.

6iii)a w

. .

Sedetyd Petroleum Engineers

SPE 36682

Determination of Relative Permeability for Steam-Water Flow in Porous Media

Willis Ambusso, Cengiz Satik and Roland N. Home, Stanford University

@P@Lml=,=-wd ~=w-=!~ individual microscopic pressure gradients and by the fraction

~tithslSSS SPENmud Todmkd Cmtawtmunl of the overall permeability that is associated with it. This
.Cdcrde. USA,6-BOddmrlSSS
fraction, normally expressed as a fraction of the mechum’s
Thispqu wm$OIBcbd larpmmtdmn byrn SPEPqmm CommStnfc4xhwj-d
canidrd
hmabatmctsdnnmdbytiaufhu(s). C0nt9nb0fttwpapu,
M permeability to a single-phase flui~ is called the relative
%*hMndb—nm.iutui byumsoddy Of Pamhun En@nmti-9@j@clto
~1-cllmdury
permeability. Relative permeability relations have been
auru4bn by!hoathorts). lhOmmuial, up’wBnwJ, donnd
ruiticmdih SocMyd P~Ew-, ibdfimf8,0r mwnbwa. Papwspmsonidal traditionally expressed as a function of saturation principally
SPEmnlinsa am8ul#sdbpblhtim - by EdtaWCanmkIo- dtlw Socidyd
r’@OblmEn@o-fwmbhb-Prb~ tom~dnc4momlhn3GU because it was believed that they depended on the pore
Wld9. llhmlldions nwyrldbcqliod m0~81muldanmncm@amu8
volume occupied by the fluids (Hassler, 1944). While a great
~IylW-~&y ~. SPE. PO
number of experiments have shown this to be true, a number
of other experiments have shown that relative permeability
depends also on several other parameters such as interracial
Abstract tension, wetting characteristics and viscosity ratios of the
A set of relative permeability relations for simultaneous flow flowing fluids etc. (Fulcher et al., 1983; Osoba et al., 1951). In
of steam and water in porous media were obtained from a addition it is necessary to define residual saturations which
steady state experiment. The experiment was conducted under normally indicate the lowest saturation value for a given phase
conditions that minimize most of the difficulties encountered to become mobile. The curves and the residual satumtions
in the past for similar experiments. A high resolution X-ray together define the relative permeability relations (Corey,
CT scanner was used to obtain three-dimensional porosity and 1954; Brooks and Corey, 1964).
saturation distributions during the experiment. Pressure and
temperature data were collected with the use of several For most applications in petroleum engineering involving the
pressure transducers and thermocouples located at various flow of oil and water as in water flooding and the flow of oil
locations along the core. Relative permeability values were and gas as in gas injection, the relative permeability relations
calculated over the core lengths identified as having flat are well known and determined routinely ftom laboratory
saturation profiles. These aspects constitute a major experiments (Osoba et al., 1951). However, for the flow of
improvement in the experimental method compared to those steam and water or for the general case of multi component
used in the past. The experimental results showed the relative multi phase flows these relations are not well known. A look
permeability to both steam and water vary linearly with at previous literature shows that there is a significant
saturation, in contrast to those obtained by Corey (1954) for discrepancy and inconsistency in the results previously
gas/water flow experiments, Comparison of the saturation obtained (Verm% 1986; Sanchez 1987; Clossman and
profiles measured by the X-ray CT scanner during the Vinegar, 1988). The main difficulties encountered in the
experiments shows good agreement with those predicted by previous experiments, as we show later in this paper, have
numerical simulation. To obtain results that are applicable to been mainly due to inaccurate measurements of fluid
general flow of steam and water in porous medi% similar saturations and inappropriate assignment of pressure gradients
experiments should be done at higher temperature and with to the individual phases.
porous rocks with different wetting characteristics and
porosity distribution. Techniques involving analysis of enthalpy transients from
producing geothermal fields have been proposed to infer
INTRODUCTION relative permeability relations (Grant 1977; Sorey et al.,
1980; Home and Ramey, 1978). However these techniques
In general, the concept of relative permeability is an attempt to
did not eliminate all of the variables and quite often the in-situ
modify Darcy’s law for single-phase flow in porous media in
fluid saturations and the overall permeability structure (i.e
order to account for simultaneous flow of multi phases. In
matrix, fkture) were ignored leaving uncertainty on the
this description the flow of each phase is governed by the
2 W. AMBUSSO, C. BATIK AND R. N. HORNE SPE 3B682

