Vous êtes sur la page 1sur 21

Transport in Porous Media 2 (1987) 497-517 497

9 1987 by D. Reidel Publishing Company.

Aspects of Geothermal Well Test Analysis


in Fractured Reservoirs
M.J. O ' S U L L I V A N
Department of Theoretical and Applied Mechanics, University of Auckland, Private Bag,
Auckland, New Zealand

(Received in revised form: 9 March 1987)

Abstract. Several numerical experiments comparing the response of a fractured geothermal reservoir
with that of a uniform reservoir are reported. A constant rate drawdown test followed by a buildup
test is investigated for both an initially two-phase reservoir and an initially liquid reservoir which
flashes during the test. Also, cold water injection into a two-phase reservoir is considered. In all cases
the fracturing of the reservoir significantly changes the results for the uniform reservoir, and makes
interpretation of the well test results more difficult.

Key words. Geothermal, well test analysis, fractured reservoirs.

1. Introduction
For isothermal reservoirs constant rate drawdown or injection tests are com-
monly analysed by means of semi-log plots of pressure versus time. The well-
known Theis solution shows that after a short initial period, such a plot asymp-
totically approaches a straight line whose slope is inversely proportional to the
mobility-thickness product of the producing layer. For drawdown/buildup or
injection/falloff tests, Theis solutions can be superimposed to give a theoretical
pressure response which again is asymptotically linear for a plot of pressure drop
versus log(t+At)/At. Here t is the shut-in time and At the time elapsed since
shut-in. These plots, commonly called Horner plots, also have a slope inversely
proportional to the mobility-thickness product for the reservoir.
For geothermal well tests the reservoir fluid is not isothermal with constant
density and viscosity. An initially compressed liquid reservoir may start to boil
near the well bore as the pressure is dropped during the well test. A reservoir
which is initially boiling may dry out near the well bore. For injection tests the
temperature of the injected fluid will probably be less than 100 ~ whereas the
reservoir fluid may be hot water or a mixture of hot water and steam at
200-300~ The applicability of conventional semi-log well test analysis to
geothermal wells has been the subject of several studies in recent years. Early
work by Moench and Atkinson (1977), Grant (1978), Grant and Sorey (1979),
Garg (1980), and Sorey et al. (1980) showed that a constant rate test in a
two-phase geothermal reservoir could be modelled using the Theis equation
provided that an effective or total two-phase kinematic viscosity and corn-
498 M.J. O'SULLIVAN
pressibility were used. A semi-analytic study by O'Sullivan (1981) confirmed
these results. It showed that during a constant rate drawdown test in a two-phase
or near boiling reservoir the discharge enthalpy rises and then stabilizes and the
effective kinematic viscosity of the two-phase fluid varies slowly. Therefore the
plot of pressure versus logarithm of time is approximately a straight line. A
similar study of injection tests by O'Sullivan and Pruess (1980) showed the
occurrence of a double straight-line pressure response with the early time
behaviour following a Theis curve for hot water at the original reservoir
temperature and the later time behaviour following a Theis curve for the injected
cooler water.
Recently, Garg and Pritchett (1981a,b, 1984) have carried out a series of
numerical studies for both two-phase reservoirs and reservoirs which flash during
the drawdown part of the test. Also they considered injection/falloff tests for
two-phase reservoirs. Their results confirmed the earlier work by showing good
straight-line plots of pressure versus log (time) for all drawdown tests and for
injection tests. However, the results were not so straightforward for buildup tests.
In the case of an initially two-phase reservoir, the buildup curve showed a single
straight-line response but for an initially compressed liquid reservoir there was a
double straight-line response. The first line corresponding to boiling conditions
established during drawdown and the second line to hot water when the reservoir
fluid had all recondensed. The falloff data obtained by Garg and Pritchett (1981a,
1984) does not conform to the isothermal Theis solution at all. The two-phase
nonlinearities are important and prevent simple analysis.
The purpose of the present study is to re-examine some of the problems
considered by Garg and Pritchett and to extend the work to include the case of a
fractured reservoir.
The effect of fracturing on the pressure response of isothermal reservoirs has
been extensively investigated (see Cinco-Ley and Samaniego-V (1981) for
example). The effect of fracturing on the behaviour of nonisothermal reservoirs
has also been studied. Semi-analytical methods were used by Bodvarsson (1982),
Bodvarsson and Lai (1982), Lai et al. (1983), Moench (1984), and Cox and
Bodvarsson (1985). Also numerical simulation was used by Bodvarsson and
Tsang (1980, 1982). Most of these studies were confined to the problem of the
injection of cold water into a hot reservoir. The effect of nonisothermal initial
states were considered for injection/falloff tests by Benson and Bodvarsson
(1982).
In the present study the fractures in the reservoir are represented by using the
MINC (multiply interconnected continua) approach developed by Pruess and
Narasimhan (1982a, b) and Pruess (1983a). This procedure is a numerical
implementation of the fracture/matrix block representation of a porous medium
used by Warren and Root (1963). It has also been used for petroleum reservoir
modelling (see Gilman and Kazemi, 1983 for example). Simulations are carried
out for both a uniform reservoir and a fractured reservoir for a number of well
ASPECTS OF GEOTHERMAL WELL TEST ANALYSIS 499
tests. The tests considered are two drawdown/buildup tests, firstly for an initially
two-phase reservoir and secondly for an initially hot water reservoir which flashes
during the drawdown part of the test. The third test is for injection into a
two-phase reservoir.

