Engineering Structures 28 (2006) 622633 www.elsevier.

com/locate/engstruct

Flexural behaviour of externally prestressed beams. Part II: Experimental investigation

Chee Khoon Ng a, , Kiang Hwee Tan b
a Faculty of Engineering, Universiti Malaysia Sarawak, 94300 Kota Samarahan, Sarawak, Malaysia b Department of Civil Engineering, National University of Singapore, Blk E1A, #07-03, 1 Engineering Drive 2, Singapore 117576, Singapore

Received 13 January 2005; received in revised form 1 September 2005; accepted 20 September 2005 Available online 26 October 2005

Abstract In Part I of this paper, a simple pseudo-section analysis method which accounts for second-order effects in a simply supported, externally prestressed beam subjected to two symmetrically applied concentrated loads was developed. In this paper, an experimental investigation of the exural behaviour is reported. A total of nine simply supported prototype beams were tested to evaluate the effect of span-to-depth ratio and second-order effects. It was found that span-to-depth ratio has no signicant effect on the exural behaviour of the beams. For beams with spanto-depth ratio of up to 22.5, a single deviator provided at midspan section is effective in minimising second-order effects, that is, maintaining higher load-carrying capacity and ensuring ductility at the ultimate limit state for the beams. However, second-order effects prevailed in a longer beam with larger span-to-depth ratio of 30.0 despite the provision of a single deviator at midspan. This type of long beams would require at least two deviators placed at one-third span sections, hence reducing the deviator spacing in order to minimise second-order effects so that the beams would achieve the desired exural performance with regard to beam strength and ductility. Theoretical predictions of the loaddeformation responses using the proposed analytical model were found to agree well with the test results in this study and experimental data of other investigations. c 2005 Elsevier Ltd. All rights reserved.
Keywords: Analysis; Beam; External prestressing; Flexure; Strength; Second-order effects

1. Introduction Experimental studies on exural behaviour of externally prestressed beams are scarce based on a literature survey carried out by the authors in Part I of this paper. Most of the theoretical investigations reported that the span-to-depth ratio of the beams has a signicant effect on the exural behaviour, particularly on external tendon stress [13]. In a separate investigation [4], span-to-depth ratio was found to be insignicant in the external tendon stress of the beams if second-order effects are minimised. Therefore, more experimental data is required to address this issue. Another aspect of the exural behaviour which is lacking in experimental data is the variation in external tendon depth, or normally termed as second-order effects. The investigation reported herein aims at providing more experimental data on the behaviour of externally prestressed
Corresponding author. Tel.: +60 82 670525; fax: +60 82 672317.

beams, particularly on the effects of span-to-depth ratio and number of deviators or second-order effects. The test results were also used for the validation of the analytical model proposed in Part I of this paper. 2. Test program Nine prototype concrete T -beams with cross-sectional dimensions and reinforcement details shown schematically in Fig. 1, were prepared (see also Table 1). In all beams, the internal longitudinal reinforcement consisted of two T16 bottom bars and four R8 top bars, with average laboratory tested yield strengths, f y and f y , of 530 MPa and 338 MPa respectively. The beams were reinforced with R6 or R8 mild steel stirrups throughout their lengths, with average laboratory tested yield strengths of 320 MPa and 338 MPa respectively. The design cube strength of the concrete used was 38 MPa at 28 days. No deviator was provided in beams T-0A and T-0B. In beams ST-1, ST-2, ST-3, ST-4 and ST-5, a 100 mm wide

E-mail address: ckng@feng.unimas.my (C.K. Ng). 0141-0296/$ - see front matter c 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.engstruct.2005.09.016

