以下步骤：
    将多个所述深度点的所述特征参数集组合为第一矩阵；所述第一矩阵的行数等于每个所述岩心特征参数集中数据的数量，所述第一矩阵的列数等于所述深度点的数量；
    将所述第一矩阵中每行的数据都减去该行数据的均值，得到第二矩阵；
    根据所述第二矩阵，计算出协方差矩阵；
    利用奇异值分解法求取所述协方差矩阵的多个特征值及对应的特征向量，并将多个所述特征向量按其对应的特征值从大到小地排序，并分别作为行向量由上至下地组成第三矩阵；
    取所述第三矩阵中的前面多行的数据组成第四矩阵；其中，所述第四矩阵的行数等于每个所述深度点所需提取的所述主成分参数的数量；
    根据所述第四矩阵和所述第一矩阵，提取出各个所述深度点的所述主成分参数。
  10. 如权利要求1所述的储层分类方法，其中，所述对多个所述深度点的所述主成分参数进行聚类分析，将多个所述深度点分为多种类型，包括：
    采用K均值聚类法将多个所述深度点的所述主成分参数进行分类，将多个所述深度点分为多种类型。
  11. 如权利要求1所述的储层分类方法，其中，
    每个所述深度点的所述主成分参数的数量为两个，分别为第一主成分参数和第二主成分参数；所述类型的种类为四种。
  12. 如权利要求1所述的储层分类方法，其中，所述根据多种所述类型，对所述待分类井段的储层类型进行分类，包括以下步骤：
    利用岩心标定测井曲线统计各种所述类型的测井参数；
    根据各种所述类型的所述测井参数，基于Fisher判别分析法建立各种所述类型的判别模型；
    提取待分类井段的测井敏感曲线，并将所述测井敏感曲线的敏感测井曲线值代入各种所述类型的所述判别模型中，从而实现所述待分类井段的储层类型的连续分类。
  13. 如权利要求12所述的储层分类方法，其中，
    所述测井参数包括测井计算的孔隙度、测井计算的渗透率、自然伽马、自然电位、声波、密度、补偿中子以及电阻率中的一种或多种。
  14. 一种储层分类装置，其中，包括：
    特征参数集获取模块，用于获取待分类井段的多个储层段的不同深度点的特征参数集；所述特征参数集包括测井特征参数集以及岩心特征参数集，所述测井特征参数集包括岩性测井特征参数、物性测井特征参数及电性测井特征参数，所述岩心特征参数集包括物性特征参数、压汞特征参数、核磁共振特征参数以及粒度特征参数；
    主成分分析模块，用于通过主成分分析法对多个所述深度点的所述特征参数集进行分析，从而提取出各个所述深度点的至少一主成分参数；
    深度点分类模块，用于对多个所述深度点的所述主成分参数进行聚类分析，将多个所述深度点分为多种类型；
    储层类型分类模块，用于根据各种所述类型的测井参数，对所述待分类井段的储层类型进行分类。
  15. 如权利要求14所述的储层分类装置，其中，所述储层类型分类模块包括：
    测井参数统计单元，用于利用岩心标定测井曲线统计各种所述类型的测井参数；
    判别模型建立单元，用于根据各种所述类型的所述测井参数，基于Fisher判别分析法建立各种所述类型的判别模型；
    连续分类单元，用于提取所述待分类井段的测井敏感曲线，并将所述测井敏感曲线的敏感测井曲线值代入各种所述类型的所述判别模型中，从而实现所述待分类井段的储层类型的连续分类。
PCT/CN2023/118717 2022-12-19 2023-09-14 储层分类方法及装置 WO2024131160A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202211632985.0A CN118228080A (zh) 2022-12-19 2022-12-19 储层分类方法及装置
CN202211632985.0 2022-12-19

Publications (1)

Publication Number Publication Date
WO2024131160A1 true WO2024131160A1 (zh) 2024-06-27

Family

ID=91499489

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2023/118717 WO2024131160A1 (zh) 2022-12-19 2023-09-14 储层分类方法及装置

Country Status (2)

