1997 Computer-aided Structural Targeting - Holyland _ Ojala.pdf

Australian Journal of Earth Sciences (1997) 44, 421^32

Computer-aided structural targeting ¡n mineral exploration: two- and three-dimensional stress mapping. P. W. H O L Y L A N D ' A N D

V. J. O J A L A ^

^ Terra Sancta Research, 48 Peoples Avenue, Gooseherry HUI, WA 6076, Australia. ~Key Centre for Strategic Mineral Deposits, Department of Geology and Geophysics, University of Western Australia, Nedlands, WA 6907, Australia.

This paper examines a geomechanical approach to predict the location of structurally controlled hydrothermal Mississippi Valiey-type Pb-Zn mineralisation and Archaean gold mineral i sation using the Stress Mapping Technique. Two-dimensional modelling used the finite difference/distinct element code in the UDEC program and three-dimensional used the indirect boundary element method in the MAP3D program. Two-dimensional. regionalscale, stress modelling of the Lennard Shelf, Western Australia shows that most of the known Mississippi Valleytype deposits and prospects are localed within or near the modelled mínimum principal-stress anomalies using the gross geology of the Lennard Shelf and an east-northeasl-west-southwest extensión direction. Three-dimensional stress modelling was used lo model stress tields around the easlern margin of the Granny Smith Granodiorite under east-wesl compression at the Granny Smith mine, Laverton, Western Australia. At a deposit scale, the patterns of the simulated mínimum principal-stress correlate well wilh the known áreas of gold mineralisation near the contact between the Granny Smith Granodiorite and sedimentary rocks. Key words: computer modelling, mineralisation, Mis.sissippi Valley-type deposits, stress, structure.

INTRODUCTION Many types of hydrothermal ore deposits are formed in structurally controlled sites in permeable fracture systems. Typically, mineralisation occurs in discrete segments of individual structures and within the mineral deposits some parts of the host structures are belter mineralised tban otbers. Epigenetic mineral deposits are related, in general, to structurally focused fluid flow during active deformation (Hodgson 1989). Hronsky et al. (1990) reviewed the common structural controls on localisation of mineralisation and oreshoots: (i) the intersection of host structure with a particular lithological unit (e.g. banded iron-formation-hosted gold deposits); (ii) intersection of two synmineralisation structures; (iii) dilationai jogs, divergent bends in faults or en echelon fault segmentation; (iv) fold hinge zones; (v) flexures in the host structure witb axes oblique to, and commonly at a high angle to, the movement direction; and (vi) zones that plunge subparallel to the stretching lineation but in which the specihc controls on their location are unclear. If only one control applies to mineralisation, it may be possible to predict the location of the next mineralised zone in a particular área or structure. However, commonly several controls interact, and the complex three-dimensional geometry of the controlling structures makes interpretation very difficult. In addition to geometry, competency contrasts and rheological properties of the host lithologies may have a signitícant influence. Therefore, even a purely kinematie approach to structural interpretation of a geological map is commonly based more on the interpreter's intuition than true geometrical relationships. Several computer codes have been devel-

oped for geomechanical purposes to model structural behaviour of complex rock masses and this paper examines the applications of two such codes in computer simulations to achieve more consistent structural interpretations and for locating structurally controlled Mississippi Valley-type Pb-Zn mineralisation and Archaean mesothermal lode-gold deposits. However, the modelling is applicable to geometrically similar deposits of other mineral commodities. This paper demónstrales that a geomechanical approach using two and three-dimensional computer simulations (Holyland i 990a; Oliver el al. 1990) is a practical way in which to attempt to predict the locations of the dilationai sites which result from deformation of a complex rock and fracture geometry. Regional twodimensional stress modelling is applied to a part of the Lennard Shelf in Canning Basin, Western Australia, which hosts significant structurally controlled Mississippi Valley-type mineralisation (Eisenlohr et al. 1994; Vearncombe et al. 1995). A deposit-scale three-dimensional modelling example is the Late Archaean gold mineralisation at the Granny Smith mine in the Eastern Goldfields Province, Western Australia.

METHODS AND T H E O R E T I C A L CONSIDERATIONS Fluid focusing Hydrothermal ore deposits are characterised by channellsed fluid flow and high fluid/rock ratios. Under low- lo medium-grade metamorphic conditions, fluid

