Growth and aggregation micromorphology of natural gas hydrate particles near gas-liquid interface under stirring condition

Qihui Hu ,Xiaoyu Wang ,Wuchang Wang, *,Yuxing Li ,Shuai Liu

1 Shandong Key Laboratory of Oil-Gas Storage and Transportation Safety,College of Pipeline and Civil Engineering,China University of Petroleum,Qingdao 266580,China

2 State Key Laboratory of Natural Gas Hydrates,Beijing 100028,China

3 Research Institute of Experiment and Detection,Petro China Xinjiang Oilfield Company,Karamay 834000,China

Keywords:Natural gas hydrate Growth Aggregation Gas-liquid interface Micromorphology Physical model

ABSTRACT To investigate the morphological evolution of the whole growth and aggregation processes of hydrate crystals near the gas-liquid interface,we used a high-pressure visual reactor with high-speed camera to capture the micromorphology of hydrate particles in a natural gas+pure water system with pressure from 2.6 to 3.6 MPa and sub-cooling from 4.7 to 6.23 °C.The results showed that under low sub-cooling conditions,the amount and size of particles increased first and then decreased in the range of 0-330 μm,and the small particles always dominated.These particles can be roughly classified into two categories:planar flake particles and polyhedral solid particles.Then,the concept of maximum growth dominant particle size was proposed to distinguish the morphological boundary of growth and aggregation.In addition,the micro model was established to better reflect the effects of particle formation process and evolution mechanism near the gas-liquid interface under stirring condition.The results of this study can provide a guidance for flow assurance in multiphase pipeline.

1.Introduction

Natural gas hydrate is a type of ice-snow crystalline compound with a cage structure and is composed of water and hydrocarbon molecules [1].Since Hammerschmidt [2] clarified the blockage phenomenon of pipeline hydrate in 1934,with the petroleum industry shifting to ultra-deep water areas and trial production of natural gas hydrate [3-6],a series of problems caused by hydrate formation and blockage had posed a serious threat to the flow assurance in industry [7-10].More and more studies have shown that the growth and aggregation of particles is one of the main causes of pipeline blockage[9,11].Therefore,it is of great significance to study the characteristics of hydrate growth and aggregation in the flowing condition.Fatnes [12] and Chenet al.[13]simulated the flow characteristics of hydrate in a pipeline under different volume fractions and flow velocities.Chenet al.[14]found micromorphology of completely undecomposed hydrate particles as spheres in a water-dominated bubbly flow.While Dinget al.[15]found hydrate agglomeration involving the un-converted water drops.Liet al.[16]used the enhanced wall function method to describe the agglomeration of hydrate particles using the elastic collision theory.However,this simulation only considered the collision of the center of gravity and neglected the particle breakage caused by collision.

In addition to the research on the flow characteristics of hydrate particles,some scholars have carried out researches on the formation of hydrate at the gas-liquid interface,which is very important for the study of hydrate production and pipeline safety [17].Hayamaet al.[18] observed the growth and formation process of methane hydrate surface,and found that the hydrate grew in the form of a polycrystalline layer and covered from the droplet surface.Colombelet al.[19]studied the process in water-in-oil emulsion by using nuclear magnetic resonance technology.It was found that the characteristics of the oil-water interface were the main factors affecting the aggregation of hydrate particles.According to Stoporevet al.[20] and Cuiet al.[21],crystals rapidly covered the water-gas or water-organics interfaces (depending on the nucleation site) at the initial stage of growth hydrate.Stoporevet al.[20] also found that at the initial stage of growth hydrate crystals firstly formed at the interface.Subsequently,large-size hydrate particles aggregated on cell walls.The entire experimental process highlighted the superiority of growth at the gas-liquid interface.

