On the prediction of the peak overpressure of ground reflected wave using a physics-informed weighted neural network

Hydrogen as one of the most promising clean energies, has received a lot of attention. Intensive research has been devoted to developing techniques for the effective usage of hydrogen energy. However, hydrogen is prone to gas explosion incidents, which may impose significant threats to people and structures in the vicinity. Blast overpressure is one of the main hazards from hydrogen explosion incidents. A simplified evaluation method for the prediction of peak overpressure from hydrogen-air cloud explosions in vented cylindrical vessels is proposed. Detailed three-dimension numerical models of vented cylindrical vessels with premixed hydrogen-air clouds were generated and validated with testing data. The numerical simulation results were used to establish the empirical correlation between peak overpressure and cylindrical vessels parameters including length and diameter, vent size and vent activation pressure. The influence of ambient temperature in the vented cylindrical vessels on the peak overpressure is also discussed. The simplified prediction method could be used for quick and reliable prediction of the peak blast overpressures in vented cylindrical vessels from accidental hydrogen explosions.

CFD simulations of hydrogen combustion in a simplified EPR containment with CFX and REACFLOW

Regulatory issues are discussed to establish Korean regulations on the safety distance between a very high temperature reactor (VHTR) and a hydrogen production facility. The major issues for the regulations concerning a gas explosion are an overpressure criteria, a regulation philosophy, and an overpressure prediction method. The overpressure can be predicted using empirical correlations of the trinitrotuluene (TNT) equivalent method and the multi-energy method (MEM). A comparison work of the predicted values using these correlations and the Japan Atomic Energy Agency (JAEA) explosion test results was performed to evaluate the applicability of these correlations to a VHTR. The MEM predicts the peak overpressure better than the TNT equivalent method because the explosion test results in a deflagration phenomenon. Thus, the MEM may be used effectively to estimate the peak overpressure for the gas explosion simply. A CFD analysis for the explosion test was also performed to establish an analysis methodology for a gas explosion. A spark ignition model to simulate an electric spark of 40 J in the JAEA explosion test was developed based on an energy conservation law. A sensitivity computational fluid dynamics (CFD) calculation was performed to elucidate the optimized pressure, temperature, and radius value of the spark ignition model. The CFD analysis results showed that the peak overpressure and the flame front time of arrival may be predicted better by the CFD analysis than by the MEM if the proper pressure and radius for the spark ignition model are chosen. So, the CFD analysis may be used as an accurate evaluation tool to provide the three-dimensional information of an overpressure and a time history of the overpressure variation. Therefore, it is recommended that the risk-informed regulation, the MEM, and the CFD analysis method should be used together to determine a safety distance.

Nonlinear seismic inversion by physics-informed Caianiello convolutional neural networks for overpressure prediction of source rocks in the offshore Xihu depression, East China

Comparative analysis of BLEVE mechanical energy and overpressure modelling

The objective of this research was to develop and to experimentally validate a method to predict the blast performance of thermobaric annular charges based on isopropyl nitrate‐aluminium and RDX‐isopropyl nitrate‐aluminium. Overpressure and thermal output were investigated in open air using piezoelectric pressure transducers, high speed visible and infrared imaging. Prediction of the explosive transformation of the annular thermobaric charge was computed by thermochemical code EXPLO5® using the Chapman‐Jouguet (ideal detonation) and non‐ideal Wood and Kirkwood (WK) detonation model. The Jones‐Wilkins‐Lee (JWL) parameters obtained in the calculations were used in hydrocode numerical simulation using AUTODYN® in order to predict the incident overpressure, total impulse and arrival time. Additional energy attributed to aluminium afterburning was used in order to simulate the post‐combustion phase. For isopropyl nitrate‐aluminium based charges, both BKW and WK models estimate correctly the peak overpressure and arrival time, while the total impulse was more accurately calculated by WK non‐ideal detonation model. For the RDX‐containing formulation, the prediction of peak overpressure, total impulse and arrival time are in agreement with the predictions obtained using WK non‐ideal detonation model with additional energy. The performances of the investigated thermobaric charges were compared with an equivalent mass charge of TNT, confirming both the expected higher total impulse and higher peak overpressures in the case of RDX‐isopropyl nitrate‐aluminium based thermobaric charges. Thermal measurements highlighted two distinct phases, an anaerobic phase and the post combustion of the thermobaric compositions. In terms of thermal output, the RDX‐isopropyl nitrate‐aluminium thermobaric based charge manifests superior performances.

