Accepted Manuscript

Effect of gas composition on dispersion characteristics of blowout gas on offshore

platform

Dongdong Yang, Guoming Chen, Jihao Shi, Xinhong Li

PII: S2092-6782(18)30232-2
DOI: https://doi.org/10.1016/j.ijnaoe.2019.02.009
Reference: IJNAOE 250

To appear in: International Journal of Naval Architecture and Ocean

Engineering

Received Date: 20 July 2018

Revised Date: 19 December 2018
Accepted Date: 19 February 2019

Please cite this article as: Yang, D., Chen, G., Shi, J., Li, X., Effect of gas composition on dispersion
characteristics of blowout gas on offshore platform, International Journal of Naval Architecture and
Ocean Engineering, https://doi.org/10.1016/j.ijnaoe.2019.02.009.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.
 ACCEPTED MANUSCRIPT
1 Effect of gas composition on dispersion characteristics of blowout gas on
2 offshore platform
3 Dongdong Yang, Guoming Chen, Jihao Shi, Xinhong Li

4 Centre for Offshore Engineering and Safety Technology, China University of Petroleum (East China),

5 Qingdao, 266580, Shandong, China

PT
6 Abstract: Gas composition has a significant impact on the dispersion behavior and accumulation characteristics

7 of blowout gas. However, few public studies has investigated the corresponding effect of gas composition.

RI
8 Therefore, this study firstly builds the FLACS-based numerical model about an offshore drilling platform. Then

9 several scenarios by varying the composition of blowout gas are simulated while the scenario with the

SC
10 composition of “Deepwater Horizon” accident is regarded as the benchmark. Furthermore, the effects of the gas

11 composition on the flammable cloud volume, the influenced area of flammable cloud, the influenced area of

12
U
hydrogen sulfide and the critical time of the hydrogen sulfide spreading to the living area are analyzed. The results
AN
13 demonstrate that gas composition is a driving factor for dispersion characteristics of blowout gas. All the results

14 can give support to reduce the risk of the similar accidents incurred by real blowouts.
M

15 Keywords: blowout accident; gas composition; offshore platform; hydrocarbon; hydrogen sulfide

16
D

17 1. Introduction
TE

18

19 Gas blowout is a catastrophic accident in the process of offshore oil and gas development since the blowouts can
EP

20 cause massive causalities and severe environmental pollution. For example, the BP Deepwater Horizon blowout

21 accident (BP, 2010) in 2010 took 11 lives, injured 17 people, destroyed the offshore platform and eventually
C

22 released the oil and gas over several months. In addition, a natural gas leakage accident occurred in the Kab-121
AC

23 platform in the Gulf of Mexico in 2007 (Zhu and Chen, 2009). Since the gas contains the hydrogen sulfide, the

24 accident killed 21 people. Therefore, studying the gas blowout is quite important to prevent and control the similar

25 accidents of offshore platforms.

26

27 The dispersion behavior analysis is the critical topic to study the gas blowout of offshore platforms. Three

28 methods have been employed to analyze the corresponding dispersion behavior, i.e. experimental method,

29 empirical model and Computational Fluid Dynamics (CFD) model. The experimental method (Dandrieux A et al.,
 ACCEPTED MANUSCRIPT
30 2002; Deng et al., 2012; Fu et al., 2016) is the most accurate while it is infeasible to be conducted because of its

31 high risk and cost characteristics. In terms of the empirical model (Liu et al., 2017; S. R. Hanna et al., 1996;

32 Zhang et al., 2011), some simplified assumptions are generally made, which thereby reduces the accuracy of the

33 model in congested facilities of offshore platforms. Relatively, CFD model (Li et al., 2016; Zhang et al., 2014; Li

34 et al., 2016) is the most widely used alternative to study the dispersion behavior of leaked gas because of its

35 superior performance, e.g. the higher accuracy, lower risk and cost. Recently, several scholars have adopted varied

PT
36 CFD models to study the dispersion behavior under different geometries by considering various wind and leakage

37 conditions. However, some scholars (Liu et al., 2015; Li et al., 2015; Tauseef S M et al., 2011; Li et al., 2018)

RI
38 simplified the gas composition, which leads to the CFD models difficultly describe the dispersion behavior of real

39 accident. That motivates this study to analyze the effect of gas composition from the real accident on the

SC
40 dispersion behaviors.

41

42
U
This study firstly builds the CFD model of gas blowout about an offshore drilling platform by using the FLACS
AN
43 (Flame Acceleration Simulator). FLACS is a leading tool for dispersion and explosion safety analysis which has

44 been validated against numerous experiments with different scales and different dispersion scenarios (SAVVIDES
M

45 C et al., 2001; Middha et al., 2009; Hansen et al., 2010; Bleyer et al., 2012). The tool has been widely utilized in

46 the process industry to simulate and analyze the characteristics of leaked gas dispersion and vapor cloud explosion
D

47 (Huser and Kvernvold, 2000; Qiao and Zhang, 2010; Li et al., 2014; Huang et al., 2017; Shi et al., 2018; Shi et al.,
TE

48 2018). Based on the FLACS, several dispersion scenarios with varied gas compositions are simulated where the

49 scenario with the composition from the BP accident is viewed as the benchmark. Furthermore, sensitivity analysis
EP

50 of different gas composition on the dispersion behavior of blowout gas, i.e. the flammable cloud volume,

51 influenced area of flammable cloud, influenced area of hydrogen sulfide and critical time of hydrogen sulfide
C

52 spreading to the living area is conducted. The sensitivity results can give a reference to prevent and control the
AC

53 similar blowout accidents.

