Gas Hydrate Tutorial PDF
Gas Hydrate Tutorial PDF
Gas Hydrate Tutorial PDF
\
'
Gas clathrates (commonly called hydrates) are crystalline compounds which occur when water
forms a cage-like structure around smaller guest molecules. While they are more commonly
called hydrates, a carefid distinction should be made between these non-stoichiometric clathrate
hydrates of gas and other stoichiometric hydrate compounds which occur for example, when
water combines with various salts.
G a s hydrates of current interest are composed of water and the following eight molecules:
methane, ethane, propane, isobutane, normal butane, nitrogen, carbon dioxide, and hydrogen
sulfide. Yet other apolar components between the sizes of argon (0.35 nm) and ethylcyclohexane
( 0 . 9 ~ 1 )can form hydrates. Hydrate formation is a possibility where water exists in the vicinity of
such molecules at temperatures above and below 273 K.
Hydrate discovery is credited in 1810 to Sir Humphrey D a y . Due to their crystalline, nonflowing nature, hydrates became of interest to the hydrocarbon industry in 1934 (Hamrnerschmidt,
1934), the time of their first observance blocking pipelines. Hydrates act to concentrate
hydrocarbons: 1 cubic meter of hydrates may contain as much as 180 SCM of gas. Makogon,
(1965) indicated that large natural reserves of hydrocarbons exist in hydrated form, both in deep
oceans and in the permafrost. Evaluation of these reserves is highly uncertain, yet even the most
conservative estimates concur that there is twice as much energy in hydrated form as in all other
hydrocarbon sources combined. While one commercial example exists of gas recovery from
hydrates, the problems of in situ hydrate dissemination in deepwater/permaftost environments will
prevent their cost-effective recovery until the next millennium.
449
The newest hydrate structure H, named for the hexagonal framework, was discovered and shown
by Ripmeester et al. (1987, 1991) to have cavities large enough to contain molecules the size of
common components of naphtha and gasoline. Initial physical properties, phase equilibrium data,
and models have been advanced (Mehta and Sloan, 1993, 1994a,b, 1996a,b,c; Udachin et al.
1996), and Sassen et al. (1994) have found one instance of in situ SH in the Gulf of Mexico.
Since information on structure H is in the fledgling stages, most of this tutorial concerns SI and
SII.
HYDRATE CRYSTAL STRUCTURES.
Table 1 provides a hydrate structure summary for the three hydrate unit crystals (SI, sII, and sH)
shown in Figure 1 . The crystals structures are given with reference to the water skeleton,
composed of a basic "building block" cavity which has twelve faces with five sides per face, given
the abbreviation 5". By linking the vertices of 512 cavities one obtains SI; linking the faces of 512
cavities results in sII; in SHa layer of linked 512 cavities provide connections.
Interstices between the 512 cavities are larger cavities which contain twelve pentagonal faces and
either two, four, or eight hexagonal faces: (denoted as 51262in SI, 51264in sII, or 5126' in sH).In
addition SH has a cavity with square, pentagonal, and hexagonal faces (435663).Figure 1 depicts
the five cavities of SI, sII, and sH. In Figure 1 a oxygen atom is located at the vertex of each
angle in the cavities; the lines represent hydrogen bonds with which one chemically-bonded
hydrogen connects to an oxygen on a neighbor water molecule.
Inside each cavity resides a maximum of one of the small, guest molecules, typified by the eight
guests associated with 46 water molecules in SI (2[512]*6[51262]*46H20),
indicating two guests in
the 5'' and 6 guests in the 5126' cavities of SI. Similar formulas for sll and SH are
( 16[512J*8[51264]*
136H20) and (3[5 12]*2[435663]*1
[51268]*34H20)respectively.
Structure I a body-centered cubic structure forms with natural gases containing molecules smaller
than propane; consequently SI hydrates are found in situ in deep oceans with biogenic gases
containing mostly methane, carbon dioxide, and hydrogen sulfide. Structure I1 a diamond lattice
within a cubic framework, forms when natural gases or oils contain molecules larger than ethane
but smaller than pentane; sII represents hydrates from thermogenic gases. Finally structure H
hydrates must have a small occupant (like methane, nitrogen, or carbon dioxide) for the 512 and
435663cages but the molecules in the 5126' cage can be as large as 0.9nm (e.g. ethylcyclohexane).
TIME-INDEPENDENT PROPERTIES FROM HYDRATE CRYSTAL STRUCTURES.
Mechanical Prooerties of Hvdrates. As may be calculated via Table 1, if all the cages of each
structure are filled, all three hydrates have the amazing property of being approximately 85%
(mol) water and 15% gas. The fact that the water content is so high suggests that the mechanical
properties of the three hydrate structures should be similar to those of ice. This conclusion is true
to a first approximation as shown in Table 2, with the exception of thermal conductivity and
thermal expansivity @avidson, 1983, Tse, 1994). Many mechanical properties of SH have not
been measured to date.
Guest: Cavitv Size Ratio: a Basis for Prooertv Understanding. The hydrate cavity occupied is a
hnction of the size ratio of the guest molecule within the cavity. To a first approximation, the
concept of "a ball fitting within a ball" is a key to understanding many hydrate properties. Figure
2 (corrected from von Stackelberg, 1949) may be used to illustrate five points regarding the
guest:cavity size ratio for hydrates formed of a &guest
component in SI or sII.
1. The sizes of stabilizing guest molecules range between 0.35 and 0.75 nm. Below 0.35nm
molecules will not stabilize SI and above 0.75 molecules will not stabilize sII.
2. Some molecules are too large to fit the smaller cavities of each structure ( e g C 3 b fits in the
5126' of SI; or i-CJIto fits the 5126' of sII).
3. Other molecules such as CH, and N2 are small enough to enter both cavities (labeled as either
512+51262
in SI or 512+5'*6' in sII) when hydrate is formed of those single components.
4. The largest molecules of a gas mixture usually determines the structure formed. For example,
because propane and i-butane are present in many thermogenic natural gases, they will cause
sII to form. In such cases, methane will distribute in both cavities of sII and ethane will enter
only the 5126' cavity of sn.
5 . Molecules which are very close to the hatched lies separating the cavity sizes appear to
exhibit the most non-stoichiometry, due to their inability to fit securely within the cavity.
450
Table 3 shows the size ratio of several common gas molecules within each of the four cavities of
SI and sII. Note that a size ratio (guest molecule: cavity) of approximately 0.9 is necessary for
stability of a simple hydrate, given by the superscript
When the size ratio exceeds unity, the
molecule will not fit within the cavity and the structure will not form. When the ratio is
significantly less than 0.9 the molecule cannot lend significant stability to the cavity.
Consider ethane, which forms in the 5i2 cavity in SI, because ethane is too large for the small 5
cavities in either structure and too small to give much stability to the large 51264 cavity in sII.
Similarly propane is too large to fit any cavity except the 51264cavity in sII, so that gases of pure
propane form sII hydrates from free water, On the other hand, methanes size is sufficient to lend
stability to the 5 cavity in either SI or sII, with a preference for SI, because CH, lends slightly
higher stability to the 5% cavity in SI than the 51264cavity in sII.
Phase Eauilibrium ProDerties. In Figure 3 pressure is plotted against temperature with gas
composition as a parameter, for methane+propane mixtures. Consider a gas of any given
composition (marked 0 through 100% propane) on a line in Figure 3. At conditions to the right
of the line, a gas of that composition will exist in equilibrium with liquid water. As the
temperature is reduced (or as the pressure is increased) hydrates form from gas and liquid water
at the line, so three phases (liquid water + hydrates + gas) will be in equilibrium. With firther
reduction of temperature (or increase in pressure) the fluid phase which is not in excess (gas in
ocean environments) will be exhausted, so that to the left of the line the hydrate will exist with the
excess phase (water).
All of the conditions given in Figure 3 are for temperatures above 273K and pressures along the
l i e s vary exponentially with temperature. Put explicitly, hydrate stability at the three-phase (LwH-V) condition is always much more sensitive to temperature than to pressure. Figure 3 also
illustrates the dramatic effect of gas composition on hydrate stability; as any amount of propane is
added to methane the structure changes (SI 3 sII) to a hydrate with much wider stability
conditions. Note that a 50% decrease in pressure is needed to form sII hydrates, when as little as
1% propane is in the gas phase.
Any discussion of hydrate dissociation would be incomplete without indicating that hydrates
provide the most industrially usefil instance of statistical thermodynamics prediction of phase
equilibria. The van der Waals and Platteeuw (1959) model was formulated after the
determination of the crystal structures shown in Figure 1. With the model, one may predict the
three-phase pressure or temperature of hydrate formation, by knowing the gas composition. For
further discussion of these details the reader is referred to Sloan (1990, Chapter 5 ) .
Heat of Dissociation. The heat of dissociation (AI%)is defined as the enthalpy change to
dissociate the hydrate phase to a vapor and aqueous liquid, with values given at temperatures just
above the ice point. For SI and sII, Sloan and Fleyfel (1992) show that to a fair engineering
approximation (*lo%) AI% is a function mostly of crystal hydrogen bonds, but also of the cavity
occupied within a wide range of components sizes. Enthalpies of dissociation may be determined
via the univariant slopes of phase equilibrium lines (In P vs. I n ) in the previous paragraphs, using
the Clausius-Clapeyron relation [AI% = -zR d(ln P)/d(lK)],
As one illustration, simple hydrates of C3H8 or i-CsIlo have similar AH., of 129 and 133 kJ/mol
A review of the literature suggests that there are two imminent challenges for future research.
First we must characterize the hydrate phase both in the laboratory and in the field. Secondly we
must characterize the kinetics hydrates formation and dissociation
451
Measurements of the Hydrate Phase. For the hydrate phase there are two measurement
techniques - diffraction and ~pectroscopic.Neutron diffraction (typified by Tse, 1994) can detect
water atoms and guest molecules, while X-ray diffraction detects oxygen positions. Recently
using X-ray diffraction Stem et al. (1996) have been remarkably successfd at converting 97% of
ice to water, by raising ice grain temperatures above the solidus while under high pressure.
Two types of spectroscopy are useful with hydrates: (1) nuclear magnetic resonance (NMR) with
cross polarization (CP) and magic-angle spinning (MAS), and (2) Raman spectroscopy. Virtually
all N M R hydrate work to date comes from the Canadian National Research Council. The first
comprehensive review of NMR studies of clathrates was written by Davidson and Ripmeester
(1984). and a thorough update has been written by Ripmeester and Ratcliffe (1991). Of Nh4R
hydrate compounds Xe has obtained prominence due to its large (ca. 100 ppm) chemical shift.
Recently Sum et al. (1996) have shown that Raman spectroscopy can be used to determine the
fraction of filled cages in hydrates, and the fraction of various components in the cages. Since
Raman appears to be both more versatile and less resource intensive, it holds substantial
Measurements of Hvdrate Kinetics. Most hydrate kinetics research has come from the laboratory
ofBishnoi, notably the work ofEnglezos et al., (1987a,b). They modeled kinetics of methane and
ethane hydrate afier nucleation, for periods up to 100 minutes. A subsequent model was made by
Skovborg and Rasmussen (1994); these workers assumed that the mass transfer of gas to the
liquid phase was the rate-controlling step, so that hydrate kinetics and diffusion to the hydrate
phase could be ignored.
However, even with these excellent beginnings, quantification of hydrate kinetics still pose
substantial challenges to hydrate researchers. For example, no model of hydrate nucleation can
acceptably fit all the data of a single experimenter; no universal nucleation model is available. As
another example, all models of hydrate growth kinetics are apparatus-dependent. While the
growth model may be a satisfactory fit for the apparatus in which the data were generated, the
model will not fit data generated in other apparatuses. Determining time-dependent hydrate
behavior s one of the most significant challenges for research.
CONCLUSIONS
This tutorial reviews hydrate crystal structures, and shows how properties such as phase equilibria
and heat of dissociation relate to molecular structure. While many time-independent properties of
SI and sII hydrates are determined, those for newer structures (e.g. sH) are just beginning to be
explored. The time-dependent characteristics of all three hydrate structures are largely
unspecified and kinetic models to date appear to be apparatus-dependent. Characterization of the
hydrate phase constitutes a current challenge.
LITERATURE CITED
Davidson, D.W., in N
I
P
, J.L. Cox, ed.,
pg 1. Butterworths, Boston (1983)
Davidson, D.W., Ripmeester, J.A.,in Inclusion Comoounds. Vol3 Chapter 3, J.L.Atwood, J.E.D.
Davies, and D.D. MacNichol,eds, Academic Press (1984)
Englezos, P., Kalogerakis, N., Dholabhai, P., Bishnoi,P., Chem. Ena. Sci., 42(11), 2647 (1987a)
Englezos, P., Kalogerakis, N., Dholabhai, P., Bishnoi,P., Chem. Enp. Sci.,42(11), 2659 (1987b)
Hammerschmidt, E.G., Ind. Ena. Chem. 26 851 (1934)
Handa, Y.P., J. Chem. Thermo., 18,891 (1986)
Makogon, Y.F., Gazov. Promst. 14 (1965)
Mehta, A.P., A Thermodvnamic Investiaation of Structure H Clathrate Hvdrates. Ph.D. Thesis,
Colorado School of Mines, Golden, CO (1996)
38,580 (1993)
Mehta, A.P., Sloan, E.D., a-,
q 39, 887 (1994a)
Mehta, A.P., Sloan, E.D., n
Mehta, A.P., Sloan, E.D., AIChE Joumal,40, 312 (1994b)
Mehta, A.P., Sloan, E.D., AIChE Journal. 42,2036(1996a)
Mehta, A.P., Sloan,E.D., in Proc. 2nd Intnl. Cod. on Natural Gas Hvdrates, pl. Monfort, J.P.,
ed., Toulouse, 2-6 June (1996b)
Meht4 A.P., sloan, ED., Structure H Hydrates: Implications for the Petroleum I n d u s t r y , b
1996 Annual Tech Conf,SPE 36742,607 Denver, CO., 6-9 October (1996~)
Monfort, J.P., ed, Proc. 2nd Intnl. Conf on Natural Gas Hvdrates, F. Foucaud, Secretariat, 18
Chemin de la Loge, 3 1078 Toulouse Cedex, France 2-6 June (1996)
452
Ripmeester, J.A., Tse, J.A., Ratcliffe, C.I., Powell, B.M., &&!& 325, 135 (1987)
Ripmeester, J.A., Ratcliffe, C.I., Solid State NMR Studies of Inclusion Comuounds, National
Research Council ofCanada, Report C1181-89S, August 17 (1989)
Ripmeester, J.A., "The Role of Heavier Hydrocarbons in Hydrate Formation." presented at 1991
AIChE Spring Meeting, Houston, April 10 (1991)
Ripmeester, J.A., Ratcliffe, C.I., Hug, D.D., Tse, J.A., , in First) International Conference on
Natural Gas Hvdrates, Annals of New York Academv of Sciences, Sloan, E.D., Happel, J.,
Hnatow, M.A., eds, 715, p161, (1994)
Sassen, R., MacDonald, I.R., Ore. Geochem. 22(6) 1029 (1994)
Skovborg, P., Rasmussen, P., Chem. Ene. Sci., 49, 1131 (1994)
Sloan, E.D., Clathrate Hvdrates of Natural Gases, Marcel Dekker, Inc., New York (1990)
Sloan, E.D., Fleyfel, F.,Fluid Phase E a 4 76, 123 (1992)
Sloan,E.D., Happel, J., Hnatow, M.A., eds, (First) International Conference on Natural Gas
Hvdrates. Annals of New
, . -oY
715, (1994)
Stem, L.A., Kirby, S.H., Durham, W.B., Science, 273, 1843 (1996)
Sum,A,, Bumss, R., Sloan, E.D., in Proc. 2nd Intnl. Cod. on Natural Gas Hvdrates, p51.
Monfort, J.P., ed., Toulouse, 2-6 June (1996b)
Tse, J.S., in First) International Conference on Natural G a s Hvdrates. Annals ofNew York
AcademvofSciences, Sloan, E.D.,Happel, J., Hnatow, M.A., eds., 715, p187, (1994)
van der Waals, J.H., and Platteeuw, J.C., "Clathrate Solutions," Adv. Chem. Phvs.,vol2,1,(1959)
von Stackelberg, M., Natunviss 36 327,359 (1949)
Table 1Geometry o f Cages in Three Hydrate Crystal Structures
Table 3 Ratios of Molecular Diameters to Cavity Diameters' for Some Molecules Including
Natural Gas Hydrate Formers
Molecule
Cavity ~ y p e = >
Guest Dmtr (A)
454
5-82
F i g u r e 1. Three Hydrate U n i t
C r y s t a l s and Constituent C a v i t i e s
594 Ha0
?.$?
cq
%He
c-c3Y
(CHZh 0
i
I
y///////////A
7% n20
F i g u r e 2 . R e l a t i v e S i z e s of
Hydrate Guest and Host C a v i t i e s
fy)
"i
4
I(d
:% '
ai
s .I-
;lo':
a , -
'-
.:
I%
.-
'
oooooDota of Dsoton
_.
+ Fmst
(lB46)
,27
3r3
4p
(OF)
55
47
67
n
/
456
Li
\
/
'I
One of the fundamental problems that links the gas hydrate resource and hazard issues is the need
for accurate assessments of the gas volumes within a gas hydrate occurrence. Most of the
published gas hydrate resource estimates have by necessity been made by broad extrapolation of
only general knowledge of local geologic conditionsz. The primary objectives of our gas hydrate
research efforts in the U.S. Geological Survey are to document the geologic parameters that
control the occurrence of gas hydrates and to assess the volume of natural gas stored within the
onshore and offshore gas hydrate accumulations of the United States. This paper begins with a
discussion of the geologic parameters that affect the stability and formation of gas hydrates, which
is followed by a description of the methodology used to assess gas hydrate resources. This paper
ends with a cumulative estimate of the in-place gas hydrate resources of the United States onshore
and offshore regions.
451
Because gas hydrates are widespread in permafrost regions and in offshore marine sediments,
they may be a potential energy resource. In-place World estimates for the amount of natural gas in
gas hydrate deposits range from 5 . 0 ~ 1 0to~1 . 2 ~ 1 0trillion
~
cubic feet for permafrost areas and
from 1.1~105to 2 . 7 ~ 1 0 8trillion cubic feet for oceanic sedimentsz. The published gas hydrate
resource estimates show considerable variation, but oceanic sediments seem to be a much greater
resource of natural gas than continental sediments. Current estimates of the amount of methane in
the world gas hydrate accumulations are in rough accord at about 7x105 trillion cubic feet2. If
these estimates are valid, then the amount of methane in gas hydrates is almost two orders of
magnitude larger than the estimated total remaining recoverable conventional methane resources,
estimated to be about 9x103 trillion cubic feet27.
The zone of potential gas-hydrate stability in each phase-diagram (figs. lA, IB, and IC) lies in the
area between the intersections of the geothermal gradient and the gas-hydrate stability curve. For
example, in figure IB, which assumes a hydrostatic pore-pressure gradient, the temperature profile
projected to an assumed permafrost base of 610 m intersects the 100 percent methane-hydrate
stability curve at about 200 m, thus marking the upper boundary of the methane-hydrate stability
zone. A geothermal gradient of 4.O0C/1O0 m projected from the base of permafrost at 610 m
intersects the 100 percent methane-hydrate stability curve at about 1,100 m; thus, the zone of
potential methane-hydrate stability is approximately 900 m thick. However, if permafrost
extended to a depth of 914 m and if the geothermal gradient below permafrost is 2.0C/100 m, the
zone of potential methane-hydrate stability would be approximately 2,100 m thick.
Most gas-hydrate stability studies assume that the pore-pressure gradient is hydrostatic (9.795
kPdm; 0.433 psgft). Pore-pressure gradients greater than hydrostatic will correspond to higher
Pore-Pressures with depth and a thicker gas-hydrate stability zone. A pore-pressure gradient less
than hydrostatic will correspond to a thinner gas-hydrate stability zone. The effect of porepressure variations on the thickness of the gas-hydrate stability zone can be quantified by
comparing each of the phase diagrams in figures IA, lB, and IC. For example, in figure IA,
458
I'
I
which assumes a 9.048 kPa/m (0.400 psi/ft) pore-pressure gradient, the thickness of the 100
percent methane-hydrate stability zone with a 610 m permafrost depth and a sub-permafrost
geothermal gradient of 2.O0C/1O0m would be about 1,600 m. However, if a pore-pressure
gradient of 1 1.3 1 I kPdm (0.500 psi/ft) is assumed (fig. 1 c)the thickness of the methane-hydrate
stability zone would be increased to about 1,850m.
The gas-hydrate stability curves in figures IA, IB, and IC were obtained from laboratory data
published in Holder and others28. The addition of 1.5 percent ethane and 0.5 percent propane to
the pure methane gas system shifts the stability curve to the right, thus deepening the zone of
potential gas-hydrate stability. For example, assuming a hydrostatic pore-pressure gradient (fig.
IB), a permafrost depth of 610 m, and a sub-permafrost geothermal gradient of 4.0"c/l00 m, the
zone of potential methane (100 percent methane) hydrate stability would be about 900 m thick;
however, the addition of ethane ( I .5 percent) and propane (0.5 percent) would thicken the potential
gas-hydrate stability zone to 1,100 m.
459
of gas and water, the pore space within the reservoir rock could be completely filled, thus making
the rock impermeable to further hydrocarbon migration. The plugging of gas pipelines and
production tubing by gas hydrates is testimony to the sealing potential of gas hydrates7. It has
also been shown that, in marine environments, gas hydrates can mechanically displace sediments
to form their own reservoir. Thus, the availability of reservoir quality rocks may not always be a
limiting factor.
IV. GAS HYDRATE RESOURCE ASSESSMENT
The major goal of this resource assessment is to estimate the gas hydrate resources in the United
States, both onshore and offshore. Similar to the assessment of the conventional resources in the
1995 U.S. Geological Survey (USGS) Oil and Gas Assessment29, this assessment of gas
hydrates is based on a play-analysis scheme, which was conducted on a province-by-province
basis. We have defined, described, and assessed all the gas-hydrate plays in the United States
regardless of their current economic or technological status. Therefore, this assessment is
concerned with the in-place gas hydrate resources--that is, the amount of gas that may exist within
the gas hydrates without reference to its recoverability. In a play analysis method, prospects
(potential hydrocarbon accumulations) are grouped according to their geologic characteristics into
plays. The geologic settings of the hydrocarbon occurrences in the play are then modeled.
Probabilities are assigned to the geologic attributes of the model necessary for generation and
accumulation of hydrocarbons. In this assessment method, geologists make judgments about the
geologic factors necessary for the formation of a hydrocarbon accumulation and quantitatively
assess the geologic factors that determine its size.
In this assessment, 11 gas-hydrate plays were identified within four offshore and one onshore
petroleum provinces (figure 2); for each play, in-phce gas hydrate resources were estimated.
Estimates for each of the 11 plays were aggregated to produce the estimate of total gas-hydrate
resources in the United States. The offshore petroleum provinces assessed consist of the US.
Exclusive Economic Zone (EEZ) adjacent to the lower 48 States and Alaska. The only onshore
province assessed was the North Slope of Alaska, which included State water areas and some
offshore Federal waters. The provinces shown in figure 2 are geographic in character; however,
their formation represents an attempt to group the individual petroleum provinces along broad
geologic lines. Maps depicting the geologic data required for this hydrate assessment have been
included in the U.S. Geological Survey 1995 National Oil and Gas Assessment CD-ROM29.
Maps of bathymetry, sedimentary thickness, total organic carbon (TOC) content of the sediments,
seabed temperature, geothermal gradient, and hydrate stability zone thickness have been published
for all four offshore provinces assessed in the US. Geological Survey 1995 National Oil and Gas
Assessment CD-ROM29. Maps depicting the thickness of the onshore gas-hydrate stability zone
in northern Alaska are also included in the Assessment CD-ROM29.
1
4
The estimates of in-place gas-hydrate resources included in this report are presented in the form of
complementary cumulative probability distributions (fig. 3). These distributions summarize the
range of estimates generated by the FASPU computer program29 as a single probability curve in a
"greater than" format (fig. 3). Our estimates are reported at the mean and at the 95th, 75th. 50th,
25th, and 5th fractiles. We consider the 95th and 5th fractiles to be "reasonable" minimum and
maximum values, respectively. In-place gas resources within the gas hydrates of the United States
are estimated to range from 112.765 to 676,110 trillion cubic feet of gas (TCFG) [3,193 to 19,142
trillion cubic meters of gas (TCMG)], at the 0.95 and 0.05 probability levels, respectively (fig. 3).
Although these ranges of values show a high degree of uncertainty, they do indicate the potential
for enormous quantities of gas stored as gas hydrates. The mean in-place value for the entire
United States is calculated to be 320,222 trillion cubic feet of gas (TCFG) [9,066 trillion cubic
meters of gas (TCMG)]. This assessment of in-place gas hydrates represents those deposits that
constitute the resource base without reference io recoverabiliry.
V. REFERENCES ClTED
1. Kvenvolden, K.A., 1988, Methane hydrate--A major reservoir of carbon in the shallow
geosphere?: Chemical Geology, v. 71, p. 41-51.
2. Kvenvolden, K.A., 1993, Gas hydrates as a potential energy resource -- A review of their
methane content, in Howell, D.G., ed., The Future of Energy Gases: U.S. Geological
Survey Professional Paper 1570, p. 555-561.
3. Makogon, Y.F., 1981, Hydrates of natural gas: Tulsa, Penn Well Publishing Company, 237
P.
4. Collett, T.S, 1993, Natural gas production from Arctic gas hydrates, in Howell, D.G.,
ed.,
The Future of Energy Gases: U.S.Geological Survey Professional Paper 1570, p. 299-312.
5. Franklin, L.J., 1981, Hydrates in Arctic Islands, in Bowsher, A.L., ed., Proceedings of a
Workshop on Clathrates (gas Hydrates) in the National Petroleum Reserve in Alaska, July
16-17, 1979, Menlo Park, California: U S . Geological Survey Open-File Report 81-1298, p.
18-21.
4
460
6.
7.
8.
9.
L
IO.
')
,1
11.
12.
13.
14.
15.
16.
17.
18.
Yakushev, V.S., and Collett, T.S., 1992, Gas hydrates in Arctic regions: risk to drilling and
production: Second International Offshore and Polar Engineering Conference, June 14-19,
1992, San Francisco, California, Proceedings, p. 669-673.
Sloan, E.D., 1990, Clathrate hydrates of natural gases: New York, Marcel Dekker, Inc., 641
P.
Makogon, Y.F., Trebin, F.A., Trofimuk, A.A., Tsarev, V.P., and Cherskiy, N.V., 1972,
Detection of a pool of natural gas in a solid (hydrate gas) state: Doklady Academy of
Sciences U.S.S.R., Earth Science Section, v. 196, p. 197-200.
Cherskiy, N.V., Tsarev, V.P., and Nikitin, S.P., 1985, Investigation and prediction of
conditions of accumulation of gas resources in gas-hydrate pools: Petroleum Geology, v. 21,
p. 65-89.
Collett, T.S., 1993, Natural gas hydrates of the Prudhoe Bay and Kuparuk River area, North
Slope, Alaska: American Association of Petroleum Geologists Bulletin, v. 77, no. 5 , p. 793812.
Collett, T.S., 1983, Detection and evaluation of natural gas hydrates from well logs, Prudhoe
Bay, Alaska, in Proceedings of the Fourth International Conference on Permafrost,
Fairbanks, Alaska: Washington D.C., National Academy of Sciences, p. 169-174.
Judge, AS., 1988, Mapping the distribution and properties of natural gas hydrates in
Canada: Proceedings of the American Chemical Society Third Chemical Congress of the
North American Continent, June 6-7, Toronto, Ontario, Abstract no. 29.
Judge, A.S., and Majorowicz, J.A., 1992, Geothermal conditions for gas hydrate stability in
the Beaufort-Mackenzie area: The global change aspect: Global and Planetary Change, v. 98,
no. 213, p. 251-263.
Brooks, J.M., Cox, B.H., Bryant, W.R., Kennicutt, M.C., Mann, R.G., McDonald, T.J.,
1986, Association of gas hydrates and oil seepage in the Gulf of Mexico: Organic
Geochemistry, v. IO,p. 221-234.
Brooks, J.M., Field, M.E., and Kennicutt, M.C., 1991, Observations of gas hydrates in
marine sediments, offshore northern California: Marine Geology, v. 96, p. 103-109.
Yefremova, A.G., and Zhizhchenko, B.P., 1974, Occurrence of crystal hydrates of gas in
sediments of modern marine basins: Doklady Akademii Nauk SSSR, v. 214, p. 1179-1181.
Ginsburg, G.D., Guseinov, R.A., Dadashev, A.A., Ivanova, G.A., Kazantsev, S.A.,
Soloviev, V.A., Telepnev, Ye.V., Askery-Nasirov, R.E., Yesikov, A.D., Mal'tseva, V.I.,
Mashirov, Yu.G., and Shabayeva, I.Yu., 1992, Gas hydrates in the southern Caspian Sea:
Izvestiya Akademii Nauk Serya Geologisheskaya, v. 7, p. 5-20.
Ginsburg, G.D., Soloviev, V.A., Cmnston, R.E., Lorenson, T.D., and Kvenvolden, K.A.,
1993, Gas hydrates from the continental slope, offshore from Sakhalin Island, Okhotsk Sea:
Geo-Marine Letters, v. 13, p. 41-48.
Kvenvolden, K.A., and Bamard, L.A., 1983, Hydrates of natural gas in continental margins,
in Watkins, J.S., and Drake, C.L., eds., Studies in Continental Margin Geology: American
Association of Petroleum Geologists Memoir 34, p. 631-641.
Shipboard Scientific Party, 1986, Sites 614-624, in Bouma, A.H., and others, Proceedings,
Deep Sea Drilling Project, Initial Reports, Washington D.C., U.S. Government Printing
Office, v. 96, p. 3-424.
Shipboard Scientific Party, 1994, Site 892, in Westbrook, G.K., and others, Proceedings,
Ocean Drilling Program, Initial Reports, College Station, Texas, v. 146, p. 301-375.
Kvenvolden, K.A., McDonald, T.J., 1985, Gas hydrates of the Middle America Trench,
Deep Sea Drilling Project Leg 84, in von Huene, R., Aubouin, J., and others, Initial Reports
Deep Sea Drilling Project, Washington, D.C., U.S. Government Printing Office, v. 84, p.
667- 682.
Shipley, T.H., and Didyk, B.M., 1982, Occurrence of methane hydrates offshore Mexico, in
Watkins , J.S., Moore, J.C., and others, Initial Reports, Deep Sea Drilling Project:
Washington D.C., U.S. Government Printing Office, v. 66, p. 547-555.
Kvenvolden, K.A., and Kastner, M., 1990, Gas hydrates of the Peruvian outer continental
margin, in Suess, E., von Huene, R., and others, Proceedings, Ocean Drilling Program,
Scientific Results, College Station, Texas, v. 112, p. 517-526.
Shipboard Scientific Party, 1990, Site 796, in Tamake, K., and others, Proceedings, Ocean
Drilling Program, Initial Reports, College Station, Texas, v. 127, p. 247-322.
Shipboard Scientific Party, 1991, Site 808, in Taira, A., and others, Proceedings, Ocean
Drilling Program, Initial Reports, College Station, Texas, v. 131, p. 71-269.
Masters, C.D., Root, D.H., and Attanasi, E.D., 1991, Resource constraints in petroleum
production potential: Science, v. 253, p. 146-152.
Holder, G.D., Malone, R.D., Lawson, W.F., 1987, Effects of gas composition and
geothermal properties on the thickness and depth of natural-gas-hydrate zone: Journal of
Petroleum Technology, September, p. 1147-1 152.
Gautier, D.L., Dolton, G.L., Takahashi, K.I., and Varnes, K.L., 1995, National assessment
of United States oil and gas resources on CD-ROM: U.S. Geological Survey Digital Data
Series 30.
~Y
20.
21.
22.
23.
I
24.
25.
26.
27.
28.
29.
461
TEMPERANRE ('C)
o.lo
10
20
30
-- -
(i.alhrmd g n d h
4.0'UlW mal*.
TEMPERATURE ('C)
o.io
20
10
30
3.2'UtOOmalan
2
Mean 320,222
F95 112,765
F5 676,110
\
0
20
10
462
POTENTIAL IMPACT
1
1.
Methane (natural gas) produced from conventional gas deposits is plentiful, easily delivered
(as a gas) t o the user by an in-place domestic distribution system, and as a fuel, methane i s
clean burning and has a respectable heat content. The prospect of methane recovery from vast
oceanic gas hydrate deposits, however, argues for an almost indefinite supply of methane, the
recovely of which will probably speed the developmentof the gas-energy economy to replace
the current oil-based economy. In addition to this development being ecologically sound, oil may
be viewed better as an industrial feedstock than as a direct fuel, so long as a convenient,
alternate source of energy such as methane is madeavailable.
Methane is particularly amenable to transport and handling as a gas in pipelines and transport
to point use in pipes within contiguous land areas. In fact, most of the early work into the
chemistry of methane hydrates was undertaken by the gas transport industry because hydrates
were forming and clogging gas pipelines even a t relatively high temperatures and moderate
pressures. Current technology frequently requires that methane fuel be moved as either
compressed gas or as liquefied gas, as when natural gas is imported to the U.S. distribution grid
from foreign gas fields. Of course, many fixed-site utilizations for natural gas (e.g. space
heating, electrical power generation, or cooking) rely exclusively on gaseousmethaneas a fuel
stock. Where technical or geographic difficulties prohibit the use of piped distribution,
however, other means of distributing gas must be developedfor use. Storage of methane (e.g.,
compressed gas) at the point of use may also a problem so long as a continuous piped supply is
not available.
i
t
Both compressedgas and liquified gas, as transport media, possess serious safety concerns
associated with the flammability of the material (compressed natural gas) or the cold
temperatures and ultimate flammability/potentially explosive nature of the liquefied medium.