A semiempirical method has been developed for predicting the peak surge-induced overpressures in the vicinity of the engine face as a function of engine cycle variables. The method is applicable to relatively long inlets, such as those typically used on supersonic airplanes. (For the shorter inlets typical of subsonic airplanes pressure relief is obtained at the inlet entrance before the peak overpressures are reached in the inlet.) A correlation of existing overpressure data is obtained by combining an analytical solution for the effect of engine bypass ratio with an empirical evaluation of the effect of over-all compressor pressure ratio. The resulting method can be used to obtain boundary conditions for inlet dynamic simulation programs that predict the effects of inlet duct geometry and auxiliary air systems on overpressure characteristics. The method predicts that higher engine bypass ratios lead to significant reductions in peak overpressures.

Many basins are dominated by vertical hydrocarbon migration. On seismic data, the vertical migration paths are generally recognized as vertically aligned zones of chaotic, often low-amplitude reflectivity. These are described variously as gas chimneys, blowout pipes, gas clouds, or hydrocarbon-related diagenetic zones. Analysis of the gas chimneys can be used for geohazard prediction, basin modeling, prospect risking, and many more applications. However, the weak expressions of gas chimneys in seismic data make them difficult to map. Thus, a method for detecting gas chimneys in poststack 3D seismic data has been developed to map their distribution and allow them to be visualized in three dimensions. This chimney probability volume is produced by a neural network from multiple seismic attributes extracted at examples of gas chimneys picked by the interpreter. Not all vertically aligned, low-amplitude, chaotic seismic reflectors represent hydrocarbon migration. Therefore, the subjective selection of training locations and the resulting neural-network predictions are validated by objective criteria before the results are used in geologic applications. Gas-chimney detection methods originally were used to highlight gas chimneys in shallow intervals to detect shallow gas reservoirs and geohazards. However, the methodology was soon used to highlight subtle, deep hydrocarbon migration pathways, hydrocarbon migration related to faulting, and expulsion from source rock. Other applications of gas-chimney analysis are overpressure prediction and prediction of the quality of gas-hydrate accumulations. Chimneys are classified based on their morphology and the relative position of the trap, faults, and chimneys. This classification provides criteria for risking top seal, vertical fault seal, and hydrocarbon charge on exploration prospects prior to drilling.

Joanne Wang, Duane Dopkin and Huw James (Paradigm Geophysical) discuss how visualization and interpretation of multi-disciplinary data volumes can improve risk assessment of predicted overpressures. Seismic data transformations are used routinely by geoscientists to enhance geometric features and physical property descriptions of the subsurface. These transformations can range from the application of one-dimensional operators to migrated and stacked seismic amplitude data to the application of complex transforms that make use of multiple dimensions (e.g. amplitude versus angle) or multiple data inputs (e.g. seismic inversion procedures). They can be extended with the application of linear or non-linear operators (e.g. neural networks) to relate the resultant elastic properties (e.g. impedance) to other rock properties or reservoir conditions making use of petrophysics and rock physics data. While these transformations exploit the dynamic properties of seismic data, other transforms make use of the kinematic properties of multi-offset seismic data to estimate or predict subsurface conditions. Pore pressure predictions and transformations that make use of seismic velocity measurements, for example, have had a huge impact on drilling safety and the economics of drilling design and well construction. The pore pressure models derived from the integration and careful calibration of wireline, petrophysical, seismic velocity and field test measurements provide the data needed to make critical pre-drill stress and overpressure predictions in order to secure a safe and economic well program. Because most exploration and development projects incorporate many data transformations, a heavy burden is often placed on the geoscientist and drilling engineer to integrate a broad range of data volumes and models. Although automated procedures (e.g. PCA analysis, neural network classification) are available to facilitate multi-volume integration, high-end visualization technologies and strategies often provide the most effective and informative data integration vehicles. The multi-disciplinary pore pressure prediction and transformation process that integrates seismic, well log, petrophysical, field test data with structural frameworks and directional well paths can return a wide range of deliverables (outputs) that are natural for creative co-visualization and concurrent interpretation by geophysicists, geologists, and drilling engineers. These co-visualizations are often the most effective way to understand the ‘interplay’ or dependencies of one transformation with another.

