Oct 10, 1998 - Thanks also to Dr. Martin Olsen, John Walsh,. Matt Jospin, and Dr. .... Williams⢠L.A., G.A.P.arks, and D.A. Crerar, Silica diagenesis, 1,. Solubility ...
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 103,NO. B10,PAGES23,951-23,967, OCTOBER 10, 1998
Precipitation sealingand diagenesis 1. Experimental results Eric Tenthorey,ChristopherH. Scholz,andEinat Aharonov Lamont-Doherty EarthObservatory of ColumbiaUniversity, Palisades, New York
Albert Ltger Departmentof EarthandPlanetaryScience,AmericanMuseumof NaturalHistory,New York
Abstract. Duringburialanddiagenesis of granularaggregates, significant permeability reduction maybe inducedby theformationof low-temperature, authigenicminerals.To quantitatively assess theimportanceof thisprocess,we haveconducteda seriesof hydrothermal flow-through experiments usingdeionizedwaterandlabradorite/quartz sand.All experiments wereconducted in a modifiedtriaxialapparatus, configured to allowcontinuous permeabilitymeasurements. Undermostof the conditionstested,significantpermeabilityreductionis observedwith no concurrent decrease in porosity.The overallpermeabilityreductionsometimes exceedsI orderof magnitude over4 daysandis positivelycorrelated to temperature anddeviatoricstress.Scanning electronmicroscope observations togetherwithdatafrom additionalexperiments showthatthe observedpermeabilityreductionis entirelya resultof secondary mineralgrowth. Si andA1 concentrations in the postexperiment fluidsarealsocorrelatedto temperature andstress, confirmingthelink betweenthechemicalstateof the systemandpermeabilitybehavior.In all experiments, permeabilityreductionis fastestearlyandlevelsoff in the latestages.To explainthe permeabilitybehaviorasa functionof time,a conceptual modelis developedin which precipitation of authigenicmineralsis rapidat earlytimeswhile dissolution of quartzand labradoriteis mostactive.As the systemapproaches equilibrium,the components necessary for secondarymineralformationare liberatedat a lower rate,therebycausingprecipitationto slow. Althoughauthigenicmineralformationdoesnotreducetotalporespacein theseexperiments, thereis a reductionin effectiveporosity,whichresultsin pccrneability reduction. 1. Introduction
The term "diagenesis" refers to the physicaland chemical changesimpartedupon sedimentssubjectedto pressures and temperatureslower than those of the metamorphicrealm. Physicalaspects of diagenesis suchascompaction havebeenwell studied,andthe mannerin whichthey affectvarioustransport properties such as permeability are largely understood. Conversely, the relationship between subsurfacechemical reaction,induration, andpermeability changeremainsunclear.
This chemically induced permeability reduction or "precipitationsealing"may drasticallyalter fluid flow in the subsurface.The formationof low-permeabilitylithologicunits can hinder the migration of hydrocarbonsand other aqueous fluids, potentially preventingor enhancingthe formation of
economically valuabledeposits or slowingthe transport of variousenvironmental pollutants.Froma rheological pointof view, largepermeability reductions may indirectlyresultin significant weakening of a rockby allowingthegeneration and maintenance of fluid pressures in excess of hydrostatic. Such Chemical reactions in the subsurface are manifestations of a fluid"overpressures" formwhenintergranular porefluidscannot system's disequilibrium.In natural systemsthe numerous be expelled during compaction or other volume-altering physical and chemical factors involved render difficult the
processes[Bredehoeftand Hanshaw, 1968; Barker, 1972; formulation of a modeldescribing permeability as a functionof Magara, 1975;Bethke, 1986; Hunt, 1990], becauseof the low chemicalreaction. However, in general,processessuchas permeability of therock.As porepressure increases, theeffective dissolution causepermeabilityandporosityto increase,while stress ontherockdecreases, therebyresulting in a weakening of precipitation of claysor otherlow-temperature mineralsusually the material. hastheopposite effect.Thefocusof thispaperis onthecoupled
effect,wheredissolution andprecipitation in a granular aggregate causecloggingof the pore network and resultin an associated
Precipitationsealingis generallyinducedby changing
temperature conditions or by primarymineraldissolution, which
results in thesupersaturation of variouslow-temperature mineral phases. A numberof experimentaland field studieshave minerals maybesointense l•hat permeability reduces tothepoint characterized theexpected equilibrium mineralassemblages and permeabilitydrop.In somecases,precipitation of secondary that hydraulicsealsform [seeHunt, 1990].
