Professional Documents
Culture Documents
Summary gas accumulations underlying the hydrate layer. This will be fol-
Reservoir behavior of a hydrate-capped gas reservoir is modeled lowed by the results for production from the hydrate layer itself,
using a three-dimensional thermal reservoir simulator. The model applying heat injection in the formation.
incorporates a description of the phase behavior of the hydrates,
heat flow and compaction in the reservoir and the hydrate cap. Modeling Natural Hydrate Associated Production
The model allows the calculation of well productivity, evaluation Attempts to model the behavior of hydrate-capped gas and hy-
of well configurations and matching of experimental data. It drate reservoirs have been documented by various authors in the
shows the potentially self-sealing nature of the hydrate cap. literature. Simple energy balance approaches are used by Kuusk-
Production scenarios were also investigated for production raa and Hammershaimb et al.2 Masuda et al.,3 Yousif et al.,4 and
from the solid hydrate cap using horizontal wells and various Xu and Ruppel5 have presented numerical solutions to analytical
ways of dissociating the gas hydrates. These investigations show models. The first two of these papers do not include thermal ef-
the role of excessive water production and the requirement for fects in their calculations. Reference 5 is specifically aimed at the
water handling facilities. A data acquisition program is needed to formation phase of hydrates in the reservoir over geological times,
obtain reservoir parameters for gas hydrate accumulations. Such and is less relevant to the production phase. An attempt at explain-
parameters include relative phase permeability, heat capacity and ing the production behavior of a possibly hydrate-capped gas ac-
thermal conductivity of the hydrate-filled formations, compaction cumulation is described by Collett and Ginsburg.6 The depth and
parameters and rate of hydrate formation and decomposition in thickness of the hydrate layer under various conditions were de-
the reservoir. scribed by Holder et al.7 and by Hyndman et al.8
All these approaches apply analytical methods to explain the
subsurface occurrence and behavior of natural gas hydrates using
various simplifying assumptions.
Introduction In earlier work1 we have shown that modeling the reservoir
Interest in natural gas hydrates is increasing with foreseen require- behavior of hydrate-capped gas reservoirs with a three-
ments in the next century for large volumes of natural gas as a dimensional 共3D兲 thermal hydrocarbon reservoir simulator allows
relatively clean hydrocarbon fuel and with increasing exploration us to account for reservoir aspects, which are disregarded in most
and production operation experience in deepwater and Arctic analytical models. Such aspects include
drilling. 䊉 well inflow pressure drop and the effects of horizontal and
While progress is being made in identifying and drilling natural vertical wells in the reservoir;
gas hydrates, there is also the need to look ahead and develop 䊉 heat transfer between the reservoir fluids and the formation;
production concepts for the potentially large deposits of natural 䊉 the geothermal gradient;
gas hydrates and hydrate-capped gas reservoirs. We are now 䊉 phase behavior and pressure/volume/temperature 共PVT兲 prop-
reaching the stage in which some of the simplifying assumptions erties of the reservoir fluids as a fluction of pressure decline;
of analytical models are not sufficient any longer for developing 䊉 internal architecture and geometry of the reservoir; and
production concepts for natural gas hydrate accumulations. For 䊉 reservoir compaction effects.
SPE Reservoir Eval. & Eng. 3 共6兲, December 2000 1094-6470/2000/3共6兲/559/8/$5.00⫹0.50 559
Fig. 1–Annihilation of 500 m thick temperature disturbance be-
low the sea floor. Fig. 2–Relationship between phases and components.
TABLE 1– INPUT PARAMETERS FOR GAS DEPLETION Initialization. The model is initialized with a geothermal tem-
MODEL perature gradient of 0.03 K/m, using a temperature vs. depth table.
In the model this leads to a nonequilibrium situation, because left
Reservoir depth 1000 m
to itself the temperature will slowly equalize in the model.
