- Email: [email protected]
Clathrate-hydrate formation from a hydrocarbon gas mixture: Compositional evolution of formed hydrate during an isobaric semi-batch hydrate-forming operation
Applied Energy 113 (2014) 864–871
Contents lists available at ScienceDirect
Applied Energy journal homepage: www.elsevier.com/locate/apenergy
Clathrate-hydrate formation from a hydrocarbon gas mixture: Compositional evolution of formed hydrate during an isobaric semi-batch hydrate-forming operation Wataru Kondo a,1, Kaoru Ohtsuka a, Ryo Ohmura a, Satoshi Takeya b, Yasuhiko H. Mori a,⇑ a b
Department of Mechanical Engineering, Keio University, Yokohama 223-8522, Japan Research Institute of Instrumentation Frontier, National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba 305-8565, Japan
h i g h l i g h t s Reports laboratory experiments of hydrate formation from synthetic natural gas. Shows how hydrate composition changes during each hydrate-forming operation. Compares the observed compositional change with thermodynamic simulations. Shows a PXRD result indicating simultaneous formation of sI and sII hydrates.
a r t i c l e
i n f o
Article history: Received 16 August 2012 Received in revised form 26 July 2013 Accepted 13 August 2013 Available online 7 September 2013 Keywords: Clathrate hydrate Gas hydrate Mixed hydrate Natural gas storage
a b s t r a c t The clathrate hydrate formation from a model natural gas, i.e., a mixture of methane, ethane, and propane in a 90:7:3 molar ratio, under a constant pressure was experimentally investigated, focusing on the compositional evolution of hydrate crystals formed inside a gas-bubbling-type reactor during each semibatch hydrate-forming operation. The experimental system used in this study was specially designed for obtaining several hydrate samples formed at different, arbitrarily selected stages during each hydrate-forming operation. Each hydrate sample was analyzed by a gas-chromatograph to determine the mole fractions of methane, ethane and propane encaged in the hydrate. These analyses revealed a monotonic increase in the methane fraction and decreases in the ethane and propane fractions during each operation until a quasi-steady state was established. Powder X-ray diffraction analyses showed that both structure-I and structure-II crystals were simultaneously formed during the quasi-steady period. The compositional evolution of the hydrates formed during the early stages before the quasi-steady state was reached deviated from corresponding predictions based on the thermodynamic-simulation scheme previously reported. A hypothetical explanation for the discrepancy between the experimental and simulation-based results was provided. Ó 2013 Elsevier Ltd. All rights reserved.
1. Introduction This paper describes an experimental study of the hydrate formation from a multi-component gas mixture. Understanding the behavior of such a hydrate-forming process is important for hydrate-based technology development for storing and transporting natural gas, separating undesirable (toxic or incombustible) species from biogases or natural gas with a high carbon-dioxide content, etc. Specifically, this paper focuses on the evolution in the composi⇑ Corresponding author. Tel./fax: +81 45 701 2054. E-mail address: [email protected] (Y.H. Mori). Present address: Plant Management Group, Takehara Power Station, Electric Power Development Co., Ltd., Tadanoumi-Nagahama, Takehara-shi, Hiroshima Prefecture 729-2329, Japan. 1
0306-2619/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.apenergy.2013.08.033
tion of the hydrocarbon guest being captured in hydrates during their continuous formation from a hydrocarbon gas mixture simulating a natural gas. Such evolution originating from the nature of fractionation of hydrocarbons in forming hydrates possibly causes the lack of compositional uniformity of the hydrate products; hence it is of great importance concerning the industrial hydrate-forming operations for natural gas storage and transport [1]. For appropriately designing and controlling such operations, we need to have an insight into how the gas-phase composition in a hydrate-forming reactor, and hence the hydrocarbon uptake into the instantaneously formed hydrate, would change during each operation. Obviously, it is extremely difficult to find a solution fully satisfying the above task, because each hydrate-forming operation includes highly complicated multi-phase dynamic processes such as the convective
W. Kondo et al. / Applied Energy 113 (2014) 864–871
mass (hydrocarbons) transfer to, and heat transfer from, hydrate crystals thereby causing their growth, each dependent on the reactor design and the system-operational parameters (pressure, temperature, liquid stirring, etc.). However, it may be still possible and, if possible, should be valuable from an engineering viewpoint to find some practical means that allows us to estimate the compositional changes in the gas phase and the hydrate product during each hydrate-forming operation. As a tool for expeditiously estimating such compositional changes, Tsuji et al. [2] first presented a thermodynamic simulation scheme, in which the fluids and, if any, a hydrate confined in the reactor are assumed to form an open, instantaneously equilibrated system undergoing a quasi-static irreversible change in response to the successive discharge of the formed hydrate from, and the supply of the feed gas and water to, the reactor. Later, a modified, more sophisticated scheme for such simulations was constructed by Ogawa et al. [3]. These simulation schemes cannot predict the hydrate-formation rate in a given hydrate-forming system, hence they cannot predict how fast the gas phase and the hydrate product change their compositions with time; this is because the schemes are entirely based on the thermodynamics-based characterization of the system, and none of the rate-controlling factors (the size and geometry of the reactor, the gas–liquid interfacial area, the dynamics of the gas–liquid mixing, etc.) are taken into account. Instead, the schemes can predict, irrespective of the size and type of the reactor, how the gas phase and the hydrate product change their compositions with the progress in the fractional replacement of the gas phase with the fresh feed gas (a hydrocarbon mixture with a fixed composition) continuously supplied to the reactor in order to compensate the loss of each hydrocarbon species from the gas phase due to the hydrate formation. Although the simulation schemes mentioned above can possibly be useful tools for designing hydrate-forming systems and their operations, we should be careful about possible deviations of the simulation-based estimates from the actual hydrate-forming behavior. Deviations possibly result from the modeling of hydrate-forming operations as thermodynamically quasi-static processes, which plays a key role in the simulation schemes [2,3]. A few attempts have been made at comparing simulation-based estimates to the corresponding experimental data concerning the change in the gas-phase composition during each hydrate-forming operation [4–6], which indicated a reasonably good agreement between them (except for the later stages of each hydrate-forming process in an unstirred semi-batch reactor containing a surfactant solution [6]). As for the compositional evolution in the hydrate product, however, no experimental examination corresponding to those for the gas-phase evolution [4–6] has been reported so far. For further clarification, this point may need a supplementary explanation, which follows. We find in the literature some previous studies in which the crystalline structures and hydrate-guest compositions in the hydrates formed from hydrocarbon mixtures were experimentally analyzed. See, for example, the papers by Uchida et al. [7], Schicks et al. [8], Hester et al. [9] and Seo et al. [10]. However, none of these studies demonstrated how the composition of the instantaneously-formed hydrate changed during each isobaric hydrate-forming process properly simulating industrial hydrateforming operations for natural-gas storage. In this paper, we describe our attempt at experimentally observing the compositional evolution of the hydrate formed in a gas-bubbling-type reactor to which a model natural gas (a mixture of methane, ethane, and propane in a 90:7:3 molar ratio) is continuously supplied to compensate for the loss of the gas due to hydrate formation, thereby maintaining the pressure inside the reactor constant. The hydrate-composition data obtained in these experiments are compared to the simulationbased predictions.
865
2. Experimental section 2.1. Experiment-design concept The experimental work performed in this study was designed to extend our previous experimental study [5], in which we investigated how the gas-phase composition inside a gas-bubbling-type reactor changed during each hydrate-forming operation by continually sampling the gas from the reactor. The reason we selected a gas-bubbling-type reactor, instead of water-spray and other types [1], was simply for the convenience of obtaining gas samples which were free from water mist and uniform in composition. This time, we intended to obtain hydrate samples for their compositional analyses using the same reactor that we used for the gas-phase analyses [5]. The major technical difficulty in obtaining hydrate samples for compositional analyses was the fact that hydrate crystals formed at various stages of each hydrate-forming operation were inseparably mixed inside the reactor, hence we could not separately remove hydrate samples formed at specified stages from the rector. To overcome this difficulty, we installed a hydrate sampler (i.e., a detachable, small-capacity reactor) in parallel with the main gas-bubbling reactor for only a short period in order to obtain a hydrate sample representing the current stage. By repeatedly attaching and detaching the sampler during each hydrateforming operation, we could collect hydrate samples representing different stages of the operation. The main reactor exclusively played a role of progressing the operation by continuously forming hydrates during every interval between successive hydrate-sample-forming procedures. The gas mixture thus aged in the main reactor due to the hydrate formation and the feed-gas supply that compensates for the gas consumption by the hydrate formation was circulated in a loop connecting the main reactor and the sampler during each hydrate-sample-forming procedure. This was to ensure that the hydrate sample was formed from essentially the same gas mixture as the one currently filling the main reactor. Technical details with which we realized the above experimentdesign concept are described in Sections 2.3 and 2.5. 2.2. Materials used The feed gas was a mixture of 90.07% methane, 6.93% ethane and 3.00% propane on a molar basis (synthesized and analyzed by Takachiho Chemical Industrial Co., Ltd., Tokyo, Japan). The deionized and distilled water was used for the hydrate formation. Sodium dodecyl sulfate (SDS) with a certified purity of 99.5% mass fraction (supplied by Nacalai Tesque, Inc., Kyoto, Japan) was added to the water at the concentration of 500 ppm (500 mg/kg) to prepare an aqueous SDS solution for use in forming hydrate samples for compositional analyses; this was just to promote hydrate formation, thereby shorten the period for forming each sample. 2.3. Apparatus Fig. 1 illustrates the experimental apparatus used in this study. This apparatus was modified from that used in our previous study [5] by installing a device for forming hydrate samples for compositional analyses. Except for this device set in parallel with the main hydrate-forming reactor on the gas circulation loop, the layout of the apparatus as well as the components assembled in it are the same as those already reported [5]. For example, the main hydrate-forming reactor is the same coolant-jacketed gas-bubblingtype 754-cm3 cell that we used in the previous study [5]. This reactor is equipped with a disk-type bubbler immersed in a water pool and a stirrer for providing a good mixing in the gas phase. The reactor is installed on the gas-circulation loop, through which the gas
866
W. Kondo et al. / Applied Energy 113 (2014) 864–871
Fig. 1. Schematic of the experimental apparatus.
