Вы находитесь на странице: 1из 14

Magazine of Concrete Research, 2001, 53, No.

6, December, 377390

Evaluation of bond strength between old and new concrete in structural repairs
J. C. T. S. Cl maco and P. E. Regan{
University of Brasilia; University of Westminster

Good bond between old and new concrete is indispensable for many repairs. Following an examination of the factors involved and the methods of testing bond, the results of an experimental programme of slant shear tests are presented. The programme comprised 223 tests designed to study test methods and the effects of bond coats, surface preparation and ageing of the base concrete. The results are analysed in terms of a Coulomb criterion and evidence is produced that good bond can be achieved by casting the repair against mature concrete with no bonding aids, provided the base surface is dry and reasonably rough.

Notation
c fc f cc f cu f ct apparent cohesion compressive strength of composite prism (slant shear strength) concrete cylinder compressive strength concrete cube compressive strength concrete uniaxial tensile strength angle between a joint and the normal to the force applied to a composite prism coefficient of friction shear stress angle of friction

Introduction
Neither joining fresh concrete to old nor the search for a reliable method of testing bond is a new problem, 1 as can be seen from Perry's work. The slant shear test illustrated by Fig. 1 appears to be a generally satisfactory method in terms of being representative of realistic stress states at the joint and also in regard to simplicity, reproducibility and sensitivity. It has been adopted in many standards but test procedures are not agreed and
Department of Civil Engineering, University of Brasilia, Campus UnB, Brasilia, D.F. 70910-900 Brazil { School of Architecture and Engineering, University of Westminster, 35 Marylebone Road, London NW1 5LS (MCR 870) Paper received 5 June 2000; last revised 8 March 2001; accepted 16 July 2001

there are significant differences in specimen dimensions, joint angles and surface preparations. 2 The French standard NFP18-872 uses a prism (100 3 100 3 300 mm) sawn at an angle of 608 to the cross section and repaired by casting fresh concrete or bonding on another half-prism of hardened concrete. 3 The Italian standard adopts composite prisms measuring 70 3 70 3 200 mm with a joint angle of 738. 4 According to the ASTM Standard C882-83, the evaluation of bond strength of epoxy systems should be carried out on 76 3 152 mm cylinders of Portland cement mortar made in two halves with the epoxy coat applied to sand-blasted or wire-brushed surfaces at 608 to the cross section. 5 BS 6319: Part 4 uses a fractured surface obtained by splitting a 150 3 150 3 55 mm concrete plaque. The plaque is placed between a pair of steel plates cut into matching trapezia and the sandwich is mounted between rubber pads in a press. When compression is applied the dilation of the rubber splits the concrete along the line defined by the cuts in the plates. The half-plaques are then repaired and sawn into 150 3 55 3 55 mm test specimens with a complete joint at 608 to the cross section. The BS offers alternative tests using off-the-mould to top cast surfaces but the fractured joint is suggested as being the most relevant for repair situations. The variety of approaches to the evaluation of bond shows that the subject is still one for discussion. In this article the bond strength in concrete repairs is examined by means of an experimental programme adopting the slant shear method to study test procedures and the 377
0024-9831 # 2001 Thomas Telford Ltd

maco and Regan Cl


fc

Joint failure = arc tan Joint

Monolithic failure envelope

fct 0 fcmin fc

Fig. 1. Failure criterion for concrete composite prisms

influence of factors such as surface preparation, joint angles (608, 6338, 708 and 908 to the cross section), bond coats (eight bonding agents), specimen dimensions (three versions), and ageing of the base concrete (23 weeks old at repair: short-term repairs; 10 months 6 old at repair: long-term repairs). Results of this programme and an analysis in terms of a Coulomb failure criterion are presented.

Analysis of the slant shear method


Stress analysis When a composite concrete prism containing a joint at an acute angle to its longitudinal axis is subjected to axial compression, the stress state at failure depends primarily on the efficiency of the bond. If the bond is effective the failure is characterised by a rupture in the concrete, either of the `double pyramid' type common in tests of monolithic specimens or on a diagonal line near the joint. If the bond is insufficient a diagonal failure occurs along the joint line. In the case of effective bond the joint is submitted to an uneven distribution of shear, compression and tensile stresses. The situation is complex and highly influ7,8 enced by the thickness of any bond coat. In the case of a joint failure the global stresses at the interface are compression and shear and the Coulomb failure criterion can be used to describe the ultimate strength c (1) where is the shear stress at the joint; is the normal stress; c is an apparent cohesion; is a coefficient of friction. This expression is represented by the solid straight line in Fig. 1, with the slope defined by what is usually called the `angle of friction' of the joint: arc tan . The friction is a function of the roughness of the 378

parent concrete surface and experimental work indicates that it is influenced by bonding agents. The cohesion c, represented in Fig. 1 by the intercept of the straight line with the shear stress axis, depends on the concrete compressive and tensile strengths, and seems to be affected by environmental conditions, surface preparation, duration of loading and the use of bond 612 coats. From the equilibrium of forces at the composite prism of Fig. 1, the normal and shear stresses at a joint with an angle to the longitudinal or uniaxial strength, f c , can be written as f c cos2 f c sin cos (2) (3)

From equations (1), (2) and (3), the compressive strength of the composite prism governed by the weak plane is fc c 1 tan2 tan (4)

Assuming a constant c value for a given concrete and conditions, the most critical joint angle corresponding to a minimum f c from equation (4) is given by p (5) crit arc tan( 2 1) With the crit from equation (5), the minimum compressive strength of the composite prism from equation (4) will be p 1 ( 2 1 ) 2 p f cmin c (6) 2 1 Table 1 gives the values of crit and f cmin from equations (5) and (6), for values of c and presented by 10 Regan. Fig. 2 shows the variation of the composite a3 prism compressive strength, related to the factor f 2 cc , proportional to the tensile strength of concrete, as a function of , according to equation (4), for different values of the friction coefficient .
Magazine of Concrete Research, 2001, 53, No. 6

