@

Pore-Volume Compressibility of Consolidated, Friable, and Unconsolidated Reservoir Rocks Under Hydrostatic Loading
G.

H. Newman,

SPE-AIME,

Chevron Oil Field Research Co.

Introduction The ~Seof pOre.vQ!Urn.~ ~~rn.nrewihilitv-norositv d----------= =_. --.-,

relations

cor___

in engineering calculations is well known. The correlations developed by Hall for both sandstones and limestones have been widely distributed. Van der Knaap published a similar correlation using limestone samples from a single well and also correlated the data with net pressure. Such correlations are attractive because of the simple relationship established. However, those correlations were intended only for well consolidated samples; correlations for friable or unconsolidated samples have not been published. This study compares our laboratory data with the published correlations of consolidated samples as well as with values for friable and unconsolidated sandstones. Compressibility values are presented for 256 rock samplesfrom 40 reservoirs ] 97 samples from 29 sandstone reservoirs and 59 samples from 11 limestone reservoirs. Porosities ranged from less than 1 percent to 35 percent. Compressibility values from the literature*3,s.$ for 79 samples are added, including Halls and Van der Knaaps.

unknown. Unconsolidated samples, on the other hand. present a much more complex problem, in that grain rearrangement is very likely during either coring or subsequent handling. The advent of the rubbersleeve core barrel much improved the chances of obtaining representative samples. We have some evidence that, if carefully handled, rubber-sleeve cores will provide reasonably undisturbed samples. However, even if the sand is captured undisturbed in the rubber sleeve, internal gas can expand the core during the trip to the surface. The history of all the samples used in this study is not complete, but most of the unconsolidated samples were obtained from rubber-sleeve cores.
Preparing the Samples

The Experiments
Sampliig

To obtain a representative sample of a formation for testing, one must avoid grain rearrangement. This problem is unlikely to occur with consolidated samples or friable samples containing some cementation, although the effect of removing the overburden is still

The consolidated and friable samples used in this study were generally plugs 1 in. in diameter and 3 in. long. and their condition ranged from well preserved to dry and weathered. The core plugs were extracted in solvent to remove water and hydrocarbons, put into a flexible jacket, and saturated with a refined oil. The unconsolidated samples of about the same dimensions were generally cored from rubber-sleeve cores that had been frozen in liquid nitrogen and for which liquid nitrogen had been used as a drilling fluid. The frozen samples were placed in a Teflon sleeve and allowed to thaw. End plates and screens were then placed on the ends of the samples. At this point the bulk volume of the sample was determined

The pore-volume compressibilities and porosities presented here were derived from 256 samples of sandstone and limestone representing 40 reservoirs. These and previously published data are in poor agreement with compressibility-porosity correlations in the literature. The salient conclusion is that to evaluate rock compressibility for a given reservoir it is necessary to measure compressibility in the laboratory.
FEBRUARY. 1973 129

from linear dimensions and the sample was placed in the test cell. A hydrostatic overburden pressure of about 50 psi was exerted on the samples before they were cleaned with solvents and resaturated with a refined oil. The change in the bulk volume at about 50 psi was recorded. At the end of the compressibility test the sample was extracted in toluene and the sandgrain volume was determined. The sand-grain volume was subtracted from the initial bulk volume to obtain a pore volume (porosity) at zero effective pressure. This porosity value at zero pressure was chosen to compare with pore-voiume compressibility because the zero pressure porosity is normally what the reservoir engineer has available from routine core analysis data.
Measuring Porosity

the following:

~=ldv. Vp
P

,..

. . .

