Modeling of a cryogenic liquid pool boiling by CFD simulation

A boiling model is developed by Computational Fluid Dynamics (CFD) code to calculate the source term of a cryogenic liquid spill. The model includes the effect of the changing ground temperature on the vaporization rate of the cryogenic liquid. Simulations are performed for liquid nitrogen. The model can describe different boiling regimes (film, transition and nucleate). The heat flux calculated for each boiling regimes are compared to the experimental data from literature. The developed numerical model seems to have a good ability to predict the heat flux for the film boiling stage. Model development is still necessary to improve the prediction of the nucleate boiling regime. Overall, the approach shows very promising results to model the complex physical phenomena involved in in the vaporization of cryogenic liquid pool spilled on ground.     

Validation of liquid nitrogen vaporisation rate by small scale experiments and analysis of the conductive heat flux from the concrete

The vaporisation of a liquid nitrogen pool spilled on concrete ground was investigated in small scale field experiments. The pool vaporisation rate and the heat transfer from the concrete ground were measured using a balance and a set of embedded heat flux sensors and thermocouples. The ability to predict the concrete's thermal properties based on these measurements was investigated. This work showed that a simple, one-dimensional theoretical model, assuming heat conduction through a semi-infinite ground with ideal contact between the cryogenic liquid and the ground, commonly used to describe the heat transfer from a ground to the LNG, can be used to match the observed vaporisation rate. Though estimated parameters, thermal conductivity and thermal diffusivity, do not necessary represent real values. Although the observed vaporization rate follows a linear trend, and thus can be well represented by the model, the overall model prediction seems to be overestimated. The temperature profile inside the concrete is slightly over-predicted at the beginning and under-predicted at later stage of the spill. This might be an effect of the dependence of the concrete's thermal properties on the temperature or may indicate an incorrect modelling and a varying temperature of the ground surface.     

Experimental study and prediction of liquid nitrogen pool evaporation process on concrete surface

The vaporization rate of pool boiling process of a liquid nitrogen spilled on concrete surface was investigated by a visual experiment platform. The boiling curve for liquid nitrogen on concrete cooling process was obtained. The shapes of bubbles in three typical boiling regimes were observed. Based on the experimental results, the coefficients of the empirical formulas for nuclear boiling and film boiling are modified, and the empirical formulas for boiling of liquid nitrogen-concrete surfaces are obtained. Combined with the calculation formula of the non-steady-state semi-infinite one-dimensional heat conduction temperature, a coupled calculation model for the heat transfer and vaporization process of liquid pool on the liquid nitrogen-concrete surface is proposed. Application of this model can better predict the quality of liquid nitrogen vaporization.     

Experimental study of the evaporation of spreading liquid nitrogen

The investigation of cryogenic liquid pool spreading is an essential procedure to assess the hazard of cryogenic liquid usage. There is a wide range of models used to describe the spreading of a cryogenic liquid pool. Many of these models require the evaporation velocity, which has to be determined experimentally because the heat transfer process between the liquid pool and the surroundings is too complicated to be modeled. In this experimental study, to measure the evaporation velocity when the pool is spreading, liquid nitrogen was continuously released onto unconfined concrete ground. Almost all of the reported results are based on a non-spreading pool in which cryogenic liquid is instantaneously poured onto bounded ground for a very short period of time. For the precise measurement of pool spreading and evaporation weight with time, a cone-type funnel was designed to achieve a nearly constant liquid nitrogen release rate during discharge. Specifically, three nozzles with nominal flow rates of 3.4 × 10⁻² kg/s, 5.6 × 10⁻² kg/s and 9.0 × 10⁻² kg/s were used to investigate the effect of the release rate on the evaporation velocity. It is noted that information about the release rate is not necessary to measure the evaporation velocity in case of the non-spreading pool. A simultaneous measurement of the pool location using thermocouples and of the pool mass using a digital balance was carried out to measure the evaporation velocity and the pool radius. A greater release flow rate was found to result in a greater average evaporation velocity, and the evaporation velocity decreased with the spreading time and the pool radius.     

