Você está na página 1de 11

Ships and Offshore Structures

ISSN: 1744-5302 (Print) 1754-212X (Online) Journal homepage: http://www.tandfonline.com/loi/tsos20

CFD simulations of gas explosion and fire actions

J. K. Paik , B. J. Kim , J. S. Jeong , S. H. Kim , Y. S. Jang , G. S. Kim , J. H. Woo , Y.


S. Kim , M. J. Chun , Y. S. Shin & J. Czujko

To cite this article: J. K. Paik , B. J. Kim , J. S. Jeong , S. H. Kim , Y. S. Jang , G. S. Kim , J. H.


Woo , Y. S. Kim , M. J. Chun , Y. S. Shin & J. Czujko (2010) CFD simulations of gas explosion and
fire actions, Ships and Offshore Structures, 5:1, 3-12, DOI: 10.1080/17445300902872028

To link to this article: http://dx.doi.org/10.1080/17445300902872028

Published online: 18 Nov 2009.

Submit your article to this journal

Article views: 553

View related articles

Citing articles: 13 View citing articles

Full Terms & Conditions of access and use can be found at


http://www.tandfonline.com/action/journalInformation?journalCode=tsos20

Download by: [Nanyang Technological University] Date: 09 October 2017, At: 01:44
Ships and Offshore Structures
Vol. 5, No. 1, 2010, 312

CFD simulations of gas explosion and fire actions


J.K. Paika , B.J. Kima, , J.S. Jeonga , S.H. Kima , Y.S. Jangb , G.S. Kimb , J.H. Wooc , Y.S. Kimc , M.J. Chunc , Y.S. Shind
and J. Czujkoe
a
Pusan National University, Pusan, Korea; b Hyundai Heavy Industries, Ulsan, Korea; c Daewoo Shipbuilding & Marine Engineering,
Geoje, Korea; d American Bureau of Shipping, Houston, TX, USA; e Nowatec AS, Asker, Norway
(Received 16 December 2008; final version received 6 March 2009)

Hydrocarbon (gas) explosions and fire are the hazardous events that present the greatest potential risk for offshore installations.
Downloaded by [Nanyang Technological University] at 01:44 09 October 2017

More refined computations of the consequences of explosions and fire within the framework of quantified risk assessment and
management for offshore installations are therefore required. Although the action characteristics of such hazardous events
must be identified prior to an analysis of their consequences, the objective of the present paper is to develop a modeling
technique for computational fluid dynamics (CFD) simulations of hydrocarbon explosions and fire. The results of this study
form part of phase I of the three-phase Joint Industry Project on Explosion and Fire Engineering in floating production,
storage and offloading systems (FPSO) (abbreviated by EFEF JIP).
Keywords: offshore installations; floating production storage and offloading unit (FPSO); gas explosion; fire; computational
fluid dynamics (CFD); quantified risk assessment (QRA) and management

