en
×

分享给微信好友或者朋友圈

使用微信“扫一扫”功能。
作者简介:

张道伟(1996—),男,湖北武汉人,在读博士研究生,从事非常规油气渗流模拟研究。E-mail:13517259746@163.com。

通讯作者:

薛亮(1983—),男,河北邢台人,副教授,博士。E-mail:xueliang@cup.edu.cn。

中图分类号:TE319

文献标识码:A

文章编号:1009-9603(2022)01-0160-08

DOI:10.13673/j.cnki.cn37-1359/te.2022.01.020

参考文献 1
邹才能,朱如凯,白斌,等.致密油与页岩油内涵、特征、潜力及挑战[J].矿物岩石地球化学通报,2015,34(1):3-17.ZOU Caineng,ZHU Rukai,BAI Bin,et al.Significance,geologic characteristics,resource potential and future challenges of tight oil and shale oil[J].Bulletin of Mineralogy,Petrology and Geo⁃ chemistry,2015,34(1):3-17.
参考文献 2
陈亮,曾诚,王丽峰,等.体积压裂新工艺在吉37井区致密油的应用[J].中国石油和化工标准与质量,2019,39(9):189-191.CHEN Liang,ZENG Cheng,WANG Lifeng,et al.Application of new volume fracturing technology for tight oil in J37 well block [J].China Petroleum and Chemical Standard and Quality,2019,39(9):189-191.
参考文献 3
于家义,李道阳,何伯斌,等.三塘湖盆地条湖组沉凝灰岩致密油有效开发技术[J].新疆石油地质,2020,41(6):714-720.YU Jiayi,LI Daoyang,HE Bobin,et al.Efficient development tech⁃ nologies for tight oil in tuffite reservoirs in Tiaohu formation of Santanghu basin[J].Xinjiang Petroleum Geology,2020,41(6):714-720.
参考文献 4
张全胜,李明,张子麟,等.胜利油田致密油储层体积压裂技术及应用[J].中国石油勘探,2019,24(2):233-240.ZHANG Quansheng,LI Ming,ZHANG Zilin,et al.Application of volume fracturing technology in tight oil reservoirs of Shengli oil⁃fleld[J].China Petroleum Exploration,2019,24(2):233-240.
参考文献 5
WARREN J E,ROOT P J.The behavior of naturally fractured res⁃ ervoirs[J].Society of Petroleum Engineers Journal,1963,3(3):245-255.
参考文献 6
DE SWANN A.Analytic solutions for determining naturally frac⁃ tured reservoir properties by well testing[J].Society of Petroleum Engineers Journal,1976,16(3):117-122.
参考文献 7
ZIMMERMAN R W,CHEN G,HADGU T,et al.A numerical dual porosity model with semianalytical treatment of fracture/matrix flow[J].Water Resources Research,1993,29(7):2 127-2 137.
参考文献 8
NOORISH Ad J,MEHRAN M.An upstream finite element method for solution of transient transport equation in fractured porous me⁃ dia[J].Water Resources Research,1982,18(3):588-596.
参考文献 9
LI L,LEE S H.Efficient Field-scale simulation of black oil in a naturally fractured reservoir through discrete fracture networks and homogenized media[J].SPE Reservoir Evaluation & Engi⁃ neering,2008,11(4):750-758.
参考文献 10
张烈辉,刘沙,雍锐,等.基于EDFM的致密油藏分段压裂水平井数值模拟[J].西南石油大学学报:自然科学版,2019,41(4):1-11.ZHANG Liehui,LIU Sha,YONG Rui,et al.EDFM-based numeri⁃ cal simulation of horizontal wells with multi-stage hydraulic frac⁃ turing in tight reservoirs[J].Journal of Southwest Petroleum Uni⁃ versity:Science & Technology Edition,2019,41(4):1-11.
参考文献 11
MAHMOOD S,ARAUJO C,KAMY S.Using embedded discrete fracture model(EDFM)in numerical simulation of complex hy⁃ draulic fracture networks calibrated by microseismic monitoring data[J].Journal of Natural Gas Science and Engineering,2018,55(1):495-507.
参考文献 12
SALERI N G,TORONYI R M.Engineering control in reservoir simulation:Part I[C].Houston:SPE Annual Technical Conference and Exhibition,1988.
参考文献 13
KAZEMI Alireza,STEPHEN Karl D.油藏数值模拟自动历史拟合方法——以Nelson油田为例[J].石油勘探与开发,2012,39(3):326-337.KAZEMI Alireza,STEPHEN Karl D.Schemes for automatic histo⁃ ry matching of reservoir modeling:A case of Nelson oilfield in the UK[J].Petroleum Exploration and Development,2012,39(3):326-337.
参考文献 14
RODRIGUES J,WAECHTER A,CONN A,et al.Combining ad⁃ joint calculations and quasi-newton methods for automatic history matching[C].Vienna:Austria SPE Europec/EAGE Annual Confer⁃ ence and Exhibition,2006.
参考文献 15
KOZA,JOHNR.Genetic programming:on the programming of computers by means of natural selection[M].Massachusetts:MIT Press,1992.
参考文献 16
苏守宝,汪继文,方杰.粒子群优化技术的研究与应用进展[J].计算机技术与发展,2007,17(5):249-252.SU Shoubao,WANG Jiwen,FANG Jie.Overview applications and research on particle swarm optimization algorithm[J].Computer Technology and Development,2007,17(5):249-252.
参考文献 17
陈国良,王熙法,庄镇泉,等.遗传算法及其应用[M].北京:人民邮电出版社,1999.CHEN Guoliang,WANG Xifa,ZHUANG Zhenquan,et al.Genetic algorithm and its application[M].Beijing:Posts & Telecom Press,1999.
参考文献 18
MASCHIO C,SCHIOZER D J.Bayesian history matching using ar⁃ tificial neural network and Markov Chain Monte Carlo[J].Journal of Petroleum Science & Engineering,2014,123:62-71.
参考文献 19
EVENSEN G,HOVE J,MEISINGSET H,et al.Using the EnKF for assisted history matching of a North Sea reservoir model[C].Houston:SPE Reservoir Simulation Symposium,2007.
参考文献 20
胡冰艳,陈旭,寇根,等.基于迭代集合卡尔曼滤波的聚合物驱油模型参数反演[J].地球物理学进展,2018,33(6):2 330-2 335.HU Bingyan,CHEN Xu,KOU Gen,et al.Inverse of parameters of core polymer flooding using iterative ensemble Kalman filter[J].Progress in Geophysics,2018,33(6):2 330-2 335.
参考文献 21
KALMAN R E.A new approach to linear filtering and prediction problems[J].Journal of Basic Engineering,1960,82(1):35-45.
参考文献 22
JAZWINSKI A H.Stochastic processes and filtering theory[J].IEEE Transactions on Automatic Control,2003,17(5):752-753.
参考文献 23
WAN E A,VAN D,MERWE R.The unscented Kalman filter for nonlinear estimation[C].Lake Louise:Proceedings of the IEEE 2000 Adaptive Systems for Signal Processing,Communications,and Control Symposium,IEEE,2000:153-158.
参考文献 24
EVENSEN G.Inverse methods and data assimilation in nonlinear ocean models[J].Physica D:Nonlinear Phenomena,1994,77(1/3):108-129.
参考文献 25
JANSEN J D,BROUWER R,DOUMA S G.Closed loop reservoir management[C].The Woodlands:SPE Reservoir Simulation Sym⁃ posium,2009.
参考文献 26
SEILER A,AANONSEN S I,EVENSEN G,et al.An elastic grid approach for fault uncertainty modelling and updating using the Ensemble Kalman filter[C].Barcelona:SPE EUROPEC/EAGE Annual Conference and Exhibition,OnePetro,2010.
参考文献 27
PANWAR A,TRIVEDI J J,NEJADI S.Importance of distributed temperature sensor(dts)data for sagd reservoir characterization and history matching within Ensemble Kalman filter(EnKF)framework[C].Mexico City:SPE Latin America and Caribbean Pe⁃ troleum Engineering Conference,OnePetro,2012.
参考文献 28
HUANG Y,ZENG F.The temperature-based localization for the application of EnKF on automatic history matching of the SAGD process[J].Computational Geosciences,2016,20(1):187-212.
参考文献 29
LI X,ZHANG D,LI S.A multi-continuum multiple flow mecha⁃ nism simulator for unconventional oil and gas recovery[J].Journal of Natural Gas Science and Engineering,2015,26:652-669.
目录contents

