摘要
砂砾岩储层由于其形成时特殊的沉积作用和构造活动使之成为有利的隐蔽油气藏,其勘探开发越来越受到重视,但由于该类储层具有埋藏深、岩性复杂、低孔低渗、孔隙结构复杂、非均质性强等特点,难以精确划分岩性并建立准确的解释模型,给储层参数的求取带来很大的困难。
最优化测井解释技术的发展为复杂岩性储层的评价开辟了一条可行之路,最优化测井解释可充分利用多种测井信息、测量误差和响应方程误差及地质资料和工作经验,根据广义地球物理反演理论,运用最优化数学算法、统计概率与误差理论及质量控制技术,从所有可能的解释结果中找出最合理的解释结果,综合求取储层参数。
通过对研究区储层的测井响应特征及“四性关系”的研究,分析了砂砾岩储层的岩性、含油性、物性及电性关系,在测井解释模型的建立上,提出了多组分模型。与多矿物模型类似,骨架的成分被细化,但由于砂砾岩储层岩性复杂,岩屑的组成多样不便细分,因此整体作为一个组分,与之类似,将斜长石与钾长石也作为一个整体,得到石英、长石、岩屑的骨架三组分。而对于测井响应方程,本文采用线性方程,简化了计算。
本文用最优化测井解释对砂砾岩储层进行评价,作为最优化技术的核心,数学算法起着决定性的作用,而遗传算法(Genetic Algorithm, GA)作为一种启发式全局搜索算法,近年来得到了广泛的应用,但遗传算法本身存在一定的缺陷—早熟,为克服这个缺点,本文提出了将具有较强的局部搜索能力的复合形算法(Complex Method)与遗传算法结合构成混合遗传算法作为最优化的搜索方法。与遗传算法相似,复合形算法也不需要目标函数的梯度信息,因此实际应用比较方便。
本文先用遗传优化算法对实际测井数据进行了处理,求取了储层的孔隙度、泥质含量及矿物组分含量,与岩心数据对比,吻合程度也较好,但由于遗传算法的早熟现象,使很多深度点并未找到最合适的解,在曲线上表现为较多的跳跃点。将复合形算法引入后,基本上没有跳跃点,且理论测井曲线与实际曲线基本重合,改进的效果很明显。另外将GA-CM算法应用到岩电参数的确定上,并以此作为求取饱和度的基础,也取得了一定的效果。
As unconventional oil resource owning to special sedimentation and tectonization, glutenitereservoir gets more and more attention in oil exploration. However it is hard to divide lithologyand set up the interpretation model accurately, due to the reservoir character of deep burial,complex lithology, low porosity and permeability, and high heterogeneity.
The optimal log interpretation paves a way for the evaluation of complex lithologyreservoir, as it can take full advantage of the logging information, and measurement error,logging response function error, geological information and working experience, lying on thegenerality geophysical inversion theory, and using optimization method, and statisticsprobability theory and error theory and quality control technology, then find the mostreasonable result from all the possible solutions.
In this paper, the relationship among lithology and oiliness and physical property andelectric nature of glutenite is studied, and a multicomponent model is proposed for welllogging interpretation. Similar with multiple mineral model, the reservoir matrix isdecomposed to many parts, however the glutenite lithology is complex, the lithic fragment isalso multicomponent, as a result, the lithic fragment is take as a whole part, and similar forplagioclase and potassic feldspar, so the matrix is make up of quartz and feldspar and lithicfragment. As to simplify the calculation, the linear well logging response equation is used.
As a core of optimal log interpretation, mathematical algorithm is deterministic part, theGenetic Algorithm is a kind of heuristic method and is applied extensively these years.However the GA method is easily converge earlier, so the paper proposed a method ofcombining the Complex Method with GA and forming a Hybrid Genetic Algorithm. TheComplex Method is a local search approach which has no require of gradient information, andit is easily used.
In this paper, genetic algorithm is first introduced to optimal log interpretation to get thereservoir parameters, such as porosity and shale content and mineral component content, whichmatched the core data well, however many catastrophe points were observed on the curves.After the CM is produced into GA, the catastrophe points disappeared and the theory welllogging curves were essentially coincident with the actual ones. Finally the method is appliedin calculating rock-electro parameters, then the desired water saturation is obtained.
引文
[1]雷群,王红岩,赵群,刘德勋.国内外非常规油气资源勘探开发现状及建议[M].天然气工业,2008,28(12):7-10.
[2]张孝珍.砂砾岩储层测井评价方法研究[D].中国石油大学(华东),2009.
[3]周磊.梨树断陷十屋油田营城组物源体系、沉积特征及储层特征研究[D].中国石油大学(华东),2010.
[4]胡复唐,等.砂砾岩油藏开发模式[M].北京:石油工业出版社,1997.
[5]李舟波.地球物理测井数据处理与综合解释[M].长春:吉林大学出版社,2003.
[6]张丽艳.砂砾岩储层孔隙度和渗透率预测方法[J].测井技术,2005,29(3):212-215.