changes to fluidchemistry underdiagenetic conditions [e.g., Bolesand Franks, 1979; Moody et al., 1985; Thorntonand
Papernumber9RJB02229.
Seyfried,1985;Vavra,1989;HajashandBloom,1991;Huang andLongo,1994].In somesediments, hydraulic sealing mayalso be partly due to masstransferresultingfrom stress-induced
0148-0227/98/98J'B-02229509.00
solubility gradients.There has been an abundanceof work
Copyright 1998bytheAmerican Geophysical Union
23,951
23,952
TENTHOREY ET AL.: PRECIPITATION SEALING, 1
hypethezz:;,other experimentswere condt•ctedin which temperature, fluidcomposition, or flow ratewerechangedduring andNur, 1977;Robin,1978;Gratier and Guiguet,1986;Tadaet theexperiment. The sandusedin all but two experiments was90 al., 1987; Hickman and Evans, 1991]. Generally, pressure wt % labradoritefeldsparand 10 wt % quartz,both with grain solutionis viewedasthe dissolutionof materialat highlystressed sizesof 210-500gm. Sandswerepreparedby placingfragments extracting graincontacts followedby precipitation of thatsamematerial of quartzor labradoritein a chipperandsubsequently immediatelyadjacentto the contact.This relocationof mass the210-500gm fractionwith sieves.Bothquartzandlabradorite combinedwith the ensuingdeformation•s believedto reduce were then run through a magneticseparatorto remove most permeability in granular media[SpruntandNur, 1977;Angevine foreignphases.Beforeeachexperimentthesandwassoakedwith and Turcotte, 1983; Walder and Nur, 1984; Gavrilenko and deionizedwaterandplacedin an ultrasonicbathto removemost fine, adheredparticles. In one experiment, 100% quartz was Gudguen,1993;LerndeandGudguen,1996]. used, and in another, 100% labradorite was used to determine the In recentyears,manytheoreticalmodelshaveattemptedto quantify the relationshipbetweenchemicalreactionand effect of sand compositionon permeability evolution. The fluid usedin all experiments wasdeionizedwater. permeabilityin porousmedia[Lichtner,1985;Dewersand staxting Ortoleva,1994;$teefeland Lasaga,1994;Boltonet al., 1996; The startingsandwasjacketedwith coppertubingthat had Aharonov et al., 1997]. However, from the experimental beenNi platedto preventanycorrosionof thejacket(Figure1). standpoint,very little work has been done. Most of the Surroundingthe jacketed samplewas a coil heater. Porous experimentalstudiespertainingto the chemicalaspectsof stainlesssteelfilters were placedat both endsof the sampleto diagenesishavenot beenequippedto measurepermeability disperse theflow of waterthroughthesampleandto preventloss changes duringreaction.The few existingexperimental studies of materialintothe porefluid system.Two thermocouples, one [Smallet al., 1992;Main et al., 1994; Scholzet aL, 1995]have on the outsideof thejacket andthe otherwithin the pore fluid shownthatnucleationandgrowthof authigenicphasescancause systemimmediatelyabovethe upperfilter, wereusedto record drasticpermeability reduction.In fact,Scholzet al. [1995]report temperature andprovidefeedback for thetemperature controllers. permeability reduction with very little concurrent decrease in Theporefluidthermocouple wasusedto controltemperature. porosity.Theseexperimental resultsare confirmedby field To ensurethat no significantmineral precipitationoccurred studies, which have shown that thin oxide/clay coatings on outsidethe sample,postexperiment filtersfrom two experiments fracturesurfaces cancausepermeabilityto decrease by an order wereexaminedwith a scanningelectronmicroscope (SEM). Two of magnitude [FullerandSharp,1992;Fu et al., 1994]. typesof mountswere prepared:perpendicular and parallel to In thispaperanditscompanion [Aharonov et al., thisissue]a flow direction.In neithercasewasany indicationof precipitation joint experimental/theoretical approachis taken to better understandthe relationshipbetweenchemicalreactionand
pertainingto thisprocess,whichis generallyreferredto as pressure solution [e.g.,Rutter,1976;De Boeret al., 1977;$prunt
permeability change.Herewe presentresultsfroma seriesof experiments in whicha quartz/feldspar aggregate equilibrates
'•---Pore Fluid Thermo
with deionized water at elevated temperaturesand stresses.
