Lei Yang,Yulong Liu,Hanquan Zhang,Bo Xiao,Xianwei Guo,Rupeng Wei,Lei Xu,Lingjie Sun,Bin Yu,Shudong Leng,Yanghui Li,*

1 Key Laboratory of Ocean Energy Utilization and Energy Conservation of Ministry of Education,Dalian University of Technology,Dalian 116024,China

2 Guangzhou Marine Geological Survey,Guangzhou 510075,China

3 China Ship Design&Research Center Co.,Ltd.,Dalian 116001,China

Keywords:Natural gas hydrate Production technique Depressurization Thermal stimulation CO2exchange

ABSTRACT Natural gas hydrate(NGH)has been widely considered as an alternative form of energy with huge potential,due to its tremendous reserves,cleanness and high energy density.Several countries involving Japan,Canada,India and China have launched national projects on the exploration and exploitation of gas hydrate resources.At the beginning of this century,an early trial production of hydrate resources was carried out in Mallik permafrost region,Canada.Japan has conducted the first field test from marine hydrates in 2013,followed by another trial in 2017.China also made its first trial production from marine hydrate sediments in 2017.Yet the low production efficiency,ice/hydrate regeneration,and sand problems are still commonly encountered；the worldwide progress is far before commercialization.Up to now,many gas production techniques have been proposed,and a few of them have been adopted in the field production tests.Nevertheless,hardly any method appears really promising；each of them shows limitations at certain conditions.Therefore,further efforts should be made on the economic efficiency as well as sustainability and environmental impacts.In this paper,the investigations on NGH exploitation techniques are comprehensively reviewed,involving depressurization,thermal stimulation,chemical inhibitor injection,CO2-CH4exchange,their combinations,and some novel techniques.The behavior of each method and its further potential in the field test are discussed.The advantages and limitations of laboratory studies are also analyzed.The work could give some guidance in the future formulation of exploitation scheme and evaluation of gas production behavior from hydrate reservoirs.

1.Introduction

Conventional fossil fuels satisfy approximately 85% of the energy needs around the whole world [1],the consumption of which is still continuously increasing.It is reported that the global energy demand will rise from 524 quadrillion British thermal unit(BTU)in 2010 to 820 quadrillion BTU in 2040[2].Conventional energy resources would not be able to fill this huge gap.Attentions are therefore put in the renewable energy involving the solar energy,wind energy,etc.[3-6].While the fact is that more than 76%of the energy is predicted to be supplied still by gas,oil and coal at 2040 with the growth of other renewable sources [7].This might well be a result of constantly discovered unconventional fossil fuels；natural gas hydrate plays a crucial role,which thereby attracts worldwide interest.

Natural gas hydrates are ice-like crystalline compounds formed by guest molecules,such as methane,and cages of framework called hosts[8,9].Worldwide attention is attracted by NGH for its wide existence in the sediments in seafloor and permafrost zones,containing approximately 10000 Gt of carbon [10-14],which is twice the total amount of carbon in proved fossil fuels all over the world[15,16].In addition,one volume of crystalline NGH could contain approximately 160 volume of methane at standard conditions [17],which makes NGH a potential material with high gas capacity.Gas hydrates were first reported by Davy in laboratory studies in 1811 [8],and first observed in oil and gas pipelines by Hammerschmidt et al.[18]until 1934.From 1960s,the research activity on NGH began to gain attention especially over the past two decades.Up to now,it is reported that over 80 countries have been involved in research activities on NGH,among which Canada,Japan,China,India,the USA,Korea,Norway and Germany have been very active recently[18-21].

The first field test worldwide was conducted in the Mallik permafrost region,Canada in 2002 through thermal stimulation[22-26].The reservoir is 1100 m deep and 650 m thick.The production lasted for 123.7 h with an accumulative gas production of 468 m3.Another trial was carried out in the same area in 2008 by depressurization；a production of 139 h and 13000 m3of gas was obtained.Significant ice generation and hydrate re-formation were encountered,resulting in a low gas production efficiency and retention of gas in the reservoir.The first trial production from marine hydrate sediments was conducted in the Nankai Trough,Japan in 2013 through depressurization[27,28].The target area was initially proposed in 1992；the occurrence of gas hydrate was confirmed based on the geophysical exploration in 1996 and well logging in 1999.In 2002,the“MH21 alliance”was established under the efforts of Japan National Oil Corporation (JNOC),Advanced Industrial Science and Technology (AIST)and Engineering Advancement Association of Japan (ENAA).32 exploration test wells were drilled along Tokai-oki and Kumano-nada area in 2004,with the depth ranging from 720 to 2033 m [29].The resource evaluation was carried out based on the well logging,core analysis and 2D seismic data.Then 3 zones of enrichment were selected according to intensive site survey on the reserves,field and reservoir conditions.The water depth is 1000 m,the 60 m thick hydrate reservoir is located 300 m below seafloor.The deep-sea drilling vessel“Chikyu”was used to conduct the field test[30].Gas and water were produced through marine riser pipe and blow-out preventer.An electrical submersible pump(ESP)was used to depressurize the borehole.The temperature in the production and monitoring wells were monitored in real-time；the pressure in the production well was measured at different locations.A complete Environmental Impact Assessment(EIA)system had been developed and monitoring equipments involving Methane Leakage Monitoring System(MLMS)and Seafloor Deformation Monitoring System (SDMS)were used for the environmental risk analysis,evaluation and management[31].On the 6th day of the test,severe sand production occurred,resulting in wellbore plugging and the termination of production.The production lasted for 6 days with a total production of 12000 m3.A GeoFoam technique was used in the second field test in 2017 to prevent sand problem；still lots of sands flowed into the production well and terminate gas production.A higher productivity was achieved (P1 well,12 days,35000 m3；P2 well,24 days,200000 m3)with depressurization.China also conducted its first field test from marine sediments in Shenhu area,South China Sea (SCS)in 2017 through depressurization [32].The water depth is 1266 m and the reservoirs are located 203-277 m below the seafloor.Three production wells and a monitoring well were drilled by“Blue Whale-I”,which is among the world's largest and most advanced ultra-deep-water semi-submersible drilling platform.The stepwise depressurization method was used in the test.The longest production so far of 60 days was reported,with an accumulative production of 300000 m3.The efficiency still requires improvements for commercialization.Another trial production was carried out in Liwan-3 area,SCS through a novel solid fluidization method.Further efforts are being made on the multi-phase flow of mixed particles-gas-liquid system and evaluation of environmental impacts.

