You are on page 1of 14

Acta Geophysica

vol. 54, no. 4, pp. 399-412


DOI 10.2478/s11600-006-0029-4

Compressibility of porous rocks:


Part II. New relationships
Ali Ahmed JALALH
Petroleum Engineering Department, Miskolc University, Miskolc, Hungary
e-mail: jalalhali@yahoo.com

Abstract
Pore volume compressibility is one of the physical properties of a reservoir
that must be specified in many reservoir-engineering calculations. The main objective of this work is to provide new general formulas for pore volume compressibility versus porosity on the basis of measured compressibilities of some limestone
and sandstone rocks in a wide range of porosity values and of varied type; the
measurements were performed on Hungarian reservoir rock samples. The obtained
laboratory results were compared with the published correlations of consolidated
limestone samples as well with values for friable and strongly consolidated sandstones. The validity of using compressibility data from the literature was investigated. The measured data showed poor agreement with the published correlations.
The first approach to find better and more accurate rock compressibility correlations consisted of combing all the data available from the literature, using the
same formula of Hornes type. However, this attempt did not give satisfying fitting
results. In the next step, by using twelve different fitting formulas, and other comprehensive nonlinear fitting regression programs, new rock compressibility correlations for limestone and sandstone rocks, with better goodness of fit, were developed. These new correlations can be generalized and used for most of oil and gas
reservoirs.
Key words: pore volume compressibility, rock compressibility, reservoir characterization, rock properties, petrophysics.
1. INTRODUCTION

Reservoir rocks become more compressible as the pore pressure drops. Rocks compressibility can provide energy to drive oil out of reservoirs. In order to evaluate the
drive energy that the rock can provide in the process of oil production, the compressi 2006 Institute of Geophysics, Polish Academy of Sciences

Unauthenticated
Download Date | 11/12/16 4:46 PM

A.A. JALALH

400

bility coefficient has been defined and measured for most reservoirs. A number of researchers, i.e., Hall (1953), Van der Knaap (1959), Newman (1973), Laurent et al.
(1993) and Jalalh (2006a, b), conducted a series of theoretical and experimental studies on the compression of reservoir rocks.
Newman Limestone
Newman Sandstone

-6

Pore volume compressibility, Cpc (10 1/psi)

100

10

1
0

10

15

20

25

30

35

40

Porosity (%)

Fig. 1. Pore volume compressibility of sandstone and limestone at 75% lithostatic pressure versus porosity (after Newman 1973, modified).

Figures 1 through 3 present a number of laboratory measurements of pore compressibility data from the literature (sources are indicated in the figures). Although the
most extensive measurements come from Newman (1973) who run tests on 256 cores
of limestone and sandstone from 40 reservoir rocks of porosity in the range between 1
and 35%, they cannot be included because the compressibility values were computed
at 75% lithostatic pressure (as assessed on the basis of the depth from which his samples were obtained). Newman (1973) also compared the results reported by other researchers, but his comparison may be not accurate; the same concerns the correlations
developed by Horne (1990) on the basis of Newmans (1973) data (Jalalh 2006a).
Data from Kohlhaas and Miller (1969) show different pore volume compressibility
values in comparison to the rest of the published data (Fig. 2). Laurent et al. (1993)
used three samples of limestone in his study. Then he repeated the compressibility test
on the same samples. In Fig. 3 there are presented other results for these three samples
and their recycling runs are included. A great scatter of data is observed.
Hall (1953) published a relation between pore volume compressibility, Cpc , and
initial porosity, , on the basis of the results of laboratory measurements. Afterwards,
it was called Halls plot; the relation simulated by the empirical formula was as follows:

Unauthenticated
Download Date | 11/12/16 4:46 PM

COMPRESSIBILITY OF POROUS ROCKS: PART II

C pc =

1.78 105

0.4358

2.587 104

C pc =

1/ psi ,

0.4358

401

1/ MPa .

(1)

Horne (1990) obtained trends of pore volume compressibility vs. initial porosity
for consolidated limestones, consolidated sandstones and unconsolidated sandstones.
The three formulas estimated for different lithology types are presented below.
For consolidated limestones
C pc = exp ( 4.026 23.07 + 44.28 2 ) 106 1/ psi .