results obtained. As discussed by Heiba et al. (1983), saturation variations caused by capillary end-effects at ,
experiments are the most reliable method to determine relative relatively low flow rates. Clossman and Vinegar (1988) are
permeability. However, laborato~ techniques also suffer from probably the fmt to report the use of X-ray CT scanner to
limitations. One of the most important limitations is imposed measure water and steam saturations in porous materials. They
by boundary effects caused by capilhuy forces. Capillarity investigated steam-water relative permeability in cores horn
introduces nonlinear effects on the pressure and saturation oil fields at residual oil saturations. The cores used for the
distribution of the wetting phase at the core exit. Thus experiments were rather small i.e., 15.4 cm in length and 2.47
experiments must be designed to minimize such effects. As cm in diameter. The flow rates were also moderate, 3.31
shown by the experiments repotied by Osoba et al. (1951) and cc/rein to 20 cc/rein. Steam quality was determined from the
by Hassler (1944), laboratory measurements of relative two temperature measurements at the inlet and exit. The
permeability can have error if capillary end-effects are not same readings were used to estimate heat losses fkomthe core.
considered. The capillary end-effects may cause pressure Relative penneabihty relations were calculatd fbm pressure
gradients and by extension saturation gradients, resulting in a measurement at the same points, Temperatures were not
nommifonn distribution of fluids in the core particularly at measured along the core but the distribution within the core
low flow rates. Ignoring this effect may result in was assumed to vary in three Poss]lde ways, linear, quadratic
underestimating the relative permeability of the wetting phase and constant. The core was enclosed in an aluminum sleeve
and attributing a permeability value for the nonwetting phase kept under vacuum conditions to minimize heat losses.
to a wrong saturation (VermA 1986). Capillary pressure Clossman and Vinegar (1988) found that the relative
effects can be overcome by use of sufficiently long cores or permeability values for the steam phase were close to those
by use of high injection rates (Osoba et al., 1951). reported by Brooks and Corey (1964) but those for the liquid
phase were somewhat smaller, More recently Piquemal (1994)
One of the earliest attempts to measure relative permeability has reported relative permeability relations for steam and
relations for srngle component two-phase flow was reported water using metiods similar to those used by Verma (1986).
by Miller (1951). In these experiments liquid propane was The experiments were conducted at 180 ‘C. Though Piquemal
allowed to flash as it moved across the core thus creating a (1994) did not discuss any emors in his measurements, the
two-phase flow with increasing gas tiction as the fluid experiments were subject to the same limitations reported by
moved fmther downstream. From the pressure and Verma (1986) who used a similar apparatus. It is interesting to
temperature measurements and application of material and notice that the results reported by Piquemal (1994) are
energy balances he determined the flowing liactions at each different fkom those obtained by Verma (1986) who observed
point and therefore estimated the relative permeability enhanced permeability of the steam phase. Piquemal (1994)
relations, It is not clear, however, whether capillary end- obtained results suggesting that steam-water flows are similar
effects were adequately eliminated since the calculated to nitrogen-water.
saturation values could not be checked by other independent
means. Among the fwst attempts to measure saturations This review shows that there is a wide range of results that
dwctly were those reported by Chen (1976) and Council have been reporte~ some of which even used similar
(1979) using a capacitance probe method. In this technique, experimental apparatus. The main reason for this has been the
the saturation was obtained ffom a calibration based on the difllculties in measuring saturations accurately and using
relation between the capacitance and the saturation within the incomect pressure gradients to compute relative permeability.
core. However the margin over which readings were obtained The investigations reported in this paper overcame these
was small, thus leaving doubts on the reliability of the relative difficulties by using an X-ray CT scanner to measure
permeability relations obtained. Later, Verma et al. (1985) and saturation accurately and by evaluating pressure gradients
Verma (1986) used a gamma-ray densitometer for actually within the wnes of constant saturation only. In this
experiments using an artificial sand pack. Though this was an work it was not possible to conduct the experiments under
improvement over the capacitance probe, the portion of the perfectly adiabatic conditions since the X-ray CT seamer
sample accessed by the densitometer was small (5Yo). In imposed restrictions on the type of materials that could be
addition, problems with the equipment during the experiments used, effectively elirniiating the use of guard heaters.
resulted in only a small part of the relative permeability curve However, heat losses were miniiized significantly by using a
being obtained. Recently, Sanchez (1987) reported the use of thick layer of high performance insulation material, In
average recovery time of a tracer injected with the fluid to addition heat losses from the core were measured at several
determine the water saturation in the core. In these locations using heat flux sensors. Numerical simulations were
experiments, pressure was measured at only two points a short carried out to determine the optimum experimental conditions.
distance ibm the either end of the core, effectively ignoring This included determining the appropriate core length, the
capillary end-effects. Sanchez (1987) estimated an average effect of various parameters on the results. In this paper, we
water saturation representing the whole core and ignored the first begin by describing our experimental apparatus and the
SPE 36682 DETERMINATION OF REIATWE PERMEABILITY FOR STEAM-WATER FLOW IN POROUS MEDIA 3