This paper suggests and examines a new application of clathrate chemistry, which could have a
significant impact on methane fuel use and distribution if implemented. We call attention to a
potential third alternative for bulk gas transport and point-of-use storage, which would be
energy dense, fairly stable, non-flammable in bulk, easy to transport, and potentially useable
as-is for motor fuels.
SYNTHETIC METHANE CLATHRATE FUEL (SMCF)
Naturally occurring methane hydrates are not stable at sea level ambient temperatures and
pressures. However, it is not intendedto use pure methane hydrate as the basis for the new fuel
transport andstorage media. Current experimental results show that hydrates can be fabricated
both from natural gas more dense than methane (de Boer e t al, 1985) with variable physical
property ranges that are stable well above the normal methane hydrate P-T stability field
(Sloan, 1990). Also, in the course of producing synthetic methane hydrate, metastabilities
about the liquidus line exist (Stern e t al., 19961, which may point toward controlling
metastability ranges of methane hydrate rather than expanding the methane hydrate stability
field. This broader stability of naturally occurring multiple gas clathrates, poorfy undemood
metastability, and relative ease with which synthetic methane (based) hydrate can be formed,
leads us tosuggestthat research fabricating special property methaneclathrates is feasible and
that research should be undertaken tofabricate a new methane fuel storage andtransport media.
463
0u)
4
;
1
100-
c
YI
k!
u)
g
d
1000-
2500500010.000-1
.20
TEMPERATURE %
60
40
40
20
40
I
)I ______i
I
Formula
Fuezt;n
It
of
Methane Gas
I CH4
,
1
1
Density g/cc
1 Energy Content 1
I
Btu/lb
Methane-water solid
(natural hydrate
potential)
Octane(gaso1ine)
JP-5'1
277
*4
1 - 10
277
*4
CH4( H20),
GH,,
C,4HN
Energy Content
Btu/ft3
7x1 0 - 4
Methane-water solid
-----
/184,000*4*6
0 70
19,000 * 3
840,000 * 3
0 77
18,500 * 3
930,000 * 3
Because it is unlikely that the energy density of a clathrate-based fuel media will ever
significantly approach that of liquid petroleum fuels, the clathrate fuel is clearly not
appropriate for all vehicles. For instance, vehicles with small volumes capacity for fuel
storage, such as private motor vehicles and aircraft, where weightlvolume is a major factor,
are not likely end-point users. Larger platforms, however, such as ships and possibly highspeed trains which could be made environmentally benign (with respect t o noise of energy
generation and exhaust), might be possible end-users, especially when the other attributes of
clathrate based fuel media, such as inhibiting uncontrolled fires and explosions in commercial
applications and explosion damping and deflecting in military applications, are taken into
consideration.
The proposed safer transport system utilizes gas hydrates (clathrates) which are physical
associations of water ice and low molecular weight gas molecules (e.g. methane, ethane, propane
or butane). These clathrates form spontaneously when water and a suitable low molecular
weight gas (e.9. methane, carbon dioxide, hydrogen sulfide, chlorine) are mixed at suitable
temperatures (generally low) and pressures (generally moderate). Indeed, the older literature
contains many references t o gas hydrates forming spontaneously in natural gas transmission
pipelines, and often blocking them; this potential situation requires the drying of gas prior t~
pipeline transportation (DOE, 1987).
Research into the low pressure species has mainly concerned todeveloping techniques that w i II
allow for industrial capability to efficiently dissolve, or gasify hydrates. Where bonding
interaction between guest and host molecules might be enhanced somewhat, gas that normally
does not hydrate, such as hydrogen, may be bound into specially formulated hydrates. If host
cavities were t o be lined with groups having a high hydrogen bonding character, such as
hydroxyl or amino groups, other factors, such as the solubility parameter of the host, would be
of less importance. Increased hydrogen bonding power might also be induced by charging guest
molecules prior t o exposure t o hydrate lattice, or through the use of magnetic field charging
(moving the fuel in a field, pulsing a field, or moving a field with respect to the orientation of
the hydrate). Release of gas could be induced through heating, lowering of pressure, o r
electronic stimulation that would produce effectssimilar t o that of microwaving food(where the
frequency of the microwave is specific to water molecules).
FUEL SYSTEM REQUIREMENTS
The proposed SMCF storage and transport system would consist of three separate components:
(1 ) Formation Module, (2) Transport Vessels, and (3) Gasseparation Unit.
(1 ).
Hydrate Formation Module (HFM). Methane hydrates are stable under moderate
pressures, and low temperatures (Fig. 1). The HFM will consist of a pressure vessel into
which are pumped stabilizer, water spray, and methane; the P-T conditions of formation are
presently uncertain but may be different from those neededfor stability of the special hydrate
during storage. Recent research shows that the methane hydrate forms immediately upon
mixing water with the gas, when the system is within the stability field of the clathrate (Peter
Brewer, MBARI, pers. comm., November, 1996). Once the hydrate is formed, the material
would be removed from the HFM, and transferred t o the transport vessel for movement to point
of use or distribution.
I
I!
(2) Hydrate Transport Vessel (HTV). The Kn/ would consist of a insulated container which
could contain the stable special hydrate at ambient to moderate pressures. The insulation would
more than likely consist of plastic foam such as is used by the refrigeration industry; vacuum
jackets would be avoidedbecauseof Cost andsafetyconcerns. The HTVcould be fabricated in any
desired shape, and might evolve t o be conformal t o the hull or some interior structural
members of the platform using the stored gasas a fuel, for example, in the double hull space of a
ship.
(3) Gas Separation Unit (GSU). The GSU could be integral tothe HTV,or separate, as mandated
by the ultimate use of the released fuel gas. The clathrates are unstable in the presence of
elevated heat; the hydrate could be decomposedby direct heating (e.g. a clathrate slush would be
transferred t o a heatedvessel: gas evolves from the slush and escapesfor use, and the water
465
from the hydrate is discarded or retained for use in making more hydrate later). Alternatively,
the hydrate slush could be sprayed with water, the heat in which would be sufficient m
decomposethe hydrate. In either case, the evolved gaswould be routed toa device (e.g. engine)
which could use the combustible gasas a fuel.
It must be notedthat there is no inherent reason why the units listed above would necessarily be
separate components. For example, the storage vessel could contain integral subsections which
would allow both formation of anddecompositionof thegas hydrate right in the KTV. Further, it
is technically possible to design and build an internal combustion engine which would use
hydrate as the only, or majority, fuel; such a system would be similar to water-injection
technology as applied t o internal combustion engines, as in some experimental fighter plane
engines andrace cars.
W CH2781-
Sloan, E.D., Jr. 1990. Clathrate Hydrates of Natural Gases. Marcel Dekker, Inc., New York and
Basel. 641 pp.
Stern, L.A., Kirby, S.H. & Durham, W.B. 1996. Peculiarities of methane clathrate hydrate
formation and solid-state deformation, including possible superheating of water ice. Science,
22& 1843-1 848.
466
Another factor that is relatively unappreciated is the fact that although the small 5 (D)
cage is common to all 3 structures, I, I1 and H, the symmetry and size of these small
cages is different, and hence their behaviour towards guest molecules should also be
quite different. This is very directly evident from the chemical shift parameters of xenon
trapped in the small cages which suggest that the structure II and H small cages are
significantly larger and less symmetric than the Structure I small cage. Experiments
with CO, have confirmed this idea.
Experimental
Bromine hydrate single crystals suitable for diffraction were grown from solutions of
different concentrations to give material of different morphologies and hydration
numbers see table 1. The crystal structures were determined on a Siemens
diffractometer equipped with a CCD detector using Mo &radiation. In all, 16 different
crystals were examined. The structures were solved with the Shextl software package.
I
,
Double hydrates of xenon or CO, were made by sealing into lOmm pyrex tubes
measured quantities of powdered ice, along with the appropriate large cage and small
1 day to several
cage guests. Samples were conditioned for times lasting from
weeks. NMR spectra were measured on an Bruker MSL 200 spectrometer equipped
with a double-tuned solenoid probe suitable for cross-polarization and dipolar
decoupling. Temperature variation was achieved with a cold gas-flow system and a
temperature controller. Spectral simulations were carried out with the Bruker Xedplot
package.
other crystals were determined to cover the different crystal morphologies. For all of
the crystals studied, the space group turned out to be the one reported by Allen and
P4Jmnm. At -100C the lattice parameters are a=23.044, ~112.075A.
Jeffrey @:
One view of the structure is shown in fig.2. The unit cell can be represented" as
follows: 2DA.8D,8TA.8T,.4P. 172H,O with a 5"6' (P) cage, two distinct 5"6' (TA, T,)
cages similar to those in str. I, and two kinds of 5" (Dh, 0,) cages. The reason for the
difficulty in finding a good structural solution becomes apparent when examining the
guest positions in fig. 2. As opposed to structures with hydrocarbon guests, in the
bromine hydrate case the scattering is dominated by the highly disordered bromine
guest: the P cage has as many as 12 possible positions for bromine; the T cages each
have 14 (T,) or 15 (T,), with the site occupancies varying from 20 down to 2 %. The
highly anisotropic site distribution is evident especially for the T cages, where the centre
of the cavity can be seen to be clear as the bromine atoms are confined to be near to
the equatorial plane of the cage. The variable hydration numbers observed for bromine
hydrate arise from the variable degree of filling of the large cages ( table 1 ). The
minimum hydration number possible is 8.6.
It will be another challenge to work out a thermodynamic model for bromine hydrate, as
the clathrate has 3 types of large cage ( P,TA, T, ) suitable for bromine, and two small
cages (DA,D,) which may contain oxygen or nitrogen from the air ( some electron
density in the small cages was indeed observed ). The bromine hydrate structure is the
I hydrate. Attempts to
only one of its type, as all other molecules of this size form str. 1
form a different hydrate with xenon as helpgas initially gave a structure II hydrate, but
with time this was seen to revert back to the bromine hydrate structure. The fact that
the large cages do not need to be full seems to be unique as well. This suggests that
guest - guest interactions may play an important role in dictating structure type.
Significant guest - guest interactions are likely for the electron-rich bromine molecule
which should have a sizable molecular quadrupole moment. Another challenge is the
understanding of the reasons for the different crystal morphologies. It appears that
some kind of "self-inhibition'' takes place to suppress the growth of certain crystal faces
when the bromine concentration varies in the growth solution.
Based on these observations, what would one expect for the Str. H small cages ?
From the "*Xe spectra, the D and D' cages should be as large as the str. II D cage and
468
469
1:10.68
1:9.41
1:9.09
1:8.94
123.62
0.805
0.914
0.946
0.962
0,998
I
I
II
II
H
H
H
cage type
5'*(D)
5I26,(T)
5"(D)
5',6'( H)
512(D)
4'5'6'(D')
5"6'(E)
symmetry
m3
42m
3m
43m
mmm
62m
6lmm
radiudk
2.50
2.93
2.50
3.28
2.50
2.50
4.1
6"
ox,(iso)b
(PPm)
(PPm)
-242
-152
-231
-80
-231
-212.4
-2 1
-16
0
-13.6
-31.8
_________
qd
0.8
_-_-__-
----_-____________----------------------------------------estimate from X-ray diffraction data; isotropic chemical shift;" 6=(2/3)Ao - chemical
shift anisotropy;dasymmetry parameter - departure of cage geometry from axial
symmetry
470
4
/
Figure 2. General
view of the bromine
hydrate structure with
the
view
approximately along
the
z
direction;
hydrogen atoms are
omitted for clarity; the
bromine atoms are
shown in their many
possible disordered
positions in the cages,
the
maximum
occupancy being one
molecule per cage
250
ma
zoo
PP"
100
50
7
.1j
1
;
:
!
?;
-2y)
471
1 9
100
Ppm
INTRODUCIION
The Gulf of Mexico continental slope is a natural laboratory for gas hydrates that
contain hydrocarbons from deeply buried thermogenic sources. Thermogenic
hydrocarbons (oil and gas) from actively generating Mesozoic source rocks (>6 km
burial depth) migrate vertically along conduits associated with actively-moving salt
structures and faults to subsurface reservoirs (2-4 km) of Tertiary age'. The
hydrocarbon trapping system is so "leaky" that large volumes of thermogenic
hydrocarbons reach the sea floor', and enter the water column*.
Although biogenic gas hydrates are abundant on the Gulf slopes, oil and gas from
deep source rocks create a geochemically complex and physically dynamic
environment for thermogenic gas hydrates at the sea floor. Structure II gas hydrate
containing C1-C4 thermogenic hydrocarbon gases was first sampled in 1984 by piston
cores in 530-560 m water depths on the Gulf slope offshore Louisiana4.
Identification of the hydrate as structure I1 was based on the relative abundance of
the C3 and i-C4 hydrocarbons4. The structural assignment was corroborated using
solid-state nuclear magnetic resonance (NMR)5. Research on gas hydrates of the
Gulf slope, however, has advanced rapidly in the last few years, and our objective
here is to summarize new results.
THE BUSH HILL STUDY AREA
The Bush Hill site in the Green Canyon area of the Gulf slope offshore Louisiana is
a well-documented site for study of thermogenic gas hydrates (27O47.5' N and
91O15.0' W). Bush Hill is a fault-related sea-floor mound about 500 m wide, with
relief of about 40 m6. Water depth of the mound crest is about 540 m, where mean
water temperature is about 7" C (range = 6 to 11" C) Phase equilibria models indicate
that Bush Hill is within the stability zone of thermogenic gas hydrates (Sloan, E.D.,
pers. communication).
Sea-floor sediments contain crude oil and related free hydrocarbon gases. Bacterial
oxidation of these hydrocarbons produces COz which precipitates as authigenic
carbonate rock with isotopically-light 8l3C values7. The crest of the mound is
colonized by seep-dependent chemosynthetic organisms including bacterial mats,
vestimentiferan tube worms, and methanotrophic mussels8. Persistent natural oil
slicks appear on satellite remote sensing images of the sea surface over Bush Hi@.
Thermogenic gas hydrates and gases that vent to the water column at the mound
crest are readily sampled by research submarines. Copious streams of gas vent
continuously to the water column where subsurface migration conduits intersect
the sea floor9Jo. Thermogenic gas hydrates form around the orifices of gas vents.
The gas hydrates at vents are not dispersed in sediments as nodules or thin seams,
but instead occur as continuous masses. Lens-shaped masses of yellow to orange gas
hydrates breach the sea-floor at numerous locations on the crest of Bush HillgJo.
The hydrates form sediment-draped mounds 30-50 cm high and up to several m in
width, with exposed gas hydrate visible at the edges of mounds'o.
Vent Gases
The C1-C5 hydrocarbons of the vent gases are dominated by methane (Cl = 91.194.7%), and 813C values of C1 are within the narrow range of -42.4 to -45.6 %o PDB
(Table 1).The Ci-CS distributions and 813C of C1 of the vent gases (Table 1)correlate
to gases from underlying subsurface reservoirs of Jolliet Fieldll.
472
i;
C1
Vent Gas 10
Vent Gas 10
Vent Gas 10
Vent Gas 10
Vent Gas 10
93.2
93.5
94.7
94.6
91.1
Hydrate (1I)'o
Hydrate(n)lo
Hydrate(I1)IO
Hydrate(H)6
813~
C2
C3
i-C4
n-C4
i-C5
-43.3
-42.5
-45.6
-43.8
-42.4
4.3
4.3
3.9
3.8
4.8
1.5
1.4
0.7
0.7
1.8
0.3
0.2
0.1
0.1
0.4
0.6
0.4
0.5
0.5
1.2
0.3
0.2
0.2
0.3
0.8
71.7
80.2
72.1
-36.3
-38.5
-39.9
10.6
9.4
12.4
12.6
7.3
11.4
2.6
1.6
2.3
1.7
1.2
1.6
0.8
0.3
0.3
21.2
-29.3
9.5
7.5
2.5
17.5
41.1
Structure I1 hydrate
Hydrocarbon compositions of massive hydrate lenses of Bush Hill are shown in
Table 1. The C1-C5 hydrocarbons of the hydrate gases are dominated by C1 (71.780.2%). The 813C values of C1 are in the range of -36.3 to -39.9 ?& PDB, somewhat
heavier than vent gases, possibly because of bacterial activitylo. The Cz and C3
hydrocarbons are both present in similar but relatively high percentages compared
with the vent gas (Table 1). Preliminary NMR of an intact hydrate sample preserved
in liquid nitrogen is consistent with structure I1 hydrate (Ripmeester, J., pers.
communication).
Structure H gas hydrate
Structure H hydrates produced in the laboratory can enclose larger molecules than
structure I or I1 hydrates, including common thermogenic hydrocarbons such as iC5. Given the widespread occurrence of petroleum, it was postulated in 1993 that
structure H hydrate could co-exist in nature with structure Il hydratel2.
Evidence for the natural occurrence of structure H gas hydrate at the Bush Hill
locality was first reported in 19946. Massive amber-colored gas hydrate breached the
sea-floor. It had been exposed when a buoyant lobe of hydrate broke free of the
sediment and floated upwards into the water column. Identification of structure H
hydrate was based on abundant i-C5, which represented 41.1% of the total C1-C5
hydrocarbon distribution of the sample (Table 1). The 8l3 C of C1 from the sample is
heavy (-29.3 %O PDB), possibly because of bacterial activity6.
Experimentally-Precipitated gas hydrate
Gas hydrate was experimentally precipitated at the crest of Bush Hill in 1995 using
natural vent gases as the starting materiallo. Water temperatures during
experiments were 9.0-9.2OC. Precipitation of white to yellow gas hydrate was noted
to occur within minutes.
The hydrocarbon compositions of experimentally precipitated gas hydrates are
similar to vent gas compositionslO. The c 1 - C ~hydrocarbons of the experimentally
precipitated gas hydrates gases are dominated by methane (C1 = 87.7-93.9%),and the
813C values of C1 are withii the 40.5 to -45.3 %O PDB range.
CONCLUSIONS
Thermogenic gas hydrates occur on the Gulf of Mexico continental slope because of
active vertical migration of oil and gas to the sea floor within their stability zone.
The Bush Hill seep site on the Gulf slope is an important case history. Massive
thermogenic gas hydrates occur in association with the orifices of hydrocarbon
vents. Both structure I1 and structure H hydrates appear to co-exist in this
473
3.
4.
Brooks J.M.; M.C. Kennicutt 11; R.R. Fay; T.J. McDonald; Sassen R. Science 1984
409-411.
5.
Davidson, D.W.; S.K. Garg, S.R. Gough, Y.P. Handa; C.I. Ratcliffe; J.A.
Ripmeester; J.S. Tse; W.F. Lawson Geochim. Cosmochim. Acta, 1986 a 6 1 9 623.
6.
7.
8.
MacDonald 1.R; G.S. Boland; J.S. Baker: J.M. Brooks; M.C. Kennicutt 11; R.R.
235-247.
Bidigare R.R. Marine Biol. 1989
9.
MacDonald, I.R.; N.L. Guinasso; R. Sassen; J.M. Brooks; L. Lee; K.T. Scott
Geoloey 1994 2 699-702.
10.
11.
12.
22 1029-1032.
1988 2 39-59.
414
Below the main frame of the vehicle is an open tool sled structure which housed the gas tank for the
methane, and mixed gas experiments. The basic system is similar to that described earlier by us (7).
A pressure regulator set to 0.7 Mpa above ambient pressure and a needle valve that limited the flow
rate to about 125 ml per minute were in line. Gas was distributed to the reaction chambers by four
hydraulically actuated pistons (Allenair) operating quarter turn valves that were controlled directly
by us through the Point Lobos control room interface. The valves and reaction tubes were mounted
on an aluminum box frame carried on the front of the vehicle and positioned for optimum viewing.
The reaction tubes were vented to the outside ocean by an overflow tube at the top of the cylinder,
arranged so as to trap a small gas bubble at all times while allowing for pressure equalization. A
peristaltic pump was attached to all reaction cylinders to flush sea water at the local temperature and
salinity through the apparatus prior to gas injection. The gas flow schematics are illustrated in Fig. 1.
For C 0 2 release we faced the problem of dispensing a liquid at the pressures and temperatures
encountered. Two systems were used: overpressuring the liquid C02 with a bubble of He gas to
expel the fluid from a vertically mounted tank; and use of a hydraulically activated piston to expel
the liquid C02 from a pressured reservoir. Once the C02 was expelled the gas flow, valving and
reaction vessel were identical to that for methane.
The gas was expelled into acrylic reaction cylinders ( 60 x 4.5 cm; volume 954 cm3 ) mounted
vertically on the frame; a second reaction chamber with a plane viewing surface, and large enough to
contain a temperature probe of five thermistors was also constructed and used in the later
experiments. The chambers contained either sea water alone, or were partially filled with sediments
of varying grain sizes. No provision was made for sample recovery on board ship at this time, and
475
the observations were purely visual, although the environmental conditions for the experiment are
well defined by the CTD sensor.
OBSERVATIONS
Methane Hydrate Formation
In our first experiment (January, 1996) we used pure methane gas ( Linde); the thermodynamic
boundary for methane hydrate formation posed by the local hydrographic (P,T,S) conditions in
Monterey Bay is close to 525 m water depth. We paused at about 500m to inject a small amount of
gas as a precaution to clear the lines, then drove Ventanato a depth of about 910m and switched on
the peristaltic pump to flush the system of trapped sea water and achieve T,S equilibrium with the
external medium ( approximately 3.9OC; 34.42%0).Once the system had flushed we injected
methane gas by bubble stream through a IO pm porous frit at the bottom of the reaction cylinders.
Methane hydrate formation occurred within a few seconds, seen easily as a bright reflective bubbly
mass at the gadwater interface at the top of the tube. The hydrate formed as a white rind on the gas
bubble surface that appeared to separate the water and gas from further rapid reaction unless some
mechanical disturbance occurred. The reaction was reproducible; an injection into a second reaction
cylinder produced an identical result. No significant induction period was observed, nor was
anything other than gas and natural sea water present.
Of the two remaining reaction cylinders one contained about 20 cm of coarse sand, and the other a
similar amount of fine grained mud. Here the hydrate formation was again first seen at the top of the
tube. But the pores of the coarse sand matrix were soon observed to be flooded with hydrate, which
sealed off further gas flow. The effect was to create cracking and then lifting of a major piece of the
solidified sand column. Gas flow through the fine mud caused channels to open up since the
capillary pressure for the gas to enter the pore spaces was higher than that required to displace the
sediment. White hydrate masses quickly formed on the walls of the channels and gas created void
spaces with an appearance and effect quite different from the coarse sand matrix. On recovery of the
vehicle the hydrates formed in our experiment dissociated during transit to the surface, and we were
not able to recover specimens for analysis.
small bore tube, was used for sample introduction in this experiment for fear of plugging the
apparatus completely. The effect was to create globules of liquid C02 which, after sticking
temporarily to the release port, rose slowly to the upper interface. There it appeared that a fine film
of accreting hydrate gave a pearly appearance to the external surface of the globules, which did not
coalesce but remained as separated units.
C
Thermal Signatures
In a modification of our apparatus we replaced one of the cylindrical reaction tubes with a plane
faced larger unit for better viewing. In this unit we placed a heat flow probe constructed by the
Woods Hole Oceanographic Institutions Alvin group. This consists of a metal rod about 1 m long
with five thermistors each separated by about IO cm. Readout from the probe was fed directly to the
control room for real time monitoring of the experiment. Working with pure C Q gas in sea water
we observed the temperature rise from the heat of formation during hydrate creation on bubble
surfaces at the gadwater interface. Disturbance of the upper boundary by bubble flow created a
mixed layer several centimeters deep which served to dissipate the heat, and it was not possible to
gain a more quantitative estimate of the amount of hydrate formed.
On termination of the experiment and on raising Ventana to shallower depths we immediately
observed a temperature drop due to quasi-adiabatic expansion cooling of the unreacted gas in the
head space. Adiabatic cooling of the sea water itself is much smaller, but can be evaluated since the
equation of state for sea water is well known (8). The temperature drop associated with gas
expansion continued on raising until the hydrate decomposition point was reached. This point was
not identical with the external oceanic boundary condition for dissociation due to the lower
temperature at equal pressure within the apparatus, but it was clearly defined by a sharp break in the
temperature trend due to cooling from the heat of dissociation.
DISCUSSION
From the experiments we have carried out to date we can make some interesting conclusions about
the manner and characteristics of hydrate formation in the deep sea, where the reaction medium
contains the .normal assemblage of suspended particles, bacteria, and trace gases which characterize
the natural environment. Firstly we have repeatedly made hydrates of several gases, each within the
period of a very few minutes or seconds, by the simple technique of direct gas injection with no
shaking or ice nucleation step whatsoever. The initial manifestation of this was the creation of
hydrate coated bubbles at the gadwater interface; but hydrate also formed in seconds to minutes
within the pore spaces of marine sediments where no provision was made for gas trapping. We
surmise that passage of the gas bubbles around the sediment grains caused sufficient surface renewal
that hydrate formed in a similar manner to the more easily visualized upper boundary, but with
smaller unit size granules. No significant induction or lag period was observed for hydrate formation
for any gas yet injected in this manner.
Once formed the hydrate structures appeared quite stable. That behavior is consistent with the idea
that the hydrate rind on bubble surfaces separates the inside gas from the outside water well enough
that further growth must occur only slowly by diffusion of the reactants through the hydrate skin.
Unless some defect or fracturing of the hydrate rind occurs, this appears to be the rate limiting step.
Growth of hydrate in marine sediments is critically dependent on the grain size of the material. In a
coarse material (sand) flooding of the pores results in cementing of the sediment into a massy unified
I
477
structure within seconds, yet yields no hydrate nodules of the kind often reported in nature (9). These
nodular structures were observed in the process of forming in the flow channels carved by gas in
experiments in tine grained mud, and the contrast between hydrate formation in the coarse and tine
matrices was dramatic.
Our work with C02 hydrate has yielded results'relevant to the proposed disposal of C02 in the deep
ocean (IO). For instance the ease with which C02 hydrate forms will pose a challenge to deep
injection facilities concerned with plugging of the system; and the observation of the relative
stability of the hydrate coated globules restricts interaction between disposed C 0 2 and the
surrounding ocean water. Furthermore, C02 hydrate did not separate spontaneously from unreacted
C02; instead it formed a mass of intermediate density between sea water and liquid COz. Our
observations were consistent with the description by Sakai et al. (1 1) of the natural venting of C02
rich fluids on the ocean floor.
Our experience with C 0 2 hydrate formation is that the liquid C02 used experimentally requires
excellent technique to handle. Post cruise analysis of our experiment carried out with He
overpressure indicated by formal calculation (using the Peng-Robinson (12) equation of state) that
the gas injected was indeed a COZiHe mixture, since under the conditions we used (about 4.40 C,
1800 psia) to prepare the gas reservoir then about I O mol% He will dissolve in the liquid CO2.
Release of this at our in situ experimental conditions will form a mixture of about 20 volume %
liquid phase, and 80 volume % vapor, accounting for our observations.
Once formed from sea water/gas (or liquid) contact, the hydrates are stable over a period of several
weeks, and quite possibly very much longer indeed, even though sea water and unreacted gas or
liquid are separated only by a thin hydrate film. The initial attempt we made to study this was
successful in separating the experimental apparatus from the vehicle, and leaving it in place. In
future experiments we will leave hydrates within sediment matrices for later recovery, and arrange
for greater sea waterihydrate contact, since water flow around the hydrates was quite restricted in the
present system.
Finally we are devising means for surface recovery of the experimental material for laboratoty
investigation, and wish to apply the knowledge we have gained to a variety of important
geochemical and gas disposal problems.
REFERENCES
1. Sloan, E.D. Jr. Clathrate Hydrates ofNatural Gases. Marcel Dekker. 1990 pp 641.
2. Stern, L.A.; Kirby, S.H.; Durham,W.B. Science 1996,273,1843-1848.
3. Dickens,G.R.; Quinby Hunt, MSGeophys. Res. Lett. 1994,8,2115-2118.
4. Sloan, E.D. In: Gas Hydrates: Relevance to World Margin Stability and Climatic Change 1996,
pp. 1-38.
5. Robison, B.H. Mar. Tech. SOC.J. 1993,26,45-53.
6. Newman, J.B.; Robison, B.H. Mar. Tech. SOC.J. 1993,26,45-53.
7. Brewer, P.G.; Orr,F.M., Jr.; Friederich, G.; Kvenvolden, K.A.; Orange, D.L.; McFarlhe, J.;
Kirkwood, W. Geology 1996 submitted.
8. Millero, F.J.; Chen, C.T.; Bradshaw, A.L.; Schleicher, K. Deep-Sea Res. 1980,27,255-264.
9. Brooks, J.M.; Field, M.E.; Kennicutt, M.C. Mar. Geol. 1991,96, 103-109.
10. Golomb, D. Energy Convers. Manage. 1993,34,967- 976.
11. Sakai, H.; Gamo, T.; Kim, Es.; Tsutumi, T.; Tanaka, T.; Ishibashi, J.; Wakita, H.; Yamano, M.;
Oomori, T. Science 1990,248, 1093-1096.
12. Peng, D.Y.; Robinson,D.B. Ind. Eng. Chem. Fundam.1976, 15,59-64.
ACKNOWLEDGEMENTS
We wish to thank the Captain and crew of the RV Point Lobos, and the pilots of the'ROV Ventana
for their outstanding work in making these experiments possible. Support was provided by the David
and Lucile Packard Foundation, and by Stanford University ( F.M.O.) and the United States
Geological Survey (K.A.K.)
478
i .
ton
A",
Pre
Relief
Val".
1. Line diagram of the experimental apparatus used for hydrate generation from the ROV Ventana; the
various pieces are not to scale.
2. Image of methane hydrate formed at the upper gadwater interface. The hydrate rind on bubble
surfaces is plainly seen. The digital information on the screen gives (top, upper left) depth, and date
and time ( lower right). The reaction cylinders are 4.5 cm. diameter.
1'
3. Image of both methane (right, white) and carbon dioxide (left, gray) hydrates approximately 3
weeks after initial formation in experimental apparatus left on the sea floor. The granular appearance
of the methane hydrate is retained; the less rounded blobs of liquid carbon dioxide have a thin veneer
of hydrate that apparently prevents surrounding sea water from further reaction,
419
'
INTRODUCTION
During last fifteen years the authors have been studying the generation and accumulation of
submarine gas hydrates. In particular. expeditions have been carried out in the Caspian, Black,
and Okhotsk seas (Ginsburg et al., 1990, 1992, 1993). and the Norwegian Sea (1996,
unpublished) The results of our investigations have been summarized in a monograph
(Ginsburg and Soloviev, 1994). That is the basis of this presentation.
RESULTS AND DISCUSSION
The analysis of the worldwide observational data suggests that submarine hydrates largely
occur in local accumulations (Ginsburg and Soloviev, 1994, 1995). All observed submarine
gas hydrates are readily divisible into two groups: associated and non-associated with fluid
vents. Hydrates of the first group, which have been observed close to the sea floor in the
Caspian, Black, Okhotsk and Norwegian seas, the Gulf of Mexico, and in several other sites
(altogether in 11 regions, Fig.1) are controlled by fluid conduits: mud volcanoes, diapirs, and
faults. As for the second group of gas-hydrates (deep-seated), their control by fluid flow may
be usually deduced from an association with indirect borehole indications of fluid flows, such
as relatively coarse-grained sediments and anomalies of pore water chlorinity (Figs.2, 3).
The generation, accumulation and disappearance of any water-soluble naturally occurring
compound in terms of water availability are governed by solubility variations of this compound.
This is true also in regard to gas hydrates. It is extremely important for natural gas hydrate
formation that the solubility of methane (which is the major component of natural hydrates) in
water in terms of hydrate stability is little affected by the general (hydrostatic) pressure
(contrary to "normal" conditions of hydrate instability) but is dictated essentially by the
equilibrium pressure of hydrate formation, which is temperature-dependent. Since the
equilibrium pressure of hydrate stability is diminished with decreasing temperature, methane
solubility in water also decreases (Fig.4, solid line). Because of this, the solubility of methane
in pore waters generally decreases towards the sea floor within the submarine gas hydrate
stability zone (Fig.5). The higher the geothermal gradient, accordingly the thinner the hydrate
stability zone, the sharper is the methane solubility decrease.
Three major mechanisms of methane transport in sediments can be distinguished: dissolved in
pore water flows, as free gas flows, and molecular diffusion. Hydrate precipitation from
ascending methane-saturatedwater is thought to be the most straightforward (Ginsburg, 1990;
Ginsburg and Soloviev, 1994). The hydrate zone forms a gas-geochemical barrier for
methane-saturated water which rises either from below or from within this zone: as the water
cools it should precipitate hydrate. The amount of precipitated hydrate obviously corresponds
to the excess of dissolved methane (Le., over the solubility). Clearly the effectiveness of this
process depends, in particular, on the rate of water flow and the water temperature; in the
case of focused flow of warm water, the thickness of the submarine hydrate zone can
decrease to zero. Gas hydrates being precipitated from infiltrated waters are progressively
filling the sediment pore space and/or fracture porosity and eventually cement them, producing
massive and vein hydrate sediment structures.
