Sie sind auf Seite 1von 20

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 105, NO.

B7, PAGES 16,749-16,768, JULY 10, 2000

Thermal conductivity of unconsolidatedsediments


with geophysicalapplications
A. Revil

CentreNationalde la RechercheScientifique,CentreEuropende Rechercheet d'Educationen Gosciences


de l'Environnement,Departmentof Geophysics,University of Aix-Marseille III, Aix-en-Provence,France

Abstract. A theoreticalmodel for the predictionof the thermalconductivityof unconsolidated


granularsediments,with applicationto sandshalemixtures,is presented.The sedimentis

modeled
asanassemblage
of grainsof thermalconductivity
s immersed
in a porefluidof

thermal
conductivity
p In order
totakeintoaccount
forthethermal
interactions
between
the
grainsa differentialeffectivemediumapproachis usedto providea relationshipbetweenthe
effectivethermalconductivityof the isotropicmixturesof grainssaturatedby the porefluid, the
porosity,andthe thermalconductivities
of the grainsandporefluid. The influenceof the topology
of the interconnected
porespaceis accountedfor throughthe useof the electricalcementation
exponentm, whichis relatedto the electricalformationfactorF by F = o-m,where is the
interconnected
porosity.The mainassumption
of themodellies in thesmallcontiguitybetween
the grains.This assumption
holdswell for unconsolidated
sandshalemixturesedimentsas
demonstrated
by comparingthe modelto availableexperimentaldata. The modeloffersthe
possibilityto derivethermalconductivityprofilesfrom downholes
measurements
of natural
radioactivity,electricalresistivity,andbulk density.

1. Introduction

purpose,a thermal conductivity model of unconsolidated


sediments is needed.

Knowledgeof the thermalconductivityis needed,for example,


to determineheat flow in sedimentarybasins [e.g., Sasset al.,
1981; Detrick et al., 1986; Yamanoand Uyeda, 1990], to get
informationon pastclimate [e.g., Murphy and Lawton, 1977], to
model heat and fluid transport and hydrocarbon formation
processes[e.g., Ortoleva, 1994], to obtain boundson mechanical
propertiesof sediments[e.g., Gibianskyand Torquato,1998], and
to evaluate possible radioactive waste disposalsites in shales.
The possibilityof gas hydratedestabilizationby global warming
due to temperatureincreaseat the seafloor and inside the first
hundredmetersof marinesedimentssuggeststhat knowledgeof
thermalconductivitymay be importantfor this purposeas well.
The thermalconductivityof a planetarysurfacecan alsoprovide
importantcluesaboutthe natureof the surfacematerials[Presley

and Christensen, 1997a, b]. Thermal conductivity of


unconsolidatedsediments is usually determined through
laboratoryexperiments
usingcoresamplesand/ordrill cuttings.
However,situationscanarisewheresamplesare not available. In
sucha case,in situmeasurements
of the thermalconductivityare
required.Thermal conductivitymeasurementsare generally
difficultto obtainin the field andparticularlyin boreholesdueto
problemsrelatedto contactresistance,convectionof the borehole
fluid, nonradial heatflow, and time lagsin the measuredthermal
responses[see Silliman and Neuzil, 1990]. However, thermal
conductivity can be inferred indirectly from downhole
measurements
and relationships
betweenthe measuredproperties
(e.g., acoustic velocities, electrical resistivity, natural
radioactivity) and the thermal conductivity. To reach this

Copyright2000by theAmericanGeophysical
Union.
Papernumber2000JB900043.
0148- 0227/00/2000JB 900043509.00

Most of the existingmodelsare unsatisfactory


to determinethe
thermal conductivityof granularmaterialsas explainedbelow in
section2. The purposeof this paper is thereforeto develop a
model to determine the effective thermal conductivity of
uncementedand unconsolidatedgranular porousmaterials. The
model should be both theoretically justified and kept simple
enoughfor geophysicalapplications.The influenceof the pore
spacetopologyis modeledwith oneparameteronly, the electrical
cementationexponent,which is obtainedeasily from electrical
resistivity measurements [e.g., Lovell, 1984]. The model
proposed
hereis basedon a differentialeffectivemedium(DEM)
approach[e.g., Sheng,1991], which appearsefficient to account
for the (thermal) interactions between the grains, and is
conceptuallybasedon the similarity betweenthe boundaryvalue
problemsfor electrical conductivityand thermal conductivityin
porousmedia. The link between these two propertiesis very
useful to computethe thermal conductivityof granularsediments
from their electricalconductiveresponse[e.g., Lovell, 1984]. The
model developed in this paper is also consistentwith those
developed by Sen et al. [1981] for the dielectric permittivity;
Revil and Glover [1998] and Revil et al [1998a, b] for the
electricalconductivity;Revil [1999a] for the membranepotential,
the thermoelectric effect, and the diffusivity; and Revil and
Pezard [1998] and Revil et al. [1999a, b] for the electrokinetic
couplingcoefficient. In section4, the thermal conductivitymodel
developedin section3 is appliedto the determinationof thermal
conductivityprofiles from conventionaldownholemeasurements
including the gamma ray count, formation bulk density, and
electricalresistivity. This model can be usedin a similar way to
derive on a routine basis thermal conductivity profiles in
sedimentarybasinsor along continentalmarginsfrom the large
amount of log data acquired by the oil industry and the Ocean
Drilling Program(ODP) over the past30 years.

16,749

16,750

2. Review

REVIL:

of Previous

THERMAL

CONDUCTIVITY

Models

In this section I briefly discussprevious modeling attempts


concerningthermal conductivity of granular porousmaterials.
Models based on classicalmixing laws (arithmetic, geometric,
and/or harmonic) poorly constrainthe thermal conductivity of
granularmaterials[e.g., Somerton, 1992, p. 83], as they do not
account for the texture of the sediment. For example, the
geometric and arithmetic mixing laws tend to overestimate
laboratorymeasurementsand the harmonicmixing law tendsto
underestimatethe measuredvalues of clay-tree sands[Woodside
and Messher, 1961; Somerton, 1992]. Another model often used
to characterizethe thermal conductivityof porousrocksis that of

Maxwell [e.g., Maxwell, 1892; Beck, 1976; Zimmerman, 1989].


This model was one of the first used to determine

the electrical

conductivityof porousmaterialsand it is basedon the assumption


that spheroidalpores,dispersedin a mineralmatrix, representthe
pore space(the oppositecase of grains dispersedin a fluid is
discussedin section 3). Other theoretical, empirical, or semi
empirical models have been developed assuminga connected
mineralphaseand/orisolatedpores(see reviewsby Beck [1976]
and Somerton[1992]). However, none of these models reflect the
pore spacegeometry of granular porousmaterials such as sand
shale mixtures. More accurate is the use of the Wyllie and
Southwickelectrical model [see Spiegler et al., 1956] appliedto
the thermal conductivity problem of unconsolidatedgranular
porousmaterials[Woodsideand Messher, 1961]. This model
assumesthat the conductivity response of an aggregate of
conductivegrainssaturatedwith a conductivefluid is compatible
with an equivalentthree-resistormodel. Two resistorsare linked
in series,one representinga heat flow paththroughthe grainsand
a secondthroughthe intergranularpores.A third resistorparallel
to the first two representsa path entirely throughthe pore space.
The main drawbackof this model is the use of threeindependent
parameters (in addition to the porosity) to characterize the
influenceof the pore spacetopologyupon the conductivityof the
mixture. This drawbackmakesthe Wyllie and Southwickmodel
unsuitable for practical applications such as downhole
measurementinterpretationas too many unknown parametersare
involved in the equations. Another set of models describes
thermal conduction in a cubic array of equal-sized spheres
[Gomaa, 1973; Ozbek, 1976; Ghaffari, 1980]. The drawbacksof
thesemodelsis to accountpoorly for the influenceof the porosity
and thermalinteractionsbetweenthe grains. Some parameters

OF UNCONSOLIDATED

SEDIMENTS

The volume averagingapproach(see Whitaker[1983, 1986],


Plumb and Whitaker [ 1990], and Quintard and Whitaker [ 1991]
for the thermal conductivityproblem) has been quite popularin
the last 20 yearsto determineformulasfor transportpropertiesin
porous media. This method has the advantage to used less
restrictiveassumptionsabout the topologyof the pore spacethan
the methodsdiscussedpreviously,and it is thereforeapplicable,in
principle,to a wide classof porousmedia. The disadvantageof
the volume averaging method is that some of the parameters
introduced in the equations are sometimes difficult to reach
experimentally.For example,as discussedby Bernabdand Revil
[1995], it is not always easy to determine the length scale A
(appearing in the electrical conductivity problem using the
volume averaging approach) from electrical conductivity
measurements.

3. Formulation

The thermal conductivityof unconsolidatedgranularporous


mediais modeledin section3.1 as a functionof the porefluid and
grain thermal conductivitiesand a grain shapeparametercalled
the electricalcementationexponent.The solutionof this problem
is usedto get a mixingformulafor clay-treesandsandpureshales
in section 3.2 and tbr sand shale mixtures in section 3.3.
3.1.

Unconsolidated

Granular

Porous Media

1 considerbelow a representative
elementaryvolume(REV) of
a statisticallyhomogeneous,
isotropic,granularporousmedium.
The grain-to-grain
"contiguity"is assumed
to be small(Figure1,
contiguityis definedby the portionof the grainsurfacebeingin
contactwith the neighboringgrains,and it takesa value between

0 and1 [Takei,1998]). The REV is assumed


to be an averaging
disk of lengthL andcross-sectional
areaA, andz is the axis in the

directionnormalto the end faces(Figure 2). The averaging


volume V = L A is muchlargerthanthe scalesof microscopic
structureand is much smaller than the scale of macroscopic
phenomena.The solid particlesformingthe matrix are assumed
to be spheroidal(the graindiameterof a particleis assumedto be
equalto thediameterof a sphereof samevolumeasthe particle).
The REV is submitted to a uniform thermal gradient

(VT) = (AT/L), where istheunitvector


in thez direction
and

anglebracketsdenotea statisticalaveragingoperatoroperating
overthe porousmedium. In this"imposedfield assumption"
the
introduced in these models are either difficult to determine
macroscopic
temperature
gradient
is
determined
by
temperature
experimentallyor usedas adjustableparameters(seediscussion
of
outsidetheREV. Whensucha differenceof temperature
Somerton[ 1992,pp. 89-102]). For furtherreferences
concerning sources
conductionthrough arrays of sphericalparticles, see Jeffrey
[1973],Batchelorand O'Brien[1977],Acrivosand Chang[1986],

is maintainedbetween the end facesof the REV, there is a flow of

and Grolier et al. [ 1991].


Empirical models are available for unconsolidatedsandsbut

are generallylimited to a given set of samples[e.g., Krupiczka,


1967;Keese,1973;Sornerton,
1992].An interesting
approach
has
alsobeensuggested
by Luo et al. [1994]. This approachis based
on fabrictheory,andit relieson an empiricalparametercalledthe
fractionof serieselements(FSE), whichcorresponds
the way the
mineralsagregateto form the mineral framework. The so-called

"fabrictheory"describedby Luo et al. [1994] represents


a mean
betweenthe so-calledWiener bounds(actuallythe arithmeticand
geometricmeans). However, the approachusedby Luo et al.
[1994] ignoresthe influenceof the shapeof the grains. We will
showbelow that it is very importantto considerthe influenceof

the grain shapeupon the effective thermal conductivityof


granularporousmaterialsandthatcanbe doneusinga differential Figure 1.
effective medium scheme.

Uncementedand uncompactedgranular porous


aggregateswith small contiguitybetweenthe grains.

