24. anelastic strain recovery and elastic properties … · this chapter describes the first...

Gradstein, F. M., Ludden, J. N., et al., 1992Proceedings of the Ocean Drilling Program, Scientific Results, Vol. 123

24. ANELASTIC STRAIN RECOVERY AND ELASTIC PROPERTIES OF OCEANIC BASALTICROCKS1

N. R. Brereton,2 P. N. Chroston,3 C. J. Evans,2 J. A. Hudson,4 and R. B. Whitmarsh5

ABSTRACT

A knowledge of rock stress is fundamental for improving our understanding of oceanic crustal mechanisms andlithospheric dynamic processes. However, direct measurements of stress in the deep oceans, and in particular stressmagnitudes, have proved to be technically difficult.

Anelastic strain recovery measurements were conducted on 15 basalt core samples from Sites 765 and 766 during Leg123. Three sets of experiments were performed: anelastic strain recovery monitoring, dynamic elastic property measure-ments, and thermal azimuthal anisotropy observations. In addition, a range of other tests and observations were recordedto characterize each of the samples.

One common feature of the experimental results and observations is that apparently no consistent orientation trendexists, either between the different measurements on each core sample or between the same sets of measurements on thevarious core samples. However, some evidence of correspondence between velocity anisotropy and anelastic strainrecovery exists, but this is not consistent for all the core samples investigated.

Thermal azimuthal anisotropy observations, although showing no conclusive correlations with the other results, wereof significant interest in that they clearly exhibited anisotropic behavior. The apparent reproducibility of this behaviormay point toward the possibility of rocks that retain a "memory" of their stress history, which could be exploited to derivestress orientations from archived core.

Anelastic strain recovery is a relatively new technique. Because use of the method has extended to a wider range ofrock types, the literature has begun to include examples of rocks that contracted with time. Strong circumstantial evidenceexists to suggest that core-sample contractions result from the slow diffusion of pore fluids from a preexisting microcrackstructure that permits the rock to deflate at a greater rate than the expansion caused by anelastic strain recovery. Bothexpansions and contractions of the Leg 123 cores were observed.

The basalt cores have clearly been intersected by an abundance of preexisting fractures, some of which pass rightthrough the samples, but many are intercepted or terminate within the rock matrix. Thus, the behavior of the core sampleswill be influenced not only by the properties of the rock matrix between the fractures, but also by how these macro- andmicro-scale fractures mutually interact. The strain-recovery curves recorded during Leg 123 for each of the 15 basalt coresamples may reflect the result of two competing time dependent processes: anelastic strain recovery and pore pressurerecovery. Were these the only two processes to influence the gauge responses, then one might expect that given theadditional information required, established theoretical models might be used to determine consistent stress orientationsand reliable stress magnitudes. However, superimposed upon these competing processes is their respective interactionwith the preexisting fractures that intersect each core. Evidence from our experiments and observations suggests thatthese fractures have a dominating influence on the characteristics of the recovery curves and that their effects are complex.

INTRODUCTION

The lithosphere is subject to a variety of forces, both internaland external, that cause it to be stressed. These stresses determinethe lithospheric dynamic response, for example faulting and earth-quakes, but at the same time reflect the motions of the lithosphericplates relative to the asthenosphere. In a recent review, Zoback etal. (1989) summarized the principal sources of tectonic stress asridge push, slab pull, trench suction, and shear traction (drivingor braking) on the underside of the plates. Less significant, andsometimes more localized, sources of stress include the resistancecaused by continent-continent collision, gravitational loads ofvolcanic and sedimentary origin, flexure in the vicinity of subduc-tion zones, fhermoelasticity (cooling), and membrane stressesassociated mainly with latitudinal plate motions.

1 Gradstein, F. M., Ludden, J. N., et al., 1992. Proc. ODP, Sci. Results, 123:College Station, TX (Ocean Drilling Program).

British Geological Survey, Keyworth, Nottingham, U.K.School of Environmental Sciences, University of East Anglia, Norwich, U.K.Department of Mineral Resources Engineering, Imperial College of Science

and Technology, London, U.K.Institute of Oceanographic Sciences, Deacon Laboratory, Wormley, Godalm-

ing, U.K.

Stress is transmitted laterally throughout the relatively rigidplates, whether they lie under oceans or continents. Therefore, instudies of global stress patterns, land-based measurements cansometimes be used to predict stress in oceanic regions. There aremany situations, however, where this approach is inappropriate,such as when determining the stress near accreting plate bounda-ries; oceanward of trenches; at multiple ridge, trench, or transformintersections; in the vicinity of seamount chains; and in regionsof anomalous oceanic intraplate tectonics. In these circumstances,observations of the oceanic lithosphere itself are required tounderstand and predict tectonic processes.

Virtually the only source of information for the oceanic crusthas been from earthquake fault plane solutions, which tend notonly to be deep-seated and relatively infrequent, but also to beconcentrated in the vicinity of the tectonically active areas of plateboundaries. There is, therefore, still a great paucity of stressinformation from the tectonically stable regions of oceanic plates.

Until recently, the experimental determination of in-situ rockstresses has been difficult. A range of rock-stress measurementtechniques are available, but most have been developed by themining industry for use in relatively shallow boreholes or inexposed rock faces and thus are inappropriate in the oceanicdomain. These techniques differ considerably in their scale ofmeasurement (i.e., the volume of rock and type of discontinuitiesover which the measurements are performed) and in the assump-

469

N. R. BRERETON ET AL.

tions behind their application. The most suitable for use in deep-ocean boreholes are hydraulic fracturing and borehole breakouts(Zoback et al., 1985; Newmark et al., 1985; Brereton and Evans,1989), while the most suitable for use on core samples are differ-ential strain analysis (Dyke, 1989) and anelastic strain recovery.The effects of scale were comprehensively reviewed by Hyett(1990).

The recently developed technique of anelastic strain recoverymay potentially determine directions and magnitudes of principalstresses by measuring changes with time in the geometry of therecovered core. The advantages of this technique are (1) that itdoes not consume extra drill-ship time and (2) that the measure-ments are relatively simple. However, a study of stress shouldinclude how the results of experimental measurements can beinfluenced by intrinsic elastic properties and rock characteristicsas well as how interpretation may be improved by a knowledgeof these properties.

This chapter describes the first attempt to use the anelasticstrain recovery technique on cores from the oceanic crust. Sitedetails and summaries of the lithologies encountered during Leg123 can be found in the Initial Reports volume for Leg 123 andwill not be repeated here. Eleven whole-round basalt core sampleswere collected from Site 765 and another four from Site 766.

ANELASTIC STRAIN RECOVERY

Background Stress

Stress is a quantity that cannot be measured directly, but mayonly be derived by way of its relation with strain or some otherindicator. The major group of stress indicators relies on corre-lation with strain relief mechanisms and associated rock deforma-tion. The release of in-situ stress can generate microcracksthrough the differential expansion of adjacent grains (Nur andSimmons, 1970). Evidence of cracks generated in this manner forrock samples recovered from deep boreholes has been summa-rized by Kowallis and Wong (1983). Teufel (1983) argued thatthese microcracks are aligned primarily by the directions of theprincipal stresses and that the total microcrack volume from strainrecovery is volumetrically proportional to the in-situ stress mag-nitudes. As a consequence of the development of such a statisti-cally aligned microcrack population, a fully relaxed low-porosityrock might exhibit a microcrack fabric, which may be reflectedby corresponding compressional (Vp) and shear (Vs) wave velocityanisotropies. Teufel (1984) suggested that values of Vp should beleast in the direction of maximum strain recovery and greatest inthe direction of minimum strain recovery and found evidence tosupport such a hypothesis. Wolter and Berckhemer (1989) attrib-uted similar observations of velocity anisotropy to the strike ofthe foliation, but also noted that the principal strain recovery isperpendicular to the foliation.

The strain associated with relaxation of in-situ stored strainenergy generally involves both an instantaneous elastic compo-nent and a time-dependent anelastic component. Voight (1968)first suggested that empirical justification exists for consideringthat the partial recoverable component of the anelastic strain willbe proportional to the total recoverable strain and, hence, thepreexisting state of stress in Theologically isotropic rocks. Hepostulated that by measuring rock cores soon after their removalfrom the borehole, strain and stress ellipsoids calculated fromthese measurements would enable one to estimate the principalin-situ stress ratios and directions at depth.

Teufel (1982) followed up Voighfs propositions by usingclip-on disk gauges and found good agreement between stressorientations determined from strain relaxation measurements andthose from hydraulic fracture measurements. However, he warnedthat a strong anisotropy in the rock prior to relief, such as a

tectonically induced microcrack fabric, may produce nonlinearcreep, which might cause a rotation of the principal axes of strainrelief with time and, hence, lead to inaccurate stress orientations.

During anelastic strain recovery monitoring in Devonian shale,Blanton and Teufel (1983) observed contraction of the rocks aswell as expansion. They attributed this to large temperaturechanges in their laboratory at the time and subsequently estab-lished temperature-strain correlations with which to correct theirfield data. However, Teufel (1986) noted that even after tempera-ture corrections had been made, contraction occurred in the mini-mum horizontal stress direction in all of 18 sandstone coresmeasured. Strain-recovery core contractions also were noted bySarda and Perreau (1989) and by Dey and Kranz (1983), whoattributed the effect to in-situ pore pressure that caused the diffu-sion of pore fluid from the sample, deflating the microcracks andcompeting with the expansion effects.

Wolter and Berckhemer (1989) measured acoustic emissionevents at the same time they measured anelastic strain recoveryand noted a strong correspondence between overall characteristicsof both the acoustic emission and the strain recovery curves. Theycommented that "bursts" of emissions probably reflect avalanche-type microcrack interactions and concluded that although thestrain recovery behavior is compatible with a rock behaving as aviscoelastic continuum, the acoustic emission behavior is clearevidence of volumetric expansion by discontinuous tensile mi-crocrack production.