NGHs are usually distributed in the porous sediments under seafloor and in permafrost regions,where high pressures and low temperatures are commonly coexisting.The conditions favorable for hydrate occurrence is majorly governed by the phase equilibrium of gas hydrate.The formation and accumulation of gas hydrate in the reservoir could play a significant role in the exploration and gas production process.In the sandy matrices,hydrate could form in the pores without contacting the sand surfaces,cement the grain contacts,grow on the sand surfaces encapsulating the sands,and even act as skeleton supporting the neighboring sand particles.The resulting behavior of hydrate in the pore spaces would significantly impact the acoustic,thermal and mechanical properties of the hydrate-bearing sediments.At the reservoir scale,the occurrence of hydrate is also inhomogeneous,which is largely controlled by the source of methane gas,local temperature and pressure conditions,and reservoir properties [33-36].The NGH deposits are mainly divided into three types:pore filling type,fractured type and massive/nodule type[37].They have been discovered and identified in many sample drilling around the world.Among them,the pore filling type deposit is the most widely present,and the field tests conducted are mostly in this type of reservoir.Considering the different reservoirs with various properties,corresponding exploitation techniques and their limitations require investigation.At present,popular production approaches involve the depressurization method,thermal stimulation method,chemical inhibitor injection method,CO2-CH4exchange method and their combinations.All these exploitation techniques aim to dissociate NGH by shifting the equilibrium conditions of the NGH reservoirs.Specifically,depressurization is considered as the simplest and is particularly suitable for the zones where free-gas is trapped beneath the methane hydrate.Yet depressurization would possibly result in a fast decomposition of hydrate,absorbing large amount of heat and carrying water and sand to flow.As for the thermal stimulation,a big concern is the economy and energy efficiency.Other methods are also suffering insufficient production efficiency or environment impacts.Thus,a detailed investigation of the production behaviors and problems of each method in laboratory is important and necessary.

In this work,a comprehensive study of different methods for gas production and its advantages and limitations are presented.The pressure vessels and simulators for mimicking gas production,operating conditions,behaviors,problems and answers are discussed.We hope more attention could be paid to NGH exploitation techniques and regulation and optimization of gas production from hydrate reservoirs.

2.Natural Gas Hydrate Resources

2.1.Structure of gas hydrate

NGH usually exists at certain conditions involving low temperature,high pressure and separated gas and water.This indicates that two natural geographic settings are suitable for hydrate occurrence,the permafrost region and the sediments under seafloor,respectively.The molecular structure of NGH has been well investigated since 1930s [16,38-40].Up to now,three main types of NGH,structure I (sI),structure II (sII)and structure H (sH),have been identified [8,15,41-44].The common feature is that water molecules are arranged as different cages under the force of hydrogen bonds.Structure I is cubic-shaped,consisting of 512cages(twelve pentagons with the diameter each about 0.790 nm)and 51262cages (diameter about 0.867 nm)with the ratio of approximately 1:3.Structure II is also cubic-shaped and contains 512cages (diameter about 0.782 nm)and 51264cages (diameter about 0.946 nm)with the ratio of about 1:3.Different from the other two structures,sH is hexagon-shaped and includes cages of 512(diameter about 0.782 nm),435663(diameter about 0.812 nm)and 51268(diameter about 1.158 nm)with their ratio of 3:2:1 [8].The size of guest gas molecules,the components and temperature and pressure conditions are considered the dominant factors of the hydrate structure[15].Especially for methane hydrate,the diameter of methane molecule (0.414 nm)is just suitable for sI hydrate with appropriate space occupancy and Van der Waals forces,especially in conditions with high pressures and low temperatures under seafloor or in permafrost region.

Fig.1.Three types of NGH Reservoir:pore filling type,fractured type and massive/nodule type[45].

2.2.Classification of gas hydrate reservoirs

According to the occurring behavior of gas hydrate in the porous sediments,three types of NGH reservoirs are commonly defined,involving the pore filling type,fractured type and massive/nodule type,respectively(see Fig.1)[37].In pore filling type reservoirs,NGH is dispersed in the pore spaces of sandstone or carbonate rocks,which is similar with the typical oil and gas reservoirs [45].In fractured type reservoirs,the natural fractures or veins present in the sediments can be suitable places for NGH formation and existence.For massive/nodule type reservoirs,NGH is gradually accumulated in the form of lump in fine mud upon long-term formation in the shallow layer sediments of the seafloor.