REVIL: THERMAL

CONDUCTIVITY

OF UNCONSOLIDATED

heat The steady state boundary value problem governing this


flow is expressedas

SEDIMENTS

16,751

sandstones, as shown later, due to the existence of a thin water

film at the grain-to-graincontacts(Figure 2) [e.g., Revil, 1999a,


b]. The contributions from thermal radiations inside the REV are

q = -A(r)VT,

(1)

V.q =0,

(2)

z=0

r(z)=T+AT,z=L,

(3)

ignoredin this paper (an extensivediscussionabout the radiative


componentis given by Presleyand Christensen[1997a]).
The volume averageof the local thermal flux densityis given
by
1

I iqdV+ZIq dV,

Q='v,, 'kv

wherethe local thermalconductivity


is A(r)=Aj- in the
interconnected
porespaceand A(r)= As in the grains,and q is

(5)

-A.!
iVTdV-As
Q=v,,
--E
j'Vr
dV, (6)

the local heat flux density. Equation (1) is the local heat
conduction equation (Fourier's law), equation (2) is the heat
conservationequationin steadystateconditions,and equation(3)
is an external boundary condition for the temperature. The
temperatureand the heat tlux are continuousat the grain/pore
fluid interface. An extensive discussion of the macroscopic
Vpistheporevolume,
V isthetotalvolume,
andVk isthe
boundary conditions for the problem of heat transfer by where
volume of the grain "k." The summation is done over all the
conduction can be found in Prat [1990]. The macroscopic
grainspresentin the REV. It is assumedthat all the grainshave
volume-averagedsolutionof suchset of equationsis given by
the samethermalconductivity(this assumptionwill be relaxedin

q=-A.f(VT)(2s
-vA.i'
)Z lvr av,

(4) section3.3). If this not the case,a weightedarithmeticaverage

Q--( q )=-A( VT ),

can be used to compute a crude estimate of the thermal

whereA is the effective thermalconductivityof the REV. The


objective of this section is to find a relationship between the
effective thermal conductivity, the thermal conductivitiesof the
pore fluid and the grains, and the microstructureof the REV,
assuming an unconsolidatedgranular porous materials with a
small contiguity between the grains (Figure 1). The model
developedbelow shouldbe also valid for compactedsandsand

(,)

conductivity
of thegrains.From (7), themacroscopic
thermalflux
densityis given by

o=-%<vr>+.<s>,

(8)

wheren is the numberdensity of grains in the REV, S is the

thermaldipole strengthof the grain, and the angle brackets


enclosing
S denotesherean averageoverall thegrainspresentin
theREV [e.g.,Batchelorand O'Brien,1977]. For example,let us
considerthe grainsto be sphericalparticlesof radiusR. In sucha

f'x

case,n--(l-O)l(4R3/3), wherepis the porosity(ratioof

,-,xa

I o./f

interconnected
porevolumeto total volume). By analogywith
the dielectricproblemthe thermaldipolestrengthof a grain k is
definedfrom (7) and(8) by (seeAppendixA)

o ',,2 c

S---(As
- Af) J'VTdV,

(9)

v,

Pore fluid

As + 2Aj.

(VT).

(]0)

Solid

particles

From the previoussolutionfor a singlesphereand usingthe selfconsistent


approximation,it is possibleto determinea relationship
betweenthe effective thermalconductivityA of a small number

density
of spherical
grains
of thermal
conductivity
As immersed
in

a fluidofconductivity
Aj, Thisrelationship
takes
thefollowing

d
/

?'"

form of the Maxwell-Clausius-Mosotti equation in dielectric


theory[e.g., Stroudet al., 1986]:

..4

Protrudin-
OH'/Si-OH
! [*
'"'
........
"''"'" "'--'-'"'"---'"'"'"'"':l
.....
x. L

- --

Figure 2. (a) Schematic representation of the representative


elementary volume (REV) used in this paper. (b) A porous
aggregate of spheroidal particles with a certain contiguity
betweenthe grains. (c) The grain-to-grain contacts of a quartz
sands maintained open due to the existence of strong steric
repulsive forcesrelated to the existence of polymeric brushes
of silicic and silanol hairs. (d) Forcesmaintainingthe grains at

anequilibrium
distance
to--20A.

A-A.f(As-A.f
)

Z+22./
=(l-)As
+2Aj'

(11)

Equation(11) is not very efficient tbr computingthe thermal


conductivityof a fluid-saturatedgranularporousmaterialbecaus.e
the thermalinteractionsbetweenthe grainsare taken into accounit
only to the first order [Batchelorand O'Brien, 1977; Sen et al.,
1981]. In other words, (11) is valid only for a dilute suspensions
of sphericalgrains.
A better way to take into account the thermal interactions
between the grains is to use a differential effective medium

16,752

REVIL:

THERMAL

CONDUCTIVITY

OF UNCONSOLIDATED

SEDIMENTS

The volumefractionof the grainsis givenby


g

(18)

vf + I/

Consequently,
(17) and(18) leadto
Figure

3.

Effective thermal conductivity of a grain of

d2,=
dqoa,
(2,s-2,)
32, 1-q0g
2,s+22`'

thermalconductivity
2`scoateduniformlywith a fluid of

thermal
conductivity
2,./,

(19)

which can be integratedas

(DEM) scheme. Following Sen et al. [1981 ] for the dielectric


problem, solid grains are imbedded in a self-similar way in the
(20)
pore fluid. First, a grain is addedto the fluid and the influenceof
A
0
this grain upon the temperature field is analyzed (as done in
Appendix A). Then this grain fluid mixture is used to coat a
where2,() is the effective thermalconductivityat porosity
secondgrain and to computethe influence of this mixture upon
This yields
the temperaturefield, and so on following an iterative process
until the desired porosity is reached. Note that with such a
processthe contiguitybetweenthe grainsis assumedto be equal
to zero at each iteration step. However, as shown later in this
section,the model works very well for consolidatedsedimentsas
For nonspherical
grains,(21) can be generalizedby analogyto the
long asthe mineralphaseremainsunconnected
(Figure 1). Let us
dielectricproblemto give

x 2,s
+22, 2,= dog

f 32,(2,s
-2`)

1-Og

2,f--2,S

=05'

2,./'2,s

=O,

(21

consider
first a spherical
graink of thermalconductivity
2,S

uniformly
coated
witha fluidofthermal
conductivity
2,./.
(Figure
3), whererk is theradiusof thecoatedgrainandRk is theradius
of thegrainitself(innersphere),
rk > Rk. The effectivethermal
conductivity of this mixture is given by analogy with the
dielectricproblemby [Van de Hulst, 1975,p. 74]

2,k
=l1-a
+2ak
/r3
/ r
3

'

(2,./,.1)(2,
S+ 22,/,.
)+ rhc
(22`f + 1)(2,
S- 2,.1-)
,

whereD is a "thermaldepolarizationfactor." The depolarization


factor dependsstronglyon the shapeof the solid grains(D = 1/3
for sphericalgrains[Sen et al., 1981]). The influenceof grain
shapeon D for the electrical conductivity problem has been
examinedby Mendelsonand Cohen[1982|, andthe sameanalysis
can be donefor the thermalconductivityproblem.Equation(22)
can be rewritten by

(12)
(13)

ak--r3(2,f
+2)(2,s
+2/t.)
+2r/
(,,f1)(2`
s-2,),.)

whereO
k is the thermalpolarizabilityof the coatedgrainand

11k--(Rk/
rk)3is the fractionof the grainsin volume.

2,= 2,f 0!-m

Consequently, from (12) and (13) the coated grain has an


effectivethermalconductivityof

2,k
=2,.t.
I2,
S+22,./.
+2r/k
(2`
S-2,.1.)]
2,s+ 22,./.-rh,(2`s-2,./')

(14)

Equation(14)is rewrittenas

2,k
-2,.I'/2,S
-2,.1'

2,+22,./
=/2,S
+22,f

(22)

(15)

1-2,12,S

(23)

wherem -=1/(1- D) is usuallycalledthe "cementation


exponent,"
which is here rigorouslyidenticalto the cementationexponent
definedin theelectricalconductivityproblemfrom firstprinciples
(seeMendelsonand Cohen [1982] and AppendixB). Note that
the term cementationexponentis very misleadingas any change
in the valueof m doesnot necessarilycorrespond
to a cementation
process. The exponentm dependson the grain shape. The m
exponentis equal to 1.5 tbr sphericalgrains,and m takes values
>1.5 for nonspherical
grains.For example,the presence
of highly
oblate (disk-shaped)grains raisesm significantlyas shownby

The thermalconductivity
at stepk+l, 2,k+l'is relatedto the Mendelson and Cohen [1982]. Below I consider m as a
thermalconductivity
at stepk, 2,k,andto thesmallamountof
phenomenological
coefficientbecauseseveralmixing of grains
grainsadded,
AVk(Figure4) by

2,
k+l--2,
k=AV
k(2,S-2,k
)

2,++22, Vi+ Vg 2,S+22,k'

with variousshapescan lead to the samevalue of the cementation


exponentas shownby Mendelson and Cohen [1982]. However,

(16)

we do not haveto worry aboutthis problemas it shouldkept in


mind that m can be obtainedeasily from electricalconductivity
measurements.

whereVfandVgarethetotalvolumes
of thefluidandgrains,

Let usdefinethekey dimensionless


conductivityratio:

respectively,
at thatstep. For an infinitesimalincrementof grains

dVg,
equation
(16)yields,
32, Vf + Vg 2,s+22,)
'

--

2,s

2,./'

(24)

(17)

Typical
values,
atroom
temperature,
for. and2,Saregiven
in

REVIL: THERMAL

CONDUCTIVITY

OF UNCONSOLIDATED

Table 1. Thermal Conductivity Data for Some Minerals

SEDIMENTS

16,753

Table 3. Thermal ConductivityData for Some Pore Fluids at


25C

Material
Quartz

2`s, W m-1C-1
8.4a, 8.0+0.5
b, 7.7+0.9
c

Fluids

2`f,Wm-1C
-1

Limestone
matrix

5.0a

Water

0.63a

Calcite
Dolomite
Anhydrite

3.26+0.23
c, 3.60
d
5.33+0.26
c, 5.51
d
6.32+0.13
c

Brine
Heptane
Freon
12

0.59
a
0.13
a
0.0096
a

Glass
1.05a
Vitreous/fused
silica 1.36d
Lead
34.3a

Argon
He
H2

0.0178a
0.14a
0.17a

Orthoclase
Plagioclase

2.32
d
1.53-2.36
d

Air
Oil

0.024
a,0.03
b
0.13
a

Biotite
Muscovite
Salt-NaC1

2.34d
2.21d
10.13
e

Alcohol
(ethyl)
Alcohol
(methyl)

0.17c
0.19c

Methane

0.029
d

Halite

6.5d

CO2

0.0145
d

aWoodside
andMessnet
[1961
].

aBeck
[1976].

bBeck
[1976].
dHorai
[1971
].

bBrigaud
[1989].
CWeast
[1970].
dMissenard
[1965].

cBrigaud
[1989].

e Missenard[1965].

The nondimensionalparameterf is a "thermal formationtactor"


Tables 1, 2, and 3. Note that (1) mineralsare usually anisotropic for the thermal conductivityproblem. This formationfactor and
and the values given are the arithmetic mean of the values the electricalformationfactor F (definedin AppendixB) are not
correspondingto the crystallographicaxis and (2) the thermal independentparameterssincerelatedby
conductivity of air depends on the water content [Missenard,
f = F1/(m-l).
(27)
1965]. For liquidsof geologicalinterest, is in the range2-100,
whereasfor gas-saturatedporousmedia, may be as high as
1000. Consequently,in the thermal conductivityproblem,both
Note thatif the porosityandthe electricalformationfactorare

the experimental and field data are located below the both independentlydetermined,f can be determinedfrom (27).
equation
isoconductivity
point
where
from(23)thecondition
(2`= 2`j)leadsEquation(25) canbe rewrittenas a second-degree

directly
tothecondition
(3,=As ) andtherefore
= 1.

Equation(23) is not very practicalas it is an equationof the

form2,/2,./.=
f(0, , 2,/2,/).A much
more
practical
relationship

would
bdoftheform2,?2`./.=
g(0, ). Such
a relationship
is

2, - 2, 20+
71(1_O)2
+02=0.

(28)

derivedbelow. For most granularmedia of interest,m -- 2.0 + 0.5,

The solutionof (28) is given by

and (23) can be written

2,!

2`!