2. Theoretical Results
The results obtained by Sorey et al. (1980) and others show that the permeability
can be deduced from the slope of pressure plots using

2.30,,
k/v, - - - (1)
47rhm

where
1/v, = ka/vl + k ~ / v v . (2)

In (1) and (2)


m = slope of semi-log plot;
k = reservoir permeability;
O,, = mass production (injection) rate;
h = reservoir thickness;
ka, k~ = liquid and vapour phase relative permeabilities.

The quantity v~ is called the total or effective kinematic viscosity. The difficulty
in using (1) and (2) is that vt is often difficult to estimate. The vapour or liquid
saturations, S~ or St, cannot be measured directly and the forms of the relative
permeability curves ka, k~ as functions of St or S~ are unknown. Grant et al.
(1982) suggest a procedure based on the work of Sorey et al. (1980) for
calculating v, if a measurement of the discharge enthalpy he is available. Because
of the stabilization of flowing enthalpy (O'Sullivan, 1981; Sorey et al., 1980) it is
reasonably easy to measure, assuming heat losses up the wellbore are negligible.
The ratio of relative permeabilities can be calculated from the relationship

kr_L = v, (h~ - he) (3)


k~ vv (he-ht)"
Sorey et al. (1980) suggest using the relationship
ka + k~ = 1 (4)
together with (3) to calculate ka, k,~ individually and thence yr. Unfortunately,
the commonly-used Corey curves do not satisfy Equation (4) but some other
common curves such as straight-line curves do. Even supposing (3) and (4) are
used for calculating vt, it is not obvious for some geothermal well tests what
enthalpy he or pressure should be used in the calculation. For drawdown tests or
injection tests the choice is obvious but for buildup or falloff tests it is not so
500 M.J. O'SULLIVAN

clear. The key question to answer in each case is what part of the reservoir is
controlling the pressure changes at the well. In drawdown or injection tests the
fluid near the well is moving most rapidly and its properties are the most
important. But for buildup or falloff tests the fluid near the well is quickly
brought to rest and the pressure drop throughout the reservoir, including near
the well, is controlled by fluid movement some distance from the wall.
The problem becomes even more complex if the reservoir is not initially
isothermal. Benson and Bodvarsson (1982) explain the effect of an initial cold
spot surrounding the well.
For the fractured reservoir model, the fluid must move out of the matrix blocks
during drawdown and back into them during buildup. These processes lead to an
even more complicated pressure response.

3. Numerical Grid
The same grid system was used for each of the problems considered. It is very
similar to that used by Garg and Pritchett (1984). The radially infinite system is
represented by a single well block (Arl=0.11) and 59 reservoir blocks of
increasing size (Ar2 =0.4, Ar3 = 1.2Ar2, At4= 1.2Ar3,..., Ar6o = 1.2Ars9). The
outer radius of the grid was chosen to be large enough so that no changes in the
outer blocks were observed in any of the simulations. For the MINC grid, a nest
of matrix blocks is imbedded in a fracture block with the whole system occupying
the same space as the reservoir block for the uniform reservoir model (see Figure
1). The number of nested matrix blocks can be varied according to an assumed

(a) UnilormReservoir

(b) Fractured Reservoir

Fig. 1. Block layout for the reservoir models.


ASPECTS OF GEOTHERMAL WELL TEST ANALYSIS 501

Table I. Rock properties used in reservoir models

Well Uniform Fractured Reservoir


Block Reservoir Matrix Fracture
Matrix

Porosity 4, 0.9999 0.10 0.02 0.10


Permeability k, m2 50 x 10 12 50 x 10-15 1 or 5 x 10-is 50 x 10 15
Rock density, Pr, kg/m3 1 2650 2650 2650
Thermal conductivity K,
W/m~ 0 5.25 5.25 5.25
Heat capacity Cr, kJ/kg~ 0.001 1.0 1.0 1.0

average fracture spacing. For the calculations reported here only one grid was
considered, with an average fracture spacing of 10 m and a nest of four matrix
blocks in each reservoir block. As the reservoir blocks increase in size moving
out through the computational grid, the size of the embedded matrix blocks also
increases corresponding to the aggregation of all the real fractured blocks within
that volume of reservoir. However, the inter-block distances between the nested
blocks is kept fixed corresponding to the average 1 0 m fracture spacing
throughout. The properties of the well block and the reservoir blocks for both the
uniform reservoir and fractured reservoir models are listed in Table I. Other
reservoir properties c o m m o n to all blocks are listed in Table II. The fractured
reservoir model uses the same permeability for the fracture blocks and the same
interblock areas as for the uniform reservoir model. Therefore, the transmissivity
is the same in each case. However, the volume of the fracture blocks is much
smaller than the original reservoir blocks. The fracture block volume was set at
2% of the original reservoir block volume and the successive matrix blocks were
assigned volumes of 8, 30, 30, 30%, respectively, of the total volume. These
figures are fairly arbitrary but are intended to be typical of actual reservoir
values. Unfortunately, the length of the numerical calculations involved pre-
cluded experimentation with variations of these values.
The simulations were carried out using the M U L K O M simulator developed at
Lawrence Berkeley Laboratory (Pruess, 1983) and another simulator developed
at the University of Auckland (O'Sullivan et al., 1983).