1. Introduction hazards within the framework of risk assessment and man-


Structural design and safety assessment requires the iden- agement is the impact of pressure on explosions and that of
tification of the characteristics of both actions and action elevated temperature on fire.
effects. Here, the term action refers to either an external The topsides of offshore platforms are most likely to be
load applied to a structure (direct action) or an imposed exposed to such hazards as hydrocarbon explosions and/or
deformation or acceleration (indirect action), and the term fire, and a number of major accidents involving them have
action effects refers to the effects of these actions on a been reported, one of which is shown in Figure 1. De-
global structure or structural components (ISO 2007). This tails of some of these accidents can be found in Vinnem
paper focuses on the characteristics of the actions that are (2007).
due to gas explosions and fires. The quantified risk assessment (QRA) and management
Although they often accompany each other in offshore framework requires that the consequences of hazards be
installations, hydrocarbon explosions and fire are differ- analysed in a more refined way. Prior to that analysis, how-
ent phenomena by nature. Hydrocarbons are organic com- ever, the characteristics of the actions that arise from them
pounds that contain only hydrogen and carbon (Nolan must be identified (Czujko 2001; Paik and Thayamballi
1996). They typically take the form of gases at ordinary 2003; Czujko 2005, 2007; Paik and Thayamballi 2007; Paik
temperatures, but change in form to a liquid or even solid and Melchers 2008).
state with an increase in molecular weight. In fact, the The computational fluid dynamics (CFD) method is
principal constituents of petroleum and natural gas are recognised as one of the most powerful tools for identi-
hydrocarbons. fying the action characteristics of hydrocarbon explosions
Hydrocarbons can be exploded through ignition when and fire. Thus, the aim of the study reported herein was
combined with an oxidiser (usually oxygen or air). Thus, to develop a modeling technique for CFD simulations of
when the temperature rises to the point at which hydrocar- hydrocarbon explosion and fire actions as part of a joint in-
bon molecules react spontaneously to an oxidiser, combus- dustry project on explosion and fire engineering in floating
tion takes place. This hydrocarbon explosion causes a blast production storage and offloading systems (FPSOs). This
or a rapid increase in pressure. project, the Joint Industry Project on Explosion and Fire
Fire, in contrast, is a combustible vapour or gas that Engineering in FPSOs (EFEF JIP), began in March 2008
combines with an oxidiser in a combustion process that is (Paik and Czujko 2008) with the main objective of develop-
manifested by the evolution of light, heat and flame. Of ing a fully QRA and management system for hydrocarbon
primary concern in terms of the actions that result from explosions and fire.


Corresponding author: Email: bonjour@pusan.ac.kr
ISSN: 1744-5302 print / 1754-212X online
Copyright 
C 2010 Taylor & Francis
DOI: 10.1080/17445300902872028
http://www.informaworld.com
4 J.K. Paik et al.

accidents. The ANSYS CFX (2008) code is employed for


this purpose.

2. Analysis of gas explosion actions


2.1. Experimental study
Gieras et al. (2006) carried out a laboratory test on methane
gas explosions. In the present study, their experimental
study was reanalysed via the ANSYS CFX code to develop
a modeling technique for CFD simulations to identify the
pressure actions of these explosions.
Figure 2 shows the layout of the test, in which a 40-dm3
volume chamber was used. Although the concentrations of
methane gas were varied in the original test, the present
Downloaded by [Nanyang Technological University] at 01:44 09 October 2017

study adopted a concentration of 9.5%, equivalent to the


Figure 1. Piper Alpha explosion and fire accident on July 6, 1988 stoichiometric condition, as an illustrative example.
(Cullen 1990).
The initial gas temperature inside the explosion cham-
ber was also varied, at 293, 373 and 473 K, and the outside
ambient temperature was kept at 293 K. A pressure gauge
The EFEF JIP comprises the following three phases: and temperature sensor were attached to the chambers outer
surface.
r Phase I (April 2008March 2009) Figure 3 shows a photo of this test chamber, which was
140 mm in diameter and 441 mm in height.
- Literature survey of previous related JIPs
- Technologies for the frequency analysis of gas ex- 2.2. CFX simulations
plosions and fire on FPSOs
- Technologies for the action analysis of gas explo- The chemical reaction of the methane gas explosion asso-
sions and fire ciated with the mixture of air can be represented by

r Phase II (April 2009March 2010) CH4 + 2(O2 + 3.76N2 )


