Академический Документы
Профессиональный Документы
Культура Документы
Permeability Measurement
Jeffrey T. Hawkins: Production Research and Development, Conoco Inc.,
Ponca City, Oklahoma
I
Hawkins
Porosity 20-2690
Permeability 9-26 md
Irreducible water saturation 5- 10%
Rock type Granular limestone
Fluid gravity 32 API
Reservoir temperature 230F
Reservoir pressure 3,000 psia
Gadoil ratio 400 scf/stb
10
waterflood. The second cycle waterflood results are often
different than the first cycle waterflood results.
Because of relative permeability hysteresis, only data
from a first cycle waterflood accurately represent the res-
ervoir waterflood. Obtaining core that has not already
been through the first cycle waterflood is difficult. The
reservoir may be under active water drive or waterflood,
or if a water-based mud is used in drilling, the core will
have been flushed with water during coring. In either case,
the core when it reaches the laboratory will already have
gone through a first cycle waterflood, thus native-state 0 .05 .1 . I 5 .2 .25 .3 .35 .4
analysis would be on the wrong cycle. POROSITY
During coring, the core is usually flushed by mud fil- Figure 1: Air permeability versus porosity.
trate. Normally, when fresh-state core plugs are taken
at the laboratory they are then flushed with brine to re-
move mud filtrate (continuing the first cycle waterflood).
The core plugs are then flushed back to S,, with oil, and are contradictory, with some studies showing significant
the relative permeability measurements are then made differences (Amaefule and Handy, 1982) between the two
on a second cycle! So, no matter to what lengths we have methods while other studies show agreement (Johnson et
gone to preserve native-wettability, the measurements al., 1959) between the two methods.
are made on the wrong cycle and are affected by hysteresis. In light of the historical controversy concerning mea-
Our experience shows that these hysteresis effects are more surement methods, relative permeabilities for this study
significant than minor alterations in wettability, espe- were measured three different ways on restored-state
cially for the non-wetting phase. The wettability resto- core. The three different methods were 1) unsteady-state
ration process, on the other hand, simulates the initial oil at ambient conditions, 2) steady-state at ambient con-
migration into the reservoir, and subsequent relative ditions, and 3) unsteady-state at reservoir conditions. In
permeabilities are measured on a first cycle waterflood. the figures used in this paper these three methods are
Cycle dependent relative permeabilities are why we abbreviated AC USS, AC SS, and RC USS. Because each
recommend restored-state analysis, even for fresh core. test was run on a separate sample, we have a comparison
We recommend using a bland, low fluid loss mud for of the methods on geologically similar plugs, rather than
coring, because the absence of wettability altering com- the same plugs.
pounds will leave the core easier to clean. We also care-
fully preserve the core, again, because the core will be
easier to clean if it is not exposed to oxygen or allowed ROCK AND FLUID PROPERTIES
to dry out. The relative permeability measurements were made on
Much work has been published in the literature con- core plugs from a granular limestone from an oil field in
cerning the effect of temperature on relative permeability the Middle East. Table 1 summarizes the pertinent rock
(Honarpour et al., 1986). However, the published data and fluid properties.
are contradictory. For example, some studies indicate a A crossplot of helium porosity versus air permeability
strong temperature effect (Edmonson, 1965), while other for the core plugs is shown in Figure 1. In general the
studies indicate no temperature effect (Miller and Ramey, core plugs possessed similar properties. Procedures for
1985). Likewise, there have been some studies published restoration to native wettability and for the three different
comparing the steady-state and unsteady-state methods measurement methods are described in the following sec-
of relative permeability measurement. Again, these data tions.
364 The Log Analyst September-October, 1989
Relative Permeability Measurement Methods
Oil dis-
placed .9
S,,, K,,@S,,, S,,, (YO
Sample no. (Yo PV) (md) (Yo PV) K,,, @ S,,, OOIP)
.8
Ambient condition unsteady-state >-
1 6.3 8.3 2.5 1.16 91.3 c-
A .7
2 -
4.9 6.0
- 11.2 0.80 87.5 m
6
Averages 5.6 1.2 6.9 0.98 92.4
Ambient condition steady-state I .6
w
3 6.8 9.1 19.4 0.43 19.2 W
4 7.2 7.1 21.6 0.46 16.7 a .5
5 6.9
- 6.3
- -
14.3 0.49 84.6 W
L .4
Averages 7.0 7.1 18.4 0.46 80.2 I-
Reservoir condition unsteady-state 4
w .3
6 10.0 8.3 24.9 0.37 12.3 E
7 9.0 8.5 21.0 0.48 76.9
8 10.0 9.6 24.2 0.53 73.1 .2
9 8.0 c
4.8
_ -
29.2 0.52
- 68.3
Averages 9.3 7.8 24.8 0.48 12.7 .I
n
W
.9
.8
>-
t
1 .7
m
I .6
oi
W
a .5
w
L .4
4
w .3
oi
.2
.I
n "
"0 .I .2 .3 .4 .5 .6 .7 .8 .9 1 0 .I .2 .3 .4 .5 .6 .7 .8 .9 1
WATER SATURATION WATER SATURATION
Figure 4: Reservoir condition unsteady-state relative perme- Figure 5: Oil relative permeabilities for all three methods with
abilities with basis of KO@ S,,,,. basis of K , @ S,,,,.