Reservoir thickness 100 m
Pressure gradients are taken from the phase densities at reser-
Reservoir length 1000 m
voir pressure and temperature conditions, with an initial pressure
Initial reservoir pressure 80 bar
of 80 bar at a datum depth of 1050 meter sub sea 共mss兲. Under
Gas composition CH4 80 mol%
these conditions and with the given phase behavior, the bottom of
Gas composition C2H6 20 mol%
the hydrate stability zone or the hydrate-gas contact is at 1050
Porosity 0.2
mss. In the hydrate-capped gas model the volume of gas initially
Permeability 200 md
in place is 0.93⫻109 m3, of which 0.44⫻109 m3 is contained in
Thermal gradient 3K/100 m
the hydrate cap. The subsurface hydrate volume is 2.96
Max well prod.rate 0.5⫻106 m3/D
Water depth 800 m
⫻106 m3. The volume of free water in the reservoir 共mainly the
connate water in the gas bearing part兲 is 1.48⫻106 m3.
560 W.J.A.M. Swinkels and R.J.J. Drenth: Thermal Reservoir Model SPE Reservoir Eval. & Eng., Vol. 3, No. 6, December 2000
Fig. 3–Phase behavior diagram.
W.J.A.M. Swinkels and R.J.J. Drenth: Thermal Reservoir Model SPE Reservoir Eval. & Eng., Vol. 3, No. 6, December 2000 561
Fig. 7–Influence of additional layers on the temperature gradi-
ent after 4 years of production.
the gas bearing part of the reservoir which are added to provide
the additional heat capacity. The horizontal leg of the well is
Fig. 6–Well pressure and temperature performance during
indicated.
hydrate-capped gas depletion.
The cross section in Fig. 8 displays the overall temperature in
the reservoir after 10 years of production. The thermal gradient in
the model is clearly visible. As can be expected there is a zone in
To obtain a maximum inflow in low pressure reservoirs multi-
the model in which the hydrate is dissociating due to the pressure
lateral wells may offer the possibility to increase the effective well
drop in the gas volume below the hydrate layer. Where this dis-
length open to gas inflow, while reducing the length of horizontal
sociation takes place, heat is taken from the surrounding forma-
pipe the gas has to flow through. Whether the extra cost of drilling
tion and this cooling effect shows up as the band of lower forma-
such wells is offset by the extra productivity will depend on a
tion temperature in the model. This temperature drop in turn limits
number of mainly economic parameters.
the dissociation process at the gas-hydrate boundary, illustrating
once more that under pressure depletion conditions the rate of
Modeling Production Options
hydrate dissociation is limited by the heat flow into the reservoir.
The major use of the reservoir simulation model is in modeling
The simulation shows that dissociation of the hydrate cap
and screening production options. The spatial configuration and
above a producing hydrate-capped gas reservoir will slow down
completion of wells in a hydrate or hydrate-capped gas develop-
due to cooling at the gas-hydrate boundary, even if large amounts
ment play an important role in determining the success or failure
of heat, stored in layers directly below the reservoir, are included
of the recovery process. It is important that injection and/or pro-
in the model. This confirms our view on top seal disintegration
duction wells are designed in an optimized manner so that the
under pressure depletion, which was mentioned as a potential
reservoir energy is conserved and drainage maximized.
showstopper by BeMent et al.11
We first model the relatively simple situation of producing gas
Simulation runs with this model show that heat conduction in
from below a hydrate layer. Use of horizontal production wells,
the formation is too slow to dissociate the hydrate cap during the
which usually have much larger productivity than vertical wells, is
producing life of the reservoir. It can be expected that leaks in the
beneficial for two reasons: they allow larger flow rates at the same
hydrate cap are in principle self-sealing, since the hydrate cap will
drawdown and they increase the reservoir contact volume, poten-
be in the hydrate stability region up to the ocean bottom. This
tially exposing a larger volume of hydrates to decomposition.
effect will also play a role around producers and reduce the risk of
massive hydrate melting around the producing wells.
Depleting a Hydrate-Capped Gas Accumulation
The well performance plot 共Fig. 6兲 gives the results of the
Fig. 6 shows the performance of the well. A constant gas produc-
model with and without the five additional layers and hardly any
tion rate of 500 000 m3/D is maintained for almost 3 years, after
difference can be seen, showing that geothermal heat flow effects
which the minimum flowing bottomhole pressure of 50 bar is hit
can be neglected during the producing life of the reservoir. This
and the well goes into a sharp decline. Over this period, the res-
also illustrates the need for heat injection if large-scale hydrate
ervoir pressure declines almost linearly from 80 to 50 bar, not
dissociation is required.11
indicating sizeable pressure support from dissociating hydrates.
