Proceedings of the 6th International Conference on Gas Hydrates (ICGH 2008),

Vancouver, British Columbia, CANADA, July 6-10, 2008.

ANALYSES OF PRODUCTION TESTS AND MDT TESTS CONDUCTED IN

MALLIK AND ALASKA METHANE HYDRATE RESERVOIRS:
WHAT CAN WE LEARN FROM THESE WELL TESTS?

Masanori Kurihara, Kunihiro Funatsu and Hisanao Ouchi

Japan Oil Engineering Company
1-7-3 Kachidoki, Chuo-ku, Tokyo, 104-0054, Japan

Yoshihiro Masuda (School of Engineering, The University of Tokyo)

Koji Yamamoto (Japan Oil, Gas and Metals National Corporation)

Hideo Narita (National Institute of Advanced Industrial Science and Technology)

Scott R. Dallimore (Geological Survey of Canada, Natural Resources Canada)

Timothy S. Collett (United States Geological Survey)

Steve H. Hancock (APA Petroleum Engineering Inc.)

ABSTRACT
Pressure drawdown tests were conducted using Schlumbergers Modular Formation Dynamics
Tester (MDT) wireline tool in the Mallik methane hydrate (MH) reservoirs in February 2002 as
well as in the Mount Elbert (Alaska) MH reservoirs in February 2007, while a production test was
conducted applying a depressurization method in one of the Mallik MH reservoirs in April 2007.
All of these tests aimed at measuring production and bottomhole pressure (BHP) responses by
reducing BHP below the MH stability pressure to estimate reservoir properties such as
permeability and MH dissociation radius. We attempted to analyze the results of these tests
through history matching using the numerical simulator (MH21-HYDRES) coded especially for
gas hydrate reservoirs. Although the magnitude of depressurization and the total duration spent
for these tests were almost identical to each other, the simulation studies revealed that there
existed significant differences in what could be inferred and could not be inferred from test results
between a MDT test and a production test.
The simulation studies mainly clarified that (1) the MDT tests were useful to estimate initial
effective permeability in the presence of MH, (2) when BHP is reduced below the MH stability
pressure at MDT tests, the pressure and temperature responses were significantly influenced by
the wellbore storage erasing all the important data such as those indicating a radius of MH
dissociation and effective permeability after partial MH dissociation, and (3) history matching of
production tests tended to result in multiple solutions unless establishing steady flow conditions.
This paper presents the results of history matching for the typical MDT and production tests
conducted in Mallik and Alaska MH reservoirs. This paper also discusses the parameters reliably
estimated through MDT and production tests, which should provide many suggestions on future
designs and analyses of short-term tests for MH reservoirs.

Keywords: MDT, production test, numerical simulation, history matching

Corresponding author: Phone: +81 3 5548 1663 Fax: +81 3 5548 1673 Email: kurihara@joe.co.jp
NOMENCLATURE formation pressure response data in the MH
B formation volume factor [m3/m3] reservoir (the Mount Elbert stratigraphic test
c compressibility [1/Pa] well). As part of an ongoing effort to compare the
D depth [m] worlds leading gas hydrate reservoir simulators
g gravity acceleration [m/s2] including MH21-HYDRES, an international group
h thickness [m] conducted history matches of one 12-hour MDT
k absolute permeability [m2] test [7].
ke effective permeability [m2]
kr relative permeability On the other hand, the gas hydrate production test
p pressure [Pa] was conducted using the depressurization methods
q fluid production rate [m3/s] in the JOGMEC/NRCan/Aurora Mallik production
~
q fluid injection rate per unit reservoir bulk program in April 2007, aiming at the continuous
volume [m3/s] MH dissociation and production [8]. The results
r radial dimension [m] of this production test were analyzed, based on all
S fluid saturation the data acquired during the test, using MH21-
t time [s] HYDRES [9].
viscosity [Pa-s]
density [kg/m3] The effort for analyzing these MDT tests and the
porosity production test suggested the difficulties in
estimation of reservoir properties such as the
subscript effective permeability to gas and water and the
g gas radius of MH dissociation. This paper briefly
init initial review the results of the analyses for the past MDT
R rock and production tests and also discusses the
t total parameters reliably estimated through MDT tests
w water and production tests, which should provide many
well wellbore suggestions on future designs and analyses of short
term tests for MH reservoirs.
INTRODUCTION
In the Mallik 2002 Gas Hydrate Production NUMERICAL SIMULATOR
Research Well Program [1], formation tests with The Research Consortium for Methane Hydrate
the MDT tool were conducted at the Resources in Japan (MH21 Research Consortium),
JAPEX/JNOC/GSC et al. Mallik 5L-38 well to which was organized to attain the exploration and
measure the production rates from test intervals in exploitation of MH offshore Japan, has been
response to reducing the bottomhole flowing implementing a variety of research projects toward
pressure and to infer reservoir properties from the the assessment of MH resources, establishment of
flow and pressure data [2]. The test results were MH production methods and examination of the
then analyzed using conventional pressure impact of MH development on the environment.
transient test analysis methods [3], which are As part of such research projects, we have been
widely applied to analyze test results in developing the state-of-the-art numerical simulator
conventional oil and gas reservoirs [4, 5]. One of (MH21-HYDRES) for rigorously predicting MH
these MDT test results were also analyzed using dissociation and production behaviors both at core
the numerical simulator (MH21-HYDRES) coded and field scales. This simulator has a capability to
especially for gas hydrate reservoirs, which deal with 3-D, 5-phase and 4-component problems
attempted to estimate the reservoir properties associated with MH dissociation kinetics. Further
through history matching simulation [6]. details on this simulator are given in our previous
papers [10, 11, 12, 13].
Another series of MDT tests was conducted in
February 2007 by the U.S. Department of Energy, THEORY OF PRESSURE TRANSIENT TEST
BP Exploration (Alaska) and the U.S. Geological ANALYSIS
Survey, in order to collect the first open-hole
The derivation of the equations used in the ke k krg
traditional pressure transient test analysis is briefly = k rw + . (6)
t w g
summarized below [4, 5].