54

55 The structure of the rest paper is organized as below: Section 2 explains the procedure of numerical dispersion

56 modeling of blowout gas by FLACS. Section 3 shows the CFD theoretical model for gas dispersion and model

57 validation. Section 4 focuses on numerical modeling of an offshore drilling platform and parameters definition. The

58 numerical simulation results and discussions are conducted in Section 5. Section 6 is devoted to the conclusions.

59
 ACCEPTED MANUSCRIPT
60 2. Procedure of numerical dispersion modeling of blowout gas by FLACS
61

62 This part is to clearly demonstrate the procedure regarding the numerical dispersion modeling of the blowout gas

63 with varied gas compositions by FLACS. Fig.1 shows the corresponding procedure. Firstly, the geometry model is

64 built according to the 3D layout and construction data of a real offshore drilling platform. Then, the input

65 parameters include the gas composition, wind condition, boundary condition, leak rate, etc. are defined. Thirdly,

PT
66 the simulation domain is set and the grid covering the domain is thereby generated based on grid refinement

RI
67 around leak position. Subsequently, sensitivity analysis of the grid size is conducted to ensure the optimal grid

68 size. Finally, several dispersion scenarios are simulated in order to assess the effect of gas composition on

SC
69 dispersion characteristics of the blowout gas.

70

U
71 It should be noted finer grid size can improve the accuracy of the dispersion modeling while also increase the
AN
72 number of grid. However, larger number of grids can bring in much computational burden. Therefore, sensitivity

73 analysis of the grid size should be conducted to keep a balance between the model accuracy and the computational

74 costs. The general process of sensitivity analysis of the grid size is that: (1) Perform dispersion simulation with
M

75 grid model (gi) and obtain blowout gas concentration field (fi); (2) Improve the grid model and perform dispersion
D

76 simulation with the improved grid model (gi+1), the blowout gas concentration field (fi+1) is obtained. (3) Compare

77 fi with fi+1, if the convergence of the dispersion simulations is met, the grid model (gi) is applicable. Otherwise,
TE

78 repeat the step (2) and step (3) until the convergence is met. Once the optimal grid size is determined, the

79 corresponding CFD model is built and can be eventually used to study the dispersion behavior of blowout gas of
EP

80 the offshore drilling platform.

C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
Fig. 1. Procedure of numerical dispersion modeling of blowout gas by FLACS
AN
81

82 3. Numerical dispersion modeling in FLACS

M

83 3.1 Basic governing equation

D

84

85 FLACS uses the finite volume method to solve the compressible RANs equations based on the three-dimensional
TE

86 Cartesian grid. The jet and dispersion processes of blowout gas follow the continuity equation, mass conservation,

87 momentum conservation and energy conservation. The control equation can be presented in general as:
EP

 
∂   ∂  ∂ 
 

∂φ 
∂t  ρφ  + ∂x j  ρu jφ  = ∂x j  ρΓφ ∂x  + Sφ

（1）

j
C

88 where φ represents the general variable, including variables such as mass, momentum, energy, turbulent kinetic
AC

89 energy and so on; ρ represents the gas mixture density; t represents time; uj represents the velocity component in

90 j-direction; Гφ represents the dispersion coefficient of general variable φ; Sφ represents the source term.

91

92 The dispersion process of the blowout gas with multiple composition follows mixture fraction and fuel mass

93 fraction transport equation, which can be expressed as:

∂ ∂ ∂  µeff ∂ζ 
（β v ρζ ）+ ( β j ρ u jζ ）=  β j 
∂t ∂x j ∂x j  σ ζ ∂x j 
（2）
 ACCEPTED MANUSCRIPT
∂ ∂ ∂  µeff ∂Y fuel 
∂t
( β v ρY fuel ) +
∂x j
( β j ρ u jY fuel ) =
∂x j
 β j
σ ∂x
 + R fuel (3)
 fuel j 
94 where βv represents volume porosity, βj represents area porosity in the j-direction, ξ represents the mixture

95 composition, ueff represents the effective turbulence viscosity, Yfuel represents mass fractions of fuel in reactants

96 and products, Rfuel represents the fuel reaction rate, which is zero in this study, σξ and σfuel are constants, which are

97 both taken as 0.7.