The flowing bottomhole temperature 共FBHT兲 of the well shows a
drop of about 1 K, due to Joule-Thomson cooling by the expand-
ing gas, a phenomenon also observed in conventional gas wells.
At the time at which the well goes in decline, lower volumes of
gas are expanding, the cooling effect will be less and, as expected,
the decline in FBHT levels off.
The observed cooling effect is beneficial for thermal behavior
of the hydrate layer around the producers, since the cooler gas
stream helps to maintain the integrity of the hydrate seal, where
this is penetrated by the wells.
Earlier analytical calculations of the Joule-Thomson cooling
effect indicated an order of magnitude of a few tenths of K/bar
pressure drop, which is in line with the current simulation results.
The pressure drop around the producing wells does not extend
into the hydrate layer, because in the model of hydrate capped gas
depletion, the hydrate layer is assumed to be fully sealing and thus
does not transmit pressure changes. Pressure distribution in the
reservoir is determined by the permeability pattern, assumed to be
homogeneous and isotropic in this case.
Fig. 7 shows the geothermal temperature distribution over a Fig. 8–Temperature distribution after depletion of hydrate-
cross section of the model with the large additional blocks below capped gas.
562 W.J.A.M. Swinkels and R.J.J. Drenth: Thermal Reservoir Model SPE Reservoir Eval. & Eng., Vol. 3, No. 6, December 2000
by a horizontal injector some 10 m below the producer. The grid-
block is refined around the wells to give improved resolution of
the pressure and temperature changes. Fig. 9 gives a cross section
of this model.
In Fig. 9 the cooling around the producing wells and the hot
fluid injected in the vertical producer are clearly visible, as is a
kind of ‘‘thermal coning’’ effect toward the producing well. This
effect is caused by slightly cooler gas moving toward the well
from the cool area higher up.
One of the major problems in simulating this situation is proper
modeling of water mobility. BeMent11 mentioned water produc-
tion as a potential showstopper that impacts the cost and complex-
ity of the development.
Water will be generated if the hydrates dissociate and consid-
erable amounts of water move through the reservoir. Obviously
the volume of water which has to be lifted will be greatly depen-
dent on the relative permeability of the water phase and also on
the vertical to horizontal permeability ratio of the formation. In
Fig. 9–Cross section of a horizontal producer in a hydrate particular, if additional water is injected in the form of steam or,
layer. even worse, hot water, then downhole gas/water separation will be
almost unavoidable for production from the layer of solid hy-
drates.
Another look at the difference between the model with and Laboratory experiments, in-situ measurements, well logs and
without additional heat capacity is the comparison of thermal gra- well tests will ultimately be required to provide data which enable
dients shown in Fig. 7. This diagram shows the distortion of the the setup of reliable models for such a specific situation.
temperature profile over the gas bearing part of the formation after
4 years of production. The only minor difference visible between Phase Distribution in the Model. Initialization of the model with
the two models is a slightly higher temperature at the bottom of solid hydrates requires assumptions of the phase distribution in
the gas bearing zone in the model with the additional heat capac- the hydrate-bearing layer. Table 2 gives an overview of the situ-
ity. The diagram also shows that the cooling around the well will ations that may occur.
not pull the well into the hydrate P, T region if it is deep enough Since we have concentrated on suboceanic hydrate occur-
共say, 50 m兲 below the hydrate cap. This supports the statement on rences, the cases that contain ice were not considered further. Of
production showstoppers 共by BeMent9兲 that hydrate formation in the remaining cases the situation in which all the phases were
the wells is a manageable problem. present seemed to be the most illustrative and a model was set up
containing hydrates, gas and liquids in the pore system. In the
Producing From the Solid Hydrate Layer model mobile gas and water are present in the hydrate stability
To investigate production from the solid hydrates a model was set zone. The same set of phases and components as that in the deple-
up of a 150 m by 90 m by 500 m block filled with hydrates in the tion case was used, i.e., a gas, a solid hydrate and a liquid phase in
pore system 共porosity of 30%兲. A horizontal producer runs across the reservoir, composed of two hydrocarbon components and wa-
the length of the block and, if injection is required, this is modeled ter. Relative permeability was defined such that gas and water are
Case Free
No. Hydrate Water Gas Ice Occurrence Consequence
W.J.A.M. Swinkels and R.J.J. Drenth: Thermal Reservoir Model SPE Reservoir Eval. & Eng., Vol. 3, No. 6, December 2000 563
Fig. 10–Performance with a producer in a hydrate layer above
the injector.
mobile down to very low saturation. Hydrates are immobile. En- Fig. 11–Saturation distribution with the injector below the pro-
thalpy in the model is treated as a component included in the ducer.
phases along with the other components.
