mô phỏng số việc khai thác gas từ HG

Ứng dụng hữu ích của Gas hydrates trong ngành công nghiệp dầu khí. Cung cấp các thông tin cần thiết cho các kỹ sư dầu khí nhằm đáp ứng cho nhu cầu năng lượng không chỉ của riêng nước ta mà còn trên thế giới.

5

6

7 Jiaxin Suna, Fulong Ninga,⇑, Shi Lib, Ke Zhangb, Tianle Liua, Ling Zhanga, Guosheng Jianga,

Faculty of Engineering, China University of Geosciences, Wuhan 430074, China

10 b State Key Laboratory of Enhanced Oil Recovery, Research Institute of Petroleum Exploration and Development, Beijing 100083, China

11 c

Guangzhou Institute of Energy Conversion, Chinese Academy of Sciences, Guangzhou 510640, China

12 d

Qingdao Institute of Marine Geology, Ministry of Land and Resources, China

13

1 6 a r t i c l e i n f o

17 Article history:

18 Received 17 September 2014

19 Revised 19 July 2015

20 Accepted 5 August 2015

21 Available online xxxx

22 Keywords:

23 Gas hydrate

24 Depressurisation

25 Overburden

26 Underburden

27 Permeability

28 Numerical simulation

29

3 0

a b s t r a c t

31 Natural gas hydrates have been investigated as a potential resource for commercially producing gas since

32 the 1990s Based on the latest available data for the Shenhu area of the South China Sea (SH7), a practical

33 two-dimensional model has been constructed to investigate the gas production potential and the

distri-34 butions of different physical properties in alternating formations by selecting a proper perforated interval

35 favouring borehole stability and gas production The effects of overburden and underburden permeability

36

on gas production are intensively discussed The simulation results indicate that the initial hydrate

dis-37 sociation mainly occurs among the upper gas hydrate bearing-sediments (GHBS) with a high

permeabil-38 ity but that in the later period, it is mainly distributed among the bottom low permeability GHBS In

39 addition, an obvious hydrate re-formation can be observed in the middle GHBS, and the dilution effect

40

in the bottom low permeability GHBS is stronger than that in the upper space with high permeability

41

A comparative study showed that the GHBS in the Shenhu area with only one permeable burden

(over-42 burden or underburden) is not the most promising target for depressurisation

43

Ó 2015 Published by Elsevier Ltd

44 45 46

47 Introduction

48 Gas hydrates are ice-like crystalline clathrates that are formed

49 when small gas molecules (mainly hydrocarbon gases) come into

50 contact with water (host molecules) under specific

low-51 temperature and high-pressure conditions They are widely

dis-52 tributed in the permafrost on land and in the ridges of active and

53 passive continental margins in the seafloor (Sloan, 1998, 2003)

54 Because of the significant associations with resources (Milkov,

55 2004), environment and climate change (Hesselbo et al., 2000;

56 Maslin et al., 2003), submarine landslides (Maslin et al., 2004)

57 and the evolution of geological history (Wang et al., 2010), gas

58 hydrates have become a hot topic for current energy and earth

59 science research The exhaustion of traditional oil and gas

60 resources, combined with a continuous increase in consumption,

61 means that unconventional energy sources, such as natural gas

62 hydrates, are considered the most promising alternative energy

63 Klauda and Sandler (2005) stated that 74,000 Gt of methane is

64 trapped in gas hydrates within marine zones, which is three orders

65

of magnitude greater than the current worldwide conventional

66 natural gas reserves Consequently, the exploration and

exploita-67 tion of marine gas hydrates has become an emphasis of current

68 and future research

69 Gas production from hydrate reservoirs at present mainly

70 includes traditional depressurisation, thermal stimulation and

71 inhibitor injection (Moridis et al., 2004; Sloan, 1998) as well as

72 the new CO2 replacement (White et al., 2011) Depressurisation

73 involves lowering the pressure below the hydrate phase

equilib-74 rium pressure at the initial temperature to cause hydrate

dissocia-75 tion, and thermal stimulation involves heating the reservoirs above

76 the hydrate dissociation temperature to induce dissociation at the

77 prevailing pressure Injecting inhibitors (such as salts and alcohols)

78 shifts the hydration pressure and temperature equilibrium and

79 results in hydrate dissociation The production mechanism of CO2

80 replacement is the exchange of CO2in situ with methane molecules

81 within a methane hydrate structure, releasing the methane

82 Previous studies (Moridis and Reagan, 2007; Zhang et al., 2010)

83 have shown that the pure thermal dissociation method and

inhibi-84 tor method have relatively high costs and limited effectiveness

http://dx.doi.org/10.1016/j.juogr.2015.08.003

2213-3976/Ó 2015 Published by Elsevier Ltd.

⇑ Corresponding author.

E-mail address: nflzx@cug.edu.cn (F Ning).

Contents lists available atScienceDirect

Journal of Unconventional Oil and Gas Resources

j o u r n a l h o m e p a g e : w w w e l s e v i e r c o m / l o c a t e / j u o g r

Please cite this article in press as: Sun, J., et al Numerical simulation of gas production from hydrate-bearing sediments in the Shenhu area by

depressuris-85 Moreover, injecting inhibitors is likely to cause environmental

pol-86 lution In addition, based on the field trials on the north slope of

87 Alaska (the well named Ignik Sikumi #1) (David et al., 2013), the

88 efficiency of CO2/CH4 exchange is relatively low, and the CO2

89 hydrate formed around the well can reduce the effective

perme-90 ability, thereby decreasing gas production Therefore,

depressuri-91 sation is still the most effective method for long-term gas

92 production from hydrate reservoirs (Moridis et al., 2009a)