However,multiple processes including formation,growth,aggregation,and deposition of hydrates generally occur simultaneously accompanied by many changes in the microstructure and properties of the hydrates,making theoretical analysis and experimental analysis of the evolution of hydrates very difficult.The particle size distribution after aggregation and fragmentation in a solid-liquid two-phase flow was further analyzed [22].The experiments of Wanget al.[23] and Liuet al.[24]proved that the van der Waals force was the main force in a pure water system and determined the particles’ stability and saturation [25].Wanget al.[26] found that the cohesive force of hydrate particles in liquid phase was independent of particle size,while was linearly correlated with the effective radius of the hydrate particle pair in the gas phase.Maet al.[27]studied the influence of a solid surface on the stability of hydrate nucleus through molecular dynamics simulation.Meanwhile,he discovered that the stability of the hydrate nucleus on the surface is closely related to the affinity for guest molecules of the solid surface.Subsequently,Lvet al.[28] conducted an experimental study on formation and coalescence in a mixture of oil-in-water emulsion in a high-pressure experimental loop of hydrate from the microscopic point of view.Songet al.[29] photographically captured the micro-morphology of three types of methane hydrate particles with the help of a high-speed camera,and focused on analyzing the small particles sheared off from the gas-liquid interface and the growth of hydrate from small bubbles.They further discussed the process of aggregation of the hydrate particles.From the perspective of decomposition,Wanget al.[30] analyzed the changes of the hydrate particle size and morphology at the main liquid bulk and the gas-liquid interface,and obtained the physical model and related parameters.

Based on the above analysis,there is little study of the microstructure of the particles,especially particle evolution near the gas-liquid interface in a flowing liquid.Considering the flow shear near the gas-liquid interface,it is particularly important to study the growth and aggregation microstructure of natural gas hydrate under flow conditions to reveal the mechanism of evolution of the hydrates and their effects on flow assurance.

Therefore,experiments to investigate the morphological evolution of the whole growth and aggregation processes of hydrate crystals near the gas-liquid interface were conducted in a cell to simulate the formation of gas hydrate.The micro-scale models of the growth and aggregation of natural gas hydrates under flow conditions were established by considering the discovered crystal structures and analysis of the evolution process.This study can provide guidance for flow assurance in multiphase pipeline.

2.Experimental

2.1.Experimental equipment

In this study,an experiment in hydrate formation was conducted by using a reaction system [24,31].The device diagram of the experiment is shown in Fig.1.

Fig.1. An experimental sketch of gas hydrate formation.

Table 1 Natural gas component parameters used in the experiments

The experimental device included (1) a gas supply system,which provided the gas needed and controlled the pressure required for the experiment;(2) temperature control system,which simulated the temperature of the environment needed through a chiller with a temperature control accuracy of ± 0.05 °C and shown as the blue part in Fig.1;(3)high-pressure cell system with visible window(?65 mm),which was made of stainless steel and had a total volume of 950 ml,to be used for hydrate formation experiments and observing hydrate particles;(4) data measurement and acquisition system,in which the temperature sensor(T2) with an accuracy of ± 0.1% on the left in Fig.1 measured the liquid phase temperature,and the middle one (T1) measured the gas phase temperature,and the pressure sensor (P) with an accuracy of±0.25%measured the pressure in the cell;(5)video camera system,which captured the phenomenon through a high-speed camera and recorded the formation and changes of hydrate in experiments.In addition,the macroscopic state was captured by a high-magnification camera (NEX-VG10E).

2.2.Experimental materials

The materials used in this experiment mainly included natural gas and deionized water.The natural gas was supplied by Beijing Nanfei Industry and Trade Company Limited.The gas composition details are listed in Table 1.The deionized water was prepared in the laboratory.

The phase equilibrium temperatures corresponding to the experimental pressure conditions were measured in the experiments as shown in Fig.2,and at the same time,more hydrate formation phase equilibrium points were calculated with help of PVTsim?software,as shown in Fig.2.Moreover,the experimental conditions were also added in Fig.2 and the corresponding subcooling was calculated as shown in Table 2.

Fig.2. Experimental conditional and the hydrate phase equilibrium curve for the natural gas.

Table 2 Distribution of sub-cooling under different experimental conditions

2.3.Experimental process

This study mainly focuses on the micro flow behavior of the hydrate particles.The experimental procedure is as follows:

First,the experimental apparatus was cleaned by nitrogen sweeping and circulating deionized water three to five times to confirm that no impurities were adhered to the cell.The tightness of the valves,gas connector,and the cell was checked.Then,650 ml of deionized water was loaded into the cell,and a vacuum was pumped to ensure that there was no air before inflation.Next,natural gas was filled until the pressure reaches the set value and maintained for a period of time.When the pressure achieved stability,the gas source was closed and the magnetic stirrer was opened to ensure that the gas and water mixed completely.In addition,the data acquisition system was used to measure and save the related data.