Theoretical Developments

Voight (1968) postulated that one could determine stress mag-nitudes and orientations from the proportional expansion of thestress ellipsoid. The orientation of the principal horizontalstresses may be calculated simply from the relative strains meas-ured by each gauge and by considering the geometrical arrange-ment of the gauges placed around the core (Frocht, 1948; Dallyand Riley, 1978; Jaeger and Cook, 1979).

Blanton (1983) developed equations for calculating stressmagnitudes on the basis of the assumptions that the rock is linearlyviscoelastic, isotropic, and also that the in-situ stresses follow aStepwise unloading function at the moment of coring, which actsas the driving force for the subsequent viscoelastic relaxation.Solutions were obtained for the anelastic recovery of isotropic andtransversely isotropic core, in which time-dependent behavior iscontained in creep compliance terms. Blanton's equations allowone to calculate the principal stress magnitudes, provided that (1)the principal recovery strains have been measured, (2) Poisson'sratio is known and is constant, and (3) the vertical principal stresscan be calculated from the overburden or gravitational loading.The constant Poisson's ratio accommodates a global creep com-pliance partitioned between distortional and dilatational compo-nents.

Blanton and Teufel (1983) further modified these equations byconsidering the influence of pore pressure. For the isotropic case,the principal horizontal stress magnitudes become,

σ, = (σz - aPD+ aPp, (1)

p

+ aPp,

where

K»

(2)

(2a)

470


and also öx, öy, and σz are the principal stress magnitudes; At,x,^y and £z are the changes in principal strains between any two

times; v is the isotropic Poisson's ratio; Kb and Ks are the bulkmodulus and matrix material bulk modulus, respectively; and Pp

is the pore pressure, α is referred to as the poroelastic constant.The x and y suffixes represent the horizontal directions and the zsuffix, the vertical direction. When the poroelastic constant istaken to be zero, Equations 1 and 2 revert to Blanton's originalequations.

Teufel (1986) conducted a parametric analysis and found thatthe increase in the minimum horizontal stress, Gh was more thansix times greater than the corresponding increase in the maximumhorizontal stress oH as Poisson's ratio increased from 0.1 to 0.4.In addition, Poisson's ratio may have a more significant effect onthe magnitude of σA than those errors associated with fluctuationsin the principal differential strains measured during recovery.Furthermore, σH is not sensitive to changes in the poroelasticconstant, but Gh significantly decreases with a decrease in α.These observations indicate that errors when calculating the prin-cipal stress magnitudes from the above equations are more sensi-tive to errors in the estimation of the mechanical properties of thecore, particularly Poisson's ratio, than to errors when measuringin the strain recoveries, and that these errors exert a much strongerinfluence on the minimum horizontal stress than on the maximumhorizontal stress.

Blanton (1989) showed that if <5h is known, or can be reliablyestimated, then both Poisson's ratio and the pore pressure termcan be eliminated from Equations 1 and 2 to permit cH to bedetermined without the need to know these parameters.

Warpinski and Teufel (1986, 1989a, 1989b) presented a re-fined isotropic viscoelastic model that retained some of the as-sumptions of Blanton's analysis, but was based on curve-fittingthe measured anelastic strain recovery data. These equations areof the form,

%. (t) = (2σx cos2 θ + 2σ sin2 θ - σx sin2 θ - σy cos2 θ - σz)

(σx- 3PP) (1 - e~"2) J2, (3)

and

(4)

where t is time; θ is the gauge angle with respect to the maximumstress orientation; J and J2 are the distortional and dilatationalcreep compliance arguments, respectively; t and t2 are deviatoricand dilatational creep time constants, respectively; and the suffixr is the radial direction in the horizontal plane. When developingthis model, Warpinski and Teufel assumed that strain relaxationtakes place through an intergranular fracturing process, the for-mation of microcracks, or grain boundary sliding, and thus thatthe matrix material bulk modulus will be neither a viscoelasticparameter, nor need the value of α be specified.

Each of the strain-gauge recovery curves is least-squares fittedto the above model to produce a set of equations relating theprincipal stress magnitudes to a series of coefficients. Warpinskiand Teufel (1989b) noted that the distortional parts of Equations3 and 4 are the more useful because they do not involve porepressure. However, these authors found that either the distortionalcompliance, /,, or the minimum horizontal principal stress mag-nitude, σ̂ , is required in addition to the vertical principal stress,σz before the maximum horizontal principal stress can be calcu-lated. The vertical principal stress was calculated in the same wayas in the Blanton model by integrating the gravitational loading,but because Jλ is an unknown or can only be estimated, σ., was

found by conducting a mini-hydrofracturing stress measurementin tandem with the anelastic strain recovery measurements. Evenso, a parametric analysis showed that over a "reasonable range ofvalues," the minimum horizontal stress magnitude was not sensi-tive to changes in Jλ and that the maximum horizontal stress wasnegligibly affected. However, the requirement for independentmeasures of both the minimum and vertical principal stress mag-nitudes, before the maximum stress magnitude can be calculated,might represent a serious restriction on the routine use of theWarpinski and Teufel model.

Nevertheless, Warpinski and Teufel (1989b) were able to usetheir model to reproduce excellent matches with measured strainrecovery curves and to reconstruct the entire strain history of acore sample since the time it had been cut. They were also able todo this for siltstone samples from the North Sea that exhibitedsustained contractions in all three axes, although they were notable to derive meaningful material properties for these examples.In addition, they were also able to demonstrate the time-dependentresponse of Poisson's ratio during anelastic strain recovery, em-phasizing the need for caution when using the Blanton model.

In a discussion of the Warpinski and Teufel (1989b) study,Blanton (1989) referred to previous work (Blanton and Teufel,1986) to show that when the two models are presented in acomparable form, they are essentially similar. One exception isthat the Warpinski and Teufel model imposes an exponentialdecay to dilatational creep, whereas the modified Blanton modeluses the actual decay, exponential or otherwise. Blanton (1989)claimed somewhat better fits to Warpinski and Teufel's experi-mental data than they had achieved themselves and also debatedthe relative merits of some of Warpinski and Teufel's assumptionsand criticisms of his model, which provoked a further responsefrom Warpinski and Teufel (1989c).

In a preliminary presentation of anelastic strain recovery re-sults from the German deep drilling project (KTB), Wolter andBerckhemer (1989) invoked a simple exponential decay model(Engelder, 1984) of the form;

(5)

where £,, = the strain at any time, t; ̂ is the asymptotic value ofstrain as time tend to infinity; and the value, x, is referred to asthe relaxation time. By plotting the logarithm of £, -1,, vs. t, theywere able to demonstrate relatively constant relaxation times formuch of their data.

Equipment Development and Practical Use

The underlying basis of the anelastic strain recovery experi-mental procedure is that upon being released from its in-situ stressenvironment, a rock core will expand in the direction of and inproportion to the magnitudes of the horizontal and vertical in-situstresses. The assumption is that the rock sample is homogeneousand will expand elliptically across its section and uniformly alongits length. By measuring the physical change in shape of thesamples at three or more predetermined positions across its di-ameter, the orientation and magnitude of the resultant strainrecovery ellipsoid can be calculated.

Early attempts to measure anelastic strain recovery (Swolfs,1975; Strickland and Ren, 1980) used standard rosette-type straingauges mounted on the surface of the core. Scientists thought thatproblems associated with this technique led to inconsistent re-sults.

To overcome some of these problems, Teufel (1982, 1983,1984) used clip-on disk gauges in a temperature-"stabilized"environment and covered the core with polyurethane wrapping toreduce moisture loss. Thereafter, Teufel and Warpinski (1984)

471


and Teufel (1986) used spring-loaded clip-on gauges that consis-ted of a support ring to hold a linear voltage-displacement trans-ducer (LVDT) and an adjustment screw (Holcomb and McNamee,1984). The support ring and adjustment screw were made fromsteel alloys chosen for their rigidity and low thermal expansivity.Under this arrangement, the core is horizontally supported at twopositions on a frame, and the gauges are mounted across thediameter of the core at two points. The rock core is not restrictedin any way, and the LVDT thus is free to respond to any changein diameter of the core without external influence.

A similar approach using LVDTs was adopted by El Rabaa andMeadows (1986) and El Rabaa (1989), except that the core wasmounted vertically within a frame that in turn held the LVDTs.Variations on this approach also were adopted by Sarda andPerreau (1989) and Wolter and Berckhemer (1989). However,these methods are open to the criticism that because the core ismounted vertically and the LVDTs are held within a rigid frame,the points of measurement will be subjected to lateral forces asthe core undergoes anelastic strain recovery along its length.

To overcome the effects of evaporation of the core fluids andalso to stabilize the temperature of the rock core and the measur-ing equipment, Sarda and Perreau (1989) placed the core andequipment in an oil bath, while Wolter and Berckhemer (1989)coated the core with wax. In each case, the core and measuringgauges were placed in a temperature-stabilized environment.

Principal stress orientations usually are calculated with respectto a scribe line marked along the length of a core. The core, inturn, must be oriented with respect to magnetic North. Ideally, thisshould be performed with an orientation device built into the corebarrel (Teufel, 1986), but it can be done from the borehole tele-viewer of formation microscanner images from logs subsequentlyrun in the borehole (Wolter and Berckhemer, 1989).

BP Research (Sunbury) U.K. followed the Holcomb andMcNamee philosophy, but with modifications to equipment de-tails and measuring techniques. The BP equipment was madeavailable to the British Geological Survey (BGS) during the earlystages of this project, and many of their design features wereincorporated into the equipment used during Leg 123.

The essentials of the BGS equipment (Fig. 1) are as follows:

1. An adjustable frame holds a core sample horizontal.2. A set of four LVDT strain transducers is mounted within

carbon-fibre support rings. Carbon fibre was chosen because ofits high stiffness, low weight, and low coefficient of thermalexpansion relative to steel. The strain-gauging transducers have arange of ±2 mm, a limit of measurement sensitivity of less than0.1 µm, a linearity of 0.25% of range, and a temperature coeffi-cient of 0.01 %/°C. The strain transducers are fitted with 60°conical tips and springs of just sufficient strength to support theweight of the ring assembly. Adjustable and lockable supportbars, diametrically opposite the LVDT, are made from Invar steelalloy having a low thermal coefficient of expansion. The corelongitudinal-axis strain-transducer support is made from carbon-fibre rods and alloy bars.

