chemengineering

Article
CFD Simulation and Mitigation with Boiling Liquid
Expanding Vapor Explosion (BLEVE) Caused by
Jet Fire
Alon Davidy 1
Israel Military Industries, Ltd., Ramat Hasharon 47100, Israel; alon.davidy@gmail.com; Tel.: +972-03-904-9118




Received: 25 September 2018; Accepted: 20 December 2018; Published: 24 December 2018


Abstract: Different kinds of explosions are driven by the internal energy accumulated in compressed
gas or superheated liquid. A well-known example of such an explosion is the burst of a vessel
with pressure-liquefied substance, known as Boiling Liquid Expanding Vapor Explosion (BLEVE).
Hot BLEVE accident is caused mainly by direct heating (pool fire or jet fire) of the steel casing at
the vapor side of the tank to temperatures in excess of 400 ◦ C. Thermal insulation around the tank
can significantly reduce and retard the excessive heating of the tank casings in a fire. This will
allow fire fighters enough time to reach the accident location and to cool the LPG (Liquid Petroleum
Gas) tank to avoid the BLEVE, to extinguish the fire or to evacuate the people in the vicinity of the
accident. The proposed algorithm addresses several aspects of the BLEVE accident and its mitigation:
Computational Fluid Dynamic (CFD) Simulation of jet fire by using fire dynamics simulator (FDS)
software by using large eddy simulation (LES); calculation of the convective and radiative heat fluxes
by using the impinging jet fire theory; performing thermochemical and heat transfer analysis on
the glass-woven vinyl ester coating of the vessel by using FDS software (version 5); and COMSOL
Multiphysics (version 4.3b) during the heating phase of composite and calculation of the time period
required to evaporate the liquefied propane by using the first and second laws of thermodynamics.

Keywords: BLEVE; heptane jet fire; fire dynamic simulation software (FDS); computational fluid
dynamics (CFD); large eddy simulation (LES); convective heat flux; radiative heat flux; glass-woven
vinyl ester lining; heat transfer and thermochemical analysis; stainless steel AISI 310 alloy

1. Introduction
Accidental releases of pressure-liquefied materials (by bursts of high-pressure tanks, processing
equipment malfunction, etc.) are one of major hazards in process industries, transportation, or
storage of flammable materials [1]. Horrific explosions caused by accidental releases of flammable
materials into the atmosphere occurred in Port Hudson (USA, 1970), Flixborough (UK, 1974), Mexico
City (Mexico, 1984), Ufa (Russia, 1989), Xian (China, 1998), Nechapur (Iran, 2004), Buncefield
(UK, 2005) [2]. Different kinds of explosions are driven by the release of internal energy accumulated
in compressed gas or superheated, liquid [3]. An example of such an explosion is the burst of
a vessel with pressure-liquefied material, known as Boiling Liquid Expanding Vapor Explosion
(BLEVE). It is most important physical explosion, i.e., not involving a chemical reaction [4]. It is
caused by bursting of a pressurized vessel containing liquid above its atmospheric boiling point [5].
The liquid stored in the vessel may be flammable or non-flammable, such as in a hot water boiler
(Steam explosion). If non-flammable, the hazard will be primarily an overpressure event with possible
vessel fragmentation. If a flammable material is ignited, it will usually produce a fireball; a secondary
effect will be a pressure wave due to the explosively rapid vaporization of the liquid. A photo of
BLEVE caused after train derailment in Casselton, North Dakota (2013) is shown in Figure 1 [6].

ChemEngineering 2019, 3, 1; doi:10.3390/chemengineering3010001 www.mdpi.com/journal/chemengineering

ChemEngineering 2018, 2, x FOR PEER REVIEW 2 of 23

usually produce a fireball; a secondary effect will be a pressure wave due to the explosively rapid
ChemEngineering 2019, 3, 1
vaporization of the liquid. A photo of BLEVE caused after train derailment in Casselton, North
2 of 22

Dakota (2013) is shown in Figure 1 [6].

Figure 1. BLEVE accident caused after train derailment in Casselton, North Dakota (2013) [6].
Figure 1. BLEVE accident caused after train derailment in Casselton, North Dakota (2013) [6].
The earliest observations of detonation waves in gases were made by Berthelot & Vielle [7] and
The earliest observations of detonation waves in gases were made by Berthelot & Vielle [7] and
Mallard & La Chatelier [8]. While studying the propagation of flames in tubes they found that under
Mallard & La Chatelier [8]. While studying the propagation of flames in tubes they found that under
certain conditions combustible gas mixtures propagate flames in tubes with speeds far greater than
certain conditions combustible gas mixtures propagate flames in tubes with speeds far greater than
had been measured previously. The propagation reaches enormous velocities, from 1000 to 3500 m/s,
had been measured previously. The propagation reaches enormous velocities, from 1000 to 3500
depending on the gas mixture, or many times the velocity of sound at ordinary temperatures and
m/s, depending on the gas mixture, or many times the velocity of sound at ordinary temperatures
pressures [9]. Detonation waves are sustained by the exothermic chemical reaction that is initiated by
and pressures [9]. Detonation waves are sustained by the exothermic chemical reaction that is
the shock compression. They develop from flames in tubes by coalescence of flame-generated pressure
initiated by the shock compression. They develop from flames in tubes by coalescence of
pulses into shock waves. Their rate of propagation is limited by the rate at which a shock can travel,
flame-generated pressure pulses into shock waves. Their rate of propagation is limited by the rate at
and hence it has been possible to develop the theory of propagation based on hydrodynamics alone
which a shock can travel, and hence it has been possible to develop the theory of propagation based
such extent that detonation velocities may be computed from the physical properties of the explosive
on hydrodynamics alone such extent that detonation velocities may be computed from the physical
medium. There are, however, other aspects of detonation phenomena which are only partially related to
properties of the explosive medium. There are, however, other aspects of detonation phenomena
hydrodynamic processes and thus are out-side the scope of classical theory. These comprise transition
which are only partially related to hydrodynamic processes and thus are out-side the scope of
from flame to detonation (DDT—Deflagration to Detonation Transformation), limits of detonation,
classical theory. These comprise transition from flame to detonation (DDT—Deflagration to
pulsation, and spin of detonation waves. In addition, various problems arise from observations on
Detonation Transformation), limits of detonation, pulsation, and spin of detonation waves. In
the ignition of explosives by weak shocks, and from other incidental observations [9]. A lot of works
addition, various problems arise from observations on the ignition of explosives by weak shocks,
have been written about mathematical modeling of gas detonations and BLEVE. With the dramatic
and from other incidental observations [9]. A lot of works have been written about mathematical
improvements of computing power, computational fluid dynamics (CFD) modeling has become an
modeling of gas detonations and BLEVE. With the dramatic improvements of computing power,
increasingly important platform for reformer modeling and design, combining physical and chemical
computational fluid dynamics (CFD) modeling has become an increasingly important platform for
models with detailed representation of the reformer geometry. When compared with first-principles
reformer modeling and design, combining physical and chemical models with detailed
modeling, CFD is a modeling technique with powerful visualization and computational capabilities to
representation of the reformer geometry. When compared with first-principles modeling, CFD is a
deal with various geometry characteristics, transport equations and boundary conditions.
modeling technique with powerful visualization and computational capabilities to deal with
Volpiani et al. [10] has carried out a Large Eddy Simulations (LES) of deflagrating flames by using
various geometry characteristics, transport equations and boundary conditions.
a dynamic flame wrinkling factor model. His model was validated from a posteriori analysis. Its model
Volpiani et al. [10] has carried out a Large Eddy Simulations (LES) of deflagrating flames by
can capture both laminar and turbulent flame regimes. At early stages of the flame development,
using a dynamic flame wrinkling factor model. His model was validated from a posteriori analysis.
a laminar flame propagates in a flow essentially at rest, corresponding to a unity-wrinkling factor.
ChemEngineering 2019, 3, 1 3 of 22

Transition to turbulence occurs when the flame interacts with the flow motions generated by thermal
expansion and obstacles. They have analyzed three distinct configurations:

(1) The first configuration presents the larger number of obstacles. The interaction between the flame
and obstacles created stronger overpressure.
(2) The second configuration presents the second highest pressure and is characterized by a long
laminar phase, once obstacles are placed far away of the ignition point.
(3) The third configuration, after the flame crossed the first baffle, the flame front is relaminarized
before reaching the central obstacle leading to the lowest observed overpressure [10].

