Laboratory Acoustic Emission Research Articles

From Labquakes to Megathrusts: Scaling Deep Learning Based Pickers Over 15 Orders of Magnitude

AbstractThe application of machine learning techniques in seismology has greatly advanced seismological analysis, especially for earthquake detection and seismic phase picking. However, machine learning approaches still face challenges in generalizing to data sets that differ from their original training setting. Previous studies focused on retraining or transfer‐learning models for these scenarios, but require high‐quality labeled data sets. This paper demonstrates a new approach for augmenting already trained models without the need for additional training data. We propose four strategies—rescaling, model aggregation, shifting, and filtering—to enhance the performance of pre‐trained models on out‐of‐distribution data sets. We further devise various methodologies to ensemble the individual predictions from these strategies to obtain a final unified prediction result featuring prediction robustness and detection sensitivity. We develop an open‐source Python module quakephase that implements these methods and can flexibly process input continuous seismic data of any sampling rate. With quakephase and pre‐trained ML models from SeisBench, we perform systematic benchmark tests on data recorded by different types of instruments, ranging from acoustic emission sensors to distributed acoustic sensing, and collected at different scales, spanning from laboratory acoustic emission events to major tectonic earthquakes. Our tests highlight that rescaling is essential for dealing with small‐magnitude seismic events recorded at high sampling rates as well as larger magnitude events having long coda and remote events with long wave trains. Our results demonstrate that the proposed methods are effective in augmenting pre‐trained models for out‐of‐distribution data sets, especially in scenarios with limited labeled data for transfer learning.

Estimating Lab-Quake Source Parameters: Spectral Inversion from a Calibrated Acoustic System.

Laboratory acoustic emissions (AEs) serve as small-scale analogues to earthquakes, offering fundamental insights into seismic processes. To ensure accurate physical interpretations of AEs, rigorous calibration of the acoustic system is essential. In this paper, we present an empirical calibration technique that quantifies sensor response, instrumentation effects, and path characteristics into a single entity termed instrument apparatus response. Using a controlled seismic source with different steel balls, we retrieve the instrument apparatus response in the frequency domain under typical experimental conditions for various piezoelectric sensors (PZTs) arranged to simulate a three-component seismic station. Removing these responses from the raw AE spectra allows us to obtain calibrated AE source spectra, which are then effectively used to constrain the seismic AE source parameters. We apply this calibration method to acoustic emissions (AEs) generated during unstable stick-slip behavior of a quartz gouge in double direct shear experiments. The calibrated AEs range in magnitude from -7.1 to -6.4 and exhibit stress drops between 0.075 MPa and 4.29 MPa, consistent with earthquake scaling relation. This result highlights the strong similarities between AEs generated from frictional gouge experiments and natural earthquakes. Through this acoustic emission calibration, we gain physical insights into the seismic sources of laboratory AEs, enhancing our understanding of seismic rupture processes in fault gouge experiments.

Spatial continuity and stress transfer: Primary and complementary factors shaping pure-swarm laboratory catalog into mixed burst-like and swarm-like

Natural earthquakes and micro-seismicity resulting from hydraulic fracturing or other engineering practices display distinctively different spatial-temporal features, like mixed burst- and swarm-like features or predominantly swarm-like features. The mechanism(s) contributing to such observations can be diverse. We present the inspections on the dynamic formation process of the single swarm-like tree in laboratory acoustic emission (AE) catalogs. Such largest swarm-like trees can contain >97% AE events from the entire catalog within a test; and all catalogs under investigation display scale-invariance features. The formation of the largest swarm-like tree correlates with the rock fracture process analogue of the source pervasive process, where its AE releases exhibit significant spatial well-organization. Comparison to other laboratory catalogs under different laboratory settings helps us identify the spatial continuity of the rock fracture process as the primary factor in forming the largest swarm-like trees at laboratory scale. The stress transfer process is involved in the rock fracture process for the tests having pre-existing spatial discontinuity. Artificial perturbations on the spatial information induced by the stress transfer process further confirm that stress transfer also serves to shift the pure swarm-like catalog into a mixed burst- and swarm-like catalog. These laboratory observations may provide inspirational insights for understanding the field-scale mechanism(s) shaping the spatial-temporal energy release features.

