Professional Documents
Culture Documents
Jones1988 Two Point Determination
Jones1988 Two Point Determination
Summary. Empirical equations are presented that accurately fit permeability, PV, or porosity data vs. net confining stress. Each
of these equations has four adjustable parameters. With little loss of accuracy, however, two of the coefficients can be preset.
Consequently, permeability, PV, or porosity measurements need to be made at only two confining stresses-e.g., at 1,500 and
5,000 psi [10.3 and 34'.5 MPa] to define the stress dependence within close tolerances completely.
With the techniques described, economical measurements of these stress-dependent properties can be made on a routine basis.
The results can be used to calculate PV compressibility and to estimate productivity declines resulting from permeability reduction
as deep, high-pressure reservoirs are drawn down.
Introduction
Many workers l - 29 have shown that the permeability, PV, and sandstone and conglomerate core plugs ranging in length from 1.6
porosity of a rock sample decrease with increasing confining or to 3 in. [4.0 to 7.6 cm]. The operation principle ofthis instrument
"overburden" stress. The decrease is most rapid at low stresses is similar to that described in Ref. 34. However, this instrument
and becomes progressively more gentle at higher stresses. Fig. 1 has been modified in several respects from the one described. First
shows three curves of the permeability ratio at a given stress to of all, the capability of making PV measurements has been added
permeability at zero net stress vs. net stress. Net stress in this case by incorporating an internal valve to the outlet end of the core
and throughout this paper is defined as the difference between holder. When this valve is open, gas (helium) can exit from the
isostatic (or hydrostatic) confining stress and the average pore plug for a Klinkenberg permeability measurement. When the valve
pressure. is closed, it seals the effluent end of the core plug, and helium is
Isostatic stresses rarely apply in an actual reservoir environment expanded from a chamber of accurately known volume into the core
because the vertical stress is generally greater than lateral stresses. plug, which was initially filled with helium at atmospheric pres-
It is difficult to justify using triaxial or even biaxial stresses in the sure. The PV of the plug is calculated from the ratio of the gas
laboratory, however, unless measurements are being made on pressure initially in the reservoir before expansion to the pressure
whole-core samples. The principal stress can be applied along the in the reservoir plus core plug after expansion, from a Boyle's law
axis of a whole-core sample by a hydraulic ram, and the two or- calculation.
thogonallateral stresses (approximated as being equal) can be ap- The second change was to increase the strength of the core holder
plied by pressure on a rubber sleeve. Core plugs, on the other hand, so that it can accommodate 10,000 psig [69 MPa] sleeve pressure.
are generally cut perpendicular to the vertical axis, and it is not Additionally, it has been provided with a pneumatically driven ram
easy to simulate reservoir stresses for plugs cut in this orientation. that applies an axial stress to the plug equal to the sleeve pressure.
Two different stresses cannot easily be applied to the rubber sleeve Sleeve pressures up to 10,000 psig [69 MPa] are applied through
unless a mechanical stress is superimposed on the hydraulic a pressure intensifier.
pressure. The third major modification was to automate the instrument fully.
Only core plugs were measured in this study, and because of the Up to 18 core plugs can be loaded into a carousel. The length, di-
difficulty of applying appropriate triaxial stresses, only isostatic ameter, and core plug identification are typed into a microcomputer
stresses were used. Teeuw 30 has shown how to translate hydrostat- that controls the instrument. The confining stresses, up to a maxi-
ic compaction data into uniaxial formation compactions if the Pois- mum of eight, are then entered. After all the plugs have been load-
son ratio of the rock is known or can be estimated accurately. ed and the information entered, the operation is completely
Andersen 31 and Andersen and Jones 32 have presented experimen- automatic. The final output includes the Klinkenberg permeabili-
tal data comparing the two stress modes. ty, Klinkenberg slip factor, Forchheimer inertial coefficient, PV,
The main difficulty with obtaining compaction data and and porosity for each confining stress selected. If two or more stress-
permeability-stress data is that mea~urements must be made at sever- es have been selected to be run on a particular plug, then
al stresses to define the curves adequately. Furthermore, in the past, permeability-vs.-stress and PV -vs.-stress curve fits are made au-
each ofthese measurements has been a tedious undertaking; hence, tomatically. In the instrument described earlier, the starting pres-
they have been expensive. This paper will demonstrate how these sure for permeability measurements was 100 psig [689 kPa]. In
curves can be generated from measurements at only two stresses the new instrument, the starting pressure has been increased to 250
and will explore the magnitude of errors that may be generated. psig [1.7 MPa]. The main reasons for the increase were to improve
Jones and Owens 25 developed a permeability-stress correlation that the accuracy of PV measurements, to increase the accuracy of the
permits a two-point fit, modified by Ostensen. 33 The equations inertial term, and to increase speed in the permeability measure-
presented in this paper should be more applicable over a wider stress ments. The minimum upstream pressure for the permeability meas-
range than the previous correlations. urements has been increased from 1 to 20 psig [7 to 138 kPa].