93 Because natural gas hydrates occur primarily in polar regions,

94 which are usually associated with onshore and offshore

per-95 mafrost, and in sediments of outer continental and insular margins

96 (Kvenvolden, 1993), conducting field trials carries a high cost

97 Therefore, numerical simulation is usually employed to investigate

98 gas production from hydrate reservoirs at this stage and to

evalu-99 ate the production potential (Li et al., 2010a,b; Moridis et al.,

100 2009a,b, 2011; Su et al., 2010; Zhang et al., 2010).Moridis et al

101 (2009a, 2011)performed a numerical simulation of gas production

102 from Ulleung Basin accumulations with impermeable confining

103 boundaries using conventional technology in vertical wells and

104 slao investigated the gas production potential from hydrate

depos-105 its in Mount Elbert by adopting horizontal wells and vertical wells

106 In addition, based on historical data and property adjustment to

107 match the numerical simulation, two-dimensional numerical

mod-108 els replicating the reservoirs of Mount Elbert were constructed by

109 Kurihara et al (2011) They forecasted the long-term production

110 performances of vertical wells in these reservoirs using the

meth-111 ods of depressurisation, a combination of depressurisation and

112 wellbore heating, and hot water huff and puff According to the

113 characteristics of gas hydrate bearing-sediments (GHBS) in the

114 Shenhu area of the South China Sea,Zhang et al (2010)established

115 a typical model of hydrate deposits to evaluate the production

116 potential and efficiency by means of depressurisation and thermal

117 stimulation using horizontal wells Based on the measurements of

118 drilling and logging from sites SH2, SH3 and SH7,Li et al (2010a,

119 2010b, 2011) investigated gas production from the Shenhu

120 hydrates by means of depressurisation and a combination of

121 depressurisation and thermal stimulation using different well

122 designs A single vertical well was simulated bySu et al (2010,

123 2012)to assess the potential of using the method of

depressurisa-124 tion and alternately producing fluid and injecting hot water (huff

125 and puff) The above research has properly evaluated the

exploita-126 tion potential of typical GHBS in the ocean and the permafrost;

127 however, the hydrate deposits are taken as a single homogenous

128 reservoir in their simulations, and the effects of the hydrate

forma-129 tion lithology distribution and saturation differences on practical

130 gas production are not fully considered.Myshakin et al (2012)

131 indicated that interbedded gas hydrate accumulations might be

132 preferable targets for gas production in comparison with massive

133 deposits Therefore, the effects of sediment lithology and

134 saturation differences should be considered Additionally, the

135 hydrates that occurring in both marine deposits and permafrost

136 are not homogeneous That is to say, an accurate prediction of

137 the gas production potential should consider various petrophysical

138 factors Here, we use available data from the in situ coring of site

139 SH7 and hydrate saturation estimated from pore water freshening

140 to construct a realistic two-dimensional model for hydrate

141 reservoirs by selecting a proper perforated interval favouring

142 borehole stability and gas production We then use this model to

143 investigate the production potential and the distributions of

144 different physical properties in alternating hydrate formations

145 using the TOUGH + HYDRATE (Moridis et al., 2008) numerical

146 simulation software that was developed by the Lawrence Berkeley

147 National Laboratory The effects of the overburden and

148 underburden permeabilities on gas production are also

149 investigated in detail

150 Simulation model

151 Background

152 The target zone is located in the southeast of the Shenhu

Under-153 water Sandy Bench area in the central part of the north slope of the

154 South China Sea, between the Xisha Trough and the Dongsha

155 Archipelago (Fig 1) The north slope of the South China Sea is a

156 passive continental margin in Cenozoic and rich of oil and gas

bear-157 ing basins The first Chinese expedition to drill gas hydrates,

158 GMGS-1, was undertaken in this area between April and June

159

2007 by Fugro and Geotek on behalf of the Guangzhou Marine

Geo-160 logical Survey (GMGS) and the Ministry of Land and Resources of

161 the PR China A total of eight sites were drilled and well-logged

162 during this project, with cores recovered at five of these sites,

163 including three sites with recovered gas hydrate samples (SH2,

164 SH3, and SH7) (Wu et al., 2007; Zhang et al., 2007) A core sample

165 analysis indicates the presence of gas hydrates at depths of

166 153–229 m beneath the seafloor, with thicknesses of 10–43 m

167 and porosities of 33–48%, in areas with water depths of

168 1108–1245 m (Nakai et al., 2007) These sI methane hydrates (i

169 e., the structures of the hydrate molecules are type I) with

170 26–48% saturation are disseminated throughout the sediment,

171 and the gas produced from these hydrates was originally derived

172 from microorganisms and consists of 96.1–99.82% methane In situ

173 measurements indicate a bottom-water temperature of 3.3–3.7°C,

174 with a geothermal gradient of 43–67.7°C km
1, corresponding to a

175 sea-bottom heat flow of 74.0–78.0 mW m
2 (average of

176 76.2 mW m
2)

177 Model construction

178 The simulations presented here are based on the GHBS at site

179 SH7, where the seafloor is at a water depth of 1108 m The GHBS

180

in this area is located155–177 m below the seafloor (mbsf) and

181 has a pore water salinity (mass fraction) of 3.05% An axisymmetric

182 cylinder with a radius of 200 m and a thickness of 82 m (i.e., the

183 thickness of the GHBS is 22 m, and the thickness of both the

over-184 burden and underburden layers is 30 m) is adopted for the model

185 domain Previous studies (Moridis and Reagan, 2007; Li et al.,

Fig 1 Location of site SH7 in the South China Sea ( Wu et al., 2009 ).

2 J Sun et al / Journal of Unconventional Oil and Gas Resources xxx (2015) xxx–xxx

Please cite this article in press as: Sun, J., et al Numerical simulation of gas production from hydrate-bearing sediments in the Shenhu area by

depressuris-186 2011) have indicated that the 30-m-thick overburden and

under-187 burden layers may be sufficient to simulate the boundary effects

188 of heat exchange and pressure propagation The borehole with a

189 radius of 0.1 m is located in the centre of the cylinder (Fig 2)

190 The top and bottom boundaries are designed as constant

tempera-191 ture and pressure boundaries The outer side of the model

192 (rmax= 200 m) is treated as a no-flow boundary

193 The drill core obtained from the field indicates that the lithology

194 varies with depth and that this area contains clay, sand clay and

195 silty clay (Fig 3a) Combining the hydrate saturation, which is

196 estimated based on the pore water freshening variation, with

197 depth (Fig 3b), the coexisting methane hydrates and water within

198 the in situ hydrate sediment can be subdivided into three layers

199 based on the lithology distribution from top to bottom, namely

200 GHBS1-4.4 m, GHBS2-9.5 m, and GHBS3-8.1 m, which have mean

201 hydrate saturation values of 0.38, 0.25, and 0.15, respectively

202 The wet thermal conductivity ks of the GHBS is taken to be

203 3.1 W m
1°C
1 Because of the difference in lithology, the dry

204 thermal conductivity kHs1, kHs2 and kHs3 of the corresponding

205 GHBS (GHBS1, GHBS2 and GHBS3) are taken to be 0.7, 0.8 and

206 1.0 W m
1°C
1, respectively The densityqof the GHBS is assumed

207

to be 2600 kg m
3, and GHBS1, GHBS2, and GHBS3 are assigned

208 porosities of 41%, 38%, and 45%, respectively, with corresponding

209 intrinsic permeabilities (K1, K2 and K3) of 7.5 10
14 (=75 mD),

210 2.0 10
14 (=20 mD) and 1.0 10
14m2 (=10 mD) (Su et al.,

211

2010; Li et al., 2011) Both the overburden and underburden are

212 assigned the same properties as the adjacent hydrate deposits by

213 considering the same formation lithology except hydrate

satura-214 tion The main modelling parameters and physical properties are

215 given inTable 1

216 The simulation uses a relative permeability model as follows

217 (Moridis et al., 2008):