Gas hydrates associated with free gas flows discharging on the sea floor were observed in the
Gulf of Mexico (Brooks et al., 1994) and in the Okhotsk Sea (Ginsburg et al., 1993). Clearly,
the gas seeping through the hydrate stability zone has no time to crystallize as a hydrate. After
a hydrate film forms at the gas-water interface, each succeeding portion of free gas, prior to
hydration, has to penetrate this film. Thus the rate of hydrate formation in the vicinity of free
gas flows is limited by the rate of this penetration (i.e.. the rate of molecular diffusion), and
hydrates are accumulated primarily from the water-dissolved gas: a solid (hydrate) phase
grows at a distance from free gas. The lateral outward diffusion of methane of the ascending
gas flow appears to be governed by the difference between chemical potentials of gaseous
and dissolved methane at common depths. The above difference is deduced from the
difference between the pressure of a free methane close to the hydrostatic pressure and the
vapour pressure of dissolved methane, which in terms of pore water saturation should be close
to the equilibrium pressure of gas hydrate formation (compare Phand P
, on Fig.6). Since this
difference decreases with increasing subbottom depth, hydrate accumulations associated with
ascending free gas flows are assumed to taper off downward. Accumulations of this type at
great water depths should be more extensive than shallow ones (other factors being equal)
because the considered difference increases with deepening water. It is self-evident that this
model simplifies the matter. In fact, the heat release caused by hydrate formation enhances
the outward methane transport and extends the diffusion aureole around ascending gas flow.
Within this aureole the hydrates are thought to result not only from outward diffusing methane
480
but also from upward diffusion, the intensity of which is controlled by high gradients of
Concentration and vapor pressure of water-dissolved methane in the hydrate zone (in terms of
methane-saturated water); these gradients greatly exceed values outside the hydrate zone
(Figs.5 and 6).
A similar pattern of methane diffusion and gas hydrate accumulation should also characterize
the vicinity of ascending flows of gas-saturated water. In particular this is possible around the
water flows which are too warm for hydrate precipitation. High gradient of temperature nearby
these flows provides favorable conditions for rapid gas hydrate accumulation.
It is generally believed that diffusion plays only a destructive role in the history of hydrocarbon
accumulations. In contrast, Egorov (1988) has put foward the concept of "directional diffusion
recondensation". This implies the diffusional transfer of hydrocarbons which saturate Water in
the presence of a temperature-controlled solubility gradient. According to this concept, the
formation and accumulation of a hydrocarbon phase in the region of lower temperatUre results
from such a transfer. We suggest that directional diffusion recondensation is just the process
which governs gas hydrate accumulation in the vicinity of free gas and gas-saturated Water
flows, as well as within and above the sediment sections where biochemical methane is
intensively generated. Relatively impervious sediments may act as a cap in this process.
DSDP-ODP data offer examples of gas hydrate occurrences close to the boundary between
relatively coarse- and fine-grained sediments (Ginsburg and Soloviev, 1994).
Thus, gas hydrates accumulate from water solutions, no matter whether methane is delivered
into the reaction zone, by infiltration or diffusion. The important distinction between two modes
of hydrate accumulation in sediments (aside from the process rate) lies in the source of
hydrate water. In the case of hydrate precipitation from infiltrated gas-saturated water this
source is flow itself; in the case of diffusional methane delivery the hydrate water is extracted
from sediment pore water in-situ.
We have proposed the term segregation to designate the mechanism of hydrate accumulation
from diffusing gas and from water being extracted from sediments (Ginsburg and Soloviev.
1994). A continuous delivery of methane and the associated formation of hydrate generates a
migration of pure water into the reaction zone from the adjacent sediments or sea water. This
mechanism of water migration is thought to be diffusion-osmotic. Hydrate inclusions of a
different shape are formed during this process due to the dewatering of surrounding sediments
if the latter are compacted. The shape of inclusions is obviously caused by the factors
controlling the fields of gas and water chemical potentials. In particular the subhorizontal
lenticular-bedded hydrate sediment structure observed in association with submarine gas
vents in the Okhotsk Sea (Ginsburg et al., 1993) may result from the subhorizontal extension
of isotherms.
As a result of water redistribution during segregational gas hydrate accumulation, the total
water content of hydrate-bearing sediments may turn out to be higher than that of the adjacent
nonhydrated ones, as has been observed in the Okhotsk Sea (Ginsburg et al., 1993). A water
content of sediments directly proportional to their hydrate content has been demonstrated in
the Caspian Sea (Ginsburg et al., 1992). Hence the hydrate accumulation in sediments may
imply not only gathering of gas but also of water. Due to hydrate water abundance, a sediment
may become fluidized upon decomposition of hydrate.
We mentioned two kinds of inhomogeneity of the geological medium exerting influence upon
gas hydrate accumulation: permeability variations, which control fluid conduits and gas hydrate
caps, and geothermal inhomogeneity (geothermal gradient), which predominantly governs gas
solubility in water. In addition two other kinds of inhomogeneity - hydrochemical and lithological
can have a pronounced effect on this process. It is well-known that water-dissolved salts inhibit
(prevent) gas hydrate formation, i.e. hydrates form more readily from fresh water. Therefore, a
gradient of water salinity within the hydrate zone under gas-saturation conditions must provoke
a diffusional flux of methane into fresh water, where this amving methane should be hydrated.
Such a situation may occur near boundaries of water flows. It is necessary to emphasize here
that the solubility of methane in the fresh gas-saturated water is known to be higher than in
saline water, whereas the corresponding methane fugacity, which actually should be
considered as a driving force of diffusion, is higher in saline water (Handa,1990).
therefore lowers the pore water chemical potential. As a result, a higher thermodynamic
concentration of methane is required for the formation of hydrate. In principle, this effect is
similar to the influence of salts dissolved in water. This surface effect was studied by many
authors and had been found negligible in terms of natural sediment water content. The kinetic
effect lies in the fact that a pore size may be less than a gas hydrate critical nucleus size at a
given temperature. In this case, for hydrate formation to start, more significant overcooling or
oversaturation is required (Chersky and Mikhailov. 1990). We suggest that the essence of both
effects (thermodynamic and kinetic) can be understood by examination of hydrate formation in
adjacent sediments having different pore sizes. It is evident that the hydrate formation in
coarse-pored sediments has an advantage over fine-pored ones - the same gas concentration
in water may turn out to be sufficient to form hydrates in the former case and insufficient in the
latter. What this means is hydrate can accumulate rather in relatively large pores in the course
of sediment compaction and/or biochemical gas generation.
CONCLUSIONS
Submarine gas hydrates mostly occur locally and are linked to fluid flows. They accumulate
from methane-saturated water, in the course of pore water infiltration and methane diffusion.
Apart from the methane availability the accumulation of hydrates is controlled by physical
factors such as temperature gradient, pore water salinity gradient and lithological variability.
The hydrates precipitate at lower temperatures and from less saline water; relatively coarsegrained sediments make better hydrate reservoirs than fine-grained sediments.
REFERENCES
Brooks, J.M., Anderson, A.L., Sassen, R.. MacDonald, I.R., Kennicutt II,M.C. and Guinasso.
slope
N.L.. Jr., 1994. Hydrate occurrences in shallow subsurface cores from continental
sediments. In: E.D. Sloan, Jr., J. Happel and M.A. Hnatow (Eds.). Int. Conf. on Natural Gas
Hydrates. Annals of the New York Acad. Sci., 715: 381-391.
Cherskiy, N.V. and Mikhailov, N.E., 1990. Size of equilibrium critical nuclei of gas hydrates.
Doklady Akademii Nauk SSSR, 312(4): 968-971 (in Russian).
Egorov, A.V., 1988. Diffusional Mechanisms of Hydrocarbons Primary Migration and
Accumulation in Offshore Sedimentary Basins, Thesis. lnstitut Okeanologii Akademii Nauk
SSSR, Moscow, 218pp. (in Russian).
Gieskes, J.M., Johnston, K., Boehm, M., 1985. Appendix. Interstitial water studies. Leg 66.In:
von Huene, R., Aubouin. J. et al. Init. Repts. DSDP, 84: Washington, D.C.: 961-967.
Ginsburg, G.D., 1990. Submarine gas hydrate formation from seeping gas-saturated
underground waters. Doklady Akademii Nauk SSSR, 313(2): 410-412 (in Russian).
Ginsburg, G.D. and Soloviev, V.A., 1994. Submarine Gas Hydrates. VNllOkeangeologia,
St.Petersburg, 199 pp. (in Russian, with English abstract).
Ginsburg. G.D. and Soloviev, V.A., 1995. Submarine gas hydrate estimation: theoretical and
empirical approaches. Proc. 27th Annu. OTC., Houston, Texas, USA, 1-4 May 1995: 513-518.
Ginsburg, G.D.. Kremlev, A.N., Grigofev, M.N., Larkin. G.V., Pavlenkin, A.D. and Saltykova,
N.A., 1990. Filtrogenic gas hydrates in the Black Sea (twenty-first voyage of the research
vessel "Evpatoriya"). Soviet Geology ahd Geophysics (Geologia i Geofizika), 31(3): 8-16,
Ginsburg, G.D.. Guseynov, R.A., Dadashev, G.A.. Ivanova, G.A., Kazantsev. S.A., Solov'yev,
V.A., Telepnev, E.V., Askeri-Nasirov, R.Ye., Yesikov,A.D., Mal'tseva, V.I.. Mashirov, Yu.G.
and Shabayeva, I.Yu., 1992. Gas hydrates of the Southern Caspian. Int. Geol. Rev,, M(8):
765782.
Ginsburg, G.D., Soloviev, V.A., Cranston, R.E., Lorenson, T.D., Kvenvolden. K.A., 1993. Gas
hydrates from continental slope offshore from Sakhalin Island. Okhotsk Sea. Geo-Marine
Letters, 13: 41-48.
Handa, Y.P., 1990. Effect of hydrostatic pressure and salinity on the stability of gas hydrates.
Journ. Phys. Chem., 94(6): 2652-2657.
Makogon, Yu.F. and Davidson, D.W., 1963. Influence of excessive pressure on methane
hydrate stability. Gazovaya promyshlennost',4: 37-40 (in Russian).
Namiot. A.Yu., 1991. Solubility of Gases in Water. Reference Textbook. Nedra, Moscow, 167
pp. (in Russian).
von Huene, R., Aubouin, J. et al.. 1985. Init. Repts. DSDP, 84: Washington, D.C.
WaWns, J.S., Moore, J.C. et al., 1981. Init. Repts. DSDP, 66: Washington, D.C.
482
FIGURES
ARCTIC
1 Fig.1.
OCEAN
Worldwide locations of
observed submarine gas
hydrates. Updated after
Ginsburg and Soloviev,
1994. 1, 2 - sea floor
seepage-associated and
non-associated
gas
hydrates, respectively.
Y-
'.."_-.___
.a,...
>
483
S,crn3/g
'
-,lo
MPa
----1
5MPa
20 t , O C
10
'
10
20
30
40
P.NlPa
H.'km
484
Fig.6.
Relationship
between
different kinds of pressure (P)
affecting diffusion of methane in
subbottom conditions. H is total
depth = water depth + subbottom
Ph is
conventional
depth.
hydrostatic pressure. Pq is
equilibrium pressure of methane
hydrate; curves 1-4 relate to water
depths 1, 2, 3,4 km, respectively.
P. is saturation pressure of
dissolved methane within sulfate
reduction
zone.
Accepted
assumptions: water is pure, gas is
pure methane (see also Fig.5).
The Pq curves are the usual PT
gas hydrate equilibrium curves but
the temperature axis is replaced
by the depth axis based on the
accepted assumptions.
,-
It should be noted that, the hydrate forming gas is assumed to be pure methane to
simplify the argument. Obviously for multi-component systems, upon hydrate
formation, some compositional variation will occur which will result in changes in
equilibrium conditions. Also, any effect due to capillary forces and the type of
rock has been ignored in this work.
THERMODYNAMICMODEL
The fugacity of each component in all fluid phases, including the salt free waterrich phase, have been calculated by an equation of state (EoS). The saline water
phase has been modelled by combining the EoS with the modified Debye-Huckel
electrostatic term, using only one interaction parameter. In optimising the watersalt interaction parameters, water vapour pressure depression data at 373.15 K and
freezing point depression data have been used. For the gas-salt interaction
coefficient, gas solubility data in single electrolyte solutions at different
temperatures and salt concentrations have been used. The model has been
extended to mixed electrolyte solutions with nine salts in its library. A detailed
485
description of modelling vapour, liquid hydrocarbon, salt free water phase, hydrate
phases, ice phase, and saline water phase is given elsewhere[6 '71.
Figure-1 shows methane solubility in distilled water at different temperatures.
There is a good agreement between experimental[8] data and predictions which
demonstrates the success of the EoS in representing highly polar systems.
Methane solubility in distilled water and in 1 and 4 molar NaCl solutions are
presented in Figure-2. The agreement between experimental[gl data and
predictions is very good, which indicates the reliability of the thermodynamic
model. (There are some deviations for pure water at higher pressures, which could
be due to the inaccuracy of experimental data, as Figure-1 shows better agreement
for the case of pure water).
In the presence of a free gas phase and under the above conditions (constant
pressure, temperature, and composition), hydrate formation would cease only when
one of the phases (Le., water or the free gas phase) disappears. However, in the
presence of sea water the increase in the concentration of salts (due to hydrate
formation) could inhibit the further formation of hydrates, as discussed later.
The effect of salt@)on the equilibrium concentration of methane in the water-rich
phase is presented in E'igure-5. As shown, the presence of salt(s) wilI reduce the
methane solubility, and inhibit hydrate formation, as the hydrate-water-methane
point for 3.5 Wt% NaCl solution is at a lower temperature compared to pure water.
Figure-6a shows the mole% hydrates formed from dissolved gas in 3.5 Wt% NaCl
solution at 280 K. The x-axis is pressure (or depth). As an example, at 20 MPa
(1993 m depth), the dissolved gas in the metastable condition is enough to form
1% hydrates. However, the amount of hydrates could be as high as 2.2% for 60
MPa pressure (5980 m depth).
As shown in Figure-6b, the 1% hydrates will increase the salt concentration by
0.03 Wt%, Le., the salt concentration will increase from 3.5 to 3.53 Wt%.
However, the effect on phase equilibria conditions is insignificant (an inhibition of
0.013 K). This means that for further hydrate formation, the controlling factor is
486
Q
'A
J
most likely to be the supply of gas. As the gas concentration in the hydrate
stability zone is significantly lower, the gas could be transported by diffusion.
Convection is also another means of supplying gas to the hydrate suability zone.
Based on the above results the following two mechanisms are proposed for hydrate
formation in subsea sediments:
- Gas released from biogenic and thermogenic sources are dissolved in sea water.
- The dissolved gas reaches the hydrate stability zone by diffusion, or the water
containing the dissolved gas reaches the hydrate stability zone by
convection/advection,
- Hydrate formation initiates at a certain degree of subcooling and the
concentration of gas in the water-rich phase is reduced, as shown in Figure-4.
Therefore, the gas concentration (in the water-rich phase) outside the hydrate
stability zone would be higher than that inside the hydrate stability zone.
More gas is provided to the hydrate suability zone by diffusion (due to the
concentration gradient) for further hydrate formation.
I
I
REFERENCES
1.
Sloan, E. D., Clathrate Hydrates of Natural Gases, Marcel Dekker Inc., New York,
(1990).
Ripmkester, J.A., Tse, I S , Ratcliffe, C.I., and Powell, B.M., A New Clathrate
2.
Hydrate Structure, Nature, Vol. 325, No. 135,, pp. 135-136, (1987).
Miller, S.L., The Nature and Occurrence of Clathrate Hydrates, Natural gases in
3.
Marine Sediments, Ed. Kaplan, I.R., Plenum Press, New York, (1974).
Makogon, Y., Gas Hydrate Formation in Porous Media, Proceed. of the 2nd
4.
International Conference on Natural Gas Hydrates, pp. 275-289, (1996).
Brown, K.M., Bangs, N.L., Froelich, Kvenolden, K.A., The Nature, Distribution,
5.
and Origin of Gas Hydrate in the Chile Triple Junction Region, Earth and
Planetary Science Letter, Vol. 139, pp. 471-483, (1996).
6.
Tohidi, B., Danesh, A., and Todd, A.C., ModellingSingle and Mixed Electrolyte
Solutions and its Applications to Gas Hydrates, Chem. Eng. Res. and Des., Vol.
73 (May), Part A, pp. 464-472, (1995).
7.
Tohidi, B., Danesh, A., Burgass, R.W., and Todd, A.C., Gas Solubility in Saline
Water and Its Effect on Hydrate Equilibria, Proceedings of the 5th International
Offshore and Polar Engineering Conference (ISOPE-95), Vol. 1, pp. 263-268,
(1995).
8.
Culberson, O.L., and McKetta, J.J., Petrol. Trans. AIME, Vol. 192, P. 223, (1951).
OSullivan. T.D.. and Smith. N.O.. The Solubilitv and
Partial Molar Vnlurne
9.
~-. _______of
_Nitrogen and Methane in Water and in Aqueous SodiumChloride from 50 to 125 and
100 to 600 Atm, J Phys Chem, Vol74, No 7, pp. 1460-1466, (1970).
~
487
350
100
50
0'
0.00
0.05
0.10
0.15 0.20
0.25
0.30
0.40
0.35
Methane, mole%
Figure-1 Experimental and predicted methane solubilities in distilled water.
4 Molar
700
1 Molar
Distilled water
600 E 500 3
@ 400 3
g 300 200 -
v)
l i
100
\ater-hpthane-l&drate
Later-meihane
40 MPa
'
60 MPa
50 MPa
c 4
8
30 MPa
.
r
20 MPa
2
10 MPa
0
270
5 MPa
2.5 MPa
1 MPa
330
350
370
390
Temperature, K
Figure-3 Predicted methane concentrations in the water-rich phases (Salt free) of
water-methane, water-methane-hydrate, and water-hydrate equilibria.
290
310
488
>
---___
270
285
290
295
300
305
Temperature, K
Figure-4 Predicted methane concentrations in the water-rich phases (3.5 Wt%
NaCl) of water-methane,water-methane-hydrate,and water-hydrate equilibria.
"I
275
280
water-hydrate
0
270
Pressure=60 MPa
I
310
330
350
Temperature, K
Figure-5 Effect of water salinity on methane concentration.
290
-5
zs
v)
cu0
390
Temperature=280 K
f 2.0
370
2.5
water-methane
'
'0
>
r
v) 1.5
al
-0
1.0
0.5
3
>
n>
Pressure, MPa
Figure-6 Percent hydrates formed from the dissolved methane and the increase in
salt concentration.
7
489
Apparatus. It is well known that the specific equipment used in the study of hydrate formation kinetics
has significant influence on the experimental results. There are basically two types o f equipment, fixed
boundary type and turbulent boundary type. The former is more suitable for simulating the hydrate
formationldissociation in situ, and the latter is closer to the conditions in the transportationpipelines and
natural gas processing equipment. The apparatus used in this work belongs to the latter type. The
schematic diagram of the experimental system is shown in Fig. 1, and the major parts are briefly
described as follows:
Transmrent samhire cell: The 2.5 cm i.d. sapphire cell was purchased from the DB Robinson Design
& Manufacturing Ltd. (Canada), the total volume and the effective volume (excluding the piston and
stirrer volume) are 78 and 59 mL, respectively. The working volume o f the cell can be adjusted by a
floating piston driven by a positive displacement pump. The maximum working pressure and
temperature are 20 MPa and 423 K, respectively.
Air bath The air bath was manufactured by Shanghai Instruments Corp., the working temperature
range i s 263 373 K and can be controlled to within f0.2 K by a digital programmable temperature
controller.
Acifation svstem: The agitation system consists o f a magnetic stirrer coupled with a permanent
magnet mounted outside o f the cell. A variable speed DC motor equipped with an rpm-controller
provides up and down reciprocating motion o f the magnet.
Pressure meusurement: The pressure in the cell was measured through pressure transducer and
pressure gauge simultaneously. A differential pressure transducer (Honeywell Inc.) was connected with
the data acquisition system. The precision of the DP transducer at the working span (0
10 MPa) is
fO.l %. A 0 25 MPa Heise pressure gauge was also installed for taking parallel pressure
readings. The pressure measurement system was calibrated against a Ruska standard dead-weight gauge,
and the precision of the pressure measurements is estimated at f0.015 MPa.
Experimental Procedure. The kinetics o f hydrate formation can be studied in two modes: the constant
temperature-constant pressure mode and the constant temperature-constant volume mode. In the former
mode, to maintain constant system pressure, the hydrate former gas consumed in the hydrate formation
process is continuously supplemented from outside. In the latter mode, the system is closed, with its
volume kept constant, and the system pressure is lowered gradually in the hydrate formation process.
The latter mode w-as applied in this study.
Prior to performing the experiment, the floating piston was lowered to the bottom o f the sapphire cell and
its position was unchanged during the measuring process. About 12 m L liquid sample was charged into
the evacuated sapphire cell. When the system temperature stabilized at the preset value, methane was
introduced into the cell until the pressure was raised to about 4.0 MPa. The gas was then discharged to
eliminate the trace amounts of residual air in the cell. Methane was again charged until the preset initial
490
system pressure was attained, then the DC motor to actuate the magnetic stirrer was started with the
s:irrer was moving up and down at a rate of four strokes per minute. The system temperature and the
change of system pressure were recorded through the data acquisition system every 30 seconds, and
displayed on the monitor screen.
Experiments Performed. The systems studied and the corresponding operating conditions are
summarized in Table 1. A total of 30 sets of kinetic data were measured for the following systems:
methane t water, methane +water + salt(s), methane + water + EG, and methane + water + salt + EG.
Experimental Results. A typical pressure vs. time ( P t) curve measured for the methane hydrate
formation process is shown in Fig. 2. The curve can be roughly divided into three zones. The first zone
(from to to ts) is called the gas dissolution zone, Ps stands for the system pressure when saturation of
the dissolved gas is established. The second zone (from Is to tr) is called the nuclearion zone, system
pressure remains nearly at constant in this zone. The time interval from Io to tr is the so called induction
period. The third zone, from rr to id, is called the crystalgrowth zone, i n this zone the system pressure
falls gradually from Pr to Pd and remains stabilized after time td. The three zones are divided rather
arbitrarily, as in fact, nucleation could proceed simultaneously with the gas dissolution process. The
relative time distribution of the three zones in the 30 experiments performed are also listed in Table I .
The detailed P t data for typical experiments are given in Table 3 and Figs. 3 5 along with the
calculated results which are discussed below.
Analysis ofthe Experimental Results. From Table I , it can be seen that the time interval of the gas
dissolution period is, in general, I 2 hours, however, the time interval of nucleation period differs
appreciably for the experiments performed, from 25 minutes (E04, E08, and E25) to 5 hours (E16),
and for some experiments (EO1 and E20) no crystal nucleus was formed even after IO hours. Since
during the nucleation period, the liquid phase is in the metastable stole, the nucleation process is
sensitive to very small perturbations to the system. This caused difficulty in obtaining repeatable results
even when the experiments were run under identical temperature and initial pressure conditions (E09a
E09c). The time period for crystal growth also differed significantly for experiments run under different
operating conditions, from 80 minutes (E04) to more than 5 hours (E28).
Under the same operating temperatures, the initial pressure has little effect on the time interval of gas
dissolution period (EO1 E04 and E05 E09), however, its influence on the nucleation period is
significant. In general, the lower the initial pressure, the longer of the nucleation period. Similar initial
pressure effect was observed in the crystal growth period (E02 E04 and E07 E09).
The temperature effect on the time interval of gas dissolution and nucleation periods (under same initial
pressure) is, in general, thebigher the temperature, the longer the time period (E10 and E12, E08 and
El I). The effect increases with the lowering of the initial pressure. Significant temperature effect was
also observed in the crystal growth period; the time interval increase almost linearly with the increase of
temperature, however, the temperature effect seemed not as sensitive to the initial pressure in this period.
The effects of inhibitors (saWethylene glycol) on the hydrate formation process are quite complex. When
the concentration of the inhibitor is less than I .O mass%, the effect of concentration is not obvious on the
time distribution ofthe three periods (El3 and E14, E17 and E18, E21 and E22).
For concentrations greater than 1 .O mass%, the time interval of gas dissolution period is little effected by
the inhibitor concentration, however, the concentration has significant effect on the time interval of
nucleation period, the higher the concentration the longer the time interval (E15 and E16, E l 9 and E20,
E23 an E24, E26 and E28). The order of inhibition effect is as follows (when concentration of inhibitor >
1 .O mass%): EG > NaHC03 > NaCl> (NaCI + NaHC03) > (NaHC03 + EG) > (NaCI + EG).
I
An interesting phenomenon observed in the experiments is that when the concentration of inhibitor is
less than 1.0 mass% (E13, E14, E17, E l 8 and E25), the induction time (gas dissolution period +
nucleation period) is significantly shorter as compared with the methane + pure water systems (run under
similar temperature and initial pressure conditions). It is in consistency with the observation of Yousif et
al. (1994), that the hydrate formation could be enhanced at low inhibitor concentration.
1
kl
e CHdaq)
CH4(g)
CHq(aq) + nH2O
(1)
k-I
k2
CH4. nH2O
k-2
491
(2)
Following Long and Sloan (1993). the coordination number n was taken as 20. Since the structure of this
labile cluster is similar to the 512 hydrate cavity (Christiansen and Sloan, 1994). we assume their size is
also similar, i.e. 0.5 nm.
Step 2; The link of clusters to form a crystal unit.
k3
m(CH4. nH20)
e==*
mCH4 r H 2 0 + (mn - r)H20
(3)
k-3
It has been well established that methane forms structure I hydrate, the structure I hydrate crystal unit
and six 51262 crystal cavities. The maximum
cell contains 46 water molecules, and consists o f two
number of methane molecules per unit cell is 8. Assume the crystal unit in Eq. (3), mCH4 . rH20, is an
ideal crystal unit cell (with its cavities fully occupied), thus m = 8, and r = 46. It is also assumed that the
size of the crystal unit cell mCH4 . rH2O is the same as the crystal unit cell o f structure Ihydrate, and is
thus taken as 1.2 nm (van der Waals and Platteeuw, 1959). Since the size o f the crystal unit is smaller
than the critical size, some o f the crystal units could be dissociated back to individual labile molecular
clusters, and the others will be further linked to form a stable crystal nucleus, its size exceeding a certain
critical size.
Step 3: Crystal units linked to form crystal nucleus N,
k4
l(mCH4. rH20)
N
(4)
Englezos et al. (1987) proposed an equation for calculating the critical size o f hydrate crystal nucleus.
Based on the proposed equation, Natarajan et al. (1994) calculated the critical size o f the methane
hydrate crystal nucleus to be approximately 10-30 nm. That means, about 8-25 unit cells with a size of
1.2 nm are required to form a crystal nucleus of the critical size, i.e. I = 8-25. Thus, approximately
64-200 methane molecules and 368-1150 water molecules are required to form a crystal nucleus of
critical size.
Step 4: Crystal nucleus growing to form hydrate crystal H,
kS
p C W 4 +N + qH20
H
(5)
During the crystal nucleus growing period, hydrate crystals H with different sizes could be formed.
Graauw and Rutten (1970) has measured the size distribution o f propane hydrate crystals (structure 11) in
a continuous stirred tank crystalizer, the results showed that the crystal size is within 10 35 pm, the
average being about 20 pm. Bylov and Rasmussen (1996). Monfort and Nzihou (1993) have also studied
the crystal size distribution. Based on the size distribution data available, we can conclude that the size of
the methane hydrate crystal is at least three times in magnitude larger than the size of the critical crystal
nucleus. The magnitude ofp and q in Eq.(5) should be IO5 and 106, respectively,
KINETIC EQUATIONS
For simplifying the derivation o f the rate equations involved in the hydrate formation process, the
following assumptions were made:
( I ) The rate o f concentration change of each component ( ri ) in the reactions shown in Eqs. (I)to ( 5 )
can be expressed in the following polynomial form,
dC;
, e,P '.'
r i = - - = e . a
(6)
df
where C; and C. represent the concentration (mol/L) of components i and j , a and p denote the order of
concentrationciange.
(2) The order o f concentrationchange is unity for all components ( ~ ~ p l . 0 ) .
(3) The water content in the aqueous phase is constant during the hydrate formation process.
(4) The volume of gas phase and liquid phase remain unchanged during the hydrate formation
process.
Based on the above assumptions, the following rate equations can be derived:
dCG
-= - k j C G
+ k.jcA
(7)
di
dCA
__
- k2cA
k - 2 C ~- mk3Cg
dr
dCD
_- k3cB - k-3CD dt
dCN
__
+ k-2CB
+ mkjCg
/k&D
dt
(8)
(9)
(10)
- k4CD - kscAcN
(11)
df
dCH
-_
- pkjcAcN
- WACN
(12)
492
where Cc; stands for the apparent mole concentration of methane in the gas phase (mole of methane in
gas phase per liter of liquid phase), CA, CB, CD, CN and CH denote the concentrations (mol/L) of
CHq(aq), CH4 .nH20, mCH4 . rH20, N and H. respectively. Based on assumption (3), the concentration
of water in the liquid phase does not appear in the rate equations. At initial conditions: I =O, CG = C@,
CA = CB = CD = C N = C H = 0, from mass balance of methane we have:
h)cH
(13)
(Jl+
Since CG, CA, CB, CD, C N and CH are restrained by Eq. (13), only five of the above six concentration
variables are independent. The concentration of the metastable molecular cluster CB was chosen as a
dependent variable. From Eq. (13) we have:
CB = c@- CG - CA - mCD - h C N - (p + h ) c H
(14)
Eq. (9) can then be removed from the rate equation set. Substituting Eq. (14) into Eqs. (8) and (IO)
yields:
dCA
k.2C@ + ( k l - k - 2 ) C ~- (k.1 + k2 + k . 2 ) C ~- mk.2 CD dt
(15)
I m k - 2 C ~ -(p + / m ) k . 2 C ~ - p k ~ C ~ c ~
dCD
_ - - k3(C@ - CG) - k 3 C ~- (mk3 + k.3 + lk4 )CD - l m k j c ~di
@ +Im)k3C~
(16)
The initial conditions are changed to: 1 = 0, CG = C @ , CA = CD = CN = CH = 0.
__ -
Eqs. (7), (II),(12), (15) and (16) coupled with the corresponding initial conditions constitute the
mathematical model of the kinetic behavior of methane hydrate formation in pure water.
THE LEAST SQUARE ESTIMATION OF THE KINETIC PARAMETERS
In the rate equations established in the previous section, there are eight unknown parameters: k l , k.1, k2,
k.2, k3, k.3, k4 and k5. As kl and k.1 are restrained by the following expression of equilibrium constant
Kc (derivation is referred to the expanded manuscript):
ki
CwOZRTVl
Kc=-=
(17)
k-l
, ,w g
where CwO, Zand Hdenote the initial water concentration, compressibility of methane and Henry's
constant of methane, respectively. k.1 can be calculated through kl as follows:
kl
klHVg
k.1 = - = ___
(18)
Kc
CwOZRTV/
Thus, only seven unknown parameters ( k l , k2, k.2, k3, k-3, 4 and k5) in the kinetic equations needed to
be determined.
The damped nonlinear least square method was used for parameter estimation, the details of the
algorithm are also given in the expanded manuscript (which is available on request). The regressed
parameter values for the methane + water systems are tabulated in Table 2, and a typical comparison
between experimental and calculated P - I data for Experiment E09b is shown in Fig. 3.
1
Experiments E05 E09 were also run at the same temperature (274.15 K) and different initial pressures.
The P I data of Experiments E08, E09b and E09c were predicted by using the parameter values
determined from E07. In the gas dissolution zone, the measured and calculated gas phase pressures are
close, the maximum relative deviation is 0.38%. In the nucleation zone, the maximum deviations B T ~
-0.059% for E08, and 0.54% for E09b and E09c. In the crystal growth zone, the maximum deviations for
EOS, E09b and E09c are 0.78%, 1.01% and 0.52%, respectively.
The test results indicate that although the kinetic data of hydrate formation depend on the initial pressure,
the model parameters determined from a specific run are capable of predicting the P I data of runs
carried out at different initial pressures (under same temperature) with good accuracy.
493
Predicrion of the P r Data for Salt/EG Containing Systems. The prediction of the kinetic data of
methane hydrate formation in brines and aqueous solution of EG are of particular interest in real
production processes and has not been previously reported. It is well known that the presence of salt(s)
and alcohol in the aqueous phase can inhibit the hydrate formation (similar to the freezing point
depression), as the solubility of methane will be significantly lowered, and the physical properties
(viscosity, density, diffusivity, interfacial tension, etc.) of the aqueous phase will in turn be significantly
changed. As a preliminary attempt to extend the proposed kinetic model to the salttethylene glycol
containing systems, we assumed the solubility of methane in the aqueous phase (expressed in terms of
the Henry's constant of methane) is the critical factor affecting the inhibition of methane hydrate
formation. The larger the Henry's constant, the greater the inhibition effect.
Among the eight parameters in the proposed klnetic model, k.1 is the only parameter related to Henry's
constant, hence, the other model parameters determined from methane + pure water system can be
applied directly to the salt/ethylene glycol containing systems. For illustration purposes, the P r data of
methane hydrate formation in 5.0 mass% NaCl solution (Experiment E16) were predicted by using the
kinetic parameters determined from Experiment E09b performed on methane + pure water system (El6
and E09b were run at the same temperature and initial pressure conditions). The Henry's constant of
methane in the 5.0 mass% NaCl solution at 274.15 K was taken from Cramer (1984). H = 3.642 x 103.
The comparison between the predicted and experimental results is presented in Fig. 5. Fairly good
prediction results were observed, the maximum deviations of the predicted gas phase pressure are
-0.13 % in the gas dissolution zone, and 0.3 I % in the nucleation and crystal growth zones.
CONCLUSIONS
( I ) The new kinetic model developed from a four-step hydrate formation mechanism and reaction
kinetics approach is capable of describing the P - t data measured in this work.
(2) Under identical temperature condition, the kinetic parameters determined for a specific initial
pressure can be applied to estimate the P r data run at other initial pressures (within the pressure range
of this study), the maximum deviation is within 0.3%.
(3) The kinetic model developed for methane + water systems can be extended to inhibitor containing
systems by replacing the Henry's constant of methane in corresponding aqueous phase.