    摘要

    致密油藏经过体积压裂形成多尺度复杂缝网,强非均质性使储层参数不确定性显著增强,精确表征裂缝和降低储层参数不确定性对精确建立油藏数值模拟模型非常重要。为此,建立了基于嵌入式离散裂缝模型的致密油藏 CO2吞吐数值模拟方法,有效表征了致密油藏压裂后的多尺度复杂缝网;结合集合卡尔曼滤波方法,对致密油藏CO2 吞吐生产数据进行了智能历史拟合,估计了储层物性和压裂裂缝参数,降低了模型参数的不确定性。结果表明,基于嵌入式离散裂缝的致密油藏模拟模型能精确表征多尺度复杂缝网,适用于处理压裂后致密油藏复杂裂缝的情况。经过等效处理的多尺度介质模型计算得到的产量曲线存在伪顶点和真实顶点,与实际致密油藏CO2吞吐生产规律认识一致;经过集合卡尔曼滤波智能历史拟合方法多次迭代拟合后,初始实现的多尺度裂缝介质模型集合参数曲线收敛性增强,J1井和J2井模拟预测产量与生产历史拟合效果良好。

    Abstract

    The multi-scale complex fracture networks are formed by volume fracturing in tight oil reservoirs. At the same time,the uncertainty of reservoir parameters is significantly enhanced due to the strong heterogeneity. It is very important to accurately characterize fractures and reduce the uncertainty of reservoir parameters for building an accurate reservoir nu- merical simulation model. Therefore,a numerical simulation model for CO2 huff-n-puff in tight oil reservoirs was construct- ed according to the embedded discrete fracture model(EDFM). By doing this,the multi-scale complex fracture network of tight oil reservoirs after fracturing was effectively characterized. With the ensemble Kalman filter(EnKF)method,the intel- ligent historical matching was carried out for CO2 huff-n-puff production data of tight oil reservoirs,and the physical prop- erties and fracture parameters of reservoirs were estimated,which reduced the uncertainty of model parameters. The results show that the EDFM-based simulation model for tight oil reservoirs can accurately characterize multi-scale complex frac- ture networks and is suitable for treating complex fractures in tight oil reservoirs after fracturing. Pseudo-and real-vertices exist in the production curve calculated by the multi-scale medium model after effective medium treatment,which is consis- tent with the actual CO2 huff-n-puff production law of tight oil reservoirs. Upon repeated iterative matching with the EnKF intelligent history matching method,the convergence of the ensemble parameter curve of the initially realized multi-scale fractured medium model is enhanced,and the simulated production and production history of two horizontal wells,i.e.,Well J1 and Well J2,are well matched.