Table II. Model parameters

Reservoir thickness 250 m


Relative permeability curves Corey
Residual liquid saturation 0.30
Residual vapour saturation 0.05
Production or injection rate 35 kg/s
Production or injection period 5.868• 105 s
Injection temperature 151~
Buildup or falloff period 1.3932 • 106s
502 M.J. O'SULLIVAN

4. Production Behaviour of a Two-Phase Reservoir


In this case, the data for model A.1 from Garg and Pritchett (1984) was used
(called A.1 here also). A production rate of 35 kg/s was maintained for t =
5.868 x 106 s followed by a buildup period of t = 1.3932 • 106 s. The initial state
for the reservoir is given in Table III. The simulations were carried out with an
increasing time step sequence (At3 = 22xt2, At4 = 2At3 . . . . ) which was occasion-
ally modified by the automatic time step control built into the programs. The
pressure decline curve for both the uniform reservoir (standard grid) and the
fractured reservoir are shown in Figure 2 and the discharge enthalpy in Figure 3.
The uniform reservoir model produces a nearly straight-line pressure decline with
a slope corresponding to v, = 1.46 x 10-6m2/s and the enthalpy stabilizes as
expected. For comparison, the case with the fracture completely isolated from the
matrix both thermally and hydrologically was run (labelled 'fracture only' in
Figures 2 and 3). This rather artificial 'fracture only' model has the same
transmissivity as the uniform reservoir model but the volume of each block is
very much smaller. The porosity is unchanged and therefore the ratio of the heat
content of the rock and heat content of the fluid is unchanged. Therefore, after
short-term well-bore storage effects, the pressure decline of the 'fracture only'
model becomes parallel to the uniform reservoir model curve (Figure 2) and the
enthalpy transients for both these models stabilise at approximately the same
value (Figure 3). Two fractured reservoir models were considered, the first with a
5MD matrix permeability and the second with a 1MD matrix permeability. The
results plotted in Figure 2 show that the pressure responses for both fractured
reservoir models initially follow that the 'fracture only' model and then gradually
asymptote towards the response for the uniform reservoir model (some numerical
oscillations resulting from the relatively coarse grid occur). As expected, the
approach towards the uniform reservoir model behaviour is slower for the model
with the lower permeability matrix.
The slope of the transition zone of these pressure curves is not an indication of
a smaller kinematic viscosity but is controlled by the flow out of the matrix into
the fracture. As shown in Figures 4 and 5, respectively, the kinematic viscosity
and vapour saturations for the uniform reservoir model are not very different
from corresponding values in the fracture blocks of the fractured reservoir model
(5MD matrix case shown). This type of pressure response is typical of that for

Table III. Initial data.

Case Pressure Temperature Steam


No. (MPa) (~ saturation

A.1 8.5917 300 0.28


B.1 8.7917 300 0.0
C.1 8.5917 300 0.28
ASPECTS OF GEOTHERMAL WELL TEST ANALYSIS 503
90

-- Uniform
80 Rk~ + Matrix 5MD
k ~ ~ MatrixlMD
,,o
7O

60

n
50

-~0 I . . . . . . . . . . . . . . . . . . . . . .

10 o 101 102 10 3 10 4 10 5 10 6

Time, secs
Fig. 2. Case A.1. D r a w d o w n test in a two-phase reservoir. Pressure vs. time.

drawdown tests in isothermal fractured reservoirs (see Earlougher (1977), p. 131,


for example).
The enthalpy transients for the two fractured reservoir models show numerical
oscillations but basically follow the 'fracture only' model at early times and the
uniform reservoir model at later times.
The Horner plots of the pressure recovery for the uniform reservoir model and
the fractured reservoir model (5MD matrix permeability) are shown in Figure 6.
The later time slope for the uniform reservoir model gives a value of ut = 1.26 x
10-6m2/s. This corresponds to the kinematic viscosity of the two-phase fluid

2300

2200

2100

2000-
if,
Q. 1900-
B
-- Uniform
e- 1800- -~ Matrix 5MD
r Matrix 1MD
uJ 1700- Fracture only
1600

1500 . . . . . . . . J . . . . . . . . I . . . . . . . . i . . . . . . . . 1 . . . . . . . . I . . . . . . . . i

10 ~ 101 10 2 10 3 10 4 105 10 6

Time, s e e s
Fig. 3. Case A.1. D r a w d o w n test in a two-phase reservoir. Enthalpy vs. time.
504 M . J . O'SULLIVAN

1B8
1],, -- 32s (unif)
1.6--~ -~ 2048s (unif)
t,o
] ~_ .-x- 131072s (unif)
I--'
1.4 - I ~ ~ --c- 32s (frac)
4<
~,'~,,..~~ .-n- 1536s (frac)
._>, 1.2~~--~~k'~"~ ~ -0- 168448s(frac)
u~
0
o
u)
> 0.8