CO2 + 2H2 O + 2(3.76N2 ) + Energy.
- Hypothetical design for FPSO topsides
- Wind tunnel tests on FPSO topside modules (e.g. After the explosion, energy was released, thus causing a
accommodation) and side-by-side offloading facil- blast, as shown in Figure 4. The peak pressure was reached
ities in association with dispersion of flame (heat after a very short time and decayed with time.
and cloud) coming from flare tower or fire with The ANSYS CFX code is now employed for an analysis
varying wind direction and strength of the explosion actions. Figure 5 shows the CFX model for
- Fire tests on plated and framed structural models the Gieras explosion test using tetrahedral elements. In con-
to identify the time-variant temperature actions and sideration of the symmetric action feature, a half chamber
structural consequence (action effects) is taken as the extent of the analysis. The total number of
- Technologies for the action effect (consequence) nodes and elements were 32,364 and 180,315, respectively,
analysis of gas explosions and fire which were determined based on a convergence study.
In CFX modeling, turbulence is modeled by the k-
r Phase III (April 2009March 2011) epsilon model and radiation by the P1 model. The total
energy model is used to model the heat transfer inside the
- Procedure of QRA and management of gas explo- chamber. The CFX code facilitates four types of wall heat
sions and fire on FPSO topsides transfer options, namely, the adiabatic, temperature, heat
- Acceptance criteria flux and heat coefficient options. The present study used
- Guidelines and engineering handbook the temperature option to model the wall heat transfer be-
haviour, so as to reflect an ambient temperature of 293 K.
The results of the study presented here form part of Figure 6 shows the variations in explosion pressure with
phase I. The focus of this study was on developing a time at (a) 0.002 s, (b) 0.012 s, (c) 0.02 s and (d) 0.132 s, and
modeling technique for CFD simulations to identify the Figure 7 shows the explosion pressure versus time history,
action characteristics of hydrocarbon explosions and fire as obtained from the CFX simulations. It can be seen from
Ships and Offshore Structures 5
Downloaded by [Nanyang Technological University] at 01:44 09 October 2017

Figure 2. Layout of the laboratory test facility used for the methane gas explosion (Gieras et al. 2006).

Figure 7 that these simulations are in good agreement with from this figure that the total energy model, which takes
the experimental results in terms of the pressure versus the kinetic energy effect into account, is more relevant to
time history, which can be characterised by four parameters, an analysis of explosion actions.
namely, time, peak pressure, type of decay after the peak Figure 7(b) presents the effects of the chamber wall
pressure has been reached and the duration of the impact heat transfer modeling. It can be seen that the ambient tem-
pressure (Paik and Thayamballi 2007). perature model that is equivalent to the test gives accurate
Figure 7(a) presents the effects of heat transfer modeling results, whereas the adiabatic model fails to transfer the
in terms of the thermal or total energy models. It is evident heat energy to the exterior of the chamber.

Figure 4. Explosion pressure versus time history, as obtained


Figure 3. Photo of the chamber used for the explosion test (Gieras from the test with a methane concentration of 9.5% (Gieras et al.
et al. 2006). 2006).
6 J.K. Paik et al.
Downloaded by [Nanyang Technological University] at 01:44 09 October 2017

Figure 5. ANSYS CFX model for the Gieras explosion test using tetrahedral elements.

Figure 6. Variations in explosion pressure with time, as obtained from the ANSYS CFX simulations: (a) 0.002 s, (b) 0.012 s, (c) 0.02 s
and (d) 0.132 s. Illustrations can be viewed in colour online.
Ships and Offshore Structures 7
Downloaded by [Nanyang Technological University] at 01:44 09 October 2017

Figure 7(a). Effects of the heat transfer model on the explosion Figure 7(b). Effects of the wall heat transfer model on the ex-
pressure versus time history of the Gieras explosion test model, plosion pressure versus time history of the Gieras explosion test
as obtained from the CFX simulations. model, as obtained from the CFX simulations.

Although the time-variant temperatures were not mea- Figure 9. Figure 10 shows a photo of the jet firing inside
sured during the test, the CFX simulations provided the test compartment.
time-variant temperature-related information, as shown in The temperature measurements at various locations
Figure 8. The peak temperature reached more than 2,200 sometime after the jet firing had begun are shown in
K around 0.1 s after ignition, which is prior to the time Figure 11. It can be seen that the maximum temperature
at which the peak pressure is meant to occur, i.e. at 0.2 s. was 1400 K and the temperature at location CD was higher
Figure 8 also indicates that the adiabatic model of wall heat
transfer behaviour also increases the peak temperature.

