SAMPLE PREPARATION TEC HN IQU ES FOR THE TRIAXIAL TESTING OF CEMENTED ROCKFILL

TECHNICAL REPORT NO. 37

L.P. GONANO

0

SAMPLE PREPARATION TECHNIQUES FOR THE TRIAXIAL TESTING OF CEMENTED ROCKFILL

TECHNICAL REPORT NO. 37

L.P. GONANO

KEYWORDS:

Triaxia.I tests, mine fills, underground me t al mining, physical models, size effects, Mount Isa Region, Qu eensland.

ABSTRACT:

Laboratory methods of producing cemented rockfiLI suitable for use as large scale triaxial test samples (380 m diameter) are discussed. It is proposed that triaxial testing of such samples will provide data for material optimization and stability analysis. A program for the final investigation of material properties is given.

The theoretical requirements for the material model.ling of cemented rockfill are not fully understood, therefore modelling is r estricted to the reproduction of the various in s i tu structures using the same components and s imilar proportions. As index properti es of the sampl es are similar to the in situ materials the major requirements of modelling have been met.

DIVISION OF APPLIED GEOMECHANICS COMMONWEALTH SCIENTIFIC AND INDUSTRIAL

RESEARCH ORGANIZATION AUSTRALIA

1977

0

•

C O N T E N T S Page

1. INTRODUCTION 1

2. AIM AND SCOPE . .2

3. BACKGROUND 2

4. PREPARATION TECHNIQUES 4

5. CONCLUSIONS

6. ACKNOWLEDGEMENTS

7. REFERENCES

8. APPENDIX

Division of Applied Geomechanics, CSIRO P.O. Box 54, Mt. Waverley. Vic. 3149

ISBN O 643 02119 1

12

14

14

15

Copyright CSIRO 1977

IJ

SAMPLE PREPARATION TECHNIQUES FOR THE TRIAXIAL TESTING

OF CEMENTED ROCKFILL

BY

L,P, GONANO

1. INTRODUCTION

In the 1100 copper orebody at Mount Isa, cemented rockfill pillars are being formed in primary stopes to allow for 100% ore extraction. These pillars, 40 metres by 40 metres in plan, and up to 250 metres high, are required to fulfil a ntunber of structural functions:

1. Laterally support rock pillars during early extraction;

2. Restrict caving following complete extraction ;

3. Minimize surface subsidence;

4. Minimize dilution in adjacent mining;

In the analysis of pillar stability, and ultimately the definition of a rational design procedure for the use of cemented rockfill as an integral part of the mining method, some knowledge of the properties of the materials in the pillars is required . •rhe assess­ment of these properties is very difficult and is complicated by a number of factors:

1. The large range of materials and structural forms developed during deposition and curing;

2. Inability to core the material or use down-the-hole in situ sampling or testing technique.s;

3. The presence of l arge particles (up to 300 mm dia.) in most of the material.

In situ testing has been conducted to obtain first hand, reliable data on some mechanical properties but such an investigation to fully deterinine the characteristics of cemented rockfill would be too expensive and restrictive owing to the number of tests required.

'.

2.

Modelling the material, either full scale reproduction or on a reduced scale, is the other alternative. Previous experience (Ref. 1) with measurement of the properties of rockfill as used in dams has shown that often, the cost and time involved in extensive tests of the full scale material is prohibitive but indicates that some degree of material modelling may be possible with .little loss of accuracy.

Provision has been made in the component "Cemented Fill Requirements" of the AMIRA project "Support and Stabilization of Stopes", for the large scale triaxial testing of cemented rockfill (CRF). Actual material components used in the filling process at Mount Isa Mines will be scaled down according to practical and theoretical require­ments and formed into 380 mm dia. x 800 mm high specimens.

