Sunteți pe pagina 1din 20
43 andi ‘ani Syrposive 2003 DEFORMATION CAPACITY OF RC MEMBERS, AS CONTROLLED BY FLEXURE OR SHEAR Michael N, Fardis and Dionysis E. Biskinis Siructures Laboratory, Deparoment of Civil Engineering, University of Patras, Greece E-mail: fardis(@upatras gr SUMMARY ‘A large database of monotonic and cyclic uniaxial test results i utilised to develop models for the deformation capacity of R/C members, The database covers beams, columns with rectangular or circular section, walls with rectangular, T- or barbelled section, and hollow rectangular piers. Most of the tests in the database are up to specimen to failure, conventionally defined as a post-peak drop in lateral load resistance by at least 20%. One type of models developed for the chord rotation (or dri) capacity for flexure-controlled failure is based on curvatures in the plastic hinge - at yielding and ultimate, calculated on the basis of first principles and of different confinement models at ultimate - and on expressions for the plastic hinge length empirically fitted to the data. Except for columns with circular section, the predictions of this type of models are characterised by unacceptably large scatter and, often, by significant bias. Purely empirical models, statistically fited to the data, are found to offer better predictive capability forthe flexure- Controlled chord rotation capacity of all types of members in the database with rectangular or quasi-rectanguler section, For members under cyclic loading ultimately failing in shear after yielding in flexure, expressions of the familiar type are developed forthe reduction of shear resistance with the chord rotation ductility ratio, These expressions, applicable over all types of members in the database, are characterised by low scatter forthe prediction of shear resistance in terms of the post-lastic cyclic displacements, but cannot be meaningfully inverted to give a shear-controlled ductile deformation capacity. A model based on fist principles with empirical corrections is also developed for the chord rotation (rift) at member yielding, asa tol forthe models of deformation capacity as controlled by flexure or shear, a8 well a forthe calculation of the secant stiffness of members at yielding 1. INTRODUCTION Recent years have seen a large interest of the international earthquake engineering community in the ‘quantification of the deformation capacity of R/C members. The emergence of procedures for seismic ‘assessment of existing structures which entail member verifications explicitly in terms of deformations [ASCE, 2000 and 2001, Comité Européen de Normalisation, 2003a] and the forthcoming codification of the design of new structures directly on the basis of nonlinear analysis with explicit checks of member deformations (Comité Européen de Normalisation, 2003b], provide strong motivation for this interes. To support the effort of quantification of the deformation capacity of R/C members, the senior author of the paper and his co-workers have been assembling atthe University of Patras over a period of almost ten years a databank of tests on R/C members. Early modes for the deformations of R/C members at yielding and at flexure-controlled failure under monotonic or cyclic loading have been developed on the basis of the databank and reported in [Panagiotakos and Ferdis, 2001), The results of more recent efforts, still on flexure-controlled members, are included in [Federation Internationale du Beton, 20032, 2003b]. The present paper represents a major stride inthis on-going effort, not only because the database has recently been re-evaluated and increased in size by almost 50%, but mainly because it has been used also for the {quantification of the resistance and deformation capacity of members which are controled by shear. su or ts6 | por sq Giz. a | os U6 aq] cee TC = oo a | ve | om sq] ox a | o o | « falel + or 0 oO 6 els £ oe a a orf sot | 0 | oe wee i [oer sie 9S 6s oO 0 oO ws ele ’ 07 - on foe] ce faze | em oe Preindueooy - unumyo: a abe SSB ar oo a Fa oat bauajnog ou Sauna ys SUDO a ‘Surpjard jesnxay puosag o11949 pee eee eee pee ‘any Jo apour pe Kzonsyy peo way s10q "Buo] Jo ais ‘(uoneyou p10ya) pao paxy 2e syxv raquiom oy j2adsox yyA weds 1¥ays Jo UoRD2yap Jo WoUTOANseIUN YEA $9699 JIqmIOUE Jo asuqEE | 248, 43: Fis ‘Ota Symposiu 2003 2, THE DATABASE OF TEST RESULTS ON R/C MEMBERS ‘The databank of tests used in this paper comprises mainly specimens subjected to uniaxial transverse (i lateral) loading with or without axial load - constant or varying. Of interest here are only specimens of that type. The main reason for assembling the database has been the development of models for the deformations that develop over the shear span, L,, of R/C members fixed at the section of maximum moment. Therefore, the databank is limited mainly to tests which report the transverse deflection at, oF near, the point of zero-moment with respect to the specimen axis at the section of maximum moment, ‘Such tests are on simple- or double-cantilever specimens, or on simply-supported beams loaded only st ‘mid-span. This deflection is used here in the form of drift (ratio or angle), @, ie. deflection divided by the distance from the point of measurement to the section of maximum moment. Normally this distance is ‘equal to the shear span or a multiple of it, and then @iis also the chord rotation atthe section of maximum ‘moment. Table 1 gives the breakdown of this type of specimens in the databank, depending on the specimen geometry (beams, columns - with conventional or diagonal reinforcement -, walls, or piers), type of loading (monotonic or cyclic), the mode of yielding or failure (due to flexure or shear, by web ‘crushing or interface shear, etc), and the occurrence or not of (bond-) slippage of longitudinal bars from their anchorage zone beyond the section of maximum moment. Normally, in simple- or double-cantilever specimens a certain amount of such slippage (pull-out) takes place, producing a fixed-end rotation that contributes to the deflection of the shear span. Due to symmetry, there is no such slippage in simply- supported beam specimens loaded only at mid-span, except when the load is applied through a bulky stub, Tong enough for rebar slippage to develop on both sides of the mid-span, Table2 Database of member tests with