Chemical hydrogen hazards are principally associated with the deflagration and detonation
phenomena that can be produced when a flammable or explosive mixture of hydrogen and an oxidant
is ignited. The minimum energy for the ignition of gaseous hydrogen in air at atmospheric pressure is
only 0.02 mJ [11]. Diéguez et al. [11] showed examples of application of CFD to safety issues such as
hydrogen leakages, hydrogen flames, detonation, and application of the simulation results to evaluate
possible physiological injuries. They performed a CFD for simulating the hydrogen leakage.
Pinhasi et al. [12] applied a numerical-solution method based on the method of characteristics
for the investigation of the BLEVE problem. The described method is applied for the solution of the
governing equations for unsteady one-dimensional of two-phase flow as well as for the airflow. This
method was also employed to calculate the fluid particle path and properties passing the shock wave.
For this purpose, a special numerical scheme was derived. The procedure is applied for the numerical
integration of the characteristic and compatibility equations, using an explicit numerical scheme.
Scholz & Wuersig [13] have performed CFD calculations of the response of a Moss containment
system (Liquified Natural Gas carrier) to pool fire exposure to simulate the transient heating up of the
different components. These tanks are insulated with 220 mm polystyrene foam to prevent the heating
of tank casings and the boiling of the liquefied natural gas (LNG) in case of exposure to external heating
such as pool or jet fires. These tanks are supported by a metal skirt, which transmits the weight of the
tank and the cargo to the lower hull.Three physical effects of the heat transfer have been simulated
during exposure to pool fire. Inside the solid components of the tank system thermal conduction has
been considered. A 2D model has been developed for the analysis of the response of the spherical tank
system under pool fire exposure. The scope of the investigation was to calculate the heat flux into the
Moss tank system. Melting of the insulation was also considered in their evaluations. The insulation
material is also combustible, and in the case of a fire exposure the thermal pyrolysis and degradation
of the insulation would be much greater than melting of the insulation. They applied ANSYS software
(Version 6.5-1).
The structure of this paper is as follows: In Section 1.1, the measures to avoid hot BLEVE are
discussed. In Section 1.2, the computational model is presented briefly. The fire dynamics simulator
(FDS) software is described in detail in Sections 2.1 and 2.2. Section 2.3 describes the calculation
method of the jet fire impinging heat flux. Section 2.4 describes the heat transfer and thermochemical
analyses on the composite coating of the tank. The thermodynamic analysis of the propane is described
in Section 2.5. Section 3 presents the FDS models and COMSOL (version 4.3b) results. The described
work contains new tools and methods. The CFD simulation of the heptane impinging jet was carried
by using the FDS (version 5). The previous works which have been written about thermal protection
of BLEVE accident, have assumed that the external surface of lining is fixed (i.e., there is not recession
of the lining). This work contains the modeling of the recession phenomena. It should be noted that
numerical modeling of thermochemical erosion is still complicated issue. The numerical modeling
requires much more knowledge than the conventional heat transfer approach such as: Arrhenius
coefficients (the data is obtained from experiments), and enthalpy of formation virgin material (the data
is obtained from calorimeter experiments). The coupling of the FDS results with COMSOL can extend
the capabilities of the algorithm and to address other issues concerned with BLEVE accident such as:
transient heating of the steel, structural analysis of the lining and steel tank as a function. The structural
ChemEngineering 2019, 3, 1 4 of 22

analysis of the steel casing is essential to verify that the structural integrity of the tank under the BLEVE
accident. The proposed algorithm is very flexible. It is possible to perform simulation of methane or
other Hydrocarbon fuels jet fires. Other insulations (rubber, PMMA etc.) can be tested. It is probably
the first time that FDS software has been applied to simulate jet fire impinging on composite lining
and causing thermochemical ablation on the tank lining.

1.1. Measures to Avoid Hot BLEVE

Hot BLEVE is caused mainly by heating of the steel casing at the vapor side of the tank to
temperatures in excess of 400 ◦ C. Thermal coating around the tank can considerably attenuate the
excessive heating of the tank casing in case of exposure to fire. This will allow fire fighters enough
time to reach the accident location and to cool the LPG (Liquid Petroleum Gas) tank to prevent the
BLEVE, to extinguish the fire or to evacuate the people in the vicinity of the tank. Important is that the
insulation (coating) attenuates the heating of the tank casing long enough [14]. At present, there are
three different types of coatings [15]:

• Intumescent fire protection coatings are based on epoxy resin. When they are exposed to fire,
at temperature of around 150–200 ◦ C, they react independently by forming foamed protective
insulating layer. The foamed coating forms a thermal barrier layer, which reduces the heat flux
input to the tank walls. Most of these coatings contain three main ingredients: An acid source,
a carbon source, and a blowing agent. The formulation of the coating must be optimized in terms
of physical and chemical properties to form an effective protective char layer [16–18].
• Sublimation fire protection coatings release gases under thermal loads. In the case of exposure
to fire, large quantities of thermal energy are needed to transform the coating directly from the
solid to the gaseous aggregate state without liquefying beforehand (phase transformation). This
transformation is used to build up a layer which absorbs and reduces the heat flow into the
tank walls.
• Ablative coatings are based on endothermic processes. The heat transfer is delayed by the
extraction of chemically bonded water from the fire protection coating.

In this work a thermochemical ablation model has been developed and carried out on glass-woven
vinyl ester coating. Glass Fiber Reinforced Plastic (GFRP) is an advanced innovation for the
transportation of flammable materials. Its light weight reduces the transportation costs [19]. A key
limitation of composite materials is their rapid loss of structural stability in fire. At temperatures
between 100 ◦ C and 200 ◦ C, the matrices consisting of thermosetting polymers such as epoxy or
polyester resin start to soften and weaken. This leads to a decrease in the structural integrity of the
composite material [20].

1.2. Algorithm Description, Assumptions and Scope

In this work, it is assumed that heptane jet fire is impinging on the external boundary of the tank
composite lining. The distance between the jet source and the tank is 3 m. The exposed diameter of
the lining is 0.8 m. The lining is made of glass-woven vinyl ester. It is assumed that the thickness of
lining is 0.02 m (20 mm). The cylindrical tank casing is made of steel AISI 310. This steel type has
higher ultimate tensile strength. It has also excellent high temperature properties with good ductility
and weldability. It is designed for high temperature service. It resists oxidation in continuous service
at temperatures up to 1150 ◦ C and has good Corrosion resistance. The tank initially contains liquid
propane. The initial temperature of the propane inside the tank is 55.6 ◦ C. The initial pressure inside
the tank is 1930 kPa. The thickness of the steel tube is 12.2 mm (w). The internal diameter of the tank is
1.2 m (Di ). The tank contains 500 kg of liquefied propane (the vapor quality of the propane is x = 0).
The left side of the tank is exposed to jet fire heating. At the liquid side (bottom) of the tank the heat of
the fire will be transferred via the steel casing wall to the liquid and cause evaporation of the liquid
propane (see Figure 2).
ChemEngineering 2018, 2, x FOR PEER REVIEW 5 of 23

of the fire will be transferred via the steel casing wall to the liquid and cause evaporation of the
ChemEngineering
liquid propane2018,
(see2,Figure
x FOR PEER
2). REVIEW 5 of 23

ChemEngineering 2019, 3, 1 5 of 22
of the fire will be transferred via the steel casing wall to the liquid and cause evaporation of the
liquid propane (see Figure 2).

Figure 2. Sketch of Propane tank exposed to Heptane jet fire.

