Application of Bayesian Regularization Artificial Neural Network in explosion risk analysis of fixed offshore platform

Computational Fluid Dynamics (CFD) is routinely used in Explosion Risk Analysis (ERA), as CFD-based ERA offers a good understanding of underlying physics accidental loads. Generally, simplifications were incorporated into CFD-based ERA to limit the number of simulations. Frozen Cloud Approach (FCA) is a frequently used simplification in the dispersion part of the CFD-based ERA procedure. However, its accuracy is questionable in the complex and congested environment such as offshore facility. Furthermore, in explosion part, some specific techniques, e.g. linear/double bin-interpolated techniques have been proposed while the corresponding accuracy is still unknown since the developers did not yet check their accuracy by considering the explosion computational data as the benchmark.This study presents a more accurate algorithm, namely Bayesian Regularization Artificial Neural Network (BRANN) and accordingly proposes the frameworks regarding BRANN-based models for the CFD-based ERA procedure. Firstly, the framework is proposed to develop the Transient-BRANN (TBRANN) model for transient dispersion study. In addition, the framework to determine the BRANN model for explosion study is developed. The proposed frameworks are explained by a case study of the fixed offshore platform. Consequently, this study confirms the more accuracy of the TBRANN model over FCA and the accuracy of BRANN model for CFD-based ERA.     

Configuration predictions of large liquefied petroleum gas (LPG) pool fires using CFD method

Liquefied petroleum gas (LPG) has potential pool fire risks due to its flammability. The configuration of pool fires plays a significant role when applying the solid flame model or point source model to assess the risks from heat radiation. However, no existing correlations can precisely predict the configuration of large LPG (100% propane) pool fires. To enhance the fundamental understanding on how pool diameter and wind velocity can influence the configuration of large LPG pool fires, an experimentally validated Computational Fluid Dynamics (CFD) model is employed to simulate fires using different burning rate models. Fire temperature profiles, flame heights, and flame tilts predicted by the CFD model were compared with empirical models and experimental data. Accordingly, new correlations for flame height and flame tilt as functions of pool diameter D and wind velocity u_w have been developed. The comparisons demonstrate that the new correlations have the best overall accuracy in the prediction of flame height and tilt for large LPG pool fires under different conditions (10 m ≤ D ≤ 20 m, 0 ≤ u_w ≤ 3 m·s⁻¹).     

Flame propagation through aluminum particle cloud in a combustion tube

The propagation of a flame is investigated experimentally and theoretically for a large, horizontal combustion tube containing a mixture of air and aluminum powder with pre-existing turbulence. One end of the tube is closed and the other is connected to a large dump-tank. Twenty dispersion systems are used on the tube to produce a uniform suspension of aluminum dust in the tube with a mean diameter of 6 μm. The characteristics of a flame front from the ignitors at the closed end are measured using photodiodes and the development of pressure is monitored by transducers. Experimental results revealed the entire process of an accelerating flame and the development of shock waves. A set of conservation equations for two-phase turbulent combustion flow is derived, using the two-fluid model, k–ε model, Hinze–Tchen model and EBU-Arrhenius model for turbulent combustion. The SIMPLE scheme usually applied to the homogeneous turbulent combustion is extended to fit this two-phase, reactive behavior. The results of calculations show the positive feedback coupling among combustion, expansion and turbulence during flame propagation. Computed and measured results are generally in good agreement.     

Extinguishment of hydrogen laminar diffusion flames by water vapor in a cup burner apparatus

Transient computations with full hydrogen chemistry were performed to reveal the flame structure and extinguishment process of co-flow, hydrogen diffusion flame suppressed by water vapor. As the concentration of water vapor was increased, the flame detached away from the burner brim and formed an edge flame at the flame base. Water vapor showed larger chemical inhibition effect than nitrogen when extinguishing hydrogen flame, which was attributed to its enhanced third body effect in the reaction H + O₂ + M = HO₂ + M. The minimum extinguishing concentration (MEC) of water vapor and nitrogen was predicted by Senecal formula and perfectly stirred reactor (PSR) model respectively. The MECs predicted by PSR model agree with the MECs calculated by Fluent, which shows that 1) the flame extinction is controlled by the flame base, and 2) radiation absorption is negligible. The measured MECs are in a reasonable agreement with the values calculated by Fluent, which demonstrates the accuracy of the CFD model. A simple model was used to investigate the relative importance of extinguishing mechanisms of water vapor. The results show that in a co-flow configuration the thermal cooling and chemical inhibition effect are the main extinguishing mechanisms in suppressing hydrogen diffusion cup burner flame.     