Source Parameters of Laboratory Acoustic Emission Events Estimated From the Coda of Waveforms

AbstractWe develop a method to estimate relative seismic moments M0 and corner frequencies fc of acoustic emission events recorded in laboratory experiments from amplitude spectra of signal's coda composed of reverberated and scattered waves. This approach has several advantages with respect to estimations from direct waves that are often clipped and also are difficult to separate in experiments performed on small samples. Also, inversion of the coda spectra does not require information about the source locations and mechanisms. We use the developed method to analyze the data of two experiments: (a) on granite from the Voronezh crystal massif and (b) on Berea sandstone. The range of absolute corner frequencies estimated in both experiments is around 70 − −700 kHz. The range of relative seismic moments covers 103.5. The relation between fc and M0 observed on the first stages of both experiments, consisted of increasing isotropic confining pressure, approximately follow scaling and the b‐value of the Gutenberg‐Richter distribution was found close to 1. This can be interpreted as rupturing of preexisting material defects with a nearly constant stress‐drop and has a similarity with observations of “natural” earthquakes. Deviations from this “earthquake‐like” behavior observed after applying axial loading and initiation of sample damaging can be interpreted as changes in stress‐drop. Lower stress‐drops prevail for sandstone and higher for granite sample respectively that can be related to the strength of corresponding material.

Study on precursor information and disaster mechanism of sudden change of seepage in mining rock mass

Abstract Changes in the stress field and seepage field of mining unloading are one of the important causes of deformation and destabilization of the rock body of the bottom slab. With the increase of mining intensity and depth, the disaster of water influx on the bottom plate under the dual action of mining unloading and pressurized water has become one of the main problems restricting the efficient and safe mining of coal. Based on the research background of confined water inrush from Ordovician limestone floor in North China Coalfield, the numerical analysis revealed that mining unloading concentrates stresses, increases deformation, and increases the plastic zone and permeability range. Based on laboratory acoustic emission and mechanical tests, it is found that the peak of acoustic emission ringing number can be one of the precursor information of limestone seepage mutation. This study reveals the evolution law of rock body deformation and permeability under the unloading path, and the research obtains the unloading seepage and water gushing disaster mechanism of the subgrade rock body.

Systematic Uncertainty Quantification of First-Polarity-Based Moment Tensor Inversion Due to Sparse Coverage of Sensor Arrays in Laboratory Acoustic Emission Monitoring

Systematic Uncertainty Quantification of First-Polarity-Based Moment Tensor Inversion Due to Sparse Coverage of Sensor Arrays in Laboratory Acoustic Emission Monitoring

An effective method for laboratory acoustic emission detection and location using template matching

In this paper, a template matching and location method, which has been rapidly adopted in microseismic research in recent years, is applied to laboratory acoustic emission (AE) monitoring. First, we used traditional methods to detect P-wave first motions and locate AE hypocenters in three dimensions. In addition, we selected events located with sufficient accuracy (normally corresponding AE events of relatively larger energy, showing clear P-wave first motion and a higher signal-to-noise ratio in most channels) as template events. Then, the template events were used to scan and match other poorly located events in triggered event records or weak events in continuous records. Through cross-correlation of the multi-channel waveforms between the template and the event to be detected, the weak signal was detected and located using a grid-searching algorithm (with the grid centered at the template hypocenter). In order to examine the performance of the approach, we calibrated the proposed method using experimental data of different rocks and different types of experiments. The results show that the proposed method can significantly improve the detection capability and location accuracy, and can be applied to various laboratory and in situ experiments, which use multi-channel AE monitoring with waveforms recorded in either triggering or continuous mode.

