Seismic strengthening of an under-designed RC structure with FRP

EARTHQUAKE ENGINEERING AND STRUCTURAL DYNAMICS Earthquake Engng Struct. Dyn. 2008; 37:141–162 Published online 24 August 2007 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/eqe.749 Seismic strengthening of an under-designed RC structure with FRP M. Di Ludovico∗, † , A. Prota, G. Manfredi and E. Cosenza Department of Structural Engineering, University of Naples Federico II, 80125 Naples, Italy SUMMARY The opportunities provided by the use of fiber-reinforced polymer (FRP) for the seismic retrofit of existing reinforced concrete (RC) structures were assessed on a full-scale three-story framed structure. The structure, designed only for gravity loads, was subjected to a bi-directional pseudo-dynamic (PsD) test at peak ground acceleration (PGA) equal to 0.20g at the ELSA Laboratory of the Joint Research Centre. The seismic deficiencies exhibited by the structure after the test were confirmed by post-test assessment of structural seismic capacity performed by nonlinear static pushover analysis implemented on the lumped plasticity model of the structure. In order to allow the structure to withstand 0.30g PGA seismic actions, a retrofit using glass fiber-reinforced polymer (GFRP) laminates was designed. The retrofit design was targeted to achieve a more ductile and energy dissipating global performance of the structure by increasing the ductility of columns and preventing brittle failure modes. Design assumptions and criteria along with nonlinear static pushover analysis to assess the overall capacity of the FRP-retrofitted structure are presented and discussed. After the retrofit execution, a new series of PsD tests at both 0.20g and 0.30g PGA level were carried out. Theoretical predictions are compared with the main experimental outcomes to assess the effectiveness of the proposed retrofit technique and validate the adopted design procedures. Copyright 2007 John Wiley & Sons, Ltd. Received 13 July 2006; Revised 16 June 2007; Accepted 4 July 2007 KEY WORDS: GFRP; full scale; RC; seismic retrofit; biaxial bending; nonlinear pushover analysis 1. INTRODUCTION The main hazard in southern European countries consists in the number of existing reinforced concrete (RC) structures which are under-designed or designed under outdated regulations or construction practice. Casualties and losses are mainly due to deficient RC buildings not suitably ∗ Correspondence to: M. Di Ludovico, Department of Structural Engineering, University of Naples Federico II, 80125 Naples, Italy. † E-mail: [email protected] Contract/grant sponsor: Italian Department of Civil Protection; contract/grant number: 2005-2008 Copyright 2007 John Wiley & Sons, Ltd. 142 M. DI LUDOVICO ET AL. designed for earthquake resistance. In the framework of the SPEAR (Seismic PErformance Assessment and Rehabilitation) research project, specifically targeted to evaluate current assessment and rehabilitation methods and at development of new assessment and retrofitting techniques, a series of full-scale bi-directional pseudo-dynamic (PsD) tests on a torsionally unbalanced threestory RC framed structure was carried out. The SPEAR structure represents a typical building in most earthquake-prone areas of Europe; thus, it is characterized by plan-irregularity, poor local detailing, scarcity of reinforcement, insufficient confinement and weak joints combined with older construction practice. The full-scale RC structure was subjected to a bi-directional PsD test in the ELSA laboratory of the Joint Research Centre (JRC) in Ispra (Italy) under the Montenegro Herceg Novi record scaled to a PGA of 0.20g. Subsequently, a post-test lumped plasticity model of the structure was implemented to assess the theoretical seismic capacity of the structure. Since both theoretical and experimental results showed that the ‘as-built’ structure was unable to withstand a larger seismic action, a retrofit intervention by using FRP laminates was designed. Once the design of the glass fiber-reinforced polymer (GFRP) retrofit was provided, the structure was subjected to a new series of two tests with the same input accelerogram selected for the ‘as-built’ specimen but scaled to a peak ground acceleration (PGA) value of 0.20g and 0.30g, respectively. The opportunity of using composite materials as an effective technique for the seismic retrofit of RC frames is herein evaluated. The background, philosophy and calculation procedures followed to carry out the design of the GFRP retrofit are presented along with the comparison between the experimental and theoretical performance of the ‘as-built’ and retrofitted structure. 2. STRUCTURAL GEOMETRY, MATERIAL PROPERTIES AND TEST SETUP The structure is regular in elevation with a story height of 3 m and 2.5 m clear height of columns between the beams; it is nonsymmetric in both directions, with 2-bay frames spanning from 3 to 6 m. The plan layout and the 3D view of the structure after the construction are shown in Figure 1. The concrete floor slabs are 150 mm thick, with a bi-directional mesh of 8 mm smooth steel rebars, spaced at 200 mm in the short span, 400 mm in the long span and 100 mm into the short span of the cantilever. Beam cross-sections are 250 mm wide and 500 mm deep. Eight out of the nine columns have a square 250 × 250 mm cross-section; the ninth (column C6) has a rectangular cross-section of 250 × 750 mm, which makes it much stiffer and stronger than the others along direction Y (i.e. the strong direction for the whole structure). The joints of the structure are one of its weakest points: neither beam nor column stirrups continue into them, so that no confinement at all is provided. Moreover, some of the beams directly intersect with other beams (see joints close to columns C3 and C4 in Figure 1) resulting in beam-to-beam joints without the support of the column. Details about the beam reinforcement for flexure and shear can be found in Negro et al. [1]. The materials used for the structure were characterized by experimental tests: the average strength of smooth steel bars was equal to f ym = 320 MPa [2]; tests performed on samples extracted during concrete casting of each floor showed an average concrete compressive strength of f cm = 25.5 MPa. Two types of FRP laminates were used: (1) uniaxial GFRP laminates at both ends of each square column (unit weight of 900 g/m2 , thickness of dry fibers of 0.48 mm/ply, modulus of elasticity of 65.7 GPa, tensile strength of 1314 MPa and ultimate strain of 0.02); (2) quadriaxial GFRP laminates for exterior beam–column joints along with the large column C6 (wrapped for its entire height) at Copyright 2007 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:141–162 DOI: 10.1002/eqe SEISMIC STRENGTHENING OF AN UNDER-DESIGNED RC STRUCTURE 5m 3m C5 C1 1m 143 0.70 m C2 B1 B2 6m 5.5 m B11 B9 B7 C9 C3 5m B12 B4 B10 C4 B8 4m B3 Y B6 B5 C8 X C6 C7 (a) (b) Figure 1. (a) Plan view and (b) 3D view of the SPEAR structure. Figure 2. Location and direction of actuators. all stories (unit weight of 1140 g/m2 , thickness of dry fibers of 0.1096 mm/ply direction, modulus of elasticity of 65.7 GPa, tensile strength of 986 MPa and ultimate strain of 0.015). It is noted that GFRP laminate properties were provided by experimental tests. Copyright 2007 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:141–162 DOI: 10.1002/eqe 144 M. DI LUDOVICO ET AL. A bi-directional PsD technique was used both in the ‘as-built’ and in the FRP-retrofitted full-scale structure. The bi-directional PsD test consisted in the simultaneous application of the longitudinal and the transverse earthquake components to the structure; a more detailed description of both the method and the mathematical approach can be found in Molina et al. [3, 4]. Four actuators per story with four associated control displacement transducers were connected to the structure; a plan view showing the positions of the actuators is depicted in Figure 2. Details about records, measurements and instrumentation can be found in Negro et al. [1]. 