【作者】 周锋德；

【导师】 唐仲华； 姚光庆；

【作者基本信息】 中国地质大学 ， 油气田开发工程， 2009， 博士

【摘要】 宝浪油田地理位置处于新疆维吾尔自治区巴音郭楞蒙古自治州焉耆回族自治县境内。构造上位于焉耆盆地博湖坳陷北部凹陷宝浪苏木构造带。主要产油层为三工河组。沉积相以上辫状河三角洲前缘水下分流河道和下三角洲前缘水下分流河道为主。岩性包括砂砾岩、砾岩、粗砂岩、细砂和粉砂岩等。岩石成分成熟度和结构成熟度较低。储层属辫状河三角洲上发育起来的粗粒低渗储层类型。单层砂体厚度从1m至6m不等，常见多层叠加。河道频繁迁移加剧了储层的非均质性。不同砂体内平均孔隙度从12%-14%，平均渗透率从2．972×10^-3μm²-27．3×10^-3μm²。储层属性和流体分布复杂，砂体厚度大（Ⅲ₁油组接近20m），同时，独特的高陡背斜构造及裂缝的发育，使该地区储层成为典型的非常规储层类型：低密度、低孔孔隙度、低渗透率、粗粒、高粘土矿物、裂缝和复杂构造。开发13年以来，一些与地质有关的问题仍然影响该地区储层的有效开发。例如：砂体的三维分布；流动单元的分布；裂缝的特征及其对流体渗流的影响；含水率上升的控制因素，等。本文主要利用地质统计学、随机模拟和数值模拟等方法，首次提出并开展了针对低渗含裂缝储层进行了综合表征和模拟的方法和实现；首次采用小尺度和大尺度数值模拟相结合的方法系统评价了不同裂缝及不同流动单元组合情况下的地质模型的开发特征及参数。具体内容主要包括以下五个方面：第一，流动单元综合研究：在地层划分、岩心分析、岩心测试、储层特征参数和岩石物理相的基础上，采用聚类方法划分了流动单元类型。通过多元回归方法建立了流动单元的预测模型。对比砂体的空间分布，分析了流动单元在平面和剖面上的分布规律。第二，裂缝单元综合表征：通过岩心观察测定了岩心裂缝的密度、倾角、走向等参数；并通过测井和构造曲率方向分别建立了低角度裂缝和高角度裂缝的预测模型，预测并绘制了不同裂缝的裂缝密度平面分布图。第三，对建立随机离散裂缝分布模型的方法和参数进行了研究，并对不同裂缝密度的随机离散分布模型进行了比较。第四，通过对预测的低角度和高角度裂缝的平面分布图进行数字化，并对裂缝的计数密度进行了面积密度的转化，最后建立了研究区的低角度和高角度裂缝随机离散分布模型。并采用多界面（多级界面主要包括油组界面、小层界面和单砂体界面）及流动单元的约束方法建立了低渗基质的地质模型。第五，对岩心规模及小尺度规模的不同流动单元及不同裂缝产状的模型进行了数值模拟和比较。并在储层地质模型的粗化基础上，对Ⅲ₁油组进行了数值模拟，在历史拟合的基础上，分析了剩余油分布和剩余可采储量的平面分布。通过研究，取得的主要成果和认识如下：（1）通过FZI=4μm、FZI=2．72μm、FZI=1．85μm和FZI=1．36μm．可以把本地区流动单元类型可以分为5类，流动单元A、流动单元B、流动单元C、流动单元D和流动单元E。在流动单元基础上建立的储层孔渗测井解释模型预测的储层物性与岩心分析的物性的相关系数大于0．7，能满足精细储层建模的要求。（2）研究区裂缝类型按成因主要分为两类：构造成因和构造-沉积复合成因。复合成因的层状裂缝最发育，断层有关的裂缝次之。将近60%裂缝属于复合成因，40%属于构造成因。裂缝按倾角主要分为两种类型，低角度裂缝和高角度裂缝。高角度裂缝走向以东北、南西和东南方向为主，平均为156°。其倾角分布范围为15°-85°之间，大多数在40°-75°之间分布，平均为54．2°。背斜的不同部位倾角不同，西北部大于东南部。低角度裂缝的倾角在0°-20°之间不等，常呈“薄饼状”，一些裂缝的宽度小于5cm。（3）三孔隙度方法能够很好的预测该地区的低角度裂缝的密度，临界值为0．3。利用SPSS软件建立了低角度裂缝密度预测的多元回归方程，利用该方法预测的裂缝密度和岩性测量的裂缝之间的相关系数为0．716。在模型建立的基础上，编写了VB代码自动计算了除部分没有补偿中子、声波和密度测井的所有井，并分析了低角度裂缝的密度平面分布。高角度裂缝的预测模型通过分析裂缝发育指数和岩心裂缝密度的关系来建立。（4）流动单元组合下的岩心尺度数值模拟结果是不同的流动单元储层单独开采下的采收率最高，其次是流动单元A和流动单元B组合，其他组合均低于这两种组合。流动单元组合下的小尺度模拟表明，在相对注入速度相等的情况下，井距对综合采出程度的影响较小；而注入速度对综合采出程度影响较大。（5）四种不同倾角的裂缝模拟结果显示，在含水量达到98%时，流动单元A、流动单元B和流动单元C中采收率的关系是135°＜45°＜0°＜90°＜无裂缝模型。（6）和没有裂缝的模型比较，具有裂缝的岩心当含水饱和度达到98%时，流动单元A和流动单元B的含水率减少，流动单元C的含水率增加。由于串通裂缝的层间流动影响，流动单元A的采收率增加，流动单元B和流动单元C的减少，总的采收率降低。（7）离散裂缝等效模拟结果显示，裂缝的等效处理网格大小对基质渗透率较低的模型模拟结果影响较大。当垂向网格大小等于模型厚度时，模拟的含水率比离散裂缝模型的结果低20%。合理的垂向网格大小和基质渗透率成正比关系。（8）多界面和流动单元共同控制储层地质建模方法能很好的控制砂体形态和砂体属性参数分布。流动单元A、B、C、D、E和泥岩的最大主变程分别为635m、429m、520m、558m、556m和783m。孔隙度在流动单元B、C、D和E中的最大主变程分别为483m、676m、516m和628m。孔隙度在流动单元B、C、D和E中的最小次变程分别为248m、200m、264m和232m。流动单元A、B、C、D和E的渗透率控制范围分别为52．5～162．6×10^-3μm²、18．3～97×10^-3μm²、6．6～51．6×10^-3μm²、2．5～17．8×10^-3μm²、1．4～16．8×10^-3μm²。最后，建立了各种储层属性分布模型。（9）运用随机模拟方法建立了不同裂缝参数条件下的裂缝分布模型，探讨了裂缝密度参数与裂缝模型结果参数的关系。当裂缝面积密度为0．05m²/m³、0．1m²/m³、0．5m²/m³、1m²/m³、2m²/m³和10m²/m³时，裂缝模型的平均孔隙度分别为0．0000、0．0002、0．0019、0．0028、0．0087和0．0461；平均渗透率分别为1×10^-3μm²、10×10^-3μm²、79×10^-3μm²、119×10^-3μm²、369×10^-3μm²、1260×10^-3μm²和1941×10^-3μm²；裂缝形状系数平均值分别为3984m^-2、371m^-2、1476m^-2、3×10⁶m^-2、8×10⁵m^-2、和1×10⁵m^-2。粗化的孔隙度和渗透率值与裂缝面密度成直线关系。（10）利用数值模拟方法对（9）的模型在基质结合不同流动单元情况下的渗流特征进行了对比。发现不同模型的采收率和累计采油量与裂缝面积密度没有直线关系。采收率在基质为流动单元A、流动单元B和流动单元C时的拐点分别为2m²/m³、2m²/m³和1m²/m³；累计产油量在裂缝面密度大于0．5 m²/m³时递增，小于0．5 m²/m³时递减。注采井连线与裂缝走向平行时的采收率和累计产油量均小于相应的在注采井连线与裂缝走向垂直下的模拟结果。而含水饱和度达到98%的时间关系为：当基质为流动单元A时，垂直模型晚于平行模型，当基质为流动单元B和C时，垂直模型早于平行模型。（11）运用随机模拟方法建立了Ⅲ₁油组的高角度裂缝和低角度裂缝离散分布模型。并利用Oda方法对该离散裂缝网络模型进行了粗化，得出了裂缝孔隙度、渗透率和裂缝形状指数。模型的平均孔隙度为0．4%，I、J和K方向的平均渗透率分别为11×10^-3μm²、22×10^-3μm²和13×10^-3μm²。I、J和K方向的平均裂缝间距分别为2．78m、4．19m和0．58m。整个模型的裂缝形状指数为18836/m²．在该模型的基础上，利用数值模拟方法对剩余油分布进行了研究。剩余油分布主要受注采井网不完善、次级断层、砂体分布和流动单元储层非均质性等因素影响控制和影响。更多还原

