ACI STRUCTURAL JOURNAL TECHNICAL

ACI STRUCTURAL JOURNAL

TECHNICAL PAPER

Title no. 101-S69

Selective Upgrade of Underdesigned Reinforced Concrete Beam-Column Joints Using Carbon FiberReinforced Polymers by Andrea Prota, Antonio Nanni, Gaetano Manfredi, and Edoardo Cosenza Seismic regions are characterized by reinforced concrete (RC) structures that have been designed without seismic provisions. As the structural upgrade of such structures becomes necessary, one of the approaches that recent design guidelines suggest is that of the hierarchy of strength. By increasing the strength and ductility of critical components, their brittle and catastrophic failure is prevented and the occurrence of more desirable mechanisms is promoted. This approach allows improving the global behavior of the structure. The key issue of strengthening design of RC frames is represented by the beam-column connection. This paper presents a technique based on fiber-reinforced polymer (FRP) composites for the seismic upgrade of RC beam-column connections that was validated with tests on 11 initially underdesigned specimens. The results of the experimental investigation are presented, and a discussion on how different parameters influenced the behavior of the samples in terms of strength or ductility, or both, is offered. Keywords: beams; columns; ductility; fibers; polymers; reinforcement.

INTRODUCTION In seismic areas, the strengthening of underdesigned reinforced concrete (RC) structures represents a crucial issue involving technical and social aspects. Because such structures were originally designed to carry only gravity loads, they lack the ductility and the hierarchy of strength that induce a global failure mechanism appropriate for seismic conditions. Typically, columns have minimum cross-sectional dimensions and their longitudinal steel reinforcement is inadequate to satisfy flexural and shear demand generated during an earthquake. This results in a weak column-strong beam construction that, under seismic loads, may lead to the formation of local hinges in the column. The associated failure mode represents the lower bound of the hierarchy of strength and is characterized by a brittle and catastrophic structural failure (Bracci, Reinhorn, and Mander 1992). Furthermore, the lack of appropriate size and spacing of column ties increases the risk of brittle and local failure mechanisms such as the collapse of the column end, resulting in crushing of the unconfined concrete, instability of the longitudinal steel reinforcing bars in compression, and pullout of those in tension when spliced. The terms “connection” and “subassemblage” are herein equally adopted to indicate the entire substructure extracted from a frame (Bonacci and Wight 1996) and given by columns, beams, and their intersection zone. The term “joint” is restricted to the portion defined by the beam-column intersection. RESEARCH SIGNIFICANCE The present research focuses on the seismic strengthening of underdesigned RC frames, which nowadays represents a ACI Structural Journal/September-October 2004

strong technical and social need in many parts of the world. The innovative aspect or the proposed technique is the combined use of FRP laminates and bars. This combination creates a synergism that was not attained previously. An experimental campaign was conducted on scaled beamcolumn connections. It aimed at validating different upgrade schemes and demonstrating that such a technique has the potential to become a sound and effective solution for the strengthening of RC frames located in seismic areas. PHILOSOPHY OF SEISMIC UPGRADE Different alternatives could be selected for the seismic upgrade of deficient RC structures: the required seismic performances can be attained by increasing the strength, or the ductility, or both. Seismic guidelines such as FEMA 273 (1997) underline that the objective can be reached by local modification of components, removal of irregularities, structural stiffening or strengthening, mass reduction, seismic isolation, or energy dissipation. Within these possible approaches, the work herein presented deals with upgrade solutions based on the local strengthening of structural components. The driving criterion is the hierarchy of strength: by boosting the strength of those members whose failure is not desirable, it is possible to attain a global performance characterized by the failure of more ductile and energy-dissipating components. Such an upgrade, even though dealing with local strength issues, enables to achieve a more ductile global performance of the system. In terms of the behavior of the underdesigned frame, the lower bound pertains to the column failure. The upgrade of columns, by providing them with higher strength by confinement and/or more flexural reinforcement, could move the failure to occur in the joint. Calvi, Magenes, and Pampanin (2001) underlined that, in the case of interior connections, moving the failure from the column to the joint can improve the global behavior of the frame, reducing the displacement demand on the column. The shear failure of the joint is brittle, however, and its influence on the global performance needs to be evaluated to understand the increase or reduction it provides in terms of energy dissipation of the entire frame. To move up along the hierarchy of strength, the joint should be strengthened next. The upgrade of both column and joint could allow movement from the intermediate level ACI Structural Journal, V. 101, No. 5, September-October 2004. MS No. 03-198 received June 12, 2003, and reviewed under Institute publication policies. Copyright © 2004, American Concrete Institute. All rights reserved, including the making of copies unless permission is obtained from the copyright proprietors. Pertinent discussion including author’s closure, if any, will be published in the July-August 2005 ACI Structural Journal if the discussion is received by March 1, 2005.