C.K. Ng, K.H. Tan / Engineering Structures 28 (2006) 622633

623

Notation d ps d ps 0 fc f cu f pe fy fy L Mcr Ms Mu My Pcr Ps Pu Py f ps tendon depth initial tendon depth concrete cylinder compressive strength concrete cube compressive strength effective prestress of prestressing or external tendons yield strength of internal tension reinforcement yield strength of internal compression reinforcement effective beam span cracking moment moment at serviceability limit state ultimate moment of resistance moment corresponding to yield load cracking load load at serviceability limit state ultimate load yield load stress increase in external tendons

top and bottom faces of the beam being tested. These clamping devices were installed to measure the midspan curvature if there is no deviator at midspan, whereas for beams with a midspan deviator, these devices were installed to measure the curvature at a section 300 mm away from the midspan, which is still within the constant moment region. Strains were measured using electrical resistance strain gauges and forces in the external tendons were measured by load cells installed at the passive anchorages. 3. Test results and discussion on effect of span-to-depth ratio From the literature review as described in Part I of this paper, it is noted that all the available equations for the calculation of stress increase, f ps , or stress, f ps , in the external tendons are functions of the span-to-depth ratio. Therefore, this part of the study was carried out to provide further validation. The ve concrete T -beams designated ST-1, ST-2, ST-3, ST-4 and ST-5, with the same cross section, internal steel reinforcement, tendon conguration and span-to-depth ratios of 7.5, 9.0, 15.0, 22.5 and 30.0 respectively as shown in Fig. 1 and described in Table 1, were tested and the results compared. With third-point loading, these span-to-depth ratios correspond to shear spanto-depth ratios of 2.5, 3.0, 5.0, 7.5 and 10.0 respectively, well within the range of slender to very slender beam regimes in which beam action or exural behaviour prevails [5]. 3.1. Loaddeformation response Fig. 2(a) and (b) show the momentcurvature curves and the loadmidspan deection curves of beams ST-1, ST-2, ST3, ST-4 and ST-5 respectively. Prior to cracking at Mcr or Pcr , all the beams assumed linear momentcurvature and loaddeection relations. The beam stiffness was reduced after the appearance of exural cracks, but the beams resumed a linear loaddeection behaviour after the crack development had stabilised. The approximate linear cracked behaviour ended when the internal steel reinforcement began to yield (at M y or Py ). Beyond this point, the curvature or deection increased dramatically without any signicant increase in external load up to the maximum moment or load carrying capacity, Mu or Pu , of the beams. All these test beams failed in exure. The larger ultimate curvatures of beams ST-1 and ST-2 shown in Fig. 2(a) were probably due to multiple large cracks passing through the curvature measuring device, resulting in larger measured deformation. The smaller ultimate curvatures in beams ST-4 and ST-5 were probably due to the beams failing at sections outside the measured regions, but still within the constant moment region. Fig. 2(b) shows the loadmidspan deection curves of the test beams. The four beams with span-to-depth ratio of up to 22.5, namely ST-1, ST-2, ST-3 and ST-4, exhibited approximately the same deection ductility ratio, which were 2.15, 2.23, 2.70 and 1.96 respectively. Deection ductility ratio is dened as the ratio of deection at the ultimate limit state

deviator at midspan was provided in each beam. In beam ST5A, two deviators were provided, one each at the one-third span sections; and beam ST-5B had three deviators, each at the quarter span sections. The initial effective depth of the external tendons, d ps 0, for all the beams was kept to 200 mm, with span that led to span-to-depth ratios, L /d ps 0, of 7.5, 9.0, 15, 22.5 and 30.0 for beams ST-1, ST-2, ST-3, ST-4 and ST-5 respectively. Beam T-0A had the same span of 4.5 m as beam ST-4 and beams T-0B, ST-5A and ST-5B had the same span of 6.0 m as beam ST-5. The beams were prestressed using external tendons one day before testing. Two straight seven-wire prestressing strands, with a diameter of 12.9 mm and an average tensile strength, f pu , of 1900 MPa, were stressed to an effective prestress, f pe , of about 0.4 f pu or 760 MPa, one on each side of the beam. Technically, beams with a deviator provided at midspan would benet from a draped tendon prole for the vertical component of the prestressing force would take effect in the load carrying capacity. However, the draping angle would vary with the variation in beam span while the tendon depth is kept constant. This is not desirable in this study as it introduces another parameter into the beam series. Therefore, all the test beams were provided with straight tendons so that the only varying parameters are the span-to-depth ratio, L /d ps 0 , and number of deviators. Each beam was instrumented to measure the deections, curvature, concrete and internal steel strains, forces in the external tendons, and crack widths. Each beam was simply supported and loaded monotonically at third-points to failure, thus creating a constant moment region within the middle onethird span segment. Deections were measured using strain gauge type displacement transducers. Curvature was measured using two clamping devices securing two displacement transducers with gauge length of 300 mm, one each near to the

624

C.K. Ng, K.H. Tan / Engineering Structures 28 (2006) 622633

(a) Reinforcement details.