Semi‐Empirical Error Ellipsoid Clustering for Identifying the Second‐Order Structural Features From a Laboratory AE Source Location Cloud: Method, Validation, and Application to a Hydraulic Fracturing Test

AbstractPoint cloud of sources in the form of acoustic emission (AE) or seismicity at both field‐ and laboratory‐scales are indicative of causative structures. The error ellipsoid clustering (EC) method, whose underlying premise is reducing the amount of discarded data, is helpful for seeking the causative structure(s) from field‐scale seismic sources. However, a proper EC method does not exist for laboratory sources. Interpretation of laboratory AE sources usually involves systematic discarding of the assumed “low‐quality” data. We seek the successful transfer of the field‐scale EC method into the investigation of laboratory‐scale sources. For that, a new semi‐empirical EC method (serendipity method) is devised, and we have systematically modified the previous EC method (standard method) in (1) error ellipsoid formation, (2) iterative clustering algorithm, and (3) clustering stopping criterion. In the literature, the standard method and its limitations were validated using synthetic data. We validate the modifications by comparing its post‐clustering results with the original source locations and with the in‐situ and post‐mortem inspections of the laboratory rock fractures; and similarly for the serendipity method. We apply the serendipity method to a laboratory hydraulic fracturing test. The serendipity method implies the existence of a 3D damage zone as the second‐order feature caused by the hydraulic fracturing in addition to the primary hydraulic fracture. This complex damage zone provides a logical explanation to the previously observed reverse permeability‐distance relationship for hydraulic fracture.

Research on Maximum Likelihood b Value and Confidence Limits Estimation in Doubly Truncated Apparent Frequency–Amplitude Distribution in Rock Acoustic Emission Tests

The b value deduced from the Gutenberg–Richter law is an important parameter for sequence and precursory analyses, both in laboratory acoustic emission tests and seismology. As the b value is a statistical value, the maximum likelihood estimation is mostly used to estimate the b value. However, traditional singly truncated maximum likelihood estimation in seismology only considers the minimum magnitude, while the acquisition device in rock acoustic emission tests will set the threshold value and maximum value of the amplitude; therefore, maximum likelihood estimation will estimate the b value in a doubly truncated size distribution, and its confidence limits need to be discussed. Here, in this study, we derive the calculation equations of the b value and the corresponding confidence limits for the maximum likelihood estimation with a narrow amplitude span in a doubly truncated frequency–amplitude distribution. The maximum likelihood b values estimated by the scheme of a singly and doubly truncated frequency–amplitude distribution are compared through acoustic emission data with the known underlying distribution. The results show that the maximum likelihood b value and confidence limits estimation scheme derived for rock acoustic emission tests with a narrow amplitude span is more reasonable. Then, the derived estimation scheme is applied to the rock dilation rupturing test; the results confirm its applicability.

Source parameter estimation of acoustic emissions induced by hydraulic fracturing in the laboratory

SUMMARYThe hydraulic fracturing technique is used for resource production, such as in shale gas/oil extraction and enhanced geothermal systems. The effects of fracturing are often monitored via induced earthquakes, and obtaining as much information as possible from those earthquakes is desirable. The stress drop—calculated from the seismic moment Mo and corner frequency fc—is an earthquake-related parameter that can help identify additional characteristics of the seismicity. To investigate the relationship between stress drops and hydraulically induced seismic events, we estimated the Mo and fc of acoustic emission (AE) events during hydraulic fracturing experiments performed in the laboratory in previous studies using 2 Eagle Ford shale and 10 Kurokami-jima granite samples. We estimated Mo by fitting the theoretical spectra to the observed spectra after correcting for the following effects: (1) frequency response of AE transducers under the installation method used in the fracturing experiment, including differences in sensitivity across every transducer used in each experiment; and (2) the difference in radiation pattern coefficients, which depends on the focal mechanisms of each AE event. This analysis used 46 857 focal mechanisms obtained from moment tensor solutions estimated using a deep learning technique. The range of the resultant Mo was found to be 2.8 × 10–5 ≤ Mo ≤ 4.5 × 10–1 [N·m], corresponding to −9.1 ≤ Mw ≤ −6.3, where Mw is the moment magnitude. We also estimated fc using the multiple-empirical Green's function method, reducing the influence of modelling errors in the AE sensor response and transfer function of the medium. Out of the 1053 events whose Mo and fc were estimated, 465 events (44.2 per cent)—regardless of their focal mechanisms—were found to have Mo and fc values consistent with the constant stress drop scaling of shear failure (i.e. shear failures have 0.1–100 MPa stress drops independent of their magnitude) that has been repeatedly confirmed in many previous studies. The remaining events showed lower fc values than those expected from the scaling law. This indicates that high pore pressure in a source region induced by fluid stimulation contributes to the occurrence of low-frequency earthquakes. Overall, we demonstrated that source parameter estimation was possible for laboratory AEs induced by hydraulic fracturing, which can improve our understanding of the characteristics of fluid-induced earthquakes.

