Gregory C. McLaskey

Fault healing promotes high-frequency earthquakes in laboratory experiments and on natural faults

Faults strengthen or heal with time in stationary contact, and this healing may be an essential ingredient for the generation of earthquakes. In the laboratory, healing is thought to be the result of thermally activated mechanisms that weld together micrometre-sized asperity contacts on the fault surface, but the relationship between laboratory measures of fault healing and the seismically observable properties of earthquakes is at present not well defined. Here we report on laboratory experiments and seismological observations that show how the spectral properties of earthquakes vary as a function of fault healing time. In the laboratory, we find that increased healing causes a disproportionately large amount of high-frequency seismic radiation to be produced during fault rupture. We observe a similar connection between earthquake spectra and recurrence time for repeating earthquake sequences on natural faults. Healing rates depend on pressure, temperature and mineralogy, so the connection between seismicity and healing may help to explain recent observations of large megathrust earthquakes which indicate that energetic, high-frequency seismic radiation originates from locations that are distinct from the geodetically inferred locations of large-amplitude fault slip.

Hertzian impact: experimental study of the force pulse and resulting stress waves.

Ball impact has long been used as a repeatable source of stress waves in solids. The amplitude and frequency content of the waves are a function of the force-time history, or force pulse, that the ball imposes on the massive body. In this study, Glaser-type conical piezoelectric sensors are used to measure vibrations induced by a ball colliding with a massive plate. These measurements are compared with theoretical estimates derived from a marriage of Hertz theory and elastic wave propagation. The match between experiment and theory is so close that it not only facilitates the absolute calibration the sensors but it also allows the limits of Hertz theory to be probed. Glass, ruby and hardened steel balls 0.4 to 2.5 mm in diameter were dropped onto steel, glass, aluminum, and polymethylmethacrylate plates at a wide range of approach velocities, delivering frequencies up to 1.5 MHz into these materials. Effects of surface properties and yielding of the plate material were analyzed via the resulting stress waves and simultaneous measurements of the balls coefficient of restitution. The sensors are sensitive to surface normal displacements down to about +/-1 pm in the frequency range of 20 kHz to over 1 MHz.

Preslip and cascade processes initiating laboratory stick slip

Recent modeling studies have explored whether earthquakes begin with a large aseismic nucleation process or initiate dynamically from the rapid growth of a smaller instability in a “cascade-up” process. To explore such a case in the laboratory, we study the initiation of dynamic rupture (stick slip) of a smooth saw-cut fault in a 76 mm diameter cylindrical granite laboratory sample at 40–120 MPa confining pressure. We use a high dynamic range recording system to directly compare the seismic waves radiated during the stick-slip event to those radiated from tiny (M −6) discrete seismic events, commonly known as acoustic emissions (AEs), that occur in the seconds prior to each large stick slip. The seismic moments, focal mechanisms, locations, and timing of the AEs all contribute to our understanding of their mechanics and provide us with information about the stick-slip nucleation process. In a sequence of 10 stick slips, the first few microseconds of the signals recorded from stick-slip instabilities are nearly indistinguishable from those of premonitory AEs. In this sense, it appears that each stick slip begins as an AE event that rapidly (~20 µs) grows about 2 orders of magnitude in linear dimension and ruptures the entire 150 mm length of the simulated fault. We also measure accelerating fault slip in the final seconds before stick slip. We estimate that this slip is at least 98% aseismic and that it both weakens the fault and produces AEs that will eventually cascade-up to initiate the larger dynamic rupture.

A Robust Calibration Technique for Acoustic Emission Systems Based on Momentum Transfer from a Ball Drop

We describe a technique to estimate the seismic moment of acoustic emissions and other extremely small seismic events. Unlike previous calibration tech- niques, it does not require modeling of the wave propagation, sensor response, or signal conditioning. Rather, this technique calibrates the recording system as a whole and uses a ball impact as a reference source or empirical Greens function. To correctly apply this technique, we develop mathematical expressions that link the seismic mo- ment M0 of internal seismic sources (i.e., earthquakes and acoustic emissions) to the impulse, or change in momentum Δp, of externally applied seismic sources (i.e., me- teor impacts or, in this case, ball impact). We find that, at low frequencies, moment and impulse are linked by a constant, which we call the force-moment-rate scale factor C F _ MM0=Δp. This constant is equal to twice the speed of sound in the material from which the seismic sources were generated. Next, we demonstrate the calibration technique on two different experimental rock mechanics facilities. The first example is a saw-cut cylindrical granite sample that is loaded in a triaxial apparatus at 40 MPa confining pressure. The second example is a 2 m long fault cut in a granite sample and deformed in a large biaxial apparatus at lower stress levels. Using the empirical cali- bration technique, we are able to determine absolute source parameters including the seismic moment, corner frequency, stress drop, and radiated energy of these magnitude −2:5 to −7 seismic events.