422

P. W. H O L Y L A N D A N D V. J . O J A L A

pressures are buffered cióse to lithostatic pressures and fluid flow is generally upward directed (Etheridge et al. 1984). Focusing of the upward fluid flow into a discrete channelway. as required to form an ore deposit. is due to lateral varialions in hydraulic head. These lateral gradienis may be induced structurally by either variations in fracture permeability of active fault zones. or by variations in mean rock stress. Mineralised extensión veins indicative of supralithostatic fluid pressures are commonly present in mesothermal gold systems and these are compatible with fluid focusing into zones of low mean rock stress, where relative fluid pressures are higher than the local rock pressure but absolute fluid pressures are lower than in surrounding rock {Ridley 1993). Sites of low mean stress can, therefore, be simuitaneously sites of fluid focusing and of low effective mean stress. At higher crustal levéis ambient fluid pressures are closer to hydrostatic pressures but fluid pressures in fluid channels might be lithostatic and both relative and absolute fluid pressures higher than in the surrounding rock. Consequently. fluid flow is strongly controlled by permeability. Although high permeability patbways can be lithologically determined, structurally controlled ore deposits are more common in crystalline rocks and, in general, they have formed in reactivated fault systems (Phillips 1972; Sibson et al. 1975; Sibson el al. 1988). A failure of a pre-existing weakness. or intact rock, can be initiated by an increase of differential stress, by an increase in the pore-fluid pressure, or by combinalion of both. In most cases, stress changes lead to decrease of mean stress or, more commonly. decrease of minimum principal stress. A model in which fluid focusing in the crust is due to variations in mean stress, or in which failure of preexisting weaknesses and, therefore, enhanced fracture permeability, is due to lowered mean or minimum stress, allows for the wide variety of structural settings of mineral deposits. Tbus, a technique that measures variations in rock stress has the potential to genérate viable exploration targets.

intact rock, will fail along that plañe of weakness. Analysis of the likely mode of failure requires consideration of the orientation of the plañe relative to tbe imposed stress. As for the case of intact rock, failure may be in one of three modes—extensión hydrofracturing. sbear failure. or extensional shear failure. Generally, there will be a range of orientations of planes for which failure will occur (Figure 2 ) before fracturing or shearing of adjacent intact rock. Puré extensional hydrofracturing in a plañe perpendicular to o"3 will occur at low differential stress states in which the Mohr circle touches the failure envelope at the normal stress axis {Figure I). In this case, the fluid pressure exceedsCT3plus the tensile strength of the fracture. With increasing fluid pressure or differential stress, the range of orientations of planes which can fail increases. As long as the intersection between the Mohr circle and the failure envelope is in the tensional field. failure will involve by dro fracturing and shearing. During deformation, failure along the plañe or Ihrough intact rock can be induced by: (i) increasing the fluid pressure and henee reducing ihe effective siresses and moving the Mohr circle to the left (Figure 2a);

03

a3_

(b)

di

normal stress

shear stress o3 al

jj a3

oí

i normal stress

Stress-fluid pressure relationship at fracture initiation Three different modes of brittie failure for homogeneous. isotropic, intact rock are shown in relation to loading stress fields on the Mohr diagram in Figure 1 under condition of high relative fluid pressure (Sibson 1989). The type of failure depends on the magnitude of the differential stress relative to the tensile strength of the rwk. Depending on the stress state and the shape of the failure envelope. failure may be by: (i) extensional hydrofracture perpendicular to 0-3; (ü) shear failure along conjúgate zones about 30° to t r l ; or (iii) extensional shear failure as conjúgate zones at, or less than, 30° to (TI (Sibson 1989).

Figure 2 shows the effect of a plañe of weakness. for instance, on failure using the Mohr-Coulomb failure criterion. The failure envelopes are constructed assuming any orientation of the plañe of weakness. A stress regime in which the Mohr circle overlaps the failure envelope for planes of weakness, but does not touch the envelope for

Figure I Minies of failure in relation to differential slress in homogeneous, isotropic rock. Mohr diagrams illuslrale ihe general failure envelope for intaci rock and the stress condilions for the three modes of failure. Modirted from (Sibson 1989). (a) Extensional fracturing (hydraulic fracturing). (b) Extensional shear (conihination of hydraulic fracturing and shear). (e) Shear fracturing. ( r l . M á x i m u m principal stress; {r3, minitnum principal stress; 6, angle o f failure to t i l ; , tensile strength ol' the rock.

STRESS M A P P I N G I N M I N E R A L E X P L O R A T I O N (ii) increasing a l and henee increasing mean and differential stresses (Figure 2b); ( i i i ) increasing the differential stress at constant mean stress [i?im = a l + (J3)/2] (Figure 2c); and/or (iv) decreasing a3 and henee decreasing mean and increasing differential stresses (Figure 2d). The diagrams in Figure 2 demónstrate that a change in stress State ffom stable to unstable results in a decrease of mean stress i n cases (a) and (d), an increase in case (b) and a constant mean stress in case (c). However, the minimum principal stress (a3) is constant only in case (b) and decreases in all other cases. Therefore, it is assumed in the analysis that variations in a3 give a better indication of proximity to failure than the mean stress.