And then,the chiller started to reduce the temperature,and the hydrate formation process was observed.When the pressure and temperature remained stable for a long time,the hydrate formation process was considered complete.Thus,with the camera capturing images of the hydrate particle morphology,the formation and attachment of the visual window was acquired every alternate hour.After the previous experiment was completed,gas was inflated again until the pressure in the cell increasing 0.2 MPa,and the aforementioned process was repeated until the dense particles could not be photographed clearly (at this experiments,the pressure was 3.6 MPa).

The actual experimental conditions were from 2.0 MPa to 4.0 MPa.When the pressure in the cell is lower than 2.6 MPa,there were not so many hydrate particles in the cell and morphology distribution can’t be easily captured.On the contrary,when the pressure in the cell is higher than 3.6 MPa,it was difficult to observe the distribution trend of single particles among too many particles in the cell.Therefore,we selected the range from 2.6 MPa to 3.6 MPa as the best experimental range.The experimental conditions were decided with the initial pressure of 2.6,2.8,3.0,3.2,3.4 and 3.6 MPa.

Moreover,to minimize the effect of the flow shear on particles and simultaneously ensure that the reaction was fully conducted,the rotation rate in most experiments was set as 100 r·min-1.The specific experimental conditions are listed in Table 3.Among them,Case 1 and Case 2 are used to select the stirring rate in the experimental process,and Case 2 to Case 7 emphasize the role of pressure condition in the entire reaction process.

Table 3 Experimental conditions

Fig.3. Temperature and pressure curves at 2.6 MPa operating conditions (the purple line represented the corresponding time when the experiment reached the hydrate phase equilibrium temperature).

When the high-speed camera was used to capture the micromorphology and flow process of hydrate,different shooting speeds should be selected according to different conditions.In this work,the shooting speed was set to 250 to 2000 fps.

3.Initial Particle Characteristics

3.1.Morphology of initial particle sizes

The formation and aggregation of hydrate particles is a continuous process,and in some stages,the microstructure changes rapidly.Meanwhile,according to the trend of temperature and pressure,taking an initial pressure of 2.6 MPa as an example,the entire system was divided into gas dissolution,induction,rapid formation,and stable generation stages,as shown in Fig.3.The division of the entire reaction process did not only depended on the change of temperature,but also integrated the liquid temperature,pressure,particle morphology and size to obtain the following change process.Among them,the preparation,gas dissolution and induction stages were greatly affected by pressure.

In this experiment,the process of maintaining a stable initial pressure and the temperature reach the phase equilibrium temperature was regarded as the gas dissolution stage.The induction stage was the process from the phase equilibrium temperature to the beginning of hydrate formation change.The rapid formation stage manifested as a descent in the pressure curve.The stable generation stage occurred when the concentration was kept in balance.

According to the distribution of statistics in each time period,it can be seen from Table 4 that with an increase in pressure,the gas dissolution stage was basically stable at approximately 60 min,which was controlled by the cooling rate during the experiments.The distribution of the induction stage is somewhat random in different experimental conditions.The rapid formation stage increased at a rate about 1.5 with the pressure.The stable generation stage was subject to the actual working time of the experiment.

Through the aforementioned phenomena,it was found that hydrate particles appear in the induction phase first,then grow and aggregate continuously.In the next stage,the concentration of hydrate is low,and the aggregation ability of small particles is affected by the shearing force.Hydrate particles always grow or aggregate into different states mainly in a pattern of single small particles,and flocculate through the van der Waals force [18,32-35]among particles.Meanwhile,particles have a tendency toward a closed aggregation shape such as a cross or triangle to promote their continued growth or aggregation with smaller particles[36-38].

3.2.Particle size distribution of initial particles

The morphology of initial particles was captured by observing the video of the experimental conditions.The particle size was analyzed by choosing the linear distance between any two points on the particle boundary (regardless of how it appeared at the interface of the visual window)as the particle size(without specialinstructions;the scale at the bottom right of the picture is a proposed scale of 200 μm,the spatial scale is 0.55,and the actual length is 110 μm),as shown in Fig.4 below.