(2)

For consolidated sandstones


C pc = exp ( 5.118 36.26 + 63.98 2 ) 106 1/ psi .

(3)

For unconsolidated sandstones (where 0.2 )


C pc = exp ( 534.01 ( 0.2 ) ) 106 1/ psi .

(4)

1000
Hall, 1953
Fatt, 1958
Van der Knaap, 1959

-6

Pore volume compressibility, Cpc (10 1/psi)

Von Gonten and Choudhary, 1969


Carpenter and Spencer, 1940
Dobrynin, 1962
100

Jalalh, 2006b
Toth and Bauer, 1988
Kohlhaas and Miller, 1969
CONSOLIDATED SANDSTONE

10

1
0

10

20

30

40

50

Porosity (%)

Fig. 2. Pore volume compressibility of sandstone versus porosity obtained in this study and
from the literature, sources as indicated.

Unauthenticated
Download Date | 11/12/16 4:46 PM

A.A. JALALH

402
1000

Hall, 1953
Van der Knaap, 1957
Von Gonten and Choudhay, 1969
Pore volume compressibility, Cpc (10-6 1/psi)

Laurent, 1993
Jalalh, 2006b
100

CONSOLIDATED LIMESTONE

10

1
0

10

15

20

25

30

Porosity (%)

Fig. 3. Pore volume compressibility of limestone versus porosity obtained in this study and
from the literature, sources as indicated.

Many reservoir managers have assumed that the compressibility of rocks calculated from Halls empirical formula has the same value as the laboratory measurement, so they have not performed core experiments, but calculated compressibility
quite easily from eq. (1). For many years, Halls equation has been the main method
for calculating rocks compressibility in many commercial softwares for well test interpretation and reservoir numerical simulation.
In laboratory, pore volume of core, Vp, can be measured at a given pore pressure,
Pp, with outer stress being constant, and then the relationship of pore volume versus
pressure can be plotted, from which pore volume compressibilities of rock, Cpc and
Cpp, can be calculated according to the following equations:
C pc =

1
Vp

V p
P ,
c Pp

C pp =

1
Vp

V p

,
Pp Pc

(5)

where Cpc and Cpp are the pore volume compressibility and the effective pore volume
compressibility, respectively, Pp is the pore pressure, and Pc is the confining pressure.

Unauthenticated
Download Date | 11/12/16 4:46 PM

COMPRESSIBILITY OF POROUS ROCKS: PART II

403

However, Halls or Hornes formulas are to be applied with caution, if core data
are not available. Below, several points are listed supporting our suspicions as to the
direct use of published Halls and Hornes solutions in reservoir analysis and well
tests (Jalalh 2006a), and encouraging us to develop new ones.
In a typical range of reservoir porosity values, the rock compressibility from
Halls plot is usually larger than the reservoir liquids compressibility. For example,
according to Halls formula, a rock of 10% porosity has compressibility of 4.8710-5
1/psi, and Hornes correlation gives 8.4310-6 1/psi, while the formation water has an
approximate compressibility of (2.763.45)10-5 1/psi.
Halls plot gives the same value of compressibility for rocks of different lithology if only they have the same porosity, regardless of the differences in rigidity. For
the same porosity, Hornes correlation gives a double value as compared to Halls one.
Halls and Hornes plots present pore volume compressibility vs. porosity. Accurate estimation of hydrocarbon reserves by using newly developed material balance
equation (Fetkovitch et al. 1991), needs information about changes in formation compressibility with declining reservoir pressure (i.e., effective stress = Pc Pp).
Hornes curves are constructed for each rock type separately, while Halls correlation represents pore volume compressibility for reservoir rocks of mixed type (i.e.,
sandstone, limestone and friable rocks).
Hall has developed his correlation on the basis of laboratory measurements,
while Horne summarized the published data for compressibility by Newman and then
developed correlation for each rock type. The original data of Newmans compressibility values were computed at 75% lithostatic pressure (on the basis of the depth from
which his samples were obtained), so the comparison with the data from other researchers may be not accurate.
The newly developed methods by Fetkovich et al. (1991) for Original Gas in
Place (OGIP) calculation and Yale et al. (1993) for Original Oil in Place (OOIP) calculation depart from the methods proposed earlier since they used a variable rather
than fixed or average value of formation volume factor and pore volume compressibility. These methods of calculation properly integrated pore volume compressibility and
formation volume factor effects over the full pressure range of investigation. Jalalh
and Bdi (2004) presented an example of using variable compressibility for improving
reservoir analysis.
In this paper, new rock compressibility correlations are presented as an alternative approach to laboratory measurements. The Cpc-original data published during
30 years have been collected (Figs. 1 through 3) and statistical analysis has been applied to get correlations similar to Hornes one, named modified Hornes correlation
(mod-Horne). The next step was to find a new empirical formula and create new rock
compressibility trend on the basis of the Cpc-original and recent results obtained by
the author (Jalalh 2006b).