218

krA¼ SA
SirA

1
SirA

;

krG¼ SG
SirG

1
SirA

;

krH¼ 0;

ð1Þ

220 221 where SirAis 0.30, SirGis 0.05, and n and nGare 3.572

222 This modelling also uses the following capillary pressure model

223 (Van Genuchten, 1980):

224

Pcap¼
PshðSÞ
1=k
1i1
k;

S¼ ðSA
S0

irAÞ

ðSmxA—S0irAÞ;

Pmax6 Pcap6 0;

ð2Þ

226 227 wherek is 0.45, S0

irAis 0.29, SmxAis 1.0, and Pmaxis 105Pa

228 The composite thermal conductivity model used in the

mod-229 elling is as follows (Moridis et al., 2005):

230

Fig 2 Schematic of the simulated Shenhu area hydrate deposits The deep yellow

area with a screen (155–159.4 mbsf) is the perforated interval (For interpretation

of the references to color in this figure legend, the reader is referred to the web

version of this article.)

(a) formation lithology (b) hydrate saturation

Fig 3 Formation lithology and hydrate saturation vs depth at site SH7 ( Nakai

et al., 2007 ).

Table 1

Main hydrate deposit properties and conditions at Site SH7.

Initial bottom temperature of GHBS 3 (T s ) 13.79 °C Grain density (q) 2600 kg/m 3

Initial bottom pressure of GHBS 3 (P s ) 13.15 MPa Geothermal gradient 43.653 K/km

Hydrate saturation in GHBS 1 (S H1 ) 0.38 Wet thermal conductivity (k s ) 3.1 W m
1°C
1

Pore water saturation in GHBS 1 (S A1 ) 0.62 Dry thermal conductivity (k Hs1 ) 0.7 W m
1°C
1

Hydrate saturation in GHBS 2 (S H2 ) 0.25 Dry thermal conductivity (k Hs2 ) 0.8 W m
1°C
1

Pore water saturation in GHBS 2 (S A2 ) 0.75 Dry thermal conductivity (k Hs3 ) 1.0 W m
1°C
1

Hydrate saturation in GHBS 3 (S H3 ) 0.15 Intrinsic permeability (K 1 ) 75  10
15 m 2

(=75 mD) Pore water saturation in GHBS 3 (S A3 ) 0.85 Intrinsic permeability (K 2 ) 20  10
15 m 2 (=20 mD)

Porosity of GHBS 1 (/ 1 ) 0.41 Intrinsic permeability (K 3 ) 10  10
15 m 2 (=10 mD)

Please cite this article in press as: Sun, J., et al Numerical simulation of gas production from hydrate-bearing sediments in the Shenhu area by

depressuris-kc¼ kHsþ ð ffiffiffiffiffiSA

p

þ ffiffiffiffiffiSH

p Þðks
kHsÞ þuSIkI: ð3Þ

232

233 There is no ice present in the South China Sea, and, as such, SIis

234 0

235 Domain discretisation

236 Fig 4shows the schematic of meshes employed in simulating

237 the gas production from the GHBS The model domain is

238 discretised into 19,688 (107 184) elements in a cylindrical

239 coordinate system (r, Z) with 19,473 active elements, and the rest

240 are assigned as boundary cells located on the top and bottom of

241 the model The scale of discretisation varies from fine

discretisa-242 tion (DZ = 0.1–0.2 m) along the Z axis in areas close to the hydrate

243 deposits and where the hydrate saturation changes to coarser

244 (DZ = 0.5–3 m) in other domains far from reservoirs, which is

ade-245 quate for accurate predictions (Moridis et al., 2007a) Considering

246 that most of the heat and mass transport, and phase change occurs

247 around the borehole, we increase the mesh grid density along the r

248 direction, which yields the grids that include 77,892 (19,473 4)

249 coupled equations that are solved simultaneously when the

250 equilibrium model of hydrate formation and dissociation is used

251 in the simulation

252 Initial conditions

253 Because the natural gas hydrates in the Shenhu area of the

254 South China Sea are distributed in poorly consolidated sediments

255 near the seafloor, pore water in the sediments could be considered

256 to exchange with the sea-bottom water, which means that the

sed-257 iment pore water pressure is hydrostatic (Hyndman et al., 1992)

258 Then, the following empirical formula can be used to calculate

259 the initial hydrostatic pore water pressure (Song et al., 2002):

260

Ppw¼ Patmþqswgðh þ zÞ  10
6; ð4Þ

262

263 where PPwis the hydrostatic pore water pressure in MPa, Patmis the

264 standard atmospheric pressure of 0.101325 MPa, h is the water

265 depth in m, z is the depth of the sediment from the seafloor in m,

266 g is the acceleration due to gravity in m s
2, andqswis the average

267 sea water density in kg m
3; this last term is a function of water

268 depth, temperature, and salinity and can be assumed to be

269

1035 kg m
3 (Li et al., 2010) The water depth at site SH7 is

270

1108 m, then the pressure distribution of the entire system,

271 including the pressure Ps(at Z = 177 m), can be determined The

272 corresponding phase equilibrium temperature (approximately

273 13.80°C) at the bottom of the hydrate reservoirs can de deduced

274

by the hydrate pressure–temperature (P–T) equilibrium curve and

275 then compared with in situ temperature measurements

276 (Fig 5and 1416°C) To ensure the stability of the gas hydrates, a

277 slight adjustment of temperature at the bottom of GHBS3is carried

278 out (Table 1 and 1379°C) This adjustment, combined with the

279 known geothermal gradient of the GHBS listed inTable 1, means

280 that the initial temperatures at the top and bottom boundaries of

281 the model can be determined In the practical process of

initialisa-282 tion, the temperature and pressure distributions of the entire model