423

mean [ { o m ^ í o l a2-i-a3)/3] or the minimum principal (CT3) stress, provided that the orientation of the farfield stress can be determined or a reasonable orientation assumed. The most severe limiting assumption of two-dimensional stress modelling is that the plañe of a map does not

Range of orientations of weak planes possible lo fail (= 35=)

S T R E S S MAPPING

leakness

Stress mapping examines the variation in strain and stress through an inhomogeneous terrain on imposition of a regional stress field. When stressed, an inhomogeneous material develops an inhomogeneous stress field whose components vary witb rheological properties and geometry. The modelling is considering only elastic and elasto-plastic stresses and strains, and, for instance, deformation dilatancy due to viscous strains is not considered. Stress mapping as a basis for prediction of hydrothermal fluid flow is based on the following reasonable assumptions: (i) low minimum principal (a3) stress indicates proximity to failure and therefore possible deformation-enhanced permeability (this is more important in modelling of high crustal level deformation); (ii) at depths of more than a few kilometres fluid pressure is buffered to be cióse to lithostatic pressure and the control on fluid pressure is mean stress; (iii) variations in mean stress will be followed by variations in fluid pressure; and (iv) fluid flow is both upwards and towards zones of low mean stress. Combined with the knowledge that structurally controlled mineralisation is commonly late in the tectonic history of a terrain and that typical Archaean lode-gold deposits and Mississippi Valley-type deposits show evidence of fluid overpressuring (Archaean: Groves el ai 1995; Mississippi Valley-type: Eisenlohr É-Í a/. 1994), this enables stress analysis of two-dimensional map patterns of rock units and faults to predict those zones of low

Figure 2 Schematic Mohr diagram,s illu.strating possible changes in the stress state which can lead lo failure of a plañe of weakness. (a) Increasing fluid pressure. for the stress state 2, tensile failure w i l l occur along fractures perpendicular to CT3, and shear failure along a wide range of orientations of pre-existing planes o f weakness. (b) Increasing a l leads to increasing differential stress and .shear failure w i l l occur when Mohr circle touches the failure envelope. (c) Increasing differential stress with constant mean stress leads to shear failure. (d) Decreasing a3 leads lo increasing differential stress and shear failure.

424

P. W. H O L Y L A N D A N D V. J. O J A L A

accurately refiect the stress pattem in an área with complex three-dimensional geometry (Holyland e! al. 1993). In many terrains. such as the Yilgam Craton, this may not be a eritieal restriction since the vast majority of the observed structures are upright and the majority of gold deposits are hosted in steep structures, although significant exceptions do occur (Hronsky el al. 1990; Libby el al. 1990). Tbe accuracy of the three-dimensional geological interpretation becomes especially important in deposit-scale modelling where the extent of interest of the vertical dimensión is similar to horizontal dimensions.

resulls in áreas of lower stress (i.e. dilation zones) which represent favoured áreas for mineralisation and therefore targets in exploration. (2) The production of a stress map requires estimates of the magnitudes and orientations of the far-field horizontal stresses. (3) RíK'k deformation properties including strength and moduli, and friclion angle and stiffness for fault deformation are required for modelling. Example of two-dimensional stress mapping: Lennard Shelf

Computer programs Two-dimensional modelling is by a distinct element code (UDEC program) and Ihe method has three distinguishing features which make it well suited for discontinuum modelling (Holyland I99()a, b). These are: (i) the method simulates an assemblage of blocks which interact through córner and edge contacts; (ii) discontinuities are regarded as boundary interactions between these blocks and discontinuity (fault) behaviour is prescribed for these interactions; and (iii) the method utilises an explicit timestepping (dynamic) algoritbm that does not limit displacements or rotations, and general non-constitutive behaviour for both the matrix and discontinuities. Modelling can be done at any scale and in addition to minimum. máximum and mean stresses, total displacement and fault displacements are computed. The total displacement is the amount of movement experienced by Ihe rock blocks. and Iheir movement directions. MAP3D® versión 1.29 (Mining Analysis Program in .íDimensions, 1993; Copyright Mine Modelling Limited, Copper Cliff, Ontario) is used in the three-dimensional stress analysis to simúlate rock-mass response under the imposed external stress. The rock-mass can include múltiple zones with different moduli, and fault slip as well as crack opening can be simulated. The formulation of the code is based on the Indirect Boundary Element Method of Banerjee and Butterfield (1981) and incorporales simultaneous use of both lictitious forcé and displacement discontinuity elements. The elastic rock-mass may contain múltiple nonhomogeneous regions and can be intersected by múltiple fault planes and joint sets. Results of the simulations are presented on user-specified grids which may slice through the three-dimensional model at any desired location and orientation.

