【作者】 蒋海昆；

【导师】 张流； 王錡； 许昭永；

【作者基本信息】 中国地震局地质研究所 ， 固体地球物理， 2000， 博士

【摘要】 为系统研究断层组合形式及温压环境对岩石变形过程中微破裂时空演化特征的影响,本文主要开展了以下几方面的工作:（1） 在前人研究成果的基础上给出声发射定位的慢度离差模型;（2） 重点进行了四个大尺度标本的双轴压缩破裂实验,分别预置拉张和挤压型雁列式断层、共线不连通断层和含圆柱型障碍体的平直断层,后两者在概念上可模拟与地震孕育有关的“强闭锁区”和断层面上的障碍体或坚固体。共进行了一万六千余次微破裂事件的时、空、强定位和一万余次微破裂事件的破裂机制反演,据此重点讨论了不同典型断层组合条件下微破裂的时空演化、破裂扩展和破裂机制问题;（3） 在不同的温压条件下开展了39个高温高压实验（固体围压介质）,重点研究不同温、压环境对AE时序特征的影响。三组实验的温、压条件分别是:室温（20℃）下围压从50MPa-850MPa改变、400MPa围压下温度从室温（20℃）-900℃改变、以及模拟地壳温压条件直至35km深度;（4） 基于实验结果及已有的研究成果,提出一个表征地震活动过程的简单物理概念模式,据此对部分地震学观测事实进行了初步的解释。 论文在以下几方面取得积极进展: 1、声发射定位方法的理论和数值试验研究 依据慢度离差模型的基本原理,给出一种仅依赖于初至到时的声发射定位及瞬时波速联合反演方法。通过数值试验对参数设置、探头数量及其与AE空间位置的关系、震相到时的噪声干扰等问题进行探讨,并对真实定位的误差分布给出统计上的圈定。数值试验结果表明:到时噪声小于测量时间单位的50%时,平均定位误差小于1.3,最大误差为5.0;97%以上的“AE”事件的定位误差小于物理不可分辨精度3mm。 该方法的最大特点在于可以不预先知道速度信息并且不需对速度结构进行均匀性假定,这对速度结构复杂并随时间（加载过程）变化的AE定位问题显得尤其重要。通过与标本破裂后表面宏观裂纹的对比分析,进一步证实了定位方法的正确性及较高的定位精度。 2、典型断层组合变形过程中声发射时空演化及破裂扩展 根据16通道AE波形记录结果,对一万六千余次AE事件进行了三维定位并确定其发生时间及能级大小。进而详细研究了四种典型断层组合变形过程中群体AE事件的时空动态演化过程,这不但丰富、深化了前人的认识,而且取得一些新的重要结论。与标本破裂后表面宏观裂纹相比较,完整地描述了变形过程中标本的动态破裂扩展图象。 2.1 从AE时序特征来看,破前随差应力的增加,不同断层组合标本AE活动均明显增强。雁列式断层和含障碍体的平直断层标本破裂前AE累积频次指数增长,而共线不连通断层标本则显示线性增长趋势。因而,AE累积频次指数或线性增长均是系统失稳前典型的微破裂前兆特征,其差异可能缘于构造组合形式的不同。相对于弹性变形阶段而言,雁列式断层标本破裂前的弱化阶段相对较长,而共线或平直断层标本则弹性变形时间较长,弱化阶段不甚明显。一般情况下峰值差应力对应最大的AER,而挤压型雁列式断层标本和含障碍体的平直断层标本在弱化阶段后期还显示破裂失稳前短时期内的相对平静现象。 2.2 从AE空间分布格局来看,预置构造对雁列式断层组合变形过程中AE空间分布格局具有较强的控制作用,破裂前局部化现象十分明显,前期的微破裂丛集图象指示并勾勒出后期宏观破裂的具体部位、扩展尺度及方向。雁列式断层组合AE主要分布于雁列区预置断层两个端点附近及其连线附近发生。结构不均匀性的影响随差应力增加逐渐显著,在变形的中后期较大事件通常集中于某一端点附近发生,拉张型雁列式断层雁列区最终破裂扩展与σ₁方向垂直,而挤压型雁列区最终破裂扩展方向则与σ₁方向平行。共线不连通断层破裂失稳前最更多还原

【Abstract】 1 Major Research ContentsThe aim of this work is to study the characteristics of acoustic emission (AE) of deforming rock sample with typical combined faults and samples at high pressure and temperature. It includes following aspects: (1) A slowness deviation model for AE location is given based on the results of predecessors. (2) The spatial-temporal evolution of AE, fracture extension and fracture mechanism during rock deformation is studied by experiment under biaxial compression. The samples of complex faults is composed compressional and extensional en-echelon, collinear fault with a non-connected area, and a regular fault with a columnar asperity. (3) With the solid confining pressure medium, thirty-nine HT-HP experiments have been performed under different pressure and temperature conditions. The focus is the influences of temperature and pressure condition on temporal features of AE sequences. The temperature and pressure conditions of three group tests are: (a) room temperature and confining pressure from 50MPa to 850MPa; (b) 400MPa confining pressure and temperature from room temperature (20°C) to 900°C; (c) simulating the temperature and pressure condition of crust up to about 35km depth. (4) Base on the experimental and theoretical results, a simple physical concept model for earthquake generation is proposed. And some observations are explained preliminarily with the model.2 Major Results and their implications(I) Based on the fundamental principle of the slowness deviation method, the specific method to determine the source location, occurring time and P-wave velocity of an AE by the Genetic Algorithm is offered. Combining with the actual experimental condition, some basic question such as AE location errors are studied by the numerical tests, and the statistic limitation for error distribution of AE location is determined. The results show that the locating errors of more than 97% of AE events are distributed within the range of 3mm, smaller than the diameter of a Tap sensor.The maximum feature of this AE locating method is that it does not need to know the velocity messages in advance and to assume the uniformity of the velocity field. This is very important for the AE location, which has a complex velocity structure and the velocity varys with the time (loading process). It is an important innovative progress for the AE location method , especially for AE location study on the structural rock samples and its results.(2) The spatial-temporal evolution of AE. fracture extension and fracture mechanism during rock deformation arc studied experimentally under biaxial compression. About more than 16000AE events are located The final images of AE spatial-temporal evolution corresponded to the macro-cracks of samples.2.1 The temporal characteristics of AE sequences show that the AE activity enhances with the increase of the differential stress before failure. Both exponential and linear increments of cumulated frequency of AE are typical precursory features before the system’s instability, the differences are mainly due to the different forms of combined faults. During the deformation of the en-echelon fault and the regular fault with a