地震油气储层的小样本卷积神经网络学习与预测

地震储层预测是油气勘探的重要组成部分,但完成该项工作往往需要经历多个环节,而多工序或长周期的研究分析降低了勘探效率.基于油气藏分布规律及其在地震响应上所具有的特点,本文引入卷积神经网络深度学习方法,用于智能提取、分类并识别地震油气特征.卷积神经网络所具有的强适用性、强泛化能力,使之可以在小样本条件下,对未解释地震数据体进行全局优化提取特征并加以分类,即利用有限的已知含油气井段信息构建卷积核,以地震数据为驱动,借助卷积神经网络提取、识别蕴藏其中的地震油气特征.将本方案应用于模型数据及实际数据的验算,取得了预期效果.通过与实际钻井信息及基于多波地震数据机器学习所预测结果对比,本方案利用实际数据所演算结果与实际情况有较高的吻合度.表明本方案具有一定的可行性,为缩短相关环节的周期提供了一种新的途径.     

基于挤喷流效应的Russell流体因子推广及应用

在油气勘探与开发过程中,寻求能够从地震资料中直接识别储层油气的流体指示因子至关重要.由于现存多数流体指示因子都是在Biot理论假设前提下建立起来的,因此在双相孔隙介质条件下不能有效地识别流体.为此,本文基于前人提出的双孔介质统一波动理论,考虑岩石裂缝间挤喷流效应,在经典流体指示因子基础上构建了一种新的流体指示因子.该流体因子能够较好地反映岩石内孔隙流体的变化对波传播产生的影响.在实际资料的应用中,采用多种方法分析、对比该流体识别因子与以Biot理论为基础的传统流体因子的优劣,理论分析与实际资料的验证表明该流体因子对于储层中油气的检测有较高的精度和灵敏度.     

多波地震深度学习的油气储层分布预测案例

有机并有效利用纵波与转换横波在油气储层敏感度上存在的差异,有助于突出地震油气储层特征,有助于提高地震油气储层分布边界刻画的精度.基于此,本文设计了一种卷积神经网络与支持向量机方法相结合的多波地震油气储层分布预测的深度学习法（Deep Learning Method）.首先,利用莱特准则剔除所生成的多波地震属性中可能存在的异常值降低网络变体数量.然后,通过能突出多波地震油气储层特征的聚类算法和无监督学习算法构建隐藏层,用于增加网络共享,提取油气特征.最后,将增加网络罚值后的井点样本作为支持向量机预测的输入样本,以降采样后的C3卷积层属性作为学习集,进行从已知到未知的地震油气储层的预测.本方案应用于HG地区晚三叠统HGR组的碳酸盐岩油气储层预测,所预测的地震油气储层边界更加清晰,预测结果与实际情况基本吻合.应用结果表明：本论文方案不仅具有可行性,且具有有效性.     

拉梅参数直接反演方法在东海N构造致密储层烃类检测中的应用

东海N构造主要目的层为强水动力环境下发育的三角洲平原分流河道砂体。平面砂层分布不连续,横向非均质性强。受埋深压实和成岩作用等影响,储层呈低孔渗特征,岩石物理规律叠置严重。此外,深部地震数据存在缺乏大角度入射信息等问题,落实研究区致密储层流体分布范围对于勘探开发设计部署具有重要意义。本文引入一种基于拉梅参数直接反演的地震流体描述方法,通过实测井数据的岩石物理定性和定量判析,优选流体敏感参数,进而结合拉梅参数的两项AVO模型参数化方程,从叠前道集中分别提取拉梅参数的AVO属性体。然后利用有色反演技术直接提取流体敏感弹性信息,以指导地震流体描述。实例应用表明,该方法的烃检结果与测井解释成果匹配度高,能够有效刻画研究区致密储层流体展布规律,可为新领域油气资源发现提供重要技术支撑。      

A new seismic attribute for ambiguity reduction in hydrocarbon prediction

Hydrocarbon prediction from seismic amplitude and amplitude‐versus‐offset is a daunting task. Amplitude interpretation is ambiguous due to the effects of lithology and pore fluid. In this paper, we propose a new attribute “J” based on a Gassmann–Biot fluid substitution to reduce ambiguity. Constrained by seismic and rock physics, the J attribute has good ability to detect hydrocarbons from seismic data. There are currently many attributes for hydrocarbon prediction. Among the existing attributes, far‐minus‐near times far and fluid factor are commonly used. In this paper, the effectiveness of these two existing attributes was compared with the new attribute. Numerical modelling was used to test the new attribute “J” and to compare “J” with the two existing attributes. The results showed that the J attribute can predict the existence of hydrocarbon in different porosity scenarios with less ambiguity than the other two attributes. Tests conducted with real seismic data demonstrated the effectiveness of the J attribute. The J attribute has performed well in scenarios in which the other two attributes gave inaccurate predictions. The proposed attribute “J” is fast and simple, and it could be used as a first step in hydrocarbon analysis for exploration.     