Based on the arrangement of different geological layers,four classes of hydrate-bearing reservoirs are classified [37,46-48].In Classes 1-3,there exist overburden and under-burden layers in geologic settings(see Fig.2).The differences are the presence/absence of water or gas layer and their arrangements.In Class 1 reservoir,hydrate layer is occurring on the top of a gas reservoir which makes the hydrate saturation in pore space very high.The highly saturated sediments will generally show a relatively low effective permeability；the boundary of the gas-hydrate layer is considered the least permeable where there is sufficient gas supply.Moreover,the thermodynamics conditions of Class 1 are fairly close to the NGH phase equilibrium conditions,hence a small amount of energy is required to induce the dissociation.From this point of view,Class 1 reservoir is appropriate for gas production.The Messoyakha Field in Russia and the Sagavanirktok Formation in Alaska are typical examples of Class 1 hydrate deposits[50].While in Class 2 reservoir,a water layer is present under the hydrate layer.This type of reservoirs has been confirmed in the Eastern Nankai Trough and Mallik sites[25,51,52].Without a continual supply of natural gas,the saturation of gas hydrate in this type of reservoir is considered to be lower than Class 1.Class 3 reservoir consists of only one hydratebearing layer without underlying gas/water zones.In this Class,NGH is formed throughout the pore spaces in the layer,sandwiched by two impermeable over-and under-burdens.Most of NGH reservoirs discovered in the Eastern Nankai Trough,Mallik site and Mt.Elbert can be categorized into this type[23,27,53-55].In Class 4 reservoir,there do not exist obvious layered structures；hydrates are dispersed in the porous sediments as different types with low saturation and unconfined geological strata in mud layers over large areas [56].In terms of recoverability,an over-and under-burden is well accepted to be necessary to maintain the mechanical stability of the reservoir,avoiding marine geological hazard；a high hydrate saturation together with a good permeability are very much required for the economy efficiency.More attentions are being paid on the connectivity of different locallyconcentrated hydrate reservoirs,in order to get a maximum gas production at a smaller cost.The produced gas-water ratio is also a factor with increasing interest in the field test,which basically represents the energy and economy efficiency.

2.3.Distribution of gas hydrate resources

Fig.2.Different Classes of hydrate reservoirs[49].

At present,the existence of NGH has been inferred or confirmed in over 230 areas worldwide(see Fig.3).Due to the particular geological structures,NGH reservoirs are normally distributed inhomogeneously.It is estimated that 97%of the reservoirs are located in the continental margin,with only a few existing in the permafrost region [32].Recently discovered marine NGH reservoirs are mainly distributed in Japan,India,the Gulf of Mexico,the Bering Strait,the South Chain Sea,Korea,Trinidad and Tobago.While in the permafrost region,Alaska in USA,the Mackenzie Delta in Canada,the Qinghai-Tibet plateau in China and Siberia in Russia are representative sites [57].The pore filling type of hydrate reservoir is commonly found in the Mallik reservoirs,the Mt.Elbert reservoirs,the Eastern Nankai Trough reservoirs [22,58,59],Shenhu area of South China sea reservoirs [60,61],the North Slope of Alaska reservoirs,the Northern Siberia reservoirs and the Qilian mountain reservoirs[24,62].The offshore reservoirs in India[63]and South Korea[64,65]are largely similar to the type of naturally fractured.Massive/nodule type is mainly found in the Gulf of Mexico and Japan Sea.

It is indicated that not all of the explored NGH reservoirs are suitable for gas production.NGH dissociation and gas production are governed by various vital parameters of reservoir,involving depth,thickness,porosity,permeability,NGH saturation and initial temperature,pressure and boundary conditions.In addition,anisotropy of permeability and inhomogeneous hydrate saturation and distribution are also essential to evaluate the gas producibility from the NGH reservoirs[66-68].All of these significant parameters and characteristics of reservoirs dominate the selection of optimized NGH exploitation method,the overall design of NGH exploitation plan and the regulation and enhancement of NGH exploitation.Therefore,a systematical evaluation of the reservoir and comprehensive laboratory investigations are crucial prior to the exploitation of the NGH.

3.Apparatus and Simulator

Large numbers of researches have been reported on simulating the gas production from hydrate-bearing sediments.Various sand particles are used to mimic the natural deposits.Different schemes of hydrate formation and gas production are proposed.The behavior of gas/water production,temperature/pressure profiles,effects of different schemes have also been discussed in detail.Yet most of the laboratory studies suffer the difficulty of scaling-up,and the gas production behaviors have been reported to be dependent on the size of the apparatus[69].In this section,the reactors used in the literature are comprehensively overviewed；the technical and operating parameters are compared,involving the volume,temperature range,pressure range and the number of transducers,etc.(see Table 1).As shown,the laboratory apparatus is improving from small to large in size,from coarse to precise in precision and from one dimension to three dimensions in performance.Basically,these systems provide the functions of withstanding high pressure,controlling temperature,monitoring the production behaviors,and simulating different production methods.

Nevertheless,the experimental apparatus is still limited by its size and boundary conditions,showing restrictions in investigating gas production behavior at field scale.Therefore,numerical simulators are widely developed for large spatial scale and long time scale.The generally used are the TOUGH+HYDRATE [88],pT+H[89],CMGSTARS[90],STOMP-HYD[91],MH21-HYDRES[92,93]and FLUENTTM[94],etc.The TOUGH+HYDRATE,a module of the TOUGH+code developed by Lawrence Berkeley National Laboratory is commonly used to simulate the nonisothermal gas recovery,phase behavior,flow of mass and heat in complex geologic media.The pT+H is a parallel numerical modeling approach developed based on the TOUGH+HYDRATE；the modeling capacity in terms of model size (larger-scale field tests)and simulation time are enhanced by 1-3 orders of magnitude.The CMG-STARS developed by Computer Modeling Group Ltd.in 1977 is a sediment simulator for heat transfer,chemical reaction and geomechanics.It is suitable for simulating recovery processes involving the injection of steam,solvent,air and chemicals.The STOMP-HYD developed by the Pacific Northwest National Laboratory is appropriate for simulation of multi-phase fluids assuming the hydrates,ice,precipitated salts and guests immobile.The MH21-HYDRES from Japan is applied to the simulation of complex system involving three dimensions,five phases and over six components on gas production from NGH via depressurization,thermal stimulation,or combination techniques.While an axisymmetric model of the three separate phases core was developed for multiphase flows during the hydrate dissociation using the FLUENTTM code；it allows variation of the porosity and permeability within hydrate sediments.