283 domain are calculated quickly using the self-balancing function of

284 the software when obtaining the temperatures and pressures at

285 the top and bottom boundaries

286 Well design and production method

287 Although previous studies (Moridis, 2008; Moridis et al., 2011)

288 suggest that using horizontal wells can greatly increase the gas

289 production for Class 2 and 3 reservoirs, it is still low in absolute

290 terms, and the use of a horizontal well substantially increases

291 the cost of installation and operation Furthermore, the mechanical

292 strength of the hydrate formation is relatively low and will

con-293 tinue to decrease with hydrate dissociation As a result, the

insta-294 bility of the borehole can lead to the subsidence and even the

295 collapse of the production platform and will affect the exploitation

296 under the condition of high pressure drawdown (Rutqvist et al.,

297

2008, 2009) Therefore, gas production from hydrate reservoirs

298

by traditional vertical wells is still the preferred alternative

How-299 ever, the borehole stability in drilling and producing must still be

300 emphasised when the vertical well design is employed (Ning,

301

2012; Yamamoto et al., 2014) According to the sediment lithology

302 (Fig 3a), the optimal perforated interval is set in sandy clay with a

303 relatively high mechanical strength and good permeability, which

304

is conducive to maintaining the borehole stability and the

dissoci-305 ated gas flowing into the production well, as shown inFig 6 To

Fig 4 Schematic of study area and meshing structure The rhombic-shaped grid in

the borehole is the perforated interval (155–159.4 mbsf) Fig 5 Temperature measured at site SH7 ( Nakai et al., 2007 ).

4 J Sun et al / Journal of Unconventional Oil and Gas Resources xxx (2015) xxx–xxx

Please cite this article in press as: Sun, J., et al Numerical simulation of gas production from hydrate-bearing sediments in the Shenhu area by

depressuris-306 avoid the theoretically correct but computationally intensive

solu-307 tion of the Navier–Stokes equation, the borehole flow is assumed

308 to be Darcian flow through a pseudo-medium describing the

inte-309 rior of the well (Moridis et al., 2007b) In the vertical well, the

310 pseudo-medium has a porosity/ = 1.0, a very high axial

perme-311 ability kZ= 5.0 10
9m2along the Z direction, a radial

permeabil-312 ity kr= 1.0 10
11m2, and a capillary pressure Pc= 0

313 As mentioned above, depressurisation is still the most

promis-314 ing method for hydrate exploitation Hence, a constant bottomhole

315 pressure (3 MPa) is adopted for production in the simulation The

316 pressure at the perforated interval is reduced by nearly 9.92 MPa,

317 which is approximately 76% of the initial pressure (12.92 MPa)

318 The constant bottomhole pressure production is applicable to most

319 hydrate formations with different permeabilities and is uniquely

320 suited to allow the gas production rate to increase to match the

321 increasing permeability (Li et al., 2011) In addition, this method

322 is beneficial to controlling the borehole pressure (for example,

323 the well pressure is higher than the pressure at the quadruple

324 point) to eliminate the possibility of secondary hydrate, and even

325 ice formation due to the temperature decrease

326 Results and analysis

327 Gas and water production

328 Fig 7 shows the evolution of the volumetric rates (a) QGof

329 methane in the gas phase produced at the well, (b) QTof total

330 methane produced, and (c) QRof methane released from hydrate

331 reservoirs in the whole domain AsFig 7 shows, both the total

332 methane production rate (QT) and hydrate dissociation rate (QR)

333 decrease sharply, and the former is less than the latter at the

begin-334 ning of the production (approximately t = 0–500 days) However,

335 both of them show a slight reduction in the later period (after

336 500 days) and tend to be consistent and stable According to the

337 free methane rate (QG) variation with time, an obvious methane

338 flow in the gas phase can be observed during the initial 50 days,

339 but its production rate decreases rapidly until no free methane

340 flows into the production well This phenomenon occurs mainly

341 because the pressure difference between the production well and

342 the formation is relatively large and because the pressure has not

343 propagated completely at the beginning of depressurisation; this

344 results in a higher pressure gradient between the production well

345 and the formation Therefore, the hydrate dissociation rate is fast

346 but is still far lower than the commercial production rate

347 (3.0 105ST m3/d) (Li et al., 2011) Almost all of the gas released

348 from the hydrate reservoirs at the beginning directly flows into

349 the production well instead of the permeable overburden because

350 the initial hydrate dissociation occurs mainly around the

perfo-351 rated interval and because the drive force caused by the

differen-352 tial pressure primarily influences this location Therefore, the

353 total production rate (QT) is relatively high, and a notable amount

354

of free gas is observed in the early production period The reasons

355 why the hydrate dissociation rate (QR) is higher than the total

356 methane production rate (QT) in the initial stage are likely that

357 (1) hydrate dissociation is endothermic, which results in the

358 increase of methane dissolving into the water; (2) hydrate deposits

359 are not confined, which means that some methane released is

dis-360 solved into the free-methane water coming from the overburden

361 and underburden However, the water due to the lag flow cannot

362 run into the well immediately; and (3) some residual free methane

363 released from the hydrate remains in the formation Because both

364 the overburden and underburden are permeable in the model, the

365 pressure gradient between the formation and the production well

366 decreases with persistent hydrate dissociation and rapid pressure

367 diffusion In addition, the hydrate dissociation itself is an

endother-368 mic process These two reasons cause the hydrate dissociation rate

369

to decrease to a large extent, and the rate then presents a notable

370 decline Certainly, the methane recovery rate (QT) in the well also

371 decreases With the gradual decrease in the pressure gradient,

372 the dissociation rate of the reservoir also reaches a lower level

373 Some of the gas released dissolves in the formation water, and

374 some dissociated gas far away from the production well may

375 escape and flow into the permeable overburden because of

buoy-376 ancy (the density of methane is lower than water) in the latter

per-377 iod This can explain why there is no free gas flowing into the

378 production well after 50 days in the simulation In addition, this

379 fact seems to indicate no continuous free gas can be produced at

380 all in an open system, regardless of how good the quality of the

381 hydrate formation is However, in the later period of production,

382 the total trapped methane comes from dissolved methane instead

383

of free gas and that dissolved in the connate formation water,

384 which leads to the total methane production rate (QT) being

385 slightly higher than the dissociation rate of hydrate formation

386 (QR) in the later stage

387

Figs 8 and 9show the total gas produced (VP) at the well, the

388 cumulative gas released (VD), the cumulative free gas (VR)

remain-389 ing in the reservoirs, the evolution of the water production rate

390 (QW) and the gas-to-water ratio (RGW) under the corresponding

391 conditions.Fig 8indicates that the total gas produced in the well

392 and the cumulative gas released continuously increase over time

393

A similar phenomenon that the cumulative gas released is higher

Fig 6 Production well design for hydrate reservoirs in the Shenhu area.

0 200 400 600 800 1000

0 200 400 600 800

Q T

Q G

Q R

3 /d)

t (d)

Q T Q G Q R

Fig 7 Volumetric rate of the total gas production (Q T ) and methane production in the gas phase (Q G ) in the well and released from hydrate dissociation (Q R ) using the constant bottomhole pressure method.