【Abstract】 Baolang oilfield is situated in Yanqi,the Hui nationality autonomous county of BayingolMonglian autonomous state,Xinjiang Vygur Autonomous Region.It is also situated on theBaolang-Sumu structural zone,north of Bohu depression,Yanqi basin.The main productionformation is Sangonghe formation of Baolang oilfield,Northwest China.The facies are,underwater distributary channel of braided delta front in upper section,and distributary channel ofbraided delta plain in lower section.Main lithologies are Gravel,Pebbled Sandstone,Coarsesandstone,Siltstone and fine sandstone,etc.The maturities of rock content and rock structure arelow.The reservoir is a coarse grain and low permeability reservoir （LPR） developed frombraided delta.The sand thickness varies from lm to 6m,with distinct multi-interlayer.Thechannel path change frequently,which caused the high heterogeneity reservoir.The averageporosity ranges from 12% to 14%,and average permeability ranges from 2.97 mD to 27.3 mD,ofdifferent oil groups,within Sangonghe formation.Distribution of reservoir properties and fluidare complex.The thickness is large,average thickness ofⅢ1-oilgroup is about 20 meters.Alsothe structure is unique with the growth of many fractures.The reservoir is a representative ofunconventional reservoirs,low density,low porosity,low permeability,coarse grain,high claycontent,fractures and complex structure.After 13 years’ production,some geological related problems still affects the efficientexploration of the reservoir.First is the 3D sand structure and flow unit distribution,second is fracture characterization and how it affects the fluid flow.third are the controlling factors thatcause increasing water production.All these problems need to be studied to enhance the oilproduction of this area.This thesis brings forward a systematic and efficient method to construct an integratedmodel for fractured low permeability reservoir,and evaluate the fluid flow performance withindifferent orientation fracture and different flow units reservoirs by using statistics,geostatistics,stochastic modeling,and large and small scale numerical simulation methods.The main contents of this thesis include five parts:First,divide the FU type based on layers correlation,core analysis,core test,reservoircharacter parameters,petrophysical facies,etc.by clustering method.Then build the FUprediction model by multi regression method.At last,figure out and analysis the FU distributionin section and plane,integrated with the distribution of sandstone.Second,make an integrated characterization for the fractures,include measure the fractureline density,dip,and azimuth,and construct the identifying and predicting model for low anglefracture and high angle fracture by log and structural curvature,respectively.At last,plot thecontour map of fracture density.Third,compare the fracture modeling results which is constructed by different stochasticmodeling parameters