Table 4 Distribution of reaction time under different experimental conditions

Fig.4. Particle morphology of hydrate particles at 283 minutes of formation(at initial pressure of 2.6 MPa).

Fig.5. The trend of the average particle size of hydrate with time at 3.0 MPa.

In addition,we carried out two repeated experiments on the same pressure condition to verify the micro characteristics of the particles.

Taking the repeated experiment at 3.0 MPa as an example,the change trend and error range of average particle size with time were shown in Fig.5.It can be seen from Fig.5 that under the operating condition of 3.0 MPa,the largest error of multiple measurements of particles size over time is around 18.4 μm,the smallest is around 10.9 μm.

Based on the definition of the particle size,particle size distribution of particles was obtained by the method of data statistics with an intermittent photography in these experiments.At each analysis moment,first record 36 seconds.And then,In the 36 seconds video screen,we selected one picture every 1000 frames,that is,10 pictures were selected from a small video screen to reflect the particle status in this video screen.And the number of particles in each particle size range,as well as the average,maximum and minimum particle size under the working condition were further analyzed.

For the morphology of the initial particles,which means particles from the first particle to the rapid formation stage,in order to eliminate the error caused by intermittent video recording,this study conducted two relatively recent video recordings at the initial moment to comprehensively analyze the particle size at the initial moment as shown in Fig.6.

Fig.6. Size distribution of initial particles under different conditions (the time selected was the two earliest shooting periods).

After counting the initial particle size,it was found that 90% of the particles were distributed in the range of 0-440 μm.Under low pressure,this was less than 10%;and under high pressure,it was closer to 25% over a range of 770 μm.The number of particles was in the form of a normal distribution (R2=0.96) across the entire range.Generally,the initial distribution was no higher than 50,and the peak value was reached around 220 μm,as shown in Fig.6.The number of initial particles increased slightly with an increase in the pressure conditions.

4.Growth and aggregation characteristics of hydrate particles

4.1.Morphological changes during growth and aggregation of hydrate particles

According to the experiments,hydrate particles are distributed as planar flakes and polyhedral solid particles.Mochizuki[39]proposed that large polygonal and transparent hydrate-crystal plates exhibit anisotropic growth at a low sub-cooling,while hydrate aggregates were found to grow isotropically at a high subcooling.This aspect provides a theoretical basis for the morphology of particles.On the other hand,it emphasizes the influence of subcooling on the morphology of particles.This is different from the typical assumption of hydrate as spherical particles,which has the smallest surface area in the same volume [40,41].However,owing to the limitations of the experiment,no spherical particles were observed.In theory,the hydrate particles would exist in a spherical shape just after nucleation.Nevertheless,the hydrate particles in this experiment were found to be larger than 10 μm.The particle morphology of planar flake hydrate can be classified into two trends.On one hand,they form a planar flake structure independently of themselves,and some free particles adhere to the surface.The irregular structure of non-trunk growth is similar to snowflakes,and they further grow into multilayer superimposed particles,as shown in Fig.7 (c).On the other hand,they can grow into another typical shape of dendrite morphology,as shown in Fig.7(a).This shape has distinct branches as well as trunks and follows certain growth law.

Fig.7. Micromorphology of hydrate under different initial pressures.

Fig.8. Typical morphological distribution of dendritic particles.

Further analysis showed that snowflake particles grow from small aggregated particles to larger ones with rounded edges.Based on the statistic of the particles distribution,it can be observed that the radial ratio basically kept about 2 floating.Based on the number of trunks,the dendritic particles can be divided into three categories,as shown in Fig.8.

As shown in Fig.8 (a),the angle between the trunks and branches of single dendritic granules was mainly 60°,which basically maintained a parallel distribution between the branches.For the dual dendritic granules as shown in Fig.8 (b),the included angle remained the same,while the angle between the two trunks was 120°.Finally,multibranched particles were synthesized by the characteristics of the first and second types of particles.

In this study,the distribution of tridendritic granules was the main one,so they were classified as multidendritic granules.There were also a small number of multitrunk particles,which mainly existed in the later stage of the experiment.This is shown in Fig.7 (e).This was caused by the continuous outward growth of the main trunks,which affected the branch area and occupied more area.