.0

(1)

dpeff

where Co = pore volume compressibility, vol/vol/psi VP = pore volume of the sample at a given effective pressure dVp = incremental change in pore volume resulting from an incremental change in effective pressure dp,ff = incremental change in effective pressure. Eq. 1 contains the assumption that most of the porevolume change results from efiective pressure difference. This is a valid approximation for higherporosity samples. A more comprehensive discussion has been given by Geertsma.*2

With the exception of the unconsolidated sandstone porosities previously discussed, initial porosities were determined by API-approved methods, which consisted of determining the pore volume by resaturation and the bulk volume by either displacement or caliber measurement.
Applying Stress Pressure. All of our data were obtained from samples under uniform hydrostatic stress. This was accomplished by transmitting the overburden pressure to the jacketed test sample with hydraulic fluid. The tests were conducted under either constant or varying overburden pressure. Overburden Pore Pressure. The pore pressure was controlled .L--.. -I. +L and omlld he varied ..== . ,CMi..kat wa]l Lll IUU&fl Llle wll~p!e ,. . . ... --independently of the overburden pressure during tests. The tests were conducted with either constant or varying pore pressures. Effective Pressure. The ability to vary the overburden

Effects of Cycling, Time, and Temperature

Cycling

Cycling is defined as a repeated application of the stress cycle. In other words, the sample is placed in the test cell and the effective pressure is increased to some predetermined value. (This value is sometimes higher than any stress the sample will ever be subjected to during reservoir depletion.) The pressure is then released and a second, third, or even fourth cycle can be performed. Except for the case of exceedingly high-strength elastic rocks, each cycle pro-

CODE _G_SANOSTONES q LIMESTONES 0

*8

00 q o

and pore pressures independently made it necessary to express the data at a common stress condition. This was established as a function of the effective pressure, defined as the difference between the overburden (lithostatic) and pore pressures.
Determining Volume Changes

.0 q O* o

00

The change in sample pore volume as a function of effective stress was obtained by various laboratory methods.: These included (1) direct measurement of fluids expelled from the samples, and (2) inferred pore-volume changes determined by measuring dimensional changes of the samples. The values were obtained during both increasing and decreasing effective stress. A pore-volume vs effective-stress relationship was found for each sample by increasing the effective stress in about 500-psi increments to an effective stress equal to or greater than lithostatic pressure based on individual sample depths. Lithogatic pressure was assumed to be 1 psi./ft.
Calculating Pore-Volume Compressibility
Fig.

( 00 0

.400-. 15
INITIAL POROSITY &T ZERO NET PRESSURE

+-

The compressibility values shown in Fig. 1 were obtained by graphically differentiating the porevolume: effective-pressure relationships by means of
130

lPorevolume compressibility at 75 percent Iithostatic pressure vs initial sample porosity for both sandstone and limestone samples. JOURNAL OF PETROLEUM TECHNOLOGY

vides a lower compressibility value as a result of an irreversible change in the rocks internal structure. Some investigators found it necessary to cycle the samples in order to form a copper jacket around the test sample. This was done so that the pore volumes expelled during subsequent loading were not affected by the penetration of the copper jacket into the annular space between the jacket and the sample, as well as into the surface pores. Other investigators, including King, have cycled the samples until they exhibit elastic behavior; that is, until there is no additional pore-volume hysteresis between cycles. Cycling, for either reason, can result in lower compressibility values on a rock that has failed internally, and the resulting condition is certainly not the condition of the sample as it is received in the laboratory. This internal failure is easily exposed with scanning electron micrographs, with thin-sections, or in the case of unconsolidated sands, with grain-size anaiysis be-1,-l;- A n .avr-.antinn tn thic wmJh+ he #--- cI1lU ....~ tZILG1 o%-. &y lUIG b,,,,,& fi,l WAW&y UJl . .s.... .. in a compacting reservoir that had failed in situ during pressure depletion or a highly fractured reservoir that had failed as a result of tectonic forces during its hi~t~ry: The effects of cycling can be even more serious on friable or unconsolidated sands because of their inelastic behavior. We have only one example where cycling may lead to a closer approximation of in-situ compressibility, and that is in testing highly disturbed unconsolidated sands, when grains have been rearranged during coring. Rearrangement of grains from that in-situ condition generally tends to provide a looser packing, which results in a higher pore-volume compressibility value during the laboratory tests. We have demonstrated this by taking sets of adjacent samples from a carefully handled and preserved rubber-sleeve core. One set of samples was carefully handled and the other set was purposely rearranged. The results have shown that the disturbed samples had much higher compressibilities. Cycling the disturbed samples- rernnv +,, eo,-1 +h.arn PICK-IV in-ck J tn .V tha ...w ... ---- nat.kino ~-- .....a rnn ---- LuLll&u .I, w,li ,tlv. e -, V=W. dition, but significant internal failure occurred. The values of pore-volume compressibility in this study were obtained by methods that did not require cycling. In addition, the reported compressibility values were obtained during the initial application of effective pressure.
Time