(4) As the dynamic behavior of hydrate formation is strongly dependent on the type of equipment and
agitation intensity, the kinetic data measured in this work can only be considered as typical for a mildly
agitated non-flowing system.
ACKNOWLEDGMENT
Financial support received from the China National Science Foundation, Postdoctoral Research
Foundation and the China National Petroleum & Natural Gas Corporation are gratefully acknowledged.
REFERENCES
M. Bylov and P. Rasmussen,'A new technique for measuring gas hydrate kinetics, Proceedings of the
2nd International Conference on Natural Gas Hydrates, June 2-6, 1996, Toulouse, France, 259-266.
R. L. Christiansen and E. D. Sloan, Mechanisms and kinetics of hydrate formation, Annals of the New
York Academy of Sciences, Vol. 715 (1994) 283-305.
S. D. Cramer, Solubility of methane in brines from 0 to 300 "C, Ind. Eng. Chem. Proc. Des. Dev., 23
(1984) 533-538.
1. de Graauw and J. J. Rutten, The mechanism and the rate of hydrate formation, Proceedings 3rd Int.
Symp. on Fresh Water from the Sea, 1970, Athens, 103-1 16.
P. Englezos, N. Kalogerakis, P. D. Dholabhai and P. R. Bishnois, Kinetics of formation of methane and
ethane gas hydrates, Chem. Eng. Sci., 42 (1987) 2647-2658.
P. Englezos, Clathrate hydrates, Ind. Eng. Chem. Res., 32 (1993) 1251-1274.
J. Long and E. D. Sloan, Quantized water clusters around apolar molecules, J. Mol. Simul., 1 1 (1993)
145-161.
Y. F. Makogon, Hydrates of Natural Gas (translated from Russian by W. J. Cieslewicz), PennWell,
Tulsa, 1983.
J. P. Monfort and A. Nzihou, Light scattering kinetics study of cyclopropane hydrate growth J. Crystal
Groyth, 128(1994) 1182-1186.
V. Natarajan, P. R. Bishnoi and N. Kalogerakis, Induction phenomena in gas hydrate nucleation, Chem:
Eng. Sci., 49 (1994) 2075-2087.
J.-H. Qiu and T.-M. Guo, Status ofthe kinetic studies on the hydrate formatioddissociation, J. Ind. &
Eng. Chem. (China), 46 (1996) 741-756.
P. Skovborg, H. J. Ng, P. Rasmussen and U.Mohn, Measurement of induction times for the formation
of methane and ethane hydrates, Chem. Eng. Sci., 48 (1993) 445-453.
E. D. Sloan, Clathrate Hydrates of Natural Gas, Marcel Dekker, New York, 1990.
J. A. van der Waals and J. C. Platteeuw, Clathrate solutions, Adv. Chem. Phys., 2 (1959) 2-57.
M. H. Yousif, R. B. Dorshow and D. B. Young, Testing of hydrate kinetic inhibitors using laser
scattering technique, Annals of the New York Academy of Sciences, Vol. 715 (1994) 330-340.
Y.-X.Zuo , S. Gommesen and T.-M. Guo, Equation of state based hydrate model for natural gas systems
containing brine and polar inhibitor, Chinese J. Chem. Eng. (in English), 4 (1996) 189-202.
494
Table I . Summary ofthe methane hydrate systems studied and the time distribution ofthree zones
Temp.
Aqueous
phase*
Exp. NO.
(K)
Crystal growth
zone (min)
273.65
4.47
IO0
645'
273.65
5.46
IO0
80
150
E03
HZ0
273.65
7.45
IO0
IO0
125
E04
HP
273.65
8.47
95
25
80
E05
HZO
274.15
4.49
IO0
579
E06
HI0
274.15
5.10
IO5
510"
E07
HZ0
274.15
5.46
95
90
205
E08
HZ0
274.15
5.96
95
25
I85
E09a
H P
274.15
6.46
105
195"
E09b
HI0
274.15
6.46
IO5
105
I60
E09c
H2O
274.15
6.47
1OS
32
183
E10
H@
274.65
6.47
90
30
I95
El I
H@
275.15
6.00
130
165
280
HzO
NaCl (0.5)+H,O
276.15
6.47
110
40
390
274.15
6.46
60
38
I60
NaCl(1 .O)+H,O
274.15
6.47
65
30
I85
NaCl(3.0)+H20
273.65
6.47
I IO
150"
NaCl (S.O)+H,O
274. I 5
6.45
I IO
325
I50
NaHCO, (0.5)+H20
274.15
6.45
65
32
168
E18
NaHCO, (1 .O)+H,O
274. I 5
6.46
65
30
200
E19a
NaHCO, (3.0)+H,O
274.15
6.47
115
210
195
E19b
NaHCO, (3.O)+H,O
274.15
6.48
115
200
200
I 1
E20
NaHCO, (5.0)+Hi0
273.65
6.46
I20
615"
E2 I
EG (0.5)+H20
274.15
6.47
I10
340"
li
E22
EG ( I .O)+H,O
274.15
6.46
I15
345"
E23
EG (5.0)+Hz0
273.65
6.47
115
25
185
EG (IO.O)+H,O
273.65
6.47
I20
505'
274.15
6.47
60
25
I85
E24
E25
i
i
E26
E27
E28
E29
E30
NaCl ( O S ) +
NaHCO, (0.5)+H,O
NaCl ( I .5)+
NaHCO, (1.5)+H1O
NaCl(2.5)+
NaHCO, (2.5)+H,O
NaCl(2.5)+
NaHCO, (2.5)+H20
NaCl(2.5)+
EG (2.5)+H20
NaHCO, (2.5) +
EG (2.5)+H,O
274. I 5
6.47
90
35
250
273.65
6.47
100
65
290
274.15
6.47
95
60
320
273.65
6.47
90
55
235
273.65
6.47
90
I12
203
Numbers in parentheses are mass percent of inhibitor; EG stands for ethylene glycol.
# No hydrate crystal formed in this time period.
Table 2. Estimated kinetic parameter
Exp.No.
klx10'
E02
E03
E04
E07
E08
E09b
E09c
El0
Ell
El2
33.99
32.95
19.46
29.05
15.54
14.15
8.539
13.74
21.28
10.22
k2x10'
135.1'
187.3
12.13
0.8026
55.59
0.03878
21.95
10.64
0.9161
94.46
kjx10'
45.37
41.57
28.48
6.543
0.3661
1.189
0.3699
34.31
4.622
3.016
0.3735
0.2669
1.470
6.374
0.01616
139.2
0.03070
1.447
1.792
0.0 2347
495
k.3~10'
k4x10
kj
5.037
8.527
5.386
9.195
3.923
0.2266
4.771
34.34
6.435
4.746
0.4297
0.1839
4.097
10.94
6.708
0.7779
4.390
28.15
8.710
7.368
6.626
5.261
10.93
4.976
0.5876
4.224
0.8824
9.487
3.455
0.9024
0.0
10.0
20.0
30.0
45.0
60.0
75.0
85.0
105.0
125.0
140.0
165 0
190.0
210.0
220.0
230.0
240.0
260.0
280.0
3 10.0
340.0
370.0
6.465
6 452
6441
6 429
6.424
6.421
6412
6.412
6.410
6.406
6.405
6.405
6.405
6.405
6.399
6.398
6.393
6.387
6.38 I
6.375
6.371
6.370
6.453
6.444
6.436
6.426
6.419
6.414
6.4 I I
6.407
6.404
6 402
6.400
6.397
6.395
6.395
6.394
6.393
6391
6.389
6.384
6.378
6.371
-0.018
- 0.040
0.0
15.0
30.0
-0.110
40.0
55.0
+ 0.036
75.0
95.0
1-0.0 I3
I10.0
135.0
- 0.034
- 0.034
'k 0.049
b
0.028
150.0
170.0
200.0
230.0
260.0
290.0
350.0
410.0
445.0
480.0
515.0
540.0
575.0
+ 0.040
+ 0.082
+0.120
+ 0.150
+ 0.065
+ 0.066
+ 0.004
- 0.066
- 0.120
-0.140
- 0.120
- 0.022
6.447
6.430
6.417
6.412
6.405
6.401
6.400
6.399
6.399
6.399
6.398
6.398
6.397
6.397
6.396
6.395
6.394
6.392
6.382
6.376
6.312
6.368
Calculations for E16 were based OH [lie parameter values determined liooi E09b.
U'I'D-resistance
pressure traiisducer
tlieriiiocouple detector
Time
496
6.43 I
6.422
6.4111
6.411
6.410
6.408
6.407
6.406
6.405
6.4US
6.403
6.401
6.399
6.396
6.392
6 378
6 312
6.367
6.361
6.356
6.350
-0.022
-0.078
-0.088
-0.13
-0.14
-0.13
-0 13
-0.1 I
-0.10
-0. I I
-0.081
-0.068
-0.028
-10.003
to. I I
+0.24
+0.3 I
t0.23
e0.23
+0.24
10.zy
6.55
experimental
6.50
4.50
- rn. - _calculated
__
-.. - _
4.35
4.30
4.25
6.30
0.0
100.0
200.0
300.0
400.0
I
500.0
Tme(min.)
600.0
'Corresponding author
Key words: hydrate inhibition, kinetic inhibitor, flowing system, performance
Introduction
The development of offshore mature basins such as the North Sea is increasingly characterized by
marginal reservoirs. Feasible economic development of these reservoirs requires a shift towards total
subsea production systems without fixed or floating production platforms. Unprocessed or minimum
processed reservoir fluids will be transported to a central processing facility or ultimately to shore.
One of the key issues of total subsea production systems is multiphase flow technology with
particular emphasize on gas hydrate control technology.
Subsea transportation of unprocessed or minimum processed well fluids over long distances today
requires the use of large amounts of methanol or glycols for hydrate inhibition. The effect of these
additives is to decrease the water activity to an extent that markedly reduces its ability to participate
in hydrate formation, and thereby in a lowering of the hydrate formation temperature. The amount of
inhibitor necessary to obtain the desired lowering of the hydrate formation temperature is substantial.
usually in the range of 20-40 weight% of the aqueous phase. This has prompted the search for new
types of additives capable of inhibiting hydrate formation at far lower concentrations (1-5).
Statoil perfoms intensive research on hydrates; methods to prevent hydrate problems as well as
studies on the formation and removal of hydrate plugs. This paper focuses on the robustness of a
commercially available additive from T. R. Oil Services (Hytreat 525) with respect to degree of
subcooling, pressure, salinity of the aqueous phase and the impact from having a defoamer or a
corrosion inhibitor in the system. The inhibitor is tested both at continuous flow conditions and at
re-starts after shut-ins. Results from tests on two different condensate systems as well as two crude
oils are summarized.
Experimental
The experiments were carried out in a high pressure loop formed as a wheel. The system is illustrated
in Figure 1. The test wheel was filled with the desired fluid at a specified temperature and pressure,
and then set under rotation. The rotation creates a relative velocity between the pipe wall and the
fluid thus simulating transport through a pipeline.
The high pressure wheel is made from stainless steel with an inner tube diameter of 52.5 mm and, a
wheel diameter of 2.0 m. The volume is 13.4 liters. The wheel includes two high pressure windows
for visual inspection, and one of these is equipped with a video camera.
The flow simulator is placed in a temperature controlled chamber. The temperature is controlled
using a programmable regulator. a heating fan and a refrigeration system. The temperature
development in the chamber as a function of time is preset in the regulator.
The wheel is attached to a motor/gearbox system enabling a variation of the peripheral velocity of
the wheel between 0.3 m / s and 5.0 m/s. A torque sensor is installed as a part of the rotational shaft
enabling torque measurements to be performed during rotation. Pressure and temperature sensors on
the wheel have ranges of 0-250 bara and -10 to +I50 OC respectively. All signals are transferred
through cables and slip rings to a real time PC-based data acquisition system.
The accuracy of the measurements is estimated to be *0.2 Nm for torque, *0.5 bar for pressure,
*O, l0C for fluid temperature in the wheel and *l.O OC for temperature in the chamber.
Experimentalprocedure
1
The wheel is rotated at a constant peripheral velocity (1.0 m / s ) as the temperature is reduced from
about 6OoC to 4OC at a given cooling rate. In the following, these experiments are referred to as
continuous flow experiments. In condensate systems without emulsifiers added, a velocity of I .Om / s
creates separated liquid phases (prior to hydrate formation). The phases are generally mixed at this
velocity when black oil systems are used.
49G
In order to minimize the number of adjustable parameters for the experiments. only one given
cooling rate is used both for the continuous flow experiments and the start-up experiments.
Results and discussion
The hydrate inhibitor tested in this work is a commercially available kinetic inhibitor concisting of a
blend of different polymer/surfactants.
In this specific study the inhibitor was tested in the high-pressure system using two different crude
oils and two gas condensates. Also the presence of a coorosion inhibitor or a defoamer on the
performance was investigated for some of the fluid systems. Results from 27 experiments performed
in the flow simulator with the different hydrocarbon fluids are presented. A summary of the
experimental conditions and observations from these experiments is given in Table 1. It should be
stressed that the subcoolings given in the Table 1 have been comected for the actual salinity of the
systems.The results are discussed in more detail below.
hdensate 4
lmpaci of corrosion inhibitor
Based on the visual information from the experiments the addition of the corrosion inhibitor changed
the physical properties of the system with respect to foaming tendency. No significant change of
performance of the hydrate inhibiting properties was observed. In all the inhibitor experiments there
was a considerable kinetic effect but there was a decrease in transporability at hydrate formation
compared to the pure system.
..r
i,
I.
lmpaci of salinity
The salinity of the aqueous phase in these experiments was varied from 0-3.5 wt%. Increased salinity
resulted in an increased delay of hydrate formation from 80 min to 20 hours. Previous studies (1,3,5)
have shown that the optimal salinity for this inhibitor is approx. 3,5 wt%. The improved performance
is due to conformational changes in the polymer systems in the inhibitor. A salinity of 0,25 wt%
corresponds to the actual salinity of the produced water from the field.
Impaci of subcooling
As seen from Table 1 the subcooling in the experiments are varied from 9'C to 13C. At continuous
flow mode an addition of 0.5 wt% of the hydrate inhibitor prevented hydrate formation at a
subcooling of approximately 9C for the test period of 36 hours when the salinity was 0,25 wt%.
Also restarts after a shut-in of 12 hours was successfullly carried out at this temperature, although
there were hydrates present in the system. However, at subcoolings above 9"C, the hydrate inhibitor
tested was not able to fully prevent the hydrate formation during continuous flow. But in all the
experiments hydrate formation was delayed compared to experiments on the blank Condensate A
fluid.
chlda&B
Impact of subcooling
Experiments were here performed with a inhibitor concentration of 0.5 wt% of the aqueous phase .
As seen from exp. 15, no hydrate formation were observed in this system at a subcooling of 1 1C at
a pressure of 70 bar within an experimental time of 80 hours. When the subcooling was increased to
13'C, hydrate formation was observed after 9 hours. When increasing the pressure to 140 bar
(exp.16) keeping the subcooling constant at 1 IoC, hydrates were formed after 6 hours, and the
flowloop plugged 1 hour after hydrate initiation. In these experiments the aqueous phase contained
3.5 wt% NaCI.
Y
The kinetic inhibitor was tested with and without the presence of a corrosion inhibitor both in
continuous flow experiments and in shut-in experiments. In exp. 17-20 also a defoamer was present
in the system.
During the continuous flow experiments without the defoamer present a significant delay of the
hydrate formation was observed both with and without the prescence of the corrosion inhibitor. At a
subcooling of 11C the presence of the corrosion inhibitor improved the performance compared to
the system with only the hydrate inhibitor present Hydrate formation was not initiated during a
period of 60 hours.
In the start-up experiments, however, the induction time was reduced. Hydrates were formed during
the stagnant period (12 hours), and the wheel was plugged shortly after restart.
The transportability of the hydrates formed was better in the presence of the corrosion inhibitor. This
is opposite to what was observed for condensate A, and it illustrates the importance of fluid effects
when this kind of technology is considered for use.
Impact of subcooling
The subcooling in the experiments was varied fom 7C to 13'C. The chemical was not capable of
fully preventing the hydrate formation in any of the experiments. However, at subcoolings below
10C the hydrates formed were transportable.
lmpact of subcooling
During continuous experiments at a subcooling of 73C and O.Swt% of hydrate inhibitor no hydrate
formation was observed for almost IO hours. The formation rate was then very slow for 2 hours
before it increased rapidly resulting in a viscosity increase of the system. No hydrate plugging was
observed.
In the start-up at a subcooling of 9.S0C, hydrates started to form very slowly after start-up. M e r
approx. 4 hours the formation rate increased drastically and flow problems were observed due to the
high viscosity of the system. However, no "solid" hydrate plug was observed in the experiment.
The inhibitor were tested in 4 different fluids with different composition and physicochemical
properties. The performance of the inhibitor is different for all these fluids. It is not possible to
extrapolate the results from one condensate to another, or one black oil system to another. It is
known tha parameters like aliphatic/aromatic ratio, amount and state of asphaltenes and resins and
also wax content will influence on the performance, and these factors will always vary between
different fluids.
Regarding the influence of flow properties it has for all the systems investigated been shown that
stagnant conditions are more difficult to handle then continuos flow. This aspect should be
investigated hrther and will be adressed more thoroughly in a forthcoming paper.
'
500
References
0 Urdahl, A.Lund, P.Msrk and T.N.Nilsen.(.:hem. Eng Sci., SO (5). 863-870, (1995).
A. Lund, 0 Urdahl and S.S. Kirkhorn, (.?hem. Eng Sci., 51(13), 3449-3458, (1996).
(2)
A. Lund, 0. Urdahl, L.H. Gjertsen, S.S. Kirkhorn and F.H.Fadnes, Proceedings from the
(3)
2nd Interriational Conjerence on Naural Gas Hydrates", p. 407, Toulouse, France, July,
(1996)
L.H. Gjertsen, T. Austvik and 0. Urdahl, Proceedings from the 2ndlnternational
(4)
Conjerence on Nauru/ Gas Hydrates", p. 155, Toulouse, France, July, (1996).
A Lund, 0. Urdahl and S.S. Kirkhorn, J. Pet. Sci. and Tech.. Submitted.
(5)
(1)
Table 1:
Exp.
Sub
501
Comments
Figure 1 :
502
ISP manufactures a full line of hydrate inhibitors including VC-7 13, PVCL, PVP and VCLNP
copolymers. These inhibitors are tested in the high pressure laboratory at realistic pipeline
conditions. We have observed that several glycol ether solvents (for example, 2-butoxyethanol)
significantly enhance the performance of the polymeric hydrate inhibitors. Better inhibitors
provide lower polymer treatment levels and lower overall cost. This paper presents the results of
our experimental study on hydrate inhibitors containing glycol ether solvents.
EXPERIMENTAL
The tests were conducted in a 300 ml stainless steel stirred reactor at high pressure and low
temperature. A diagram of the apparatus is shown in Figure 1. Following the procedure of Long,
et a1: 30 stainless steel balls are placed in the bottom of the reactor to increase nucleation sites.
The reactor is immersed in a refrigerated bath, which normally maintains temperature to within
0.5'F. The pressure in the reactor is controlled to within 5 psi by a programmable syringe pump.
The pump displaces hydraulic oil into a piston cylinder which contains the hydrate-forming gas
on one side and hydraulic oil on the other. The volume of oil displaced by the syringe pump to
maintain constant pressure indicates gas consumption in the reactor.
The inhibitors were tested at 0.5 wt% dry polymer and 0.75 wt% glycol ether in the salt solution.
In a typical experiment, 0.6 g dry polymer and 0.9 g glycol ether liquid were added to 120 g of a
3.5 wt%, filtered, synthetic sea salt solution and mixed for at least one hour. The resulting
solution was transferred to the 300 ml reactor, sealed, and immersed in the temperature bath at
39.2'F (4OC). The pressure was then increased to 1000 psig with green canyon gas and held
constant to within about 5 psi with the syringe pump. After the pressure reached 1000 psig, the
reactor stirrer was turned on to 1000 rpm. The gas volume, as measured by the syringe pump, and
the reactor pressure and temperature were recorded electronically at 1 minute intervals
throughout the experiment.
503
Temp.
w
Temperature bath
Refrigeration
Figure 1: Gas hydrate test apparatus. The 300 ml reactor was charged with 120 g of sea
salt solution containing 0.5% dissolved inhibitor. Tests were conducted at
constant 39.2F and 1000 psig for 20 hours. See text for more detail.
MATERIALS
Gaffix VC-713 is a terpolymer of VCL, W,and DMAEMA. For consistency, all experiments
reported here were conducted with the same manufacturing lot of VC-713.
Butyl Cellosolve, or 2-butoxyehano1, has the formula n-Cd-I90C&bOH. It is an industrial
solvent with a boiling point of 171OC manufactured by Union Carbide. This material and the
other glycol ethers listed in Table 2 were obtained from Aldrich and have a purify of about 99%.
The synthetic sea salt corresponds to ASTM Standard Specification for Substitute Ocean Water
and was purchased from Marine Enterprises of Baltimore, Maryland.
Green canyon gas is a typical natural gas mixture. It has the composition listed in Table 1.
Figure 2 shows the result of adding 0.75 wt% butyl Cellosolve to a mixture of 0.5 wt% VC-713
in sea salt solution. For comparison, the figure also shows the test results for 3.5 wt% sea salt
solution with no inhibitor, 0.5 wt% VC-713 in sea salt solution, and butyl Cellosolve in sea salt
solution. Each test was conducted at 39.2OF and 1000 psig. At these conditions green canyon
gas has an equilibrium dissociation temperature of 64.7F in deionized water, giving a total
subcooling of 25S0F. The gas consumption was calculated from measured volume change with
the real gas law (compressibility factor = 0.83).
504
AS the figure shows, this particular lot of VC-713 inhibits hydrates for only about 40 minutes at
the test Conditions. Adding 0.75 wt% butyl Cellosolve dramatically increases the performance of
the inhibitor, to the extent that no detectable hydrates form for the duration of the 20 hour test.
The figure also shows that butyl Cellosolve does not inhibit hydrates without polymer present.
0.5
.
200
400
600
butyl Cellosolve
1000
800
1200
Time. minutes
Glycol ether
Formula
2-butoxy ethanol
n-CdH9OC2hOH
2-isopropoxy ethanol
CHjCH(CH,)OC2&0H
800
1-propoxy-2-propanol
C,H70CH2CH(CH,)OH
600
2-(2-butoxyethoxy) ethanol
n-C4H90C2H40C2hOH
440
I-butoxy-2-propanol
~-C~H~OCH~CH(CHI)OH
450
2-propoxy ethanol
n-C3H70C2H40H
350
C2H40C2hOH
10
>I200
2-ethoxy ethanol
1-methoxy-2-propanol
CHjOCH2CH(CHj)OH
10
40
Table 2: Induction times for 0.75 wt% glycol ether plus 0.5 wt% VC-713 in 120 g of sea
salt water at 39.2"F and 1000 psig.
Butyl Cellosolve also showed strong synergism with other kinetic inhibitors. Table 3 compares
the induction times for polyvinylcaprolactam homopolymer (PVCL) and 50/50 VCLNP
copolymer with and without butyl Cellosolve. Test conditions were identical to those described
above.
505
Table 4 lists the surface tension o f aqueous solutions of glycol ethers as reported in the
manufacturers literature. The data indicate that the higher homologs are surface active. If this
hydrophobicity of the hydrocarbon chain also causes the chain to associate with the dissolved
polymer, then the glycol ether may allow the polymer conformation to expand in solution. This
could occur if the surfactant breaks the weak bonds between polymer segments which pull the
coils together and tighten the conformation. An extended polymer would presumably have more
of its length available for interaction with the crystal surface, which may account for its improved
performance as a hydrate inhibitor.
Induction times (minutes)
Inhibitor (0.5%)
>1200
Gaftix VC-713
>I200
50150 VCLNP
Table 3: Comparison of induction times for kinetic inhibitors with and without butyl
Cellosolve added, at 39.2OF a n d 1000 psig in sea water solution.
Solvent
methyl Cellosolve
Cellosolve
124.5
90.1
134.9
propyl Cellosolve
104.2
150.1
32.3
butyl Cellosolve
118.2
171.2
28.9
Table 4: Surface tension and other properties of the 2-alkox ethanol homologs. All
data was obtained from manufacturers literature. Y
REFERENCES
1.
Lederhos, J. P.; Long, J. P.; Sum, A.; Christiansen, R. L.; Sloan, E. D., Jr., Effective
Kinetic Inhibitors for Natural Gas Hydrates, Chemical Engineering Science, 51 (8),
2.
Notz, K.;Bumgartner, S. B.; Schaneman, B. D.; Todd, J. L., The Application of Kinetic
Inhibitors to Gas Hydrate Problems, 27th Annual Offshore Technology Conference,
Houston, TX,1-4May 1995,719.
Bloys, B.; Lacey, C.;Lynch, P., Laboratory Testing and Field Trial of a New Kinetic
Hydrate Inhibitor, 27th Annual Offshore Technology Confirence, Houston, TX, 1-4May
(1996) 1221.
3.
1995,691.
4.
5.
Long, J.; Lederhos, J.; Sum, A.; Christiansen, R.; Sloan, E. D., Kinetic Inhibitors of
Natural Gas Hydrates, Proceedings of the Seventy-Third GPA Annual Convention, New
Orleans, LA, 7-9March 1994,85.
Union Carbide, Glycol ,Qhers, (1993).
506
RESULTS - SERIES 1
Before the first series the cell was cleaned in the same way as used by Skyum ( I ) and Andersen (2).
This means rinsing with distilled water and drying using compressed air. The cell was evacuated and
loaded with 20 cm' of distilled water and ethane at 21C. Series 1 was started with a temperature
setting of 3 "C. Figure 2 shows the pressure transient and part of the temperature transient of the
first experiment (M6). The pressure initially drops from 15.3 bar to 14.0 bar due to the reduction in
temperature from 21 "C to 3 "C. The pressure and the temperature then remain constant until the end
of the induction time. As hydrates start to form the pressure drops dramatically due to gas
consumption and there is a small increase in the temperature due to the heat evolved. The course of
events in experiment M6 is representative for all the experiments in series 1 and 2.
A total of 11 experiments is included in series 1 and the results are shown in table 1 and in figure
3. The induction time is here defined as the time that elapses from the cooling flow is started and
until the pressure starts to drop due to hydrate foimation. To calculate the amounts of hydrate formed
within 600 and 3600 seconds respectively after the induction period, the computer program HYLAB
501
has been used. The HYLAB program is essentially based on the gas hydrate kinetics model proposed
by Skovborg (3):
dddt
k
A
cw
x,,,
:
:
:
:
xb
In principle the model requires knowledge of the product k*A to be able to calculate the gas
consumption rate and the pressure drop with time.
When analysing experimental data as here the model is rearranged to allow ameasured pressure drop
to be converted into a gas consumption rate giving as the results the amount of hydrate formed and
the product k*A.
The induction time varies between 38209 seconds and 341430 seconds whereas the growth rates are
much more uniform.
RESULTS - SERIES 2
After discovering that simply keeping the cell closed between experiments thus keeping the
composition constant did not yield reproducible induction times it was decided to try to eliminate
the effect of presence of impurities. The cell was consequently cleaned much more thoroughly using
only double distilled water and drying with dust free paper. Similar conditions to those used in series
1 were established. More than three weeks (-2,000,000 seconds) passed after pressurizing the cell
and no hydrates were formed. The pressure was then increased to 21 bar. Approximately 38500
seconds later hydrates eventually formed. Series 2 comprises this and five more experiments at 3
"C and 21 bar. The results are shown in table 2 and in figure 3. Not surprisingly the growth rates
were higher in series 2 than in series 1. The induction times were on the average lower (table 4) but
still very scattered
RESULTS - SERIES 3
The results in series 1 and 2 showed that even minor amounts of impurities have some effect on the
induction time. Consequently it was decided to add a large amount of impurities to the system to
make it insensitive to addition of further impurities. It was decided that the "composition" of the
added impurities should to some degree match the solid material found in multiphase pipelines. The
impurities consist of equal amounts of CaCO,, BaSO,, rust and asphaltenes. Before addition of water
and gas to the cell, 20 mg impurities was added and conditions similar to those of series 2 were
established. The course of events in the experiments of series 3 differs drastically from that of the
other experiments. In series 3 the pressure initially not only drops due to reduction in temperature
but also because hydrates were formed shortly after the hydrate equilibrium temperature is reached.
The observed induction time in table 3 is the time from the beginning of cooling until some hydrate
particles were seen. In table 3 is also listed the time it takes before the conditions in the cell are
favourable for hydrate formation from an equilibrium model (4) point of view. It can be seen that
hydrates were often observed before the conditions for their formation were favourable. This cannot
be contributed to low accuracy in the gas hydrate model used, but rather to the fact that there is
some time delay in the temperature measurement. In all experiments performed in the cell the
pressure start dropping for approximately 30 seconds before any change in the temperature is seen.
For that reason it seems likely that gas hydrates start to form shortly after the actual conditions in
cell are favourable for their formation.
DISCUSSION
The results in series 1 and 2 indicate that induction times are very hard to reproduce even under
seemingly identicalconditions. Series 3 indicates that addition of large amounts of impurities almost
eliminates the induction time. It is however difficult from the present data to say whether the
standard deviation on the induction times in series 3 is as large as for the other experiments.
Comparison between the observed induction times in series 3 and the time after which the hydrate
formation is favourable could indicate that the induction times also in this case are scattered, though
very short.
The growth rates are for all series much less scattered than the induction times and thus easier to
compare. It is no surprise that the growth rate in series 2 is larger than in series 1 since the driving
508
force for hydrate formation is considerably higher. Expressed in terms of the Skovborg model, the
than in series 1 (-1.8'
difference x,,, - xb is larger in series 2 (-3.7*
When looking at the average A*k which is needed to describe the experiments in series 1 and 2, it
is interesting to note that a smaller value applies for series 2. One must assume that the interfacial
area is the same in both series, i.e. k must differ. This observation has two possible explanations.
Either the mass transfer coefficientis pressure dependent or the mass transfer of ethane from the gas
phase to the water phase is not the only significant kinetic step. If for example the building up of gas
hydrate crystals is also a significant kinetic step, one must assume that xb will be higher than
predicted from a gas hydrate equilibrium point of view. If xb is in fact higher than predicted by the
model, an analysis of the data assuming equilibrium between the water phase and the hydrate phase
will result in a higher value of A'k as it is the case in series I compared to series 2. This could
suggest that when the driving force for hydrate formation is small, the building up of crystals is a
significant step, whereas it becomes less significant for larger driving forces.
When analysing the amount of hydrate formed in series 3 it is seen that during the first 600 seconds
it is even lower than the results obtained in series 1, whereas at later times the results are comparable
to those of series 2 (table 4). The explanation is probably that the driving force for hydrate formation
is small in the beginning of the series 3 experiments and the conditions of the experiments resemble
those of series 2 at later stages. In series 3 the standard deviation on the amount of hydrates formed
is higher compared to the previous series, indicating that impurities in some way make the growth
rate more random.
One can also observe that the average A*k gets smaller as more hydrates are formed. This is most
likely due to the reduction of the interfacial area caused by formation of a hydrate slurry layer
between the two fluid phases.
CONCLUSIONS
Reproducing induction times in "clean" laboratory conditions seems very difficult if not impossible.
Deliberate addition of impurities to the system seems to eliminate the induction time. Extreme
caution should therefore be taken when operating a multiphase pipeline under conditions favourable
for hydrate formation. Kinetic inhibitors that supposedly prolong the induction time should be
evaluated also in the presence of considerable amounts of impurities.
The experiments reveal that the growth rate depend on the driving force. This may either be
interpreted as a pressure dependence of the parameter k in the model by Skovborg or it may indicate
that the building up of crystals is also a significant kinetic step. From the present experiments it is
not possible to say which explanation is correct.
When analysing kinetic experiments one should consider that the system investigated may not be
in thermal equilibrium at all times.
ACKNOWLEDGMENT
We are grateful to KSLA-ST/2 , Amsterdam, The Netherlands for lending us the measurement cell,
and to K.S. Pedersen, Calsep NS for many productive discussions.
REFERENCES
Skyum L., 1995, "Hydrate Kinetics", M.Sc. Thesis, Department of Chemical Engineering,
The Technical University of Denmark.
Andersen B.D., 1995, "Formation of Gas Hydrates", MSc. Thesis, Department of Chemical
Engineering, The Technical University of Denmark.
Skovborg P. and Rasmussen P., 1994, "A Mass Transport Limited Model for the Growth of
Methane and Ethane Gas Hydrates", Chem. Eng. Sci., 49, 113 I-i 142.
Munck J., Skjold-Jwgensen S. and Rasmussen P.,1988, "Computations of the Formation of
Gas Hydrates", Chem. Eng. Sci., 43, 2661-2672.
509
~~
Series 1
Experiment No.
Induction
time I s
Moles o f hydrate
formed in 600 s.
Moles o f hydrate
formed i n 3600 s.
M6
38209
0.0247
0.0804
M7
4 I827
0.0229
0.0792
M8
60066
0.0205
II
1I
1
M9
45509
0.0768
0.0245
155361
0.0222
0.0786
MI2
I04444
0.0230
0.0797
~~
MI3
298416
0.0134
II
0.0802
MI0
I
I
(1
0.0631
MI4
341430
0.0216
0.0785
MI6
106390
0.0233
0.0804
MI8
77961
0.0 I88
0.0733
1:
1
1I
L
Series 2
Experiment No.
Induction
timels
Moles o f hydrate
formed in 600 s.
Moles o f hydrate
formed in 3600 s.