In general, the flow of a certain fluid in a porous Equation (3) can be analytically solved assuming a
medium is expressed rigorously as [14] constant production rate of q with appropriate
initial and boundary conditions, such as those for
kk S an infinitely acting reservoir, bounded circular
r (p + gD ) + q~ = . (1) reservoir, and constant-pressure outer boundary.
B t B
For example, in an infinitely acting reservoir, the
initial and boundary conditions are defined as
In conventional well test analysis methods,
assuming that the rock and fluid properties are p = pinit , for all r at t = 0 (initial condition) (7)
independent of time and space throughout the test
period and that the fluid flows in the single-phase p pinit , for r at t > 0
(8)
state without the effect of gravity and under a (boundary condition)
small pressure gradient, Equation (1) is simplified p qB
to [4] = , for r = rwell at t > 0
r 2khrwell . (9)
k p (boundary condition)
2 p = , (2)
ct t
The solution of Equation (3) is then given by the
where ct denotes a total compressibility given by following exponential-integral equation, when
ct = c f + cR . For the radial flow expected in a kt
25 :
ct r 2
well test, Equation (2) is further reduced to

1 p ct p qB 1 ct r 2
r = . (3) p(r , t ) = pinit Ei , (10)
r r r k t 2kh 2 4kt

For gas-water two-phase flow, which is the most where the exponential-integral is defined by
likely flow condition expected during a well test in
a MH reservoir, Equations (1) for the gas and e u
water phases are combined mathematically, Ei( x ) = du . (11)
ignoring the capillary pressure, saturation gradient
x u
and generation of gas and water, to yield
kt
When > 100 , Equation (10) is
1 p ct p ct r 2
r = , (4)
r r r ke t approximated as

t
qB kt
p(r , t ) = pinit ln + 0.80907 , (12)
2
where the total compressibility ct and the total 4kh ct r
k
mobility e are defined, neglecting the
t which is applied most popularly to the
conventional pressure transient test analysis
dissolution of gas into water, as
methods.
S w Bw S g Bg
ct = + c R = S w cw + S g c g + c R The solution of Equation (4) can be also given in
Bw p Bg p the same manner as the above. Since Equations
(5) (3) and (4) are linear and homogeneous, the
superposition of the solutions with simple
boundary conditions in time and space leads to the sets of pressure drawdown-buildup (1-hour first
solutions with the complicated boundary flow at a constant total fluid production rate of
conditions, such as those associated with the 0.173 m3/d, 3-hour first shut-in, 1-hour second
multiple rates and multiple wells. flow at a constant total fluid production rate of
0.432 m3/d and 5-hour second shut-in). These
NUMERICAL EXPERIMENTS FOR simulation-derived pressure responses are then
INVESTIGATING THE APPLICABILITY OF examined by conventional well test methods as
CONVENTIONAL WELL TEST ANALYSIS illustrated in Figure 1.
METHODS TO MDT TESTS IN MH
RESERVOIRS Model area
500 m radius in the vicinity of the
well
Although MH saturation, and hence effective Thickness 1m
permeability to fluids, significantly change along Grid system 50 x 1 x 1 (radial)
with the dissociation of MH during the test in a Grid size (r)
0.020, 0.024, 0.028, ..., 77.7 m
(max. r = 500 m)
MH reservoir, reservoir parameters such as Wellbore radius 0.108 m (8.5 inches)
permeability and distances to boundaries are Absolute
1000 mD
assumed to be constant and are estimated in permeability
accordance with the solutions of Equation (3) or Porosity 38%
Initial pressure 9.3 MPa
(4) (e.g., that given by Equation (12)), whenever Initial temperature 286 K
conventional pressure transient test analysis Table 1: Specifications of radial model for
methods are applied. Therefore, where MH numerical experiments
saturation changes are large during the course of a
well test, it is difficult to assess the reliability and Simulated
Simulated well
well test
test behavior
certainty of conventional well test analysis
methods in determining a single average value for
the well test parameters.
Radial model
A series of numerical experiments were conducted simulation (MDT)

to investigate how the solutions of Equation (3) or Conventional

Conventional well
well test analysis
(4), therefore the conventional well test analysis Reservoir
Reservoir parameters
parameters
k, kee, Shh, Sgg, Sww, Rdis
methods, could be applied to the analyses of MDT dis