3. Analysis of fire actions


3.1. Experimental study
The HSE (1995) carried out a laboratory test on a jet fire
accident caused by the combustion of propane gases.
The thermally insulated compartment shown in Figure
9 was used for this test. This compartment, which is approx-
imately 135 m3 in volume, has an open door with a breadth
of 2.5 m and a height of 2 m for ventilation purposes. This
door also allows the heat and flame that result from the fire
to flow out.
The jet fire was activated through a 380-mm height by
90-mm diameter nozzle located at the bottom centre of the
compartment. It was supposed that the mass flow of the jet
fire at the nozzle would be 0.33 kg/s, which is equivalent to
a flow velocity of 190 m/s. The outside ambient temperature
of the compartment was kept at 281 K during the test.
The temperatures during the fire were measured via Figure 8(a). Effects of the heat transfer model on the temperature
temperature sensors at various locations or elevations, versus time history of the Gieras explosion test model, as obtained
i.e. at AB and CD and the opening, as indicated in from the CFX simulations.
8 J.K. Paik et al.

Figure 10. Photo of the jet fire test (HSE 1995).


Downloaded by [Nanyang Technological University] at 01:44 09 October 2017

The ANSYS CFX code is now used for an analysis of the


Figure 8(b). Effects of the wall transfer model on the temperature jet fire actions. Useful CFD simulations of this test have
versus time history of the Gieras explosion test model, as obtained also been performed in the literature (e.g. Wen and Huang
from the CFX simulations. 2000).
Figure 12 shows the CFX model of the jet fire test,
which was developed in the present study using tetrahedral
than that at AB. This is because the heat flow moves out elements. In consideration of the symmetric action feature,
via the opening (ventilation). half of the compartment is taken as the extent of the analysis.
It is generally considered that the heat flow actions out-
side the compartment may also affect the inside action char-
acteristics. Therefore, a certain amount of open air around
3.2. CFX simulations the ventilation door was also included in the CFX model,
Although the imperfect combustion of propane gases may as shown in Figure 12(b). The total number of nodes and
produce some CO and C2 H2 , the chemical reaction associ- elements were 30,918 and 165,353, respectively.
ated with the perfect combustion of these gases when mixed The thermal energy model was applied in the CFX mod-
with air can be represented by eling for heat transfer behaviour, the eddy dissipation model
for combustion behaviour, the P1 model for radiation be-
C3 H3 + 5O2 3CO2 + 4H2 O. haviour and the k-epsilon model for turbulence behaviour.

Figure 9. The jet fire test facility (HSE 1995).


Ships and Offshore Structures 9
Downloaded by [Nanyang Technological University] at 01:44 09 October 2017

Figure 12(b). Extent of the ANSYS CFX model analysis of the


jet fire test for a half of the test compartment with an extended
outside domain.

As the test compartment was thermally insulated, it was


Figure 11. Variations in temperature at different locations or
elevations, as obtained in the test (HSE 1995). considered that a partial, rather than full, wall heat transfer
should be allowed. The total heat flux via the wall may be
given by
In terms of the wall heat transfer modeling, three types
qw = hc (To Tw ),
of options were considered, i.e. the adiabatic, temperature
and transfer coefficient models, as discussed previously.
The adiabatic model does not allow any heat flow transfer where qw = the total heat flux; hc = the heat transfer coeffi-
out of the compartment via the wall, whereas the temper- cient, which was taken to be hc = 30 W/m2 K in the present
ature model allows it to transfer directly to the ambient fire test; To = the external or outside (ambient) tempera-
temperature. The transfer coefficient model is capable of ture, which was equivalent to 281 K in the present test and
dealing with partial heat transfer behaviour. Tw = the internal temperature.
Figure 13 shows the variations in temperature with time
at (a) 0.02 s, (b) 1.0 s, (c) 2.0 s and (d) 9.0 s after the jet
firing had begun, as obtained from the CFX simulations. It
is evident from this figure that the heat flow moves with time
through the opening. Figure 14 presents a comparison of
the temperatures at various locations or elevations between
the test measurements and the CFX simulations.
Although the time-variant temperature characteristics
were not measured in the test, the CFX simulations provided
such information, as shown in Figure 15.
Figure 16 shows the effects of the wall heat transfer
model on the temperature versus time history at location
CD at an elevation of 1.7 m, as obtained from the CFX
simulations. It can be seen from this figure that the adi-
abatic model overestimates the fire temperatures, whereas
the ambient temperature model underestimates them. The
proposed CFX model with a wall heat transfer coefficient of
30 W/m2 K gives very accurate solutions when compared
to the experimental results.