  • 中国致密油藏储量丰富,但陆相成因占比高、储层致密、天然裂缝高度发育且非均质性强[1]。作为从钻井和压裂方面有效提高致密油藏产能的技术,体积压裂水平井技术近年来已被广泛应用于油田开发[2-4]。体积压裂后在压裂改造区内形成复杂缝网,裂缝形态复杂多样和油藏参数具有较强的不确定性,精确表征多尺度裂缝和降低油藏参数不确定性是精细化建立致密油藏数值模拟模型的关键。

  • 裂缝渗流模拟模型主要包括等效连续介质模型、双重介质模型、多重基质模型和离散裂缝模型。 WARREN 等对双重介质模型提出了完整的数学阐述和模型建立方法,NOORISH等进一步通过忽略裂缝开度,将带有厚度二维连续裂缝单元采用一维双节点的形式进行表示,并提出离散裂缝模型的概念[5-8],但是这些模型难以兼顾致密油藏复杂缝网表征和渗流模拟的计算精度和计算效率。LI 和 LEE 在离散裂缝网络的基础上将裂缝片进行适度简化,结合双重介质和离散裂缝网络模型的优点,提出嵌入式离散裂缝模型(EDFM),以期提升求解效率[9]。张烈辉等引入 EDFM 方法,建立了考虑重力和应力敏感效应的三维致密油藏分段压裂水平井模型,对比结果显示EDFM方法对裂缝分布形态表征的适应性更好[10]。MAHMOOD 等基于微地震监测数据,将 EDFM 方法应用于裂缝实际几何形状数值模拟,生成和校正了水力压裂裂缝网络模型[11]

  • 历史拟合通过不断优化调整油藏属性参数,使模拟模型计算与实际生产数据相匹配[12]。智能历史拟合方法对于降低油藏属性参数不确定性,使建立的数值模拟模型符合油田生产实际具有重要意义[13]。油藏辅助智能历史拟合是以油藏数值模拟模型为载体进行更新,通过参数化不确定性油藏属性,采用优化算法对目标函数进行优化的过程。现阶段主要的智能历史拟合方法包括梯度类方法、进化算法、人工神经网络和集合卡尔曼滤波(EnKF)方法,EnKF 方法因无需计算梯度的特点而得到广泛应用[14-20]

  • EnKF及其改进方法是在卡尔曼滤波(KF)方法的基础上发展和扩展得到的,目前在油藏辅助智能历史拟合中得到广泛应用。KALMAN 在求解维纳滤波时提出了卡尔曼滤波线性预测方法[21]。此后,诸多学者对 KF 方法进行改进,使其具有处理非线性问题的能力,包括扩展卡尔曼滤波和无迹卡尔曼滤波等[22-23]。EVENSEN 提出 EnKF 方法,在处理非线性模型中表现优异[24],因此有学者将 EnKF 引入到油藏模型参数的更新,进行智能历史拟合。JAN⁃ SEN 等采用 EnKF 方法对油藏模型进行连续更新,以修正油藏模拟器的参数设置[25]。SEILER 等采用 EnKF 方法更新了弹性网格,处理不确定性断层油藏模型的几何不确定性[26]。PANWAR等运用EnKF 方法对SAGD的分布式温度传感器测量数据进行了油藏辅助历史拟合[27]。HUANG 等对 EnKF 方法在 SAGD 过程中的局部化应用进行研究,通过提出基于温度的局部化方法对SAGD开发过程进行自动历史拟合[28]

  • 致密油藏 CO2吞吐需要通过组分模型进行模拟,其自身天然裂缝发育和非均质性较强,导致油藏参数之间相关性较差,对拟合生产数据造成较大的困难。EnKF 方法相对于传统梯度类历史拟合方法,避免了计算梯度造成的局部最优问题,同时由于其序贯式数据处理的特点,能够对产量动态数据进行较好的拟合。此研究在考虑多组系天然裂缝和压裂主裂缝与次级裂缝基础上,采用矫正 Oda 法表征天然裂缝和储层改造体积(SRV)区域中小尺度裂缝网络,嵌入式离散裂缝表征压裂主裂缝,处理致密油藏多尺度裂缝效应。通过对上述多尺度裂缝表征进行参数化,结合模型基质参数的特征,构建完整的不确定性参数集合,应用 EnKF 方法对参数集合进行智能历史拟合,降低其不确定性,使其反演后的模型更接近实际油藏状况,为未来生产动态提供更准确的预测。

  • 1 嵌入式离散裂缝模型

  • 嵌入式离散裂缝模型的假设条件主要包括:① 模型是基于状态方程的等温组分模型,考虑致密油藏非达西渗流条件。②模型中存在油、气、水三相,其中水不与油或气互溶,考虑油、气两相的相平衡条件。③考虑储层的应力敏感效应,以及重力和毛管压力的影响。

  • EDFM建模中油、气、水三相流体渗流满足质量守恒方程,即流动项+源汇项=累积项。其中各相流体因流动导致的组分变化为流动项,井筒采出和注入为源汇项,相饱和度和体积变化为累积项。

  • 致密油藏组分模型基质系统中组分 c 在油、气相渗流的表达式为:

  • a=0,g xc,amρaKmKramμapam-a=0,g xc,amqamf=tϕma=o,g Samxc,aρa
    (1)
  • 基质水相渗流方程为:

  • ρwKmKrwmμwpwm-qwmf=tϕmSwmρw
    (2)
  • 裂缝系统中 c 组分在油、气相渗流的完整表达式为:

  • a=0,g xc,afρaKfKrafμapaf+a=o,g xc,afqamf-qawell=tϕfa=o,g Safxc,aρa
    (3)
  • qawell=a=o,g xc,a2πKfKraflnrerwell +S×paf-pwellμa
    (4)
  • 裂缝水相渗流方程为:

  • ρwKfKrwfμwpwf+qwmf-2πKfKrwflnrerwell+S×pwf-pwellμw=tϕfSwfρw
    (5)
  • EDFM方法是以非相邻连接的形式将裂缝嵌入基质系统,其窜流方程为:

  • qamf=qannc=i=1Nnnc AinncKinncKraμapa-γaDa-pa-γaDainncdinnc
    (6)
  • 2 集合卡尔曼滤波方法

  • 定义集合状态向量 Y 和均值Y¯的表达式分别为:

  • Y=y1,y2,,yN
    (7)
  • Y¯=1Ni=1N yi
    (8)
  • 从而可以得到EnKF方法的递推过程:

  • ①预测阶段:通过嵌入式离散裂缝模拟模型计算预测状态向量为:

  • yf=U[y]+εw
    (9)
  • 集合预测状态向量为:

  • Yf=y1f,y2f,,yNf
    (10)
  • ②对第n个迭代时间步集合预测状态向量中的元素进行更新,得到更新后的状态向量和其集合为:

  • yna=ynf+Kndn-Hnf
    (11)
  • Yna=yn,1a,yn,2a,,yn,Na
    (12)
  • 为保证集合中各项观测数据的差异性,一般需要对实际观测数据添加随机扰动误差以区分集合成员,表达式为:

  • dn=dobs+εn
    (13)
  • 卡尔曼增益矩阵表达式为:

  • Kn=CYfHTHCyfHT+CD-1
    (14)
  • 因此,集合更新状态向量为:

  • Yna=Ynf+KnD-HYnf
    (15)
  • 3 智能历史拟合应用

  • 3.1 致密油藏EDFM数值模型建立

  • 致密油藏EDFM数值模型是基于角点网格模型建立,综合考虑储层平面面积大小、计算量控制和所需达到的模拟精度,设置模型参数(表1)。

  • 采用 Petrel 进行地质建模,非常规油气藏数值模拟软件 UNCONG[29]建立致密油藏 EDFM 数值模型(图1),完成人工主裂缝的嵌入式离散裂缝表征处理和次级裂缝的SRV区域表征处理。

  • 3.2 离散天然裂缝等效渗透率模型

  • 生成 2 个组系的离散天然裂缝,采用 Fisher 分布原理生成离散裂缝模型(图2),将离散裂缝模型采用等效连续介质方法进行等效处理,得出图3 的 xyz方向天然裂缝的等效渗透率复合场,将所得等效附加渗透率与原始均质基质渗透率进行叠加,完成对天然裂缝的处理。将处理后的模型附加到地质模型中,作为智能历史拟合的载体模型。

  • 根据上述数学模型建立致密油藏 CO2吞吐模型,产量结果如图4。致密油藏基质渗透率较低,压裂后形成的压裂改造区与未改造区渗透率差异性大,因此CO2在扩散过程中速度会发生变化,导致在开井生产后返排阶段压裂改造区内的混相原油迅速采出,远端未改造区返排速度小于近井地带,因此出现原油补充滞后,近井地带原油快速采出,形成产量伪顶点,后续远端原油到达井筒后,达到产量真实顶点。结果显示,开井生产时出现伪顶点和真实顶点,说明 CO2吞吐过程中的基质低渗透性所引起的扩散延迟效应,在处理后的 EDFM 模拟中得到了较好的体现,适应性也较好,计算结果可靠。

  • 表1 致密油藏EDFM数值模型的基本参数

  • Table1 Basic parameters of EDFM-based numerical model for tight oil reservoirs

  • 图1 致密油藏EDFM数值模型

  • Fig.1 EDFM-based numerical model for tight oil reservoirs

  • 图2 采用Fisher分布原理生成离散裂缝模型

  • Fig.2 Generation of a discrete fracture model using Fisher distribution

  • 图3 xyz方向天然裂缝等效渗透率复合场

  • Fig.3 Complex equivalent permeability fields of natural fractures in xy,and z directions

  • 致密油藏储层经过体积压裂后形成围绕主裂缝的次级裂缝网络,常规获取裂缝几何和分布参数方法为微地震监测,但是地震数据只能大致上获得一个主干裂缝位置和参数以及压裂改造区范围的信息。由于致密油藏裂缝规模巨大,EDFM 是通过计算裂缝系统与基质系统的非相邻连接来表征系统间窜流量的,因此若所有裂缝均采用显式的 ED⁃ FM 方法进行表征,则会出现识别非相邻连接数目及计算量过大的问题。通过均质化基质、天然裂缝等效连续介质处理、等效SRV区域次级裂缝网络和主裂缝 EDFM 方法显式表征等方式,降低不确定因素的处理难度。裂缝处理方法将形成3类多尺度裂缝的表征形式:第 1 类为主干裂缝 EDFM 显式表征法,通过微地震响应数据进行刻画;第2类为SRV区域化处理次级裂缝网络,对压裂改造区域内的次级裂缝和微裂缝连通网络进行等效均质化处理,形成等效渗透率压裂改造区;第 3 类为等效渗透率表征大规模天然裂缝,通过等效将随机性极强的天然裂缝进行处理,由于天然裂缝的参数信息难以获取,一般采用的方法是通过裂缝属性的随机概率密度分布函数进行获取,本文采用的是 Fisher 模型生成随机天然裂缝。通过 3 类裂缝处理方法,实现对致密油藏多尺度裂缝的定量表征,为模型建立提供合理的参数,完善多尺度裂缝致密油藏模型的建立理论,计算效率和适应性有了大幅提升,同时为后续参数反演提供精确的油藏模拟模型。

  • 图4 基于EDFM的多尺度致密油藏CO2吞吐产量示意

  • Fig.4 CO2 huff-n-puff production curve of multi-scale tight oil reservoirs based on EDFM

  • 3.3 致密油藏的EnKF智能历史拟合

  • 设置CO2吞吐工作制度后即可计算模型动态产量。对载体模型进行智能历史拟合需要设置目标函数:

  • E(θ)=12dobs-H(θ)TCd-1dobs-H(θ)+θ-θprTCθ-1θ-θpr
    (16)
  • 待拟合参数集合如(17)式所示,相应的各类参数的初始分布设置如表2。