A 1:250 (XX) solid geology interpretation of an área including Mississippi Valley-type deposits at Cadjebut, Blendevale and Twelve Milc Bore produced by Dorling (1995) as a part of his PhD study was used in the twodimensional simulation (Figure 3a is a simplified scale reduction of this map). Structures were divided according to their size as hrst or second order. First-order structures include regional faults such as the Pinnacles Fault and have strike lengths over 10 km; second-order structures are generally under 10 km in strike length. First-order structures were modelled as having lowest stiffness. Rock types include conglomérate, shale, platform limestones, and basement granite. The basement granites were treated as the most competent rock type. followed by the platform limestones. shale and hnally unconsolidated conglomérate (for commercial reasons the exact parameters are not published here). The máximum principal stress (a 1) direction used in the simulation is at 110-290°, equivalent to an extensión al 020-200° based on the regional geology of the Fitzroy Trough and Lennard Shelf. Sites of low minimum principal stress ((r3) simulated in the model for the southeast Lennard Shelf are shown in Figure 3b. Low-stress anomalies show a cióse correlation with known sites of Mississippi Valley-type mineralisation and a significant anomaly occurs near Blendevale. The Cadjebut deposit has a small, less intense low-stress anomaly, and at Twelve Mile Bore there is again a prominent anomaly. In addition to Blendevale, Cadjebut and Twelve Mile Bore, other sites of known mineralisation which correspond with lowstress anomalies include the Brooking Spring, Brooking Springs Station, Fossil Downs. Virgin Hills, and Gap Creek prospects.

Input for stress mapping

Example of three-dimensional stress mapping: the Granny Smith mine

Stress mapping requires tbe following input data: (I) An interpreted, accurate geological map or threedimensional geological model. Geological map data are converted to a solid geology interpretation which provides continuous lithological and structural information (Figure 3a). The study área is treated as a mosaic of polygonal blocks (rock units) and joins (faults and shear zones). When external stress is applied to this system the blocks are juggled and internally deform until equilibrium is attained. This juggling and intemal deformation

At Granny Smith, gold mineralisation is located along a north-south-striking, moderately east-dipping, reverse fault. which purtiy follows a gran i lo id-sed i me ntary rock contact (Figure 4). In different sections of the fault, mineralisation may be developed in the sedimentary rocks, in the granitoid and/or along Ihe contact between them. The orientations of conjúgate sets of mineralised fractures in the granitoid are variable, indicating that the ItK'al stress field was heterogeneous and the orientation of the máximum principal stress í
Figure 3 (a) Digiiised solid geology interpretation of the Lennard Shelf between Fitzroy Crossing and Cadjebut (after Dorling 1995). The geology is simplified to four mechanical units, the basenieni granites are treated as the most competent rock type, followed by the platform limestones. shale and finally unconsolidated conglomérate. First-order structures are modelled as having lowest stiffness. The m á x i m u m principal stress r r l was simulated at 110" based on the regional geological history of the Fitzroy Trough and Lennard Shelf (b) Plot showing contoured valúes of minimum principal stress (CT3); only the contours below far-field minimum principal stress are shown. Anomalously low valúes o f rr3 show a cióse correlation w i l h known sites of Mississippi Valleytype mineralisation. A significant anomaly occurs near Blendevale. with the deposit positioned approximately in the northwest .segment of the low stress anomaly. Other sites o f Mississippi Valley-type mineralisation which correspond with low (r3 anomalies include the Brooking Springs prospect, Brooking Springs Station prospect, Cadjebut, Fossil Downs prospect, Virgin Hills prospect, Twelve M i l e Bore deposit and the Gap Creek prospect.

i

S T R E S S MAPPING IN M I N E R A L E X P L O R A T I O N

425

+ + + + + + + + + + + + + + +

+ + + + + + + + + + + + + + + + + + + + + + -

'/

^ " ^ + + + + + + + + + H • + + + + + + + H

& + + + + +-i

20 km Lithofogical contact

\

First order fault —

Second order fault Faull continualion

Platform facies

Twelve Mrle Bore

Conglomérate Basin facies

Zn-Pb prospect Zn-Pb deposit

^ + ^ Precambrian granite

426

P. W. H O L Y L A N D A N D V. J . O J A L A

from the far-field east-norlheasl-west-southwest orientation. It i.s interpreted that these variations were controlled by the geometry of the irregular granitoid contact (Ojala 1995; Ojala el al. 1993a, b). After mining and extensive drilling at the Granny Smith mine, it has been possible to construct a detailed three-dimensional geological model of the mineralisation and the eastern contact of the Granny Smith (Figure 5). As mentioned above, the structural observations indícate that the local stress field was heterogeneous, therefore, this deposit should be a good test for three-dimensional stress analysis on the scale of a single mineral deposit. The main aims of the computer simulation were to test correlation between: (i) computed minimum principal stress lows and dilations and the áreas of mineralisation at the granitoid margin; and (ii) computed variations in local stress fields under

east-northeast-west-southwest compression and observed variations of orientations of mineralised structures near the granitoid margin. GEOLOGICAL