In the process of branch formation,dendritic particles tended to grow in two directions.Under the action of the flow shear force,the reverse growth fell off,leaving suitable growth branches.The growth direction which might have the smallest resistance,similar to the streamlined design of automobiles and the growth trend of some plants in nature.However,the top particle was sharper,and the length of the trunk gradually lengthened from the top to the bottom because the flow resistance at the top was the largest and more growth driving force was consumed.In addition,compared with large resistance at the top of the main trunk,the lower resistance is favorable for the growth of the branch.

According to the statistics results,the dendritic granules first formed in the early stage,and the main trunk grew gradually with a size basically no more than 550 μm.When approaching 550 μm,the particles further grew into another trunk from the bottom,that is,a dual structure.Therefore,it is difficult to observe the existence of single dendritic particles in the later stage of the experiment.As the experiment time grows,the multi-dendritic structure gradually grew into another trunk on the basis of the dual-dendritic structure.This shows that branches are on the main trunk,which grow together with the trunks,and there are branches or small particles attached to the branches.When the branches meet the main trunk or branches,they stop growing.

In addition,it can be seen that the growth trends were different.When the trunks grew,the length increased first,and then the root position of the trunk gradually became thicker,as shown in Fig.9.This indicates that growth is transmitted upward through the roots,similar to the growth process of leaves [42,43].

Fig.9. Formation of dendritic granules.

Fig.10. The flake particle morphology under 100 r·min-1 rotation rate.

Fig.11. The chain particle morphology under 150 r·min-1 rotation rate (128 min).

To further verify the process of the abovementioned particles,the decomposition of the dendritic granules was also studied in the experiments,and it was easier to observe the changes in typical micro-particle morphology.The process of the decomposition was as follows:the particles attached to the branches first fall away,the branches follow,and finally,only the main trunks remain,which seldom break directly.In light of this phenomenon,it can be inferred that the main trunk growth is based on the growth of the particles themselves.

The polyhedral solid particles were further formed on the basis of the above planar flake particles.The particle growth process might be the collision and aggregation of the flake particles,or the continuous growth and aggregation from themselves.This is distributed in different forms,but basically maintains the form of internal aggregation.Under this growth condition,the particles have more opportunities to contact with the free particles,and the rough surfaces of the particles help increase the particle size.

In addition,it was found that increasing the stirring rate has a great influence on the distribution and micromorphology of hydrate particles.Taking 2.6 MPa of pressure as an example,the distributions of two rotation rates (100 r·min-1and 150 r·min-1)were studied.It is obvious that the distribution of flake particles is dominant at low rotational speeds,as shown in Fig.10.Nevertheless,when the rotation rate is increased,the distribution of chain particles is dominant,as shown in Fig.11.

In addition to recording the changes in the micromorphology,the macro-aggregation was also observed,as shown in Fig.12.With the breakup of the small bubbles near the gas-liquid interface,the original gas was released.Under the condition of phase equilibrium of hydrate formation,hydrate particles were formed at the gas-liquid interface firstly.Then,they gathered near the gas-liquid interface owing to the density difference with water and grew upward in the gas phase space,while the liquid phase space developed downward.With an increase in time,the liquid bulk gradually became turbid,and the color became darker.That is,the concentration and number of hydrate particles increased and caused the transmittance to become weaker.

By comparing the aforementioned processes,we observed a difference in the window liquid level,which was caused by the formation of hydrate.After the liquid level dropped,the upper-part occupancy increased and hydrate production increased.As time went on,the liquid bulk became more turbid,and the thickness of the deposit layer at the gas-liquid interface increased gradually.

Therefore,it can be seen that small particles exist for a long time.This is because when hydrate particles are first formed,they mainly grow as single particles or aggregate into different aggregation states under the van der Waals force [35,44-46].Stirring results in small particles near the gas-liquid interface.It was observed that the main morphologies of particles were small particles and flake particles with main trunks,while snowflake particles only existed in the form of multilayer superposition.This is because the connection between snowflake particles is weak,and unstable snowflake particles are broken into small ones and carried to the interface under the shear force.Then,because snowflake particles are generally large and receive more flow resistance,it is difficult to carry a large number of them by stirring;thus,the formation of snowflake particles is very likely to separate from the flocculation layer.

In summary,the main distribution of dendritic particles occurred owing to the tightness of the trunk connections and their smaller shape.In the later stage of the experiment,the particles were mainly manifested in the form of multitrunk particles.

4.2.Particle size in growth and aggregation of hydrate particles

Analysis of the particles near the interface showed that both the initial pressure and the rotational rate had impacts on the particle distribution.

4.2.1.Effects of pressure on particle size

Through an analysis of the particle size distribution,it can be seen that the total distribution trend of particles in the range of 0 to 330 μm first increased and then decreased,during which the proportion of small particles was still larger.In the meantime,the number of large particles increased synchronously,but the proportion was small.

Fig.12. Distribution of flocculation layer at gas-liquid interface (2.6 MPa).

Fig.13. Distribution trend of particles in different stages (2.6 MPa).

Table 5 Particle size distribution at gas-liquid interface under different experimental conditions

The distribution trend under other working conditions was consistent with that under 2.6 MPa,so in this section only 2.6 MPa was taken as an example.Fig.13 shows that the number of particles from 0 to 330 μm was higher,accounting for 80% of the total particle number,while the overall distribution showed a fluctuating upward trend.In addition,a greater number of particles were captured in the small size range under low-pressure conditions.

In addition to observing the distribution of particles,the distributions of the average,maximum,and minimum particle sizes under different experimental conditions were obtained by statistics across the entire time period.The specific distribution is shown in Table 5.

Although the pressure was not high,the hydrate production in the experiment was basically kept at about 0.2 mol after calculation.However,the morphology of hydrate particles as well their size are different under different initial pressures,so the micromorphology of hydrate particles is more sensitive to pressure changes.

The effect of the initial pressure on the size is shown in Fig.6 and Table 4.The results showed that under a pressure of 2.6 MPa,the main particles were mainly consisted of small planar flakes,with a maximum particle size of 933 μm,of which flake particles can reach a maximum of 605 μm.When the particles with main trunks were at 2.8 MPa and 3.0 MPa,they continued to grow close to 1000 μm.Then the difference between the length of main branches increased,and the maximum diameter of the dendritic granules can reach more than 825 μm.When the experiment was at 3.2 MPa,the snowflake particles grew into multilayer particles with uneven surfaces and curved horizontal surfaces,which were based on the growth of the planar flake structure.The dendritic granules started to grow from single-model to multiple trunk distributions,forming a larger particle size.When the particle size was 3.4 MPa,the trunk structure was more clearly visible,and the number of planar flake structures was lower.The maximum particle size was 1758 μm,while the maximum size of polyhedral solid particles was more than 1210 μm.At 3.6 MPa of pressure,the particles were basically filled with dendritic granules and polyhedral solid particles,and the maximum particle size was more than 1883 μm.

4.2.2.Effect of rotation rate on particle size

At different rates,the trend of particle size distribution is clearer,as shown in Fig.14 and Fig.15.It can be seen from Fig.14 (a) (R2=0.96) and Fig.15 (a) (R2=0.99) that the variation of the particle number with the size range was generally a normal distribution,and the corresponding particle size was between 165 and 275 μm when the curve reached the peak value.From the distribution trend in Fig.14 (b) and Fig.15 (b),the maximum size of hydrate kept changing,basically fluctuating from 385 to 1210 μm regardless of where the particles were located.The distribution of the smallest particle size was a horizontal line on the whole;the average size of the system fluctuated in the range of 220-248 μm.

Fig.14. The particle distribution at gas-liquid interface under 100 r·min-1.

Fig.15. The particle distribution at gas-liquid interface under 150 r·min-1.

At a certain rotation rate,the thickness of the hydrate layer on the interface increased with time until the flocculation layer gradually aggregated and extruded into a stable sedimentary layer.When the rate was suddenly increased,the increased shear force sloughed the original flocculated particles.This brought the particles near the gas-liquid interface first,increased the number of particles,and promoted the aggregation process at the new speed.Moreover,the above changes explain the reasons for the changes in particle morphology at different rotation rates.The formation of chain particles is caused by dendritic granules at a low rotation rate.This is owing to the trend in which the particles fall off the branches and continue to develop under the action of high shear,leaving only the trunk colliding with other trunks,forming chain particles and attaching some free particles.

