收稿日期:2005-01-18基金项目:国家教育部高等学校优秀青年教师教学科研奖励计划(T RA POYT )资助作者简介:冯国庆(1974-),男,山东荷泽人,博士,主要从事油藏描述和油藏数值模拟研究.文章编号:1673-064X(2005)05-0009-03利用多点地质统计学方法模拟岩相分布Stochastic simulation of lithofacies distribution using multi -point g eostatistics冯国庆1,陈浩2,张烈辉1,李允1(1.西南石油学院油气藏地质及开发工程国家重点实验室,四川成都610500;2.新疆油田分公司陆梁油田作业区,新疆克拉玛依834000)摘要:文中介绍了多点地质统计学的基本原理,并利用该算法模拟了我国东部某砂岩油藏的岩相分布.通过对储层非均质性模拟研究,揭示储层在岩性、物性和含油气性的各向异性或非均质性,揭示砂体展布、连通程度以及在横向、纵向上的变化规律,有利于开发方案的制定及注采井网的布署.模拟结果表明,多点地质统计学模拟方法可以很好地再现砂体局部的变异性和非均质性.由此可得出以下结论:snesim 模拟算法可以快速灵活地进行多点地质统计模拟,多点地质统计学比两点地质统计学更具有优势.关键词:多点地质统计学;随机模拟;训练图像中图分类号:TE19;T E121.1+5 文献标识码:A 传统的地质统计学是利用变差函数描述地质变量的相关性和变异性,通过建立在某个方向上两点之间的地质变量的变化关系来描述空间的变化特性.但是,建立在两点统计关系上的变差函数本身在描述储层非均质性上有很大的不足,它反映的仅仅是空间两点之间的相关性,不能充分描述复杂几何形状砂体如河道砂体和冲积扇砂体空间的连续性和变异性.当井资料较少时,用于计算实验变差函数的点对很少,它也就不能正确反应空间两点之间的相关性.建立在两点基础上的变差函数在储层地质建模中存在一定的不足,而多点地质统计学[1-4]是建立在多个点的相关关系上,所以它在解决描述空间变量的连续性和变异性方面得到越来越广泛的应用.斯坦福大学的Journel 教授曾指出多点地质统计学是今后地质统计学发展的方向,它的优势已越来越显著.1 多点地质统计学的原理在阐述多点地质统计学之前,首先回顾一下变差函数的地质统计学方法是如何模拟储层岩相分布的,以序贯指示模拟算法为例进行说明.该方法的基本原理[5]简述如下:假设在模拟区域有k 种岩相s 1,s 2,,s k ,对于模拟目标区域内的每一相,定义指示变量:I (u )=1 Z(u)I s k0 Z(u)|s k,对于任一待模拟点,其出现第k 种相的概率为:P (I k =1|Z(u A )=s A ,P A ),A 为待估点所包括的条件区域,利用两点地质统计学方法计算该概率是采用克里格方法:P (I k =1|Z(u A )=s A ,P A )=E (I k )+E nA =1KA [1-E (I k )].2005年9月第20卷第5期西安石油大学学报(自然科学版)Journal of Xi c an Shiy ou U niversity(N atural Science Edition)Sep.2005V ol.20No.5其中,K A 为克里格方法确定的权系数,它通过求解由变差函数或协方差函数建立的克里格方程组来确定.多点地质统计学与两点地质统计学的主要区别在于上面的概率的确定方法不同,它首先引入一训练图像,通过在训练图像中寻找与待估点内条件数据分布完全相同的事件的个数来确定概率分布,因此它可以反映出多个位置的联合变异性.例如,计算图1(a)中u 点的概率时,相应的条件数据场为d n ={z (u 1),z (u 2),z (u 3),z (u 4)},其基本方法是首先要在训练图像(b)中寻找与图(a)中数据分布完全相同的事件的个数,即要在训练图像中找出与图(a)几何完全相同的区域,同时在该区域中相同的位置处z (u 1),z (u 2),z (u 3),z (u 4)的值完全相同.在训练图像中一共找到4个既能满足条件数据u 1,u 2,u 3,u 4数值,同时又能满足它们分空间几何形状的事件,在这4个事件中,3个事件的u 点的值为0,只有1个事件中u 点值为1,因此u 点岩相为1的条件概率为P {u =1|d n }=1/4,而P {u =0|d n }=3/4,这样便可求出了u 点的条件概率.图1 多点统计方法估计条件概率的原理上述方法不仅考虑了区域内条件数据的值,而且也考虑了条件数据的几何形状.而两点地质统计学只是依靠z (u 1),z (u 2),z (u 3),z (u 4)的值及各点与u 点距离通过求解克里格方程组来确定u 点的概率,并没有考虑d n 的几何形状和各条件数据的配位关系.基于上述原理,Sebastien Strebelle 提出了snes -im 模拟算法[2],利用该算法可以快速、灵活地模拟岩相分布.该方法的具体步骤为:(1)利用非条件模拟建立三维训练图像;(2)定义通过所有待估结点的随机路径;(3)对随机路径中的任意待估点l(=1,2,,,l):¹定义查找范围内的条件数据;º保留邻区的数据点;»在训练图像中寻找与该区域内条件数据完全相同的事件,计算该点岩相的分布概率.¼由Monto Carlo 法得到位置处的一个模拟值;½将模拟结果归入条件指示数据集中.(4)重复上一步模拟,直到所有的点全被模拟.训练图像既可以通过非条件模拟求出,也可以通过该地区的地质露头资料分析得出.对训练图像的条件非模拟可以选择非条件的布尔模拟方法,其方法和原理参见文献[6].2 实例分析对于开发中后期的砂岩油藏储层参数模拟采用两阶段模拟方法可以较为准确地反映储层的非均质性,而/两步建模0的第一步就是要建立储层结构或流动单元模型,模拟沉积体在空间排列的复杂性;利用多点统计学模拟方法可以较好地完成砂体骨架模拟.模拟区域选择我国东部某砂岩油藏第15小层,在该层一共有64口井,测井资料解释结果表明有26口井钻遇砂体,另外38口井钻遇泥岩,砂体比例为40%.对岩相进行编码,砂岩为1,泥岩为0,图2为该层井位分布图.图2 第15小层井位和井点岩相图采用上述方法模拟砂体的分布.首先建立训练图像,运用布尔模拟方法,把砂体比例40%输入,为保证训练图像数据充足,网格划分为250@250@1,一共由62500个模拟数据组成,布尔模拟结果见图3.把条件数据和布尔模拟生成的训练图像,输入到snesim 模拟算法中进行模拟.根据该区域的特点,椭圆最大搜索半径选为300m,搜索半径内最多)10)西安石油大学学报(自然科学版)的条件数据设为30,搜索主方向选择物源方向5b ,得到该层的砂泥岩分布(图4).从模拟结果看出,它很好地满足了条件数据,即在各井点处的模拟结果与数据相一致,这表明该方法为条件模拟.同时,模拟的砂体展布方向和趋势与依靠地质经验手工绘制的砂体展布图(图5)比较吻合,在模拟的左下角与左上角砂体的展布与手工勾绘的几乎完全一致,但该方法在局部区域表现出砂体展布的非均质性和不确定性,与手工勾画砂体展布的平滑而唯一的表现是具有一定差别的,它充分体现了砂体局部的变异性和非均质性.图3布尔模拟生成的训练图像图4 多点地质统计方法模拟的砂泥岩展布图图5 手工绘制的该小层的砂体展布图3 结 论(1)多点地质统计学是今后地质统计学发展的主要方向,它可以联合反映空间多个位置点的几何形状和相互配位关系;在模拟具有复杂形状地质体分布时,它比两点地质统计学方法具有更大的优势.