MODEL

For the geological model, a simple linear triangulation was used to interpólate the granitoid contact between the dnllhole intersections and mapped contact locations to créate the eastern contact of the Granny Smith Granodiorite. To minimise boundary effects in the área of interest, the granitoid block was extended from the present erosión surface 420 RL (metres above sea level) to 600 RL and from the deepest drillhole intersection at 54 RL to -800 RL. The dimensions of the granitoid model used in the simulations are about 1.4 km high, 2.1 km long and 1.4 km wide, and it contains 938

6814000N-I-

Fault Plunging aniilorm Plunging synform Bedding Foliation Youging direction Lamprophyre dyke

TU

Granitoid Porphyry dyke Banded iron formation Clastic sedimentary rocks Conglomérate

Figure 4 Geology of environs o f 500 m

ihc Granny Smith mine showing the distribution of main rock types and major structures.

Figure 5 Contours o f gold grade thickness (sum of gold assays above 0.5 ppm along the hole, contact ± 50 m) and inferred máximum principal stress direction w i t h respect to the structure contour map of the eastdipping contact between the granitoid and sedimentary rocks at Granny Smith. The fracture vein orientations discussed in this study have been measured from oriented drillcores and from the south and west walls of the Granny pit. Diamond drillcores have been oriented using downhole spear, and holes have been surveyed using Eastman single-shot camera to record dip and azimuth o f the hole. Oriented drillcores intersect most of the eastern contact of the granitoid to a deplh o f about 1.50 m.

STRESS M A P P I N G I N M I N E R A L

EXPLORATION

427

Structure contours of the granitoid surface

6813000N

Au gram metres (0.5 ppm cut off) 30om-

6812600N

6812600N

6812800N

6812800N,

_

e8i2200N

6812200N

6812400N

6812400N

-

-

6812000N

6811800N

-

6811600N

-

-

443300E

PROPERTIES

6812000N

6811800N

_

-

6811400N

6811200N

Windich

6811600N

6811400N

6811200N

.A20-__

1

6811OOON

surfaces defined by 237 control points (i.e. diamond drillhole intersections and mapped contacts). MATERIAL

The rock properties used for the simulation are shown in Table 1. Physical properties of rocks were estimated using the compilations of Clark (1966) and Engelder

443506f'

443700E

6811OOON ' 443900E

4441OOE

(1993). All materials were modelled as Coulomb materials. Young's modulus of 50 x 10^ Pa was used for the Granny Smith Granodiorite and is typical of granilic rocks. Assuming Young's modulus ratio of two, as used by Stromgárd (1973) in his photo-elastic and finiteelement models of a strong body within a weak matrix, this gives a valué of 25 x lO'' Pa for sedimentary rocks. This valué can be considered a reasonable approximation

428

P. W. H O L Y L A N D A N D V. J. O J A L A

Table 1

Sumniiiry of materiiil properties for the stress simulation of a granitoid intrusión in a sedimentary rix;k host at Granny Smilh. Valúes are approximated from the compilations of Clarke (1966). Granitoid Young's tiiodulus (MPa) Poisson's ralio Nonnal stiffness (MPa) Shear stiffness (MPa) Tensile strength (MPa) Compressive strength (MPa) Cohesión (MPa) Hriction angle (")

50 (XK) 0.25

-

-4 30 6 20

Sedimentary rock

25 000 0,20

-

-

3 20 4 20

Lithological conlact*

_

-

5

omn 0 0 0 10

Contact between the granitoid and sedimentary rock.

as it is within the lower end of the range measured for sbaies and siltstones (Clark 1966). These valúes define a relative rheological scale. such that the granitoid was stronger than the surrounding sedimentary rocks. The rock units were assumed to be homogeneous and their material properties isotropic and constant. The lithological contact was modelled to be weaker than the surrounding rocks.

INITIAL

STRESS

STATE

The far-field minimum principal stress (a3) was defined as vertical and the máximum and intermedíate principal stresses (al and a2) to be horizontal and oriented 080° and 170°. respectively. These are ba.sed on tbe orientations of regional foliations and the dominant mineralised fracture vein orientation within the Granny Smith Granodiorite (Ojala 1995). The valúes for a l , a2 and a3 were 100 MPa, 80 MPa and 50 MPa, respectively. Assuming a hydrostatic pore pressure, the 50 MPa confining pressure (i.e. a3) corresponds approximately to the effective stress at a depth of about 4 km (Engelder 1993). A far-field differential stress of 50 MPa, used to simúlate deformation during mineralisation, is within the lower end of the range of the differential stress estimated for ibrust faulting (Engelder 1993). This was considered as a reasonable assumption considering that the observed displacements along mineralised fractures in the granitoid are at most on a decimetre scale (Ojala 1995). When a much higher differential stress was u.sed, the computer model became very unstable and simulation a lower differential stress yielded similar results but witb longer computing times.