Evaluation of CFD simulations of transient pool fire burning rates

Fire is the most commonly occurring major accident hazard in the chemical and process industries, with industry accident statistics highlighting the liquid pool fire as the most frequent fire event. Modelling of such phenomena feeds heavily into industry risk assessment and consequence analyses. Traditional simple empirical equations cannot account for the full range of factors influencing pool fire behaviour or increasingly complex plant design. The use of Computational Fluid Dynamics (CFD) modelling enables a greater understanding of pool fire behaviour to be gained numerically and provides the capability to deal with complex scenarios.This paper presents an evaluation of the Fire Dynamics Simulator (FDS) for predictive modelling of liquid pool fire burning rates. Specifically, the work examines the ability of the model to predict temporal variations in the burning rate of open atmosphere pool fires. Fires ranging from 0.4 to 4 m in diameter, involving ethanol and a range of liquid hydrocarbons as fuels, are considered and comparisons of predicted fuel mass loss rates are compared to experimental measurements.The results show that the liquid pyrolysis sub-model in FDS gives consistent model performance for fully predictive modelling of liquid pool fire burning rates, particularly during quasi-steady burning. However, the model falls short of predicting the subtleties associated with each phase of the transient burning process, failing to reliably predict fuel mass loss rates during fire growth and extinction. The results suggest a range of model modifications which could lead to improved prediction of the transient fire growth and extinction phases of burning for liquid pool fires, specifically, investigation of: ignition modelling techniques for high boiling temperature liquid fuels; a combustion regime combining both infinite and finite-rate chemistry; a solution method which accounts for two- or three-dimensional heat conduction effects in the liquid-phase; alternative surrogate fuel compositions for multi-component hydrocarbon fuels; and modification of the solution procedure used at the liquid-gas interface during fire extinction.     

细水雾扑灭油池火过程中火焰强化现象的实验研究

在细水雾扑灭油池火的初期,某些油池火焰会有一个突然强化、剧烈燃烧的情况发生。通过实验研究证明,酒精等与水互溶的液体不会发生强化现象,汽油和柴油等不溶于水的液体则有明显的强化现象;同时,发生强化现象的根本原因是共沸使得细水雾到达燃料表面后急剧沸腾而强化燃烧。     

氨分解装置中液氨储罐火灾爆炸危险性分析

以某金属处理企业氨分解装置中液氨储罐罐区为例,对液氨泄漏后火灾爆炸事故及其伤害范围进行了研究,用池火、蒸气云爆炸和沸腾液体扩展蒸气爆炸模型进行计算分析,给出火灾、爆炸事故的人员伤害和财产损失范围。结果表明:围堤堤内池火或罐内池火时,罐区建构筑物内的汽化器、管道等设备会因直接过火或热辐射导致损坏,建筑内人员死亡,但难以波及罐区之外;蒸气云爆炸产生相当于1192.72kgTNT爆炸的当量,爆炸的后果严重,应重点防范,防范的重点为液氨泄漏、点火源;沸腾液体扩展蒸气爆炸的火球半径56.1m,持续时间8.7s,死亡半径27.2m,其源于储罐受热或系统突然失效,液体瞬时泄漏汽化并遇点火源而发生,具有突发性且后果严重,企业应高度重视并严格储罐及系统的定期检验与校验、密切关注系统的有效运行。     