Please cite this article in press as: Sun, J., et al Numerical simulation of gas production from hydrate-bearing sediments in the Shenhu area by

depressuris-394 than the total gas produced in the early stage (approximately t = 0–

395 1800 days) can be observed, whereas the trend reverses in the later

396 stages However, the total gas trapped over 20 years is only

397 approximately 1.3 106ST m3, which indicates very low gas

398 recovery The free gas remaining in the reservoirs increases before

399 starting to drop, but compared with the total methane released,

400 the amount of gas trapped is also lower This fact again shows that

401 the trapped gas comes primarily from the dissolved methane in the

402 water in the later period of production The water production rate

403 (QW) is shown inFig 9; it increases quickly at the beginning until it

404 reaches a relatively high rate (4.0 105m3/d) and then stays

con-405 stant The gas-to-water ratio (RGW= VP/VW) is considered a relative

406 criterion for evaluating the efficiency of production, which is

407 mainly used to economically assess the production potential

408 According to Fig 9, RGWdrops sharply over time and is close to

409 0.5 ST m3of CH4/m3of H2O In conclusion, using this method is still

410 not economically feasibile

411 Spatial distributions of SHand SG

412 The dynamic evolution of hydrate and gas during production

413 can be determined by comparing the hydrate and gas spatial

distri-414 butions at different times.Figs 10 and 11show the evolution of the

415 SHand SGdistributions, respectively, 155–177 m below the seafloor

416 at different times (1 year, 5 years, 10 years and 15 years)

417 According toFig 10, the initial hydrate dissociation occurs

pref-418 erentially around the perforated interval and then gradually moves

419 forward in the GHBS Comparing the simulation results at different

420 moments, we can find that the movement rate of the dissociation

421 front in the upper high permeable GHBS (GHBS1) is relatively rapid

422

in the early period (1 year) Subsequently, its movement rate slows

423 down Then, the hydrate decomposition in the large-scale zone is

424 mainly located in the bottom low permeability formation,

espe-425 cially the interface between the GHBS (GHBS3) and the

underbur-426 den (177 m), resulting in the emergence of a lower dissociation

427 interface Hydrate dissociation in the bottom low permeability

for-428 mation is continuous, and the movement rate of the dissociation

429 front is significantly faster than the upper high permeability

for-430 mation with sustainable exploitation, which causes the mergence

431

of the lower dissociation interface and the cylindrical interface

432 around the perforated interval When the interfaces are completely

433 melded, the methane produced at the well is inferred mainly from

434 the bottom low permeability GHBS (GHBS2 and GHBS3) because

435 the hydrate decomposition zone in the lower part with low

perme-436 ability is significantly larger than in the upper part with high

per-437 meability in the later period (10 and 15 years) No obvious

438

‘‘secondary hydrate” is formed in the dissociation front before

439 the mergence of the two dissociation interfaces, although it can

440

be observed in the dissociation front of the middle GHBS2,

Fig 10 Evolution of spatial distribution of S H

Fig 11 Evolution of spatial distribution of S G

0.0

t (d)

V P V D V R

Fig 8 Volumes of total gas production (V P ) at the well, methane released from

hydrate dissociation (V D ), and cumulative free gas (V R ) remaining in the reservoirs

using the constant bottomhole pressure method.

0.0

t (d)

Q W

3 /d

0 1 2 3 4 5

4 / m

2 O)

Fig 9 Evolution of volumetric rate of water production (Q W ) and gas-to-water ratio

(R GW ).

6 J Sun et al / Journal of Unconventional Oil and Gas Resources xxx (2015) xxx–xxx

Please cite this article in press as: Sun, J., et al Numerical simulation of gas production from hydrate-bearing sediments in the Shenhu area by

depressuris-441 especially in the interface between the upper GHBS1and the

mid-442 dle GHBS2 in the later production stages The phenomenon

443 described above occurs because the upper hydrate formation with

444 high permeability is closer to the perforated interval

Conse-445 quently, the initial hydrate dissociation occurs primarily in the

446 upper part However, because the permeability of the upper

447 hydrate formation is better than that of the others and because

448 the overburden is permeable, the fluid can move rapidly to balance

449 the variation of the formation pressure at the later stage As a

450 result, the pressure gradient between the production well and

451 the upper hydrate formation decreases dramatically Furthermore,

452 because of the high hydrate saturation in the upper formation, the

453 effective permeability decreases to some extent and goes against

454 the fluid flow in GHBS1, which causes the late decomposition zone

455 to decrease significantly (because the pressure does not vary

visi-456 bly if the fluid cannot flow rapidly) Because there is an obvious

457 heat and mass transfer between the hydrate formation (GHBS3)

458 and the underburden, i.e., the fluid with a higher temperature in

459 the underburden will run into the hydrate formation, a second

dis-460 sociation interface will arise The large-scale hydrate dissociation

461 that can be observed in the bottom low permeability formation

462 occurs mainly for the following reasons: (a) the pressure

transmis-463 sion is relatively slow because of the low permeability and the

464 pressure in the bottom formation is originally higher than that at

465 the top; thus, the pressure gradient between the formation and

466 the production well is higher and causes significant hydrate

disso-467 ciation; and (b) hydrate dissociation is endothermic, and the fluid

468 with a high temperature below the hydrate formation is sufficient

469 to provide the heat required for decomposition Because the

470 hydrates in the bottom formation decompose rapidly, the

corre-471 sponding free gas generation is also obvious In addition, hydrate

472 dissociation can lead to the improvement of the formation effective

473 permeability such that a gas flow channel is formed in the

dissoci-474 ation front of the hydrate Meanwhile, most of the heat carried in

475 the fluid (the overburden and underburden) is consumed by the

476 upper (GHBS1) and the lower (GHBS3) hydrate formation; thus,

477 there is little heat obtained by the middle hydrate formation

478 (GHBS2), adding absorption of heat in dissociation, which leads to

479 a relatively low temperature in the front of GHBS2 When the gas

480 released underneath flows through the region, an evident

‘‘sec-481 ondary hydrate” can be formed because of the suitable pressure

482 and temperature Furthermore, because the hydrate saturation in

483 the upper formation is higher, the effective permeability is

signif-484 icantly low The migration gas accumulates here due to buoyancy

485 and results in a local pressure increase, which contributes to form

486 the ‘‘secondary hydrate”