'

(25)

2`=--f+(1-O)1-O+4(1-O)
2+4f
. (29)

2,s,Wm-1C-1

The validity of (29) by comparisonwith the original equation,


Equation (29) is the
fundamental equation derived in this paper. This is a new
equation,which relatesthe effective thermal conductivityto the
grain and pore fluid thermal conductivities,the porosity,and a
grain shape parameter m, which is easily determined from
electrical conductivity measurements. In the limit >> 1 the

Smectite

1.88+0.15a

asymptotes
oforder
0 and1in -1of(23)(using
(26),m= 2,and

Kaolinite
Chlorite

2.64+0.20a
2.77
b
3.26_+0.25
a ,4.4-4.9
b

Illite

1.85+0.23a

Clay (unspecified)
Mixedlayerclay
Natrolite
Chabazite
Stilbite

1.30-1.45
c
1.85+0.54
a
2.00b
1.22b
1.18b

f --O !-m

(26) (23), is analyzed in Appendix B.

Table 2. Thermal Conductivity Data for Clay Minerals and


Zeolites

Material

aBrigaud
[1989].

bHorai
[1971].

Cpoelchau
etal. [1997].

Newton'sbinomialexpansion)yield

O>>1

(31)

>>1[,
J 1+2(f--l)+0(0-2)+
....

Equation (30) shows that in the limit >> 1 the thermal


conductivitybecomesindependentof the grain conductivityand
thusthe thermalconductivityof an evacuatedporousaggregateis
zero,or at leastmuchsmallerthanthe thermalconductivityof the

16,754

REVIL: THERMAL CONDUCTIVITY OF UNCONSOLIDATED SEDIMENTS


0.8

Table 4. Cementation Exponent From the Effective Thermal


ConductivityData of SandsSaturatedWith Low Thermal
Conductivity Gases

Sample

#1, quartz sand

24 + 4

0.19

#2, lead shot

15 + 4

0.38

1.54

#3, quartzsand

12 + 2

0.36

1.71

#4, glassbeads

8+ 3

0.38

1.90

--

a,

-'

air

2.10
0.4

freon 12
.

#5, quartz sand

6+ I

0.55

1.48

0.2

grains. This is confirmedby the experimentaldataof Woodside


and Messmer[1961, Figure 11] determinedon unconsolidated
sands.Woodsideand Messmet [1961, p. 1698] noted "The

0
0

effectiveconductivities
in vacuoof all five packsstudiedwere
lessthanone-hundredth
of the conductivities
of the respective

0.01

0.02

0.03

0.04

Thermal
conductivity
ofsaturant,
f (Wm'1C
'1)

solid particles." Equation(31) showsthat the influenceof the Figure 5. Variation of effective thermal conductivityof
materialswith the conductivityof saturating
thermalconductivity
of the grainsinfluences
theeffectivethermal unconsolidated
conductivity
of thesediment
in a linearwayin thedomainof "low gasfor low thermalconductivitygases(experimentaldatafrom
and Messmer[ 1961]). The linesrepresents
the model
porefluidthermalconductivity."
Thereis an analogybetween Woodside

this equationand the linear "highsalinityasymptote"


in the derivedin the main text in the limit O >> 1. Each point
electrical
conductivity
problem
(equation
(B9),Appendix
B; see representsa thermal conductivity experiments, which are
alsoRevilet al. [1998a]).From(29) in thelimit p-->1 we have performedon five differentsamples(numbered1 to 5; see Table
4).
f -->1 andin thelimit -->0 wehavef -->+oo;therefore

lim = j.,

(32)

lim = s,

(33)

values,reported
in Table4, arein the samenarrowrangethan
thosedetermined
on sandsandsandstones
with low claycontent
(Figures
6a and6b). Thisshowsthatthecementation
exponent
determinedfrom electricalconductivitymeasurements
is in the
asrequiredfor internalconsistency
of themodel.
Thedatabase
of Woodside
andMessner[ 1961] is usedto test samerange(1.5-2.0) as the cementationexponentdetermined
the previous model. The measurementswere made on three fromthethermalconductivity
equation(equation(29)). The use
-->0

quartzsandpacks,a glass-bead
pack,anda leadshotpack ofthethermal
conductivity
ratio,k//j.
plotted
asa function
ofthe
saturated
withvariousfluidsof differentthermalconductivity.dimensionless
conductivity
ratio -=AS/ 2,.t.(Figure
7) tends
to
Thethermal
conductivities
of thematerial
grains
andtheporosity normalize the data according to the general theoretical
of thesamples
aregivenin TablesI and4, respectively.
I first considerations
made
byStroud
etal.,[1986].
Inother
words,
testthe "low fluid conductivity"limit of the model. In thislimit

= g(p,), wherethe functiong dependson the modelused. It

>> 1, equation(30) predictsthat the dependence


of the followsthat the geometricmodel,oftenusedto computethe
effectivethermalconductivityupon the pore fluid thermal effectivethermalconductivity
of granularporousmedia[e.g.,
arguments,
and
conductivity
is linear.Thisisin agreement
withtheexperimentalLovell,1984],cannotbejustifiedfromtheoretical
dataof Woodside
andMessnet
[1961](Figure
5). According
to thismodelis notableto explainwhythedataof Figure7a (with
(30) theproportionality
factoris equalto thethermalformation very differentgrain thermalconductivities)
plot on the same
factor
f, whichisrelated
totheporosity
by(26). Usingthevalues curve.ThefactthattheratioZ/2,fdepends
onlyontheratio
off determined
fromFigure
5 andtheporosity
of eachsample,
I --As/Zf andporosity
is a verygeneral
property,
whichalso
determined
the valuesof the cementation
exponent
m. These existsfor theelectricalconductivity
anddielectricproblems.The
agreement between the experimental data of Woodside and

k+
1

Messnet[ 1961] andtheprediction


of themodeldeveloped
in this
paperis very good. None of the modelsreportedin Table 5 are

ableto fit theexperimental


dataovertheirfull range(Figures
7b
and 7c).

A secondtestof the modelis givenby usingthe datafrom


Somerton[1992,p. 68]. The effectivethermalconductivity
of
sandsversusporosity for dry/solvent-saturated/brine-saturated

naturaloil sandsis presented


in Figure8. Again,the present
Figure 4. During the iterative processof the DEM scheme, modelfits the experimental
dataquitewell withinthe porosity
at each step a new grain is addedto the mixture defined at the
range0.28-0.50,exceptfor gas-saturated
sampleswith high
previous
step.If k is thethermalconductivity
of themixture porosities.Thermal conductivityof gas-saturated
granular
at stepk, thismixtureis usedto coata new grain at step (k+ 1),
materials
is low dueto thesmallcontiguity
betweenthegrains.
andtheeffectivethermalconductivity
at this step,k+l, is The modelpredictsthatthermalconductivity
of brine-saturated
determinedin a self-consistent
way and thenusedto coat a new
grain,and so on until the desiredporosityis reached.

sand increasesfourfold to sixfold over the same sand saturated

with air, in agreement


with the experimental
datapreviously

REVIL: THERMAL

CONDUCTIVITY

OF UNCONSOLIDATED

12 Pape
etal.
(1987)
I........'

lO

a.

SEDIMENTS

3.2. Clean Sand and Pure Shale

Both sandsand shales can be consideredas granular porous


materialswith an unconnectedmineral phaseeven if the grains
aretouchingeachother. The thermalconductivityof a cleansand
or a pureshaleis givenfrom (29) by

2,= f+--(1-)1-+ 1

1.5

O--;I,c
//;I,j(Pure
shale)
'
(35)
2,q
Z.f
(Clean
sand)

2.5

Cementation exponent, m

16,755

i I i i i I i i

f --

,
cpstms,/(l-ms,)
(Clean
sand)

[Sh
mh/(l-rash)
(Pureshale)
'

(36)

where
)Lqand2,c arethethermal
conductivities
ofthequartz
and
clayminerals,
respectively
(givenin Tables1 and2), Psa
andCPs
h
are the porosities of the clean sand and the pure shale

respectively,
andmSdandmSharethecementation
exponents
of
.

1.5

2.5

Cernentation exponent, rn

Figure 6.
Histogram of the cementation exponent m
determined from electrical conductivity measurements. (a)
Forty-seven samples of clay-free sands and sandstoneswith
low clay content,from Pape et al. [1987]; (b) Ninety-two clayfree to shaly sandstonesamplesfrom Sen et al. [1990].

clean sandand pure shale. The cementationexponentof clean


sandis in the range 1.4-2.1 (Figures6) with a meanvalue of 1.7.
For shales,the cementationexponentis in the range2.0-4.0 with a
mean value on the order 2.5 (Figure 10). For mudstones,Henry
[1997], Lauer-Lerrede et al. [1998], and Erickson and Jarrard
[1998] reportedcementationexponentsin the range 2.3-2.5, in
agreementwith the previousrange.
3.3. Sand Shale Mixtures

In deltaic-typeenvironments,for example,clay particlesand


sand grains are mixed in a broad range of sand shale ratios
[Matlack et al., 1988]. The resulting mixtures looks like those
shownin Figure 11. In this sectionan expression
for the thermal
discussed(Figures7 and 8). Note that for gas-saturated
granular
conductivityof a sandshalemixtureis derivedfrom the thermal
porousmaterialsthe thermalconductivityincreaseswith the gas
conductivity
of the two end-members
(cleansandandpureshale)
pressure
andgrainsize.The thermalconductivitycontribution
due
andclay content. In sand-shalemixtures,porosityis not a simple
to gas in the pore spaceis proportionalto the mean free path of
functionof the shalefraction. Porositiesof pure sandand shale
the gas molecules present in the pore space and inversely
are higherthanthe porosityof a sandshalemixture. Porosityis
proportionalto the densityof the gas. At high gaspressures
(say,
reducedas clay particlesfill the poresof a sandor as sandgrains
>0.5 MPa) the two effects cancel each other [Presley and
are dispersed
in a shalematrix. For mixturesof sandandclay the
Christensen,1997a, b], and the influenceof the gas pressureupon
porosityis givenby [Marion et al., 1992]:
the effective thermal conductivity can be neglected. The
influenceof grain size is not well understood,but the increaseof
P= Psa- qv(1- Psh
), 0 < q0v< Psa,
(37)
the effectivethermalconductivitywith grain size doesnot appear
rp= q)vrpsh, 1_>q)v ->(PSd,
(38)
to change much with gas pressure[Presley and Christensen,
1997a, b].

Equation (29) is also tested over the entire porosity range


(0.01-0.50) for cleansandsand sandstones
(Figure 9). The model
predictioncompareswell with the variousexperimentaldata from
Brigaud and Vasseur[1989] and Somerton[1992], with a mean
cementation exponent identical to that given from electrical
conductivitymeasurements(see Figure 6) and in the range 1.7 +
0.3. This agreementis surprisingbecausethe contiguitybetween
the grainscannotbe neglectedin sucha case. However, a closer
description of the grain to grain contacts of a quartz sand
undergoingdeformationby pressuresolutionshowsthe existence
of a thin fluid film at the grain-to-grain contacts(Figure 2 and
Revil [1999b]). This thin film is probablyresponsibleof a small
effective contiguity between the grains, which can explain the
goodagreementbetweenthe presentmodel and the experimental
data (otherwisemost of the heat would flow throughthe mineral
contact points and the present model should be modified to
accountfor this effect, seeBatchelor and O'Brien [ 1977]).

whereq0
v is the clayvolumefractionand Psaand cps
h are the
porositiesof the clean sand and pure shale end-members,
respectively(Figure 11). A sedimentwith a shalecontentwhich

ranges
fromq0
v = 0 to thatwhichjustfills all thespacebetween
thesandgrains(q0v=Psa)is calleda "clayeysand."A sediment
with a greatershalecontentthan this (q0v > Psa)is calleda
"sandyshale". The porosityof sandshalemixturehasa minimum
at the clayey sand/sandyshale boundarygiven by p= cp&tcPs
h.
Koltermannand Gorelick [1995] showthat (37) and (38) are good
first-order approximationsto the porosity of a binary mixture
assumingideal packing(seeKohermannand Gorelick[1995], for
nonidealpacking). Ideal packingis a good approximationwhen
the ratio between the coarsestand the finest grain size is large
(say, >10). This is the casefor sandshalemixtureswhere sand
grain diametersare >50 gm and clay grain diameter is <5 gm.
The relationships
betweenthe shalecontentby volume q0v andby
weight q0w is then[Marion et al., 1992]

16,756

REVIL:THERMALCONDUCTIVITY
OFUNCONSOLIDATED
SEDIMENTS
lOO

lOO

Thismodel

b.