A method for simulation of vapour cloud explosions based on computational fluid dynamics (CFD)

The effectiveness of the application of CFD to vapour cloud explosion (VCE) modelling depends on the accuracy with which geometrical details of the obstacles likely to be encountered by the vapour cloud are represented and the correctness with which turbulence is predicted. This is because the severity of a VCE strongly depends on the types of obstacles encountered by the cloud undergoing combustion; the turbulence generated by the obstacles influences flame speed and feeds the process of explosion through enhanced mixing of fuel and oxidant. In this paper a CFD-based method is proposed on the basis of the author’s finding that among the various models available for assessing turbulence, the realizable k-? model yields results closer to experimental findings than the other, more frequently used, turbulence models if used in conjunction with the eddy-dissipation model. The applicability of the method has been demonstrated in simulating the dispersion and ignition of a typical vapour cloud formed as a result of a spill from a liquid petroleum gas (LPG) tank situated in a refinery. The simulation made it possible to assess the overpressures resulting from the combustion of the flammable vapour cloud. The phenomenon of flame acceleration, which is a characteristic of combustion enhanced in the presence of obstacles, was clearly observed. Comparison of the results with an oft-used commercial software reveals that the present CFD-based method achieves a more realistic simulation of the VCE phenomena.     

Efficient simulation of flame acceleration and deflagration-to-detonation transition in smooth pipes

Evaluation of accident scenarios including flame acceleration and deflagration-to-detonation transition (DDT) in chemical plant piping systems increases the need for an efficient numerical simulation tool capable of dealing with this phenomenon. In this work, a hybrid pressure-density-based solver including deflagrative flame propagation as well as detonation propagation is presented. The initial incompressible acceleration stage is covered by the pressure-based solver until the flame velocity reaches the fast flame regime and transition to the density-based solver is done. The deflagration source term is formulated in terms of a turbulent flame speed closure model incorporating various physical effects crucial for flame acceleration at low turbulence conditions (Katzy and Sattelmayer, 2018). Modelling of the detonation source term is based on a quadratic heat release function (Hasslberger, 2017). The presented numerical approach is validated in terms of DDT locations and pressure data from Schildberg (2015) as well as recently completed flame tip position measurements. For this purpose, H₂/O₂/N₂ mixtures ranging from 25.6 vol-% H₂ to 29.56 vol-% H₂ in two different pipe geometries are considered. The focus of the current work is on predicting the DDT location correctly and good agreement is observed for the investigated cases.     

Applications of dust explosion hazard and disaster prevention technology

Powdered materials are widely used in industrial processes, chemical processing, and nanoscience. Because most flammable powders and chemicals are not pure substances, their flammability and self-heating characteristics cannot be accurately identified using safety data sheets. Therefore, site staff can easily underestimate the risks they pose. Flammable dust accidents are frequent and force industrial process managers to pay attention to the characteristics of flammable powders and create inherently safer designs.This study verified that although the flammable powders used by petrochemical plants have been tested, some powders have different minimum ignition energies (MIEs) before and after drying, whereas some of the powders are released of flammable gases. These hazard characteristics are usually neglected, leading to the neglect of preventive parameters for fires and explosions, such as dust particle size specified by NFPA-654, MIE, the minimum ignition temperature of the dust cloud, the minimum ignition temperature of the dust layer, and limiting oxygen concentration. Unless these parameters are fully integrated into process hazard analysis and process safety management, the risks cannot be fully identified, and the reliability of process hazard analysis cannot be improved to facilitate the development of appropriate countermeasures. Preventing the underestimation of process risk severity due to the fire and explosion parameters of unknown flammable dusts and overestimation of existing safety measures is crucial for effective accident prevention.     