0.6 . . . . . . . . I . . . . . . . . I . . . . . . . . I . . . . . . . . I

1 -1 10~ 101 102 103


Radius, m
Fig. 4. Case A.1. Drawdown test in a two-phase reservoir. Kinematic viscosity vs, radial distance.

which is still moving (as shown in Figures 7(a), 8(a) and 9(a)). The condensed
water surrounding the well has a much smaller kinematic viscosity but because it
is stationary, does not control the pressure response.
The pressure build-up response of the fractured reservoir model in Figure 6
eventually follows the uniform reservoir model (for t > 2 x 105 s approximately)
but at earlier times is different. At early times wellbore storage effects, caused by
the relatively large volume well block, dominate and then the slope flattens as the
fluid moves slowly back into the matrix blocks. This part of the curve and the
later convergence to the uniform reservoir response is similar to that for an
isothermal fractured reservoir (see Earlougher, (1977), p. 132, for example), but

0.45
_ -- 32s (unif)
~.~,~-,,<,..,,~"~-~e,~ ~ 2048s (unif)
~ ~ ' ~ x ~ '~ -~ 131072s (unif)
t- 9~ ~ ,,~ -~ ..c- 32s (fra.c)
O \ "~, "~6 ~ ~ 1536s (frac)
0.35 . ~ a c )

O)

0.25 . . . . . . . . I . . . . . . . . I . . . . . . . . i . . . . . . . . I

10-1 10o 101 102 103


Radius, m
Fig. 5. Case A.1. Drawdown test in a two-phase reservoir. Vapour saturation vs. radial distance.
A S P E C T S OF G E O T H E R M A L W E L L TEST ANALYSIS 505
90

80

~ ~ F r a c t u r e d
-Q 70

...,1 Uniform ~
~ 60

D. 50

40 ........ , ........ , ........ , ........ , ........ , ........ ,


100 101 102 103 104 105 106
(T+DT)/DT
Fig. 6. Case A. 1. Buildup test in a two-phase reservoir. Pressure vs. (t + D t/D t).

1.50
A
1.25 - - 255s
r
--~ 7679s
O 1.00 ~ --*- 89599s
4~
2010s
0.75

O 0.50
O
0.25

0.00 . . . . . . . . I . . . . . . . . I . . . . . . . . I . . . . . . . . i

0-~ 10~ 10~ 102 103


Radius, m

0.51 B - - 28s
-~ 1024s
0.4 t -4- 21244s
183039s
C
O 0.3

0.2
r
0.1

0.0
10-1 100 101 102 103
Radius, m
Fig. 7. Case A. 1. Buildup test in a two-phase reservoir. Kinematic viscosity vs. radial distance. (a)
Uniform reservoir, (b) Fractured reservoir.
506 M. J. O'SULLIVAN

0.5 A -- 255s
-~- 7679s
0.4 --4- 89599s
lOs
t-
O
. m
0.3

0.2
O~
0.1

0.0
1
-1 10 ~ 101 10 2 103

Radius, m

1.50 ] B
/ 28s
1.25 ] -~ 1024s
-~- 21244s
1.00 - -o- 183039s

0,75 -

'~0 0.50 -
o
o25-

0.00 , . . . . . . . i . . . . . . . . i . . . . . . . . i . . . . . . . . i

10 -1 10 ~ 101 10 2 10 3

Radius, m
Fig. 8. Case A.1. Buildup test in a two-phase reservoir. Vapour saturation vs. radial distance. (a)
Uniform reservoir. (b) Fractured reservoir.

the thermal effects cause a longer transition period. At the end of the drawdown
part of the test in the fractured reservoir, there are significant differences (up to
10~ between the temperatures of the fracture blocks and their associated
matrix blocks. As the results in Figure 3 show, similar amounts of total energy are
r e m o v e d from both the uniform reservoir and fractured reservoir during produc-
tion. However, the energy loss is spread further out in the fractured reservoir
case. Therefore, during buildup more heat near the wall is available in the
fractured reservoir to maintain two-phase conditions there (see Figure 7(b)). T h e
flow rates for the fractured reservoir in Figure 9(b) cannot be directly compared
with those for the uniform reservoir because of the slow rate of return of fluid
from the fracture blocks into the matrix blocks. T h e approximately flat part of the
buildup curve for the fractured reservoir when ( t + At)/t lies between 5 and 100
ASPECTS OF G E O T H E R M A L WELL TEST ANALYSIS 507

40

-- 255s
7679s
30
-4- 89599s
1312010s
20

0
10

_Z L_
U.

0-1 100 101 102 103

Radius, rn

40
B

-- 28s
t/) 30
--~ 1024s
r
2r -~- 21244s s
20

0
10
U.

0
10-1 10 o 101 10 e ] 03

Radius, m
Fig. 9. Case A.I. Buildup test in a two-phase reservoir. Flowrate vs. radial distance. (a) Uniform
reservoir. (b) Fractured reservoir.

(or 500 s < At < 200 000 s), approximately, gives a spurious value of vt =
0.398 X 1 0 - 6 m 2 / s . Therefore, if a v t appropriate for the uniform reservoir case
was used in estimating kh from the fractured reservoir curve in Figure 6, an error
of approximately 300% could be expected.