  • θ=Km,Kfl,Kf2,Ks1,Ks2T
    (17)
  • 表2 智能历史拟合参数初始分布

  • Table2 Initial parameter distribution of intelligent history matching

  • 由图5可知,初始实现的集合产量曲线经过En⁃ KF迭代收敛于拟合中心,虽然观测点波动较大且致密油藏非线性强,但是通过剔除异常点后曲线仍然能收敛,说明 EnKF 方法良好的数据同化能力。传统的历史拟合方法一般采用计算梯度的方法逼近目标函数的最小值,易陷入局部最优的困境,特别是针对致密油藏CO2吞吐中的参数多且差异大的组分模型进行历史拟合,更增加了参数梯度计算的难度。而 EnKF 相较于传统梯度类历史拟合方法,既不需要采用计算梯度的形式,同时由于其序贯式数据处理的特点,能够实时处理生产动态数据。智能历史拟合结果显示,初始100组模型经过EnKF更新迭代后产量曲线收敛于实际数据,J1和J2井的计算数据与实际数据拟合性较好、收敛性增强。

  • 4 结论

  • 体积压裂致密油藏复杂缝网表征困难,在考虑多组系天然裂缝、主裂缝和中微尺度次级裂缝基础上,结合矫正Oda法和等效连续介质理论,表征天然裂缝和微尺度压裂缝、储层改造体积(SRV)区域表征中尺度裂缝网络、嵌入式离散裂缝表征主干压裂缝,以此完成致密油藏多尺度裂缝处理。所建立的 EDFM 模型不仅能较好地表征裂缝形态分布,而且大规模降低计算量,有效提高模型计算效率,同时在致密油 CO2吞吐模型应用上表现出较好的适应性。

  • 图5 基于EDFM的致密油藏CO2吞吐模型中J1井和J2井拟合前后效果对比

  • Fig.5 Performance comparison before and after matching of Well J1 and Well J2 in CO2 huff-n-puff model of tight reservoirs based on EDFM

  • 建立基于 EDFM 的角点网格致密油藏 CO2吞吐模型,通过多尺度裂缝表征方式进行叠加,得到所需智能历史拟合的载体模型,不仅使致密油藏复杂裂缝得到良好的表征,也有利于从模拟效率角度处理复杂裂缝的多尺度效应。在此模型基础上,结合基于贝叶斯理论目标函数和 EnKF 优化方法,对不确定性强的初始参数进行反演,很好的降低了初始参数的不确定性。

  • 通过智能历史拟合应用,对 2 口水平井日产油量进行随机智能历史拟合,经过迭代更新后,提高了实际历史数据的拟合程度,参数不确定性显著降低,说明 EDFM/EnKF 结合的方法对于模拟实际致密油藏有很好的拟合效果,为体积压裂致密油藏复杂裂缝的表征、储层参数不确定性的降低和产量预测精度的提升,提供了有效的技术手段。

  • 但是单纯使用 EnKF方法仍然对大部分非线性极强的油藏拟合效果不佳,由于其面对大规模油藏智能历史拟合时具有仿真时间长的缺点,因此将 EnKF 方法与其他人工智能算法结合是改良智能历史拟合方法的有效途径,未来的工作方向可以在多尺度裂缝模型表征基础上通过与混合算法的结合,实现天然裂缝发育致密油藏的智能历史拟合。

  • 符号解释

  • a——流体相类型,可表示气相或油相;

  • Annci ——基质中第i个非相邻连接的接触面积,m2

  • c ——组分;

  • C-1 d ——观测数据的误差协方差矩阵的逆;

  • CD ——集合观测向量的误差协方差矩阵;

  • C-1 θ ——模型参数先验误差协方差矩阵的逆;

  • Cγf——集合预测状态向量的误差协方差矩阵;

  • dnnc i ——基质中单个离散裂缝片第 i 个非相邻连接的特征距离,m;

  • dn ——第n个集合成员带误差的观测向量;

  • d obs——实际生产数据构成的观测矩阵;

  • D ——集合观测向量;

  • Da ——流体某相a的深度,m;

  • Eθ)——EnKF拟合的目标函数;

  • g——气相;

  • H ——单位扩展矩阵,H = [ O|I],维度由 yndn 决定来修正矩阵维度;

  • Hθ)——预测数据构成的预测动态矩阵;

  • i ——相应的非相邻连接索引号;

  • Kf ——裂缝绝对渗透率,mD;

  • K f1——J1井主裂缝绝对渗透率的猜测中心,mD;

  • K f2——J2井主裂缝绝对渗透率的猜测中心,mD;

  • Knnc i ——第i个非相邻连接的绝对渗透率;mD;

  • K m——基质绝对渗透率,mD;

  • K n——第n个迭代时间步的卡尔曼增益矩阵;

  • K ra——介质中流体某相a的相对渗透率,mD;

  • K f ra——裂缝中流体某相a的相对渗透率,mD;

  • K m ra——基质中流体某相a的相对渗透率,mD;

  • Kf rw——裂缝中水相的相对渗透率,mD;

  • K m rw——基质中水相的相对渗透率,mD;

  • K s1——J1井SRV压裂改造区的等效渗透率,mD;

  • K s2——J2井SRV压裂改造区的等效渗透率,mD;

  • θ——油藏模型参数的后验估计值;

  • θpr ——油藏模型参数的先验估计值;

  • N——EnKF方法实现时的集合成员总数;

  • N nnc——所有非相邻连接的总个数;

  • p well ——井底流压,MPa;

  • pa ——介质中流体某相a分压,MPa;

  • pf a——裂缝中流体某相a分压,MPa;

  • pm a——基质中流体某相a分压,MPa;

  • pf w——裂缝中某点的水相分压,MPa;

  • pm w——基质中水相分压,MPa;

  • qmf a ——基质与裂缝通过窜流作用交换的流体流量,与非相邻连接的数量有关,kg/s;

  • qnnca ——非相邻连接交换传导流量速度,kg/s;

  • qwell a ——裂缝流入井中某相a的流量,kg/s;

  • q mf w ——基质与裂缝通过窜流作用交换的水流量,kg/s;

