Académique Documents
Professionnel Documents
Culture Documents
5
4
3
Intro
C. E. Neuzil
Start
U.S. GeologicalSurvey,Reston,Virginia
Search
poorlyunderstood,
is commonly
a criticalparameter
in analyses
of subsurface
flow.
Datanow available suggesta regularrelation betweenpermeabilityand porosityin
claysandshalesandpermeabilities
that, evenat largescales,are significantly
lower
thanusually
assumed.
Permeabilities
between10-23and10-17m2 havebeenobtained
at porosities
between0.1 and 0.4 in both laboratoryand regionalstudies.Althoughit is
clearthat transmissivefractures or other heterogeneitiescontrol the large-scale
hydraulic
behaviorof certainargillaceous
units,the permeabilityof manyothersis
apparently
scaleindependent.
Theseresultshavesignificant
implications
for
understandingfluid transport rates and abnormal pressure generation in basins, and
permeability
of argillaceousunitsis thereforea parameterof
considerableimportance for analyzing groundwater flow in
sedimentaryenvironments, with particular importance attachingto the relation between permeability and porosity
because of the sediment compaction that accompanies
burial. Unfortunately, it is usually infeasible to measure
permeabilityor its variation in argillaceousunits and few
reliableguidelinesexist for estimatingit. Specifically, there
arefew indicationsof (1) how permeability relates to porosity in thesemedia and (2) how small the permeability of these
sediments can be at various scales. Researchers have used
Until the 1970s, little was known about clay and shale
hydraulic properties beyond permeability data obtained as a
by-product of consolidationtesting. Although attention has
been focused on the problem since then, the difficulty of
measuring small permeabilities has continued to limit the
acquisitionof data. As a result, extensivepermeabilitydata
for these materials do not exist. In addition, a large fraction
of the laboratory data that are available are not suited for
hydrogeologicapplications.Many studiesusedpurifiedclays
or reconstituted sediments, leaving in question the applicability of the resultsto naturalsystems.Othersfailed to report
porosity(or void ratio) or gaveno indicationof how the data
were obtained. Only a handfullof laboratory determinations
(1) usednatural media in an undisturbedstate, (2) monitored
appropriatemethodologyand carefultechnique.Laboratory
data I have found that meet these criteria, including data
andshalepermeability
in flow simulations.
The relatively
low permeabilities
thesedata imply at typical porosities
indicatethat fluid fluxes in someflow systemscould be
significantly
smallerthan analysesindicate;elucidationof
thesefluxesis crucialfor understanding
phenomena
suchas
abnormal
pressure
generation,
petroleumandore emplace-
Thispaper
isnotsubject
to U.S. copyright.
Published
in 1994bythe niques(region6) or with particularcaredevotedto the test
American
Geophysical
Union.
equipment
andprocedures
(regions2, 4, 7, 9, and 10).
Papernumber93WR02930.
Figure 1 suggeststhat a log-linearrelation betweenper145
146
NEUZIL'
Loghydraulic
conductivity,
m.s'l
-16
-14
-12
................
-10
I
-8
I
5
4
3
0o8-
Intro
Start
Search
0.0
I
I,
r'
I
-20
.11
,
I
Effect of Scale
I
-18
....I
,,
-16
permeability
of 10-16 m2 [Bredehoeft
et al., 1983].Scale
dependencealso has been detectedin lacustrineclay [Rudolpheta!., 1991]and clay till [Keller et al., 1988]andis
impliedin numerousinstanceswhen in situ permeability
not unexpected
for clay-richmedia.Severaltheoreticalbackgroundinformation.
The datain Figure2 wereobtainedfrom inverseanalyses
expressions
relatepermeability
to porosityand poresize;
of
a varietyof flow systems.
Inverseanalysisyieldsestione derived by Kozeny [1927] and modifiedby Carman
properties,
generally,
by numerically
simu[1937],whichusesspecific
surfaceareaasa measure
ofpore matesof system
size,appearsto provideusefulinsightin thepresentcontext.
Althoughthe Kozeny-Carman
equationis basedon rather
restrictiveassumptions
[Scheidegger,
1974],it predictspermeabilityof kaoliniteclay cakeswith reasonableaccuracy
[Olsen,1962].The predictedrelationis nearlylog linear,
with a slopesimilarto that suggested
by Figure1.
