【作者】 郭玲莉；

【导师】 刘力强；

【作者基本信息】 中国地震局地质研究所 ， 固体地球物理学， 2013， 博士

【摘要】 地震过程伴随着岩体的破裂与断层的摩擦滑动失稳过程。断层的扩展与失稳过程一直是地震学研究的基本问题，观测变形过程物理场演化特征，建立失稳模式，对于理解和认识地震机制具有重要的参考和应用价值。地震过程可以简单地划分为地震孕育阶段，地震发生阶段，与震后调整阶段。三阶段中，地震发生过程瞬间的记录尤其缺乏，特别是对地震瞬间的近场力学过程所知甚少。但是，短时间内，大面积提高现场地震力学过程的瞬态场的观测具有相当难度。因此，在实验室模拟断层失稳滑动过程，对近断层瞬态场进行高频细微研究，对理解现场地震过程有更重要的意义。本文以典型粘滑失稳模型和预制断层扩展模型为研究对象，实验样品选用房山花岗闪长岩，使用新研制的多通道高频高精度应变观测系统和速度测量系统，配合声发射系统和数字散斑系统进行观测，在双轴伺服加载系统上进行模拟实验。首先描述了断层带局部应变、断层扩展的声发射信号、断层宏观错动速度等结构参数，分析瞬态高速滑动过程中各种力学参数的变化规律。其次研究了断层破裂的启动，扩展、停止等阶段上各参数的变化特征、作用以及他们之间的相互关系，从微观、瞬态角度认识断层失稳滑动的物理本质。接下来在应力与应变空间上描述了地震过程的区域应力路径和局部应变路径，讨论自发地震与诱发地震的应变变形阶段，从另一角度为现场地应力观测以及区域应力分析提供参考。同时探讨了粘滑类型、应力降大小与震级的关系，为研究摩擦粘滑与天然地震的关系提供基础。研究内容和主要结论如下：1）三维断层扩展模型的实验结果显示，岩桥区断层贯通是一个快速过程，先多点局部扩展，后跳跃式连接。在断层贯通之后，样品整体崩跨之前，存在一个相对稳定的阶段，持续时间约几十毫秒。这为全程研究断层扩展过程提供机会。2）通过多组双剪模型、斜面剪切模型和拐折模型的粘滑实验的实验模拟，发现了断层粘滑失稳过程的演化具有特定的模式，可以划分为三阶段：预滑动阶段、高频震荡阶段、低频调整止滑阶段。失稳应力降不等同于地震，地震是应力降过程中的高频振荡阶段。该模式不受加载方式影响，是粘滑失稳过程的固有特征。系统一旦进入预滑状态，就进入了不可逆的地震过程，会依次完成各个阶段进程。3）粘滑模型实验结果显示，一次应力降过程可能出现1～3次高频振荡，形成“单震”事件、“双震”事件和“三震”事件。双剪模型的失稳模式单一，只发生“单震”事件，斜面剪切模型和拐折模型则可能产生“双震”事件和“三震”事件，显示了多点震源错动现象。从速度频谱看，双剪模型含有三个主频峰值，斜面剪切模型具有双峰特征，拐折模型只有一个主频率峰；从子事件持续时间看，斜面剪切模型的高频振荡持续时间最短，双剪和拐折模型类似；从各子事件发生间隔看，斜面剪切模型几乎是连续发生，拐折模型则具有几十毫秒的“平静期”。根据声发射的定位，各子事件的震源出现于断层的不同位置。4）从几何结构上看，双剪模型、斜面剪切模型和拐折实验所模拟的断层为近似平直断层。从变形结构看，这些貌似简单的断层模型呈现出一种复杂而稳定的变形图像和非均匀的波动。在断层失稳之前，其预滑在空间上分布是不均匀的，会产生局部剪应变集中。在现场地应变观测中应注意这一基本现象。在总体应变能或应变强度基本不变的情况下，局部主应变轴的瞬时转动可以造成在断层带上剪应变与正应变的分配变化，破裂先从一个小区域开始，它使得临近区域的应力增加，导致破裂扩展，引发断层失稳。弹性回跳模型的力学机制可以用应变轴旋转过程来表达。5）使用双剪粘滑模型模拟自发地震和诱发地震的区域加载过程，利用应变观测系统多点连续观测发震断层附近的局部应变变化。在应力与应变空间上描述了地震过程的区域应力路径和局部应变路径。结果表明，远场动力学过程或者断层加载的方向与幅度难以从近断层带的应变观测中推测反演，近断层带的变形状态主要受构造部位控制，各点的应变路径明显不同。宏观加载应力路径与局部应变路径响应的转换阶段一致,存在一定映射关系，由多个局部应变获得的平均应变路径可能推测区域加载阶段，但加载主方向与近断层平均应变方向存在系统偏差。断层局部变形路径具有明确的物理意义，在γ-⊥空间的走向标明了断层所处的变形阶段。自发地震的应变路径可以划分为三个部分：应变积累阶段、剪应变的线性偏离阶段和失稳滑动阶段。诱发地震的应变路径包括四个阶段：正斜率的应变积累阶段、负斜率的稳态滑动阶段、亚稳态应变僵持阶段、扰动失稳滑动阶段。自发地震与诱发地震的应变路径差异较大，可以考虑从应变路径上判别断层稳定性与可能的地震类型。6）通过三种结构模型的粘滑地震实验模拟，利用高频速度连续观测系统获得了地震失稳过程的速度特征，讨论了最大位移量的选取方法，估算了实验室粘滑型地震的矩震级,探讨了粘滑类型、应力降大小与震级的关系。结果表明，实验室粘滑型地震的震级范围为-4.4～-3级，断层构造面的差异对各种粘滑模型的地震震级分布有明显影响。通过对比分析实验室地震、矿山微震、诱发微震及小震震群，确认在小尺度破裂滑动的范围内，应力降与地震震级没有明显相关性。决定地震震级的主要因素应当是震源尺度。更多还原

【Abstract】 It is widely accepted that an earthquake results from rock failure and unstablefrictional sliding on the fault plane. Thus fault propagation and instability is anfundamental issue in earthquake science. Observations to characteristics of physicalfield evolution during fault deformation and establishment of instability models are ofgreat significance for understanding earthquake mechanisms.The earthquake process can be simply divided into three stages: seismicgestation, earthquake occurrence, and post-seismic adjustment. Of them, there are fewrecords to the earthquake process, in particular little is known on the near-fieldmechanical process when earthquake happens in an instant. The laboratoryexperiments on the transient process of fault unstable sliding make it possible toconduct a detailed study on such transient process which would be helpful tointerpretations of field observations of earthquakes.In this work, three typical stick-slip models and fault propagation models havebeen selected as the objects of experimental research. A specially designed dynamicstrain observation system was employed to acquire data continuously with a samplingrate of3,400Hz. A velocity observation system with the sampling frequency of96KHz and an acoustic emission system with the sample rate of1MHz were used tomonitor the slip rate of stick-slip and failure signals. A series of simulationexperiments on granodiorite samples have been conducted at a biaxial horizontalloading apparatus with the electro-hydraulic servo control system. Firstly, thestructural