  • r e——有效井径,表明裂缝控制的有效区域,m;

  • r well ——井筒半径,m;

  • S ——表皮因子,实数;

  • Sf a——裂缝中流体某相a的饱和度,小数;

  • Sm a——基质中流体某相a的饱和度,小数;

  • Sf w——裂缝中某点的水相饱和度,小数;

  • Sm w——基质中某点的水相饱和度,小数;

  • t ——时间,s;

  • U ——数值模拟模型;

  • xc,a——组分c在某相中的含量,小数;

  • xf c,a——裂缝中组分c在某相中的含量,小数;

  • xm c,a——基质中组分c在某相中的组分含量,小数;

  • y ——状态向量;

  • yf ——预测状态向量;

  • yi——集合成员i的状态向量;

  • yN——集合状态向量;

  • ya n——第n个迭代时间步的更新状态向量;

  • yan,N——第 N 个集合成员在第 n个迭代时间步的更新状态向量;

  • yf n——第n个迭代时间步时N个集合预测状态成员的均值;

  • yf N——第N个集合成员的预测状态向量;

  • Y ——集合状态向量;

  • Y-——集合状态向量的均值;

  • Yf ——集合预测状态向量;

  • Yan——第n个迭代时间步集合更新状态向量;

  • Y f n——第n个迭代时间步集合预测状态向量;

  • εn——观测向量噪声,产生集合成员差异;

  • εw ——模型运算过程中产生的周围噪声;

  • μa——流体某相a的黏度,mPa·s;

  • μw——水相黏度,mPa·s;

  • ρa——流体某相a的密度,kg/m3

  • ρw——水相密度,kg/m3

  • γa——流体某相 a 的重度,即密度与重力加速度的乘积,N/m3

  • ϕm——基质孔隙度,小数;

  • ϕf ——裂缝孔隙度,小数。

  • 参考文献

    • [1] 邹才能,朱如凯,白斌,等.致密油与页岩油内涵、特征、潜力及挑战[J].矿物岩石地球化学通报,2015,34(1):3-17.ZOU Caineng,ZHU Rukai,BAI Bin,et al.Significance,geologic characteristics,resource potential and future challenges of tight oil and shale oil[J].Bulletin of Mineralogy,Petrology and Geo⁃ chemistry,2015,34(1):3-17.

    • [2] 陈亮,曾诚,王丽峰,等.体积压裂新工艺在吉37井区致密油的应用[J].中国石油和化工标准与质量,2019,39(9):189-191.CHEN Liang,ZENG Cheng,WANG Lifeng,et al.Application of new volume fracturing technology for tight oil in J37 well block [J].China Petroleum and Chemical Standard and Quality,2019,39(9):189-191.

    • [3] 于家义,李道阳,何伯斌,等.三塘湖盆地条湖组沉凝灰岩致密油有效开发技术[J].新疆石油地质,2020,41(6):714-720.YU Jiayi,LI Daoyang,HE Bobin,et al.Efficient development tech⁃ nologies for tight oil in tuffite reservoirs in Tiaohu formation of Santanghu basin[J].Xinjiang Petroleum Geology,2020,41(6):714-720.

    • [4] 张全胜,李明,张子麟,等.胜利油田致密油储层体积压裂技术及应用[J].中国石油勘探,2019,24(2):233-240.ZHANG Quansheng,LI Ming,ZHANG Zilin,et al.Application of volume fracturing technology in tight oil reservoirs of Shengli oil⁃fleld[J].China Petroleum Exploration,2019,24(2):233-240.

    • [5] WARREN J E,ROOT P J.The behavior of naturally fractured res⁃ ervoirs[J].Society of Petroleum Engineers Journal,1963,3(3):245-255.

    • [6] DE SWANN A.Analytic solutions for determining naturally frac⁃ tured reservoir properties by well testing[J].Society of Petroleum Engineers Journal,1976,16(3):117-122.

    • [7] ZIMMERMAN R W,CHEN G,HADGU T,et al.A numerical dual porosity model with semianalytical treatment of fracture/matrix flow[J].Water Resources Research,1993,29(7):2 127-2 137.

    • [8] NOORISH Ad J,MEHRAN M.An upstream finite element method for solution of transient transport equation in fractured porous me⁃ dia[J].Water Resources Research,1982,18(3):588-596.

    • [9] LI L,LEE S H.Efficient Field-scale simulation of black oil in a naturally fractured reservoir through discrete fracture networks and homogenized media[J].SPE Reservoir Evaluation & Engi⁃ neering,2008,11(4):750-758.

    • [10] 张烈辉,刘沙,雍锐,等.基于EDFM的致密油藏分段压裂水平井数值模拟[J].西南石油大学学报:自然科学版,2019,41(4):1-11.ZHANG Liehui,LIU Sha,YONG Rui,et al.EDFM-based numeri⁃ cal simulation of horizontal wells with multi-stage hydraulic frac⁃ turing in tight reservoirs[J].Journal of Southwest Petroleum Uni⁃ versity:Science & Technology Edition,2019,41(4):1-11.

    • [11] MAHMOOD S,ARAUJO C,KAMY S.Using embedded discrete fracture model(EDFM)in numerical simulation of complex hy⁃ draulic fracture networks calibrated by microseismic monitoring data[J].Journal of Natural Gas Science and Engineering,2018,55(1):495-507.

    • [12] SALERI N G,TORONYI R M.Engineering control in reservoir simulation:Part I[C].Houston:SPE Annual Technical Conference and Exhibition,1988.

    • [13] KAZEMI Alireza,STEPHEN Karl D.油藏数值模拟自动历史拟合方法——以Nelson油田为例[J].石油勘探与开发,2012,39(3):326-337.KAZEMI Alireza,STEPHEN Karl D.Schemes for automatic histo⁃ ry matching of reservoir modeling:A case of Nelson oilfield in the UK[J].Petroleum Exploration and Development,2012,39(3):326-337.