method used together with the conditions that have to be met COZ and to saturate eventually it completely. This step
in order to overcome some of the errors associated with continued until the core was completely saturated with water,
laboratory measurements. Next the results of the numerical at which time the core was X-ray scanned again at the same
investigations are discussed. Finally, we shall present the locations to obtain wet-core CT (CTWJ values an~ pressure
experimental investigations and a discussion of the results. and temperature readings were taken at this time. Steady-state
relative permeability experiments involved injection of
varying fractions of steam and water into the core.
Measurements done at each step resulted in a single data point
EXPERIMENTAL APPARATUS AND PROCEDURE on the relative permeability vs. saturation curve. Starting
from a completely water-saturated core and injecting steam at
The description of the apparatus for these experiments was
increasing tkctions will give rise to a drainage curve while
discussed in Satik et al. (1995). In general, it consists of an
the opposite procedure gives rise to an imbibition curve. Each
injection Unigan epoxy core holder and a production unit, The
step continued until steady-state conditions at which injeetion
injection unit consisted of two furnaces to generate steam and
and production rates became the same for both steam and
hot water, two temperature controllers to control the
water and also pressures and temperatures stabilized. At the
temperatures of the two fiunaces, and two pumps to inject the
onset of steady-state conditions, another X-ray scanning was
fluids. Temperatures were measured by the thermocouples
performed along the core at the same locations to obtain CT
embedded within the outer most layer of the epoxy core
(CT=P) values corresponding to the particular steam-water
holder. A 12-channel thermometer unit was connected to a
fraction. Next, the steam-water fraction was change~ keeping
computer for collecting the temperature data. The
total flow rate constant and the entire procedure was repeated.
thermometer gathered readings from the eleven J-type
thermocouples, eight of which were located on the core while
the other three were on the steam line, the water line and the
mixing point for the steam and water at the injection end of
the core. Pressures were measured by using eleven pressure
transducers. Direct monitoring of pressures and temperatures
during the experiment enabled us to determine when steady
state conditions had been reached. Heat losses on the core
body were measured by using heat flux sensors placed at
various locations along the core body.

The core (Berea sandstone) samples used for these

experiments have been described in detail in Ambusso (1996) Figure 1: Schematic of the experimental apparatus,
and had the following properties; permeability 600md,
porosity 20%, length of 38 cm and diameter of 5.04 cm. The After the experiment was completed, an interpretation
core sample was fust heated to 450”C for twelve hours to sofhvare was used to calculate the porosity and saturation
deactivate clays and to get rid of residual water. The two ends distributions tim the CT values obtained with the X-ray CT
of the core were then covered by the end plugs fitted with scanner. To calculate porosity the following expression was
nipples for injection and production of fluid. Eight po~ to used:
measure temperatures and pressures were then fitted at the
freed intervals along the edge of the core before the rest of tie += :;:;,--; ...................................... (1)
core was covered completely by high temperature epoxy to am

complete the construction of the core holder. The core holder where CTwarer, CTair are CT numbers for water and air,
was tested for leaks before being covered with an insulation respectively. Similarly, the expression used to calculate
material made of ceramic blanket. The core holder was then saturations is:
placed on the motorized bench of the X-ray CT scanner so CTW, - CTWP
that it could be scanned as required. Finally, saturations were s. = ................................... (2)
measured by using a high resolution X-ray CT scanner. A CTW, - CTh
schematic of the experimental apparatus is shown in Figure 1. and
Sw=1-s. .................................................. (3)
The experimental procedure was as follows. Finsl air inside where S~t and Sw denote steam and water saturations,
the pore spare was displaced out by injecting several pore respectively.
volumes of COZthen the core was scanned at predetermined
locations to obtain dry-core CT (CT*) values. Nexg several The starting point of the derivation are the conservation
pore volumes of water were injected into the core to remove
4 W. AMBUSSO, C. SATIK AND R. N. HORNE SPE 36682

equations for mass and energy fluxes: investigation. This program is a multicomponent thesmal .
simulator specifically designed to handle heavy-oil operations
mt =mv+m, .............................................. (4) such as surfactant flooding, steam injection and in-situ
combustion in single and dual porosity medi% and flactured
reservoirs. Three main aspects were investigated; the effect of
m,h, = mvhv+ mlh, + Q ............ ..... .......(5) the type of relative permeability curves, the effect of flow
rates and flowing &actions on pressure, temperature and
where m and h refer to mass flow rate and enthalpy, saturation and the effect of heat losses on fluid distribution
respectively and the subscript t refers to total, v to vapor phase along the core and fluid segregation due to the combined
and 1 to the liquid phase. Q is the total heat lost upstream of effects of gravity and condensation. In all of the cases the
the point being considered. physical dimensions of the models were sinilar to those used
for the experiments (a core of 5.08 cm in diameter and 43.2
Then using flat interface thermodynamics the steam fkaction cm in length). Permeability and porosity values were set to
(x) in the flow at any time would be given by: 600 md and 20Y0,respectively. The injection and production
points were fwed at the centers of the end plates. In each
m,(h, -h, )-Q simulation run, the parameters of interest were saturation,
x= ................................... (6) pressure and temperature.
ml htv
Three types of relative permeability curves were used in the
h,, is the latent heat of vaporization at the prevailing numerical simulations: the widely used Corey (1954), the
temperature and pressure. linear curves and the curves derived byVerma(1986). These
curves are shown in Figure 2. In particular, the curves
Then the relative permeabilities to steam and water can he reported by Verma (1986) were of interest since these curves
calculated by the corresponding Darcy’s equations for each
phase in terms of the mass flow rates: 1