measurements of curvatures slip of Toad history and mode of failure Il tests] All longitudinal. onoronic|_eyelie Beyond flexural yielding | beyond] texual ‘bars from ‘d flexural failures beyor ‘flexural fail ember type and cross anchorage | flexural ine’ _Wvielding| ion AI] yielding - fa pes|20) cpl xara Be | 09" yey failure +8) | OHO) mea [a om lumn - Rectang- (@ [iter convertionatfi3a 36] 174) 49 | 1 | 70 6 | 76] 126 | 12s reinforced © gularorT [27/162] 199] 181 | 0 | 6 o | 6 | 187 | 187 i beams Nofolumns — with 7s198] 373) 230 | 1 | 76 6 | s2| 313 | 312 angular web @ [Wall - Rectangular 5[O] 5 [0 o 3 2 seins 5 Wall - T- of © parbelled 7 7 Oo oO 1 o 1 1 1 OF pvas-au—[efefe lo [ol + 2 [ol 6 | 6 (ee (AN ‘columns, \eynco) apne ear 1871198] 385 | 230 1 80 8 88) 319 | 318 ty Pi =a 4 oO} 14 0 oO 8 0 8 8 8 or members kore Mt iber 01198 399) 230 i 88 8 96 | 327 | 326 Tn some tests of the databank the relative rotation between the section of maximum moment and a nearby section within the plastic hinge region has been measured and translated into an average curvature, 4, of the plastic hinge region, including or not effects of reinforcement slippage from its anchorage beyond the 513 3 Faria ‘Otani Symposia 2003 section of maximum moment (Table 2). Some of these tests are not included in Table 1, because deflections with respect to the specimen axis atthe section of maximum moment were either not reported, Or could not be derived from the measurements (e.g. in beams with loading other than at mid-span alone). Most of the tests in the databank were continued up to specimen ultimate conditions (failure), identified with a distinct change in the measured lateral force-deformation response: in monotonic loading, with a noticeable drop of lateral force after the peak (at least 20% of the maximum resistance); in eycling loading, with a distinct reduction of the reloading slope, or ofthe area of the hysteresis loops, or of the peak force, compared to those ofthe preceding eycle(s), typically associated with a drop in resisting force ‘greater than 20% of the maximum resistance. Reinforcing ste! in the tests has been classified into three grades. Most specimens have ductile hot-olled steel with hardening ratio, 4, in the order of 1.5 and strain at peak stress, Gu around 15%, In European tests heat-treated tempeorestcel has been used after the early “90s, with a value of f/f around 1.2 and of & of the order of 8%. In about 60 monotonically tested European specimens, brittle cold-worked steel thas been used, with a value of ff about 1.1 and of éy around 4%, ‘Table 3 Mean‘, median* and coef. of variation of ratio experimentabto-predicted values at yielding, FET cane foodiane| soefficient a Frew tin RE 20/4) Beams, colurans w/ rectangular section - w/o slip 198] 1.05 | 1.00 | 29.5% eo/G.ec) Beams, columns w/ rectangular section - w/ slip 1151] 1.05 | 0.995 | 40.0% ow/4.e4a) Beams, colurans w/ rectangular section - All 1349] 1.05 | 0.995 | 38.6% 0/920) Walls (all w/ slip) 145] 1.015] 0.99 | 32.5% Columns w/ circular section (all w/ stip) 160| 1.05 | 0.99 | 33.4% a0/ Petar Beams, columns w/ rect. section = wo slip 198 [1.325 1.275 | 29.3% o/s Beams, columns w/ ree. section - w/ slip 175] 1.205} 1.06 | 37.6% sri-lnscnss BEAMS, columns w/ rect. section - All 373| 1.27 | 1.205 | 33.4% ha) My t-te recip Beams, columans & walls w/ rectangular section [1513] 1.025] 1.015 | 16.2% IM, ‘Columns w/ circular section 18i| 1.015] 1.005 | 16.7% (M,L-l3 0, exg(MyLo/3@)ora Beams, columns & walls wi rectan. section [1412] 1.10 | 1.035 | 40.9% Columns w/ circular section 152] 1.07 | 1.035 | 31.2% is igh, the median is more representative of the average tread than the mean, asthe ‘median ofthe ratio predicted-to-experimental value is always the inverse ofthe median ofthe ratio experimentalto- ‘predicted value, whereas the mean of both ratios is typically greater than the median, 3. DRIFT (OR CHORD ROTATION) AT MEMBER YIELDING ‘The value ofthe chord rotation at yielding at the corresponding end of the member, 8, is important as the baseline for the plastic component of the ultimate chord rotation (plastic rotation capacity), as well as the normalizing factor of chord rotation (total or plastic component), whenever this latter is expressed as ductility ratio, up. More importantly, 6, determines, through Ba.1), the value of the secant stiffness of the shear span, L,, at member yielding, often taken as the effective clastic stfThess in a bilinear force- deformation model ofthe shear span under monotonic loading: M, Mats 0 38, ‘here M, is the yield moment inthe bilinear M-8 model of the shear span. Eg ‘The following expressions were derived from those tests in the databank with yielding in flexure: For beams or columns with rectangular section: S14 43 Fads ‘tani Symposium 2003 24S, 4. 40,00275 +a, — #000275 404-7 al @ For walls with rectangular, T-shaped, ot H-shaped section, or for barbelled walls: 0 ~4, 140.0025 +04 292 Ls Pea ere ea Neat cd For columns with circular cross-section: L Od phoney at w fg, denotes the yield curvature atthe section of maximum moment and ifthe moment diagram is linear ‘over the shear span, the 1" term in Eqs.(2)(4) is the contribution of flexural deformations to 0, The 2" term represents the magnitude of shear deformations within the shear span at flexural yielding and has been found to be practically independent of any factor other than the type of member in Eqs.(2)-(4). The 3° term isthe fixed-end rotation due to bar pull-out from the anchorage zone and does not appear in Eqs. (@+{4) when such pull-out is not physically possible. In that case ay=0 is used in Eqs. (2)-(4) forthe zero- ‘one variable ay, whereas a,=1 applies if such pull-out is considered as possible. Inthe 3 term, g, is the yield strain of the tension reinforcement, d or d’ denote the effective depth to the tension or to the ‘compression reinforcement, respectively, dy isthe diameter of the tension reinforcement and ff. the yield stress of tension reinforcement and the compressive strength of concrete (both in MPa). The 3 term ‘corresponds toa linear reduction of steel stress from f, to zero over a development length deriving from a mean bond sess of 0.625 /f, (which is low in comparison othe ulimate bond sess of about 2/7, oF 25 AF; in unconfined or confined concrete respectively, but gives the bes fit of Egs.2)(4) tthe data) FL iz meeaatee 7] Lead i 1, t| et at | | at | t ae i d 5 : i 4 i TT ‘ [Te i= / . «| 484 _. an (a) beams or columns; ((b) walls; rectangular, T- or H- (¢) columns; rectangular section, Eq,(2) section, Eq.