mixture forming a continuous gas phase inside the reactor is sucked out and, having been mixed with a fresh feed gas supplied from an external high-pressure cylinder, sent back to the reactor through the bubbler with the aid of a high-pressure fluid-circulation pump. The gas flow rate through the pump, i.e., the rate at which the gas mixture is sucked from the reactor into the loop, is monitored by a mass flow meter. On the downstream side of this mass flow meter, the feed-gas supply line from a high-pressure gas cylinder merges with the loop, such that the feed gas with the fixed composition is allowed to flow, together with the circulated gas mixture, into the reactor, thereby compensating for the loss of the gas mixture due to the hydrate formation inside the reactor. A mass flow meter installed on the feed-gas supply line allows us to continuously measure the instantaneous gas flow rate within an accuracy of ±10.0 cm3/min NTP (converted to a volumetric flow rate under the ‘‘normal temperature and pressure’’ defined as 273.15 K and 101.3 kPa). A bypass line was newly attached to the gas circulation loop such that we could make, by manipulating a three-way valve on the loop, the gas pumped through the loop bypass the lower part of the main reactor containing the water pool and the bubbler and flow into a hydrate sampler, i.e., a cylindrical, small-capacity (10 cm3 inside volume) gastight borosilicate-glass tube equipped with stop valves at its top and bottom ends (a Hyper Glass Cylinder manufactured by Taiatsu Techno Corp., Ltd., Tokyo). This sampler is initially charged with 7 cm3 of an aqueous SDS solution (500 ppm in SDS mass fraction) and maintained at the same pressure–temperature condition as that in the main reactor. Due to the well-recognized SDS effect for promoting hydrate formation from hydrocarbons (see, for example, Refs. [11–13]), we can expect that a high water-to-hydrate conversion ratio will be achieved during a relatively short period, as compared to the time span for the entire evolution of the hydrate-forming system confined in the main reactor, while the gas is being forced to pass the bypass line. The gas flowing out of the sampler without having been consumed by the hydrate formation is then discharged into the upper gas-phase space inside the main reactor to be mixed with the gas remaining there before flowing into the circulation loop. Once the hydrate formation inside the sampler is deemed to be completed, we can
switch the gas flow back to the ordinary path through the main reactor and remove the sampler from the bypass for a gas-chromatographic analysis of the hydrate filling the sampler. The assembly that we used for sampling the gas phase in the main reactor and analyzing the samples by a high-speed gas chromatograph in our previous study [5] was used again with no changes in this study. This time, we applied this assembly for the purpose of analyzing the gas released by dissociating each hydrate sample; this is described later in some detail. 2.4. Procedure In advance of each experimental run, the main reactor and the gas-circulation loop were evacuated while the loop was closed from the bypass line. The reactor was then charged with 300 cm3 of water, which formed a 63-mm deep pool on the surface of the bubbler. The feed gas was supplied to fill the remaining 454-cm3 space inside the reactor plus the 262-cm3 space inside the loop until the inside pressure had increased to 2.9 MPa. Besides the above preparation of the main reactor and the gas-circulation loop, the sampler charged with 7 cm3 of the aqueous SDS solution was mounted on the bypass line and immersed in a thermostated water bath. After the temperature inside the main reactor and that of the bath holding the sampler became constant at about 275 K, the pump on the gas-circulation loop and the stirrer in the main reactor were turned on. Throughout every experimental run, the gas flow rate through the loop and the rotation rate of the stirrer were controlled at 360 cm3/min and 400 rpm, respectively, while the pressure and temperature inside the chamber were held at 2.9 ± 0.05 MPa and 275 ± 0.5 K, respectively. Once the hydrate formation was detected by monitoring the feed-gas supply to the reactor as well as the visual observations through the sight window of the reactor, we proceeded to prepare for the hydrate sampling operations, the details of which are described in the next subsection. 2.5. Hydrate sampling/analysis details During each experimental run, we typically carried out the hydrate-sampling operations four times at arbitrary intervals in order
W. Kondo et al. / Applied Energy 113 (2014) 864–871
to obtain hydrate samples formed at different stages over a transient hydrate-forming process asymptotically approaching a quasi-steady state.2 The procedure of each hydrate-sampling operation is described below. When a specified, though arbitrarily selected, time had passed after the start of hydrate formation inside the main reactor, we switched the gas flow in the circulation loop from the ordinary line passing through the main reactor to the bypass line by manipulating the three-way valve placed at the upstream-side junction of the ordinary and bypass lines (see Fig. 1) and, at the same time, opening the stop valves at the inlet and the outlet of the hydrate sampler. When we noted, based on the visual observation and the gasflow monitoring on the feed-gas supply line, that the hydrate formation inside the sampler had almost ceased, we closed the above stop valves, and then manipulated the three-way valve to turn the gas-flow path to the ordinary line to enable the further progress of the hydrate-forming process. After the resumption of the hydrateforming process inside the main reactor, we removed the sampler from the bypass line, and attached a new sampler, which had been pre-cooled to about 275 K in advance, to the bypass line to prepare for the next hydrate-sampling. The hydrate sample obtained by each sampling operation outlined above was specified by ng, the moles of the feed gas supplied to the hydrate-forming system (the fluids + hydrate contents in the main reactor, the sampler, and the gas-circulation loop including the bypass line) over the time span between the inception of hydrate formation in the main reactor and the formation of the hydrate sample inside the sampler3, or by the ratio of ng to ng0, the moles of the feed gas with which the hydrate-forming system was initially charged. The rest of the procedure of handling the hydrate sample is outlined below. For temporary use, a pressure gauge was attached to the sampler just removed from the bypass line. The stop valve at the top of the sampler was then opened to discharge the gas, that had occupied a small proportion of the inside volume of the sampler, until the inside pressure read by the pressure gauge decreased to 0.1 MPa. This gas discharge operation was performed holding the sampler in the water bath maintained at 275 K. After discharging the gas, the sampler was removed from the bath into the laboratory air to make the hydrate inside the sampler dissociate. After completion of the hydrate dissociation, the sampler containing the gas released by the hydrate dissociation was connected to a branch of the gas sampling line that extended from the main reactor to the gas chromatograph (see Fig. 1). The subsequent procedure for analyzing the gas supplied from the sampler was just the same as that we used for analyzing the samples taken from the gas phase inside the main reactor and described in detail in our previous paper [5]. 2.6. Powder X-ray diffraction measurements In addition to the gas-chromatographic analyses of hydrate samples throughout each hydrate-forming process, we performed powder X-ray diffraction (PXRD) analyses of the hydrate samples formed during the quasi-steady asymptotic regime (ng/ng0 > 0.55) in which the gas-chromatography-based hydrate composition remained almost constant. The hydrate samples for the PXRD mea2 The term ‘‘quasi-steady state’’ defines an asymptotic regime for each hydrateforming operation in which any systematic change in composition was no longer observed with the gas phase remaining inside the reactor or with the formed hydrate. Note that any ‘‘steady state’’ in its rigorous sense cannot be achieved as long as the operation is the semi-batch type, instead of the continuous type. 3 It took 15–30 min to form each hydrate sample. Hence, ng increased, though only slightly, from its initial value, ng,1, to its final value, ng,2, during the formation of one hydrate sample. For simplicity, we employed the arithmetic mean of ng,1 and ng,2 as the ng value representing this hydrate sample.
867
surements were prepared through a two-step freezing procedure which is outlined below. The hydrate sampler removed from the hydrate-forming apparatus (Fig. 1) was first immersed in a bath of an aqueous ethyleneglycol solution controlled at 30 °C to freeze the contents of the sampler except for the gas mixture occupying the top portion inside the sampler. After discharging the gas mixture by opening the valve at its top port, the sampler was immediately placed into a liquid-nitrogen bath for further cooling of its contents. The sampler was then removed from the bath, its bottom screw cap was removed in a cold nitrogen-gas atmosphere, and the hydrate sample (to be more exact, a hydrate + ice mixture) in the amount of several grams was scraped from the sampler. This sample was finely ground in a nitrogen-gas atmosphere at a temperature below 100 K, then top-loaded onto a copper-made specimen holder for PXRD analysis. The loaded sample was exposed to Cu Ka radiation generated by an Ultima III diffraction system (Rigaku Corp., Tokyo, Japan) in a parallel-beam optics alignment. Each measurement was performed in the h/2h scan mode with a step width of 0.02° at 123 K. 2.7. Note on the use of SDS The objective of this note is to clarify the readers’ possible concern that the use of SDS in forming the hydrate samples for compositional analyses might have added an unnecessary complexity to the hydrate-forming system dealt with in this study. One may be concerned about the difference in the hydrate-forming conditions between the SDS-containing sampler and the SDS-free main reactor. Such concerns are quite groundless. Due to its molecular size, SDS cannot be encaged in any types of clathrate hydrates. Therefore, SDS does not have any crystal-structural effect on the hydrate formation. Moreover, it has been experimentally confirmed that the SDS addition to the orders of 100–1000 ppm in water results in no practical change in the thermodynamic hydrate-forming conditions for methane, methane + carbon dioxide, and methane + ethane + propane [14,15]. Based on these facts, we can expect that the SDS addition altered neither the crystalline structures nor the guest compositions of the hydrate samples while it significantly promoted the hydrate nucleation and growth, thereby significantly shortening the time required for forming each hydrate sample. 3. Results and discussion We have performed nine experimental runs under nominally the same operational conditions as those specified in the preceding section. Each run included an isobaric hydrate-forming operation, which extended to 24–96 h. The obtained results are shown below, particularly focusing on the evolution of the hydrate-phase composition through each hydrate-forming operation. (A complete set of experimental data relevant to the description which follows is provided in the electronic Supplementary data.) It should be noted that the rate of hydrate formation is not a subject of interest in this study, similar to our recent studies about hydrate formation from hydrocarbon mixtures [3,5]. Fig. 2 summarizes the data obtained in the nine experimental runs in the form of xi,hyd versus t plots, where xi,hyd denotes the water-free-base mole fraction of each hydrate-guest species (i = 1 for methane, i = 2 for ethane, and i = 3 for propane) contained in each hydrate sample and t denotes the time lapse after the onset of the hydrate formation inside the main reactor. The xi,hyd values for the three individual species were deduced by normalizing the P relevant gas-chromatographic data such that ixi,hyd = 1. Along the t axis, each data point is located at the midpoint of the
868
W. Kondo et al. / Applied Energy 113 (2014) 864–871
operation from the real time t to ng/ng0 that expresses, in a normalized form, the degree of replacement of the gas mixture filling the system with the feed gas continuously supplied to the system from the outside. As previously discussed [5], ng/ng0 is considered to be a universal index, independent of the system scale as well as the hydrate-formation rate, of the progress in compositional evolution of the gas mixture inside the system during each hydrate-forming operation. Fig. 3 shows the xi,hyd versus ng/ng0 plots of the same experimental data that we used in preparing Fig. 2. Also shown for comparison are the relevant predictions by the thermodynamic simulations based on the scheme that we previously developed [3].