(1 - XM%MVI
kd=- ................................0(n
4P
kA~ I ao

I
0.4

and

‘ml ~. ‘S o
kn=- ......................................... (8] o 02 0.4 0.6 a9 1
4P -—
M-
AX Figure 2: Relative permeability curves used for the numerical
simulation.
Thus a knowledge of the values of flowing mass fictions in
the above equations and pressure drop along a column of the
represent a more recent measurement and the methods used to
core with constant or flat saturation provides a value for the
obtain them were similar to ours. In order to reduce the
relative permeability.
number of variables between the curves the irreducible
saturations for the liquid phase tiom the curves obtained by
RESULTS
Verma (1986) were used for all of the curves. In order to
reproduce the end-effects it was necessary to incorporate a
NUMERICAL capillary pressure in the simulator. These fimctions are well
known for oil snd water under static conditions. In the case of
Prior to designing the experiments, numerical simulations
steam-water flow in porous media however, these fimctions
were canied out to determine the optimum dimensions of the
are cumently not known<We therefore used the relations given
core required to overcome capillary end-effects and to
in a pammetric form -C*b@w), where C is a constant. To
evaluate the effect of various parameters on the reauhs. The
mimic the capillary end-effects, the core was divided into
details of the simulations were described in Ambusso et al.
several small blocks while the first and the last blocks were
(1996).
assigned zero capillary pressure. In real situations the capillary
pressure may be small but will always be non-zero in the
The STARS software was used for the numerical
injection lines.
SPE 36682 DETERMINATION OF RELATIVE PERMEABILITY FOR STEAM-WATER FLOW IN POROUS MEDIA 5

9
Figure 3 shows the numerical simulation results of saturation,
0.9
pressure and temperature distributions obtained using different
0.8
relative permeability curves. The total injection rate is 15
0.7
cc/rein of water and the steam quality is 0.1. In all of the cases
0.s
the flow was modeled as adiabatic. It is clear that the linear
0.5

-’---WI
relative permeability curves predict lower steam saturations. I
They also give lower pressure drops across the core for all of 0.4
I
the injection rates. This is consistent with the higher mobility 0.3

predicted (equal to unity for all saturations) for the combined 0.2
flow steam and water. In all of the cases the steam saturation 0.1
increases marginally towards the production end until the end- 0
effects reverses the trend. ‘l%is, too, is consistent with the
flashing of water into steam as the pressure declines. In all of
the cases, the capillary end-effects are very strongly expressed
(a)
but decrease as flow rate increases. The results also show that
it is possible to have a substantial flat saturation profile even
for modest injection rates, These curves suggest clearly that
the type of relative permeability curve has a significant 60

influence on the results obtained,

40

To investigate the effect of flow rate, the relative permeability t

20
curves obtained by Verma (1986) were used. In each case the
steam quality (in mass) was kept constant at 0.1. The injection 1! 20
rates were 5, 15 and 30 cc/rein. Figure 4 shows the numerical
simulation results of saturation distributions at these three
flow rates. These results show the expected behavior. lle
portion of the curve affected by the capillary end-effects
decreases as the flow rate increases. The pressure and o 10 20 30 40 ao
temperature also rise to higher values as the flow rate
increases. These results show that the appropriate length for (b)
the core which is not to be affected by end-effects is about 30
160
cm. Therefore a core length of 43.2 cm was selected to be
used for the experiments.