487 Fig 11shows that gas saturation in the formation is relatively

488 low, with the highest saturation reaching only approximately 0.1,

489 which is similar to previous results (Li et al., 2011) In the early

490 stage of exploitation, the free gas with a high saturation is mainly

491 distributed in the vicinity of the production well, but the free gas

492 around the well drops significantly, thereby reducing the gas

satu-493 ration with the production process It can be observed that the gas

494 in the dissociation front of the bottom low permeability hydrate

495 formation continuously decreases in the later period This occurs

496 because the hydrate dissociation is faster in the early stage and

497 mainly gathers near the production well Because both the

over-498 burden and underburden are permeable, the pressure diffusion is

499 quick, and the effect of depressurisation thus decreases

signifi-500 cantly in the later stage Therefore, the amount of gas released in

501 both dissociation fronts drops accordingly On the one hand, the

502 decomposition zone in the bottom hydrate layer (GHBS3) is wider

503 than that in the other two layers (GHBS1 and GHBS2) (Fig 10)

504 However, the GHBS1has relatively high hydrate saturation, even

505 for the secondary hydrate formation at the top of the lower

disso-506 ciation front This can greatly reduce the effective permeability,

507 and prevent gas migration from moving to the perforated interval,

508 resulting in free gas accumulation in the lower dissociation front

509

On the other hand, hydrates have completely dissociated around

510 the borehole (approximately 10 m), as shown in Fig 10, which

511 can provide flowable channels for fluid (because the effective

per-512 meability of the formation increases after hydrate dissociation)

513 Both the free gas and water from GHBS3flow into the production

514 well along these channels due to the driving force caused by the

515 differential pressure Therefore, the free gas mainly accumulates

516

in the lower dissociation front in GHBS2 under the influence of

517 both the density difference and differential pressure When flowing

518 into the perforated interval, the free gas that is accumulated in the

519 lower dissociation front dissolves in the flowing water because of

520 low saturation However, there is no barrier for the upper gas

521 released, so it can migrate into the overburden and directly

dis-522 solve in the water because of buoyancy Therefore, free gas can

523

be clearly observed in the dissociation front of the bottom

forma-524 tion, but not in the upper space

525 Spatial distribution of T

526

Fig 12 shows the evolution of the spatial temperature

527 distribution over time for the simulated 20 years, which also

528 indirectly reflects the decomposition of the hydrate formation

529 Based on Fig 12, the temperature around the dissociation front

530 decreases significantly because cooling occurs as the hydrate

531 dissociation and production proceed in the early stage of the

532 simulation An inversion of the geothermal gradient as a result of

533 dissociation-induced cooling in the GHBS can be observed within

534

365 days, which is mainly the consequence of the rapid

535 endothermic decomposition of the hydrate As production

536 proceeds, the dissociation front is always accompanied by a low

537 temperature However, the temperature ‘‘subsidence” is obvious

538

in the upper dissociation front, but not at the bottom This is

539 mainly attributed to the fluid in the underburden having a higher

540 temperature, which can relieve the effect of the endothermic

541 hydrate dissociation on the temperature distribution in the bottom

542 hydrate layer when it gradually flows into the GHBS, as opposed to

543 the fluid with a lower temperature in the upper layer This also

544 explains why hydrate dissociation zone in the bottom is wider than

545 that in the upper part In addition, these phenomena only appear

546 near the production well is because the hydrates around the well

547 have completely decomposed This significantly improves the

548 effective permeability, and the fluid within the overburden and

Fig 12 Evolution of spatial distribution of T.

Please cite this article in press as: Sun, J., et al Numerical simulation of gas production from hydrate-bearing sediments in the Shenhu area by

depressuris-549 underburden can flow fluently Most fluids of different sources and

550 temperatures, including the original free water in the formation

551 and the gas and water released from hydrate dissociation, converge

552 toward the well, which causes the confluence of the isotherm

553 Spatial distribution of Xi

554 Fig 13shows the distribution of the salt concentration over

555 time in the GHBS The dilution effect on the salinity in the

dissoci-556 ation front is obvious (Fig 13) because fresh water is released from

557 hydrate dissociation and reduces the water salinity in situ Unlike

558 the cases with impermeable boundaries, the variation of salinity

559 is not significant because of the permeable overburden and

under-560 burden (Moridis et al., 2009a) This finding is consistent with the

561 spatial distributions of the other physical properties The results

562 shown inFig 13can verify that the bottom hydrate formation with

563 low permeability decomposes more rapidly than the upper based

564 on the spatial distribution of the salinity This may also be related

565 to the formation permeability because salt water is produced

con-566 tinuously from the formation and because the bottom hydrate

for-567 mation with low permeability will prevent water flow from other

568 regions to significantly affect the salinity distribution The low

569 salinity distribution is not obvious because it is opposite the upper

570 hydrate formation

571 Effect of permeability in overburden and underburden

572 The distribution of the hydrates in in situ reservoirs is actually

573 very complicated, and both permeable and impermeable burdens

574 may occur A previous field trial in the Nankai Trough

575 (Yamamoto et al., 2014) indicated that a hydrate deposit with

576 one permeable underburden is still the most promising target

577 Therefore, four different cases are investigated in this study to

578 evaluate the production potential, as shown inTable 2 For the

con-579 venience of comparison, the absolute permeability of all permeable

580 burdens is set to 10 mD, and the other physical properties of the

581 formation remain unchanged The simulation results are shown

582 in the following figures over 4000 days

583 Fig 14shows that for the Shenhu area in Case 4, the hydrate

584 dissociation rate is the highest and it increases rapidly until it

585 reaches its maximum in the first 400 days Then, it shows a slight

586 decrease and slowly returns back to the former maximum in the

587 later period When the hydrate formation only has an impermeable

588 overburden (Case 2), the decomposition rate is far lower than that

589

in Case 4, but it is significantly higher than the other two cases (i.e.,

590 Cases 1 and 3) This can also be verified from the hydrate

dissoci-591 ation zones at t = 1825 days (Fig 15) In addition, the dissociation

592 rate of the hydrate formation in Case 3 is slightly higher than that

593 with both permeable burdens (Case 1) at the beginning, but they

594 tend to gradually become identical In short, for the Shenhu area

595

if there is only one permeable burden, the dissociation rate, which Fig 13 Evolution of spatial distribution of X i

Table 2 Four different cases are investigated in the simulations.

Conditions Case 1

(GHBS P )

Case 2 (GHBS LP )

Case 3 (GHBS UP )

Case 4 (GHBS IP ) Permeable

overburden

Permeable underburden

102

103

104

10-1

100

101

102

103

104

102

103

104

3 /d

t (d)

GHBSP GHBSUP GHBS

IP

Q G

3 /d

3 /d

Fig 14 Volumetric rates of total gas production (Q T ) and methane production in the gas phase (Q G ) from the well and that released from hydrate dissociation (Q R ) under conditions of different permeable burdens.