Fig. 2. Plot of xi,hyd, the measured water-free-base mole fractions of the three hydrocarbon species (i = 1 for methane, i = 2 for ethane, and i = 3 for propane) contained in each of the hydrate samples formed at different stages during each semi-batch hydrate-forming operation under the constant pressure of 2.9 ± 0.05 MPa. The axis of time t represents the time lapse after the inception of hydrate formation in the main reactor.
15–30 min period which elapsed during the formation of the relevant hydrate sample inside the sampler.3 Inspecting the xi,hyd versus t plots in Fig. 2, we can estimate that the composition of the newly formed hydrates became almost constant within 20 h following the inception of the hydrate formation. However, this observation is dependent on the apparatus and operating conditions employed in the present experiments, and hence lacks generality. Besides, the experimental results arranged in the form of xi,hyd versus t plots cannot be compared to the predictions given by the thermodynamic simulation scheme that we previously developed [3], because the scheme is inherently free from any quantitative time scale. In order to provide the experimental data with a generality such that one can compare them to appropriate simulations as well as, though not yet available at present, corresponding data generated by other hydrate-forming equipment which may be different in size and/or type from ours, we need to convert the index for the aging of the hydrate-forming system during each hydrate-forming
Fig. 3. Plot of the same xi,hyd data as those shown in Fig. 2 with the replacement of the time scale t by the gas-phase replacement index, ng/ng0, i.e., the ratio of ng, the moles of the feed gas having been supplied to the hydrate-forming system (the fluids + hydrate contents in the main reactor, the sampler, and the gas-circulation loop including the bypass line) after the inception of hydrate formation in the main reactor, to ng0, the moles of the feed gas with which the hydrate-forming system was initially charged. The horizontal bar for each data point indicates the ng/ng0 increment during the relevant hydrate-sample formation. The solid and dotted curves show the evolution of the xi,hyd values for the instantaneously formed hydrates predicted by two thermodynamic simulations for an isobaric (at 2.90 MPa), quasi-static hydrate-forming operation. These simulations were provided by the same simulation scheme [3] into which two different PECPs, CSMGem [16] and HWHydrate [17], were alternatively incorporated.
W. Kondo et al. / Applied Energy 113 (2014) 864–871
Two different predictions (plotted by solid and dotted curves, respectively) are discriminated, with respect to the phase-equilibrium calculation program (PECP) incorporated in the simulation scheme. Following the way of our previous studies of this series [3,5], we alternatively incorporated two different PECPs, CSMGem [16] and HWHydrate [17], into the simulation scheme such that we can take an objective view of the PECP-dependent results of the simulations. To help our understanding of the results of these simulations, we show in Fig. 4 how these simulations predict the occupancies, by the three guest species, of the 51264 cages in structure II (sII) or 51262 cages in structure I (sI) of the instantaneously formed hydrates throughout the entire hydrate-forming process. As demonstrated in Fig. 3, the experimental xi,hyd data showed a simple asymptotic change with an increase in ng/ng0. That is, x1,hyd increased from its initial level, 0.8, to its asymptotic level, 0.9, before ng/ng0 increased to 0.5. In parallel with this change in x1,hyd, x2,hyd and x3,hyd monotonically decreased from 0.13 to 0.07 and from 0.07 to 0.03, respectively. It should be noted that, over the entire ng/ng0 range above 0.5, x1,hyd, x2,hyd and x3,hyd were almost constant at about 0.9, 0.07 and 0.03, respectively, in good agreement with the composition of the feed gas. This fact means that, during such an asymptotic region, the hydrate-forming system was in a quasi-steady state thereby yielding an instantaneous mass balance between the feed-gas supply to the system and the hydrate formation with respect to each hydrateguest species. In Fig. 3, we find significant differences in the xi,hyd evolution behavior between the experimental results and the simulations. Whichever PECP (CSMGem or HWHydrate) we used, the simulations predicted a sharp initial decrease in x3,hyd from 0.3 to 0.03 and a simultaneous increase in x2,hyd as a result of the competitive occupation of the 51264 cages of the sII hydrates by propane and ethane molecules (see Fig. 4). That is, the preferential occupation of the 51264 cages by propane molecules during the earliest stages results in a rapid depletion of propane in the gas phase (see Fig. 5 in Ref. [5]), thereby allowing ethane molecules to be subsequently included in the 51264 cages. As a result of the propane depletion and the subsequent ethane depletion in the gas phase, x2,hyd has a peak at ng/ng0 0.2. However, neither the initial sharp decrease in x3,hyd nor the peak of x2,hyd at ng/ng0 0.2 were
Fig. 4. Evolution of the occupancies, by the three guest species, of the large cages (51264 cages in sII or 51262 cages in sI) of the instantaneously-formed hydrates during each hydrate-forming process. The predictions by the two simulations, one using CSMGem and the other using HWHydrate, are plotted by the solid and dotted lines, respectively. Note that the simulation using HWHydrate predicts an alternate formation of the sII and sI hydrates over the quasi-steady region of the process (ng/ng0 > 0.55).
869
Fig. 5. PXRD profile (measured at 123 K) of a hydrate sample formed in the ng/ng0 increment from 1.13 to 1.15. The plus signs (+) denote the observed intensities; the solid line on the plus signs represents the intensities calculated from the best-fit model of the Rietveld refinement. The bottom curve shows the deviation between the observed and calculated intensities. The tick marks on the three rows represent the calculated peak positions for the sI hydrate, the sII hydrate, and the hexagonal ice (Ih), respectively.
observed in our experimental results. We later discuss this discrepancy between the simulations and the experimental observations. Another issue to be raised here is the difference between the CSMGem-based simulation and the HWHydrate-based simulation about their xi,hyd predictions over the quasi-steady region (ng/ng0 > 0.55). The CSMGem-based simulation predicts the exclusive formation of sII hydrates throughout the entire hydrate-forming process, hence suggesting smooth xi,hyd versus ng/ng0 curves as demonstrated in Fig. 3. In contrast, the HWHydrate-based simulation predicts an alternate formation of sII hydrates in two different compositions and also an sI hydrate (see Fig. 4), resulting in periodical jumps in xi,hyd. Integrating such discrete xi,hyd values over more than one cycle of successive formation of the three classes of hydrates (i.e., two classes of sII and one class in sI) along the ng/ng0 axis, we obtain the averaged values of x1,hyd, x2,hyd and x3,hyd very close to 0.90, 0.07 and 0.03, respectively, which are consistent with the mass-balance requirement for the hydrate-forming system under a steady condition. Whether an sI hydrate actually forms along with the sII-hydrate formation during the quasi-steady region, in which the methane fraction in the gas phase is the highest throughout the entire hydrate-forming process, is an intriguing problem. According to the Gibbs phase rule, sII and sI hydrates can simultaneously form in the present experimental system containing four chemical species, i.e., three hydrocarbons and water. (Readers may consult Section 4.2 in Ref. [2] concerning the phase-rule-based consideration on this point.) In fact, the coexistence of sII and sI hydrates in a four component (methane + ethane + propane + water) or five component (methane + ethane + propane + isobutene + water) experimental system has already been reported in the literature [8,10]. Thus, the formation of an sI hydrate together with an sII hydrate in our experimental system would, if actually observed, no longer be a new scientific finding by itself. However, no reliable means is currently available that allows us to predict if hydrates in both structures simultaneously form or a hydrate in either structure exclusively forms under a given experimental condition. Inevitably, we have to rely on some experimental means to know the actual crystalline structures of formed hydrates, which is an indispensable piece of knowledge for designing industrial natural-gas-hydrate production processes. Thus, we later discuss on this issue based on the PXRD measurements of the hydrate samples obtained in the quasi-steady region for each hydrate-forming process.