The numerical approaches take the advantages of scaling-up,providing field-scale information during gas production from hydrate reservoirs.Yet common assumptions are made involving isotropy of thermal conductivity and permeability,homogeneity of component distribution,and immobility of hydrate patches.Some guidance could be provided through large-scale simulation,yet the actual production from natural deposits could be much more complicated and unpredictable.Further efforts are still required in terms of gaining more knowledge on the geological settings,and improving the accuracy of predictions.

4.Exploitation Technologies

Compared with conventional natural gas production techniques,the exploitation for NGH are quite different.The techniques for gas production from conventional natural gas reservoirs generally focus on creating fractures to release the natural gas in the sediments.However,gas production from NGH reservoirs involves shifting the reservoir conditions to break the three-phase equilibrium of gas hydrate.So far,various gas production methods have been proposed,including depressurization method[95-97],thermal stimulation method [98,99],chemical inhibitor injection method[100-103],CO2-CH4exchange method [104-106],combination method [83,107-109],and some other novel methods [110,111].The common idea in these methods are the process of breaking hydrate phase equilibrium(see Fig.3).To be specific,depressurization and thermal stimulation directly shift the conditions of the reservoir from hydrate stable zone to the unstable zone (see Fig.3).The chemical injection process moves the phase equilibrium curve to a more rigorous condition,allowing the decomposition of NGH.The CO2exchange method takes the advantages of the difference of the phase equilibrium conditions between methane and CO2hydrate.The advantages and limitations of each method will be discussed in this section.

4.1.Depressurization method

The depressurization method mainly shifts the local pressure conditions to the area below the phase equilibrium curve of gas hydrate,where NGH will automatically dissociate to gas and water,as shown in Fig.3.Upon dissociation,the NGH saturation decreases gradually,resulting in a rising effective permeability of the reservoir.Yet,due to the endothermic process of NGH dissociation[112],large amount of heat is required for fast and continual dissociation.A local temperature reduction normally occurs resulting from insufficient heat supply；this will hinder the NGH dissociation or even trigger the secondary hydrate formation and ice generation [113].From 1960s,the depressurization method has been widely discussed,and this method has been successfully applied in the field tests[95,114].Even up to now,the depressurization is still well accepted as a promising method from the perspectives of energy efficiency and productivity[115-117].In the last decade,various depressurization scenarios have been performed to test their effects on the production behavior,as partly summarized in Fig.4.As shown,the temperature conditions vary from 0.5 to 12.5 °C and the pressures change from 0.6 to 13.6 MPa.The production pressure and pressure gradients are considered the two key parameters governing the production behavior.In the following,the progress on the investigationson gas production using depressurization are introduced from the aspects of laboratory experiments and numerical simulations.

Table 1 The parameters of several apparatus,including the size of rector,the transducer quantity of pressure and thermal and the range of pressure and temperature

4.1.1.Laboratory experiments

In 1969,the depressurization method was already utilized for gas production at Messoyakha gas hydrate field test [122].Since then,numerous laboratory studies have been conducted to investigate the behaviors of gas production from hydrate sediments through depressurization.The idea of a dissociation front was proposed and considered as a function of time；a moving boundary model was used to describe hydrate dissociation[123,124].A temperature decrease below 273 K was then observed in hydrate-bearing ocean drilling cores upon depressurization process[125].This indicates the possible obstacle from insufficient heat supply on the dissociation of gas hydrate.Efforts were also made on the kinetics of hydrate decomposition during production；the decomposition rates were obtained at the conditions of 273.5 K and the production pressure of 2.72 MPa [126].Additionally,the control mechanisms of gas hydrate production via depressurization method have been investigated[71].A depressurization to the lowest pressure was shown to result in a fastest dissociation rate；kinetics of hydrate dissociation was found to significantly affect the gas production behavior at a laboratory-scale.A dissociation model has been proposed to describe the gas production rate coupling heat and mass transfer[127].The results agreed well with the experiments；heat transfer from surroundings was considered to play a crucial role in gas production process.To scale up to the reservoir scale,it is deduced that the sensible heat of the surrounding sediment could be the major heat source during initial gas production；an efficient heat transfer from the ambient would govern the following hydrate decomposition.Further studies conducted at isothermal temperature and varying production pressures verify the significant role of pressure drop and heat supply on the gas production rate in NGH reservoir[118,128].

Fig.3.Schematic diagram of breaking the phase equilibrium of gas hydrate.The horizontal green arrow with △T represents the thermal stimulation method；the vertical green arrow with △P indicates the depressurization method and the diagonal green arrow with △T+△P shows the combination method；the blue arrows represent the inhibitor injection method and the yellow zone allows the CO2-CH4exchange me4d[19].

Li et al.[77]investigated the gas production behavior from methane hydrate in porous sediments by depressurization in a novel 3-D cubic simulator；the production pressures vary from 4.5 to 5.6 MPa and the production temperature was controlled at 281.15 K.Three main stages were first divided throughout the gas production process:free gas release stage,mixed gas production stage and hydrateoriginated gas stage.As shown in Fig.5,the A-B process is the free gas release stage,where the pressure of the system is sharply dropped；the B-C stage proceeds along the phase equilibrium curve,with hydrate decomposing under heat supply from the sensible heat of the reservoir；in the C-D stage,the temperature will recover upon heat transfer from the ambient.Further studies in a larger simulator also indicated the crucial role of the heat and mass transfer in the gas production[120].In terms of sensible heat,the porous material and the water content were considered to be the significant factors[76,129]；a high thermal conductivity of the sediments was found to promote the dissociation initially,but has little influence on the final percentage of gas production[74].Recently,attentions were also paid on the effects of various classes of hydrate reservoir on the gas production behavior.Water-excess hydrate examples of Classes 2 and 3 were dissociated through depressurization[119]；the differences between the gas production from the two deposits were discussed.Moreover,special interest was also put on the effects of particle size.The specific surface area was considered to significantly impact the gas production rate[121].