Bond strength between old and new concrete


Table 1. Critical joint angles and minimum compressive strength of prisms
Interface conditions Roughy Medium Smooth Cohesion c
a3 0X25 f 2 cc a3 0X25 f 2 cc X 0 5 Namm2

10

Friction 14 09 07

crit 7228 6638 6258

f cmin (Namm2 )
a3 1X56 f 2 cc a3 1X12 f 2 cc X 1 92 Namm2

Composite prism minimum compressive strength f (Namm2 ) cc y Implies that aggregate is exposed at the interface

35 30 25 Ratio, fc /fcc2/3 20 15 10 05 0 0 1 2 3 4 5 6 tan 7 8 9 10 11 12 = 14 = 10 = 07 Drawn for c = 025 fcc2/3 :N/mm2 55 60 65 70 75 80 : degrees

Fig. 2. Composite prism compressive strength 3 joint angle

and having a diameter f cmin. This envelope is represented by a solid line in Fig. 1. Test results of this research show that the assumption of concrete zero tensile strength is conservative for rough joint surfaces and a modification on the initial curved part of the failure envelope is to be proposed. If bond is effective at the joint, failure occurs by monolithic concrete crushing. The ultimate normal and shear stresses at the interface can be expressed by equations (2) and (3) in terms of the prism compressive 12 strength f c , and represented graphically by the partcircle shown in Fig. 1 as a chain-dotted line, defining the region where monolithic crushing failure should prevail. The part-circle diameter is the compressive strength of monolithic concrete prisms, f c , for which 10 an usual value, if end restraints are negligible, is f c 0X95 f cc 9 0X75 f cu (9)

13

By means of trigonometrical relationships and making tan , equations (4) and (6) can take the fol13 lowing forms, common in the plasticity theory fc c cos cos sin( ) 2 c cos 1 sin (7) (8)

f cmin

According to the Coulomb criterion, sliding failure at the joint takes place on a section subjected to combination of shear and normal stresses, described by equation (1), represented graphically by the straight line in Fig. 1. If a limited concrete tensile strength is assumed, the failure envelope in the tension zone becomes a part-circle tangent to the straight line, shown broken in Fig. 1, with its centre on the -axis and containing the 12 point with coordinates ( f ct , 0). Since the straight line has an inclination and intercept c, it can be proved that this line is tangential to the Mohr's circle with diameter f cmin, the minimum compressive strength given by equations (6) or (8) and shown dotted in Fig. 1. If the concrete tensile strength is neglected, which is usual in design, the modified Coulomb criterion for joint failures is simplified to a straight line of inclination and intercept c and an initial arc of circle tangential to both the straight line and the vertical axis
Magazine of Concrete Research, 2001, 53, No. 6

Surface preparation To assess repair techniques the interface should simulate the most usual texture, i.e. rough with aggregate exposed. The cracked surface specified by BS 6319: Part 4 is extremely rough and not representative of most repair substrates. Proposals of using sawn sur2,9 faces seem unsuitable as their exaggerated smoothness is not realistic for concrete repairs or even for most precast concrete. Moreover, it was consistently 14 demonstrated by Paulay, Park and Phillips and 15 Daschner that bond strengths of rough joints are undoubtely greater than those of smooth joints but the different methods to obtain rough surfaces do not pro16 duce very different results. Foure also showed at laboratory conditions that effective concrete repairs are obtained even by leaving the concrete surface in its natural condition after vibration. At the same time it does not seem to be difficult to define standard conditions for a rough surface. Standard classifications of surface treatments are given in reference 17 and the depth of treatment defined as `medium' for sandblasting (sufficient to generally expose coarse aggregate with slight revealmaximum reveal 6 mm) seems realistic for repair situations and not difficult to obtain with simple manual or mechanical scarification. The maximum aggregate size can be specified for any particular situation. 379

maco and Regan Cl Joint angle Equation (5) defines the critical joint angle if failure is governed by the Coulomb criterion. In spite of the great influence of this angle on the slant shear strength, several works neglect this parameter. The critical angle depends upon the friction, , and different authors give 10 different values. Regan proposes 1X4 for rough surfaces with exposed aggregates, making crit 72X28. 11 While Franke gives 0X7 for sandblasted surfaces. For this crit 62X58 but he adopted 608 in his own tests. 18 ACI 318M-95 stipulates 1X0 for concrete placed against hardened concrete with the surface roughened to a full amplitude of about 6 mm. In this case, crit 67X58. 19 Johnson's research showed crit to be about 708. He made tests at 508, 608 and 708 and found the joints to be stronger than the parent concrete for the first two angles while for 708 the two strengths were roughly equal. doubts as to the ability of the test method to evaluate 20 bond. Series 1, 2, 3 and 4: slant shear tests of large prisms Following the unsatisfactory results from the BS tests, alterations to the slant shear test method were introduced aiming for a better assessment of the factors involved. Specimen dimensions. To allow better simulation of real repair surfaces the prism size was increased to 1000 mm height and 150 3 150 mm test cross section. The geometry of the base concrete part was made as in Fig. 3(a) to avoid the surface preparation damaging an acute angle wedge. Joint angle. The angle between the joint and the normal to the load was changed to 708 (Series 1) to be closer to the theoretically critical angle. Additional tests were made for the angles 6338 (series 2) and 908 (series 3). Surface preparation. The surface of the base concrete was roughened with a pneumatic needle gun to expose coarse aggregate and provide a rough surface with a depth of treatment that could be defined as medium according to reference 17. This preparation seems more representative of repair situations than either splitting or sawing. Ageing of the base concrete. For test series 1, 2 and 3 the repair concrete was placed when the base