3. The effects of any slight temperature variation on the sup-port-ring/strain-transducer/ adjustment-bar assembly are moni-tored by a sixth strain-transducer set across a glass ceramic diskhaving a low coefficient of thermal expansion (0.05 × 10~6/°Cover a 0°-50°C temperature range). Any apparent strain changesrecorded across the disk are taken to reflect temperature and otherenvironmental effects on the equipment itself and are sub-sequently subtracted from the recorded output of each of thecore-measurement gauges.

4. A platinum temperature probe is used to record the tempera-ture immediately adjacent to the core sample.

5. The rock sample is placed on the frame, together with thestrain gauging transducers and temperature sensor, inside a ther-mally insulated container filled with an inert silicon fluid. Thehigh thermal inertia of the silicon fluid together with the thermallyinsulated container serve to minimize the effects of laboratorytemperature fluctuations and also to prevent loss of pore fluidfrom the core by evaporation. In addition, a sufficiently high fluidviscosity is chosen to dampen out potential vibrations transmittedfrom the ship to the sensitive strain transducers.

6. The strain-transducer and temperature-sensor control unit isinterfaced with a portable microcomputer that acts as a datalogger. The data are captured directly into a spreadsheet usingLotus Measure as the communication software.

Up to three rock cores can be monitored at a time, and theequipment can be set to collect and record data at intervals of from1 min for a period of up to several days without attendance. Atany time during data collection, a realtime graphical displayallows for monitoring of progress and data quality.

A tee-delta strain-transducer arrangement (Dalley and Riley,1978) is used, whereby the four gauges are positioned at 0°, 60°,90° and 120° around the core, referred to as positions A, B, D,and C, respectively. Position A is arranged to correspond with thescribe line. The use of four gauges, rather than three, effectivelygives more data than necessary and thus greater confidence in theresults.

The principal horizontal strains and their orientations are cal-culated from the following relationships:

(6)

and

(7)

where, ‰.B.CD a r e t n e strain recoveries measured by the respectiveLVDTs, %uh are the maximum and minimum principal horizontalstrains, respectively, and Φ//, is the negative angle of t,H (followingthe sign convention adopted by Dally and Riley, 1978) frommeasurement position A.

Equations 6 and 7 assume an idealized elliptical deformationof the rock. Yasunaga (1989) showed that this hypothesis can betested with the following relationship:

(8)

The nearer the left side of Equation 8 is to zero, the nearer therock deformation approximates an ellipse.

Leg 123 Shipboard Measurements

Sample Selection and Preparation

None of the sediment cores recovered from Holes 765A, 765B,and 765C were sufficiently lithified to be suitable for anelasticstrain recovery experiments. The first sample for this study wasthus taken from the first complete basement core at Hole 765C(Bas 1). The subsequent poor recovery of concomitant whole-round rock core resulted in only 10 additional samples beingrecovered from Hole 765D before drilling ceased. In most cases,the overall core was so broken that the samples that were recov-

472


to

Figure 1. Anelastic strain recovery equipment and core sample Bas 1.

473


ered were the only suitable ones available. Similarly, none of thesediment cores recovered from Hole 766A were suitable foranelastic strain recovery experiments. The diabase dolerite sam-ples from Site 766 were less invaded by calcite veins than thosefrom Site 765 and were selected for being as free as possible fromsuch features. The last two samples (Bas 14 and Bas 15) wererecovered from the same core section and were immediatelyadjacent to one another. The break in the rock between these twosamples could be matched, so that one continuous scribe linecould be drawn from one to the other. This permitted repro-ducibility of the results for assessment between two halves ofwhat is essentially one sample.

An outline of the sample preparation is presented in the InitialReports volume of Leg 123 and in a more detail by Brereton et al.(1990) and will not be repeated here.

Results

The equipment for measuring stress recovery of the core per-formed without fault, and the silicon fluid prevented any dryingout of the sample during measurement periods. The thermal insu-lation afforded by the immersion of the sample within the siliconfluid inside the insulated container restricted the temperature driftduring the experimental periods to a maximum of 0.8°C, butgenerally to within 0.3°C.

We have been unable to conduct a full parametric analysis ofthe errors in the calculated maximum and minimum strain recov-eries and in strain recovery angles that can be directly attributedto equipment measurement errors, and clearly, this will dependupon the magnitude of the recovered strains, but we think theseerrors are small in relation to the geological and geomechanicalfactors.

The displacements recorded by each of the gauging transduc-ers were converted into strains, and the maximum and minimumprincipal horizontal strains and the angle of the maximum princi-pal horizontal strain relative to the scribe line, Φw (a negativeangle, following the sign convention adopted by Dally and Riley,1978), were calculated from Equations 6 and 7. The values of Φ#,and £,H,h'V (shown in Tables 1 and 2) were taken as the means overthe final few hours of strain recovery when approximateequilibrium had been achieved.

No core orientation device was used when drilling during Leg123, and thus the orientation of the stress ellipsoid relative tomagnetic North was determined from paleomagnetic measure-ments. Shipboard scientists also hoped to use the borehole tele-viewer to help orient the samples, but the quality of the televieweroutput during the cruise was not good enough. The relative posi-tion of magnetic North during the Early Cretaceous for the west-ern Australian margin has been estimated as 270° ±30° (Emble-ton, 1981,1984). If the mean magnetic orientation with respect tothe scribe line is θ (Table 1), then the position of the scribe linerelative to present-day North is 270 - θ, and the orientation of themaximum principal horizontal strain relative to present-day North(AH) is 270 - θ - Φtf These orientations are shown in Table 1.

As discussed earlier, instrumental errors when determining thestrain recovery orientations are thought to be small. Deviationsabout the mean are shown in Table 1 and are generally less than2°. The instrumental error in the paleomagnetic measurements isestimated as ±5° (James Ogg, pers. comm., 1988), and the sys-tematic error in the position of North during the Early Cretaceousis ±30°. However, as mentioned in the previous section, the lasttwo samples (Bas 14 and Bas 15) were recovered from the samecore section immediately adjacent to one another and can beregarded as duplicates. The angles of maximum strain relief,relative to the scribe line marked onto these two samples, were20° apart, while the angles of paleomagnetic measurements were41° apart. Therefore, the total error in AH including both instru-

mental errors and "geological" errors, is difficult to estimate, butmay be significant.

In an initial appraisal of the Leg 123 strain recovery data,Yasunaga (1989) used Equation 8 to estimate how closely therecorded core sample deformations approximated an ellipse. Hefound that values for the left side of the equation were 0.2 for coreBas 11; 22 for Bas 5; between ±40 and 81 for Bas 1, 3, 6, 7, 10,13, and 15; and between ±121 and 205 for Bas 2, 4, 8, 9, 12, and14. The implication from these results is that for many of the coresamples, the four horizontal gauge recordings are not repre-sentative of elliptical deformation. However, the application ofthis check should be more rigorously validated before it can beused quantitatively.

The vertical and maximum and minimum principal horizontalstrains are shown in Table 2. Also shown are the elapse times fromthe onset of strain relief monitoring to approximate equilibrium.The graphical results of the principal strain recovery curves, theLVDT measured strain recovery curves, the principal angles, Φ//,and the recorded temperatures for each core sample have beendescribed by Brereton et al. (1990), but examples for the samplesBas 4, 8,9, and 12 are shown in Figure 2. Uncorrected preliminaryprincipal-strain recovery curves for the samples Bas 1, 2, and 8were shown in the Initial Reports volume of Leg 123.

Only the core samples Bas 7 and 11 exhibit the classicalanelastic strain recovery expansion curves in all three axes, whileBas 5 and 9 exhibit similar initial trends that then reverse tobecome contractions in the minimum horizontal or vertical axes.The core samples Bas 1, 2, 6, 8, and 10 mainly show smooth,relatively large-magnitude expansions in the maximum horizontaland vertical axes and contractions in the minimum horizontal axis.For all but one of these samples (Bas 1), the minimum horizontalaxis contractions were derived from the response of only one ofthe five LVDTs, with the other four recording expansions. Thesamples Bas 13, 14, and 15 exhibit consistently smooth, relativelylarge contractions in the vertical axes, while Bas 13 and 14contracted in the maximum and minimum horizontal axes as well.All the LVDT responses for the sample Bas 12 (Fig. 2D) werecontractions for the first 2 hr, but thereafter one of the horizontaland the vertical measuring LVDTs changed to record relativelylarge-magnitude expansions. Bas 3 was subject to disturbance inthe laboratory, and thus these results are best ignored.

The most startling result is from sample Bas 4 (Fig. 2A) from1005.78 mbsf at Site 765, where the strain relief was large andconsistently negative in all three directions and continued torecover for almost 80 hr without showing signs of reachingequilibrium. A similar result has since been reported by Warpinskiand Teufel (1989b), where a North Sea siltstone contracted con-sistently in all three axes for almost 40 hr.

The relaxation time analysis (Eq. 5) described by Wolter andBerckhemer (1989) was applied to the derived principal straincurves. In general, we found that most of the core samples exhib-ited some degree of linearity on the semi-log plots, but this wasusually over fairly limited time ranges and also over those por-tions of the displacement-time curves that "looked" exponential.Only Bas 2 exhibited consistent linearity over most of the timerange for all three axes, while the Bas 4 plots showed consistentcurvature. No apparent correlation or trend among the derivedrelaxation times, either for each core sample or between coresamples, was observed.

Thermal Stabilization of Core Samples

We recognized from the outset that thermal stabilization of thecore samples might be an imponderable. The temperature at 1195mbsf in Hole 765D was 38.9°C (Ludden, Gradstein, et al., 1990).During the coring process, the core barrel was flushed withdrilling fluid and was subsequently hauled up through almost

474


Table 1. Principal strain orientations derived from paleomagnetic andanelastic strain recovery angles.