#2, Lead shot


#3, Quartz sand
#4, Glass beads

--

#1, Quartz sand

-'

#5, Quartz sand

&&

Porosity = 0.19
m = 2.02

Porosity :
m:

Porosity = 0.36-0.38
m=

10

1.63

1O0

1000

lOO

10000

10

100

1000

0.51-0.59

1.60

Arithmetic

10000

Arithmetic
Krupiczka

Krupiczka

This model

Harmonic
'

Porosity
= 0.19

m=200

.
10

100
S

1000

104

10

. m= 2.00
100
S

1000

104

Figure
7.(aand
b)Variation
ofeffective
thermal
conductivity
ratio
A/A,.
with
theconductivity
ratio
= AS/
27.
(experimental
data
from
Woodside
andMessmet
[1961
]). Thesolid
linrepresents
themodel
prediction.
(c
andd)Comparison
between
thepresent
model
andvarious
models
taken
fromtheliterature
(Figures
7cand7d

canbe comparedto the Figures7a and7b).

Table5. MixingFormulaeUsuallyUsedto Computethe


EffectiveThermalConductivityas a Functionof
ConductivityRatio andPorosity
Model

(PW=

(pv(
1_CPs
h)ph
Sh

Sd'

(PV-<CPSd,(39)

(PV(1-CPsh)Pg
+(1-CPsa)p

Equation

(l_Osa)pa
+CPsa(l_CPsh)ph,
(PV
=CPsa,
(40)

Arithmetic A/At = p+(1-qS)

Maxwell A/Aj=2p
+(3- 2p)
3-+o

(PW=

(PV
(1-CPsh
)PJ
h
Sh

Sd'

(PV(1-CPSh)Pg
+(1-(pv)Pg

Pv->cPSd
. (41)

A/A.!
=OA+Blg
O
Krupiczka

A = 0.280- 0.757log10p;
B =-0.057

Here ph and p"/are the graindensities


of clayparticles
(without
bound
water)andquartz,
respectively.
From(37)-(41),
theporosity
isrelatedto (Pwby

Harmonic

2,/2,./=
o+(]-)

Sd

c)= CPS
d-

(1
-Sh
cPSd
)Pg
(PW
_
, (PW
<
fig (1- (PW)

'

(42)

REVIL: THERMAL CONDUCTIVITY OF UNCONSOLIDATED SEDIMENTS

0o

- oilsands(m= 2, Xs = 4.45W m'1C'1)

2.5o

16,757

Shale

' 2.00
'o

1.50

Brine
saturated

?_

Solvent
saturated

1.00

0.50 _l
I

0.00

[ ,

2.5

3.5

Cementation exponent, m

0.28

1.5

0.32

0.36

0.4

Figure 10.
Histogram of cementation exponent rn
determinedfrom electrical conductivitymeasurementsfor shale
and very shaly sediments. Data are from Waxman and Smits
[ 1968], Brown [1988], and Sen et al. [ 1990].

0.44

Porosity

Figure 8. Variation of measuredthermal conductivityof oil


sandscomparedto the prediction from the model developedin
this paper (solid lines). The experimental data are from
Somerton[1992, p. 68]. The mineral compositionof the oil

conductivityresponseof suchmixturestomthe resultobtainedin


section 3.2 for fluid-saturated

sands
yields
Zs = 4.45W m-lC
-1using
a weighted
arithmetic
average[seeSomerton, 1992, p. 68, Table V-5].

Parameters

used
are
dry
oZf
0.055
Wm-1C-I,
solvent-saturated
Z),.
=
0.125
W m-1 C-, andbrine-saturated
Zf= 0.590Wm-1C-,

sands:

2,s -->2,q,
2,I. - 2,Sh,

(44)
(45)

--->q0v,

(46)

f '-->
(q0v)ms,t/(I-ms,t)
=(q0V)-3,

(47)

cementationexponentm = 2.

whereZq is thethermalconductivity
of quartz,Zshis the
thermalconductivityof the shalematrix given by (34)-(36), and

Sd

OShPg
OW
Sh
'
Pg PW
+(I-Osh)Pg'(I-PW)

0W_>
0it.

0 = Sd

exponent
associated
with thesand
(43) mSd= 1.5is thecementation

In the sandy shale domain, and accordingto Figure 12, I use


the following changeof variable to deducedirectly the thermal

Quartzsand

(brine

The thermal

2,=
(0V)
32,ShI(O
V)-3
O+
1(1-13)
x(1-13+-\/(1-13)
2+4(qOv)-313)l,l>qo
d,(48)

saturated)

r=0.93

:;, 6

grains alone [see Revil and Cathles, 1999].


conductivityof a sandyshaleis thereforegivenby

'--Sd

Quartz

:i.%..i:i:i:.:i:..i.....:
Clay
-- Sh

1
-

q0v- 0

Best fit of the rn

Zs
=7.2
+0.4
Wm"
C"
I

m
=1.70
+0.06I

,,,,I

....

0.1

I,,,,I,,,,

0.2

0.3

Sand

Sandy shale

Shale

Figure 11. Sand-shale mixtures with increase of the shale


content from the left to the right. For a clay-free sand (first

I,,,,I

0.4

q0v< )sa q0v- q)saq0v> q)saq0v- 1


Clayey Sand

0.5

Porosity,

Figure 9. Effect of porosity on the thermal conductivity of


brine-saturated
clean sandsand sandstones. Fluid conductivity

end-member),the noncompacted
porosityis equalto Pst.For
a clayey sandthe porosity decreasesowing to the presenceof

clay particlesin the pore space. This decreasecontinues until


the critical point whereall the pore space of a clean sand is
used
isZf=0.59Wm-] C
-1.Thebest
fitofthemodel
(plain occupiedby clay particles, that is when the shale content is
line) leads to a cementation exponent of 1.70 in agreement equalto the porosity of a clean sand. After this point, the
with cementation exponent of clean sands determined from sedimentis a sandy shale, and an increase in shale content is
electrical conductivity measurements.Experimental data from only possible through replacementof quartzgrains by clay
Zamora et al. [1993], circles; Somerton [ 1992, Table B 1],
particles, and the porosity increaseswith the shale content.
squares;Brigaudand Vasseur[1989], triangles; and Somerton The secondend-memberis a pure shale with no quartzgrains
[1992, Table B3], inverted triangles.
andwith a porosityequalto Psh.

16,758

REVIL: THERMAL

Shale matrix

(tpv)
%

CONDUCTIVITY

OF UNCONSOLIDATED

sand,i.e., v =sa, and 2,(s


a) can be calculatedfrom (48).

Sandgrains

SEDIMENTS

The predictionof the previousequations(equations(48)-(50)), are


shownin Figure 14 togetherwith experimentaldatafromBrigaud
[1989, p. 74]. There is a quite a good agreementbetweenthe
model and thesedata. The thermal conductivity of a sand shale
mixture reachesa maximum when the shale fraction entirely fills
the pore spaceof a sandand two minima for the clay-freesand
and pure shale end-members. Unfortunately, measurementsof
thermal conductivity for sand shale mixtures with known
porosity,shale content,and compactionstate were not found in
the literatureto furthertestthe presentmodel.
A simpler versionof the previousmodel is to use (29) and a
clay contentweightedarithmeticmean for the grain conductivity.
This leadsto the following set of equations:

(-tpv)

Figure 12. Sandgrains dispersedin a shale matrix (sandy


shale). The situationis analogousto a suspensionof spherical
particlesin water in the limit of no contiguity betweenthe
grains.

O-- 2,q/ 2,sh,

Volumic clay content

0.60'

0.2

0.4

0.6

0.8

'''I'''!'''I'''I',

(49)

where/Sh
isgivenby (34)-(36).
In the clayey sand domain the influence of the clay content
uponthe effectivethermal conductivityis very similar to that of
the brine for partially brine-saturatedquartz sands(seeFigure 13
and section 3.4). Because the clays are attached to the sand
grains, the thermal conductivity of the clay assemblagehas a
greater effect than the thermal conductivity of the pore fluid.
Owing to the analogy between the two problems,an empirical
relationship,given by Somerton[1992] in the contextof partially
brine-saturated
quartzsands,can be usedto estimatethe influence
of the clay contentin the clayey sanddomain. This yields

2,=2,sa
+[2,(sa)-2,sa](q)
v/sa)
i/2,for0<v<sa,
(50)

0.1
................:

o..
E

where2,(Sd) is the thermalconductivity


at the criticallimit for
which the clay particles entirely fill the pore spaceof a clean

The clayey sand problem:

......

(4)

I'"'-x...
:'.,,......,i.....!;,;i!
-.:: : quartz
shale
I-l brine

toy= 0
Cleansand

0 < toy< sa
Clayey
sand

toy=

..c:

LU

(2)

Clay-filled
sand

..t4*:'

$w=0
Gas-saturated

0_<$w_<1
Partiallysaturated

02.

0.4

0.6

0.8

Volumic clay content

The partiallybrine-saturatedsand problem:

i:..-'.'':"'.,.::
.......... . '*.......

:'"'-:'
quartz
brine
I-l gas

..:::2.i,'-';:;'";(-.

$w = 1
Brine-saturated

Figure 13. Similarity between the clayey sandproblem and


the partially gas-saturatedsand problem.

Figure 14.
Effect of clay content on the thermal
conductivity of brine-saturatedsand/shale mixtures. Fluid

conductivities
used
are27=0.60Wm- C-I,q= 8 Wm- C, mSd = 1.8,mSh =2.5, Sd
i, 2,c- 3 Wm-l C-
=0.40, and

sh= 0.60 (curve 1), 0.50 (curve2), 0.40 (curve 3), 0.30
(curve4). Experimentaldataare from Brigaud[1989, p. 74] for
kaolinirewith variablequartz content (open circles) and quartz
sandswith little or no clay fraction (plain circles) and from
Somerton [1992, p. 237, Table B8] for silty quartzsandstones
(squares).

REVIL: THERMAL

CONDUCTIVITY

OF UNCONSOLIDATED

+4fO)1,
(51) ? 3.5

(52)

'E

Quartz

SEDIMENTS

16,759

sand

- 0.335

(53)
2.5

Notethattpw(andnottpv)is usedherebecause
tpvrefersto the
shalecontentin volume with its associatedpore volume, whereas

tpwreferscorrectlyto the appropriate


clay mineralframework
contentbecauseclay minerals (without their bound water) and
quartzgrainshave roughlythe samegrain density. From these
equations the thermal conductivity of the sediment can be

computed
withthehelpof theporosity0, theclaycontenttpw(in
weight) and the electrical cementationexponentm. If electrical
conductivitydata are not availableto determinethe cementation
exponent,a crude estimateof this parametermay be obtainedby
the tollowing empiricalrelationship:
(54)

m = msd(1- tPW) + mShq)W,

-o

1.5

,E

Modelprediction,
1

m - 1.6' ,

- 8 W m' C'

._>

( 0.5
0

0.2

0.4

0.6

0.8

Brinesaturation
Sw
Figure 16. Measured thermal conductivity of partially brine
saturatedquartzsands. Parametersusedfor the model are m =

2,s= 8.0Wm-C-,
2,j(brine)
= 0.59Wm- C-I,2,j(air)
=
wheremsc
t = 1.8andmsh= 2.5 (Figures5 and9). Theprediction 1.6,

of (51)-(53) are in fair agreementwith the availableexperimental 0.030 W m- C-1 Experimentaldata are from Somerton
data (Figure 15). Unfortunately, no experimental data are [1992, p.72].
available to further test our model.

suggested
the following empiricalrelationshipto determinethe
3.4.