M20
38489
0.0294
0.1112
M21
147025
0.0262
0.1097
M22
67392
0.0275
0. I003
M23
0.0313
0.1139
M24
47379
0.0222
0.1027
M25
10096
0.0269
0.0960
Moles of hydrate
formed in 600 F
Molzs o f hydrate
formed in 3600 s
280
285
00183
0 0892
M32
229
230
0.0248
0.1038
M33
I74
205
0.0193
0.0990
_ _ _-.
M34
M35
M36
227
235
0.0168
0.0778
220
215
0.0243
0.01095
233
185
0.0239
0.01029
510
~ I (
OhFcNed
Inducuon time6
Series 3
Experiment ho
1:
122524
PI 100
Sensor
Pressure
Transducer
Goa I
Phaseil
Outline of
Sapphire Window
4
- Cell
-2 -Magnet
Cooling Jacket
7I
IC
Temperature
-1
3.4
\
\
3.3
3.2
\
\
'.--
3.1
3.0
1mm300004MMO50000600007000080000
Tirnekeconds
0.032
v)
a,
0
0.028
v)
0
0
r-l
0
series1
series2
r,
.- 0.024
.+
a,
R 0.020
Lc
100000
200000
Induction Timekeconds
Figure 3. Growth rate vs. induction time for series 1 and 2.
511
300000
a,
I-
-- -_
40000
Experimental
The apparatus for studying the precipitation and growth of solids is a variation of that described
elsewhere (Figure 1 Holder and Enick, 1995). The apparatus consists of two coaxial cylinders.
Hydrates form in the annular space by growing radically inward from the cooled outer wall. To
form hydrates, the temperature controlled vessel is pressurized with the appropriate gas, 100-150
ml of water are injected and the inner cylinder rotated. The vessel can be operated in a vertical
position as shown in Figure 1 or in a horizontal position. For these experiments a horizontal
512
position was used exclusively. The fundamental improvement of this apparatus relative to cells
used in the past is that gas will flow in the annular space over the hydrate formation surface
rather than remaining static. This will provide a means of investigating the effects of gas shear
stress at the hydrate forming surface in pipelines.
I\
Temperature gradients within this vessel are established by flowing methanoVwater mixtures
from separate isothermal baths through cooling coils on the outside of the vessel and the inner
cooling chamber. The water in the vessel would coat the rotating inner cylinder, evaporate, and
condense on the inside of the outer cylinder.
A temperature gradient was established across the annular gap. The surface of the water was
then in direct contact with the flowing methane at 273K - 275K. Hydrates began to form inside
the outer cylinder. Pressure decreased as the methane from the gas phase entered the hydrate
phase and was used to deliver the amount of water converted to hydrate.
Experiments continued until either 1) the pressure of the methane remained constant, indicating
that hydrate formation had ceased or 2 ) the rotation of the inner cylinder ceased because small
amounts of hydrates clogged the bearings and friction surfaces. The system was then
depressurized and the hydrate crystals examined before they had an opportunity to completely
dissociate. The hydrates formed a relatively uniform layer of frost-like solid in the annular gap.
Results
We have measured the linear growth rate of hydrates formed from pure methane, pure carbon
dioxide, two mixtures of methanetpropane whose compositions were (95% methane and 5%
propane) and (97% methane and 3% propane). The important variables in these studies were
gas flow rate, gas composition, temperature, and pressure. Table 1 lists the results.
Gas flow rate: Higher gas flow rates (reported as RPM) tend to produce higher rates of hydrate
formation. This is because the higher gas flow rates dissipate the considerable heat release
generated dying hydrate formation (50-100 kJ per mole of hydrated gas) and because the higher
gas flow rates improve the mass transfer of water to the hydrate forming surface. It is still not
clear which of these factors is most significant. However, the effect of gas flow rate appears to
level off at the highest rates. This means that mass and heat transfer are no longer limiting and
a true kinetic value for hydrate growth is obtained. The higher gas flow rates used here are
comparable to pipeline Reynolds numbers in excess of 10,000 and thus these conditions are
those that might be obtained in an actual gas pipeline.
Cas composition: It was observed that no clear difference in growth rates for the 95% methane
and 97% methane mixtures were observed, but rates for gas mixtures and for carbon dioxide
were faster than for pure methane. The methane hydrate is the least thermodynamically stable
and it appears that the thermodynamic driving force (difference between the equilibrium
temperature and the actual experimental temperature at the hydrate surface) affects the rate at
which hydrates form. Another factor which may be important is the ability to stabilize the large
cavity of the hydrate structure. Propane stabilizes the large cavity of structure I1 better than
methane and carbon dioxide stabilizes the large cavity of structure I better than methane. The
ability to stabilize the large cavity may play a role in the kinetics. The current experimental
evidence is not conclusive on this issue.
Another variable of interest for the methanetpropane mixtures is that the gas composition
changes as the hydrates form since the propane concentration in the hydrates is much higher than
in the gas phase. As more hydrates form a eutectic mixture of methane and propane containing
less than 1% propane should be present. This mixture should result in the simultaneous
formation of both structure I and structure I1 hydrate.
Temperature: In general, temperatve is thought to increase the rate of any kinetic process.
However, higher gas temperatures decrease the thermodynamic driving force and will tend to
impede the rate.
513
Pressure: The rangeof pressures used was small, but the results indicate that higher pressures
increased the rate of hydrate formation.
Inhibitors: Both WAX and PVP reduced the rate of hydrate formation to negligible values
when present. Both were applied to the inside surface of the outer cylinder prior to hydrate
formation.
The overall correlation for growth rate for both methane + propane mixture is
linear rate = 0.001535(RPM)+ 9.3x 10 (PIKPA)
0.0178(Cooling Coil (TK).
This correlation has a R2 of 0.75.
Conclusion
The rates of hydrate formation along pipe walls will likely be comparable to the rates measured
in this study. Linear growth rates of 0.2cmflv are likely to represent the maximum growth rate
that could be expected in gas transportation lines. As the hydrates thicken, they can serve as
insulators of the line which will result in slower cooling of the produced fluids ( which come out
of the earth at higher temperatures than exist in the transportation line).
The insulation will
produce higher transportation temperatures and could either enhance or inhibit hydrate
formation rates.
Based upon the rates measured here, transportation lines could be operated for hundreds or
thousands of hours prior to their blockage by hydrates.
Acknowledgment
We would like to thank GRI for their partial support of the early stages of this research under
contract GRI5091-260-2121.
References
Holder, G.D. and R.M. Enick, Solid Deposition in Hydrocarbon Systems-Kinetics of Waxes,
Ashpaltenes and Diamondoids, Final Report, Gas Research Institute, GG, 1995.
Holder, G.D., S.Zetts, and N. Pradhan, Reviews in Chemical Engineering, S(1), 1,1988.
Holder, G.D.; Zele, S.;Enick, R.; LeBlond, Cdnn. N Y. Acad. Sci. 1994,715,344.
M.N. Lingelern, A.I. Majeed, and E. Stange, Industrial Experience in Evaluation of Hydrate
Formation, Inhibition, and Dissociation in Pipeline Design and Operation, in Natural
Cas Hydrates, Annals of the NY Academy of Science, E.D. Sloan et al, editors, New
York, p7S, 1994.
Sloan, E.D., Clathrate Hydrates of Natural Gas, Marcel Dekker, New York, 1990.
514
Figure 1.
2. Gas Hydrate
3.
4. Gas Cooling Fluid Flowing in Cylinder
5. Ice Cooling Fluid Rowing in Coils
6. RotatingCylinder
7. High Pressure Cell
8. Insulation
T - Thermocouples
P Pressure Transducer
Bulkgas
Temp 00
1Oo%C,
1Oo%C,
lOo%C,
lOo%C,
1Oo%C,
IOo%CI
IOoICI
100%C,
IOoIC,
S%C,:%%C,
S%C,:%%C,
S%C,:%%C,
5%C,:%%CI
5%q:%%C,
3%C,:97%CI
3%C+97%Cl
3%C3:97%cl
3%C3:97%C1
3%C,:97%C1
3%C3:9S%C,
cq
cwm
CHJPVP
283.7
283.5
282.9
282.1
281.9
+I- 0.4
+I- 0.5
+I- 0.4
+I- 0.4
+I-0.3
283.0 +I- 0.5
282.3 +I- 0.7
279.7 +I- 0.6
276.5 +0.2
I291.1 +/- 0.3
288.9 +I-0.4
288.7 +I- 0.6
288.4 +I-0.7
287 +0.9
I284.7 +/-0.7
288.1 +/-0.1
284.5 +/-O.I
286.8 +I- 0.2
287.5
286.4 +/-0.1
287. +0.3
I277.8 +/-0.7
278.6 +0.6
I-
Cooling coil
Temp 00
W)
6M9
6039
6021
6023
5987
6ow
6019
6861
6618
7014
7110
693
+i-o.i
8
8
8
30
2
8
25
I5
52%
5482
IS
7089
6781
7036
67Q
7016
b!m
7011
3555
271.4 +I-0.1
271.6 +I- 0.2
m.1 +I- 0.2
RPM
7
I5
60
I
30
7
7
60
2
2
30
8
1s
6wo
no
Initial
515
In 1991, a report appeared' which suggested that methanol, when co-deposited with
water in a vacuum at low temperatures,forms a clathrate hydrate of structure II with a
lattice constant of 16.3 A. This observation poses a number of questions regarding
the ability of methanol to both inhibit and promote hydrate formation. If one is a high
temperature property, and the other a low temperature property, at which temperature
do they cross over and what are the responsible interactions ? On the other hand,
there are some general questions regarding the interpretation of the results. For
instance, the usual structure II lattice parameter is 17.1 f 0.1 A',so it isn't easy to see
how the structure II lattice can adapt to a 12% volume reduction.
patterns were taken at 80K. Samples were annealed for 10 min. at the appropriate
temperatures then cooled for the recording of the XRD pattern. The various phases
which appear on crystallization of the amorphous deposits and mixtures were identified
by comparison with data obtained for the pure crystalline materials or from literature
p'ata.
H NMR powder patterns were recorded at a frequency of 30 MHz on a Bruker MSL
200 NMR spectrometer equipped with a 5mm solenoid probe. Quadrupole echo
sequences were used with a 90" pulse length of 2.5 psec and an echo delay time of
30 p e c . Samples of 1:1 stoichiometry were quenched and conditioned in the probe at
temperatures somewhat above 150K.
Quadrupole coupling parameters were
determined by a fitting procedure using the Bruker Winfit package.
With the identification of the methanol monohydrate phase, all of the reflections can be
accounted for for all of the samples considered in this study. The crystallization
processes which transform the lowest temperature amorphous phases are outlined
below for a range of sample compositions.
Ice
Ice IC
methanoVwater ( 1:20, 1:lO ) =) methanol monohydrate + Ice Ih + Ice IC(trace)
methanol/water ( 1: 2 ) * methanol monohydrate + Ice Ih + methanol
methanol 3 a - methanol
Perhaps the most unusual observation is the near disappearance of cubic ice as a
distinct phase in the presence of methanol. Figure 1 shows a typical XRD run on a
codeposit (1.2 methanol- water ) initially annealed at 120K. The ice Ih reflections
clearly are present for all temperatures. The calorimetric data is in agreement with
this, as the only evidence for the crystallization of cubic ice is from a very weak
exotherm near 148K in the 1O:l deposit ( see table 1 ), and the subsequent
transformation of ice ICto ice Ih occurs near 179K; these events occur near 142K and
19% for pure ice. Methanol seems to act as a catalyst in the direct conversion of
amorphous ice to ice Ih.
How do our results relate to those reported for the " methanol clathrate '' ? The
material prepared was thought to be a clathrate mixed with some cubic ice based on
the assignment of 10 reflections to a 16.3 A clathrate lattice plus 1 reflection due to
cubic ice. By comparing the diffraction data for the two studies it becomes apparent
that all of the observed reflections in both studies can be accounted for in terms of ice
517
Ih and methanol monohydrate. It should be noted that the small angle reflections ( hkl
= 11 1, 220 and 31 1 for str. II ) which are essential for assigning the clathrate structures
in mixed-phase systems were not observed. Certainly in light of our data there is no
need to propose a structure II clathrate hydrate with an unusual lattice parameter.
Although the XRD powder data have given a good indication of the symmetry and size
of the unit cell in the crystal, the detailed structure remains as yet unknown. Some
additional information on the methanol monohydrate can be obtained from NMR
spectroscopy. The 'H NMR quadrupole coupling parameters obtained for the water
lattice are useful, as these can be used to give information both on the strength of the
hydrogen bonds and the dynamics of the water lattice. The 'H NMR lineshape for a
sample of CH,OH. D,O yielded a quadrupole coupling constant x ( = e2Qq/h ) of 206.9
kHz and an asymmetry parameter q= 0.09. These values are very near to those for
ice Ih ( 215 kHz, O.l"), and are characteristic of a fully hydrogen-bonded network. The
x value is an average for the D,O and CH,OD deuterons, as exchange must produce
the deuterium-substitutedmethanol. The small decrease in x indicates that the 0 - 0
distances are, if anything, on average slightly shorter than those in ice Ih. The
temperature dependence of the lineshape ( fig. 2 ) gives a rough indication of the rate
of the dynamic processes in which water molecules are involved. Above about 140K,
the lineshape develops a central component which is a manifestation of slow
reorientation 'of the water molecules within the lattice. A comparison with the
lineshapes obtained for pure ice Ih indicates that the water molecules in the methanol
hydrate at 150K have the same motional correlation time as those in ice Ih at -260K.
Since the water reorientation in ice is defect-driven, the implied low activation energy
for water reorientation in methanol monohydrate must reflect the ease with which
defects can be generated. A more detailed analysis of D,O dynamics is complicated by
the fact that besides the water molecules there are two kinds of methanol 0 - D bonds
that must be involved in restricted dynamic processes ( see below ). Examination of a
corresponding CD,OD.H,O ( fig. 2, below ) sample showed that there are two
dynamically inequivalent CD, groups, as there are two overlapping powder doublets.
Both CD, groups show rapid rotation about their 3-fold axes, one of the two shows little
or no additional motion, the second shows the presence of another process which
formally can be explained by a jump between two positions with 40" between the
directions of the CD, group symmetry axes. The central line ( fig. 2, right ) arises from
the more mobile OD and HOD deuterons. The NMR measurements show quite clearly
that molecular motion is possible at much lower temperatures than in a pure ice lattice.
CONCLUSIONS
On the basis of results reported here it appears unnecessary to propose the existence
of a clathrate hydrate of methanol. In vapour-deposited or quenched water - methanol
mixtures the crystallization products include mainly methanol monohydrate and Ice Ih
on the water-rich side of the composition diagram. Methanol seems to act as a
catalyst for the direct conversion of glassy ices to ice Ih. On the basis of XRD it is
proposed that the methanol monohydrate is tetragonal with unit cell dimensions a=
4.660, c= 13.813 A. NMR results indicate that both water and methanol are part of a
fully hydrogen-bonded network, but that water reorientation takes place much more
easily than in pure ice Ih and in most clathrate hydrates. Finally, one can say that
methanol by all accounts remains an inhibitor of hydrate formation.
Issued as NRCC no:39128
' Sloan, Jr., E. D. "Hydrates of Natural Gas , Marcel Dekker, New York, N. Y. 7997
' Davidson. D. W., Gough. S. R.. Ripmeester,J. A.. Nakayama, H., Canad. J. Chem. 7981 59.2587
' Davidson. D. W. , in "Water a ComprehensiveTreatise. Franks, F., Ed., Plenum, N. Y.. 7973. V01.2
' Calvert, L D., Srivastava, P.. Acta Crystallogr. Sect.A 7969, 25, S131
"
' Ripmeester, J. A., Ratcliie, C. I.,J. Phys. Chem. J . Phys. Chem. 7990 94.8773
*Blake, D., Allamandoh. L.,Sandford, S., Hudgins. D., Friedemann, F., Science 7997 254,548
'Davidson. D.W.. Handa. Y P , Ratcliffe. C. I , Ripmeester,J. A,. Tse, J. S., Dahn. J. R., Lee.,F.,
Calvert. L. D.. Mol. Cryst. Liq. Cryst. 7986, 141, 141
Wallqvist. A., J. Chem. Phys. 7992 96. 5377
'Vuilhrd. G.. Sanchez, M., Bull. Cham. Soc. Fr. 7967,1877
'IHanda. Y. P., HawMns, R. E., Murray, J. J.. J. Chem. Thermodyn. 7984,16, 623; Y. P. Handa, J.
!hem. Thermodyn. 7986,18,8gi
Ripmeester, J. A., Ratcliffe, C. I., Klug, D. D., J. Chem. P h p . 1992,96,8503
518
.....................................................................................................
Starting material
ice (a)
methanol (a)
......................................................................................................
ice Ih
liquid mixture
liquid
795K( en )
783K( en; w )
.178K( en )
179K( en; w)
Ice Ih + liq
773K(en)
Ice Ih + liq
174K(en )
p phase + liq.
Ice IC
MeOH.H,O
Ice Ih
MeOH.H,O +
Ice Ih +eutectic
mixture
758K(en,w)
768K (en)
a phase + eutectic
mixture
758K (en.w)
748K(ex, w)
MeOH.H,O +
Ice Ih +eutectic
mixture (s)
742K (ex)
736K (9)
a phase +eutectic
mixture (s)
737K(ex)
722K (ex )
Ice (a)
amorphous
mixture
7 72K (9)
L',
amorphous
mixture
702K (ex)
methanol ( a )
.................................................................................................................
a = amorphous; ex = exotherm; en = endotherm; w = weak; g = glass transition, s =
solid
519
\
!>?-I
:aiai
-;BY
-MX
n
1
120 K
4Cm33
xoocn
b.
520
-2wooo
-400300
APPROACH
The approach used to develop the procedure for estimating hydrate forming conditions for
gases over electrolytes is similar to that used by Moshfeghian and Maddox (1990) for their
work on inhibited water solutions. They predicted the conditions for hydrate formation over
pure water and then calculated an adjustment, or change, in those conditions to account for
the presence of the inhibitor.
In the present work the method of Holder, Cohin and Papadopoulos (1980) is used to
calculate the conditions for hydrate formation over pure water, and equations are developed to
adjust those conditions for the presence of electrolyte. Holder and mworkers used
experimental measurements to generate chemical potential, enthalpy and heat capacity
functions. No gas species dependent adjustable parameters are required.
THERMODYNAMIC MODEL
The equations used for predicting the influence of weak electrolyte solutions on natural gas
hydrate forming conditions were developed from the work of van der Waals and Plalteeuw,
(23). The value of the Langmuir constant depends on temperature and potential energy
function parameters and was evaluated using the mathematical expressions of Parrish and
Prausnik (18).
521
At equilibrium the chemical potential of water is equal in all phases present. If the free water is
present as ice, none of the ice will be incorporated within the hydrate structure, and the
chemical potential of the water in the hydrate will be equal to that of ice. If liquid water is
present the free water and the water in the hydrate will have the same chemical potential.
The way in which the activity of water is evaluated depends on the components present in the
system under consideration. If pure liquid water is present Holder, et al. (8) suggest that gas
solubility in the water phase will be so slight that ,x, the mole fraction of water, may be used
without creating significant error. If. on the other hand, water is present in an electrolyte
solution with an appreciable concentration of salt present, the model suggested by Piker and
Mayorga (22) can be used to estimate the activity of water in the electrolyte. Following
Englezos and Bishnoi (5) and Tohidi, et al. (26) this work has used the Piker and Mayorga
activity model for predicting conditions necessary for hydrate formation in the presence of
electrolyte solutions.
i'
In case there is more than one electrolyte present in the solution, the procedure proposed by
Patwardhan and Kumar (19) is used to estimate the water activity.
Formation of hydrate from gas and liquid water molecules where G molecules of gas involve N
molecules of water can be represented by the following chemical reaction:
G + N(Hz0) t+ GN(H20)
For this representation Maddox. et al. (14) showed that the effect of a non electrolyte inhibitor
on the hydrate temperature of a natural gas could be explained as:
(1)
The derivation of equation (1) is discussed in detail by Pieroen (21). The electrolyte can be
treated as an "inhibitor" if the procedure developed is used for estimating electrolyte activities
and other parameters. In this work (AHINR), the enthalpy of hydrate formation per water
molecule in the hydrate lattice in equation (1) is assumed to be a function of pressure and the
ionic strength of the electrolyte solution and to take the following form:
AH
-=
NR
e,P
I+e,P+e,(hP)
with el, e2. e3 and e, being adjustable parameters determined from experimental electrolyte
solution hydrate data.
Experimental measurements of the hydrate temperature of methane (1, 2, 23) and ethane
(6,26) in the presence of liquid water and electrolyte solutions and propane (8. 10. 13, 26) in
the presence of ice. liquid water and electrolyte solution were used to evaluate the parameters
in equation (2). The values are: el= 597.33; 82=-0.0409; e3=0.0000227; e4=-0.0751. These
are global parameters that apply for any gas and any single or mixed electrolyte solution.
They reproduce the measurements of Blanc and Tournier-Lassenre (l), Dholabhai, et ai. (3).
and Roo, et al. (23) for methane (CH,) with an average absolute temperature deviation of
0.33%; The experimental determinations by Englezos and Bishnoi (6) and Tohidi. et al. (26)
for ethane are reproduced with an average absolute deviation of 0.56% across all data points.
Experimental data for propane by Englezos and Ngan (e), Holder and Godbole (10). Kubota,
et al. (13), and Tohidi, et al. (26) are reproduced with an average absolute deviation for all
propane data points of 0.35%. A summary of these results is shown in Table 1.
CALCULATION PROCEDURE
Based on the discussion above a procedure for calculating the hydrate forming temperature
and pressure for natural gas components in contact with water containing one or more
electrolyte salts can be suggested. Assuming that the pressure is fixed and the hydrate
forming temperature is required, the sequential steps in the procedure are:
1. Assume that the hydrate forming temperature in the presence of water with no electrolyte
present is above 273.15K. the freezing point of water. If the hydrate temperature is lower than
this, that fact will become evident and the assumed temperature can be changed. Use
equations 3, 7 and 13 to evaluate P.
2. Calculate the activity of water and AWNR.
3.Calculate the hydrate temperature in the presence of electrolyte.
4. If the temperature calculated in step 3 is greater than 273.15K,all is well; if it is lower than
273.15K retum to step 1, assume the temperature is less than 273.15K. and repeat steps 2
and 3.
522
RESULTS
The procedure outlined has been used to predict hydrate forming conditions for carbon dioxide
(Cod over electrolyte solutions. COz is a different gas than light hydrocafbons in that it
displays appreciable solubility in water, even at moderate pressures. There are cases in which
gas solubility is high enough that the mole fraction of water in the water phase departs
substantially from 1.O. The solubility of CO, in water can be expressed as:
where:
gr = -725919.22
g2 = 2890.54
Q3 =
0.05127
g, =
-0.11228
Calculations of COzsolubility using equation (3) match the experimental measurements of Con
solubility made by Stewart and Munjal (25) within an average absolute mole fraction W I
deviation of 0.00042 over temperatures from 259 to 281K and pressures from 1.O to 4.25 MPa.
Using equation (3) for CO, solubility in water and the hydrate prediction procedure developed
here, the hydrate forming conditions for GOz over electrolyte solutions have been calculated"
and compared with experimental determinations made by Dholabhai, et al. (4). and Englezos
(7). The results for CO, are summarized in Table 1. The 161 COz data points show an
average absolute temperature deviation of 0.46% over the full temperature, electrolyte
composition and pressure range of the data. ,
CONCLUSION
The model developed for predicting hydrate forming conditions in the presence of electrolyte
solutions does an excellent job of reproducing experimental measurements. It also has the
capability to make accurate predictions of hydrate formation in cases where the gas shows
appreciable solubility in the water phase. It represents a significant step in predicting hydrate
forming conditions for constituents of natural gas.
NOMENCLATURE
= activity of water
at = activity of water in the single salt solution defined by m:
A, = Debye-Huckelcoefficient = 0.392 for water at 25C
= parameter in equation (10)
PI = parameter in equation (10)
P2 = parameter in equation (10)
C
,, = Langmuir constant
CP = specific heat, caVg-mole-K
f,. = gas-phase fugacity of the Afh gas species
h = molar enthalpy, caWg-mole
m = molality of electrolyte solution
mk = molality of electrolyte k in mixed electrolyte solution
mi = molality of electrolyte kin a solution containing only electrolyte k and that has
the same ionic strength as the mixed solution
n = formula of electrolyte
n, = number of positive ions in electrolyte formula
n. = number of negative ions in electrolyte formula
nc = total number of components in gas phase
R = gas constant. 1.987 cal/g-mole-K
T = absolute temperature, K
V = molar volume cm3/g-mole
p t = chemical potential of water in the gas occupied lattice, cal/g-mole
a,
REFERENCES
Tabla 1
Component
Electrolyte
Concentration
rnoVL
NaCl KCI CaCI2
Temperature Pressure
Range
Range
K
MPa
AATD'
K
CH4
261-281
2.39-92.0
0.33
GHe
265-283
0.50-2.0
0.56
GHe
248-278
0.1-0.54
0.35
Coo
259-281
1.0423
0.46
524
INTRODUCTION
During the past five years we have been measuring phase equilibrium data in
gas hydrate forming systems in our laboratory. The broad objective of the work is to
provide thermodynamic data which will be used either directly in process design of
relevant operations in the oil and gas industry or can be used to test the validity of
computational methods for phase equilibrium. We have studied the effect of glycols,
water soluble polymers and electrolytes in hydrates from natural gas components. In
the present work, we provide measurements for two systems: First, phase equilibrium
data for the propane-water-triethylene glycol (TEG) system and second data for the
methane-carbon dioxide-2,2-dimethyl butane (neohexane) system.
It has been known since the 1930s that natural gas and water can form a solid
ice-like compound commonly called gas hydrate (Hammerschmidt, 1934). This may
take place at temperatures above the normal freezing point of water. Because the
formation of hydrates has severe economic consequences, in oil and gas operations,
prevention of formation is major concern. The most common method to prevent hydrate
formation is to inject methanol, glycol, or electrolytes (inhibiting substances). There is a
growing interest to replace thermodynamic inhibitors with kinetic inhibitors i.e. chemicals
which could perhaps prevent the agglomeration of gas hydrates after they have been
formed (Muijs. 1991; Sloan et al. 1994; Englezos, 1996). In spite of this growing effort
as well as the progress that has been made in hydrate thermodynamics, equilibrium
data for gas hydrates are still needed not only for process design but also for the
development and testing of predictive methods for hydrate equilibria (Sloan, 1990;
Englezos, 1993; Sloan et al. 1994).
Triethylene glycol is an industrially used chemical to inhibit the formation of gas
hydrates. Ross and Toczylkin (1992) have presented data on the effect of TEG on
methane and ethane gas hydrates. These are known to be structure I hydrates, Hence,
one of the objectives of this work is to report incipient equilibrium data for propane
hydrate in aqueous triethylene glycol solutions. Propane hydrate is known to form
structure I1type hydrate crystal lattice.
Following the report from the National Research Council (NRC) of Canada in
1987 on a new hydrate structure, Sloan and co-workers reported the first structure H
hydrate phase equilibrium data in 1992 (Ripmeester et al. 1987; Ripmeester and
Ratcliffe,l990; Lederhos et al. 1992). The possibility of forming structure H hydrates in
gas and oil reservoirs provides the motivation to obtain phase equilibrium data for
structure H hydrates. Subsequently, additional data and a method to predict structure H
equilibrium were reported from Sloans laboratory (Lederhos et al. 1993; Mehta and
Sloan,1993; 1994a; Mehta and Sloan,1994b; Makogon et al. 1996; Mehta and Sloan,
1996). Additional data were also reported by other laboratories (Danesh et a1.1994;
Hutz and Englezos, 1996).
Thus far only methane, nitrogen and argon have been used as light components
in the formation of structure H hydrates. Carbon dioxide in conjunction with neohexane
and ice also forms structure H hydrates (Ripmeester, 1996). In our laboratory, we
attempted to prepare such hydrate but in liquid water. However, we were not able to
form hydrates which could be of structure H. At a given temperature, the hydrate that
was formed was stable at the carbon dioxide structure I hydrate equilibrium pressure.
525
of the hydrate. We plan to analyze the gas and the solid phase in order to elucidate the
structure.
CONCLUSIONS
The effect of triethylene glycol (TEG) on the formation of propane hydrate was
studied at 0, 10 and 20 wt % aqueous TEG solutions. TEG was found to have a
significant inhibiting effect comparable to glycerol but weaker than methanol or NaCI.
The experiments with a 80-20 % methane carbon dioxide mixture together with
neohexane in liquid water were not conclusive with respect to the structure formed.
However, the incipient equilibrium formation conditions for this system were
determined.
REFERENCES
1. Breland, E., and P. Englezos 1996. Equilibrium Hydrate Formation Data for Carbon
Dioxide in Aqueous Glycerol Solutions. J. Chem. Eng. Data. 41(1), 11-13.
2. Danesh, A,, B. Tohidi. R.W. Burgass and A.C. Todd, 1994. Hydrate equilibrium data
of methyl cyclo-pentanewith methane or nitrogen. Trans I Chem E (Chem. Eng.
Res. Des.), 72, PartA: 197-200.
3. Englezos, P., 1993. Clathrate hydrates. lnd. Eng. Chem. Res., 32(7): 1251-1274.
Englezos. P. 1996. Nucleation and Growth of Gas Hydrate Crystals in Relation to
Kinetic Inhibition. Revue de llnstitut francais du petrole, (in press).
4. Englezos, P. and Ngan, Y. T.,1994. Effect of polyethylene oxide on gas hydrate
phase equilibria. FluidPhase Equilibria,92: 271-288.
5. Hammerschmidt, E.G. 1934. Formationof Gas Hydrates in Natural Gas
Transmission Lines. lnd. Eng. Chem. 26 (8), 851-855.
6. Hutz. U., and P. Englezos. 1996. Measurement of Structure H Hydrate Phase
Equilibriumand the Effect of Electrolytes. Fluid Phase Equilibria, 117, 178-185.
7. Lederhos,J.P., Christiansen, R. L. and Sloan, E.D., 1993. A first order method of
hydrate equilibrium estimation and its use with new structures. FluidPhase
Equilibria, 83: 445-454.
8. Lederhos,J.P., Metha, A.P., Nyberg, G.B., Warn, K.J. and Sloan, E.D., 1992.
Structure H clathrate hydrate equilibria of methane and adamantane. AlChE J.,
38(7): 1045-1048.
9. Metha, A.P. and Sloan E.D.Jr., 1993. Structure H hydrate phase equilibria of
methane + liquid hydrocarbon mixtures. J. Chem. Eng. Data, 38: 580-582.
lO.Metha, A.P. and Sloan E.D.Jr., 1994a. A thermodynamic model for structure-H
hydrates, AlChE J., 40(2): 312-320.
1l.Metha, A.P. and Sloan E.D.Jr., 1994b. Structure-H phase equilibria of paraffins,
naphthenes, and olefins with methane, J. Chem. Eng. Data. 39, 887-888.
12.Metha. A.P. and Sloan E.D.Jr., 1996. Improved thermodynamic parameters for
prediction of structure-H hydrate equilibria,AlChE J., 42(7): 2036-2046.
13.Muijs, H.M. 1991. Surfactants in Oil Production, in Chemicals in the Oil Industry:
Developments and Applications; Ogden, P.H.,Ed., Roy. SOC.Chem. pp.277-297.
14.Ripmeester. J.A. 1996. Personal Communication.
15.Ripmeester,J.A., Tse, J.S., Ratcliffe,C.I. and Powell, B.M., 1987. A new clathrate
hydrate structure. Nature, 325: 135-136.
16.Ripmeester.J.A. and Ratcliffe, C.I., 1990. 129 Xe NMR studies of clathrate
hydrates: new guests for structure II and structure H. J. Phys. Chem., 94: 87738776.
17.Ross, M.J. and L.S. Toczylkin, 1992. Hydrate Dissociation Pressures for Methane or
Ethane in the Presence of Aqueous Solutions of Triethylene Glycol, J. Chem. Eng.
Data, 37. 488-491.
18.Sloan, E. D. Jr.. 1990. Clathrates Hydrates ofNatural Gases. Marcel Dekker, New
York.
19. Sloan, E.D.; Happel, J.; Hnatow. M.A. 1994. International Conference on Natural
Gas Hydrates,Annals of the New York Academy of Sciences, Vol. 715, New York.
ACKNOWLEDGMENT
The financial support from the Natural Sciences and Engineeing Reseerch
Council of Canada (NSERC) is greatly appreciated.
527
Figure 1.
Glycol-Water Bath
Experimentalapparatus
500
A
0
.
0
Thbwork
Methenbneoheranehydrate smdure H
Carton dioxide hydrate SINdure I
04
270
271
272
273
274
275
276
277
270
; 0
279
Temperature (K)
Figure 2.
.)
.
X
X
Ax
1000
01
Temperature (K)
Figure 3.
Gas hydrates were, until recently, considered to be a phenomenon associated with small diameter
hydrocarbon molecules, For systems relevant to the oil industry, the upper limit was set by n-butane.
In 1987, Ripmeester et al(1,Z) discovered a new hydrate structure called structure H.This structure has
cavities larger than those of the former known structures I and I1 and can therefore be stabilized by
heavier guest molecules, Of components foud in real hydrocarbon fluids, isopentane,
methylcyclopentane, methylcyclohexaneand 2,3ðylbutane were identified as potential structure H
formers while benzene, cyclohexane and cyclopentane were identified as potential structure I1 formers.
Recently, Thoidi et. al (3) presented experimental and prediction results on the effect of isopentane,
methylcyclopentane,cyclopentane and cyclohexane on the hydrate equilibrium properties of two ~ t u d
gas mixtures,They concluded that stlucture I1 heavy hydrate formers increase the hydrate stability of
the two hydrocarbon gas mixtures, where as structure H heavy hydrate formers do not have sigruficant
effect on the hydrate phase boundary.