Fig.4.Depressurization scenarios in gas production process.Data shown are from Sun et al.[78],Lee et al.[118],Zhao et al.[113],Song et al.[75],Zhao et al.[77],Yang et al.[119],Castaldi et al.[81],Liu et al.[80],Li et al.[120]and Li et al.[121].

4.1.2.Numerical simulation

Laboratory experiments always suffer the difficulty to scale up in the space and time aspect.Therefore,numerical simulations are required to predict gas production behavior and to provide guidance to the field test.The intrinsic model for the kinetics of hydrate dissociation was developed by Kim et al.based on experiments[130],laying the foundation of the following simulations.The results indicated that the gas production rate was positive correlated to the particle surface area and the fugacity difference of methane；the Arrhenius temperature dependence was represented by proportionality constant.The flow equations were then taken into consideration,and three-phase 1D model to simulate the process of gas production from sandstone sediments was proposed[123].The results showed the synchronous production of water together with gas production.The gas-water ratio is currently a very crucial parameter evaluating the productivity and energy efficiency of the gas production.A three-dimensional,multi-phase flow finitedifference numerical model was developed to investigate simultaneous flow through hydrate reservoirs[131].The dissociation front was found to quickly propagate benefiting from the increasing permeability upon free gas fraction；this makes a greater effect on the gas production than the depressurization gradient.A parametric study through a onedimensional linearized model indicated that the gas production rate is governed by three factors:well pressure,reservoir temperature and zone permeability[132].Further attentions are also paid on the permeability of the reservoir[133,134].New models were introduced to describe the effect of hydrate presence on the wettability properties of porous media；the relative permeability is indicated to play a critical role in gas production.The capillary pressure was found to trigger banded structures with alternating high-low hydrate saturation in different hydrate deposits.Studies based on a 2-D axisymmetric simulator have shown that the depressurization could be more effective than other techniques during gas production[135].Yet,a molecular dynamic simulation study inferred that hydrate decomposition proceeds layer by layer；gas production by depressurization was found to be slower as compared to thermal stimulation and chemical inhibitor injection techniques[136].

Fig.5.Schematic diagram illustrating the three stages of gas production in hydratebearing sediment induced by depressurization.The inset represents the typical temperature curve during production[74].

4.2.Thermal stimulation method

The thermal stimulation method shifts the local temperature condition down below the phase equilibrium curve by heating the reservoirs (see Fig.3).External heat is usually supplied by warm water or steam injection,electrical heating or microwave heating.Considering energy efficiency,the input energy should not exceed the potential energy we can recover from the reservoir.Therefore,an efficient heat transfer in the reservoir matters；the thermal conductivity makes significant contribution in this process[137-139].Warm fluids in different phases and with various temperatures have been used to decompose the hydrate (see Fig.6).The injection temperature varies from 12.3 to 180.0 °C,and the local pressure ranges from about 2 MPa to about 7.5 MPa.In this section,the thermal properties of hydrate sediments are introduced；the results from laboratory studies and numerical simulations investigating gas production from NGH through thermal stimulation are discussed.

4.2.1.Thermal properties

In spite of the local heat convection induced by the gas and water migration,the heat for large-scale hydrate decomposition is considered to be provided largely by heat conduction from the ambient.Thus,the thermal properties of the reservoirs and the various components in the sediments play a crucial role.The thermal conductivity,thermal diffusivity and specific heat of sI methane hydrate have been measured using a needle probe technique[149-151].The thermal diffusivity of methane hydrate was measured over twice that of water,while the thermal conductivity is quite close between the two；the specific heat of hydrate is only about half that of water [149].Moreover,experiment data have shown that the effective thermal conductivity of hydrate-bearing sediments increases with decreasing sediment porosity[151-154].Thermal conductivity of methane hydrate was measured to be in the range of 0.5-0.65 W·m-1·K-1[155-157],while the thermal conductivity of water at 0 °C and ice were 0.56 W·m-1·K-1and 2.14 W·m-1·K-1[158,159].Besides,the in-situ thermal conductivity of hydrate sediments in seafloor and permafrost are also estimated [52,160,161]；this could provide some guidance in the simulations of gas production,as thermal conductivity is a crucial parameter in the equations.

4.2.2.Laboratory experiments

Several types of thermal stimulation methods have been proposed,including hot water injection,well-bore heating,hot water huff and puff,thermal flooding,etc.In early 1980s,hot injection was used as an effective technique for producing gas from Class 2 NGH reservoirs with high permeability[99].Investigations on hot water injection also indicated that the temperature and pressure in the sample fluctuated between stability and the decomposition region of hydrate during the injection[162].Different temperatures and rates of hot injection have been used,showing significant impact on the energy ratio of thermal stimulation production；gas production rate was found to initially increase and then decrease with time,with water production constant during the process[163].In 1991,the feasibility of using electromagnetic heating to decompose hydrate was investigated[164].A significantly higher energy efficiency compared with steam and hot brine injection was claimed；the heating radius was not necessary to be very large for a continuous gas production.Moreover,microwave heating method are also evaluated in terms of production behavior by Li et al.[165]and Zhao et al.[166]；the gas production rate was found to increase with higher microwave power and a function was developed to describe the relationship between temperature increasing and microwave radiation time.

Fig.6.The thermal stimulation scenarios in gas production process.Data are from Phirani et al.[140],Lee et al.[141],Sun et al.[142],Li et al.[143],Song et al.[75],Fizgerald et al.[144],Song et al.[145],Feng et al.[146],Wang et al.[147],and Li et al.[148].