test behavior in MH reservoirs. It was also Analysis results

examined through these experiments how the Comparison
k , k1 , k 2 , s, r
estimates of reservoir properties such as Figure 1: Procedure of numerical experiments
permeability and the distances to the boundary by
the conventional methods are close to those
assigned to the numerical models [6]. Analyses assuming infinitely acting model
The simulated bottomhole pressure behavior at the
Procedure of numerical experiments second shut-in was analyzed based on water
A one-dimensional radial MH reservoir model was production rates, using an infinitely acting
constructed, in accordance with the specifications reservoir model solution with a zero skin factor.
listed in Table 1, targeting the vicinity of a Figure 2 shows the log-log plots, along with the
hypothetical well where MDT tests are conducted. analytical solution curves with appropriate
The reservoir properties such as porosity, absolute permeability, for initial MH saturations of 70 and
permeability and initial pressure and temperature 90%. Note that the difference between the
were assigned to the model so as to make it similar simulated pressure and optimal analytical solution
to the Mallik reservoir where the actual MDT tests is significant in the case of the initial MH
were conducted. saturation of 90%, which indicates the limitation
of the applicability of the infinitely acting model.
The MDT test behaviors were simulated with this As summarized in Table 2, the calculated effective
model, assuming a variety of reservoir properties permeabilities to water suggested by these
such as initial MH saturation and permeability analytical curves are 3.12 and 0.05 mD for initial
reduction exponent. Pressure responses were MH saturations of 70 and 90%, respectively.
predicted in accordance with the schedule of two
 Since the bottomhole pressure behavior for initial
On the other hand, the predicted effective MH saturations of 80% or less indicates the
permeabilities to water at the short radius (where existence of a constant pressure boundary (CPB)
the minimum MH saturation is detected after the or no-flow boundary (NFB) and that for initial MH
second flow), medium radius (where the MH saturation of 90 % shows the trend diagnostic of
saturation is equal to the average MH saturation laterally decreasing permeability, these pressure
over the MH dissociation area) and long radius behaviors were analyzed using other models in an
(where the MH saturation is less than the initial attempt to find a better matched solution and
MH saturation by 1%), as listed in Table 2, information on the radius of MH dissociation. As
indicate that the effective permeability to water shown in Table 3 and Figure 3, the calculated
estimated by conventional well test analysis permeability derived from other models is not
methods is similar to that at the medium radius significantly different from that estimated by the
distance. infinitely acting reservoir model for many of the
initial MH saturations, although the analytical
solutions from some of these models show much
~ Pseudopressure change
Pseudopressure derivative
better agreement with the simulated bottomhole
Pressure derivative (kPa)

pressure behaviors as shown in Figure 3.

Pressure change (kPa)

Initial Effective Analysis results

CPB model Radial Radius
NFB model Composite model
hydrate permeability
portion (m)
saturation to gas (mD) kew (mD) R (m) kew (mD) R (m) kew (mD) R (m)
0.4 S, M, L 0.118 77.670 79.701 11.214 79.215 15.350 78.930 14.250
0.5 S, M, L 0.118 31.334 34.472 6.483 32.160 10.000 32.058 9.400
0.6 S, M, L 0.118 11.160 11.088 3.913 10.931 6.560 10.895 6.170
S 0.118 4.552
0.7 M 0.138 2.962 3.087 1.795 2.966 4.830 2.988 4.550
L 1.751 2.427
S 0.118 3.478
0.8 M 0.231 0.612 0.660 1.755 0.639 11.190 0.639 9.530
L 1.249 0.319
S 0.118 3.274
0.9 M 0.272 0.023 0.055 0.895 0.052 4.240 0.105 0.380
Equivalent time (hours) L 0.454 0.009
Abbreviations: S, short radius; M, medium radius; L, long radius; CPB, constant-pressure boundary;
(a) Initial MH saturation of 70% NFB, no-flow boundary; R, distance to boundary

Table 3: Comparison of numerical model

~ Pressure change parameters and those analyzed by various models
Pressure derivative
Pressure derivative (kPa)
Pressure change (kPa)

~ Pressure change
Pressure derivative
Pressure derivative (kPa)
Pressure change (kPa)

Equivalent time (hours)

(b) Initial MH saturation of 90%

Figure 2: Log-log plots of pressure and pressure Equivalent time (hours)
derivative with analytical solution lines
(a) Initial MH saturation of 70% (constant pressure
Model initial SH 0.7 0.9 boundary model)
condition kew (mD) 2.43 0.01
Radial portion Short Medium Long Short Medium Long ~ Pressure change
r (m) 0.1179 0.1383 1.7508 0.1179 0.2719 0.4542 Pressure derivative
SH 0.6423 0.6725 0.6988 0.6109 0.8738 0.8990
Model
Pressure derivative (kPa)
Pressure change (kPa)

parameters S g 0.0231 0.0202 0.0017 0.0884 0.0109 0.0039

after Sw 0.3346 0.3073 0.2995 0.3007 0.1153 0.0972
2nd flow k* (mD) 5.8584 3.7690 2.4782 8.9230 0.0320 0.0105
krw 0.7770 0.7859 0.9792 0.3669 0.7090 0.8630
kew=k*krw (mD) 4.5519 2.9622 2.4266 3.2739 0.0227 0.0091
Analysis k (mD) 3.1220 0.0530
results k/kew 0.6859 1.0539 1.2866 0.0162 2.3359 5.8354

Table 2: Comparison of numerical model

parameters and those analyzed by infinitely acting
model
Equivalent time (hours)