and methodsFourth,build 3D discrete fracture modeling for low angle fracture and high angle fracture,based on digitize and convert the fracture line density into fracture area density.And,construct ageo-model for the low permeability matrix,controlled by multi-surface and flow units.Fifth,make comparisons of the fluid flow performance for models with different flow unitsand different fracture character by large- and small-scale simulation models.Also,make a casenumerical simulation for the III1 oil layer group,and analyze the oil distribution character basedon well history lnatch.The main results as follows:1.The FU are divided into 5 types,FU-A,FU-B,FU-C,FU-D,and FU-E,by FZI=4μm,FZI=2.72μm,FZI=1.85μm and FZI=1.36μm,respectively.The correlation coefficientbetween predicted permeability,controlled by FUs,and core measured permeability ishigher than 0.7.The predicted permeability can meet the demand for fine reservoircharacterization.2.The fractures in the studied area can be divided into two types,tectonic and complex（include tectonic and sedimentary） formation.Complex formation fracture is frequentlyobserved,fault-related fracture is second.Complex formation fractures account for 60%,while tectonic fracture account for 40%.Ahnost,the fractures are divided into two types,low angle fracture and high angle fracture by the fracture dip.The azimuth of HAF,has three directions,NE,SW,and SE,with average of 156°.Dip angles are varied from15-85°,most of them between 40-75°,and the average is 54.2°.It is varied in differentsections of the anticline,NW higher than SE.Dip of LAF ranges from 0°to 20°,havingno constant trend direction.They growth like“thin biscuit”,some fracture space is smallthan 5cm.3.Tri-porosity can be used to predict the fracture intensity （FI） of LAF.The critical valueis 0.3.Multiple regression method was used to construct the fracture intensity calculationmodel by software SPSSTM.The correlation coefficient between predict FI and cored FIis 0.716.With this model,VB programs were written and used to calculate the FI ofmost wells except some wells which having no logs of CNL,AC,and DEN.Then thecontour maps of predicted fracture density distributions were plotted.HAF distribution ispredicted by calculating and comparing the fracture growth index with the cored fractureintensity.4.The FU grouped core-scale simulation results show that it got the highest recovery witheach FU produced individually,and it got an equivalent recovery with grouping FU-Aand FU-B and separating FU-C.Other group-models got lower recovery.Pilot studyresults show that there are no big changes of the total oil recovery for different wellspacing model when the relative pore volume injection speed is equivalent;but there arebigger differences for different injection speed models with same well spacing.5.Four simulation results,each with a different orientation fracture,show that when watercut increased up to 98%,the relationship of oil recovery is,fracture oriented 135°againstdisplacing direction＜45°＜0°＜90°＜Non-fracture model,with FU-A,FU-B and FU-Cmatrix.6.Compared with non-fracture model,when water cut increasing up to 98%,the water cutof FU-A