4. Concluding remarks
Figure 12(a). ANSYS CFX model of the jet fire test using tetra- The aim of the study reported herein was to develop a
hedral elements. CFD modeling technique for the analysis of hydrocarbon
10 J.K. Paik et al.
Downloaded by [Nanyang Technological University] at 01:44 09 October 2017

Figure 13. Variations in temperature with time at (a) 0.02 s, (b) 1.0 s, (c) 2.0 s and (d) 9.0 s.

Figure 14(a). Comparison of temperatures at AB in the test Figure 14(b). Comparison of temperatures at CD in the test
measurements and CFX simulations. measurements and CFX simulations.
Ships and Offshore Structures 11
Downloaded by [Nanyang Technological University] at 01:44 09 October 2017

Figure 14(c). Comparison of temperatures at the opening in the Figure 16. Effects of the wall heat transfer model on the tem-
test measurements and CFX simulations. perature versus time history at CD at an elevation of 1.7 m, as
obtained from the CFX simulations.

explosion and fire actions. Existing test data on a methane


gas explosion and propane gas jet fire were reanalysed using or heat may play an important role. A related study is thus
the ANSYS CFX code. on-going as part of the EFEF JIP.
It is concluded that the CFD simulations proposed in
this study are in good agreement with the experimental
results. Acknowledgements
This study was limited to laboratory test data obtained This work was supported for two years by a Pusan National Uni-
versity research grant.
in a closed chamber or compartment, although the topsides
of FPSOs are more likely to be open or partially closed
environments, in which the dispersion behaviour of a blast References
ANSYS. 2008. ANSYS CFX release 11.0. Canonsburg (PA):
ANSYS Inc.
Cullen L. 1990. The public inquiry into the piper alpha disaster.
London (England): HMSO.
Czujko J. 2001. Design of offshore facilities to resist gas explosion
hazard: engineering handbook. Oslo (Norway): CorrOcean.
Czujko J. 2005. Computational methods in the analysis of ex-
plosion resistant barriers. Proceedings of International Con-
ference on Computational Methods in Marine Engineering
(MARINE 2005) Oslo, Norway, June 2729.
Czujko J. 2007. Consequences of explosions in various indus-
tries. Safety and reliability of industrial products, systems
and structures (SAFERELNET 2007).
Gieras M, Klemens R, Rarata G, Wolanski P. 2006. Determination
of explosion parameters of methane-air mixtures in the cham-
ber of 40 dm3 at normal and elevated temperature. J Loss Prev
Process Indust. 19:263270.
HSE. 1995. Large scale compartment fires: experimental details
and data obtained in test comp-27. London (England). Health
and Safety Executive report-OTO94024.
ISO. 2007. International standards ISO 18072-1. Ships and marine
technologyship structures Part 1: General requirements
for their limit state assessment. Geneva (Switzerland).
Nolan DP. 1996. Handbook of fire and explosion protection engi-
Figure 15. Temperature versus time history at various locations neering principles for oil, gas, chemical, and related facilities.
at an elevation of 1.7 m, as obtained from the CFX simulations. Westwood (NJ): Noyes Publications.
12 J.K. Paik et al.

Paik JK, Czujko Z. 2008. Joint industry project on explosion and Paik JK, Thayamballi AK. 2007. Ship-shaped offshore installa-
fire engineering of FPSOs. Proceedings of the workshop on tions: design, building, and operation. Cambridge (England):
Design of FPSO Structures; 2008 Nov. 2021; Geoje, Korea. Cambridge University Press.
Paik JK, Melchers RE. 2008. Condition assessment of aged struc- Vinnem JE. 2007. Offshore risk assessment. London (England):
tures. New York: CRC Press. Springer-Verlag.
Paik JK, Thayamballi AK 2003. Ultimate limit state design of Wen JX, Huang LY. 2000. CFD modeling of confined jet fires
steel-plated structures. Chichester (England): John Wiley & under ventilation-controlled conditions. Fire Saf J. 34:1
Sons. 24.
Downloaded by [Nanyang Technological University] at 01:44 09 October 2017

Você também pode gostar