Development of heat transfer and evaporation model of LNG covered by Hi-Ex foam

Natural gas is a kind of clean, efficient green energy source, which is used widely. Liquefied natural gas (LNG) is produced by cooling natural gas to −161 °C, at which it becomes the liquid. Once LNG was released, fire or explosion would happen when ignition source existed nearby. The high expansion foam (Hi-Ex foam) is believed to quickly blanket on the top of LNG spillage pool and warm the LNG vapor to lower the vapor cloud density at the ground level and raising vapor buoyancy. To identify the physical structure after it contacted with LN₂ and to develop heat transfer model, the small-scale field test with liquid nitrogen (LN₂) was designed. In experiment, three layers including frozen ice layer, frozen Hi-Ex layer and soft layer of Hi-Ex foam were observed at the steady state. By characterizing physical structure of the foam, formulas for calculating the surface of single foam bubble and counting foam film thickness were deduced. The micro heat transfer and evaporation model between cryogenic liquid and Hi-Ex foam was established. Indicating the physical structure of the frozen ice layer, there were a certain number of icicles below it. The heat transfer and evaporation mathematical model between the frozen ice layer and LNG was derived. Combining models above with the heat transfer between LNG, ground and cofferdam, the heat transfer and evaporation mathematical model of LNG covered by Hi-Ex foam was developed eventually. Finally, LN₂ evaporation rate calculated by this model was compared with the measured evaporation rate. The calculated results are 1.2–2.1 times of experimental results, which were acceptable in engineering and proved the model was reliable.     

Quantification of turbulence in cryogenic liquid using high speed flow visualization

A high speed flow visualization experiment was conducted to characterize the boiling induced turbulence when a cryogenic liquid is released on water. The advective transport of turbulent structures traversing through the liquid was captured and reconstructed using image processing software to obtain information on velocity components. The numerical results obtained from image processing were used to determine turbulence parameters like turbulent intensity, turbulent kinetic energy and eddy dissipation rate. An interesting aspect of the study was the formation of wavy structures called ‘thermals’ which were characteristic of turbulent convection. The thermals were found to act as a catalyst in increasing heat transfer and turbulence between water and cryogenic pool. The turbulent intensity was influenced by the turbulent velocity and had direct effects on the vaporization flux. Among the turbulence parameters, increase in turbulent kinetic energy resulted in faster vaporization of cryogenic liquid through enhanced mixing, whereas variations in the eddy dissipation rate had weak dependence on vaporization. Additionally, the initial height of cryogenic liquid was also found to strongly affect the vaporization mass flux.     

Modeling of pool spreading of LNG on land

This paper presents a source term model for estimating the rate of spreading of LNG and other cryogenic mixtures on unconfined land. The model takes into account the composition changes of a boiling mixture, the varying thermodynamic properties due to preferential boiling within the mixture and the effect of the various boiling regimes on conductive heat transfer. A sensitivity analysis is conducted to determine the relative effect of each of these phenomena on pool spread. The model is applied to continuous and instantaneous spills. The model is compared to literature experimental data on cryogenic pool spreading.     

低温液体在水中快速相变过程流动与传热数值模拟

用试验和理论分析的方法对快速相变爆炸强度的预测缺乏定量模型,因此建立了一种欧拉-欧拉双流体多相流模型与传热模型相互耦合的数值模型,并通过与Clarke H将液氮喷射入水的快速相变试验数据对比来验证模型的可靠性和正确性。通过数值计算得出快速相变过程中流场、压力场、温度场随时间变化的情况,探讨了快速相变的传热机理。结果表明:快速相变是强制对流、膜态沸腾、爆发沸腾和核态沸腾之间的转换过程;相间换热系数随时间的无量纲变化关系可以用3个高斯分布的叠加来描述。     

