简介:Anewmulti-moderesistivityimagingsonde,withtoroidalcoilsassource,canconductthreeresistivitymeasurements:azimuthalresistivity,lateralresistivity,andbitresistivitymeasurements.Thus,theloggingtimeandcostaregreatlysaved.Thetoroidalcoilsaresimplifiedasanextendedvoltagedipoleandtheresponseequationsarederivedforahomogenousformation.Basedon3DFEM,thedepthofinvestigation(DOI),verticalresolution,circumferentialazimuthalcapacity,boreholediameter,mudresistivity,thicknessoftargetformation,andtheresistivityofthesurroundingformationandmudinvasionaresimulated.TheresultssuggestthatthethreemeasurementmodesofthenewsondearedifferentinverticalresolutionsandDOIs.Thecircumferentialdetectionabilityoftheazimuthbuttondependsonthecontrastbetweentheanomalyandformationresistivityandtheopenangleoftheanomaly.Whethertheboreholeistruncatedatthebitornothasagreatinfluenceonthesimulationresults.Theboreholeandmudinvasionaffecttheapparentresistivityinallmodes,buttheeffectsofresistivityofsurroundingformationandthicknessofthetargetformationareonlycorrectedforlateralresistivitymeasurement.
简介:随机有限断层法作为半经验半理论的地震动合成方法,综合考虑了震源、传播路径以及场地条件对地震动的影响,可在工程关注的频率范围内模拟地震动时程,为实现较大区域地震动模拟提供了一种有效的方法。本文系统阐述了随机有限断层法的基本原理、静力学拐角频率模型及动拐角频率模型的发展;分析了主要模型参数(断层尺度、破裂速度、剪切波速及应力降)、Kappa因子和介质品质因子的取值原则,并结合中国大陆的实际情况给出了相应的取值范围;最后讨论了现有模拟方法在考虑参数取值、子断层划分和子断层之间相互作用等方面的不足以及能量处理方式上的缺陷,明确了随机有限断层法发展的方向。
简介:川东北元坝地区二叠系上统长兴组储层具有储层薄、非均质性强、裂缝发育且类型多样的特点。常规叠后波阻抗反演技术难以展示岩性非均质性及识别薄储层,因此,在叠后波阻抗反演基础上,充分利用叠后地震数据及井点处的岩相信息,在井点处建立地震数据与岩性的对应模式,引入深度神经网络预测技术,将井点岩相训练结果推广至井间,再通过贝叶斯判别其对应的岩性概率,实现井间岩性预测;对比相干法裂缝预测,开展了最大似然法裂缝预测,计算每个样点的相似性,保留最小相似值。结果表明:①深度神经网络技术预测岩相与实钻吻合度高,能提高储层横向分辨能力;②最大似然法预测裂缝与相干法预测裂缝相比,可展示裂缝发育细节,表征中小尺度裂缝,与实钻结果具有更高的吻合度。
简介:Finite-difference(FD)methodsarewidelyusedinseismicforwardmodelingowingtotheircomputationalefficiencybutarenotreadilyapplicabletoirregulartopographies.Thus,severalFDmethodsbasedonthetransformationtocurvilinearcoordinatesusingbody-fittedgridshavebeenproposed,e.g.,standstaggeredgrid(SSG)withinterpolation,nonstaggeredgrid,rotatedstaggeredgrid(RSG),andfullystaggered.TheFDbasedontheRSGissomewhatsuperiortoothersbecauseitsatisfiesthespatialdistributionofthewaveequationwithoutadditionalmemoryandcomputationalrequirements;furthermore,itissimplertoimplement.WeusetheRSGFDmethodtotransformthefirstorderstress–velocityequationinthecurvilinearcoordinatessystemandintroducethehighprecisionadaptive,unilateralmimeticfinite-difference(UMFD)methodtoprocessthefreeboundaryconditionsofanirregularsurface.Thenumericalresultssuggestthattheprecisionofthesolutionishigherthanthatofthevacuumformalism.Whentheminimumwavelengthislow,UMFDavoidsthesurfacewavedispersion.WecompareFDmethodsbasedonRSG,SEM,andnonstaggeredgridandinferthatallsimulationresultsareconsistentbutthecomputationalefficiencyoftheRSGFDmethodishigherthantherest.