2. AIM AND SCOPE

Before production testing can begin, the most practical methods of sample preparation have to be investigated. The theoretical require­ments for material similitude are difficult to formulate and it is doubtful if any but the simplest of such requirements would be practi­cable. Thus the investigation has concentrated on finding the best methods of achieving particular types of structures while ensuring uniform and consistent samples. In most cases, only particle size distribution is required to be modelled. 'I'he insitu structures have been properly modelled by maintaining fidelity in properties such as density and component proportions. This suggests that the principal theoretical requirements of similitude have been satisfied.

3. BACKGROUND

CRF pillars are formed through the simultaneous deposition of graded rockfill and cycloned mill tailings (HF) free falling from the top of the stope. · The hydraulic fill has a pulp density of 70% and contains about 6% equivalent Portland ·cement content. The rockfill forms a conical rill, while the hydraulic fill lies in the lower portions of the open stope, percolating through the rockfill where possible (Ref. 2). The components may suffer degradation, compaction, segregation and cementation and widely differing materials may be produced. For example, the intention of cementing the rockfill with CHF is not always realized and often the rill cone does not extend to the peri­meter of the stope. The materials produced vary continuously from one extreme to the other. For convenience, they have been classified as in Fig . 1. Uncemented structures 2a to 6 are not expected to be very common and no attempt has been made to reproduce samples of these materials for this study. Structures 1 to 5 are most common; their relative proportions and position within a pillar being primarily dependent upon

1. Stope geometry;

2. Proportion of rockfill to CHF introduced into the stope;

3. Location of stope fill inlets;

4. Rockfill gradation (this has an effect on the ability .of CHF to percolate through the mass ).

r-

I'.-.'· :•,:.:: ·. :: . . ·· .. . .. ,; . :.-: . . .. .. .. .. . . . · .... : : . .-...

. . . : ·. : ·.· .... . . ·: ·

3.

No . 1. Cemented sandfill - may be layered but no rocks pres ent .

No. 2. Rock 'float er s ' in a cemented sandfill matrix - no rock to rock contact.

No. 3.

No. 4.

No . 5.

Rockfil 1 with rock to rock contact with inter rock vo ids fully occupi ed with cemented sandfill - no air vo ids.

Rockfill with rock to r ock contact with inter rock voids partially occupied wi t h cemented sandfi ll.

Rockfill - uncemented. No fin es .

No. 2a. Rock 'floaters ' in uncemented matrix - no rock to rock contact.

No. 3a. Rockfill with rock to rock contact with inter rock voi ds filled with · uncemented material.

No. 4a . Rockfil l with rock to rock contac t with inter rock voids partially filled with uncemented material.

No. 6. Fine grained unc ement ed material e.g. rockfill fines.

2235

Fig .1 MATERIAL STRUCTURE LEG END

,,

c.!) z V) VI <t CL

w c.!)

~ z w C> 0:: w CL

100

90

80

70

60

50

40

4.

For any one basic s t ructure , rockfil l size gradation and cementing ability of the CHF may vary somewhat due to segregation and the hydraulic effect of excess water on cement fines. The effect of these changes within any one structura l type will not be considered. Instead, grading curves which have been found to be r epresentative for each material will be adopted (Fig. 2) . Structure ' s Na's. 1 and 5 are relatively easy to reproduce. Hence most effort has been directed towards developing methods of preparing structures 2, 3 and 4 . The properties of structure No. 1 type material have been det e rmined in a separate investigation (Ref. 3).

4. PREPARATION TECHNIQUES

4.1 General. It is stressed that while a prototype fill pillar may be very inhomogeneous, a triaxia l t est sample is attempting to model the material in any one region or zone (not the pillar structure ). Hence the sample should be as ·uniform as possible if the properti es of t he particular material are to be accurately measured.

In the preparation of the materi als, it is desirable to have f airly stiff mixes (low water content) and relatively high cement/aggregat e ratios. The reason for this is that very fluid mi xes which essentially lose a lot

72 8 S SIEVES

200 100 52 36 25 14 7 3/16 3/s 3/4 1112 3 6

JFf ffif[~ t= _ H 1 __ ;_ _ I ,_,_._- ..... I_,___._·-+-+-+---' ,},..,, ~

-+-1-+-++-1- '--··-'---'---'--'-' .. E=Nfffiff B_,f-H-H+H---1-+-·1-+-++Hl-- --1-+-n-.11--1~ '7_-

t----+---+--1-+-1-++1-+--- 11 ~ t~q I A~-~ _l"_B_~~I~U R fiJto.:.J~ __ ,_ _ __ ---•-+-++-++H--- ) ~~ 1· f-j ~-- =----~'. ___ J.1---.~ ---+-,_+-,_+-, ... µ+.; .. --·1-----=·-l---·-'-- '- ,/", --1

t--·

- kV I I

-1-·-- - ·1- ·:_ ,____

1--- ,_ -~-,-... ..

>----

--

-- - ... -

r--·--- --

30 >----+--t,-- , ...

- ' __ - - --- - -- --- -~ - " - 3 -~ / -• ___ -~f- ___ ·-'-'- ~ I- [;7Lt -·+--- +,ff-_,_

. ~--- --+-+-+ ++-- ,_, ~-==/ ,--=:: ,.:~- ''\"~---!-++H+-H--1 / - ~'<-i-t-+-f-H-

.1----. ---1-+- H -+ -l- ---- '·- - ' . .. V ---·->-· i1 ~ ~_..~!>'!'~ 1----+-+-l--1-H-H·I--

. -· -- . .... - -- - --- ,- ~ ~ .. ::-..~ >- . ,- -+--+--+-+-1-H t--

-· .. -- ·--·--- ... .. ------- p·--- ... · - ~ _ ;_ ···-- - 1----+-+--H-+H · '---

,__

20 ~ -

1-- -- ---

10 - ---·- --·

t---- -- _,_ ,•~

0 0 ·0001 0·001 0·01 O·I 10 100

PARTICLE SIZE ( rnm)

CLAY ·j- SAND GRAVEL ·----'---2236

. Fig. 2.- Grading curves for in s itu structures (f or CRF found in S32 stope)

'\

5.

of water in curing and setting tend to be quite variable both within and between samples. Nevertheless, for some structures, a very fluid mix may be an essential feature of the preparation. High cement proportions (of the order of 10% or more versus 6% in situ ) are considered desirabl e in enabling early testing of manageable sampl es. Because completely differ­ent water/cement ratios and cement contents have been adopted in the samples as compared to insitu conditions, indexation of strength will be required. Water/cement ratio and cement content effectively control cohesion of the material. Other important factors which influence cohe­s ion are the curing conditions (surrounding pressure, t emperature and moisture conditions ). 'rhese should be as found ins:Ltu but this is not essential since s imilitude of cohesion has not been attempted. An index fac tor for cohesion may _be derived from a comparison of the uniaxial strength of the samples with the known strengths of corresponding struc­tures ins i tu or with strengths of more faithfully reproduced samples . This index factor will be valid only for separat e particular sets of preparation, curing and te.sting conditions for both sampl e and prototype. The other aspect of material strength modelling i s similitude of inter­nal frictional resistance donated by 0. This is related to a number of factors:

1. Particle size distribution;

2. Particle shape ;

3. Structural fabric of the materi al ;

4. Particle breakage;

Reproduction of particle shape will not be a problem since the sample components wil l be as used in the prototype. Great care will be taken to simulate the structural fabric. However , particle size distribution will require scaling to increase the ratio of sample diameter to maximum particle size to greater than 5: 1. If this ratio is l ess than 5: 1, specimen size effects occur (as found for example by Fumagalli (Ref. 4)) preventing the proper measurement of true "properties" of the material .

In the past , t wo methods have been used for scaling rockfill in the laboratory .

1. Scalping the particles above a certain size range. This is often done by the Snowy Mountains Engineering Corporation (A.D. Hosking, pers. comm.).

2. Shifting the whole grading curve to a smal l er size range such that the maximum particle size is again 5 or 6 times smaller than the sample diameter .

With the second method, difficulty has been experienced in reproducing the density of the prototype material. •rhe first method has been adopted here and appears to be fundamentally sound for two reasons:

1. 'rhe large particles in an aggregate serve merely to bulk the material and do not greatly affect the phys :i.cal processes of particle struc-· turing and cementation.

2. Because of the important effects of the high specific surface of the fines fraction on workability , the cement reaction , and surface ten­sion effects , the lower portion of the grading curve should be accurate ly r eproduced.

..

6.

Since scalping removes the large .particles which require l es s than proportionat e cementation , it may be necessar y to increase water and cement contents s lightly in the s amples to ma inta in matrix strength . This effect is already accommodated by the indexation of cohesion described earlier . Provided scalping do.es not change the structure of the material, its effect on the frictional characteristic of the material is expected to be minimal.

If confining pressures are sufficiently high and t he mode of failure such that aggregate particle breakage occur s , both the strength and deformation response of the mode l mat erial may differ from that of the prototype. Nevertheless, in the present investigation , i t may not be significant . It may be shown that i f the particles are inhomogeneous ly stressed and frac ture in a brittle manner , the fracture stress is de­pendent upon the particle s i ze. Analogous behaviour exists in t he degradation of rock particl es by impact and in the size effects obser ved in point load tests (Ref .~) . The breakdown in s imilitude is related to the energetics of brittle fracture (Ref. 6) and is diff icult to correct unless the aggregate minerals are changed.

The methods of modelling or indirectly accounting strength prope r ties have been described. Provided prototype component s , the ir proportions and s tructural fabric are maint a ined i n the model materi al, similitude of deformation characteristics will al. so occur . However , by r educing the maximum particl e size as has been advocated , the propor tions of rockfill to CHF will be reduced if the same structura l f abr ic is main­tained . It is suggested that where possible, because of the difference in moduli between rock particles and CHF , the proportion of these compo­nents should be maintained at the expense of slightly altering the grading. For s tructures 2 and 3, it i s proposed to maintain the correct component proportions and structures by altering the grading in orde r t o f aithfully model t he deformation char act eristi cs. For structur e 4 , the f abric and frictional strength are strongly dependent on the grading of the particles , therefore grading will be reproduced and the small er ror in de f orrnabi l ity will be tolerated. It is expected that this error will be smal l (deformation modulus will be underestimat ed in the samples ) since the scaling factor is quite small.

The size distribution for the components used in the preparation of the samples are as gi ven in 'I'able 1. 'l'hese are r epresentative gradings determined from ins itu material. Portland cement was used in lieu of the mixture of Portland cement and copper reverbatory furnace s·lag used in practice at Mount Isa. All proportions are by dry we ight and account of specific gravities should be t aken when other components are used. The quantity of water r equired has been estimated from the s lump of the mix. All rock particles should be l ess than 3 inches (75 mm) in mean diameter .

4.2 Structure No. 2. This s t ructure is basical l y cemented hydraulic fill with rock "float ers". Scalped grading curve No. 2 should be appr oxi­mately matched , but similit ude of this structure is more dependent upon proportions of rock and CHF than particle gradings. Three trial s amples, TS1, TS2 and TS4 were prepared . The propor tions and resulting particle size distributions t r i ed are shown .in Table 2. Proportions are expressed as percentages of granular components. Cement is expressed as an "added" percentage of the hydraulic fill component only , to con form with existing

L

. n

7.

practice. The ratio of rock to l ydraulic fill is also shown . All the components we re mixed in a concrete type mixer and placed in the mould with no vibration or rodding to prevent "settling out " of the heavier rock particles. 'I'S 2 (slump 5 ems ) was much firmer than TS1 (slump 15 ems) by virtue of a lower water content.

Large :rock particles appeared to settle out in TS1 but not in TS2. In both s amples, str~cture was solid with no voids l al'."ger than 5 mm. The structure of '.P.S2 was very much the "floater " type with no segregation (Fig. 3) . In 'l'S4, a stiff mix was again used but the rock:hydraulic fill ratio was reduced to 1:1 •5 . Type 2 structure was still evident . Because of the large content of hydraulic fill in all these mixes, shrinkage upon drying was considerable .

Density of TS4 at an age of 4 days was measured a t 2 . 1 Mg/m3 (133 J.bs/ft 3) which compares favourably wi th mea.,,urements of i ns i tu density (Fi g. 4). For structures 1, 2 , 3 and 4 , dens ity increases proportionally with per­c entage of rock particles.. When structure 5 is deve loped , t he high void f orces a reduction in density again .

Recommendat ions for the attainment of s tructure 2 are :

1. Rock (no f i nes ) to hydraulic f ill rati o should be in the range f rom: 0 . 4: 1 t o 0 :1 .

2 . 'rhe components should be mixed as l ean concret e ,:md p l aced wi th no vibration or compaction.

3. The fi rmest mi x consistEmt with a reasonably so l id and dense spec i men to avoi d settling of " f l oaters" shoul d be used.

4. Gradi ng curve No. 2 should be approximated .

5. Quantities of mater i a ls for the casti ng of one sample as f or •rs4 (Volume = 0 .09rn3) are :

[Coarse 17 Kg ROCK M~dium 13 Kg

Fine 40 Kg TAI LINGS (HF) 100 Kg CEYJEN'r 10 Kg

6. The mater i a ls may be mi xed in a numbe r of ba t ches i f necessary.

4.3 Structure No. 3. I n t his mat erial, the matrix of CHF should not contain nny significant air voids yet provide good rock-to-rock cont act. Two mi xes TS5 and TS7 were tried in an effort to produce a dense uni form mix o f t he correct proportions (Tab l e 3) . The components we r e mixed as a l ean concret e with very little s lump (0-2 ems ) and well compacted and vibrated during placement. When broken up after curing, sampl e 'l'S5 did not reveal f ul l y developed rock-to-rock contact . TS7 was better i n t his respect and is suggested as the best mix t o simulate str ucture No. 3 . Density was found to be only s lightly higher than for the in situ values ; agreement is expected t o be be t ter as t he sample drie s out.

Re commendati ons fo r the preparation of s t r ucture No. 3 a r e

1. Rock to hydraulic fi ll ratio should be about 3 : 1.

2. Grading curve No. 3 should be matched.

3. Components may be mi xed in one batch i f des i red and placed with good vibra tion .

•

8.

GRADING CURVES FOR COMPONENTS

96 PASSING

ROCK (Kennedy Siltstone)

B. S. Sieve Size (ins )

Coarse Medium Fine Mill Ta ilings (H.F.)

3

1~ 3/4 3/8 3/16 7

14

100 7 .5 0

100 100

22 100 0 70

28 3

36 100 52 · 76

100 36 200 30

TABLE l. PARTICLE SIZE DIS'I'RIBUTIONS - FILL COMPONENTS

Test Sample 'l'Sl TS2 TS4 Rock

Coarse : Medium:Fi~1:H.F . (10: 7 : 17 :66) (10: 7 : 17 :66 ) (10 : 7 : 23 :60 )

Rock: H.F . l · ') 1:2 2:3 B.S. Sieve Size (ins ) 3 100 100 . 100

1~ 91 91 91 3/4 85 85 85 3/8 78 78 76 3/16 71 71 66 7 66 66 60

14 66 66 60 36 66 66 60 52 50 50 46

100 24 24 23 200 20 20 18

Percentage Passing

·-----------·---TABLE 2 . PARTICLE SIZE DISTRIBUTION - STRUC'r URE NO . 2

Scalped Grading Curve No. 2

100 78 65 59 56 54 53 49 46 38 16

. ,., ~. ,. ~ • ..•. ~ - •

" ,.··. ..

+• .. .. , JI • . . .. ..

Fig. 3.- Sample No. TS2 (Structure 2) showing 'floaters' (painted white for clarity)

9 •

-------Test Sampl e

Rock .-. - -- ·---,

Coarse:Medium:Fine:H .F. Rock: H.F .

Grading B.S. Sieve Si ze (ins ) 3

1\ 3/4 3/8 3/16 7

14 36 52

100 200

TS5

( 2 5: 2.0: 2 5: 3 0) 7:3

100 77 59 47 37 30 30 30 23 11

9

TS7

(25:25: 25:2 5) 3 :1

100 77 55 41 33 26 25 25 18

9 8

Percentage Passing

TABLE 3. PARTICLE SIZE DISTRIBU'l1 ION - STRUCTURE NO . 3

•rest Sample Rock

,------.:....·--, Coarse:Medium:Fi ne :H.F. Rock:H.F .

Grading B. S. Sieve Size (ins ) 3

11.:i 3/4 3/8 3/16 7

14 36 52

100 200

TS3 TS6

(45:45:0:10 ) (30 : 40 : 20:10 ) 9:1 9:1

100 100 58 72 20 39 10 24 10 16 10 11 10 10 10 10

8 8 4 4 3 3

Percent age Passing

TABLE 4. PAR'l'ICLE SIZE DIS'l'RIBUTION - STRUCTURE NO. 4

ll.

Scalped Grading Curve No. 3

100 74 58 47 38 31 26 22 20 17

9

Scalped Gradi ng Cur ve No. 4

100 50 30 18 14 12 10

9 8 7 5

4. Quant ities of mater i als reqt1ired for thci casting of one Sillnple (volume = 0.09 m3) as for TS7

ROCK

Tailings Cement

[

Coarse M~dium Fine

are

42 Kg 35 Kg 42 Kg 50 Kg

5 Kg

12.

4.4 Struc ture No. 4. This structure is dry rockfill with parti a l cementation by percolation of cemented hydraul ic fill. 'J:he void ratio is high and the material weak in the unconfined condition even though the rock particles may be densely packed . Grading curve No. 4 is representa­tive for this mater i al. Two samples were prepared 'TS3 and TS6 (Table 4). No fine rock wa.s used in TS3 to ensure good percolation. Procedure was to place the r ock in the mould and then pour very fluid hydraulic fill (with 10% cement) onto the 'top and vibrate it down through the mass.

The CHF t ended to settle out into a hard packed cohesive mass with a subsequent reduction in permeability. While al l rock particles became coated with CHF, the specimen was far from uniform (Fig. 5) . The more realistic particle size distribution used in 'rS6 resulted in some areas being completely uncemented despite the use of very aqueous CHF. To improve both cementation and uniformity, the sample will need to be formed in l ayers using the t echnique of "raining" CHF onto rock parti­cles which a r e being progressively added to the mould. Suggested quanti·­ties for the preparation of one samvle are:

[Coarse 45 Kg ROCK Medium 60 Kg

Fine 30 Kg Tailings 15 Kg Cement 1.5 Kg

5. CONCLUSIONS

An approach to the reproduction in the laboratory of a number of struc-· tural forms of cemented rockfill has been described. Because of the limited knowledge of the behaviour of cemented rockfill , it has not been considered worthwhile at this stage to fully investigate the theoretical similitude requirement for material modelling. Nevertheless, reasonable similitude of mechanical properties is expected. This can be attributed to

1. faithful reproduction of structural fabric,

2. use of the same components as in the prototype ,

3. the us e of a small scale ratio (based on particle size ) of approximately 21~ to minimize distortion effects ,

4. successful approximation of a number of index properties such as particle grading, dens ity and component proportions.