Results of an Injection Case As in the previous injection case, the bottomhole temperature
This model, with a horizontal producer on top of a parallel injec- in the producing well first drops due to expansion cooling, then
tor, was run assuming injection of steam into the formation. This stabilizes at low production rates and finally after breakthrough of
resulted in the performance shown in Fig. 10. Various phenomena the heat front stabilizes at somewhat below the hot water injection
clearly show up in this diagram and they are the following. temperature.
The well, producing at a constant rate, goes into decline after After the breakthrough of water and heat in the producer, the
about 450 days when the minimum flowing bottomhole pressure gas and water production rates also stabilize.
is reached. Note that this time is a direct function of the volume of Because of the domal geometry, mobile gas in the hydrate layer
gas contained in the model. Expansion cooling around the well- in this model is in pressure communication with the underlying
bore then ends and the bottomhole temperature in the well stabi- free gas. For this reason some of the depletion effects of the
lizes. hydrate-capped gas case are also visible in this model, in particu-
The injector is switched on and the production decline is lar the cooling due to hydrate dissociation of the boundary layer
halted. Gas and water rates in the producer increase again. between the hydrate and the hydrate-capped gas.
Water and gas production stabilize due to lift constraints in the
well. Modeling Reservoir Charging
Because of the gas mobility and the good permeability, the
A simulation run was performed to try to model the formation of
bottomhole pressure in the well is a close reflection of the overall
hydrates by gas percolating through water-filled sediment. In the
reservoir pressure in the model.
model the gas was injected through five wells in the bottom layer
The pressure stabilizes at the minimum flowing bottomhole
of the reservoir. Although the model is only a crude approxima-
pressure.
tion, a hydrate was formed. Whether in the end an effective cap is
The bottomhole temperature in the producing well first drops
formed has not yet been simulated. The aim of these simulations
due to expansion cooling, then stabilizes at low production rates
will be, first, to try to model the experiments by de Boer et al.10 In
and finally increases when the hot injection fluid breaks through at
their laboratory experiments on core material they showed the
the producer.
formation of a very competent hydrate seal, which could with-
Fig. 11 illustrates how in this configuration the hot injection
stand a 2.4 bar pressure differential over a hydrate thickness of 5
fluid drains down into the reservoir away from the producer. This
cm.
also illustrates the impact of the assumed vertical permeability in
the model.
The performance plot in Fig. 12 of a case with an overlying
water injector shows the characteristics of this production sce-
nario.
The well, producing at a constant rate, goes into decline after
about 450 days. Expansion cooling around the wellbore then ends
and the bottomhole temperature in the well stabilizes.
The injector is switched on and the production decline is
halted. Gas and water rates in the producer increase again and a
short peak in the gas production shows up.
The water rate in the producing well increases early on to a
high level of 300 m3/D, which is temporarily higher than the
injection rate, when the water front, including water from the
dissociated hydrates, reaches the producer.
The flowing bottomhole pressure in the well is a close reflec-
tion of the overall reservoir pressure in the model. It shows the
linear pressure decline due to the production of 0.5 Mm3/D of gas.
The pressure stabilizes at the minimum flowing bottomhole Fig. 12–Performance with the producer in the hydrate layer be-
pressure. low the injector.
564 W.J.A.M. Swinkels and R.J.J. Drenth: Thermal Reservoir Model SPE Reservoir Eval. & Eng., Vol. 3, No. 6, December 2000
Discussion Dissociation of the hydrate cap caused by pressure depletion in
Further Work Required. Reservoir parameters required as input a hydrate-capped gas reservoir will be slowed down considerably
for this type of simulation are routinely available for conventional due to cooling at the gas-hydrate interface.
oil and gas reservoirs. For accurate modeling of gas hydrate and Production from solid hydrate requires large numbers of pro-
hydrate-capped reservoirs a dedicated laboratory and field data ducers, handling of considerable water volumes and large
collection program is needed to obtain the following parameters. amounts of energy input.