The huff and puff method,also known as cyclic steam stimulation (CSS),was accidentally discovered by Shell Oil Company in 1960 during a Venezuela recovery project；it is widely used in the oil industry to enhance oil recovery [167-169].This method has also been introduced in the gas hydrate production process.Investigations on its performance has been conducted in a single vertical well in the three-dimensional cubic hydrate simulator (CHS)[84,170].Three stages were identified in the huff and puff cycle:heat injection,soaking and gas production,respectively.It is shown that with the injected heat isotropically spreading out from the injection point,the pressure sharply rose initially followed by a slower increase.The dissociation process was also considered as a moving boundary ablation process via heat injection.Further studies in a pilot-scale hydrate simulator(PHS)indicate that the thermal diffusion range was limited around the well with a constant hot water injection rate；the gas production process is largely dominated by depressurization rather than thermal stimulation [84,146].The average gas production rate was found to be much higher compared with the results in CHS.

Studies using the inverted five-spot water flooding method indicate that the gas production is initiated simultaneously with the hot water injection [148]；it could be a promising method based on the evaluation of energy efficiency ratio.Further investigations show the three stages of the water flooding:reservoir temperature rise,uniform dissociation and frontal brim breach [171].The heat transfer analysis during thermal stimulated hydrate dissociation exhibits temperature transfer delay and a significant temperature difference along the radial direction is observed[172].Besides,heat was considered to be mainly transferred through conduction from the dissociated zone to the dissociating zone.Moreover,the thermal stimulation method(hot water circulation)has been applied in the production test at the Mallik site in Canada.In spite of its effectiveness and speediness,the applicability of the thermal stimulation method is still challenging due to the large energy input and low energy efficiency[173].

4.2.3.Numerical simulation

A number of researches have been conducted on the gas production behavior through thermal method,considering the heat and mass transfer,hydrate dissociation kinetics and phase transition；special attentions are paid on the recovery behaviour and energy efficiency ratio[107,174-176].

A three-dimensional finite-difference simulation on the gas production from NGH reservoir containing stratified layers has shown that the thermal capacity of the reservoir was sufficient to provide the energy necessary for hydrate dissociation [98].A two-phase,three-dimensional numerical model was then developed to simulate the dissociation of hydrate for addressing the issue of gas production feasibility based on the Holder's model[177].Selim and Sloan[178]assumed decomposition of a pure hydrate block using a constant heat on semi-infinite medium and presented a mathematical model to describe the hydrate dissociation.The model was later verified to show an uncertainty of about 10%with experimental data [179].Moreover,simulations coupling the intrinsic kinetics with heat transfer rates have shown that the gas production rate is controlled by heat transfer and intrinsic kinetics[180].Further studies also indicate the vital role of the kinetic limitation on modeling short-term dissociation process；a neglect could result in a significant under-prediction of recoverable hydrate.Yet it has little influence on large-scale systems undergoing thermal stimulation[181].Moreover,a significant difference in dissociation patterns was observed when the thermal boundary conditions were shifted from adiabatic to constant-temperature[182].

In recent years,special interest is put in the simulation of largescale natural reservoirs.The simulation results of a thermallyinduced gas hydrate dissociation in the Mallik 5L-38 well were highly consistent with the field test；only a small amount of gas could be produced through single thermal stimulation or depressurization in conventional well configuration [183].Therefore,it requires a combination method or other innovative approaches to produce gas from Class 3 hydrate reservoir.The gas production potential from the laminar hydrate deposit through a single horizontal well at drilling site SH7 in the Shenhu Area,China by means of stream huff and puff was also examined [184].It is indicated that hydrate re-generation could occur during the injection stage；reasonable injection and production rates were considered necessary to avoid an over pressurization and depressurization.Further studies on simulating gas production from hydrate reservoirs in Shenhu Area using thermal stimulation also show similar results[185-187].

4.3.Chemical inhibitor injection method

Recently,the inhibitor injection method also draws industrial attentions,though appearing not so promising in large-scale use[141,188-194].Actually,the chemical inhibitors have already shown their effectiveness and are being widely used in the gas/oil pipelines to avoid hydrate blockage.As a potential method to decompose hydrate,it works by shifting the equilibrium curve towards a more rigorous condition,breaking the stable state of hydrate(see Fig.3 for details).Two main types of chemical inhibitors are generally classified:thermodynamic inhibitors changing the hydrate equilibrium conditions and kinetic inhibitors slowing down the hydrate formation rate [143,195-214].The thermodynamic inhibitors are more effective and secure,among which methanol and ethylene glycol are two common ones widely used[215].Ethylene glycol is considered to show a higher availability,lower toxicity and better performance in inducing gas production from hydrate reservoirs[216].The physical properties of the inhibitor are of crucial importance in its application,governing its effective diffusion and permeation into the reservoirs.In this section,properties of the two thermodynamic inhibitors are introduced；the laboratory investigations on gas production through injecting chemical inhibitors are discussed.

4.3.1.Properties of inhibitors

The low boiling point,high vapor pressure,low freezing point and low viscosity of methanol make it possible to be operated in a wide temperature ranges.Generally,part of the injected methanol is evaporated into gas phase together with methane recovery during gas production；the remaining continually cycle with aqueous solution.This makes it difficult to recycle in solution considering the feasibility,availability and economy.As for ethylene glycol,the characteristics of non-poisonous,relatively high boiling points ensure its recycling；yet the relatively high viscosity of its aqueous solution generally causes difficulty in separation at low temperature conditions.Moreover,as the inhibitors are working in the form of fluids,the sufficient flow of which could play a crucial role in the efficient hydrate decomposition and gas production.Consequently,similar to thermal stimulation,the key issues for chemical inhibitor injection also involves the effective diffusion and permeation into the reservoirs.In addition,the concentration of inhibitor solution,injection rate,interfacial area along with temperature and pressure are the important factors controlling hydrate dissociation rate.