699

Andrea Prota is an assistant professor of structural engineering at University of Naples Federico II, Naples, Italy. He received his Master of Science in civil engineering at the University of Missouri-Rolla, Rolla, Mo., and his PhD in structural engineering at the University of Naples Federico II. His research interests include the seismic behavior of reinforced concrete and masonry structures, the use of advanced materials for new construction, and retrofit of existing structures using innovative techniques. Antonio Nanni, FACI, is the V&M Jones Professor of Civil Engineering at the University of Missouri-Rolla. He is a member of the Concrete Research Council, ACI Committees 437, Strength Evaluation of Existing Concrete Structures (current Chair); 440, Fiber Reinforced Polymer Reinforcement (founding Chair); 544, Fiber Reinforced Concrete; 549, Thin Reinforced Cementitious Products and Ferrocement; and Joint ACI-ASCE-TMS Committee 530, Masonry Standards Joint Committee. His research interests include the performance of concrete-based structures. Gaetano Manfredi is a professor of structural engineering at the University of Naples Federico II, where he received his PhD in structural engineering. His research interests include seismic engineering and the use of advanced composites in civil structures. Edoardo Cosenza is a professor of structural engineering at the University of Naples Federico II, where he received his PhD in structural engineering. His research interests include seismic engineering, steel-concrete composite structures, and composite materials for construction.

additional tension reinforcement. The simultaneous presence of FRP confinement by the laminates prevents the NSM reinforcement from becoming ineffective as a result of load reversals. A selective upgrade results from choosing different combinations and locations of externally bonded laminates and bars to obtain different structural subassemblage performances. To assess the validity of this strengthening methodology, an experimental program was carried out. Its objectives were to investigate the effects of FRP reinforcement on the behavior of the beam-column connection, its failure mechanism, and its ductility. The aim was to demonstrate that, by means of a targeted upgrade, it is possible to establish a hierarchy of strength in the subassemblage. Extensive literature is available on installation and performance of externally bonded FRP laminates using wet manual layup. In particular, the reader is referred to a recent ACI publication (ACI Committee 440 2002) that deals with all aspects of this concrete strengthening method. Even though less developed, the complementary strengthening method of NSM reinforcement has been well documented (Nanni and Faza 2002; Prota, Parretti, and Nanni 2003). This strengthening method consists of cutting shallow grooves in concrete and embedding the FRP bars in an epoxy or cementitious paste. NSM reinforcement installation is relatively simple in the case of flexural strengthening of beams and slabs, particularly in the case of negative moment regions. For the case of interest in this paper, a continuous bar has to be placed along the side faces of two consecutive columns and through the beam. As a result, the field installation would entail a combination of concrete cutting and drilling; particular attention should be paid during the grooving process to avoid cutting steel. Within the presented research, the procedure of drilling holes through the joint and making grooves on the column faces was eased and sped up by nailing wood strips directly to the forms prior to casting (Prota et al. 2000). Interviews with contractors and observations of other projects highlighted that field installation is possible with conventional tools. The authors believe that once the upgrade methodology has been fully validated in terms of structural performance, further studies may be necessary on the technological side to optimize the installation procedure of NSM rods and to improve constructibility.