Analyses assuming other models

 (b) Initial MH saturation of 90% (composite of 0.5 m was divided into five grid layers, only the
model) centre (third) layer of which was assumed to
Figure 3: Log-log plots of pressure and pressure connect with the well by perforation. The reservoir
derivative with analytical solution lines properties, such as porosity and MH saturation,
were assigned to the model in accordance with the
For initial MH saturation of 90%, the boundary to well log interpretation results [14]. Note that the
the permeability reduction estimated by the wellbore storage effect was neglected and that the
composite model may suggest the medium or long fluid flow was assumed to stop at the sand face
radius. For initial MH saturations of 70 and 80%, instantaneously after shut-in.
the estimated distances to a CPB seem to indicate
the long radii. For initial MH saturations of less Model area 500 m in the vicinity of the well
Thickness 1.2 m
than 70%, however, the distances suggested by the Grid system 50 x 1 x 7 (radial)
models accounting for these boundaries are far 0.020, 0.024, 0.028, ..., 77.7 m
Grid size (r)
greater than those observed in the simulation (max. r = 500 m)
models and are not appropriate for inferring radii Wellbore radius 0.108 m (8.5 inches)
Absolute permeability 400 mD
of MH dissociation, as shown in Table 3. Porosity 41.5%
Permeability reduction
3
ANALYSES OF MDT TEST IN MALLIK index (N)
Initial gas hydrate
The second MDT test actually conducted on the saturation
85.1%
Mallik 5L-38 well in February 2002 [3], for the Initial water saturation 14.9%
interval between 1089.5 and 1090.0 m, was Initial pressure 11.568 MPa
Initial temperature 287.05 K
reproduced through numerical simulation to
examine how the actual well test analysis was Table 4: Specifications of radial model for history
effective at inferring MH reservoir parameters [6]. matching of Mallik MDT test behavior
This MDT test was composed of two parts, namely
a productionshut-in part and an injectionshut-in Well
part, as shown in Figure 4. The first part of the test
was selected as the target for the simulation and
examination. 0.35 m
( )
19744.2 0.338487

18299.2 0.26374
Pressure
Changes
Rate changes
Rate schedule
Schedule Perforation
Gas flow rate (x 103 m3/d)

16854.3 0.188993
in 0.50 m
Gas Flow Rate (m3/day (*1E-03) (st))

numerical
(kPa)

15409.4
Simulation 0.114247

model
Pressure (kPa)
Pressure

13964.4 0.0394999

12519.5 -0.0352468

11074.5 -0.109993 0.35 m

9629.57 -0.18474

8184.63
0.0002 0.979606 1.95901 2.93842 3.91782 4.89723
Time (hours)
Time (hours)
5.87664 6.85604 7.83545 8.81485 9.79426
-0.259487
10.7737 Figure 5: MTD test#2 performances
Figure 4: MTD test#2 performances
History matching
Note that the gas flow rate shown in Figure 4 were Figure 6 (a) depicts the bottomhole pressure
estimated based simply on the bottomhole simulated with the above model specifying the
pumping rates assuming that only gas flowed from total fluid rate (pumping rate) as a boundary
the reservoir [3]. condition, together with the actual bottomhole
pressure observed during the test. There is a
Reservoir modeling significant difference between the observed and
As shown in Table 4 and Figure 5, the two- simulated pressures, especially for the second and
dimensional radial model was constructed to third flows. The parameters of the reservoir model,
replicate the interval where MDT test#2 was especially absolute and relative permeabilities,
actually carried out. In the model, the test interval were then adjusted to reproduce the observed
bottomhole pressure [6]. As shown in Figure 6 (b),
a satisfactory agreement between observed and 0.171m 0.408m
simulated pressure behavior was accomplished.
The production rates of gas and water during the
test, and the distribution of MH saturation after the
third flow period, simulated with the history
matched model are given in Figures 7 and 8,
respectively. It was inferred that the gas flow was
not dominant at the bottomhole even during the
third flow period and that the medium and long
radii after the third flow were as small as about
0.171 and 0.408 m, respectively.

12000
Figure 8: Simulated MH saturation distribution
after third flow
11000

10000 Examination of actual analysis results

Pressure (kPa)

Measured
The parameters of the history matched model were
9000 Base case
then compared to the actual MDT test analysis
8000 results. Figure 9 shows the log-log plot of
pressure and pressure derivative for the third shut-
7000
in period, simulated with the history matched
6000
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5
model, together with the measured values.
Time (hours) Although there is a noteworthy difference between
(a) Initial run the simulated and measured values in the very
early part of the shut-in, the match could have
12000
been significantly improved if a wellbore-storage
11500 coefficient was incorporated into the simulation.
11000
The simulated pressure derivative behavior
corresponding to the medium radius is masked due
Pressure (kPa)

Measured
10500
Final case either to too small elapsed shut-in time or to the
10000 effect of partial penetration. The simulated
9500 pressure derivative at the middle part of the shut-in
9000
depicts the normal radial flow followed by the
indication of a lateral decrease in permeability
8500
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 beyond the long radius. The simulated pressure
Time (hours)
derivative drastically increases during the late part
(b) History matched of the shut-in, similar to the analytical model with
Figure 6: Simulated bottomhole pressure behavior sealing faults or laterally decreasing permeability.
40
On the other hand, the measured pressure
Water rate (m3/d at SC)
35