PT
98

99 3.2 Turbulence model

RI
100

SC
101 In FLACS, the Reynolds stress tensor is introduced to describe the turbulence of the dispersion process for the

102 blowout gas. In order to solve the Reynolds stress tensor, the k–ε model presenting the turbulent kinetic energy

U
103 transport equation and turbulent kinetic energy dissipation rate equation is employed according to Boussinesq
AN
104 eddy viscosity assumption. The corresponding Reynolds stress tensor, turbulent kinetic energy transport equation

105 and turbulent kinetic energy dissipation rate equation are shown as:

 ∂u ∂u j  2
M

− ρ ui''u ''j = ueff  i +  − ρ kδ ij

 ∂x
 j ∂xi  3
（4）
D

∂ ∂ ∂  ueff ∂k 
( βv ρ k ) + ( β j ρ u j k ) =  β j  + β v Pk − β v ρε
∂t ∂x j ∂x j  σ k ∂x j 
（5）
TE

∂ ∂ ∂  ueff ∂ε  ε2
( βv ρε ) + ( β j ρu jε ) = β
 j + β
 v εP − C β
2ε v ρ
∂t ∂x j ∂x j  σ ε ∂x j 
（6）
k
EP

106 where ui and uj represent the velocity component in i-direction and j-direction, respectively; k represents the

107 turbulent kinetic energy; ε represents turbulent kinetic energy dissipation rate; δij represents the stress tensor; Pk
C

108 and Pε represent the production of turbulent kinetic energy and the production of dissipation, respectively; C2ε is a
AC

109 constant, which is taken as 1.92; σk and σε represents Prandtl–Schmidt number of k and ε, which are taken as 1.0

110 and 1.3 respectively.

111

112 3.3 Model validation

113

114 The CFD software FLACS, which has been widely used in the oil and gas industry, has been used to simulate the

115 dispersion behavior and accumulation characteristics of blowout gas in this study. (Chris Savvides et al., 2001)
 ACCEPTED MANUSCRIPT
116 validated the reliability and accuracy of CFD software FLACS against experimental measurements of flammable

117 gas releases in an offshore module. A large number of scenarios with variations of leakage conditions, wind

118 conditions, perimeter confinement are studied numerically and experimentally. The comparison results have

119 shown that FLACS predict the dispersion behavior and accumulation characteristics of leaked gas with good

120 accuracy. (Dadashzadeh M et al., 2013) used FLACS to simulate the dispersion of complex blowout gas and the

121 subsequent explosion consequences against the “Deepwater Horizon” accident. The results of the study are in

PT
122 good agreement with the accident investigation results of the BP report, so it is feasible to use FLACS to study the

123 dispersion behavior of blowout gas with the different composition. Olav R. Hansen et al (2010) performed

RI
124 validation of FLACS against experimental data for LNG vapor dispersion, a total of 33 experiments were

125 simulated using FLACS, the comparison between simulation results and experimental data demonstrated that

SC
126 FLACS is an effective tool to simulate the dispersion of vapor from an LNG release. In a study conducted by

127 Prankul Middha, et al., (2013), validation of FLACS against hydrogen dispersion experiments was performed. A

128
U
range of different kinds of release conditions, including subsonic jets, sonic jets, impinging jets and liquid
AN
129 hydrogen releases, were considered. In general, the simulations results correlated well with experimental data.

130
M

131 4. Numerical modeling of an offshore drilling platform

D

132 4.1 Initial condition

TE

133 Several parameters, e.g. leak source parameters (leakage position, leakage direction, and leak rate), wind field

134 parameters (wind direction and wind speed) and the composition of the blowout gas etc., can affect the dispersion
EP

135 characteristics of blowout gas. In this study, the leak position is assumed to the wellhead while the leakage

136 direction is vertical upward. The leak rate is calculated according to the unimpeded flow of the well which is 1
C

137 million cubic meters per day. Furthermore, according to the distribution of crew in this platform, as seen in Table
AC

138 1, the influence of blowout gas on the living area which located in the very east side should be attached great

139 importance. Therefore, the west to east wind direction which blowing in the living area is adopted, and the annual

140 average wind speed 5.28m/s is adopted. In addition, since this study focuses the influence of the blowout gas

141 composition on the dispersion characteristics, varied compositions contain the hydrogen sulfide are set by

142 considering the composition of BP ‘Deepwater Horizon’ accident as the benchmark.

143

144 A total of 6 different scenarios including 6 compositions are accordingly simulated. Table 2 shows the

145 corresponding scenarios. From the Table 2, one can see the concentration of hydrogen sulfide is fixed to 1000ppm
 ACCEPTED MANUSCRIPT
146 while the composition and content of hydrocarbon gas vary with different scenarios. The gas composition of

147 blowout gas in scenario 1&2&3 is about light hydrocarbon gas, the gas composition of blowout gas in scenario

148 4&5 contains heavy hydrocarbon gas. And the composition and content of hydrocarbon gas in scenario 6 is from

149 the “Deepwater Horizon” accident. The composition and content of blowout hydrocarbon gas in BP “Deepwater

150 Horizon” accident are outlined in Table 3.

151

PT
152 Table 1 Drilling platform personnel distribution

Living Drilling Compressor Process Electric Aided Wellhead

Staff groups Number

RI
area area area area room process area area
Drilling 80 40 40 — — — — —
Production 10 9 0.2 0.2 0.2 0.2 — 0.2

SC
Maintenance 15 10.5 — 0.7 1.2 1.7 0.7 0.2
Management 5 4.1 0.2 0.2 0.2 0.1 — 0.2
Logistics 10 10 — — — — — —