As the liquid propane Figureevaporates

2. Sketch ofcompletely
Propane tank (vapor
exposed quality is x = 1)
to Heptane jetthe
fire.temperature and vapor
pressure will be increased Figure 2. Sketch
(for propane of Propane
19 bar at tank55.6 exposed
°C). At to Heptane
the phase jet of
fire.
superheated vapor, the
pressure
As theofliquid
the propane
propane inside
evaporatesthe completely
tank will increase(vapor quality with the
is x =temperature.
1) the temperature However, if the
and vapor
As the liquid propane evaporates completely (vapor quality is x = 1) the temperature and vapor
temperature
pressure willofbethe steel on (for
increased the gas side of19the
propane bartank exceeds
at 55.6 ◦ C). 450–550 °C theof
At the phase steel will lose its
superheated integrity
vapor, the
pressure will be increased (for propane 19 bar at 55.6 °C). At the phase of superheated vapor, the
and the tank
pressure of thewill rupture
propane [14].the
inside This will
tank also
will occur with
increase if a tank is equipped with
the temperature. a Pressure
However, if the Relief Valve
temperature
pressure of the propane inside the tank will increase with the temperature. However, if the
(PRV).
of Theon
the steel setthepoint of the
gas side of thePRVtankof exceeds
the tank450–550
vehicles ◦ Cisthegenerally
steel willsetloseatits1.93 MPa and
integrity for propane
the tank
temperature of the steel on the gas side of the tank exceeds 450–550 °C the steel will lose its integrity
mixtures.
will ruptureThe material
[14]. stress
This will alsocan be calculated
occur if a tank iswithequipped[14]: with a Pressure Relief Valve (PRV). The set
and the tank will rupture [14]. This will also occur if a tank is equipped with a Pressure Relief Valve
point of the PRV of the tank vehicles is generally set at 1.93 MPa for propane mixtures. The material
(PRV). The set point of the PRV of the tank vehicles pD
stress can be calculated with [14]:   i i is generally set at 1.93 MPa for propane (1)
mixtures. The material stress can be calculated with p2i D[14]:
w i
σ= (1)
2w
where  is the stress in (MPa), p is the internal p D
i i pressure of the tank in (MPa), Di is the
where σ is the stress in (MPa), pi is thei internal  pressure of the tank in (MPa), Di is the diameter(1) of
diameter of the tank and w is the of the tank wall. 2 w results in a material stress
This of 189 MPa. The
the tank and w is the of the tank wall. This results in a material stress of 189 MPa. The propane tank
propanebetank should be able this
to withstand
where  able
should is the stress in (MPa),
to withstand pi is this
pressure. the
If pressure.
the materialIf
internal
the material
pressure
stress stressstrength
of the the
exceeds exceeds of
tank in (MPa),
thethe
strength
Dtank
i is the
of
the
the tank
tank will the
fail. tank
Intankwill fail.
Figure 3 the In Figure 3ofthe strengths ofareseveral steels are shown as a function of
diameter
temperature
of the
[21].
and
The
w strengths
material
is the of the several
stress
tank wall.
exerted on
steelsThis
the by
shown
results
the
inas a function
a material
pressurized
of temperature
stress
propane
of 189 MPa. [21].
is shown
The
for
The material
propane tankstress
shouldexerted
be ableontothe by the pressurized
withstand this pressure. propane
If theismaterial
shown for stress comparison.
exceeds the strength of
comparison.
the tank the tank will fail. In Figure 3 the strengths of several steels are shown as a function of
temperature [21]. The material stress exerted on the by the pressurized propane is shown for
comparison.

Figure 3.
Figure 3. Plot
Plotofof
ultimate tensile
ultimate strengths
tensile of several
strengths stainless
of several steelssteels
stainless compared to thetomaterial
compared stress
the material
exertedexerted
stress by thebypressurized Propane
the pressurized (horizontal
Propane line) shown
(horizontal as a function
line) shown of temperature
as a function (The
of temperature
thermomechanical data were taken from Reference [21]).
(The thermomechanical data were taken from Reference [21]).
Figure 3. Plot of ultimate tensile strengths of several stainless steels compared to the material stress
exerted by the pressurized Propane (horizontal line) shown as a function of temperature (The
thermomechanical data were taken from Reference [21]).
ChemEngineering 2019, 3, 1 6 of 22

Figure 3 shows that the Ultimate Tensile Strength of the Stainless Steels decreases with the
temperature. The AISI 310 has the maximal tensile strength.
In this work a CFD simulation and mitigation study BLEVE has been performed. The proposed
computational work is composed of five phases:

(a) Simulation of jet fire by using FDS software

(b) Calculation of the convective and radiative heat fluxes by using the impinging jet fire theory
(c) Heat transfer analysis on the composite coating of the vessel by using FDS software (version 5)
and COMSOL Multiphysics (version 4.3b) during the heating phase of composite
(d) Calculation of the time period required to evaporate the liquefied propane
(e) Structural analysis on the tank steel casings

It is important to be able to design thermal coatings for gas pipes and storage tanks that can
withstand such catastrophic events. Explosive pressure and structural damage can be estimated with
analytical and numerical tools. By using the described computational tools, it is possible to postpone
or even to prevent this accident. Important issue which should be considered is that the insulation
(coating) attenuates the heating of the tank wall long enough.

2. Materials and Methods

2.1. Fire Dynamic Simulation Modeling of the Jet Fire

In flames formed by fuel jets in the atmosphere the combustion process is coupled with the process
of mixing. An analysis of such flames requires consideration of the mixing process only. Flames of fuel
jets range from slow combustion exemplified by candle flames where the stream filaments are laminar
and mixing occurs by molecular diffusion only to rapid fuel combustion in industrial furnaces and
engines of various types, where mixing is accomplished by turbulent eddy motion [9]. In this work
CFD study of Jet Flames has been performed by FDS software. The FDS has been developed at the
National Institutes of Standards and Technology (NIST), e.g., McGrattan et al. [22,23]. This software
calculates simultaneously the velocity, pressure, temperature, density, and chemical composition
(mole fraction) within each numerical grid cell at each discrete time step. The FDS software also has a
visual post-processing image simulation program named "smoke-view".
This software performs conjugated heat transfer analysis on the solid walls i.e., it solves
simultaneously the temperature of the gaseous products and the temperatures at the external surfaces
of the solid walls.
It also calculates the temperature, heat flux, and mass loss rate of the enclosed solid surfaces
(thermochemical pyrolysis). We shall see later that the mass loss rate is controlled by Arrhenius
equations. The latter is used in the case where the fire heat release rate is unknown. The major
components of this software are:
Radiation Transport—The FDS software solves the Radiation Transport Equation (RTE) for a
non-scattering gray gas. In a limited number of cases, a wide band model can be used in place of the
gray gas model. The RTE is solved by using the Finite Volume Method (FVM) [23].
Combustion Model—FDS uses a mixture fraction combustion model. The mixture fraction is a
conserved scalar quantity that is defined as the fraction of gas at a given point in the flow field that
originated as fuel. This software assumes that combustion is mixing controlled, and that the reaction
of fuel and oxygen is infinitely fast. The mass fractions of all the major reactants and products can be
derived from the mixture fraction by means of “state relations,” empirical expressions arrived at by a
combination of simplified analysis and measurement [24].
Hydrodynamic Model—FDS code is formulated based on CFD of fire-driven fluid flow (Natural
convection). The numerical solution can be performed by using either a Direct Numerical Simulation
(DNS) method or LES. The latter is relatively low Reynolds numbers and is not severely limited in
ChemEngineering 2019, 3, 1 7 of 22

grid size and time step as the DNS method. This software cannot perform CFD simulations on high
Mach number applications.
In addition to the classical conservation equations solved by FDS software (including mass species
momentum and energy), it also solves thermodynamics-based state equation for a perfect gas.

2.2. Governing Equations of FDS Software

The following sections introduce the conservation equations for mass, momentum, and energy
for a Newtonian fluid.

2.2.1. Mass and Species Transport

Mass conservation is expressed terms of the density, ρ [22]:

∂ρ .
+ ∇ · ρu = m000 b (2)
∂t

where ρ denotes the density (kg/m3 ), and u denotes the velocity field (m/s). This equation is expressed
terms of each gaseous species, Yα :

∂ . .
(ρYα ) + ∇ · ρYα u = ∇ · ρDα ∇Yα + m000 α + m000 b,α (3)
∂t

Dα denotes the diffusion coefficient of α component of the mixture (m2 /s).

2.2.2. Momentum Transport

The momentum equation in conservative form is written [22]:

∂
(ρu) + ∇ · ρuu + ∇ p = ρg + fb + ∇ · τij (4)
∂t
Here fb denotes the force term (Pa/m). The stress tensor, τij (Pa), is defined [22]:
  ( !
2 1i= j 1 ∂ui ∂u j
τij = µ 2Sij − δij (∇ · u) ; δij = ; Sij = + i, j = 1, 2, 3 (5)
3 0 i 6= j 2 ∂x j ∂xi

The term Sij denotes the symmetric rate-of-strain tensor. The symbol µ denotes the dynamic
viscosity of the fluid. In this work the overall computation is carried out by using LES, in which the
large-scale eddies are calculated directly, and the subgrid-scale dissipative processes are modeled.
The numerical algorithm is designed so that LES becomes DNS as the grid is refined. It should be
noted that in most fire simulation applications, FDS uses LES. For example, in simulating the flow of
smoke through a large, multi-room enclosure, it is not possible to obtain the combustion and transport
processes directly.