Unauthenticated
Download Date | 11/12/16 4:46 PM

A.A. JALALH

404

2. MODIFIED HORNES CORRELATION

Figure 4 displays the collected published limestone rock compressibility data,


which are referred to as the Cpc-original, and Hornes trend curve versus porosity.
Two dashed lines represent 68% of the upper and lower confidence intervals from
standard deviation of Hornes correlation.
Motulsky and Christpoulos (2003) have presented an extensive overview of practical applications for linear and nonlinear regression. They concluded that the sum-ofsquares procedure is a useful statistic calculation to compare models. The sum-ofsquares (SS) is the sum of the squares of vertical distances of the points from the
curve. It is expressed in the units used for the Y-values, squared. Standard (least
squares) nonlinear regression calculation is performed by varying the values of the
model parameters to minimize the SS.
Using the equation below, we can calculate the value of root mean square (Sy,x)
from the sum-of-squares and degrees of freedom (NP):
S y,x =

SS
,
N P

(6)

-6

Pore volume compressibility, Cpc (10 1/psi)

where Sy,x is the standard deviation of the vertical distances of the points from the line,
N is the number of data points, and P is the number of constant parameters used in the
model.
Figures 4 and 5 show a comparison of Hornes and mod-Hornes models with the
upper and lower confidence intervals (dashed lines). The confidence interval is calculated using the standard deviation (Sy,x) of sample means obtained from eq. (6). On
80
75
70
65
60
55
50
45
40
35
30
25
20
15
10
5
0
-5
-10

Horne
68% Upper confidence interval
68% Lower confidence interval
Cpc -original

10

15

20

25

Porosity (%)

Fig. 4. Published limestone pore volume compressibility data, Cpc-original, and Hornes
trend curve versus porosity with upper and lower 68% confidence interval.

Unauthenticated
Download Date | 11/12/16 4:46 PM

-6

Pore volume compressibility, Cpc (10 1/psi)

COMPRESSIBILITY OF POROUS ROCKS: PART II


80
75
70
65
60
55
50
45
40
35
30
25
20
15
10
5
0
-5
-10

405

mod-Horne
68% upper confidence interval
68% lower confidence interval
Cpc -original

10

15

20

25

Porosity (%)

Fig. 5. Published limestone pore volume compressibility data, Cpc-original, and mod-Horne
curves versus porosity with upper and lower 68% confidence interval.

both figures, the data are very scattered and points of porosity of less than 5% (i.e., 8
points out of the dashed lines) are laying out of the confidence interval of Hornes and
mod-Hornes regression lines. Table 1 shows a comparison of the statistical correlation data.
T ab le 1
Descriptive statistics for data from Horne and mod-Horne models (limestone)

Summation of data
Mean of data set

Cpc original

Cpc Horne

Cpc mod-Horne

5.3110-4
1.2910-5

6.68110-4
1.62910-5

5.3810-4
1.3110-5

9.96410-6

7.90310-6

Average of standard deviation (Sy,x average)

The descriptive statistic data in Table 1 show that the mean and summation data
from Hornes model are higher than the original rock compressibility data. It means
that Hornes trend does not fit the data used in this study. Using Hornes model formula, and incorporating the latest rock compressibility data available in the literature
and our laboratory measurement results for limestone and sandstone, the better fitting
correlation constants for limestone rocks have been found, and Hornes formula type
was preserved. This correlation will be referred to as modified Hornes correlation,
mod-Horne. This modified Hornes correlation can serve as a better empirical formula for limestone and sandstone:

Unauthenticated
Download Date | 11/12/16 4:46 PM

A.A. JALALH

406

C pc -mod -limestone = exp ( 3.9952 33.933 + 98.040 2 ) 106 1/ psi ,

(7)

C pc -mod -sandstone = exp ( 3.4895 15.249 + 31.599 2 ) 106 1/ psi .