Gas rate (m3/d at SC) derivative plots on a unit slope line during the
30
early part of the shut-in, indicating a dominant
Production rate (m3/d)

Total rate (m3/d at RC)

25
wellbore storage effect, then deviates from this
20

RC (m3/d)
1st DD
0.000
2nd DD
0.008
3rd DD
0.188
line and becomes constant, depicting a radial flow
Gas rate
15 SC (m3/d)
3
0.000
0.290
1.071
0.642
20.882
2.072
portion of the buildup. Finally, the measured
Water rate RC (m3/d)
10 SC (m /d) 0.290 0.642 2.072
pressure derivative significantly increases during
5 the late part of the shut-in, indicating the presence
0
0 1 2 3 4 5 6
of sealing faults or laterally decreasing
Time (hours) permeability.
Figure 7: Simulated gas and water flow rates
Table 5 summarizes the reservoir parameters
analyzed using conventional well test analysis
method with sealing faults, as well as those
observed in the history matched model. For the keg (mD)
1st BU 2nd BU 3rd BU
0.029 0.067 0.100
third shut in period, the effective permeability to Analysis Lb (m) - - 0.812
water was analyzed to be about 33 mD assuming results kew (mD) 2.000 5.192 32.639
keg corrected with Q, , ct 0.000 0.00727 0.0722
100% of water flow, which is very similar to 39 Absolute permeability (mD) 20004000
mD predicted at the middle radius in the history S 0.000 0.00526 0.0912
keg (mD)
matched model. On the other hand, the effective M 0.000 0.00203 0.0184
L 0.000 0.00062 0.0045
permeability to gas was analyzed to be 0.1 mD S 8.151 49.487 190.258
Model
assuming 100% of gas flow, while that observed in parameters
kew (mD) M 8.151 22.058 39.180
L 8.151 11.885 11.735
the history matched model is much smaller. This S 0.000 0.119 0.119
difference must result from the assumption of Gas hydrate
dissociation radius (m)
M 0.000 0.143 0.171

100% gas flow and from the application of gas L 0.000 0.204 0.408
Distance to boundary (m) 7.328 7.328 7.328
properties for the fluid physical properties of Abbreviations: S, short radius; M, medium radius; L, long radius; Lb, distance to boundary
viscosity, formation volume factor and Q, production rate; , viscosity; ct, total compressibility; BU, buildup

compressibility to the analysis. The effective Table 5: Comparison between history matched
permeability to gas is corrected to be 0.07 mD by model parameters and analyzed parameters
applying the gas production rate shown in Figure 7
and the properties of the gas-water mixture, which ANALYSES OF MDT TEST IN MT ELBERT
is a good estimate of the effective permeability at MDT tests were conducted in the Mt Elbert MH
the short or medium radius obtained from the reservoirs in February 2007. The first through the
modeling. However, it should be noted that the third flow and shut-in periods of the C2 MDT
accurate estimation of gas and water rates at the experiments were selected for history matching
bottomhole is very difficult, as no fluid flows to simulation in the MH Simulator Code Comparison
the surface. Study [7]. The C2 MDT experiments involved
Since the shut-in time corresponding to the long alternating flow periods of various durations, using
radius was masked by wellbore-storage effects in a positive displacement pump, and build-up phases,
the actual test, NFB like actual pressure response during which there was no pumping, as shown in
suggests not the radius of MH dissociation but the Figure 10.
sealing faults or laterally decreasing permeability,
probably caused by the significant permeability The pressure and temperature were measured
reduction associated with the change in facies or directly during the various flow and buildup
MH characteristics. Since this distance to the periods of the MDT test, while produced fluid
boundary was calculated based on the effective volumes were not measured directly. As shown in
permeability to gas and the physical properties of Figure 10, during the first flow period the
gas in the actual test analysis, it was estimated to bottomhole pressure was kept within the MH
be as low as 0.8 m, which is much smaller than 7.3 stability region, while during the second and third
m estimated from the history matching simulation. flow periods, the bottomhole pressure was lowered
below the MH-gas-water equilibrium pressure,
10000 which led to the dissociation of MH.
Pressure and pressure derivative (kPa)

1750
Measured p
Packer Set
Wellbore-storage effect
in measured pressure behavior 1500
Initial
Hydrostatic
Measured
t (dp/dt) Final
Simulated p Hydrostatic
1250 End 1st Build-up
39.8 min
1000
End 2nd Build-up End 4th Build-up
97.6 min 60.7 min
1000 End 3rd Build-up
FBHP, psi

266.4 min
Gas Sample
Event
Log approximation invalid or
effect of partial perforation 750

Simulated Effect of outer

500 Estimated Hydrate Stability Pressure
t (dp/dt)
Lateral decrease in k* by End 4th Flow
near-well MH dissociation k* reduction 14.2 min
100 End 1st Flow End 2nd Flow End 3rd Flow
250 15.5 min 15.7 min 116.9 min
0.0001 0.001 0.01 0.1 1
rM Equivalent Time (hours)
rL
0