4.3.2.Laboratory studies

In early days,researches on electrolyte inhibitor,such as NaCl solutions,KCl and CaCl2solutions[217,218],have gained many attentions.The inhibiting effects of various similar salts were investigated and classified by Makogon [39].Then the depression effect of KCl and CaCl2on cyclopropane hydrate formation was reported[219].Further studies have found that approximately 3.45%brine concentration was able to induce hydrate dissociation and significantly increase the gas production；excessively high brine concentration could yet result in a reduction of gas production rate as gas flow is hindered by the solution [220].Recently,MD work has preliminarily explained the mechanism of the inhibiting effects of halogen ions；the salt ions could lead to a change of water positions in hydrate cavities,making cage structure fractured easily[221].

Electrolytes are non-volatile chemical inhibitors,while the commonly used methanol and ethylene glycol are volatile.Studies have found that the inhibiting effect decreases with an increasing volatility[222].This indicates that besides the recyclability,inhibitors with a high volatility would not be efficient in the large-scale test.The instantaneous rate of gas production has been investigated to depend on the inhibitor concentration,injection rate,pressure/temperature of inhibitor solution and hydrate-inhibitor interfacial area[223].Fan et al.[224]also indicated that the dissociation rate depends on the concentration and flow rate of ethylene glycol [182,215].Ethylene glycol was considered to contribute to reducing the dissociation heat,and less energy is required for dissociation at a higher ethylene glycol concentration [215].Moreover,it has been inferred that,dissociation using inhibitor injection was spontaneous and occurred throughout the entire hydrate sample,compared with localized dissociation through depressurization [216].However,the application of chemical inhibitor injection in the field tests is still very challenging,due to the potential environmental impacts,high cost and low efficiency resulting from the low permeability of hydrate sediments.Yet this method could contribute to reducing the energy input,making it an alternative option to combine with depressurization or thermal stimulation.

4.4.CO2-CH4exchange method

Recently,CO2-CH4exchange method has drawn worldwide attention,due to its advantages in sequestrating CO2underground by forming CO2hydrate while synchronously recovering methane gas from hydrates[225,226].The newly-formed CO2hydrate would be relatively stable at the geological conditions,the mechanical stability of which would also help maintaining the strength the residual reservoir[227,228].The molecular-scale exchange process is briefly shown in Fig.7.At certain conditions,the CO2molecules can diffuse into the hydrate structures and replace the CH4molecules in the hydrate cages,forming CO2hydrate and release CH4gas.The micro-mechanism of this process still remains partly unclear.In this section,the motivations of CO2-CH4replacement method are introduced,followed by discussions on the laboratory progress and limitations.

4.4.1.Motivations and challenges

The phase equilibrium curves of methane hydrate and CO2hydrate are intersectant(see Fig.3)；there exists a certain region where CO2hydrate is stable and CH4hydrate is not[104].When the temperature and pressure are located in this region,the CH4molecules will escape from the water cages,while CO2molecules are possible to fill this cage.Both methane and CO2molecules typically form sI hydrate [8].Some studies have claimed that no transformation of hydrate structure occurs during the CO2-CH4exchange process[8,15,229].In addition,the heat released during the exothermic CO2hydrate formation process(57.98 kJ·mol-1)is higher than that required in the endothermic CH4hydrate dissociation process((54.49 kJ·mol-1)[230].This indicates that the exchange process could possibly proceed spontaneously.Further MD investigation reports that the Gibbs free energy is-12 kJ·mol-1for the exchange reaction at the condition of pure water,pure gas and fully hydrate saturation[231].Yet,the molecule diameter of CO2(0.512 nm)is larger than the size of the small cage of clathrate hydrate structure I.Therefore,the idea of mixture gas exchange has been proposed.The molecule diameter of N2is sufficiently small to fill the small cage；the CO2/N2mixture might enhance the CH4exchange rate,provided that the conditions are favorable for the N2to go into the small cages[8].Moreover,the exchange process can also help reduce water production；that water released during hydrate dissociation process is then consumed for CO2hydrate formation.

However,problems still remain ahead the application of CO2-CH4exchange at a large-scale.The micro-mechanism of the exchange process is not very clear.Some researches claim that the exchange consists of two independent stages:methane hydrate decomposition and CO2hydrate formation [232-235].The water cages are considered to completely destruct,followed by CO2hydrate formation from liquid water.Yet other studies also report that CO2molecules displace CH4molecules directly without changing hydrate structure[229].The exchange is directly driven by gas molecule diffusion in solid hydrate phase,which could be a very slow process.Investigations have indicated that there exists a strong limitation effect of the outer mixed hydrate shell on the mass transfer of guest molecules [236].To scale up,the natural geological settings of the sediments could be very complicated coupling endothermic/exothermic reactions and mixed gas molecules migration；the effective diffusion and permeation of CO2fluid in the reservoir would significantly govern the exchange efficiency.Another concern is the potential impacts on the deep-sea ecological environment upon CO2injection[237].In spite of the existing challenges,the exchange method still remains interesting,and could possibly be applied on basis of careful control.

4.4.2.Laboratory studies

Fig.7.The schematic diagram showing CO2-CH4exchange process.