4.3.Conversion between growth and aggregation of hydrate particles

Fig.16. Gauss fit of particle distribution (2.6 MPa).

Through the aforementioned changes in particle size distribution and morphology,the growth of large particles was accompanied by shearing and sloughing in addition to growth and aggregation [47-49].Hydrate forms different shapes under different dominant actions[50,51].From the time of particle appearance and the connection among particles,the snowflake particles are the dominant morphology of growth.

After recording the specific process of hydrate particle aggregation,the particle size distribution obtained at different time intervals was fitted by Gauss fitting.Since the variation in particle size with time basically conformed to a normal distribution,the particle size corresponding to the maximum difference between the corresponding curves in different time periods fitted by Gauss was regarded as the maximum particle size that could be achieved by the growth.This simplifies the formation process of particle aggregation,as shown in Fig.16.

These two fitting lines represented the change of the particle number in the rapid increase stage and the later stable stage in the experiment.Fitting line a (R2=0.94) with a peak value of the normal distribution was regarded as the small particle maximum number of the growth,and line b (R2=0.99) was regarded as the particle number curve when the hydrate aggregated and reached a stable state.The size corresponding to the maximum difference between the two fitting lines was considered to be an indicator that the function of the hydrate particle growth began to weaken.That is,after this size,the process of particle growth gradually changed from growth-dominated to aggregation-dominated.

According to the statistics of the maximum-growth dominant particle size under different experimental conditions,the change of the particle number with a size range can be represented by a normal distribution curve.At the same time,when the maximum difference between the two fitting lines was reached and the corresponding range increased from 165 to 413 μm with pressure increasing from 2.6 MPa to 3.6 MPa,the particle size range became larger and gradually became stable,as shown in Table 6.

Table 6 Aggregated particle size distribution under different experimental conditions

Fig.17. Microscopic physical model for the whole process of gas hydrate particle growth,aggregation and adhesion.

Therefore,with an increase in pressure,the maximum-growth dominant particle size will increase,which improves the possibility of forming larger particles.This is because the higher pressure increases the sub-cooling degree and the driving force of formation[52,53].In addition,the formation period is longer,resulting in a delay of appearance of the aggregation-dominant particle size,so the maximum-growth dominant particle size increases.Moreover,the particles tend to adhere to the flocculation layer,which makes the trend of particle aggregation and growth more insignificant.The gradual increase in the average particle size indicates that the aggregation ability among particles is becoming stronger,which promotes the formation of particles with larger sizes.Therefore,an increase in pressure increases the possibility of aggregation among particles and promotes the development of particles to large sizes until the entire system reaches an equilibrium state.

From the above experiments and data analysis,it can be seen that when growth was dominant in the early stage,the initial aggregated particles and planar flake particles were mostly distributed.After reaching the maximum-growth dominant particle size,owing to the transformation into an aggregation-dominant role,the polyhedral solid particles dominated at this stage.According to the experiments,it can be seen that the aggregation of dendritic granules was dominant,and their distribution was greater throughout the experimental process.Therefore,we can regard the maximum-growth dominant particle size of the dendritic granules as the size-critical value of the dominant role in the process of particle growth.

5.Micro-Physical Model of Growth and Aggregation Near the Gas-Liquid Interface Under Stirring Condition

Snowflake particles have larger contact areas,which is conducive to the attachment of free particles.Dendritic granules have a trunk structure,making it easy to intercept small particles in the fluid.Therefore,both planar and solid particles are prone to aggregation.After reaching the maximum-growth dominant particle size,the particles continue to grow toward themselves and aggregate into large particles.In this study,snowflake particles are mainly distributed in the form of dendritic particles owing to their large resistance and mass distribution.