基于浓度敏感性分析和遗传算法的甲烷燃烧机理简化与优化

采用良搅拌反应器模型和层流预混火焰模型计算甲烷/空气燃烧过程,通过元素流通法和浓度敏感性分析法,对甲烷燃烧详细化学动力学机理GRIMECH 3.0进行简化。利用遗传算法,以甲烷/空气详细机理获得的组分浓度和一维层流火焰速度为目标,对简化机理进行优化。结果表明,相比于优化之前的简化机理,优化后的简化机理在描述甲烷/空气燃烧反应的组分浓度、层流火焰速度以及反应物和产物的时空分布方面,具有更高的精度。     

A quantitative risk assessment tool for the external safety of industrial plants with a dust explosion hazard

A quantitative risk assessment (QRA) tool has been developed by TNO for the external safety of industrial plants with a dust explosion hazard. As a first step an industrial plant is divided into groups of modules, defined by their size, shape, and constructional properties. Then the relevant explosion scenarios are determined, together with their frequency of occurrence. These include scenarios in which one module participates, as well as domino scenarios. The frequency is partly based on casuistry.

A typical burning velocity is determined depending on the ignition type, the dust properties and the local conditions for flame acceleration. The resulting pressure development is predicted with the ‘thin flame model’. Module failure occurs when the explosion load exceeds thresholds, which are derived from single degree of freedom (SDOF) calculations for various types of modules. A model has been developed to predict the process of pressure venting after module failure and the related motion of launched module parts.

The blast effects of the primary explosion are based on results from calculations with BLAST3D. The blast and flame effects of the secondary external explosion due to venting are calculated using existing models. The throw of fragments and debris is quantified with a recently developed model. This model is based on trajectory calculations and gives the impact densities, velocities, and angles as output. Furthermore the outflow of bulk material is taken into account. The consequences for external objects and human beings are calculated using existing models. Finally the risk contours and the Societal risk (FN curve) are calculated, which can be compared to regulations.     

液化天然气加注船安全定量风险分析

液化天然气（LNG）加注船是一种为LNG动力船提供燃料的新型船舶,国内目前尚处于起步阶段,缺乏相关安全标准和规范。采用国际定量风险评价（QRA）的通用理念,研究适用于我国LNG加注船的安全评价方法,提出具体实施步骤和依据准则,并以国内某LNG加注船作为实例分析说明。建立加注船LNG火灾事故树,确定相关火灾事故概率;研究适合加注船火灾事故后果的方形火焰模型,以及LNG火灾热辐射对人体伤害的计算方法;参考国际海事组织（IMO）的风险准则,确定LNG加注船个人风险和社会风险;最后与按NFPA-59A计算的防火间距作对比分析。通过计算,例中加注船的风险位于须采取相关安全措施的ALARP区域,风险控制区域半径20m。若按NFPA-59A要求计算防火间距,该加注船对外部须划定半径52m区域作为安全区域,且须将船身增长37m以满足内部防火要求,在实际工程中无法实现,相比较QRA方法更适合我国LNG加注船的安全评价工作。     

Interaction of flame instabilities and pressure behavior of hydrogen-propane explosion

The flame destabilization mechanism of hydrogen-propane-air mixture is firstly revealed. The effects of unstable flame formation on pressure rise rate and burning rate are quantified. Finally, the theoretical prediction of explosion pressure behavior is performed by considering diffusive-thermal and hydrodynamic instability. The results demonstrated that before the explosion pressure starts to climbe, as the propane fraction increases, the effective Lewis number of lean and stoichiometric mixture undergoes the transition from Le_eff < 1.0 to Le_eff > 1.0, the stabilizing effect of diffusive-thermal instability continues to reduce for the rich mixture. After the explosion pressure starts to climbe, the hydrogen-propane flame becomes more unstable, which is mainly attributed to enhancing hydrodynamic instability. The maximum rate of pressure rise and burning rate should be augmented by unstable flame formation, the flame instabilities must be considered in the explosion pressure evaluation.     

Scale effect of flame instability mechanisms on propagation velocity

Blast pressure is a major source of the damage caused by an accidental gas explosion. Its magnitude depends on the thermal expansion rate and hence on the flame propagation velocity. This paper discusses the influence of two flame instability mechanisms, i.e., diffusive-thermal and hydrodynamic instabilities, on flame propagation velocity with an emphasis on their scale effects as the flame continuously increases its size during an explosion. The Sivashinsky equation is numerically solved to simulate flame propagation behaviors. It is found that flame propagation velocity, V_f, is independent of flame size under the influence of diffusive-thermal instability, whereas V_f increases with flame size under the influence of hydrodynamic instability. The latter result is understood as a result of flame’s fractal structure. Fractal dimension is determined from the dependency of V_f on flame size, and the obtained fractal dimension is close to the known experimental value.     