(3) cireular section, Eq.(4) Figure 1 Comparison of experimental chord rotations at yielding to values from Eqs.(2)(8) ‘The ft of Eqs.(2)(4) tothe data is shown in Figure 1, while the statistics of the ratio of experimental to predicted values are given in Table 3. This fit corresponds to a yi derived from fist principles. More specifically, for rectangular or T-sections, 9, is derived on the hypothesis, of equilibrium, and of linear o- laws up to a stes! by the tension reinforcement, or up to a concrete strain of 0.9f2/E« i compression (see [Panagiotakos and Fardis, 2001] for expressions for 9). Similar assumptions are ‘employed for circular sections, except that the concrete o-e law is taken parabolic up toa strain of 0.002 and recourse has to be made to an iterative algorithm [Biskinis et al, 2002}. Possibly due to the way in which experimental curvatures are derived from the relative rotations of two sections within the plastic hinge, the expressions developed for g, from first principles do not provide unbiased estimates of the yield curvature in the 373 tests where such curvatures were measured. This is ‘lear from Figure 2 and from the fact that median ratios of experimental-to-predicted ¢,, listed in Table 3, sis Faris ‘Oxi Symposium 2003 are well above 1.0 (interestingly, they are greater in the 198 tests in which measured curvatures were not affected by slippage of the rebars from the anchorage zone, than in the 175 tests where they were affected). Despite this discrepancy between measured values of g, and the ones calculated from first principles, a closer fit to the data is possible through expressions of the type Eqs.(2)-(4) if these latter values are used instead of the empirical expressions: p, = 2.16 /d or g, = 1.9e/h which were found to provide an unbiased fit to the measured values of g, in the 373 tess, without & significantly increased scatter over that of the fundamental expressions for gy 2 om mee ar Hee baw : ~ i . | on 1 : LL fa aeceatectit cane (@) beams and columns with rectangular section; _(b) beams and columns with rectangular section; to slip of longitudinal bars slip of longitudinal bars Figure 2 Comparison of experimental curvatures at yielding to values calculated from 1" principles My tm 4 i oe (@) rectangular sections (®) columns with circular section Figure 3 Comparison of experimental yield moments to values calculated from 1" principles To complete the picture regarding the application of Eq(1) 1 estimate the secant span, Ly at member yielding, the yield moment 44, calculated from first principles (with same assumptions and references mentioned above forthe calculation of the yield curvature, p,)is compared Figure 3 to the experimental “yield moment”, taken as the moment at the comer of a bilinear Mf-8 curve fited to the envelope of the measured 4-0 response of the shear span. Satstcs of the ratio of the experimental yield moment to the calculated value are listed in Table 3. The median value of this ratio exceeds 1.0, as the corner of the bilinear approximation to the experimental MM-9 curve lags behind first 516 45: Fans (ani Symposiaen 2003 yielding of the tension reinforcement or strong nonlinearity of the extreme compression fibers. The difference is greater in circular sections, as there the experimental M-0 response curves down more gradually. So, for such sections the experimental value of M, is compered in Figure 3 and in Table 3 to the average of M, and ofthe theoretical ultimate moment, My of the section, both computed (through iterations) onthe basis of first principles as described in [Biskinis etal, 2002]. Statistic ofthe ratio of the ‘experimental secant stiffness at member yielding to the value calculated from Eq(1) on the basis of the yield moment M, from first principles and the values of @, from Eqs.(2}-(4) are also listed in Table 3. For ‘members with rectangular or circular section, the experimental secant stiffness at member yielding is on average 25% or 30%, respectively, of that of the uncracked gross section; but its scatter about this later ‘values is far greater (around 70%) than about the value given by Eq.(1)- ‘Table 4 Mean, median, coefficient of variation of ratio experimental-to-predicted values at ultimate vant ee cer! Paved. tse srincipler Mandertq (6) Monotonic loading - Rect. sections 230} 1.12 | 0.995. I Pure iii Pssiny Eq 7) Monotonic loading - Rect. sections _|230) 1.055) 0.715 Perm srencen se Monon loading - Reet. sections 230] 2.36 | 1.97 nsstasin Monotonic loading -Rect. sections _|230| 1.09 | 0.995. arin Mn (9 C¥eHE loading ~ Rectangular sections | €9 | 138 | 1.13 89 |o.9ss| 0.545 Pa fraermsincor ce Cl ong Revs os 89 |1.905| 1.33 ren Cyelic loading -Rectang, sections _| 89 | 1.28 | 0.995 cea) C¥elie- Conforming members w/ es. web. Ea(10)(12)(I3) 888] 1.125] 1.005 0, sa Cyclic. Conforming beams, rect. columns. Eqs.(10),(12),14) 823] 1.10 | 1.00 cai) Cyclic. Conforming circular columns. Egs.(10),(12),15) | 76 [1.035] 1.00 6, a Cyolic. Nonconform. rect. columns, walls Eqs.10),(12),16) | 36 10,995] 0.995 8.) Monotone, Reams, restangular colons. as (10)(V1)(17) [276] 135 [1.005 11 Monotonic. Beams, rectangular columns. Eqs.(10)(11),(18) |276| 1.36 | 1.00, [@vexs/A.x9(s) Cyclic. Conforming beams, rect. columns, walls. Eq.(8),(19)| 888] 1.10 | 0.995 ne Gcen(s Cyclic. Conforming beams, rectangular columns. Eq.(8)(20)| 823 1.09 | 1.00 wc O.4\s) Cyclic. Conforming circular columns. Eqs.(8),(21) 76 | 1.10 | 0.995 1 Cyclic. Non-conforming rectang, columns, walls. Eq.8),22) 36 | 1.12 | 1.10 6,c4c2 Cyclic. Conforming beams, rectan. columns, walls, Eq.(23) [880 [1.025] 0,995 zee) Monotonic. Conform, beams, rect. columns, walls. E23) | 279 1.095] 1.005 /Gazacxy Monotonic or cyclic. Conforming walls. Eq (23) 59 | 0.98 | 0.995 zac) Monotonic or cyclic. Conf, beams, columns, walls Eq(23) |1159|1.045] 1.00 G,es:o, Cyclic, Non-conforming rectang. columns, walls. £q(23)_| 36 [0.855] 0.835 e/8.2q(u) Cyclic. Conforming beams, ectan. columns, walls. Eq,(24) | 80] 1.05 | 1.00 /Bxacam Monotonic. Conform. beams, rect. columns, walls. B.