U
Total 120 73.6 40.4 1.1 1.6 2 0.7 0.6
AN
153
154 Table 2 Leakage accident scenarios
Scenario Methane Ethane Propane Nonane Decane Carbon dioxide hydrogen sulfide
M

1 59.14% 20% 20% 0 0 0.76% 0.1%

2 79.14% 10% 10% 0 0 0.76% 0.1%
3 99.14% 0 0 0 0 0.76% 0.1%
D

4 79.14% 0 0 10% 10% 0.76% 0.1%

5 59.14% 0 0 20% 20% 0.76% 0.1%
TE

The composition and content of hydrocarbon is in accordance with

6 0.76% 0.1%
the BP “Deepwater Horizon” accident

155
EP

156 Table 3 Composition and content of hydrocarbon gas in Deepwater Horizon

Composition Methane Ethane Propane Butane Pentane Hexane Heptane Octane Nonane Decane
C

Concentration 57.18% 5.53% 3.85% 2.60% 1.62% 1.16% 1.68% 1.81% 1.33% 22.38%
AC

157

158 4.2 Grid model of the drilling platform

159

160 The pre-processor of FLACS-Computer Aided Scenario Design (CASD) is used to build the geometry of an actual

161 offshore drilling platform. The geometry mainly focuses on the spatial layout and external structure, the interior

162 cabin structure, pipe gallery and pipeline are simplified. The simulation dimensions are 90m × 110m × 65m in x, y

163 and z directions in the grid model, respectively. The computation domain adopts the non-reflecting boundary
 ACCEPTED MANUSCRIPT
164 condition and is divided into the core area and the expanding area. The core area mainly includes the upper deck

165 module of the platform, such as drilling floor, living area, etc., while the rest is located in the expanding area. The

166 1m grid of the core area is applied for the following dispersion simulations according to the criteria (grid size is

167 1m) given by FLACS manual (GexCon, 2015) and the grid sensitivity analysis in Section 5. The stretched factor

168 1.2 is adopted from the core domain to simulation domain boundary. Besides, “local grid refinements” around the

169 leak areas is performed in the non-leakage direction in order to improve calculation precision and calculation

PT
170 stability (GexCon, 2015). The whole drilling platform model and grid model are shown in Fig. 2.

RI
U SC
AN
Fig. 2. Offshore platform and corresponding grid model
M

171 5. Numerical simulation results and discussions

D

172 5.1 Effect of gas composition on the flammable cloud volume

TE

173 With the above numerical model, the effect of blowout composition on the flammable cloud volume is analyzed.

174 It should be noted that the flammable cloud incurred by the blowouts usually distributes non-homogeneously, it is
EP

175 difficult to quantitatively identify the volume of such flammable cloud. Accordingly, FLACS defines the

176 ‘Equivalent Stoichiometric Cloud’ (ESC) to represent such flammable cloud. The idea regarding the ESC is that
C

177 the large non-homogeneous gas cloud volume can be approximated by a smaller, more reactive, stoichiometric gas
AC

178 cloud, i.e. the ESC (Hansen O R et al., 2013; Qi X et al., 2017). The introduction of the ESC can linearize the

179 expected hazards from arbitrary non-homogeneous and dispersed flammable gas clouds. The precision of the ESC

180 is much better than alternative simplifications, e.g. some faster and less accurate consequence models. In this

181 paper, the most widely used Q9 (Gupta S and Chan S, 2016; Li J et al., 2017) and relatively conservative FLAM

182 (Kees van Wingerden and Nicolas Salaun, 2016) are chosen to evaluate the hazards of flammable gas clouds.

183

Q9 ( m3 ) = ∑VolumeFuel × ( SE ) / ( S × E )max (7)

 ACCEPTED MANUSCRIPT
FLAM ( m3 ) = ∑VolumeLFL< Fuel <UFL (8)

184 where ∑VolumeFuel represents all flammable volumes, S represents laminar burning velocity for the actual

185 concentration, E represents the volume expansion of the actual mixture. The FLAM takes the entire flammable

186 gas volume as the equivalent cloud.

187

188 Before analyzing the effect of the gas composition on dispersion characteristics of blowout gas, the grid sensitivity

PT
189 analysis on the ESC is conducted. This part takes the influence of varied grid sizes on the Q9 under scenario 6 as

RI
190 the example to illustrate. Fig. 3 demonstrates the corresponding results. As can be seen, the Q9 history curve

191 under grid size 1m is closer to that under 0.75m compared to that under 1.2m and 1.5m. That means the Q9

SC
192 history curve converges with decreasing the grid size. Furthermore, the grid size 1m is used for the following

193 dispersion analysis in order to keep the balance between the computation costs and result accuracy, The 1m grid

U
194 can also meet the criteria given by FLACS manual (GexCon, 2015), which ensures the reasonability of the
AN
195 following dispersion results.