2.2.3. LES
One of the most dominant features of any CFD model is its treatment of turbulence. Among the
three main techniques of simulating turbulence i.e., Reynolds Averaged Navier Stokes (RANS), LES,
and DNS, FDS uses only LES and DNS. LES is a technique used to model the dissipative processes
(viscosity, thermal conductivity, material diffusivity) that occur at length scales smaller than those that
are explicitly resolved on the numerical grid. This means that these parameters in the equations above
cannot be used directly in most practical simulations. They must be replaced by surrogate expressions
that “model” their impact on the approximate form of the governing equations. This sub section
contains a simple explanation of how these terms are modeled in FDS. There is a small term in the
ChemEngineering 2019, 3, 1 8 of 22

energy equation known as the dissipation rate, ε, (Pa/s) the rate at which kinetic energy is converted
to thermal energy by viscosity [22]:
"  
∂u 2
 2
∂w 2
   
2 2 ∂v
ε = τij · ∇u = µ 2Sij · Sij − (∇ · u) = µ 2 +2 +2 +
3 ∂x ∂y ∂z
 # (6)
∂u 2 ∂w ∂v 2 ∂u ∂w 2 2 ∂u ∂v ∂w 2
      
∂v
+ + + + + − + +
∂x ∂y ∂y ∂z ∂z ∂x 3 ∂x ∂y ∂z

The viscosity µ is modeled

 1
¯ ¯ 2 2
µ LES = ρ(Cs ∆)2 2Sij : Sij − (∇ · u)2 (7)
3

where Cs denotes an empirical constant and ∆ is a length on the order of the size of a grid cell.
The bar above the various quantities denotes that these are the resolved values, meaning that they are
computed from the numerical solution sampled on a coarse grid (relative to DNS). The other diffusive
parameters, the thermal conductivity and material diffusivity, are related to the turbulent viscosity
by [22]:
µ LES c p µ
k LES = ; (ρD)t,LES = LES (8)
Prt Sct
The turbulent Prandtl number Prt (defined as the ratio of momentum diffusivity to thermal
diffusivity) and the turbulent Schmidt number Sct (defined as the ratio of momentum diffusivity
to mass diffusivity) are assumed to be constant for a given scenario. The model for the viscosity,
µ LES , serves two roles: first, it provides a stabilizing effect in the numerical algorithm, damping out
numerical instabilities as they arise in the flow field, especially where vorticity is generated. Second,
it has the appropriate mathematical form to describe the dissipation of kinetic energy from the flow.

2.2.4. Energy Transport

The energy conservation equation is written in terms of the sensible enthalpy, hs (J/kg) [22]:

∂ Dp . . .
(ρhs ) + ∇ · (ρhs u) = + q000 − q000
b − ∇ · q00 + ε (9)
∂t Dt
The sensible enthalpy (J/kg) is a function of the temperature [22]:

hs = ∑ Yα hs,α (10)
α

where sensible heat of each component in the mixture is calculated by [22]:

ZT
c p,α T 0 dT 0


hs,α ( T ) = (11)
T0

. .
here q000 denotes the volumetric heat release rate produced from a heptane oxidation (W/m3 ), q000 b
.
denotes the energy transferred to the evaporating heptane liquid (W/m3 ) and q00 represents the
conductive and radiative heat fluxes (W/m2 ) [22]:
.
q00 = k ∇ T − ∑ hs,α ρDα ∇Yα + qr
00
(12)
α
ChemEngineering 2019, 3, 1 9 of 22

2.2.5. Equation of State

The pressure is computed by applying the ideal gas equation of state (Frequently the jet fire occurs
at atmospheric pressure which is much less than the critical pressure of the gaseous mixture):

ρRT
p= (13)
W

where p denotes the pressure in (Pa), T is the temperature in (K), R is the gas constant and W is the
molar mass of the gaseous mixture in (J/mole).

2.3. Calculation Method of the Jet Fire Impinging Heat Flux

Impinging flame jet has been extensively investigated due to their importance in a wide range of
applications [25]. Two heat transfer mechanisms have been considered in this work: forced convection
and radiation. The total heat flux was considered in FDS transient thermochemical ablation and
conduction calculations [24].

2.3.1. Convective Heat Transfer

Heat is transferred between the jet impingent turbulent flow and the solid surface [26].

2.3.2. Radiative Heat Transfer

In the jet fire, most if the heptane burns in turbulent diffusion flame as fuel and air mix together.
The flame is highly luminous, and soot particles are formed during the combustion process [26].
The radiative heat transfer within the heptane jet fire is composed of the thermal radiation from high
temperature burned gases and the radiation from soot particles.

2.3.3. Calculation of the Maximal Heat Flux Value

The maximal total heat flux (composed of radiative and convective) is calculated by using the
following Equation [25]:
 0.3
q00 max,conv+rad = 10 q f (14)

It is based on a thermal input of q f = 60 − 1600 kW. The values of the thermal input were taken
from FDS calculation results.

2.4. Heat Transfer and Thermochemical Analyses on the Composite Coating of the Tank
The proposed approach can be divided into two simulation parts.
In first part, the analysis of composite has been carried out under jet fire thermal loads by using
the FDS (see Section 2.5). The FDS code is used to solve for the recession of the surfaces as a function
of time. The temperature and heat flux of the interior structural surface profiles are used and applied
to subsequent simulation part. In the second part, COMSOL heat transfer module (version 4.3b) has
been applied to calculate the temperature profiles through the thickness of the composite by importing
the instantaneous surface temperatures and the recession obtained from the FDS model. In this work,
we have assumed that the composite decomposed to a pyrolysis gas plus char. The reaction scheme is
shown [27]:
P → αG + (1 − α)C (15)

During the pyrolysis reaction, mass of the polymer is consumed and produces a fraction, α, of gas
and the remaining char. The first order reaction rate for the composite is:

dr p
= −k0 r p (16)
dt
ChemEngineering 2019, 3, 1 10 of 22

where r p denotes the instantaneous thickness of the polymer in (m), t is the time in (s) and k0 denotes
the rate constant for pyrolysis reaction in (1/s). The rate constant in for the pyrolysis reaction, k0 , is a
function of temperature and is better described by the Arrhenius relationship [27]:
 
E
k0 = A0 · exp − A0 (17)
R·T

where A0 is the pre-exponential factor of pyrolysis reaction (1/s). E A0 is the activation energy of
pyrolysis reaction (kJ/kmol), R is universal gas constant ((J/mole)/K) and T is the temperature in (K).
The pre-exponential factor and activation energy can be found by thermogravimetric analysis. A 1D
heat conduction equation for the composite temperature Ts ( x, t) is applied in the direction x pointing
into solid (the point x = 0 represents the surface) [22]:
 
∂Ts ∂ ∂Ts .
ρs c p,s = λs + qs (18)
∂t ∂x ∂t
.
The source term, qs , consists of chemical reactions (production or loss) rate given by the pyrolysis
models for different types of solid and radiative absorption and emission in depth [22].
. . .
qs = qs,c + qr (19)

The chemical source term of the heat conduction equation consists of the heat of the reaction [22]:
.
qs,c = −ρs k0 ∆Hr (20)

∆Hr is the heat of reaction. The thermal properties (thermal conductivity, heat capacity and
density) and the rate constants of the composite used in the calculation of FDS and COMSOL are
shown in Table 1 [28].

Table 1. Thermophysical and thermochemical properties of E-Glass vinyl ester used in calculations [28].

Material Property Value

rho 1140 kg/m3
Cp 760 J/(kg·◦ C)
λ 0.43 w/(m·◦ C)
E 2.13 × 105 kJ/kmol
A 5.59 × 1013 1/s
n 1
∆H 378,800 J/kg

As was stated before the receding surface is exposed to radiative and convective heat flux
(see Equation (14)). FDS software calculates the convective heat flux by using the following Equation [22]:
.

qc = h Tg − Tw (21)

According to the LES approach, the convective heat flux to the surface is calculated by combining
the natural and forced convection empirical correlations [22]:
 
 1 λg 4 1
 
0.037Re Pr w/ m2 · K

h = max C Tg − Tw ,
  3 5 3 (22)
L

C denotes the coefficient for natural convection. It is equal to 1.52 for horizontal surface and equal
to 1.31 for vertical surface, L denotes the characteristic length related to the size of the obstruction, λ g
denotes the thermal conductivity of the gaseous products mixture. Re is the Reynolds number and Pr
is the Prandtl number. They are based on the gas flowing past the obstruction correlation.
ChemEngineering 2019, 3, 1 11 of 22

The recession output coordinate, which was obtained earlier by FDS software, has been exported
to COMSOL finite element model. In this model, the composite external surface recession is specified
with a moving mesh boundary condition enabled by using pre-packaged feature described as the
Arbitrary Lagrangian-Eulerian (ALE) module. This module permits moving boundaries without the
need for the mesh movement to follow the material [29].

Mechanical Properties of the Composite Degradation

The temperature dependence of mechanical properties can be expressed as the hyperbolic tangent
function temperature. For Young’s modulus, the degradation law is written as [30]:

E1 + E2 E − E2
E( T ) = + 1 tan h[Φ( T − Tk )] (23)
2 2
here E1 is the Young’s modulus at initial temperature in (MPa), E2 is the residual modulus in (MPa),
Φ is 0.026 1/K, and Tk = 88 ◦ C. The initial structural properties of the coatings are influenced from the
environmental conditions (ageing). It should be noted that moisture diffusion into the resin affects the
thermomechanical properties of the GFRP. According to Huang et al. [31] moisture diffused inside
the epoxy resins will decrease the glass transition temperature, the elastic modulus, and the static
strengths, as well as fatigue performance. Vinyl Ester resin exhibits better performance under exposure
to moist environment. According to Mouritz & Mathyz [32] vinyl ester resin is applied frequently
in marine composites because of its higher heat-distortion temperature, better water resistance and
slightly higher mechanical properties.