(24) |279| 1.135] 1.025 Asa Monotonic o eycic. Conforming wals.9,(24) '59 | 1.02 | 1.005 /Bseaian Monotonic or eyclic. Conf. beams, columns, walls. Eq.(24) |1189}1.075| 1.00 |ozea/ Ox sai34 Cyclic. Non-conforming rectang. columns, walls. £q(24) | 36 10.875] 0.865 Waew/Pasas) Ductile shear. Colurans with rectangular section. Ea25) ]146]0.995| 0.99 Waew/Vasyas, Ductile shear. Colurans with circular section. Eq(25) | 45 |1.045] 1.008 \ae/Vaeacs, Ductile shear. Walls with rectangular or T-section. Eq.25)| 6 | 0.99 | 0.99 ‘xexg/Vi.0q;25) Ductile shear. Piers with hollow rectang. section. Eq.(25) | 11 | 1.10 | 0.995. Vara/Vesuc Ductile shear. Columns, walls or piers. Eq. (25) 208| 1.01 | 0.995 Vaew/Pasacn Ductile shear. Columns with rectangular section. Eq.(26) | 146] 1.00 | 1.00 Vasw/Vaeqco) Ductile shear. Columns with circular seetion. Eq.(26) | 45 [1.015] 0.975 Wacw/Vasacao Ductile shear. Walls with rectangular or T-section, Eq(26)| 6 | 1.08 | 1.06 VV Due she. Ps with lio rest seston, B26) | 11 | 185/148 ‘tain Ductile shear. Columns, walls or piers, Fa.(26) 208 | 1,015] 1.00 , Web crushing. Walls with rectangular or T-section, Eq.(29} 37 | 1.02 | 1,00 si7 43: Frdie ‘Otani Syempasiu 2003 4, ULTIMATE DRIFT (OR CHORD ROTATION) FOR FLEXURE-CONTROLLED FAILURE 4,1 Introduction ‘Specimens that failed after yielding in flexure were distinguished ftom those failing in shear before flexural yielding, on the basis of the following two criteria: (a) the reported behaviour before and at failure and (b) a comparison of the measured lateral resistance with the resistance derived from the value Of the yield moment, M,, calculated from first principles (taking into account the scatter exhibited in Figure 3). Specimens characterised in this way as yielding in flexure, were further categorised as ultimately failing in flexure or in shear on the basis of the reported experimental behaviour at failure. ‘When the available information was incomplete, unclear or unconvincing, recourse was also made to comparisons of the measured ultimate deformation withthe predictions of models such as those described in the rest ofthis paper for flexure- or shear-controlled ultimate behaviour. 42 Formulations based on curvatures and plastic-hinge length If inelastic behaviour and failure is controlled by flexure, the familiar description of the plastic ‘component of drift ratio (or chord rotation) over the shear span Las the product of the plastic component Of (he ultimate) curvature, uo), and a plastic-hinge length, Ly. is very appealing: aatott ot tsb ‘Notwithstanding its mechanical and physical appeal, the real criterion for the value of Eq.(5) is its ability ‘to predict the experimental ultimate drift ratio or chord rotation, @,. Empirical expressions for Ly needed to this end cannot be developed independently of the models used for the other variables entering into Eq(5), namely for 8, 9, andg,, To maintain the apparent rationality of Eq.(5), priority should be given to models based on rational mechanics. For 4, the model based on first principles and outlined in Section 3 isanaturl choice, despite the unsatisfactory agreement with “measured” values displayed in Figure 2. ost, Ss) 3) The natural choice for gy i a model wit similar basis as that for, namely the plane sections hypothesis and equilibrium, but with nonlinear o-¢ laws. [Panagiotekos and Fardis, 2001] presented analytical expressions for g, for sections with rectangular compression zone, unsymmetric reinforcement concentrated at the two flanges and uniformly distributed (web) reinforcement in-between, ‘These expressions are based on: (a) o-¢ law for steel taken as elatic-perfectly plastic for relatively low steel strains, such as those associated with section ultimate conditions due to crushing of the concrete, oF clasti-linearly strain-hardening from f, t 610 the ultimate strength at strain é. for the large steel strains typical of section failure duc to stec! rupture; (b) a o-¢ law for concrete which is parabolic up to ultimate strength and horizontal thereafter up tothe ultimate concrete strain. The expressions for take into account spalling of the unconfined concrete cover and confinement of the concrete inside the hoops thereafter. The only important open parameter is the model to be used for the ultimate strength, the associated strain, age and the ultimate strain of concrete, gas these are affected by confinement. The vvalue of 9, is sensitive mainly - if not only - to the ultimate strain of confined concrete, 6... The options considered for the confinement model are: 8) The original Mander model [Mander et al, 1988). It comprises: (a) an increase of f, and é with confining pressure p, which is in good agreement with triaxial test results on confined concrete in ‘concentric compression and (b) @ concrete ultimate strain derived from a postulated conservation of strain energy. giving, if the ox law of conerete (confined or not) is taken as horizontal after ultimate strength: ae Fn t Enel PS! Sec) © where &, may be taken equal t0 0.004 and gis the ukimate strain of transverse reinforcement. b) The Mander model, with Eq4(6) as modified by Priestley {Paulay and Priestley, 1992]: 318 43: rds (Otani Symposia 2008 Bone = 0.00441 46 6 20 Spe! Soc) om ©) The model in CEB/FIP Model Code 90 {Comité Eurointemational du Beton, 1993}. This is the reference model in Europe for confinement, as it has been adopted in the 2003 version of Eurocode 2 (the European concrete design standard) and, therefore, by Eurocode 8 (the European seismic design standard) as well. This model provides for a more modest increase of f.and fo with p, and for an ultimate strain éay. of: boas “Fou + 0:2pI f= 0.0038 + 0.200, fy! fo ®) where 9, isthe transverse reinforcement ratio (minimum among the two transverse directions), fw its ryield stress and crthe confinement effectiveness factor according to [Sheikh and Uzumeri, 1982]: a), 58 ee Be) ° ‘with 5, the centerline spacing of stirups, b. and h the dimensions of the confined core to the inside ff the hoop and b; the centerline spacing along the perimeter of the cross-section of those longitudinal bars (indexed by i) which are laterally restrained by a stirrup comer or by a cross-tie. 4) The following expression for the strength of confined concrete, giving slightly lower strength ‘enhancement than the Mander model: ay fot 1fos ee) } aw along with the following variations of Eqs.