3. EXPERIMENTAL BEHAVIOR OF THE ‘AS-BUILT’ STRUCTURE: 0.20g PGA LEVEL Identifying the most appropriate ground motions to be used for experimental tests is not an easy task; a procedure specifically targeted at such an objective has been recently reported in [5]. In the case of the full-scale SPEAR structure, accelerograms obtained from the Montenegro 1979 Herceg Novi ground motion record were used as the input signal for the PsD tests. After extensive analytical study (see Jeong and Elnashai [6]), this record was selected among many different earthquake scenarios because: (1) the analysis of inter-story drift time histories (that provides more accurate results than static pushover analysis in the case of irregular buildings) showed that it could induce no pronounced peaks in terms of inter-story drifts especially in the earlier part of the response, thus allowing the collection of considerable experimental results before subjecting the structure to the maximum demand; (2) this record is Eurocode 8 [7] spectrum compatible. As the retrofit phase was intended to consist of a ‘light’ intervention, the appropriate intensity of PGA was chosen in order to obtain a level of damage in the first round of tests significant but not so severe as to be beyond repair. Thus, it was decided to run the test in the ‘as-built’ configuration Table I. Experimental outcomes. X -direction Y -direction Copyright Test Total absorbed energy (kJ) Max base shear (kN) Max top displ. (m) ‘As-built’ 0.20g 44.00 195 0.1057 FRP retrofit 0.20g 42.20 211 0.1088 FRP retrofit 0.30g 83.36 196 0.2053 ‘As-built’ 0.20g 65.00 276 0.1031 FRP retrofit 0.20g 68.66 287 0.1125 FRP retrofit 0.30g 104.38 281 0.1266 2007 John Wiley & Sons, Ltd. Level Max I –S displ. (m) 1 2 3 1 2 3 1 2 3 0.0246 0.0570 0.0358 0.0320 0.0554 0.0343 0.0594 0.1060 0.0635 1 2 3 1 2 3 1 2 3 0.0306 0.0472 0.0326 0.0397 0.0476 0.0311 0.0423 0.0559 0.0507 Earthquake Engng Struct. Dyn. 2008; 37:141–162 DOI: 10.1002/eqe SEISMIC STRENGTHENING OF AN UNDER-DESIGNED RC STRUCTURE 145 with a scaled PGA level of 0.20g. The results of the first test showed that the major damage concerned the ends of the square columns with crushing of concrete at all stories. The level of damage was more significant at the second story. For each floor, the most damaged members were the columns, where torsional effects produced inclined cracks on the compressive sides. During tests, significant cracks opened on the tensile side of the columns at the beam–column interface. The damage on the rectangular column C6 was less significant even though crushing of concrete and cracks at the interface with beams were observed (see Negro et al. [1]). The experimental outcomes in terms of total absorbed energy, maximum base shear and top displacement along with the maximum inter-story displacement for directions X and Y are summarized in Table I: the maximum base shear was reached along direction Y (276 kN) rather than X (195 kN). This was consistent with the arrangement of the wall-type column C6 placed with its strong axis in direction Y . In contrast, much larger top displacements were reached in direction X (0.1057 m) rather than Y where a maximum top displacement of 0.1031 m was achieved. On the basis of the damages detected on the structure, Table I shows that the maximum inter-story drifts were reached at the second story (0.0570 m in X and 0.0472 m in Y ). 4. POST-TEST ASSESSMENT OF THE ‘AS-BUILT’ STRUCTURE Numerical analysis is performed with the aim of reproducing a typical design process that can be adopted by a structural engineer to assess an existing building. The purpose was mainly to use a typical rehabilitation design methodology and verify its outcome by a qualitative comparison with the experiment; the analysis was not aimed at verifying the analytical model against the experimental results. Thus, a finite element analysis program, SAP2000 [8], very commonly used by structural engineering practitioners, was utilized to run the numerical analyses. An assessment procedure based on a pushover analysis was adopted; indeed, this method was considered more appropriate to a practitioner’s approach. 4.1. Lumped plasticity model of the structure A post-test assessment of structural global capacity was performed by nonlinear static pushover analysis on the ‘as-built’ structure. Pushover analyses in the longitudinal and transverse directions were performed by subjecting the structure to a monotonically increasing pattern of lateral forces proportional to the 1st and 2nd modes of vibration (in directions X and Y , respectively) and mass distribution. Lateral loads were applied at the location of the center of masses in the model. Center of mass at each story, mass values, modal displacements of each center of mass in directions X and Y , along with the corresponding normalized lateral loads, are summarized in Table II. In the analytical model slabs were omitted and their contribution to beam stiffness and strength was considered, assuming a T cross-section for the beams with the effective flange width equal to the rectangular beam width (250 mm) plus 7% of the clear span of the beam on either side of the web [9]. This assumption provides flange width values between the conservative flange width indicated in Eurocode 8 [7] for design purposes and the width recommended for gravity load design. Moreover, to take into account the effect of the slabs, a rigid diaphragm was assumed at each story of the model. For a comprehensive study of the seismic response of existing RC buildings, shear failure of members should be taken into consideration; however, in the present case it was not considered because shear demand was significantly lower than both beam and Copyright 2007 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:141–162 DOI: 10.1002/eqe 146 M. DI LUDOVICO ET AL. Table II. Geometrical characteristics, mass times modal displacement and normalized lateral loads for directions X and Y . Story 1 2 3 Center of mass∗ (x; y; z) (m) Masses (ton) Mass × modal displacement (1st mode) direction X (ton m) (4.55; 5.30; 2.75) (4.55; 5.30; 5.75) (4.58; 5.34; 8.75) 65.86 65.86 63.28 0.669 1.460 1.847 Normalized lateral loads direction X (kN) Mass × modal displacement (2nd mode) direction Y (ton m) Normalized lateral loads direction Y (kN) 