    • [14] RODRIGUES J,WAECHTER A,CONN A,et al.Combining ad⁃ joint calculations and quasi-newton methods for automatic history matching[C].Vienna:Austria SPE Europec/EAGE Annual Confer⁃ ence and Exhibition,2006.

    • [15] KOZA,JOHNR.Genetic programming:on the programming of computers by means of natural selection[M].Massachusetts:MIT Press,1992.

    • [16] 苏守宝,汪继文,方杰.粒子群优化技术的研究与应用进展[J].计算机技术与发展,2007,17(5):249-252.SU Shoubao,WANG Jiwen,FANG Jie.Overview applications and research on particle swarm optimization algorithm[J].Computer Technology and Development,2007,17(5):249-252.

    • [17] 陈国良,王熙法,庄镇泉,等.遗传算法及其应用[M].北京:人民邮电出版社,1999.CHEN Guoliang,WANG Xifa,ZHUANG Zhenquan,et al.Genetic algorithm and its application[M].Beijing:Posts & Telecom Press,1999.

    • [18] MASCHIO C,SCHIOZER D J.Bayesian history matching using ar⁃ tificial neural network and Markov Chain Monte Carlo[J].Journal of Petroleum Science & Engineering,2014,123:62-71.

    • [19] EVENSEN G,HOVE J,MEISINGSET H,et al.Using the EnKF for assisted history matching of a North Sea reservoir model[C].Houston:SPE Reservoir Simulation Symposium,2007.

    • [20] 胡冰艳,陈旭,寇根,等.基于迭代集合卡尔曼滤波的聚合物驱油模型参数反演[J].地球物理学进展,2018,33(6):2 330-2 335.HU Bingyan,CHEN Xu,KOU Gen,et al.Inverse of parameters of core polymer flooding using iterative ensemble Kalman filter[J].Progress in Geophysics,2018,33(6):2 330-2 335.

    • [21] KALMAN R E.A new approach to linear filtering and prediction problems[J].Journal of Basic Engineering,1960,82(1):35-45.

    • [22] JAZWINSKI A H.Stochastic processes and filtering theory[J].IEEE Transactions on Automatic Control,2003,17(5):752-753.

    • [23] WAN E A,VAN D,MERWE R.The unscented Kalman filter for nonlinear estimation[C].Lake Louise:Proceedings of the IEEE 2000 Adaptive Systems for Signal Processing,Communications,and Control Symposium,IEEE,2000:153-158.

    • [24] EVENSEN G.Inverse methods and data assimilation in nonlinear ocean models[J].Physica D:Nonlinear Phenomena,1994,77(1/3):108-129.

    • [25] JANSEN J D,BROUWER R,DOUMA S G.Closed loop reservoir management[C].The Woodlands:SPE Reservoir Simulation Sym⁃ posium,2009.

    • [26] SEILER A,AANONSEN S I,EVENSEN G,et al.An elastic grid approach for fault uncertainty modelling and updating using the Ensemble Kalman filter[C].Barcelona:SPE EUROPEC/EAGE Annual Conference and Exhibition,OnePetro,2010.

    • [27] PANWAR A,TRIVEDI J J,NEJADI S.Importance of distributed temperature sensor(dts)data for sagd reservoir characterization and history matching within Ensemble Kalman filter(EnKF)framework[C].Mexico City:SPE Latin America and Caribbean Pe⁃ troleum Engineering Conference,OnePetro,2012.

    • [28] HUANG Y,ZENG F.The temperature-based localization for the application of EnKF on automatic history matching of the SAGD process[J].Computational Geosciences,2016,20(1):187-212.

    • [29] LI X,ZHANG D,LI S.A multi-continuum multiple flow mecha⁃ nism simulator for unconventional oil and gas recovery[J].Journal of Natural Gas Science and Engineering,2015,26:652-669.

  • 参考文献

    • [1] 邹才能,朱如凯,白斌,等.致密油与页岩油内涵、特征、潜力及挑战[J].矿物岩石地球化学通报,2015,34(1):3-17.ZOU Caineng,ZHU Rukai,BAI Bin,et al.Significance,geologic characteristics,resource potential and future challenges of tight oil and shale oil[J].Bulletin of Mineralogy,Petrology and Geo⁃ chemistry,2015,34(1):3-17.

    • [2] 陈亮,曾诚,王丽峰,等.体积压裂新工艺在吉37井区致密油的应用[J].中国石油和化工标准与质量,2019,39(9):189-191.CHEN Liang,ZENG Cheng,WANG Lifeng,et al.Application of new volume fracturing technology for tight oil in J37 well block [J].China Petroleum and Chemical Standard and Quality,2019,39(9):189-191.

    • [3] 于家义,李道阳,何伯斌,等.三塘湖盆地条湖组沉凝灰岩致密油有效开发技术[J].新疆石油地质,2020,41(6):714-720.YU Jiayi,LI Daoyang,HE Bobin,et al.Efficient development tech⁃ nologies for tight oil in tuffite reservoirs in Tiaohu formation of Santanghu basin[J].Xinjiang Petroleum Geology,2020,41(6):714-720.

    • [4] 张全胜,李明,张子麟,等.胜利油田致密油储层体积压裂技术及应用[J].中国石油勘探,2019,24(2):233-240.ZHANG Quansheng,LI Ming,ZHANG Zilin,et al.Application of volume fracturing technology in tight oil reservoirs of Shengli oil⁃fleld[J].China Petroleum Exploration,2019,24(2):233-240.

    • [5] WARREN J E,ROOT P J.The behavior of naturally fractured res⁃ ervoirs[J].Society of Petroleum Engineers Journal,1963,3(3):245-255.

    • [6] DE SWANN A.Analytic solutions for determining naturally frac⁃ tured reservoir properties by well testing[J].Society of Petroleum Engineers Journal,1976,16(3):117-122.