Influence

of the Gas Content

influenceof thebrinesaturation
Sw upontheeffectivethermal

conductivity:
In this section the effect of gas content upon the thermal
conductivity of an initially brine-saturated granular porous
2,(Sw)
= 2,(0)
+[2,(1)-)4o)]sw
/2,
(55)
materialis studied. Becausebrine is the wetting fluid, the thermal
conductivityof the brinehasa greatereffectthanthatof gasupon
where2,(0)and2,(1)referto theeffectivethermalconductivity
of
thermalconductivityof the sediment. Furthermore,the brine has
the gas-saturatedand brine-saturatedsediment,respectively,
a higher thermal conductivitythan any other fluid which might
inthispaper
from
(29)inwhich
. corresponds
tothe
occupythe interconnectedpore space. Somerton[1992, p. 71] determined
thermalconductivityof the gasandthe brine,respectively.There
is goodagreementbetweenthesedata and the predictionof (55)
linkedto (29), within the accuracyof the experimentaldata(+5%)

i
C)

(Figure16). Note thata smallamountof waterin a gas-saturated


granularporousmaterialcanincreasethe thermalconductivityby
more than 100%, whereasa small amount of gas in a watersaturatedgranularporousmaterialhas only a small effect upon
theeffectivethermalconductivityof the material.

Porosity < 0.25


Porosity > 0.25
Predicted

4. GeophysicalApplications
/

0=0.05

The modelderivedin section3 offersthe possibilityto derive


thermalconductivityprofiles from the analysisof compaction
profiles(section4.1), to investigatethe role of gasin the change
of the geothermalgradient in sedimentarybasins(section4.2),
and to deriveeffectivethermalconductivityprofilesfrom in situ
downholemeasurements
suchas naturalradioactivity,electrical
resistivity,andbulk density(section4.3).

0=0.60
o

0.2

0.4

0.6

0.8

Clay content
Figure 15.
Effect of clay content and porosity on the
thermal conductivity of brine- and water-saturatedsand shale
mixturesand silty sandsones. The solid lines representthe

predictions
ofthemodel.
Parameters
used
are27= 0.60Wm']
c-l,2,q
=8Wm-1C'I,2,c,= 3Wm'] C'](anupper
limitfor
thethermalconductivity
of theclays),mSd= 1.8, mSh= 2.5.

Experimentaldataare from:Brigaud [1989, p. 74], circles, and


Somerton[1992, p. 237, Table B8], squares.

4.1. Compactionand Thermal ConductivityProfiles


In the first hundredmetersbelow the seafloorin sedimentary
basins the porosity decreases strongly with depth from a
depositional porosity in the range 0.70-0.80 to porosities
approximatelyin the range0.40-0.50. It is expectedthat thermal
conductivityincreasesin this depth interval as a result of the
sedimentcompaction.
Consolidation
of the soupysediments
presentin the first 300 m
below the seafloor (mbsf) is mainly the result of what is called

16,760

REVIL:

THERMAL

CONDUCTIVITY

OF UNCONSOLIDATED

"mechanicalcompaction." Mechanicalcompactionresultsfrom
slippageand rotationsof grainswhich changetheir positionand
orientation,but not their shape(Figure 1), in order to reach a
denserarrangement[e.g., Rieke and Chilingarian, 1974]. We
consider as a reasonableassumptionthat during mechanical
compaction,the porositychangeis proportionalto the effective
stress
change,dO'ef
t , andto thedifferencebetweenthe porosity
andtheresidualporositycresultingfrom the first consolidation
step. The proportionalityconstantis a pseudo-poremechanical

SEDIMENTS

We note 0theporosityin a referencestatewith no stressapplied


(the referencestate is taken at the top of the sedimentarybasin
andtherefore0is the depositional
porosity). We integratenow
(56) to get the porosity/depthrelationshipin the hydrostatic
pressuredformations. The changesin lithostaticand pore fluid
pressures
underhydrostaticpore pressureconditionsover a depth
incrementdz are givenby

aV(z):

compressibility
noted r,, (tim is consideredas a constant

dO'ef
t
dt

> 0.

(56)

Unit II

(58)

wherepj. isthedensity
of thefluidin theinterconnected
pore
space
(assumed
tobeconstant),
pg isthegraindensity,
andg is
accelerationof gravity. Since the differential of the effective

Unit I

lOO

(57)

dp(z)= p/ gdz,

independent
of temperature,porosity,and effectivestress).Under
thepreviousassumption,
mechanical
compaction
is described
by

de=-flm(-c)dO'eff,as -

['''I'''--''I'''I'''

200
ISubunit[

Ilia

. 300
v

) 400

mputed thermal

._

S11bl
nit

> 500

6OO
Smectite

ODP
0

....
10

....
20

' ....
30

Site

....
40

clay
fraction

of the

860

....
50

'
60

70

2.2

2.4

2.6

2.8

Thermal
conductivity
oftheclayfraction
(W m" C")

Compositionof the clayfraction(%)

ODP
Site
860
/ unitl
::::::
! :?!ii
ili\Shaly
sand
line

100

e Unit
II :i"'"'i
:'}V
m
..:::
.::,::
':....

200
300

conductivity

Chlori

700 ....

Subunit
Ilia
-

400

Subunit

500

6OO

s= 2 Wm"C"Predicted
' i1,,I,,,,I,,,,I

7OO
o

0.1

,
0.2

0.3

range

,,I,,,,I,,,,I,,,,I

0.4

0.5

0.6

Porosity
(fraction)

0.7

0.0

0.50

1.0

1.5

2.0

2.5

3.0

Thermal
conductivity
(Wm"C'1)

Figure 17a. Clay fraction, thermalconductivityof the clay fraction, porosity versusdepth, and computed
conductivityprofilesat ODP Site 860.

REVIL: THERMAL CONDUCTIVITY OF UNCONSOLIDATED SEDIMENTS

16,761

stressis givenby do.ef


t do'-dp = dP-dp, equations(57) and to the way the sedimentcompactswith depth. This is illustrated
in Figures 17a and 17b, in which two setsof data from two ODP
boreholesare analyzed[Behrmannet al., 1992].
ODP Site 860 is located in the region of the Chile triple
junction, near the seaward site of a small forearc basin. Three
Integrationof (59) from the top surfaceof the sedimentarybasin lithologicalunits are identified at this site. Unit I (0-87.7 mbsf) is
composedby clayey to silty clay with nannofossilsand with
(z = 0, = 0) to depthz yields
gradedsilt and sandinterbeds,and one 10-m thick massivesand
[Behrmannet at, 1992, p. 159]. Unit II is formedby claystoneto
silty claystone plus sandstonesand thin conglomerate beds.

(58) are combinedwith (56), yielding

d=-(-c)(l-)mg(t9g-Oj.)dg.
(59)

(z)=
0
-.
+(10).
exp(z/Zc)(60)
,

P0
-c+(1- 0)exp(z/Zc)

Subunits IIIA and IIIB (242.5-309.8

I/Zc--(l-c)Smg(pg
-pf).

(61)

Equations (60) and (61) can be useful to parametrize


compactionprofiles and to derive thermal conductivityprofiles
from the mineralogicalcompositionof the formationsin addition

mectiteChlodte

mbsf and 309.8-617.8 mbsf)

includes clayey siltstone, silty claystone (with or without


nannofossils)plus sandy silty claystonewith thin conglomerate
beds. The clay contentis -50% and varies from 0 to 80%. The
carbonate content is -2%

and varies from 0 to 10%.

Unit
IA

UnitlB -

lOO

Unit IIA .

o
o

200

300

Unit lib -

x: 400

._

500

600

UnitIIC

/-

Computedthermal

conductivityof the

7OO

8OO

,,,,

I,,,,I,,,,

I,

10

20

30

,,,

40

sift

50

60

I ....

I ....

I ....

clay
fraction
.

2.2

I '-'='

''1

2.6

UnitlB

e
/

200

2.8

'''1 .... I .... I .... I'''

ODP
Site
863_ Unit
IA

lOO

2.4

Thermalconductivity
oftheclayfraction
(W m" C4)

Compositionof the clay fraction(%)

....

70

Unit
,,A

1.8 (sandstone)

300

..tl, Unit
lib

400
500

' .
600

;;::::
.........

m = 3.0

'

...?73?
'"'":;;:::'

(shale)

Unit IIC

7OO

,,,I,

800
0

, , I1,,,,

0.1

0.2

I ,,,

0.3

, I,,

, , I ,,

0.4

0.5

Porosity
(fraction)

Predicted
range

0.6

The sand

fraction is mainly composedof quartz plus some feldspar. The

=1,,,,I,,,,I,,,,1,1,,,,I,,,,I,,,

0.5

1.5

2.5

3.5

Thermal
conductivity
(Wm4 C
4)

Figure 17b. Same as Figure 17a, except for Site 863.

16,762

REVIL: THERMAL CONDUCTIVITY OF UNCONSOLIDATEDSEDIMENTS

clay fraction is mainly composedof chlorite (-50 + 5%), illite

field data supportthe existenceof gas capillarysealingas a


powerfulway to maintainpore fluid overpressure
over time in
sedimentary
basins
filledby siliciclastic
sediments.
Thepresence
assemblage
is 2.5 W m'1C'1. Takingall themineralogical
of gas has a strong influence upon the effective thermal
composition
into accountleadsto a meanvalue for the grain conductivity
of granularmaterials.Indeed,a gassandwouldhave
thermal
conductivity
equal
to4.0W m-1oC-1.Fitting
theporositya thermalconductivity
in therangeto thatusuallyassociated
with
(-44 + 5%), andsmectite(-6 + 4%). Usingthe valuesgivenin
Table2, a lowerboundfor thethermalconductivity
of theclay

datawith(60)(Figure
17a)leads
toP0
= 0.567,Pc
= 0.33,zc= 208 a water-saturatedshale(compareFigures14 and 16). Therefore
m. With theknowledge
of thethermalconductivity
of thegrains
andtheporosity,
we cancompute
a thermalconductivity
profile
using(29) assuming
that the cementation
exponentlies in the
range 1.8-3.0. This defines an enveloppefor the thermal
conductivityversusdepth, which is in agreementwith the
availableexperimental
data(Figure17a).A similarapproach
is
usedto compute
thethermalconductivity
profilesversus
depthat
ODP Site 863 (Figure 17b). Site 863 is locatedat the baseof the

trenchslopeof theChiletrenchat thepointwheretheChileRidge


is beingsubducted.
Themineralogical
variations
withdepthare
fairlywellknownat thissite[Behrmann
et al., 1992],andthey
wereusedto computea thermalconductivity
of thegrainfraction
at eachdepthratherthanto usea meanvaluelike in theprevious

thecorrelation
between
thetopof porefluidoverpressure
andthe
changein the geothermalgradientcouldbe both associated
with

thepresence
of freegascapillaryseal. Thisassumption
hasstill
to be tested further as more and more field casesare documented
in the Gulf Coast of Mexico.