196
M
D
TE
C EP
AC

Fig. 3. The variations of Q9 with different grid sizes

197

198 Fig.4 shows the variation of the ESC (FLAM, Q9) formed by the blowout gas in different scenarios. The ESC

199 increased rapidly at the initial stage and then reached a relatively stable state after a period of fluctuation. As is

200 shown in Fig.4, the ESC varies significantly in different scenarios. And the ESC of steady state in different

201 scenarios, from big to small, are scenario 5, scenario 6, scenario 4, scenario 1, scenario 2, and scenario 3. The

202 maximum value of the ESC (FLAM, Q9) in varied scenarios are shown in Fig.5. Through the data extracted from
 ACCEPTED MANUSCRIPT
203 Fig.4 and Fig.5, the ESC (FLAM/Q9) formed by dispersion of blowout gas increases with the increase of heavy

204 composition in blowout gas. The main reason for this result is that the more the heavy composition is, the greater

205 the density of the blowout gas is, the blowout gas is prone to gather easier owing to larger gravity. Moreover, the

206 heavy composition gas increases the frictional resistance, leading to the decrease of the kinetic energy of light

207 composition gas, which further conducive to flammable gas accumulation. Comparing scenario 1 (scenario 2,

208 scenario 3) with scenario 6, the equivalent cloud volume of scenario 6 is larger than that of the scenario 1

PT
209 (scenario 2, scenario 3). Therefore, simplifying the blowout gas into light composition gas will cause the

210 evaluation result of flammable cloud seriously underestimated when involved in blowout accident of gas field

RI
211 with complex composition.

SC
Scenario 1 Scenario 2 Scenario 1 Scenario 2
Scenario 3 Scenario 4 Scenario 3 Scenario 4
Scenario 5 Scenario 6 Scenario 5 Scenario 6
2000 500

U
400
1500
AN
Volume/m3
Volume/m3

300
1000
200
M

500
100

0 0
D

0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
Time/s Time/s
TE

(a) Variation of FLAM with different scenarios (b) Variation of Q9 with different scenarios
212 Fig. 4. Variation of FLAM and Q9 with different scenarios
EP

1800

1600 FLAM Q9
1400
C

1200
Volume/m3

AC

1000

800

600

400

200

0
Scenario 1 Scenario 2 Scenario 3 Scenario 4 Scenario 5 Scenario 6

Fig. 5. Maximum FLAM and Q9 with different scenarios

 ACCEPTED MANUSCRIPT
213 5.2 Effect of gas composition on the influenced area of flammable cloud
214 To study the influenced area of the flammable cloud in different scenarios, it is necessary to make clear that the

215 explosion limits vary with different scenarios. So a unified concentration range cannot be used to monitor the

216 dispersion range of hydrocarbon gas with different scenarios. In this part, the explosive limits are utilized as the

217 lower and upper limits of monitoring intervals for different scenarios. And the explosive limits of blowout gas are

218 calculated according to Lechteilier Principle (seen in Equation 9).

PT
L = 1（Y1 L1 + Y2 L2 + ⋅⋅⋅ + Yn Ln ） （9）

RI
219 where L represents the explosion limit of blowout gas, n represents gas categories, Yn represents the volume

220 fraction of gas n, Ln represents the explosion limit of gas n.

SC
221

222 Table 4 shows the explosion limits of different hydrocarbon gas, and the calculated explosion limits of the mixed

U
223 gas in each scenario are summarized in Table 5. The distribution of hydrocarbon gas within explosion limits of
AN
224 each scenario is shown in Fig.6. As shown in Fig.6, the influenced area of hydrocarbon gas within explosion limits

225 in different scenarios, from big to small, are scenario 5, scenario 6, scenario 4, scenario 1, scenario 2, and scenario
M

226 3. This phenomenon is also consistent with the distribution rule of the ESC (FLAM, Q9) in 5.1. Although

227 spreading towards the downwind area of the wellhead, the blowout hydrocarbon gas of each scenario is mainly
D

228 distributed in the upper space of the drilling floor. Therefore, a violent explosion accident may be triggered once
TE

229 an ignition source shows up, which endangers the crew and equipment in drilling rig area.

230 Table 4 Explosion limit of different hydrocarbon gas

EP

Gas composition Explosion limit Gas composition Explosion limit

Methane 5%~15% Hexane 1.1%~7.5%
Ethane 3%~15.5% Heptane 1.1%~6.7%
C

Propane 2.1%~9.5% Octane 1%~6.5%

Butane 1.9%~8.5% Nonane 0.7%~5.6%
AC

Pentane 1.4%~7.8% Decane 0.6%~5.5%

231

232 Table 5 Explosion limits of different scenarios

Scenario 1 Scenario 2 Scenario 3 Scenario 4 Scenario 5 Scenario 6

Explosion limit 3.55%~13.52% 4.15%~14.22% 5%~15% 2.13%~11.19% 1.35%~8.92% 1.63%~9.92%

233
 ACCEPTED MANUSCRIPT

(a) Scenario 1 (b) Scenario 2

PT
RI
U SC
(c) Scenario 3 (d) Scenario 4
AN
M
D
TE

(e) Scenario 5 (f) Scenario 6

Fig. 6. Spatial dimension of hydrocarbon gas within the explosion limit with different scenarios
EP

234

235 5.3 Effect of gas composition on the influenced area of hydrogen sulfide
C

236 The leakage and dispersion of hydrogen sulfide threaten the oil and gas industry seriously, especially in offshore
AC

237 exploration and development operations, the limited space of the platform and the difficulty of emergency rescue

238 increase the severity of the blowout accident. Being exposed to 30 mg/m3 of hydrogen sulfide more than 1h, the

239 respiratory tract and eyes of people will be injured. While being exposed to 150 mg/m3 of hydrogen sulfide, it will

240 cause irreversible effects on the human body. According to the threshold mass concentration (15mg/m3),

241 safety-critical mass concentration (30mg/m3) and dangerous critical mass concentration (150mg/m3) of hydrogen

242 sulfide, the hydrogen sulfide dispersion region is divided into the safety zone, warning area, dangerous area and

243 seriously dangerous area.