Development of novel combustion risk index for flammable liquids based on unsupervised clustering algorithms

Current liquid flammability classification mainly relies on flash point and its risk is largely dependent on consequence and probability. However, combustions of liquefied marine fuels have their uniqueness, leading to a less consistent with the common classification. This work aims at classifying flammable liquids in compression ignition engines for further safety evaluation. Besides liquid flammability characteristics, flame propagation and aerosol formulation are considered. Two unsupervised machine learning clustering algorithms, k-means and spectral clustering, are applied to the collected liquid compounds database. To consider both cluster cohesion and separation, the global mean silhouette value is used to find the optimal number of clusters and to evaluate the clustering performance. The results show that the spectral clustering outperforms k-means on classifying the risk ratings for all proposed models, while the clustering accuracy of the optimal model has been doubled by employing spectral clustering algorithm. Moreover, principal component analysis and star coordinate diagrams are presented to visualize high dimensional data to 2-D graphs. Finally, the overall liquid safety performance is evaluated by a novel combustion risk index via the weight values determined by the information entropy approach. This index can be used to explore inherently safer fuels in the process industries.     

Analytical modelling of hydrocarbon pool fires: Conservative evaluation of flame temperature and thermal power

As well known, risk is a combination of probability and consequences of an accident. In analyzing the consequence of accidental hydrocarbon fires and the potential for domino effects, the evaluation of the flame extent and temperature are of the utmost importance. Since the primary effects of pool fires are connected to thermal radiation and issues of interplant/tank spacing employees’ safety zones, firewall specifications are to be addressed on the basis of a proper consequence analysis. By means of real scale experimental tests it was verified that both the thermal power and the flame temperature, T_f, increase as the pool area increases, up to reach maximum values in connection with a “critical pool dimension”. Dealing with pool areas higher than the critical one, experimental results, performed by different researchers at different scales, show a decrease of T_f. An in-depth analysis of the different concurring phenomena connected to a pool fire development allowed identifying the limiting step controlling the flame temperature. In fact, the trend of T_f is mainly determined by the increasing difficulty of oxygen diffusion within the internal bulk of gaseous hydrocarbons. In this article, we propose a novel pool fire modelling approach based on the simplified physical phenomena occurring in a circular turbulent diffusion fire and suitable to provide a theoretical insight into the above-mentioned experimental trends and to obtain the maximum values of the flame temperature and of the thermal power.The geometry of the pool is dictated by the surroundings (i.e., diking) and the analytical models here presented were successfully applied to the common situation of circular pools.However, it must be remarked that the developed model, matching fairly well experimental data for different hydrocarbons, can be applied in modelling similar scenarios characterized by different geometric or environmental conditions (e.g. road and rail tunnel fires).     

Large eddy simulation of methane–air deflagration in an obstructed chamber using different combustion models

In this paper, simulations of methane–air deflagration inside a semi-confined chamber with three solid obstacles have been carried out with large eddy simulation (LES) technique. Three sub-grid scale (SGS) combustion models, including power-law flame wrinkling model by Charlette et al., turbulent flame speed closure (TFC) model, and eddy dissipation model (EDM), are applied. All numerical results have been compared to literature experimental data. It is found that the power-law flame wrinkling model by Charlette et al. is able to better predict the generated pressure and other flame features, such as flame structure, position, speed and acceleration against measured data. Based on the power-law flame wrinkling model, the flame–vortex interaction during the deflagration progress is also investigated. The results obtained have demonstrated that higher turbulence levels, induced by obstacles, wrinkle the flame and then increase its surface area, the burning rates and the flame speed.     