2.5. Calculation of the Time Required to Evaporate the Propane and the Calculation of the Explosion Energy
Calculation of the energy released by superheated liquid was investigated by many investigators.
Among them is Giesbrecht et al. [33]. They defined the explosion energy as the work done by the fluid
on surrounding air as it expands isentropic (reversible adiabatic process). The changes in internal
energies are calculated from thermodynamic properties for the fluid. During this process, the system
expands isentropically (no heat transfer, no irreversibilities) from state 2 (the intermediate state) to
state 3, with p3 equal to the ambient pressure p0. In this work, it is assumed that the liquid has been
heated by the jet fire. After expansion, the final internal energy is U3 .The work which the system can
perform is the difference between its intermediate and the final internal energies. The specific work
done by expanding fluid is calculated by [5,34]:

Eex,wo = U2 − U3 (24)

where: Eex,wo is the work performed in expansion from state 2 to 3 in (kJ), U2 is the internal energy of
the fluid after the isobaric heating in (kJ) and U3 is the internal energy of the fluid at the final state in
(J) (At the final state, the propane expands into the atmosphere, it is in equilibrium with surroundings,
thus the temperature and the pressure of the propane is equal to the environment temperature and
pressure). According to the first law of thermodynamics, the heat interaction is equal to the change in
the internal energy [35]:
q00 max,conv+rad A∆t = U2 − U1 = m(u2 − u1 ) (25)

where: qmax 00 is the heat flux which is absorbed in the steel in (kW/m2 ) (This term is calculated in the
following section). In this work the maximal heat flux has been taken to estimate the minimal time
period. A is the cross-section area of the impinging jet in (m2 ). ∆t is the time period which is the steel
exposed to the impinging jet (after the composite lining eroded completely) in (s). U1 is the internal
energy of the fluid at the initial state in (J). It is assumed that the propane at the initial state is liquid.
The propane undergoes isobaric heating from state 1 to state 2.
  conv rad At  U 2  U1  m  u2  u1 
qmax, (25)

where: qmax is the heat flux which is absorbed in the steel in (kW/m²) (This term is calculated in
the following section). In this work the maximal heat flux has been taken to estimate the minimal
time period.
ChemEngineering A 3,
2019, t is the time period which is12 of 22
is 1the cross-section area of the impinging jet in (m²).
the steel exposed to the impinging jet (after the composite lining eroded completely) in (s). U1 is the
internal energy of the fluid at the initial state in (J). It is assumed that the propane at the initial state is
3. Results
liquid. The propane undergoes isobaric heating from state 1 to state 2.
This section contains five parts. Section 3.1 contains the thermal results of FDS software for
3. Results
heptane jet flame. Section 3.2 contains the calculation results of the heat flux. In Section 3.3 the FDS
This section
and COMSOL contains results
multiphysics five parts.
forSection 3.1 contains
the coating and thethepropane
thermal tank
results
areofpresented.
FDS software for 3.4
Section
heptane jet flame. Section 3.2 contains the calculation results of the heat flux. In Section 3.3
presents the calculation of the time period required to evaporate the liquefied propane. Section 3.5 the FDS
and COMSOL multiphysics results for the coating and the propane tank are presented. Section 3.4
presents the structural analysis of the tank steel casing after the complete erosion of the coating.
presents the calculation of the time period required to evaporate the liquefied propane. Section 3.5
presents
3.1. FDS the structural
Software analysis
Results for of Model
Jet Fire the tank steel casing after the complete erosion of the coating.

3.1.
FDSFDS Software Results
simulation for Jet Fire
can provide much Model
detailed information on the heptane jet fire. This information
includes FDS
the simulation
local andcantransient
provide much detailed gas
gas velocity, temperature,
information speciesjetconcentration,
on the heptane solid wall
fire. This information
includes the
temperature, fuellocal and rate,
burning transient gas velocity,
radiative gasconvective
heat flux, temperature, species
heat concentration,
flux and Heat Release solid wall
Rate (HRR).
temperature, fuel
The temperature fieldburning rate,
at t = 78.6 s isradiative
shown inheat flux, 4.
Figure convective heat flux and Heat Release Rate
(HRR). The temperature field at t = 78.6 s is shown in Figure 4.

Figure
Figure 4. 4. Temperaturefield
Temperature field ((°C)
◦ C) of
ofthe
theHeptane
Heptanejetjet
firefire
at tat= 78.6 s. s.
t = 78.6

The The maximal

maximal temperatureatatt t==78.6
temperature 78.6 ss approaches
approaches toto1000
1000°C.◦ C.
The velocity
The fieldfield
velocity at t =at79.9
t = s79.9
is s is
shown
shown in in Figure
Figure 5. 5.
ChemEngineering 2018, 2, x FOR PEER REVIEW 13 of 23

Figure
Figure 5. 5. Velocityfield
Velocity field(m/s)
(m/s) of
ofthe
theHeptane
Heptanejetjet
firefire
at tat= t79.9 s. s.
= 79.9

The soot concentration field at t = 5.8 s is shown in Figure 6.

ChemEngineering 2019, 3, 1 13 of 22
Figure 5. Velocity field (m/s) of the Heptane jet fire at t = 79.9 s.

The
The soot
soot concentration
concentration field at tt == 5.8
field at 5.8 ss is
is shown
shown in
in Figure
Figure 6.
6.

Figure 6. Soot concentration field created by Heptane jet fire at t = 5.8 s.

Figure 6. Soot concentration field created by Heptane jet fire at t = 5.8 s.
To check the FDS results which have been obtained for heptane jet fire (see Figures 4 and 5 in
To check
the paper), the FDS FDS
Additional results which
model for have been obtained
simulating propane for
theheptane jet fire
jet fire was (see Figures
developed. 4 and 5 in
The numerical
the paper), Additional FDS model for simulating propane the jet fire was developed.
results of this model were validated against experimental results taken from the report of Wighus & The numerical
ChemEngineering 2018, 2, xwere
results of this
Drangsholt [36]. FigureFOR
model PEER
7 shows REVIEW
validated against experimental
the temperature results jet
field of propane taken
fire from
at t =the
3.2 report
s. 14 of&
of Wighus 23

Drangsholt [36]. Figure 7 shows the temperature field of propane jet fire at t = 3.2 s.

Figure 7. Temperature field (◦ C) of the Propane jet fire at t = 3.2 s.

Figure 7. Temperature field (°C) of the Propane jet fire at t = 3.2 s.
As can be from Figure 7, the maximal temperature obtained for propane jet fire is 1015 ◦ C.
As can be
The maximal from Figure
temperature 7, the maximal
reported in abovetemperature obtained
report (see Table 3.1 infor
thepropane
mentionedjet fire is 1015
report) °C. The
is 1150 ◦ C.
maximal temperature reported in above report (see Table 3.1 in the mentioned report)
Thus, the relative error between the measured and calculated temperature is 11.7%. FDS software has is 1150 °C.
Thus,validated
been the relative error
against between
several the measured
experimental and calculated
test cases. temperature
They are described is 11.7%. FDS software
in [37].
has been validated against several experimental test cases. They are described in [37].

3.2. Heat Flux Calculation Results

The combined heat flux (radiative and convective) as a function of time is presented in Figure 8.
 Figure 7. Temperature field (°C) of the Propane jet fire at t = 3.2 s.

As can be from Figure 7, the maximal temperature obtained for propane jet fire is 1015 °C. The
maximal temperature reported in above report (see Table 3.1 in the mentioned report) is 1150 °C.
Thus, the relative
ChemEngineering 2019, 3, 1error between the measured and calculated temperature is 11.7%. FDS software
14 of 22
has been validated against several experimental test cases. They are described in [37].

3.2.3.2. Heat
Heat Flux
Flux CalculationResults
Calculation Results
The
The combined
combined heatflux
heat flux(radiative
(radiativeand
and convective)
convective) as
as aa function
functionofoftime
timeisispresented
presentedinin
Figure 8. 8.
Figure

ChemEngineering 2018, 2, x FOR PEER REVIEW 15 of 23

Figure8.8.Combined
Figure Combined radiative
radiative and
and convective
convectiveheat
heatflux.
flux.

3.3. FDS Software Results for the Composite and Steel Casing Heating
AsAscan
canbebeseen
seenfrom
from Figure 8, the
Figure 8, themaximal
maximal value
value of the
of the combined
combined heat heat flux approaches
flux approaches to 170 to
kW/mat2 tat= t100
170kW/m²
3.3.1. Validations.of the Model
= 100 s.

3.3. FDS An ablation

Software analysis
Results has been and
for the Composite performed on the
Steel Casing composite coating of the vessel. The
Heating
thermosphysical and thermochemical of woven glass/vinyl ester laminate were taken from Table 1.
3.3.1.Figure
Validation of the
9 shows theModel
surface temperatures of the composite at different values of heat fluxes: 10
kW/m², 25 kW/m²
An ablation and 50
analysis haskW/m².
been performed on the composite coating of the vessel. The thermosphysical and therm

Figure 9. The surface temperatures of composite as a function of time for different heat fluxes.
Figure 9. The surface temperatures of composite as a function of time for different heat fluxes.

Figure 9 indicates that the surface temperature of the composite increases with heat flux.
Similar results for the surface temperature of the composite have been reported in Reference [28].
Figure 10 shows the decrease in composite thickness as a function of time.
ChemEngineering 2019, 3, 1 15 of 22
Figure 9. The surface temperatures of composite as a function of time for different heat fluxes.

Figure
Figure9 indicates that
9 indicates thethe
that surface temperature
surface temperatureofofthe
thecomposite
compositeincreases
increaseswith
withheat
heatflux.
flux. Similar
results for the surface temperature of the composite have been reported in Reference [28].
Similar results for the surface temperature of the composite have been reported in Reference [28]. Figure 10
Figure
shows the10decrease
shows theindecrease
compositein composite
thicknessthickness as a function
as a function of time.
of time.

ChemEngineering 2018, 2, x FOR PEER REVIEW 16 of 23

Figure10.
Figure 10Predicted
. Predictedthickness ofthe
thickness of thecomposite
compositeas as a function
a function of time
of time for different
for different heat fluxes.
heat fluxes.
From
From Figure
Figure 10 10 it can
it can bebe seen
seen thatthe
that thecomposite
compositeeroded
eroded completely
completely after
after time
timeperiod
periodofof800
800s s
for heat flux of 50 kW/m². For lower values of heat fluxes the erosion of the composite thickness is
for heat flux of 50 kW/m2 . For lower values of heat fluxes the erosion of the composite thickness is
much smaller.
much smaller.
3.3.2. Model Predictions for the Glass-Woven Vinyl Ester Coating
3.3.2. Model Predictions for the Glass-Woven Vinyl Ester Coating
Figure 11 shows the predicted thickness of the glass-woven vinyl ester as a function of time
Figure 11 shows the predicted thickness of the glass-woven vinyl ester as a function of time

Figure 11. Predicted thickness of the composite as a function of time for heat flux of 170 kW/m2 .
Figure 11. Predicted thickness of the composite as a function of time for heat flux of 170 kW/m².
From Figure 11 it can be seen that the composite eroded completely after time period of 338 s
From Figure 11 it can be seen that the composite eroded completely after time period of 338 s
(about 6 min) for heat flux of 170 kW/m2 . This time period will allow to extinguish the fire or to
(about 6 min) for heat flux of 170 kW/m². This time period will allow to extinguish the fire or to
evacuate the people in the vicinity of the accident. It should be noted that during this time period the
evacuate the people in the vicinity of the accident. It should be noted that during this time period
the temperature at the interface of propane/steel remains at 55.57 °C (The propane has not been
heated). Since the propane is in liquid state, its convective coefficient is much larger. As was
mentioned earlier the recession data of the lining have exported to the COMSOL Multiphysics
software (version 4.3b). The heat transfer analysis has been performed for coarse and fine grids.
Figure 12 shows the calculated temperature on the steel/lining interface for coarse and fine mesh.
ChemEngineering 2019, 3, 1 16 of 22

temperature at the interface of propane/steel remains at 55.57 ◦ C (The propane has not been heated).
Since the propane is in liquid state, its convective coefficient is much larger. As was mentioned earlier
the recession data of the lining have exported to the COMSOL Multiphysics software (version 4.3b).
The heat transfer
ChemEngineering 2018,analysis has been
2, x FOR PEER performed for coarse and fine grids. Figure 12 shows the calculated
REVIEW 17 of 23
temperature on the steel/lining interface for coarse and fine mesh.
ChemEngineering 2018, 2, x FOR PEER REVIEW 17 of 23

Figure 12. Calculated temperature at the steel/lining interface obtained for coarse and fine mesh.
Figure 12. Calculated temperature at the steel/lining interface obtained for coarse and fine mesh.
Figure 12. Calculated temperature at the steel/lining interface obtained for coarse and fine mesh.
Figure 12 12
Figure shows
showsthat
thatthe
thetemperatures obtainedatatthe
temperatures obtained the steel/glass-woven
steel/glass-woven vinylvinyl
ester ester interface
interface for for
Figure
coarse and 12
fineshows
mesh that
are the
almosttemperatures
identical. obtained
Figure 13 at the
shows steel/glass-woven
the calculated vinyl ester
temperatures
coarse and fine mesh are almost identical. Figure 13 shows the calculated temperatures at the atinterface
the for
interface
coarse and fine
interface
of glass-woven meshester/steel
are vinyl
of glass-woven
vinyl almost identical.
ester/steel
and and
at the Figure 13 shows
at the propane/steel
propane/steel the calculated temperatures at the
interface.
interface.
interface of glass-woven vinyl ester/steel and at the propane/steel interface.

Figure 13. Calculated temperatures at the interface of glass-woven vinyl ester/steel and at the
Figure 13. Calculated
Propane/Steel interface. temperatures at the interface of glass-woven vinyl ester/steel and at the
Propane/Steel interface.

Figure 13. Calculated temperatures at the interface of glass-woven vinyl ester/steel and at the
Propane/Steel interface.
ChemEngineering 2019, 3, 1 17 of 22
ChemEngineering 2018, 2, x FOR PEER REVIEW 18 of 23

◦ C.
Ascan
As canbebeseen
seen
thethe temperature
temperature riserise at interface
at the the interface of glass-woven
of glass-woven vinyl ester/steel
vinyl ester/steel is 10
is about about
10 °C. According
According to Figureto3 Figure 3 the in
the decrease decrease in theofstrength
the strength the steel of
canthe
be steel can beFigure
neglected. neglected. Figurethe
14 presents 14
presents
3D the 3Dfield
temperature temperature field of the
of the composite composite
lining and thelining and the steel casing.
steel casing.

Figure14.
Figure 14.3D
3Dtemperature
temperaturefield
fieldof
ofthe
theComposite
Compositelining
liningand
andthe
theSteel
Steelcasing.
casing.

Figure14
Figure 14shows
showsthe thecavity
cavitycreated
createdbybythe
theof
ofthe
thejet
jetfire
fireimpingement
impingementon onthe
thelining.
lining.At
Atthis
thisstage,
stage,
thesteel
the steelcasing
casingisisexposed
exposedto tothe
theflames
flamesand
andthe
thepropane
propanewillwillstart
starttotoboil.
boil.The
Thetemperature
temperatureononthe
the
composite surface is lower than the temperature on the steel surface. This is because the thermal
composite surface is lower than the temperature on the steel surface. This is because the thermal
conductivityof
conductivity ofthe
thecomposite
compositeisislow.
low.

3.4.
3.4.Structural
StructuralAnalysis
AnalysisResults
Resultsfor
forthe
theSteel
SteelCasings
CasingsAISI
AISI310
310
AA coupled
coupled heat and structural
heat and structuralanalysis
analysishas hasbeenbeenperformed
performed ononthethe steel
steel casing
casing of the
of the tank.tank.
The
The thermophysical and thermomechanical properties of steel AISI 310 have been
thermophysical and thermomechanical properties of steel AISI 310 have been taken from [21]. The taken from [21].
The
3D 3D Von-Mises
Von-Mises stress
stress distribution
distribution ofof
the tankatatt t= =1010s s(after
thetank (afterthe
thecomposite
composite lining
lining has
has been
been eroded
eroded
completely and the propane was evaporated) is shown in
completely and the propane was evaporated) is shown in Figure 15. Figure 15.
Figure 15 indicates that the maximal Von-Mises stress is greater than ultimate tensile strength
shown at Figure 3 (189 MPa). The high heat flux increases the temperature of the steel wall and
enhances the thermal stresses (produced by the thermal expansion). It also increases the internal
pressure of the propane inside the tank. The 3D displacement field of the tank at t = 10 s (after the
composite lining has been eroded completely and the propane was evaporated) is shown in Figure 16.
ChemEngineering 2019, 3, 1 18 of 22
ChemEngineering 2018, 2, x FOR PEER REVIEW 19 of 23

Figure 15. 3D plot of the Propane tank Von-Mises stress field obtained at t = 10 s.