Country Link
CN (1) CN118228080A (zh)
WO (1) WO2024131160A1 (zh)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20090259446A1 (en) * 2008-04-10 2009-10-15 Schlumberger Technology Corporation Method to generate numerical pseudocores using borehole images, digital rock samples, and multi-point statistics
CN102011583A (zh) * 2010-11-30 2011-04-13 中国石油天然气股份有限公司 一种电成像结合礁滩地质模型识别储层的方法
CN107315203A (zh) * 2017-08-04 2017-11-03 中国石油化工股份有限公司江汉油田分公司勘探开发研究院 一种致密砂岩储层甜点识别及综合评价方法
WO2017197059A1 (en) * 2016-05-11 2017-11-16 Saudi Arabian Oil Company Nuclear magnetic resonance and saturation well logs for determining free water level and reservoir type
CN113642772A (zh) * 2021-07-13 2021-11-12 重庆科技学院 基于机器学习的测井储层识别预测方法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20090259446A1 (en) * 2008-04-10 2009-10-15 Schlumberger Technology Corporation Method to generate numerical pseudocores using borehole images, digital rock samples, and multi-point statistics
CN102011583A (zh) * 2010-11-30 2011-04-13 中国石油天然气股份有限公司 一种电成像结合礁滩地质模型识别储层的方法
WO2017197059A1 (en) * 2016-05-11 2017-11-16 Saudi Arabian Oil Company Nuclear magnetic resonance and saturation well logs for determining free water level and reservoir type
CN107315203A (zh) * 2017-08-04 2017-11-03 中国石油化工股份有限公司江汉油田分公司勘探开发研究院 一种致密砂岩储层甜点识别及综合评价方法
CN113642772A (zh) * 2021-07-13 2021-11-12 重庆科技学院 基于机器学习的测井储层识别预测方法

Also Published As

Publication number Publication date
CN118228080A (zh) 2024-06-21

Similar Documents

Publication Publication Date Title
CN112526107B (zh) 裂缝型致密砂岩储层甜点识别与定量表征的方法
CN103670384B (zh) 一种储层分类方法及系统
CN102053268B (zh) 一种蚀变火山岩岩性识别方法及装置
US10969323B2 (en) Systems and methods for special core analysis sample selection and assessment
Zhu et al. Rapid identification of high-quality marine shale gas reservoirs based on the oversampling method and random forest algorithm
CN104747185A (zh) 非均质油藏储层综合分类评价方法
Saporetti et al. Extreme learning machine combined with a differential evolution algorithm for lithology identification
CN115659245A (zh) 一种基于机器学习的砂岩型铀矿岩层类型识别方法及装置
US20140297186A1 (en) Rock Classification Based on Texture and Composition
CN105064986A (zh) 利用常规测录井资料构建储层四性关系谱的方法
CN106599482A (zh) 一种非常规超压致密气有效储层的识别方法
CN105986811A (zh) 基于常规测井资料五维度岩性判识方法
CN115390155A (zh) 针对非均质储层的测井解释方法、装置、电子设备及介质
Bagheri et al. Reservoir rock permeability prediction using SVR based on radial basis function kernel
Zhao et al. Data-driven diagenetic facies classification and well-logging identification based on machine learning methods: a case study on Xujiahe tight sandstone in Sichuan Basin
Yao-Jun et al. Unsupervised seismic facies analysis using sparse representation spectral clustering
WO2024131160A1 (zh) 储层分类方法及装置
CN116540316B (zh) 基于svm分类算法及聚类算法的地质土层测试方法
CN110847887B (zh) 一种细粒沉积陆相页岩裂缝识别评价方法
CN110399649A (zh) 一种基于成岩指示元素的变质岩量化识别方法
CN116821786A (zh) 一种低渗储层分类方法、装置及介质
US6272444B1 (en) Method for characterizing the coherence of an environment characteristic measurements
CN115700321A (zh) 一种基于测井复合参数的复杂岩性识别方法
CN109580678A (zh) 利用数字岩心技术快速识别评价低阻油气层方法
Sun et al. Base on temporal convolution and spatial convolution transformer for fluid prediction through well logging data

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 23905337

Country of ref document: EP

Kind code of ref document: A1