In the oil industry the hydrate prevention strategy makes extensive use of hydrate prediction programs.
To our knowledge, none of the present commercially available programs take account of the heavy
hydrate formers. If some heavy hydrate formers, found in a real hydrocarbon fluid, have a significant
effect on the hydrate equilibrium temperature these have to be included in the prediction programs.
In this work the hydrate equilibrium properties of isopentane, cyclopentane and cyclohexane in a
synthetic hydrocarbon fluid have been determined. The vapor phase was simulated by either pure
methane - a structure I former, or a synthetic gas mixture assuring formation of structure 11. The
hydrocarbon liquid phase was in all experiments Exwsol D60, which is a paraffinic C9 - C13 distillation
cut.
Also presented is a survey of the content of cyclopentane in a selection of real hydrocarbon fluids from
the North Sea area.
EXPERIMENTAL
All experiments were performed in a high pressure sapphire PVT-cell. The cell is placed in a
temperature controlled air bath in which the temperature can be varied between -40 and +ZOO 'C. The
temperature stability is 0.1 OC and the resolution is O.OIoC. The cell has a maximum worlung pressure
of 500 bara. The accuracy of the pressure measurement is estimated to be within 0.5 bar and the
resolution is 0.1 bar. The cell volume is controlled and varied using a piston directly coupled to a
computerized brushless motor. Volumes are read with a resolution of 0.0001 an3. The estimated
accuracy is 0.005 cfi? Maximum cell volume is 100 cm3.
Stirring is provided by a magnetically coupled stirrer, driven by a computer controlled, variable speed
motor. Maximum speed is 1000 rpm. Rheology changes of the experimental fluids are continuously
monitored by measurement of the effect required to keep the motor running at c o m t speed.
Experimental procedures. The sapphire cell was cleaned and evacuated prior to filling of the
experimental fluids. All fluids were added gravimetrically in the following sequence: Water (purified by
reversed osmosis), hydrocarbon liquid phase (D60 - added various amounts of heavy hydrate formers)
and finally the hydrocarbon vapor phase (CI or synthetic gas mixture). The water cut was, in all
experiments, approximately 50%. The hydrocarbon phase were recombined and the saturation point, at
ambient temperature, were determined.
Hydrate formation was initiated by cooling the system at a constant rate, 3 - 5 "Chours, while
continuously stirring the cell. AAer a period of time, allowing for equilibrium to be established, the
system was reheated at a rate of 0.25 "Chours until the hydrates were completely melted. The
experiment gives information of the hydrate equilibrium temperature, the degree of sub cooling and the
visual appearance of the formed hydrates. Hydrate formation and decomposition are indicated by
deflections in a volume vs. temperature plot (the isobar) and by a change in the rheology of the system
(the apparent viscosity). The experiments were performed at isobaric conditions, at 100, 200 and 300
bara.
All experiments were documented by video recordings.
Experimental fluids. The synthetic gas mixture were composed of CI, C2 and C3 in the ratio 74.89,
16.47 and 8.64 mole %, respectively. All components were minimum 99.9 % pure. Analysis of the
mixture did not show significant contamination of any other components.
529
Methane were 99.95 % pure. Isopentane, cyclopentane and cyclohexane were all of analflcal grade
purity.
Exsol D60 is a commercial paraffmic solvent. The composition of D60 is given in table I .
The compositions of the different experimental systems are given in tables 2 and 3. The measured
saturation pressure and the recombination gadoil ratio are given along with the composition.
530
REFERENCES
Ripmeester, J.A., Tse, J.S., Ratcliffe, C.I. and Powell, B.M.:
(I)
"A New Clathrate Hydrate Structure", Nature (1987) 325.
Ripmeester, J.A., Ratcliffe, C.I. and McLaurin, G.E.:
(2)
"The role of heavier Hydrocarbons in Hydrate Formation",
AlChE Spring Meeting (1991)
Thohidi, B., Danesh, A., Burgass, R.W.andTodd, A.C.:
(3)
SPE35568 "Effectsof Heavy Hydrate Formers on the Hydrate Free Zone of Real
Reservoir Fluids", European Production Operations Conference & Exhibition (1996)
Carbon no.
Naphtenes
(wt.%)
Mw
Density
Table
(kg/m3)
126.2
768
140.8
778
155.4
788
169.8
797
183.7
805
(mole%)
41.16
38.02
(mole%)
9.05
8.36
(mole%)
4.75
4.39
(mole %)
10.97
D60
(mole%)
45.03
38.26
GOR
(Sm3/Sm3
144
139
Psat
(Bara)
131 Q23.1 "C 125 6323.9 "C
Hydrate equilibrium (T):
Q 100 Bara
Q 200 Bara
19.7
22.0
25.3
27.4
39.9
8.78
4.80
1.07
45.66
132
141Q29"C
19.9
22.2
, Q 100 Bara
I
13.7
Q 200 Bara
Q 300 Bara
i
531
12.9
25.7
29.4
29.8
Table 3 cont The methanem60 system Composition and hydrate equdibnum data
c 4
c 5
C6
System
C1
(mole%)
59 02
59 75
(mole%)
C2
Cy-C5 (mole %)
4 61
0 92
Cy-C6 (mole %)
D60
(mole Oh)
36 37
39 33
GOR
(Sm3/Sm3)
180
176
Psat
(Bara)
247 @25 4 C 247 @21 7 C
Hydrate equilibnum (C)
Q 100 Bara
22 4
13 1
Q 200 Bara
18 6
@ 300 Bara
20 9
Table 4. Conte
3
4
5
3
7
B
9
10
11
12
13
8 01
33 52
182
244 @24 C
13 6
Cyclopentane
(M%)
0.01
0.03
0.07
0.07
0.12
0.0:
0.05
0.05
0.04
0.04
0.05
0.06
0.05
0.04
0.05
0.04
0.09
0.08
0.02
0.03
0.09
0.08
0.04
0.07
56 47
0.11
350
300
A
de 250
c, m
Cl Em
5 200
0
ClrSZW.8bUR.b
.. .. .. .. .
v)
u)
2 150
ci
100
5
10
15
20
TEMPERATURE (C)
25
532
400
50L 8
'
12
'
16 ' 20 ' 24
TEMPERATURE ("C)
'
28
'
d2
d 250
~i
1200
CH,CLATHRllE
'Oo0
534
-. .
ofthe
rotational peak w,th neutron momentum
Fig,
transfer (a)
L\
I
8"
'
.I0
I6
'J
..
$
Ee
0-1
10111.01b1 ,11,*,10*
IO
10
JO
T ~ r n p u ~1KJ
w~
535
'06
Fig
moves the guest vibrations out of the lattice acoustic region. This fact is shown in the
calculations on the hypothetical hydrate. The correlation time between the *'light
xenon" and the lattice is extremely short and less than 1 psec. The corresponding
"coupled" vibrations are located from 250 to 600 cm.'. which are outside the
translational region of the lattice vibrations. To summarized, it is shown from the
present molecular dynamics calculations, the coupling between the guest and water
lattice vibrations only occurs at the low frequency region. In practice, the hydrocarbons
enclathrated in the hydrate structure all possess low frequency librational and rattling
motions thus permitting the exchange of thermal energy with the host lattice.
-50.0
1000.0
500.0
0.0
lrequenq (crn^-l)
Fig. 4 The fourier transform of the cross velocity correlation function for
xenon and "light xenon" hydrate.
t
0
q -vector ---+
2nla
Realizing the basic mechanism for the coupling between the guest and host lattice,
a simple model based on a modified Einstein model can be used to predict the thermal
conductivity of gas hydrate [16,17]. One important feature of this thermal conductivity
536
model is that only the molecular properties are needed as input parameters for the
calculation of the thermal conductivity In this model, the localized vibrations are
assumed to be heavily damped with lifetimes of half a period of the oscillation and the
distribution of the localized modes in a solid can be approximate by the Debye model.
Using the equivalent of the gas kinetic equation, an expression for the minimum thermal
conductivity Amincan be derived.
At high temperature when T >> OD , the transport integral approaches unity and the
limiting thermal conductivity Am is given by
i(
A co = 2 E)5kBn3(2vt
6
+ v,)
Where k, is the Boltzman constant, n is the number density and v, and v, are the
longitudinal and transverse sound velocity of the hydrate. Therefore, once the
experimental density and acoustic sound velocities are known, the thermal conductivity
at any temperature can be estimated. Previous calculations show that the calculated
thermal conductivities for several gas hydrates employing these are in good accord with
experiment [17].
The electrostatic field created by the water froming the cavities in a hydrate has
important effects on the vibrations of the guest molecules. In view of the potential use
of vibrational - infrared and Raman spectroscopies as an alternative means of the
determination of the hydration number, it is imperative to understand these electrostatic
effects. An electrostatic potential map (MEP) for the hydrate cavity can be computed
from quantum mechanical method. We have computed the MEP for the small and large
cage of a structure I hydrate. The MEP can be used to derive appropriate point charge
model for the water molecules for the future simulation of the interactions between the
guest and the cavities. Details of the computational results will be published elsewhere.
REFRERENCES
1. ROSS,R.G., P. Anderson, G. Backstrom, 1981, Nature, 290, 322.
2. Tse, J.S., White, M.A., 1988, J. Phys. Chem., 92, 5006.
3. Holder, G., Angert, P.F., Pereira, 1983, "Natural Gas Hydrates", Cox, J.L. ed.,
Butterowrlh Publishers.
4. McMullan, R.K., Jeffrey, G.A., 1965, J. Chem. Phys., 42, 2725.
5. Mak, T.C.W., McMullan, R.K.. 1965, J. Chem. Phys., 42, 2732.
6. Tse, J.S.. Powell, B.M., Sears, V.F., Handa, Y.P., 1993, Chem. Phys. Lett., 215,
383.
7. Shpakov, V.P., Tse, J.S., Belosludov, V.R., Belosludov, R.V., 1996, J. Phys.,
Conden. Mat., (submitted).
8. Tse, J.S., Klein, M.L., McDonald, I.R., 1983, J. Phys. Chem., 92, 5006.
9. Berendsen, H.J.C., Postma, J.P.M., van Gunsteren, Hermans, J., 1981.
" Intermolecular Forces", Pullman, B. ed., Reidel Dordrecht. the Netherlands.
10.Jorgenson. W. 79 L., Chandraskhar, J.D., Madura, J.D., Impey, R.E., Klein, M.L.,
1983, J. Chem. Phys.,79, 926.
i i . T s e , J.S., Klein, M.L., McDonald, I.R., 1987, J. Chem. Phys., 91, 5786.
12.Garg, S.K., Gough, S.R., Davidson, D.W., 1975, J. Chem. Phys., 63,1646.
531
13.Tse, J.S., Ratcliffe, C.I., Powell, B.M., Sears, V.I., Handa, Y.P., 1997, J. Phys.
Chem., in press.
14.Tse, J.S., Klein, M.L., McDonald, I.R., 1983, J. Chem. Phys., 78, 2096.
15.Tse, J.S., Klein, M.L., McDonald, I.R., 1984, J. Chem. Phys., 81, 6124.
16.Cahil1, D.G., Pohl, R.O., 1988, Ann. Rev. Phys. Chem., 39, 93.
17.Tse, J.S., 1994, J. Incl. Phenmon., 17,259.
538
UK.
Abstract.
In this work we present an efficient method for calculation of free energies for molecular
crystals. This method is a generalization of the local harmonic approximation and allows full
wmiinate free energy minimization at finite temperatures and pressures. In terms of gas
hydrates, this method provides a first principles route to the chemical potential of water in
the hydrate lattice. This quantity has been calculated for different levels of cavity occupancy
for the type I hydrate of methane. The values obtained indicate that the number of occupied
cavities has a significant effect on the chemical potential of water. Further, we have used this
method to calculate the total free energy of methane hydrate and ice. Using the integrated
form of the equation of state for a Lennard-Jones fluid we have also calculated the free
energy of the free guest species. With these three values the methandicdmethane hydrate
three-phase co-existence line can be obtained.
I\
1. Introduction
The ability to calculate free energy in an efficient manner is of paramount importance in the
structural and thermodynamic study of gas hydrate systems. In principle, it is possible to
fully characterize the structural and thermodynamic properties of the system from a
knowledge of the free energy. In most cases, this involves calculating the structural or
thermodynamic property as a function of temperature and pressure. If this is the case, the
free energy must be calculated at many different state points. Thus the eficiency of the free
energy calculation becomes the limiting factor and determines the scale of the calculations
which can be undertaken.
The development of theories for free energy calculations on atomic solids based on
local atomic vibrational behaviour has been an important contribution in this area'. In the
Local Harmonic Model (LHM) the atoms are modelled as Einstein oscillators which vibrate
in the field created by the other atoms, but there is no interatomic vibrational coupling. The
LHM is computationally inexpensive and has been shown to give a good description of the
thermodynamic properties of atomic solids'. It should be mentioned that analogous theories
exist, notably the second moment model of Sutton2,which is equivalent to the LHM only the
nature of the approximation to the density of states differs. In this work we have further
developed the LHM for complex molecular crystals and applied the theory to gas hydrates.
At present most attempts to explain the stability and properties of gas hydrates are
based on the van der Waals and Platteeuw cell theory. According to this model the water
molecules form a wee-defined crystal lattice containing cavities into which the guest
molecules may be absorbed. The theory also assumes that the free energy of the water lattice
is independent of which molecules, if any, occupy the cavities. Thus the contribution of the
water lattice to the total free energy of the system must be the same when all the cavities are
empty as when all the cavities are occupied. Recent computer simulations by Rodger'
indicate some fundamental difficulties with the van der Waals and Platteeuw theory. His
results indicate that the empty lattice is unstable rather than metastable. If this is the w e , the
guests must serve to dampen out the critical lattice vibrations that lead to the rearrangement
of the host lattice. Tanaka" has considered distortion of hydrate cages around xenon and
carbon tetrafluoride guests. The work showed that the smaller xenon atoms did not distort
the hydrate cages, while the carbon tetrafluoride caused significant distortion of the small
cages. This deformation gave rise to a change in the water chemical potential and casts
further doubt on the validity of the primary assumption of the cell theory.
In this paper we present the extension to the LHM for molecular crystals and its
application to gas hydrates. The method provides a simple, computationally cheap tool for
the investigation of the lattice relaxation in gas hydrates and their structural and
thennodynamic properties. We present optimum cell lengths over a range of temperatures
and pressures for methane hydrate obtained using a single co-ordinate free energy
minimisation and a range of other properties. Also we present values of the free energy
difference between ice and the P-hydrate, and between the P-hydrate and the water lattice of
hydrates of various occupancies. We demonstrate how the three phase line (ice-hydratevapour) can be calculated using this method. Finally, we show the necessity for a full coordinate (i.e. all atomic co-ordinates) free energy minimisation rather than the single coordinate (i.e. unit cell length) calculation using the dissociation pressure calculated using
each method compared to the experimental value.
539
Method.
The essence of the LHh4 is the ease of calculation of the vibrational partition function, qvlb
1
q,,b = 7
1 - e- k d
which is simple to calculate from the vibrational frequency, u
a2u~ I u2 =
ar2 4nmz
The Helmholtz free energy can then be calculated from
A = -k,7'lnq,,
(1)
(2)
(3 )
In the classical limit, where k,7is much greater than hu, equation ( I ) simplifies to
k,7'
qtih
(4)
For a perfect crystal with a unit cell of N atoms. the quasi-harmonic approximation gives the
Helmholtz free energy as
where U is the potential energy and the vibrational frequency of atom i, u,, is obtained from
the dvnamical matrix with elements
as the square root of the eigenvalues of the matrix MID, where M is the mass matrix. In the
LHM, I),, is set to zero unless i and j refer to co-ordinates of the same atom. Thus D is
reduced from a 3Nx3N matrix to N 3x3 matrices and the diagonalisation becomes
substantially easier. Equation ( 5 ) is, therefore, a 3-dimensional, many-atom representation of
equation (3).
However, for a molecular system, this neglect of interatomic coupling is only valid
for atoms in different molecules. Within the same molecule the vibrations of the atoms are
strongly coupled through the presence of covalent bonds. So, in the Molecular Local
Harmonic Model (MLHM), the dynamical matrix can still be reduced to block diagonal
form, except in this case each block represents a set of molecular co-ordinates rather than a
set of atomic co-ordinates. Thus D is reduced to A4 9x9 matrices. The expression for the
Helmholtz free energy in the MLHM is
(6)
where N,is the number of atoms in molecule i. m is the number of molecules and (o,,)'
are
the eigenvalues of the molecular matrices M,"D,.
We have used the SPC model' for water and a single Lennard-Jones site for the
methane. Using a grid search method we have performed a single co-ordinate minimisation
by calculating the free energy over a range of unit cell lengths. This is a simple application
of the MLHM and provides optimum cell lengths and gradient properties such as thermal
expansivity, isobaric compressibility and heat capacities. We have also performed full
atomic co-ordinate free energy minimisation using a conjugate gradient-type approach, where
the atoms are moved according to the free energy force. For the calculation'of hydrate
dissociation pressures we have used an equation of state for a Lennard-Jones fluid6 to
describe the properties of the fluid guest.
111.
Resuits.
In figure I(a) we show the effect of temperature on the unit cell length of methane hydrate at
1 atmosphere. It is noticeable that the effect is linear and that the cell length values obtained
are comparable with those of experiments'. In figure I(b), we show the effect of pressure on
the unit cell length of methane hydrate at 260 K. Again, the effect is linear and the cell
length values are similar to experimental values'. The thermal expansivity calculated using
the values shown in figure I(a) is 1.78x10-' K-'.This compares favourably with 0 . 7 7 ~ 1 K0~
' obtained experimentally by Tse ef al.'. The compressibility calculated from figure I(b) is
3 . 3 ~ 1 0 "Pa.', which compares well with the estimate of 14x10" Pa'' given by Sloan'". The
heat capacities calculated from our work are of the order of 50 to 55 J moll K ' , which is are
a factor of 4 to 5 smaller than the experimental values of Handa".
540
Initial observations of these results suggest that there is some effect of the guest
molecules on the host lattice. The difference in Gibbs free energy between the P-hydrate and
the water lattice of the occupied hydrates is approximately I . I kJ/mol. These differences are
very similar to empirical estimates obtained by the cell theory for the free energy difference
between the hydrate water lattice and ice (1.2-1.3 kJ/mol at 273 KI2). There are also
differences between the Gibbs free energies of the water lattices of the occupied systems.
This difference is approximately 0.1 kJ/mol. On closer inspection, inclusion of guest
molecules appears to stabilise the water lattice. However, it seems that enhanced stability
and guest inclusion have a complex relationship. The fully occupied methane hydrate may
have the lowest total Gibbs free energy, but it does not have the most stable water lattice.
Occupation of the two 5 cavities has the greatest stabilising effect on the water lattice.
There seems to be some degree of variation of the effect of guest molecules on the
host water lattice with temperature and pressure. The biggest differences, around 1 . 1 kJ/mol,
are experienced at the higher temperatures and lower pressures that we have studied. These
are precisely the conditions of interest in industrial applications. Given that the magnitude of
these changes in the free energy of the water lattice, AG,,,.is comparable with A h p , it must
be expected that the accuracy of the cell theory predictions will vary with composition. At
low temperatures and higher pressures the difference is considerably smaller, about 0.2
kJ/mol.
From the full atomic co-ordinate free energy minimisation we observe similar trends,
such that the fully occupied hydrate is always the most stable hydrate, but the hydrates of
intermediate occupancy have the more stable water lattice than either the P-hydrate or the
fully occupied hydrate. The magnitude of A(&, seems to be temperature dependent, so that at
higher temperatures the difference is larger. This is a reflection of the fact that the free
energy minimisation is entropy driven. Our calculations show that AG,, can be as high as 2.3
kJ/mol at higher temperatures. Such differences in the thermodynamic properties are
currently ignored in the cell theory, which assumes that the properties of the water lattice of
occupied hydrates are the same as the P-hydrate. Thus Apa~in the cell theory ignores any
occupancy-dependent properties. Holder and Hand, for example, used an optimum value
for Apa- of I . I 1 5 kJ/mol. It is clear that occupancy-dependent changes in Ala-! of about 2.3
kJ/mol will be very significant and inclusion of such guest perturbation of the host lattice will
be necessary to correct errors experienced when using the cell theory to determine hydrate
dissociation pressures.
The MLHM, when coupled with an appropriate equation of state to describe the
thermodynamics of the bulk guest phase, can be used to calculate hydrate dissociation
pressures by calculating points on the ice/gas/hydrate phase line. We have used results from
the unit cell optimisation and the full co-ordinate minimisation in order that we may
determine whether the full co-ordinate minimisation is necessary to correctly describe gas
hydrate properties, or whether the single co-ordinate minimisation is sufficient. In figure 2,
we illustrate the dissociation pressure calculated using the unit cell optimisation. The point
where the two lines cross indicates the dissociation pressure and in this case, methane hydrate
at 270 K, it is approximately 40 MPa. In figure 3, we illustrate the case for full co-ordinate
minimisation. Here we calculate the dissociation pressure to be 2.5 M a . This is in excellent
agreement with the experimental value* of2.32 MPa. The quality of this agreement must be
contemplated given the errors within the SPC m ~ d e l although
~,
SPC water has been shown
to yield the correct melting temperatures for methane hydrateJ5.
IV. Conclusions.
In this paper we have shown the development of a simple model for free energy minimisation
of complex molecular crystals. We have applied this model to gas hydrates and shown that
calculation of thermodynamic properties using this method yields values in good comparison
to experiment. We have demonstrated the importance of lattice relaxation in the calculation
541
of some of these properties and illustrated this using the calculation of dissociation pressure
as an example.
I I.90
I Plrn
H l l X X l Plm
II.85
2g
z,
I I xo
-l
11.75
I
11.85
H Z t NK
H 2 8 l K
C~iicCell
Opiimiraiion.
-55.f
I
-.;i
2.- 9 0
c
-hl 0
Figure 3.
D i s s o c i a t i o n P r e s s u r e o f Methane H y d r a t e .
Using f u l l c o o r d i n a t e op tirn i sa t i o n.
- 77
h
i
-70
2
3
-79
-80
-81
4
0
10
,'
543
MPa
15
20
S Y N T H E S I S OF POLYCRYSTAL1,INE M E T H A N E HYDRATE, A N D I T S P H A S E
STABILITY A N D M E C H A N I C A L P R O P E R T I E S A T E L E V A T E D P R E S S U R E
Laura A. Stem, Stephen H. Kirby (both at: USGS. Menlo Park, CA 94025)
and William B. Durham (UCLLNL, Livermore, California, 94550)
Kei wordp. gas hydrote synthesis; reaction kinerics; mechunicol properties
Abstract
Test specimens of methane hydrate were grown under static conditions by combining cold,
pressunzed CHq gas with H2O ice grains, then warming the system to promote the reaction CH4
(g) + 6H2O (s+) + CHq.6HzO. Hydrate formation evidently occurs at the nascent ieeiliquid
water interface, and complete reaction was achieved by warming the system above 27 I .5 K and up
to 289 K, at 25-30 MPa, for approximately 8 hours. The resulting material is pure methane hydrate
with controlled grain size and random texture. Fabrication conditions placed the H20 ice well above
its melting temperature before reaction completed, yet samples and run records showed no evidence
for bulk melting of the ice grains. Control experiments ucing Ne, a non-hydrate-forming gas,
verified that under otherwise identical conditions, the pressure reduction and latent heat associated
with ice melting is easily detectable i n our fabncation apparatus. These results suggest that under
hydrate-forming conditions. H2O ice can persist metastably at temperatures well above its melting
point.
Methane hydrate samples were then tested in constant-stmn-rate deformation experiments at
T= 140-200 K.Pc= 50-100 MPa, and& 10-'-10-6s-l. Meawremenls in both the bnttle and
ductile fields showed that methane hydrate has measurably different strength than H2O ice, and
work hardens to a higher degree compared to other ices as well as to most metals and ceramics at
high homologous temperatures. This work hardening may be related lo a changing stoichiometry
under pressure during platic deformation: x-ray analyses showed that methane hydrate undergoes a
process of solid-state disproportionation or exsolution dunng deformauon at conditions well within
its conventional stability field
INTRODUCTION
Methane hydrate is a nonstoichiometnc compound consisting of a network of H20 molecules
that are hydrogen-bonded in a manner similar to ice and interstitially encaging CHq gas molecules
(I). Distributed globally in shallow marine and permafrost environments, methane hydrate harbors
a significant yet virtually untapped hydrocarbon source (2.3,4). Despite scientific interest in this
compound and potential commercial importance. many of the physical and material propenies of
methane hydrate are as yet poorly constrained or unmeasured, and a full understanding of these
properties will eventually be needed to turn potential energy projrcrions into practical plans for its
recovery. We have now established optimal growth parameters for efficient synthesis of methane
hydrate suitable for such testing. and have measured Ihcse samples in deformation experiments to
determine fracture and flow charactenstics. The results revealed some anomalous behavior in the
formation, plastic flow behavior, and stability of methane hydrate at elevated pressure (5).
SAMPLE SYNTHESIS
Our objective was to synthesize large-volume, cohesive, low-porosity, polycrystalline hydrate
aggregates with controlled, fine grain size and random crystallographic grain onentatton. Our
technique differs from previous studies (6). most of which involve continuous agitation of reaction
mixtures. resulting in strongly textured matenal unsuitable for materials testing. We produced
samples of virtually pure methane hydrate by the general reaction CHq (g) +6H2O (SA)
+
CH4.6H2O (s), by [he mixing and subsequent slow, regulated heating of sieved granular ice and
cold, pressurized CHq gas in an approximately constant-volume reaction vessel (Figs. I , 2, & 3A).
Sample fabrication details are as follows: CHq gas from a source bottle is initially boosted in
pressure ( P ) by a gas intensifier and routed into sample molding vessels housed in a deep freezer.
The sample assembly (Fig. 2) consists of two steel vessels immersed in an ethyl alcohol bath
initially held at freezer temperature (7) of 250 K. One vessel serves as a reservoir to store and chill
pressurized CHq gas, and !he other houses the sample mold. The mold consists of a hollow splitcylinder that encases an indium sleeve filled with 26 g of H20 ice "seed" grains, packed to 40%
porosity. Seed matenal is made from a virtually gas-free, single-crystal block of triply distilled H20
ice, ground and sieved to 180-250 pm grain size. Initially, the sample chamber with seed ice is
closed off from the reservoir and is evacuated. A loosely fitting top disk inserted on top of the
packed seed ice grains (Fig. 2) prevents displacement of the packed ice grains during evacuation.
The reservoir vessel is first charged with pressurized CHq gas to 34 MPa, and cools to 250 K.
The reservir is then opened to the pre-evacuated sample chamber, and C b pressure drops to
roughly 22 MPa. These steps serve to fill the porosity between the ice grams at a molar ratio of CHq
to H2O in the sample vessel well in excess of that required for full hydnte formation (7, 8). The
bath T is then slowly raised by means of the hot plate situated beneath the alcohol bath (Fig. 2). As
the sample and reservoir warm, they self-pressunze. Pressure increases steadily with increasing T
until reaction initiates, at which point consumption of C b gas by hydrate formation slows the rate
of P increase. Data-acquisition software (LabVIEW, National Instruments) was used to monitor and
record P and T conditions throughout each run, and the extent of reaction was determined by the
measured PCH4 offset from the reversible CH4 expansion curve.
Following full reaction, the heat source is turned off and the system slowly cools back down to
250 K. The sample chamber is then quenched i n liquid nitrogen. isolated from the reservoir,
vented, disconnected from the apparatus. and opened. The inner, hollow split-cylinder contajning
the sample is pushed from the mold and pried off the jacketed sample. Samples are then stored in
liquid nitrogen until mechanical testing.
544
\\
\\
i
I!
i!
however, grew methane hydrate on disks of melting ice to measure hydrate growth rates at constant
temperatures under static conditions. They observed two stages of methane hydrate formation, an
initial "nucleation" period during which the formation rate increased with time, followed by a
"growth" period, during which the formation rate decayed with time until no more ice remained on
the disks. Hydrate growth rates were shown not only to be determined by the rate of the supply of
the hydrate-forming species to the growth surface, but also by the rate of removal of the exothermic
heat of formation from the forming surface (14). They concluded that the onset of melting ice along
exposed surfaces not only promoted hydrate formation by providing a "template" for the formation
of hydrates, but moreover, provided a heat sink for absorbing the heat of formation during hydrate
growth. Once a rind of hydrate has encased an ice grain, the most likely process of continued
hydrate formation involves solid-state diffusion of methane gas through the hydrate shell to the ice
core (8, 15).
Our observations are in accord with the interpretations of Hwang et al. (8), and additionally we
conclude from our sample textures and run records that this surface layer of hydrate encasing each
seed ice grain not only rate limits reaction in the grain interior, but also serves to shield the ice grain
from nucleating melt by removing the existence of a free external ice surface. A similar
superheating effect has been measured in gold-plated silver single crystals, and results suggest that
either a free external surface or internal defects or dislocations are critical for melting to take place at
the normal "thermodynamic" melting point (16). In our experiments, methane hydrate may be
producing a similar effect by shielding the ice cores from nucleating melt and from establishing a
liquid-solid H20 interface, by rapid reaction of incipient melt nuclei with CHq gas to form hydrate.
We note that our method of seed ice preparation produces grains with few internal grain boundaries,
and additionally, the ice grains are likely to anneal at the warm temperatures during fabrication, thus
removing many of the internal defects for melt to nucleate on.
i
1
1
I
I
\
SUMMARY
M e t h a n e hydrate displays exceptional characteristics that merit further investigation i n t o t h e
nature a n d b e h a v i o r of this important c o m p o u n d . In t h e course of establishing o p t i m a l g r o w t h
parameters for synthesizing h y d r a t e samples suitable for rheological testing, we d e m o n s t r a t e d that
under conditions favorable t o h y d r a t e formation, t h e r a t e of H 2 0 ice melting may b e suppressed t o
allow short-lived superheating of ice t o temperatures well a b o v e its melting point. D e f o r m a t i o n tests
showed t h a t n o t o n l y does methane hydrate have a measurably different rheology than H 2 0 ice, but
that it also undergoes e x t e n s i v e work hardening accompanied by a process of solid-state
disproportionation during deformation a t conditions well within its e q u i l i b r i u m stability field. Such
unexpected consequences of m e t h a n e hydrate formation a n d deformation may affect the physical,
mechanical, a n d geochemical properties of hydrate-bearing s e d i m e n t s in ways n o t previously
appreciated. In particular, h y d r a t e instability under nonhydrostatic stress m a y affect e n v i r o n m e n t s
such as t h o s e underlying continental shelves or in associated accretionary prisms prone to regional
tectonic influences, where t h e presence of hydrates influences the strength, stability, porosity, porefluid composition, a n d migration p a t h w a y s of hydrate-cemented sediments.
b
4
hydrate (1.2 nm cubic unit cell, space group Pm3n ), constructed from 46 H 2 0 molecules and eight cavities
available for C Q gas molecules.
2. E. Sloan, CIarhrare HydraresofNarural Gases, Marcel Defier, Inc.. New York, 641 p.. 1990.
3. K. Kvenvolden. Chemical Geology, 71.41-51, 1988.
4. Because hydrates concentrate methane by a factor of 170 with respect to STP gas and as little as 1090 of the
recovered energy is required for dissociation, hydrate reservoirs are considered a substantial future energy resource; it
has been estimated that the total amount of gas in this solid form may surpass the energy content of the total
fossil fuel reserves by as much as a factor of two [(2), (3), also Claypool, G.E.. and 1. R. Kaplan, in: Kaplan. I.
R.. (ed), Natural Gases in Marine Sedimenrs, Plenum Press, New York, 99-139, 19741.
5 . L. Stern. S. Kirby, and W. Durham, Science. 273, 1843-1848, 1996.
6. R. Barrer and D. Ruzicka. Trans. Faraday SOC., 58, 2253. 1962; R. Barrer and A. Edge, Proc. Roy SOC.(London),
A300. 1 . 1967; B. Falabella and M. Vanpee I d . Eng. Chem. Fund.. 1 3 , 2 2 8 , 1974; K. Aoyogi. K. Song. R.
Kobayashi. E. Sloan, and P. Dharmawardhana. Gas Processors Assn. Research Reporr No.45. Tulsn, OK, 1980;
see Sloan (2) for full review of fabrication techniques.
7. Hwang et a1 (8)noted that for hydrate formation from melting ice, higher gas P yields higher formation rates.
Makogon ( I S ) had earlier suggested that as hydrate formation is an interfacial process. high concentrations of
hydrate-forming species are required at the interface.
8. Hwang. M.I., D.A. Wright. A. Kapur. and G . D. Holder. J . Inclusion Phenorn., 8. 103.1 16. 1990.
9. The volume of an empty structure I hydrate lattice is 16% greater than the equivalent mass of ice I [the empty
structure 1 lattice has a density of 0.78, and stoichiometric methane hydrate has a density near ice (0.90 vs 0.92 for
ice)], but there is a large -AV associated with hydrate formation due to the volume reduction of the gas phase into
the hydrate structure. Here, we start with 26 g of seed ice, and the actual molar reaction IS: 1.4H20 + 0.23 C Q
(g) + 0.23(CHq.6.1H20). The 3.8 g of CH4 uptake measured after sample synthesis confirms this hydrate
stoichiometry and is consistant with (IO). Independent measurement of C Q collected from a dissociating sample
also verified this stoichiometry (K. Kvenvolden andT. Lorenson, personal communication). While AV of the
reaction is nearly 2 1 8 . we only measure a 6.4% associated drop from the starting P due to the large volume of the
combined reservoir plus sample chamber.
IO. Gas hydrate number n varies with P; increasing P maximizes guest-molecule site occupancy. At sample
synthesis conditions (-28 MPa) n for methane hydrate should be 6.1 i 0.1, and at 100 MPa n = 5.85 iO.05 (S.