percent were measured at 74F; the remainder were measured between 130 and 275 F. While we have not made a systematic study of the effect of temperature, a statistical analysis of pore-volume compressibility conducted on a suite of sandstone and limestone samples at various temperatures within the indicated temperature range showed no significant temperature effects. The results, however, were not conclusive, since the scatter of the compressibility data at any one temperature was as great as any observable temperature effects, or greater. Von Gonten and Choudhary, in discussing the effects of temperature on compressibility, show increases as high as 12 percent at 400F. We recommend, therefore, that all compressibility measurements be made at reservoir temperature.
Presentation
ihe

of Data

compressibility vaiues shown in this ~~ld~y ~~~, in !n.o~t ~a~e~, r.---_.nressure dependent. To compare samples that had been obtained from various depths. which means the samples were subjected to various effective stresses under reservoir conditions, a common effective pressure base of 75 percent of the lithostatic pressure was used. This value was seiected as the most probal.ie average efiective stress the sample would encounter during reservoir depletion. The lithostatic pressure was assumed to be 1 psi per foot of depth. The values obtained at this pressure base, plotted against the initial sample porosity, are shown on Fig. 1, along with Halls correlation. Compressibilityporosity values obtained from the literature, for both sandstones and limestones, are shown on Fig. 2. pore-voiume

Analysis of Data
Limestones

The limestone values shown on with both Halls and Van der on Fig. 3. No attempt was made Or ~?~~~ ~~rn.pi~~ by geography
Sandstones

Fig. 1 are compared Knaaps correlations to separate the limelitholoe~.

Our pore -volume compressibility values presented here were obtained from pore-volume: effective-stress relationships that had been obtained using pressure increments of about 30 minutes (30 min/500 psi). This time was generaiiy sufficient to reach a practicai stress equilibrium for most samples. We are aware that true stress equilibrium cannot be obtained in the laboratory in any practical time. However, the most -:--: c--- . ..1 ..,. -1------- .,.1.,. h., +L - G-t f=,,, slgnlll~iint ~diiine Gilttll&GS 1-G pl~d 111 LUG 111OL Iew minutes of applied stress. It is not within the scope of this report to investigate these time effects; we only point out that they exist.
Temperature

Of the compressibility values presented in Fig. 1, 81

FEBRUARY, 1973

To analyze further the porosity and pore-volume compressibility of the sandstone samples shown in Fig. 1, we used a qualitative rock-typing system. The samples were grouped as consolidated, friable, or unconsolidated: 1. Consolidated samples consisted of hard rocks (thin edges could not be broken off by hand). 2. Friable samples could be cut into cylinders, but the edges could be broken off by hand. 3. Unconsolidated samples would fall apart under --,. ...,.1 ---- SpCLldI their own weight unit?% they had iiiidergull~ treatment such as freezing. Each rock type was replotted under this crude classification system (Figs. 4 through 6). HaiYs corhk .anrlctnne data nnint~ are alcn chmw~-. -~lmt;- 99A . . . y--... -----.. 1b,,=.. . w SJ>.7 Ji-11..s>..w The results of this classification system are compared in Fig. 7 by class averaging the compressibility values from Fig. 1 and Figs. 3 through 6 in porosity increments of 5 percent. For example, the compressibilities for each rock type having between O and 5 percent porosity were averaged and piotteci
131

at 2.5 percent porosity. Halls correlation is also shown. These class-averaging curves are intended for internal comparison of the data only. They are not for correlation, because of the wide variations in the compressibility values. This wide variation can be seen on Fig. 8, which shows the range of the data points making up the class average for the consolidated sandstones.