Test programme
Initial tests to the BS 6319: Part 4 carried out in this research showed the standard specification to be rather inappropriate, probably due to the extreme roughness of the cracked concrete surface and the adoption of an unsuitable joint angle. The high bond strengths obtained for cracked and dry-jointed specimens and even for concrete prisms with waxed interfaces raise serious

'Large' specimens Series 1 and 4 Series 3

150 150 294 305 150

150 'Small' specimens 6 mm steel plate 20

New concrete (repair)

Series 5 to 7

Joint 100 100 1000 1000 305

25

92

102 Repair

294

305

Rollers 102

200

200

Rollers 150 (a) 150 (b) (c)

Fig. 3. Shear tests of concrete composite prisms (measurements in mm)

380

Magazine of Concrete Research, 2001, 53, No. 6

25

70

Old concrete (base)

706

675

20

Base 70

Bond strength between old and new concrete material was about two weeks old. The age of the base at testing ranged from three to 26 weeks. To explore the effects of differential shrinkage, composite prisms with 708 joints were made with the same procedures, except that the base concrete was about eight months old when the repair material was added and the specimens were not tested until the repair concrete was two months old (series 4). The concrete mix design was as proposed in BS 5 6319 (10: 20: 16, cement: aggregate: sand with a w/c ratio of 04), to give a 28 day average cube strength of 65 Namm2 , except that the aggregate was a mixture of 10 mm and 20 mm natural gravel in the ratio 1:2. The base concrete half-specimens were cast in steel moulds with oiled wooden inserts to form the slant surfaces. After curing, scarification and cleaning with compressed air, the bases were replaced in the same moulds (minus inserts) with their joint faces upward to receive a specific bond treatment and the repair concrete. Rapid-hardening Portland cement was used, concrete was compacted on a vibrating table and the moulds were covered in plastic immediately after casting. Specimens were demoulded after 24 h and covered with wet hessian to cure for a minimum of three days. In all series, each set comprised three test specimens together with three control cubes of the base and repair concretes. In series 3, 6 mm-thick steel plates were attached, with epoxy resin and expanding bolts, to the 150 mm wide opposite faces of the specimens to prevent tension failures originating from the eccentric compression (Fig. 3(b)). In series 4 additional joint treatments were included and some sets were repeated to confirm the increase in the strength of the specimens without bond coat (abbreviation `R' for the repeating sets). The slant shear tests and compressive tests of control prisms were performed in a 10 000 kN testing machine and the control cubes of the base and repair concrete were tested on the same days as the composite prisms. The influence of platen restraint was studied and roller pads were used sandwiched between steel plates below 12,15 the specimen to reduce this restraint. Series 5, 6 and 7: slant shear tests of small prisms These series aimed to investigate the size influence and to define a specimen of small dimensions, quicker and easier to test, and able to duplicate the results obtained from real scale prototypes (Fig. 3(c)). Additionally, some extra bonding systems available on the market were tested. Test specimens were produced in 4 0 3 4 0 3 20 0 (102 3 102 3 508 mm) metal moulds, standard formwork for concrete flexure tests, available in most laboratories, with joint angles of 708 (series 5 and 7) and 608 (series 6). Procedures were similar to those for the large composite prisms. Half-specimens were cast in steel beam
Magazine of Concrete Research, 2001, 53, No. 6

moulds with oiled wooden inserts in their bottoms to form the slant surfaces and the repair concrete was cast against the previously treated surface. Series 7 was designed to study the effects of the ageing of the base concrete in small composite prisms. The procedures were as above except that the base concrete was about 8 months old when the repair was cast and the specimens were tested when the repair concrete was two months old. Four joint treatments were used in this series. Compressive tests were performed in a 2000 kN testing machine, with roller pads below the specimens, and the control cubes of the base and repair concretes were tested on the same days of the slant specimens. Joint surface treatments The test programme comprised 15 different joint treatments, the characteristics of which are described below. ( a) No bond coat, joint surface wetted and free water removed before casting the repair (sets 11, 41, 51 and 61). ( b) No bond coat, joint surface dry (sets 12, 21, 31, 42, 52, 62, and 71). ( c) Portland cement mortar bond coat of equal parts by weight of rapid-hardening cement and fine dry sand (passing no. 14 sieve), w/c ratio about 04 to give a creamy consistency (sets 13, 22, 32, 43, 53, 63 and 72). ( d ) One bond coat of epoxy resin (Sikadur 32 supplied by SIKA) a two component resin applied by brush to the substrate in a thin layer 1 to 2 mm (sets 14, 23, 33, 44, 54, 64 and 73). ( e) One bond coat of a prepacked SBR system (Resibond SBR and Resipatch Primer Powder, supplied by CeltiteSelfix) applied by brush to a saturated substrate after removal of free water. New concrete cast while the bond coat was still wet (sets 15, 45, 55 and 65). ( f ) Two bond coats of SBR, the first applied as for set 15 and the second applied after the first was dry (set 16). ( g) Two bond coats of epoxy resin, the first applied as for set 14 and the second applied after the first was dry (set 17). ( h) No bond coat, surface dry and no roller pads below the specimens at test (set 46). ( i) One epoxy mortar bond coat (Ep.m.A, proposed by 19 Johnson with proportions of 1:4 (resin: sand silica flour) and 1:23 (silica : sand passing sieve 14) applied by trowel in a layer about 3 mm thick, with no primer coat of plain resin (set 47). ( j) One pre-packed epoxy mortar bond coat (Sikadur 41 supplied by SIKA, a two component resin added to a pre-mixed powder, Ep.m.B) applied by trowel in a layer about 3 mm thick with no primer coat (sets 48 and 74). 381

maco and Regan Cl ( k) Joint surface left smooth as cast (sets 56 and 66). ( l ) One bond coat of an elastic 60% solid pure acrylic emulsion (VDM758, supplied by Harlow Chemicals Co Ltd) mixed with water at a ratio of 1:4 by volume (sets 57 and 67). ( m) One bond coat of a 55% plasticiser free dispersion of vinyl acetate/ethylene co-polymer (Vinnapas-Dispersion EP400, supplied by Wacker Chemicals Ltd) mixed with water at a ratio of 1:1 (sets 58 and 68). ( n) One bond coat of a pure acrylic ester dispersion (Mowiton M370, supplied by Harlow Chemicals Ltd) mixed with sand, cement and co-polymer in water (1:4 co-polymer: water) (sets 59 and 69). ( o) Monolithic control specimens (sets 18 and 510).