Saito, D. Marshall, and R. Kobayashi. AlCE 1.. 10, 734, 1964; also see ( I I ) , p.54.)
11. Handbook ofgas hydrare properlies and occurrence. US.DOE Publication DOWMUL9239-1546,234p., 1983.
12. The reaction CH4 (9) + 6H20 (ice) + CHq.6H20 liberates a small amount of latent heat (= 2 W . 3 kllmal at
273 K and a CH4 P of 28 MPa, determined from the Clapeyron slope (T.Makogon and E. Sloan, 1. Chem. a n d
Eng. Dara 39, 2,351-353. 1994). the measured enthalpy of formation at standard conditions (Y. Handa, Chem
Thermodynamics 18.915-921, 1986). AVr (9).and its variation with P ( l l ) . )This heat is not reflected as a T
anomaly (Fig. 5B).evidently because reaction occurs over a period of 8 hours and such heat would be small
compared with the exchange of heat of the sample with its surroundings by conduction. (The standard enthalpy
for melting of ice is -6.01 kJ1mol. or -36 HI 6 moles for comparison with the hydrate-forming reaction).
13. The importance of vigorous agitation to renew icdwater surfaces for hydrate formation was established by Villard
(P.Villard. Compr. Rend., 106, 1602, 18118) and IS also discussed by Sloan (2) and Hwang et al (8).
14. Hwang et al. (8)note that as hydrate formation is an exothermic process, the heat released by the phase change
increases the T at the formation interface. This effect is greater for hydraie formation from liquid water than from
ice since the heat of formation is partially absorbed by the melting ice.
15. Y. Makogon, Hydrares ofNarural Gases, W.H. Cielewicz Translation. PennWell Publishing, Tulsa OK, 1981
16. I. Daeges. H. Gleiter and 1. Perepezko Phys. Lcrr. 119A 79, 1986. See also: S. Phillpot J. Lutsko D.
Wolf. and S. Yip, Piys. Rev. E , 40 (5),'283-2840, i9S9. and S. Phillpot, S. Yip, and D. Wolf, Comdurers in
Physics, 3. 20-31. 1989, for further discussion of results.
17. H. Heard, W. Durham, C. Boro. and S. Kirby. in 7he Brirrle-Ducrile Transition in Rocks3Geophysical
Monograph 56. ed. by A. G . Duba et al.. American Geophysical Union, Washington. D. C.. 225-228, 1990.
18. W. Durham. S. Kirby, and L. Stern, 1. Geophys. Res. 97, E12, 20,883-20.897. 1992; also S. Kirby, w.
Durham, M. Beeman, H. Heard, and M. Daley. J. Phys., 48, suppl.. 227-232, 1987.
19. This work was supported under NASA order W-18927, and was performed in part under the auspices of the USGS
and in part by the U.S. DOE by the Lawrence Livermore National Laboratory under contract W-7405-ENG-48.
547
Run#
(step)
281 1
(K)
160
160
160
and results.
Comments
Strain hardening.
2
Strain hardening.
3
Brittle failure. ~ 2 5 %
H20 iceb.
282 I
140
Failure. multiple events.
2
140
Failure, multiple events. -25% iceb.
168 100 ----Pressurization & compaction onlyc;
366 1
0.138 --7I
Strain hardening a t
step.
185 100 3.5 x
367 1
0.215
96
90
2
185 100 3.5 x I O 4
=30% ice post-deformation.
0.185 --102
Strain hardening. 25% iceb.
168 100 3.5 x
368 1
168 100 3.5 x
0.16
--100 Identical run as 36gd.
369 1
No evolved CH4 gas.
zoo loo 3.5 10-5 0.120 --62
Strain hardenin at 10-5.
370 I
No evolved gas%.
2
200
100 3.5 x
0.230 85
80
=30% ice post-deformation.
a Pc is confining pressure gas medium; Q is total strain: ayis yield strength; oSsis steady-state strength.
Post-deformation, detcrmined by x-ray diffraction.
Samples 361, 368, 369. & 370 all underwent identical pressurization and compaction as 366 prior to testing.
Runs 369 & 370 hnd a gas collection system attached throughout testing to detect evolved CH4 gas.
___
___
Tcrnpcrature.K
123
173
366.'
m h
368
367 370
369
-90 .'
IW
80
70
..
.~
.'
I"
u
282
223
'
273
*
323
281
25
-I50
-125
-lW
-75
-50
-25
25
50
Figure 2:
Apparatus for
fabricating
cylindrical test
specimens of
methane hydrate
from CH4 gas and
melting ice. The
sample assembly is
housed in a freezer
at 250 K, and
consists of two
steel vessels
immersed in an
ethyl alcohol bath.
One vessel stores a
reservoir of cold,
pressurized CH4
gas at 35 MPa and
250 K, and the
second contains
the sample mold
with pre-jacketed
and pre-evacuated
H20 "seed ice.
Two-way valves
allow isolation of
any component of
the assembly, and a
vacuum pump
connected to the
sample chamber
permits evacuation
of the system. The sample chamber is warmed by a hot plate situated beneath the alcohol bath and
controlled remotely with a variable autotransformer. Temperature is monitored by thermocouples
enplaced in the base of the sample mold and in the bath, and pressure is measured by the gauge and
transducer. as shown. Procedures promoting methane hydrate crystallization are described in the text.
548
Figure 3:
X-ray powder diffraction
patterns for methane hydrate as
grown (A) and after mechanical
testing (B). Methane hydrate
deformed under nonhydrostatic
stress undergoes a partial solidstate disproportionation, as
evidenced by H20 ice peaks
(dotted lines) found in postdeformation x-ray diffraction
patterns.
io
ib
io
zj
is
40
io
is
28. d e p s
Figure 4
(A) Temperature-pressure profile
of sample fabrication conditions
promoting the hydrate-forming
reaction: CH4 (g) + H2O (ice) +
CHp5H20. Warming the ice + gas
mixture above the H2O solidus
(dot-dashed line, point A) initiates
reaction. Increasing temperature
slowly to 289 K, over an 8 hour
span, accelerates full reaction.
Complete reaction in our
apparatus is marked by a I .8 MPa
pressure drop (AP,) from start to
finish. Squares A-E correspond to
individual samples that were
quenched at specific intervals
during hydrate formation to
determine hydrate content as a
function of APr and time.
(B),Temperature-time profile
dunng hyrate formation. Hydrate
content (vel.%) of samples A-E
given on top scale bar, and show
how the rate of hydralc formation
decays with time under static
growth conditions.
265
275
280
285
290
Temperature,K
270
Figure 5:
(A) Temperature-pressure
record of neon gas + H20 ice
experiment demonstrates full
melting and refreezing of H2O
ice near its solidus when in the
presence of non-hydrateforming gas. The Ne + ice run
shows no net pressure drop
associated with melting and
refreezing, so start-finish
conditions are coincident.
(B) Detail of temperature-time
history of Ne (g) + H20 ice in
the region of ice melting,
showing the lag of the sample
temperature compared to the
bath temperature associated with
the absorption of heat by the
endothermic melting of ice. No
such effect is displayed by the
themal history of the methane
hydrate experiment, also shown
(grey open circles).
time, hours
280
275
M
b-
270
265:s.
'
'
"
3 '
'
' . 3.5
' '
'
time, hours
549
'
'
4'
'
'
4.5
Figure 6:
Schematic of triaxial gas
deformation apparatus set up
for methane hydrate testing at
cryogenic temperatures. The
indium-jacketed sample sits
within a cylindrical pressure
vessel in which N2 or He gas
provides the confining
medium. A sliding piston
moves through dynamic seals
from below to impose
constant axial shortening.
Samples are mounted on to a
"venting" internal force gauge
permitting sample
communication with room
conditions and allowing initial
hydrostatic pressurization to
elliminate residual porosity
prior to deformation. The gas
collection system (shown at
top) was attached during
several tests to monitor
possible loss of methane gas
during deformation.
Figure 7:
(A) Strength
measurements of methane
hydrate show that it has
measurably different
strength than H20 ice.
Ice flow constants are
from (18). Methane
hydrate data points with
arrows indicate faulting
behavior.
(B)Stress-strain curves of
deformed methane
hydrate (run 368)
compared to "standard"
polycrystalline H 2 0 ice.
While the strengths of the
two compounds are
comparable, methane
hydrate undergoes
systematic strain
hardening to an extreme
degree (over 18% strain)
while H 2 0 ice typically
displays an ultimate yield
strength followed by
relaxation to steady state
behavior.
100
h6
v
10
e!
;a
Y
E:
E?
ca
10
240
200
180
160
Temperature, K
220
140
120
140
h
120
E:
100
10
&'
40
20
0.00
0.05
0.10
0.15
0.20
Strain, %
550
h
\
I
t
INTRODUCTION
Though the existence of hydrates was demonstrated by Davy (1) in the early part of
the nineteenth century, current interest dates from 1934 when Hammerschmidt (2)
discovered that hydrates were responsible for plugging natural gas lines. This
discovery stimulated numerous studies to determine the hydrate structure and its
formation and decomposition conditions. The authors have recently employed a
temperature-ramped, isobaric (constant pressure), variable-volume technique that is
capable of providing continuous details of hydrate formation and decomposition.
Furthermore, the method enables a straightforward calculation of solubility of the
hydrate former in the host phase which may be pure water or aqueous solutions.
Information on the solubility of gases like CH4 in pure water is very. useful for the
calculation of some derived thermo-physical properties such as the enthalpy of
solution, the enthalpy of formation, and the entropy change of the solution. Gas
solubility has been extensively studied (5-12) and found to be extremely low. It has
been generally reported for temperatures above ambient. However, the same
information at the low temperatures and high pressures is very scarce. The solubility of
hydrate formers, such as methane, ethane, carbon dioxide, etc., is not easily
measurable due to the appearance of the hydrate solid phase, metastable phases, etc.
The increase of the solubility with decreasing temperature can be explained by the
formation of an ice-like structure (Le. pentagonal dodecahedra) in the solvent (13-16).
Another explanation is that displacement of solvation equilibrium occurs with changes
in temperature ( AHsolv<O ) and that the solute introduces low entropy structures in
water (15-17). The solvation of gas, R, is considered as a relaxation to the equilibrium
process described as follows:
R + nHfl = R(Hfl)n.
First introduced by Pauling (3) in 1957 and expanded in recent studies (12-18), a
concept has developed that the hydrate structure has a geometry similar to basic water
structure and the liquid forms its own "buckyballs". The 'buckyball" includes 21 water
molecules, 20 of which form a pentagonal dodecahedron with one molecule in the
middle to add stability to the cage. At ambient temperature, it was proved that these
structures are metastable and flicker in and out of existence (19). Sloan and Fleyfel (4)
in 1991 proposed a kinetic model of gas hydrate formation from ice assuming that
during the nucleation period considerable metastability occurs because of the forming
and breaking of structures. Although many more detailed studies are required, our
experiments have made it possible to detect the different steps suggested by Sloan et
al. (4).
The solubilities of methane and ethane in water at low pressures (3.45 and 0.66
MPa, respectively) have been reported earlier (20). This paper reports results for
methane obtained at higher pressures (10.48 and 13.93 MPa) in the temperature
range 291.2 to 278.2 K, which includes the hydrate formation conditions, and
divergence of these measurements from solubility predicted by Henry's law is
presented.
EXPERIMENTAL SECTION
Experimental Apparatus
551
Calorimeter
A differential calorimeter, which is a variation of the common heat flux calorimeter
(21), used in these studies consists of two symmetrical containment vessels, both
thermally insulated from the surrounding aluminum block. One containment vessel
serves as the sample cell while the other vessel serves as a reference. The
surrounding block temperature is ramped at a fixed rate, (by using heating andlor
refrigeration), allowing a steady-state heat flux between the sample vessel and the
surroundings. But the calorimeter can also be used in an 'isoperibolic" operation
where the surrounding block is held at a constant temperature (22). The calorimeter is
equipped with an internal electrical conductivity cell to track the amount of water in the
hydrate and liquid phase by monitoring the conductivity of a dilute KCI solution during
hydrate formation.
Moreover, any heat exchange between the containment vessel and the
surrounding block occurs almost exclusively by conduction and is measured by two
thermopiles. The resulting differences in voltage for the two thermopiles represent the
differential heat flux for the two containment vessels. Integrating this voltage over time
gives the total heat transfer associated with "the event." However, for this work, the
heats of dissociation have been calculated from the solubility of methane since the
thermopile values were not reliable enough to provide us with consistent data. Figure
1 shows the apparatus and the pressure-maintaining system.
Computerldata acquisition
Performance of the calorimeter strongly depends on the control and data
acquisition program for the computer. Our program is capable of handling various data
acquisitions while controlling the pressure and the temperature precisely. The
program has been written in 'Visual Basic" and can set heater load and pump position
(Le. volume of gas added to the cell during the ramping experiments). The pressure is
controlled every four seconds, and the data are collected every minute, allowing
precise control of the temperature-ramping, isobaric experiments. Measurements were
made during the data acquisition from two pressure transducers, three PRTs, a
thermopile, and an electrical conductivity device.
The stepping motor, Genrad RLC Digibridge, HP multimeter, and Keithley digital
multimeter are controlled via an IEEE interface. Pressure transducers and a
temperature controller are interfaced via an RS232. Figure 2 shows a block diagram of
the control and data acquisition systems.
Experimental Procedure
The right cell of the calorimeter shown in Figure 1 is first charged with roughly 650
grams of the dilute aqueous potassium chloride solution (-0.004 normal) prepared
with ultra-pure water (17.0 megaohm-cm resistance) and Baker Analyzed Reagent
grade salt. This solution fills approximately two thirds of the cell, in order to insure that
the electrical conductivity cell be immersed, and magnedrive propellers provide
vigorous mixing of the cell contents. After the system has been evacuated to remove
air, methane gas is introduced to the system comprising the calorimeter cell, the right
pump, and the pressure lines.
The fundamental measurement is the change in volume of the digital pump during
the temperature-ramping experiments while the pump is controlled by a stepping
motor to maintain the pressure constant. The stepping motor is actuated by a digitalbased driver which is controlled by a computer through the IEEE interface. The system
consists of the stepping motor, Digidrive, transformer, and 21 00 Indexer with an IEEE
interface. Flow rates range from 5 to 96 cclmin has been achieved. The pump is able
to add to or withdraw gas from the cell at very precise rates. But precise control of the
isobaric operation strongly depends on the pressure transducer that closes the loop.
They are rated at 10,000 psia with an accuracy of 0.07%. The displacement of the
plunger of the pump gives us the volume of methane gas added to the cell, hence the
solubility of methane in water-rich phase.
RESULTS AND DISCUSSION
Henry's law.
(2) Solubility of the gas begins to increase beyond that accounted for by Henry's
law.
(3)The gas intake by the water increases, and this point is commonly called a
catastrophic temperature(Tc). The solubility continues to increase.
(4) Catastrophic hydrate formation occurs, and the amount of solid present in the
water has drastically increases.
(5) Solidification starts but the magnetic stirrer is still running.
b) Heating
(6) Dissociation of the hydrates begins. The hydrate crystals start melting and the
volume maintains a constant value.
(7) The volume drops very fast and the hydrates are almost completely
decomposed. The volume returns to its initial value.
Figures 4 and 5 show calculated methane solubility of (log[mole fraction CH4 in
water]) at 13.93MPa (2020 psia) plotted vs. (Im and Ln (T), respectively, while Figure
6 presents the solubility of methane (1000 XCH4) vs. temperature at 10.48 MPa
(1520 psia). Table 1 provides the solubility of methane gas in pure water obtained
during the temperature-ramping experiment at a rate of 1.P"C/hr, for both pressures of
10.48 and 13.93MPa. Table 2 presents the changes of enthalpy and entropy obtained
by plotting Ln(x) versus 1 / l (T in K) from the relation of
d Ln(x) I d (1lT) = - AHIR
and Ln ( X ) vs. Ln (T) ( again, T in K) by the relation:
d Ln(x) I d In(T) =+ AS/R
553
CONCLUSION
A fully automated calorimeter with some modifications has facilitated isobaric
(constant pressure), variable-volume, and temperature-ramped experiments, and the
experimental procedure has enabled us to elucidate discrete steps involved in the
hydrate formation and decompostion for a high pressure methane-water system in a
continuous manner.
Simultaneously, the measured volumes were utilized to determine directly
methane gas solubility in the water phase in a way that minimizes uncertainties
associated with the appearance of the solid hydrate phase.
Of importance, it has been confirmed that the solubility of methane gas in water in
the vicinity of the incipient hydrate formation temperature is much greater than that
would be predicted by Henry's law, a frequently-used conventional calculation
procedure.
Finally, the obtained solubility was used to calculate derived thermo-physical
properties, i.e. changes of enthalpy and entropy of the solution.
LITERATURE CITED
1993,25,6722.
(19)Bernal, J.D. Royal Sodefy on Physics of Water and Ice published in Hydrogen
Bonding, Pergamon Press 1957.
(20)Song, K.Y.; Feneyrou, G.; Fleyfel, F.; Martin, R. Lievois, J.S.; Kobayashi, R.
Solubility Measurements of Methane and Ethane At and Near Hydrate
Conditions, in press, Fluid Phase Equilibria 1996.
(21)Calvet, E.; Prat, H. Recents Progres en Microcalorimetrie Dunod Edition Paris
1958.
(22)Lievois, J.S. Development of an Automated, High Pressure Heat flux
Calorimeter and its Application to Measure the Heat of Dissociation of
Methane Hydrate. Ph-D Dissertation, Rice University, 1987.
(23) Frank, H.S.; Wen, W.Y. Discussion Faraday Society 1957,24, 133.
(24)Himmelblau, D.M. PartialMolal Heats and Entropies of Solution for Gases
Dissolvedin Water from the Freezing Point to Near fhe Critical Point. J. Phys.
Chem. 1959,63,1803.
(25)Rettich, T.R.; Handa, Y.P.; Battino, R.; Wilhem, E. Solubilityof Gases in Liquids.
High-Precision Determinationof Henry's Constants for Methane and Ethane
in Liquid Water at 275 to 328 K J. Phys. Chem. 1981 85. 3230.
(26)CSMHYD - hydrate program developed by the Colorado School of Mines
(Dendy E. Sloan), available through the Gas Processors Association (GPA),
Tulsa, Ok.
~
554
Table I. Solubility of Methane Gas in Water (Obtained from TemperatureRamping (1.2 oC/hr),Variable-Volume,Isobaric Experiments)
sol. of cy,
XCH4 'lo00
pressure.
MFaa
@ia)
temp.,
K ("c)
10.48
(1520)
291 (18.0)
289 (16.0)
288 (15.0)
285 (12.0)
285 (11.6)
284 (11.0)
283 (10.0)
282 (9.0)
280 (7.0)
279 (6.0)
278 (5.0)
Ta=11.6"C
Tb=l6.0 "C
r =i3.80c
2.04
2.41
2.61
3.62
4.09
4.45
5.17
5.92
7.54
9.17
11.41
temp.,
K O
13.53
(2020)
292 (19.0)
290 (17.0)
289 (16.0)
289 (15.5)
288 (14.6)
287 (14.0)
286 (13.0)
285 (12.0)
283 (10.0)
280 (7.0)
279 (6.0)
278 (5.2)
4.23
5.03
6.68
11.29
6.72
25.70
37.m
44.95
52.44
298 (25.0)C
262
P=14.6"C
t=l8.WC
T =16.24"C
296
3.22
3.92
1.98
298 (25.0)'
a
b
*
c
sol. Of c y ,
XCH4 '1 OOO
pressure,
MPa a
@a)
range of temp.,
g:
change of e m y ,
AH, KcavmclaOfgaS
change of entropy,
AS, CaVmole.K
W
W
10.483
(1520)
?=ll.S"C
Tb 4 6 . 0 "C
T' =13.8 "C
16.0 to
12.0 to
11.6 to
7.0 to
12.0
11.6
7.0
5.0
-16.8
-47.7
-21.1
-32.0
686
-170.5
-74.7
-115.9
13.931
(2020)
P=14.6OC
Tb = 18.0 "C
T'= 16.24 "C
19.0 to
16.0 to
14.6 to
12.0 to
17.0
14.6
13.0
6.0
-7.1
-27.5
-87.8
46.1
-24.4
-95.2
-308.9
555
-92.3
1
,
."."3.4,
3.k
3:s
3.55
10OOlT. K-1
-2.5
-3.0
-3.5
-4.0
-4.5
-5.0
-5.5
-6.0
5.62
5.63
5.64
5.65
5.66
5.61
5.68
5.69
Lnm
Figure 5. Logarllhm of solublltty (mole traction) of methane In the water
phase vs. bgemhm of absolute temg. for a methane and water SyJtem at
13.931 MPa (2020 psla).
Temperature, *C
flgure 6. Solubility (mole fracllon) of methane in the water phase va
temp. for a methane and water system at 10.483 Upa (1520 pala).
551
558
i
!I
RESULTS
\\
I;
t
I
A
I
g.
i,
'
\ '
DISCUSSION
Crystal growth planes which are exhibited macroscopically are the slowest growing planes (fastergrowing planes grow out of existence). Studying molecular models of the SUhydrate, it is evident that the
6-membered rings of the large cavities all lie in the ( 111) planes. This suggests a hypothesis for the normal
growth habit, appealing to a presumed higher energy barrier against producing these rings compared to the
5-membered rings. The H-bonds behvm water molecules are strained more from their natural angle in
forming 6-manbemi rings than in forming 5-membered ones. We therefore believe this process is slower,
and may result in the planes colltaining these rings being the slowest growing. This hypothesis is
strmgthened when notmg that 5-membered rings sean to be naturally occming structures in water
(Rahman and S W i , 1973). This contrasts with Smelik and King (1996) who describe a mechanism
where forming of the 512 cages is viewed as the controlling factor. Our hypothesis does not transfer directly
to the SI hydrate, but a similar argument can be made about specific { 110) planes having a higher number
of hexagonal rings per unit area (although not parallel to the plane) than e.g. the (100) or the (111)
planes. No clear evidence for either ofthese hypotheses has been p r e s d to date.
The exponmtial shape of the curye for growth rate vs. supercoohg would indicate that the socalled Jackson a-factor is greater than 3 (Myerson, 1993), suggesting that the surface ofthe growing THF
hydrate is molecularly smooth, and that the growth mechanism is creation and propagation of steps on the
faces. We believe that surface nucleation is the most probable mechanism for this, as invoking screw
dislocations as the dominating factor does not explain orimtation preference and homogeneity. However,in
the very few cases where other planes than (11 1 ) are seen, screw dislocations on (111) might be invoked
as an explanationfor speeding these planes up and exhibitmg otherwise outgrown facets.
We have no completely satisfymg explanation for the 2 - d i m m s i d growth at intermediate
~~~centratiicms
of inhibitor. The question is complicated by the fact that in some cases, the edges of these
559
plates have been ideaifid as also being ( 111), and this is puzzling, as it implies that some { 111) planes
grow faster than others. Fundamentally there seems to be only two options to explain this, since just at an
edge one expects the physical conditions in the liquid Comaining the crystal to be virtually identical on the
two faces very close to their common edge. One possibility is an imperfection of some kind that stimulates
growth on one of the faces (the thin edge), and the other is some kind of timedependent adsorption efFect
for the inhibitor.
The first of these explanations has some precedent in he &ect caused by stacking faults (van de
Waal, 1996). Such faults would not be unexpected on the sU (111) planes. We do not rule out this
mechanism, but we are skeptical of it because of the complete lack of macroscopic morphological evidence
of such faults. The second possible explanation appeals to a mechanism where a fast-growingplane pushes
aside the inhibitor as it grows, quickly mot& that the polymer can not reorient and fmd its st~duralfit
and bond onto the leading surface. l l i s phenomenon has some precedence also, in the effect of kinetic
inhibitors on ice growth (Harrison et al., 1987). We do not f e d confident in choosing one of these
explanationsover the other at present, and may evmtually have to appeal to a combination of the two. The
results indicate preferential adsorption on (1 1l ) , but this question is not complaely resolved, as the
crystals also only exhibit (1 1 1 ) in their uninhibited state.
We believe that the complete growth inhibition is a result of polymer adsorption to the crystal
surface, with the adsorbed molecules ading as barriers to furlher g r d . When the concmtration is high
mot&, polymer molecules will sit closer on the surface than twice the critical radius for crystal growth at
the corresponding temperature, and the clystal will not be able to grow b e t w m the polymer strands. The
adsorption process is fairly rapid, as no measurable growth takes place atter a crystal is transferred to an
inhibited solution, and the minimum no-growth cmcmtration is probably close to the umcmtration needed
at the surface, as the difkivity of the polymer is much lower than any other component in the systm.
However, there has to be some time involved in diffusion and onmtation ofthe inhibitor, if the latter o f the
above hypotheses for 2-dimmsional growth is physically correct.
The tests with non-inhibitors show that it is not enough to have long molecules ading as difFusion
barriers or molecules with a high capacity for H-bondmg. We think that the pendant groups of our
polymers are important in achiwing strong adsorption. One possible explanation is that the pendants fit as
pseudo-guest molecules in unfinished large cavities, with extra bin@ to the surface caused by H-bonds
from the carbonyl groups on the pendants. There is some evidence from molecular simulations s d g
that this might happen (Makcgon (1997), Carver et al. (1995)). The experiments where inhibited crystals
were transferred back into uninhibited solutions, show that the adsorption is practically irreversible. Each
pmdant p u p or H-bonding site on its own probably shows equilibrium adsorption and desorption, bul in
the case where numerous sites along a polymer chain are engaged in this process, desorption of some of
them would have little influence on the overall molecule, and these sites would be kept close to their
adsorption area, and could easily readsorb. For the atire polymer to desorb, all the adsorption sites would
have to "let loose" at the same time, an event which is statistically unlikely after a certain numbs of
adsorption points has been achieved for each molecule. When desorption and further growth w a s found to
occur, it happened in areas where it is easy to imagine the polymer fit to be less than perfect: at the v d c e s
and at the interface bemeen the crystal and the glass pipettes.
The EO hydrates grown with inhibitor have not yet been studied in as much detail as the THF
systan, but the preliminary results show some parallels to the SU hydrates. PVCap and VC-713 show
differmt results than PVP, indicating that the difference in inhibition performance might be more
fundamadalthan just a differmce in degree of efFediveness. We believe that this is mainly due to the
pendant group of PVP being smaller and not having the same stabilizing &ect to provide strong
adsorption. The dramatic small-scale branching of the EO hydrate crystals with inhibitor is somRHhat
similar to what is hown in the crystallographic literature as spheruliticgrowth. However, our experiments
with constant orieaatim throughola, whereas for
indicate that these crystals are still single
spherulites, the orientation will be off by some degree for each new branch. We believe that this is a new
phmommon, as we have not been able to tind repom of such growth in the literature, and remains a topic
for firther investigation. '
ACKNOWLEDGMENTS
The foUowing companies are gratefully achowledged for their finanaal support of this work
Amoco,m c o , Chevron, Conoco, Exxcm, Mobil, 0% Petrobras, Phillips, Statoil and Te~aco.
REFERENCES
C a m r , T.J., Drew, M.G.B., Rodger, P.M.,L Chem Soc. Faraday Trans.,91(19), pp 3449-3469 (1995)
W u n d s s o n , J. and Wdmg,A., Proc. 2nd Int.C d . on Nat.Gas Hydrates, pp 415-422, Toulouse,
France, June 2-6 (19%)
H a m s o R K . ~ l e t t J . , B u r c h a m , T . S . , F e e n g . Y . Nature,
,
32, p 241 (1987)
Englezos, P., I&ECReseurch, vol. 32, pp 1251-1274 (1993)
Kvenvolden, K.A., Ann New YorkAcad Sci., vol. 75, pp 232-246 (1994)
-0%
J.P., h% J.P., Sum,A., Christiansea, R.L., Sloan, ED.,Chem Eng. Sci., 51, (1996)
thesis, Colorado School ofh%nes(1997).
Makogon, T.Y, Ph.D.
MyaOn, A.S., Handbook OfIndUstrial Ctystallizatiation,BI&IWOI~~-H~~YMNI,
Boston (1993)
Rahman, A., Stillinger, F.H.LAmChemSoc.,
,
95, pp 7943-7948 (1973)
S l m E.D.,
ClathrateHydrates ofNatural Gases, Mard-Dekker, New YO& (1990)
m for publication (1996)
Smelik, E.A. and King, H.E., Am Minemlogisf,a
van & W d , B.W.,L ofCiyskdGrowth, vol. 158, pp 153-165 (19%)
560
Figure 1
n e experimental 4.Water or water-glycol mixtures are used as coolant. The test tubes
are -2.5 on outer diameter, screwcapped Pyrex glass.
Figure 2
Octahedral s II ( 11 1} crystal of THF hydrate grown without any additives at AT= 3.4 K.
Pipette end is approximately 2 mm across.
Figure 3
Dodecahedral s I {110) crystal ofEO hydrate grown without any additives at AT= 0.5 K.
p i p e end is approximately0.2 mm across.
561
Figure 4
THF hydrate crystal growing in solution with 0.25 wt% VC-713 atAT= 2 K. Original
crystal outline is seen, with the induced 2-d plates sproutingfiom it. pipette md is
f,
approximately 2 mm across.
I
0.m
R-squand = 0.999182
0.00 .
Figure 5
,r
200
4.00
am
om
8.00
Figure 6
'
Abstract.
In this paper we present the results of our iu-sifu x-ray and neutron diffraction experiments
during the formation of gas hydrates under industry operating conditions. We have
performed energy dispersive x-ray diffraction to investigate the crystalline nature of species
formed during the hydrate formation process. These experiments have been performed on
carbon dioxide and propane hydrate. We show that Bragg peaks, indicating crystal structures,
appear during the formation process. Some of these peaks appear in the final hydrate
diffraction pattern and others do not. However, in the intermediate stages there i s a lot of
amorphous structure which could not be interpreted. In an effort to understand this part of
the formation process we have performed neutron diffraction experiments on the Small Angle
Neutron Diffraction for Amorphous and Liquid Samples (SANDALS) instrument at the
Rutherford Appleton Laboratory. This instrument is designed to look at short range structure
and provides information on the water structure around the guest molecules in the liquid
phase. We present examples of the solvation sphere around methane molecules during the
formation of methane hydrate, together with the average size and variance of the coordination sphere
Introduction.
A substantial body of information is n o d available on the equilibrium properties of clathrate
hydrates [1-5]. This data has been collected over several decades during which it has been
used to support the development of fundamental models to explain chemical and engineering
aspects of gas hydrate knowledge [6]. However, these properties and the information
currently available fail to identify the mechanisms through which gas hydrates nucleate, grow
or decompose, and in some cases fail to accurately determine the thermodynamic properties,
Hydrate research has recently concentrated on the kinetics of gas hydrate formation. The
methods for the prevention of hydrate formation have also concentrated on kinetic control
rather than thermodynamic control.
The work described in this paper uses x-ray and neutron diffraction to investigate the
structure in liquid water before and during hydrate formation. We have used the Energy
Dispersive X-Ray Diffraction (EDXD) instrument at the Daresbury Laboratory to monitor the
formation and decomposition of gas hydrates. We have also used the SANDALS (Small
Angle Neutron Diffraction of Amorphous and Liquid Samples) at the Rutherford-Appleton
Laboratory to monitor structure in the liquid phase before and during hydrate growth.
Methodoloav.
I
I
Both types of experiment have been conducted in specially designed. high pressure cells.
Temperature, pressure and composition were selected and controlled throughout the
experiments. The experiments were conducted iii-siitf, such that gas consumed during hydrate
formation is replaced by gas from the cylinder and the pressure maintained.
The x-ray diffraction experiments were carried out over a 250 to 300 K temperature
range and 0.1 to 3.5 MPa pressure range [7,8]. The detector angle was fixed at 5.042'. hisitir experiments were conducted for carbon dioxiddwater and propandwater systems,
Diffraction patterns were recorded every 200 second throughout the experiments, Data was
collected as a series of intensity IK energy spectra recorded with time throughout the
experiments.
The neutron diffraction experiments on the methanelwater system were carried out at
constant pressure (2100 psi) and the temperature varied using a ramping procedure. Two
scattering patterns were recorded at each temperature. The first stage of the experiment was
carried out at a temperature in the water/gas region of the phase diagram (291 K). The
scattering pattern was recorded over a 30 minute period. The temperature was reduced to a
p i n t within the hydrate region of the phase diagram (283 K). Further scattering patterns
were obtained. The temperature was further reduced to 277 K and subsequently reduced to
263 K were the complete sample was frozen. Any ice obtained was melted by reheating to
277 K, where the final scattering patterns were obtained.
Spherical Harmonic Analyses were performed to obtain detailed views of the local
Reverse Monte Carlo (RMC)
intermolecular orientational correlation function [9-1 I].
563
simulations were performed on one scattering pattern at each of the four temperature set
points (291 K, 283 K, 277 K (cooling) and 277 K (heating)). The original simulation cell
contained 12 methane molecules and 1500 water molecules arranged in a random
configuration. The RMC simulation were carried out over 5 million trials, using an empirical
water model. The results for the RMC simulations were deemed to be more realistic than our
previous results using a Spherical Harmonic Reconstruction [9-1 I].
The EDXD experiments lead to some interesting preliminary results. In figure 1, we show
the diffraction patterns taken during an experiment At time zero, the diffraction pattern
indicates an amorphous sample, i.r. the liquid and gas mixture. As the experiment
progresses, sharp peaks appear in the diffraction pattern. These peaks are due to Bragg
reflections from a crystalline material. The final pattern is comparable with the diffraction
pattern of the complete hydrate structure. That is, the calculated unit cell length is in
agreement with literature values and the indexed peaks correspond to Bragg reflections
obtained from the single crystal diffraction patterns.