(2)利用snesim 模拟算法可以快速灵活地进行多点地质统计模拟,模拟的岩相展布图具有一定的真实性,它为储层参数的两阶段模拟奠定了基础.参考文献:[1] Strebelle S.Condit ional simulation of complex geologicalstructures using multiplepoint statist ics[J].M athematical G eolog y,2002,34(1):1-22.[2] Strebelle S,Journel A G.Reser voir modeling using mult-i ple -point stat i stics[C].SPE71324,2001.[3] Jef Caers.Stochast ic Reser voir Characterization U singM ultiple -po int statistics [A ].In:Proceedings o f the I AM G 99,F ifth Annual Conference of the I nternational Associat ion for M at hemat ical Geology [C].L ippars S G,et al ,1999.467-472.[4] Jef Caers.G eostatistical history matching under tr aining -image based geo logical model constraints [A ].SP E77429,2002.[5] 冯国庆,李允.应用序贯指示模拟方法模拟沉积微相[J].西南石油学院学报,2001,23(2):1-4.[6] Deutsch ,Journel.GSL IB:G eostatistical Softw ar e Libraryand U ser .s Guide,2nd ed[M ].New Yor k:Ox ford U n-iversity Press,1998.104-109.编辑:国伍玲)11)冯国庆等:利用多点地质统计学方法模拟岩相分布ABSTRACTS OF THE PRESENT ISSUE(JXSYU ISSN1673-064X)Determination of the geologic age of the bottom interface of Quaternary of Caidamu Basin Abstract:Up to now there are different viewpoints on the geolog ic age the bottom interface of Quaternary. T he determination of the geolog ic age of the bottom interface of Quaternary of Caidamu Basin is not only favor-able to the theoretical study of Quaternary but also of importance to the observation of biog as reservoirs.The ge-ologic age is determined as2.5M a B.P.based on the variation of sporopollen assemblages,the evolution of palaeoclimate,the unconformity formed by intense tectonic movement,the contact state of sediment and the v ariation of archaeom agnetism.The bottom interface is identical w ith that of Quaternary loess in the northwest of China determined by Liu Tun-sheng of our country and that of Quaternary g lacial age beginning determ ined by Shakleto N.T.based on the profile of DSDP site552A hole in North Atlantic Ocean.It is a unconformity surface in w hich the basin g raduates into conformity contact.The palaeoclimate becomes cold suddenly across it.A lot of strong first appears here.Key words:Caidam u Basin;fluvia-l lacustrine sedimentation;bottom interface of QuaternaryZ H U X iao-m in1,K AN G A n2,K AN G Qiang3(1.Key Laboratory of Education M inistry for Hydrocarbon Accumulation M echanism,China University of Petroleum,Beijing,102249,China; 2.China Offshore Petroleum Exploration Corporation,Beijing,100021,China;3.China Oil&Gas Exploration Development Cor-poration,Beijing,100011,China)JXSYU2005V.20N.5p.1-4Effects of volcanic activities on the distribution of accommodation space during the development of volcanic se-quence in Xinyang DepressionAbstract:The effects of the volcanic activities on the formation and the distribution of accommodation space among regressive system tracts of volcanic sequence are discussed w ith Eogene volcanic sequence in Xinyang De-pression being taken as an ex ample.The features of volcanic sequences and the patterns of the accommodation space in regressive system tracts show that,the local accommodation space of the fault-depression lake basin