(b) External tendon and deviator conguration. Fig. 1. Test set-up, reinforcement details and external tendon conguration of test beams.

C.K. Ng, K.H. Tan / Engineering Structures 28 (2006) 622633 Table 1 Details of test beams (a) Beams for the study on effect of span-to-depth ratio Beam ST-1 ST-2 ST-3 ST-4 ST-5 Tendon area, A ps (mm2 ) 201 201 201 201 201 Initial tendon depth, d ps 0 (mm) 200 200 200 200 200 Effective prestress, f pe (MPa) 764 771 750 757 760 f pe / f pu 0.40 0.41 0.39 0.40 0.40 Tendon prole Straight Straight Straight Straight Straight

625

Span-to-depth ratio, L /d ps 0 7.5 9.0 15.0 22.5 30.0

(b) Beams for the study on second-order effects Beam ST-5A ST-5B T-0A T-0B Tendon area, A ps (mm2 ) 201 201 201 201 Initial tendon depth, d ps 0 (mm) 200 200 200 200 Effective prestress, f pe (MPa) 762 742 745 742 f pe / f pu 0.40 0.39 0.39 0.39 Number of deviators at equal spacing 2 3 0 0 Tendon prole Straight Straight Straight Straight Span-to-depth ratio, L /d ps 0 30.0 30.0 22.5 30.0

(a) Momentcurvature curves.

(b) Loaddeection curves. Fig. 2. Effect of span-to-depth ratio on loaddeformation curves.

to that at rst yielding of steel reinforcement. Beam ST-5 with large span-to-depth ratio of 30.0 however had the lowest deection ductility ratio of 1.69. This low ductility ratio is due to second-order effects at failure and lower concrete strength of the beam as discussed in the impending sections. 3.2. Stress in external tendons Fig. 3 shows the average stress in the external tendons with the applied load respectively and Table 2 shows the average change in stresses in external tendons, f ps , up to the cracking moment, Mcr , service moment, Ms , and ultimate moment, Mu . The service moment was dened as the ultimate moment divided by a factor of 1.7. The ultimate moment was dened as the maximum observed moment resisted by the beams. The rate of increase in stresses was faster in the external tendons after cracking compared to before cracking. The difference among the four beams, ST-1, ST-2, ST-3 and ST-4, was insignicant as these beams failed in exure (see Table 2). Beam ST-5 however registered a lower stress increase at the ultimate limit state due

Fig. 3. Effect of span-to-depth ratio on stress increase in external tendons.

to lower concrete strength and second-order effects as discussed in the impending sections. After the internal steel reinforcement had yielded, the external tendons began to register rapid increase in stress. At the ultimate limit state, it is noted that the stress increase in

626

C.K. Ng, K.H. Tan / Engineering Structures 28 (2006) 622633

Table 2 Stress increase in external tendons of the beams with different span-to-depth ratio Beam ST-1 ST-2 ST-3 ST-4 ST-5 Concrete strength, f c a (MPa) 34.5 29.9 33.2 28.3 25.1 Stress increase, 0 Mcr 17 18 15 18 17 f ps (MPa) 0 Ms 94 89 85 92 82 0 Mu 443 381 409 366 270 Normalised stress increase [4], 0 Mu 41.8 39.5 39.6 39.4 31.5 f ps /( f c )2/3

a Calculated from measured cube strength using f = 0.8 f . cu c

Table 3 Comparison of experimental ultimate moment to theoretical predictions Beam ST-1 ST-2 ST-3 ST-4 ST-5 Concrete strength, f c a (MPa) 34.5 29.9 33.2 28.3 25.1 Ultimate moment, Mu (kN m) Experimental 97.0 93.6 94.2 87.8 80.0 Mu ,expt. / Mu ,theor. Theoretical 95.0 92.4 94.0 90.4 86.5 1.02 1.01 1.00 0.97 0.92