The broadrangein permeabilityat a givenporosityin
Figure1 (evenif regions11 and 12 are discounted)
could
havemultiplecauses.First, very smallpermeabilities
are
measuredat or near the limits of instrumentalresolution,
which can introduce error. Second, effects of anisotropy
latingflowandadjusting
thevalueof uncertainparameters
to
obtaincomputed
hydraulic
conditions
thatbestmatchthose
observed.inverse methodswork reliably for well-posed
problems
andgenerally
aretheonlywayto evaluate
small
permeabilitiesat large scales.
Certaininverseanalysesyieldedonly maximumvalues;
theseareindicated
by arrowsin Figure2. In thecase
of
regions
6 and7, thisisduesimply
to ambiguity
inherent
in
the analyses;
a rangeof permeabilities
lowerthanthose
indicated
giveequally
goodmatches
to observed
conditions.
Theuncertainty
in region5 existsbecause
inverse
analyses
sometimes
incorporate
significant
changes
in permeability
overgeologic
time.Region5 is derivedfrom an inverse
analysis
ofgeopressures
insediments
oftheU.S. GulfCoast
byBethke
[1986a].
These
sediments
mayhaveexpelled
excess
fluid
through
natural
hydrofractures
during
parts
of
clustering
of clay particles.Experiments
with claycakes
their
history
[Engelder,
1993,
p.
41;
Capuano,
1993].
If
[Mitchellet al., 1965]shownearlythreeordersofmagnitude
microstructure,
whichOlsen[1962]considered
to arisefrom
indeed
varied
intimewiththepresence
and
variationin permeability
because
of microstructural
differ- permeability
a
ences.Comparable
variationisexpectable
in naturebecause absenceof transmissivefractures,region 5 represents
average
of thetwoconditions.
Onlythe
of thephysically
andchemically
diverse
environments
in time-integrated
whichargillaceous
sediments
aredeposited.
No data are known to me that indicate clay or shale
unfractured
permeability,
which
must
besmaller,
isdirectly
comparable
withthelaboratory
measurements
inFigure
1.
147
Table1. Background
Information
forLaboratory
Permeability
DataPlotted
in Figure1
Region
Orienta-
in
Figure1
1
Formation
(Location)
bottomdeposit
(North Pacific)
Lithology
(Mineralogy)
Typeof Test*
(Permeant)
bottommud(illite, mechanical
smectite)
Intro
Number or Effective
tionto
Measure- Stress,
Bedding ments
MPa
normal(?) 48
0.04--0.4
transient,
Source
Silva et al.
[1981]
steady flow,
and quasisteady flow
(seawater)
bottom deposit
(North Pacific)
bottomdeposit
(North Pacific)
Gulf of Mexico
unconsolidated
sediment;
varying
proportions of
clay, silt, and
sand
bottom deposit
bottom mud (illite,
(North Pacific)
chlorite)
Pleistoceneto
marine and
Recent (Quebec, lacustrineclay
Mississippi
Delta, Sweden)
Sutherland Group
(Saskatchewan)
glacial till
(montmofillonite,
illite, kaolinite)
quasi-steady
flow (seawater)
steadyflow
normal
22
(seawater)
steadyflow and normal (9.) approxiquasi-steady
mately
600
flow (natural
pore water and
distilled water)
mechanical
various
approxitransient
mately
Morin
and Silva
[1984]
Silva
et al.
[19811
Morin
and Silva
[1984]
0.04-0.3
Tavenas
et aI.
[1983]
Bryant et al.
[1975]
250
(seawater)
mechanical
transient
normal
27
0.06-2.0
and
Keller
et al.
[1989]
quasi-steady
flow (natural
Pierre Shale
(central South
Dakota)
claystone (mixed
layer,
montmorillonite,
illite)
mechanical
transient,
hydraulic
transient,
steady flow
(pore water
normal
85
0.1-50
C. E. Neuzit
(unpublished
data, 1987)
and
parallel
duplicate and
distilled water)
9t
10
1!
12
5
4
3
25
0-40
Young et al.
[1964]
23
33
Magara[1978]
Neglia [1979]
Lower Pliocene
(Italy)
*Mechanicaltransienttestsincludestandardconsolidationtestsand similarproceduresinvolvingtransientdeformation
under
mechanical
loads.Hydraulictransient
tests,whichincludepulseandinjection
tests,involvevaryingthehydraulic
boundary
conditions
ofthesample
andanalyzing
thetransient
pressure
response.
Quasi-steady
flowtestsarealsoknownas
falling
headtests.Steadyflowtestsare"standard"
permeability
tests.Forfurtherdiscussion
of low-permeability
testing
methodologies,
seeNeuzil[1986].
t?orosity
estimated
fromdescription
of samples.