This study aimed at elucidating some characteristics of the shock wave overpressure generated by a non-traditional layered charge comprising an inner high-energy explosive and an outer polymer matrix composite. Two models for predicting the peak overpressure (Δpm) of the charge were established, namely, a model based on the initial parameters of the blast wave, and a model considering the weakening of the explosion energy through the introduction of polymer matrix cladding. The overpressure of a typical layered charge was experimentally measured for model validation. It was found that the difference between the Δpm predicted by the two models and the experimental data is less than 15.12% and 14.17%, respectively. The model that was established based on the conservation of energy law, is in best agreement with the experimental data under different cladding/charge mass ratios (αm). The model that was based on the initial parameters of the blast wave obtained a low predicted value when αm was 0.4-0.8, which is attributed to the non-uniformity of the gas-solid mixture during the explosive dispersion stage.

Two methods of sonic boom predictions are used to compare predicted overpressures to measured sonic booms to quantify the effects of weather and flight variability on the lateral propagation of the booms. The sonic boom measurements were collected by the US Air Force at Edwards AFB, CA using a variety of supersonic aircraft. The testing included 43 flights from eight different aircraft at several Mach number and altitude combinations. The flights were planned for steady and level operation but some maneuvering of the aircraft caused focusing of the propagated booms at several measurement locations. The comparison of measured to predicted peak overpressures uses both Carlson and Full ray-tracing algorithms. The comparison examines the lateral spread of the boom and the effects on local weather profiles and variability of the flight profile on the propagated boom. This analysis demonstrates that, in general, predictions are in good agreement with the measured values but tend to overestimate the peak overpressures near the lateral cutoff.

Prediction of near field overpressure from quarry blasting

Physics-informed machine learning model for prediction of ground reflected wave peak overpressure

In order to predict the effects of gas volume and mass inertia of explosion doors on overpressure for venting of gas explosion, a numerical method are used to simulate some situations of venting devices such as panels or explosion doors. The numerical data are in good agreement with the experiments of panels given an overpressure peak. A set of experiments were carried out in a 10 m3 silo and methane is used as fuel. The results of measurement in experiments indicated that the inertial mass of explosion door has a possibility of causing considerable damage and has effected on the rate of opening vented door and the pressure rise in such the experimental silo.

The role of manned space flight in the field of space exploration and utilization is growing. However, the security system of the manned spaceflight is still imperfect. In the case that the rocket explodes, crew modules maybe damaged by the blast wave, which will threaten the safety of the crews. This research aims to obtain the necessary data and information to enable the designers of the launch vehicles and crew modules to develop safer launch systems. To this end, this paper proposes a numerical method using LS-DYNA to study the propagation law of blast waves caused by rocket explosion on the launch pad and to quantify the impact of the blast wave on crew module. The numerical results indicate that the final blast waveform of the model with rocket is conical in the upper and lower parts, and spherical in the middle. At the same time, the third-stage explosion is the most harmful to the crew module, while the first-stage explosion is the least. Furthermore, the model with rocket has a marked effect on explosion strength: the pressure enhancement factor is about 4–17 times. Most importantly, overpressure prediction formula acting on the crew modulesof explosion on the launch pad is established for quick peak overpressure predicting and damage evaluating.