(6) or (7): bau =0.004+ 0.66 mul2Pifin! foe) ay Fane 20.004 +1 Sn P.Sye! Le) «2 ‘As shown in Figure 4 and $ and by the statistics of the ratio of experimental fo predicted values in Table 4, in members with rectangular section confinement model option (d) and Eqs.(11) and (12) provide a better average fit to the measured values of ¢g, in monotonic or cyclic tests, respectively, than the other three alternatives, and with less scatter. Alternative (a), with Eq.(6) derived from the original Mander model, provides almost the same average agreement as confinement model option (d), especially with monotonic data, albeit with significantly larger scatter, Considering the comparison with test results as a vindication of confinement model (d) and of Eqs.(11) and (12) (Whatever the value of the measured data may be, in view of the scatter and the uncertainty introduced by the gauge length), empirical expressions are derived for the plastic hinge length, Ly, to fit E4(6) to the data on ultimate chord rotation, @,, of all members failing in flexure, using the value of ‘derived from first principles and confinement model option (4) above. The value of @, used in Ea.(5) is. also the one derived from first principles. It was found that a better overall fit of Eq.(5) to the data on 6, is possible, if Eqs.(2)-(4) are used forthe chord at yielding, 6, instead of their 1" (flexural) term alone. It ‘was also found that the same expression for Lyi cannot fit both the monotonic and the eyclic data. ‘Moreover, for eyclic loading different expressions are appropriate for members with or without detailing. {for earthquake resistance (often called “conforming” vs. “non-conforming” detailing, also in Tables 1 and 2), Expressions tried for L, are linear combinations of the shear span, L,, and/or of the section depth (or D in circular sections). A term proportional tothe product ofthe diameter ofthe tension reinforcement, dy, and its yield stress, f, is added, to account for the effect of slippage of the longitudinal reinforcement from its anchorage zone beyond the section of maximum moment. This term is multiplied with a4, where ‘a= 0 if such slippage is not physically possible, while au 1 if it is. A term inversely proportional to {Fe +88 in the 3 term of Eqs.(2)-(4) for dependence on ultimate bond stress, is not to advantage. 519 43: Faris Otc Symposivn 2003 omni © ‘reesttm| © Figure 4 Measured ultimate curvatures in monotonic or eyelie tests of members with rectangular section, compared to predictions derived from 1" principles using various confinement models: (a) Mander model, Eq,(6); (b) Priestley model, Eq(7); (c) CEB/FIP MC90, Eqs.(8),(9); (d) Eas. (10)-(12) ‘The expressions found to provide the best overall fit to 6, for members failing in flexure are: For “conforming” beams, columns and walls with rectangular web, under eyclic loading (see Figure S(a)): Lyris "0.0261, + 0.3h +0 22 MPO bs 7 ° (13) {Alternative expression for “conforming” beams and columns with recangular section (not for walls), under cyclic loading (see Figure 5(b)): “AAMPa) Lptaga “03h +0 ds nok o “ For “conforming” columns with circular section, under cyclic loading (see Figure 6(a)): A,(MPa) Igserey =OATSL, +44 OPO a, ens uy fe 520 43: Fandis ‘Otani Symposium 2003 — nett senate T | ooo 5 ° a TO ne @ ae Figure $ Comparison of experimental ultimate chord rotations in cyclic tests to the predictions of Eq.G5) for confinement according to Eqs.(10),(12): (@) “conforming” beams, columns or walls with rectangular section, plastic hinge length per Eq.(13); (b) “conforming” beams and columns with rectangular section, plastic hinge length from Eq.(14). fe ron a 1 - dan Phe 7 e" 2 k pra 3° t tp attt e é, é, of eA |__ TAT hee ‘ . te : i | ae © ae ) Figure 6 Comparison of experimental ultimate chord rotation in cyclic tests of columns with circular section, with the predictions of Eq.(6):() for confinement according to Eqs-(10)(12) and plastic hinge length from Eq(15); or b) for confinement according to Eq.(8) and plastic hinge length from Eq.(21) For “non-conforming” columns and walls with rectangular section, under eyclic loading (see Figure 7(a)): Ia = 0.0251, +0.125h4 0, POO, dy a6) For beams and columns with rectangular section (not walls), “conforming” or not, under monotonic Toading (see Figure 8(a)): Lunia *007L, +05h+0y 207? 4, an ‘of, as an almost equivalent alternative (se Figure 8(0): Lpimen =08h+ 0, 22 g, as) 2 521 43 Fars (Otani Symposium 2003 : wCe I I eh hemlet z i V : i y i: > a si bucate a a | an? ae eho @ ) Figure 7 Comparison of experimental ultimate chord rotation in cyclic tests of columns or walls with rectangular section and “non-conforming” detailing, with the predictions of Eq.(S): (a) for confinement according to Eqs.(10)(12) and plaste hinge length from Eq,(16)s (©) for confinement according to Eq(8) and plastic hinge length from Eq.(22) ot ea tare Ineo oti Se 8) @ cee © Figure 8 Comparison of experimental ultimate chord rotations in monotonic tests of beams or columns with rectangular section, “conforming” or not, tothe predictions of Eq.(5) for confinement ‘according to Eqa.(10}(11) and plastic hinge length from: (a) Eq.(17) oF (6) EqAl8)- Because option (c) (ie. the model in CEBV/FIP Model Code 90) is now the reference model in Europe for confinement - as it has been adopted in Eurocode 2 and, therefore, is the basis for Eurocode 8 as well - expressions parallel to the ones ahove are developed for Fy, for use in Eq(S) along with the value of @, derived from first principles and confinement option (c), including Eq.(8) forthe ultimate strain. For “conforming” beams, columns and walls with rectangular web, under cyclic loading (see Figure 9(a)): (MPa L pl CEB,cy, Lh = 0.06L, +0.035h + ay i » ay a9 ‘Alternative for “conforming” beams and columns with rectangular section (not for walls), under cyclic loading (see Figure 9()}: MPa LPO, (20) Lp cansyh = 05K ay 322 4-3: Pantie ‘Otani Symposium 2003 oe LL ete] * 7 7 WL j i» ree ee a ae @ eee ® Figure 9 Comparison of experimental ultimate chord rotations in cyclic tests to the predictions of Eq.(S) for confinement according to Eq.(8) (CEB/FIP MC90): (2) “conforming” beams, columns or walls with rectangular section, plastic hinge length per Eq.(19); (b) “conforming” beams and columns with rectangular section, plastic hinge length from Eq.