RESULTS

Minimum stress (a3) patterns and a l orientations produced after 10 load steps. which resulted in about 0.5 m máximum displacement along the contact, are shown in Figures 6 and 7, respectively. The most obvious result of the deformation simulations is, as expected for a thrust-fault regime. that a3 valúes are low in the áreas of sballower dipping

granitoid contad. TheCT3valúes are also low within the granitoid cióse to áreas of steeper dipping contact and a high a3 anomaly is developed in Ihe sedimentary rock adjacent to the contact at these siles. Discrete low-a3 anomalies also occur in locations where the change in dip is large in steeply dipping arcas. In addition, within the áreas of a shallowly dipping contad, variations in the magnitude of a3 are correlated witb small changes in the shape of the surface. Within the shallowly dipping but smooth parts of the contact. the a3 anomalies are not as strong as in the áreas in which the contact is concave or convex. The simulated stress field is also heterogeneous wilh résped lo a l stress orientations. These helerogeneities for the deformation simulation at the 380 RL are shown in Figure 7. Variations in the calculated a l orientations are greatest in the áreas within the Granny pit área where the contact is shallowly dipping and irregular. These áreas correlate with áreas of low a3 valúes. At the Windich deposit, the contact shape is less complex and, although there are up lo 90° variations in the modelled CTI orientation, variations are restricted to very cióse to the contad and cover smaller volumes than in the Granny deposit. Importantly, the modelled a l orientations correlate very well with a l orientations inferred from fracture vein orientation measurements (Figure 7). Figure 6 shows ihat low a3 correlate with the gold mineralisation over the eastern contact of the Granny Smith Granodiorite. The anomaly east of the Windich pit is at a deeper level than the pit and it is solely a result of the small área of shallowly dipping contact in the threedimensional geological model which is outside of the drilling data. There is an especially good correlation between the wide low-stress áreas in the granitoid and the wide low gold-gradc halo. In cross-section, this broad low-stress volume. which develops in the granitoid where the contact sharply .steepens, correlates with a wide mineralised zone of fracturing of lower average grade than deeper levéis in the deposit (Figure 8). In summary, the deformation simulation suggests that: (i) the gold mineralisation is structurally controlled, especially by the shape of the granitoid surface; (ii) the máximum compressive far-field stress was oriented at about 080-260°; and (iii) the variations in the orientations of mineralised

Figure 6 Contours o f CT3 (shaded) and gold grade thickness illustrating the good correlation of the simulated low-stress áreas and gold mineralisation. Pit outlines show the broad extent of economic mineralisation.

STRESS M A P P I N G I N M I N E R A L E X P L O R A T I O N 443300E

443500E

443700E

443900E

429

444100E

_

6813000N

6812800N

6812400N

-

6812600N

-

6812200N

6811800N

-

6812000IM

_

6811600N

6811400N

6811200N

6811400N

_

6811200N

6811OOON

6811OOON 443300E

fracture veins might be related to heterogeneous local stress fields which could have resulted from irregularities in the shape of the granitoid surface.

DISCUSSION The two-dimensional, regional-scale stress modelling of the Lennard Shelf between Fitzroy Crossing and Cadjebut

^

443500&Í —

443700E

^

443900E

4441 OOE

shows the ability of stress-mapping technology to simúlate zones of low minimum principal stress (dilation zones) which correspond with known Mississippi Vallcy-typc deposits, suggesting a structural control on fluid focusing. The modelling of palaeo-stress patterns on the Lennard Shelf supports the model in which the fluid flow and location of mineralisation is largely the result of gross regional structure and rheological variation of different rock types during 020-200°-directed extensión.

430

P. W. H O L Y L A N D A N D V. J. O J A L A 443300E

6813000N

443500E

443700E

443900E

-

6812800N/^

-!.

6812600N

6811400N

_

444100E

-

6813000N

_

6812800N

-

6812600N

-

6812400N

-

6812200N

-

6812000N

_

6811800N

-

6811600N

-

6811400N

-

6811200N

Figure 7 CTI orientations at 3 8 0 6811200N

6811OOON

443300E

443500E

443700E

443900E

At Granny Smith, the genera! patterns of the simulated three-dimensional minimum principal stress correlate very well with gold mineralisation. Also the computed variations of the a l orientations correlate with the observed variations of a l as determined by orientations of mineralised conjúgate fractures. This also suggests that the important input parameters (three-dimensional geo-

444100E

6811OOON

RL calculated using the M A P 3 D ® program and cr 1 orientations inferred from conjúgate fracture vein measurements. Note that the greatest variations in inferred and calculated orientations occur in the same áreas.

logical model, rock parameters. stress orientations) are realistic. In some respects. this correlation is quite surprising as the modelling was completed only to simúlate the stress pattem before failure. This indicates that the low-stress zones, where failure is likely to occur first, remain as low-stress zones even after failure and during deformation. It also suggests Ihal no through-

Figure 8 Seclion through the northern part o f the Windich deposit showing the gold mineralisation and low cr3. Note the widening of the gold mineralisation and the low-o-3 área within the granitoid where the dip of the contact steepens. The low-(r3 anomalies deeper in the section are the result of a curve on the modelled surface caused by interpolation beyond drillhole intersections.