(8)

A comparison of the descriptive statistic data in Table 1 for limestone and in Table 2 for sandstone indicates an improvement in the fitting of the curve to the data by
using mod-Horne model. The mean and summation of data in mod-Horne model are
much closer to the original data. This is clearly seen by the lower value of the standard
deviation of the mod-Horne model (Tables 1 and 2, Figs. 4 and 5).
T ab le 2
Descriptive statistics for data of Horne and mod-Horne models (sandstone)

Summation of data
Mean of data (average)

Cpc original

Cpc Horne

Cpc mod-Horne

5.57510-4
8.57710-6

2.70610-4
4.16410-6

4.64710-4
7.14910-5

8.50410-6

6.95810-6

Average of root mean square (Sy,x average)

3. NEW ROCK COMPRESSIBILITY CORRELATION

The use of the modified formulas gives more accurate compressibility values for most
reservoir rocks, better than the original Hornes formulas. After the modification of
Hornes formulas (eqs. (2) and (3)) we attempted to find a simple and accurate formula which would give more precise pore volume compressibility values, considering
all our measured compressibility data and the data available from the literature. We
used 12 different formulas combining porosity and pore volume compressibility values with two matching parameters and used spreadsheet of Microsoft Excel to find the
best match for the used sandstone and limestone compressibility data. Then we transferred the data into professional fitting regression program CurveExpert 1.38 (Daniel
2003).
In the last phase of our calculation we used statistical program Prism-4 (GraphPad Software Inc. 2005). By this program, we were able to enhance the fitting parameters obtained from the CurveExpert program.
The twelve formulas adopted in the correlation calculations are the standard
equations applied in each professional regression software. In the CurveExpert program, two types of fitting curves were used. The first group falls under the category of
regression curves (which can be further subdivided into linear and nonlinear), which
attempt to minimize the difference between fitted curve and data points. The second
group, interpolations, ensure that the curve fit passes exactly through each data point.
In the first group, the linear fit, quadratic fit, polynomial, exponential, logarithm, reciprocal logarithm, power and several other modifications of fitting were applied. In
the second group, the Lagrange interpolation, linear spline, quadratic spline and others
were used (Jalalh 2006b).

Unauthenticated
Download Date | 11/12/16 4:46 PM

COMPRESSIBILITY OF POROUS ROCKS: PART II

407

A new compressibility correlation for limestone rocks

The best fitting result was obtained by the Harris model, y = 1/(a + bxc) (Daniel 2003).
This model gives correlation coefficient R = 0.95. Variables y and x in this model represent the pore volume compressibility and porosity values, respectively, while a and b
values are fit constants. The new limestone compressibility relationship is as follows:

1
C pc -new-limestone =
10-6 1/ psi .
1.05

2
2
1.022 + 1.681 ( )

(9)

Our newly obtained pore volume compressibility trend for limestone rocks with
porosity in X-axes is presented in Figs. 6a, b. Most of the limestone data lay within
the lower and upper band of 68% confidence interval. The descriptive statistics in Tables 3a, b, and plots in Figs. 6a, b and 7 support the new trend of our correlation for
the pore volume compressibility of limestone rocks. An additional support is provided
by the agreement of new fitting line and the descriptive statistical data, as presented in
Tables 3a, b.
Goodness of fitting
Degrees of freedom
R
Absolute sum of squares
Sy.x

-6

Pore volume compressibility, Cpc (10 1/psi)

100
38
0.95
1392
6.052

10

1
0

10

15

20

25

Porosity (%)

Fig. 6a. The new fitting curve: pore volume compressibility for limestone rocks versus porosity. The goodness of fit parameters are presented.
T ab le 3 a
Statistical data of the new fitting curve
for limestone rocks
Runs test
Points above curve
Points below curve
Number of runs
P value (runs test)
Deviation from model