0.362 0.790 1 0.533 1.325 1.794 0.297 0.738 1 Elastic period T (s) = 0.62 ∗ Coordinate referred to the coordinate system of Figure 1. column shear capacities (as also reported by Jeong and Elnashai [6]). Therefore, only the inelastic flexural behavior of elements was considered by modeling the structural members with lumped plasticity at both ends; a bilinear moment–rotation relationship was used for each plastic hinge. The moment–rotation relationship was obtained based on moment curvature analysis carried out considering section properties and constant axial loads on the elements (axial loads on beams, due to gravity loads, were assumed equal to zero); as a design hypothesis, a parabolic–rectangular stress–strain diagram was assumed for concrete and elasto perfectly plastic for steel. Yielding curvature, y and moment My , corresponded to the attainment of the tensile steel yielding strain; the ultimate curvature, u , and ultimate moment, Mu , corresponded to the attainment of ultimate strains in concrete or steel (conventionally assumed equal to 3.52 for concrete and 402 for steel, respectively). Yielding and ultimate rotation, y and u , as well as plastic hinge length, L pl. , were computed according to Eurocode 8 [10] type equations dbL f y (1) y = flex. y L V + shear + slip √ fc    0.5L pl. u =  y + (u − y )L pl. 1 − (2) LV L pl. = flex. L V + shear h + slip dbL f y (3) where L V is the shear span, dbL is the diameter of longitudinal bars, f y and f c are the average steel and concrete strength, respectively, and h is the cross-section depth; factors flex. , shear , slip along with flex. , shear , slip and  were provided by the latest seismic guideline developed by the Italian Department of Civil Protection, Ordinance 3431 [11] flex. = 0.1 where el. Copyright flex. = 31   h 1 shear = 0.17, shear = 0.0013 1 + 1.5 , = LV el. 0.24 slip = 0.13y slip = √ fc is a coefficient equal to 1.5 or 1 for primary or secondary members, respectively. 2007 John Wiley & Sons, Ltd. (4) Earthquake Engng Struct. Dyn. 2008; 37:141–162 DOI: 10.1002/eqe 147 SEISMIC STRENGTHENING OF AN UNDER-DESIGNED RC STRUCTURE NX-NY PX-PY 350 Base shear [KN] 280 C C 210 C NY 140 C PX NX C CM PY_AS-BUILT C NY_AS-BUILT 70 PY PX_AS-BUILT NX_AS-BUILT C -0.15 LSSD C C -0.10 -0.05 0 0.00 0.05 0.10 0.15 Top displacement [m] Figure 3. Pushover curves for the assessment of the ‘as-built’ structure capacity. The simplified assumption of shear span L V = 0.5L during the horizontal loading process was adopted in modeling the structure [12]. Given that the original detailed construction drawings were known and the comprehensive material testing was performed, a knowledge level equal to 3, KL3, was assumed (according to the Ordinance 3431 [11]) corresponding to a confidence factor (i.e. CF) of 1. As a consequence of this knowledge level, the average strength values for materials were assumed in the analysis. 4.2. Theoretical capacity vs demand The significant damage limit state (LSSD), which corresponds, according to Ordinance 3431 [11], to attainment of the 0.75u in one of the plastic hinges, was investigated to assess the structural capacity. On the basis of such a limit state, pushover analyses on the ‘as-built’ structure were performed in the longitudinal direction (positive and negative X directions, named PX and NX , respectively) and in transverse direction (positive and negative Y directions, named PY and NY , respectively) (see Figure 3). The theoretical results in terms of maximum base shear, Fmax , top displacement, dmax , and absolute inter-story displacement are summarized in Table III. Seismic demand was computed with reference to the Ordinance 3431 [11] design spectrum (soil type c, 5% damping) which provides a pseudo-acceleration spectrum compatible with that obtained by the experimental ground motion record, Montenegro Herceg Novi (see Figure 4). Although the ‘as-built’ structure was tested under a maximum PGA level of 0.20g, theoretical analysis was also performed for a seismic level of 0.30g to evaluate the theoretical structural performance under a larger seismic action intensity. Seismic demand was calculated by combining the pushover analysis of an equivalent multi-degree-of-freedom (MDOF) model with the response spectrum of an equivalent single-degree-of-freedom (SDOF) system. The results in terms of maximum top Copyright 2007 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:141–162 DOI: 10.1002/eqe 148 Copyright ‘As-built’ structure Capacity FRP-retrofitted structure Demand Capacity Demand Push Limit 0.20g 0.30g 0.20g 0.30g direction state Level Fmax (kN) dmax (m) IS displ. (m) dmax (m) dmax (m) Fmax (kN) dmax (m) IS displ. (m) dmax (m) dmax (m) Earthquake Engng Struct. Dyn. 2008; 37:141–162 DOI: 10.1002/eqe PX LSSD 1 2 3 232 0.0690 0.0124 0.0505 0.0043 0.0623 0.0934 235 0.1182 0.0164 0.0973 0.0044 0.0626 0.0939 NX LSSD 1 2 3 232 0.0626 −0.0093 −0.0485 −0.0040 0.0618 0.0927 235 0.1076 −0.0100 −0.0935 −0.0041 0.0618 0.0927 PY LSSD 1 2 3 251 0.0962 0.0287 0.0344 0.0326 0.0607 0.0910 253 0.1201 0.0364 0.0428 0.0409 0.0610 0.0917 NY LSSD 1 2 3 292 0.0740 −0.0284 −0.0323 −0.0125 0.0603 0.0904 294 0.0908 −0.0365 −0.0411 −0.0131 0.0604 0.0906 M. DI LUDOVICO ET AL. 2007 John Wiley & Sons, Ltd. Table III. Summary of the results in terms of capacity and demand for the ‘as-built’ and the FRP-retrofitted structure. 149 SEISMIC STRENGTHENING OF AN UNDER-DESIGNED RC STRUCTURE Montenegro 1979 Herceg Novi Ground Acceleration Y Direction 1g PGA 1 0.8 0.6 0.4 0.2 0 -0.2 -0.4 -0.6 -0.8 -1 Ground Acceleration [g] Ground Acceleration [g] Montenegro 1979 Herceg Novi Ground Acceleration X Direction 1g PGA 0 1 2 3 4 5 6 (a) 0.8 0.6 0.4 0.2 0 -0.2 -0.4 -0.6 -0.8 -1 0 7 8 9 10 11 12 13 14 15 Time [s] 1 Acceleration response spectra 3.50 ag = 1g 3.0 Sa [g] Sa [T]/ag 2.0 1.5 0.5 0.50 0.0 (c) 2.5 T [sec.] 3 3.5 7 8 9 10 11 12 13 14 15 Time [s] 4 4.5 a g =1g 1.50 1.00 2 6 2.00 1.0 1.5 5 2.50 Herceg y 1 4 3.00 Herceg x 2.5 0.5 3 Spectra in Acceleration-Displacement format 3.5 0 2 (b) 5 ag = 0.30g ag = 0.20g 0.00 0.00 (d) 0.10 0.20 0.30 0.40 Sd [m] Figure 4. (a) Longitudinal; (b) transverse component of Herceg Novi records, PGA 1g; (c) acceleration response spectra (5% damping) of X and Y components and Ordinance soil c spectrum; and (d) spectra in AD format. displacement required for each investigated PGA level are summarized in Table III, which shows that the ‘as-built’ structure is able to satisfy the LSSD in each direction for 0.20g PGA level even if, especially in both positive and negative X directions, the capacity slightly exceeds the demand. Moreover, increasing the seismic action up to a PGA of 0.30g, such verification is satisfied only in direction PY ; at this PGA level, the maximum gap in terms of maximum top displacement is provided in direction NX where the difference between seismic demand and displacement capacity is 0.0301 m (0.0927 m vs 0.0626 m) corresponding to a performance gap of 48%. The capacity spectrum approach (CSA) was also used for the seismic verification [13]. Thus, both the elastic acceleration and displacement spectrum were scaled at PGA levels of 0.20g and 0.30g and plotted in acceleration–displacement (AD) format (see Figure 4). In Figure 5, the seismic demand for the equivalent SDOF system is determined for the two levels of ground motion analyzed by using a CSA [13]; demand is computed for direction NX , where the maximum capacity–demand gap was recorded. By using the same graph to plot the demand spectra and capacity, it is possible to determine the elastic acceleration and the corresponding elastic displacement demand (named Sae and Sde , respectively) required in the case of elastic behavior. They are computed by intersecting the radial line corresponding to the elastic period of the idealized bilinear system, T ∗ , with the elastic demand spectrum. Once the ductility demand,  = Sd /Dy∗ , is computed (depending on whether T ∗ is greater or less than TC ), the inelastic demand in terms of accelerations and displacements is provided by the intersection point of the capacity diagram with the demand spectrum corresponding to . Figure 5 highlights that the ‘as-built’ structure in direction NX , hardly able to satisfy the Copyright 2007 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:141–162 DOI: 10.1002/eqe 150 M. DI LUDOVICO ET AL. As Built Structure (Push NX) 1.00 ag = 0.30g 0.20g Elastic Demand 0.90 As-Built Capacity Diagram 0.20g Inelastic Demand 0.80 0.30g Elastic Demand 0.30g Inelastic Demand T* = 0646 s S ae 0.70 ag = 0.20g Sa [g] 0.60 0.50 µ = 5.2 0.40 0.30 µ = 3.5 0.20 S ye 0.10 0.00 0.00 PERFORMANCE GAP (48%) * Dy 0.05 Sd = S de 0.10 Sd [m] 0.15 0.20 Figure 5. ‘As-built’ structure elastic and inelastic demand spectra vs capacity diagram. demand due to the 0.20g PGA level, totally lacks the appropriate capacity to resist the 0.30g PGA level. Indeed, the requested ductility is  = 5.2 against the available structural ductility of  = 3.5. The displacement demands in Figure 5 refer to the equivalent SDOF system; thus, to obtain the displacement demands of the MDOF system (reported in Table it is necessary to multiply the  III),  SDOF system demand by the transformation factor  = m i i / m i i2 = 1.23 (where m i is the mass in the ith story and i are the normalized displacements). The results of theoretical analysis closely approximated those of the experiment, indicating the first attainment of the significant damage limit state (i.e. 0.75u in the plastic hinge) at the column ends of the second floor (i.e. at columns C3 and C4 in directions PX and NX , respectively) where the most significant damage was found during the test. Moreover, according to the damage detected on the structure after the test, it provided 0.20g as a limit acceleration value for the verification of the LSSD. 5. DESIGN OF THE REHABILITATION WITH COMPOSITES The selection of fiber texture and retrofit design criteria were based on deficiencies underlined by both the test on the ‘as-built’ structure and the theoretical results provided by the post-test assessment. They indicated that a retrofit intervention was necessary in order to increase the structural seismic capacity; in particular, the theoretical results showed that the target design PGA level of 0.30g could have been sustained by the structure if its displacement capacity were increased by a factor of 48%. In order to pursue this objective, the retrofit design strategy focused on two main aspects: (1) increasing the global deformation capacity of the structure and thus its dissipating global performance and (2) fully exploiting the increased deformation capacity by avoiding brittle collapse modes. Thus, the retrofit design was aimed at maximizing the benefits of the externally bonded FRP reinforcement along the direction of dominant stresses by increasing either the column confinement or the shear capacity of exterior beam–column joints and of the wall-type column, C6. The design Copyright 2007 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:141–162 DOI: 10.1002/eqe SEISMIC STRENGTHENING OF AN UNDER-DESIGNED RC STRUCTURE 151 principles of the rehabilitation strategy are outlined in the following sections with reference to two main issues: (1) design of column confinement; (2) design of exterior beam–column joints and wall-type column shear strengthening. 5.1. Confinement of columns Both the experimental activity and theoretical assessment of the ‘as-built’ structure showed that the columns’ cross-sectional dimensions and the amount of longitudinal steel reinforcement were inadequate to satisfy the demand generated by the biaxial bending associated with the axial load; the weak column–strong beam condition led to the formation of plastic hinges in the columns. With a view to a ‘light’ strengthening intervention, it was decided to increase the ductility of the plastic hinges at column ends without changing their position rather than establishing a correct hierarchy of strength by relocating them. Indeed, in this latter case, flexural strengthening of the columns with proper anchorage at their ends would have been necessary. This objective was pursued by GFRP columns’ confinement, which allows the ultimate concrete compressive strain to be enhanced. This corresponds to an increase in the curvature ductility that, assuming a plastic hinge length not significantly affected by the retrofit intervention, leads to a proportional increase in the plastic hinge rotation capacity. In order to compute the axial strain of the FRP-confined member, the equation provided by the latest guideline developed by the Italian National Research Council, CNR-DT 200/2004 [14] was used  f l,eff ccu = 0.0035 + (5) f cd where the ultimate axial strain for FRP-confined concrete, ccu , is computed as a function of design compressive concrete strength, f cd , and the effective lateral confining pressure, f l,eff ( f l,eff = keff f l , where keff is the coefficient of effectiveness depending on the cross-section shape and FRP configurations, and f l is the confining lateral pressure depending on the geometric strengthening ratio, f = 2tf · (b + d)/b · d (tf is the FRP thickness, b and d are cross-section dimensions), the FRP modulus of elasticity and design strain). The equations to compute the coefficient of effectiveness, keff , and the confining lateral pressure, f l , are given in CNR-DT 200/2004 [14]. Considering that calculations are referring to an existing structure, the design compressive concrete strength was assumed as the average compressive concrete strength obtained by field tests, f cm = 25.5 MPa. To quantify the amount of FRP to be installed, the central column, C3, was selected for calculations since it carries the maximum axial force due to the gravity loads (P = 409 kN at first story); thus, it has the minimum rotational capacity. In Table IV, theoretical results in terms of concrete ultimate axial strain provided by Equation (5), along with the ultimate curvature (calculated based on section analysis), are reported for one, two and three plies of uniaxial GFRP or CFRP confinement (with unit weight of 900 and 300 g/m2 and thickness of 0.48 and 0.166 mm/ply, respectively). The last two columns summarize increases in the ultimate rotation and the percentage rotation increase with respect to the original configuration, abs . The ultimate rotation values