Another change was to provide three helium reservoirs instead
Experimental Measurements of one. The three reservoirs consist of a manifold, which has a rela-
All measurements in this study were made with an unsteady-state tively small volume, and a small tank and a large tank that can be
Klinkenberg permeameter-porosimeter on I-in. [2.5-cm] -diameter selectively connected to the manifold through valves. The manifold
alone is used as the reservoir for the Boyle's law expansion for
PVand for permeability measurements with tight core plugs. The
'Now with Core Laboratories (Division of Western Atlas Intl.).
. small tank is connected to the manifold for permeability measure-
Copyright 1988 Society of Petroleum Engineers ments on somewhat higher-permeability plugs. The largest tank is
SPE Formation Evaluation, March 1988 235
1000r---r--'r--.--~--~---r---.---.---r--~
800
6·2
600
400
200
100
80
60
.,;
E 40 Core Plug 8-2
~ -"
C
"- .S 0
+
"" s
20
.4
10
.3 8
6
.2 4
.1
2
o .2 4 .6 1.0
- e- c / c*
a. kpsi
Fig. 2-Straight-line transformation of permeability-stress
Fig. 1-Permeability VS. net stress data with best fit of Eq. 1. data.
connected to the manifold for yet higher permeability plugs, and inlet end of the plug is somewhat less than 250 psig [l. 7 MPa].
both tanks are connected for very permeable plugs. With the avail- At its outlet end, the pore pressure is atmospheric (0 psig). At the
able tank combinations, permeabilities ranging from about 0.01 to termination of the permeability determination, the inlet and outlet
3,000 md can be measured. Lower-permeability plugs « 10 /-td) pore pressures are about 20 psig [138 kPa] and 0, respectively. Be-
can be measured, but the time requirement becomes excessive. The cause these pore pressures are changing with both time and posi-
time required for each permeability and PV measurement varies tion in the plug, the net stress also varies accordingly. With a plug
from about 30 seconds to 5 minutes, except for the lowest- that demonstrates a considerable variation in permeability with net
permeability plugs, which require more time. stress, this means that the plug's permeability is also varying with
For the measurements described in this paper, all plugs were care- time and position. However, the mathematics used in the curve fit
fully cut into square-ended cylinders, then cleaned in a Soxhlet ex- of the pressure-time data generated during a permeability meas-
tractor with toluene and dried in a vacuum oven. The rubber sleeve urement assumes a constant permeability during that measurement.
used to confine the plugs was a 70-durometer Buna-n tube with Thus, if the permeability varies significantly during a single meas-
3/16-in. [0.48-cm] wall thickness. All measurements were made urement (these changes are greatest at the lowest stresses), then
in the order of increasing stress. The PV measurement was made the constant-permeability assumption distorts all of the parameters
first, followed by the permeability measurement without relieving resulting from that curve fit (k, b, and (3). Consequently. the mini-
the stress between the two measurements. The stress was then in- mum recommended confining stress is 1,000 psig [6.9 MPa] (1,500
creased to the next higher level without relieving it, and so on. psig [10.3 MPa] is the preferred minimum value).
We consider the lowest stress measurements to be the least reliable
with this instrument, and in general, routine measurements at con- Empirical Equations
fining stresses less than 1,000 psig [6.9 MPa] are to be avoided. The data of Fig. 1 have been fit by the empirical equation
The low-stress PV measurements suffer because a heavy-walled
rubber sleeve does not completely conform to microscopic irregular-
ities of the core surface at low stresses. Consequently, PV com- k=ko exp{ak[exp(-ala*)-I]}/(1+Ca) . .............. (1)
pressibility measurements at these low stresses tend to be too high.
Low-stress permeability measurements also tend to be less reliable The curves connecting the data points represent a "best fit" for
than the higher-stress measurements because at the start of an each core plug in that all four of the coefficients in Eq. 1 (k o, ak'
unsteady-state permeability measurement, the pore pressure at the a*, and C) were fit using least-squares techniques. Fig. 1 shows
.82
'~~O--~--~---L--~---L--~--~--~--~---Jl0'
a. kpSi
10
9 6-2
O. kpoi
8
.,
. A semilogarithmic plot ofEq. 2 results in a straight line with inter-
u
cept ko and a slope of (-ak)' Fig. 2 shows such a plot. These are
+ 3
the same data shown in Fig. 1. Notice that the abscissa ranges from
zero to one. The zero corresponds to a net stress of zero, and
,..a. one corresponds to infinite stress. Note that the right point on each
line corresponds to a net stress of nearly 10,000 psi [69 MPa]
(see Fig. 1).
PV reduction with increasing stress exhibits the same general be-
havior as the permeability reduction. Fig. 3 shows the ratio of the
2
stressed PV to that at zero stress as a function of net stress for the
same three core plugs. The curves connecting the data points
represent least-square fits by use of Eq. 3:
Fig. 4-Straight-line transformation of PV-stress data. Permeability and PV data from a number of core plugs were fit
by use of Eqs. 1 and 3. For a majority of the cases, the decay cons-
- 6-2
~
--
90
\
,\~
\
\\'%
600 t-
500 -- 9-1
-- --.-
-
---
400 - 7-1
\
">iE ". .....\ ~
~""
300
~
7-2.