870
W. Kondo et al. / Applied Energy 113 (2014) 864–871
The significant discrepancies between the experiments and simulations concerning the xi,hyd versus ng/ng0 relations during the early transient region (ng/ng0 < 0.55) are possibly dependent on the gas-bubbling-type hydrate-forming scheme that we used in our experiments. We describe below our hypothetical interpretation of the above discrepancies, taking the observed hydrate-formation behavior into account. In either of the main reactor and the hydrate sampler (see Fig. 1), the gas mixture was released into a liquid pool in the form of discrete bubbles typically several millimeters in diameter, resulting in the hydrate formation on the surface of each bubble buoying up in the liquid pool besides the hydrate formation on the free surface of the pool. As already described in detail by Mori and Mori [18], such hydrate formation turned the gas bubbles into the form of hydrate balloons, each composed of a thin hydrate shell and a gaseous core, which often agglomerated into a hydrate foam layer lying on the liquid pool thereby intervening between the pool and the gas phase at the top of the reactor or the sampler. It should be noted that each bubble during its rise in the bulk of the pool or resting on the top of the pool was practically isolated from the gas mixture forming the continuous phase in the reactor, the sampler, or the gas-circulation loop. As the hydrate formation progressed, preferentially consuming propane and ethane, on the surface of each bubble, the gas phase inside the bubble should have rapidly depleted propane and ethane, thereby increasing the fraction of methane. Thus, the gas mixture actually involved in the hydrate formation inside the bubble must have been increasingly methane-rich as compared to the gas mixture which would have been available if complete gas-phase mixing inside the entire hydrate-forming system (as assumed in the simulations) was occurring. Concerning their hydrate-forming experiments by bubbling a seven-component hydrocarbon mixture into seawater confined in a cylindrical cell, Hester et al. [9] pointed out a possible deviation in composition between the bubbles and the continuous gas phase located at the top of the cell as a result of preferential hydrate formation on the bubble surface. Essentially the same phenomenon must have occurred in our gas-bubblingtype reactors. Consequently, the hydrates formed from the gas mixture confined in such discrete bubbles must have been more methane-rich than those which would be formed from the gas mixture ideally mixed at each instant. The in-bubble hydrate-formation process assumed above possibly accounts, at least in part, for the discrepancies in the evolution of xi,hyd during the early transient regime (ng/ng0 < 0.55) between the experimental observations and the simulations.4 It seems intriguing to investigate if such an evolution of xi,hyd varies when using a different hydrateforming scheme in which the feed gas makes contact with the aqueous phase taking the form of a dispersed phase instead of a continuous phase. 4 The hydrate formation primarily consuming the gas mixture confined in the bubbles, instead of that in the continuous gas phase occupying the top portion of the main reactor, may also account for some discrepancies in the evolution of xi,gas, the mole fractions of methane (i = 1), ethane (i = 2) and propane (i = 3) in the continuous gas phase, between the experiments and the simulations, which are demonstrated in Figs. 5 and 7 in our previous paper [5]. In these figures, we find that the increase in x1,gas and the simultaneous decreases in x2,gas and x3,gas with an increase in ng/ng0 up to 1.0 experimentally observed are less steep than those predicted by the simulations. This fact may be ascribable to the absence of gas mixing between the bubbles and the continuous gas phase from which the gas mixture was periodically sampled for the chromatographic analyses. Presumably, the continuous gas phase, which was initially at a composition equal to that of the feed gas (a mixture of methane, ethane, and propane in a 90:7:3 molar ratio), still remained richer with ethane and propane than the gas phase inside each bubble during the early transient region because the area over which the former gas phase was in direct contact with water, thereby allowing hydrate formation, was small (relatively to its volume) and hence its composition was less apt to change than that of the gas phase inside each bubble. Consequently, the compositional change in the continuous gas phase was possibly delayed compared to the change that the gas phase in an ideally mixed hydrate-forming system (as assumed in the simulations) would exhibit.
Hydrate samples formed in three different increments regarding ng/ng0, 1.13–1.15, 1.23–1.28 and 1.45–1.54, in the quasi-steady regime (see Fig. 3) were subjected to the PXRD measurements to confirm their crystalline structures. Similar PXRD profiles were obtained with these samples, and one of them is shown in Fig. 5. This PXRD profile, likewise those for the other two samples, indicates that this hydrate sample was a mixture of two types of hydrates, sI with the lattice constant of 11.852(3) Å and sII with the lattice constant of 17.113(1) Å, and the hexagonal ice Ih which must have been formed from the residual liquid while the sample scraped out from the sampler was chilled for the subsequent PXRD measurement. The sI-to-sII ratios for the three samples were estimated from the relevant PXRD profiles by the Rietveld method using the Rietan-FP program [19]; they were 0.25:0.75, 0.33:0.67, and 0.68:0.32, respectively. The significant scatter in these ratios was presumably ascribable to the spatial non-uniformity of the sI-tosII mixing in each sample, only a small portion of which was used in the PXRD measurement, rather than to the sample-to-sample difference in the sI-to-sII ratio representing each sample. Despite such a scatter in the estimated sI-to-sII ratios, we can reasonably claim that, in the hydrates formed in the quasi-steady asymptotic regime in each hydrate-forming process, the proportion of sI crystals was not minute but substantial, almost comparable to that of the sII crystals. This finding is qualitatively consistent with the prediction by the HWHydrate-based simulation. 4. Concluding remarks This study has experimentally investigated the hydrate formation from a simulated natural gas (a mixture of methane, ethane and propane in a 90:7:3 molar ratio) in an isobaric, gas-bubblingtype reactor, focusing on the compositional evolution of the hydrate instantaneously formed in the reactor throughout each semi-batch hydrate-forming operation under constant pressure (2.9 MPa). Using an auxiliary reactor, i.e., a detachable, smallcapacity gas-bubbling-type reactor, along with the main reactor, for obtaining a hydrate sample formed within a short period at an arbitrary stage during the above operation, we have succeeded in collecting compositional hydrate-product data over the entire process of the semi-batch hydrate-forming operation. These data revealed how the guest-molecule composition of the hydrate product changes from its initial state to an asymptotic (quasi-steady) state, with the gas-replacement progress in the hydrate-forming system resulting from the simultaneous gas uptake into the formed hydrate and the replenishment of the system with fresh feed gas (the ternary mixture simulating natural gas). Based on PXRD analyses, we have also revealed that both the sI and sII crystals were mixed in the hydrates formed during the asymptotic regime of the hydrate-forming process. The thermodynamic simulation scheme that we previously reported [3] was found to be effective for predicting the stage beyond which the hydrate-forming system falls in the asymptotic regime, i.e., the hydrate product maintains the constant guest-molecule composition equal to that of the feed gas. However, the scheme fails to predict the evolution of the guest-molecule composition during the earlier transient regime presumably due to the compositional nonuniformity in the gas mixture dispersed in the actual hydrate-forming system. It is desirable that this point should be further examined using different types of hydrate-forming reactors. Acknowledgments This study was supported in part by a Grant-in-Aid for the Global COE Program for the ‘‘Center for Education and Research of Symbiotic, Safe and Secure System Design’’ from the Ministry of
W. Kondo et al. / Applied Energy 113 (2014) 864–871
Education, Culture, Sports, Science and Technology, Japan. We thank Eiichiro Tachikawa, a former student at Keio University, for his help in the experimental work during the early stage of this study. Sanehiro Muromachi, a current graduate student in Keio University, is also acknowledged for his valuable technical advice about the preparation of hydrate samples for the PXRD analyses. Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.apenergy.2013.08.033. References [1] Mori YH. Recent advances in hydrate-based technologies for natural gas storage – a review. J Chem Ind Eng (China) 2003;54(Suppl.):1–17. [2] Tsuji H, Kobayashi T, Okano Y, Ohmura R, Yasuoka K, Mori YH. Thermodynamic simulations of isobaric hydrate-forming operations: formulation of computational scheme and its applications to hydrate formation from a methane + ethane + propane mixture. Energy Fuels 2005;19:1587–97. [3] Ogawa H, Imura N, Miyoshi T, Ohmura R, Mori YH. Thermodynamic simulations of isobaric hydrate-forming operations for natural gas storage. Energy Fuels 2009;23:849–56; Ogawa H, Imura N, Miyoshi T, Ohmura R, Mori YH. Thermodynamic simulations of isobaric hydrate-forming operations for natural gas storage. Energy Fuels 2011;25:1333. [4] Takahashi M, Kanda N, Sano K, Iwasaki T. Formation characteristics of multicomponent mixed gas hydrates that simulates natural gas. Mitsui Zosen Tech Rev 2009;198:12–8 (in Japanese). [5] Kondo W, Ogawa H, Ohmura R, Mori YH. Clathrate hydrate formation from a hydrocarbon gas mixture: evolution of gas-phase composition in a hydrateforming reactor. Energy Fuels 2010;24:6375–83. [6] Ando N, Kodama T, Kondo W, Mori YH. Clathrate hydrate formation from a methane + ethane + propane mixture in an unstirred surfactant-containing system. Energy Fuels 2012;26:1798–804.
871
[7] Uchida T, Moriwaki M, Takeya S, Ikeda IY, Ohmura R, Nagao J, et al. Two-step formation of methane–propane mixed gas hydrates in a batch-type reactor. AIChE J 2004;50:518–23. [8] Schicks JM, Naumann R, Erzinger J, Hester KC, Koh KA, Sloan Jr ED. Phase transitions in mixed gas hydrates: experimental observations versus calculated data. J Phys Chem B 2006;110:11468–74. [9] Hester KC, Dunk RM, Walz PM, Pelzer ET, Sloan ED, Brewer PG. Direct measurements of multi-component hydrates on the seafloor: pathways to growth. Fluid Phase Equilb 2007;261:396–406. [10] Seo Y, Kang S-P, Jang W. Structure and composition analysis of natural gas hydrates: 13C NMR spectroscopic and gas uptake measurements of mixed gas hydrates. J Phys Chem A 2009;113:9641–9. [11] Kutergin OB, Mel’nikov VP, Nesterov AN. Surfactant effect on the mechanism and kinetics of gas hydrate formation. Doklad Akad Nauk 1992;323:549–53 (in Russian). [12] Zhong Y, Rogers RE. Surfactant effects on gas hydrate formation. Chem Eng Sci 2000;55:4175–87. [13] Okutani K, Kuwabara Y, Mori YH. Surfactant effects on hydrate formation in an unstirred gas/liquid system: an experimental study using methane and sodium alkyl sulfates. Chem Eng Sci 2008;63:183–94. [14] Gayet P, Dicharry C, Marion G, Graciaa A, Lachaise J, Nesterov A. Experimental determination of methane hydrate dissociation curve up to 55 MPa by using a small amount of surfactant as hydrate promoter. Chem Eng Sci 2005;60:5751–8. [15] Lee S, Zhang J, Mehta R, Woo T-K, Lee JW. Methane hydrate equilibrium and formation kinetics in the presence of an anionic surfactant. J Phys Chem C 2007;111:4734–9. [16] CSMGem. A phase-equilibrium calculation program package accompanying the following book. In: Sloan Jr. ED, Koh CA, Clathrate hydrates of natural gases, 3rd ed., Boca Raton, FL: CRC Press; 2007. [17] HWHydrateGUI (ver. 1.1), a phase-equilibrium calculation program developed at the Centre for Gas Hydrate Research, Heriot-Watt University, Edinburgh, UK; 2005. [18] Mori T, Mori YH. Characterization of gas hydrate formation in direct-contact cool storage process. Int J Refrig 1989;12:259–65. [19] Izumi F, Momma K. Three-dimensional visualization in powder diffraction. Solid State Phenom 2007;130:15–20.
Contents lists available at ScienceDirect
Applied Energy journal homepage: www.elsevier.com/locate/apenergy
Clathrate-hydrate formation from a hydrocarbon gas mixture: Compositional evolution of formed hydrate during an isobaric semi-batch hydrate-forming operation Wataru Kondo a,1, Kaoru Ohtsuka a, Ryo Ohmura a, Satoshi Takeya b, Yasuhiko H. Mori a,⇑ a b
Department of Mechanical Engineering, Keio University, Yokohama 223-8522, Japan Research Institute of Instrumentation Frontier, National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba 305-8565, Japan
h i g h l i g h t s Reports laboratory experiments of hydrate formation from synthetic natural gas. Shows how hydrate composition changes during each hydrate-forming operation. Compares the observed compositional change with thermodynamic simulations. Shows a PXRD result indicating simultaneous formation of sI and sII hydrates.
a r t i c l e
i n f o
Article history: Received 16 August 2012 Received in revised form 26 July 2013 Accepted 13 August 2013 Available online 7 September 2013 Keywords: Clathrate hydrate Gas hydrate Mixed hydrate Natural gas storage
a b s t r a c t The clathrate hydrate formation from a model natural gas, i.e., a mixture of methane, ethane, and propane in a 90:7:3 molar ratio, under a constant pressure was experimentally investigated, focusing on the compositional evolution of hydrate crystals formed inside a gas-bubbling-type reactor during each semibatch hydrate-forming operation. The experimental system used in this study was specially designed for obtaining several hydrate samples formed at different, arbitrarily selected stages during each hydrate-forming operation. Each hydrate sample was analyzed by a gas-chromatograph to determine the mole fractions of methane, ethane and propane encaged in the hydrate. These analyses revealed a monotonic increase in the methane fraction and decreases in the ethane and propane fractions during each operation until a quasi-steady state was established. Powder X-ray diffraction analyses showed that both structure-I and structure-II crystals were simultaneously formed during the quasi-steady period. The compositional evolution of the hydrates formed during the early stages before the quasi-steady state was reached deviated from corresponding predictions based on the thermodynamic-simulation scheme previously reported. A hypothetical explanation for the discrepancy between the experimental and simulation-based results was provided. Ó 2013 Elsevier Ltd. All rights reserved.
1. Introduction This paper describes an experimental study of the hydrate formation from a multi-component gas mixture. Understanding the behavior of such a hydrate-forming process is important for hydrate-based technology development for storing and transporting natural gas, separating undesirable (toxic or incombustible) species from biogases or natural gas with a high carbon-dioxide content, etc. Specifically, this paper focuses on the evolution in the composi⇑ Corresponding author. Tel./fax: +81 45 701 2054. E-mail address: [email protected] (Y.H. Mori). Present address: Plant Management Group, Takehara Power Station, Electric Power Development Co., Ltd., Tadanoumi-Nagahama, Takehara-shi, Hiroshima Prefecture 729-2329, Japan. 1
0306-2619/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.apenergy.2013.08.033
tion of the hydrocarbon guest being captured in hydrates during their continuous formation from a hydrocarbon gas mixture simulating a natural gas. Such evolution originating from the nature of fractionation of hydrocarbons in forming hydrates possibly causes the lack of compositional uniformity of the hydrate products; hence it is of great importance concerning the industrial hydrate-forming operations for natural gas storage and transport [1]. For appropriately designing and controlling such operations, we need to have an insight into how the gas-phase composition in a hydrate-forming reactor, and hence the hydrocarbon uptake into the instantaneously formed hydrate, would change during each operation. Obviously, it is extremely difficult to find a solution fully satisfying the above task, because each hydrate-forming operation includes highly complicated multi-phase dynamic processes such as the convective
W. Kondo et al. / Applied Energy 113 (2014) 864–871
mass (hydrocarbons) transfer to, and heat transfer from, hydrate crystals thereby causing their growth, each dependent on the reactor design and the system-operational parameters (pressure, temperature, liquid stirring, etc.). However, it may be still possible and, if possible, should be valuable from an engineering viewpoint to find some practical means that allows us to estimate the compositional changes in the gas phase and the hydrate product during each hydrate-forming operation. As a tool for expeditiously estimating such compositional changes, Tsuji et al. [2] first presented a thermodynamic simulation scheme, in which the fluids and, if any, a hydrate confined in the reactor are assumed to form an open, instantaneously equilibrated system undergoing a quasi-static irreversible change in response to the successive discharge of the formed hydrate from, and the supply of the feed gas and water to, the reactor. Later, a modified, more sophisticated scheme for such simulations was constructed by Ogawa et al. [3]. These simulation schemes cannot predict the hydrate-formation rate in a given hydrate-forming system, hence they cannot predict how fast the gas phase and the hydrate product change their compositions with time; this is because the schemes are entirely based on the thermodynamics-based characterization of the system, and none of the rate-controlling factors (the size and geometry of the reactor, the gas–liquid interfacial area, the dynamics of the gas–liquid mixing, etc.) are taken into account. Instead, the schemes can predict, irrespective of the size and type of the reactor, how the gas phase and the hydrate product change their compositions with the progress in the fractional replacement of the gas phase with the fresh feed gas (a hydrocarbon mixture with a fixed composition) continuously supplied to the reactor in order to compensate the loss of each hydrocarbon species from the gas phase due to the hydrate formation. Although the simulation schemes mentioned above can possibly be useful tools for designing hydrate-forming systems and their operations, we should be careful about possible deviations of the simulation-based estimates from the actual hydrate-forming behavior. Deviations possibly result from the modeling of hydrate-forming operations as thermodynamically quasi-static processes, which plays a key role in the simulation schemes [2,3]. A few attempts have been made at comparing simulation-based estimates to the corresponding experimental data concerning the change in the gas-phase composition during each hydrate-forming operation [4–6], which indicated a reasonably good agreement between them (except for the later stages of each hydrate-forming process in an unstirred semi-batch reactor containing a surfactant solution [6]). As for the compositional evolution in the hydrate product, however, no experimental examination corresponding to those for the gas-phase evolution [4–6] has been reported so far. For further clarification, this point may need a supplementary explanation, which follows. We find in the literature some previous studies in which the crystalline structures and hydrate-guest compositions in the hydrates formed from hydrocarbon mixtures were experimentally analyzed. See, for example, the papers by Uchida et al. [7], Schicks et al. [8], Hester et al. [9] and Seo et al. [10]. However, none of these studies demonstrated how the composition of the instantaneously-formed hydrate changed during each isobaric hydrate-forming process properly simulating industrial hydrateforming operations for natural-gas storage. In this paper, we describe our attempt at experimentally observing the compositional evolution of the hydrate formed in a gas-bubbling-type reactor to which a model natural gas (a mixture of methane, ethane, and propane in a 90:7:3 molar ratio) is continuously supplied to compensate for the loss of the gas due to hydrate formation, thereby maintaining the pressure inside the reactor constant. The hydrate-composition data obtained in these experiments are compared to the simulationbased predictions.
865
2. Experimental section 2.1. Experiment-design concept The experimental work performed in this study was designed to extend our previous experimental study [5], in which we investigated how the gas-phase composition inside a gas-bubbling-type reactor changed during each hydrate-forming operation by continually sampling the gas from the reactor. The reason we selected a gas-bubbling-type reactor, instead of water-spray and other types [1], was simply for the convenience of obtaining gas samples which were free from water mist and uniform in composition. This time, we intended to obtain hydrate samples for their compositional analyses using the same reactor that we used for the gas-phase analyses [5]. The major technical difficulty in obtaining hydrate samples for compositional analyses was the fact that hydrate crystals formed at various stages of each hydrate-forming operation were inseparably mixed inside the reactor, hence we could not separately remove hydrate samples formed at specified stages from the rector. To overcome this difficulty, we installed a hydrate sampler (i.e., a detachable, small-capacity reactor) in parallel with the main gas-bubbling reactor for only a short period in order to obtain a hydrate sample representing the current stage. By repeatedly attaching and detaching the sampler during each hydrateforming operation, we could collect hydrate samples representing different stages of the operation. The main reactor exclusively played a role of progressing the operation by continuously forming hydrates during every interval between successive hydrate-sample-forming procedures. The gas mixture thus aged in the main reactor due to the hydrate formation and the feed-gas supply that compensates for the gas consumption by the hydrate formation was circulated in a loop connecting the main reactor and the sampler during each hydrate-sample-forming procedure. This was to ensure that the hydrate sample was formed from essentially the same gas mixture as the one currently filling the main reactor. Technical details with which we realized the above experimentdesign concept are described in Sections 2.3 and 2.5. 2.2. Materials used The feed gas was a mixture of 90.07% methane, 6.93% ethane and 3.00% propane on a molar basis (synthesized and analyzed by Takachiho Chemical Industrial Co., Ltd., Tokyo, Japan). The deionized and distilled water was used for the hydrate formation. Sodium dodecyl sulfate (SDS) with a certified purity of 99.5% mass fraction (supplied by Nacalai Tesque, Inc., Kyoto, Japan) was added to the water at the concentration of 500 ppm (500 mg/kg) to prepare an aqueous SDS solution for use in forming hydrate samples for compositional analyses; this was just to promote hydrate formation, thereby shorten the period for forming each sample. 2.3. Apparatus Fig. 1 illustrates the experimental apparatus used in this study. This apparatus was modified from that used in our previous study [5] by installing a device for forming hydrate samples for compositional analyses. Except for this device set in parallel with the main hydrate-forming reactor on the gas circulation loop, the layout of the apparatus as well as the components assembled in it are the same as those already reported [5]. For example, the main hydrate-forming reactor is the same coolant-jacketed gas-bubblingtype 754-cm3 cell that we used in the previous study [5]. This reactor is equipped with a disk-type bubbler immersed in a water pool and a stirrer for providing a good mixing in the gas phase. The reactor is installed on the gas-circulation loop, through which the gas
866
W. Kondo et al. / Applied Energy 113 (2014) 864–871
Fig. 1. Schematic of the experimental apparatus.
mixture forming a continuous gas phase inside the reactor is sucked out and, having been mixed with a fresh feed gas supplied from an external high-pressure cylinder, sent back to the reactor through the bubbler with the aid of a high-pressure fluid-circulation pump. The gas flow rate through the pump, i.e., the rate at which the gas mixture is sucked from the reactor into the loop, is monitored by a mass flow meter. On the downstream side of this mass flow meter, the feed-gas supply line from a high-pressure gas cylinder merges with the loop, such that the feed gas with the fixed composition is allowed to flow, together with the circulated gas mixture, into the reactor, thereby compensating for the loss of the gas mixture due to the hydrate formation inside the reactor. A mass flow meter installed on the feed-gas supply line allows us to continuously measure the instantaneous gas flow rate within an accuracy of ±10.0 cm3/min NTP (converted to a volumetric flow rate under the ‘‘normal temperature and pressure’’ defined as 273.15 K and 101.3 kPa). A bypass line was newly attached to the gas circulation loop such that we could make, by manipulating a three-way valve on the loop, the gas pumped through the loop bypass the lower part of the main reactor containing the water pool and the bubbler and flow into a hydrate sampler, i.e., a cylindrical, small-capacity (10 cm3 inside volume) gastight borosilicate-glass tube equipped with stop valves at its top and bottom ends (a Hyper Glass Cylinder manufactured by Taiatsu Techno Corp., Ltd., Tokyo). This sampler is initially charged with 7 cm3 of an aqueous SDS solution (500 ppm in SDS mass fraction) and maintained at the same pressure–temperature condition as that in the main reactor. Due to the well-recognized SDS effect for promoting hydrate formation from hydrocarbons (see, for example, Refs. [11–13]), we can expect that a high water-to-hydrate conversion ratio will be achieved during a relatively short period, as compared to the time span for the entire evolution of the hydrate-forming system confined in the main reactor, while the gas is being forced to pass the bypass line. The gas flowing out of the sampler without having been consumed by the hydrate formation is then discharged into the upper gas-phase space inside the main reactor to be mixed with the gas remaining there before flowing into the circulation loop. Once the hydrate formation inside the sampler is deemed to be completed, we can
switch the gas flow back to the ordinary path through the main reactor and remove the sampler from the bypass for a gas-chromatographic analysis of the hydrate filling the sampler. The assembly that we used for sampling the gas phase in the main reactor and analyzing the samples by a high-speed gas chromatograph in our previous study [5] was used again with no changes in this study. This time, we applied this assembly for the purpose of analyzing the gas released by dissociating each hydrate sample; this is described later in some detail. 2.4. Procedure In advance of each experimental run, the main reactor and the gas-circulation loop were evacuated while the loop was closed from the bypass line. The reactor was then charged with 300 cm3 of water, which formed a 63-mm deep pool on the surface of the bubbler. The feed gas was supplied to fill the remaining 454-cm3 space inside the reactor plus the 262-cm3 space inside the loop until the inside pressure had increased to 2.9 MPa. Besides the above preparation of the main reactor and the gas-circulation loop, the sampler charged with 7 cm3 of the aqueous SDS solution was mounted on the bypass line and immersed in a thermostated water bath. After the temperature inside the main reactor and that of the bath holding the sampler became constant at about 275 K, the pump on the gas-circulation loop and the stirrer in the main reactor were turned on. Throughout every experimental run, the gas flow rate through the loop and the rotation rate of the stirrer were controlled at 360 cm3/min and 400 rpm, respectively, while the pressure and temperature inside the chamber were held at 2.9 ± 0.05 MPa and 275 ± 0.5 K, respectively. Once the hydrate formation was detected by monitoring the feed-gas supply to the reactor as well as the visual observations through the sight window of the reactor, we proceeded to prepare for the hydrate sampling operations, the details of which are described in the next subsection. 2.5. Hydrate sampling/analysis details During each experimental run, we typically carried out the hydrate-sampling operations four times at arbitrary intervals in order
W. Kondo et al. / Applied Energy 113 (2014) 864–871
to obtain hydrate samples formed at different stages over a transient hydrate-forming process asymptotically approaching a quasi-steady state.2 The procedure of each hydrate-sampling operation is described below. When a specified, though arbitrarily selected, time had passed after the start of hydrate formation inside the main reactor, we switched the gas flow in the circulation loop from the ordinary line passing through the main reactor to the bypass line by manipulating the three-way valve placed at the upstream-side junction of the ordinary and bypass lines (see Fig. 1) and, at the same time, opening the stop valves at the inlet and the outlet of the hydrate sampler. When we noted, based on the visual observation and the gasflow monitoring on the feed-gas supply line, that the hydrate formation inside the sampler had almost ceased, we closed the above stop valves, and then manipulated the three-way valve to turn the gas-flow path to the ordinary line to enable the further progress of the hydrate-forming process. After the resumption of the hydrateforming process inside the main reactor, we removed the sampler from the bypass line, and attached a new sampler, which had been pre-cooled to about 275 K in advance, to the bypass line to prepare for the next hydrate-sampling. The hydrate sample obtained by each sampling operation outlined above was specified by ng, the moles of the feed gas supplied to the hydrate-forming system (the fluids + hydrate contents in the main reactor, the sampler, and the gas-circulation loop including the bypass line) over the time span between the inception of hydrate formation in the main reactor and the formation of the hydrate sample inside the sampler3, or by the ratio of ng to ng0, the moles of the feed gas with which the hydrate-forming system was initially charged. The rest of the procedure of handling the hydrate sample is outlined below. For temporary use, a pressure gauge was attached to the sampler just removed from the bypass line. The stop valve at the top of the sampler was then opened to discharge the gas, that had occupied a small proportion of the inside volume of the sampler, until the inside pressure read by the pressure gauge decreased to 0.1 MPa. This gas discharge operation was performed holding the sampler in the water bath maintained at 275 K. After discharging the gas, the sampler was removed from the bath into the laboratory air to make the hydrate inside the sampler dissociate. After completion of the hydrate dissociation, the sampler containing the gas released by the hydrate dissociation was connected to a branch of the gas sampling line that extended from the main reactor to the gas chromatograph (see Fig. 1). The subsequent procedure for analyzing the gas supplied from the sampler was just the same as that we used for analyzing the samples taken from the gas phase inside the main reactor and described in detail in our previous paper [5]. 2.6. Powder X-ray diffraction measurements In addition to the gas-chromatographic analyses of hydrate samples throughout each hydrate-forming process, we performed powder X-ray diffraction (PXRD) analyses of the hydrate samples formed during the quasi-steady asymptotic regime (ng/ng0 > 0.55) in which the gas-chromatography-based hydrate composition remained almost constant. The hydrate samples for the PXRD mea2 The term ‘‘quasi-steady state’’ defines an asymptotic regime for each hydrateforming operation in which any systematic change in composition was no longer observed with the gas phase remaining inside the reactor or with the formed hydrate. Note that any ‘‘steady state’’ in its rigorous sense cannot be achieved as long as the operation is the semi-batch type, instead of the continuous type. 3 It took 15–30 min to form each hydrate sample. Hence, ng increased, though only slightly, from its initial value, ng,1, to its final value, ng,2, during the formation of one hydrate sample. For simplicity, we employed the arithmetic mean of ng,1 and ng,2 as the ng value representing this hydrate sample.
867
surements were prepared through a two-step freezing procedure which is outlined below. The hydrate sampler removed from the hydrate-forming apparatus (Fig. 1) was first immersed in a bath of an aqueous ethyleneglycol solution controlled at 30 °C to freeze the contents of the sampler except for the gas mixture occupying the top portion inside the sampler. After discharging the gas mixture by opening the valve at its top port, the sampler was immediately placed into a liquid-nitrogen bath for further cooling of its contents. The sampler was then removed from the bath, its bottom screw cap was removed in a cold nitrogen-gas atmosphere, and the hydrate sample (to be more exact, a hydrate + ice mixture) in the amount of several grams was scraped from the sampler. This sample was finely ground in a nitrogen-gas atmosphere at a temperature below 100 K, then top-loaded onto a copper-made specimen holder for PXRD analysis. The loaded sample was exposed to Cu Ka radiation generated by an Ultima III diffraction system (Rigaku Corp., Tokyo, Japan) in a parallel-beam optics alignment. Each measurement was performed in the h/2h scan mode with a step width of 0.02° at 123 K. 2.7. Note on the use of SDS The objective of this note is to clarify the readers’ possible concern that the use of SDS in forming the hydrate samples for compositional analyses might have added an unnecessary complexity to the hydrate-forming system dealt with in this study. One may be concerned about the difference in the hydrate-forming conditions between the SDS-containing sampler and the SDS-free main reactor. Such concerns are quite groundless. Due to its molecular size, SDS cannot be encaged in any types of clathrate hydrates. Therefore, SDS does not have any crystal-structural effect on the hydrate formation. Moreover, it has been experimentally confirmed that the SDS addition to the orders of 100–1000 ppm in water results in no practical change in the thermodynamic hydrate-forming conditions for methane, methane + carbon dioxide, and methane + ethane + propane [14,15]. Based on these facts, we can expect that the SDS addition altered neither the crystalline structures nor the guest compositions of the hydrate samples while it significantly promoted the hydrate nucleation and growth, thereby significantly shortening the time required for forming each hydrate sample. 3. Results and discussion We have performed nine experimental runs under nominally the same operational conditions as those specified in the preceding section. Each run included an isobaric hydrate-forming operation, which extended to 24–96 h. The obtained results are shown below, particularly focusing on the evolution of the hydrate-phase composition through each hydrate-forming operation. (A complete set of experimental data relevant to the description which follows is provided in the electronic Supplementary data.) It should be noted that the rate of hydrate formation is not a subject of interest in this study, similar to our recent studies about hydrate formation from hydrocarbon mixtures [3,5]. Fig. 2 summarizes the data obtained in the nine experimental runs in the form of xi,hyd versus t plots, where xi,hyd denotes the water-free-base mole fraction of each hydrate-guest species (i = 1 for methane, i = 2 for ethane, and i = 3 for propane) contained in each hydrate sample and t denotes the time lapse after the onset of the hydrate formation inside the main reactor. The xi,hyd values for the three individual species were deduced by normalizing the P relevant gas-chromatographic data such that ixi,hyd = 1. Along the t axis, each data point is located at the midpoint of the
868
W. Kondo et al. / Applied Energy 113 (2014) 864–871
operation from the real time t to ng/ng0 that expresses, in a normalized form, the degree of replacement of the gas mixture filling the system with the feed gas continuously supplied to the system from the outside. As previously discussed [5], ng/ng0 is considered to be a universal index, independent of the system scale as well as the hydrate-formation rate, of the progress in compositional evolution of the gas mixture inside the system during each hydrate-forming operation. Fig. 3 shows the xi,hyd versus ng/ng0 plots of the same experimental data that we used in preparing Fig. 2. Also shown for comparison are the relevant predictions by the thermodynamic simulations based on the scheme that we previously developed [3].
Fig. 2. Plot of xi,hyd, the measured water-free-base mole fractions of the three hydrocarbon species (i = 1 for methane, i = 2 for ethane, and i = 3 for propane) contained in each of the hydrate samples formed at different stages during each semi-batch hydrate-forming operation under the constant pressure of 2.9 ± 0.05 MPa. The axis of time t represents the time lapse after the inception of hydrate formation in the main reactor.
15–30 min period which elapsed during the formation of the relevant hydrate sample inside the sampler.3 Inspecting the xi,hyd versus t plots in Fig. 2, we can estimate that the composition of the newly formed hydrates became almost constant within 20 h following the inception of the hydrate formation. However, this observation is dependent on the apparatus and operating conditions employed in the present experiments, and hence lacks generality. Besides, the experimental results arranged in the form of xi,hyd versus t plots cannot be compared to the predictions given by the thermodynamic simulation scheme that we previously developed [3], because the scheme is inherently free from any quantitative time scale. In order to provide the experimental data with a generality such that one can compare them to appropriate simulations as well as, though not yet available at present, corresponding data generated by other hydrate-forming equipment which may be different in size and/or type from ours, we need to convert the index for the aging of the hydrate-forming system during each hydrate-forming
Fig. 3. Plot of the same xi,hyd data as those shown in Fig. 2 with the replacement of the time scale t by the gas-phase replacement index, ng/ng0, i.e., the ratio of ng, the moles of the feed gas having been supplied to the hydrate-forming system (the fluids + hydrate contents in the main reactor, the sampler, and the gas-circulation loop including the bypass line) after the inception of hydrate formation in the main reactor, to ng0, the moles of the feed gas with which the hydrate-forming system was initially charged. The horizontal bar for each data point indicates the ng/ng0 increment during the relevant hydrate-sample formation. The solid and dotted curves show the evolution of the xi,hyd values for the instantaneously formed hydrates predicted by two thermodynamic simulations for an isobaric (at 2.90 MPa), quasi-static hydrate-forming operation. These simulations were provided by the same simulation scheme [3] into which two different PECPs, CSMGem [16] and HWHydrate [17], were alternatively incorporated.
W. Kondo et al. / Applied Energy 113 (2014) 864–871
Two different predictions (plotted by solid and dotted curves, respectively) are discriminated, with respect to the phase-equilibrium calculation program (PECP) incorporated in the simulation scheme. Following the way of our previous studies of this series [3,5], we alternatively incorporated two different PECPs, CSMGem [16] and HWHydrate [17], into the simulation scheme such that we can take an objective view of the PECP-dependent results of the simulations. To help our understanding of the results of these simulations, we show in Fig. 4 how these simulations predict the occupancies, by the three guest species, of the 51264 cages in structure II (sII) or 51262 cages in structure I (sI) of the instantaneously formed hydrates throughout the entire hydrate-forming process. As demonstrated in Fig. 3, the experimental xi,hyd data showed a simple asymptotic change with an increase in ng/ng0. That is, x1,hyd increased from its initial level, 0.8, to its asymptotic level, 0.9, before ng/ng0 increased to 0.5. In parallel with this change in x1,hyd, x2,hyd and x3,hyd monotonically decreased from 0.13 to 0.07 and from 0.07 to 0.03, respectively. It should be noted that, over the entire ng/ng0 range above 0.5, x1,hyd, x2,hyd and x3,hyd were almost constant at about 0.9, 0.07 and 0.03, respectively, in good agreement with the composition of the feed gas. This fact means that, during such an asymptotic region, the hydrate-forming system was in a quasi-steady state thereby yielding an instantaneous mass balance between the feed-gas supply to the system and the hydrate formation with respect to each hydrateguest species. In Fig. 3, we find significant differences in the xi,hyd evolution behavior between the experimental results and the simulations. Whichever PECP (CSMGem or HWHydrate) we used, the simulations predicted a sharp initial decrease in x3,hyd from 0.3 to 0.03 and a simultaneous increase in x2,hyd as a result of the competitive occupation of the 51264 cages of the sII hydrates by propane and ethane molecules (see Fig. 4). That is, the preferential occupation of the 51264 cages by propane molecules during the earliest stages results in a rapid depletion of propane in the gas phase (see Fig. 5 in Ref. [5]), thereby allowing ethane molecules to be subsequently included in the 51264 cages. As a result of the propane depletion and the subsequent ethane depletion in the gas phase, x2,hyd has a peak at ng/ng0 0.2. However, neither the initial sharp decrease in x3,hyd nor the peak of x2,hyd at ng/ng0 0.2 were
Fig. 4. Evolution of the occupancies, by the three guest species, of the large cages (51264 cages in sII or 51262 cages in sI) of the instantaneously-formed hydrates during each hydrate-forming process. The predictions by the two simulations, one using CSMGem and the other using HWHydrate, are plotted by the solid and dotted lines, respectively. Note that the simulation using HWHydrate predicts an alternate formation of the sII and sI hydrates over the quasi-steady region of the process (ng/ng0 > 0.55).
869
Fig. 5. PXRD profile (measured at 123 K) of a hydrate sample formed in the ng/ng0 increment from 1.13 to 1.15. The plus signs (+) denote the observed intensities; the solid line on the plus signs represents the intensities calculated from the best-fit model of the Rietveld refinement. The bottom curve shows the deviation between the observed and calculated intensities. The tick marks on the three rows represent the calculated peak positions for the sI hydrate, the sII hydrate, and the hexagonal ice (Ih), respectively.
observed in our experimental results. We later discuss this discrepancy between the simulations and the experimental observations. Another issue to be raised here is the difference between the CSMGem-based simulation and the HWHydrate-based simulation about their xi,hyd predictions over the quasi-steady region (ng/ng0 > 0.55). The CSMGem-based simulation predicts the exclusive formation of sII hydrates throughout the entire hydrate-forming process, hence suggesting smooth xi,hyd versus ng/ng0 curves as demonstrated in Fig. 3. In contrast, the HWHydrate-based simulation predicts an alternate formation of sII hydrates in two different compositions and also an sI hydrate (see Fig. 4), resulting in periodical jumps in xi,hyd. Integrating such discrete xi,hyd values over more than one cycle of successive formation of the three classes of hydrates (i.e., two classes of sII and one class in sI) along the ng/ng0 axis, we obtain the averaged values of x1,hyd, x2,hyd and x3,hyd very close to 0.90, 0.07 and 0.03, respectively, which are consistent with the mass-balance requirement for the hydrate-forming system under a steady condition. Whether an sI hydrate actually forms along with the sII-hydrate formation during the quasi-steady region, in which the methane fraction in the gas phase is the highest throughout the entire hydrate-forming process, is an intriguing problem. According to the Gibbs phase rule, sII and sI hydrates can simultaneously form in the present experimental system containing four chemical species, i.e., three hydrocarbons and water. (Readers may consult Section 4.2 in Ref. [2] concerning the phase-rule-based consideration on this point.) In fact, the coexistence of sII and sI hydrates in a four component (methane + ethane + propane + water) or five component (methane + ethane + propane + isobutene + water) experimental system has already been reported in the literature [8,10]. Thus, the formation of an sI hydrate together with an sII hydrate in our experimental system would, if actually observed, no longer be a new scientific finding by itself. However, no reliable means is currently available that allows us to predict if hydrates in both structures simultaneously form or a hydrate in either structure exclusively forms under a given experimental condition. Inevitably, we have to rely on some experimental means to know the actual crystalline structures of formed hydrates, which is an indispensable piece of knowledge for designing industrial natural-gas-hydrate production processes. Thus, we later discuss on this issue based on the PXRD measurements of the hydrate samples obtained in the quasi-steady region for each hydrate-forming process.