Results
Series 1, 2 and 3: large prismsshort-term repairs Table 2 presents the test results of series 1, 2 and 3, including cube strengths of base and repair concretes and failure modes. Fig. 4 compares the efficacy of different bond systems in terms of the ratio f cm a f cumin (average slant shear strength of composite prisms/lower base or repaircube strength) for series 1 and 2, as a proportion of the control specimens ratio f cm a f cu (set 18). From Table 2 and Fig. 4 it can be seen that. ( a) The modified slant shear test is sensitive to the

Table 2. Slant shear testsseries 1, 2 and 3: large prismsshort-term repairs


Series Set no. Joint system Test ages (days) f cu (Namm2 ) concrete Base Repair 7 68 65 62 57 58 48 466 643 677 638 683 699 632 605 565 660 579 732 654 679 626 170 175 199 261 258 273 329 322 297 370 388 365 271 288 272 158 193 186 353 352 349 348 342 324 325 323 307 336 320 310 313 329 73 68 69 81 61 135 127 141 Prism strength f c (Namm2 ) Average strength f cm 181 264 316 374 277 179 353 346 Failure mode Standard deviation (Namm2 ) 16 08 17 12 10 19 01 04

Base Repair 1 Joint angle 708 11 12 13 14 15 16 17 18 No b. coat wet joint No b. coat dry joint Portland cement mortar b. coat Epoxy resin b. coat SBR latex b. coat 2 SBR b. coats 2 epoxy resin b. coats Monolithic control prisms No b. coat dry joint Portland cement mortar b. coat Epoxy resin b. coat No b. coat dry joint Portland cement mortar b. coat Epoxy resin b. coat 21 82 80 78 74 78 80 50

(s) (s) (s) (s) (s) (s) (s) (s) (s) (c) (c) (c) (s) (s) (s) (s) (s) (s) (s/c) (s/c) (m) (m) (m) (m) (m) (s) (s) (m) (s) (s) (s) (s) (s) (s) (s) (s) (s) (c) (c) (c)

2 Joint angle 6338

21 22 23

50 47 129

35 33 115

606 673 687

627 652 685

324 321 317

01 15 10

3 Joint angle 908

31 32 33

136 201 198 194

122 187 184 179

637 703 710

656 656 678

71 71 134

03 10 07

Failure modes: (s) shear failure along the joint; (c) failure of concrete adjacent to the joint; (s/c) shear failure along the joint with simultaneous concrete failure adjacent to the joint; (m) monolithic failure with concrete crushing in the upper part of the specimen (section 150 3 150 mm).

382

Magazine of Concrete Research, 2001, 53, No. 6

Bond strength between old and new concrete


120 Series 1 (joint angle 70) 103% 100 Ratio, fcm /fcumin: % of control 93% 95% 86% 80 72% 68% 60 21 22 23 46% 40 Set 11 20 12 13 14 15 16 17 18 81% 74% Series 2 (joint angle 633) 99% 100%

No bond coatwet joint

No bond coatdry joint

Portland cement mortar bond coat

Epoxy resin bond coat

SBR latex bond coat

2 SBR latex bond coats

2 epoxy resin bond coats

Monolithic control prisms

Joint systems

Fig. 4. Slant shear ratioslarge prismsshort-term repairs

( b)

( c)

(d )

( e)

different bond treatments and consistent in terms of the low standard deviations of the strengths in each set. The strengths of joints made without bond coats increased as the joint angle changed from 708 to 6338 (72% of set 12 compared to 93% of set 21). The predominantly monolithic failures for the 6338 joint also suggests the 708 angle to be critical for rough joints without bond coat. The best adhesion was in set 14 with one bond coat of epoxy resin giving a ratio of 103% with no bond failures. The results for set 17 with two bond coats of epoxy were also good, reaching a 99% ratio, although showing partial bond failures. The 95% strength of set 13, specimens with PC mortar bond coats, should also be noted although all the specimens failed at the joints. The use of one bond coat of SBR mortar did not produce a significant improvement over results for specimens without bond coats. Set 16 with two bond coats of SBR mortar gave the lowest bond strength (46%), in agreement with previous 21 work. It should be mentioned that the manufacturer's instructions advise that if, for any reason, the slurry coat is allowed to dry it must be removed and a new application made. The reductions of bond strengths as the joint angle changed from 708 to 6338 are noticeable for the PC mortar (reduction of 9% on the slant shear ratio) and particularly for the epoxy resin bond coats (reduction of 22%). This suggests that these joints under higher normal stresses develop differential interface strains, which break down the bond sooner than when no bond coat is used. These

results agree with Eyre and Domone's conclusion that bonding systems can show different performances as the joint angle changes. ( f ) The compressive strength of the monolithic prisms is low ( f cm 0X57 f cumin of set 18) compared to the relationship f c 0X75 f cu , commonly accepted when end restraints are negligible. The only apparent reason for this was the geometry of the test specimens, with the lower part width 50 mm larger than the upper one. This fact can induce uneven stress distributions on the lower prism reducing compressive strengths. ( g) In the direct shear tests (series 3) the strengths of prisms without bond coat and with PC mortar bond coat are virtually the same. Epoxy bond coats provided a major improvement of about 90% upon the other two systems with concrete failures adjacent to the joint instead of pure bond failures. Series 5 and 6: small prismsshort-term repairs Tests results are shown in Table 3. Fig. 5 compares the different bond systems for the 708 and 608 joint angles, in terms of the ratios f cm a f cumin as a proportion of f cm a f cu for the solid prisms. In sets 51, 54 and 64 the value of f cumin was replaced by the average for the two concretes, as the low strength of the repair concrete would artificially produce high values for the slant shear ratios. Results from Table 3 and Fig. 5 show that. ( a) Strengths and failure modes of large and small composite prisms compare favourably, indicating that the results from the smaller version may well be as realistic as those produced with the larger 383