(20). For “conforming” columns with circular section, under cyclic loading (see Figure 6()): S(MPa) .135L, +038D+a,. Lpceseey dy ey It proved to be almost impossible to find an expression for Lu providing good average ft to the results of cyclic tests on “non-conforming” columns and walls with rectangular section, when confinement option (©) is used for the calculation of g, The best expression found, Eq.(22) below, underestimates test results con average by 10% (see Figure 7(b)): Sf,(04Pay Upsceaoid = O2L, + ay dy @2) ‘Simirarly, no expression could be found for Ly that provides an acceptable fit to the monotonic test results, when calculation of ¢, is based on the CEB/FIP confinement model (option (c), Eq.) ‘The search exercise for expressions for L, that provide satisfactory average fit of Eq,(5) to the chord rotation at flexural failure, , has shown that such fitting is associated with large scatter. The scatter is reflected by the large coefficients of variation listed in Table 4. It is also evident from Figures 5 to 9, ‘which also show the line corresponding to the S%-fractle of the experimental value, given its prediction from Eq,(5). The larger the scatter, the lower is the 5%-fractile line. Figures 5 and 9 show also a tendency for overprediction of cyclic data in “conforming” members with rectangular section, while Figure 8 shows certain underprediction of high experimental values in monotonic loading. Circular columns are @ notable exception regarding scatter. For them Eq.(5), along with confinement options (d) or (e) and -Eqs.(12) and (15) or (8) and (21), respectively, provide acceptable scatter. ‘Another lesson leamed from this exercise is that the expressions for La in Eqs.(13)-(22) are not the only possible answer for each particular ease, Other linear combinations of dy, and/or h (or D) may provide ‘an almost equally good average fit to the data, albeit usually with lager seater. 'A third lesson is that, it is often feasible to achieve almost equivalent final fits with g-values calculated from very different confinement models (obviously using each time the appropriate expression for Ly), mo matter the agreement (or lack of it) between the calculated g, values andthe experimental values. $23 4-3 ard (rani Symposium 2003 . ; aa » ; a Lf ore » | ow) Figure 10 Comparison of experimental ultimate chord rotation of beams, columns or walls with ‘rectangular compression zone and “conforming” detailing, with the predictions of Eq.23) 4.3 Empirical formulations Except for circular columns, the predictive ability of Eq(S) - along with the corresponding expressions for ly - was found to be unsatisfactory. As an altemative, empirical expressions are sought for the chord rotation at flexural failure, @, of members other than circular columns, and are developed via methods of statistics. The earlier work by [Panagiotakos and Fardis, 2001], based on about two-thirds ofthe present data, provides the baseline. That work has pointed out the necessity of using the monotonic and the cyclic data together, as wel asthe primary variables on which 8, depends and a possible functional dependence. One of the main conclusions was that @, depends on whether loading to failure is monotonic or fully- reversed (cyclic), but is rather insensitive to the number of major deflection cycles preceding failure, ‘Two alternative - and almost equivalent - expressions are developed here forthe chord rotation at flexure- controlled failure, , of members wit rectangular cross-section (or rather rectangular compression zone) and “conforming” detailing, The fist, £q(23), is for the total ultimate chord rotation, @, while the other, q,(24), is for its plastic component, "= 0-8, withthe elastic component, 6, given by Eqs.(2) or (3). 24 43 Faris (Otani Symposiv 2003 Toad ta nen tn a va as: ot Rane a ey | 7 N [. % | Baro =| m8) Figure 11 Comparison of experimental ultimate chord rotation of beams, columns or walls with rectangular compression zone and “conforming” detailing, with the predictions of Eqs.(2), (3) and (24) N15, yaa fe) deast-oayiron,(ienor{enenielf"(f'ad"Flaem an 225, 037s (ap, Le) 00,0455 0-Ob af} (MBE }"[) AO ay ‘max0.01,0) A where: a, and a: coefficients forthe type of steel, equal to aa~ 0.0194 and cy ~ 0.015 for ductile hot-rolled or for heat-treated (tempcore) steel and to 2, 0.0125 and a”, = 0.0065 for cold-worked steel; ay: zero-one variable for type of loading, equal to 0 for monotonic loading and to 1 for cyclic loading: ‘aq: zer0-one Variable for slip, equal to | if there is slip of the longitudinal bars from their anchorage beyond the section of maximum moment, o to 0 if there is not (of. Eqs.(2}(4), (13}-(22)}; ‘wat! 2er0-one variable for walls, equal to 1 for shear walls and to 0 for beams or columns; ‘=Nbhg. (with b= width of compression zone, N = axial force. positive for compression); ©, a: mechanical reinforcement ratio, aff, of longitudinal reinforcement which isin tension (including 525 43: Fanci (Otani Symposiun 2003 ‘web reinforcement between the two flanges) or in compression, respectively; fe uniaxial (cylindrical) concrete strength (MPa) TEJfe=-MiVh: shear span ratio atthe section of maximum moment; ta/bu ratio of transverse ste] parallel tothe direction of loading; Sout yield stress of transverse steel, ‘& confinement effectiveness factor according to [Sheikh and Uzumeri, 1982], given by Eq (9); 4: steel rato of diagonal reinforcement in each diagonal direction. Figures 10 and 11 show the ft of Egs.(23) and (24) tothe data for members with “conforming detailing”, andthe lines corresponding to the 5% fractile of the predictions for the full set of experimental data (1159 tests) from which Eqs.23), (24) derive, The S%-fractile forthe plastic part of 4, alone, 6", is one-third of the predictions of Ea.(24). The statistics ofthe fitting listed in Table 4 suggest thet Eqs.(23) and (24) are practically equivalent in terms of predictive ability, with a slight overall edvantage of Eq.(23). Figure 12 compares the data for members with “non-conforming” detailing to the predictions of Eqs.(23) ‘and (24), with the confinement effectiveness factor a taken equal to zero, if the stirrups are not closed with 135® hooks, These specimens are too few to support an independent statistical analysis. Their qualitative difference from the “conforming” ones prevented their inclusion in the data set that yielded Eqs.