 ACCEPTED MANUSCRIPT
244

245 Fig.7 shows the distribution of hydrogen sulfide in the stable state with different scenarios. The dispersion range

246 of hydrogen sulfide in scenario 1& 2&3 is larger than that in scenario 4&5&6. The distance between the far end of

247 gas cloud and the leakage source is defined as dispersion distance. The vertical dispersion distance and horizontal

248 dispersion distance of hydrogen sulfide with different scenarios are shown in Fig.8, in which safety critical mass

249 concentration (30mg/m3) is adopted as the monitoring lower limit. As shown in Fig.8, the vertical dispersion

PT
250 distance of hydrogen sulfide (higher than 30mg/m3) does not change with the change of the gas composition

251 obviously while the horizontal dispersion distance decreases with the increase of heavy composition in blowout

RI
252 gas. And according to Fig.7, hydrogen sulfide with high concentration is mainly gathered in the drilling floor.

253 Since there are many workers engaged in drilling operations in the drilling floor, adequate gas masks should be

SC
254 repaired in case of mass poisoning accident. In addition, the effect of hydrogen sulfide on living area needs to be

255 attached great importance. The distance between hydrogen sulfide and living area under different scenarios is

256
U
summarized in Table 6. The data of Table 6 indicates that the distance between hydrogen sulfide cloud and living
AN
257 area decreases with the decrease of heavy composition in blowout gas. And according to the reference by Yang D

258 et al., (2017), the vertical dispersion distance of hydrogen sulfide decreases with the increase of wind speed.
M

259 Therefore, under the high-speed wind coming from the stern, the hydrogen sulfide in blowout gas without heavy

260 composition has the relatively low dispersion height and is easy to gather in the upper of the living area. Under the
D

261 action of forced ventilation at the top of the living area, the hydrogen sulfide is easily spread into the living area,
TE

262 which increases the possibility of poisoning accidents in the living area. And in scenario 6, the hydrogen sulfide

263 rarely spreads into living area in such wind direction and wind speed. Therefore, when assessing the toxic risk of
EP

264 blowout accident of sour gas field, the assessment results will be relatively conservative if simplifying the

265 blowout gas into light composition gas.

C
AC

(a) Scenario 1 (b) Scenario 2

 ACCEPTED MANUSCRIPT

(c) Scenario 3 (d) Scenario 4

PT
RI
U SC
(e) Scenario 5 (f) Scenario 6
AN
Fig. 7. Steady Spatial dimension of H2S with different scenarios

266
M

Horizontal direction Vertical direction

30
D

25

20
TE

15

10
EP

0
Scenario 1 Scenario 2 Scenario 3 Scenario 4 Scenario 5 Scenario 6
C

Fig. 8. Dispersion distance of H2S above 30mg/m3

AC

267
268 Table 6 Distance between H2S and living area with different scenarios
Scenario Scenario 1 Scenario 2 Scenario 3 Scenario 4 Scenario 5 Scenario 6
Distance/m 1.5 0 0 2.5 7.5 5.5
269

270 5.5 Effect of gas composition on the critical time of hydrogen sulfide spreading to the

271 living area

272 According to the drilling platform personnel distribution, the crew is mainly concentrated in the living area. The
 ACCEPTED MANUSCRIPT
273 critical time of the hydrogen sulfide spreading to the living areas is the significant issue since it can support the

274 emergency response program in time. Therefore, the effect of the gas composition on the critical time of hydrogen

275 sulfide spreading to the living area is analyzed in this part.

276

277 For the purpose of implementing the emergency response program in time, the detection and alarm instrument of

278 hydrogen sulfide on drilling platform should be activated when the concentration of hydrogen sulfide reaches 20%

PT
279 of threshold mass concentration. So 3mg/m3 is chosen as monitoring lower limit to study the time of hydrogen

280 sulfide spreading to the living area. The critical time required for monitoring hydrogen sulfide with different

RI
281 scenarios is summarized in Table 7. The detection and alarm instrument of hydrogen sulfide on the top of the

282 living area can monitor hydrogen sulfide in all scenarios except scenario 5. And the more the heavy composition

SC
283 in blowout gas, the longer the time of hydrogen sulfide being monitored. For example, the time that hydrogen

284 sulfide being monitored for scenario 3 in which methane accounts for a large proportion of blowout gas is 16s.