[7]高博禹,彭仕宓,刘红歧.蒙古林砾岩油藏储层测井精细解释模型[J].测井技术,2005,29(1):55-58.
[8]罗利,姚声贤,任兴国,等.神经网络及模式识别技术在测井解释中的应用[J].测井技术,2002,26(5):364-368.
[9]胡松,张超谟,胡瑶,文得进.基于主成分分析的砂砾岩储层孔隙度计算方法研究[J].石油天然气学报(江汉石油学院学报),2010,32(3):100-104.
[10]夏宏泉,张贤辉,范翔宇等.基于神经网络法的逐点渗透率测井解释研究[J].西南石油学院学报,2001,23(1):11-13.
[11]范宜仁,葛新民,汪海龙,邓少贵.非均质性砂砾岩储层渗透率预测方法研究[J].西南石油大学学报(自然科学版),2010,32(3):6-10.
[12]傅爱兵,王善江,张豆娟,倪自高.含水饱和度指数n值计算方法探讨[J].测井技术,2007,31(5):438-440.
[13]雍世和.最优化测井解释[M].北京:石油大学出版社,1996.
[14]雍世和,孙建孟,毕卓新.用最优化方法评价复杂岩性储集层[J].测井技术,1988,12(4):18-28.
[15]肖立志.测井资料最优化解释方法求解结果的评价[J].石油物探,1989,28(2):65-72.
[16]肖立志,钟兴水.GLOBAL测井解释方法Ineoherence函数性质研究[J].石油学报,1990,11(2):49-57.
[17]邹长春,尉中良,等.利用遗传算法实现最优化测井解释[J].测井技术,1999,23(5):361-365.
[18]冯国庆,等.最优化测井解释的遗传算法实现[J].天然气工业,2002,22(6):48-51.
[19]Paulo A. Pereira,et.al..A Hybrid Genetic Algorithm for Constrained CombinatorialProblems:An Application to Promotion Planning Problems[J].Operations ResearchProceedings,2011,Part9:301-306.
[20]董勇,郭海敏,张闪.自适应混沌粒子群算法在油水两相流动分析中的应用[J].石油天然气学报(江汉石油学院学报),2011,(33)2:81-84.
[21]欧阳健,王运位,胡淑芬.复杂岩性优化测井数字处理方法[J].石油与天然气地质,1983,4(1):76-87.
[22]严成信,邹长春,张红燕.孔南火成岩测井评价方法探讨[J].测井技术,1998,22(2):101-106.
[23]王拥军,夏宏泉,陈科贵,赵良孝,张树东.塔里木碳酸盐岩储层最优化测井解释方法[J].西南石油学院学报,2000,22(4):18-21.
[24]潘保芝,闫桂京.遗传最优化测井解释方法及应用研究[C].中国地球物理学会,2001:98.
[25]肖亮,毛志强,孙中春,等.最优化方法在复杂岩性储集层测井评价中的应用[J].断块油气田,2011,18(3):342-345.
[26]林梁,张玉金.测井优化解释程序系統GCDQ[J].大庆石油学院学报,1986,2:41-52.
[27]袁秀婷.新疆塔河石炭系油气储层测井最优化评价[J].工程地球物理学报,2004,1(6):514-519.
[28]冯庆付,等.火成岩天然气储层最优化测井数字处理方法[J].天然气工业,2007,27(8):38-40.
[29]王新峰.东营凹陷砂砾岩体测井特征[J].工程技术,2011,14:704.
[30]李金铸,刘琮,卢春辉,等.东庄储层四性关系研究[J].油气田地面工程,2011,30(4):23-25.
[31]王小平,曹立明.遗传算法——理论、应用与软件实现[M].西安:西安交通大学出版社,2002.
[32]Goldberg D.E. Genetic Algorithms in Search, Optimization and MachineLearning[M].Addison-Wesley Publishing Company,1989.
[33]Holland J.H.Adaptation in Nature and Artificial Systems[M].Ann Arbor:The Universityof Michigan Press,1975.
[34]崔珊珊.遗传算法的一些改进及其应用[D].中国科学技术大学,2010.
[35]樊骥,吴红梅.基于最优化理论几种算法的比较[J].才智,2011:49.
[36]王银年.遗传算法的研究与应用[D].江南大学,2009.
[37]潘正君,康立山.演化计算[M].北京:清华大学出版社,1998.
[38]Rasoul Irani, Reza Nasimi.Evolving neural network using real coded genetic algorithmfor permeability estimation of the reservoir.Expert Systems with Applications,2011,(38):9862-9866.
[39]田云英,夏宏泉.基于多矿物模型分析的最优化测井解释[J].西南石油学院学报,2006,(28)4:9-11.
[40]Zoya Heidari,Carlos Torres-Verdín.Improved estimation of mineral and fluid volumetricconcentrations in thinly-bedded and invaded formations.SPWLA51st Annual LoggingSymposium,2010:19-23.
[41]Oberto Serra.Fundamentals of Well-log Interpretation:The acquisition of loggingdata[M].Developments in Petroleum Science,1986.