5. P r o d u c t i o n B e h a v i o u r of a Liquid R e s e r v o i r
In this case, the data for model B.1 from Garg and Pritchett (1984) was used
(called B. 1 here also). The data used was identical to that for the earlier case A. 1
except for the initial pressure which was set at 0.2 MPa above the boiling
pressure (see Table III). The pressure decline for the uniform reservoir, shown in
508 M. J. O'SULLIVAN
90
Uniform
"-4-- Matrix5MD
t~
r
80 ~ e only

t~
t~

70 ........ , ........ , ........ , ......... 9......., ........ ,


100 101 102 103 104 105 106
Time, secs
Fig. 10. Case B.1. Drawdown test in a hot water reservoir. Pressure vs. time.

Figure 10, follows the expected f o r m for a reservoir which flashes during the test
(see O'Sullivan and Pruess, 1980, for example). T h e slope of the linear part of the
decline for t > 1000s gives a kinematic viscosity of 4.14 x 10 -7 m2/s which is
appropriate for the two-phase fluid surrounding the well (see Figures 11 and 12).
T h e 'fracture only' model used for case A.1 was also run for case B.1. As
expected, the pressure decline for the 'fracture only' model follows a parallel but
lower path. Also, the enthalpy transients are similar for both cases (see Figure
13). T h e pressure response for the fractured reservoir model is m o r e complex in
this case than for case A.1. It does not follow the 'fracture only' curve at early
times but lies between it and the uniform reservoir curve. Because of the very

0"61..,,.
~-L -- 32s (unif)
0.5-] ~ -~- 2048s(unif)
''~_ "-"- 131072s(unif)
O
T" 'x. "%. -- 32s (frac)
-It 0.4 ~ "~ ~ 2048s(frac)
~"~"".~ ~'~"*~ ~k "-~ 223232s(frac)
. i
0.3 "~~...__~... ~
o
o
0.2

0.1 . . . . . . . . I . . . . . . . . i . . . . . . . . . i . . . . . . . . i

101 10~ 101 102 103


Radius, m
Fig. 11. Case B.1. Drawdown test in a hot water reservoir. Kinematic viscosity vs. radial distance.
ASPECTS OF GEOTHERMALWELLTEST ANALYSIS 509
0.3
- - 32s (unif)
--,'- 2048s (unif)
131072s (unif)
t- 0.2 -=- 32s (frac)
O
~ - - A - 2048s (frac)
-'1
o~ 0.1 (frI

0.0
10 -1 10 ~ 101 10 2 10 3

Radius, m
Fig. 12. Case B.1. Drawdowntest in a hot waterreservoir.Vapour saturationvs. radial distance.

large compressibility of a two-phase steam/water mixture, as soon as the fluid in


the fracture flashes, the pressure diffusivity in the two-phase fluid will be smaller
than that in the hot water still in the neighbouring matrix blocks, even though the
matrix permeability is much smaller. Therefore, the flow between the fracture and
the matrix affects the pressure decline earlier in this case than for case A. 1. The
late time behaviour is also different. Because of the higher liquid content initially
in the matrix, it is not so easily forced out of the matrix by boiling. The flows
between the matrix and the fracture are much smaller at late times in case B.1
than in case A.1. Therefore, the fractured system behaves more like the 'fracture
only' system than the uniform reservoir at late times. The extra heat available
from the matrix blocks dries out the fluid in the fracture slightly more than for the
'fracture only' case (see Figure 12). As shown in Figure 13 this effect pushes up

1390

1380 - - Uniform ~T
-+- Matrix 5MD j ~ T ~.
--j
1370 ~

1360

e- 1350
e-
I.tl
1340

1330 ......... 9 ........ . ....... , ........ , ........ , ........ ,


100 101 102 103 104 105 106

Time, sees
Fig. 13. Case B.1. Drawdowntest in a hot waterreservoir.Enthalpyvs. time.
510 M.J. O'SULLIVAN
88

t~
86
~ o r m
m
m Fractured ~
84
Q.

82 . . . . . . . . i . . . . . . . . I . . . . . . . . i

0~ 101 102 103


(T+DT)IDT
Fig. 14. Case B.1. Buildup test in a hot water reservoir. Pressure vs. (t + Dt)/Dt.

0,4
A

-4- 31s
8575s
0.3 - - 262527s
- - 1245570s ,

u)
o 0.2
o
.~_
>

0.1 '[ . . . . . . . . i . . . . . . . . i . . . . . . . . i , , ,"r'~ '"1

10-1 10~ 101 102 103


Radius, m
0.5 B
31s
0.4 ~ 5247s
132223s
-- 1139840s
>, 0.3

0
0
._~ 0.2

0.1 . . . . . . . . i . . . . . . . . i . . . . . . . . i . . . . . . . . i

10-1 10o 101 102 103


Radius, m
Fig. 15. Case B.1. Buildup test in a hot water reservoir. Kinematic viscosity vs. radial distance. (a)
Uniform reservoir. (b) Fractured reservoir.
ASPECTS OF G E O T H E R M A L WELL TEST ANALYSIS 51 1