STRESS M A P P I N G I N M I N E R A L E X P L O R A T I O N West

431 East

• i . \ S . - S S S \

+ + + + -1- -1- -1-1- 4-

4+

+ + + -1-

-1- - i -

Drill hole fQQ \,\\

Gold mineralisation (> 0.7g/t) Plagiodase porphyry

^

\ * * * \d E

Sedimentary rocks

100

goiíig failures, which would have significantly changed the geometries of structures or rock units, formed during the mineralisation. The results further supporl the conceptual model that low rock stress focuses the ore fluids (Holyland 1990c, d; Ridley 1993). Two main explanations for this correlation can be drawn: (i) variations in mean rock stress result in gradients in fluid pressure; or (ii) rising fluid pressure decreases effective stress and tbe tensile or extensional shear failure occurs first in the áreas of low o"3, which, following failure, are more permeable. Numerical modelling of rock-mass deformation is not an exact science and there will always be parameters of the rock-mass which are not understood. In other words, rock mechanics models are 'data-limited' and there is seldom enough data to simúlate the rock-mass behaviour unambiguously (StarHeId & Cundall 1988). However. two-dimensional modelling of the Lennard Shelf indicates that good results, which are useful in exploration, can be obtained with a quite limited geological datábase. Furthermore, the results of three-dimensional stress modelling using the Granny Smith data show that it is possible to simúlale a stress distribution that is very similar to the stress distribution which is inferred from field observations of fracturing. Critical to the successful modelling of stresses during epigenetic mineralisation and deformation are reasonably accurate input parameters (rock propenies, far-field stress field, and especially the three-dimensional geometry).

m

Three-dimensional

stress

modelling

formed

part

of

V J O ' s P h D p r o j e c l w h i c h was s u p p o r t e d b y U n i v e r s i t y o f W e s t e r n A u s t r a l i a and A u s t r a l i a n Overseas Postgraduale R e s e a r c h s c h o l a r s h i p s , s u p p l e m e n t e d by an

Outokumpu

S a a t i o s t i p e n d . T h e p e r m i s s i o n t o p u b l i s h these data f r o m Placer Pacific L i m i t e d and D e l t a G o l d N L is g r a t e f u l l y acknowledged.

REFERENCES BANERJEE P. K . & BUTTERFIELD R . 1981. ¡ioimJary Element Meíhods in Enf>ineering Science. McGraw-Hill Book Company L t d . London. C L A R K S. R 1966. Handbook of Physical Conslants. Geological Society o f America. New York. DORLING S. L . 1995. Structural Evoluiion of Fit/roy Trough and Lennard Shelf during the Devonian-Early Carboniferous Pillara Extensión Pha,se: Implications for facies distribution and Mississippi Valley-type deposit formation. PhD thesis, University o f Western Australia. Penh (unpubl.). EISENLOHR B . N . , T O M P K I N S L . A . , CATHLES L . M . . B A I Í L E V M . E .

& GROVES D . I . 1 9 9 4 . Mississippi Valley-type deposits: producís of brine expulsión by eustatically induced hydrocarbon generalion? A n example from northwestem Australia. Geoloí-y 22.315-318. ENGELDER T . 1993. Stress Redimes in the Lithosphere. Princeton University Press. Princeton. ETHERltx¡F. M . A.. W A L L V. & Cox S. F 1984. High Huid pressures during regional meiamorphism and deformation: implications for mass transpon and deformation mechanisms. Journal of Geophysical Research 89, 4344--f358. GROVES D . I . . R I D L E Y J. R., B L O E M H . M . J. ICTAI.. 1995. Lode-gold deposits of the Yilgam Block: producís of Late-Archaean crustal-scale overpressured hydrothermal systems. Geological Society of London Special Puhlication 95, 155-172.

ACKNOWLEDGEMENTS Thanks are due to Simón Dorling for providing the Lennard Shelf solid geology map for the mtxielling. Comments by Mark Barley, William Power and John Ridley improved the manuscript considerably. We thank David Groves for the support al the Key Centre, Department of Geology and Geophysics and Terra Sancta Research for the u.se of equipment and permission to publish the results.