To investigate the effect of heat losses on fluid segregation, a

.i----
three-dimensional numerical model was constructed by
dividing the core into a 100x3x3 grid in the x, y and z
directions, respectively. Insulation around the core was added
as an additional layer of low thermal conductivity in the y, and
z directions. The thickness of the insulation was set to 2.54 cm
and the porosity and permeability of the insulation were set to
zero. In the simulator, the thermal properties were set to those
provided by the manufacturer. The curves obtained by Verma
26

0
0 10 20
EEl
so 40 60
(1986) and a flow rate of 12 cc/rein were used to generate the ObmnalmmHcttm
results shown in Figure 5. The results are for the middle three
layers flom the uppermost to the lowest. As expected, the (c)
temperature and pressure are practically the same for all of the Figure 3: (a) Saturation, (b) Pressure and (c) Temperature
blocks at a given cross-section except at the end blocks where distributions for different relative permeability relations.
there exists a non-axial flow. Saturations vary only marginally
in the vetiical direction.
6 W. AMBUSSO, C. SATIK AND R, N. HORNE SPE 36662

1
0.s n= mn
0.3 ~n)=16 cdmln
lnJ@O cchln
0.7
0.6
0.6
I
0.4
I
0.s
0.2
~
0.1 o 10 so 30 40 63
0 Q km)
o 10 so So 40 so
C#sulafmmwccul
Figure 6: Saturation profiles for dlabatic and non-adiabatic
Figure 4: Saturation distributions for three different injection cases, obtained flom the numerical simulation.
rates, obtained fkomthe numerical simulation.
EXPERIMENTAL
A steady state flow experiment in a Berea sandstone core was
1 conducted to measure steam/water relative pcrmeabilities. The
0.s single core dynamic method was used, This method required
0.8 that a two-phase mixture of steam and hot water be injected
0.7 into the core. Changing the flowing fractions of each phase
and letting the system adjust itself to steady-state conditions,

I0.6

0.5
the relative permeability relations were determined horn the
knowledge of the flowing fiw%ons, pressures and
temperatures. To determine the flowing &actions along the
core it was necessary that the enthalpy of the injected fluid be
known accurately, which the required the knowledge of the
tkactions of the injected fluids before entering the core.
Although it has been suggested that in the porous media the
0 10 so 30 40 m process of boiling and phase change may require more energy
Oishnc4ffom Inht. cm due to capillary forces (Udell, 1982), experiments by Miller
Figure 5: Saturation dkributions for three vertical layers, (195 1) with light gasoline showed that the temperature and
obtained tlom the numerical simulation. pressure follow values close to those for flat surface
thermodynamics, For this reason steam table values were
To compare the results obtained with and without heat losses a used to compute the flowing phase fictions in the core.
one-dimensional model was constructed with the same
dimensions. The comparison of results fkomthese two models We used two methods of injection of fluids with known
is shown on Figure 6. ‘l’here still is a flat saturation profile enthalpy. The fwst was to inject a fluid that was heated to high
over the most of the early part of the curve but the capilky temperature but always keeping the pressure above saturation
end-eff’ are more strongly expressed and start earlier for so that the fluid always remained in the liquid state upsfream
the non-adiabatic case. Also, the meam saturation does not of a throttle valve, set to release fluid only after some
show the marginal increase observed for the adiabatic cases threshold pressure has been reached. The enthalpy of the two-
but is rather simply flat. Ilk is an important aspect of these phase mixture would be the same as that of the hot liquid
results since only one value of saturation was computed per water comected for heat losses along the injection line, the
section, effectively making the experiments one-dimensional. kinetic energy being negligible in this case. T& method is a
Recognition of the variation in saturation was an important modification of that used by Miller (1951) and Arihara
confirmation of acceptable results. This indeed was the case. (1976). This method was used to inject fluids at low enthalpy
These results indicate that heat losses will affect the to obtain relatively low steam fractions and it proved useful
measunsmentsbut the main features will be unaffected. llms a since steam table values could be used to determine the
flat saturation profile, which is required to evaluate relative enthalpy of the injected fluids, using either the temperature or
permeability accurately, is still present and is of sufficient the pressure In the second method, the streams of steam and
length. hot water were first mixed before entering the core. Due to
SPE 36882 DETERMINATION OF RELATIVE PERMEABILllV FOR STEAM-WATER FLOW IN POROUS MEDIA 7