4.3. DownholeMeasurementsAnalysis

Unconsolidated
coresamples
canbe alteredby drillingand/or
gas decompression. It follows that core measurementsare not

alwaysreliableto getan ideaof thein situthermalconductivity


profilein the poorlyconsolidated
part of sedimentary
basins.
Another possibility is to rely on downhole measurements.

theirworkby
case.
Fitting
theporosity
datawith(60)(Figure
17a)leads
toP0
= However,Blackwelland Steele[1989]concluded

0.48,Pc= 0.04, zc = 569 m. Thereis a verygoodagreement statingthat in fact there is no enoughinformationto estimate
effectivelyfor a section
of sedimentary
rock.
between
themodelandtheexperimental
data,asshown
in Figure thermalconductivity
17b.
In thissectionI will showthatthemodeldeveloped
in section3
may be usedto determinethermalconductivityprofilesfrom
4.2. Influence of Gas Saturation on the Geothermal Gradient

In deltaic-typesedimentarybasinslike in the Gulf Coast of


Mexico, an important volume of silicoclastic sedimentscan be
depositedover shortperiodsof time. In theseenvironmentsit has

electricalresistivity,bulkdensity,andgammaray measurements,
whichproperties
areeasilymeasured
in boreholes.
Suchtypesof
analysiswould complementdirect measurements
of the thermal

conductivity
in sedimentary
basins,whicharetechnically
difficult

been observeda correlationbetweenthe top of overpressured andtime consuming[e.g., Sillimanand Neuzil, 1990,Somerton,
sediments
anda brutalchangein thetemperature/depth
curve(the 1992, and referencestherein]. Previous works have correlated
conductivity
datawithvarious
logs(e.g.,neutron
porosity
slopeof this curve representsthe geothermalgradient)[e.g., thermal
Blackwelland Steele,1989, Figure2.10]. Beboutet al. [1979] index,sonicintervaltraveltime,bulkdensity,andgammaray
reported,for example,a summaryof "equilibrium"bottomhole
temperature
obtainedin the Harris County(Texas) showinga

geothermal
gradient
of 29 + 3Ckm-1 in thehydrostatically

Mineralogicalinversionfrom downhole
measurements

pressured
sectionof the sedimentary
columsanda temperature

gradient
of theorderof 43 + 4Ckm-1 in theoverpressured
section.Suchan observation
is veryclassicalall alongtheGulf
Coastof Mexico(L.M. Cathies,personal
communication,
1997).
Blackwelland Steele [1989] have shownthat the maintenanceof

hightemperature
gradients
throughthetopof overpressure
cannot
be attributed
to fluid flow in the overpressured
section.They
suggested
thatthevariations
mustbepartlyor predominantly
due
to thecontrast
in thermalconductivity
between
theoverpressures Electrical
and hydrostaticallypressuredsections. Indeed, it has been

suggested
by numerous
authorsthattheoccurrence
of porefluid
pressure
in the Gulf Coastis associated
with the presence
of
massive(low permeability)shale units. As shale have a lower

thermalconductivity
than sands(as shownin section3), it is
expected therefore that the geothermal gradient in the
predominantlyshale sectionwould be higher than in the
predominantlysandsection.However, severalcasestudieshave

beenreportedin whichthe transitionbetweenhydrostatic


to
overpressures
is notcorrelated
withmajorlithologicchanges
in
theshalecontent
andthereis stilla strongchange
in thevalueof

resistivity
Bulk

density
Figure

18.

General flow chart used to determine the

effective thermal conductivity of poorly consolidated


sediments
fromdownholemeasurements.
In a generalscheme,
downhole measurements can be inverted to determine the

mineralogy(for example,the shalecontentandclay type in

thegeothermal
gradient[e.g.,Coelho,1997].
sandshalemixtures),the electrical formation factor F, and the
Here I proposean explanationfor thesefield observations. porosity. The porosity and the electrical formation factor can

Thisexplanation
is based
onthemodeldeveloped
in section
3 and be usedto computethe thermalformationfactorf and the
thepossibility
of capillary
sealing
beingresponsible
of porefluid mineralogy to determine a mean value for the thermal
overpressure
[e.g.,Watts,1987;Revilet al., 1998b].A capillary conductivity of the grain framework. Then the effective
sealis a flowbarriergenerated
whenfreegas(oroil) ispresent
in thermal conductivityof the sediment mixture is determined
layered
sediments
withgrainsizevariations.
Bothlaboratory
and from the modeldevelopedin the main text.

REVIL: THERMAL

CONDUCTIVITY

OF UNCONSOLIDATED

SEDIMENTS

16,763

p= (l-rp)pg+rppf,

logs),but thesecorrelations
havebeenratherempiricalin nature,
andtheydid not providedvery convincingresults[e.g., Vacquier

a.f[!+2(F-l)/]

(63)

(64)

et al., 1988; Somerton, 1992, and referencestherein]. Empirical


o'-' --,
correlations between well logs and thermal conductivity are
determinedfor one suite of samples from a given geological in order to determinethe porosity p,the clay content(in weight)
W,andthe cementation
exponentm. Here},is the gammaray
environment,and they usually fail for other suites of samples q0

fromothergeological
environments.
Williamsand Andersonreading,
YSd
is thegamma
rayvalueof a cleansand(-10 gamma
[1990] attemptedto developa phononconductionmodelfor low- rayunits),
}tSh
isthegamma
rayvalue
ofpure
shale,
pgisthegrain
(-2650 kg m-3),o' is theelectrical
conductivity
of the
porositycrystallinerocks, which yields thermal conductivity density
profilesdirectly from the densityand sonicvelocity well logs. formations
(inverse
oftheelectrical
resistivity),
o).istheelectrical
This method has been shown to be accurate only in specific conductivity of the pore fluid, and F is the electrical formation
environments[e.g., Pribnow et al., 1993] and is not appropriate factor. The gamma ray of a pure shale and the cementation
for high-porosity granular sediments. Here we propose to exponentare computedfrom [e.g., Revil et al, 1998a]:
determine the effective thermal conductivity from the general
YSh= Z XiYi,
(65)
flow chart shown in Figure 18. As we specializedhere to sand
shale mixtures,

we show below

that the effective

conductivity can be derived from electrical resistivity/bulk


density/gammaray data (commonlymeasuredin boreholesin the
Ocean Drilling Program and in the oil industry), on the basis of
the model developedin section3.
Downhole measurementsof the naturalradioactivity,formation
density,and electricalresistivityin shallowsandshalemixtfires
can be interpretedwith the following set of relationships[Revil et
al, 1998a]:
(62)

7 = (1- q)w)YSd+ OW}/Sh,

,
!

thermal

Shale

Porosity

F = p-m,

whereXi aretheclayratiosof eachclaymineralto thetotalshale


fractionand7i is thegammaraylevelfor eachclaymineral.In
additionto the downholedata, the dependenceof the grain and
pore fluid thermalconductivitieson temperatureis needed. They
are providedin AppendixC.
An example of such an application is given for downhole
measurements
madein a boreholelocatedin SouthEugeneIsland
(SEI) salt withdrawal minibasin (offshore Louisiana, Gulf Coast

content

(weight)

Thermal
conductivity
(W m' C')

800

lOOO

12oo

> 14oo

16oo

I
Clayey
i]
Sandy
'(;(1 ' ,
, t, , , , , , , , ,

"'"'
Sandylayers
sand
l,]'--"
shale
domain
[',,
domain

2000
''''''''''''''
0.20

0.30

0.40

0.50

(66)

50

Figure 19. Thermalconductivityprofile. Porosityandclay contentare determinedfrom the density and


gammaray logsaccording
to equations
(62)-(63). The porositydecreases
with depthas a resultof compaction
High frequencyvariations in the porosity profile results from clay content variations. The thermal
conductivity
profiledeterminedfrom the porosity,the clay content,andthe modelderivedin section3 exhibits
two kind of spatial variation: (1) the global trendcorrespondsto the effect of compactionwith low thermal
conductivityat shallowdepthsand (2) the high-frequencyvariationscorrespondto variations in shale content
with high thermalconductivitiesat the critical limit betweenthe clayey sandand sandyshaledomains.

16,764

REVIL: THERMAL CONDUCTIVITY

OF UNCONSOLIDATED

of Mexico). The SEI basin is a young Pleistocenepassiveshelfmargin minibasin, which contains very prolific oil and gas
reservoirs. During the last 2.8 Myra sequenceof several
kilometersof sandshalemixtureswas depositedover pre-Tertiary
sediments[Holland et al., 1990]. The sand fraction is mainly
composedof quartzsandswith minor amountsof carbonates
and
the clay fractionis composedof mixed layer (ML) clays(64.5%),
kaolinite (K, 15.5%), chlorite (C, 18.25%) and illite (I, 1.75%).

SEDIMENTS

Appendix A: Maxwell Formula


The purposeof this appendixis to review briefly the classical
problem of the temperature distribution around a sphere of

thermalconductivity
A,
2 andradiusR immersed
in a fluid of
thermalconductivity
A,1. The system(sphereplusfluid) is
submitted
toa uniform
thermal
gradient
{VT)= (AT/L), where
3, is the unit vector in the z direction. The temperature
distributionafter introduction of the sphere obeys the Laplace

Using
s(ML)= 1.85W m-1C-l,,s(K)= 2.7Wm-1C-l,,s(C) equation,
V2T=0. It follows
thatin a spherical
coordinate
= 4.0W m-1C-l,,s(I)= 1.85W m-1C-1(seeTable2),the system(r, 0, go)with axis such as 0 = 0 coincides with the
mean value of the thermal conductivity of the clay mineral

of theapplied
thermal
gradient{VT), thetemperature
fraction
isA,
c= 2.4W m-l c-l Thethermal
conductivity
ofthe direction
distribution takes the form

quartz
sand
mineral
fraction
isA,q
= 7.0Wm-1C'I(accounting
for the small amountof carbonates). The pore fluid conductivity
versus depth is determined from equation (C3) (Appendix C)
using a temperature-depthprofile obtained from temperature
measurementsin this area and in this depth interval (which

corresponds
to thehydrostatically
pressured
section):
T(z) = TO+

Gz where
TO= 20C
andG= 25Ckm-1andzisthetruevertical
depth (in km). The clay content, porosity, and cementation
exponent are calculated using the density, gamma ray, and
electrical resistivity downhole measurements and (62)-(66)
(Figure 19). The thermalconductivityprofile is computedusing
(51)-(53). The thermalconductivityprofile exhibitsa large-scale
trend due to compaction of the sediments and small-scale
variationsdue to variationsin the shalecontent(Figure 19).

(A1)

r>R'T(r,O)=
(VT).r
+A(VT).-53,
r _<R' r(r,O)=B (Vr).r,

(A2)

<VT>.r=(----)rcosO.
(A3)

The first term of (A1) correspondsto the farfield temperature


gradient and the second term representsthe influence of the
sphere(a =-A represents
the thermalpolarizabilityof the grain).
The boundary conditions correspond to the continuity of the
temperatureandheatflux,

T(R+,O)=T(R-,O),

(A4)

8T(R+,O)
ST(R-,O)
A,2
= Aq,

5. Concluding Statements

3r

(A5)

3r

A model for the effective thermal conductivity of granular whereR+ andR-representthe externaland internallimit of
sediments has been developed using a differential effective
r --->R. This leadsdirectly to
medium scheme. For such sediments,the thermal conductivity
depends on the mineral and fluid thermal conductivities, the
porosity,and the cementationexponent,which representsa grain
shapeparameter. The cementationexponentcan be determined
from the electricalconductivityformation factor and thereforeby
2 + 2
electricalconductivitymeasurements.The pore spaceis assumed
to be fully connected,and the contiguity betweenthe grains is
assumedto be small. The model performswell on laboratorydata The thermal dipole strength is defined by (see section 3.1,
with natural and artificial granular porous media. The main equation(10))

-l
+A,2
A=R3/
/l-/2
/'

(A6)

B= 3,

(A7)

S---(A,
2- , )I VT(r,0)dV.

(A8)

conclusions reached here are as follows:


1. Electrical

and thermal

conductivities

can be unified

into a

singleframework,and the influenceof the grain shapeupon the


effective thermal conductivity of a sediment can be evaluated
with thehelp of electricalconductivitymeasurements.
2. New analyticalexpressionsfor the thermalconductivityof
clay-free sands,pure shales,clayey sands,and sandyshale are
derived as a function of the porosity, cementationexponent,
thermal conductivitiesof the grains and pore fluid, and clay
content for sand shale mixtures.