Feldspardissolutionand accompanyingsecondarymineral growthresultin significantpermeabilityreduction. During these experimentstherewaslittle or no mechanicalcompaction.Hence the permeabilitychangesdid not result from net porosityloss; only the natureand shapeof porositywere alteredas a resultof precipitation.Basedon the chemicalandpermeabilitydatafrom the experiments,a conceptualmodel of fluid-rock equilibrium and kineticswasdevelopedand tested. In the companionpaper, Aharonov et al. [this issue] describe a theoretical model of
BrassFitting
Ni-Plated
Cu Tubing Copper
Heating C
permeabilityevolutionasa functionof chemicalequilibriumstate andattemptto fit thoseresultsto theseexperimentaldata.
Pore Fluid Conduit
SteelFilter Quartz-Feldspar Sand
External
2. Experimental Method All reportedexperiments wereconductedin a triaxialpressure apparatusat the Rock MechanicsLaboratory of the LamontDoherty Earth Observatory. The experimentsconsistedof subjecting a fluid-saturatedsandto different pressureand temperatureconditionsandmeasuringpermeabilityevolutionof thesand over the courseof approximately4 days. All experimentswere run at a confiningpressure(Pc) of 100 MPa and a porepressure
(Pp) of 50 MPa. Eachexperiment wasrun at a constant
Thermocouple ;tainlessSteelSpacer
A1203Spacer
O-Ring
temperatureanddeviatoricstress(Od) varyingfrom 25ø to 275øC and25 to 90 MPa, respectively.Two seriesof experimentswere Figure 1. Samplecolumnassemblyusedin all experiments. conducted,one with varyingtemperature(T) and the otherwith Sandis contained withincentralportionof Ni-platedcoppertube. varying Od (see Table l), to isolatethe effect of T and Od on Stainless steelspacersnearsampleallow dispersal of heat,while permeability evolution.Pc,Pp, andOdwereall maintainedaluminaspacersat endsof tubingpreventlossof heatto bodyof constantduringeachexperimentby servo-control.The Od values triaxialpress.Axial holesdrilled throughthe centersof spacers were measuredby using an external load cell. To test several allow fluid flow thoughsample.
TENTHOREYET AL.:PRECIPITATION SEALING,1
23,953
Sampl• • Pressure
Colum
Vessel
Pore Pressure
Transduce
Vacuum
Pump Main
Ram Pore Pressure
Pore Fluid/ Sampling Valve
Intensifier Bottom
Argon Tankrce• [ (Confining Pressure) Water
Sou
nand PumpS[ [
Top
3
Differential
Pore
Pressure Transducer
Figure 2. Simplifiedillustrationof porefluid system.Two porefluid intensifiers circulatewaterbackandforth throughsampleundera knownpressurehead.Flow rateallowsa continuous calculationof permeabilityusing Darcy'slaw.