Microseismic Temporal-Spatial Precursory Characteristics and Early Warning Method of Rockburst in Steeply Inclined and Extremely Thick Coal Seam

Early warning of a potential rockburst risk and its area of occurrence helps to take effective and targeted measures to mitigate rockburst hazards. This study investigates the microseismic (MS) spatial-temporal precursory characteristic parameters in a typical steeply inclined and extremely thick coal seam (SIETCS) with high rockburst risk and proposes three spatial/temporal quantification parameters and a spatial-temporal early warning method. Analysis results of temporal parameters show that the sharp-rise-sharp-drop variation in total daily energy and event count can be regarded as a precursor for high energy tremor. The appearance of peak values of both energy deviation (≥20) and event count deviation (≥1) can be regarded as precursors that indicate imminent rockburst danger. A laboratory acoustic emission (AE) experiment reveals that precursor characteristics obtained from the study can be feasibly used to warn the rockburst risk. The spatial evolution laws of spatial parameters show that the high energy density index of MS (EDIM), velocity, velocity anomaly regions correlate well with stress concentration and rockburst risk areas. The field application verifies that the temporal-spatial early warning method can identify the potential rockburst risk in a temporal sequence and rockburst risk areas during the temporal early warning period.

AEnet: Automatic Picking of P-Wave First Arrivals Using Deep Learning

First arrival time picking is one of the critical processing steps of acoustic emission (AE)/microseismic (MS) monitoring for studying rock fracture processes. Because of massive monitoring data, the automatic arrival time picking technique is particularly desired. Inspired by recent successful applications of machine learning (ML) in earthquake phase identification, we propose a deep learning (DL)-based P-wave first arrival time picking method named AE Network (AEnet) for laboratory AE monitoring data. Our approach consists of two steps: classification and picking. The convolutional neural network (CNN) is used to classify each sample point of acoustic waveforms into either noise or signal. Different from prior DL-based phase picking studies using raw waveforms, we combine the waveform and high-order statistics as the input to enrich the input data features and accelerate the CNN model learning process. Our approach is examined using the laboratory AE monitoring data and the performance of each component of AEnet is also analyzed. The results show that the CNN model can classify the sample points accurately for the picking procedure. With this classification result, we pick the first arrival time of each trace using the curve fitting method and an unsupervised clustering algorithm. To evaluate the performance of AEnet, we apply Akaike Information Criterion-Short Term Averaging/Long Term Averaging Method (AIC-STA/LTA), one of the most popular and traditional picking methods, on the same waveforms and use the manual picks as the reference. Error analysis results show that AEnet outperforms AIC-STA/LTA.

Acoustic Emission Events Interpreted in Terms of Source Directivity

During inversion for the source mechanisms of laboratory acoustic emission events, relatively high misfit values (expressed as the RMS of the inverted equations) were observed. Our experiment was performed on Westerly Granite. A processed set of data consisting of 2405 acoustic emission events was used and a semi-homogeneous velocity model was considered. A correction for sensor radiation patterns and individual sensor constants was taken into account, and an acausal attenuation model was assumed. Source mechanisms were inverted for the moment tensor. The application of a more sophisticated medium model improved inversion quality only for some events. Introducing the source directivity, a standard approach for earthquakes with magnitudes larger than approximately 4, increased the number of successfully inverted events. Directivity was introduced using a Haskell source model; optionally unilateral and bilateral versions of the source were considered. Lower values of RMS for the Haskell source model were considered to justify the directivity approach. This formalism enables us to select the preferable fault from the two nodal planes within the fault plane solution. The rupture directions were observed to tend to the dip direction of the preferred fault. They were found to be preferably subparallel to slip directions for acoustic emissions with a significant DC component. The source time functions retrieved from the seismograms are in agreement with finite source theory.