Start
Search
148
-14
I
-2
'
'
-10
1 .......... !
-8
i
0.8
.....
iii;;
0.6
0.4
5
4
3
analysis[Bredehoeft
et al., 1983]yieldeda relativelylarge
0.2
permeability
of 10-16m2 (!0-9 ms-l);thisresultapplies
to
an area with dimensions of hundreds of kilometers. A more
0
-18
-16
recent
study
ofafewkm2portion
ofthebasin[NeuziI,
1993]
-14
Logpermeability,
m2
Table 2.
Region
in
Figure 2
1
2
Implications
The increasing speed of digital computers has stimulated
efforts to analyze subsurface fluid flow using numerical
simulation. For example, significant effort has been devoted
to analyzing paleoflow in sedimentary basins becauseof its
relevance to economic minerals and waste disposal.Re.
cently, we have seen studies that have simulated paleof10w
Lithology
(Mineralogy)
Type of
Analysis
(Barbados
clay, calcareous
transient flow
Accretionary
mudstone
Ridge complex)
Sutherland Group glacial till
transient flow
(Saskatchewan)
(montmorillonite,
Orientation
to Bedding
various
Vertical and
Horizontal
Dimensions
1 and 15 km
Source
Screaton et al.
[ 1990]
Keller et aI.
[ 1989]
kaolinite)
Pierre Shale
(Central South
Dakota)
4*
Colorado Group
and Upper
Manville
claystone (mixed
transientflow
normal
Neuzil [1993]
layer,
montmorillonite,
illite)
claystone, shale
transient flow
normal
Corbet
and
Bethke [1992]
Shales
(Alberta)
5
Gulf of Mexico
6*
Pierre, Cafiile,
clay, shale
transient flow
normal
claystone and shale steady state flow normal
!0 and >300 km
3 and 800 km
Belitz and
Bredehoeft
[1988]
Graneros
Shales (Denver
Basin)
(Siberia)
Bethke [1986a]
"argillaceous
transient flow
normal (?)
rock"
Nesterov and
Ushatinskii as
reportedby
Brace [1980]
*Porosity
range
estimated
fromdepth
andthickness
ofsediments
(T.F.Corbet,
Jr.,Sandia
National
Laboratory,
personal
communication, 1993, and K. Belitz, Dartmouth College, personalcommunication, 1993).
149
regimes
drivenby compaction
[Bethke,1985;Harrisonand Figures 1 and 2, with the latter based on flow at moderate to
5
4
3
Intro
[Senger
etal., 1987].Suchanalyses
areinherently
uncertain
Start
because
the systemhydraulicpropertiesin the geologicpast Conclusions
mustbe estimated.Permeabilitiesor rangesof permeability
Much remainsto be learned about the hydraulic properties
assumed
for argillaceousunits, as in the studiescited above, of clays and shales.For example, we are unable to predict
arecommonly
between10-9 to 10-16m2 (hydraulic
con- how and where heterogeneity(due to depositionalarchitecductivity
of 10-12to 10-9 ms-1) forporosities
lessthan0.4. ture or fracturing)will affectlarge-scalepermeabilityin these
Figures1 and2 suggest
that argillaceous
formations
canbe media. For the present,when explicit data are not available,
much
lesspermeable,
withvaluesin therangeof 10-23to flow analysesmust considera range of permeabilityvalues
quently
incorporate
relativelypermeable
subunits
andtheir Figures 1 and 2 should be considered possible even at
permeability
canbe correspondingly
high.Nevertheless,
as regional scales. Values lower than shown, however, probaFigure2 shows,they apparentlycanbe dominated
by their bly need not be considered.Becauselaboratorypermeabilcanvary by a factorof 103,
low-permeability
components.
This probablyis mostoften itiesat similarporosities
true for flow normal to stratification, which is controlled by
theleastpermeablehorizons.
Adoptionof the relatively low permeabilitiesindicatedby
Figures1 and2 wouldresultin reinterpretation
of someflow
byleakagethrougha shale,simulationsusingloweredshale
permeabilities
at basinscale(hundreds
of kilometers)would
indicate reduced rates of fluid transport throughout the
Acknowledgments.
I wishto thankGrantGarven,Ward Sanford,
Dave Rudolph,Craig Bethke, and two anonymousreviewersfor
their helpful commentson this paper.
References
Explanationof subnormal
fluidpressure,Am. Assoc.Peri. Geol.
Bull., 72(11), 1334-1359, 1988.