( 3A'l
I(VT)
rcosO,
(A9)

r_<
R'T(r,O)
= 2/+2
2

S=---4
rR313
2q
(22-/1
)](VT).
3

2 + 21

(A0)

3. The thermal conductivity of sand shale mixtures reachesan


uppervaluewhenthe clay particlesfill entirelythe porespaceof a

The REV is now composedof a dilutesuspension


of spherical
particlesof radiusR with a densityn. The suspension
is dilute
enoughto assumethatthetemperature
field arounda sphereis not
4. The new modelhaspotentialuseto determinein situthermal
perturbedby the presenceof other spheres.The effectivethermal

clean sand.

conductivity profiles in boreholes from the downhole


conductivityis determinedfrom:
measurements
of the naturalradioactivity(gammaray), electrical
resistivity,andbulk densityof the formations.
Q = -,,1(vr) + n(S)= -A,(VT),
The theoreticalbasisof the model standsonly for high-porosity
unconsolidated
sediments.However, someextrapolationsmadein
this paper to densersediments(with porositybelow 0.20) show a
goodagreementwith experimentaldata.
A2+ 2Al

(All)

with,
n(S)=-(10)(
3'1
('2
-Aq))(VT).
(AI2)

REVIL: THERMAL

CONDUCTIVITY

OF UNCONSOLIDATED

This leadsdirectlyto the Maxwell formula:

3(2-/1
)/

=11+(1--)
2+21'

(AI3)

SEDIMENTS

16,765

whichis formallyexactfor m = 2, and a goodapproximation


for
1.5 <_m <_2.5, whichis the rangeof cementation
exponentusually
obtained(seeFigures6 and 10). In the electricalconductivity

problem,experimental
dataandfield dataareusuallyabovethe

which is valid to first order in (1-0), that is, for very dilute

isoconductivity
point(corresponding
to thecondition
tys),
i.e., = 1. Equation
(B5)canberewritten
asa second
degree

suspensions
of spheres
immersedin the fluid. Equation(A13) is

equationin ty,whichcanbe solvedto give

similarto the first orderin (1-0)

to (11).

ty=

AppendixB: Electrical Conductivity

F+

'

(1-) 1 +

+4F

(B8)

'

of equation
(B8) << I is
Usingthe DEM approach,Senet al. [1981] give the complex Thehighsalinityasymptote
dielectricconstante* of a mixture of grainsimmersedin water:

ty= F (1+ 2(F- 1)).


=0,

(B9)

(B1)

e/ * -e S*

We test below the approximationmade aboveand leadingto


theapproximate
formula(B8). The exactformulais givenby

whereef * and es* arethecomplex


dielectric
constants
of
water and solid particles,respectively,0 is the porosity of the
mixtures,and D, called the "depolarizationfactor" (0 <_D <_1),
dependsstronglyon the shapeof the insulatingsolid particles.
Equation(B1) is valid over the entireporosityspectrumassuming
a smallcontiguitybetweenthe grains. Taking the imaginarypart
of (B 1) in the DC frequencylimit leads to the Baron's equation
[seeHelfferich, 1995, p. 489]:

10

25

Relative

m=1.5

15

10

os = 0.1S/m

G-IriS

crf - crS

=0,

0.11 10

(B2)

of(inS/m)
/

where ty is the macroscopicelectrical conductivityof the porous

material,o S is the electricalconductivityof the insulatinggrains


coatedby the electrical double layer [e.g., Revil et al., 1999a, b],

andcrf istheelectrical
conductivity
of theporefluid. Equation
(B2) is derived for a strictly discontinuoussolid phase and
predicts zero macroscopic conductivity when the electrical
conductivity of the pore fluid is zero. Equation (B2) can be
rewrittenby two equivalentforms:

,
1-CrS
/O'./
I"

cr= or./(p
m 1-O's/O-

(B3)

0.1

4-' 7 (D
= 0.20 Analytical
? 6 m=1.5 result

(B4)

4
oe
=

/////- Exact
result

as-O
,._

Typicalvaluesof m for clay free sandslie in the range1.5 to 2.1


with a mean value of 1.8 (Figure 6). Above the isoconductivity
point, > 1, Equation(B3) canbe approximated
by

o.t

10

where m -- 1/(1 - D) is called the electrical cementationexponent.


The electricalformationfactor F is definedby

Porefluidconductivity
(ofinS/m)

or=tyrO
]-m
1-/s
_m.
iI_o.f
/O.
sii_m

Exact
resulta.

0.1

-/(olo.t.) '

(B6)

II

=_

//,:
1
0.01

0.01

0.1

.....,
0.1

-1 -1
)
. ! , 11

Porefluidthermal
conductivity
(VVm'1C
'1)
Figure B1. Comparisonbetweenthe exact result from the
DEM analysisandthe analyticalapproximationdevelopedin
the main text for the electrical conductivity of (a) a brine-

(B7)

saturated granular porous medium and (b) the thermal


conductivity.

16,766

REVIL: THERMAL CONDUCTIVITY

o=

OF UNCONSOLIDATED

(B10)

F 1-/(o'/o'/)
The cementationexponent,n rangesbetween 1.5 and 2.5. Note
that(B8) and(B 10) are equalfor m = 2. The mostimportanterror
made by using (B8) instead of (B10) is obtained for m = 1.5.
Comparisonbetweenboth solutionsis done in Figure B). The
maximumerror is --,10%,so we consider(B8) to be a fairly good
approximationof (B10). The same type of analysis can be
performed for the thermal conductivity model. The exact and
approximateequationsare
m

2,=2,.1.
f 1-(2,12,].)10
'
2,=

f+

(1-) 1 +

+4f

''

whereTOis a reference
temperature
(e.g.,298.15K) andT is in
kelvins. An alternativerelationshipis given by Davis [1984]
(citedby Luo et al. [1994]):

1 )(2.142
- 0.08184-)
+0.01720.2978
2,s(r)
= 2,s(ro
'
(c2)

whereTO=298.15K andT is in kelvins.Thethermal


conductivity
of a salt electrolyteis analyzedusingthe experimentaldata from
Ozbek and Phillips [1979] (see Figure C1). In the range 20300Cthe temperaturedependence
of water and salt electrolyte
thermal conductivity is well reproducedby the following
relationship:

:v r)=

(BI2)

Again, the most importanterror is for m = 1.5. The maximum


error is -20%, so we consider (B12) to be a fairly good
approximationof (B l l) only when the cementationexponentis
not too far from m = 2 (say,n in the range 1.7-2.3).

SEDIMENTS

r0)D
+=(r-r0)+
='(r-r0)'-], (C3)

where
Cfisthesalinity,
T isinC,TOisa reference
temperature

(e.g.,20C),a = 2.38x 10-3 C- anda'=-9.97 x 10-6 C'2.


The influenceof fluid pressureuponthe thermalconductivityof
watercanbe neglected[Missenard,1965].

Acknowledgments. I thank Dominique Grauls, Larry Cathies, and

Appendix C: Influence of Temperature


The temperature
dependence
of the thermalconductivityof the
mineralgrainsfor quartzand clay mineralsis expressedby [e.g.,
Chapmanet al. 1984;Brigaud, 1989]

ChuckShearerfor fruitfuldiscussions;
Christophe
Vergniautfor helpin
gettingtheODP data;andGraemeCairnsandMaria Zamorafor reading
an earlier versionof this manuscript. I thank the two anonymous
reviewersfor very constructivecomments. This project was made
possibleby fundingfrom Elf Aquitaineand the generalsupportof the
CorporateSponsorsof the Global BasinResearchNetwork (GBRN) at
CornellUniversity
andCNRSin France.Thisworkbenefited
froma grant

2,s(T)=2,s(To)(-),
(C1)

givenby theGasResearchInstitute(GR15097-260-3787).

References
0.70

'''1''''

I''''''''1''''

a.

I'''

NaCl

Acrivos,A., and E. Chang, A model for estimatingtransportquantitiesin


two-phasematerials,Phys.Fluids,29(1), 3-4, 1986.
Batchelor, G. K., and R. W. O'Brien, Thermal and electrical conduction

througha granularmaterial, Proc. R. Sot'. London,Ser. A, 355, 313333, 1977.

'-,,,
,7l::0.65
i,,,
,,,,
,,,,
,,,i
._>

0.60

o 0.55

0.50
0

I,,,

50

100

150

200

250

300

Temperature (C)

Station, Tex., 1992.

Bernab&Y., and A. Revil, Pore-scaleheterogeneity,energydissipation,


and the transportpropertiesof rocks, Geophys.Res. Lett., 22, 1529-

>,

1532, 1995.

Blackwell, D. D., and J. L. Steele, Thermal conductivityof sedimentary


rocks: measurements and significance, in Thermal Histor3, of
SedimentaryBasins, Methods and Case Histories, editedby N. D.
Naeserand T. H. McCulloh, pp. 13-36, Springer-Verlag,New York,

_.1.00
' 0.95

1989.

Brigaud,F., Conductivit6thermiqueet champde temp6raturedansles


bassinss6dimentaires
h partir des donnfesde puits, Ph.D. thesis,276
pp., Univ. of Montpellier,Montpellier,France,1989.
Brigaud,F., and G. Vasseur,Mineralogy,porosity,and fluid controlon
thermalconductivityof sedimentary
rocks, Geophys.J., 98, 525-542,

0,85 ,,,!,,,,I,,,
0

Bebout,D. G., B. D. Weise, A. R. Gregory, and M. B. Edwards,Wilcox


sandstones
in the deep subsurfacealong the Texas Gulf Coast:Their
potentialfor productionof geopressured
energy,DOE Rep. ET28461,
Tex. Bur. of Econ. Geol., Austin, 219 pp., 1979.
Beck, A. E., An improvedmethodof computingthe thermalconductivity
of fluid-filled sedimentaryrocks,Geophysics,41, 133-144, 1976.
Behrmann, J. H., et al., Proceedings of the Ocean Drilling Program,
Initial Reports,Vol. 141, 708 pp., Ocean Drill. Program,College

,,,,I,,,

Temperature (C)

Figure C1. Thermal conductivity of aqueousNaC1 solutions


for various salinity (in the range 0-0.25 weight fraction, see
symbols) at saturatedvapor pressure(experimental data from
Ozbek and Phillips [1979]). (a) Non-normalized data. (b)

Normalized
data2,j(C?T)/2,j(Cp
20C).The solidline
correspondsto a second-orderpolynomial.

1989.

Brown, G. A., The formation porosity exponent: The key to improve


estimatesof water saturationin shaly sands,Trans. of SPWLA II, 29th
Annu. LoggingSymp., San Antonio,Texas, PaperAA, 1988.
Chapman,D. S, T. Keho, M. Bauer,and M.D. Picard,Heat flow in the
Uinta basin determinedfrom bottom hole temperature(BHT) data,
Geophysics,
49, 453-466, 1984.
Coelho,D., Three dimensionalanalysisof the temperaturefield in Block
330, South Eugene Island, Gulf of Mexico, Ph.D. thesis, 292 pp,
Cornell Univ., Ithaca, N.Y.,

1997.

REVIL: THERMAL

CONDUCTIVITY

Davis, B. W., Thermal conductivitiesof reservoir rocks, Chevron Oil


Field Res.Co. Tech. Mem., TM 84000293, 32 pp., La Habra, Calif.,
1984.

Detrick, R. S., R. P. Von Herzen, B. Parsons, D. Sandwell, and M.

Dougherty,Heat flow observations


on the Bermudarise and thermal
modelsof midplateswells,J. Geophys.Res.,91, 3701-3723, 1986.
Erickson,S. N., and R. D. Jarrard,Porosity/formationfactor relationships
for high-porositysiliciclasticsedimentsfrom AmazonFan, Geophys.
Res. Lett., 13, 2309-2312, 1998.

Ghaffari, A., Model for Predictingthermal conductivityof rock/fluid


systems,
Ph.D. thesis,135pp., Univ. of Calif., Berkeley,1980.
Gibiansky,L., and S. Torquato,Rigorousconnectionbetweenphysical
properties
of porousrocks,J. Geophys.
Res.,103,23911-23923,1998.
Gomaa,E., ThermalBehaviorof PartiallyLiquid SaturatedPorousMedia,
Ph.D. thesis,313 pp., Univ. of Calif., Berkeley,1973.
Grolier, J., A. Fernandez, M. Hucher, and J. Riss, Les Propridtds
Physiquesdes Roches.Thdorieset Modbles,451 pp., Masson,Paris,
1991.

Helfferich,F., Ion Exchange,624 pp., Dover, New York, 1995.


