Sie sind auf Seite 1von 7

The effect of aggregate size on the use of the cylinder splitting test as a measure of tensile strength

D.J. H A N N A N T (1), K.J. B U C K L E Y (2), j. C R O F T ( a )

In the design of concrete mixes for tensile strength it is often assumed that the cylinder splitting strength is a constant proportion of the direct tensile strength and is therefore an acceptable alternative. It is shown herein that aggregate size can have an important effect on the split cylinder/direct tensile strength ratio which, for the range of mixes examined varies between 0.95 and 1.28 for 102 mm diameter cylinders. It is suggested that this is caused by the complex compressive stress system immediately beneath the packers and experimental work to testthis hypothesis is described.

true tensile strength or a constant proportion of it. Techniques for the measurement of the " true " or direct tensile strength [6, 7, 8] have now become sufficiently sophisticated for this value to be accepted as an accurate estimate of the material property known as tensile strength. Any other test which purports to give an estimate of the same value can therefore be judged against the direct tensile strength.

Comments on previous work Introduction

The cylinder splitting test has been used for testing concrete for the past 25 years and has become sufficiently well established to be included as an indirect test for tensile strength in the recommended testing procedures of several countries [1, 2, 3, 4]. The main reasons for its introduction have been that the test is easy to carry out on a standard compression testing machine and it usually gives a value for the tensile strength of normal concretes between that of the direct tensile test and the modulus of rupture [5]. These reasons in themselves are worth very little if the measured value for strength is not, in fact, the (1) B.Sc., Ph.D., C.Eng., M.I.C.E., Lecturer in Concrete Technology, University of Surrey. (2) B.Sc., University of Surrey. (3) B.Sc., now Engineer with Balfour Beatty.

Excellent surveys of published work relating to the cylinder splitting test have been compiled by Popovics [9] and Franca and Pincus [10]. The bulk of established work has been concerned with the effects on the cylinder splitting strength of various specimen sizes, packer types and sizes, rates of loading, moisture condition and all the other variables which are known to affect concrete strength. A few papers [10, 11, 12] have described investigations into the strain pattern in the cylinders in order to estimate whether the strains predicted using the usual elastic theory [5] are achieved in practice. However, due to the length and width of the strain gauges and the effects o f aggregate size, it has not been possible either to accurately measure the biaxial strain distribution immediately beneath the packer or the peak strains in the high strain gradient across the cylinder diameter (Fig. 1). 15

VOL.

6 - N ~ 31 -

1973 - N A T I I : R I A U X

ET

CONSTRUCTIONS

Region
biaxial

of

high

compression

&/__.
Fig.
1. --

Theoretical stress distribution in the split cylinder test [5]. (Cylinder length

" L ").

Some recent work [8] has compared the direct tensile strength with the split cylinder strength and this has shown that aggregate size and volume concentration can have a considerable influence on the ratio split cylinder/direct tensile strength. The effect on this ratio of aggregate size and type has also been observed in studies on mortar and lightweight aggregate concrete for which it has been shown that the strength measured using the split cylinder test may be equal to, or less than, the direct tensile strength [ 13]. The reason for the effects of these aggregate parameters on the split cylinder/direct tensile strength ratio has not been entirely clear although a rational explanation for the effect of aggregate size has been proposed by Johnston and Sidwell [8] as follows... " The introduction of large particles of aggregate of greater tensile strength than the concrete as a whole means that diametral failure must occur through the stronger aggregate or else the failure path must lengthen by going round the aggregate through an area of lower stress ". Both these effects would increase the resistance to fracture. The outstanding omission in all the published work known to the authors is the lack of a critical study of the role played in the failure mechanism of the cylinder by the multiaxial stress system immediately underneath the packer (Fig. 1). This stress system according to elastic theory is one of high biaxial compression in the plane of the diameter but the sign and magnitude of the stress parallel to the cylinder axis is unknown and is likely to be dependent on the material of the packing strip. Also, the distribution of stresses in the biaxial compressive zone will be dependent on the stiffness and size of the packing strip and it has been shown [14, 15] that softer, thicker, packing strips generally cause an increase in the failure load of the cylinder. Since the effect of the type of the strip is likely to die away rapidly at a distance from the strip, the change in strength with packer type may well be a function of the change in the stress distribution immediately beneath the packer. Packing strip effects are also likely to be less significant as the cylinder diameter increases because the area affected is a reducing proportion of the total area under stress. 16

It is suggested that the lack of appreciation of the effects of this localised biaxial compressive zone on the mechan;sm of failure in the cylinder has assisted the general acceptance of this test. Another criticism of the published work is the general acceptance that although the compressiontension stress system in the cylinder splitting test should result in a strength lower than the direct tensile strength [16] this is not generally the case for normal concretes. Lack of an adequate explanation for this fact has generally been disregarded.

Suggested mechanism of failure of the cylinder


Figure 2 shows the suggested failure sequence. a) The tensile crack is initiated first and starts to propagate. b) Crack growth is inhibited by the biaxial compressive zones underneath the packing strip. c) Failure is only possible after the crack has propagated around the biaxial compressive zone or after this zone has failed in shear or diagonal tension. A single wedge is generally left behind but a secondary wedge can form at the other end as the cylinder rotates and breaks open. The alternative explanation for the failure mechanism is that the wedge forms first and then initiates the central tensile crack. If this is in fact the case, the standard formula for calculating the tensile stress at failure is even less relevant than in the case shown in Figure 2. Also, close observation of the crack development for concrete containing steel fibres suggests that this explanation is unlikely to apply.

Techniques for examining the importance with respect to failure load of the biaxial compressive zone under the packer
The importance of the biaxial compressive zone was assessed by testing 102 mm diameter cylinders which had zero tensile strength across the diametral plane,

D.J. H A N N A N T

. K.J. B U C K L E Y

- J. C R O F T

ltl
a) Crack initiation b) Crack growth c) Failure
Fig. 2.---[Proposed ~ i l ~ e s ~ u e n c e ~rthecylmder

but normal strength within the biaxial zone under the packing strip (Fig. 3). In Figure 3 (a) the tensile strength in the zone over the central 5/6 of the diameter was reduced to zero by means of a cast in paper insert. This long thin crack remained stable under the standard test procedure [1] until a wedge formed under one packer at a load approaching the normal failure load. This implied that the crack was unable to propagate through or around the compression zone until a wedge had formed. In Figure 3 (b) the technique was carried to the limit with two complete half cylinders placed together for the normal cylinder splitting test [1] using standard packing strips. Although there was zero tensile strength throughout the whole diameter, collapse of the cylinder could not occur until a wedge formed at one end o f the diameter at a high proportion of the normal failure load. Thus it is apparent that the tensile strength of the central section need not necessarily be a major parameter in the load carrying capacity of the cylinder,

or, as a corollary, the tensile strength of the material of the cylinder cannot necessarily be deduced from the failure load using the normal elastic analysis. The effect of maximum aggregate size on the measured strength using the test shown in Figure 3 (b) was investigated for concretes of similar direct tensile strengths and thus the effect of the aggregate size on the strength of the biaxial compressive zone could be examined in isolation from the effects of aggregate size in the tensile zone such as those postulated by Johnston and Sidwell [8].
E x p e r i m e n t a l work

The purpose of the experimental work was two fold : a) To examine the effect of aggregate size on the ratio, split cylinder/direct tensile strength for mixes of similar direct tensile strength. Both 102 mm diameter and 152 mm diameter cylinders were used for these tests with 12 mm wide by 3 mm thick plywood packing strips.

JJl

a)

Zero tensile strength for 5/6 diameter

b)

Two halves.

Zero

tensile strength across the diameter

Fig. 3. - - Techniques used to assess the importance of the biaxial compressive zone under the packer.

17

VOL.

6 - N ~ 31 - 1973 - M A T E R I A U X

ET CONSTRUCTIONS

b) To determine the contribution o f the biaxial compressive zone under the packing strip to the cylinder splitting strength using aggregates of different maximum size. This was investigated using cylinders cast as separate halves but tested in the usual way as whole cylinders (Fig. 3 b). the results being compared with whole cylinders and direct tensile tests. The majority of this work was carried out on 102 mm diameter cylinders. In addition to the above studies the effect of width of packing strip was removed almost completely in a few tests by the use of near knife edges in the form of hacksaw blades held in clamps. The horizontal stress distribution was then, in theory, tensile throughout the vertical diameter although the localised vertical compressive stress immediately beneath the blade was extremely high, and in fact caused local crushing of the material. The indirect tensile strengths measured using this test were in all cases lower than the direct tensile strengths which contrasted with the results when standard packing strips were used. The direct tensile strengths were measured on prisms I02 mm 102 mm 506 mm long using a scissor grip technique (See Figure 4) of a similar type to that described in reference [8]. The whole system was checked for eccentricity and consistency of results by Kolias [17] and the within batch coefficient of variation was known to be less than 6 % with an apparently insignificant eccentricity as determined from electrical resistance strain gauges. The cylinder size was either 102 mm diameter 204 mm long or 152 m m diameter x 150 mm long and the mix proportions are shown in Table I. The mixes were designed to have similar direct tensile strengths in order that differences in the ratio, split cylinder/direct tensile strength could not be attributable to different strength levels and this aim was broadly achieved.

Test results

a) Effect of aggregate size on the ratio split cylinder/direct tensile strength (Tables II et III). (i) 102 mm diameter cylinders. The cylinder splitting test underestimated the tensile strength of the mortars by about 5 %, whereas the same test overestimated the tensile strengths of the 9.5 mm and 19 mm aggregate concretes by 11 and 28 ~o respectively. It is therefore suggested that comparisons of the tensile strengths of concretes containing aggregates of different size cannot be reliably obtained from the 102 m m diameter cylinder splitting test.

(ii) 152 mm diameter cylinders. The differences between the split cylinder and direct tensile strengths were rather less dependent on aggregate maximum size than for the 102 mm diameter cylinders, possibly due to reduced packer effect. The 9.5 mm and 19 mm aggregate concretes showed similar results, with the split cylinder overestimating the strength by about 10 %. Similar tests were carried out using 38 mm maximum sized aggregate but as this was more than 88 of the minimum prism section for direct tension the detailed results are not reported here. However, the split cylinder strength was, on average, 17 % higher than the measured direct tensile strength as compared with the 35 % - - 70 % overestimate for 38 mm aggregates suggested by Johnston and Sidwell [8]. The authors' overestimate would have been less than 17 % if the large aggregate had, in fact, reduced the measured direct tensile strength below the true value. An important difference between the authors' test procedure and that of Johnston and Sidwell [8] is that in the authors' case the tensile stress in the prisms was applied at right angles to the direction of gravity during casting whereas in reference 8 the tensile stress was applied parallel to the direction of gravity during casting. Thus the effects of bleeding on the direct tensile strength [18] were probably less for the results shown herein.

TABLE I. - - Mix proportions by weight Maximum aggregate size 5 mm Cement Water added to air dry aggregate
5 mm I

9.5 mm
1

--?----

19 m m

0.57 3.42

0.40 1.30
1.58

Aggregate size 9.5 mm 19 mm


Fig. 4. - -

1.26

Scissor grip technique for measuring direct tensile

strength.

T h e s a n d grading was within zone 2 ot~B.S. 882 a n d ] 201 : 1965. All aggregates were T h a m e s Valley Gravels.

18

D.J.

HANNANT

- K.J. BUCKLEY

- J. C R O F T

TABLE II. - - Strength results for cylinder splitting (102 mm diameter cylinders) and direct tensile tests.

Maximum aggregate size mm

Average direct tensile strength MN/m 2 2.68 (11)

Average split cylinder strength MN/m 2 2.55 (14) 3.16 (18) 3.14 (12)

Ratio

split cylinder/ direct tensile

0.95 1.11
1.28

9.5 19

2.84 (17) 2.45 (11)

Note : Figures in brackets ( ) denote the number of specimens tested.

TABLE III. - - Strength results for cylinder splitting (152 mm diameter cylinders) and direct tensile tests

Maximum aggregate size

(ram)

Average direct tensile strength MN/m 2 *2.28 (9)

Average split cylinder strength MN/m 2 2.33 (9) 3.26 (13) 2.98 (13)

Ratio

split cylinder/ direct tensile

* 1.02 1.10

9.5 19

2.97 (15) 2.69 (13)

1.11

Note : Figures in brackets ( ) denote the number of specimens tested. * This series of tests had a high incidence of failures within the grips which did not occur with the other aggregate sizes. Any stress concentrations or multiaxial effects within the grip would tend to reduce the measured direct tensile strength and hence this ratio, (Column 4) is likely to be an overestimate.

Fig. 5. - - Failure types for half cylinders and whole cylinders (102 mm diameter) for maximum aggregate sizes of 5[mm, 9.5 mm and 19 ram.

b) Contribution of the biaxial compressive zone to measured cylinder splitting strength. As noted earlier, this investigation was carried out using 102 m m diameter cylinders cast as separate halves but tested in the usual way as whole cylinders. The types of failure for the half cylinders and whole cylinders for the three aggregate sizes are shown in Figure 5. It can be seen from the photograph that the failure pattern for a particular aggregate size is similar, whether or not the material has any tensile strength across the vertical diameter. A single shear wedge at least as wide as the packer is usually formed at collapse and it is noticeable that the width of the cracked zone adjacent to the packer, increases with maximum aggregate size. This was observed throughout the tests, the larger aggregates tending to give wider, rougher, wedges than the mortars for which the broken surfaces tended to be relatively smooth. Similar effects of aggregate size on the wedge were

also observed for whole 152 m m diameter cylinders (Fig. 6). It does not seem unreasonable to propose, therefore, that if the wedge is critical to the failure mechanism, then changes in wedge shape, for materials with the same tensile strength, will alter the maximum load sustained by the cylinder. A summary of the test results for the 102 m m diameter whole cylinders, half cylinders and direct tensile tests is given in Table IV. It can be seen from Table IV that the measured indirect tensile strength of the half cylinders, which was known to be zero across their diameters, was about 50 % of the normal cylinder splitting strength for mortars and was about 75 % of the cylinder splitting strengths for the larger aggregates. Also, the proportion of the direct tensile strength carried by'the biaxial compressive zone, in the half cylinders increased with increasing aggregate size up to 96 % for the 19 mm aggregate c o n c r e t e . 19

VOL.

6 - N ~ 31 -

1973 - M A T I ~ R I A U X

ET

CONSTRUCTIONS

Fig. 6. - - Effects of aggregate size on wedge shape for 152 mm diameter cylinders. Aggregate sizes were 5 mm, 9.5 mm, 19 mm and 38 n u n .

These results indicate that a high proportion of the load measured during a cylinder splitting test on a 102 mm diameter cylinder could be associated with the resistance to failure of the wedge zone immediately beneath the packerand relatively little of the measured load may be associated with the tensile strength of the material across the cylinder diameter. Thus, the tensile strength of the material as given by the standard cylinder splitting formula, may be only indirectly related to the true tensile strength, via the strength of the wedge when subjected to a complex multiaxial stress system. This may be affected by other parameters, such as aggregate size and concentration in a different way from that of tensile strength and hence it would appear that there can be no straightforward relationship between direct tensile and split cylinder strengths.
Conclusions

indirect tensile strength of the material as estimated from the usual elastic analysis. 2. The stress distribution in (1) above, affects concretes with different maximum aggregate sizes, to different degrees and hence there can be no simple correlation between the cylinder splitting strengfh and the direct tensile strength of a wide range of concretes. 3. As a result of (i) and (2) above, great care should be exercised when basing mix design for tensile strength on the split cylinder test. 4. The split tensile strength obtained from the 102 mm diameter cylinders allowed by B.S.1881.1970 [1] is particularly affected by aggregate size and consideration should be given to deleting this test from the standard. Acknowledgements The authors would like to express their appreciation of the help and guidance provided throughout this work by Mr. R.I.T. Williams, Reader in Construction Materials, at the University of Surrey.

1. The complex compressive stress system immediately beneath the packers in the cylinder splitting test can have a considerable influence on the failure load of the cylinder and hence on the

TABLE IV. - - Strength results for splitting whole and half cylinders (102 mm diameter) and direct tensile tests Maximum aggregate size
1TIII1

Average split cylinder strength (whole cyls.) MN/m ~ 2.61 (8)

Average split cylinder strength (half cylinders) MN/m ~ 1.33 (8) 2.52 (9) 2.40 (9)

Strength ratio Half cyls./whole cyls. 0.51 0.75 0.75 ! I i

Average direct tensile strength MN/m 2 2.76 (8) 3.03 (9/ 2.50 (9)

Strength ratio Half cyls./direct 0.48 0.83


0.96

9.5 19

3.36 (9) 3.22 (9)

Note : Figures in brackets ( ) denote the number of specimens tested. 20

D.J.

HANNANT

- K.J. BUCKLEY

- J. C R O F T

RI~SUM1~ Effet de la dimension des granulats sur la mesure de la r6sistance en traction par ressai de fendage sur cylindre. - - Lorsqu'on dose le bdton pour une certaine rJsistance & la traction, on admet souvent que la rJsistance au fendage sur cylindre est en rapport constant avec la rJsistance en traction directe et que, par consdquent, elle constitue une alternative acceptable. On montre ici que la dimension des granulats peut avoir un effet important sur le rapport rJsistance au
fendage sur cylindre/r~sistance en traction directe qui, pour les mdlanges dtudiOs, varie de 0,95 d 1,28 pour des cylindres de 102 m m de diamktre. On sugg~re que la cause peut en ~tre le s y s t k m e complexe de contrainte en compression qui se situe imm~diatement sous les bandes d'interposition ; on ddcrit l'~tude exp~rimentale entreprise pour v~rifier cette hypothkse.

REFERENCES
[1] B.S. 1881. 1970. Part 4. - - Methods o f testing concrete f o r strength. [2] Department of the Environment. - - Specification for road and bridge works. London, 1969. (H.M. Stationery Office). (4th Edition as amended by Technical Memorandum H_ 10/71). [3] A.S.T.M. C496-62T. - - Method o f test for splitting tensile strength o f moulded concrete cylinders. [4] International recommendations f o r the design and construction o f concrete structures. Principles and Recommendations. June 1970. CEB-FIP Sixth Congress, Prague. [5] WRIGHT P.F.J. - - Comments on an indirect tensile test on concrete cylinders. Magazine of Concrete Research No. 20, July 1955, pp. 87-96. [6] ELVERY R.H., HAROUN W. - - A direct tensile test for concrete under long or short term loading. Magazine of Concrete Research Vol. 20. No. 63. June 1968, pp. 111-116. [7] HUGHES B.P., CHAPMAN G.P. - - Direct tensile test for concrete using modern adhesives. Bulletin Rilem, No. 26, March 1965, pp. 77-80. [8] JOHNSTON C.D., SIDWELL E.H. - - Testing concrete in tension and in compression. Magazine of Concrete Research, Vol. 20, No. 65, Dec. 1968, pp. 221-228. [9] PoPovtcs S. - - Relations between various strengths o f concrete. Highway Research Board Record, No. 210, pp. 67-94. [10] FRANCAG., PINCUS G. - - The distribution o f concrete strains in the split cylinder test. Journal of Materials, Vol. 4, No. 2, June 1969, pp. 393-407. [11] SEEFRIED K.J., GESUND H., PINCUS G. - - A n experimental investigation o f the strain distribution in the split cylinder test. Journal of Materials, Vol. 2, No. 4, Dec. 1967, pp. 703-718. [12] ADDINALL E., HACKETT P. - - The effect o f platen conditions on the tensile strengths o f rock-like materials. Civil Engineering and Public Works Review, Oct. 1964, pp. 1250-1253. [13] WARD M.A. - - Journal of Materials, Vol. 2, No. 1, March 1967, pp. 191-195. (Written discussion on paper by Malhotra V.M. and Zoldners N.G.) [14] SINGH-CHANNEJ. - - The effect o f testing conditions on the apparent indirect tensile strength o f concrete. Thesis. Battersea College of Technology (Now University of Surrey), 1964/65, London.
[15]

SPOONERD.C. - - Measurement o f the tensile strength o f concrete by an indirect method ~ the cylinder splitting test. Cement and Concrete Association. T R A 419 May 1969.

[16] NEWMAN K., NEWMAN J.B. - - Failure theories and design criteria f o r plain concrete. Paper 83. International Conference on Structure, Solid mechanics and engineering design. Southampton University, April 1969. [17] KOLIAS S. - - Thesis in preparation. University of Surrey. [18] HUGHES B.P., ASH J.E. - - Anisotropy and failure criteria for concrete. Rilem. Materiaux et Constructions, Vol. 3, No. 18, 1970, pp. 371-373.

21

Das könnte Ihnen auch gefallen