Fig. 1—Test setup. of the hierarchy of strength (that is, shear failure of the joint) to its upper bound (that is, beam failure). Inducing such failure would be the best result of seismic upgrade. Formation of plastic beam hinges would mean that a ductile and very effective energy dissipating mechanism be achieved, maintaining the integrity of the global structure. PROPOSED SOLUTION FOR UPGRADE OF BEAM-COLUMN CONNECTIONS For upgrading RC beam-column connections, the proposed technique is based on the combined use of externally bonded fiber-reinforced polymer (FRP) laminates and nearsurface-mounted (NSM) FRP bars. The externally bonded laminates may be used to confine the column and to improve ductility and enhance performance of the compressed concrete. Laminates can also be used in the joint zone and beams to improve the shear capacity. The NSM bars are used in the column to improve flexural capacity by providing 700

TEST PROGRAM AND SETUP The experimental program consisted of 11 tests on interior RC beam-column connections (two beams only). The investigated parameters were: the constant axial load level on the column, P; the type of FRP reinforcement (laminates and bars); and the amount of FRP reinforcement applied. At one of the column ends away from the joint, a constant axial load P was applied by means of a hydraulic jack (that is, Location (1) in Fig. 1) independently operated. At the end of the other column, a load cell was placed to record the total reaction provided by the testing frames. To simulate pin connections, two steel rollers were used at both ends of the column. Two additional loads were applied on the ends of each beam with independently operated jacks (that is, Locations (2) and (3) in Fig. 1). A load cell on each jack connected to the beam recorded the applied force. As the specimen rested on the laboratory structural floor, a greased plywood sheet between the specimen and the floor limited friction and allowed for the free movement of the specimens. ACI Structural Journal/September-October 2004

Fig. 2—Loading arrangement: (a) gravity loads; (b) seismic loads before moment inversion on beams; and (c) seismic loads after moment inversion on beams. First, forces representing gravity loads were applied on the column and both beams (Fig. 2(a)); the direction of such loads is north-to-south in Fig. 1. Then, the two forces (shear forces) on the beams changed cyclically to simulate a seismic action inducing shear and bending moment on the columns while the sum of gravity loads (that is, axial load on the top column and beam shear forces) remained constant. Figure 2(b) and (c) depict the load pattern before and after moment inversion on beams, respectively; they allow to underline that the entity of seismic action on columns can be calculated based on equilibrium considerations for any given combination of shear forces applied on the beams. Dashed arrows in Fig. 2(b) and (c) indicate the reversal forces (that is, reversal cycles). To select the axial load P, similar tests performed on interior and exterior subassemblages, and reported in the literature, were considered. Three different axial load ratios (that is, axial load P divided by the product of the gross concrete area times its compressive strength) equal to 0.1, 0.2, and 0.3 were selected for the column; in this paper, they are denoted with codes L, H, and M, respectively. TEST SPECIMENS Typical frame and geometrical ratios of underdesigned structures were taken into account in choosing the dimensions of tested connections. Laboratory constraints required to adopt some scaling to maintain the specimen size and weight to a manageable level. The objective was to design beam-column connections typical of gravity load frames built during the 1960s without seismic provisions. For this, specimen design was carried out following the recommendations of the ACI 318-63 Building Code, predating the current code by 39 years. The limited number of planned tests imposed to eliminate three of the seven typical construction details (Beres et al. 1996) that have been found to be critical during a seismic event, namely, lapped splices of column reinforcement, discontinuities of beam reinforcement, and construction joints. The amount and spacing of longitudinal reinforcement as well as tie size and spacing were determined according to ACI 318-63 recommendations for 10 of the specimens. The lack of proper confinement was analyzed with the remaining last specimen by omitting some ties in the bottom column. The percentage of longitudinal reinforcement in the columns was equal to 1.92% of the concrete column, which was still within the typical range of 1 to 2%. The specimens were made without transverse reinforcement in the joint region and with dimensions yielding a weak-column strongbeam construction. ACI Structural Journal/September-October 2004

Fig. 3—Geometry and internal reinforcement of specimen. Table 1—Test matrix Column

Joint/beam

Axial NSM Laminate NSM Specimen load, kN bar type wrapping bars

|| to beam axis

⊥ to NSM || beam to beam axis axis

L1 L2

124.5 124.5

— —

— x

— —

— —

— —

— —

L3 L4

124.5 124.5

1 2

x x

x x

— —

— x

— x

H1 H2

249 249

— —

— x

— —

— —

— —

— —

H2U H3

249 249

— 1

x x

— x

— —

— —

— —

H4 H5

249 249

2 1

x x

x x

— x

x x

x —

M3

373.5

2

x

x

—

—

—

A square column with side equal to 200 mm and a rectangular beam with a 200 x 355 mm cross section was selected. Based on a design compressive strength of concrete equal to 30 MPa, the selected axial load ratios were achieved by applying axial loads of 124.5, 249.0, and 373.5 kN to the top column of specimen types L, H, and M, respectively, as reported in Table 1. They correspond to average stresses equal to 3, 6, 701

Table 2—Summary of experimental results fc′ , Specimen MPa L1

Fig. 4—Type 1 CFRP bars.

38.9

Failure mode Compression failure of columns

Story drift angle Ultimate Ulticolumn At cracking, mate, shear, kN % % 41.18

0.30

3.11

L2

39.8 Tension failure of columns

44.21

0.27

2.76

L3 L4

38.9 36.5

Shear failure of joint Column-panel interface

57.24 56.60

0.56 N/A

3.30 5.38

H1

31.7

Compression failure of columns

38.45

0.30

2.82

H2 H2U

36.5 36.5

Combined column-joint Combined column-joint

49.70 51.19

0.35 0.35

3.50 3.53

H3 H4

31.7 39.8

Shear failure of joint Column-joint interface

62.35 70.42

0.36 N/A

2.42 4.27

M3

39.8

Shear failure of joint

56.17

0.62

3.27

Note: N/A = not available.

transverse reinforcement of the beams. In all columns and beams, the concrete cover was equal to 38 mm. The geometry of a typical specimen is depicted in Fig. 3.

Fig. 5—Type 2 CFRP bars.

Fig. 6—Upgrade schemes: Scheme 2 (Specimens L2, H2, and H2U) and Scheme 3 (Specimens L3, M3, and M3). and 9 MPa, respectively. The cross-shaped specimen was 2.64 m long in the column direction and 3.05 m wide in the beam direction, respectively. Four Φ16 bars were placed as the longitudinal column reinforcement and Φ10 ties, spaced at 200 mm on center, were used as transverse reinforcement. According to the ACI recommendations, the first tie of each column was placed at 100 mm from the face of the beam. Three Φ22 and two Φ18 bars were used as negative and positive longitudinal reinforcement of the beams, respectively. U-shaped Φ10 stirrups, spaced at 100 mm on center, were used as 702

MATERIAL PROPERTIES Specimens were fabricated in four different placements having concrete cylinder compressive strengths fc′ as shown in Table 2. Tensile tests were performed in accordance with ASTM A 370 on three coupon specimens for each different diameter of steel bar. The yield strength was calculated averaging the results of each set of three samples. For Φ16, Φ18, and Φ22 bars, the average yield strength was 449, 559, and 511 MPa, respectively. Carbon FRP (CFRP) was selected for both bars and laminates. Two different types of CFRP bars were used. Type 1 bars (Fig. 4), with diameter equal to 9.5 mm, showed an average tensile strength equal to 2155 MPa, elastic modulus equal to 113.9 GPa, and an ultimate strain equal to 1.89%. These average values were obtained by performing three tensile tests as described in Micelli and Nanni (2001). For Type 2 bars (Fig. 5), with diameters equal to 8 mm, the average tensile strength was equal to 2014 MPa, elastic modulus equal to 108.3 GPa, and an ultimate strain equal to 1.86%. Figure 4 and 5 allow observation of the different surface properties of Type 1 and 2 bars. The former (Fig. 4) has lugs similar to a steel bar; the sequence of four samples depicted in Fig. 4 shows how surface configuration varies along the circumference. Type 2 bars are characterized by a smooth and sandblasted surface (Fig. 5); the four samples of Fig. 5 show the helicoidal winding of fibers that confine the core of the bar. Unidirectional CFRP laminates were adopted with the following nominal properties: ultimate tensile strength equal to 4323 MPa, modulus of elasticity equal to 264.0 GPa, and thickness equal to 0.165 mm (Yang et al. 2002). UPGRADE SCHEMES Table 1 summarizes upgrade schemes of tested subassemblages. Specimens L1 and H1 were used as controls. Subassemblages L2, H2, and H2U represent the first level of upgrade aimed at moving the failure from the column to the joint. For this, wrapping the end of each column (close to the joint) for a length of 380 mm was carried out using two plies of CFRP on each column end (Fig. 6). Specimen H2U was ACI Structural Journal/September-October 2004

Table 3—Theoretical analysis on column strength and failure mode Column

Vm (from Design fc′ , Pn, kN Mn, kNm Mn), kN MPa

Vn, kN

Expected failure

Control

30

329

40.25

36.59

66.3

F* - CC†

Wrapped

30

329

42.22

38.38

139.1

F* - CC†

Wrapped + NSM

30

329

73.42

66.74

139.1

F - NSM‡

*

Flexural failure. Concrete crushing. ‡ Breaking of NSM bars. †

equal to H2 in terms of upgrade solution; however, it was characterized by a confinement defect generated by omitting two ties in the bottom column. For Specimens L3, H3, and M3, NSM bars were installed on the columns prior to wrapping them with CFRP laminates. The application of such FRP bars, continuous through the beam, provides additional reinforcement fully anchored and effective in the maximum moment region of the column (Fig. 6) (Prota et al. 2001b). In Specimens L4 and H4, the joint was also strengthened along with the columns. Because the major concern for the joint is constituted by the presence of shear stresses, a diagonal reinforcement should ideally be used. Considerations based on the practical installation of CFRP reinforcement suggested strengthening the joint zone in both directions parallel and perpendicular to the beam axis. In the parallel direction, NSM bars were used, while in the perpendicular direction, CFRP laminates were applied (Fig. 7). Three CFRP bars, each 1520 mm long, were installed; the CFRP laminate (one ply) covered only the area of the joint without extending to the columns to simulate field conditions (that is, the presence of slab would limit access to top column). To anchor the NSM bars, CFRP single-ply U-wrapping (covering the bottom surface) of the beams for a width of 510 mm from the beam-column interface was also used. The Upgrade Level 5 is equivalent to Level 4 except for the presence of laminates in the direction parallel to the beam axis that substituted the CFRP bars of Scheme 4 (Fig. 7) (Prota et al. 2001b). The four FRP bars per side installed into the column of Schemes 3, 4, and 5 increased the reinforcement ratio from 1.92 to 2.28% (that is, L3, H3, and H5) and 2.16% (that is, L4, H4, M3). depending on the type of FRP bar used in each specimen (Table 1). The design of the upgrade schemes was checked by performing a strength assessment of the columns; the objective was to verify the validity of the designed strengthening configurations prior to proceeding with FRP installation. The bottom column was analyzed because it represents the worst situation as compared with the top column. Calculations were run assuming a design compressive strength of concrete equal to 30 MPa and considering the load condition corresponding to the axial load ratio of 0.2 (that is, type Subassemblages H). Summing the gravity load applied on the beams (that is, 40 kN on each side) to that acting on the top column (Table 1), an axial load Pu of 329 kN was obtained for the analysis performed according to the numerical procedure developed by Realfonzo et al. (2002). Three columns were considered: control (Scheme 1), wrapped (Scheme 2), strengthened with Type 1 NSM bars and wrapped (Scheme 3); the contribution of NSM in compression was disregarded. The calculated nominal moment Mn and ACI Structural Journal/September-October 2004