a Calculated from measured cube strength using f = 0.8 f . cu c

the external tendons was almost proportional to the concrete strength, except for beam ST-5, regardless of the span-todepth ratio (see Table 2). The normalised stress increase in the external tendons [4] shows only 6% of difference at the ultimate limit state for the four beams with span-to-depth ratio up to 22.5. For beam ST-5 with a long span, hence large span-to-depth ratio of 30.0, second-order effects prevailed and the results were not sufcient in describing the null effect of span-to-depth ratio. In this case, the large deviator spacing caused the beam to fail at a section adjacent to the loading point, where second-order effects were the most prevailing within the constant moment region. Test results of beams with more deviators as described in the impending sections are more representative in describing the null effect of span-to-depth ratio in long beams. The lower stress increase in external tendons of ST-5 was partly due to the lower concrete strength of the beam, as the neutral axis at the ultimate limit state is larger leading to lower strain in the external tendons, and hence stress, at failure. This effect is further discussed in the impending section on comparison of test results to theoretical predictions. 3.3. Ultimate strength and failure mode All the test beams failed in exure with satisfactory ductility at the ultimate limit state. Yielding of internal steel reinforcement followed by crushing of concrete on the top face in the constant moment region of the beams was observed. All these beams registered roughly the same ultimate moment capacity (see Table 3). The small differences in beam strength were due to different concrete strengths of the beams. Beam ST5 registered a relatively lower strength and deection ductility

Fig. 4. Effect of span-to-depth ratio on failure pattern (top to bottom: ST-1; ST-2; ST-3; ST-4; ST-5).

ratio as compared to the other beams. This was due to the beam failing at a section adjacent to the loading point, at about one-third span section, where the effective tendon depth was considerably smaller attributed to second-order effects. It is deduced that a deviator at midspan is not sufcient in minimising second-order effects at this failure section of beam ST-5, for which the span-to-depth ratio is 30. The lower beam strength and deection ductility ratio of beam ST-5 was also partly due to the lower concrete strength of the beam as discussed earlier and elaborated later in the section on comparison of test results to theoretical predictions. The appearance of the test beams after being tested to failure is shown in Fig. 4. 3.4. Comparison of test results to theoretical predictions Test results of the beams were compared with the theoretical predictions using the analytical method proposed

C.K. Ng, K.H. Tan / Engineering Structures 28 (2006) 622633

627

(a) Momentcurvature curves.

(b) Loaddeection curves.

(c) Stress in external tendons. Fig. 5. Comparison of test results to theoretical predictions of beams with different span-to-depth ratios.

in Part I of this paper, and the typical comparison of momentcurvature response illustrated by beams ST-2 and ST4 are shown in Fig. 5(a). Fig. 5(b)(c) show the comparisons of loaddeection response and stress in external tendons for all beams ST-1, ST-2, ST-3, ST-4 and ST-5. In this series of beams, no second-order effects were theoretically expected in the beams at midspan section due to the provision of a deviator. Very good agreement between the analytical predictions and the test results was obtained. The lower ultimate stress in external tendons of beam ST5 due to lower concrete strength was also correctly predicted as shown in Fig. 5(c). The smaller ultimate deformation or strain distribution across a cross section of a beam with lower concrete strength is due to the fact that the neutral axis is larger. This larger neutral axis is required as the concrete compression zone of a cross section needs a larger cross-sectional area to gain sufcient compression in order to achieve equilibrium with the tensile forces in the steels, both internal reinforcement and external tendons. This large neutral axis in turn creates a smaller deformation or strain distribution across the tension zone of the cross section, hence leading to smaller strain, and hence stress, in the external tendons at the ultimate limit state. Table 3 compares the test results for the ultimate moment of resistance with the theoretical predictions. The test values were well within 3% of the predicted values, except for beam ST-5 which failed at a value which was 8% lower than the predicted value, due to second-order effects. If a further number of

(a) Momentcurvature curves.

(b) Loaddeection curves. Fig. 6. Effect of number of deviators on loaddeformation curves for beams with L /d ps 0 of 22.5.

628

C.K. Ng, K.H. Tan / Engineering Structures 28 (2006) 622633

(a) Momentcurvature curves.