parameters during the unstable slip failure were described, such as localstrain along the fault, acoustic emission signals, fault macroscopic slip rate. And theevolution features of these mechanical parameters during the transient process of faultunstable slip were analyzed. Secondly, how the fault rupture initiates, propagates andstops is studied, and the interaction and relationship between the parameters wasdiscussed, which provide insights into the physical nature of fault unstable slip on amicrocosmic scale and from the transient field. Next, the regional stress path and localstrain path during the whole earthquake process were described in the stress space andstrain space. The strain deformation stage on spontaneous earthquakes and inducedearthquakes was discussed, which can provide a new perspective for the in-situobservations and regional stress analysis. Finally, the calculation method of maximumdisplacement was discussed for estimating the moment magnitude of stick-slip events.The relationship of stick-slip models and stress drops and magnitude was alsoanalyzed in this thesis, which provides the basis for studying the relation betweennatural earthquakes and stick-slip friction.The research content and the main conclusions are summarized below.1) The experimental results on the three-dimensional fault propagation modelshow that the coalescence of the bridge area occurs at the last stage and is a rapidprocess. The crack initiation occurs at several points which combine each otherrandomly. There is a stable period present between the fully propagation and coalescence and sample failure, and the duration time is several tens milliseconds.2) The experiments on three typical stick-slip model show that the evolution ofthe fault stick-slip unstable process has a relatively stable feature characterized bythree typical phases (precursory slip, rapid slip incorporated with high-frequencystrain vibration and terminal adjustment), which is independent of dynamical loadingconditions and is the inherent feature of fault unstable slip. Main energy release takesplace at rapid slip with high-frequency oscillation. Stress drop is not the same as theearthquake, and the earthquake happens in the rapid slip with high-frequencyoscillation. When a stick-slip event occurs, the rock deformation enters the precursorslip phase, which means the fault enters the irreversible and unstable process, thestrain along the fault will go through the three phases in order.3) The experiment results of three typical stick-slip models indicate that theseveral high frequency fluctuation events are included in a stress drop process and canbe considered as a single stick-slip event, double stick-slip event and triple stick-slipevent. There are only single stick-slip events in the double shear model experiments.Double stick-slip events and triple stick-slip events are recorded in the both inclinedplane shear model and5°-bending model in which multi-point dislocation happens.FFT spectrum of slip rate shows that three dominant frequencies are present in doubleshear model, double dominant frequency is included in the inclined plane shear modeland5°bending model has only one dominant frequency. The durations of highfrequency oscillation are40ms in the inclined plane shear model,70ms in doubleshear model and5°bending model, respectively. The high-frequency oscillationduration of the inclined plane shear model is shorter than others. From the substick-slip event intervals, several sub events in inclined plane shear model happen oneby one, while there are tens of milliseconds "quiet period" between two sub stick-slipevents in the5°bending model. According to the acoustic emission