difficulty in keeping both streams close to saturation, steam temperature and pressure. Typically stabilization took three to
was superheated by a few degrees and liquid water was kept a five hours, though the measurements reported here were taken
few degrees below the boiling point. l%is too enabled the use after at least eight hours. Once a steady state had been
of steam table values for computation of the phase &actions. confmne~ the measurements of temperature and pressure
This approach was used to obtain high steam fraction values. were recorded together with the heat flux sensors readings.
The X-ray CT scans were then taken at locations where the
After assembling the core and the auxiliaries, the experiment dry and wet scans had been taken to obtain CT=P values.
was initiated by first determining the porosity of the core. This These scans were then processed into saturation images using
was done by taking X-ray CT scans of the core at various Equations 2 and 3. Figure 8 shows the saturation distribution
locations when it was dry and again when it was fully obtained from the X-ray CT scanning at a location along the
saturated with water. The initial scan, referred to as the dry core. The saturation profiles presented in this paper were
SCtJ?l,WSS pWfOSltled to obtain CT values. Following this, a obtained by averaging the saturation values over a cross
steady stream of water at low % ow rate (5 cclmin) for sectional area of the core. In general all of the images showed
sufficiently long time (12 hours) to saturate the core very uniform saturation distributions for the most sections at
completely with water. A second scan, referred to as the WI all flow rates.
scan, was then conducted to obtain CTWt values at the same
locations as the&y scan was performed. By using Equation 1
and the two sets of images obtained at every location scanned 0.4
it was possible to determine the porosity distribution of the
core. The avesage porosity was found to be about 20°/0.After
the porosity distribution had been determined the absolute
permeability was determined by flowing water at different
flow rates and measuring pressures along the core. After
determining the absolute permeability, the core was brought to 0,2
experimental conditions by injecting hot water at slowly
increasing temperatures. The heating stage at low flow rates
provided an additional opportunity to check the permeability
of the core at higher temperatures. The permeability values at
higher temperatures were also found to be close those e
measured at the room temperature. (Werago: t3. 2eee97
Figure 7: A selected image for porosity distribution obtained
In general all of the images gave an average porosity of
from the X-ray CT scars.
20i0.5Y0. In spite of this uniform value some images had
regions of local variations in porosity. Figure 7 shows the
Table 1: Summ q of important experimental data.
porosity distribution obtained ffom the X-ray CT scanning at a
location along the core. Some of the images show zones with step Waler, qsteamy &t
somewhat different porosity. It is not clear whether the cclmin cclmin %0
anomalous zones are due to larger pores or due to a different 1 ’15 0 15-10
packing of sand grains. Otherwise, the porosity over the most 2 13<6 414.6 22
of the core length is very close to the average porosity. This 4 7.78 274 38
core can therefore be considered a close approximation to a 5 7.82 1062 50
uniform porous medium. 6 3.46 625.7 53
Once the target experiment temperature had been reached, E7 H3.49 762.6 64
the core was allowed to attain thermal equilibrium before any % 10.16 [1007.8 68
readings were taken. During the experiment the phase 9 10!19 I1OO7.7 68
fictions of the injected fluids were changed 14 times while
attempting to increase the steam flaction (and steam saturation
in the core). Each of these 14 attempts will be referred as steps
in the rest of the paper. For the fmt four steps only the hot
water line was used, The steam ffaction was adjusted by
changing the injection temperature and the flow rate. For the
subsequent ten steps both the steam and hot water lines were
used. A summary of the important results are given in Table 1.
Steady-state conditions were recognized by the stabilization of
8 W. AMBUSSO, C. SATIK AND R. N. HORNE SPE 38882 .

1 phase these tubes were filled completely with water. By this

method water in these tubes was assumed to be in contact with
water in the core. In general all of the pressure measurements
reflected the expected behavior i.e. decreasing values along
the core from the injection end.

- ~, ●-
II
.-
.*
<s9> : 8.307358 .*
.-
Figure & A selected image for steam saturation distribution
.mnm
obtained fiwmthe X-ray CT scan.

Figure 9 shows all of the saturation profiles obtained during

the experiments. In general all of the saturation profiles show Figure 9: Saturation profiles for all of the steps conducted
a decreasing trend horn the injection end to the production
during the experiment.
end which was also observed in the numerical simulation
results for the non-adiabatic case. The fnt and second steps of
injection show a few irregular trends at 17 cm and at 25 cm
horn the injection point. These trends are also repeated to a
lesser degree at the same points at higher saturations. These
anomalies are attributed to the inhomogeneities existing in
permeability or porosity. These are however minor and the
saturations still reflect the general trend. In addition, the
j; 1+ El
values of saturation are never really constant but change +Tmthln
gradually. Thus the flat saturation profiles are not always m ~
-a- hw-ui

“jlar”. However the saturation values change only slightly Oswt,aam maa
~bomhinuem
over the most of the core length and can be averaged over an
interval to a representative value. In addition, from previous Figure 10: Temperature profiles for all of the steps conducted
experience with other experiments (e.g. oil and water) relative during the experiment.
permeability typically changes monotonically with saturation
by small amounts. Therefore, relative permeability computed
over regions where saturations vary by less than M ‘Acan be
considered to be constant. The saturation profiles shown in .-w
*fin
Figure 9 reveal a number of other interesting features. The +*
capillary end-effkcts are observed at low steam flow rates with +Snu
.-
high steam fraction. This can be seen for steps that are —mm.
—Ten
different in rates but have the same steam tkactions (e.g. Steps --
+rn
4 and 5), This suppmts some of the results obtained from the +~
+~
simulation where the capillary end-effect is very strong at
small flow rates.
Figure 11: Pressure profiles for all of the steps conducted
Figures 10 and 11 show steady-state temperature and pressure
during the experiment.
profiles obtained during the experiment respectively. As
described in the experimental apparatus section, the
thermocouples were inserted in ceramic tubes within the outer
The experiment was not conducted under perfect adiabatic
most layer of the epoxy. Thus the thermocouples probably did
conditions since it was not possible to use guard heaters. Thus
not make gcwd contact with the core. This might have led to
the interpretation of the results must consider heat losses and
lower tempemture readings than expected. The pressure
the flowing fkactions must be comcted accordingly. Critical
readings were taken using teflon tubes attached on the core
to the evaluation of the flowing tlactions is the knowledge of
body. To ensure that the pressure readings were for the water the injected enthalpy and the heat losses. The heat losses on
SPE 36682 DETERMINATION OF REtAWVE PERMEABILITY FOR STEAM-WATER FLOW IN POROUS MEDIA 9