Based on the r esults of this initia l inves tigation, a program for the preparation and tes ting of production samples has been drawn up. This is presented in the Appendix.

I - I

I

13.

Fig. 5.- Sample No. TS3 (Structure 4) showing partial non-uniform cementation and voids.

l,

14.

6. AC KNOWLEDGEMENTS

~:'he work descr ibed in this report fonns part of an integrat.ecl geomechanics study which i s being car ried out under t he joint sponsorship of CSIRO , Division of l\pplied Geomechan ics and ·i.:he Australian Minera l Industries Research As!,ociation.

The author wishes to r ecord his appreciation to Mr. M. Worswick for his assistance in the experimental phase of the work and Mr. R. Kirkby from Mount Is a Mines for his adv i ce and co- operation.

7. REFERENCES

l. MARSAL , R.J. (196.7) .- Large Scale Testing of Rockfill Materials. ASCE, J. Soil Mech. Fotu1d. Div., Vol. 93, No. SM2 , pp. 27-43 .

2. BARRET'l' , J .R. (1973) .- Structural Aspects of Cemented Fill Behaviour. Proc. J"ubilee Symp. in Mine Filling, Mount I sa , Aust . Inst. Min . Metall. , pp. 97-104.

3. GONANO, L.P . (19 75) . - The Mechanical Properties of Cemented Hydraulic Fil l Pillars at Mount I sa Mines . CSIRO , Aust. oiv. of }\p_pl:i.ed Gcomechanic s . 'J.'echnical Report No . 36.

4. FUMAGALLI, E. (1969). - Tests on Cohesionless Materials for Rockfiil Dams. ASCE, J. Soil Mech. Found . Div. Vol. 95 , No. SM2, pp. 313-330.

5. BROCH, E. and FRANKLIN, J·.A. (197 2) . - The Point Load Strength 'I'es t. I nt. J . Rock Mech. Min. Sci., Vol. 9, No. 6, pp . 669- 697.

6. BROWN, E.T . and GONJ\.NO , L.P. (1975 ) . - Analysis of Size Effect Behaviour in Brittle Rock . Proc . 2nd Au st . ··N.Z. Conf. on Geomechanics, Brisbane, ,July .

'r

r

8. APPENDIX

PRODUCTION SAMPLE PREPARATI ON AND TRIAXIAL TESTING

8 . 1 AIMS OF THE TRIAXIAL TESTING Pi: OGR.i'\M

1. Initial apprai,;;al of the restrictions and pos;sibilities to satisfy similitude in rnodo.L; of in c;i t1i c0rncnted fj :11.

2. Simulate various material structures found in sit u and evaluate their mechanical properties. Sim.il .i t ude will be r equired in proportions, cement content , s t r uctural f abric and dens ity .

3 . Maximisation of strength properties - specifically, to det ermine improvernent E; :i n pillar strength when favourable methods of composition and placement are used with a fixed cement content. Thi s essentially involves an optimizat i on of structure 3 type mater i al.

To some extent, a ll three activities are t rial -and·-error processes and hence only a limited number of samples will be used initially.

8. 2 SPECIMEN PREPARA'l'ION

Three s ets of samples will be prepared.

'A ' samples basically for activity 2.

'B' s ampl es - f or activity 1 and to index the results f r om activity 2 .

'C' samples - for activity 3, i.e. optimizati on.

Specimens will be designated according to their purposes (activity) and structure (Fig. 1). The program is as f ollows:

Structure No.

1

2

3

4

ACTIVI'rY

A

6

6

6

6

NO. OF' SPECIMENS

B C

3

3 6

The proportions of materials for t he preparation of 'A' and ' B' samples . are (cement and CRF are extra percentages as% of tailings only):

• 16.