285
U
When the composition and content of blowout gas being in accordance with the blowout gas in “Deepwater
AN
286 Horizon” accident, the time that hydrogen sulfide being monitored is 30s. Therefore, when assessing the

287 poisoning risk of blowout accident of sour gas field with complex composition, it may cause a relatively
M

288 conservative result if simplifying the blowout gas into light composition gas. But the conservative results are

289 beneficial for the design of prevention and control barrier.

D

290 Table 7 Time need for H2S being monitored

TE

Scenario Scenario 1 Scenario 2 Scenario 3 Scenario 4 Scenario 5 Scenario 6

Time/s 18 17 16 19 — 30

291
EP

292 6. Conclusions
C

293 In this paper, taking an offshore drilling platform as the research object, the FLACS-based numerical model is

294 built, and several simulations by varying the compositions of blowout gas are conducted. The main purpose is to
AC

295 study the dispersion behavior and accumulation characteristics of blowout gas with different gas compositions.

296 The main conclusions are as follows:

297

298 (1) Gas composition affects the dispersion behavior and accumulation characteristics of the flammable cloud. It is

299 to say the influenced area of flammable cloud increases as the blowout gas becomes heavier. In addition, the

300 flammable cloud is mainly gathered in the upper space of the drilling floor.

301
 ACCEPTED MANUSCRIPT
302 (2) Gas composition affects the dispersion behavior and accumulation characteristics of hydrogen sulfide. As the

303 heavy composition in blowout gas decreases, the influenced area and high concentration area of hydrogen sulfide

304 increase while the dispersion height of hydrogen sulfide decreases, which reduces the critical time of the hydrogen

305 sulfide spreading to the living area.

306

307 (3) Overall, the detailed composition should be considered when assessing the consequence of the similar blowout

PT
308 accident. That is because simplifying the composition can make the consequences caused by the flammable cloud

309 underestimated while that by the hydrogen sulfide overestimated. Additionally, all the above results can give a

RI
310 reference for the decision making process in order to prevent and control the similar blowout accident.

311

SC
312 Acknowledgement

U
313 The authors gratefully acknowledge the financial support provided by the National Key R&D Program of China
AN
314 (No: 2017YFC0804501) and Graduate Student Innovation Projects of China University of Petroleum (No:

315 YCX2017054). Acknowledgements are also given to the valuable comments supplied by the anonymous

316 reviewers.
M

317 References
D

318 BP, 2010. Deepwater horizon accident investigation report. http://www.washingtonpost.com/wp-srv/politics/documents/Deepwater

TE

319 Horizon Accident Investigation Report Executive summary.pdf.

320 Zhu, Y., Chen, G. Analysis of H2 S dispersion in the congested working environments of offshore platform. Journal of Safety and
EP

321 Environment, 2009, 9(6): 140-143.

322 Liu, D., Wei, J. Modelling and simulation of continuous dense gas leakage for emergency response application. Journal of Loss
C

323 Prevention in the Process Industries, 2017, 48:14-20.

AC

324 S.R. Hanna., P.J. Drivas. Guidelines for use of vapor cloud dispersion models (Second Edition). New York: Center for Chemical

325 Process Safety, AIChE, 1996.

326 Zhang, J., Lei, D., Feng W. Analysis of Chemical Disasters Caused by Release of Hydrogen Sulfide-bearing Natural Gas. Procedia

327 Engineering, 2011, 26:1878-1890.

328 Dandrieux, A., Dusserre G., Ollivier J. Small scale field experiments of chlorine dispersion. Journal of Loss Prevention in the Process

329 Industries, 2002, 15(1):5-10.

330 Deng, H., Chen, G., Zhu, Y., Fu, J., Liu, D. Simulation and experiment of gas leakage and dispersion in complex topography. Journal

331 of China University of Petroleum, 2012, 36(1):122-126.

 ACCEPTED MANUSCRIPT
332 Fu, J., Zhao, Z., Chen, G., Luo, H., Zhang, B., Zhang, B., Ye, C., Zhu, Y. Influences of liquid pipeline flow and pressure on

333 small-hole leakage rate. Acta Petrolei Sinica, 2016, 37(02): 257-265.

334 Li, J., Ma, G., Abdel-Jawad, M., Huang, Y. Gas dispersion risk analysis of safety gap effect on the innovating FLNG vessel with a

335 cylindrical platform. Journal of Loss Prevention in the Process Industries, 2016, 40:304-316.

336 Zhang, J., Lei, D., Feng, W. An approach for estimating toxic releases of H2S-containing natural gas. Journal of Hazardous Materials,

337 264 (2014):350–362.

PT
338 Li, X., Zhou, R., Konovessis, D. CFD analysis of natural gas dispersion in engine room space based on multi-factor coupling. Ocean

339 Engineering, 2016, 111:524-532.

RI
340 Liu, K., Chen, G., Wei, C. Combustible gas diffusion law and hazardous area of FPSO. Acta Petrolei Sinica, 2015, 36(8): 1018-1028.

341 Li, X., Chen, G., Meng, H., Zhu, Y., Shi, J. Diffusion characteristics of blowout combustible gas in semi-submersible drilling

SC
342 platform in South China Sea. China Offshore Oil and Gas, 2015, 27(01): 111-115+120.