Since 1980s,CO2-CH4exchange technique for gas production has been proposed and a number of researches on thermodynamic and kinetics have been performed[238-240].Till 1996,the first experimental evidence for the possibility of CO2injection for methane gas production from NGH sediments was reported[104].Then the trials on different states of CO2were successively conducted.Experiments on the gaseous CO2have shown the preference of CO2going into the cages；the average ratio between CH4and CO2in gas phase after exchange was 2.5,indicating a higher tendency of CO2to be in hydrate phase[104].The in-situ results on the exchange rate using Raman showed that the exchange process could be extremely slow[241]；the freshly formed CO2hydrate shell acts as a mass transfer barrier[242].A higher reaction temperature was found to enhance this process[243].Further effects of initial CO2mole fraction,system pressure and diffusion coefficient of CH4were also investigated[244].Moreover,liquid CO2is considered to be helpful in the exchange as the chemical potential of CO2molecules is significantly raised,providing a higher driving force[245,246].Actually,the temperature and pressure conditions of NGH in nature are mostly located in the region of liquid CO2state[230].Experiments on the exchange of CH4hydrate soaked in liquid CO2have shown that the formation of CO2hydrate almost consumed all the H2O molecules released from CH4hydrate[245]；a mathematical model based on non-equilibrium thermodynamics was also proposed to describe the observed phenomena.A kinetic study on the exchange process using liquid CO2found that the decomposition of the middle-sized cage proceeded faster than that of the small cage；this might be attributed to the stability difference of CH4molecules in different cages[247].A kinetic model considering the rate difference between middle-sized cages and small cages was then proposed.Compared with gaseous CO2injection,liquid CO2could help reduce water production[248].The exchange rate was also found to be not dependent on the region of the temperature-pressure conditions；the fugacity differences between methane hydrate decomposition and carbon dioxide hydrate formation could play a more crucial role[249].In terms of improving the exchange rate,a scenario using CO2-water emulsion was first proposed by Mcgrail et al.[250]；it was shown to be helpful compared with liquid [251]and gaseous CO2[252].Further advantages of the emulsion-based technique could also been found in the following literature [253].Besides,new ideas are coming out using supercritical CO2[254,255]and CO2mixed with other gases[229].Techniques involving material balance(MB)[256,257],X-ray diffraction (XRD)[258-265],neutron diffraction [266],Raman[267],nuclear magnetic resonance(NMR)[268],magnetic resonance imaging(MRI)[269-273]and particle size analysis(PSA)[274]are being used,in order to qualify and quantify the complicated exchange process[275].

4.5.Combination method

Concerning the limitations of the individual technique discussed above,the idea of a combination method is also proposed,taking the advantages of each method.The common combination is huff and puff in conjunction with depressurization,which has been proved to be more efficient[276]；or chemical inhibitor injection combined with thermal stimulation,showing great advantages on avoiding the secondary formation of gas hydrate and enhancing the permeation of chemical inhibitors[277,278].A number of investigations on gas production via a combination method have been carried out[75,276-282].

Experiments on the combination of thermal stimulation and depressurization have shown that the energy efficiency could be raised from 1.8 to 2.88 compared with single thermal stimulation[279].Comparison on the single depressurization,two-cycle warm-water injection and the combination method of the two has also indicated an increase of gas production by 18.63% and 31.19% [75,283]；the combined method showed advantages in all energy efficiency,percentage and average rate of gas production.Further studies through in-situ magnetic resonance imaging(MRI)showed similar results as well [73,284,285].Moreover,the hot water injection is commonly combined with chemical inhibitor injection(normally salts).Investigations on hot brine injection has shown that the dissociation could be shortened with an increasing salinity [143].Compared to hot water injection,the thermal efficiency and energy ratio could be enhanced by hot brine injection.Attentions are also paid on the injection patterns；the continuous hot brine injection shows better performance than intermittent injection[286].Hydrate dissociation with NaCl solution is also found to be faster compared with Na2SO4solution and ethylene glycol[248].Moreover,models are also used to describe the migration of the thermal front during the injection[287].Yet few researches are found on the combination with chemical inhibitors,possibly resulting from the environmental concerns.A comparison on the gas production behaviors through depressurization,thermal stimulation and combined method is presented in Fig.8.As shown,the combination method reveals a remarkably better performance in total gas production and peak gas production rate.

4.6.Other methods

The in-situ combustion(ISC)was first proposed by Cranganu[290]in 2009.A specially designed heating apparatus is used to decompose the hydrates；an air/gas fuel mixture are introduced into a combustion vessel via an injection tubing.It was shown that only about 1.1 to 1.7%of the produced gas is sufficient to maintain the energy input for gas production.Based on this idea,a simulator with 425 L was developed[85,86].Studies have shown that a high porosity and high hydrate saturation could help the combustion[291].However,safety concerns of the uncontrolled process limit its further application.In-situ warm sea water injection is also proposed as a potential method to recover gas hydrate[292],showing advantages on the availability of the decomposition triggers.

5.Conclusions

Fig.8.The comparison of gas production behaviors through depressurization,thermal stimulation and combination method.Shis hydrate saturation.Data in (a),(b)and(c)are from Song et al.[75]；(d)from Li et al.[288]；(e)and(g)from Falser et al.[289]；(f)from Feng et al.[278]；and(h),(i)and(j)from Zhao et al.[281].

NGH has been well accepted as a potential source of clean carbon with enormous reserves.However,the characteristics of NGH reservoirs in the world vary a lot；to identify an effective and safe method for gas production still remains challenging.In this paper,the status of NGH exploitation techniques is comprehensively reviewed；efforts in laboratory experiments and numerical simulation on the depressurization method,thermal simulation method,chemical inhibitor injection method,CO2exchange method,and the combination method are introduced and compared.The advantages and limitations are briefly discussed,and the potential for scaling-up and application in the field test is analyzed.How to connect the laboratory-scale mechanisms with the field-scale behaviors would be the issue of concern in the near future.The limited experience in the exploitation of NGH from marine sediments would guide the following fundamental researches.Problems still remain involving sand production,inhomogeneous distribution of hydrate,low permeability,insufficient heat supply,severe water production,potential mechanical instability,environmental impacts etc.Further efforts should be focused on the existing engineering problems and find the efficient solution.