Reservoir and hydrate cap permeability, which determines res- For reliable modeling of hydrate-associated reservoirs, basic
ervoir flow and well productivity. Permeability in the vertical di- data measurements which accurately describe the distribution of
rection plays a role in the drainage of water, which results from natural gas hydrates in the pore system and flow of gas and water
dissociation of the hydrate cap in the reservoir. This water could in the presence of natural gas hydrates are required.
potentially drain into the wells.
Relative permeability of gas and water in the presence of hy- Nomenclature
drates.
Thermal conductivity of a hydrate-filled formation. d ⫽ thickness of the zone with disturbed temperature
Heat capacity of the hydrate filled formation. n ⫽ Corey exponent
Dissociation speed/rate of decomposition of the hydrate in the S wc ⫽ connate water saturation
formation. Although this parameter can be calculated from the
above measurements, it is desirable to obtain independent experi- Acknowledgments
mental confirmation of its value in specific formations. For experi- This work is a result of the Natural Gas Hydrates Production
ments on formation and dissociation of hydrates see the work of Team at Shell International Exploration and Production, and the
de Boer et al.10 and of Kamath et al.12 authors thank team members Owen BeMent, Klaas Bil, Ulfert
Compaction parameters, in particular pore volume compress- Klomp, Leo Roodhart, Bob Sambell, Dan White, and Charles
ibility of the reservoir. Zoons for their enthusiasm and support.
Thermal and pressure behavior of wells, both theoretical de-
scriptions and field measurements. References
Reliable production data for model calibration at a reservoir 1. Drenth, A.J.J. and Swinkels, W.J.A.M.: ‘‘A Thermal Reservoir Simu-
scale, including flow rates, pressures and temperatures as a func- lation Model of Natural Gas Hydrate Production,’’ Proc., Intl. Sym-
tion of time. posium on Methane Hydrates, organized by JNOC, Chiba 共October
Well test data. 1998兲 .
2. Kuuskraa, V.A. and Hammershaimb, E.C.: ‘‘The Energy Balance of
Model Limitations. The performance modeled here is based on a Hydrate Recovery,’’ Proc., 1984 Intl. Gas Research Conference, 63.
conceptual model of some of the processes that play a role in 3. Masuda, Y. et al.: ‘‘Numerical Calculation of Gas Production Perfor-
mance From Reservoirs Containing Natural Gas Hydrates,‘‘ paper
production from natural gas hydrates. Such a model is only the
SPE 38291 presented at the 1997 SPE Asia Pacific Oil and Gas Con-
starting point in a process to find the best development options ference and Exhibition, Kuala Lumpur, 14–16 April.
and well configurations. The simulation outcomes very much re- 4. Yousif, M.H. et al.: ‘‘Experimental and Theoretical Investigation of
flect the assumptions put into the model and in a real situation Methane-Gas-Hydrate Dissociation in Porous Media,’’ SPERE 共Feb-
some of these parameters or conditions may be different. In par- ruary 1991兲 69.
ticular phase presence and mobility in the pore system, described 5. Xu, W. and Ruppel, C.: ‘‘Predicting the Occurrence, Distribution and
in Table 2, is a crucial parameter, which should be determined Evolution of Methane Gas Hydrate in Porous Marine Sediments,’’ J.
from core and laboratory studies. Geophys. Res. 共1999兲 104, 5081.
A consequence of the choice of the presence of mobile water is 6. Collett, T.S. and Ginsburg, G.D.: ‘‘Gas Hydrates in the Messoyakha
Gas Field of the West-Siberian Basin: ‘A Re-Examination of the Geo-
that injection is modeled without hydraulic fracturing. Fracing
logic Evidence,’ ’’ Proc., Seventh Intl. Offshore and Polar Confer-
may well occur in a real injection situation and rock mechanics ence, Honolulu, Hawaii 共May 1997兲.