Fig. 8. Effect of number of deviators on stress in external tendons. (b) Loaddeection curves. Fig. 7. Effect of number of deviators on loaddeformation curves for beams with L /d ps 0 of 30.0. (a) L /d ps 0 = 22.5 (top to bottom: T-0A; ST-4).

deviators were provided in this beam to minimise these effects, then the failure would probably occur within the constant moment region, as shown in the tests reported in the next section. 4. Test results and discussion on effect of number of deviators 4.1. Loaddeformation response For beams with span-to-depth ratio, L /d ps 0 , of 22.5, Fig. 6(a) shows that the section curvature at a particular load was reduced by the provision of a deviator at midspan section. Beam T-0A registered an ultimate section curvature of about twice that for beam ST-4. The loaddeection curves in Fig. 6(b) indicate no differences in behaviour prior to cracking. After cracking, beam ST-4 exhibited higher stiffness than beam T-0A. For beams with span-to-depth ratio, L /d ps 0 , of 30, the momentcurvature curves in Fig. 7(a) also indicate that the beam without any deviator had larger curvatures at a given load. The section curvatures increased with the number of deviators but the differences were small. The measured ultimate section curvatures of beams ST-5 and ST-5A were very low due to the beams failing at sections away from the instrumented sections. The beams were expected to fail at sections near to the midspan where the beams were instrumented to measure curvatures. However, the failure sections were still within the constant

(b) L /d ps 0 = 30.0 (top to bottom: T-0B; ST-5; ST-5A; ST-5B). Fig. 9. Effect of number of deviators on failure pattern.

moment region as expected. Beam T-0B, which had no deviator, registered larger curvatures at failure. The loaddeection curves in Fig. 7(b) show no signicant difference in stiffness for beams ST-5, ST-5A and ST-5B in the uncracked and cracked regimes up to the yield load. Beam T-0B showed no difference in loaddeection response prior to cracking compared to those beams having deviators, but exhibited smaller stiffness in the cracked regimes. The ultimate beam deection was lower in beam ST-5 than in beams ST5A and ST-5B, as the failure section occurred away from the midspan section, and the tendon depth at the failure section was considerably smaller due to second-order effects. The deection ductility ratios as calculated for ST-5, ST-5A and

C.K. Ng, K.H. Tan / Engineering Structures 28 (2006) 622633 Table 4 Stress increase in external tendons of the beams with different numbers of deviators Beam T-0A ST-4 T-0B ST-5 ST-5A ST-5B Span-to-depth ratio, L /d ps 0 Number of deviators 0 1 0 1 2 3 Stress increase, 0 Mcr 17 18 16 17 17 24 f ps (MPa) 0 Ms 68 92 59 82 93 95

629

0 Mu 166a 366 147a 270 376 413

22.5

30.0

a Ultimate load occurred just after yielding of internal steel reinforcement.

Table 5 Comparison of experimental ultimate moment to theoretical predictions for the study on second-order effects Beam T-0A ST-4 T-0B ST-5 ST-5A ST-5B Concrete strength, f c a (MPa) 31.3 28.3 29.3 25.1 31.7 26.4 Ultimate moment, Mu (kN m) Experimental 68.2 (73.7b ) 87.8 52.7 (64.9b ) 80.0 81.8 84.2 Mu ,expt. / Mu ,theor. Theoretical 61.1 (76.8b ) 90.4 N/A (70.7b ) 86.5 84.1 87.9 1.12 (0.96b ) 0.97 N/A (0.92b ) 0.92 0.97 0.96

N/A = not available due to the predicted variation in tendon depth being larger than the headroom (the vertical distance from the initial external tendon level to the bottom face of the beam ange) for relative movement of the external tendons. a Calculated from measured cube strength using f = 0.8 f . cu c b Ultimate load occurred just after yielding of internal steel reinforcement.

(a) Momentcurvature curves.

(b) Loaddeection curves.

Fig. 10. Comparison of theoretical predictions to test results of beams with span-to-depth ratios of 22.5 and 30.

630

C.K. Ng, K.H. Tan / Engineering Structures 28 (2006) 622633

(c) Stress in external tendons.

