Empirical Acoustic Source Model for Chemical Explosions in Air

Abstract

Chemical explosions generate pressure disturbances in air that radiate as nonlinear shock waves near the source and transition into acoustic waves with distance. Because low-frequency acoustic waves generally travel large distances without significant loss of energy, they are often used for explosion monitoring and yield estimation. However, quantitative relationships between acoustic energy and explosion yields are required for accurate yield estimation. Here, we develop an empirical acoustic source model for chemical explosions from experimental data. The empirical model returns the acoustic pressure waveform for the detonation of 1 kg of trinitrotoluene, which is conventionally used to represent the explosive release of 4.184 MJ of explosion energy. The full-waveform model can be used to predict acoustic signals for an arbitrary yield of a high-explosive detonation based on the standard scaling law and to estimate acoustic energies in a specific frequency range. We evaluate the accuracy of the acoustic source model independently by estimating the yield of other explosive events that are not included in the model development. Statistical characteristics of the model and their implications for the uncertainty quantification of estimated yields are discussed.

Yield Estimation of the August 2020 Beirut Explosion by Using Physics‐Based Propagation Simulations of Regional Infrasound

“…This result is somehow surprising. The K21 source model was developed by local data (<10 km) obtained from chemical explosion experiments (Kim et al., 2021), and we expected the source model variance would be smaller than the variance of numerical simulations for long‐range propagation. The larger source uncertainty may be explained by three factors.…”

“…The acoustic source energy is calculated by the empirical acoustic source model proposed by Kim et al. (2021), hereafter called K21. The model provides full waveforms as a function of the yield of surface explosions allowing us to calculate acoustic energy excited in designated frequency bands (Text S1 in Supporting Information ).…”

“…We predict acoustic energy (Ep) between two frequencies ( 𝐴𝐴 𝐴𝐴1 and 𝐴𝐴 𝐴𝐴2 ) by multiplying acoustic source energy (S(W, f, x 0 )) at a position x 0 by the transmission loss ( 𝐴𝐴 𝐴𝐴 (𝑓𝑓 𝑓 𝑓𝑓; 𝑓𝑓0) ) from x 0 to x. The acoustic source energy is calculated by the empirical acoustic source model proposed by Kim et al (2021), hereafter called K21. The model provides full waveforms as a function of the yield of surface explosions allowing us to calculate acoustic energy excited in designated frequency bands (Text S1 in Supporting Information S1).…”

“…Particularly, Fourier amplitudes between 0.1 and 1.0 Hz show the highest signal‐to‐noise ratio and consistency across the five elements with small variance. Larger surface explosions tend to produce lower‐frequency infrasound due to their spatial scales (Kim et al., 2021), and this substantially higher energy in the lower frequencies indicates the considerable size of the Beirut explosion. Acoustic amplitudes at relatively higher frequencies (>1.0 Hz) fluctuate across the array.…”

Volcano infrasound: progress and future directions

“…Integrated seismo-acoustic ML analysis will be particularly beneficial. Additionally, the volcano infrasound community is well positioned to synthetically create data given the various source time function models (Kinney & Graham 1985;Kim et al 2021), atmospheric models (Schwaiger et al, 2019), and propagation software 2020) currently available. Along with other physics-based data augmentation strategies, this may supplant the need to collect and label large datasets and will help generalize ML methodologies.…”