Recent Advances and Challenges of Waveform‐Based Seismic Location Methods at Multiple Scales

AbstractSource locations provide fundamental information on earthquakes and lay the foundation for seismic monitoring at all scales. Seismic source location as a classical inverse problem has experienced significant methodological progress during the past century. Unlike the conventional traveltime‐based location methods that mainly utilize kinematic information, a new category of waveform‐based methods, including partial waveform stacking, time reverse imaging, wavefront tomography, and full waveform inversion, adapted from migration or stacking techniques in exploration seismology has emerged. Waveform‐based methods have shown promising results in characterizing weak seismic events at multiple scales, especially for abundant microearthquakes induced by hydraulic fracturing in unconventional and geothermal reservoirs or foreshock and aftershock activity potentially preceding tectonic earthquakes. This review presents a comprehensive summary of the current status of waveform‐based location methods, through elaboration of the methodological principles, categorization, and connections, as well as illustration of the applications to natural and induced/triggered seismicity, ranging from laboratory acoustic emission to field hydraulic fracturing‐induced seismicity, regional tectonic, and volcanic earthquakes. Taking into account recent developments in instrumentation and the increasing availability of more powerful computational resources, we highlight recent accomplishments and prevailing challenges of different waveform‐based location methods and what they promise to offer in the near future.

A fuzzy comprehensive evaluation methodology for rock burst forecasting using microseismic monitoring

Rock bursts have become one of the most severe risks in underground coal mining and its forecasting is an important component in the safety management. Subsurface microseismic (MS) monitoring is considered potentially as a powerful tool for rock burst forecasting. In this study, a methodology for rock burst forecasting involving the use of a fuzzy comprehensive evaluation model was developed, which allows for a more quantitative evaluation of the likelihood for the occurrence of a rock burst incident. In the fuzzy model, the membership function was built using Gaussian shape combined with the exponential distribution function from the reliability theory. The weight of each index was determined utilising the performance metric F score from the confusion matrix. The comprehensive forecasting result was obtained by integrating the maximum membership degree principle (MMDP) and the variable fuzzy pattern recognition (VFPR). This methodology has been applied to a coal mine in China to forecast rock bursts. To select MS indices for rock burst forecasting using the fuzzy evaluation model, laboratory acoustic emission (AE) measurements of coal samples collected from the mine were performed. The model parameters were first calibrated using historical MS data over a period of four months, during which six rock burst incidents were observed. This calibrated model was able to forecast the occurrence of a subsequent rock burst incident in the mine.

Microseismicity of an Unstable Rock Mass: From Field Monitoring to Laboratory Testing

AbstractThe field‐scale microseismic (MS) activity of an unstable rock mass is known to be an important tool to assess damage and cracking processes eventually leading to macroscopic failures. However, MS‐event rates alone may not be enough for a complete understanding of the trigger mechanisms of mechanical instabilities. Acoustic Emission (AE) techniques at the laboratory scale can be used to provide complementary information. In this study, we report a MS/AE comparison to assess the stability of a granitic rock mass in the northwestern Italian Alps (Madonna del Sasso). An attempt to bridge the gap between the two different scales of observation, and the different site and laboratory conditions, is undertaken to gain insights on the rock mass behavior as a function of external governing factors. Time‐ and frequency‐domain parameters of the MS/AE waveforms are compared and discussed with this aim. At the field scale, special attention is devoted to the correlation of the MS‐event rate with meteorological parameters (air temperature and rainfalls). At the laboratory scale, AE rates, waveforms, and spectral content, recorded under controlled temperature and fluid conditions, are analyzed in order to better constrain the physical mechanisms responsible for the observed field patterns. The factors potentially governing the mechanical instability at the site were retrieved from the integration of the results. Abrupt thermal variations were identified as the main cause of the site microsesimicity, without highlighting irreversible acceleration in the MS‐event rate potentially anticipating the rock mass collapse.