w ith volcanic cones is nonhomogeneous,w hich is different from common fault-depression lake basins.By studying se-quence stratigraphy in Xinyang Depression,it is realized that the volcanic cones in reg ressive system tracts lead to the nonhomogeneous distribution of the accom modation space in this depression,w hich enriches the content of terrigenous sequence stratigraphy.Key words:Xinyang Depression,volcanic activity;sequence;accommodation space;nonhomog eneity L I Ji-shan1,JIANG Zai-x ing2(1.Institute of Porous Flow&Fluid M echanics,Chinese Academy of Sc-i ences,Langfang065007,H ebei,China;2.Faculty of Earth Resource and Information,China U niversity of Petroleum,Dongying257061,Shandong,China)JXSYU2005V.20N.5p.5-8Stochastic simulation of lithofacies distribution using mult-i point geostatisticsAbstract:The research of reservoir heterogeneity is one of the main contents of reservoir description,the heterogeneity of lithology,the heterogeneity of the physical parameters of reservoirs,and the distribution,con-nectivity and vertical and horizontal variations of sandbodies can be revealed by the simulation of reservoir hetero-geneity,which is useful for deploy ing reasonable development plan and arranging w ell netw ork.But traditional g eostatistics describes the correlation and variability of geologic variables by m eans of variate difference function, and it can not fully reflect the correlation and variability of the geologic bodies with com plicated geometric shape since it is only based on two-point geostatistics.M ult-i point geostatistics is based on multiple-point correlation relationship.It has great advantages in reflecting the joint variability of the geolog ical variables.The theory of multiple-point geostatistics is introduced in this paper and a case is also g iven.T he lithofacies distribution of the sandstone reservoir in a oilfield in eastern China is simulated,and the result shows that using multiple-point geo-statistics can reveal the local heterogeneity and variability of sandbodies.Key words:multiple-point geostatistics;stochastic simulation;training im ageÔFEN G G uo-qing1,C H EN H ao2,Z H ANG L ie-hui1,LI Yun1(1.State Key Laboratory of Oil and Gas Reservoir Geology and Exploitation,Southw est Petroleum Institute,Chengdu610500,Sichuan,China; 2.Lu-liang Oilfeild Operational Zone,Xinjiang Oilfield Company,Kelamayi834000,Xijiang,China)JXSYU2005 V.20N.5p.9-11Inorganic geochemical explanation of abnormal vitrinite reflectance:taking Niu-38well in Dongying Sag, Jiyang Depression as an exampleAbstract:The v itrinite reflectance of N iu-38well is restrained,it decreases from0.6%of the upper to 0.45%of the low er of cored interval.T he causes leading to the abnormality of the vitrinite reflectance are ana-lyzed from organic geochemistry and inorg anic geochemistry separately.The explanation from organic g eochem-istry is that,vitrinite reflectance decreases w ith the increase of the contents of sapropelic components,and hy-drogen-rich org anic matters restrain the increase of the vitrinite reflectance.From inorganic geochem istry,the study results in this paper