were computed with reference to Equation (3). On the right-hand side of Figure 6, the moment–curvature relationship is plotted for the original C3 column cross-section (continuous line) under the axial load acting at first story (P = 409 kN, due to only the gravity loads); the dashed line shows how the moment–curvature relationship changes as one ply at a time of GFRP confinement is added. The same graph is plotted in the Copyright 2007 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:141–162 DOI: 10.1002/eqe 152 Copyright Earthquake Engng Struct. Dyn. 2008; 37:141–162 DOI: 10.1002/eqe FRP type FRP thickness tf (mm) Original 1 GFRP 2 GFRP 3 GFRP 1 CFRP 2 CFRP 3 CFRP — 0.480 0.960 1.440 0.166 0.332 0.498 ply plies plies ply plies plies FRP volumetric ratio frp = 2tf (b + d)/bd — 0.00768 0.01536 0.02304 0.00266 0.00531 0.00797 Ultimate strain ccu (2) Neutral axis depth xc (mm) 3.50 7.30 8.87 10.08 7.12 8.62 9.77 80.9 72.07 70.81 70.12 72.26 70.98 70.28 Ultimate curvature u (rad/mm × 105 ) 4.325 10.129 12.527 14.376 9.854 12.145 13.902 Ultimate rotation u (rad) Ultimate rotation absolute increase abs. (%) 0.0125 0.0248 0.0298 0.0337 0.0242 0.0290 0.0327 0 98 138 169 93 131 161 M. DI LUDOVICO ET AL. 2007 John Wiley & Sons, Ltd. Table IV. Influence of GFRP and CFRP confinements on concrete ultimate axial strain, ultimate curvature and ultimate rotation. 153 SEISMIC STRENGTHENING OF AN UNDER-DESIGNED RC STRUCTURE 60 ORIGINAL ORIGINAL 50 Moment (kNm) 40 3 PLIES2 PLIES 1PLY 1 PLY 2 PLIES 3 PLIES 30 CFRP UNI-AX300 g/m 2 GFRP UNI-AX900 g/m2 20 10 0 15 10 -10 5 -5 0 5 10 15 Curvature (rad/mmx105) Figure 6. Moment–curvature for original, GFRP and CFRP upgraded C3 column cross-section. left-hand side of the diagram with respect to CFRP confinement. Figure 6 shows that both GFRP and CFRP confinements cause a negligible increase in the cross-section ultimate moment (from a value of Mu = 51.14 kN m in the original configuration to Mu = 51.48 kN m in the retrofitted one, whether for GFRP or CFRP confinement). In contrast, theoretical calculations clearly show that, with reference to the glass and carbon fibers selected, the curvature increase and the related ultimate rotation increase (see Table IV) are very significant but are not substantially affected by the two different kinds of laminates. Once it had been established that both materials were able to increase almost equally the ultimate concrete axial strain, hence both the ultimate curvature and ultimate rotation, given that in the case of interior application in buildings, durability performance is not the driving design criterion, the choice of the fibers to be utilized was essentially governed by costs. Comparing the application costs per m2 , it was calculated that by using uniaxial glass fibers with unit weight of 900 g/m2 , instead of uniaxial carbon fibers with unit weight of 300 g/m2 , the costs were reduced by about 30%. This was the reason for selecting glass laminates. By using GFRP laminates, the ultimate rotation increase goes from 98% for one GFRP ply installed and becomes about 138 and 169% for two and three GFRP plies, respectively (see Table IV). Since the design goal was to allow the structure to withstand a 0.3g PGA level and given that the theoretical analysis indicates that a 48% structural deformation capacity increase was necessary to pursue this objective, it was estimated that an increase in the local rotation capacity of the plastic hinge at least twice that of the original member could have been necessary. Importantly, the local increase in the rotation capacity is not proportional to the increase in the global deformation capacity. Thus, based on such considerations, the first trial in the design of the GFRP confinement was chosen as two plies of laminates with a unit weight of 900 g/m2 applied to all the square columns and extended by a length greater than that of the effective plastic hinge (about 380 mm) Copyright 2007 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:141–162 DOI: 10.1002/eqe 154 M. DI LUDOVICO ET AL. computed by Equations (3) and (4). Furthermore, in order to validate the design choice, a nonlinear static pushover on the FRP-retrofitted structure was provided at the end of the design process by using the SAP2000 [8] analysis program. 5.2. Design of shear strengthening: beam–column joints and wall-type column To avoid the attainment of shear strength of exterior joints caused by increasing the ductility of columns, which is brittle and could be detrimental to the global performance, further FRP was designed on beam–column joints corresponding to the corner square columns C2, C5, C7 and C8. The original shear strength of the exterior joints was computed by using equations provided by Ordinance 3431 [13]. Using this seismic guideline, the principal tensile stress of an exterior joint, nt , may be determined by using the following equation:     2  2   N Vn  N  0.3 f c − + (6) nt =  2Ag Ag   2Ag where N is the axial force in the upper column, Ag is the horizontal joint area, Vn is the acting shear on the joint due to the contributions of both shear force on the upper column and tensile reinforcement on the beam and f c is the compressive concrete strength. By assuming nt equal to √ 0.3 f c , it was possible to compute, for each exterior joint of the structure, the horizontal ultimate shear force and the corresponding shear strength, 0,max (Vn /Ag ), under which tensile joint failure is achieved. Theoretical results, in terms of original joint shear strength, 0,max , with reference to the external joints at each story, along with the axial force due to only gravity loads, are summarized in Table V. Since theoretical simulations of the first round of tests predicted shear stresses on the exterior joints comparable with those reported in Table V (e.g. 1.87 and 2.01 MPa vs 1.82 and Table V. Shear strength of the un-strengthened and GFRP-retrofitted corner joints. GFRP-retrofitted joint shear strength max (MPa) Exterior joint column Axial force, N (N) Original joint shear strength 0,max (MPa) 1 ply 2 plies 3 plies 1st story C5 C8 C2 C7 59 100 44 280 154 090 91 520 1.92 1.82 2.44 2.11 3.40 3.26 3.67 3.53 4.46 4.48 4.81 4.72 5.34 5.47 5.72 5.43 2nd story C5 C8 C2 C7 28 010 20 060 72 740 43 360 1.71 1.65 2.00 1.81 3.25 3.16 3.41 3.39 4.43 4.39 4.56 4.52 5.27 5.37 5.38 5.44 3rd story C5 C8 C2 C7 23 590 15 650 68 320 38 940 1.68 1.62 1.97 1.78 3.23 3.14 3.42 3.37 4.41 4.37 4.60 4.50 5.26 5.35 5.44 5.42 Floor Copyright 2007 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:141–162 DOI: 10.1002/eqe 155 SEISMIC STRENGTHENING OF AN UNDER-DESIGNED RC STRUCTURE 50 Inclination of principal tensile stress (˚) GFRP QUADRI-AX 1140 g/m2 40 30 Original Shear Strength νo,max = 1,62 MPa 20 1 Ply 2 Plies 3 Plies 10 0 0 1 2 3 Shear Stress ν (MPa) 4 5 6 Figure 7. Principal tensile stress inclination vs shear stress relationship for different amounts of external GFRP reinforcement (corner joint C8—third story). 