Paleomagnetic Anelastic recovery StrainDepth angle angle orientation

Sample (mbsf) (θ) (Φ//) (Δ//)a

Site 765

Bas 1765C-63R-2,

34-54Bas 2

765D-3R-1,110-125Bas 3

765D-5R-6,110-129Bas 4

765D-7R-3,90-110Bas 5

765D-13R-1,31-50Bas 6

765D-18R-3,37-54Bas 7

765D-19R-2,0-27Bas 8

765D-20R-1,77-99Bas 9

765D-22R-2,0-16Bas 10

765D-24R-5,73-90Bas 11

765D-26R-1,73-92

Site 766

Bas 12766A-51R-5,

93-114Bas 13

766A-52R-2,131-147Bas 14

766A-55R-6,33-57Bas 15

766A-55R-6,57-76

937.58

965.47

990.76

1005.78

1054.71

1103.51

1111.60

120.27

1138.60

1163.09

176.73

487.14

492.72

524.56

524.80

219

23

320

-54

87

260

166

211

354

350

293

45

289

286

327

-67.3 ± 0.3

29.2 ± 0.2

85.7 ± 4.5

-38.71 0.2

-20.1 ±0.7

8.2 ±0.9

-2.6 ± 2.2

-11.7 + 0.1

36.5 ± 0.2

-12.5 ±0.4

82.8 ± 1.3

2.7 + 0.1

88.0 ± 1.2

-58.9± 3.0

-79.1 ± 1.0

118

38

44

23

107

71

59

113

74

42

73

43

22

a AH = 270 - θ - Φ//•

6 km of ocean water, thereby cooling the rock core to about 4°Cfor several hours. The average temperature of the silicon fluidwithin which the sample was immersed in the shipboard labora-tory was about 26°C. Therefore, the core samples, over the spaceof a few hours, were subjected to thermal cycling that imposed anindeterminate temperature profiled through the rock. How longthe cores then took to reach equilibrium with the silicon fluiddepended upon the thermal characteristics of the basalt.

An examination of the temperature curves and some of theLVDT responses indicated that core samples had retained a pro-portion of their in-situ heat, but that during the first hour or so,the cores had became sufficiently thermally stabilized so that theeffects of temperature thereafter were negligible (Brereton et al.,1990).

This conclusion was confirmed by subjecting three core sam-ples (Bas 4, 8, and 12) to a heating-cooling cycle designed to

replicate their experiences following coring. This experiment wasconducted some months later, when the samples had fully relaxedfrom the effects of in-situ stresses and any surface moisture hadevaporated. The results from this thermal cycling showed thatwhile the strain recovery curves may be influenced by thermalstabilization processes taking place during the first 2 hr, thereafterthese processes can be discounted.

Thermal Azimuthal Anisotropy

It became clear during equipment construction and testing thatnot only was there a strong correlation between core sampledeformation and temperature, but that thermal expansion-contrac-tion invariably exhibited strong azimuthal anisotropy. Charlez etal. (1986) debated whether microcracking of a rock is a memoryof its initial state of stress, while Lacy (1987) conducted bothanelastic strain recovery and thermal expansion testing and re-

475


Table 2. Anelastic strain recovery principal strains.

Sample

Site 765

765C-63R-2,34-54Bas2

765D-3R-1,110-125Bas3

765D-5R-6,110-129Bas4

765D-7R-3,90-110Bas 5

765D-13R-1,31-50Bas 6

765D-18R-3,37-54Bas 7

765D-19R-2,0-27Bas 8

765D-20R-1,77-99Bas 9

765D-22R-2,0-16Bas 10

765D-24R-5,73-90Bas 11

765D-26R-1,73-92

Site 766

Bas 12766A-51R-5,

93-114Bas 13

766A-52R-2,131-147Bas 14

766A-55R-6,33-57Bas 15

766A-55R-6,57-76

Depth(mbsf)

937.58

965.47

990.76

1005.78

1054.71

1103.51

1111.60

1120.27

138.60

1163.09

1176.73

487.14

492.72

524.56

524.80

Final principal strains

(m strains)

31.5+0.7

46.3 ±0.6

-2.1 +0.7

-70.5 +0.5

49.7 +0.6

5.0 ±0.6

14.5 ±0.8

139.0 ±0.8

81.9 ±0.8

20.2 ±0.7

25.5 ±0.9

144.8 ±1.2

2.0 ±1.2

-24.6 ±1.3

11.8 ±1.7

(m strains)

-47.6 ±0.9

-41.1 ±0.9

-6.7 ±0.9

-205.9 +0.6

5.6 ±1.1

-24.9 ±0.5

2.5 ±0.8

-3 ±0.4

6.1 ±0.7

-9.9 ±0.6

12.6 +1.0

-51.3 ±0.7

-12.5 ±0.9

-41.6 ±1.6

-3.1 ±1.3

(m strains)

1.1 ±0.8

57.9 ±0.6

5.9 ±0.4

-102.9 ±0.3

1.5 +0.2

14.1 ±0.4

6.6 ±0.2

28.5 ±0.2

21.4 ±0.3

18.3 +0.1

16.1 ±0.4

-5.6 ±0.2

-36.5 ±0.3

-30.3 ±0.3

-33.4 ±0.2

Elapsetimea

38.6

22.5

11.6

79.5

25.0

11.5

5.7

46.3

20.1

19.9

9.9

19.7

14.8

19.5

19.5

Hours from onset of strain relief monitoring.

ported that the results from the same cores were usually consis-tent.

The core samples, Bas 4, 8, and 12, were instrumented as ifstrain recovery was to be recorded, but were placed on the BGSlaboratory bench and allowed to respond to air temperaturechanges cycling up and down between 15.5° and 26.5°C overseveral days. First, an obvious and marked correlation was seenbetween apparent strain and temperature, but a considerable de-gree of anisotropic behavior also was noted. The displacement-temperature curves ranged from being virtually linear with a highcorrelation coefficient and little or no hysteresis through exhibit-ing hysteresis by following different displacement curves, de-pending on whether the temperature was increasing or decreasing,to exhibiting an incoherent response with little correlation at all.Some of the gauges across both Bas 8 and Bas 12 showed consis-tently negative slopes to the curves. The maximum slopes for Bas4, 8, and 12, respectively, were 19, 21, and 26, microstrains/°C,

while the minimum slopes were 7, -5 and -7 microstrains/°C. Theresponses of the strain gauge assemblies mounted over the glassceramic disks were generally somewhat erratic with a low corre-lation coefficient and a maximum of no more than 6 micro-strains/°C for short periods of time. The orientations of maximumthermal expansion showed a tendency to switch through 90°,depending upon changes in the relative magnitudes of measuredstrains with temperature for individual LVDTs. However, apartfrom Bas 4, no apparent consistency existed between the thermalexpansion orientations and the strain recovery orientations.

INTERPRETATIVE STRATEGYOne can clearly see from previous sections that the strain

recovery curves, obtained on board the ship, exhibited largevariations, both in the character of the relaxation curves and inthe calculated directions of the principal horizontal stresses. Mostcurves show expansion, as expected, but also contractions, while

476


the orientations of maximum compressive stress showed no con-sistent trend (Table 1). Factors that can explain these potentialvariations are grouped under the following topics:

1. Instrumental problems.2. Core orientation inaccuracies.3. Complex in-situ stress conditions.4. Anisotropy of elastic properties in the basalt samples.

In general, we considered instrument errors were small andnegligible in comparison to the magnitudes of the strains re-corded. Concerning core-orientation errors, we have not, as yet,been able to check out more rigorously the paleomagnetic results,but any error in the estimation of the pole direction during theEarly Cretaceous, of course, will be systematic and common toall the results and thus cannot account for variations among theorientations of core samples.

The possibility of complex stress conditions in the in-situ rockmass (perhaps involving significant localized pore-water pressurevariations and macro anisotropic characteristics that invalidatesome of the theoretical assumptions) remains. The problem ofelastic anisotropy is potentially of major importance, because theassumption of isotropy is fundamental to the anelastic strainrecovery method. Therefore, information about the elastic prop-erties or factors that influence elastic properties should be consid-ered. These properties are also important for providing clues tothe mechanisms of strain recovery.

The elastic properties of the sample will be controlled by avariety of factors, including visible fractures and voids on amacro-scale, and mineralogy, crystal orientation, and cracks on amicro-scale. Information was sought by a variety of methods,including the following:

1. Macrocrack characterization (with bulk sample descrip-tion).

2. Mineralogy (from thin section analysis).3. Scanning electron microscopy.4. Dynamic measurements of seismic velocities.

Not all of the recovered basalt samples for which anelasticstrain recovery measurements were recorded could be studied indetail, and some selectivity was necessary.

CHARACTERIZATION OF CORE SAMPLES

Macro-Characterization StudyIt was clear from a superficial examination that core samples

were intersected by an abundance of fractures, some of whichpassed right through the samples and were cemented by calcite orother minerals. Therefore, before any destructive work was doneon the cores, we considered it prudent to construct a qualitativedescription of the fracture structure of each core.

It is difficult to express three-dimensional structures in twodimensions, and various attempts to accomplish this included thefollowing:

1. Photographing at 90° intervals around the core.2. Photographing with vertical and horizontally polarized

light.3. Photographing with ultraviolet and fluorescent dye.4. X-ray photography.5. Surface tracing and feature moulding.

All samples were photographed at 90° intervals as a generalrecord, but we concluded that none of the above methods im-proved upon the visual tracing by hand on transparent sheeting as

a means of recording small but visible cracks. Tracings for eachof the cores have been described by Brereton et al. (1990), butexamples for the samples Bas 4, 8, 9, and 12 are shown in Figure3. In each case, the angular orientation around the core is relatedto the scribe line.