Magazine of Concrete Research, 2001, 53, No. 6

maco and Regan Cl

Table 3. Slant shear testsseries 5 and 6: small prismsshort term repairs


Series Set no. Joint system Test ages (days) f cu (Namm2 ) concrete Base Repair 575 720 536 644 603 684 520 741 548 602 480 710 619 481 540 687 823 708 605 239 274 201 293 308 250 423 306 321 387 344 360 229 258 218 11 22 44 111 121 103 176 189 214 210 148 134 490 470 410 246 244 354 342 293 265 460 330 380 399 422 400 26.7 210 266 103 171 145 96 105 94 222 189 192 192 186 152 Prism strength f c (Namm2 ) Average strength f cm 238 284 350 364 235 26 112 193 164 457 Failure mode Standard deviation (Namm2 ) 30 24 52 18 17 14 07 16 33 41

Base Repair 5 Joint angle 708 51 52 53 54 55 56 57 58 59 510 No b. coat wet joint No b. coat dry joint Portland cement mortar b. coat Epoxy resin b. coat SBR latex b. coat Smooth as cast joint no b. coat Acrylic emulsion b. coat Vynil acetate/ ethylene b. coat Acrylic ester b. coat Monolithic control prisms No b. coat wet joint No b. coat dry joint Portland cement mortar b. coat Epoxy resin b. coat SBR latex b. coat Smooth as cast joint no b. coat Acrylic emulsion b. coat Vynil acetate/ ethylene b. coat Acrylic ester b. coat 81 77 75 85 84 80 76 75 74 42 47 43 46 42 42 43 42 40 41

(s) (s) (s) (s) (s) (s) (s) (s) (s) (s/c) (s/c) (s/c) (s) (s) (s) (s) (s) (s) (s) (s) (s) (s) (s) (s) (s) (s) (s) (m) (m) (m) (s) (s) (s) (s) (s) (s) (s) (s) (s) (c) (c) (c) (s) (s) (s) (s) (s) (s) (s) (s) (s) (s) (s) (s) (s) (s) (s)

6 Joint angle 608

61 62 63 64 65 66 67 68 69

80 76 75 85 84 81 79 76 75

37 39 39 42 42 38 42 40 41

606 741 548 644 605 606 536 520 741

628 613 613 481 540 628 823 708 605

281 300 390 407 248 140 98 201 177

59 32 54 11 27 28 05 15 17

See notes to Table 2

384

Magazine of Concrete Research, 2001, 53, No. 6

Bond strength between old and new concrete


120 Series 5 (joint angle 70) 100 Ratio, fcm /fcumin: % of control 86% 80 61% 65% 53% 51% 36% 30% 28% 57 24% 67 58 68 59 69 39% 39% 510 Monolithic control prisms 57% 60% 94% 85% 95% Series 6 (joint angle 60) 100%

60

59%

40 Set 51 61 52 62 53 63 54 64 55 65

20 56 5% 0 No bond coatwet joint No bond Portland Epoxy SBR coatdry cement mortar resin latex joint bond coat bond coat bond coat Smooth as cast joint Acrylic Vynil acetate Acrylic emulsion ethylene ester bond coat bond coat bond coat Joint systems 66

Fig. 5. Slant shear ratiossmall prismsshort-term repairs

( b)

( c)

(d )

( e)

(f)

( g)

specimens. The variability of results from the small prisms (maximum standard deviation 5X9 Namm2 ) is higher than that obtained with large prisms (maximum standard deviation 1X9 Namm2 ). Epoxy resin and PC mortar exhibited the best adhesion capabilities out of all the bond coats examined, with very similar slant shear ratios although all the prisms with PC mortar failed at their joints. Results of sets 7 to 9 (pure acrylic emulsion, vynil acetate/ethylene copolymer and pure acrylic ester dispersion bond coats) typify a weaker bond group and the use of SBR bond coat did not produce any significant improvement over the results without bond coats. Test results of smooth joints highlight the need for adequate surface preparation. Removing the cement latence from the joint to a depth of around 6 mm increased the strength ratios from set 56 to 52 (708 joints) by 48% and from set 66 to 62 (608 joints) by 35%. The strengths of the 708 jointed prisms are consistently lower than the 608 prisms, the only exception being for the pure acrylic VDM758 set of very poor bond strengths. The f cm a f cu ratio was 057 for the larger solid prisms and 075 for the small prisms. This difference significantly decreased the slant shear ratios of the small prisms which suggests that a better comparison of results between small and large prisms should be given by their absolute strengths. The reduction of 19% in the average strengths shows the effect of wetting the parent concrete prior to casting the new material to be detrimental to bond strengths as in series 1. This fact can not

be noticed in Fig. 5 because of the low strength of the repair concrete of set 51. Series 4 and 7: long-term repairs Results are summarised in Table 4. Fig. 6 presents a comparison between the short-term and long-term data for large prisms with the same angles and surface treatments. It can be seen that: ( a) In absolute terms the strengths of joints made without bond coats increased with age (average ratio of 72% in series 1 to 995% in series 4). Monolithic failures occurred in all the specimens of set 49. This high strength reduces the range of possible improvements due to the use of a bond coat. ( b) Test results confirm the detrimental effect of wetting the parent concrete prior to casting the new material. ( c) The tests without rollers below the specimens (set 46) shows a 10% increase in slant shear strength as compared with those of set 42. Previous works which used such rollers to encourage joint failures 15 include the tests by Daschner and those by Clark 12 and Gill. ( d ) The low strengths of set 47 (79%) can be attributed to the dryness of the epoxy mortar Ep.m.A. The performance of the prepacked epoxy mortar Ep.m.B (91%set 68) was inferior to that of specimens without any bond coat (dry joint). Problems with the workability of these epoxy mortar formulations can mar their use as bond coats without the application of a primer coat of pure resin. ( e) The average increase of 52% in the ratios f cm a f cumin from the short-term (series 5) to long-term repairs (series 7) ratifies in the small 385

Magazine of Concrete Research, 2001, 53, No. 6

maco and Regan Cl


Table 4. Slant shear testsseries 4 and 7: long-term repairs
Series Set no. Joint system Test ages (days) f cu (Namm2 ) concrete Base Repair 589 622 567 566 562 581 559 851 730 717 615 63.6 610 371 734 702 634 643 585 627 650 696 255 260 299 332 337 312 260 243 259 323 344 330 261 331 348 352 327 369 258 222 272 334 273 312 363 380 385 374 317 355 414 384 391 386 342 413 408 418 393 374 331 339 325 325 306 Prism strength f c (Namm2 ) Average strength f cm 271 327 254 332 313 349 251 306 376 349 396 Failure mode Standard deviation (Namm2 ) 24 13 10 11 46 21 26 25 09 24 13

Base Repair 4 Large prisms Joint angle 708 41 42 43 44 45 46 47 48 49 410 411 No b. coat wet joint No b. coat dry joint Portland cement mortar b. coat Epoxy resin b. coat SBR latex b. coat No b. coatdry joint no rollers below prisms at tests Epoxy mortar b. coatA Epoxy mortar b. coatB No b. coat dry joint (Ry ) Portland cement mortar b. coat(Ry ) Epoxy resin b. coat(Ry ) No b. coat dry joint Portland cement mortar b. coat Epoxy resin b. coat Epoxy mortar b. coatB 314 334 321 314 320 323 313 313 321 315 317 57 64 61 59 61 63 54 55 85 76 82

(s) (s) (s) (s) (s) (s) (s) (m) (m) (c) (s/c) (s/c) (s/c) (s) (s) (s) (s) (s) (s) (s) (s) (s) (s) (s) (m) (m) (m) (m) (m) (m) (c) (m) (m) (s) (s) (s) (m) (s) (s) (s/c) (s/c) (s/c) (c) (c) (c)

7 Small prisms Joint angle 708

71 72 73 74

321 315 317 313

85 76 82 55

730 717 615 851

627 650 696 585

381 406 348 319

29 10 19 09

See notes to Table 2 y Repeating sets

prisms tests the finding that the strengths of joints made without bond coat increase significantly with age.

Discussion
Ultimate stresses at the interfaces of composite prisms Test results for short-term repairs are presented in Fig. 7 in terms of the ultimate stresses at the interface, and , calculated by equations (1) and (2). Stresses 386

a3 are related to the factor f 2 cc , f cc being the minimum value in each set calculated from the cube strengths by f cc 0X8 f cu . In the previous sections, systems were compared in terms of the `slant shear ratios', defined as the ratio f cm a f cumin (average strength of composite prisms/lowerbase or repaircube strength) related to the control specimens ratio f cm a f cu . Bond systems with slant shear ratios lower than 50% were excluded in Fig. 7, as they typify `slip joints'. In Fig. 7(a), the data for the large prisms of series 1 and 2, including the results affected by the detrimental

Magazine of Concrete Research, 2001, 53, No. 6

Bond strength between old and new concrete


120 Series 1: Short-term repairs 100 Ratio, fcm /fcumin: % of control 99% Series 4: Long-term 103% 95% 94% 108% 97% 100% Joint angles: 70

80 68% 60

80% 72% 74%

40 Set 11 20 42 12 42 13 43 14 44 15 45 18

No bond coatwet joint

No bond coatdry joint

Portland cement mortar bond coat Joint systems

Epoxy resin bond coat

SBR latex bond coat

Monolithic control prisms

Fig. 6. Large prismsshort- and long-term repairs ( joint angle 708)

3 Series 2 - joint 70 Series 3 - 633 Series 4 - direct shear Control 2 /fcc2/3 fcc = 50 N/mm2 fcc = 49 N/mm2 Series 6 - joint 70 Series 7 - joint 60 Control

1 arctan = 14 0 0 1 2 /fcc2/3 (a) 3 0 1 2 /fcc2/3 (b) 3 4 arctan = 14

Fig. 7. Ultimate stresses at jointshort-term repairs: (a) series 1, 2 and 3large prisms; (b) series 5 and 6small prisms

effect of pre-wetting the joint without bond coat, show good agreement with the straight line of the joint failure envelope, that can be expressed by
a3 2 0X25 f 2 cc (Namm )

(10)

This straight line assumed the values of c and 10 presented by Regan for rough interface conditions, presented in Table 1. As can be seen in Fig. 7, Regan's proposal 1X4 for the line inclination showed a good 11 agreement with experimental results. Franke's value, 0X7, for sandblasted surfaces and the ACI 318M18 95, 1X0, for concrete placed against hardened
Magazine of Concrete Research, 2001, 53, No. 6

concrete with the surface roughened to a full amplitude of about 6 mm, are rather conservative. The value of the line intercept in equation (10), a3 2 c 0X25 f 2 cc (Namm ), was compared with results from the direct shear tests (series 3points on the vertical axis) and showed to need correction. Only the epoxy bonded prisms presented ultimate shear stresses in accordance with that value and results of six prisms with no bond coat or PC mortar coat are virtually identical and consistently below that value. Results from these tests will be used again in this section to propose a different line intercept. In Fig. 7(b) it can be seen that the data for the 708 387