基于地震属性约简的深度学习储层物性参数预测:以莺歌海盆地乐东区为例

储层物性参数作为描述储层特性、储层建模和流体模式的重要指标,其准确估算可以为储层预测提供有力参考依据,但传统储层物性参数反演方法无法兼顾反演精度及空间连续性。针对上述问题,本文引入地震属性作为深度学习算法输入,针对地震属性之间存在的信息冗余特征,利用随机森林-递归消除法对地震属性进行约简预处理,最终建立一种基于地震属性约简的储层物性参数预测方法。实际数据测试结果表明,地震属性约简的深度学习储层物性参数预测结果具有良好的精度及横向分辨率,证实本文方法的有效性。      

Rock physics model for seismic velocity & fluid substitution in sub-resolution interbedded sand–shale sequences

The measured geophysical response of sand – shale sequences is an average over multiple layers when the tool resolution (seismic or well log) is coarser than the scale of sand – shale mixing. Shale can be found within sand – shale sequences as laminations, dispersed in sand pores, as well as load bearing clasts. We present a rock physics framework to model seismic/sonic properties of sub-resolution interbedded shaly sands using the so-called solid and mineral substitution models. This modelling approach stays consistent with the conceptual model of the Thomas–Stieber approach for estimating volumetric properties of shaly sands; thus, this work connects established well log data-based petrophysical workflows with quantitative interpretation of seismic data for modelling hydrocarbon signature in sand – shale sequences. We present applications of the new model to infer thickness of sand – shale lamination (i.e., net to gross) and other volumetric properties using seismic data. Another application of the new approach is fluid substitution in sub-resolution interbedded sand–shale sequences that operate directly at the measurement scale without the need to downscale; such a procedure has many practical advantages over the approach of “first-downscale-and-then-upscale” as it is not very sensitive to errors in estimated sand fraction and end member sand/shale properties and remains stable at small sand/shale fractions.     

应用地震数据体结构特征法预测塔河油田八区油层分布规律

地震数据体结构特征,指对二维或三维地震数据体中每一地震道离散数据点按时间顺序排列所显示的波形特征。应用地震数据体结构特征法对储层进行油气预测,是近年来新兴的一项储层预测技术。塔河油田八区的主产层为碳酸盐岩裂缝一缝洞型储层,具有很强的非均质性,给储层的油气预测带来了一定的困难。针对塔河油田八区特殊的地质条件,采用了地震数据体结构特征法对该区奥陶系储层进行了油气预测,并对油区内、外的区域进行了有利区块的划分。在有利区块中设计的18口钻探井位,经钻井证实均获得了较高的油气产量,取得了很好的经济效益。     

Predicting the distribution of reservoirs by applying the method of seismic data structure characteristics： Example from the eighth zone in Tahe Oilfield

Seismic data structure characteristics means the waveform character arranged in the time sequence at discrete data points in each 2-D or 3-D seismic trace. Hydrocarbon prediction using seismic data structure characteristics is a new reservoir prediction technique. When the main pay interval is in carbonate fracture and fissure-cavern type reservoirs with very strong inhomogeneity, there are some difficulties with hydrocarbon prediction. Because of the special geological conditions of the eighth zone in the Tahe oil field, we apply seismic data structure characteristics to hydrocarbon prediction for the Ordovician reservoir in this zone. We divide the area oil zone into favorable and unfavorable blocks. Eighteen well locations were proposed in the favorable oil block, drilled, and recovered higher output of oil and gas.     

多分量地震资料预测松辽盆地兴城地区深层火山岩与有利含气带

松辽盆地断陷期天然气有利带预测是扩大该盆地勘探领域的一个重要组成部分.为在油气资源基地松辽盆地开展大范围深层油气预测，选择徐家围子断陷内局部构造—升平-兴城构造进行地面三分量地震勘探，得到一个可行的预测方案.在深探井控制下，利用三分量地震资料，进行P波、P-SV波层位标定、震相对比，得到研究工区断陷期火山岩分布的局部预测；经过目标层段振幅比、频率比等物性参数的计算对比，得到该工区断陷期营城组有利含气区预测.进一步分析了研究工区区域构造对断陷期地层的控制作用.从烃源岩种类、成熟程度以及火山岩储层性能几方面综合分析，认为在更大范围内存在深层有利含气区带是可能的.最后指出在深层有利含气区带预测中采用三维三分量技术的必要性，以及无井控预测的基本方法，包括构造特征、地层特征预测知识的延用，多参数特征间相似性和综合比较等.     