Blast overpressures from medium scale BLEVE tests

The measured blast overpressures from recent tests involving boiling liquid expanding vapour explosions (BLEVE) has been studied. The blast data came from tests where 0.4 and 2 m³ ASME code propane tanks were exposed to torch and pool fires. In total almost 60 tanks were tested, and of these nearly 20 resulted in catastrophic failures and BLEVEs. Both single and two-step BLEVEs were observed in these tests. This paper presents an analysis of the blast overpressures created by these BLEVEs. In addition, the blast overpressures from a recent full scale fire test of a rail tank car is included in the analysis.The results suggest that the liquid energy content did not contribute to the shock overpressures in the near or far field. The liquid flashing and expansion does produce a local overpressure by dynamic pressure effects but it does not appear to produce a shock wave. The shock overpressures could be estimated from the vapour energy alone for all the tests considered. This was true for liquid temperatures at failure that were below, at and above the atmospheric superheat limit for propane. Data suggests that the two step type BLEVE produces the strongest overpressure. The authors give their ideas for this observation.The results shown here add some limited evidence to support previous researchers claims that the liquid flashing process is too slow to generate a shock. It suggests that liquid temperatures at or above the Tsl do not change this. The expansion of the flashing liquid contributes to other hazards such as projectiles, and close in dynamic pressure effects. Of course BLEVE releases in enclosed spaces such as tunnels or buildings have different hazards.     

Prediction models for the flash point of pure components

The flash point for a liquid is defined as the lowest temperature at which its vapor forms a spontaneously ignitable mixture when brought in contact with air. Having a good estimation of the flash point temperature at normal conditions is relevant because it is one of the main properties used to characterize fire and explosion hazards for liquids. Most of predictive correlations in the literature include a physical effect by including the normal boiling temperature in their formulation. To achieve combustion, sufficient evaporation is required and then a part of the liberated energy from the combustion is used to support the evaporation. Thus the evaporation energy and a chemical effect given by the heat of reaction are incorporated in this work. It is firstly verified that the relation between the flash point temperature and the normal boiling temperature tends to be a constant. Thus a heuristic approach based on 611 chemical species of diverse families indicates that this relation is around 0.75. The dispersion of the error has been reduced by using two proposed correlations where both physical and chemical properties are included in the model. In particular, the second equation is based on the group contributions method, which has been developed for alkanes. This method is in fact a combination of the group contributions method and the first model to overcome the difficulties in predicting isomeric differences.     

The boiling liquid expanding vapour explosion

The boiling liquid expanding vapour explosion (BLEVE) has existed for a long time and for most of this time it has been cloaked in mystery. Several theories have been put forward to explain this very energetic event but none have been proven. This paper describes a series of tests that have recently been conducted to study this phenomenon.

The study involved ASME code automotive propane tanks with nominal capacities of 400 litres. The tanks were exposed to a combination of pool and/or torch fires. These fire conditions led to thermal ruptures, and in some cases these ruptures resulted in BLEVEs. The variables in the tests were the pressure-relief valve setting, the tank wall thickness, and the fire condition.

In total, 30 tests have been conducted, of which 22 resulted in thermal ruptures. Of those tanks that ruptured, 11 resulted in what we call BLEVEs. In this paper, we have defined a BLEVE as the explosive release of expanding vapour and boiling liquid following a catastrophic tank failure. Non-BLEVEs involved tanks that ruptured but which only resulted in a prolonged jet release.

The objective of this study was to investigate why certain tank ruptures lead to a BLEVE rather than a more benign jet-type release. Data are presented to show how wall temperature, wall thickness, liquid temperature and fill level contribute to the BLEVE process.     

On models of spreading pools

The spread of liquid pools floating on water is commonly modelled by the assumption of a constant (densimetric) Froude number at the front. This can be understood in terms of a balance between the pressure difference across the front of the spreading pool, and the resistance at the front from having to push displaced water out of the way. Some early models of pools spreading on land also assumed constant Froude number, but in this case there is absolutely no credible justification for such an assumption. This was highlighted by Webber and Jones (1987) who noted that resistance will come from friction with the ground under the whole base of the pool, resulting in a very different spreading law. Nevertheless, the assumption of constant Froude number spreading of pools on land continues in some circles to this day, and a recent paper by Raj (2011) even goes as far as to assert that Webber and Jones (1987) assumed the spreading law, which in actual fact they were at pains to debunk. This paper is intended to set the record straight, with a detailed discussion of the physical phenomena controlling the way pools spread on land.     