A persistent fracture plane through a core is characterized, ona two-dimensional representation, by a sine curve. Evidence ofsuch sine curves was clear (Bas 12), but most of the fracturing ismore complex. Most of the cores exhibit a considerable numberof fractures, some intensely so (Bas 8). Only the cores Bas 12, 14,and 15 from Site 766 are relatively free of fractures. Some frac-tures have been intercepted by others, and some terminate withinthe rock matrix. Some are smoothly curved, while others areangular and rough. Many are filled and cemented. Two generalobservations about fracture structures can be made: (1) the per-sistency of the fractures is relatively small so that most terminatein the core, either by interception or by disappearing into the rockmatrix; and (2) four dominant fracture sets seem to be suggestedas being almost vertical or almost horizontal and have dips ofabout 60° and 40°.

Core structures also were traced on board the Resolution bythe USGS-Stanford team (Ludden, Gradstein, et al., 1990). Thesescientists recorded a population of fractures from 27 cores fromSite 765 that were generally symmetrically oriented, some ofwhich were filled with calcite, smectite, and other clay alterationminerals. The dominant dip angle of through-going fractures fellnear 45°, with additional nodes near 30°and 60°, but the rangecovered from 15° to 90°. A similar study was conducted usingseven cores from Site 766. Again, the dominant dip angle fell near45°, with a range of from 5° to 80°, but no clearly definedadditional nodes. Most of these fractures were filled with smectiteand minor amounts of pyrite. Secondary crack growth was oftenassociated with a trend that cross-cut the preexisting fractures atoblique angles. These late-forming fractures were often, thoughnot always, filled with calcite. An interesting feature of the basaltcores at Site 766 was a systematic increase in dip angle fromclustering near 45° above about 505 mbsf, to growing progres-sively steeper to about 80° by 515 mbsf, and then decreasing againto about 20° below about 520 mbsf. This change in dip angle wasconsidered to result from thermoelastic stresses that occurredwithin the cooling body of the intrusive basalt, coupled with thetectonic stresses prevalent at the time of intrusion.

Many of the fractures that split the cores from Site 766 intosmaller pieces showed the morphological features of core diskingdescribed by Dyke (1989) and by Sarda and Perreau (1989).

Micro-Characterization StudyThin sections were made from subcores drilled from samples

Bas 1, 4, 8, 11, 12, and 15 to ascertain the nature of the crackfillings and to see if any mineral orientation or banding might havecontributed to elastic property anisotropy. The pyroxene/plagio-clase crystals showed no apparent orientation, nor was there anyobvious flow structure or banding on this scale. Most cracks arefilled with calcite, but others contain green clay minerals oropaque minerals. Relict vesicles are present, usually filled withcalcite. Although the thin sections showed some cracks (whichcannot be seen without the microscope), they do not provide anymore evidence of crack orientation than could be seen with thenaked eye on the hand specimens.

In addition to the optical study, pore structure was examinedusing scanning electron microscopy (SEM). A slice from each ofthe thin-sectioned cores was taken using a slow-cutting saw tominimize crack damage and then polished with a 1 µm paste.

The observations from the thin section and the SEM studiesshowed that, in general, the cracks could be categorized as fol-lows:

477


Max

Long

Min

Time (hours)

Figure 2. Graphical results from the anelastic strain recovery measurements. Examples forcore samples Bas 4 (A), 8 (B), 9 (C), and 12 (D) are shown with three diagrams for each: (1)the calculated principal strains showing the maximum and minimum horizontal strains andthe vertical or longitudinal strain; (2) the LVDT-measured strains where numbers 1, 2, 3, and4 correspond to A, B, C, and D around the core; 5 = the vertical measurement; 6 = the responsefrom the glass ceramic disk; (3) the calculated angle of the maximum strain direction relativeto the scribe line marked on the core sample and the temperature recorded immediatelyadjacent to the core during the course of the measurements.

478

Long


Max

50

Figure 2 (continued).

479


Max

2 4 β 8 10 12 14 16 18 20

Long

Mln

- 26.2

26.1

8 10 12

Time (hours)

14 16 18 20


480

Long

0 2 4 6 8 10 12 14 16 18 20

6 8 10 12 14 16 18 20


Max

23.6


481


A

BASALT 4

B

0 [20 40 60 80 |iOO 120 140 160 I80 200 220 240 260 280 300 320 340

20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340

Figure 3. Macro-characterization of the core samples. Tracings for Bas 4 (A), 8 (B), 9 (C), and 12 (D) are shown. In each case, the angular orientationaround the core is related to the scribe line.

1. Vesicles: These high-aspect ratio pores are usually filled.Only a few are open, some of which show signs of new crystalgrowth within. An example of a typical vesicle is shown in Figure4A.

2. Larger cracks: Cracks visible in thin section and also ob-served using the SEM. These include (1) completely sealed cracksfilled with calcite, but sometimes with opaques or a green trans-lucent mineral; (2) partly sealed cracks that are now open; (3)open cracks with no evidence of healing (Figs. 4B and 4C). It ispossible that the cracks in (2) and (3) may have opened orre-opened as a result of stress release or because of the generalstresses of coring. All of the larger cracks traversed the thinsection or SEM slice. All have low-aspect ratios, and the opencracks thus will close at low confining pressures.

3. Microcracks: These are observed only with the SEM and arecomparable to the basalt crystal sizes or smaller. These include(1) grain boundary cracks, most of which appear to be healed, butsome are discontinuous, with evidence of bridging or healing;these can have variable aspect ratios (up to 1.0), but overallrepresent a minute fraction of the total porosity of the sample(Figs. 4D and 4E); (2) irregular discontinuous fractures that ap-pear to be of "secondary" origin (Fig. 4F); (3) cracks withincrystals, which may be associated with crystal structure (Fig. 4G).In this category, the widths of the cracks are typically about 1 µm,and their contribution to the overall porosity is almost negligible.While some cracks are clearly related to the primary formation ofthe basalt, others may represent localized stress release, but onecannot say when this occurred.

D


180

270

) 1 * i 1 V

0 [20 [40 |βθ |βθ |iOO |i20 |i4θ|i6O |18O |20θ]22θ|24θ|260 |28θ|3Oθ|32θ[3'40

0 I 2 0 I 4 0 60 80 100 120 140 160 180 200 220 240 260 280 BOO 320 340


The micro-characterization study was not comprehensive, butresults indicate that the overall microcrack density and porosityare small. This is in agreement with the findings of Brereton (thisvolume), where a mean porosity of 3.5% (ranging from 1.6% to8%) from 61 determinations was found. Although a range ofaspect ratios is evident, it appears that the porosity is containedeither within low-aspect ratio cracks (less than 0.05) or withincracks having high ratios. The origin of some of the cracks isundoubtedly primary, in that they relate to the formation of thebasalt (e.g., vesicles) and to thermal contraction along grainboundaries. Fractures that are filled, or partly so, presumablyrelate to ocean-floor processes, but the clean cracks or fracturesmay be related to coring processes, or to strain recovery.

DYNAMIC ELASTIC PROPERTIES

Subcore Selection and Measurement

Facilities at the UEA allowed us to measure seismic velocitiesin rock samples at confining pressures of up to 0.4 GPa withcontrolled pore pressures. From measurements of density andcompressional (P-wave) and shear (S-wave) velocities, the elasticproperties (bulk modulus, shear modulus, and other parameters)can be determined. The velocities themselves, measured at differ-ent orientations, can provide an indication of fracture anisotropy.

Subcores having a diameter of 22 mm were removed from thecore samples Bas 1, 4, 8, 11, 12, and 15. In almost all cases, five

483


B

200 µm 300 µm

20 µmFigure 4. Electron photomicrographs showing examples of the microcrack features observed. A.Partly filled vesicles with open boundary crack. Other high aspect ratio cracks may result fromplucking during sample preparation. B. Typical clean fracture. At this scale, no porosity is visibleother than the fracture (and vesicles). C. Detail from B, showing open crack with no evidence ofhealing. D, E. Grain boundary cracks that are substantially healed, but still show a narrow opencrack. F. Localized low-aspect ratio crack. G. Aligned low-aspect ratio cracks associated with crystalstructure.

subcores were taken from the same positions where the anelasticstrain recoveries were measured, including one from the verticaldirection. The ends were trimmed, and all were oriented relativeto their position within the core sample. They were vacuum-dried,saturated with seawater, and "dry" and "saturated" densities andapparent porosity were calculated. Thin sections were oriented foreach subcore from the trimmed material.

Seismic velocities were measured by a pulse transmissionmethod similar to that of Birch (1960), using lead zirconiumtitanate transducers having a frequency of 1 MHz. The transit timeof the pulse across the sample was measured to an accuracy ofabout 10 ns for the P-wave, resulting in an overall velocity error

of about 1%. The 5-wave velocities (measured separately) wereof poorer quality, and some were rejected. Even so, the measure-ment error is probably about 5%. The vibration direction for theS-wave transducer was vertical along the line of the sample,except for the vertical subcore, where it was parallel to the coreorientation mark. All measurements were performed on saturatedsamples, with pore pressure kept at atmospheric pressure.

The in-situ confining pressure on the core materials at Site 765is about 0.1 GPa. Velocities were typically measured at pressuresup to a maximum of 0.4 GPa, because at these higher pressuresmost of the microcracks will be closed, which allows one todetermine the matrix velocities and elastic properties. At low

4X4


2.31 µm 75 µm

Figure 4 (continued). 15 µm 20 µm

pressures (less than about 0.025 GPa for P-waves and 0.05 GPafor 5-waves) the wave onsets were poor, and thus the velocitieswere less accurately determined.

Variation of Velocity With PressureFigure 5 depicts the suite of velocity-pressure curves from the

core sample Bas 8. The characteristics of this data set are typicalof all of the core samples measured and indicate small increasesin velocity with pressure, except for a slightly greater increase atthe very low end of the pressure range (0 to about 0.025 GPa).The velocity-pressure curves indicate tight rocks, with only lowporosities in the form of low-aspect ratio microcracks. Any addi-tional porosity must be in the form of high-aspect ratio pores,which do not close at the pressures applied (Walsh, 1965; Toksozet al., 1976). Velocities from the reducing part of the pressurecycle were always near those on the increasing part, and theabsence of any velocity hysteresis was confirmed by repeatedvelocity-pressure cycling on some cgres. We had intended toexamine pore aspect ratio spectra using the inversion method ofCheng and Toksoz (1979), but this would have been of no valuegiven the small velocity variation in these samples. Results from

these velocity measurements have been tabulated by Brereton etal. (1990) and are summarized in Table .3.