columnar asperity, the cumulated frequency increases exponentially before failure, but it increases linearly for the collinear fault with a non-connected area. Comparing with to the elastic stage, weaken stage is relatively long for the en-echelon faults. On contrary the elastic stage is very long for the collinear fault with a non-connected area and the regular fault with a columnar asperity, weaken stage is not obvious for these two types of fault combination. Generally, the peak differential stress corresponds to the maximum AER, and the relative quiet phenomena of short time appear in the anaphase of weaken stage before the instability for compressional en-echelon and regular fault with a columnar asperity.2.2 Viewing from the space distribution of AE events, the prefab faults have a big influence on the AE distribution pattern in the en-echelon area. Most of AE are distributed between the two points of prefab faults and big AE generally occurr around one point. The fracture localization is very clear. The evolation of AE space distribution corresponds to the extensional process and crack image of macrofractures. The extending direction of the fracture for extensional and comresssional en-echelon are perpendicular and parallel with the a,, respectively. For the collinear fault with a non-connected area, the most distinct feature is that with progress of deformation, a clear evolation process of gap revealed around the non-connected area, which develop in time and concentrates in space gradually. Due to complex action between the hanging and lower wall and asperity, the alternate AE activity in different areas is the most remarkable feature of the regular fault with a columnar asperity. Taking asperity as a center, compressional risen area and extensional area are formed on the cross regions around the asperity. The AE events occur alternately in the compressional or extensional region with the increment of difference stress. Among them, big AE mainly occurr in the extensional area and most of small AE events concentrate in the compressional area.2.3 The variations of the b values display a common feature. The b value reduced gradually in the prophase for a long time, and then rises back quickly before the instability. The b value dropping processes generally happen during the processes of difference stress increment. It indicates that the size of micro-fractures augment with the increment of difference stress. This dropping tendency of b values may continue to the weaken stage. And the quick comeback processes mainly happen in the anaphase of the weaken stage and at the moment before the fracture instability. The studies also show that after the failure, the dropping process of b value in en-echelon region correspond to the fracture extending process outside the en-echelon area (the hanging wall) in the time. This is a typical separation phenomenon of "source" and "precursory". Its basic essential is that the fractured region has a lower strength than other regions and has a more sensitive reaction for tinier variation of regional stress field.2.4 Considering the frequency and magnitude characteristics of AE sequences, the total AE number of the collinear fault with a non-connected area and the regular fault with a columnar asperity are larger than that of en-cchelon. On the other hand, the big AE events of en-echelon samples are slightly larger than that of collinear fault sample and regular fault sample The ratiobetween the number of MM>\.5 events and the total AE number for extensional and compressional en-echelon are 66% and 69% , respectively. But this ratio for collincar fault with non-connected area and regular fault with a columnar asperity are only 21% and 42% , respectively. The b values of G-R relationship of the collinear fault or regular fault are a little larger than that of en-echelon. This means that the small AE events occupies more larger ratio during the deformation period for the collinear fault with a non-connected area and the regular fault with a columnar asperity. Meanwhile, it indictes that the differences of b value due to the differences of structure are larger than the variation of b value caused by the increase of differential stress. That is, relative to the mechanical condition, the structure difference has more strong influence on the b values.3 <. Based on the AE locating results, the fracture mechanisms of more than 10000 events are inversed and statistical characteristics of AE fracture mechanism during the deformation process of typical combined faults.3.1 For micro-fracture types, the dip and oblique slip are dominant. Among them more than 54% micro-fractures are dip slip in the compressive en-echelon area, indicating the strong compressive action and a complex faulting type. The