基于随机森林算法的陆相沉积烃源岩定量地震刻画:以东海盆地长江坳陷为例

烃源岩的定量地震刻画对于勘探开发区块的优选、盆地油气资源量的估算都具有重要意义.陆相沉积环境下的浅湖或半深湖相的烃源岩横向变化快,其空间展布需要依靠钻井约束下的反射地震进行刻画,但是其地震弹性特征与岩性和有机质含量的映射关系呈现高度非线性化,因而很难利用传统基于地震岩石物理模型驱动的烃源岩地震预测方法进行有效刻画.本文以低勘探区的东海盆地长江坳陷为例,提出了一种在数据驱动的机器学习框架下,综合利用地质约束、钻井录井、测井、地球化学和叠前地震数据进行烃源岩的定量地震刻画的工作流程.其核心思想是利用随机森林集成学习算法对小样本数据表现优异的特征,以井位处的测井弹性数据(纵波速度和密度)、岩性、地球化学标定的总有机碳含量(TOC)为样本标签数据,在地质导向约束下通过随机森林算法生成学习网络,并将该网络与叠前地震反演结果相结合,采取先预测泥岩再预测总有机碳含量的“两步走”策略,完成对烃源岩空间分布及其非均质性的定量地震刻画,并对预测结果的不确定性进行评价.测试结果显示,随机森林算法相较于其他的机器学习算法能够更准确的识别陆相沉积地层的泥岩,并比传统的利用阻抗转化方法获得更可靠的总有机碳含量预测结果.     

基于方位地震数据的地应力反演方法

在页岩油气藏的开发和勘探阶段,需要对储层进行水力压裂改造,形成有利于油气聚集和运移的裂缝.地应力是进行水力压裂改造的重要参数,能够决定裂缝的大小、方向以及分布形态,影响着压裂的增产效果,且最大和最小水平应力差异比（ODHSR,Orthorhombic Differential Horizontal Stress Ratio）是评价储层是否可压裂成网的重要因子.本文探讨了基于地震数据估算地应力的方法,以指导页岩气的水力压裂开发.首先,利用叠前方位地震数据反演得到地层的弹性参数和各向异性参数;其次,基于正交各向异性水平应力差异比近似公式,利用反演得到的弹性参数和各向异性参数估算地层的ODHSR;最后,选取某工区的裂缝型页岩储层的叠前方位地震数据对该方法进行实际应用.实际工区地震数据应用表明,基于叠前方位地震数据反演得到的ODHSR能够有效的识别储层中易于压裂成网的区域.     

泥页岩岩石物理建模研究

泥页岩由于其复杂的岩石特性（主要是裂缝及有机质的存在）,目前还没有有效的岩石物理模型可以较为精确的模拟其性质.本文在自洽模型和微分等效介质模型的基础上,引入Berryman三维孔隙形态及Brown-Korringa固体替代技术,建立适用于富有机质泥页岩的新型岩石物理模型.在此基础上进行正演分析,讨论不同孔隙形态对于自洽模型的临界孔隙度以及岩石速度的影响.正演分析的结果表明即使将未知的混合岩石作为背景岩石,微分有效介质模型的引入使得固体相和流体相仍然不是对称的,临界孔隙度不一定要落在0.4到0.6之间.且不同的孔隙形状对于自洽模型的临界孔隙度以及岩石的速度具有明显的影响.此外,基于岩石物理模型,文章讨论了不同孔隙形态、不同泥质含量时有机质对于岩石弹性性质的影响.最后利用一口页岩气井对该模型进行验证,预测的纵横波速度与测井结果吻合的很好,证明了该模型对于富有机质泥页岩的适用性.     