浸油沙层火焰传播速度研究

使用0#柴油和沙层作为燃料和地面模型,通过实验研究了泄漏液体燃料渗透在地面之后的火焰传播现象.详细研究了燃料床中沙粒直径、燃烧盘宽度及一端施加辐射热对火焰传播速度的影响,并使用热电偶测量了沙层表面火焰前沿到达时的温度以及沙层表面能够达到的最高温度.结果表明,燃料渗漏在地面的火焰传播速度明显低于液池火灾的火焰传播速度,且改变沙油质量比、燃烧盘的宽度及外界热辐射直接影响火焰传播速度.     

Sizing of safety valves for multi-purpose plants according to ISO 4126-10

Multi-purpose plants are frequently protected with mechanical safety devices like safety valves or bursting disks. Due to many changes of recipes it must be checked regularly whether the safety devices are sufficiently sized. But the sizing procedure of individual safety devices can be very tedious. Therefore energy specific relief areas (effective relief area per kW of energy input) have been determined for approx. 60 typical solvents. They are indicated for reactors with safety devices which have a set pressure of 7 bar (abs) or 11 bar (abs). These values are independent of the size of the reactors for vaporizing systems and arbitrary safety valves. The energy specific relief areas allow the minimum required relief area quickly to recalculate if the energy input of the reactor is known. In addition, the application of solvents in multi-purpose plants can be evaluated from a safety point of few.The energy specific relief areas are calculated based on a relief of two-phase gas/liquid mixtures. The data have been determined with the non-equilibrium HNE-DS method, which takes into account the boiling delay of the liquid in the safety device and the slip between gases and liquids. The method is recommended in the international standard ISO 4126 part 10. In addition, practical advice and possible improvements are outlined. The method leads to significantly smaller relief areas than according to the API 520. For multi-purpose plants with available safety devices this method allows for a considerable expansion of the application range of reactors.     

An experimental study on the burning rate of a continuously released n-heptane spill fire on an open water surface

Spill fires are common during oil product storage and transportation after a loss of containment. Since the burning fuel is moving and the fuel depth is quite shallow, the burning rate in a spill fire is different from that of a pool fire with a static burning zone. Unlike pool fires, which have been studied for decades and have well-established correlations for burning rate, research on spill fires is inadequate. In this paper, continuously released n-heptane spill fire experiments were conducted on open water surfaces with varying fuel discharge rates. The pool diameters were measured, and the spill fire burning rates were estimated based on a dynamic balance between fuel supply and combustion. The burning rates in n-heptane pool fires from the literature were reviewed and compared with the estimated burning rates in spill fires of the same dimension. The spill fire burning rate was found to be close to that in a pool fire during the initial burning phase but lower than that in a bulk burning pool fire and that in a “fuel-level-controlled” pool fire. The distinction between the burning rates of spill fires and pool fires is explained by the heat balance analysis of the fuel layer. A model for the spill fire burning rate was proposed accordingly. The results calculated with the presented model are closer to the measured data than those calculated with pool fire models.     

水面LNG液池扩展模型的分析与对比研究

为了评价在开阔水面上的液化天然气（LNG）火灾和蒸气云爆炸灾害后果，分析了LNG水面扩展动态过程；对比分析了Fay模型、FERC模型和计算流体力学软件FLACS的计算结果，探讨了LNG液池面积随时间的动态变化过程，分析了泄漏量、泄漏速率等参数对LNG液池扩展半径的影响；根据液池扩展模型的计算结果，确定了LNG液池的最大面积，并以此分析了LNG流淌火灾的辐射危害。研究结果表明:对于相同的泄漏条件，3种方法模拟的泄漏LNG水面扩展动态过程相似，一般情况下，FLACS模型，FERC模型和Fay模型所计算的最大液池半径依次增大;由于FERC模型与FLACS软件的模拟结果接近且偏于保守，故此在一般的工程应用时，采用FERC模型即可方便快捷地获得较为准确的结果。