- ..-
80
~'~ ......
"'<~.~.~ -----
......
8-1
9-2 ..-
---
- --
200 ~
10-1
..... -c
E
"" ...... ~..
,-::
70 b
0 ,
\0
w
a,kpsi M 100 t- -
+ 90
Fig. 6'"-Sensitivity of curve with low a k to c:r*. 80~ -
70
60r- -- 10-2
....
-
10 50
40,=-
-- 8-2
--
-
30 --
201- -
.2 .4 .6 8 1.0
1_e- a !3000
8~--~~~~~-.~ __j
6
8 ..
4
7
.. -
2
...
61-
..
0
~
1.0 t>
ID
.8 ......
+
I
-0 .6
~
E ~
>
...
t> .4
......•
\0
t--~"----~-a---e-~~-----!~--e-__~.J
9-2
+
.....
-'"
3
--
11-2
---- .....
I I I I
.4 .6 .8 1.0
1 _ e- a /3000
.04
Fig. 10-Two-polnt fits of PV-stress data .
.02
when the points at 1,500 and 5,000 psi [10.3 and 34.5 MPa] are
fit: Fig. 8 shows typical results for plugs ranging in permeabilities
from about 40 to nearly 700 md. In every case, the points chosen
.01L-~ __ ~__-L__-L__ __ __
~ ~ ~ __L-~~~
for the fit are the third and fifth points from the left side. These
o .2.4 .8 1.0 correspond to confining stresses of 1,500 and 5,000 psi [10.3 and
1_e- a /3000 34.5 MPa] and the corresponding net stresses of approximately
1,390 and 4,890 psi [9.6 and 33.7 MPa], respectively. For the
Fig. 9-Two-point fits of permeability-stress data-lower per- majority of the points, the fits are quite acceptable.
meabilities. Fig. 9 shows two-point fits for lower-permeability plugs. Lines
M-2 and M-3 were calculated from data presented by McLatchie
et at. II Note that, in general, the slopes of these lines tend to be-
come more negative with decreasing permeability.
2,000 and 5,000 psi [13.8 and 34.5 MPa] are the "true curves"
Fig. 10 demonstrates that PV data can also be successfully fit
which we are trying to approximate by using a u* of 3,000 psi [20.7
from data taken at two confining stresses, 1,500 and 5,000 psi [10.3
MPa] as discussed for Fig. 5, then the errors generated at a net
and 34.5 MPa]. The same arbitrary constants, 3,000 psi [20.7 MPa]
stress of 3,000 psi [20.7 MPa] are only 0.5%. This is the maxi-
for u* and 3 x 10 -6 psi -I [0.44 x 10 -6 kPa -I ] for C, were used
mum interpolation error between the measured stresses of 1,500
for these fits. Note that in Fig. 10, the point farthest to the left on
and 5,000 psi [10.3 and 34.5 MPa]. The maximum extrapolation
each line seems to lie slightly above the line. We believe that this
error at 9,000 psi [62.0 MPa] is 1.7%.
is a result of an experimental error rather than a flaw in the two-
Because u* values of 2,000 and 5,000 psi [13.8 and 34.5 MPa]
point technique. As discussed previously, this is probably a result
generally represent the extremes for the "best fit" curves for perme-
of incomplete conformance of the rubber sleeve to microscopic ir-
ability and PV data, we can conclude from the examples shown
regularities on the surface of the core at the lowest net stress, which
in Figs. 5 and 6 that interpolation errors are well within experimental
is approximately 390 psi [2.7 MPa].
accuracy, and extrapolated values are also quite accurate provided
The decrease in porosity with increasing stress can also be han-
that the extrapolation is not too big.
dled by this technique. First of all, porosity is related to PV by
We will now investigate the sensitivity of the curves to C. Fig.
the following relationship:
7 shows three curves in which u* is held constant at 3,000 psi [20.7
MPa], but C is varied from 0 to 10 x 10 -6 psi -I [0 to 1.45 x 10- 6
</>= Vp/(Vp + VG), .................................. (5)
kPa -I]. Fig. 7 shows that Chas an appreciable effect only at high
stresses. The deviation of the two outer curves from the center one
at 9,000 psi [62.0 MPa] is about ±1.3%. or, if rearranged,
.40f- 5
~
...
3:t:)
•<0
.2Ot-
- 9-2
iii
C-
W
I
.., ... ... - W-1
-
'&
x
..
+ >
Q.
c .10 - 0
... -- -
.08
.06
-.il
.......
- W-2
-
W-3
-
I I I I I
.04 00
0 .2 .4 .6 .8 1.0 2 4 6 8 10
1-e
-0"/3000 a. kpsi
Fig. 11-Two-point fits of porosity-stress data. Fig. 12-PV-compressibillty curves.
tional to PV. Thus, this group must replace Vp in the previous Eq. 4 may be differentiated with respect to stress to yield PV
treatment. This results in compressibility: