Sie sind auf Seite 1von 12

Int. J. Rock M~ch. Mining Sci. Vol. 1, pp. 535-546. Pergamon Press 1964. Printed in Great Britain.

ON THE VALIDITY OF THE 'BRAZILIAN' TEST FOR BRITTLE MATERIALS


C. FAIRHURST School of Mineral and Metallurgical Engineering, University of Minnesota, Minneapolis, Minnesota, U.S.A. 55455

(Received 18 February 1964)


Abstract--Failure in the Brazilian tensile test (diametral compression of a disc) is analysed on the basis of a Gritiith-type fracture criterion, empirically generalised for an arbitrary ratio between uniaxial compressive strength and uniaxial tensile strength. Results indicate that failure may occur away from the centre of the test disc for small angles of loading contact area with materials of low compression-tension ratios. In such cases the 'tensile strength' as usually calculated from test results, is lower than the true value. INTRODUCTION THE SO-CALLED ' B r a z i l i a n ' test is used as a means o f d e t e r m i n i n g the tensile strength o f brittle materials. (Fig. 1). A thin circular disc is c o m p r e s s e d (to failure) across a d i a m e t e r .

FIG. 1. The Brazilian Test, showing notation used. T h e c o m p r e s s i o n induces tensile stresses (oar) n o r m a l to the diameter, which are sensibly c o n s t a n t over a region a b o u t the centre. (Fig. 2). T h e tensile strength is calculated on the a s s u m p t i o n s t h a t failure occurs at the p o i n t o f m a x i m u m tensile stress (i.e. at the centre) 535

536

C. FAIRHURST

and that the radial compressive stress (~rru) has no influence on failure. This note examines the validity of these assumptions.

-1'6

-1"4

-1"2

-I'0

-0"8 - 0 " 6

-0-4

-0-2

y'

0"2

0"4

Compressio~

Tension

Stresses ore multiples of 2p/'n" F[~. 2. Stresses along line of loading YY' in Brazilian Test for strip load angle a = tan-l~%

STRESS D I S T R I B U T I O N I N T H E L O A D E D DISC HONDROS [1] has analysed the stress distribution in the Brazilian test for the case of a thin disc loaded by a uniform pressure, radially applied over a short strip of the circumference at each end of a diameter. (Fig. 1). The stresses of interest i.e. those acting at points along the diameter YY', are given by the expressions:

ary - -

1 - - 2 r 2 / R 2 cos 2a -~ r 4 / R 4 - - tan-~

-- 7 ~ ]

__

~ ~ 4 / R

4 + tan -a

] tan a ]

tan a ,
.

(1)
(2)

(Tensile stresses are considered to be positive throughout this note). where: O . o y - - tangential normal stress at a point on the YY' axis (vertical in Fig. I), O'ry ~ radial normal stress at a point on the Y Y ' axis, p= load per unit area of strip ( p ~ P l a t ) , P: applied load, ~ l ~ projected width of the loaded section of the rim, t ~ thickness of the disc, 2 a ~ angle subtended at the centre of the disc by the loaded section of the rim, r ~ radial distance of a point from the centre of the disc, R~ radius of the disc. Fig. 2 shows the variation of stresses, a0u ,and aru along the diameter YY' for the case

ON THE VALIDITY OF THE ~BRAZILIAN' TEST FOR BRITTLE MATERIALS

537

where a -----t a n - l ~ . Stresses are expressed as multiples of 2pfir. (Calculated values are given in Table 1). At the centre:
%u = + _2_p(sin 2a -- ~) -"- 2p~a
~" ~g

(a < R/5).

Substituting P/ Dt -= pa,
2P c~u -- ~rDt"

(3)

This expression [equation (3)] is used to calculate the tensile strength of the material. It is seen that equation (3) is the same as that given by point loads P applied at the ends of the diameter. [2] C R I T E R I O N OF F A I L U R E FOR BRITTLE MATERIALS Griffith criterion It is generally accepted that Griffith's hypothesis [3-5] is the most satisfactory explanation of fracture of brittle materials. This hypothesis assumes that fracture occurs by the rapid extension of sub-microscopic, pre-existing flaws randomly distributed throughout the material, The fracture criterion is defined as follows: (i) If 3~rl + ~a > / 0 Failure occurs when ~rl = K. (ii) If 3~1 + ~ra ~< 0 Failure occurs when (g~ -- ~ra) ~ + 8K(~r~ -4- ~ra) ---- 0. (4) (5)

where: ~rl and ~ra are the major and minor principal stresses respectively, K is the uniaxial tensile strength of the material. (It is assumed here that the intermediate principal stress gz has no influence on fracture. [6]) Plotted as a Mohr envelope (i.e. on ~r, r cartesian co-ordinates), the criterion is a parabola:
r~ = 4 K ( K - ~), where: r is the shear stress across the plane of failure at the point of failure, ~r is the stress normal to the plane of failure at the point of failure.

(6)

The Griffith criterion is often criticised on the grounds that the predicted 8 : 1 ratio* between the uniaxial compressive strength and the uniaxial tensile strength is usually lower than ratios observed experimentally. MCCLINXOC~ and WALSrt [7] (who suggest the discrepancy may be due to the development of frictional stresses across the faces of Griffith cracks closing in a compressive stress field) quote ratios ranging from 5 to 22 derived from analysis of data reported by WtrERt~ER. [8] Experimental results do seem to confirm, however, the parabolic shape of the envelope. [9]
Empirical generalisation o f the Griffith criterion The above criticism of the Griffith criterion may be overcome for present purposes, by empirical generalisation of the criterion to allow for an arbitrary compression: tension

* This ratio is, of course, readily determined by setting trl = 0 in equation (5).

538

c. FAIRHURST

~ ~ ~ 6 ~ ~ o ~ 6 ~ 6. .6 . . ? ~ ~ e ~ ~ o o o ~ o o o ~~ ~ o o o o o o o ~~ 660 666o

0 ~ e ~~e~a e~ ee~e~ ~
~ ~

. ~

0
~

. . . . ~ ~

~ ~

~ o ~ 0000000000~

"

o ~

~ - ~ 6 6 . 6 . 6. . ~O000~O0~m~ ~ ~
~ 6 0

&~O

~,,1~ o~-p,D~,~.:.4.. a ~
~ 6~66666666~

-~o,.~

~ ' ~ o~.~ ~ ~~ . . . . & ~ & 6 6 6 6 6 ~ 6660

~6o

I ] I1 I I I I ~ ~~llll
< <

I I I -~ ~~

] 1 11 I I , ~ ~
Z

. . . .~ . I I I ] I I . .~

0 <

IIII

~ l ~ ~ -s ~ l

I I I ~ ~ s ~os ~ ~

~1~ m

$
m ;~ M <

,-: ~1 <

~ ~

~ ~ s ~ ~ ~ 2 ~ g ~
~ ~ ~

~ o

m &

~ ~
~ ~ m ~

~ ~

~ 6

@ ~

~ ~

~ @@

~ ~ ~@ ~~ @ ~

~l~

~o~

6 6 6 6 6 ~ 6 6 6 6 6 ~ 6 6 ~ 6 ~ 6 6 ~ IIIII111111111111111111111

~1 ~
~t~

66666~666~6~66666666~6~ III
~ ~ ~ ~ ~ ~ o o~6666666666~o66666666666~ --.y..--__ ~ _ _ _ ~ - ~ ~ T ~

IIII

O N T H E V A L I D I T Y OF T H E ' B R A Z I L I A N '

TEST F O R BRITTLE MATERIALS

539

strength ratio. The generalised form may be stated as follows (A detailed derivation is given in the Appendix): (i) If m(2m -- 1)~rl ~rz >~ 0, failure occurs when ~r~ = K. (ii) If m(2m -- 1)ol q- aa ~< 0, failure occurs when
(f~ ~s)~
_ _

(7)

o'i -~- fill

2(m--1)2K[1-k~r~-~fa~k~2--

2K f ~ m - 1~ ] -

1}],

(8)

where: rn = ~/(n q- 1), n = absolute value of the ratio uniaxial compressive strength uniaxial tensile strength

The condition for failure at some point along the vertical diameter in the Brazilian test can now be formally determined by substitution of equation (1) and equation (2), putting fog = O'1, and fru = ell, into equation (7) and equation (8), determining the minimum value of the applied radial pressure p as a function of n, a and K, for which either of equation (7) and equation (8) is satisfied. The general relationship is not, however, a simple one. Attention has, therefore, been confined to consideration of two specific cases from which the main general practical implications are readily deduced. DETERMINATION OF STRESS SEVERITY AT A POINT ALONG THE VERTICAL DIAMETER The above procedure determines the theoretical point of failure in an ideal, homogeneous, isotropic material. In practice, failure may actually first occur at some other point where, although the calculated stress condition is not the most severe, the material is locally weaker due, for example, to a flaw, inclusion, grain boundary etc. It is thus of interest to obtain a measure of the 'stress severity' at points along the diameter in orde~ to assess the likelihood of such unpredicted failure. The 'stress severity' will be defined as the ratio of the (theoretical) load at failure of the specimen to the load which would (theoretically) have been required to cause failure at the point. Conditions (i) and (ii) of equation (7) and equation (8) will, in general, both exist along the diameter. We will refer to that part of the diameter for which equation (7) holds as Region (i), and that part for which equation (8) holds as Region (ii). One of two situations may exist at failure, viz. : (a) Failure in region (i), when:
O'lm~x ~ K"

(9)

in region (i), and

--(~r~_--~rll)2
~r1+~3

<2(m--

1)~K 1 q - ~ l q _ O s [ \

2K_f{m--l]~
2

] -- 1

}]

in region (ii). substituting for K, from equation (9): --(~rl--~r~) z ~rlrnax(~rl+~r~) < 2 ( m R.M.~2N

2flmax/[m--1~ ~
fa[\- 2~]

}]
-- 1 ,

1)~ 1 + ~ i ~

0o)

540

C. FA1RHURST

where: 0.1max is the maximum value of the major principal (tensile) stress in region (i). Thus, when condition (I0) obtains everywhere in region (ii) failure occurs in region (i) at the point given by equation (9). The stress severity (S) is: In region (i): S ....
0.1 O'lmax
--

(11)

--(~1
In region (ii): S
= -

0.~)~

(r-~l-max(0.~}+ 0"-a) -- -2O'lmaxJ'~/m_- 1] 2 2(m--1)z[1-t-0.1+oz~l, 2 ' --1}]

(12)

(b) Failure in region (ii) when: O'lmax = in region (i) and, --(o'l--0.a) 2 2(m--l)2[ 20.1max f ~ m - - 1~ ~ 0.1max (0.1 + o~l) -)~ 1 + fl(~11~ 0 . ~ ) ~ , - - ~ ] -)]
I

fl

0 < fl < 1

(13)

(14)

in region (ii), [Equation (14) is obtained by substitution for K from equation (I 3) in equation (8)]. Failure will occur at that point in region (ii) for which ~ in equation (14) is a minimum. The stress severity is: In region (i) In region (ii) S
= flmln ~1 . fflmax (15)

S =

Brain'

(1~

where B has the value required to satisfy equation (14) at the point. Note that two regions may be defined, from the limits in equation (7) and equation (8) as: Region (i) -- -- < m ( m -- 2),
ffl if3

(17) Region (ii)


-- 0"3 0.1

> m ( m -- 2).

Values of --~3/0.1 for various compression : tension ratios are given in Table 2.
TABLE 2. VALUES OF --~ra/e I FOR VARIOUS COMPRESSION : TENSION RATIOS (n)
n --%fir: 8 3 9 3.66 I0 4.35 11 5.07 12 5-79 13 6-53 14 7.29 15 8.00

EXAMPLES The effect of the compression : tension ratio n on the position of failure and the indicated tensile strength [i.e., that given by equation (3)] was examined for the cases ~ ~ tan -1 ~,2,

ON THE VALIDITY OF THE 'BRAZILIANs TEST FOR BRITTLE MATERIALS

541

a = tan -1 ~t. Results are given in Table 1 and shown graphically in Figs. 3 and 4. It is seen that both principal stresses are compressive close to the rim load. In this region ol, on the faces of the disc at least, is the zero stress ~ = 0, acting normal to the disc faces. The values shown in brackets in columns 6, 11, 12, 13, 14 of Table 1, and plotted as dotted lines in Fig. 4, are calculated on the assumption that ~ = ~r~ = 0. It is seen that failure should then occur first under the loaded rim. However, frictional restraint at the load contact surface will produce ~z compressive stresses in the interior of the disc close to load, so that the exact state of stress is unknown. Although spalling does often occur, it is usually limited to thin flakes on each face preceding major failure by vertical splitting. This would indicate that the assumption ~r~ = 0 everywhere under the load is too severe.

= =tan-'-~

-2K

-IOK

-SK Normol $1ress,

-K

(,)

~ 0.85 "~

a =ton-,~

~.

Oo
0"75 0

"~

-IOK

-SK N o r m l ~mss, ~

(b) E[G. 3 (a), (b). ~ o ~ dia~ams of stressesat poims =lo~ vc~i~] di~ctcr at failure lot materials with various compression : to.ion ratios.

542

c. FAIRHURST

o.8

1
-IOK

I
-$K

I ,I

Ill

I
O K

Normal stress, o"


(c)

"~bOlo
~
_ ~
. ~ .~" /~ ~ ~ . -

a: mn-'~ ]

/
5~

~
d

~.---~__ ~ . -. ~
/ ...... ~ .

~- .__

~
. -

..... ~//,~

o.
. ." ~
-SK Normal

/ /,

-, ....

~k.', ~

x,

/ ,
-IOK

.." ~

///:.X/ ~ (~*~
s~ress, ff

', I'x&NN ~I ~
O K

(d)

Era. 3 (c), (d). Mohr diagrams of stresses at points along vertical diameter at Nilure for materials with various compression : tension ratios.

A S S U M P T I O N OF U N I F O R M R A D I A L L O A D I N G The above analysis has assumed uniform radial loading (Fig. 1). The actual distribution of the radial load will have little influence on the stresses in the central region and should not, therefore, affect test results if failure occurs in this region. In general, however, appreciable tangential stresses may be developed across the load contact area. Simple calculation (Appendix 2) indicates that the average tangential stress (e~), assuming no slip between the contact surfaces, is

ez=--p~

[,'st ,'r]l~l s-E., . . . . . . .E . .r.]. l ~

-- vs~ E,

1 -- vr] Er ]'

(18)

where: - - p = applied normal pressure, = effective Poisson's ratio for loading platens, vsl

ON THE VALIDITY OF THE 'BRAZILIAN' TEST FOR BRITTLE MATERIALS

543

= Poisson's ratio for test specimen, = modulus of elasticity for loading platen, Er ---- modulus of elasticity for test specimen. The term 'effective Poisson's ratio' is introduced to take account of the restraining effect of the unloaded area of the platens on the lateral extension of the loaded area. (It is assumed
vr Es
1.0,

~.~'/ n : l Z 1 ~" ~ f~ ~ ~"--~ ~'~n = g,

~0

a:tan
~I~ 0.5

-i I

~ 0.5

n : l ~

I a " -to

n ~1 g

J E o
r I i i [ 0.5 i | i =1

~n=l~ n:~O ~~ : ~

f - r~:8

~ I.O

1"5

~ Stress

0.5
severity

I.O

Stress

severity

(S)

(S)

c o 0.5
0"5
--

x~ ~N ~

IoC

~.~

F~. 4. Stress severity at points along vertical diameter at failure for materials with various compression: tension ratios. that the platens have a larger face area than the platen-specimen contact area), and is defined as: the Poisson's ratio required by material of total face area equal to the platen specimen contact area to produce the same lateral restraint on the specimen as under actual loading conditions. Accurate calculation of the restraint is difficult due to the complicated (i.e., for purposes of analysis) arrangement of the bearing area across which pressure is applied to the platen. Uncertainty may, of course, be avoided simply by using platens of area equal to the contact area. This may be accomplished by placing short prisms of appropriate elastic properties between the testing machine platens and the specimen. Tangential stresses are, according to equation 18, eliminated, when vsl Er ----- vr Es. It is to be noted that indiscriminate insertion of low modulus materials (rubber, blotting paper etc.) may produce unknown tangential tensions, and the use of lubricants (grease, graphite etc.) may result in undesirable effects such as wedging by penetration of lubricant into the specimen. CONCLUSIONS 1. The uniaxial tensile strength of materials with low compression : tension ratios is underestimated (when calculated in the usual manner, i.e. equation (3)] by Brazilian tests in which radial loading is applied over small strip angles. A Griffith material, for example,

544

c. FAIRHURST

tested with strip angle c~ ~ tan -1 ~ would fail when the centre tensile stress ---=0.70 K. A s s u m i n g that the uniaxial compressive strength is accurately k n o w n (8K) then the indicated compression tension ratio would be 11.5 : 1, i.e., no material tested in this w a y could indicate Griffith behaviour. 2. The dependence o f indicated tensile strength on strip angle is reduced as the compression : tension ratio increases. 3. A greater region is critically stressed (Fig. 4) as the larger strip angle (a = tan -1 ~) is a p p r o a c h e d , so that tensile strength values become m o r e representative o f the whole specimen rather than a p o i n t in it. 4. A strip angle o f the o r d e r o f ~ = tan-1 ~ w o u l d a p p e a r preferable to the value a = ~2 used b y H o n d r o s [1] ( R a d i a l loading m a y be m o r e difficult to realise as the strip angle is increased.) 5. Care must be t a k e n to minimise tangential stresses along the l o a d e d rim, since these m a y significantly m o d i f y the stresses induced in the disc, m a k i n g any analysis based on r a d i a l loading, i.e. equation (1) a n d e q u a t i o n (2), invalid.* 6. C o n s i d e r a t i o n s similar to those c o n t a i n e d in this note should be given to o t h e r tests used in strength d e t e r m i n a t i o n s o f brittle materials.
Acknowledgements--The author wishes to acknowledge helpful conversations with Drs. D. S. BERRYand

P. HACKETT,and Mr. E. ADDINALL of Nottingham University Mining Department, and thanks to Mr. M. L. LARSONwho carried out some of the computations in Table 1, and the American Petroleum Institute for financial support of this research. REFERENCES 1. HONDROSG. Aust. J. appl. Sci. 10, No. 3, 243-268 (1959). 2. TIMOSHEmOS. and GooomR J. N. Theory o f Elasticity, p. 107, McGraw-Hill, New York (1951). FROCHT M. M. Photoelasticity 2. p. 127, Wiley, New York (1957). 3. G R i r r m t A. A. Proc. First Int. Congr. App. Mech. pp. 55q53 (1924). 4. AND[RSONO. L. Fracture (Editors: B. L. Averbach, D. K. Felbeck, G. T. Hann, D. A. Thomas), pp. 331-353, Wiley, New York (1959). 5. ODE H. Rock Deformation (Editors: D. Griggs, J. Handin), Mere. geol. Soc. Amer. 79, 293-321 (1960). 6. MUemSLL S. A. F. Rock Mechanics (Editor: C. Fairhurst), pp. 563-577, Pergamon Press, Oxford (1963). 7. McCLINTOCKF. A. and WALS~,J. B. Proc. Fourth U.S. Nat. Congr. App. Mech. pp. 1015-1021 (1962). 8. WtmR~C[RR. G. Annotated Tables of Strength and Elastic Properties of Rocks. Amer. Inst. Min. Metall. Engrs., Pet. Branch, Paper 663G (1956). 9. MtmR[L~ S. A. F. Mechanical Behaviour o f Non-Metallic Brittle Materials (Editor: W. Walton), Butterworth, London (1958). APPENDIX DERIVATION OF THE GENERALISED Let K ~ uniaxial tensile strength, uniaxial compressive strength n~ uniaxial tensile strength
-

1 FRACTURE CRITERION

First, we will determine the equation of the parabolic Mohr envelope which enclosed (a) the uniaxial tensile strength circle, touching it at the vertex (K, O) and (b) the uniaxial compressive strength circle

(A1)

* See note added in proof, following App:ndix 2

ON THE VALIDITY OF THE 'BRAZILIAN' TEST FOR BRITTLE MATERIALS where: ~ is the normal stress co-ordinate on the Molar diagram (Fig. 5) ~- is the shear stress co-ordinate on the Mohr diagram.

545

-nK Normal stress, cr

FIG. 5. Mohr diagram showing generalised parabolic envelope of failure. The general equation of the parabola with vertex at (K, O) is:
~ = 2f(K~r), (A2)

where f i s the radius of curvature at the vertex. Eliminating ~-~between equation (A1) and equation (A2) we obtain:
,~ ( n K -

2.f)~ q- 2 f K = 0.

(A3)

For points of tangency the roots of equation (A3) will be equal, therefore: 2 f = (m ~ 1)~K, where: m = a/(n q- 1). Substitution of 2 f = (m q- 1)2K in equation (A3) yields imaginary values of ~r, so we choose: 2f= (m -- 1)'K. (A6) (A5) (A4)

Thus the required equation of the parabola is, from equation (A2) ~,z = (m -- 1)~K. (K -- ~), where: m = x/(n + 1). The general equation for Mohr's circles of failure is (A5) (A7)

[o
where el, ~3 are respectively, the major and minor principal stresses at failure. Solving, as before, for points of common tangency between the envelope equation (AT) and the failure circles, equation (AS) we obtain the relationship:
(~1 + ~2)'
~'1 ~- (~3

2(m -

1)~x 1 + . - - _ _ ~ _

}]

(Ag)

between the principal stresses at failure. It is to be noted that there is a family of failure circles which are tangent to the envelope at the vertex

546

c. FAIRHURST

(K, O). Since the radius of curvature of the envelope at this point is (m -- 1)~K, the family of circles is limited to those with radii not greater than this value, for example: aa -- az 2
~

( m - - 1)~K
-

Since ~ = K, this implies: ~3 < [1 -- (m -- I)2]K = m(I -- 2m)K, or, alternatively:

m(2m -- 1)~ q % - 0

(AIO)

Within the limits of inequality (A10) failure occurs when o~ = K and is independent of oa. Outside these limits the failure circles will be tangent to the parabola at some point other than the vertex, the maximum and minimum principal stresses at failure being given by equation (A9). We may, therefore, formally describe the Generalised Fracture Criterion as follows: (i) iftn(2m -- 1)cq ~a ~> 0 failure occurs when: ~t = K. (ii) If m(2m -- l)e~ + 33 ~ 0 failure occurs when:
( % __ %)3

(AI 1)

~ + %

2 ( m _ I)3K[1 @

2K ~ { m - - 1~ 3 }1 ~r~ -!- % ~ 2 ] -- 1 .

(A12)

It will be seen that this reduces to the Griffith Criterion equation (4) and equation (5) when m -- 3 (i.e n = 8). APPENDIX EFFECT AND OF DIFFERENCE SPECIMEN 2 BETWEEN PLATENS ACROSS

IN ELASTIC PROPERTIES STRESSES SURFACES CONTACT

ON TANGENTIAL

DEVELOPED

It is assumed that the total platen face area is in contact with the specimen. Consider a small element of the contact surface with pressure p acting normal to it. Then: *~=,~=~ ~ x ( 1 - Vr) + vrp

1
(A13)
....

=~slo:c(lor, after rearrangement:

v~l)+ ~,,p]

--p

-~s

--

Er]

Er

'

(A14)

where ~ , *u are strains along the principal directions in the plane of the contact surface and other symbols are as defined after equation (18) in the main text.

Note added in proof--Since this paper was submitted, the author's colleagues at Nottingham have studied
the effect of tangential stresses on the stress distribution. It appears that the effect is negligible over the central ] of the disc. Thus, provided failure occurs in this region, the tensile strength may be computed on the assumption of radial loading. The analysis of tangential stresses will be discussed in a paper to be presented by Dr. Hackett and Mr. Addinall, at the Sixth Rock Mechanics Symposium, Rolla, Missouri U.S.A. in October 1964.

Das könnte Ihnen auch gefallen