Data
28
13
14
0.06028
not significant

Unauthenticated
Download Date | 11/12/16 4:46 PM

A.A. JALALH

408

T ab l e 3 b
Statistical data of nonlinear regression of new fitting curve
for limestone rocks
Cpc original

Cpc new fitting

-4

Pore volume compressibility, Cpc (10-6 1/psi)

Summation of data
5.3110
Mean of data (average)
1.2910-5
Average of standard deviation (Sy,x average)

80
75
70
65
60
55
50
45
40
35
30
25
20
15
10
5
0
-5

5.07210-4
1.23710-5
6.05210-6

Cpc new fitting


68% upper confidence interval
68% lower confidence interval
Cpc -original

10

15

20

25

Porosity (%)

Fig. 6b. The new fitting curve: limestone compressibility data Cpc-original, versus porosity
with upper and lower 68% confidence interval.

A new compressibility correlation for sandstone rocks

For the sandstone compressibility, we found the best fitting result using the same
model type as for limestone rocks (Jalalh 2006b). This model gives correlation coefficient R = 0.87. The following equation presents the new sandstone compressibility relationship:

1
C pc-new-sandstone =
10-6 1/ psi .
2.1412 + 4.064 2 ( )0.4652

(10)

Bates and Watts (1988) stated that it is not appropriate to use R as a main criterion whether the fit is reasonable. A high value of R tells us that the curve is coming
very close to the points, but does not tell us whether the fit is sensible in other respects. It may happen that the best-fit values of the parameters make no sense (e.g.,

Unauthenticated
Download Date | 11/12/16 4:46 PM

Pore volume compressibility, Cpc (10-6 1/psi)

COMPRESSIBILITY OF POROUS ROCKS: PART II


80
75
70
65
60
55
50
45
40
35
30
25
20
15
10
5
0
-5
-10

409

New fitting
68% lower confidence
68% upper confidence
Horne
68% lower confidence
68% upper confidence
Cpc -original

10

Porosity (%)

15

20

25

Fig. 7. The standard deviation of new fitting and Horne curves for the limestone compressibility data Cpc-original versus porosity with upper and lower 68% confidence interval.

negative rate constants) or the confidence intervals are very wide. Descriptive statistics data presented in Tables 4a, b and plot in Fig. 8 support the new correlation trend
of the pore volume compressibility of sandstone data with porosity.
T ab le 4 a
Statistical data of the new fitting curve
of sandstone rock compressibility
Runs test

Data

Points above curve


Points below curve
Number of runs
P value (runs test)
Deviation from model

31
34
29
0.162
not significant
T ab l e 4 b

Statistical data of nonlinear regression of the new fitting curve (sandstone)


Cpc original
-4

Summation of data
5.5710
Mean of data (average)
8.5810-6
Average of root mean square (Sy,x average)

Cpc new fitting


5.5810-4
8.5810-6
4.7610-6

Unauthenticated
Download Date | 11/12/16 4:46 PM

A.A. JALALH

410

Table 5 summarizes the summation of squares and degree of freedom of all used
models introduced in this study. It confirms the previously discussed results.
T ab le 5
Comparison of the sum of squares, SS, and degree of freedom, DF,
of the models used for limestone and sandstone samples
Lithology
Model name
Horne
mod-Horne
New

Limestone
SS
3.7710-9
2.3710-9
1.3910-9

Sandstone
DF
38
38
38

SS
7.8810-9
3.0010-9
1.3910-9

DF
62
62
62

Fig. 8. The new fitting curve of pore volume compressibility for sandstone rocks versus porosity. The goodness of fit parameters are presented.
4. CONCLUSIONS

1. The data presented in this paper are the first published results of pore compressibility measurements performed in such a wide range of porosity values and rock
types, including rocks from Hungarian hydrocarbon fields.
2. It was found that the pore volume compressibility values reported in this study
are in poor agreement with the published compressibility correlations. This is also
confirmed by values from the literature. Therefore, there is a necessity for laboratory
compressibility measurements in evaluating rock compressibility for a given reservoir.