fracture type has a large difference between the compressive and extensional areas in the sample of the regular fault with a columnar asperity, the dip slip is dominant in the compressive areas, meanwhile the oblique and dip slip are similar in the extensional areas. The rate of extensional or compressive dip slip is some extent different during the different periods before or after failure for the en-echelon fault samples. For the compressive en-echelon sample, when it approaches to the failure, the remarkable dilatancy-diffusion weaken process leds to increase of the rate of extensional fracture obviously in the future macro-fracture area. This means that it is possible to estimate the occurrence time of big earthquakes to some extent by comparing the proportional variation of extensional or compressive component in the mechanism solutions. The study also shows that there is a visible increase of strike slip component of big events with M4￡>2.() when the failure approaches, even if the dip and oblique slip also were dominant. At the same time, the horizontal or nearly horizontal forces for these big events are dominant too. Its major reason is that the big events controlled by existing faults strongly, while small events produced mainly by the fracture of intact rock. There is no obvious relation between the small events and existing faults.Generally, the dominant direction of A-plane and B-plane of mechanism solutions are parallel or perpendicular to existing faults and the al is roughly their symmetry axis. But they looklike rather dispersed. Indicating the strong influence of existing faults and loading process on the AE mechanism solutions. Results also show that the mechanism solutions (specially for stress field direction) of AE which happened inside the intact rock and far from existing faults can reflect the realistic characteristics of regional stress field.3.2 Local mechanical circumstance from the AE mechanism solutions is very complex. For principal stress a,, oblique and nearly horizontal slips are dominant, and the vertical force is lack. The direction of principal extensional or compressive stress reflects the direction and acting mode of additional tectonic stress, the direction of P- or T-axis is close to that of additional loading force roughly but generally had a large deviation. Almost at the same point inside the intact rock, the mechanism solutions also vary with time. This means that under the intact medium condition, the stress level may be taken an important effect on the mechanism solutions of AE. This also reveals that in seismicity it is likely understandable for middle or small earthquakes, which occur in thesame small area and different stage, but their seismic mechanism solutions have statistical stability of short time. They are acted commonly by both the local medium asymmetry and regional tectonic stress, very different from large earthquakes which are controlled by existing faults. Comparing with the experiment results, the remarkable variation of mechanism solutions of these middle or small earthquakes probably reflect changes of regional stress field.3.3 The results of partial experiments show that the tectonic deformation process has big influence on the direction of local principal stress. The dominant direction of P-axis has a believable change before and after the failure. The results also show that the influence of additional force would increase gradually if the loading strength is big enough. Maybe this is the major reason that seismic activity displays similar images in the different tectonic structure areas. For the sample of the regular fault with a columnar asperity, the mechanism solutions are different in the initiative and passivity wall during the process of fault relative movement. The dominant direction of P- or T-axis is coherent in the lower wall (passivity wall), but it is dispersive in the hanging wall (initiative). Meanwhile, the force type is relatively coherent in compressive region and dominated horizontal or nearly horizontal force.4-. The Fourier spectrum of the AE wave in a infinitesimal area (approximately in the same point) shows that during the elastic deformation stage, the high frequency component reveals an increasing tendency with the development of deformation (time). And during weaken stage after the peak strength, besides the wide frequency belt of energy distribution, a lower frequency peak also appears for some AE events. The wavelet analysis on the AE wave indicates that the dynamic fracture processes are not similar for any two AE events. The calculations of seismic parameters on four big AE events (M^^.O) show that the fracture size of AE is mainly between 1.24mm to 1.66mm.5, At different temperatures and different confining pressures as well as temperature and pressure conditions corresponding to in different crust depths, the temporal characteristics of AE sequences have been stress studied. The results show that the temperature is the major influence factor for granite brittle-ductile transition. And the failure forms of granite mainly depend on the confining process. For rock strength, the influence of confining pressure is stronger than that of temperature in the brittle and semi-brittle field, but opposite in the semi-brittle and ductile field.5. / At the room temperature, the rock strength becomes large with the increasing confining pressure, and the failure types also become from incremental failure into abrupt failure. Meanwhile, the failure types of medium become from tensile-shear fracture to shear fracture and further to the shear fracture and stick-slip. At the low confining pressure, the system keeps its stability state when the sample fractures and there are a few AE events in this process. And the AE temporal distribution is stochastic before and after failure. This means that it can not accumulate huge energy to produce a big fracture or instability under the pressure condition of earth surface (about zero depth). Under the major confining pressure, the system is instability, showing stick-slip features and obvious stress drop. With increment of confining pressure, the number of AE before and after failure increases and the time when AE are detected is eralier. Showing an obvious cluster feature in time means that a tiny stress disturbance (increment of difference stress) can trigger the microfracture or urge it to extend continuously at high confining pressure. Before the failure, the AE cumulating frequency increases accord with exponent model and the exponential increasing rate increase with the increasing of confining pressure In the quasi-periodic process of stress drop, amplitude, releasing rate and time interval of large stress dropincrease with the increase of confining pressure. These can be analogy to such an observed fact qualitatively: that in quasi-periodic seismic activity, the high environment stress can generate a strong earthquake and the active period of strong earthquake will be longer. Dividing stages by the occurrence time of big stress drops, a remarkable distinguish is, along with the drown on of the huge instability, the cumulating frequency of AE increases with the exponential model or linear model before or after failure, respectively.5.2 Under the condition of 400MPa confining pressure, the rock strength decreases gradually with the rising of temperature. The sample failure types become from abrupt instability at low temperature into the quasi- abrupt instability at middle temperature and incremental failure at high temperature. The transition temperature is about 150°C and 550°C. The medium failure types become from brittle tensile-shear fracture and stick-slip (about 20~250°C) , through the semi-brittle fracture (350°C) to the semi-brittle flow (650°C) . At 850°C, plastic flow and partial smelting appear. With the rising of temperature, the number of AE decreases quickly, and the level of difference stress also has a little increase. Below 250°C, the cumulating number of AE before the failure correspond to the exponential increasing model with the failure draw on, but the rate of exponential increasing minishes with the rising of temperature. There is no AE event been recorded at the high temperature (650°C and 850°C).5.3 Under the experiment conditions that model changes with crust depths, the granite rock strength increases with the depth until 30km. In shallow crust (about 3km), abrupt rock failure or quasi-abrupt instability happen at lower pressure and result in tensile-shear fractures. AE are distributed almost uniformly before or after failure. Downward to the depth range (about 6-1 Okm) with progressive failure as the main feature, there are no or only a few number of AE before or after failure. In deeper range (about 14-22km), rock failure shows some features of quasi-abrupt instability at high pressure. There are still few AE before failure, but with the stick-slip, much more AE events are detected post-failure. At the temperature and pressure condition of more deeper crust (about 26km), rock failure takes abrupt instability with at pressure, there are dense AE activity before failure and cumulating frequency of AE increases exponentially before the failure. At about 35km depth range, rock strength decreases quickly with the depth and the sample semi-ductile or ductile deforms and progressive fails, there are no AE detected before and after failure.5.4 At different temperatures and confining pressures, the frequency-magnitude relationships of AE accord with the famous G-R relation. At the