evident.Thetemperature gradients encountered by thefluidasit passesfrom hot sampleto cold tubingoutsidethe vesselalso pose some question as to mineral precipitation outside the sample.It is likely thatsomeprecipitation occurred in thepore
through20-23 g of sampleat constantpore pressuredifferential. The volume of water circulatedthroughthe samplewas large enoughto allow mixing with fluid in eachintensifier. Once flow hadbeeninitiated,the porepressuredifferenceacrossthe sample Typically, fluidlinesasa resultof supersaturation of thefluidwithdropping was measuredfrom the two porepressuretransducers. temperature. However,the fact thatpostexperiment fluidscould the measuredpore pressuredifferencesacrossthe samplewere be storedfor up to a monthand still yield near-equilibrium between0.15 and0.35 MPa, smallpressuredeviationsin comparconcentrations (seebelow)suggests that mineralprecipitation isonwith the ambientporepressureof 50 MPa. Sincethe crossoutsidethe samplewas very limited as a result of sluggish sectionalarea of the sample(A), the lengthof the sample(AL), kineticsat roomtemperature. the porepressuredifferenceacrossthe sample(AP), the viscosity In all experiments the sample was first subjected to of water (g) and the flow rate (Q) were all known parameters, precompaction, afterwhichtheheaterwasretightened. Thisstep absolutepermeabilitycouldbe calculatedat any point duringan was necessarybecauseapplicationof Pc resultedin significant experimentby usinga modifiedform of Darcy'slaw:
radial compactionand decouplingof the sampleand heater. K= tt ALO A AP Oncethe samplewasreinserted intothevessel,it wassubjected to an axialcompression of approximately 1 MPa to affectO-ring In mostexperiments theinitial"cycletime"(timerequired for a seals. This was followedby saturationof the sampleand slugof fluid to flowbackandforththroughthe sample)was evacuationof any air from the porefluid system. Pc was then approximately 60 s with subsequent increasesin cycle time raised to 100MPawithPpalways maintained athalfof Pc so corresponding to permeabilityreductionof the sample. In thataneffectivestress of 50 MPawasneverexceeded.ThenOd severalexperiments,testswere conductedwith different values wasappliedat a rateof 1 MPa/sandtime-dependent compaction of AP,whichshowed thatAPvariedlinearlywithflowrate(Q). was allowed to stabilize(approximately45 min). Fluid flow This result indicated that flow was laminar and that the use of throughthe specimenand temperaturewere then applied Darcy'slaw wasjustified. simultaneously. Raisingthetemperature resultedin an additional phaseof compaction, whichwill be discussed in greaterdetail 3. Analytical Methods below. Oncethisfinal compaction wascomplete, permeability measurementsbegan and continueduntil the conclusionof each The ARL scanning electronmicroprobe quantometer at the experiment. AmericanMuseumof NaturalHistory(AMNH) wasusedto The flow of waterthroughthe sandwasdrivenby two pore determine thecomposition ofthestarting materials. Thefeldspar fluid intensifiers(Figure2), bothconnected to porepressure used is labradoriteand has a measuredcompositionof transducers. The total volume of water in the system was
An49.5Ab48.3Or2.2 . Eventhough thelabradorite wasplaced
approxiinately 35cm3,6 cm3 ofwhich wasdriven back andtbrth throughthe magneticseparator, someforeignmineralphases,
23,954
TENTHOREY ETAL.:PRECIPITATION SEALING, 1
totalinglessthan2%, wereobserved in thestarting material.The
Odto 17.7%at the higheststresses.Uponthe application of
natureof thesephaseswill be describedin a later section. Pro-
temperature andflow, all experiments exhibitfurtherstrainof 47%, with the exceptionof the experimentconducted at 25øC.