2.44 MPa for exterior joint at columns C8 and C2 on the first floor, respectively), as confirmed by shear cracks observed on joints after the tests, it was decided to preserve the corner joints by installing FRP laminates. The shear improvement provided by FRP laminates was assessed according to the approach proposed by Antonopoulos and Triantafillou [15] which, on the basis of equilibrium considerations, allows following the possible states of joint behavior up to failure. Once geometric, bond and material properties are given and the acting axial forces are evaluated, the equations provide the inclination of the principal tensile stress, , and the shear stress, , corresponding to any given state of joint strains. Failure of the FRP strengthened joint occurs when either the concrete crushes (i.e. the principal compressive stress attains the crushing strength of concrete) or the FRP fails (i.e. the ultimate stress is attained or debonding occurs). In order to take account of the fact that by increasing the joint strains, the inclination of principal tensile stresses, , changes considerably, it was decided to upgrade the exterior joints by using quadriaxial laminates; according to the column retrofit, glass fibers were chosen. As the Antonopoulos and Triantafillou [15] model referred to uniaxial laminates, only fibers placed along the axial direction of columns and beams and those with a component on them were taken into account for calculations. The amount of the FRP needed on the joints was designed with reference to the weakest joint of the structure at column C8 (i.e. the original shear strength was 1.82, 1.65 and 1.62 MPa at first, second and third story, respectively). The target design was to improve its shear strength up to at least 4.00 MPa, about 2.5 times more than the original shear strength at the third story. With reference to this joint at the third story (axial load P = 15 650 N), Figure 7 shows the relationship between the inclination of the principal tensile stress, , and the shear stress, , corresponding to any given state of joint strains for one ply of FRP reinforcement installed (continuous line) and its progress by adding one ply at a time of GFRP quadriaxial laminates up to three plies (dashed line). Copyright 2007 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:141–162 DOI: 10.1002/eqe 156 M. DI LUDOVICO ET AL. Importantly, the theoretical failure mode was always concrete crushing, assuming that proper anchorage would be ensured to prevent FRP debonding. Figure 7 clearly shows that the amount of external FRP necessary to pursue the proposed target shear strength was two plies of GFRP quadriaxial laminates with a unit weight of 1140 g/m2 . The results in terms of shear strength, max , with reference to each exterior joint, obtained by installing one, two and three plies of quadriaxial GFRP laminates, were computed and reported in the last three columns of Table V. The results reported in the same table confirm that, in every case, two plies of GFRP laminates are adequate to achieve a shear strength of at least 4.00 MPa. Since rectangular column C6 has a sectional aspect ratio equal to 3, shear rather than flexure could have controlled its behavior. Hence, shear FRP retrofit was considered necessary. It was computed (by using CNR-DT 200/2004 [14] provisions) that totally wrapping the rectangular column C6 along its entire length with two plies of the same quadri-axial GFRP laminates used for the above joints was able to increase the sectional shear strength by about 50% (i.e. the shear strength is 196 kN by considering concrete and stirrups shear contribution only, and this values increases up to 286 kN by considering the GFRP effect). Importantly, only fibers placed perpendicular to the longitudinal axis of the column and those with a component on that direction were taken into account for calculations. Thus, the same equations provided for uniaxial laminate shear strengthening were used in calculations. 6. ASSESSMENT OF THE RETROFITTED STRUCTURE Nonlinear static pushover analysis was performed on the FRP-confined structure in order to estimate the effectiveness of the proposed retrofit technique on the global structural behavior. Assuming that the story mass remains constant after the FRP retrofit intervention, the modal displacements of each center of mass in directions X and Y and the corresponding normalized lateral loads are the same as those referred to the ‘as-built’ structure (reported in Table II). FRP confinement was taken into account by modifying the inelastic flexural behavior of the elements at the member ends, where the lumped plasticity is assumed. The bilinear moment–rotation relationship used for each plastic hinge was modified to account for the increases in ultimate curvature u (and the related increase in ultimate rotation capacity) due to FRP confinement. In particular, yielding curvature, y , and moment, My , were not modified by FRP confinement, while the ultimate curvature, u , and ultimate moment, Mu , were determined on reaching the increased ultimate strains in concrete, ccu , (determined from Equation (5)) or in steel reinforcement, assumed equal to 402 as in the ‘as-built’ structure. Yielding rotation, ultimate rotation and plastic hinge length were computed by using Equations (1)–(4). The knowledge level was again assumed equal to 3, KL3, with a corresponding confidence factor, CF, equal to 1. The significant damage limit state, (LSSD), was investigated to assess the structural capacity at both 0.20g and 0.30g PGA level in directions PX –NX and PY –NY , respectively. The pushover curves on the FRP-retrofitted structure for each direction analyzed are reported in Figure 8. The theoretical results in terms of maximum base shear, Fmax , top displacement, dmax , and absolute inter-story displacement are summarized in Table III. Seismic demand was computed with reference to the same design spectra analyzed in the ‘as-built’ configuration (see Figure 4) scaled at 0.20g and 0.30g PGA level. The results in terms of maximum top displacement required for each investigated PGA level and direction are summarized in Table III, which shows that the FRP retrofitted structure is able to satisfy the LSSD in each direction at both 0.20g and 0.30g PGA level. In particular, Copyright 2007 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:141–162 DOI: 10.1002/eqe 157 SEISMIC STRENGTHENING OF AN UNDER-DESIGNED RC STRUCTURE NX-NY PX-PY 350 Base shear [KN] 280 210 NY 140 PY_FRP NX PX NY_FRP CM PX_FRP 70 PY NX_FRP LSSD -0.15 -0.10 0 0.00 -0.05 0.05 0.10 0.15 Top displacement [m] Figure 8. Pushover curves for the assessment of the FRP-retrofitted structure. FRP Retrofitted Structure (Push NX) 1.00 ag = 0.30g 0.90 0.80 Sae 0.70 T* = 0646 s ag = 0.20g 0.20 g Elastic Demand FRP Structure Capacity Diagram 0.20 g Inelastic Demand 0.30 g Elastic Demand 0.30 g Inelastic Demand "AS Built Capacity Diagram" Sa [g] 0.60 0.50 0.40 0.30 µ = 5.19 µ = 3.46 0.20 Sye 0.10 0.00 0.00 * y D 0.05 Sd = S de 0.10 Sd [m] 0.15 0.20 Figure 9. FRP-retrofitted structure elastic and inelastic demand spectra vs capacity diagram. verification is also satisfied in direction NX where the maximum gap in terms of displacement demand was recorded for the ‘as-built’ structure: capacity is increased up to 0.1076 m (0.0626 m in the ‘as-built structure’) while demand at the target seismic