and FU-B decrease,but that of FU-C increase.The recovery of FU-A increase,of FU-B and FU-C decrease,and the total recovery decrease,since the cross flowthrough fracture.7.The simulation results for discrete fracture with equivalent method show that the impactis bigger by fracture equivalent grid size when the permeability of matrix is lower;whenthe cell size in vertical equal the reservoir thickness,the decrease of water cut more than20 %;the suitable grid size in vertical is proportionally with matrix permeability.8.The multi-surfaces and FU controlled modeling method can simulate the sandstoneaccurately.Also the distribution of properties can be simulated properly.The biggestmajor range of Mud,FU-A,FU-B,FU-C,FU-D and FU-E are 635m,429m,520m,558m,556m,and 783m,respectively.The maximum major range of porosity in FU-B,FU-C,FU-D and FU-E are 483m,676m,516m and 628m,respectively.The minimumMinor range of FU-B,FU-C,FU-D and FU-E are 248m,200m,264m and 232m,respectively.The transformations of permeability versus FU-A,FU-B,FU-C,FU-D and FU-E are 52.5 to 162.6,18.3 to 97,6.6 to 51.6,2.5 to 17.8 and 1.4 to 16.8,respectively.Several properties models have been generated.9.Discrete fracture models with different fracture parameters were constructed andcompared.It shows that when fracture area density are 0.05m²/m³,0.1m²/m³,0.5m²/m³,1 m²/m³,2m²/m³,and 10m²/m³,the average porosity are 0.0000,0.0002,0.0019,0.0028,0.0087 and 0.0461,respectively;the average sigma are 3984m^-2,371m^-2,1476m^-2,3x106m^-2,8x105m^-2,and 1x105m^-2,respectively;the average permeability are lmD,10mD,79mD,119mD,369mD,1260mD and 1941mD,respectively.The upscaledporosity and permeability of the fracture model have linear relationship with fracturearea density.10.The simulation results,using above models,show that there are no linear relationshipbetween oil recovery and cumulative oil production and fracture area density.Theinflexions of the oil recovery trend for FU-A,FU-B and FU-C are 2m²/m³,2m²/m³ and1 m²/m³,respectively.Cumulative oil production decrease when fracture area density lessthan 0.5m²/m³,and increase when it bigger than 0.5m²/m³.When the production andinjection well situated in fracture azimuth （Parallel model）,the oil recovery andcumulative oil production both less than that when they situated perpendicular withfracture azimuth （Perpendicular model）.The time,when water cut up to 98%,ofperpendicular models is later than parallel model when matrix is FU-A,andperpendicular model is earlier than that of parallel model when matrix is FU-B andFU-C.11.DFN model of HAF and LAF have been generated with stochastic modeling methodrespectively.And then the DFN model was upscaled for porosity,permeability,sigmawith Oda method.The average porosity of total model is 0.4%.The average permeabilityin I-,J-,and K-direction,of total model are l lmD,22mD and 13mD,respectively.Themean of fracture space in I-,J-,K-direction are 2.78m,4.19m,and 0.58m respectively.The average sigma of total model is 18836 m^-2.The residual oil is controlled by,inefficient network between production and injection,the little fault,sandstonedistribution and the heterogeneity properties of FU.更多还原