experiment andpostexperiment sands wereexamined by usinga ZeissDSM 950 SEM, alsoat AMNH. Two typesof mountswere
Although temperature isresponsible forthisstrain, thereispoor
preparedfor examination:loosegrain mountsof the sandand
correlationbetweentemperature or stressand the amountof
polished sections. Loosegrainmounts allowpreservation of any fragilemineralgrowthandtherefore providetexturalinformation. Polishedsectionswere used for quantitativeanalysisand characterization of two-dimensional (2-D)spatial relationships. Followingeachexperiment, the fluid fromeachporefluid intensifier wasextracted andrefrigerated for subsequent chemical analysis.Chemicalanalyses on thefluidsweremadeby using
strain. Scholzet al. [1995]calledthisphaseof compaction
the direct currentplasma(DCP) spectrometer at the LamontDohertyEarthObservatory.Majorelements analyzedwereNa, K, Ca, Si, andAl. Fe andNi werealsoanalyzed to ensurethatno
significant dissolution of theNi platingor stainless steelpore pressuretubinghad occurred.In later analyses,K, Fe, and Ni were not measured, since earlier measurements indicated concentrations below detection limits. Solutions for chemical
"hydrothermal consolidation" andspeculated thatit wasdueto increased dissolution at highlystressed graincontacts followed bygrainboundary sliding.Furthertestsusingsands of different compositionand shape,however,now suggestthat this
phenomenon is notchemically induced butis likelycaused by temperature-induced annealingof the strain-hardened jacket, perhaps accompanied by somethermomechanical rearrangement of packing.However, thenatureof thishydrothermal consolidationis notimportant in theframework of thisstudy,sinceall permeability measurements aremadeuponthecompletion of this strain.Axialstrainduringtheremainder of theexperiments is characterized by creep,whichamountsto no morethanabout
1.5%strain after4 days.However, muchorallofthisstrain may
analysiswereprepared by diluting5 mL aliquotsof sampled be dueto bulgingof the sampleandmaynotrepresent real solutionwith45 mL of 2 N traceelementgradenitricacid. This compactionof the sand. procedure wasdoneprimarilyto ensure thatall dissolved species Although thecirculation ofwaterisinitiated concurrently with wouldrems. i:: in solutionandalsoto preservesampledfluid. theapplication of temperature, permeability measurements donot Furthermore, dilution of the primary fluids resulted in beginuntilallT-induced strainhasstabilized andthesystem has concentrations well withinthelineardynamicrangeof theDCP. attained thermalequilibrium.Thisstepensures thatmechanical Elementalconcentrations weredetermined by usingthreefluid compaction orchanges oftemperature arenotresponsible forany standards,which were preparedby using SPEX standard of theobserved permeability changes. Tobeconsistent, permesolutions.
abilitymeasurements inalltheexperiments commence 1.3x 104 s afterflowhasbeeninitiated (Figure3). Fourdayswaschosen
4. Observations
asthedurationof mostof theexperiments because it wasfound
4.1 BasicObservationsof PermeabilityReduction
In a typical experiment,three separatephasesof axial compactionare observedin the early stages(Figure3). The applicationof confiningpressureresultsin 5-10% axial strain. Axial strainfromthe application of 6d variesfrom3.2% at low
25
thatpermeability changes followeda near-exponential form(see below)witha timeconstant thatcaused permeability toflattenoff in lessthan4 days. The resultsof the permeabilitymeasurements are shownin
Figures 4a and4b. In thefourhighest-temperature experiments, significant reductions in permeability areobserved (Figure4a).
-
*'•,,,Permeability Measurements 20
Begin Approximately at this Point
-
Raising of Temperature and
Circulation
g 10-5
•
of Fluid
Initiated
Deviatoric Stress Applied
d •..........• Confining Pressure Applied
0
0
I
I
I
10000
20000
30000
I
40000
I
50000
Time (seconds) Figure3. Compaction curve fromanexperiment conducted atT=275øC anddeviatoric stress of75MPa.Early portion ofcurveischaracterized bymechanical compaction caused bytheapplication ofconfining pressure and deviatoric stress. Further strain isobserved oncetemperature andcirculation ofwaterisinitiated, probably aresult oftemperature-induced re-annealing ofthecopper jacket.Once thisT-induced strain hasstabilized, permeability measurements begin,withlessthan1.5%strain beingobserved duringthefinal4 days.
TENTHOREY ET AL.: PRECIPITATION SEALING, I
23,955
A)
10-2 •:
T=25øC
6
4
10-3•.
125øC
6
•6
"X X I
I
0
50
4X-m-t•l-•• X +.4-
I
100
I
4-
+++
I
150
200
I
250
o 175 C(II) I
I
300 350x103
Time (sec)
B)
••
ß
Diff. Stress= 2i MPa 3•a
• •
-
•
-
•
-
•
Quartz-60 MPa
104_
""
...
....
75MPa(I)
'fi•_•,•OO0