Figure 9: Log-log plots of pressure and pressure 0.00 2.00 4.00 6.00
Test Time, hours
8.00 10.00 12.00

derivative, comparing simulated and measured Figure 10: Measured bottomhole pressure
data
Construction of near wellbore model
Two-dimensional radial reservoir model was other hand, the bottomhole pressure and
constructed for the area of 10 m from the test well temperature were simulated specifying the fluid
with the thickness of 10 m. Thirty-six grid blocks flow of zero at the fluid inlet of the MDT tool.
with a minimum grid size (r) of 3 mm were Since the bottomhole pressure and temperature
allocated in the radial direction, incorporating two performances predicted were significantly
innermost grid blocks replicating the wellbore and different from those observed (Figure 12 (a)), the
the MDT tool so as to rigorously simulate the following parameters were tuned as matching
wellbore storage effect as shown in Figure 11. parameters in the course of history matching:
Sixty-six grid layers having a thickness of 15.24
cm were assigned for the vertical direction. The Initial effective permeability to water
initial reservoir properties such as porosity, water (irreducible water saturation for relative
saturation and MH saturation were defined to each permeability calculation)
grid layer as listed in Table 6. Rock compressibility
Relative permeability to gas
Gas volume initially dissolved in water phase
Well

intensity for MH re-formation

0.111 m
As shown in Figure 12 (b), excellent matching was
attained between simulated and measured
0.06 m
bottomhole pressure. Although the simulated
10 m 2149.5 ft
temperature depicted in this figure is slightly
2152.5 ft

MDT
different from the observed one, the temperature in
Fluid Production Point the annulus varies by location (depth); namely, the
temperature predicted at the upper part of the
MDT tool was higher than observed one due to the
1m
effect of MH re-formation, while that predicted at
Figure 11: Grid system of the near wellbore model the lower part of the MDT tool was much lower
than observed one.
Parameters Value
Modeling area 10.1 m around the well Figure 13 shows the gas and water rates at the
Thickness (m) 10.1
bottomhole predicted by the history matched
Grid system r-z radial coordinate
36 (r-direction) model, which indicates that the water production
Number of grid blocks
66 (z-direction) was dominant at the bottomhole conditions even
Horizontal absolute 1,000 (outside well) during the third flowing period. The gas and MH
permeability (mD) 100,000 (inside well)
Vertical absolute 100 (outside well) saturation distributions at the end of each flowing
permeability (mD) 100,000 (inside well) and shut-in period are depicted in Figures 14 and
29.6-36.7 (outside well) 15, respectively. It is clarified that most of gas
Porosity (%)
100.0 (inside well)
Initial pressure (MPa) 6.78 (@model center) was accumulated not in the reservoir but in the
Initial Temperature (K) 275.95 (@model center) annulus inducing the large wellbore storage effect
Initial MH 34.9-72.7 (outside well) and that the radius of MH dissociation was as
saturation (%) 0.0 (inside well)
Initial water 27.2-65.1 (outside well) small as about 5 cm from the sand face. Although
saturation (%) 0.0 (inside well)
-9
the initial effective permeability of about 0.12 mD
Rock compressibility (1/Pa) 1.0x10 was definitely inferred from the pressure data of
Gas solubility into water accounted
MH reformation neglected the first flow, other interesting parameters such as
Table 6: Reservoir model parameters the radius of MH dissociation, permeability after
MH dissociation were not indicated in the
Analysis by numerical simulation bottomhole pressure behavior due to this wellbore
Using the reservoir model constructed in the above, storage effect.
the gas and water production rates and the
bottomhole temperature were simulated for the
flow periods, specifying the observed bottomhole
pressure profile as a boundary condition. On the
 12 5.000
0.44 m
Pressure (mesured)
Pressure (calc : initial)
Temperature(mesured)
10 Temperature (calc : initial)

3.000

8
2.0 m
Pressure (MPa)

Temperature (C)
6 1.000 Wellbore Open Space

4 After (1st Flow) After (1st Shut-in)

-1.000

0 -3.000
0.000 2.000 4.000 6.000 8.000 10.000 12.000
Time (hour)

(a) Initial run

After (2nd Flow) After (2nd Shut-in)
12 5.000

Pressure (mesured)
Pressure (calc : matched)
10 Temperature(mesured)
Temperature (calc : matched)
3.000

8
Pressure (MPa)

Temperature (C)

6 1.000 After (3rd Flow) After (3rd Shut-in)

-1.000

0 -3.000
0.000 2.000 4.000 6.000 8.000 10.000 12.000
Time (hour) After (4th Flow) After (4th Shut-in)
(b) History matched Figure 14: Gas saturation distribution predicted by
Figure 12: Simulated bottomhole pressure and the history matched model
temperature behavior
0.44 m
0.06
Cumulative fluid production at reservoir (m3)

Total fluid (estimated)

0.05 Total fluid (simulated)
Gas (simulated) 2.0 m
Water (simulated)
0.04
Wellbore Open Space

0.03
After (1st Flow) After (1st Shut-in)
0.02

0.01

0
0 2 4 6 8 10 12
Time (hour)
After (2nd Flow) After (2nd Shut-in)
Figure 13: Gas and water rates at the bottomhole
predicted by the history matched model

After (3rd Flow) After (3rd Shut-in)

After (4th Flow) After (4th Shut-in)