room temperature, the b value in entire deformation process decreases with the increasing of confining pressure. This implies that tendentious fall of b value may predict an enhancement of environment stress. At the 400MPa confining pressure and different temperatures, the b value before failure is a little larger than that of after failure. In the case of before or after failure, the b-value is high at the low temperatures and is low at the high temperatures. The lowest b value appears at 350°C, which is the temperature condition when failure types transform from abrupt instability to incremental failure under the 400MPa confining pressure. Under the condition changing with the crustal depth, the b value of AE sequence before failure seems a little smaller than that in post-failure stage, similar to natural earthquake sequences. An important fact is that the b value tends to decrease with depth. The results of after failure, show that the lowest b value appears at the temperatures and pressures equivalent to 14~18km deep crust.5.5 The study results demonstrate that the AE temporal sequences have an obviousmultifractal characteristic which is mainly controlled by the temporal cluster feature of AE sequences. The numerical ranges of index a decrease with the increase of the confining pressure, and becomes wide with the rise of temperature. The former means that the scaling types decrease with the increase of confining pressure, and predicts that the temporal structure of AE sequence tends to be simple and in order. The later means that the scaling types increase with the elevation of temperature and implies that the temporal structure of AE sequence tends to be complex and chaos with the rise of temperature. The experiment results under the conditions that change with the crustal depth show that the numerical range of index a is the widest at about 18km depth and becomes narrow at shallower or deeper crust. So, when the temperature and pressure condition simulating the real environment of focal depth changes from shallow to deep in the earth crust, the range of a of microfracture sequence is likely to go tlvrough such an evolvation process that a changes from narrow to wide and then to narrow again.6, Based on the properties and mechanical behaviors of rock deformation and failure, as well as the rock strength features with the crust depth, a simple physical model for earthquake generation is proposed. In the epicenter area, it is assumed the medium component is the same and almost uniformity from shallow to deep, and assuming the tectonic stress loading at an fixed strain rate in horizontal direction and there is no relation to depth. Under this medium and mechanical condition, in the upper crust brittle and semi-brittle field, the rock strength increases with the depth. The differential stress increases gradually with time. For somewhere in the crust, when the differential stress is larger than the rock strength, the rock will fail. But the rock failure does not always lead to abrupt instability (earthquake). It also relates to the mechanical behavior of rock failure at different temperatures and pressures. According to the model, during a loading circle, tectonic stress increases gradually with time and the foci move to the deeper crust gradually. The relocation results of some earthquake sequences support this concept model.In brittle field, when temperature and pressure condition simulating the real environment of crust changes from shallow to deep, the mechanical behavior of granite failure will go through the section of abrupt instability7 at low pressures, the section of incremental failure, the section of quasi-abrupt instability and the section of abrupt instability at high pressures. The medium failure and correlative microfractures temporal distribution before or after failure are very different when the mainfracture occurs in different sections. The range of a of microfracture sequence is likely to go through such an evolvation process that a changes from narrow to wide and then to narrow again. With the development of earthquake generation, the fracture position becomes deep gradually. Due to above intrinsic attributes, the phenomena of "enhancement-quiet-active again" and other seismic images will emerge in the epicenter area before the mainshock. A long period dropping of b values before the mainshock may be the result of small earthquakes moving to the deep in the epicenter area, and sudden drop of b values, close to the mainshock may be the reflection of rock dilatancy weaken or fault displacement weaken. The range of a, multifractal index spetrum, of microshock sequence is going through to wide or to narrow before the strong earthquake may depends on the focal depth distribution of the mainshock.更多还原