343 Tauseef, S.M, Rashtchian, D., Abase, S. A. CFD-based simulation of dense gas dispersion in presence of obstacles. Journal of Loss

344
U
Prevention in the Process Industries, 2011, 24(4):371-376.
AN
345 Li, X., Chen, G., Zhu, H., Xu, C. Gas dispersion and deflagration above sea from subsea release and its impact on offshore platform.

346 Ocean Engineering, 2018, 163, 157-168.

M

347 SAVVIDES, C., TAM, V., KINNEAR, D. Dispersion of fuel in offshore modules: comparison of predictions using FLACS and

348 full-scale experiments. // Proceedings of Major Hazards Offshore Conference. London: ERA Technology Ltd., 2001.
D

349 Middha, P., Hansen, O.R, Storvik, I.E. Validation of CFD-model for hydrogen dispersion. Journal of Loss Prevention in the Process
TE

350 Industries, 2009, 22(6):1034-1038.

351 Hansen, O.R, Gavelli, F., Ichard, M., Davis, S.G. Validation of FLACS against experimental data sets from the model evaluation
EP

352 database for LNG vapor dispersion. Journal of Loss Prevention in the Process Industries, 2010, 23(6):857-877.

353 Bleyer, A., Taveau, J., Djebaïli-Chaumeix, N., Paillard, C.E., Bentaïb, A. Comparison between FLACS explosion simulations and
C

354 experiments conducted in a PWR Steam Generator casemate scale down with hydrogen gradients. Nuclear Engineering &
AC

355 Design, 2012, 245(3):189-196.

356 Huser, A., Kvernvold, O. Explosion risk analysis - Development of a general method for gas dispersion analyses on offshore

357 platforms. Parallel Computational Fluid Dynamics, 2001:517-524.

358 Qiao, A., Zhang, S. Advanced CFD modeling on vapor dispersion and vapor cloud explosion. Journal of Loss Prevention in the

359 Process Industries, 2010, 23(6):843-848.

360 Li, J., Madhat Abdel-jawad, Ma, G. New correlation for vapor cloud explosion overpressure calculation at congested configurations.

361 Journal of Loss Prevention in the Process Industries, 31(2014):16-25.

 ACCEPTED MANUSCRIPT
362 Huang, Y., Ma, G., Li, J. Multi-level explosion risk analysis (MLERA) for accidental gas explosion events in super-large FLNG

363 facilities. Journal of Loss Prevention in the Process Industries, 45(2017): 242-254.

364 Shi, J., Khan, F., Zhu, Y., Li, J., Chen, G. Robust data-driven model to study dispersion of vapor cloud in offshore facility. Ocean

365 Engineering, 2018:98–110.

366 Shi, J., Li, J., Zhu, Y., Hao, H., Chen, G., Xie, B. A simplified statistic-based procedure for gas dispersion prediction of fixed offshore

367 platform. Process Safety & Environmental Protection, 2018, 114:48-63.

PT
368 Dadashzadeh, M., Abbassi, R., Khan, F., Hawboldt, K. Explosion modeling and analysis of BP Deepwater Horizon accident. Safety

369 Science, 2013, 57(8):150-160.

RI
370 GexCon, 2015. FLACS v10.4 User’s Manual, Norway.

371 Hansen, O.R, Gavelli, F., Davis, S.G., Middha, P. Equivalent cloud methods used for explosion risk and consequence studies. Journal

SC
372 of Loss Prevention in the Process Industries, 2013, 26(3): 511-527.

373 Qi, X., Wang, H., Song, X., Chen, G. Threshold of equivalent gas cloud size based on explosion load of control room. CIESC Journal,

374 2017, 68(12): 4857-4864.

U
AN
375 Gupta, S., Chan, S. A CFD based explosion risk analysis methodology using time varying release rates in dispersion simulations.

376 Journal of Loss Prevention in the Process Industries, 2016, 39:59-67.

M

377 Li, J., Ma, G., Hao, H., Huang, Y. Optimal blast wall layout design to mitigate gas dispersion and explosion on a cylindrical FLNG

378 platform. Journal of Loss Prevention in the Process Industries, 2017, 49: 481-492.
D

379 Kees van Wingerden, Nicolas Salaun. LNG-Engine Safety: Design of Protective Measures Using CFD. Chemical Engineering
TE

380 Transactions, 2016, 48: 31-36.

381 Yang, D., Chen, G., Shi, J. Research on dangerous region of H2S-containing natural gas diffusion resulting from offshore platform
EP

382 blowout. Journal of Safety Science and Technology, 2017, 13(08): 114-120.
C
AC
 ACCEPTED MANUSCRIPT

Effect of gas composition on dispersion characteristics of blowout gas on offshore platform

Dongdong Yang, Guoming Chen, Jihao Shi*, Xinhong Li

Centre for Offshore Engineering and Safety Technology, China University of Petroleum (East

China), No.66, Changjiang West Road, Qingdao, China

PT
Corresponding author: Jihao Shi. E-mail: shi_jihao@163.com (J. Shi)

RI
U SC
AN
M
D
TE
C EP
AC