ISRM Suggested Method for Laboratory Acoustic Emission Monitoring

Acoustic emission (AE) is defined as high-frequency elastic waves emitted from defects such as small cracks (microcracks) within a material when stressed, typically in the laboratory. AE is a similar phenomenon to microseismicity (MS), as MS is induced by fracture of rock at an engineering scale (e.g., rockbursts in mines), that is, in the field. Thus, seismic monitoring can be applied to a wide variety of rock engineering problems, and AE is a powerful method to investigate processes of rock fracture by detecting microcracks prior to macroscopic failure and by tracking crack propagation. A basic approach involves using a single channel of data acquisition, such as with a digital oscilloscope, and analyzing the number and rate of AE events. Perhaps the most valuable information from AE is the source location, which requires recording the waveform at several sensors and determining arrival times at each. Thus, investing in a multichannel data acquisition system provides the means to monitor dynamics of the fracturing process. The purpose of this suggested method is to describe the experimental setup and devices used to monitor AE in laboratory testing of rock. The instrumentation includes the AE sensor, preamplifier, frequency (noise) filter, main amplifier, AE rate counter, and A/D (analog-to-digital) recorder, to provide fundamental knowledge on material and specimen behavior in laboratory experiments. When considering in situ seismic monitoring, the reader is referred to the relevant ISRM suggested method specifically addressing that topic (Xiao et al. 2016).

Effects of Anisotropy for P-Wave Velocity on Locating Accuracy of Acoustic Emission Sources in Sandstone

The accurate measurement of wave velocity has a significant effect on the locating accuracy of acoustic emission sources, especially in health assessment and determination for rock structures. There commonly exists some deviation between the measured velocity value and the actual velocity value due to the anisotropy of rock velocity, which causes the large locating error. In this paper, the locating error caused by the anisotropy for P-wave velocity is discussed quantitatively. The locating experiments were carried out on the sandstone sample, whose length, width, and height are 100, 100, and 200 cm respectively. The mineral composition and micro structures for the sample were observed. AE sources were triggered by lead breaks and the arrival times were detected through eight sensors. Then, the sources can be located based on time difference locating method. Results show that the difference of wave velocity is large in different directions, which reaches 2400 m/s. The errors of wave velocity affect the locating accuracy seriously, which is up to 20.5%. A total of 135 groups of velocity values in different directions were measured. It is found that the locating accuracy of repeatedly measuring velocity through diagonal line is higher than accuracy of traditional used axial velocity. In addition, the source locating accuracy can be improved greatly by measuring and solving the average wave velocity in multiple directions. However, the average velocity value corresponds to the locating result is not the best. Tests clarify that the optimal velocity is higher than the average velocity. It is suggested that the wave velocity should be measured in multiple directions for several times when performing the laboratory acoustic emission (AE) sources locating experiments. The conclusions provide important theoretical guidance for the current traditional AE sources locating experiments with only average value of axial velocities.

A laboratory acoustic emission experiment and numerical simulation of rock fracture driven by a high-pressure fluid source

In order to improve our understanding of rock fracture and fault instability driven by high-pressure fluid sources, the authors carried out rock fracture tests using granite under a confining pressure of 80 MPa with fluid injection in the laboratory. Furthermore, we tested a number of numerical models using the FLAC3D modeling software to find the best model to represent the experimental results. The high-speed multichannel acoustic emission (AE) waveform recording system used in this study made it possible to examine the total fracture process through detailed monitoring of AE hypocenters and seismic velocity. The experimental results show that injecting high-pressure oil into the rock sample can induce AE activity at very low stress levels and can dramatically reduce the strength of the rock. The results of the numerical simulations show that major experimental results, including the strength, the temporal and spatial patterns of the AE events, and the role of the fluid can be represented fairly well by a model involving (1) randomly distributed defect elements to model pre-existing cracks, (2) random modification of rock properties to represent inhomogeneity introduced by different mineral grains, and (3) macroscopic inhomogeneity. Our study, which incorporates laboratory experiments and numerical simulations, indicates that such an approach is helpful in finding a better model not only for simulating experimental results but also for upscaling purposes.

Laboratory acoustic emission study for earthquake generation process

Since the similarity in size distribution of earthquakes and acoustic emissions (AE) was found in the 1960s, many laboratory studies have been motivated by the need to provide tools for the prediction of mining failures and natural earthquakes. This paper aims, on the one hand, to draw an outline of laboratory AE studies in the last 50 years, which have addressed seismological problems. Topics include the power laws in which the similarity between AEs and earthquakes is involved and progress that has been made in AE technology and laboratory AE study. On the other hand, this study will highlight some key issues intensively discussed, especially in the last three decades, such as aspects related to the pre-failure damage evolution, fault nucleation and growth in brittle rocks and discuss factors governing these processes.