Fig 15 Evolution of spatial distribution of S H at t = 1825 days.

8 J Sun et al / Journal of Unconventional Oil and Gas Resources xxx (2015) xxx–xxx

Please cite this article in press as: Sun, J., et al Numerical simulation of gas production from hydrate-bearing sediments in the Shenhu area by

depressuris-596 is still far lower than the commercial production rate, cannot be

597 significantly increased This also verifies the previous research on

598 the significance of impermeable burdens (Moridis et al., 2009b)

599 The pressure can only be transferred in the GHBS when both

600 burdens are impermeable Therefore, hydrate dissociation occurs

601 in a large area (Figs 15 and 16), which causes the hydrate

decom-602 position rate to increase The slight decline observed in the later

603 period occurs because the pressure in the formation reduced

604 accordingly when using the depressurisation method and because

605 hydrate decomposition is an endothermic process, which results in

606 the decrease of the formation temperature When the reservoir

607 pressure is approximately equal to the equilibrium pressure at

608 the corresponding temperature, the hydrate dissociation rate is

609 significantly reduced With the later continuous heat transfer

610 between the formations, the decomposition rate increases again

611 and trends toward stability Presumably, the dissociation rate will

612 gradually decrease at the end of production because little hydrate

613 can decompose during the continuous depressurisation Thus, the

614 production in Case 2 is better than in the other two conditions

615 (i.e., Cases 1 and 3) because the pressure can only diffuse

down-616 ward and the released gas cannot escape (Fig 16) That is the rising

617 gas is trapped by the impermeable overburden and eventually

618 flows into the well as a result of the pressure driving force In

addi-619 tion, both the fluid pressure and temperature in the underburden

620 are higher than those of the overburden, which can accelerate

621 the hydrate decomposition, thereby increasing the hydrate

622 dissociation rate In contrast, the pressure cannot diffuse through

623 the bottom impermeable formation in Case 3 Thus, it will transfer

624 radially along the GHBS, thereby advancing the hydrate

dissocia-625 tion For Case 1, on the one hand, the fluid temperature is relatively

626 high in the underburden, which is conducive to heat conduction;

627 on the other hand, the convective heat transfer is significant, which

628 also accelerates hydrate dissociation From Fig 14, there is no

629 difference between Cases 1 and 3 in promoting the dissociation

630 rate of hydrate Because only a slight difference can be observed

631 initially, the dissociation rates of the hydrate and the cumulative

632 volumes of the dissociated hydrate tend to be the same (Figs 14

633 and 17) Therefore, both the cumulative volume of methane

634 produced in the well and the amount of hydrate dissociation in

635 Case 4 are much higher than in the other three cases (Figs 15

636 and 17) The production curve also presents a slight increase in

637 Case 2 at the beginning because of the ‘‘barrier effect” of the

638 overburden Once the upper and lower dissociation fronts merge

639 together, the free gas released and fresh water can flow into the

640 production well fluently (because hydrate dissociation will

641 increase the effective formation permeability) This is why the

pro-642 duction rate curve of methane, including free gas, initially

643 increases and then starts to decrease (Fig 14) Meanwhile, the

cor-644 responding water production rate curve presents an obvious

ten-645 dency to increase, which can also be observed in Case 1 (Fig 18);

646 this is similar to previous research results (Su et al., 2010)

647 Based on the foregoing analysis, the hydrate formation in Case 4

648 decomposes quickly and the dissociated gas cannot escape due to

649 the ‘‘barrier effect” Therefore, more free gas flows into the

produc-650 tion well than in the other situations, and it shows an increasing

651 trend In comparison with the other three cases, we find that the

652 free gas rate in Case 1 is significantly lower than that in Case 2 (this

653

is mainly due to the barrier effect) but is slightly higher than that

654 with only a permeable overburden (Case 3) In the later period,

655 there is no free gas flowing into the production well (Fig 14) In

656 addition, the total methane production rate shows almost the same

657 rule mentioned above under the condition of both permeable

Fig 16 Evolution of spatial distribution of S G at t = 1825 days.

103

104

105

106

107

104

105

106

107

102

104

106

108

t (d)

GHBSP GHBSUP GHBSLP GHBSIP

V RE

Fig 17 Volumes of total gas production (V P ) at the well, methane released from hydrate dissociation (V D ) and cumulative free gas (V R ) remaining in the reservoirs under conditions of different permeable burdens.

0.0 5.0x104 1.0x105 1.5x105 2.0x105 2.5x105 3.0x105 3.5x105

GHBSP GHBSUP GHBSLP GHBSIP

t (d)

0.1 1 10 100

Q W

3 /d

4 / m

2 O)

Fig 18 Evolution of the volumetric rate of water production (Q W ) and gas-to-water ratio (R GW ) under conditions of different permeable burdens.

Please cite this article in press as: Sun, J., et al Numerical simulation of gas production from hydrate-bearing sediments in the Shenhu area by

depressuris-658 burdens (Case 1) and is slightly lower than that of the formation in

659 Case 3 (Fig 14) The reason may be that the fluid temperature is

660 relatively higher in the underburden, which is not conducive to

661 free methane gas dissolution when flowing into the GHBS

Further-662 more, most free gas dissociated at the bottom will flow into the

663 production well along the dissociation front instead of escaping

664 (Fig 16) (the hydrate formation effective permeability has

signifi-665 cantly improved after dissociation) However, most of the free gas

666 will escape because of the long distance between the dissociation

667 front and the production well with the increase of the hydrate

dis-668 sociation area in Case 3 Judging from the remaining free gas in the

669 whole formation, the amount is still the largest in Case 4 (Fig 16)

670 The amount of free gas remaining in Case 3 is larger than that in

671 Case 2, followed by Case 1, where both the overburden and

under-672 burden are permeable (Figs 16 and 17) This can explain why more

673 free gas is produced in the well in Case 1 than in Case 3 (because

674 most of the free gas in Case 3 remains in the hydrate layers) A

rel-675 atively rapid dissociation can also be observed in Case 2 (Fig 14),

676 which is higher than that of the others (i.e., Cases 1 and 3)