level intensity, 0.30g, is equal to 0.0927 m. These results are illustrated in Figure 9 where the seismic demand and structural capacity of the FRP-retrofitted structure are determined (by using the CSA) in direction NX , for the two levels of ground motion analyzed. However, in the case of the retrofitted structure at the 0.30g PGA level, the most critical verification is in direction NY for which the capacity Copyright 2007 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:141–162 DOI: 10.1002/eqe 158 M. DI LUDOVICO ET AL. FRP Retrofitted (PX) As Built Structure (PX) 1.00 1.00 ag = 0.30g 0.90 0.90 0.80 0.80 0.70 0.70 Elastic Demand 0.60 0.50 µ = 5.22 0.40 Elastic Demand 0.60 Sa [g] Sa [g] ag = 0.30g 0.50 µ= 5.11 0.40 0.30 0.30 0.20 0.20 Capacity Diagram 0.10 0.00 0.00 0.05 Inelastic Demand 0.01 Sd [m] 0.10 0.15 0.20 0.05 0.90 0.80 0.80 Sa [g] Sa [g] Elastic Demand 0.60 0.50 µ =5.25 ag = 0.30g Elastic Demand 0.60 0.50 µ = 5.19 0.40 0.30 0.30 0.20 0.20 Inelastic Demand 0.10 0.10 Capacity Diagram 0.00 0.00 0.05 0.01 0.15 0.20 0.05 ag = 0.30g 0.80 0.70 0.70 Elastic Demand 0.60 Sa [g] Sa [g] 0.20 0.90 0.80 µ = 4.95 0.40 Elastic Demand 0.60 0.50 µ = 4.87 0.40 0.30 0.30 0.20 0.20 Capacity Diagram 0.10 0.00 0.00 0.05 Inelastic Demand 0.01 Inelastic Demand Capacity Diagram 0.10 0.15 0.00 0.00 0.20 0.05 Sd [m] 0.01 0.15 0.20 Sd [m] FRP Retrofitted (NY) As Bullt Structure (NY) 1.00 1.00 ag = 0.30g 0.90 0.90 0.80 0.80 ag = 0.30g 0.70 0.70 0.60 Elastic Demand Sa [g] Sa [g] 0.15 FRP Retrofitted (PY) 1.00 ag = 0.30g 0.90 0.01 Sd [m] As Built Structure (PY) 1.00 Inelastic Demand Capacity Diagram 0.00 0.00 Sd [m] 0.50 0.20 0.70 0.70 0.50 0.15 FRP Retrofitted (NX) 1.00 ag = 0.30g 0.90 0.40 0.01 Sd [m] As Built Structure (NX) 1.00 Inelastic Demand Capacity Diagram 0.00 0.00 µ = 4.28 0.30 0.30 0.20 0.00 0.00 µ = 4.24 0.50 0.40 0.40 0.10 Elastic Demand 0.60 Capacity Diagram 0.20 Inelastic Demand Inelastic Demand Capacity Diagram 0.10 0.00 0.05 0.01 0.15 Sd [m] 0.20 0.00 0.05 0.01 0.15 0.20 Sd [m] Figure 10. Theoretical seismic performance comparison at 0.3g PGA between ‘as-built’ and FRP-retrofitted structure. displacement is 0.0908 m, whereas the demand is equal to 0.0906 m. In order to show the increase in deformation capacity provided by FRP confinement in each direction, LSSD verification at 0.30g PGA by using the CSA is reported in Figure 10: on the left-hand side, the ‘as-built’ structure Copyright 2007 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:141–162 DOI: 10.1002/eqe SEISMIC STRENGTHENING OF AN UNDER-DESIGNED RC STRUCTURE 159 is analyzed, while on the right the theoretical predictions for the FRP-retrofitted structure are plotted. Figure 10 clearly shows that the column confinement provides the structure with significantly enhanced ductility, allowing it to achieve the theoretical demand by only playing on the plastic branch of the base shear top displacement curve. 7. TESTS ON THE RETROFITTED STRUCTURE Once testing of the ‘as-built’ structure was completed, prior to laminates installation, unsound concrete was removed in all the parts of the elements where crushing was detected. The original cross-sections were then restored using non-shrinking mortar. In addition, all cracks caused by the first round of testing were epoxy injected. Then, according to the design of the retrofit illustrated above, the eight square columns were all confined at the ends by using two plies of GFRP uniaxial laminates, each with a unit weight of 900 g/m2 . At each story, GFRP confinement was extended for 800 mm from the beam–column interface. In some cases, this length was increased up to 1000 mm to account for the more extended concrete damage (see Figure 11(a)). Beam–column joints corresponding to the corner square columns (C2, C5, C7 and C8) were strengthened using two plies of quadriaxial GFRP laminates having each a unit weight of 1140 g/m2 . This joint reinforcement was extended on the beams by 200 mm on each side in order to U-wrap it and to ensure a proper bond. The joint strengthening intervention scheme along with the joint internal and external view after the retrofit is presented in Figure 11(a). The GFRP panels on the joints were not connected to the columns. Indeed, the continuity of external reinforcement can vary the strength hierarchy of the connection and reduce the contribution of fixed end rotation to the rotation capacity of the column. Therefore, the plastic hinge length of the rehabilitated columns was assumed comparable with that of unconfined columns. The shear strength scheme of column C6 and an overview of the whole structure after the retrofit intervention are presented in Figure 11(b). Once FRP retrofitted, the structure was first tested with a PGA level of 0.20g, to have a direct comparison with the previously executed experiment, then with a PGA level of 0.30g. The experimental activity showed that the retrofitting intervention provided the structure with a very significant enhanced deformation capacity with respect to the ‘as-built’ configuration, which almost totally lacked the appropriate capacity to resist even the 0.20g PGA level of excitation. After the vertical elements and the joints were wrapped with glass fibers, the retrofitted structure was able to withstand the higher (0.30g PGA) level of excitation without exhibiting significant damage. After tests, FRP was removed and it was shown that the RC core was neither cracked nor damaged. The comparison of the column damage after tests on both the ‘as-built’ and the FRP-retrofitted structure is reported in Figure 12. The experimental outcomes are summarized in Table I for both PGA levels of 0.20g and 0.30g for directions X and Y , respectively; very similar behavior emerges between the ‘as-built’ and retrofitted structure at the same seismic level intensity (0.20g). Indeed, the maximum base shears recorded were 195 and 211 kN in direction X , and 276 and 287 kN in Y , for the ‘as-built’ and retrofitted configuration, respectively (increase of about 8 and 4%, in directions X and Y , respectively). The same trend was recorded in terms of top displacement: the maximum difference recorded was about 9% in direction Y (0.1031 m vs 0.1125 m for the ‘as-built’ and retrofitted structure, respectively) confirming that, as masses and strength do not significantly change, the retrofit intervention does not modify the structural response. Moreover, on increasing the seismic level intensity up to 0.30g, the maximum base Copyright 2007 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:141–162 DOI: 10.1002/eqe 160 M. DI LUDOVICO ET AL. 2 plies of QUADRI-AX 1140/48 (a) (b) Figure 11. (a) Column confinement and shear strength of exterior joints and (b) shear strength of column C6 and retrofitted structure overview. shear recorded on the retrofitted structure was slightly lower (7 and 2% in directions X and Y , respectively) than that achieved in the retrofitted structure at 0.20g. In contrast, the maximum displacement of the structure was significantly enhanced, especially in direction X ; the maximum top displacement recorded was 0.2053 m, roughly twice that reached during the previous tests. This confirms that the FRP retrofit