(d) Variations in external tendon depth. Fig. 10. (continued)

ST-5B were 1.69, 3.02 and 2.78 respectively. With more deviators provided in beams ST-5A and ST-5B, the ductility of the beams had improved to the one observed in beams ST1, ST-2, ST-3 and ST-4. It is deduced that a single deviator is not sufcient in minimising second-order effects in such long beams where the span-to-depth ratio, L /d ps 0 , is 30. More deviators are required to ensure that both beam strength and ductility are not affected by second-order effects in long beams. Beam T-0B showed larger deection near failure but the ultimate deection was not recorded. The beam could not be tested to total failure as the external tendons had come into contact with the lower face of the ange of the beam. 4.2. Stress in external tendons The stress in external tendons was not signicantly affected by the number of deviators for beams with one or more deviators up to the yield load, as shown in Fig. 8. Beam ST-5 registered a lower stress at failure as compared to beams ST5A and ST-5B due to second-order effects and lower concrete strength as discussed earlier. Beams T-0A and T-0B showed no signicant difference in the stress increase in external tendons up to the yield load level compared to the corresponding beams with deviators. The stress increase in external tendons for these two beams at the ultimate limit state, however, was much lower due to reduction in effective tendon depth.

The average change in stresses in external tendons, f ps , up to the cracking moment, Mcr , service moment, Ms , and ultimate moment, Mu , are summarized in Table 4. For beams with span-to-depth ratio, L /d ps 0 , of 22.5, the stress increase was almost identical before cracking, indicating that the second-order effects were minimal. Beyond cracking, the stress increases in the external tendons were in general smallest in beams without deviators and largest in beams with larger numbers of deviators. Comparing beams T-0B, ST-5, ST-5A and ST-5B which had span-to-depth ratio, L /d ps 0, of 30, the stress increases in the external tendons were again almost identical before cracking, except for beam ST-5B which showed larger stresses at the cracking load that was substantially higher than those for the other beams. After cracking, the stress increase was generally smallest in beam T-0B, and largest in beam ST-5B. It is also noted that beams T-0A and T-0B registered very low increase in external tendon stresses at maximum load carrying capacity which occurred immediately upon the yielding of internal steel reinforcement. 4.3. Failure mode and ultimate strength Beams ST-4, ST-5, ST-5A and ST-5B failed in exure with yielding of internal steel reinforcement followed by crushing of

C.K. Ng, K.H. Tan / Engineering Structures 28 (2006) 622633

631

Fig. 11. Comparison of theoretical predictions to test results of beams tested by Mutsuyoshi et al. [3].

concrete on the top face in the constant moment region. These beams showed satisfactory ductility at failure. Beams ST-5A and ST-5B failed nally with relatively large deections. In beams T-0A and T-0B, the maximum moment of resistance corresponded to the yielding of internal steel reinforcement. The appearances of the beams after being tested to failure are shown in Fig. 9(a) and (b). The ultimate strengths of the beams are shown in Table 5. Beam T-0A had lower ultimate load than beam ST-4 due to the absence of deviators. Beam ST-5B which had three deviators registered the highest strength among the four beams with L /d ps 0 ratio of 30, although the concrete strength of this beam was the lowest. Comparing T-0B with ST-5, it is deduced that beam strength can be reduced by 19% if no deviator is provided for beams with a span-to-depth ratio of 30.

4.4. Comparison of test results to theoretical predictions Fig. 10(a)(d) show the comparison of test results of beams T-0A, T-0B, ST-4, ST-5A and ST-5B with the predictions of the proposed analytical model. These beams had span-todepth ratios of 22.5 and 30 and different numbers of deviators. Generally, the theoretical predictions agreed with the test results and the decrease in load-carrying capacity of beam T-0A after the yield load was also well predicted by the method. In beam T-0B, the stiffening effect in the predicted curve after the total load of 50 kN as shown in Fig. 10(c) is attribute the underestimation of tendon stress due to second-order effects. The point of the ultimate limit state was also not obtainable as the external tendons were predicted to move beyond the top surface of the beam. This is not practically possible as observed

632

C.K. Ng, K.H. Tan / Engineering Structures 28 (2006) 622633

Fig. 12. Comparison of theoretical predictions to test results of beams tested by Yaginuma [6].