677 With the combination of the water withdrawal and

678 gas-to-water ratio over time, as shown inFig 18, we find that a

679 gas hydrate formation with only one permeable burden is

con-680 ducive to reducing the yield of water but that the potential is still

681 much larger than that of a formation without permeable burdens

682 According toFig 18, the water withdrawal rate only increases

ini-683 tially in Case 4 before declining continuously to a lower value,

684 whereas the gas production rate gradually increases This is why

685 the gas-to-water ratio rapidly decreases at first and then gradually

686 increases to far exceed 10 ST m3of CH4/m3of H2O However, the

687 case in which the underburden is permeable is superior to that

688 of the permeable overburden, but they are both much lower than

689 the hydrate formation without permeable burdens When an

690 impermeable burden exists on top of the hydrate formation, RGW

691 can be improved to a certain extent, whereas it cannot be

692 improved in the case of a permeable overburden

693 Conclusions

694 This study used drilling and pore water freshening data from

695 site SH7 in the Shenhu area of the South China Sea to construct a

696 two-dimensional model that are more similar to reality than other

697 models (Li et al., 2010a, 2010b, 2011; Su et al., 2010, 2012; Zhang

698 et al., 2010) The production potential and the physical property

699 distributions in alternating hydrate formations during extraction

700 are analysed by numerical simulation The effects of burden

per-701 meability on gas production are also investigated in detail and

702 yield the following results:

703 (1) Under the condition of constant depressurisation, the total

704 methane production rate is slightly lower than the gas

705 release rate at first, but the situation reverses later as they

706 both trending toward stability The free gas flowing into

707 the production well is only evident at the beginning of

pro-708 duction In other words, no continuous free gas can be

pro-709 duced at all in an open system, regardless of how good the

710 quality of the hydrate formation is Because of the lack of

711 impermeable burdens, the production rate is very low as a

712 great amount of free water flows into the production well

713 and is accompanied by gas escape Therefore, the

investi-714 gated method is still not an effective way to exploit gas from

715 hydrate reservoirs in the Shenhu area

716 (2) The different physical property distributions in alternating

717 hydrate formations at different times shows that the initial

718 hydrate dissociation preferentially occurs around the

perfo-719 rated interval and then gradually spreads outward in the

720 GHBS According to the simulation results, the upper high

721 permeable GHBS (GHBS1) dissociates more rapidly in the

722 early period However, later on, the trend reverses, and an

723 obvious ‘‘secondary hydrate” can be observed in the

dissoci-724 ation front of the middle GHBS (GHBS2) In the early stage of

725 exploitation, the free gas is mainly distributed in the vicinity

726

of the production well, but in the later period, it can only be

727 observed in the dissociation front of the bottom low

perme-728 ability hydrate formation with decreasing saturation The

729 corresponding temperature decreases and geothermal

gradi-730 ent reversion can occur in the dissociation front In addition,

731 there is a significant dilution effect in the dissociation front,

732 and that of the bottom hydrate formation with low

perme-733 ability is obviously stronger than the upper high

permeabil-734 ity hydrate formation

735 (3) By comparing the effect of the different burden

permeabili-736 ties on gas production, we find that when there is only one

737 permeable burden (overburden or underburden) in the

738 hydrate formation in the Shenhu area, the production rate

739 still cannot be increased significantly by using a constant

740 bottom hole pressure However, a hydrate reservoir with a

741 permeable underburden is superior to that with only a

per-742 meable overburden or both permeable burdens based on the

743 simulation results

744

745 Acknowledgements

746 The authors would like to think Dr Matthew Reagan for

valu-747 able suggestions about our modelling This work was supported

748

by the National Natural Science Foundation of China (51274177),

749 the Program for New Century Excellent Talents in University

750 (NCET-13-1013), the Enhanced Oil Recovery State Key Laboratory

751 Project (2011A-1002), the Fok Ying Tong Education Foundation

752 (132019), a Key Project of the Natural Science Foundation of Hubei

753 Province (2012FFA047) and the Fundamental Research Funds for

754 the Central Universities (No 120112)

755 References

756 David Schoderbek, H.F., Keith, H., James, H., et al., 2013 ConocoPhillips gas hydrate

757 production test final technical report Oil & Natural Gas Technology.

758 Hesselbo, S.P., Gröcke, D.R., Jenkyns, H.C., et al., 2000 Massive dissociation of gas

759 hydrate during a Jurassic oceanic anoxic event Nature 406, 392–395.

760 Hyndman, R., Foucher, J., Yamano, M., et al., 1992 Deep sea

bottom-simulating-761 reflectors: calibration of the base of the hydrate stability field as used for heat

762 flow estimates Earth Planet Sci Lett 109, 289–301.

763 Klauda, J.B., Sandler, S.I., 2005 Global distribution of methane hydrate in ocean

764 sediment Energy Fuels 19, 459–470.

765 Kurihara, M., Sato, A., Funatsu, K., et al., 2011 Analysis of formation pressure test

766 results in the Mount Elbert methane hydrate reservoir through numerical

767 simulation Mar Pet Geol 28, 502–516.

768 Kvenvolden, K.A., 1993 Gas hydrates—geological perspective and global change.

769 Rev Geophys 31, 173–187.

770

Li, G., Moridis, G.J., Zhang, K., et al., 2010a Evaluation of gas production potential

771 from marine gas hydrate deposits in Shenhu area of South China Sea Energy

772 Fuels 24, 6018–6033.

773

Li, G., Li, X.S., Chen, Q., et al., 2010b Numerical simulation of gas production from

774 gas hydrate zone in Shenhu Area, South China Sea Acta Chim Sin 68, 1083–

775

1092 (in Chinese).

776

Li, G., Li, X.S., Zhang, K.N., et al., 2011 Numerical simulation of gas production from

777 hydrate accumulations using a single horizontal well in Shenhu Area, South

778 China Sea Chin J Geophys 54, 2325–2337 (in Chinese).

779 Maslin, M.A., Thomas, E., 2003 Balancing the deglacial global carbon budget: the

780 hydrate factor Quat Sci Rev 22, 1729–1736.

781 Maslin, M., Owen, M., Day, S., et al., 2004 Linking continental-slope failures and

782 climate change: testing the clathrate gun hypothesis Geology 32, 53–56.

783 Milkov, A.V., 2004 Global estimates of hydrate-bound gas in marine sediments:

784 how much is really out there? Earth Sci Rev 66, 183–197.

785 Moridis, G.J., 2008 The use of horizontal wells in gas production from hydrate

786 accumulations In: Proceeding of 6th International Conference on Gas Hydrates,

787 Vancouver, BC.

788 Moridis, G.J., Reagan, M.T., 2007 Strategies for gas production from oceanic Class 3

789 hydrate accumulations In: Proceedings of Offshore Technology Conference,

790 Houston, TX, USA.

10 J Sun et al / Journal of Unconventional Oil and Gas Resources xxx (2015) xxx–xxx

Please cite this article in press as: Sun, J., et al Numerical simulation of gas production from hydrate-bearing sediments in the Shenhu area by