maco and Regan Cl small prisms also satisfy the joint failure line. However, some results of 608 prisms are slightly below the straight line although their strengths were higher than the 708 prisms, suggesting that analysis only in terms of absolute strengths may be insufficient, agreeing with 9 Eyre and Domone's conclusion. A comparison of the control prism data, points on the horizontal axes in Fig. 7, again shows the significantly lower strengths of the larger solid prisms. Figure 8 shows the test data from short- and longterm repair series, comprising tests of large and small composite prisms, for three relevant bond systems of this research: `no bond coat' with the parent concrete surface kept dry before pouring the new concrete; PC mortar; and epoxy resin bond coats. The initial curved part of the failure envelope described in Fig. 1, based on the concrete having zero tensile strength, was reported to be adequate for the 12 prediction of the strength of smooth concrete joints. However, for roughened joint surfaces it appears to be conservative as indicated by the average ultimate strengths of the direct shear tests (Table 2series 3). In Fig. 8 the initial arc is replaced by a straight line between the point where the arc was tangential to the joint failure envelope line, corresponding approximately to the vertical coordinate 04, and the point on a3 2 the vertical axis given by 0X10 f 2 cc (Namm ). This proposal is assumed to be safe for any system producing a bond strength equal or superior to that of the new concrete cast directly against the old without a 10 bond coat. Regan proposed an identical expression for the cohesion at crack surfaces (width 1 mm) in monolithic gravel concrete, which is probably more unfavourable. The analysis of Fig. 8 shows the data for the three bond systems in good agreement with the failure envelope expressed by
a3 2 X X 2a3 0X25 f 2 cc 1 4 for > 0 1 f cc (Namm ) (11) a3 2 X X 2a3 0X10 f 2 cc 3 0 for 0 < < 0 1 f cc (Namm ) (12)

Daschner carried out an extensive experimental programme concerning the influence of the parent concrete surface preparation, concrete strength and joint angle on the compressive strength of composite prisms. Tests were made with precast saw-tooth joint surfaces, no bond coats and 150 3 150 3 600 mm prisms. A wide range of angles were used, from 5858 to 908 between the joint and the normal to the load. Zelger 22 and Ru sch tested 80 3 250 3 700 mm prisms also with saw-tooth joints and no bond coat, with angles 5948, 4898, 3698, 2428, 1318 and 08. Fig. 9(a) presents these test results, the ultimate stresses and a3 being related to the factor f 2 cc . The data show good agreement with the proposed failure criteria. 11 Franke collated a reasonable amount of data of slant shear and direct shear tests on epoxy jointed concrete specimens with the surfaces roughened by sandblasting and with variable joint angles. Results from Franke and other authors' data are displayed in Fig. 9(b) (symbols are mean values) and a satisfactory agreement with the criteria is achieved. The analysis also suggests that a better correlation for epoxy joints might be achieved by using a higher value of c and a lower value of , as represented by the broken line corresponding to cohesion and friction values a3 X c 0X50 f 2 cc and 1 1. Although a similar value was found for the friction in this research, for the cohesion it was about half this value, as can be seen in

15

20 No bond coat 16 PC mortar bond coat Epoxy bond coat

12 /fcc2/3 08

arctan = 14 04

arctan = 14

arctan = 14

01 0 05 /fcc2/3 0 05 /fcc2/3 0 05 /fcc2/3 10

Fig. 8. Proposed failure criterion for composite prismsshort- and long-term repairs

388

Magazine of Concrete Research, 2001, 53, No. 6

Bond strength between old and new concrete


3 Daschner: fccmin = 14 N/mm2 fccmin = 25 N/mm2 fccmin = 32 N/mm2 fccmin = 25 N/mm2 Diaz: Buchting & Moosbrugger: Eibl et al.: Rehm & Franke: Ladner & Flueler: fcc = 57 N/mm2 fcc = 46 N/mm2 fcc = 3254 N/mm2 fcc = 3980 N/mm2 fcc = 1826 N/mm2 fcc = 60 N/mm2

Zelger & Rusch: 2 /fcc2/3

1 arctan = 14 04 fcc = 14 N/mm2 01 0 1 2 /fcc2/3 (a)

fcc = 32 N/mm2 arctan = 14 fcc = 30 N/mm2

2 /fcc2/3 (b)

Fig. 9. Proposed failure criterioncomparison with other authors: (a) no bond coat jointsdata collated by Daschner; 11 epoxy bond coat jointsdata collated by Franke

15

(b)

Fig. 8, suggesting that the adoption of a correlation for steep joints ( 9 908) is still needed.

Conclusions
An experimental programme of slant and direct shear tests was developed to evaluate the strength of bond between old and new concrete with different bonding systems and surface preparations. The tests provided evidence that a sound bond can be achieved by casting the repair against mature concrete without the use of bonding aids, provided the base surface is dry and roughened to an extent that the aggregate is exposed and no damage is caused to the concrete near the joint. Casting new concrete directly against the old without a bond coat produced very high slant strengths in the long-term tests, with very little or no improvement left to be achieved by bonding agents. The very dry surface of the aged concrete probably improves the adhesion at the interface by migration and penetration of hydrated cement crystals into the scabbled surface of the old concrete probably in combination with a decrease in the w/c ratio of the concrete near the joint. Efficient curing is certainly essential to guarantee a good hydration of the new concrete and to minimise early shrinkage. For rough surfaces without any bond coat the angle of 708 between the joint line and the normal to the load was shown to be more critical than the 608 recom4,5 mended by American and British standards. The influence of bond coats was found to be greater for steeper joints and repairs cast against relatively new base concrete. In the slant shear tests with 708 joints, epoxy resin coats substantially increased bond strengths. In the direct shear tests the epoxy coats even
Magazine of Concrete Research, 2001, 53, No. 6