Discussion
Fig. 1 shows that our lower-porosity limestone and sandstone samples follow the general trend obtained by Hall: the pore-volume compressibility values increase with decreasing porosity. This is more pronounced on Fig. 3, where only the consolidated limestone data are compared with Halls and Van der Knaaps correlations, and on Fig. 4, where only the consolidated sandstones are compared with Halls data. The individual compressibility curves for these consolidated samples showed substantially elastic behavior; most of the volume was recovered when the pressure was released. The samples with the higher porosity in Fig. 1 tend to be unconsolidated, and behave contrary to HalIs general trend; compressibility tends to increase with

porosity. This is evident on Fig. 6 as well. The unconsolidated samples also show significant inelastic behavior (permanent volume reduction with pressurization, resulting from internal grain failure). The friable samples in Figs. 1 and 5 also show this inelastic behavior, but there is apparently very little correlation between compressibility and initial sample porosity. Besides showing wide variations in compressibility as a function of porosity -and rock type, our results -- ..; are in poor agreement wth Halls correlation. ~ltemture values of compressibility for 79 samples (including Halls data), shown in Fig. 2, support our results and have about the same scatter. Van der Knaap obtained a good correlation for 23 limestone samples taken from a single well; but his values are also in poor agreement with Hall%. We believe the poor agreement between our data and Halls is in part because Halls are based on only 12 samples 7 limestones and 5 sandstones in the porosity range of 2 to 26 percent. Our data are based on 256 samples, 194 in the same porosity range as Halls.

Conclusions 1. The pore-volume compressibility-porosity

values obtained in this study are in poor agreement with

!000 .

I I I o 14ALL1. SANDSTONES q I+ALL1, LIMESTONES

~ D ~ ~ fi VAN OER KNAA$.

SANDS1ONES 100 I I

q VAN DER KNAA#, LIMESTONES

FA7T3, SANDSTONES DOBRVN1N5, SANDSTONES KOHLHAAS AND MILLERS. SANDSTONES VDN GONTEN ANO CHWOHARY7, SANOSW)NES (rlj

I i

I ~ ?

A VON GONTEN ANO CHOUDMARY7. LIMESTONES %

+ CARPENTER ANO SPENCERS. SANDSTONES e

VAN OSR KNAPPS CORRELATION

,,00,.o~ 0540152(1253rJ
Hg. 2Fore-volume obtained from compressibility literature vs source

.~.+
15

20

25

35

354rJ45
INITIAL poRos17Y AT ZERO NET PRE5fWRE Fig. initial as sample indicated.

INITIAL POROSITY AT ZERO NET MESSURE 3-Pore-volume compressibility for limestones.

at 75 percent
porosity

Iithostatic

pressure

vs initial sample

porosity

132

JOURNAL

OF PETROLEUM

TECHNOLOGY

.-

~o
CONSOLIDATED SAMCSTONES

A:
0 0
0

1 1

k
? : x

,0 i\

o
q

I
,..-O
Fig. 4-Pore-volume pressure Iithostatic

o 0

Ooo 0 0.O 0
I
20
compressibility I

I I
35
for

1! INITIAL POROSITV AT ZERO NET PRESSURE Fig. 5-Porewolume rwessure compressibility vs at 75 percent for

25
at 75

30
percent

INITIAL POROSITY AT ZERO NET PRESSURE Iithostatic initial sample sandstones. porosity

vs

initial

sample

porosity

consolidated

sandstones.

friable

I
o

;8
L+

O.
c 0 0 uNCONSOLIOATEO

UNCONWLIOATEO SANDSTONES 0 o

00

00

0 0 o
o

Wc)

%0

00

@e
o
TOTAL ALL SAMPLES 0 c1 lALLS ORRELATION

00
HALL%

CORRELATION

00

\
L. ~

CDNSULIOATED SANDSTONES

-\A
,.., ~
INITIAL POROSITY AT ZERO NET PRESSURE Fig. GPorwrolume pressure unconsolidated compressibility vs initial sample sandstones. at 75 percent for Fig. 7-Class averages vs initial of pore-volume sample porosity. compressibility