    • [7] ZIMMERMAN R W,CHEN G,HADGU T,et al.A numerical dual porosity model with semianalytical treatment of fracture/matrix flow[J].Water Resources Research,1993,29(7):2 127-2 137.

    • [8] NOORISH Ad J,MEHRAN M.An upstream finite element method for solution of transient transport equation in fractured porous me⁃ dia[J].Water Resources Research,1982,18(3):588-596.

    • [9] LI L,LEE S H.Efficient Field-scale simulation of black oil in a naturally fractured reservoir through discrete fracture networks and homogenized media[J].SPE Reservoir Evaluation & Engi⁃ neering,2008,11(4):750-758.

    • [10] 张烈辉,刘沙,雍锐,等.基于EDFM的致密油藏分段压裂水平井数值模拟[J].西南石油大学学报:自然科学版,2019,41(4):1-11.ZHANG Liehui,LIU Sha,YONG Rui,et al.EDFM-based numeri⁃ cal simulation of horizontal wells with multi-stage hydraulic frac⁃ turing in tight reservoirs[J].Journal of Southwest Petroleum Uni⁃ versity:Science & Technology Edition,2019,41(4):1-11.

    • [11] MAHMOOD S,ARAUJO C,KAMY S.Using embedded discrete fracture model(EDFM)in numerical simulation of complex hy⁃ draulic fracture networks calibrated by microseismic monitoring data[J].Journal of Natural Gas Science and Engineering,2018,55(1):495-507.

    • [12] SALERI N G,TORONYI R M.Engineering control in reservoir simulation:Part I[C].Houston:SPE Annual Technical Conference and Exhibition,1988.

    • [13] KAZEMI Alireza,STEPHEN Karl D.油藏数值模拟自动历史拟合方法——以Nelson油田为例[J].石油勘探与开发,2012,39(3):326-337.KAZEMI Alireza,STEPHEN Karl D.Schemes for automatic histo⁃ ry matching of reservoir modeling:A case of Nelson oilfield in the UK[J].Petroleum Exploration and Development,2012,39(3):326-337.

    • [14] RODRIGUES J,WAECHTER A,CONN A,et al.Combining ad⁃ joint calculations and quasi-newton methods for automatic history matching[C].Vienna:Austria SPE Europec/EAGE Annual Confer⁃ ence and Exhibition,2006.

    • [15] KOZA,JOHNR.Genetic programming:on the programming of computers by means of natural selection[M].Massachusetts:MIT Press,1992.

    • [16] 苏守宝,汪继文,方杰.粒子群优化技术的研究与应用进展[J].计算机技术与发展,2007,17(5):249-252.SU Shoubao,WANG Jiwen,FANG Jie.Overview applications and research on particle swarm optimization algorithm[J].Computer Technology and Development,2007,17(5):249-252.

    • [17] 陈国良,王熙法,庄镇泉,等.遗传算法及其应用[M].北京:人民邮电出版社,1999.CHEN Guoliang,WANG Xifa,ZHUANG Zhenquan,et al.Genetic algorithm and its application[M].Beijing:Posts & Telecom Press,1999.

    • [18] MASCHIO C,SCHIOZER D J.Bayesian history matching using ar⁃ tificial neural network and Markov Chain Monte Carlo[J].Journal of Petroleum Science & Engineering,2014,123:62-71.

    • [19] EVENSEN G,HOVE J,MEISINGSET H,et al.Using the EnKF for assisted history matching of a North Sea reservoir model[C].Houston:SPE Reservoir Simulation Symposium,2007.

    • [20] 胡冰艳,陈旭,寇根,等.基于迭代集合卡尔曼滤波的聚合物驱油模型参数反演[J].地球物理学进展,2018,33(6):2 330-2 335.HU Bingyan,CHEN Xu,KOU Gen,et al.Inverse of parameters of core polymer flooding using iterative ensemble Kalman filter[J].Progress in Geophysics,2018,33(6):2 330-2 335.

    • [21] KALMAN R E.A new approach to linear filtering and prediction problems[J].Journal of Basic Engineering,1960,82(1):35-45.

    • [22] JAZWINSKI A H.Stochastic processes and filtering theory[J].IEEE Transactions on Automatic Control,2003,17(5):752-753.

    • [23] WAN E A,VAN D,MERWE R.The unscented Kalman filter for nonlinear estimation[C].Lake Louise:Proceedings of the IEEE 2000 Adaptive Systems for Signal Processing,Communications,and Control Symposium,IEEE,2000:153-158.

    • [24] EVENSEN G.Inverse methods and data assimilation in nonlinear ocean models[J].Physica D:Nonlinear Phenomena,1994,77(1/3):108-129.

    • [25] JANSEN J D,BROUWER R,DOUMA S G.Closed loop reservoir management[C].The Woodlands:SPE Reservoir Simulation Sym⁃ posium,2009.

    • [26] SEILER A,AANONSEN S I,EVENSEN G,et al.An elastic grid approach for fault uncertainty modelling and updating using the Ensemble Kalman filter[C].Barcelona:SPE EUROPEC/EAGE Annual Conference and Exhibition,OnePetro,2010.

    • [27] PANWAR A,TRIVEDI J J,NEJADI S.Importance of distributed temperature sensor(dts)data for sagd reservoir characterization and history matching within Ensemble Kalman filter(EnKF)framework[C].Mexico City:SPE Latin America and Caribbean Pe⁃ troleum Engineering Conference,OnePetro,2012.

    • [28] HUANG Y,ZENG F.The temperature-based localization for the application of EnKF on automatic history matching of the SAGD process[J].Computational Geosciences,2016,20(1):187-212.

    • [29] LI X,ZHANG D,LI S.A multi-continuum multiple flow mecha⁃ nism simulator for unconventional oil and gas recovery[J].Journal of Natural Gas Science and Engineering,2015,26:652-669.