--------------- ·----· ROCK 'rAILING CEMEN'r CRFS

Structure Coarse Medium Fine

- For 'A ' Samples

Al 100 % 10% A2 10 7 23 60 10 A3 25 2·· _:) 2.S 25 10

A4 30 35 20 15 10

For 'B ' Samples

Bl 100 3.5 7 B3 30 35 20 15 3.5 7

For 'C' Samples

Basical l y ' A3 ' but modified as given below.

' A' samples will be prepared in the manner described earlier in the report , a different method being used for each structure .

Samples will be cured at 35°C for 28 days i n a moi st environment. The specimens will be tested at the following confinin9 pressures :

2 at 0 MPa 2 c1.t 0.7 MPa 2 at 2.0 MPa

The specimens wi ll be unsaturat ed and tests will be of the ' undrained ' type.

'B ' samples will be prepared in a manner similar to that for in situ placement L e . water will. be 30 96 of CHF component. Curing will be at in situ temperat ure f or 56 days in a moist condit ion. All ' B' s amp l es wiJ.l be t est ed unconfined a nd the r esults compared to the unconfined 'A' samples . The ma in purpose of the compar ison between ' A' and 'B' will be essentially to account for differences in cohes ion (hence unconfined strength) . 'B' samples should aim t o reproduce the in situ ' materia l (structures 1 and 3 only) in every poss i ble way.

'C ' samples are essenti ally for compar ison with the A3 samples and hence have the same curing and t es ting conditions .

Cl (2 samples ) - same grading as A3 but with very low wat r content . C2 (2 samples ) - same as Cl except CHF proportion a lso reduced. C3 (2 samples) - s ame as C2 but proportion of rock fines increased to produce a smoother grading curve . In this way it is pos s ible to check if rock fines are detrimental.

Because both p l atens .in the triaxial ce ll are of fixed type , the ends of the specimen will need to be pl ane and parallel t o within 0.3 mm . This may be achieved by usiny the levelling rods to accurately position the top platen (which is floating on a layer of grout ) relative to the bottom p l aten.

0

• n

8. 3 TES'rING

·rhe data collected from each specimen series will include:

cohesion internal friction modulus Poisson's ratio density

C

<P.

Ev··J

y water cont -at W

unconfined

non- l inearity of stress-strain curve A particle size distribution] f . ·rom preparation cement content

8.4 ANALYSIS OF RESULTS

1. Interna l comparison of 'A' sample results will indicate the varia­bility of mater i al. between zones and provide data on the mechanical properties of in situ structures for numerical stability analyses. Similarly they will give a r ange of properties to be used in the numerical analys is of the effect of pillar inhomogeneity and they w:i.11 contribute towards an understanding of the mechanics of cemen­t ed granular systems.

2; Internal comparison of ' C' samples and with A3 samples w:i.11 shuw how the material strength may be optimized using variations in gradation, water content and placement method only. From experience, material structure 3 is the strongest , so only this type will have to be optimized .

3. Type ' B' samples serve t wo purposes . By comparison with in situ properties they provide a check on the fidelity of modelling and t hey present an indexation of type 'A ' specimens. Note that 'A ' spe_imens have of necessity a higher than normal cement content, and different preparation t echniques .

'I'wo factors FIB (in situ versus 'B' samples ) and FBA _(B samples versus

A samples) will be required to relate ' A' strengths to in situ strengths.

4. Comparison of A3 sample strengths with Al sample strengths will show how t he u se of rockfiJ.l may improve the quality of the fill even though cement content has been reduced .

8.5 NOTES

1. In the selection of grading curves, sizings as found in S32 have been scalped. Attempts to shift the grading curve to smaller sizes, results in incorrect density of material and also neglects the impor­t ance of fine s (especially specific surface area and surface tension effects).

2. During preparation, CHE' may obsorb l arge quantities of water whi his released in the early i,tages of setting. This effect ively produces a high water/cr~ment rat· o t.hou9h the mix is barely wo1~kable. Hence it appears that to make any significant gains in strength , th " CIIF com­ponents may need to be reduced . The effect of this will be measured in the 'C ' samples .