(23), (24) (whereas they were included in the analysis that yielded Eqs. (2) and (3), as they were not considered to have fundamentally different pre-yield behaviour). According to Figure 12 and to the statistics listed in Table 4 for these specimens, the predictions of Eqs.(23), (24) should be divided by 1.2 ‘or by 115, respectively, to yield the expected ultimate chord rotation for “non-conforming” detailing. Taking into account the small size of the data set the seatter about the median is considered about the same as that ofthe eyclic data on “conforming” members. As a result, the lines corresponding to the SY fractile of the predictions forthe full set of experimental data for members with “conforming” detailing (1159 tess) are taken to apply also to the ones with “non-conforming” detailing, T oni nt no) ‘1 A! NY ; 1. i |_| | ee Hee be Coo ft A Paste! | wile! ones eae Fate? § 7 e i. asa eee Figure 12 Comparison of experimental ultimate chord rotation of columns or walls with rectangular ‘cross section and “non-conforming” detailing, with the predictions of (a) Eq.(23) (b) Eqs.(2), (3) a 526 4-3: Faris ‘tan syrposivn 2003 5. DRIFT (OR CHORD ROTATION) AT SHEAR FAILURE AFTER FLEXURAL YIELDING CConerete members that yield first in flexure, but their ultimate failure under cyclic loading clearly shows strong effects of shear, are considered to exhibit a “ductile shear” failure mode [Kowalsky and Priestley, +2000], as opposed to a “brittle shear” mode, in which failure takes place at relatively low deformations before flexural yielding. “Ductile shear” failure is controlled by diagonal tension and yielding of web reinforcement, rather than by web crushing. It has by now prevailed 10 quantify ths failure mode via a shear resistance Va, (as this is controlled by web reinforcement according to the well-established M&rsch truss analogy) that decreases with the (displacement) ductility ratio under cyclic loading (Ascheim and Moehle, 1992, Kowalsky and Priestley, 2000, Mochle et al, 2001]. As the number of available cyclic tests that led to “ductile shear” failure isnot sufficient to support development ofan independent (statistical or ‘mechanical) model for the deformation capacity of R/C members as affected or controlled by shear, the present work also adopts the solid base of the MOrsch analogy for shear, to describe in force terms & failure mode which is controlled by deformations. More specifically, the formulations proposed in [Kowalsky and Priestley, 2000] and in {Moehle et al, 2001] on the basis of limited number of tests on circular or rectangular columns, respectively, ae extended herein, using a much larger dataset of “ductile shear” failures of columns with both types of section and walls. ‘The outcome of the present effort is two models for Vas a function of the plastic chord rotation (or Aisplacement) ductility ratio, Ap, defined asthe ratio of the postelastic chord rotation at “ductile shear” failure, to the chord rotation at yielding, @, as this is computed from Eqs.(2)(4). (Inthe application ofthe models for predictive purposes, the post-tastic chord rotation at “ductile shear” failure will be obtained by subtracting from the total chord rotation the value of 8, from Eqs.(2)(4). In the development of the models, though, the experimental yield chord rotation was subtracted from the total experimental value, to climinate parasitic elastic flexibility effects observed in some tests). In both models the effect of axial force, N, on Vp is accounted for through a separate term, asin [Comité Eurointerational du Beton, 1993, Kowalsky and Priestley, 2000]. A 45° tru inclination is considered, as in (Moehle etal, 2001], because truss inclinations other than 45° are normally taken when only the web reinforcement is considered to contribute to Vp, (F,, term), without a separate concrete contribution (V. term). In the first model, Eq.(25), only the V, term decreases with #/s, as in [Kowalsky and Priestley, 2000], while in the second, Eq.(26), both the ¥. and the P,, terms are taken to decrease with 4#, as in [Mochle ot al, 2001): 095ml, fax 124 (I-an6mi( 44) 4, 05) ni 0354)+016 085s ine -n16ni(s 42] +] 26) min, 0.554, f,)}+0.16 where: zk: depth of cross-section equal tothe diameter D for circular sections) x: compression zone depth, "N: compressive axial force (postive, taken as zero for tension); ‘Ljhe=M/Vi: shear span ratio at member end; ‘Ac: eross-seotion are, taken equal to byd for cross-sections with rectangular web of width (thickness) dy, and structural depth dof to xD. 74 (where D. isthe diameter ofthe concrete core tothe inside Of the hoops) for circular sections; fe concrete sirength (MPa); ‘Pos: total longitudinal reinforcement ratio; Vai contribution of transverse reinforcement to shear resistance, taken equal to: 4) for cross-sections with rectangular web of width (thickness) dy: Ve= pebuye en where and f., are the ratio and the yield stress of transverse reinforcement and z isthe length ‘of the intemal lever arm (taken equal to ded’ in beams and columns, of 0 0.7Sh in walls), 527 3 Par (tani Symposium 2003 ‘b) for circular cross-sections: 4, V4, = 2A 5,(D=20) 08) Fe fp(D-20) es) where 4, is the cross-sectional area of a circular stirrup, », is the centerline spacing of stirrups and c the concrete cover to reinforcement. - "TI L ra SEC) story - a i = Ty = ce | Oe a |e one tan aaeRee ceetgee ese Saeed gee ee Figure Compan torial shang inch ar hh prediction of: (a) Eq. 25); (b) Eq.(26) hed rede etn @ w Figure 14 . Ratio of experimental to predicted shear strength for “ductile shear” failure, as a function of (total) chord rotation ductility ratio at failure, for predictions from: (a) Eq.(25); (b) Eq(26) Inthe fitting of Eqs.(25), (26), the experimental shear strength is taken equal tothe shear force at flexural yielding, and not to the exact value of lateral force resistance atthe instant “ductile shear” failure is taken ‘to occur (In the application of Eqs.(25), (26) for the prediction of “ductile shear” failure, their right-hand- side is set equal io the - theoretical - value of M/L,). This is also the experimental shear strength compared to the predictions of Eqs.(25), (26).in Figures 13 and 14 and near the bottom of Table 4 Eqs.(25), (26) are practically equivalent: Eq.