Slip-pulse rupture behavior on a 2 m granite fault

We describe observations of dynamic rupture events that spontaneously arise on meter-scale laboratory earthquake experiments. While low-frequency slip of the granite sample occurs in a relatively uniform and crack-like manner, instruments capable of detecting high-frequency motions show that some parts of the fault slip abruptly (velocity > 100 mm s−1, acceleration > 20 km s−2) while the majority of the fault slips more slowly. Abruptly slipping regions propagate along the fault at nearly the shear wave speed. We propose that the dramatic reduction in frictional strength implied by this pulse-like rupture behavior has a common mechanism to the weakening reported in high-velocity friction experiments performed on rotary machines. The slip pulses can also be identified as migrating sources of high-frequency seismic waves. As observations from large earthquakes show similar propagating high-frequency sources, the pulses described here may have relevance to the mechanics of larger earthquakes.

High-fidelity conical piezoelectric transducers and finite element models utilized to quantify elastic waves generated from ball collisions

Experimental studies were performed using high-fidelity broadband Glaser-NIST conical transducers to quantify stress waves produced by the elastic collision of a tiny ball and a massive plate. These sensors are sensitive to surface-normal displacements down to picometers in amplitude, in a frequency range of 20 kHz to over 1 MHz. Both the collision and the resulting transient elastic waves are modeled with the finite element program ABAQUS and described theoretically through a marriage of the Hertz theory of contact and a full elastodynamic Greens function found using generalized ray theory. The calculated displacements were compared to those measured through the Glaser-NIST sensors.

Acoustic emission beamforming for enhanced damage detection

As civil infrastructure ages, the early detection of damage in a structure becomes increasingly important for both life safety and economic reasons. This paper describes the analysis procedures used for beamforming acoustic emission techniques as well as the promising results of preliminary experimental tests on a concrete bridge deck. The method of acoustic emission offers a tool for detecting damage, such as cracking, as it occurs on or in a structure. In order to gain meaningful information from acoustic emission analyses, the damage must be localized. Current acoustic emission systems with localization capabilities are very costly and difficult to install. Sensors must be placed throughout the structure to ensure that the damage is encompassed by the array. Beamforming offers a promising solution to these problems and permits the use of wireless sensor networks for acoustic emission analyses. Using the beamforming technique, the azmuthal direction of the location of the damage may be estimated by the stress waves impinging upon a small diameter array (e.g. 30mm) of acoustic emission sensors. Additional signal discrimination may be gained via array processing techniques such as the VESPA process. The beamforming approach requires no arrival time information and is based on very simple delay and sum beamforming algorithms which can be easily implemented on a wireless sensor or mote.

Integrating Broad-Band High-Fidelity Acoustic Emission Sensors and Array Processing to Study Drying Shrinkage Cracking in Concrete

Array processing of seismic data provides a powerful tool for source location and identification. For this method to work to its fullest potential, accurate transduction of the unadulterated source mechanism is required. In our tests, controlled areas of normal-strength concrete specimens were exposed to a low relative humidity at an early age to induce cracking due to drying shrinkage. The specimens were continuously monitored with an array of broad-band, high-fidelity acoustic emission sensors contrived in our laboratory in order to study the location and temporal evolution of drying shrinkage cracking. The advantage of the broadband sensors (calibration NIST-traceable) compared to more traditional acoustic emission sensors is that the full frequency content of the signals are preserved. The frequency content of the signals provides information about the dispersion and scattering inherent to the concrete, and the full unadulterated waveforms provide insight into the micromechanisms which create acoustic emissions in concrete. We report on experimental and analytical methods, event location and source mechanisms, and possible physical causes of these microseisms.

Absolute Calibration of an Acoustic Emission Sensor

Calibrated sensors are essential for quantitative comparisons of acoustic emission source mechanics. We describe experimental techniques and mathematical models for implementation of an absolute sensor calibration scheme using glass capillary fracture and ball impact, two sources which can be easily implemented. Additionally, we describe the specific experimental procedures for absolute sensor calibration using the both the fracture of a glass capillary tube, and drop of a 0.4 mm ruby ball, as the calibration source. The mathematical formulation is based on a Green’s function formalism. The Glaser-type conical piezoelectric sensor, used as an example in this study, has a noise floor of approximately 1 picometer displacement when coupled to steel. The amplitude of the sensor response is flat within 3 dB from 50 kHz to 2 MHz at a level of 0 dB relative to 1 V/nm.

Mechanisms of sliding friction studied with an array of industrial conical piezoelectric sensors

We use a new design of high-fidelity nanoseismic sensors to detect the stress waves produced at the initiation of sliding during stick-slip friction. The piezoelectric sensors can detect radiated waves just a few pm in amplitude in the frequency range of 10 kHz to over 2 MHz. The reported experiments are designed to provide insights that may be applicable to both fault scales and micro contact junctions. The sensors used are packaged in a hardened steel case to facilitate their use in the field. The transducers small size (14 mm threaded body, 30 mm long) permits a dense population of sensors to be installed on laboratory-sized samples, or surrounding localized centers of damage on structural applications. The closely spaced sensor array facilitates the localization of individual load releases from tiny asperities on a cm-scale frictional interface. At the same time, the broadband response of the conical piezoelectric sensors makes possible the study of source dynamics using theory developed for the study of earthquake source mechanisms via radiated seismic waves.