in the experiment, in which this was prevented by the bottom face of the ange. The ultimate strengths of the beams predicted by the proposed analytical model are given in Table 5. All the test results were well within 8% of the predicted values, except for beam T-0A due to the over-estimation of reduction in external tendon depth. 5. Validation of theoretical predictions In order to validate the proposed analytical model further, it was used to predict the exural response of two other series of beams, Series M and Y, tested by Mutsuyoshi et al. [3] and Yaginuma [6] respectively. Comparison was also made with the member analysis approach proposed by Mutsuyoshi et al. [3]. Series M [3] comprised T-beams No. 1 and No. 2, each with an effective span of 5.2 m, and tendons draped at two deviators, which were symmetrically placed along the span with spacings of 1.8 m and 3.0 m in beams M-1 and M-2 respectively. Series

Y [6] consisted of rectangular beams OA-1, SA-1 and NA1, having straight tendons, an effective span of 4.0 m, and tested under two symmetrical concentrated loads spaced at 600 mm apart. Beam OA-1 had no deviator; beam SA-1 had two deviators provided under the point loads and beam NA1 had internal unbonded tendons. The concrete strengths were 39.3 MPa and 31.7 MPa for Series M and Series Y respectively. The test results of Series M beams were compared to the predictions of both the proposed approach and the member analysis approach by Mutsuyoshi et al. [3] in Fig. 11. In general, both methods predicted the loaddeection curves, external tendon stress and variations in external tendon depth well. The present analytical model however, gives slightly more exible response for loading above the yield load. Fig. 12 compares the test results of series Y beams with the predictions of the proposed analytical model. The loaddeection response and the external tendon stress for the linear elastic uncracked regime were well predicted. Beyond the uncracked regime, the proposed analytical model overestimated

C.K. Ng, K.H. Tan / Engineering Structures 28 (2006) 622633

633

the deection and external tendon stress but these were still well within reasonable values. In beam OA-1, the predicted ultimate tendon stress was lower than the tendon stress at yielding due to overestimation of second-order effects. From the above comparison with test results and other analytical methods based on member analysis, it is seen that the proposed method provides a simple approach which gives reasonably good predictions of the exural behaviour of externally prestressed beams that are subjected to second-order effects. 6. Conclusions

in satisfactory ultimate load behaviour, that is, strength and ductility. For beams with larger span-to-depth ratios of more than 22.5, the provision of two deviators, one each at the onethird span section, is recommended. Theoretical calculations based on the proposed pseudosection analysis method considering second-order effects were found to be in good agreement with the experimental results with regard to ultimate strength as well as the deections, stress increase in the external tendons, and the variations in external tendon depth. References

Tests conducted on simply supported, externally prestressed beams with a deviator at midspan and span-to-depth ratios ranging from 7.5 to 22.5 indicated that span-to-depth ratio has no signicant effect on the stress increase in the external tendons. However, for a beam with span-to-depth ratio of 30, it was found that one deviator provided at midspan is not sufcient in minimising second-order effects. Based on tests carried out on simply supported beams with straight tendons and span-to-depth ratios of 22.5 and 30, with or without deviators, it was found that second-order effects due to changing tendon eccentricity which leads to lower loadcarrying capacity and ductility, can be effectively minimised by the provision of deviators along the span of the beam. For beams with span-to-depth ratios of up to 22.5, the provision of a single deviator at the section of maximum moment resulted

[1] Naaman AE, Alkhairi FM. Stress at ultimate in unbonded post-tensioning tendons part 2: proposed methodology. ACI Structural Journal 1991; 88(6):68392. [2] Alkhairi FM, Naaman AE. Analysis of beams prestressed with unbonded internal or external tendons. ASCE Journal of Structural Engineering 1993; 119(9):2680700. [3] Mutsuyoshi H, Tsuchida K, Matupayont S, Machida A. Flexural behaviour and proposal of design equation for exural strength of externally pc members. Journal of Materials, Concrete Structures and Pavements 1995; 508(26):6776. [4] Ng CK. Tendon stress and exural strength of externally prestressed beams. ACI Structural Journal 2003;100(5):64453. [5] MacGregor JG. Reinforced concrete: mechanics and design. 3rd ed. Upper Saddle River (NJ): Prentice-Hall; 1997. [6] Yaginuma Y. Non-linear analysis of ultimate exural strength of beams with external tendons. Journal of Prestressed Concrete, Japan 1995;37(3): 5465.