Download as pdf
Download as pdf
You are on page 1of 21
210.5 Technical Report 543 Some aspects of modelling the behaviour of reinforced concrete under shear loading P.S. Chana Bsc(Eng) CEMENT AND CONCRETE ASSOCIATION Contents 0 0 un R B B 16 18 a 21 Summary Introduction Review of recent developments in ideas about the behaviour of reinforced concrete beams in shear Effect of beam depth upon the shear strength Experimental work Geometric similitude ‘Scaling of the concrete mix Scaling of reinforcement Size of control specimens ‘Testing of beams in series 2 to 5 Manufacture and testing of series 6 beams iscussion of results Experimental results Cracking similitude Strength similitude Conclusions Acknowledgements References Summary Introduc: This report is intended for research workers and model-testing engineers involved in investigating the behaviour of reinforced concrete structures under shear forces, it describes the results of a series of 36 tests on 400 mm deep by 200 mm wide proto type beams with no shear reinforcement, similar half-depth beams and 1:2, 1:33 and 1:8:5 seale model beams. The size of the aggregate was also scaled. The structural behaviour, including cracking and ultimate strength, of the models and prototype is compared. The implications of the results obtained for the analysis of model tests are discussed and reasons for any scale effects observed are proposed. Finally, Possible improvements in model techniques where shear is being considered are discussed. Model tests have been carried out, in the last 20 years, on a variety of structures such as bridges, cooling towers and multi-storey buildings. This period has also seen the development of a model concrete, or microconcrete, which realistically simulates the mechanical properties of structural concrete. Some examples of the early tests carried out at the Cement and Concrete Association are given in reference 1. These models, besides being useful architectural aids, were used to confirm theoretical design calculations and were not intended for development work. In recent years, the role of models has undergone a decisive metamorphosis. Models are now widely used as a design tool for yielding information on a whole variety of structural responses such as deflections, ultimate loads, cracking and modes of failure. Small-scale models are being increasingly utilized in research on structural concrete and for development work in complex structures which cannot be rationally analysed by theory. This trend has been accentuated by the adoption of limit-state methods of design. Itis therefore essential to know the modes of failure and ultimate strengths of complex structural systems or individual members. Limit-state methods of design are now also concerned with serviceability conditions, such as cracking and deflections, as well as ultimate. Strength models obviously have great potential in this field, provided they can adequately simulate both the elastic and the inelastic response of the structure. As a result, models are now being called upon to check and develop design criteria for numerous concrete structures, including off-shore platforms, pile caps, bridge structures and shell structures”, Clearly, the model-testing engineer must, at some stage, convert his model-test results into values reflecting the behaviour of the prototype. Unfortunately, a proper scientific basis has not yet been established for modelling all aspects of structural behaviour; in particular, the behaviour in bond and shear needs examination, The situation is further complicated by the phenomenon of ‘size effects’. Many researchers have observed that the strength and mode of failure of models varies as the size of test specimen changes". This problem can be overcome if the actual magnitude of the size effect is known and suitable adjustments are made to the predicted prototype behaviour. Hence, there is a need to develop for each structural type ~ e.g. beam. plate, shell - suitable experimental data to correlate the structural behaviour of prototype structures with that of differential scale models. There is also a need for fundamental research on modelling, on a single-element basis, to determine why size effects occur and to advance sensible hypotheses as to how these can be elimin- ated, This work could then be applied to more complex systems such as slabs or pile-caps. Tn view of the above considerations, recent work on modelling at the Cement and Conerete Association has concentrated on fundamental studies in preference to complete structural model tests. Three parallel studies were initiated to look at the modelling aspects of shear, bond and flexure. This paper considers the results of the shear study. It is hoped that this work will help research workers and model-testing engineers to choose their scaling parameters effectively and relate their model results to prototype behaviour. ‘Tests were carried out on model beams of different scales with a view to correlating the structural behaviour under a shear mode of failure with particular emphasis on ultimate strength, These beams were reinforced with a main steel percentage of 1-8 land had no shear reinforcement. The maximum aggregate size was also scaled. Crack patterns, strains, deflections and ultimate loads were recorded for each series of beams. These are compared and possible reasons for any lack of similitude are given. Finally, the implications of the results obtained on mode! analysis are discussed. Review of recent developments in ideas about the behaviour of reinforced concrete beams in shear Before the results obtained are presented, it will be useful to summarize present ideas about the behaviour of reinforced concrete under shear loading. Fundamental research on the behaviour of reinforced concrete beams under the combined action of shear forces and bending moments during the last decade has radically changed many ideas about the mechanisms of shear force transfer. In particular, this research hhas highlighted the transfer of shear forces across cracks by aggregate interlock action and dowel action. The free-body diagram in Figure 1a shows the shear force in a cracked section of a reinforced concrete beam to be resisted by the compression zone, aggregate interlock action and dowel action, Some work has been done to try ‘and quantify the relative contributions of these actions by monitoring displacements in a conventional beam test and relating these to model tests with similar stress states", Work by Taylor® at the Cement and Concrete Association showed that the relative proportions of the three mechanisms of carrying shear force in a typical beam were: compression zone 20-40% terlock 33-50% dowel action 15-25% ‘At this stage, it is worth while considering the behaviour of a reinforced concrete beam without any links as itis loaded right up to shear failure. Before any cracking, the beam acts as a composite member composed of concrete and steel. As the beam. is loaded up, flexural cracks form on the tension face of the beam at intervals along the span depending on the bending moment and the steel ratio. A redistribution of stress due to the modified action of the member takes place. Measurements show that there is an increase in the rate of deflection with 2 corresponding change in the steel and concrete stresses. The flexural cracks extend a short distance up and then start to become inclined, primarily owing to the action of the shear stresses induced. A series of these cracks can form at approximately equal spacing with decreasing inclination towards the support. The appearance of diagonal cracks leads to further redistribution of stresses. At this stage, the aggregate interlock mechanism ominates and the shear force carried by the longitudinal reinforcement by dowel action starts to increase rapidly. The appearance of the critical diagonal crack leads ' ae (2) tmcemal force disribution (6) diagonal cracking ate : (C1 (0) flexural cracking (4) erack pater at fllure Figure 1: Internal fore dstnbution and development of eracking under shear loading. to an unstable force system resul illustrated in Figures 1b-d. The term ‘shear failure’ has often been loosely applied to many different failure : ‘modes, resulting in considerable confusion. The mode of failure considered in this paper is sometimes called a flexural shear failure, in which a diagonal crack usually develops from a previously formed flexural crack accompanied by the splitting of the concrete at the level of the steel. This type of failure is common for rectangulat beams with moderate steel percentages (0:5~2-5%) and shear span to depth ratios of greater than 2-5. Most of the fundamental work on shear has been concentrated on this particular failure mode. 2 in failure. The development of cracking is Effect of beam depth upon the shear strength Test work has ‘shown that the relative shear strength of a reinforced beam without shear reinforcement decreases as the beam depth increases", Most of this work was carried out before recent ideas on the mechanisms of shear force transfer had been fully developed. Hence, litle attention was paid to scaling concrete cover and reinforcement layout (affecting dowel strengths) and maximum aggregate. size {affecting interlock). Taylor argued that the increased shear strength of shallow beams was due to lack of scaling of the aggregate size. Since crack widths in shallow beams are smaller, they would have enhanced interlock strengths leading to greater shear capacity. Taylor carried out a limited number of tests on beams ranging in depth from 250 mm to 1000 mm with the size of aggregate scaled and observed a lesser scale effect in comparison to previous investigations. However, he accepted that, even when the aggregate is scaled, beams tend to lose their relative strength as their size increases. In practice, the concrete mix used would probably be the same, regardless of beam size. CP110'* recognizes this phenomenon of size effects and allow a depth factor &, to enhance the shear strengths of beams less than 300 mm deep. Since the design implications of the size effect are not the main concern of this paper, these will not be considered any further. The reader is referred to Taylor's paper, Experimental work “The details of the beams tested are given in Table 1. Five series of beams were tested." All the beams except series 5 were designed to be exact geometrical models of the prototype beams in series 2. The beams had a main stect percentage of 1-8 with no shear reinforcement, The shear span : depth ratio was 3-0. The beam dimen- sions are shown in Figures 2 and 3. This test series was designed to conform to the Jaws of similitude for modelling shear behaviour which are summarized below. Table 1 Details of beams. Maximum Beam | Overall Bttecive — Diameter | Maximum No. depth epi of steel (om) (mm) (em) (mum) (mm) (am) SERIES 2 (PROTOTYPE) 2aa 20 2b | 20 228 10 pee ae 356 203 20 aa 2a | 20 230 2» 5 senies 3 (1i2scatey 3a Bub 328 202 7 100 10 10 SERIES (113.99) aia 4b 428 4b 430 4b 4a | 4b rer) 106 60 6 5 SERIES 6 (156.5 SCALED 64 62 63. 64 65 a a 235 232 24 66 67 68. 63. I! SERIES 5 (HALF-DEPTH BEAMS) sa Sub AM 200 170 200 20 ee 526 20 “It was originally intended to test some larger specimens called series 1, but it was decided not to Proceed wilh ths series. 7 —— ae ky R in eoee noo “ e ° SER a ee Figure 3: Cross-secional dimensions of beams tested (10 sole). Geometric similitude In addition to scaling the over-all dimensions, particular attention was given to modelling the cover and the layout of the main steel. These two parameters have been found to have a significant influence upon the dowel strength. The beams had over-all depths of 400, 200, 120 and 47 mm with corresponding bar diameters of 20, 10, 6 and 2-35 mm. The cover to the main steel was twice the bar diameter in each case. Scaling of the concrete mix For modelling the behaviour of reinforced concrete in shear, the concrete strengths in compression and tension must be correct. The need to scale the mix has been ‘discussed previously. The complete aggregate grading curve cannot be scaled, especially when small aggregate particles are being considered. However, it is acceptable to scale just the maximum aggregate size, since it is this parameter and not the complete grading curve that has the greatest effect upon aggregate interlock action. Finally, when post-cracking behaviour is being considered, a necessary similitude requirement is for the complete stress-strain curve to be identical with that of the prototype concrete. ‘The concrete mix for series 2 and 3 beams was designed to give a 28 day strength of 40 Nimm? (actual strengths in Table 3). The microconecrete mix for beam series 4 and 6 was designed by using the figures in reference 15. The grading curves for all the beams tested are shown in Figure 4. Irregular gravel aggregate and sand were used in each case. Table 2 gives the water/cement and the aggregate/cement ratios for each of the mixes used. Table 2 Details of concrete mixes. beter) Maximum aggregate Wiaterlcement ‘Aaeregatelcement Re ft a wate Dads 20 050 4 ‘ $ o2 to ‘ 23 on 30s ‘Table 3 Strengths of control specimens. Susot fom [a2 eae rem ben eee ie 24 475 S04 524 48.9 48:2 484 | 493 | 360 337 3:47 342 244 339 335 22 | 150 |ao0 dow a7 aaa 430 ate | are | 290 286 313 298 257 263 | 285 3a Meteeeseeeecescsscccseeecce eee 297293 28 — — — 292 3 doa ear — = = | dee | 200 3m 208 — — — | aor 33 too [406 395 402 — = 401 | 242 273 28 — — — | 267 Di amen: = = = | 36] gear as 2 > > | aS D2 fai aig are > DL | ha | dean ae — = x | an bs [asa ag ae — =~ > | 7 | Sts 303 360 — 3s a Jara 308 324 322 326 300 | 309 | 332 351 308 321 301 301 | 322 2 S38 512 sie ses 500 S20 | s22 | 366 361 348 396 347 ass | 366 si | qq |e @9 a4 — = [ws | 330 3% a — — — | am 33 bes 3 a2 — — — |r| diam am — — > | am 61 |a58 384 358 344 365 327 | ase | 336 325 265 377 341 306 | 325 oa 363. 392 373 371 389 302 | a0 | 206 281 88 203 2m 292 | oe 63 Wea as 343 392 382 a60 | ot | fon gon sae tan os Gul | 379 oa SG Ga ard Ga S21 | Goo | 3m San tan den aoe 352 | UR os as |sre 313 459 603 42 et | Geo | dor aan tw an ta sis | 40 6s G9 ora on6 643 $86 coo | oso) 326 342 t93 Sa an 34s | 499 67 441 406 474 432 406 427 | 431 | 328 274 346 304 286 310 | 303 sa 452 383 aoa doe a52 art | age | a91 301 3a2 346 308 29s | 308 Figure 4: Aggregate grading curves Scaling of reinforcement ‘The stress-strain requirements of the tensile reinforcement in the model beams may be relaxed if shear is the failure mode under consideration. It is more important to reproduce the concrete properties accurately. However, the two properties of the reinforcement which must be similar to the prototype are the modulus of elasticity and the bond properties. Similarity of the modulus of elasticity ensures deformation similitude; it is necessary for model beams to simulate prototype deformations since this affects shear force transfer in the compression zone as well as that due to aggregate interlock and dowel action. The bond properties of steel also affect the behaviour of reinforced concrete beams in shear. Poor bond quality will result in a shear crack spreading further upwards, leading to greater shear rotations and premature failure. The tensile reinforcement in beam series 2, 3, 4 and 5 consisted of high-yield G K Torbar with diameters of 20, 10 and 6 mm. The mode! beams in series 6 were reinforced with 6BA steel studding with an ‘effective diameter’ of 232 mm. The stress-strain curves of the various diameter steel reinforcement are given in Figure 5. ‘The longitudinal steel in beam series 2, 3, 4 and 5 was provided with L-shaped end anchorages. Size of control specimens ‘The model-testing engineer must satisfy himself that the size of the control specimen used will not lead to misleading values of concrete strength. Sabnis* has suggested the following empirical rules to select the model cylinder or cube size. (1) The strength of the model material is equal to that of a test cylinder with diameter equal to the minimum dimension of the structure in the region of failure, e.g. the shell thickness or the width of the beam. (2) The size of the largest sand particle used in the mix may not be larger than ‘one-fifth to one-eighth of the diameter of such a eylinder, or 80% of the distance between reinforcing bars in the model 1 2 - Jaw STRAIN te STRAIN x 1" Figure Sb: Sirest-strain curve for 6 mm dia. reinforcement Figure Sd: Stess-stain curve for 20 mm dia reinforcement ‘The strength of the concretes for the respective beam sizes was determined from 150 mm, 100 mm, 70 mm and 25 mm cubes. The splitting strengths were determined from control specimens of similarly scaled sizes. Testing of beams in series 2 to 5 ‘The beams in series 2 to 5 were cast in wooden moulds and compacted by internal vibration. Six cubes and six cylinders were cast as control specimens for each beam. Alll the test specimens were stripped after 2 days and similarly cured under polythene until ready for testing. The beams were tested, one at a time, in an inverted position supported by a jack (see Figure 6). The upward force from the jack was balanced by Figure 6: Test arrangement the two end reactions, which were monitored by separate load-cells. Therefore, no correction for self-weight effects need be applied in determining the shear force. ‘The manufacture and testing of the model beams of series 6 will be described separately ‘The beams were all tested to failure under a point load. The loading and support plates were scaled as far as possible in order to minimize any effects due to stress concentrations. The load was applied in uniform predetermined increments. After each increment of loading, deflections, strains and crack patterns were recorded while maintaining a constant deflection. 50 mm Demec gauges were used to read the strains, Manufacture and testing of series 6 beams ‘The mixing for the model beams in series 6 was carried out in a small planetary mixer for 30 seconds dry and 2 minutes wet. The beams and control specimens were cast in specially adapted metal moulds and compacted on a mains-frequency vibrating, table. The steel studding used as reinforcement was held in place during compaction by passing through holes drilled into the end-plates and clamped by nuts on either side. These model beams had an adequate length beyond the supports to satisfy bond requirements. The model beams were tested in the rig used by Evans and Clarke for comparing \ the flexural behaviour of small-scale microconcrete beams‘"?, The test arrangement is shown in Figure 7, A two-point load arrangement is used with a constant moment zone 112 mm long. The load is applied by means of a micrometer screw-jack acting through a load-cell, a spreader beam and a captive ball and roller system. The ends of the beams are supported similarly. Strain profiles were monitored by 12:5 mm. electrical-resistance strain gauges stuck to the surface of the beam. The model beams were tested in a similar manner to the beams in the other series. Figure 7: Microconcrete beam (series 6) under ts. Discussion of results Experimental results ‘All the beams tested failed in shear. Typical crack pattems at failure for each series, of beams tested are shown in Figures 8 to 12. The failure loads and shear strengths of the beams are summarized in Table 4. To allow for the differences in the cube strengths, the shear resistance has been standardized to a 40 Nimm! cube strength by assuming that itis directly proportional to the square root of the cube strength. The t size of the control specimens used to determine the concréte properties were given in Table 3. The observed variation in shear strength with model size is shown in | Figure 13. ‘Crack spacings, first cracking load and the total number of cracks within the shear span were determined for each beam tested. Intermediate cracks which did not extend to the level of the reinforcement have been ignored. Table 5 summarizes the average values obtained for each series of tests. The first cracking load is presented as a percentage of ultimate load. Figure 9: Typical crack pattern at failure for beam series 3 (depth = 200 mm). Figure 11: Typical erack patter t failure for beam series § (depth = 200 mm; rot ite geomeric model). Figure 12: Typical crack pater at fllure for beam series 6 (depth = 47 mm) Table 4 Shear strengths of beams. Bam | Failure No. j toad | oo) 2ie 360 2 | on 22a ar 22 94 23a 994 23 96-4 Bia 28 134 ia Bab 29 135 145 32a 285 138 bas 32 255 vas 150 33a 265 | 150 150 33 22 131 131 Di m1 125 rat ba 234 | 132 va7 D3 a4 1a hs ‘ae 992 ise 175 5 865 136 135 428 an Ma 160 42 on 133 176 438 1.70 184 161 430 1235 194 70 aia 962 ist 132 a 1047 165 has Sia we rat 140 Sub a8 rat 140 52a 550 162 163 5.26 56-0 165 166 6a 2a 27 229 62 192 195 200 63 210 213 215 64 209 212 173 65 26 264 24 66 24 278 220 67 236 239 230 68 204 227 220 69 229 232 219 jo cae pt nn gent od) 6 0 ei 2.3.4.8 AVERAGE SHEAR STRESS.¢=Vibd —Niew Figure 13: Relationship benwcen ultimate shear strength and scaled depth. Cracking similitude Figures 8 to 12 show that the general crack patterns at failure are remarkably similar, with the exception of series 5. Major cracks only have been considered in determining, cfack spacings. These have also been scaled reasonably as shown in Table 5. However, the correlation between crack spacings would be poor if intermediate cracks were also considered. Investigations into the flexural behaviour of model beams have usually revealed a wider spacing of cracks in models", It is likely that the good correlation between the major crack spacing obtained in this investigation is due to the use of threaded rod as reinforcement for the microconcrete beams. This is confirmed by recent study of the flexural behaviour of mode! beams carried out at the C&CA by Evans and Clarke", who showed that the use of threaded rod as reinforcement led to crack spacings similar to those of the prototype beams, whereas plain-wire or rolled-wire reinforcement results in larger crack spacings, possibly owing to variable bond between the steel and the concrete, The cracking similitude for series 5 beams is poor. This is not surprising, since these beams were not true geometric models, having only the right proportion of steel but a different aspect ratio. These results ‘emphasize the need for correct scaling in model testing, Table 5 shows that the onset of cracking is at a consistently higher load stage than the model beams. It has been suggested that the later onset of cracking in models is due to the difficulty in detecting thin cracks in models and hence that ‘visible’ cracks seem to appear at a later load stage". It is the author's opinion that this is not so. ‘A powerful microscope was used to observe cracking in the microconcrete beams. ‘The beam was loaded up slowly at loads approaching the cracking load. Cracks tended Sto appear suddenly between two load stages; these cracks were clearly visible through the microscope although there had been no previous sign of any gradual opening up of the crack. It would seem that any crack has a finite minimum width which can casily be picked up by a powerful microscope. This isan area that needs further study. ‘The effect of “delayed” cracking in model beams upon their ultimate strength will be considered in the next section. At this stage, the influence of strain gradient upon the tensile strain capacity will be discussed. Strain measurements showed that the model beams had a strain state similar to that of the prototype for a corresponding load stage; it follows then that the strain gradient in the model is magnified by the scale factor. The sensitivity of concrete to a strain gradient is well known and has been put forward as the reason for the discrepancy between the tensile strength as obtained from a direct tensile test and a modulus of rupture test. The effect of strain gradient upon tensile strain capacity has been studied quantitatively by Blackman et al”. To overcome the problem of isolating the strain gradient effect from a size effect, Blackman carried out tests on concrete prisms of the same size but subjected to different combinations of axial, eccentric and flexural loads giving different strain gradients through the specimen (see Figure 14). The relation between the ultimate tensile strain and the strain gradient is shown in Figure 15. Blackman concluded by stating that “there seems to be little Table § Cracking ilitude (average values). Beam | Avenge padre (mm) | Nonberot oad at onc na (ced pace cactsin, | _crtking ara percentage No | __inpereaese} shearspan orate tnd 2 7 “ 2 | 6 ” 4 | 5 585 + | s « » | 4 ” fot true geometrie models. (a) uniform tensile strain distribution produced by axial loading (4) equal tensile and compressive sirans produced by flexure (0) non-uniform este srain dsrbuton produced by -{— + ‘een loading 7 i (e) compressive strain larger than tensile strain produced by axial loading and flexure (6) tensile sain larger thon compressive sain produced by (compressive stain larger than tensile strain produced by ‘axial loading and flexure eccente loading Figure 14: Sorin disribuion obzained by various loading system (afer Blackman etal"), = slope of stain disibuton ULTIMATE TENSILE STRAIN 1 SLOPE oF stmAIN oistmiBuTION— S201 Figure 18: Effect of sain distribution upon theultimate tensile strain doubt that the strength of concrete is related in some way to the rate of change of stress across the section". His results clearly show the increase in tensile strain capacity with increasing strain gradient. No satisfactory explanation has yet been given for the mechanism associated with this behaviour. However, it is clear that the later onset of, cracking in model beams can be attributed to the high strain gradients present in these beams, Strength similitude tis apparent from Figure 13 that a significant scale effect has been observed despite scaling the aggregate size. The results of the series 5 beams have not been taken into account when fitting a best-fit curve through the experimental points. As mentioned before, these beams were not exact geometrical models, having 20 mm diameter bars as reinforcement as compared to 10 mm bars in the series 4 beams of the same depth; hhence crack widths would be different, leading to varying interlock strengths and dowel stresses. The 47 mm deep beams are approximately 90% more resistant in shear than the 400 mm deep beams. These beams, however, represent an extreme case of scaling; the scale effect observed for normal laboratory-size specimens which rarely tend to be less than 120 mm deep is less severe, being approximately 40%. How can this seale effect be explained? The common argument about lack of similitude of aggregate interlock is unacceptable because the aggregate was scaled. The higher strengths observed for the model tests described in this report can be attributed to the high strain gradients present and partly to the high tensile strengths Sof model concretes. The effect of a high strain gradient upon the onset of flexural cracking in model ‘beams has been discussed in the previous section. In all the beams tested the majority of shear cracks formed as extensions of existing flexural cracks; a few web shear cracks were observed. Flexural cracking and the consequent redistribution of stresses have profound effect upon the formation of flexural shear cracks. It follows, then, that delayed flexural cracking will lead to the formation of a shear crack at a comparatively later load stage. Table 6, which is derived from Figure 13 and Table 2, demonstrates that a major proportion of the ultimate shear strength scale effect can be attributed to ‘elastic’ loading, ie. from zero load to first cracking, The ‘residual shear strength’, or ‘post-cracking shear strength’, defined as the difference between the ultimate load and the first cracking load, is almost constant irrespective of scale. In view of the above discussion, it is useful to make comparisons between the behaviour of model beams in flexure and shear. High strain gradients in model beams will not have a major effect upon the ultimate strength in flexure which is determined primarily by the yield strength of the steel, and only to a limited extent by the concrete capacity. On the other hand, when shear behaviour is being considered, failure is initiated by the formation of a diagonal crack in the concrete which in turn is influenced by the strain gradient; the steel does not necessarily yield. Hence, itis reasonable to expect the scale effect to be greater than for flexure. Experimental results confirm this conclusion. Swamy" and Walraven have both mentioned the influence of high strain gradients upon the shear strength of model beams ‘The behaviour of a model beam in shear is also dependent on the tensile strength of the concrete. This has been the subject of a great deal of research effort in recent Table 6 Comparison of shear stress at first cracking and ultimate load. A Load at onset 5 Resa Beam timate of eacking Shear stress shear stress series shear stres asa percentage atonet of TAB No. | (Grom Figure 13) of ulimate cracking (Ninn) toad (Simm!) (nm) 2 108 “ 048 060 3 135 3 066 069 ‘ 156 585 091 065 6 2.08 37 137 0-67 years. It is well known that model coneretes generally have higher tensile strengths, which will aso lead to higher eracking loads); in addition, in the post-cracking regime, the ultimate interlock stresses will be enhanced. Recent work at Stuttgart __ University” has made it possible to make the fundamental properties of concrete, .e. tensile strength, compressive strength and modulus of elasticity, independent of one another by the use of different admixtures. Thus, the tensile strength could be reduced while the compressive strength and the modulus of elasticity were kept the same. In this series of tests, the average value of the ratio of cube strength to tensile strength was 14:7 for the 150 x 300 mm eylinders (20 mm aggregate) as compared to 11-9 for the 25 x $0 mm cylinders (2-4 mm aggregate). This represents a 23-54% relative increase in tensile strength for the 1:8:5 scale model beam as compared to a 90% increase in shear strength. The scale effect observed can therefore be partly attributable to the high tensile strength of model concretes. The effect of the use of a low-tensile-strength microconcrete upon the shear strength remains to be seen; however, itis unreasonable to expect that the scale effect will be completely eliminated. BO selene y -7E 08H C—O lamnemcraapete cont y= O74 = 085% =-O samen savegnecmeta, p= N69 150% Figure 16: Ulimate nominal shear stresses as a function ofthe effective beam depih as obtained by Walraven™. “The posteracking shear stresses, where aggregate interlock starts to become the dominant mechanism of shear force transfer, are derived in Table 6. It can be seen that these are almost constant irrespective of scale. This analysis appears to contradict the normally accepted explanation of scale effects, i. lack of interlock similitude of aggregates. This explanation has also recently been challenged by Walraven'*? who tested lightweight-aggregate concrete beams of different depths and compared these results with beams of normal-aggregate concretes. The same concrete mix was used for each beam depth with only the ype of aggregate as a variable for each series in test. For the tests on lightweight concrete beams, expanded clay was used. The cube strengths of all concretes was approximately 35 N/mm*, The results obtained are shown in Figure 16. It is clear that, despite the markedly different aggregate interlock characteristics of normal gravel aggregates and lightweight aggregates, the scale effect is identical for the two sets of beams. From these results, Walraven inferred that it was highly unlikely for the enhanced shear strengths of shallow beams to be due to high interlock strengths. ‘A similar comparison can be made of the results presented in this report with those ‘obtained by Leonhardt‘. Leonhardt carried out tests on completely similar beams of depths 80, 160, 240 and 320mm with steel diameters 6, 12, 18 and 24 mm respectively. However, the conerete mix used was the same in each case (maximum aggregate 1 0 eee ae Figure 17: Comparison of test results with those of Leonkardt size = 15 mm). Figure 17 shows the results obtained and a comparison with those discussed in this report. It is clear that the scale effect observed with the aggregate scaled is only slightly less than that obtained without scaling the aggregate. This also suggests that aggregate interlock effects are only partly responsible for the scale effect. The discussion above has focussed on some of the problems associated with ‘modelling the behaviour of reinforced concrete beams under shear forces and bending ‘moments. It is clear that the scaling of aggregate and use of low-tensile-strength microconcrete will result in only slight improvement of model behaviour. Strain- gradient effects present the model engineer with a truly difficult situation, It is really the sess gradient that is important rather than the strain gradient. The two terms can be interchanged in the context of this paper since the values of the modulus of clastcity are similar. It is possible to obtain similar siress gradients by employing a model material with compressive and tensile strengths similar to those of the proto- type, but with the modulus of elasticity of both concrete and steel reduced by the scale factor. The practical difficulties in the development of such a material are formidable. Conclusions (On the basis ofthe results obtained, the following conclusions are drawn with reference to the use of models in simulating the behaviour of reinforced concrete elements under the action of shear forces. (1) Satisfactory similitude of the over-all crack pattern and failure. mechanism was achieved for all true geometric models, (2) Provided due attention is paid to similitude requirements, elastic strain states similar to those of the prototype can be obtained on model beams. (3) Model beams are relatively stronger in shear than their prototypes. It seems unlikely that this scale effect can be wholly attributed to the lack of similitude of aggregate interlock action. Even if the aggregate is scaled, a significant scale effect is observed. (4) Models exhibit higher cracking loads, possibly owing to the influence of high strain gradients. tis thought that this, in turn, retards the formation of diagonal cracks leading to higher shear strengths. Model-test results, from small-scale beams in ticular, should be treated with caution when a shear mode of failure is being concerned. The scale effect observed may cast doubts on the validity of certain small. scale model tests, since a possible shear failure mode of the prototype may not be detected in the model. However, the strength enhancement for ‘normal’ laboratory- size specimens is not serious, being less than 40%. Nevertheless, it is useful to select, the model size to be as large as possible (5) The complete elimination of scale effects is difficult. Geometric scaling leads to strain-gradient effects. Scaling of the mix results in changes in the fundamental mechanism of crack propagation which in turn is influenced by variations in cement content, angularity of aggregate and the actual size of the crack. (6) The scatter of results from model tests is large. Where possible, any model test concerned with shear behaviour should be repeated to check the reliability of the data obtained. ‘This study has also identified areas that need further investigation, These are listed below: (1) the nature of the mechanisms of microcracking in microconerete and a com- parison with normal concrete; (2) the effect of using a low-tensile-strength microconcrete on the shear strength of model beams; (3) tests similar to those described in this report on two-dimensional structures such as pile caps and slabs, with a view to correlating the structural behaviour of models at different scales. Acknowledgements References The author wishes to thank Mr G. H. Beaven for assistance in carrying out the experimental work, Mr C. F. Cook for providing the instrumentation and Mr D. J. Evans for his help and useful comments on the testing of the microconcrete beams. ‘Thanks are also due to Dr W. B. Cranston and Dr J. L. Clarke for their guidance throughout the duration of this project. 1. canner ano concRETE ASsociATION, Model est: 6. TAYLOR, HF. .Invesigaion of the dowel shear Ing ~ Proceedings ofaanesday meeting, Landon, 1? March 1964. London, 1964. pp. 38 2. casas, .k-and ARMER G5. (Editors). Rein forced and presiresied microconcrete model. ‘Laneaster, Construction Press 1980. pp. 387 3. tire, w. A. and paPanont, m. Size effect in small-cale models of reinforced concrete ‘beams Jounal of the American Concrete nat ‘we, Proceedings Vol. 63, No. 11. November 1966, pp. 1191-1204 4, NEMILLE, A. 4 A general relation for strengths ‘of concrete specimens of different shapes and sizes. Joumal of the American Concrete Inst- tue. Proceedings Val. 63, No. 10. October 1966, pp. 10951110. 5. tavton, HF. Furer tests to determine shear sireses in reinforced concrete beams. London, = Cement and Concrete Association, February 1970. pp.27. Technical Report 438 (publics tion 427438) forces carved by the tensile stel in reinforeed ‘concrete beams. London, Cement and Concrete ‘Xwociation, November 1969. pp. 24, Techaleal ‘Report 431 (publication 42.431), ‘TAYLOR, HJ Investigation ofthe forces cared across craks in reinforced concrete beams in ‘shear by interlock ofeeeregate Londen, Cement and Concrete Association, November 1970. pp. 22. Technical Report 447 (publication 22.487), Fenwick, Rc. and PAULAY, T. Mechanisms of, shear resistance of eonerete beams. Proceedings ofthe American Society of Cin! Engineers Vo. 84, No. 10. October 1968, pp. 2325-2350, ‘TAYLOR, PJ. The fundamental behavior of reinforced concrete beams in bending and shear. Shear in reinforced concrete. Vol. 1 Detroit, American Concrete Institute, 1974 Publication SP-42. pp. 43-77 1 10. H 1. 13. 14 15, 16. ‘rarton, x. #2. Shear strength of large beams Proceedings of the American Society of Chil Engineers Vol. 98, No. S11, November 1972, p- 2473-2490. xan, 6.3. How safe are our large reinforced concrete beams? Journal ofthe American Con- crete Insite. Proceedings Vol. 64, No. 3 ‘March 1967. pp. 128-141 ‘tonsa, ¥. and WALTER, x. The Snatigart ‘shear test 1961. Translated by €.¥. ANERONGEN {rom the German in Beton- und Siahlbetonbau. ‘Vols. $6 and $7. London, Cementand Concrete Association, 1964, pp.'134. C&CA Library ‘Translation Cj, 111. swat, ws Uber den Einfuss der Balkenhohe ‘auf die Schubiragithigheit von cinfeldrigen Stahibetonbalken mit und ohne Schubbeweh= rung, (On the influence of beam height ofthe ‘hear sitength of single-span reinforced con- crete beams with and without web reinforce iment.) Thesis submited to the University of Stuttgart forthe depree of Dring. 1968, Durst STANDARDS INSTITUTION. The structural tue of concrete. Pat I: Detgn, materials and ‘workmanship. Londoa. pp. 184. P 110: Part 121972, JOHNSON, K.P. Strength tests on scaled down coneretes suitable for models, with a note on Imix design. Magazine of Concrete Research Vol. 18, No, 40: March 1962. pp. 47~83. snows, 6. and MIRZA, 5.4. Size effects in ‘model soncretes? Proceedings of te American Society of Engineers. Vol, 108, No, STS, June 1979. pp. 1007-1020. v7. 18, 20. 24 2, 2 24, VANS, D. 1 and CLARKE, JL. A comparison Denween the flcural behaviour of small micro~ concrete beams and that of prototype beams. ‘Wexham Springs, Cement and Concrete Asso- ciation, February 1981. pp. 32, Technical Report $42 (publication 4.542), ‘naw, Z.¥. and PEKOUSON, Pa. Accuracy of ‘models used in research in reinforced concrete Journal ofthe American Concrete Institute Pro- ‘ceedings Vol. 60, No. 11. November 1963. pp- 1683-1661, Swany, Nand QUEESMY sa. Strength, cracking fand deformation similtude in reinforced T= ‘beams under bending and shear. Joural of te American Concrete Insinde, Proceedings Vol 68, No. 3. March 1971. pp. 187-195 fcr... Factors infiencing “concrete Strength, Joumal of the American Concrete Inatite, Proceedings Vol. 87, No. 6. February 1951. pp. 417-432. BLACKAULN . SMITH, 0. Mand YOUNG, LEE Stress distribution affects ultimate tensile strength, Joumal of the American Concrete Insite, Proceedings Vol 8, No.6. December 1958. pp. 679-684. WWALKAVEN, J. ¢. The influence of depth on the shear strength of lightweight concrete beams ‘thous shear reinforcement. Delft, University Of Delft, 1978. pp. 36. Stevin Laboratory Report $-78-4, : Srl scale direct models of reinforced and pre- ‘essed concrete structures. Ithaca, N'Y. ‘Cornell University, September 1966. School of Givi! Engineering Report No, 326. aise, & Reinforced and prestresed micro- ‘concrete models. pp. 17-25 of reference 2

You might also like