Unauthenticated
Download Date | 11/12/16 4:46 PM

COMPRESSIBILITY OF POROUS ROCKS: PART II

411

3. Using equations of the same shape and type as Hornes equation, modified
Hornes formulas were constructed, which give better compressibility estimations.
4. A new rock compressibility correlation for limestone and sandstone rocks was
developed, based on the rock compressibility data available in the literature and our
own laboratory measurements.
5. To increase the accuracy of reservoir engineering calculations, the rock compressibility measurement should be one of the routine core measurements in the laboratory.

References
Bates, D.M., and D.G. Watts, 1988, Nonlinear Regression Analysis and Its Applications, Wiley
& Sons Inc., New York, 432pp.
Carpenter, C.B., and G.B. Spencer, 1940, Measurements of Compressibility of Consolidated
Oil-Bearing Sandstones, RI 3540, USBM.
Daniel, H., 2003, A comprehensive curve fitting system for Windows: CurveExpert program,
Version 1.38.
Dobrynin, V.M., 1962, Effect of overburden pressure on some properties of sandstones, Soc.
Pet. Eng. J. 360-366.
Fatt, I., 1958, Pore volume compressibilities of sandstone reservoir rocks, Trans. AIME 213,
362.
Fetkovich, M.J., D.E. Reese and C.H. Whitson, 1991, Application of a general material balance for high-pressure gas reservoirs, Paper SPE 22921 presented at the SPE Annual
Technical Conference and Exhibition, Dallas.
GraphPad Software Inc., 2005, Prism 4 for Windows, version 4.03 GraphPad Software, San
Diego California USA.
Hall, H.N., 1953, Compressibility of reservoir rocks, Trans. AIME 198, 309-311.
Horne, R.N., 1990, Modern Well Test Analysis: A Computer-Aided Approach, Petroway Inc,
Palo Alto, CA.
Jalalh, A.A., 2004, Pore volume compressibility measurement. In: Intellectual Services for
Oil & Gas Industry: Analysis, Solutions and Perspectives, vol. 3, University of Miskolc/ UFA State Petroleum Technological University, Miskolc.
Jalalh, A.A., 2006a, Compressibility of porous rocks: Part I. Measurements of Hungarian reservoir rock samples, Acta Geophys. 54, 3, 319-332.
Jalalh, A.A., 2006b, Determination and analysis of porous rock compressibility of Hungarian
rock samples, Ph.D. Dissertation, Miskolc University Library, Miskolc.
Jalalh, A.A., and T. Bdi, 2004, Effect of compressibility in calculation of Original Gas In
Place (OGIP), Intern. Sci. Confer. MicroCad, University of Miskolc.
Kohlhaas, C.A., and F.G. Miller, 1969, Rock-compaction and pressure-transient, analysis with
pressure-dependent rock properties, SPE 44th Annual Fall Meeting, Denver, CO.

Unauthenticated
Download Date | 11/12/16 4:46 PM

412

A.A. JALALH

Laurent, J., M.J. Bouteca, J.P. Sarda and D. Bary, 1993, Pore pressure influence in the poroelastic behaviour of rocks: experimental studies and result (SPE 20922).
Motulsky, H.J., and A. Christopouls, 2003, Fitting models to biological data using linear and
nonlinear regression (A practical guide to curve fitting), GraphPad Software Inc., San
Diego, CA.
Newman, G.H., 1973, Pore-volume compressibility of consolidated, friable, and unconsolidated reservoir rocks under hydrostatic loading, Soc. Pet. Eng. J. 129-134.
Tth, J., and K. Bauer, 1988, Deformation of porous rock structures 2. Pore volume deformation and pore compressibility, Oil and Natural Gas J. 65-69 (in Hungarian).
Van der Knaap, 1959, Nonlinear behavior of elastic porous media, Trans. AIME 216, 179-187.
Von Gonten, W.D., and B.K. Choudhary, 1969, The effect of pressure and temperature on
pore-volume compressibility, SPE 44th Annual Fall Meeting, Denver, CO (SPE 2526).
Yale, D.P., G.W. Nabor and J.A. Russell, 1993, Application of variable formation compressibility for improved reservoir analysis (SPE 26647).
Received 4 July 2006
Accepted 9 August 2006

Unauthenticated
Download Date | 11/12/16 4:46 PM

You might also like