The intermediate stages of this experiment are more difficult to interpret (see figure
2). These diffraction patterns contain a mixture of amorphous and crystalline species. At
present, we cannot determine whether Bragg peaks in these intermediate stages which do not
appear in the the final hydrate are due to a preferred orientation effect or a crystalline
intermediate, which may or may not exist in some form in the product. Methods for
interpreting these intermediate stages are currently under investigation.
The EDXD experiments are primarily designed to look at definite crystal structures, as
the x-rays are diffracted by planes of electron density in the crystal. Neutrons, on the
otherhand, are diffracted by atomic nuclei and therefore provides the possibility of
investigating short-range order in liquids. The data shown here were subjected to a spherical
harmonic analysis to m e s s t h e degree of orientational correlation of water around the
methane. The first term in this series expansion of the data is simply thccentres correlation
function, methane carbon to water oxygen. This is shown in Figure 3 for the four cases. A
pronounced co-ordination sphere is found in all instances, but changes quite abruptly for case
(4) when a significant amount of methane hydrate is formed. It can be seen from cases ( I ) (3), the co-ordination sphere peaks at an average distance of about 3.6 A for case ( I ) and then
gradually moves to larger radius values as the waterhethane system is pushed towards the
formation of hydrate by decreasing of temperature. The fits that were obtained are shown in
figures 4 and 5 .
The RMC simulations aim to reproduce the total corrected scattering pattern of the
sample. In figure 3, we show the CH,-0 pair correlation function obtained from the
scattering patterns obtained at each of the four temperatures described above. The changing
correlation function from (a) to (d) indicates an increase in the average number of water
molecules in the co-ordination sphere around methane. This is taken from the area under the
first peak in the pair correlation function in each case. The position of the peak also indicates
that the co-ordination sphere contracts in radius as the experiment progresses. These
distances have been used to extract examples of the co-ordination spheres from the final RMC
structures. The regions extracted from these structures give some indication of the water
structure around the methane molecules during each stage of the experiment. Firstly, we
should point out that the number of water molecules within the required distance of the
methane molecules is not the same for each methane at each stage of the experiment. The
area under the first peak in the pair correlation function indicates the average number of water
molecules in the first co-ordination sphere. In Table 1 we present the average and variance of
the number of water molecules in the co-ordination spheres for each stage of the experiment.
It is interesting to note that there are distorted ring tetramers, pentamers and hexamen in
these structures. It is noticeable that the average number of water molecules in the coordination spheres increases from part (a) to part (d). This is a consequence of hydrate
formation where the expected average number of water molecules in the co-ordination sphere
is 23 assuming full conversion to hydrate.
Conc Iusion s
We have demonstrated that the use of energy dispersive x-ray and small angle neutron
diffraction techniques can provide complementary information on the formation of natural
gas clathrate hydrates.
564
Acknowledgements
The authors would like to acknowledge the management and financial support of the GRI and
funding from the Rutherford-Appleton Laboratory, ISIS Facility. Thanks are also due to C.C.
Tang and A. Neild at the Daresbury Laboratory.
References
t
I
565
Figure 1:
x-ray d i f f r a c t o g r a m f o r
carbon dioxide hydrate.
\\
F i g u r e 2:
x-ray d i f f r a c t o g r a m s f o r
carbon dioxide hydrate growth.
/
1
Methane in water
0
C - 0 radial distribution
1
F i g u r e 3:
radial distribution functions for
methane c a r b o n and w a t e r o x y g e n d u r i n g
methane hydrate growth.
6-
18'C
566
[AI
- Cooling
nvary =
v [AI
kJ,3shmop
h t Date- 19-FEE-96
0956
i
.i L;
nvary =
y [A)
5.0
-2.5
2.5
00
5.0
Ikoh40hrnml
75
0956
I,
-7 S
-5.0
E
l
567
4
-2 5
0.0
2.5
5.0
7.5
Keywords: Gas Hydrate, Methane Hydrate, Phase Equilibria, Formation and Dissociation
INTRODUCTION
Large quantities of gas hydrates, clathrate compounds of water and gases formed under high
pressure and low temperature, are found in marine sedimenu and in cold regions. To produce large
amount of methane gas from the reservoir with a reasoniable way, it is nccessary to obtain
fundamentals on the mechanism of formation and dissociation and the properties under the practical
situations, including phaw equilibria and kinetics of gas hydrates. In addition, we have proposed
the displacing method of gas hydrates, so that methane gas is extracted from the reservoir by
displacing methane with carbon dioxide in molecular level. The main purposes of this research are
to acquire the further understanding for the mechanism of hydrate formation/dissociation, and also
to accumulate engineering data for completing the original concept of displacing method. We have
carried out a experiment on the formation and dissociation of methane and carbon dioxide, using an
apparatus consisting of extremely high pressure vessel and observation windows. This paper
presents the experimental results on the properties of gas hydrates formation and dissociation under
the condition of three phase equilibrium, and some discussion on kinetics and the mechanism of gas
hydrate formation.
EXPERIMENTAL
The experimental apparatus, in which the formation and dissociation of g a ~hydrate could be
observed, was designed and manufactured, so that the several conditions of pressure, temperature
and concentration of g a e s could be precisely controlled. Fig. 1 illustrates the schematic diagram of
the apparatus and the measuring system of the experiment. The pressure cell (I), made of stainless
steel and with 90 ml in internal volume, can be used under the pressure condition of up to 40 MPa.
It contains three glass windows (2) for observation, a magnetic type mixing equipment (3), and
some nozzles (4) for introducing gas and liquid component. The pressure cell is installed within a
constant temperature bath ( 5 ) filled with cooling agent of ethylene glycol, where the temperature
can be controlled with the accuracy of d . 2 C. Gas and liquid fluids are introduced into the cell
using a high pressure pump (6) and a fluid supplier system (7), and then are adjusted so that the
pressure would be kept constant or free to change by means of two cylinder type controllers (8).
The accuracy of controlled pressure is designed to be d.05MPa. Four transducer for detecting and
controlling the pressures are equipped, and five thermocouples are installed for measuring the
temperatures inside the cell as shown in Fig.1. The observation system including a optical fiber
borescope (9) and a CCD camera (10) can be utilized for this apparatus.
The procedure of the experiment on the three phase equilibrium of gas hydrate are described in the
following. First, a quantity of pure water is supplied by the high pressure pump into the pressure
cell, and other portion of the volume is filled with pressurized gas component by the fluid supplier
system. The quantities can be calculated as equivalent volume portions needed for hydrate structure,
The internal pressure can he adjusted, using both the cylinder units in liquid phase and the supplier
system in gas phase. The formation of gas hydrate is observed during the temperature of whole
system going down, while the pressure is no longer controlled at this situation. The accumulation of
hydrate would complete for about a couple of hours, after which the temperature is controlled to
go up with the constant rate for getting the equilibrium data of dissociation. The optical cell and the
spectrometric system are equipped at the glass window for detecting the nucleation and the change
of liquid phase structure. A lot of experimental data, such as temperatures, pressures and gas
concentrations, and other experimental conditions, are measured and analyzed using the data
acquisition system.
RESULTS AND DISCUSSION
The solubility of gases and gas transport into liquid phase are important factors to promote the
formation of gas hydrates in the three phase equilibrium condition. In this experiment, two types of
mixing arid bubbling operations were adopted for the promotion of nucleation. Fig2 shows
photographic figures for the formation of methane hydrate, observed through the glass window.
These results were obtained under the same condition as the pressure was approximately 10 MPa
and the temperature was 4.0 to 6.0 C. As shown in the left figure, a lot of fine fragments of
568
1
I
phase droplets of methane hydrate at the interface of gas and liquid phase. The following
experimental results are mainly related to the mixing operation, because it enables the equilibrium
properties to become more favorahle and reproductive. For keeping sufficient saturation o r
dissolved condition of gases into liquid phase, each experiment requires previous mixing operation
\
i
f
conditions, such as gas component, initial pressure and temperature, and restating hysteresis
situation. Fig.3 shows the trend curves for temperature and pressure, obtained using the gas
component of CH4 (100%). TEMPI represents the temperature measured by N o . 1 thermocouple at
the center of the cell, and TEMP2 the controlled one outside the cell. PRES means the pressure of
liquid phase inside the cell. This figure shows that the formation of gas hydrates eventually started
during the process of temperature decreasing, because slight increases of temperature appeared after
the nucleation point due to the heat of formation. The rapid decrease of pressure was observed
around the formation point, while the pressure gradually decreased a? the temperature became
lower due to the change of solubility. On the other hand, the dissociation of gas hydrate was
observed during the process of temperature increasing, from which the increase of temperature
started to decline. Fig.4 shows the trend curve of differential temperature measured by N o . I and
No.5 thermocouples as a heat balance in the experiment. The first peak on the curve indicates the
exothermic heat of formation, and the last one the endothermic heat of dissociation. According to
these relations and the observation, critical temperatures and pressures necessary for gas hydrate
formation and dissociation can he determined as in Table 1. These results include the properties for
Fig.5 shows the three phase equilibrium relation obtained through the series of experiment for COz
and CH4 hydrate. Upper two lines are the formation and dissociation equilibrium curves for CH4
gas hydrate, the relation between critical temperatures T and pressures P, and lower lines for COz
gas hydrate. In other words, gas hydrate can he present as solid phase with gas and liquid
components at higher position from curves for formation, and it no longer exists at lower position
from curves for dissociation. From the results it is clear that the relation between P and T is
approximately linear in semi-log plotting, and that the critical pressures of CH4 are relatively higher
than those of C02 assuming the same temperature condition. This suggests that the formation of
CH4 hydrate requires much higher potential o r activation energy rather than in the case of C o t
hydrate. In addition, it is found that the equilibrium curves for the formation of hydrates are placed
at upper position compared with those for the dissociation. Further, the equilibrium data measured
in case of dissociation agree well with estimated values using theoretical methods of kinetics, hut
those in case of formation largely differ from the theoretical values. These large differences may
include interesting phenomena and fundamentals on the mechanism of gas hydrate formation. Thus,
further experiments were carried out on the behavior of history and hysteresis observed the process
of formation and dissociation of gas hydrates.
i'
Fig.6 illustrates the history curve of P and T obtained in the experiment No.39 for CH4 hydrate.
The formation process proceeds on the oscillated curve between A and D, and the dissociation
process on the curve between E and F. The differences of temperature and pressure between
formation and dissociation equilibria are approximately 3.5 9: and 0.5 MPa, respectively. The
history of P and T is usually regarded to he the super cooling effect in the process of formation.
The temperature differences obtained by the experiment using COz were quite smaller than that in
case of CH4. In addition, the super cooling effect clearly appeared in the case that the g a ~
component was introduced by the way of bubbling. These results suggest thai the supper cooling
effect in the history behavior might he largely related to the interface conditions, the way of g a ~
introduction, and the component of ga?and liquid.
Fig.7 shows the hysteresis curves for demonstrating the effect of restarting situations, obtained by
continuous three experimental runs in the same conditions. The second and third runs were
restarted immediately after completing the previous run. In the first run of experiment, the
temperature of initiating formation was 15.2 C at the pressure of 17.5 MPa. However, the
formation temperature was shifted to 16.6 C in the second run, and up to 16.8 9: in the third run.
This means that the formation temperatures increase in case of restarting situations, so that they
approach the dissociation temperatures or the theoretical temperature for three phase equilibrium.
Another experimental run showed that the formation temperature of second run shifted by 2.5 to 3
9: compared with that of the first run, hut that of third run recovered to the level of first run if the
liquid with dissolved gas was left as the final condition of second run for 12 hours. These results
569
Exp.No.
EXMH09
EXMHOX
EXMH24
EXMH I2
Initial
Pressure
(MPa)
5.80
8.05
10.20
Formation Equilibrium
Pressure
(MPa)
$:?
9.4
Temperature
("c)
1 # 1
Dissociation Equilibrium
Pressure
(MPa)
%
8.6
Temperature
("c)
5.0
10.6
11.7
18.8
14.2
18.0
2.4
4.0
5.0
Dlhaua
wala **a"$,
Fig. 1 The schematic diagram of the experimental apparatus and the measuring system.
(Upper figure) Main system. (Lower figure) Pressure cell and measuring sensors.
15
16
a
n
5
0
0
30
fin
0
90
1 2 0 1 5 0 180 2 1 0 2 4 0 2 7 0 3BQ
TIPIE ( m i " )
F i g 3 The trend curves of pressure and temperature obtained in the experiment for methane gas
hydrate formation and dissociation.
2.5
k,
1.5
3
4-
I-
0.5
eat of dissoclatlon
.m
0
c
;
5 -0.5
.Yc
-1
0
60
120
180
240
300
Time (inin)
Fig.4 The change in differential temperature, and heat valance of gas hydrate formation
and dissociation.
571
Fig.5 The relations between pressure and temperature in three phase equilibrium condition
for COZand CH4 gas hydrates.
17.5
17
16.5
16
12
14
16
18
20
22
Temperature ('C)
Fig6
The history curve of pressure and temperature in the process of gas hydrate formation
and dissociation.
18
17.4
17.8
m
17.6
v
17.2
17
16.8
16.6
14
15
16
17
18
19
20
21
22
Temperature ('C)
Fig.7
The hysteresis curves of pressure and temperature in the process of gas hydrate formation
and dissociation.
572
INTRODUCTION
Gas hydrate is the species that guest molecules (CO,, CH, etc.) are included in the cage of hydrogenbonding network of water molecules. Recently, it was found that there are great quantity of methane
hydrate in the sediment below the sea bottom. Since Japan has no natural energy resources, we pay
much attention to the natural gas hydrates as one of the hopeful energy source for the future. To
establish elemental technologies for mining methane hydrate, we staned fundamental and also practical
study of gas hydrate.
The observation of real crystallization process is indispensable to study formation I dissociation
mechanism precisely. On the other hand, it is important for establish methane hydrate recovering
method to get pure hydrate crystals and measure accurate physical and chemical property of them.
To achieve these objects, we are constructing high pressure crystallization apparatus with optical
window. In this presentation, we will report the resent result of hydrate crystal formation and
recovery experiment with this apparatus.
As a fundamental research on the gas hydrate formation, we are interested in molecular clustering
structure of aqueous solution. As for the interaction of magic numbered water cluster, H+(H20),,,
which has dodecahedral structure, with organic molecules, it was observed that tetrahydrofuran (THF)
did not dissociate the hydrogen-bonding network of H+(H20)*' cluster; however, methanol
dissociated it. In the mass spectra of clusters generated from THF-water and methanol-water mixtures,
H+(H,O),, clusters contact with THF, H+(H20)21(THF)n:n=I ,2,3..., and H+(H20)21 clusters
substituted by methanol. H+(H20)21-n(CH30H)n: n=l,2,3..., were observed, respectively'). Such
molecular structures in aqueous solutions seem to be related with the nucleation mechanism of hydrate.
That is, THF makes hydrates, but methanol works as gas hydrate inhibitor. We also intend to carry
out methane hydrate formation experiment with any additives such as THF or CH30H,etc., and
discuss nucleation mechanism of gas hydrate with the clustering structure of water molecules which
were observed by cluster beam mass spectrometer.
EXPERIMENTAL
High pressure crystallization apparatus with optical vessel
Fig.1 shows the schematic diagram of the high pressure crystallization apparatus with optical
vessel. Photographs of this apparatus and main vesscl are also shown in Fig.2 and Fig.3 respectively.
As shown in Fig.1, main vessel (c 1) and sub vessel (C2) are piston-cylinder type high pressure
vessels. Inner diameter and volume of these vessels are 15 mm and 20 mB respectively. The main
vessel has a pair of sapphire windows to observe crystallization process on both side and stainless
filter to separate liquid phase from solid phase on the bottom. Main vessel is soaked in a silicon oil
bath[Temperature control range ; -3oc-+13Oc].
This oil bath also has optical wlndows and
crystallization process can be observed through these windows with microscope or multi channel
spectrophotometer. Compression of vessels is carried out by using oil pressure equipment (19). Main
vessel is connected with gas supplying system, which can supply host gases continuously under high
pressure(max. under 400kg/cm2). Control of gas flow rate is carried out by flow controller for high
flow rate (3) [0.5-5N Q/min] and low flow rate (4) [O.Ol-O.IN O/min].
crystallization unit is 4000kg/cm2.
573
Maximum pressure of
(C 1) and compressed to form hydrate crystals. The excess liquid phase is removed from the crystals
in the main vessel (C 1) to the sub vessel (C2). keeping the pressure in the main vessel constant and
slightly decreasing the pressure in the sub vessel. After separation process is finished, stop valve
between two vessels (VH1) is shut and crystals are recovered.
To confirm performance of the apparatus, crystallization and separation of pure indole from
indole/isoquinoline mixture was carried out. Then, we started hydrate formation experiment of
THF/water system and CH4/water system. In case of methane/water system, to prevent dissociation
of methane hydrate, the main vessel is cooled to about 80C by dry ice/methanol bath before the
decreasing of main vessel pressure and taking out of hydrate crystal.
high
purity
indole
crystal
by
high
pressure
Fig.4 shows the typical operation diagram of high pressure crystallization. Sample solution is
indole-iscquinoline mixture (80mol%-indole). Separation process is as follows.
I ) Inject sample mixture into main vessel ( C 1 ) and compressed to 1000kg/cm2step wise under 50
"C. In this process, pure indole crystal is formed and impurities concentrated into liquid phase
2) Hold the pressure of main vessel at 1000kg/cm2 until solid-liquid phase equilibrium is achieved.
3) Open the stop bulb ( V H l ) between main and sub vessel and remove liquid phase from the crystal.
In this process, pressure in the main vessel is constant (1000kg/cm2) all the way.
However,
as sown in Fig.4, after almost stop the piston displacement, the pressure which is indicated in
pressure gauge ( P H I ) gradually decreased and finally reach almost atmospheric pressure. This
means that almost all the liquid is removed and pressure in the main vessel can not transmitted
to pressure gauge (PH 1). At that time, crystal in main vessel is squeezed under 1000k~crn2
and remained liquid is perfectly separated.
4) Pressure of main vessel is decreased to atmospheric pressure and separated crystal is
recovered.
The punty of recovered solid was measured by gas chromatography. Indole concentration of that
crystal was 98mol% and it was known that high purity crystal can be separated by this high pressure
crystallization apparatus.
\
I
4"C,400kg/cm2. After separation of methane hydrate from water phase was finished, hydrate crystal
wa$ cooled to about -8O'c by methanol/dry ice bath. (Under -80C. the dissociation pressure of
methane hydrate become lower than atmospheric pressure. ) Then, pressure of main vessel was
decreased to atmospheric pressure and formed crystal was recovered. Photograph of recovered solid
is shown in Fig.6.
Recovered crystal is flammable white solid and dissolve with discharging
babbles. Though It thought to be a methane hydrate, detailed analyses was still not performed and
.
purity of crystal is not confirmed for the moment.
CONCLUSION
We constructed the gas hydrate formationheparation apparatus using high pressure crystallization
method and confirmed a formation and recovery of gas hydrate by this equipment. For the present,
adjustment of gas supplying unit was not finished perfectly and strict control of crystal formation is
not achieved. Also, the system for observation of hydrate crystal formation is under adjustment.
From now on, we try to improve high pressure mass controller in gas supplying unit and structure
of optical system and establish the technique for formation of pure hydrate and recover it. With the
separated gas hydrate, we are going to measure physical and chemical property of it precisely.
We are also canying out methane hydrate formation experiment with THF or CH,OH. we will also
make a discussion about influence of these additives to methane hydrate formation with the
comparison of the clustering structure of water molecules which was observed by cluster beam mass
spectrometer.
ACKNOWLEDGEMENT
This study is financially supported by NED0 (New Energy and Industrial Technology Development
Organization ).
REFERENCE
1. Yamamoto, Y and Wakisaka. A, Proc.Int.Conf.on Natural Gas Hydrate, Toulouse,
J u n e , 1996, 355
2. Yamamoto,Y, Sato,Y , Ebina, T., Yokoyama, Ch., Tak-ahashi, S., Nishiguchi, N. and
Tanabe,H., Nenryou kyoukashi ,1991, 7 0 , 533
3. Yamamot0.Y. Sato,Y, Mito,Y., Tanabe, H., Nishiguchi, N. andNagaoka, K. Proc.
Int. Symp. on Prep. of Func. Mat., 1989, 195
d
/
I
515
576
Fig.4
i000kg/cm~)
577
Depth is a key factor in ocean disposal of CO,. To avoid premature escape of the.CO, from
surface waters to the atmosphere, injection below a depth of about 500 m would be required.
From approximately 500-m to 3000-m ocean depth, undissolved C02 would exist as a buoyant
liquid. At greater depths, the liquid CO, would sink. In the absence of hydrate formation, the
minimum depth for effective CO, sequestration would be around 700 m. Drops released at this
depth would completely dissolve before reaching a depth of 500 m (1).
CO, clathrate hydrate (C02-nH20, 6 < n < 8 ) is a crystalline compound that can form under
temperature and pressure conditions associated with C Q disposal in the ocean below a 500-111
depth. The hydrate can form either as solid crystalline particles or as a coating on the surface
of liquid C02 drops. The solid hydrate particles should sink in the ocean, facilitating
sequestration; however, thin hydrate shells on liquid C q drops would limit dissolution of the
C02 and complicate sequestration attempts. During transport to depth and injection, hydrate
formation may clog submerged conduits, erode and foul injector nozzles, and negatively impact
C02 dispersion. The U.S. DOE supports experimental and theoretical research at FETC and
at UHM to address these concerns. A small high-pressure viewcell at FETC and a large
pressurized tank at UHM are being utilized in the experimental programs. Mathematical
modeling of these phenomena is also being performed at both FETC and UHM.
RESEARCH AT FETC
The work at FETC was initiated in 1993. All of the experimental work has been performed in
a high-pressure, variable-volume viewcell (HVVC) that has a maximum working volume of
about 40 cm. A description of the HVVC system and basic experimental procedures have been
published (2,3). The HVVC system can operate at temperatures down to near 0C and at
pressures up to 69 MPa (l0,OOOpsig). The HVVC system can therefore be used to simulate
ocean depths down to 6900 m. This is more than adequate for studying the behavior of CO, at
the depths currently being considered for unconfined release of C Q in the ocean (lo00 m to
1500 m) (4).
518
To enable more accurate prediction of the fate of C Q injected into the ocean, experiments at
FETC have focused on determining the relative density of the hydrate in water (2,5) and
seawater and on the formation of hydrate shells on drops of COz and their effect on drop
dissolution (2,6). With respect to the relative density of the hydrate, observations in two-phase
systems with water and either gaseous or liquid COz showed that the hydrates which formed at
the COJHzO interface were initially snowlike in appearance and buoyant in the water-rich phase.
With time, the hydrates became icelike (transparent) in appearance and sank. Trapped,
unconverted C02 may have caused the bulk density of the initially formed hydrates to be less
than that of the water-rich phase. Eventually, this trapped COz either escaped or was converted
to hydrate, causing the density to increase and the appearance to change. In contrast, when
formed from dissolved COz, the hydrates were initially icelike in appearance and sank. Buoyant
hydrate particles would frustrate sequestration in the ocean by causing the COz to rise to
unacceptably shallow depths. On the other hand, sinking hydrate particles would facilitate
sequestration by causing the COz to descend to greater depths before dissolution and thus
increase its residence time in the ocean.
Some of the above experiments with water and C Q have recently been repeated using General
Purpose Seawater (GPS) from Ocean Scientific International Ltd. As in the fresh-water
experiments, hydrates formed from COzdissolved in the seawater were icelike in appearance and
sank in the seawater-rich phase.
Observations of the rate of hydrate shell formation on C Q drops in water and seawater have
also been performed at FETC. In these experiments, a COz drop is introduced into the viewcell
and comes in contact with either existing hydrate particles or the glass or stainless steel parts of
the viewcell itself. In all such cases, hydrate shell formation began at the point where the
bubble or drop contacted the crystalline hydrate or viewcell, then rapidly spread out along the
bubble or drop surface. Others have also reported similar phenomena (7). Specific examples
from our experimental work have been previously described (6). The rate of growth of the
hydrate shell on CO, drops (0.5 cm to 1 cm in diameter) in water has been estimated at 0.5 to
1.0 cmz/s. Recent observations in seawater gave similar results.
The rate of dissolution of hydrate-covered COz drops has also been studied in water (2) and
more recently in seawater. In these experiments, the rate of decrease in drop radius was
measured. Rates in the range of 0.0045 to 0.02 cmlh have been observed for hydrate-covered
drops. These rates are slower than those obtained by other workers (2). The differences
between the results are likely due to dissimilar experimental conditions and equipment. Data
from the recent experiments in the viewcell indicate that the rate of shrinkage of C Q drops in
seawater appears to be slower than in fresh water for drops of similar size under similar
conditions. The reason for the slower rate in seawater is the topic of current investigations.
To overcome the limitations of the viewcell and more realistically simulate the environment that
a COz drop encounters in the ocean, a high-pressure water tunnel facility has been planned.
This device is patterned after a similar apparatus developed by others for the study of methane
hydrates (8). A low-pressure model is currently under construction to verify the flow patterns
in the proposed test section of the tunnel.
Recent mathematical modeling efforts at FETC have been directed at determining the thickness
of the hydrate shell that forms on COz drops under conditions expected for ocean disposal (6).
The model was developed to estimate both the thickness of the initially-formed shell and the
bounds on the ultimate thicknesses of shells after reaching steady state in saturated and
unsaturated environments. The degree of saturation is determined relative to the equilibrium
COz concentration at the hydrate equilibrium pressure, C,, at the temperature of the system.
Under anticipated ocean disposal conditions, the system can actually be oversaturated owing to
the induction period that often accompanies hydrate formation (6).
The model assumes that the ultimate thickness of the shell is governed by the diffusion of the
C 0 2 through the hydrate shell and diffusion or convection of dissolved COz away from the
hydratecovered particle. It was demonstrated that a very thin hydrate shell ( < 0.1 cm) would
initially form around drops of injected CO,. If injected into unsaturated water, a stable hydrate
thickness on the order of 1@* to lo4 times the radius of the drop would form. The model
therefore implies that the initially formed shell would become thinner in an unsaturated
environment. The thinning of the hydrate shell after formation has been experimentally observed
in the viewcell experiments and is reflected by changes in both the texture and transparency of
r"
579
hydrate-covered drops. Initially, the shell has rough texture and is opaque. Within a few
minutes the shell becomes smooth and relatively transparent.
For a CO, drop injected into saturated water, the model predicts that with time the hydrate shell
would thicken, possibly approaching 10' cm in thickness for growth periods well in excess of
100 hours. Such conditions could occur in the vicinity of the injection. Since the water is
saturated with respect to hydrate-forming conditions, the hydrate shell serves only to slow the
diffusion of C02 and thus limit the formation of additional hydrate from the injected CO,.
Results for this scenario are shown in Figure 1. The diffusivity values for CO, through the
hydrate shell, D,, used in Figure 1 are in the range of values for diffusivities in solids.
Experimental determination of this value would be required for validation of this portion of the
model.
In water oversaturated with C Q relative to CH, the shell could also thicken by addition to the
hydrate layer from the CO, dissolved in the water. This mode of growth was the subject of an
earlier paper (9). The above model also did not take into consideration the effect of varying salt
concentration at the surface of drop as hydrates form. Current modeling efforts at FETC are
directed at incorporating this effect.
RESEARCH AT UHM
A %-month research grant to investigate ocean disposal of C 0 2 was awarded by DOE to the
University of Hawaii at Manoa in August 1995. The laboratory study is being conducted by the
Hawaii Natural Energy Institute of the School of Ocean and Earth Science and Technology. The
principal objective of the study is to obtain data on liquid C Q discharge jet instability and on
drop dispersion, interactions, and dissolution under conditions representative of the deep ocean.
These data will be applied to the development and validation of predictive models to perform
(ocean)environmental hazard assessments and to devise injection methods that ensure effective
containment of the C02 from the atmosphere.
Experiments at UHM employ hunique High-pressure C02 Mixing Facility (HCMF), designed
specifically to investigate the oceanic C02disposal process. The HCMF comprises a cylindrical
pressure vessel, systems to hold and supply liquid C Q and chilled (ambient to (PC) water, and
diagnostics and data acquisition equipment. The insulated steel pressure vessel has an I.D. of
0.55 m and is 2.46-m tall. During experiments, it is partially filled with fresh or seawater and
pressurized with an inert gas to simulate conditions in the ocean down to depths of
approximately 600 m. A photograph of the pressurized test vessel is shown in Figure 2.
Numerous viewports provide access for quantitative optical probes and for flow visualization.
Details of the construction and operation of the HCMF have been reported (10).
Two types of tests will be conducted using the HCMF: (1) continuous discharging of liquid C02
through a variety of orifices over a range of conditions to study effluent breakup and injector
performance; and (2) monitoring of single CO, droplets or droplet pairs as they dissolve and
interact during simulated buoyant rise through the ocean. In the continuous discharge
experiments, a non-intrusive, laser scattering diagnostic will be employed to map droplet size
distribution spectra, velocity, and number density in the region immediately downstream from
the injector. These data on initial C02 droplet size distributions and spatial dispersion are
needed to model accurately the disposal process. The primary experimental variables include
jet velocity, simulated depth of discharge, and injector orifice size and geometry.
In the droplet dissolution experiments, a transparent diffuser will be submerged in the pressure
vessel aligned with its vertical axis. Water from the vessel is pumped downward through the
diffuser to stabilize in space buoyant liquid CQ droplets that have been bled into the diffuser.
By this means, the unrestrained rise. through the ocean can be simulated in a facility of finite
height. Close-up image acquisition and analysis will be employed to document droplet
dissolution and interaction phenomena.
In both the continuous discharge and dissolution tests, experiments will be repeated under
ambient water conditions that either foster or preclude the formation of the C02 clathrate
hydrate. As of this writing, facility preparations and diagnostic development are being
completed.
Complementary mathematical modeling to date has focused on theoretical analyses of CQ jet
instabilities and the generation of the dispersed droplet phase. The effects of hydrate formation
on these instabilities have been considered (11). Results suggest that if hydrate formation
580
kinetics proceed more rapidly than the amplification of the jet instability, then breakup may be
modified and the dispersed droplet phase size distribution altered. At the extreme, a solid
hydrate layer could grow and deposit around the injection orifice, encasing the jet and possibly
closing it off. While calculations indicate that hydrate formation effects will be restricted to
situations involving relatively large injector orifices (> l-cm diameter) and low jet velocities
(< 6 cm/s), the uncertainty in some parameters used in these calculations, notably the rate
constant for hydrate formation, warrant experimental confirmation.
Recently, the DOE-funded experimental research at UHM has attracted international interest.
Additional funds have been committed to the project by ABB Management, Ltd. in Switzerland
and Statoil (Norway) has offered the use of instrumentation and the professional assistance of
research personnel from the University of Bexgen. Interest in collaborative studies has also been
expressed by Japanese investigators.
SUMMARY
DOE-sponsored research at FETC and UHM is addressing important issues related to the effect
of hydrates on the ocean disposal of C Q . These issues include: 1) the relative density of
hydrate particles and how to either preclude hydrate formation or form particles that would sink
in the ocean and thus facilitate sequestration efforts; 2) the behavior of hydrates during formation
either as particles or as shells around liquid C02 drops, and how their formation would impact
C Q dissolution and jet dynamics; and 3) the dissolution behavior of hydrate-covered C a drops
and the fate of these drops in the ocean. Data are being sought to help develop and validate
predictive models which can be employed to identify effective sequestration strategies and to
evaluate impacts on the marine environment.
DISCLAIMER
Reference in this report to any specific product, process, or service is to facilitate understanding
and does not imply its endorsement or favoring by the United States Department of Energy.
REFERENCES
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
I
.. .
(available NTIS).
Waninski, R. P.; Cugini, A. V.; Holder, G. D. Proc. Int. Conf. Coal Sci. 1995, 2,
1931-1934.
Waninski, R. P.; Lee, C.-H.; Holder, G. D. 1. Sumrcrit. Fluids 1992, 5, 60-71.
Dioxide: WorkshoD 2 Omerod, B.; Angel, M. Ocean Storape of C&n
, IEA Greenhouse Gas R&D Programme Report, June 1996.
Holder, G. D.; Cugini, A. V.; Warzinski, R. P. Environ. Sci. Tech. 1995, B, 216-278.
Holder, G. D.; Waninski, R. P. PreDr. P a n -Am. Chem. Soc.. Div. Fuel Chem 1996,
41(4), 1452-1457.
Bumss as reported by E.D. Sloan, Jr. in International Conference on Natural Gas
Hvdrates; E.D. Sloan, Jr.; J. Happel; M.A. Hnatow, Eds.; Annals of the New York
Academy of Sciences; Vol. 715; p 17.
Maini, B. B.; Bishnoi, P. R. Chem. Enme. Sci, 1981, 3,
183-189.
Holder, G. D.; Cugini, A. V.; Warzinski, R. P. Environ. Sci.Tech. 1995,a,276-278.
Masutani, S. M.; Kinoshita, C. M.; Nihous, G. C.; Ho, T.; Vega, L. A.
W e n . M.g& 1993, 3(9-11), 865-872.
Teng, H.; Masutani, S. M.; Kinoshita, C. M.; Nihous, G. C. Prepr. Pao. -Am. Chem,
&c.. Div. Fuel Chem. 1996, 41(4), 1447-1451.
c
r
li'
581
0.025
: 0.020
v1
F:
2
2
0.015
F
4
0.010
v,
50
150
100
200
250
300
Time, h
figure 1. Thickening of the hydrate shell in a saturated reservoir as a
function of time and solid-phase diffusivity, DH (values shown in figure), and
the initial thickness of the hydrate shell (indicated at time = 0).
R'
Figure 2. Photograph of the pressurized test vessel used in the High Pressure
C q Mixing Facility at the University of Hawaii at Manoa.
582