Compaction trend versus seismic anisotropy in shaly formations

Shales comprise more than 60% of sedimentary rocks and form natural seals above hydrocarbon reservoirs. Their sealing capacity is also used for storage of nuclear wastes. The world's most important conventional oil and gas reservoirs have their corresponding source rocks in shale. Furthermore, shale oil and shale gas are the most rapidly expanding trends in unconventional oil and gas. Shales are notorious for their strong elastic anisotropy, i.e., so‐called vertical transverse isotropy. This vertical transverse isotropy, characterised by a vertical axis of invariance, is of practical importance as it is required for correct surface seismic data interpretation, seismic to well tie, and amplitude versus offset analysis. A rather classical paradigm makes a clear link between compaction in shales and the alignment of the clay platelets (main constituent of shales). This would imply increasing anisotropy strength with increasing compaction. Our main purpose is to check this prediction on two large databases in shaly formations (more than 800 samples from depths of 0–6 km) by extracting the major trends in the relation between seismic anisotropy and compaction. The statistical analysis of the database shows that the simultaneous increase in density and velocity, a classical compaction signature, is quite weakly correlated with the anisotropy strength. As a consequence, compaction can be excluded as a major cause of seismic anisotropy, at least in shaly formations. Also, the alignment of the clay platelets can explain most of the anisotropy measurements of both databases. Finally, a method for estimating the orientation distribution function of the clay platelets from the measurement of the anisotropy parameters is suggested.     

琼东南盆地崖南区少井岩性建模技术研究

崖南区是琼东南盆地已证实的富生烃区,几口已钻井都已证实主要目的层段为高压地层,利用常规的压力预测方法预测新钻井的压力会出现较大的误差.若是从区域应力角度入手预测新钻井的压力误差会减小,其预测基础为岩性模型.对于已开发的油气田,利通常规的岩性建模方法可以建立较好的岩性模型;但是对于崖南区而言,由于地震资料品质不是很好,同时本区钻井较少,很难通过常规的建模方法建立岩性模型,所以本区研究重点是如何利用少井建立岩性模型.通过研究认为若完成崖南区的岩性建模必须改进建模流程,改进的岩性建模流程克服了常规岩性建模在崖南区存在的问题,主要有三方面的优点:1)不采用相模型约束岩性建模,解决了由于研究区相模型划相较粗很难约束岩性模型建立的问题;2)属性模型控制岩性模型的横向变化趋势,解决了几种常规属性与岩性间没有较好关系的问题;3)利用泥质含量结合岩性资料建立岩性体,得到的岩性模型比较接近实际情况.C井钻前完成岩性模型建立,利用C井井点位置提取岩性数据与本井钻后录井岩性数据对比,发现预测岩性与录井岩性的吻合程度很高,证明改进的岩性建模思路在崖南少井区可用.     

三维AVO解释技术在洪泽地区岩性油藏勘探中的应用

洪泽地区由于沉积的特点，储层横向变化快，油藏受构造、岩性、油源多因素控制。在对该区三维AVO属性体解释中，利用多元回归方法求取了横波曲线，分岩性和含油气性统计了纵、横波、泊松比参数分布规律，建立了本区的含油砂岩的正演模型，从而降低了AVO解释的多解性。通过井-震结合对四种AVO属性数据体进行了标定，并确定了各属性体应用范围，进而进行了储层和含油气检测。实践表明，该方法能有效地利用AVO属性数据体进行储层预测及油气检测，具有一定的推广价值。     

Generalized effective medium resistivity model for low resistivity reservoir

With the advancement in oil exploration,producible oil and gas are being found in low resistivity reservoirs,which may otherwise be erroneously thought as water zones from their resistivity.However,the evaluation of low resistivity reservoirs remains difficult from log interpretation.Since low resistivity in hydrocarbon bearing sands can be caused by dispersed clay,laminated shale,conductive matrix grains,microscopic capillary pores and high saline water,a new resistivity model is required for more accurate hydrocarbon saturation prediction for low resistivity formations.Herein,a generalized effective medium resistivity model has been proposed for low resistivity reservoirs,based on experimental measurements on artificial low resistivity shaly sand samples,symmetrical anisotropic effective medium theory for resistivity interpretations,and geneses and conductance mechanisms of low resistivity reservoirs.By analyzing effects of some factors on the proposed model,we show theoretically the model can describe conductance mechanisms of low resistivity reservoirs with five geneses.Also,shale distribution largely affects water saturation predicted by the model.Resistivity index decreases as fraction and conductivity of laminated shale,or fraction of dispersed clay,or conductivity of rock matrix grains increases.Resistivity index decreases as matrix percolation exponent,or percolation rate of capillary bound water increases,and as percolation exponent of capillary bound water,or matrix percolation rate,or free water percolation rate decreases.Rock sample data from low resistivity reservoirs with different geneses and interpretation results for log data show that the proposed model can be applied in low resistivity reservoirs containing high salinity water,dispersed clay,microscopic capillary pores,laminated shale and conductive matrix grains,and thus is considered as a generalized resistivity model for low resistivity reservoir evaluation.     