(25) has slightly better overall statistics, while Eq. (26) gives better average agreement to the data for each one ofthe four types of members included inthe fitting. With very few exceptions, the experimental value of in the 208 tests which were considered 0s, controlled by shear and were used for the fitting of Eqs (25), (26) is less than the flexure-controlled ultimate chord rotation predicted from Eq523) ot (24), and, as a matter of fact, significantly less. To pursue the possibility of using Eqs.(25), 26) as a means to predict the shear flexure-contolled ultimate chord rotation, their right-hand-side was set equal tothe the theoretical value of M4y/L, and Eqs(25), (26) were solved for u/s. Figure 15, which compares the resulting values of i= 4f'g+1 to the experimental cones, clearly shows that, despite the relatively low seatter associated with them, Eqs.(25) and (26) do not 528 43: Fads ‘Otani Symposium 2003 lend themselves for a meaningful prediction of a shear-ontrolled deformation capacity. The reason for this failure is the low sensitivity of Vq to 42's. ol pra ee ae @) oo) Figure 15 Experimental ductility ratio, compared to value predicted by inverting: @) E425); 0) Ea.25) Itis reminded that “ductile shear” failure, described by Eqs.(25), (26) is considered to be associated with diagonal tension and with yielding of the web reinforcement. The large majority (27) of 33 walls in the database considered to have failed in shear after having yielded in flexure, did so by web crushing at a shear force normally lower than the predictions of Eqs.(25) or (26) and ata chord rotation much less than the value corresponding to flexure-controlled failure according to Eqs.(23), (24). tis noteworthy thatthe shear resistance of these 27 walls as well as that ofthe 10 cyclically loaded ones that experienced web ‘rushing before yielding in flexure, seems to be insensitive to the megnitude of the deformation at which ‘web crushing took place (see also Figure 16) and to follow the expression yy. 4 4 Fone 00{ 0085-87] (-o) hid @ Despite the low scatter associated with Eq.(29), the data behind it are not sufficient to support proposing ‘tas an upper limit forthe shear strength of walls under cyclic loading. Nonetheless, the low magnitude of its results, compared to those given by current code mules is disconcerting, The - normalized by bud ~ shear resistance ofthe few short colurnns in the database that fail by web crushing after flexural yielding is about double that obtained from Eq,(29), and decreases with iA, Those data, though, are even fewer. Figure 16 Ratio of experimental shear strength to prediction of Eq.(29) in shear walls (rectangular, T ‘of barbelled) failing by web crushing, as a function of (total) chord rotation ductility ratio at failure 529 3: Fale ‘Oxi Symposium 2003 6. ACKNOWLEDEMENTS: ‘The European Commission provides financiel support to this work under project SPEAR (Seismic Performance Assessment and Rehabilitation, contract No: G6RD-CT2001-00525) of its GROWTH programme. Prof. Otani participates in this project, within its component for international collaboration. ‘The contributions of Dr. T-B. Panagiotakos to the on-going development of the database, as well to earlier phases ofthis research, are gratefully acknowledged. 7. REFERENCES ASCE (2000), Prestandard for the seismic rehabilitation of buildings. American Society of Civil Engineers (FEMA Report 356), Reston, Va. ASCE, (2001), Seismic evaluation of existing buildings, ASCE draft Standard, American Society of Civil Engineers, Reston, Va. Ascheim, "M.A. and J.PMoehle(1992), "Shear Strength and Deformability of RC Bridge Columns Subjected to Inelastic Cyclic Displacements” University of California, Earthquake Engineering Research Center, Report UCB/EERC-92/04, Berkeley, CA Biskinis, D., G. Roupakias, and MN. Fardis(2002), "Stiffness and Cyclic Deformation Capacity of Circular Concrete Columns", in: Befestigungstechnik Bewehrungstechnik und ...Festschrift 21 Ehren von Prof. Dring. Rolf Eligehausen anlisslich seines 60. Geburtstages (W. Fuchs, H.-W. Reinhardt, eds.), Aktuelle Beitrage aus Forschung und Praxis, Ibidem-Verlag, Stuttgart, pp. 321-330. Comité Eurointernational du Beton(1993), CEB/FIP Model Code 1990. T. Telford (ed.), London. Comité Européen de Normalisation(2003a), Draft European Standard prEN1998-3:200x Eurocode 8: Design of structures for earthquake resistance. Part 3: Strengthening and repair of buildings. Revised final PT Draft (Stage 34) Doc. CEN/TC250/SC8/N37I. July 2003, Brusels Comité Européen de Normalisation(2003b), Draft European Standard prEN1998-1:200x Eurocode 8: Design of structures for earthquake resistance. Part 1: General rules, seismic actions and rules for buildings. Final Version (Stage 49), Doc. CEN/TC2SO/SCR/N335A, January 2003, Brusells Federation Internationale du Beton(2003a), Seismic Assessment and Retrofit of RC Buildings fib Bulletin 1No.24, Lausanne. Federation Intemationale du Beton(2003b), Displacement-based Design of RC Buildings Chapter 5: “Displacement Capacity of Members and Systems”, fib Bulletin No.25, Lausanne Kowalsky, M. and M.J.N. Priestley(2000), "Improved Analytical Model for Shear Strength of Circular Reinforced Concrete Columns in Seismic Regions” Siructural Journal, American Concrete Institute, Vol. 97, No. 3, pp.388-396. Mander, J.B. M.J.N, Priestley, R. Park(1988), "Theoretical Stress-strain Model for Confined Concrete”, Journal of Structural Engineering, American Society of Civil Engineers, V.114, No.8, pp-1827-1849. Moehle J., A. Lynn, K. Elwood, H, Sezen(2001), "Gravity Load Collapse of Building Frames during Earthquakes" 2nd US-Japan Workshop on Performance-based Design Methodology for Reinforced Conerete Building Structures, Pacifie Earthquake Engineering Research Center, Richmond, CA. Panagiotakos T. and M.N. Fardis (2001), "Deformation of R.C. Members at yielding and ultimate”, Structural Journal, American Concrete Institute, Vol. 98, No. 2, pp. 135+ Paulay, T. and MLN Priestley(1992), Seismic Design of Reinforced Concrete and Masonry Buildings, J. Wiley, New York, N.Y. Sheikh, S.A. and S.M, Uzumeri(1982), "Analytical Model for Concrete Confinement in Tied Columns", Journal of Structural Division, American Society of Civil Engineers V.108, No. ST12, pp.2703-2722. 530

S-ar putea să vă placă și