show that,the information w ater in the studied interval becomes fresh and its acidity becomes a little strong from the low er to the upper.The acidic condition is favorable to the m aturation of organic matters.In the studied interval,the contents of K,Na,Mg and Li vary a little,therefore,there are no correla-tions between them and vitrinite reflectance;there are neg ative correlations betw een the contents of Ca,Mn,Sr and B and vitrinite reflectance,they maybe restrain the increase of vitrinite reflectance;there are positive corre-lations betw een the contents of Fe,Ba,Co,Zn,Ni and Rb and vitrinite reflectance,they maybe promote the increase of vitrinite reflectance.Key words:Dongy ing Sag;vitrinite reflectance abnormality;inorganic geochemistry;therm al evolution of organic matterXIE Ming-j u1,2,QI U N an-sheng1,2(1.Key Laboratory of Education Ministry for H ydrocarbon Accumula-tion M echanism,China U niversity of Petroleum,Beijing102249,China; 2.Basin&Reservoir Research Cen-ter,China University of Petroleum,Beijing102249,China)JXSYU2005V.20N.5p.12-16Application of nonlinearly constrained reservoir inversion technique in W eicheng Oilfield Abstract:Linear seismic inversion can only predict the distribution feature of thick sandbodies,but it can. t obtain the know ledge of the thin layer such as substratum to meet the demands of progressive exploration. Based on processing the seismic and log data of target,the distinct know ledg e of the structure of the reservoir can be gained by means of the nonlinear seismic inversion constrained by structure and log conditions,w hich can g uide the adjustment of oilfield deployment.T his study method is used in the progressive ex ploration of Weibei area in Weicheng Oilfield.The target processing of the seism ic and log data improves the resolution of the old data,and then the thin sandbodies are clearly recognized by nonlinearly constrained inversion,w hich are verified by drilling data.The fact demonstrates that nonlinearly constrained inversion has important function to the know ledge of reservoirs and the interpretation and verification of structures,and it is also an important tool to g uide progressive exploration.Key words:nonlinear constraint;reservoir inversion;w avelet frequency-division;logg ing curve reconstruc-tion;prog ressive exploration and developmentL I Zong-tian1,LI U Wei2(1.Department of Petroleum Engineering,China University of Geosciences, Beijing100081,China; 2.No.3Production Plant,Zhong yuan Oilfield Company,Puyang457001,Henan, China)JXSYU2005V.20N.5p.17-21Method of computing the mechanical parameters of formation and its application in Sulige area of Changqing OilfieldAbstract:A set of models for calculating three formation pressure profiles(form ation pore pressure,caving pressure and fracture pressure)and other formation mechanical parameters are established based on the study of borehole wall destabilization,logging data and ordinary empirical formulas.T he calculated m echanical parame-ters can help us design the safe density lim it of drilling fluid to make sure that the borehole w all is stable duringÕ。