positioning, thesources of the sub stick-slip events appear at different locations on the fault.4) The fault simulated by three stick-slip model experiments is nearly straightand flat in geometry. But there is a complicated and non-uniform fluctuation strainfield along the fault and the release of strain energy in the rock along the fault isuneven during the stick-slip process. The spatial distribution of precursor slip is notuniform before the fault instability with local shear strain concentration. When thestrain energy or strain intensity is largely unchanged, instantaneous rotation of localprincipal strain axes can cause the change of the distribution of the normal strain andshear strain along the fault, and then unstable slip starts from a small area first andmakes nearby area stress increase, which causes fault failure and unstable. Theprincipal strain axis rotation process can be used to express the mechanicalmechanism of the elastic rebound model.5) Double shear stick-slip model experiments were conducted to simulate theregional loading process of spontaneous earthquakes and induced earthquakes, and astrain observation system was employed to acquire data continuously to monitor thelocal strain changes near the fault under the loading process. Regional stress path andlocal strain path during the whole earthquake process was described in stress spaceand strain space. Research results show that the far-field dynamic process or the direction and amplitude of the regional loading process are difficult to speculatedirectly from the near fault strain observations. Deformation of the fault zone ismainly controlled by the tectonic position. Each location has different strain paths.Although the morphology of a local strain path is different greatly from themacroscopic stress path, there is a certain mapping relation on the correspondingtransformation stages between the stress path and strain path. The average strain pathcan be used to speculate the regional loading stage, but there is deviation between themain loading direction and average strain direction. The evolution trend of the localstrain path indicates the possible stage of the fault deformation. The strain path ofspontaneous earthquakes can be divided into three parts: strain accumulation, lineardeviation of shear strain and unstable slip. The strain path of induced earthquakesincludes four stages: strain accumulation with positive slope, steady state slip withnegative slope, metastable strain stalemate and unstable slip under a disturbance.Spontaneous earthquakes and induced earthquakes have their own inherent and steadypath modes, so that the fault stability and the possible earthquake types could bejudged according to the special strain path.6) Experiments on three typical stick-slip models were conducted for simulatingearthquake processes, and a specially designed velocity observation system wasemployed to acquire data continuously to monitor the slip rate of stick-slip. Thecalculation method of maximum displacement was discussed for estimating themoment magnitude of stick-slip events. The relationship of stick-slip models andstress drop and magnitude has been analyzed. The data show that the magnitude rangeof stick-slip under laboratory experiments is-4.4～-3. The magnitude of laboratoryearthquakes depends on the different fault structures and the macroscopic loadingprocess. By comparison of laboratory quakes, mine micro earthquakes, inducedearthquakes, and small quake swarms, this work determined that the stress drop is notsignificantly correlated with the value of the magnitude in a range of small-scalerupture. And the earthquake magnitude is primarily determined by the volume of theseismic source.更多还原