Figure 15: MH saturation distribution predicted by
the history matched model
ANALYSES OF PRODUCTION TEST IN COMPARISON BETWEEN MDT AND
MALLIK PRODUCTION TESTS
As discussed in our paper published in this volume The analysis of pressure transient tests in MH
[9], successful history matching was accomplished reservoirs is imprecise using simplified analytical
for reproducing the behavior of 15-hour techniques and may be subject to error due to the
continuous MH dissociation and production test by large number of uncertainties, such as the
depressurization conducted on one of the Mallik distribution of MH saturation near the wellbore
MH reservoirs in April 2007. Although the total and the production rates of water and gas phases
duration spent for the test was almost the same as during the test period. Moreover, the informative
that for the above mentioned MDT tests, the radius pressure responses to the MH dissociation front at
of MH dissociation was much larger than those by the time of pressure propagation may be masked
MDT tests, which reduces the well bore storage by the wellbore storage effects and/or errors in
effect providing more insights into the mechanism semilog type pressure approximation, which
for MH dissociation and production. makes the analysis more complicated and
erroneous. Since a short-term MDT test somehow
However, since the flow conditions were not distorts the system response, a longer test is
stable in this production test, the recommended for a more representative data set.
fluctuated/scattered data measured during the test
made the analysis very complicated, leading to Taking account of the very low initial effective
many solutions and hypotheses for the MH permeability and the very small area expected for
dissociation and production mechanism. MH dissociation, the durations requested for the
pressure transient test, production test and pilot
10000 1000
test for MH reservoirs should be totally different
Gas rate (measured)
from those for a conventional oil and gas reservoir.
Cumulative gas production (m )

Gas rate (simulated)

3

8000 800
As summarized in Table 7, it may be necessary to
Gas production rate (m /d)

Cumulative gas production (measured)