Craig's (197 1) rules of thumb, the curves have very low from the relative permeability curves, is the same for all
values of S,,, the crossover point (where oil and water three methods. This indicates that the mineral oils used
relative permeabilities are equal) occurs at a water satu- in the ambient condition tests (Blandol and Isopar) did
ration less than 50%, and the curves show a high value not alter the wettability of the restored-state core. Results
of K,, Q So,. In fact, one of the curves is actually off the from all three methods will now be compared. Because
graph and goes to a value of 1.16 for K,, Q So,. there are so many curves, the oil and water curves will
be compared separately.
Ambient Condition Steady-State
Figure 3 shows the ambient condition steady-state rel- Oil Relative Permeability Curves
ative permeabilities. Again, there is some scatter in the Figure 5 shows all of the oil relative permeability curves
data, and the curves for the most part are oil-wet. Note, for all three methods. While there is some spread among
however, that the So, values are not as low, and the K,, the curves, they generally have the same shape, with dif-
Q So,values are not as high as for the ambient condition ferent values of So,.At this point, remember that these
unsteady-state measurements. are not repeat tests on the same exact piece of rock, but,
rather, rocks from the same formation that are miner-
Reservoir Condition Unsteady-State alogically similar. Therefore, the major differences among
The reservoir condition unsteady-state relative perme- the curves can probably be attributed to differences in
abilities are shown in Figure 4.There is some scatter in rock characteristics. Also, the differences in So,correlate
the data and, again, the curves appear oil-wet in nature. well with differences in test conditions, as will be ex-
Note that, like the ambient condition steady-state mea- plained later.
surements, the reservoir condition data have higher So,
values and lower K,, Q So, values than the ambient con- Water Relative Permeability Curves
dition unsteady-state curves. All of the water relative permeability curves for all three
methods are shown in Figure 6 . Again, there is some
CoMPARISoN OF ALL THREE spread in the data. The ambient condition unsteady-state
The wettability of the rocklfluid pairs, as estimated results generally give a slightly higher value of waier rel-
I I
Hawkins
4
w .3
Ambient condition unsteady-state
Ambient condition steady-state
250 x
47 x 10-7
6.9
18.4
nc Reservoir condition unsteady-state 7 x 10-7 24.8
.2
1 1
.8 .8
c >.
k
-
7
m
-
-I
m
6 a
I .6 z .6
E c
W w
a a
W W
L .4 1 .4
4
w
&
4
W
gr
.2 .2
n
OO .2 .4 .6 .a 1 -0 .2 .4 .6 .8 1
ferences in Abrams numbers along with differences in relative permeability curves uses the equation (Schneider,
pore volumes injected caused the differencesin So,.While 1987)
the values of So, vary in general, the curve shapes are
very similar over most of the saturation range. Labastie
s*= ( S W - S w i )
(1 - Swz - s o r ) .
and co-workers (1 980) found a similar phenomenon while
varying flow rate in relative permeability experiments. This method starts and ends all the curves at normal-
They found that while So,changed with flow rate in gen- ized saturations of 0.0% and 100.0%, respectively. Figure
eral, curve shape remained the same. 7 shows two sets of data from each of the three mea-
surement methods, normalized with the standard meth-
SELECTION OF APPROPRIATE RESIDUAL od. Comparing Figure 7 with Figure 6 we see that this
OIL SATURATION normalization technique actually spreads the water rel-
The reservoir condition tests were selected as the best ative permeability curves farther apart. By making all of
model for field performance. These tests most closely the curves start at 0% normalized saturation we are mak-
approximate the field conditions both in terms of flow ing the assumption that they all have similar S , values,
rate and viscosity ratio and, also, the number of pore Examining Table 2, we see that this is a fairly reasonable
volumes injected. Averaging the reservoir condition re- assumption in that all the plugs had similar Sw,values.
sults we come up with an So,of 24.8%. However, forcing all the curves to end at a normalized
saturation of 100% is making the assumption that they
NORMALIZING OF THE RELATIVE all have the same So, value. Examining Table 2, again,
PERMEABILITY CURVES we see that this assumption is not valid. Because of dif-
If core plugs are of similar wettability, similar lithology, ferences in test conditions and differencesin pore volumes
and similar pore structure, and they have similar Sw,Val- injected, the different methods gave substantially different
ues, it is reasonable to expect that they would have similar So, values.
relative permeability curves. To compare the relative per- To take into account varying So,values, a new equation
meability curves on an equivalent basis, the curves need was developed for normalizing relative permeability
to be normalized. The standard method of normalizing curves. The new equation starts all the curves at a nor-