Velocity AnisotropyThe velocity differences among the subcores might potentially

indicate anisotropy in the large sample, but the results should beregarded with some caution. First, all the anelastic strain recoverymeasurement points are not on the same plane, but are severalcentimeters apart along the sample. The macro-scale fractures,often discontinuous, can be distributed unevenly along the coreand thus the elastic properties, and hence seismic velocities, ofany subcore would be heavily influenced by local irregularities,which may not be a component in an overall simple model ofanisotropy. Second, as each subcore is taken from the actualdirection of the anelastic strain recovery measurement it shouldreflect ideally the elastic properties giving rise to the strain recov-ery curves at that point. However, because the subcores are shorterthan the full diameter of the core samples, all components thatcontributed to the strain recovery curves will not be present. Thus,the macro-characteristics of the cores will be accentuated by thevisible fractures, but the dynamic properties will be influenced

485


o•J r- 1 i r

ODP 8-5

> 30

confining pressure (G Pa)

Figure 5. Velocity-pressure curves for core sample Bas 8. The symbol x = a measurement at increasing pressure,o = measurement at decreasing pressure, and Δ = both measurements having same value. Numbers 8-1,2, etc.,indicate subcores. The curves for this sample are typical and show little velocity variation with pressure.

486


Table 3. Variation of P- and S-wave seismic velocities with confining pressure for all the subcores measured.

Sample

BAS01-1BAS01-1BAS01-2BAS01-3BAS01-3BAS01-4BAS01-4BAS01-5BAS01-5BAS04-1BAS04-1BAS04-2BAS04-3BAS04-3BAS04-4BAS04-5BAS04-5BAS08-1BAS08-1BAS08-2BAS08-2BAS08-3BAS08-3BAS08-4BAS08-4BAS08-5BAS08-5BAS11-1BAS11-1BASH-2BASH-2BASH-3BASH-3BASH-4BASH-4BASH-5BASH-5BAS12-1BAS12-1B AS 12-2B AS 12-2BAS12-3BAS12-3B AS 12-4BAS12-4BAS12-5BAS12-5BAS15-1BAS15-1BAS15-2BAS15-2BAS15-3BAS15-3BAS15-4BAS15-5B AS 15-5

Propagation

PSPP

sP

sP

sP

sPP

sPP

sPsPsP

sP

sP

sP

sP

sPsP

sP

sP

sP

sP

sP

sPsP

sP

sP

sPPs

Saturateddensity(g/m3)

2.8352.8352.8572.8662.8662.7892.7892.8412.8412.8692.8692.8652.8162.8162.8602.9702.9702.8782.8782.8822.8822.8782.8782.9302.9302.9092.9092.8132.8132.7122.7122.8322.8322.7912.7912.8302.8302.8192.8192.8372.8372.8472.8472.8502.8502.8402.8402.9282.9282.9332.9332.9182.9182.9222.9292.929

0.0

5.603.135.545.753.145.402.995.522.835.902.985.735.352.805.355.90

-5.603.155.433.185.552.985.752.855.752.904.982.805.022.555.283.035.352.725.572.915.35

-5.15

-4.95

-5.352.985.252.955.773.406.003.405.983.355.995.75

-

0.01

5.723.145.565.863.225.503.005.572.855.943.005.755.442.805.686.11

-5.613.185.653.205.642.985.782.875.752.915.152.825.072.725.453.035.392.765.602.915.53

-5.392.955.152.505.583.025.522.935.983.426.023.406.003.326.006.003.00

0.025

5.723.165.705.893.255.553.025.602.865.953.025.785.552.805.736.15

-5.633.205.703.205.682.995.822.885.792.925.232.825.152.735.533.055.422.825.622.905.65

5.452.965.253.035.633.035.582.936.073.436.053.426.033.306.046.033.15

0.05

5.773.185.715.933.255.583.055.632.865.973.035.805.602.805.786.18

5.703.245.733.205.723.005.872.895.802.955.292.855.212.755.583.055.452.865.652.935.72

.5.502.975.303.055.683.045.632.936.123.456.103.456.063.286.066.063.17

0.075

5.803.195.725.933.255.593.065.632.865.993.045.835.602.815.806.203.105.723.255.783.195.753.035.902.895.822.935.322.875.262.765.603.035.482.885.672.945.74

5.522.985.363.065.703.055.652.926.133.496.123.496.103.306.086.083.20

0.10

5.833.205.725.963.275.603.085.642.876.003.055.875.652.835.836.233.125.783.255.813.225.763.055.912.905.832.935.352.905.282.775.623.035.502.885.682.955.75

5.553.005.383.075.723.065.662.926.143.516.143.516.133.336.106.103.22

Pressure (GPa)

0.125

5.873.225.735.973.295.623.105.672.906.003.065.895.692.865.906.253.185.803.275.813.235.783.105.932.935.852.935.362.935.302.785.643.045.522.895.702.965.78

5.593.025.423.075.723.075.692.946.153.526.163.526.153.366.136.143.25

0.15

5.883.235.745.973.305.623.125.672.926.023.085.905.702.875.926.283.225.823.285.823.245.803.155.952.955.852.925.382.935.322.795.653.045.542.905.702.975.80

5.603.055.453.095.733.075.702.976.173.556.183.546.163.386.146.163.28

0.175

5.913.245.755.983.315.633.135.672.946.073.125.905.722.885.946.293.245.853.295.833.275.823.175.982.965.892.925.392.965.352.805.683.045.552.935.723.005.802.855.623.075.473.115.743.075.723.006.183.556.203.556.183.406.156.183.30

0.20

5.933.265.765.993.325.643.155.682.956.053.155.925.752.915.956.303.255.853.325.833.305.833.186.002.975.912.945.402.975.372.805.703.055.582.965.743.005.822.935.623.105.503.135.753.085.733.026.203.576.223.576.193.406.176.193.32

0.25

5.963.275.795.983.315.683.175.693.006.083.175.965.782.936.006.303.285.893.355.853.355.883.216.022.985.952.955.422.995.422.825.723.035.622.995.783.035.872.985.683.155.583.155.793.105.763.056.203.556.263.596.223.426.206.233.35

0.30

5.973.285.805.983.315.703.205.703.026.113.206.005.802.956.026.303.335.923.385.893.385.903.256.073.005.982.965.453.005.482.845.773.015.653.005.803.035.893.015.703.205.653.205.823.125.783.086.233.606.303.606.253.456.246.283.35

0.35

5.983.305.805.983.305.733.225.743.036.153.246.035.832.976.156.323.365.953.405.903.405.923.276.103.006.022.975.463.005.542.855.863.015.703.005.843.035.903.045.723.225.723.255.863.145.823.126.253.646.333.646.283.486.276.303.37

0.40

6.013.305.826.003.305.743.225.753.056.183.286.055.872.986.086.333.385.953.435.903.435.963.286.153.006.052.985.503.005.572.875.833.055.733.005.873.035.923.055.733.255.803.285.883.155.853.106.253.676.353.666.303.506.306.353.38

heavily by the matrix properties and any larger fractures that cutthe subcore. This is a scale effect that has been introduced amongthe core samples and the subcores.

A summary of the P and S-wave velocities (Vp and V̂ ) withrespect to their orientation is shown in Figure 6. For the cores Bas1, 4, and 11, the P-wave velocity differences among the subcoresis greater than the likely error, and the velocity distributionsuggests that these differences can be substantially explained bya simple velocity ellipse. Bas 8 shows velocity variation, butwithin the possible error, and for Bas 15, the velocity differencesare not significant. Bas 12 subcores show variations that areslightly greater than the errors, but the four directions (perpen-dicular to the axis of the core) show alternating high and low

values. Such a pattern of anisotropy may be explained by inter-secting fracture planes, whereas the simple "anisotropy" shownby the cores Bas 1, 4, and 11 may reflect one dominant fractureorientation parallel to the high-velocity direction (for example,see Crampin et al., 1980).

Such dominant fracture orientations were not seen on thesecores, and the velocity measurements suggest only some depend-ence on the macro-crack characteristics. However, a comparisonof these velocity anisotropy results with the anelastic strain recov-ery results does provide some evidence to support the view thatvalues of Vp should be least in the direction of maximum strainrecovery and greatest in the direction of minimum strain recovery.This is entirely true for sample Bas 8, while for samples Bas 1 and

487


ODP CORE 1 ODP CORE 8

P-WAVEP-WAVE

5.63 ', 5.75 ( 2 C * 5.74

5.82

6.01

0.4 GPa

5.77

5.92

5.75

5.75

0.1 GPa

S-WAVE

P-WAVE

j ^ r 5.80618 ( S T * * 5•78

' 5.60

5.97

0.05 GPa

ODP

6.23

6

CORE

T\*I. 00

0.1 GPa

4

5.87

5.83

5.65

ODP CORE 11

6.05

6.08

5.87

6.18

0.4 GPa

S-WAVE

Figure 6. Measured P- and 5-wave velocities at 0.05, 0.1, and 0.4 GPa pressures, plotted according to subcore orientations. The symbol o on the 0.05GPa plot = direction of scribe line, and other symbols refer to subcore numbers. Unreliable S-wave velocities have not been included.

Bas 4, the minimum values of Vp correspond to the LVDT posi-tions that recorded the maximum strain recovery. The opposite isthe case for Bas 12, but this sample produced anomalous strainrecovery results that may be attributed to the prominent sine wavefracture noted during macro-characterization. Results from Bas11 and 15 were less conclusive.

Complete core shear-wave velocity anisotropy measurementswere performed on samples Bas 13 and 14 at Imperial College.The results for Bas 14 were inconclusive, but those for Bas 13showed a clear correlation between the orientation of minimumVs and the orientation of maximum strain recovery.