Application of Multiboost-KELM algorithm to alleviate the collinearity of log curves for evaluating the abundance of organic matter in marine mud shale reservoirs: a case study in Sichuan Basin,China

The total organic carbon (TOC) content reflects the abundance of organic matter in marine mud shale reservoirs and reveals the hydrocarbon potential of the reservoir. Traditional TOC calculation methods based on statistical and machine learning have limited effect in improving the computational accuracy of marine mud shale reservoirs. In this study, the collinearity between log curves of marine mud shale reservoirs was revealed for the first time, which was found to be adverse to the improvement of TOC calculation accuracy. To this end, a new TOC prediction method was proposed based on Multiboost-Kernel extreme learning machine (Multiboost-KELM) bridging geostatistics and machine learning technique. The proposed method not only has good data mining ability, generalization ability and sound adaptivity to small samples, but also has the ability to improve the computational accuracy by reducing the effect of collinearity between logging curves. In prediction of two mud shale reservoirs of Sichuan basin with proposed model, the results showed that the predicted value of TOC was in good consistence with the measured value. The root-mean-square error of TOC predicting results was reduced from 0.415 (back-propagation neural networks) to 0.203 and 1.117 (back-propagation neural networks) to 0.357, respectively; the relative error value decreased by up to 8.9%. The Multiboost-KELM algorithm proposed in this paper can effectively improve the prediction accuracy of TOC in marine mud shale reservoir.     

Mapping deep targets based on integrated 3D MT-gravity interpretation: a case study

Mapping deep geological hydrocarbon targets is of significant importance in basin exploration. In areas lacking reliable seismic data, magnetotelluric (MT) and gravity explorations are helpful to delineate the distribution of potential deep geological hydrocarbon targets. Here we investigate the effectiveness of the integrated 3D MT and gravity explorations for mapping the potential deep hydrocarbon source rocks. The result based on the data from the W Basin (part of the Ordes Basin) of China demonstrates that the method is efficient and economical for basin exploration. The method is particularly useful in target areas which are of great interest for oil and gas exploration but lack high quality seismic data. In our method, we first use the high-precision 3D small-bin MT data acquisition to improve the data accuracy. Then we perform datum static correction method and apply 3D inversion to obtain the3D resistivity distribution. We also develop a layered resistivity model based on resistivity logging to assist the interpretation of the inverted 3D resistivity data so as to derive an initial 3D geological model. Starting from the initial model, we use 2D gravity data to update the model via 2D inversion line by line, and then pass the updated model for the next round of the 3D MT inversion. The integrated inversion is implemented iteratively so the model converges to satisfy the need of final geological analysis. The application to the W Basin shows that we could successfully delineate the geological distribution of the potential deep hydrocarbon source rocks within the basin and map the thickness of the upper Paleozoic.     

异常地层压力声波衰减分析及地层压力地震数据外推反演

澳大利亚西北大陆架卡拉汶盆地含有丰富的油气资源, 但其地质条件复杂, 普遍存在异常地层高压分布,特别是该区区域盖层-中深层巨厚的Muderong页岩层内存在局部异常高压带,给石油钻井带来巨大困难.本文根据卡拉汶盆地北部地区10余口井的实测地层压力资料、声波测井数据和联井地震剖面, 研究该页岩层内异常压力千赫兹尺度声波响应的衰减特征,及其与地层有效压力之间的关联计算模型.通过联井地震数据的井-震相关性分析进行井-震过渡,研究井旁地震数据在异常高压带的衰减特征,及其与地层有效压力之间的关联计算模型,从而实现井震资料联合进行地层压力地震外推反演.研究表明,Muderong页岩层地层有效压力与井中声波速度/阻抗品质因子和井旁地震阻抗品质因子具有良好的正向相关性,根据二者交汇数据拟合经验公式反算得到的各井有效压力结果基本反映出实测有效压力曲线的变化特征.卡拉汶盆地北部地区地震数据具有良好的井-震相关性,地震资料品质相对可靠,确保了声-压相关分析的井-震过渡.最后,通过地震阻抗品质因子与地层有效压力之间的关联计算模型,进行该区异常地层压力地震外推反演,预测结果与钻井证实的异常压力分布特征基本符合.本研究为卡拉汶盆地中深部区域盖层局部异常高压带的地层压力外推反演提供了一种有效途径.