the core body were computed from the measurement of the sustaining steam quality at very high and very low values.
heat flux directly. Determining heat losses along the injection These end-point values are, however, inferred from the
lines were however a major challenge. They were estimated relative perrneab]lity curves and are about 200/0for the water
born the temperature drop while injecting water during the and less than 10% for the steam phase.
heating process. To determine the flowing fractions at a
particular poin~ the heat losses upstream of the point under
1
consideration were evaluated and subtracted 6om the total
energy at the injection point. The heat lost in the injection Iine
was estimated from the plot of heat loss rate vs. injection 0.8
temperature (see Ambusso, 1996 for details). To compute the
flowing fictions, Equation 6 was used after Q had been
calculated from the summation of the heat lost on the body
and the heat lost along the injection line, These were then I
■

8
converted into volumetric flow rates for the prevailing
temperature and pressure. Since the pressure and therefore the
specific volume of steam changed along the core, the
volumetric flow rates were computed for all of the points
along the core. The volumetric flow rates were surprisingly
similar and generally did not differ by more than 20/0over 5
0.2

0 LLLsL-
0 0.2 0.4 0.6
Wanr amtumal
0.8 1

cm intervals.
Figure 12: Relative permeability for steam and water.
‘Ile next parameter of interest was the temperature dependent
viscosity for water which varied between 252 and 211 x 104 Several relative permeability relations for flow of steam and
kg/m-s. The arithmetic mean of the values at the two end water in porous media derived horn experiments have been
points was used for a given interval. A final comection to the proposed in the past (Chen et al., 1978; Council and Ramey,
results was to include the errors due to pressure 1979; Vennz 1986 Clossman and Vinegar, 1988). In all of
measurements. This was done for all of the intervals. The the curves reported in the past the relative permeability for
error was estimated to be order of +X1.5psi. Finally, the one or more of the phases have tended to follow the relations
relative permeability values were computed from the obtained by Corey (1954) for nitrogen and water. However
experimental data using Equations (7) and (8). l%e results are none of the previous investigators have measured saturation
plotted on Figure 12. The residual limits are not well defined directly as in the experiment reported here. As a result none
in the experimental results, because it was not possible to of them has measured the pressure of a single phase alone
inject steam at 100°/0quality due to excessive condensation over any interval. Unlike previous investigations, these
aIong the injection lines. It was also not possible to estimate results show that the relative permeability for both phases are
the steam relative permeability at low saturations as the enhanced in comparison to relations obtained by Corey
correction for the enthalpy of the injected fluid was very close (1954).
to the correction in heat lost tlom the injection line and the
core body. These end points are however inferred horn the The principal feature of these measured relative permeability
relative pcrmeabili~ curves and are about 20°A for tbc water curves is their close similarity to the so-called “X curves”. Usc
and less than 10°/0 for the steam phase. The relative of the “X curves” for geothermal simulation has been
permeability for the steam and water phases vary common, but until now has been based only on philosophical
approximately linearly with saturation. In view of the arguments.
common usage of so-called “X curves” in numerical
simulations of steam-water flow, this is a rather fommate ACKNOWLEDGE NTS
result.
This work was supported by DOE contracts DE-FG07-
901D12934 and DE-FG07-951D13370, the contribution of
CONCLUSION
which is gratefully acknowledged. Tbe authors also would like
The relative permeability curves presented in this paper have to thank Louis Castanier of Stanford University for his
been derived from experiments in which the saturations within assistance in the design of the experimental apparatus and our
the core have been measured by using an X-ray CT scanner. undergraduate researchers Nia Fong and Thembile Mtwa for
Fwtlwrmore the saturation profiles have been shown to follow their assistance in conducting experiments. The thermal
very closely what is expected from the numerical simulation. simulator (STARS) was made available by Computer
The residual saturations are not well defined in the Modelling Group, Calgary, Alberta. This paper is a
experimental results due to the difficulties encountered in modification of an earlier work presented at the 1996
10 W. AMBUSSO, C. SATII( AND R. N. HORNE SPE 38882