Elastic ModuliFor each subcore, the apparent elastic moduli (and other para-

meters) can be calculated from the respective P and S-wavevelocities and densities. The moduli are not necessarily truevalues because only simple equations were used that assumeisotropic elastic behavior. The determined velocities and derivedparameters for each subcore at varying pressures have been tabu-lated by Brereton et al. (1990) and are summarized in Table 4 for0.1-GPa confining pressure.

DISCUSSION AND CONCLUSIONSThree sets of experiments were conducted on the basalt sam-

ples collected during Leg 123: anelastic strain recovery monitor-

ing, dynamic elastic property measurements, and thermal azi-muthal anisotropy observations. In addition, a range of other testsand observations was recorded to characterize each sample. Onefeature common to the experimental results and observations isthat no apparently consistent orientation trend occurs eitheramong the different measurements from each core sample oramong the same sets of measurements for the various core sam-ples. However, some evidence of correspondence between veloc-ity anisotropy and anelastic strain recovery exists, but this is notconsistent for all the core samples investigated.

The thermal azimuthal anisotropy observations, althoughshowing no conclusive correlations with the other results, were ofsignificant interest in that they clearly exhibited anisotropic be-havior. The apparent reproducibility of this behavior may pointtoward the possibility that rocks retain a "memory" of their stresshistory, which might be exploited to derive stress orientationsfrom archived core.

Thin sections do not provide any evidence of crack orientation,other than that seen by eye, but electron microscopy analysisindicates (1) partly sealed cracks that have become open, (2) opencracks with no sign of healing, and (3) irregular discontinuousmicrocracks, some of which may represent localized stress re-lease. Such a mechanism of stress release should accord with theacoustic emission events observed by Wolter and Berckhemer(1989). In contrast, the rock matrix appears to be isotropic, with

488


ODP CORE 12

P-WAVE

5.60

5.72

5.38

5.75

0.1 GPa

S-WAVE3.25

3.15

3.28

3.05

ODP CORE 15

P-WAVE

6.08 ( g — - !

6.12

0.05 GPa

, 6.10

•* 6.06

k 6.06

>

6.10 ( ? [

i6.14

0.1 GPa

, 6.14

•* 6.10

* 6.13

Λ

6.35 ( g ^ —

r6.25

0.4 GPa

r 6.35

• 6.30

6.30

S-WAVE


no evidence of crystal alignment. Even though the microcracksexhibit a range of aspect ratios, the observed porosity is generallylow. Very low- (less than 0.01) or high- (vesicles) aspect ratiosdominate. Velocity variations with pressure confirmed that thebasalts are "tight," while f-wave velocities indicate significantanisotropy.

The strain recovery measuring equipment developed for thisstudy proved to be most effective and behaved without fault. Wewere able to measure strains to within about 1.5 microstrains overlong periods and to reduce to a minimum the effects of tempera-ture variation on both the equipment and the core samples. Wehave demonstrated that the effects of thermal stability within thebasalt cores are generally small relative to the magnitudes of thetotal strains recorded and also that these effects can ostensibly beignored after about the first hour of measurement. Therefore,systematic equipment errors and core-sample thermal stabiliza-tion effects can be discounted as explanations for the anomalousbehavior of some of these basalt cores.

Anelastic strain recovery is a relatively new technique. Whenexperimental results were first reported, much of the work wasbeing done using hydrocarbon reservoir rocks, and close agree-ment with other stress-measurement methods generally wasfound. Most of the early data presentations tended to follow thesomewhat "classical" power law strain recovery curves wherebycore samples expanded asymptotically with time. These and themore recent studies seem to indicate that when the method is

applied to relatively straightforward, isotropic sediments, theresults can be regarded as yielding reliable and reproduciblemeasures of stress magnitude and orientation.

Because the method has been applied to an increasingly widerange of rock types, the literature has begun to include examplesof rocks that contract with time. Explanations for this "nonclassi-cal" behavior have been difficult to verify, but strong circumstan-tial evidence exists to suggest that core-sample contractions resultfrom the slow diffusion of pore fluids from a preexisting mi-crocrack structure, permitting the rock to deflate at a greater ratethan the expansion caused by anelastic strain recovery. Scientistshad generally assumed that drilling takes place under near-bal-anced conditions, such that pore fluid pressures remain atequilibrium with drilling fluid pressures and that the permeabilityof the core material is sufficiently high to allow for maintainingthis equilibrium during core recovery. This will not be so if a lowcore material permeability prevents these differential pressuresfrom maintaining equilibrium during and subsequent to core re-trieval. Therefore, for low-permeability materials, such as silt-stones, mudstones and basalts, it is possible that the two processesof anelastic strain recovery and pore pressure recovery will com-pete with one another. The characteristics of the resulting strainrecovery curve will depend upon the magnitudes of the rockstresses, the permeability, and the pore pressure.

One of the basic assumptions of anelastic strain recovery isthat the rock is Theologically isotropic; that is, it exhibits the sametime-dependent behavior in all directions. It is evident that thesebasalt cores have been intersected by an abundance of fractures,some of which pass right through the samples, but many have beenintercepted or terminate within the rock matrix. The behavior ofthese core samples thus will be influenced, not only by theproperties of the rock matrix between the fractures, but also byhow these macro- and micro-scale fractures mutually interact.

The behavior of three core samples is worthy of note:

1. Bas 1: It is possible that the negative responses of all thegauges may be related to one subvertical and impersistent set offractures that terminate within the rock matrix, located at about180° from the scribe line.

2. Bas 12: The atypical responses of the first horizontal andvertical gauges, which initially contract but then expand, may beattributed to the fact that only they are set across a prominentsinusoidal fracture intersected by a subvertical fracture.

3. Bas 15: Although the macro-characterization of this coredescribed little sign of fracturing, the response curve of thevertical gauge showed persistent contraction.

Although correlations between the strain recovery curve re-sponses and the more complex fracture characteristics are diffi-cult, a general observation may be that an anomalous response canbe associated with a measurement across, or close to, a fracture.Yasunaga (1989) discussed the possibility that shear displacementalong an impersistent fracture may result in apparent contractionsand apparent expansions for different sets of gauges, dependingupon the locations of the gauges relative to the fracture plane.

The theoretical models reported in the literature were notapplied to the Leg 123 strain recovery results at this stage for tworeasons: (1) one must assume too many unknowns to apply theBlanton or the Warpinski and Teufel models in their current stateof development; and (2) the basalt fracture anisotropy introducesthe significant uncertainty that the responses of the individualtransducer gauges behave semi-independently and, collectively,do not represent elliptical deformation of the surface of the rockcore. However, one might find it useful to investigate the use ofsome of these models in an attempt to gain further insight into thebehavior of fractured rocks under stress relief conditions.

489


Table 4. Summary of seismic velocities and apparent elastic properties for eachsubcore at 0.1 GPa confining pressure.

Sample

BAS01-1BAS01-2BAS01-3BAS01-4BAS01-5BAS04-1BAS04-2BAS04-3BAS04-4BAS04-5BAS08-1BAS08-2BASO8-3BAS08-4BAS08-5BAS11-1BAS11-2BAS11-3BAS11-4BAS11-5BAS12-1B AS 12-2BAS12-3BAS12-4BAS12-5BAS15-1BAS15-2BAS15-3B AS 15-4B AS 15-5

Density(g/cm3)

2.842.862.872.792.852.872.872.822.862.972.882.892.882.932.912.822.722.842.802.832.822.842.852.852.842.932.942.922.932.93

Vp(km/s)

5.835.725.965.605.646.005.875.655.836.235.785.815.765.915.835.355.285.625.505.685.755.555.385.725.666.146.146.136.106.10

Vs(km/s)

3.20

3.273.082.873.05

_2.83

-3.123.253.223.052.902.932.902.773.032.882.95

.3.003.073.062.923.513.513.332.953.22

V/>/Vs

1.822

1.8231.8181.9651.967

1.996-

1.9971.7781.8041.8892.0381.9901.8451.9061.8551.9101.925

-1.8501.7521.8691.9381.7491.7491.8412.0681.894

S

0.284

0.2850.2830.3250.326

0.333-

0.3330.2690.2780.3050.3410.3810.2920.3100.2950.3110.315

0.2940.2590.3000.3190.2570.2570.2910.3470.307

K(GPa)

57.7

61.052.259.267.7

-59.9

-76.855.757.559.869.565.649.947.954.853.658.5

.53.446.757.758.762.362.466.574.968.5

µ(GPa)

29.1

30.726.523.426.7

.22.6

-28.930.429.926.824.725.023.720.826.023.224.6

.25.526.826.724.236.136.232.425.430.4

L

38.3

40.534.643.649.9

-44.8

-57.535.437.542.053.148.933.234.037.538.242.1

-36.328.839.942.638.338.345.057.948.3

E(GPa)

74.6

78.867.962.170.8

-60.1

-77.177.276.569.966.166.561.254.667.460.764.8

-66.167.669.463.990.890.983.668.679.4

Note: L = Lame's constant; S = Poisson's ratio; E = Young's modulus; K = bulk modulus; µ = shearmodulus.

In summary, the strain recovery curves recorded for each ofthe 15 basalt core samples during Leg 123 reflect the result of twocompeting time-dependent processes: anelastic strain recoveryand pore pressure recovery. If these were the only two processesto influence the gauge responses, then one might expect that,given the additional information required, the Warpinski andTeufel theoretical model could be used to determine consistentstress orientations (assuming accurate paleomagnetic core orien-tations) and reliable stress magnitudes. However, superimposedupon these competing processes is their respective interactionwith the preexisting fractures intersecting each core.

Evidence from our experiments and observations suggests thatthese fractures have a dominating influence on the characteristicsof the recovery curves and that their effects are complex.

Therefore, until such time as theoretical models can be devel-oped to cope with anelastic strain recovery of fractured rocks,results from basalt cores should be interpreted with considerablecaution. If basalts are likely to be the only suitably lithifiedmaterial available from ODP drilling sites, then core samples needto be selected carefully to exclude those exhibiting signs ofpreexisting fracturing.