Bethke, C. M., A numericalmodel of compaction-drivengroundwater flow and heat transfer and its application to the paleohydrologyof intracratonicsedimentarybasins,J. Geophys.Res.,
90(B8), 6817-6828, 1985.
readilythanpresentlysupposed.
Thisis consistent
with the Brace,W. F., Permeabilityof crystallineandargillaceous
rocks,Int.
J. Rock Mech. Mitt. Sci., 17(5), 241-245, 1980.
widespread
occurrenceof features suchas thrust faulting
andmultiplegenerationfracturing,which are thoughtto Bredehoeft,J. D., C. E. Neuzil, andP. C. D. Milly, Regionalflow
in the Dakota Aquifer: A studyof the role of confininglayers,
requirehighfluid pressures[Engelder, 1990].
U.S. Geol. Surv. Water SupplyPap., 2237, 45 pp., 1983.
The permeabilitydata presentedhere also have implica- Bryant,W. R., W. Hottman,and P. Trabant,Permeability
of
tionsfor subsurfacewaste isolation, In a broad sense, the
R. M., Evidence
offluidflowin microfractures
ingeopressaryfor assessing
the risk of advectingtoxic wastefrom a Capuano,
suredshales,Am. Asso.Petl. Geol.Bull., 77(8),1303-1304,1993.
repository
to the biosphere.Figures1and2, however,have Carman,P. C., Fluidflowthroughgranularbeds,Trans.Inst. Chem.
groundwater
flowin the westernCanadasedimentary
basin,J.
Geophys.Res., 97(B5),7203-7217,1992.
T., Smoluchowski's
dilemma
revisited:
A noteon the
laboratory
andinversepermeability
estimates
evidentin Engelder,
Search
5
4
3
Intro
!50
NEUZIL:
HOW PERMEABLE
fluidpressure
history
of thecentral
Appalachian
fold-thrust
belt, temperature
on somephysical
properties
of oceansediments,
j.
Geophys.Res., 89(B1), 511-526, 1984.
in TheRoleof Fluidsin CrustalProcesses,
editedby J. D.
Bredehoeft
andD. L. Norton,pp. 140-147,
National
AcademyNeglia,S., Migrationof fluidsin sedimentary
basins,Am.Assoc.
Press,Washington,D.C., 1990.
Start
Search
Engelder,
T., Stress
Regimes
intheLithosphere,
457pp.,PrincetonNeuzil,C. E.,Groundwater
flowinlow-permeability
environments,
UniversityPress,Princeton,N.J., 1993.
Garven,G., Theroleofregional
fluidflowin thegenesis
ofthePine 2020, 1993.
Pointdeposit,WesternCanadasedimentary
basin,Econ.Geol., Olsen,
H. W., Hydraulic
flowthrough
saturated
clays,Clays
Clay
80(2), 307-324, 1985.
Garven,
G., A hydrogeologic
model
fortheformation
ofthegiantoil Rudolph,D. L., J. A. Cherry,andR. N. Farvolden,Groundwater
sandsdeposits
of the WesternCanadasedimentary
basin,Am. J.
Sci., 289, 105-166, 1989.
flowandsolutetransportin fracturedlacustrineclaynearMexico
City, Water Resour. Res., 27(9), 2187-2201, 1991.
A. E., ThePhysics
of Flow Through
PorousMedia,
Ge, S., andG. Garven,Hydromechanical
modeling
of tectonicallyScheidegger,
3rd ed., 353pp., Universityof TorontoPress,Toronto,Ont.,
drivengroundwater
flowwithapplication
to theArkomaForeland
Basin,J. Geophys.Res., 97(B6),9119-9144,1992.
1974.
1968.
fluidpressures,
andflow ratesin the Barbadosridgecomplex,
J.
Senger,
R. K., andG. E. Fogg,Regional
underpressuring
in deep
brine
aquifers,
Palo
Duro
Basin,
Texas,
!,
Effects
of
hydrostratig.
Keller,C. K., G. vanderKamp,andJ. A. Cherry,Hydrogeology
of
raphy and topography,Water Resour.Res., 23(8), 1481-1493,
Mexico Basin,Am. Jr. Sci., 291, 109-176, 1991.
two Saskatchewan
tills, I, Fractures,bulk permeability,and
spatialvariabilityof downwardflow,J. Hydrot.,101(1-4),97-121,
1988.
1987.
effectof Cenozoicbasindevelopment,
WaterResour.Res.,25(8),
1494-1504, 1987.