Experimental study on the competing effect of ceramic pellets on premixed methane-air flame propagation in a duct

It is urgent to explore effective suppression methods for gas fires and explosions to ensure the safe utilizations of combustible gases in industrial processes. In this work, experiments are performed to study the effect of spherical ceramic pellets on premixed methane-air flame propagation in a closed duct. High-speed schlieren photography and pressure transducers are used to record the flame propagation and pressure transient, respectively. Behaviors of the flame propagating through a section of the duct filled with ceramic pellets in mixtures at different equivalence ratios are scrutinized. Three different diameters of pellets are considered in the experiments. The result shows that the flame can be quenched in the case with a smaller pellet diameter (3 mm) for a wide range of equivalence ratios from fuel-lean to fuel-rich mixture. For larger pellet diameter (5 or 10 mm), flame extinction occurs in fuel-rich mixtures (e.g. Φ = 1.1, 1.2). For the cases of flame surviving through the pellets bed, the pellets show a significant influence on the flame structure and behavior. The flame propagation depends on the porosity and the mean void diameter of the porous media in the pellets bed. Small void diameter is beneficial to flame quenching, while large porosity can accelerate the flame propagation. The pressure dynamics evolution is closely related to the interaction of flame with the pellets, and it depends on whether the flame quenches in the pellets bed. Overall, d = 3 mm ceramic pellets display the best suppression effect on flame propagation and pressure buildup in this study. The results of this study are of great significance to guide the safety design of spherical suppression materials in engineering applications for process safety researchers and engineers.     

Coupling effects of foam ceramics on the flame and shock wave of gas explosion

The coupling of gas explosion flame and shock wave is analyzed. In the gas explosion process, shock wave is affected by the flame directly, and shock wave also induces the flame. Inhibiting explosion can be achieved by the interference between the flame and shock wave propagation. If the coupling effects can be damaged, the adverse effects caused by the explosion should be mitigated and controlled. According to the structure characteristics of foam ceramics, the coupling effects mechanism of ceramic foam on gas explosion flame and shock wave is researched. When the explosion goes through the structure of foam ceramics, the flame can be quenched and the shock wave be attenuated. After the flame is quenched, the supply of precursor shock wave energy is cut off. Due to lack of energy supply, the destructive effects of blast wave will be reduced effectively. Coupling effects of the flame and shock wave can be damaged by the special structure of foam ceramics. Studies suggest that a certain function to represent the structure characteristics of foam ceramics must exist. For a certain material of foam ceramics, the sure porosity δ and the pore diameter d also can be get, which is the key to research and develop foam ceramic suppression technology of gas explosion.     

湍流模型和壁面函数对有障碍半开口空间氢气燃烧模拟的影响

为了进一步验证有障碍半开口空间内氢气燃烧数值模拟的准确性,使用对比验证的方法对FLUENT软件湍流模型和壁面函数数值模拟结果的准确性进行了研究。研究结果表明:可实现k-ε湍流模型和非平衡壁面函数使燃烧压力和火焰形状与实验结果相比偏差较大,RNG k-ε湍流模型和可伸缩壁面函数可更准确地预测燃烧压力变化及火焰传播行为。     

Experimental study on vented deflagration of hydrocarbon fuel-air mixtures in a 20-L semi-confined cylindrical vessel with a slight static activation pressure

The overpressure peaks and flame propagation characteristics of hydrocarbon fuel-air mixtures vented deflagration in a 20-L cylindrical vessel with a slight static activation overpressure (P_ST = 2.5 kPa) and five vent opening ratio were studied by a series of experiments. The experiments focused on the effect of vent opening ratio on the overpressure peaks and flame propagation characteristics of hydrocarbon fuel-air mixture vented deflagration. The internal overpressure-time profiles and high-speed photographs of flame propagation processes were obtained. The results showed that three overpressure peaks were distinguished in the internal overpressure-time profiles, caused by the burst vent cover (p_burst), the acceleration of burnt gas (p_fv), and the fierce external deflagration of vented unburned fuel (p_ext), respectively. The changing of the vent opening ratio had almost no effect on the value of p_burst and (dp_burst/dt). With increasing vent opening ratio, the values of p_fv, p_ext, (dp_fv/dt) and (dp_ext/dt) showed a decreasing trend while the values of p_burst and (dp_burst/dt) were nearly constant. The flame presented a hemispherical shape before the vent cover ruptured then developed as a mushroom shape after accelerated to external field. There were three flame speed peaks during flame propagation process, resulted from venting flow acceleration, external deflagration, and axial heat flux formed by internal combustion. With the increase of vent opening ratio, all of the maximum flame speed, external average flame speed, maximum flame distance and external flame duration showed a downward trend, excepting for the internal average flame speed almost remained constant.