HODGSON C , J. 1989. The structure of shear-relaled. vein-lype gold deposits: a review. Ore Geology Reviews 4. 231-273. H O L Y L A N D P. W . 1990a. The nature of the lithosphere: cracks and blocks? Terra Sancta, Perth. H O L Y L A N D P. W . 1990b. Simulation of the dynamics of Archaean deformation in the Yilgam Block. Western Australia, ¡n: Glover J. E . & Ho S. E . eds. Third ¡niernaiiontil Airlitiean

432

P. W. H O L Y L A N D A N D V. J. O J A L A

Syinpri.siiim. Perth. Abstraéis (WA) Inc.. Perih.

pp. 347-349.

Geoconferences

HOLYI.AND P. W. I99()c. Stress mapping in Ihe M i . Isa región. In: Mí ¡su Inlier Geolof-y Confereme, Melhourne, Absiracis. pp. Monash University. Melbourne. H O L Y L A N D P. W. 199üd. Targeting o f epithermal ore deposits using stress mapping techniques. In: Proceeilings Pacific Rim 90 Congress. Melbourne. I I I , pp. 337-341. Australasian Institute of M i n i n g and Melallurgy, Melbourne. H O L Y L A N D P . R I D L E Y J . R . & V E A R N C O M B E J . R . 1993. Stress

mapping technology, In: Parnell J., Ruffel A . H , & Molos N . R . eds. Geofiuids '93:Contributions to au Internatiomd Conference on Fluid Evoluiion, Migration and Inleracíion in Rocks. Torquay. England, Ahsiracts. pp. 272-275. Geological Socieiy of London. HRONSKY J . M . A . , CA,SSIDY K . R . GRIGSON M . W . ICT AI.. 1990.

Deposit- and mine-scale structure. In: Ho S. E . , Groves D . I . & Bennet J. M . eds. Gold Deposits of the Arvhuean Yilgam Block. Wí'.vfíTíí Australia: nature. génesis and exploration guides 2 0 , pp. 38-54. University of Western Australia, Perth. LIHRY

J- W . . B A R L E Y

M . E . , EISENLOHR B . N . . GROVES

D.L.

HRONSKY J, M , A . & V E A R N C O M B E . J . R . 1990. Cralon-scale

deformation /ones. In: Ho S. E., Groves D. I . & Bennet J. M , eds. Gold Deposits of the Archaean Yilgam Block. Western Australia: nature. génesis and e.xploration giiides, 2 0 , pp. .30-37. University of Western Australia. Perih. OJALA V , J. 1995. Siruclural and depositional controls on gold mineralisation at the Granny Smilh Mine. Laverton, Western Australia- PhD thesis, University o f Western Australia. Perth (unpuhl.). OJALA V . J.. R I D L E Y J . R . . GROVES D . I . & H A L L G . C . 1993a. The

lode-gold deposit. Geological 3 4 . 55-56.

Society

of Australia

O i . l V E K N , H . S - VALKNTA R , K . & W A L L V . J , 1990. The effect o f

heterogeneous stress and sirain on metamorphic fluid flow, Mary Kalhleen, Australia, and a model for large-scale fluid circulalion. J
R . H . . MOORE

R. M . & RANKIN

A . H . 1975. Seismic

pumping—a hydrothennal Huid transpon mechanism. ofthe Geological Society of Umdtm 1 3 1 . 653-659. SIBSON

R . H . . ROBERT F. & POULSEN K . H . 1988, High

Archaean

angle

STAREIELD A , M . & C U N D A L L P, A . 1988, Towards a meihodology

for rock mechanics modelling, Iniernaiional Journal Mechanics and Mining Sciences and Geomechanics 25.99-106.

of Rock Abstraéis

SlROMílARD

K, E, 1973, Stress distribution during deformation o f boudinage and pressure shadows. Tectonophysies 16, 21.5-248,

V E A R N C O M B E J . R , . D E N T I T H M , C . D O R L I N G S, L , ET M.. 1995.

Regional- and prospect-.scale controls on Mississippi Valleytype Zn-Pb minerali/alion at Blendevale, Canning Basin, Western Australia. Economic Geology 9 0 , 181-186.

OJALA V . J., R I D L E Y J . R . . GROVES D . I . & H A L L G . C . 1993b.

Smith: an example o f a graniloid-hosted

Journal

reverse faults. fluid-pressure cycling. and mesothermal goldquartz deposits, Geology 1 6 . 551-555.

Granny Smith gold deposit: the role o f helerogeneous stress distribution al an irregular granitoid contact in a greenstone facies lerranc. Mineraliuni Deposita 2 8 , 4 0 9 ^ 1 9 .

Granny

Abstraéis

Received

t

22 April

¡996: aecepied 2 July

¡996