ACKNOWLEDGMENTSThe authors gratefully acknowledge the financial support

provided to each institute by the NERC Ocean Drilling ProgramSpecial Topic Grants Committee, without which this work wouldnot have been possible. Also, the technical assistance and advicegiven by the staff at the BP Research Centre during the earlystages of this project are fully acknowledged. Much of the macro-characterization and dynamic elastic property measurement wasconducted by Kevin Jones, for which we are duly grateful. Thanksalso to John Ludden and Felix Gradstein, the two Co-Chief Sci-entists; Andrew Adamson, the ODP Staff Scientist; James Oggand Kazuto Kodama, who performed the paleomagnetic core

orientation measurements; and all the Leg 123 scientific party andcrew of JOIDES Resolution.

REFERENCES

Birch, F., 1960. The velocity of compressional waves in rocks to 10kilobars. / . Geophys. Res., 65:1083-1102.

Blanton, L., 1983. The relationship between recovery deformation andin-situ stress magnitudes. Soc. Pet. Eng. Pap., No. 11624.

, 1989. Discussion of "A viscoelastic constitutive model fordetermining in-situ stress magnitudes from anelastic strain recoveryof core."Soc. Pet. Eng. Prod. Eng. Pap., 281-287.

Blanton, L., and Teufel, L. W., 1983. A field test of the strain recoverymethod of stress determination in Devonian shale. Soc. Pet. Eng. Pap.,No. 12304.

, 1986. A critical evaluation of recovery deformation measure-ments as a means of in situ stress determinations. Soc. Pet. Eng. Pap.,No. 15217.

Brereton, N. R., Chroston, P. N., Evans, C. J., Hudson, J. A., andWhitmarsh, R. B., 1990. Stress in the oceanic crust: anelastic strainrecovery and elastic properties of basaltic rocks. Br. Geol. Surv. Tech.Rep., WK/9O/10.

Brereton, N. R., and Evans, C. J., 1989. Rock stress orientations fromborehole breakouts. In Louwrier, K., Staroste, E., and Garnish, J. D.(Eds.), European Geothermal Update. Proc. 4th Int. Sent. Results ECGeotherm. Energy Res. Demonstr. (April 1989). Florence (KluwerAcademic Publishers), 213-231.

Charlez, P., Hamamdjian, C , and Despax, D., 1986. Is the microcrackingof a rock a memory of its initial state of stress? Proc. Int. Symp. RockStress Measurements (Stockholm), 341-350.

Cheng, C. H., and Toksoz, M. N., 1979. Inversion of seismic velocitiesfor the pere aspect ratio spectrum of a rock. / . Geophys. Res.,84:7533-7543.

Crampin, S., McGonigle, R., and Bamford, D., 1980. Estimating crackparameters from observations of P-wave velocity anisotropy. Geophy-sics, 45:345-360.

Dally, J. W., and Riley, W. F., 1978. Experimental Stress Analysis: NewYork (McGraw-Hill).

490


Dey, T. N., and Kranz, R. L., 1988. State of stress and relationship ofmechanical properties to hydrothermal alteration at Valles Calderacore hole 1, New Mexico. / . Geophys. Res., 93:6108-6112.

Dyke, G. C , 1989. In situ stress indicators for rock at great depth. [Ph.D.thesis]. Univ. of London.

El Rabaa, A. W., 1989. Determination of the stress field and fracturedirection in the Danian Chalk. In Maury, V., and Fourmaintraux, D.(Eds.), Rock at Great Depth: Rotterdam (Balkema).

El Rabaa, A. W., and Meadows, D. L., 1986. Laboratory and fieldapplication of the strain relaxation method. Soc. Pet. Eng. Pap., No.15072.

Embleton, B.J.J., 1981. A review of the Paleomagnetism of Australia andAntarctica. In McElhinny, M. W., and Valencio, D. A. (Eds.), Paleo-reconstruction of the Continents. Am. Geophys. Union Geodyn. Ser.,2:77-92.

, 1984. Continental Paleomagnetism. In Veevers, J. J., et al.(Eds.), Phanerozoic Earth History of Australia: Oxford (Oxford Univ.Press), 11-16.

Engelder, T., 1984. The time-dependent strain relaxation of Algeriegranite. Int. J. Rock Mech. Min. Sci. Geomech. Abstr., 21:63-73.

Frocht, M. M., 1948. Photoelasticity: New York (Wiley).Holcomb, D. J., and McNamee, M. J., 1984. Displacement gauge for the

rock mechanics laboratory. Sandia Nat. Lab. (Tech. Rep.) SAND, No.84-0651.

Hyett, A. J., 1990. The potential state of stress in a naturally fracturedrock mass. [Ph.D. thesis]. Univ. of London.

Jaeger, J. S., and Cook, N.G.W., 1979. Fundamentals of Rock Mechanics(3rd ed.): New York (Chapman and Hall).

Kowallis, B. J., and Wong, H. F., 1983. Microcrack study of granitic coresfrom Illinois deep borehole UPH3. J. Geophys. Res., 88:7373-7380.

Lacy, L. L., 1987. Comparison of hydraulic fracture orientation tech-niques. Soc. Pet. Eng. Form. Eval. Pap., 66-76.

Ludden, J. N., Gradstein, F. M., et al., 1990. Proc. ODP, Init. Repts., 123:College Station, TX (Ocean Drilling Program).

Newmark, R. L., Anderson, R. N., Moos, D., and Zoback, M. D., 1985.Structure, porosity and stress regime of the upper oceanic crust: sonicand ultra-sonic logging of DSDP Hole 504B. Tectonophysics, 118:1-42.

Nur, A., and Simmons, G., 1970. The origin of small cracks in igneousrocks, int. / . Rock Mech. Min. Sci. Geomech. Abstr., 7:307-314.

Sarda, J. P., and Perreau, P. J., 1989. Determination of in-situ stressparameters from measurements on cores. CEC Rep., Brussels-Luxem-bourg, EUR 12514.

Strickland, F., and Ren, N., 1980. Predicting the in-situ stress for deepwells using the differential strain curve analysis. Soc. Pet. Eng. Pap.,No. 8954.

Swolfs, H. S., 1975. Determination of in-situ stress orientation in a deepgas well by strain relief techniques. Unpubl. Terra Tek Rept., TR-75-43.

Teufel, L. W., 1982. Prediction of hydraulic fracture azimuth from anelas-tic strain recovery measurements of oriented core. In Goodman, R.,and Heuze, F., (Eds.), Proc. 23rd U.S. Natl. Rock Mech. Symp.,238-246.

, 1983. Determination of in situ stress from anelastic strainrecovery measurements of oriented cores. Soc. Pet. Eng. Pap., No.11649.

_, 1984. Determination of the principal horizontal in situ stressdirections from anelastic strain recovery measurements of orientedcores from deep wells: applications to the Cotton Valley formation ofeast Texas. Geomechanics, 57:55-63.

, 1986. In-situ stress measurements in inclined holes in the NorthSea: applications to water flooding and enhanced oil recovery. Soc.Pet. Eng. Pap., No. 13986.

Teufel, L. W., and Warpinski, N. R., 1984. Determination of in situ stressfrom anelastic strain recovery measurements of oriented cores: com-parison to hydraulic fracture stress measurements in the Rollins Sand-stone, Piceance Basin, Colorado. Sandia Nat. Lab. (Tech. Rep.) SAND,No. 84-0512C.

Toksöz, M. N., Cheng, C. H., and Timur, A., 1976. Velocities of seismicwaves in porous rocks. Geophysics, 41:621-645.

Voight, B., 1968. Determination of the virgin state of stress in the vicinityof a borehole from measurements of a partial anelastic strain tensorin drill cores. Feismech. Ingenieurgeol., 6:201—215.

Walsh, J. B., 1965. The effect of cracks on the compressibility of rock./ . Geophys. Res., 70:381-389.

Warpinski, N. R., and Teufel, L. W., 1986. A viscoelastic constitutivemodel for determining in situ stress magnitudes from anelastic strainrecovery of core. Soc. Pet. Eng. Pap., No. 15368.

, 1989a. In-situ stresses in low permeability nonmarine rocks. / .Pet. Tech., 405^14.

_, 1989b. A viscoelastic constitutive model for determining in situstress magnitudes from anelastic strain recovery of core. Soc. Pet.Eng. Product. Eng. Pap., 272-280.

, 1989c. Authors reply to "Discussion of 'A viscoelastic consti-tutive model for determining in situ stress magnitudes from anelasticstrain recovery of core'." Soc. Pet. Eng. Product. Eng. Pap., 287-289.

Wolter, K. E., and Berkhemer, H., 1989. Time dependent strain recoveryof cores from the KTB-deep drill hole. Rock Mech. Rock Eng.,22:273-287.

Yasunaga, K., 1989. Anelastic strain recovery [M.Sc. thesis]. Univ. ofLondon.

Zoback, M. D., Moos, D., Mastin, L., and Anderson, R. N., 1985. Wellbore breakouts and in situ stress. / . Geophys. Res, 90:5523-5530.

Zoback, M. L., Zoback. M. D., Adams, J., Assumpëio, M., Bell, S.,Bergman, E. A., Blümling, P., Brereton, N. R., Denham, D., Ding, J.,Fuchs, K., Gay, N., Gregersen, S., Gupta, H. K., Givishiani, A., Jacob,K., Klein, R., Knoll, P., Magee, M., Mercier, J. L., Müller, B. C ,Paquin, C , Rajendran, J., Stephensson, O., Suarez, G., Suter, M.,Udas, A., Xu, Z. H., and Zhizhin, M., 1989. Global patterns ofintraplate stress: a status report on the world stress map project of theInternational Lithosphere Program. Nature, 241:291-298.

Date of initial receipt: 15 May 1990Date of acceptance: 19 August 1991Ms 123B-132

491

24. anelastic strain recovery and elastic properties … · this chapter describes the first...

Documents