the production enthalpy from the fractured reservoir a small amount above that
from the other two reservoir models.
The Horner plots for case B.1 are shown in Figure 14 for both the uniform
reservoir and the fractured reservoir (5MD matrix permeability). Both plots are
similar and show two distinct linear regions. The earliest straight line gives
kinematic viscosity vt = 3.02 x 10 -7 m2/s which corresponds to two-phase fluid
which continues to move at some distance from the well (see Figures 15, 16, 17).
The later straight line gives a kinematic viscosity v, = 1.78 • 1 0 -7 m 2 / s which
corresponds to the movement of hot water after all the boiling fluid has
recondensed. The difference between the uniform reservoir and fractured reser-
voir results in this case is much less than for case A. 1. This can be explained by
the same mechanism as was used to explain the late time pressure decline during

0.25
B
-4-- 31S
--~ 5247S
132223S
C
0.15
O -- 1139840S

0.05

-0.05 9 . . . . . . . i . . . . . . . . i . . . . . . . . i . . . . . . . . i

10 q 10 o 101 102 103

Radius, m

0.25 ] A

9 -~- 31s
-~- 8575s
~- 0.15.,/, / ~ -- 262527s

-0.05 ................. , ........ , .........


10 -1 100 101 102 103

Radius, m
Fig. 16. Case B.I. Buildup test in a hot water reservoir. Vapour saturation vs. radial distance. (a)
Uniform reservoir. (b) Fractured reservoir.
512 M. J. O ' S U L L I V A N

35
A
=~ -4- 255s
7679s
25
89599s
O1
1312010s
15

O
LI.

-5 . . . . . . . . i . . . . . . . . i . . . . . . . . I . . . . . . . . I . . . . . ',, ,I

10 -1 10 0 101 10 2 10 3 10 4

Radius, m

cn
r
35

25
~ -4- 31s
5247s

15

O 5
14.

-5 . . . . . . . . I . . . . . . . . I . . . . . . . . I . . . . . . . q . . . . . . . . I

01 10 ~ 101 10 2 10 3 10 4

Radius, m
Fig. 17. Case B.I. Buildup test in a hot water reservoir. Flowrate vs. radial distance. (a) Uniform
reservoir. (b) Fractured reservoir.

the production test. T h e flow between the fracture and matrix was smaller at
the end of the production test for case B.1 than for case A.1. Similarly, during
buildup less m o v e m e n t of fluid back into the matrix blocks is required for case
B.1 as they have remained wetter than for case A.1. T h e slightly hotter matrix
blocks adjacent to the fracture slow the recondensation process (see Figure 14)
for the fractured reservoir, otherwise the two curves are very similar.

6. Injection of Warm Water Into a Two-Phase Reservoir


In this case (called C.I here and by Garg and Pritchett, 1984), water at
approximately 151 ~ is injected at 35 kg/s into a reservoir with identical proper-
ties and initial conditions to those used for case A. 1. The pressure buildup results
are shown in Figures 18 and 19 for a uniform reservoir and two different
ASPECTS OF G E O T H E R M A L WELL TEST ANALYSIS 513
95

~ 90

85 ,...~
" ~ [ "~ Matrix 5MD
~ %.. /~ -- Matrix IMD
0. 80

75 ........ , . . . . . . . . , . . . . . . . . , ........ , . . . . . . . . , . . . . . . . . ,
10 o 101 10 2 10 3 10 4 10 5 10 6

Time, sees
Fig. 18. Case C.1. Injection test in a two-phase reservoir. Pressure vs. time.

fractured reservoirs with 5MD and 1MD matrix permeability, respectively. For
an idealised uniform reservoir problem with a zero well radius, the pressure
buildup should show two linear sections, the first at early times with a slope
corresponding to hot water at the reservoir temperature and the second at late
times with a slope corresponding to the injected cold water (see O'Sullivan and
Pruess, 1980, for example). In Figure 18, the early hot water straight line is
masked by the wellbore effects introduced by the relatively large well block in
the model. The enlarged late-time plot shown in Figure 19 shows the cold water
straight line clearly, with a slope giving a kinematic viscosity v , =
2.01 x 10 -7 m2/s.