Figure 15 indicates that the maximal Von-Mises stress is greater than ultimate tensile strength
shown at Figure 3 (189 MPa). The high heat flux increases the temperature of the steel wall and
enhances the thermal stresses (produced by the thermal expansion). It also increases the internal
pressure of the propane inside the tank. The 3D displacement field of the tank at t = 10 s (after the
composite lining has been eroded completely and the propane was obtained
evaporated) is shown in Figure
Figure15.
Figure 15.3D
3Dplot
plotof
ofthe
thePropane
Propanetank
tankVon-Mises
Von-Misesstress
stressfield
field obtainedatatt t==10
10s.s.
16.
Figure 15 indicates that the maximal Von-Mises stress is greater than ultimate tensile strength
shown at Figure 3 (189 MPa). The high heat flux increases the temperature of the steel wall and
enhances the thermal stresses (produced by the thermal expansion). It also increases the internal
pressure of the propane inside the tank. The 3D displacement field of the tank at t = 10 s (after the
composite lining has been eroded completely and the propane was evaporated) is shown in Figure
16.

Figure16.
Figure 16.3D
3Dplot
plotof
ofthe
thePropane
Propanetank
tankdisplacement
displacementfield
fieldobtained
obtainedatatt t==10
10s.s.

Figure 16 indicates that the maximal displacement is 11.744 mm. The maximal displacement
shown in this figure is 96% from the steel initial thickness. This value is greater than the yield point of
the steel. Thus, the cover of the tank will rupture.

Figure 16. 3D plot of the Propane tank displacement field obtained at t = 10 s.

ChemEngineering 2019, 3, 1 19 of 22

3.5. Calculation Results of the Time Period Required to Evaporate the Liquefied Propane
The thermodynamic properties of the propane were calculated from thermodynamic tables [38].
They are summarized at Table 2.

Table 2. Thermodynamic properties of Propane [38].

Thermodynamic Property State 1 State 2

Pressure (p) 1930 kPa 1930 kPa
Temperature (T) 55.57 ◦ C 55.57 ◦ C
Vapor quality (x) 0 1
Internal energy (u) 246.1 kJ/kg 477.4 kJ/kg
Entropy (s) 0.88 kJ/(kg·◦ C) 1.70 kJ/(kg·◦ C)

According to Equation (24), the time period required to evaporate the liquefied propane is:
22.6 min (1353.4 s). Table 2 indicates that the entropy increases from state 1 to state 2. Thus, this process
fulfills the second law of thermodynamics.

4. Discussion
Accidental releases of pressure-liquefied materials (by bursts of high-pressure tanks, processing
equipment malfunction, etc.) are one of major hazards in process industries, transportation, or storage
of flammable materials. Horrific explosions caused by accidental releases of flammable materials into
the atmosphere occurred in Port Hudson (USA, 1970), Flixborough (UK, 1974), Mexico City (Mexico,
1984), Ufa (Russia, 1989), Xian (China, 1998), Nechapur (Iran, 2004), Buncefield (UK, 2005). Different
kinds of explosions are driven by the release of internal energy accumulated in compressed gas or
superheated, liquid. An example of such an explosion is the burst of a vessel with pressure-liquefied
material, known as BLEVE. It is caused by bursting of a pressurized vessel containing liquid above its
atmospheric boiling point. The main cause of a hot BLEVE is heating of the steel wall at the vapor side
of the tank to temperatures in excess of 400 ◦ C. Thermal insulation around the tank can significantly
retard the excessive heating of the tank wall in a fire. This will allow fire fighters enough time to reach
the accident location and to cool the LPG tank to avoid the BLEVE, to extinguish the fire or to evacuate
the people in the vicinity of the accident. Important is that the insulation (coating) retards the heating
of the tank wall long enough. At present, there are three different types of coatings: Intumescent fire
protection coatings, Sublimation fire protection coatings and Ablative coatings
In this work a CFD simulation and mitigation study of BLEVE has been performed. It is assumed
that heptane jet fire impinging on the external boundary of the tank composite lining. The distance
between the jet source and the tank is 3 m. The lining is made of glass-woven vinyl ester. The thickness
of lining is 0.02 m (20 mm). The cylindrical tank casing is made of steel AISI 310. The tank initially
contains liquid propane. The initial temperature of the propane inside the tank is 55.6 ◦ C. The initial
pressure inside the tank is 1930 kPa. The thickness of the steel tube is 12.2 mm (w). The diameter of the
tank is 1.2 m (Di ). The tank contains 500 kg of liquefied propane. The left side of the tank is exposed to
jet fire heating (the vapor quality of the propane is x = 0). At the liquid side (bottom) of the tank the
heat of the fire will be transferred via the steel casing wall to the liquid and cause evaporation of the
liquid propane. As the liquid propane evaporates completely (vapor quality is x = 1) the temperature
and vapor pressure will be increased (for propane 19 bar at 55.6 ◦ C, At the phase of superheated steam
the vapor pressure will increase with the temperature increase). However, if the temperature of the
steel on the gas side of the tank exceeds 450–550 ◦ C the steel will lose its integrity and the tank will
rupture [14]. This will also occur if a tank is equipped with a PRV. The set point of the PRV of the
tank vehicles is generally set at 1.93 MPa for propane mixtures. The proposed computational work is
composed of four phases:
ChemEngineering 2019, 3, 1 20 of 22

(a) CFD Simulation of jet fire by using FDS software

(b) Calculation of the convective and radiative heat fluxes created by impinging jet hitting the
composite coating
(c) Performing heat transfer analysis on the composite coating of the vessel by using FDS software
and COMSOL Multiphysics to simulate the heating and pyrolysis processes of composite lining
(d) Calculation of the time period required to evaporate the liquefied propane
(e) Structural analysis on the steel casing of the tank

It is probably the first time that FDS software has been applied to simulate jet fire and the
composite lining pyrolysis process of the tank lining in order to simulate BLEVE accident scenario.
As far as I know, this work is the first coupled CFD and structural analysis study on the BLEVE
accident of pressure tank. The structural analysis of the steel casing is essential to verify that the
structural integrity of the tank under the BLEVE accident.
It was found that the maximal value of the combined heat flux approaches to 170 kW/m2 .
An ablation analysis has been performed on the composite coating of the vessel. The thermophysical
and thermochemical of woven glass/vinyl ester laminate have been applied in the model. The surface
temperatures of the composite at different values of heat fluxes: 10 kW/m2 , 25 kW/m2 and 50 kW/m2
have been evaluated and validated. The model indicates that surface temperature of the composite
increases with heat flux. Similar results for the surface temperature of the composite have been
reported in the literature. It was found that for lower values of heat fluxes the erosion in the composite
thickness is much smaller.
The composite eroded completely after time period of 338 s (about 6 min) for heat flux of
170 kW/m2 . This time period will allow to extinguish the fire or to evacuate the people in the vicinity
of the accident. It should be noted that during this time period the temperature at the interface of
propane/steel remains at 55.57 ◦ C (The propane has not been heated). Since the propane is in liquid
state, its convective coefficient is much larger. The temperature rise at the interface of glass-woven
vinyl ester/steel is about 10 ◦ C. The decrease in the strength of the steel can be neglected. According
to the thermodynamic analysis it is shown the time period required to evaporate the liquefied propane
is 22.6 min (1353.4 s). A coupled heat and structural analysis has been performed on the steel casing of
the tank. The results indicate that after 10 s (after the composite lining has been eroded completely and
the propane was evaporated), the maximal displacement is 11.744 mm. The maximal displacement
obtained is 96% from the steel initial thickness. This value is greater than the yield point of the steel.
Thus, the cover of the tank will rupture.

5. Conclusions
A new tool has been developed in this work to analyze the pressure vessel exposed to jet fire.
The described work contains new tools and methods. The CFD simulation of the heptane impinging
jet was carried by using the FDS (version 5).
In this work I have also modeled the recession phenomena. It should be noted that numerical
modeling of thermochemical erosion is still complicated issue. The numerical modeling requires
much more knowledge than the conventional approach such as: Arrhenius coefficients (the data is
obtained from thermogravimetric experiments), and enthalpy of formation virgin material (the data
is obtained from calorimeter experiments). As far as I know this is the first time, that the recession
of the composite lining is applied on thermal protection of tank. Moreover, the FDS results have
been coupled with COMSOL to extend the capabilities of the algorithm and to address other physical
phenomena such as transient heating of the steel, structural analysis of the lining and steel tank as a
function. The proposed algorithm is very flexible. It is possible to perform simulation of methane or
other Hydrocarbon fuels. This model can be implemented for simulating other kinds of composite
linings (such as PMMA, PEEK). This work can be implemented on BLEVE caused by pool fire. It is
ChemEngineering 2019, 3, 1 21 of 22

possible to add to the proposed model additional coating which will be on the glass Vinyl Coating
such as intumescent coating.

Funding: This research received no external funding.

Conflicts of Interest: The author declares no conflict of interest. No funding sponsors had any role in the
numerical analyses, or interpretation of data; in the writing of the manuscript, and in the decision to publish
the results.