3

6000
Cumulative gas production (simulated;
corrected [cum @4/3 15:00=0.0])
600
spend a couple of days only for a single MDT test.
Cumulative gas production (simulated:
before correction) For a long term production test, it must be the
4000 400
most essential to keep the bottomhole condition as
stable as possible. Bottomhole assemblies for the
2000 200 test well should be designed giving the
establishment of a stable flow the heist priority.
0 0
2007/4/2 2007/4/2 2007/4/2 2007/4/2 2007/4/3 2007/4/3 2007/4/3 2007/4/3 2007/4/3 2007/4/3 2007/4/3
12:00 PM 3:00 PM 6:00 PM 9:00 PM 12:00 AM 3:00 AM 6:00 AM 9:00 AM 12:00 PM 3:00 PM 6:00 PM
Time Duration for
Duration for MH
Well test conventional
Figure 16: Gas production predicted by the final reservoirs
reservoirs
history matched model Pressure transient
5-10 hours 1-2 days
test (RFT, MDT)
Well Sort tem flow test
15 m 1-2 days 5-10 days
(DST, etc.)
Long term
1-2 months 1-6 months
production test
Perforation interval Table 7: Test durations requested for various test
Pressure Temperature
(MPa) (K)
CONCLUSIONS
Through the numerical simulation studies for
analyzing MDT tests and production test
conducted in MH reservoirs, the following insights
MH saturation Gas saturation were obtained for the parameters learned from
F (fraction) (fraction)
pressure transient tests.
igure 17: Reservoir properties at the end of the test
predicted by he final history matched model The permeability value suggested by the
conventional well test analysis methods is
close to the average effective permeability to
a certain fluid over the region of MH
dissociation.
 When initial the initial effective permeability States Geological Survey and APA Petroleum
is large enough for the fluid flow, decrease in Engineering Inc. for their technical support.
pressure derivative at the last stage of a test
period looks indicating the presence of a REFERENCES
constant pressure boundary (CPB). [1] Dallimore SR, Collett TS, Uchida T, Weber
When initial effective permeability is very M. Overview of the science program for the
small, pressure derivative may continue to Mallik 2002 Gas Hydrate Production
increase even at the last stage of a test period Research Well Program; in Scientific Results
due probably to the insufficient test period. from the Mallik 2002 Gas Hydrate
The radius to the composite boundary may Production Research Well Program,
suggest the radius to the location in between Mackenzie Delta, Northwest Territories,
wellbore and hydrate dissociation front. Canada, In: S.R. Dallimore, T.S. Collett,
editor. Scientific Results from the Mallik
The simulation studies also suggested the 2002 Gas Hydrate Production Research Well
following as difficulties and limitations of Program, Mackenzie Delta, Northwest
MDT/production tests in MH reservoirs: Territories, Canada, Geological Survey of
Canada Bulletin 585, 2005.
Accurate flow rates of gas and water, which [2] Satoh T, Dallimore SR, Collett TS, Inoue T,
should vary with time, is unknown. Hancock SH, Moridis G, Weatherill B.
Distributions of saturation for each phase Production-test planning for the
(water, gas and MH) and hence the saturation JAPEX/JNOC/GSC et al. Mallik 5L-38 gas
dependent parameters such as the relative hydrate production research well; in
permeability to each phase, effective viscosity Scientific Results from the Mallik 2002 Gas
and total compressibility are unknown. Hydrate Production Research Well Program,
The time for the pressure propagation to the Mackenzie Delta, Northwest Territories,
MH dissociation front is very early. The Canada, In: S.R. Dallimore, T.S. Collett,
informative pressure responses at this time editor. Scientific Results from the Mallik
are muffled by the wellbore storage effect 2002 Gas Hydrate Production Research Well
and/or errors in log-log type pressure Program, Mackenzie Delta, Northwest
approximation. Territories, Canada, Geological Survey of
Other than the above, pressure behavior Canada Bulletin 585, 2005.
affected by the MH dissociation may mislead [3] Hancock SH, Dallimore SR, Collett TS, Carle
the analysis. D, Weatherill B, Satoh T, Inoue T. Overview
For a long term production test, it must be the of pressure-drawdown production-test results
most essential to keep the bottomhole for the JAPEX/JNOC/GSC et al. Mallik 5L-38
gas hydrate production research well; in
condition as stable as possible.
Scientific Results from the Mallik 2002 Gas
Hydrate Production Research Well Program,
ACKNOWLEDGMENTS
Mackenzie Delta, Northwest Territories,
This work was financially supported by the
Canada, In: S.R. Dallimore, T.S. Collett,
Research Consortium for Methane Hydrate
editor. Scientific Results from the Mallik
Resources in Japan (MH21 Research Consortium)
2002 Gas Hydrate Production Research Well
on the National Methane Hydrate Exploitation
Program, Mackenzie Delta, Northwest
Program by the Ministry of Economy, Trade and
Territories, Canada, Geological Survey of
Industry (METI). The authors gratefully
Canada Bulletin 585, 2005.
acknowledge them for the financial support and
[4] Earlougher RC Jr. Advances in Well Test
permission to present this paper. The authors wish
Analysis, Society of Petroleum Engineers of
to thank Japan Oil Engineering Company, the
AIME, Dallas, Texas, Monograph 5, 1977.
University of Tokyo, Japan Oil, Gas and Metals
[5] Sabet MA. Well Test Analysis, Gulf
National Corporation, National Institute of
Professional Publishing Co., Houston, Texas,
Advanced Industrial Science, Geological Survey
Contributions in Petroleum Geology and
of Canada, Natural Resources Canada, United
Engineering, Volume 8, 1991.
[6] Kurihara M, Funatsu K, Kusaka K, Yasuda M, Production Research Well Program,
Dallimore SR, Collett TS, Hancock SH. Well- Mackenzie Delta, Northwest Territories,
test analysis for gas hydrate reservoirs: Canada, Geological Survey of Canada
examination of parameters suggested by Bulletin 585, 2005.
conventional analysis for the [12] Masuda Y, Naganawa S, Ando S, Sato K.
JAPEX/JNOC/GSC et al. Mallik 5L-38 gas Numerical calculation of gas-production
hydrate production research well, In: S.R. performance from reservoirs containing
Dallimore, T.S. Collett, editor. Scientific natural gas hydrates. Paper SPE 38291,
Results from the Mallik 2002 Gas Hydrate Proceedings of Western Regional Meeting,
Production Research Well Program, Long Beach, California, 1997.
Mackenzie Delta, Northwest Territories, [13]Masuda Y, Fujinaga Y, Naganawa S, Fujita K,
Canada, Geological Survey of Canada Sato K, Hayash, Y. Modeling and
Bulletin 585, 2005. experimental studies on dissociation of
[7] Anderson BJ, Wilder J, Kurihara M, White M, methane gas hydrates in Berea sandstone
Moridis G, Wilson S, Pooladi-Darvish M, cores. Presented at 3rd International
Masuda M, Collett TS, Hunter R, Narita H, Conference on Gas Hydrates, Salt Lake City,
Rose K, Boswell R. Analysis of Modular Utah, 1999.
Dynamic Formation Test Results from the [14] Aziz K, Settari, A. Petroleum Reservoir
"Mount Elbert" stratigraphic test well, Milne Simulation, Applied Science Publishers Ltd.
Point, Alaska, Proceedings of the 6th (Chapman & Hall), London, United Kingdom,
International Conference on Gas Hydrates, 1979.
Vancouver, CANADA 2008.
[8] Numasawa M, Dallimore SR, Yamamoto K,
Yasuda M, Imasato Y, Mizuta T,. Kurihara M,
Masuda Y, Fujii T, Fujii K, Wright JF, Nixon
FM, Cho B, Ikegami T, Sugiyama H.
Objectives and Operation Overview of the
JOGMEC/NRCan/Aurora Mallik Gas
Hydrate Production Test, Proceedings of the
6th International Conference on Gas Hydrates,
Vancouver, CANADA 2008.
[9] Kurihara M, Funatsu K, Ouchi H, Masuda Y,
Yasuda M, Yamamoto K, Numasawa M, Fujii
T, Narita H, Dallimore SR, Wright JF.
Analysis of the JOGMEC/NRCan/Aurora
Mallik Gas Hydrate Production Test through
Numerical Simulation, Proceedings of the 6th
International Conference on Gas Hydrates,
Vancouver, CANADA 2008.
[10] Masuda Y, Konno Y, Iwama H, Kawamura T,
Kurihara M, Ouchi H. Improvement of Near
Wellbore Permeability by Methanol
Stimulation in a Methane Hydrate Production
Well, Paper OTC 19433 presented at 2008
Offshore Technology Conference held in
Houston, Texas, U.S.A., 58 May 2008.
[11] Kurihara M, Ouchi H, Inoue T, Yonezawa T,
Masuda Y, Dallimore SR, Collett TS.
Analysis of the JAPEX/JNOC/GSC et al.
Mallik 5L-38 gas hydrate thermal-production
test through numerical simulation. In: S.R.
Dallimore, T.S. Collett, editor. Scientific
Results from the Mallik 2002 Gas Hydrate