92

~
9 91

-',I 90
Uniform

~ 89 -- Matrix 1MD

88 . . . . . . . . . i . . . . . . . . i . . . . . . . . I

102 103 104 105 106

Time, sees

Fig. 19. Case C.1. Injection test in a two-phase reservoir. Pressure vs. time (late time behaviour).
514 M.J. O'SULLIVAN
The corresponding curves for the two fractured reservoir cases eventually
approach the cold water straight line but, for a period between approximately
5000-500 000s, follow another straight line with the slope giving kinematic
viscosity u, = 1.28 x 10-7m2/s, which is equal to the value for water at the
reservoir temperature of 300~ for the 5MD matrix case and a similar value of
ut = 1.00 x 10 -7 m2/s for the 1MD matrix case. The fractured reservoir results in
Figure 19 are very similar to results obtained by Benson and Bodvarsson (1982)
and Cox and Bodvarsson (1985) for injection into a uniform reservoir with a
'coldspot' surrounding the well at the start of the test. Cox and Bodvarsson (1985)
also considered cold-water injection into an isothermal fractured reservoir but
their model was different from that used here, as they assumed zero matrix
permeability and used a semi-analytic method for modelling heat transfer be-
tween the fracture and the rock matrix. Their results show a similar transition
from a hot water slope to a cold water slope. These results indicate that, for a
fractured reservoir, the kinematic viscosity of hot water at the original reservoir
temperature may be the appropriate one to use in well-test analysis. In analysing
field data it may be difficult to determine whether or not the transition from the
hot slope to the cold slope shown in Figure 19 has occurred. Grant et al. (1982)
also suggest using the properties of the hot water in analysing injection tests but
their argument is based on the behaviour of an initially isothermal uniform
reservoir and does not appear to agree with theoretical results obtained here and
elsewhere for this case (see Garg and Pritchett, 1984; Benson and Bodvarsson,
1982; and O'Sullivan and Pruess, 1980). Careful examination of the results in
Figures 20, 21 and 22 show that in the 'fracture' of the fractured reservoir, the
cold front moves approximately the same distance as in the uniform reservoir but
it is surrounded by a more extensive band of warm and hot water in the fractured
reservoir. Also during intermediate times (1000-100 000 s) fluid is being

300 q ~ -- - ................

256s (unif)
250 ~ + 16384s (unif)
"0 / ~ 586800s (unif)
/ --e- 208s (frac)
200 ~ , , , . ~ - ~ - 1 6 3 3 6 s (frac)
~. "-~ 586800s (frac)

I.- 150 . . . . . . . i . . . . . . . . i . . . . . . . . i

10-1 100 1 01 102 103


Radius, m
Fig. 20. Case(2.1. Injectiontest in a two-phasereservoir. Temperaturevs. radial distance.
ASPECTS OF GEOTHERMAL WELL TEST ANALYSIS 515

0.3"

-- 256s (unif)
e,, 0.2
0
, m
16384s (unif)
586800s (unif:
2~ ..-u- 208s (frac)
0.1 16336s (frac)
586800s (frac

0.0 "1 . . . . . . . . I

10 q 100 101 102 103


Radius, m
Fig. 21. Case C.1. Injection test in a two-phase reservoir. Vapour saturation vs. radial distance.

squeezed out of the fracture into the matrix with the low matrix permeability
partially controlling the pressure changes. Both these factors contribute to the
lower effective kinematic viscosity indicated by the fracture reservoir curves in
Figure 19.
Benson and Bodvarsson (1982), for a uniform reservoir, and Cox and Bod-
varsson (1985), for a fractured reservoir, analysed pressure fall-off data obtained
following cold-water injection tests. These results showed an early time cold
water straight line quickly changing to a hot water straight line as the cold zone
around the well becomes stationary. However, Garg and Pritchett (1984) found
that their results were confused by wellbore storage effects and, even at late
times, did not produce a straight line on a Horner plot. The present investigation
confirmed their results. The results for the fractured reservoir case were similarly
confusing and are therefore not reproduced here.

_,T
.~
-10

-20
/
/
/ M~ j
'
/ 1 1 /f
/"
/
I
1

/~
~

/"
,/ ~
/
(unif)
~ I ~ 16080s(unif)
'-~ 586800s(unif)
§
~
2o8s (frac)
16080s(frac)
~ 586800s (frac)
I -30
n-

-40 . . . . . . . . I . . . . . . . . t . . . . . . . . I . . . . . . . . i

10 -1 100 101 102 103


Radius, m
Fig. 22. Case C.1. Injection test in a two-phase reservoir. Flowrate vs. radial distance.
516 M.J. O'SULLIVAN

7. Conclusions
T h e numerical experiments conducted for this investigation show that geother-
mal well test results for fractured reservoirs m a y be confusing. For drawdown
tests, the a p p r o a c h to a straight line whose slope corresponds to the true reservoir
properties is slow. It is preceded by an initially rapid pressure decline followed by
a fiat transition period where fluid m o v e s f r o m the fracture into the matrix. If
the reservoir is initially all liquid but flashes during the test, the effect of
fracturing on the slope of the pressure decline curve is less pronounced. T h e
buildup phase of the well test is also considerably affected by fracturing with the
straight-line response of the uniform reservoir distorted so m u c h that it would be
difficult to choose a slope for a k h calculation.
For the injection of cold water into a hot two-phase reservoir, the fractured
reservoir produces a response which has a long transition region which is
approximately straight on a semi-log plot but which corresponds to the kinematic
viscosity of hot water, whereas the uniform reservoir or the later time behaviour
of the fractured reservoir follows a straight line with a slope corresponding to
cold water properties.
T h e numerical experiments reported here are not comprehensive. Only two
cases of permeability contrast between the fractures and the matrix and two sets
of initial conditions were considered and only one basic set of reservoir
parameters were used. Unfortunately, the results do not help greatly in explaining
how geothermal well test results should be analysed but they do indicate some of
the deviations f r o m standard results that fractures in the reservoir may produce.