doubled the shear strengths compared with joints without bond coat. However, an opposite tendency was manifested for prisms with 6338 angles: joints made with epoxy bond coats exhibited a considerable reduction in the slant shear strengths. The use of a Portland cement mortar bond coat improved adhesion by about 30% in comparison with specimens with no bond coat in the short-term tests, although it did not prevent bond failures. As for epoxy joints, PC mortar bond coats exhibited poorer performance when the angle between the joint and the cross section decreased in relation to the modified angle of 708. No improvement was achieved in the direct shear tests of PC mortar joints in relation to the no bond coat joints. Some materials marketed as having `bonding properties' showed in the present work detrimental effects on bond strengths (formulations of pure acrylic emulsion, vinyl-acetate/ethylene co-polymer and pure acrylic ester dispersion). SBR latex bond coats did not improve bond strengths. The small specimens developed and tested in this investigation generally reproduced the main results from the large prism tests. The preparation required is minimal and this type of specimen can be useful to evaluate bond strength performance in quicker and easier tests. A modified Coulomb criterion was shown to provide a satisfactory prediction for the ultimate stresses at the interface for any joint system producing a bond strength equal or superior to that of the new concrete cast directly against the old without a bond coat. The failure criteria adopted in this analysis exhibited good agreement with experimental data from this and other work. However, it should be remarked that test results indicated that the constants in the equation may be 389

maco and Regan Cl affected by the bond coat material. For epoxy joints, for example, lower values of should be considered possibly in combinations with higher values of c.
12. CLARK L. A. and GILL B. S. Shear strength of smooth unreinforced construction joints. Magazine of Concrete Research, 1985, 37, No. 131, 95100. 13. JENSEN B. C. Lines of discontinuity for displacements in the theory of plasticity of plain and reinforced concrete. Magazine of Concrete Research, 1975, 27, No. 92, 143150. 14. PAULAY T., PARK R. and PHILLIPS M. H. Horizontal construction joints in cast-in-place reinforced concrete. ACI Special Publication SP42-27, Detroit, 1974, pp. 601616. 15. DASCHNER F. Notwendige schubbewehrung zwischen betonfertigteilen und ortbeton (Necessary shear reinforcement between precast and insitu concrete). Lehrstuhl fu r Massivbau, Technische Universita t Mu nchen, Mu nchen, 1976, p. 129. B. Comportement des surfaces de reprise de betonnage 16. FOURE vis-a-vis du cisaillement (Behaviour in shear of construction joints in concrete). Annales de l'Institut Technique du Batiment et des Travaux Publics, 1988, 462, 109135. 17. AMERICAN CONCRETE INSTITUTE. Guide to cast-in-place architectural concrete practice, Chapter 9 Treated architectural surfaces. ACI Committee 303. R-74. ACI Manual of Concrete Practice, Part 3, 1982, pp. 2225. 18. AMERICAN CONCRETE INSTITUTE. Building code requirements for structural concrete (ACI 318M-95) and Commentary (ACI 318RM-95), Detroit, 1995, pp. 371. 19. JOHNSON R. P. The structural properties of an epoxy mortar and its use for structural joints. The Structural Engineer, 1970, 44B, No. 6, 227233. MACO J. C. T. S. and REGAN P. E. Evaluation of bond strength 20. CLI between old and new concrete. Proceedings of the Fourth International Conference on Structural Faults & Repair. London, Vol. 1, 1989, pp. 115122. 21. DIXON J. F. and SUNLEY V. K. Use of bond coats in concrete repair. Concrete, 1983, 17, No. 8, pp. 3435 [Discussions: 17, No. 1012; 18, No's 3 and 5]. SCH H. Der einfluss von fugen auf die festig22. ZELGER C. and RU keit von fertigteilschalen (The influence of joints on the strengths of precast walls). Beton-und Stahlbetonbau, 1961, 10, 234237.

References
1. PERRY R. B. Tests on bonding new concrete to old. Engineering News, 1908, 60, No. 7, 167168. 2. PAILLERE A. M. and RIZOULIERES J. Criteria for evaluation of materials for concrete repair. Proceedings of an International Conference on Gestion des Ouvrages d'art, 1981, Vol. II. Editions Anciens ENPC, Paris, pp. 479484. 3. CAMOMILLA G. General report. Proceedings of an International Conference on Gestion des Ouvrages d'art, 1981, Vol. II. Editions Anciens ENPC, Paris, pp. 105136. 4. AMERICAN SOCIETY FOR TESTING AND MATERIALS. Bond strength of epoxy-resin systems used with concrete. Standard Test Method C-882. 1983. 5. BRITISH STANDARDS INSTITUTION. Testing of resin compositions for use in construction. Method for measurement of bond strength (slant shear method), BS 6319: Part 4, 1984, pp. 8. MACO J. C. T. S. Repair of structural concrete involving the 6. CLI addition of new concrete. PhD thesis, Polytechnic of Central London, 1990, pp. 234. 7. HRANILOVIC M. Failure criteria for structural joints. Proceedings of the RILEM International Symposium on Adhesion between Polymers and Concrete. Chapman and Hall, London, 1986, pp. 650660. 8. WALL J. S. and SHRIVE N. G. Factors affecting bond between new and old concrete. ACI Materials Journal, 1988, 85, No. 2, 117125. 9. EYRE J. R. and DOMONE P. L. J. The slant shear testing of bond repair materials for concrete structures. Proceedings of the Second International Conference on Structural Faults & Repair, Engineering Technics Press, Edinburgh, 1985, pp. 141148. 10. REGAN P. E. Consistent design and detailing of concrete members and connections Part 2: Struts, ties and nodes. Structures Research Group. Polytechnic of Central London, 1986, pp. 25. 11. FRANKE L. The dimensioning of adhesive-bonded joints in concrete building components. Proceedings of the RILEM International Symposium on Adhesion between Polymers and Concrete, Chapman and Hall, London, 1986, pp. 461473.

Discussion contributions on this paper should reach the editor by 1 May 2002.

390

Magazine of Concrete Research, 2001, 53, No. 6

Вам также может понравиться