is able to greatly increase the global deformation capacity of the structure, affecting its strength only slightly. Further experimental evidence is obtained if the results in terms of absolutely inter-story drift are analyzed. Table I shows a significant increase in absolute inter-story drifts at each floor if values recorded at the 0.3g and 0.20g PGA tests are compared. In particular, an increase of about 85% was recorded at the second story in the weak direction X (0.1060 m vs 0.0570 m). The experimental behavior of the rehabilitated structure was very close to that expected according to the rehabilitation design: (1) very ductile behavior of the columns was observed and (2) no brittle mechanisms occurred (i.e. shear failure or significant damage of joints). Copyright 2007 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:141–162 DOI: 10.1002/eqe SEISMIC STRENGTHENING OF AN UNDER-DESIGNED RC STRUCTURE As-Built Structure (PGA = 0.20g) Column view after the Column after damaged test concrete removing 161 FRP retrofitted Structure (PGA = 0.30g) Column view after the test Concrete core after FRP removing Figure 12. Damage on columns: comparison after the test on the ‘as-built’ and FRP-retrofitted configuration. 8. CONCLUSIONS The paper deals with full-scale tests on an under-designed RC structure in the ‘as-built’ and FRP retrofitted configurations. The retrofit criteria and calculation procedures used to design the amount and layout of FRP required to improve the seismic performance of the structure are presented and discussed. The experimental results provided by the structure in the ‘as-built’ and GFRP-retrofitted configurations highlight the effectiveness of the FRP technique in improving the global performance of under-designed RC structures in terms of ductility and energy dissipation capacity. In the present case study, this goal was achieved by confining the column ends and preventing brittle mechanisms (i.e. exterior joints and rectangular column shear failure). The experimental results confirmed that this seismic upgrade approach, outlined by the Italian guideline CNR-DT 200/2004, could be effective. The design equations used for shear strengthening of exterior beam–column joints and of the wall-type column were found effective to quantify the GFRP laminates needed to enable the structure to fully exploit its improved deformation capacity given by the increased ductility of the FRP-confined columns. Pushover analysis provided results qualitatively close to the experimental outcome, confirming the effectiveness of the FRP retrofit in increasing the global deformation capacity of the ‘as-built’ structure by strongly improving its displacement capacity at a significant damage limit state. The experimental results confirmed the theoretical predictions, showing that the FRP retrofit allowed the structure to withstand a level of excitation, in two directions, 1.5 times higher than that applied to the ‘as-built’ structure, without exhibiting significant damage or structural deterioration. ACKNOWLEDGEMENTS The SPEAR project was co-ordinated by Dr Paolo Negro from the Joint Research Centre with the administrative co-ordination of Prof. Michael Fardis from the University of Patras. Professor Fardis also provided the original design of the structure. The cooperation of members of the EU Joint Research Centre at Ispra is gratefully acknowledged. The SPEAR consortium for the preliminary numerical analyses and Copyright 2007 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:141–162 DOI: 10.1002/eqe 162 M. DI LUDOVICO ET AL. the whole staff of the ELSA Laboratory of the JRC, where all experimental activities were carried out, are gratefully acknowledged. The analysis of the test results was developed within the activities of Rete dei Laboratori Universitari di Ingegneria Sismica—ReLUIS (Research Line 8) funded by the Italian Department of Civil Protection—Executive Project 2005-2008. The retrofit of the structure was supported by MAPEI S.p.a., Milan, Italy. REFERENCES 1. Negro P, Mola E, Molina FJ, Magonette GE. Full-scale PsD testing of a torsionally unbalanced three-story non-seismic RC frame. Proceedings of 13th WCEE, Paper No. 968, Vancouver, Canada, 1–6 August 2004. 2. Cosenza E, Prota A. Experimental behaviour and numerical modelling of smooth steel bars under compression. Journal of Earthquake Engineering 2006; 10(3):313–329. 3. Molina FJ, Verzeletti G, Magonette G, Buchet Ph, Geradin M. Bi-directional pseudodynamic test of a full-size three-story building. Earthquake Engineering and Structural Dynamics 1999; 28(12):1541–1566. 4. Molina FJ, Buchet Ph, Magonette GE, Hubert O, Negro P. Bidirectional pseudodynamic technique for testing a three-story reinforced concrete building. Proceedings of 13th WCEE, Paper No. 75, Vancouver, Canada, 1–6 August 2004. 5. Dhakal RP, Mander JB, Mashiko N. Identification of critical ground motions for seismic performance assessment of structures. Earthquake Engineering and Structural Dynamics 2006; 35(8):989–1008. 6. Jeong S-H, Elnashai AS. Analytical assessment of an irregular RC frame for full-scale 3D pseudo-dynamic testing—Part I: analytical model verification. Journal of Earthquake Engineering 2005; 9(1):95–128. 7. European Standard, EN 1998-1. Eurocode 8 ‘Design of Structures for Earthquake Resistance’, Part 1: General Rules, Seismic Actions and Rules for Buildings. Final Draft, Brussels, December 2003. 8. CSI. SAP2000 V-7.1. Integrated Finite Element Analysis and Design of Structures Basic Analysis Reference Manual. Computer and Structures Inc.: Berkeley, CA, U.S.A., 2002. 9. Fardis MN. Analysis and design of reinforced concrete buildings according to Eurocode 2 and 8. Configuration 3, 5 and 6, Reports on Prenormative Research in Support of Eurocode 8, 1994. 10. European Standard, EN 1998-3. Eurocode 8, ‘Design of Structures for Earthquake Resistance’, Part 3: Strengthening and Repair of Buildings. Doc CEN/TC250/SC8/N343, Draft No. 3, January 2003. 11. Ordinance n. 3431. ‘General Criteria for Seismic Classification of National Territory and Technical Guidelines for Structures in Seismic Zones’, 3rd May 2005. (Ordinanza n. 3431, 3 maggio 2005, Ulteriori modifiche ed integrazioni all’ordinanza del Presidente del Consiglio dei Ministri n. 3274 del 20 marzo 2003, recante ‘Primi elementi in materia di criteri generali per la classificazione sismica del territorio nazionale e di normative tecniche per le costruzioni in zona sismica’). 12. Park R, Paulay T. Reinforced Concrete Structures. Wiley: New York, 1975; 769. 13. Fajfar P. A nonlinear analysis method for performance based seismic design. Earthquake Spectra 2000; 16(3): 573–592. 14. CNR-DT 200/2004. Guide for the Design and Construction of Externally Bonded FRP Systems for Strengthening, 2004 (Downloaded free from: http://www.cnr.it/sitocnr/IlCNR/Attivita/NormazioneeCertificazione/ NormazioneeCertificazione file/IstruzioniCNR DT200 2004 eng.pdf). 15. Antonopoulos CP, Triantafillou TC. Analysis of FRP-strengthened RC beam-column joints. ASCE Journal of Composites for Construction 2002; 6(1):41–51. Copyright 2007 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:141–162 DOI: 10.1002/eqe