should then be taken into account. 7. Holder, G.D., Malone, R.D., and Lawson, W.F.: ‘‘Effects of Gas
Another situation, which may occur in nature, is the condition Composition and Geothermal Properties on the Thickness and Depth
where the hydrates are not contained in a consolidated formation of Natural-Gas-Hydrate Zones,’’ JPT 共September 1987兲 1147; Trans.,
with an established self-supporting pore system. Hydrates could AIME, 283.
well be mixed with fine silt in a slurry, which changes in compo- 8. Hyndman, R.D. et al.: ‘‘Deep Sea Bottom-Simulating-Reflectors:
sition and compacts heavily when the hydrates are removed. Such Calibration of the Base of the Hydrate Stability Field as Used For
a situation cannot be modeled with the type of simulator de- Heat Flow Estimates,‘‘ Earth Planet. Sci. Lett. 共1992兲 109, 289.
9. Regtien, J.M.M. et al.: ‘‘Interactive Reservoir Simulation,’’ paper
scribed, which is based on Darcy flow of fluids through a compe-
SPE 29146 presented at the 1995 SPE Symposium on Reservoir
tent pore system. Simulations, San Antonio, Texas, 12–15 February.
Finally, the model used for this injection case is a simple cross 10. de Boer, R.B., Houbolt, J.J.H.C., and Lagrand, J.: ‘‘Formation of Gas
section. In an actual situation the reservoir geometry may play an Hydrates in a Permeable Medium,’’ Geol. Mijnbouw 共1985兲 64, 245.
important role as well. 11. BeMent, W.O. et al.: ‘‘Are There Showstoppers to Commercial Gas
Hydrate Production?’’ Proc., Intl. Symposium on Methane Hydrates,
organized by JNOC, Chiba 共October 1998兲 .
Conclusions 12. Kamath, V.A. et al.: ‘‘Experimental Study of Brine Injection and
Depressurization Methods for Dissociation of Gas Hydrates,’’ SPEFE
Current thermal simulation tools in standard use in the oil and gas 共December 1991兲 477.
industry provide the functionality required to model pressure and
temperature behavior of hydrate and hydrate-capped reservoirs.
The simulator used here was the Shell in-house MoReS simulator, SI Metric Conversion Factors
with no specific alterations to make it suitable for hydrate simu-
bar ⫻ 1.0* E⫹05 ⫽ Pa•s
lation. Any thermal simulator with suitable PVT handling, e.g.,
using K values, could be used successfully. Obtaining correct *Conversion factor is exact. SPEREE
PVT input is not a trivial exercise, however, and it requires a
suitable PVT property simulation package.
Wim (J.A.M.) Swinkels is currently a petroleum engineer with
The same tools can in principle be used to model generation of Shell Technology E&P in Rijswijk, The Netherlands. e-mail:
in-situ hydrates over geological times. W.J.A.Swinkels@siep.shell.com. Previously, he was the Reser-
Modeling the depletion of a hydrate-capped gas reservoir voir Engineering Program Director for Shell E&P Training and
through a horizontal well shows the importance of Joule-Thomson held various petroleum engineering positions in Shell compa-
cooling around the wellbore in this type of reservoir. nies in Malaysia and The Netherlands, as well as in the Shell
W.J.A.M. Swinkels and R.J.J. Drenth: Thermal Reservoir Model SPE Reservoir Eval. & Eng., Vol. 3, No. 6, December 2000 565
Global studies team. He also held an advisory position on oil has worked for 24 years as a reservoir engineer in various po-
and gas licensing matters with the Ministry of Economic Affairs sitions for Shell in Holland, Norway, Nigeria, and the U.K. e-mail:
in The Netherlands from 1984–85. His special interests include rd@petroventures.com. His special interests include hydrate
production, horizontal wells, EOR, and acreage evaluation,
reservoir simulation and modeling, reserves and resource
especially in the Middle East. Currently, he is working as an
management, and handling uncertainty in E&P decision mak- independent consultant through Petroventures Intl. Drenth
ing. Swinkels holds an MS degree in applied mathematics from holds an MS degree in theoretical and technical physics from
Eindhoven Technical U., The Netherlands. Rik (A.J.J.) Drenth Groningen U., The Netherlands.
566 W.J.A.M. Swinkels and R.J.J. Drenth: Thermal Reservoir Model SPE Reservoir Eval. & Eng., Vol. 3, No. 6, December 2000