Henry, P., Relationshipbetween porosity, electrical conductivity, and
cationexchangecapacityin BarbadosWedge sediments,
Proc. Ocean
Drill. Program,Sci. Results,156, 137-149, 1997.
Holland, D .S., J. B. Leedy, and D. R. Lammelin, EugeneIsland Block
330 field - USA, offshoreLouisiana,in StructuralTraps,III, Tectonic
Fold and Fault Traps, Am. Assoc. of Petr. Geol., Treatise of
PetroleumGeologyAtlas of Oil and Gas Fields, compiledby E.A.
Beaumont,andN.H. Foster,pp. 103-143,Tulsa,Okla., 1990.
Horai, K., Thermal conductivityof rock-formingminerals, J. Geophys.
Res., 76, 1278-1308, 1971.

Jeffrey,D. J., Conductionthrougha randomsuspension


of spheres,Proc.
R. Soc. London, Set. A, 335, 355-367, 1973.

Keese, J., Thermal conductivityof unconsolidatedoil sands,M.S. thesis,

72 pp.,Univ. of Calif., Berkeley,1973.


Koltermann, C. E., and S. M. Gorelick, Fractional packing model for
hydraulicconductivityderivedfrom sedimentmixtures,WaterResour.
Res., 31, 3283-3297, 1995.

OF UNCONSOLIDATED

SEDIMENTS

16,767

Poelchau,H. S., D. R. Baker, T. Hantschel,B. Horsfield, and B. Wygrala,


Basin simulationand the design of the conceptualbasin model, in
Petroleumand Basin Evolution, edited by D. H.Welte, B. Horsfield,
andD. R. Baker,Springer-Verlag,Berlin,pp. 5-70, New York, 1997.

Prat,M., Modellingof heattransferby conductionin a transitionregion


betweena porousmediumand an externalfluid, Transp. in Porous
Media, 5, 71-95, 1990.

Presley, M. A., and P. R. Christensen, Thermal conductivity


measurements
of particulatematerials, 1, A review, J. Geophys.Res.,
102, 6535-6549, 1997a.

Presley, M. A., and P. R. Christensen, Thermal conductivity


measurements
of particulatematerials,2, Results,J. Geophys.Res.,
102,6551-6566,

1997b.

Pribnow, D., C. F. Williams, and H. Burkhardt, Well log-derived


estimatesof thermalconductivityin crystallinerockspenetratedby the
4-km deep KTB vorbohrung, Geophys. Res. Lett., 20, 1155-1158,
1993.

Quintard,M., and S. Whitaker (Eds.), ICHMT Seminar on "Heat and


Mass Transferin PorousMedia", Dubrovnik,Hemishere,Washington,
D.C., 1991.

Revil, A., Ionic diffusivity, electrical conductivity, membrane and


thermoelectricpotentialsin colloids and granularporousmedia: A
unifiedmodel,J. Colloidlnte.rfaceSci., 212, 503-522, 1999a.
Revil, A., Pervasivepressuresolution transfer: A poro-visco-plastic
model,Geophys.
Res.Lett.,26, 255-258, 1999b.
Revil, A., and L. M. Cathies,Permeabilityof shalysands,Water Resour.
Res., 35, 651-662,1999.

Revil, A., and P. W. J. Glover, Nature of surfaceelectricalconductivityin


naturalsands,sandstones,
and clays, Geophys.Res.Lett., 25, 691-694,
1998.

Revil, A., and P. A. Pezard, Streamingelectrical potential anomaly along


faultsin geothermalareas,Geophys.Res.Lett., 25, 3197-3200, 1998b.
Revil, A., L. M. Cathies,S. Losh, and J. A. Nunn, Electricalconductivity
in shaly sandswith geophysicalapplications,J. Geophys.Res., 103,
23,925-23,936, 1998a.
Revil, A., L. M. Cathies, J. D. Shosa, P. A. Pezard, and F. D. de

Krupiczka,R., Analysisof thermalconductivityin granularmaterials,Int.


Larouziere, Capillary sealing in sedimentarybasins: a clear field
Chem.Eng., 7(1), 122-144, 1967.
example,Geophys.Res.Lett., 25, 389-392, 1998b.
Lauer-Leredde,C., P. A. Pezard, F. Touron, and I. Dekeyser, Forward Revil, A., P. A. Pezard,and P. W. J. Glover,Streamingpotentialin porous
modellingof the physicalpropertiesof oceanicsediments:Constraints
media, 1, Theory of the zeta-potential,J. Geophys.Res., 104, 20,021from core and logs, with paleoclimatic implications, in Core-Log
20,031, 1999a.
Integration,edited by P. K. Harvey and M. A. Lovell, Spec. Publ. Revil, A., H. Schwaeger,L. M. Cathies,and P. D. Manhardt,Streaming
Geol. Soc., 136, 115-127, 1998.

Lovell, M. A., Thermal conductivity and permeability assessmentby


electrical resistivity measurements in marine sediments, Mar.
Geotechnol., 6, 205-240, 1984.
Luo, M., J. R. Wood, and L. M. Cathies, Prediction of thermal

conductivityin reservoirrocksusingfabrictheory,J. Appl. Geophys.,


32, 321-334, 1994.

Marion, D., A. Nur, H. Yin, and D. Han, Compressionalvelocity and


porosityin sand-claymixtures,Geophysics,
57, 554-563, 1992.
Matlack, K. S., D. W. Houseknech,and K. R. Applin, Emplacementof
clayinto sandby infiltration,J. Sediment.Petrol., 59, 77-87, 1988.
Maxwell, J. C., Treatiseon Electricityand Magnetism,vol. 1,440 pp., 3rd
ed., Clarendon,Oxford, England, 1892.
Mendelson,K. S., and M. H. Cohen,The effectof grainanisotropyon the

electricalproperties
of sedimentary
rocks,Geophysics,
47, 257-263,
!982.

Missenard,A., Conductivitd
ThermiquedesSolides,Liquides,Gaz, et de
leursMdlanges,Eyrolle, Paris, 1965.

Murphy,H. D., and R. G. Lawton,Downholemeasurements


of thermal
conductivityin geothermalreservoirs,J. PressureVesselTechnol.,99,
607-611, 1977.

Ortoleva, P. J., Basin compartmentsand seals,AAPG Mere. 61, 3-26,


1994.

Ozbek, H., Thermal conductivityof multi-fluid saturatedporousmedia,


Ph.D. thesis,106 pp., Univ. of Calif., Berkeley,1976.
Ozbek, H., and S. Phillips, Thermal conductivity of aqueous NaCI
solutions from 20C to 330C, report prepared for U.S. Dep. of
Energy,46 pp., LawrenceBerkeleyLab., 1979.
Pape,H., L. Riepe,andJ. R. Schopper,Interlayerconductivityof rocks.A
fractal model of interface irregularities for calculating interlayer
conductivityof naturalporousmineral systems,ColloidsSurf., 27, 97122, 1987.

Plumb, O. A., and S. Whitaker, Diffusion, adsorption,and dispersionin


heterogeneous
porousmedia:the methodof large-scaleaveraging,in
Dynamicsof Fluids in Hierarchical Porous Media, edited by J. H.
Cushman,pp. 149-176, Academic,San Diego, Calif., 1990.

potentialin porousmedia,2, Theoryand applicationto geothermal


systems,
J. Geophys.Res.,104, 20,033-20,048,1999b.
Rieke, H. H., and G. V. Chilingarian, Compaction of Argillaceous
Sediments,Dev. Sedimentol.,vol. 16, 424 pp., Elsevier Sci., New
York, ! 974.

Sass,J. H., D. D. Blackwell,D. S. Chapman,J. K. Costain,E. R. Decker,


L. A. Lawer, and C. A. Swanberg,Heat flow from the crustof United
States,in PhysicalPropertieso.[RocksandMinerals,vol. II-2, edited
by Y. S. TouloukianandC. Y. Ho, pp. 503-548,McGraw-Hill,New
York, 1981.

Sen, P. N., C. Scala,and M. H. Cohen,Self similar model for sedimentary


rockswith applicationto the dielectricconstantof fusedglassbeads,
Geophysics,
46, 781-795, 1981.
Sen, P. N., C. Straley,W. E. Kenyon, and M. S. Whittingham,Surface-tovolume ratio, charge density, nuclear magnetic relaxation, and
permeabilityin clay-bearingsandstones,
Geophysics,
55, 61-69, 1990.
Sheng,P., Consistentmodelingof the electricalandelasticpropertiesof
sedimentary
rocks,Geophysics,
56, 1236-1243, 1991.

Shipboard
Scientifc
Party,Site891,Proc.OceanDrill. Progrant,Initial
Reports,146, 241-300, 1994.
Silliman, S. E., and C. E. Neuzil, Borehole determination of formation

thermal conductivity using a thermal pulse from injected fluid, J.


Geophys.Res., 95, 8697-8704, 1990.
Somerton,W. H., ThermalPropertiesand Temperature-Related
Behavior
of Rock/FluidSystems,257 pp., ElsevierSci., New York, 1992.
Spiegler,K. S., R. L. Yoest, and M. R. J. Wyllie, Electrical potential
acrossporousplugsand membranes,Discuss.Faraday Soc., 21, 174185, 1956.

Stroud, D., G. W. Milton, and B. R. De, Analytical model for the


dielectricresponseof brine-saturatedrocks, Phys. Rev. B, 34, 51455153, 1986.

Takei, Y., Constitutivemechanicalrelationsof solid-liquidcomposites


in
terms of grain-boundarycontiguity, J. Geophys.Res., 104, !8,18318,203, 1998.

Vacquier,V., Y. Mathieu,E. Legendre,and E. Blondin,Experimenton

16,768

REVIL: THERMAL CONDUCTIVITY OF UNCONSOLIDATED SEDIMENTS

estimatingthermalconductivityof sedimentaryrocksfrom oil well


logging,AAPG Bull., 72, 758-764, 1988.

Van de Hulst,H. C., Light-Scattering


by SmallParticles,JohnWiley,
New York, 1957.

Watts,N. L., Theoretical


aspects
of caprockandfaultsealsfor single-and
two-phasehydrocarbon
columns,Mar. Pet. Geol.,4, 274-307, 1987.

Woodside,W., and J. H. Messnet,Thermal conductivityof porousmedia.


I. Unconsolidatedsands,II, Consolidatedrocks, J. Appl. Phys., 32,
1688-1706, 1961.

Yamano, M., and S. Uyeda, Heat-flow studies in the Peru trench


subductionzone, Proc. Ocean Drill. Program, Sci. Results,112, 653661, 1990.

Waxman, M. H., and L. J. M. Smits, Electrical conductivities in oil-

Zamora,M., D. Vo-Thanh, G. Bienfait, and J.-P. Poirier,An empirical


bearingshalysands,Soc.Pet. Eng.J. , 8, 107-122,1968.
relationshipbetweenthermalconductivityand elasticwave velocities
Weast,R. C., Handbooko.['Chemistryand Physics,Chem.RubberCo.,
in sandstone,
Geophys.Res.Lett., 20, 1679-1682, 1993.
Ohio, 1970.
Zimmerman, R. W., Thermal conductivityof fluid-saturatedrocks, J.
Whitaker,S., FundamentalPrincipleso.f Heat Trans.
fer, Krieger,
Petrol. Sci. Eng., 3, 219-227, 1989.
Melbourne, Fla., 1983.

Whitaker,S., Localthermalequilibrium:an application


to packedbed
catalyticreactordesign,Chem.Eng.Sci.,41, 2029-2039,1986.
Whitaker,S., Heattransferin catalyticpackedbedreactors,in Handbook
A. Revil, CNRS-CEREGE, Departmentof Geophysics,BP-80, Fo.['Heat and Mass Trans.
l'kr, vol. 3, editedby N. P. Cheremisinoff, 13545Aix-en-Provence,
Cedex4, France.(revil@cerege.fr)
Chap. 10, Gulf, Houston,Tex., 1989.

Williams,C. F., andR. N. Anderson,


Thermophysicai
properties
of the
Earth's
crust:
In situmeasurements
fromcontinental
andocean
drilling, (ReceivedApril 19, 1999; revisedJanuary13, 2000;
J. Geophys.Res.,95, 9209-9236, 1990.

acceptedFebruary7, 2000.)

Das könnte Ihnen auch gefallen