References
Benson, S. M. and Bodvarsson, G. S., 1982, Nonisothermal effects during injection and falloff tests,
SPE paper 11137 presented at the 57th Annual Technical Conference and Exhibition of the
Society of Petroleum Engineers, New Orleans, Louisiana.
Bodvarsson, G. S. and Tsang, C. F., 1980, Thermal effects in well tests of fractured reservoirs, Proc.
3rd Invitational Well-testing Symposium, Lawrence Berkeley Laboratory Report LBL-12076,
Berkeley, California, pp. 110-119.
Bodvarsson, G. S., 1982, Mathematical modelling of the behaviour of geothermal systems under
exploitation, PhD thesis, University of California, Berkeley.
Bodvarsson, G. S. and Lai, C. H., 1982, Studies of injection into naturally fractured reservoirs, Trans.
Geothermal Res. Council 6, 245-248.
Bodvarsson, G. S. and Tsang, C. F., 1982, Injection and thermal breakthrough in fractured
geothermal reservoirs, J. Geophys. Res. 87, 1031-1048.
Cinco-Ley, H. and Samaniego-V., F., 1981, Transient pressure analysis for fractured wells, Soc. Pet.
Eng. J. 21, 1749-1766.
Cox, B. L. and Bodvarsson, G. S., 1985, Nonisothermal injection tests in fractured reservoirs, Proc.
lOth Workshop on Geothermal Reservoir Engineering, Stanford, pp. 151-162.
Earlougher, R. C., 1977, Advances in well test analysis, Soc. Pet. Eng. AIME, New York.
Garg, S. K., 1980, Pressure transient analysis for two-phase (water/steam) geothermal reservoirs, Soc.
Pet. Eng. J. 20, 206-214.
Garg, S. K. and Pritchett, J. W., 1981a, Buildup analysis for two-phase geothermal reservoirs, Trans.
Geothermal Res. Council 5, 287-290.
ASPECTS OF GEOTHERMAL WELL TEST ANALYSIS 517

Garg, S. K. and Pritchett, J. W., 1981b, Cold water injection into two-phase geothermal reservoirs,
Proc. 7th Workshop on Geothermal Reservoir Engineering, Stanford, pp. 175-178.
Garg, S. K. and Pritchett, J. W., 1984, Pressure transient analysis for two-phase geothermal wells:
some numerical results, Water Resour. Res. 20, 963-970.
Gilman, J. R. and Kazemi, H., 1983, Improvements in simulation of naturally fractured reservoirs,
Soc. Pet. Eng. J. 23, 695-707.
Grant, M. A., 1978, Two-phase linear geothermal pressure transients - A comparison with single-
phase transients, N Z Y. Sci. 21, 355-364.
Grant, M. A. and Sorey, M. L., 1979, 'The compressibility and hydraulic diffusivity of a water-steam
flow, Water Resour. Res. 15, (3), 684-686.
Grant, M. A., Donaldson, I. G., and Bixley, P. F., 1982, Geothermal Reservoir Engineering, Academic
Press, New York.
Lai, C. H., Bodvarsson, G. S., Tsang, C. F., and Witherspoon, P. A., 1983, A new model for well test
data analysis for naturally fractured reservoirs, SPE paper 11688 presented at California Regional
Meeting of the Society of Petroleum Engineers, Ventura, California.
Moench, A. F. and Atkinson, P. G., 1978, Transient pressure analysis in geothermal steam reservoirs
with an immobile vaporising liquid phase, Geothermics 7, 253-264.
Moench, A. F., 1984, Double-porosity models for a fissured ground-water reservoir with fracture
skin, Water Resour. Res. 20, 831-846.
O'Sullivan, M. J. and Pruess, K., 1980, Analysis of injection testing of geothermal reservoirs, Trans.
Geothermal Resour. Council 4, 401-404.
O'Sullivan, M. J., 1981, A similarity method for geothermal well test analysis, Water Resour. Res. 17,
390-398.
O'Sullivan, M. J., Zyvoloski, G., and Blakeley, M. R., 1983, Computer modelling of geothermal
reservoirs, School of Engineering Report 318, University of Auckland, New Zealand.
Pruess, K. and Narasimhan, T. N., 1982a, A practical method for modelling fluid and heat flow in
fractured porous media, SPE paper 10509 presented at the 6th Symposium on Reservoir Simulation
of the Society of Petroleum Engineers, New Orleans, Louisiana.
Pruess, K. and Narasimhan, T. N., 1982b, On fluid reserves and the production of superheated steam
from fractured, vapor-dominated geothermal reservoirs, J. Geophys. Res. 87, 9329-9339.
Pruess, K., 1983a, Heat transfer in fractured geothermal reservoirs with boiling, Water Resour. Res.,
19, 201-208.
Pruess, K., 1983b, Development of the general purpose simulator MULKOM, in 1982 Annual
Report, Report LBL-15500, Earth Sciences Division, Lawrence Berkeley Laboratory, Berkeley,
California.
Sorey, M. L., Grant, M. A., and Bradford, E., 1980, Nonlinear effects in two-phase flow to wells in
geothermal reservoirs, Water Resour. Res. 16, 767-777.
Warren, J. E. and Root, P. J., 1963, The behaviour of naturally fractured reservoirs, Soe. Pet. Eng.
J., September, 245-255.

Vous aimerez peut-être aussi