Meanwhile,according to long-term observations,the thickness of the initially formed hydrate film was only a few microns [54-56].The volume of hydrate in the film was very small owing to the thin film,and it was found that the flocculation layer occurred mainly in the form of film thickening.The flocculation layer is formed by the continuous aggregation and adhesion of particles,and the speed is faster at the beginning because the position is directly contacting with the gas.During the formation of the flocculation layer,the rate gradually decreases because the aggregated hydrate hinders the direct contact between gas and liquid.The shearing action causes the attached particles to fall off or disperse particles with weak connectivity among them,resulting in a sharp increase of the particles number,which affects the thickness of the flocculation layer in a small dynamic adjustment process.

The initially formed hydrate film is somewhat porous or fractured (cannot absolutely isolate gas from water) [57].Since hydrate is hydrophilic and the film is porous,water can go through and wet both sides of the hydrate film.The driving force for hydrate formation recovers after water migration,which forms a new hydrate at the film interface from gas and water.The newly formed hydrate is also porous or fractured and can adsorb water from its surroundings owing to osmosis.The flocculation layer formation,water adsorption,and new hydrate particles formation processes are repeated over and over,resulting in a thickening of the hydrate flocculation layer [58-60].Based on the aforementioned distribution and morphological changes in particle size and the one-dimensional hydrate model proposed by Chen [58],a physical model of hydrate growth and aggregation was obtained as shown in Fig.17.This is convenient for analyzing the hydrate particle morphology and change trends.

Based on experimental observation,it is showed that large particles generally exist in the later stage.The proportion of small particles is largest in the early stage and then gradually decreases.It can be concluded that the existence and growth of large particles are mainly dominated by aggregation,while the existence and growth of small particles is dominated by growth.The transition point of the proposed growth mechanism mainly depends on the maximum growth dominant particle size.This ranged from 165 to 413 μm in this study,which corresponds to the active growth range of snowflake and dendritic granules.

Moreover,there is not a clear boundary between growth stage and aggregation stage.Therefore,in this experiment,the particle size and the time when the dendritic granules began to dominate the distribution were regarded as the largest-growth dominant particle size and the starting time when aggregation played a dominant role.In summary,the maximum growth dominant particle size was between 220 and 495 μm,and the aggregation time was between 200 and 500 min,which were positively correlated with the pressure value.

6.Conclusions

This work focused on the morphological evolution of growth and aggregation of natural gas hydrate near the gas-liquid interface under stirring condition.Experiments were conducted in a highpressure visual reactor with high-speed camera to capture the micromorphology of hydrate particles in a natural gas+pure water system with pressure between 2.6 MPa and 3.6 MPa and subcoolings range of 4.7 to 6.23 °C,which revealed the following conclusions:

At First,the initial aggregated structure was formed at a low volume concentration of 1%.The initially formed natural gas hydrate particles appeared as tiny aggregated particles with a low volume concentration.It was observed that the gas hydrate particles were mainly dendritic and polyhedral.Under low subcooling conditions,the total distribution trend of particles in the range of 0 to 330 μm first increased and then decreased,during which the proportion of small particles remained larger.

Secondly,the pressure and rotational rate had great influence on the morphology and size distribution.Increasing pressure improved the driving force of the hydrate formation process and increased the number of hydrate particles and the gap of the particle numbers in different particle size ranges.Speeding up the rotation rate fractured the formed flake particles into chains or smaller particles,which weakened the connection between the falling particles and formed a more stable particle shape at another high rotational speed.

Thirdly,to better distinguish the morphological boundary between particle growth and aggregation,we used the maximum dominant size of dendritic granules as the size-critical value of the dominant role.This helped to better specify each link in the hydrate formation process and was conducive to controlling the occurrence of hydrate aggregation.

Finally,a micro-physical evolution model of hydrate growth and aggregation was proposed.The physical model was improved by combining the change in the flocculation layer thickness with the phenomenon of crystal growth and aggregation,which described the change process of particle size and morphology and better reflected the changes in hydrate particles during the entire process.The results of this study can provide a guidance for flow assurance of multiphase pipeline,and further studies on stirring and other influencing factors are needed.Moreover,taking into account the limitations of intermittent photography and the rotary shear in the autoclave,more experiments should be conducted in pipelines to supplement the microstructural and kinetic data and improve the microscopic model.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (51974349,U19B2012,51991363),the Natural Science Foundation of Shandong Province(ZR2017MEE057),which are gratefully acknowledged.