Geothermal Resources Council Annual Meeting, Portland 1944,

Oregon.
[14] A.A. Heibq H.T. Davis and L,E. Striven, 1983, Effect of
REFERENCES Nettability on Two-Phase Relative Penneabilities and
Capillary Pressure, SPE paper 12172.
[1] W.J. Ambusso, 1996. Experimental Determination of
Steam-Water Relative Permeability Relations. MS Thesis,
[15] R.N. Home and H.J. Ramey, 1978. Steam/Water
Stanford University, Stanford. CA
Relative Permeabilities From Reduction Da@ GRC Trans.
Vo]. 2 pp 291.
[2] W.J. Arnbusso, C. Satik and R.N. Home, 1996. A Study
of Relative Permeability For Steam-Water Flow in Porous
[16] F. Miller, 1951. Steady Flow of Two-Phase Single-
Media. Proc. Of 21st on Stanford Workshop on Geothermal
Component Fluids Through Porous Medi& Trans. AIME, 192,
Reservoir Engineering.
205.
[3] N. Arih~ 1976. A Study of Non-Isothermal Sigle and
[17] J.S. Osob& J.J. Richardson, J.K. Kerver, J.A. Hafford and
Two-Phase Flow Through Consolidated Sandstones, PhD
P.M. Blair, 1951. Laboratory Measurements of Relative
dissertation, Stanford University, Stanford. CA.
Permeability. Transactions, AIME, VO1.192,pp 47-55.
[4] R.H. Brooks and A.T. Corey, 1964. Hydraulic Properties
[18] J. Piquemal, 1994. Saturated Steam Relative
of Porous Medi% Colorado State University, Hydro paper
Permeabilities of Unconsolidated Porous Media. Transport in
No.5.
Porous Media 17, pp 105-120.
[5] H.K. Chen, 1976. Measurement of Water content in
[19] J.M. Sanche~ 1987. Surfactant Effects on the two-
Porous Media Under Geothermal Fluid Flow Conditions. Ph.
Phase Flow of Steam/Water and Nitrogen/Water in an
D. Thesis, Stanford University, Stanford, California
Unconsolidated Permeable Medium. Ph. D. Thesis, University
of Texas, Austin, Texas.
[6] H.K, Chen, J.R. Council and H,J. Ramey, Jr., 1978.
Experimental Steam-Water Permeability Curves. GRC Trans.
[20] C. Satik, W. Ambusso, L.M. Castanier and R.N. Home,
vol. 2 pp 102-104.
1995. A Preliminary Study of Relative Permeability in
Geothermal Rocks, GRC Trans. Vol. 19 pp 539.
[7] P.J. Clossman and H.J. Vinegar, 1988. Relative
Permeability to Steam and Water at Residual Oil in Natural
[21] M.L. Sorey, M.A. Grant and E, Bradford, 1980.
Cores; CT Scan Saturation. SPE Paper 174449.
Nonlinear Effects in Two Phase Flow to Wells in Geothermal
Reservoirs. Water Resources Research, VOL. 16 No. 4 pp
[8] A.T. Corey, 1954. The Interrelations Between Gas and Oil
767-777. ‘
Relative Permeabilities, Producers Monthly Vol. 19 pp 38-41.
[22] K.S. Udell, 1982. The Thermodynamics of Evaporation
[9] J.R. Counsil, 1979. Steam-Water Relative Permeability.
and Condensation in Porous Media. SPE Paper 10779.
Ph. D. Thesis, Stanford University, Stanford California.
Resented at SPE California Regional Meeting San Francisco,
CA.
[10] J.R. Council and H.J. Ramey, Jr., 1979. Draiiage
Relative Permeabilities Obtained From Steam Water Boiling
[23] M.A. Venn& K. Ruess and P.A. Whherspoon, 1985. An
Flow and External Gas Drive Experiments, GRC Trans. Vol. 3
Experimental Investigation of Two-Phase Steam/Water
pp 141-143.
Relative Permeability in Unconsolidated Porous Medium.
Eurth Science (Berkeley California): Vol 8,5-8,
[11] RA. Fulcher Jr., T. Ertekin and C.D. Stahl, 1983. The
Effects of Capillary Number and Its Constituents on Two-
[24] M.A. VermA 1986. Effects of Phase Transformation of
Phase Permeability Curves. SPE Paper 12170.
Steam-Water Two-Phase Relative-Permeability. Ph.D. Thesis,
University of Califomiq Berkeley.
[12] M.A. Grant 1977. Permeability Reduction Factors at
Wairakei, Paper presented at AIChE-ASME Heat Transfer
Conference, AIChE, Salt Lake City, Utah, August pp 15-17.

[13] G.L. Hassler, 1944. Method and apparatus for

permeability measurements. U.S. Pat. Off., Washington, D.C.,