TY - BOOK
T1 - Blast Array Optimization for Vibration Reduction in Heterogeneous Models
AU - Trabi, Bernd
N1 - no embargo
PY - 2024
Y1 - 2024
N2 - Predicting Peak Ground Velocity (PGV) is crucial for blast mining operations to ensure that the charge weights are designed to stay within safe limits and to prevent surrounding structures from potential damage. The problem is often complicated by significant scatter in observed PGV, mainly due to unknown complexities of seismic wave propagation. It is difficult to estimate the contribution of each wave type to the PGV, as it involves a mix of overlapping seismic phases. The data generally suggest that the direct P-wave is the primary contributor. Traditional PGV prediction models often rely on empirical approaches, and the Scaled Distance (SD) method is particularly popular. This method, which has the fewest parameters to calibrate, is also effective for single-sensor setups. A dataset of 55 mining production blasts recorded with an array of 81 seismic sensors is used to compare the performance of different methods. The large array also allows to apply a multi-sensor inversion that provides more insight into the physical parameters. The findings reveal that the classical SD approach underperforms at this particular site, mainly because the empirical relationship between the radial amplitude decay constant b and charge weight exponent c does not hold in this context. The determined charge weight exponent value of 0.5 likely reflects a fundamental physical relationship between charge, energy, and amplitude, suggesting its potential applicability across various sites. A new numerical prediction method is presented based on forward-modelled waveforms on a 3D velocity model. The numerical simulations were done with the Barcelona Subsurface Imaging Tools for 80 predefined blast positions. The final data is then further processed to simulate future blasts at arbitrary positions with a realistic source time function and amplitude spectrum. The numerical method can simulate waveforms that match well with the observed waveforms, at least in the P-wave window. The observed data show no clear S-wave but a considerable amount of surface waves, which cannot be reflected by the numerical waveform modelling method. Even though the method can predict PGV, which is a highly non-linear measure, PGV prediction is outperformed by simple engineering methods. With the interference of seismic waves from spatially separated production blasts, a reduction in impact on specific target zones should be achievable. While modifying the blast pattern in a homogeneous model to minimize the effect on a specific target zone is relatively straightforward, it is challenging to find such optimal configurations in a complex numerical model that includes velocity variations and topography. A Markov Chain Monte Carlo (McMC) method is used to find such a configurations. The cost function of the McMC seeks to minimize the PGV in certain target zones. These derived configurations were later implemented in real production blasts.
AB - Predicting Peak Ground Velocity (PGV) is crucial for blast mining operations to ensure that the charge weights are designed to stay within safe limits and to prevent surrounding structures from potential damage. The problem is often complicated by significant scatter in observed PGV, mainly due to unknown complexities of seismic wave propagation. It is difficult to estimate the contribution of each wave type to the PGV, as it involves a mix of overlapping seismic phases. The data generally suggest that the direct P-wave is the primary contributor. Traditional PGV prediction models often rely on empirical approaches, and the Scaled Distance (SD) method is particularly popular. This method, which has the fewest parameters to calibrate, is also effective for single-sensor setups. A dataset of 55 mining production blasts recorded with an array of 81 seismic sensors is used to compare the performance of different methods. The large array also allows to apply a multi-sensor inversion that provides more insight into the physical parameters. The findings reveal that the classical SD approach underperforms at this particular site, mainly because the empirical relationship between the radial amplitude decay constant b and charge weight exponent c does not hold in this context. The determined charge weight exponent value of 0.5 likely reflects a fundamental physical relationship between charge, energy, and amplitude, suggesting its potential applicability across various sites. A new numerical prediction method is presented based on forward-modelled waveforms on a 3D velocity model. The numerical simulations were done with the Barcelona Subsurface Imaging Tools for 80 predefined blast positions. The final data is then further processed to simulate future blasts at arbitrary positions with a realistic source time function and amplitude spectrum. The numerical method can simulate waveforms that match well with the observed waveforms, at least in the P-wave window. The observed data show no clear S-wave but a considerable amount of surface waves, which cannot be reflected by the numerical waveform modelling method. Even though the method can predict PGV, which is a highly non-linear measure, PGV prediction is outperformed by simple engineering methods. With the interference of seismic waves from spatially separated production blasts, a reduction in impact on specific target zones should be achievable. While modifying the blast pattern in a homogeneous model to minimize the effect on a specific target zone is relatively straightforward, it is challenging to find such optimal configurations in a complex numerical model that includes velocity variations and topography. A Markov Chain Monte Carlo (McMC) method is used to find such a configurations. The cost function of the McMC seeks to minimize the PGV in certain target zones. These derived configurations were later implemented in real production blasts.
KW - Prediction of blast vibrations
KW - numerical modelling
KW - peak ground velocity
KW - PGV
KW - PGV prediction
KW - SLIM
KW - mining blasts
KW - Markov Chain Monte Carlo
KW - McMC
KW - Markov chain optimization
KW - reduction of blast vibrations
KW - Eisenerz
KW - seismic waveform modelling
KW - Vorhersage von Sprengerschütterungen
KW - numerische Modellierung
KW - maximale Bodengeschwindigkeit
KW - PGV
KW - PGV-Vorhersage
KW - SLIM
KW - Sprengungen im Bergbau
KW - Markov-Kette-Monte-Carlo
KW - McMC
KW - Markov-Ketten-Optimierung
KW - Reduzierung von Sprengerschütterungen
KW - Erzberg
KW - seismische Wellenformmodellierung
U2 - 10.34901/mul.pub.2025.025
DO - 10.34901/mul.pub.2025.025
M3 - Doctoral Thesis
ER -