,.o~
o
5 10
15

20

25

30

INITIAL POROSITY AT ZERO NET PRE~RE

Iithostatic

porosity

FEBRUARY,

1973

133

to initial sample porosity. These data- suggest that correlations might be obtained for both well consolidated limestones and sandstones with similar lithologies. That is, correlations may be obtained from samples within a given reservoir, provided lithologic variations are sm@l. This would be similar to Van der Knaaps correlation for limestones from a single well. Much the same approach is recommended for the friable and unconsolidated samples, but here the pore-bolume compressibility is not merely porosity dependent; other stress parameters need to be investigated.
References
of Reservoir Rocks, 1. Hall, H. N.: Compressibility Tnzm., AIME ( 1953) 198, 309-311. 2, Van der Knaap, W.: Nonlinear Behavior of Elastic Porous Media. Trans., AIME ( 1959) 216, 179-187. 3. Fatt, I.: Pore Volume Compressibilities of Sandstone Reservoir Rocks: Trans., A-ME ( i95tl j 213, 362-364. 4. King, M. S.: Wave Velocities in Rocks as a Function of Changes in Overburden Pressure and Pore Fluid Saturation: Geophysics ( 1966) 31, No. 1. 5. Dobrynin, V. M.: Effect of Overburden Pressure on Some Properties of Sandstones: Sot. Pet. l%?. J. (Dec.,
1962) 360-366. 6. Kohlhaas, C. A. and Miller, F. G.: Rock-Compaction and Pressure-Transient Analysis with Pressure-Dependent Rock Properties, paper SPE 2563 presented at SPE 44th ;;6xd Fall Meeting, Denver, Colo., Sept. 28-Ott. 1,

\ \\

1.0 o
Fig.

I
II-Class

I
5

I 10

INITIAL POROSITYAT ZERO NET PRESSURE average porosity of pore-voiume for consolidated

\/ IT
AVERAGE FOR CONSOLIDATED SANDSTONES I .
I

RANGE OF THE CLASS AVERAGE

15

I 20

I 25

34

ccmipwssibiiitj sandstones.

initial

sample

published compressibility-porosity correlations. This is also supported by values in the literature. There is a need, therefore, for laboratory compressibility measurements in evaluating rock compressibilityy for a given reservoir. 2. Pore-volume compressibilities for a given porosity can vary widely according to rock type. 3. Attempts to correlate the data showed that consolidated sandstones differed greatly from limestones and friable and unconsolidated sands, but the data are too widely scattered for correlations to be nhle. we!! ~efin~~ trends were found only in the .-1; L IQ. consolidated sandstones and limestones and in the unconsolidated sands, whereas the compressiiiiiity of the friable samples showed little or no relationship

7. Von Gonten, W. D. and Choudhary, B. K.: The Effect of Pressure and Temperature on Pore-Volume Com9C9C FIw<...w.. . . . . -t..+ U. *t -CPF_ AA~~ .&& pressibiiity, Paper @nc ~r~ /-.- nual Fall Meeting, Denver, Colo., Sept. 28-Ott. 1, 1969. 8. Carpenter: C. B. and Spencer, G. B.: Measurements of Compressibility of Consolidated Oil-Bearing Sandstones, RI 3540, USBM (Oct., 1940). 9. Jennings, H. Y.: How to Handle and Process Soft and Unconsolidated Cores, World Oil (June, 1965 ) 116-119. 10. API Recommended Practice for Core-A nalysis Procedure, API RP40, 1st cd., New York (Aug., 1960). 11. Mann, R. L. and Fatt, I.: Effect of Pore Fluids on the Elastic Properties of Sandstone Geophysics ( 1960) 25, 433-444. 12. Geertsma, J.: The Effect of Fluid Pressure Decline on Volume Changes of Porous Rocksfl Trans., AIME (1957) 2?1 720 -la .m~ ALU, JJ, -J2Z.
Original manuscript received in Society of Petroleum Engineers ... -. .,, ----...-+. -.-m;.rl.-W . F.@ omce uec. 44 19JJ. --..: IWVISW ,,l-{lb=u, ,w, ,--. . . . ~~, ~Q7~, Paper (SPE 3S35) was presented at SPE Rocky Mountain Regional Wzeti rig, he!i! in 2!enver, (%!0., April 10-12, 1972, Q Copyright 1973 American Institute of Mining, Metallurgical, and Petroleum Engineers, Inc.

134

JOURNAL

OF PETROLEUM

TECHNOLOGY