References
1. Marshall, V. Major Chemical Hazards; Ellis-Horwood: Chichester, UK, 1987.
2. Yakush, S.E. Model for blast waves of boiling liquid expanding vapor explosions. Int. J. Heat Mass Transf.
2016, 103, 173–185. [CrossRef]
3. Gelfand, B.E. Features and simulations of non-ideal explosions. In Proceedings of the 3rd International
Seminar on Fire and Explosion Hazards (ISFEH 2000), Preston, UK, 10–14 April 2000; pp. 43–56.
4. Van den Schoor, F. Influence of Pressure and Temperature on Flammability Limits of Combustible Gases in Air,
Division of Applied Mechanics and Energy Conversion Section; Katholieke Universiteit: Leuven, Belgium, 2007.
5. Baker, Q.A.; Tang, M.J.; Pieroazio, A.P.; Birk, A.M.; Woodward, S.E.; Geng, J.; Ketchum, D.E.; Parsons, P.J.;
Thomas, J.K.; Daudonnet, B. Guidelines for Vapor Cloud Explosion, Pressure Vessel Burst, BLEVE, and Flash Fire
Hazards, 2nd ed.; John Wiley & Sons, Inc.: Hoboken, NJ, USA, 2010.
6. Marsden, E. Consequence Modelling: Overview of Hazards. Available online: https://risk-engineering.org
(accessed on 1 October 2018).
7. Berthélot, M.; Vieille, P. Sur la vitesse de propagation des phenomenes explosifs dans le gaz. C. R. Acad.
Sci. Paris 1881, 93, 18.
8. Mallard, E.; Le Chatelier, H.L. Sur la vitesse de propagation de l’inflammation dans les melanges explosifs.
C. R. Acad. Sci. Paris 1881, 93, 145–148.
9. Lewis, B.; Von Elbe, G. Combustion Flames and Explosion of Gases, 2nd ed.; Academic Press Inc.: New York, NY,
USA; London, UK, 1961.
10. Volpiani, P.S.; Schmitt, T.; Vermorel, O.; Quillatre, P.; Veynante, D. Large eddy simulation of explosion
deflagrating flames using a dynamic wrinkling formulation. Combust. Flame 2017, 186, 17–31. [CrossRef]
11. Diéguez, P.M.; López-San, M.J.; Idareta, I.; Uriz, I.; Arzamendi, G.G.; Luis, M. Hydrogen Hazards and Risks
Analysis through CFD Simulations. In Renewable Hydrogen Technologies: Production, Purification, Storage,
Applications and Safety; Gandia, L., Arzamedi, G., Diegues, P., Eds.; Elsevier: Oxford, UK, 2013; Chapter 18;
pp. 437–452.
12. Pinhasi, G.A.; Ullmann, A.; Dayan, A. 1D plane numerical model for boiling liquid expanding vapor
explosion (BLEVE). Int. J. Heat Mass Transf. 2007, 50, 4780–4795. [CrossRef]
13. Scholz, B.; Wuersig, G.M. Consequences of Pool Fires to LNG Ship Cargo Tanks. In Process and Plant Safety
Applying Computational Fluid Dynamics; Schmidt, J., Ed.; Wiley-VCH Verlag & Co.: Weinheim, Germany, 2012;
Volume 12, Chapter 8; pp. 123–137.
14. Molag, M.; Reinders, J.; Elbers, J. Measures to Avoid a Hot BLEVE of a LPG Tank; IChemE: Rugby, UK, 2007.
15. Sklorz, C.; Otremba, F.; Balke, C. BLEVE performance of fire protection coating systems on dangerous goods
tanks in a test fire. Int. J. Mater. Sci. Eng. 2013, 1, 90–93. [CrossRef]
16. Jimenez, M.; Duquesne, S.; Bourbigot, S. High-throughput fire testing for intumescent coatings. Ind. Eng.
Chem. Res. 2006, 45, 7475–7481. [CrossRef]
17. Jimenez, M.; Bellayer, S.; Naik, A.; Bachelet, P.; Duquesne, S.; Bourbigot, S. Topcoats versus durability of an
intumescent coating. Ind. Eng. Chem. Res. 2016, 55, 9625–9632. [CrossRef]
18. Geoffroy, L.; Samyn, F.; Jimenez, M.; Bourbigot, S. Intumescent polymer metal laminates for fire protection.
Polymers 2018, 10, 995. [CrossRef]
19. Pötzsch, S.; Krüger, S.; Sklorz, C.; Borch, J.; Hilse, T.; Otremba, F. The fire resistance of lightweight composite
tanks depending on fire protection systems. Fire Saf. J. 2018, 100, 118–127. [CrossRef]
20. Mouritz, A.P.; Gibson, A.G. Fire Properties of Polymer Composite Materials; Springer: Dordrecht, The
Netherlands, 2006.
ChemEngineering 2019, 3, 1 22 of 22

21. American Iron and Steel Institute. High Temperature Characteristics of Stainless Steels, A Designer’s Handbook
Series; AISI: Washington D.C., NY, USA, 1979.
22. McGrattan, K. Fire Dynamics Simulator (Version 5)—Technical Reference Guide Volume 1: Mathematical Model;
NIST Special Publication 1018; National Institute of Standards and Technology U.S. Department of Commerce:
Washington D.C., NY, USA, 2010.
23. McGrattan, K.; Forney, G.P. Fire Dynamics Simulator (Version 5)—User’s Guide; NIST Special Publication 1019;
National Institute of Standards and Technology U.S. Department of Commerce: Washington D.C., NY,
USA, 2010.
24. McGrattan, K. Numerical Simulation of the Caldecott Tunnel Fire, April 1982; NISTIR; National Institute of
Standards and Technology; U.S. Department of Commerce: Washington D.C., NY, USA, 2005.
25. Baukal, C.E. Heat Transfer in Industrial Combustion; CRC Press LLC: New York, NY, USA, 2000.
26. Heywood, J.B. Internal Combustion Engine Fundamentals; McGraw-Hill Book Company: New York, NY,
USA, 1988.
27. Statler, D.L., Jr. A Mechanistic and Modeling Study of Recycled and Virgin Flame Retarded Polycarbonate.
Ph.D. Thesis, College of Engineering and Mineral Resources, West Virginia University, Morgantown, WW,
USA, 2008.
28. Feih, S.; Mathys, Z.; Gibson, A.G.; Mouritz, A.P. Modelling the tension and compression strengths of polymer
laminates in fire. Compos. Sci. Technol. 2007, 67, 551–564. [CrossRef]
29. COMSOL Multiphysics—Modeling Guide, Version 4.3b; COMSOL AB: Stockholm, Sweden, 2013.
30. Gu, P.; Chen, W. Influence of thermal distortion to compression failure of polymer matrix panels in fire.
Compos. Struct. 2012, 94, 2174–2180. [CrossRef]
31. Huang, Y.; Zhu, Y.; Cimini, C.A., Jr.; Ha, S.K. Characterization of moisture effect on static and fatigue
performance of epoxy resin using thin-film specimen on dynamic mechanical analyzer. J. Compos. Mater.
2016, 51, 303–314. [CrossRef]
32. Mouritz, A.P.; Mathys, Z. Post-fire mechanical properties of marine polymer composites. Compos. Struct.
1999, 47, 643–653. [CrossRef]
33. Giesbrecht, H.; Hess, K.; Leuckel, W.; Maurer, B. Analyse der potentiellen explosionswirkung von kurzzeitig
in de atmosphaere freigesetzen brenngasmengen. Chem. Ing. Tech. 1980, 52, 114–122. [CrossRef]
34. Van Doormaal, J.C.A.M.; Van Wees, R.M.M. Rupture of Vessels. In Methods for the Calculation of Physical
Effects, CPR 14E (Yellow Book), 3rd ed.; Van den Bosch, C.J.H., Weterings, R.A.P.M., Eds.; TNO: The Hague,
The Netherlands, 2005; Chapter 7; pp. 697–775.
35. Çengel, Y.A.; Boles, M.A. Thermodynamics: An Engineering Approach, 5th ed.; McGraw-Hill: New York, NY,
USA, 2006.
36. Wighus, R.; Drangsholt, G. Impinging jet Fire Experiments—Propane 14 MW Laboratory Tests, STF25 A92026;
SINTEF NBL—Norwegian Fire Research Laboratory: Trondheim, Norway, 1993.
37. McGrattan, K.; Hostikka, S.; Floyd, J.; McDermott, R. Fire Dynamics Simulator (Version 5)–Technical Reference
Guide Volume 3: Validation; NIST Special Publication 1018-5; National Institute of Standards and Technology,
U.S. Department of Commerce: Washington D.C., NY, USA, 2010.
38. Moran, M.J.; Shapiro, H.N.; Boettner, D.D.; Bailey, M.B. Fundamentals of Engineering Thermodynamics, 7th ed.;
John Wiley and Sons: Hoboken, NJ, USA, 2011.

© 2018 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).