870
W. Kondo et al. / Applied Energy 113 (2014) 864–871
The significant discrepancies between the experiments and simulations concerning the xi,hyd versus ng/ng0 relations during the early transient region (ng/ng0 < 0.55) are possibly dependent on the gas-bubbling-type hydrate-forming scheme that we used in our experiments. We describe below our hypothetical interpretation of the above discrepancies, taking the observed hydrate-formation behavior into account. In either of the main reactor and the hydrate sampler (see Fig. 1), the gas mixture was released into a liquid pool in the form of discrete bubbles typically several millimeters in diameter, resulting in the hydrate formation on the surface of each bubble buoying up in the liquid pool besides the hydrate formation on the free surface of the pool. As already described in detail by Mori and Mori [18], such hydrate formation turned the gas bubbles into the form of hydrate balloons, each composed of a thin hydrate shell and a gaseous core, which often agglomerated into a hydrate foam layer lying on the liquid pool thereby intervening between the pool and the gas phase at the top of the reactor or the sampler. It should be noted that each bubble during its rise in the bulk of the pool or resting on the top of the pool was practically isolated from the gas mixture forming the continuous phase in the reactor, the sampler, or the gas-circulation loop. As the hydrate formation progressed, preferentially consuming propane and ethane, on the surface of each bubble, the gas phase inside the bubble should have rapidly depleted propane and ethane, thereby increasing the fraction of methane. Thus, the gas mixture actually involved in the hydrate formation inside the bubble must have been increasingly methane-rich as compared to the gas mixture which would have been available if complete gas-phase mixing inside the entire hydrate-forming system (as assumed in the simulations) was occurring. Concerning their hydrate-forming experiments by bubbling a seven-component hydrocarbon mixture into seawater confined in a cylindrical cell, Hester et al. [9] pointed out a possible deviation in composition between the bubbles and the continuous gas phase located at the top of the cell as a result of preferential hydrate formation on the bubble surface. Essentially the same phenomenon must have occurred in our gas-bubblingtype reactors. Consequently, the hydrates formed from the gas mixture confined in such discrete bubbles must have been more methane-rich than those which would be formed from the gas mixture ideally mixed at each instant. The in-bubble hydrate-formation process assumed above possibly accounts, at least in part, for the discrepancies in the evolution of xi,hyd during the early transient regime (ng/ng0 < 0.55) between the experimental observations and the simulations.4 It seems intriguing to investigate if such an evolution of xi,hyd varies when using a different hydrateforming scheme in which the feed gas makes contact with the aqueous phase taking the form of a dispersed phase instead of a continuous phase. 4 The hydrate formation primarily consuming the gas mixture confined in the bubbles, instead of that in the continuous gas phase occupying the top portion of the main reactor, may also account for some discrepancies in the evolution of xi,gas, the mole fractions of methane (i = 1), ethane (i = 2) and propane (i = 3) in the continuous gas phase, between the experiments and the simulations, which are demonstrated in Figs. 5 and 7 in our previous paper [5]. In these figures, we find that the increase in x1,gas and the simultaneous decreases in x2,gas and x3,gas with an increase in ng/ng0 up to 1.0 experimentally observed are less steep than those predicted by the simulations. This fact may be ascribable to the absence of gas mixing between the bubbles and the continuous gas phase from which the gas mixture was periodically sampled for the chromatographic analyses. Presumably, the continuous gas phase, which was initially at a composition equal to that of the feed gas (a mixture of methane, ethane, and propane in a 90:7:3 molar ratio), still remained richer with ethane and propane than the gas phase inside each bubble during the early transient region because the area over which the former gas phase was in direct contact with water, thereby allowing hydrate formation, was small (relatively to its volume) and hence its composition was less apt to change than that of the gas phase inside each bubble. Consequently, the compositional change in the continuous gas phase was possibly delayed compared to the change that the gas phase in an ideally mixed hydrate-forming system (as assumed in the simulations) would exhibit.
Hydrate samples formed in three different increments regarding ng/ng0, 1.13–1.15, 1.23–1.28 and 1.45–1.54, in the quasi-steady regime (see Fig. 3) were subjected to the PXRD measurements to confirm their crystalline structures. Similar PXRD profiles were obtained with these samples, and one of them is shown in Fig. 5. This PXRD profile, likewise those for the other two samples, indicates that this hydrate sample was a mixture of two types of hydrates, sI with the lattice constant of 11.852(3) Å and sII with the lattice constant of 17.113(1) Å, and the hexagonal ice Ih which must have been formed from the residual liquid while the sample scraped out from the sampler was chilled for the subsequent PXRD measurement. The sI-to-sII ratios for the three samples were estimated from the relevant PXRD profiles by the Rietveld method using the Rietan-FP program [19]; they were 0.25:0.75, 0.33:0.67, and 0.68:0.32, respectively. The significant scatter in these ratios was presumably ascribable to the spatial non-uniformity of the sI-tosII mixing in each sample, only a small portion of which was used in the PXRD measurement, rather than to the sample-to-sample difference in the sI-to-sII ratio representing each sample. Despite such a scatter in the estimated sI-to-sII ratios, we can reasonably claim that, in the hydrates formed in the quasi-steady asymptotic regime in each hydrate-forming process, the proportion of sI crystals was not minute but substantial, almost comparable to that of the sII crystals. This finding is qualitatively consistent with the prediction by the HWHydrate-based simulation. 4. Concluding remarks This study has experimentally investigated the hydrate formation from a simulated natural gas (a mixture of methane, ethane and propane in a 90:7:3 molar ratio) in an isobaric, gas-bubblingtype reactor, focusing on the compositional evolution of the hydrate instantaneously formed in the reactor throughout each semi-batch hydrate-forming operation under constant pressure (2.9 MPa). Using an auxiliary reactor, i.e., a detachable, smallcapacity gas-bubbling-type reactor, along with the main reactor, for obtaining a hydrate sample formed within a short period at an arbitrary stage during the above operation, we have succeeded in collecting compositional hydrate-product data over the entire process of the semi-batch hydrate-forming operation. These data revealed how the guest-molecule composition of the hydrate product changes from its initial state to an asymptotic (quasi-steady) state, with the gas-replacement progress in the hydrate-forming system resulting from the simultaneous gas uptake into the formed hydrate and the replenishment of the system with fresh feed gas (the ternary mixture simulating natural gas). Based on PXRD analyses, we have also revealed that both the sI and sII crystals were mixed in the hydrates formed during the asymptotic regime of the hydrate-forming process. The thermodynamic simulation scheme that we previously reported [3] was found to be effective for predicting the stage beyond which the hydrate-forming system falls in the asymptotic regime, i.e., the hydrate product maintains the constant guest-molecule composition equal to that of the feed gas. However, the scheme fails to predict the evolution of the guest-molecule composition during the earlier transient regime presumably due to the compositional nonuniformity in the gas mixture dispersed in the actual hydrate-forming system. It is desirable that this point should be further examined using different types of hydrate-forming reactors. Acknowledgments This study was supported in part by a Grant-in-Aid for the Global COE Program for the ‘‘Center for Education and Research of Symbiotic, Safe and Secure System Design’’ from the Ministry of
W. Kondo et al. / Applied Energy 113 (2014) 864–871
Education, Culture, Sports, Science and Technology, Japan. We thank Eiichiro Tachikawa, a former student at Keio University, for his help in the experimental work during the early stage of this study. Sanehiro Muromachi, a current graduate student in Keio University, is also acknowledged for his valuable technical advice about the preparation of hydrate samples for the PXRD analyses. Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.apenergy.2013.08.033. References [1] Mori YH. Recent advances in hydrate-based technologies for natural gas storage – a review. J Chem Ind Eng (China) 2003;54(Suppl.):1–17. [2] Tsuji H, Kobayashi T, Okano Y, Ohmura R, Yasuoka K, Mori YH. Thermodynamic simulations of isobaric hydrate-forming operations: formulation of computational scheme and its applications to hydrate formation from a methane + ethane + propane mixture. Energy Fuels 2005;19:1587–97. [3] Ogawa H, Imura N, Miyoshi T, Ohmura R, Mori YH. Thermodynamic simulations of isobaric hydrate-forming operations for natural gas storage. Energy Fuels 2009;23:849–56; Ogawa H, Imura N, Miyoshi T, Ohmura R, Mori YH. Thermodynamic simulations of isobaric hydrate-forming operations for natural gas storage. Energy Fuels 2011;25:1333. [4] Takahashi M, Kanda N, Sano K, Iwasaki T. Formation characteristics of multicomponent mixed gas hydrates that simulates natural gas. Mitsui Zosen Tech Rev 2009;198:12–8 (in Japanese). [5] Kondo W, Ogawa H, Ohmura R, Mori YH. Clathrate hydrate formation from a hydrocarbon gas mixture: evolution of gas-phase composition in a hydrateforming reactor. Energy Fuels 2010;24:6375–83. [6] Ando N, Kodama T, Kondo W, Mori YH. Clathrate hydrate formation from a methane + ethane + propane mixture in an unstirred surfactant-containing system. Energy Fuels 2012;26:1798–804.
871
[7] Uchida T, Moriwaki M, Takeya S, Ikeda IY, Ohmura R, Nagao J, et al. Two-step formation of methane–propane mixed gas hydrates in a batch-type reactor. AIChE J 2004;50:518–23. [8] Schicks JM, Naumann R, Erzinger J, Hester KC, Koh KA, Sloan Jr ED. Phase transitions in mixed gas hydrates: experimental observations versus calculated data. J Phys Chem B 2006;110:11468–74. [9] Hester KC, Dunk RM, Walz PM, Pelzer ET, Sloan ED, Brewer PG. Direct measurements of multi-component hydrates on the seafloor: pathways to growth. Fluid Phase Equilb 2007;261:396–406. [10] Seo Y, Kang S-P, Jang W. Structure and composition analysis of natural gas hydrates: 13C NMR spectroscopic and gas uptake measurements of mixed gas hydrates. J Phys Chem A 2009;113:9641–9. [11] Kutergin OB, Mel’nikov VP, Nesterov AN. Surfactant effect on the mechanism and kinetics of gas hydrate formation. Doklad Akad Nauk 1992;323:549–53 (in Russian). [12] Zhong Y, Rogers RE. Surfactant effects on gas hydrate formation. Chem Eng Sci 2000;55:4175–87. [13] Okutani K, Kuwabara Y, Mori YH. Surfactant effects on hydrate formation in an unstirred gas/liquid system: an experimental study using methane and sodium alkyl sulfates. Chem Eng Sci 2008;63:183–94. [14] Gayet P, Dicharry C, Marion G, Graciaa A, Lachaise J, Nesterov A. Experimental determination of methane hydrate dissociation curve up to 55 MPa by using a small amount of surfactant as hydrate promoter. Chem Eng Sci 2005;60:5751–8. [15] Lee S, Zhang J, Mehta R, Woo T-K, Lee JW. Methane hydrate equilibrium and formation kinetics in the presence of an anionic surfactant. J Phys Chem C 2007;111:4734–9. [16] CSMGem. A phase-equilibrium calculation program package accompanying the following book. In: Sloan Jr. ED, Koh CA, Clathrate hydrates of natural gases, 3rd ed., Boca Raton, FL: CRC Press; 2007. [17] HWHydrateGUI (ver. 1.1), a phase-equilibrium calculation program developed at the Centre for Gas Hydrate Research, Heriot-Watt University, Edinburgh, UK; 2005. [18] Mori T, Mori YH. Characterization of gas hydrate formation in direct-contact cool storage process. Int J Refrig 1989;12:259–65. [19] Izumi F, Momma K. Three-dimensional visualization in powder diffraction. Solid State Phenom 2007;130:15–20.