Fig. 7—Upgrade schemes: Scheme 4 (Specimens L4 and H4) and Scheme 5 (Specimen H5).

Fig. 8—Interaction diagrams for unstrengthened and strengthened columns.

Fig. 9—Concrete crushing in column of Specimen L1. the shear force Vm that Mn would induce in each column are reported in Table 3. The nominal shear capacity Vn for each scheme was also computed. The comparison between Vm and Vn highlights that the member fails for flexure in all cases; the expected failure mode is reported in Table 3. The interaction diagrams were also computed for the three considered columns and are depicted in Fig. 8; they underline how the effect of CFRP wrapping is significant only above the balanced condition (Scheme 2). The addition of NSM bars (Scheme 3), whose contribution in compression was disregarded, provided a strong benefit below the balanced 703

Fig. 12—View of failed joint in Specimen L3: front (left) and lateral (right). Fig. 10—Tensile column failure in Specimen L2.

Fig. 13—Failed Specimen H4 from top (left) and bottom (right) of beam. Fig. 11—Combined failure of Specimen H2U: joint (left) and column (right). condition and does not affect the behavior of the column for high axial load as compared with the only-wrapped. For the axial load ratios of 0.1 and 0.2 (that is, axial load on bottom column equal to 204 and 329 kN, respectively), the wrapping yields a moment increase equal to 7.44 and 109.37%, respectively; due to the presence of NSM bars and wrapping, an increase of column flexural capacity equal to 4.89 and 82.42%, respectively, is expected. Such values confirm that a small strength increase is provided by the wrapping below the balanced point; the NSM bars are very effective in that region and their contribution decreases as the axial load increases. OBSERVED FAILURE MODES Experimental outcomes are summarized in Table 2. Both control connections (L1 and H1) showed column failure due to concrete crushing (Fig. 9). Wrapping of the column ends moved the failure from the compression to the tension side for the low axial load (L2): wide cracks at the column-joint interface were observed, indicating an advanced deformation of the yielded steel, while light damage of the joint occurred as shown by shear cracks (Fig. 10). In the case of higher axial load on the columns (H2 and H2U), a combined column joint failure was experienced: still-wide cracks at the column-joint interface denoted a high level of steel deformation, but the damage on the joint was more pronounced than for connection L2 (Fig. 11). The installation of NSM bars as flexural 704

reinforcement for the column along with wrapping (L3, H3, and M3) allowed for the movement of the failure from the column to the joint (that is, shear failure) (Fig. 12). Regardless of the axial load level, no damage was observed on the tension side of the column at the interface with the joint for these three specimens. Strengthening of the joint (L4, H4, and H5) induced the failure to occur at the column joint interface (Fig. 13). This type of failure is related to the layout of the FRP reinforcement; namely, in the direction perpendicular to the beam axis, the laminate was terminated at the column joint interface to account for the presence of the floor system. The computer file containing the recorded data for Specimen H5 was not saved; for this reason, values are not shown in Table 2. This specimen provided only additional information concerning the failure mode. STRENGTH AND DUCTILITY PERFORMANCE Table 2 reports values of column shear corresponding to the failure of the subassemblage, illustrating how strength increases with the level of upgrade. For the lowest axial load ratio of 0.1, connection Schemes 3 and 4 showed almost the same strength, increasing the performances of the control specimen by approximately 39%. An increase in strength equal to approximately 7% was achieved by only wrapping the columns (Specimen L2). In the series with an axial load ratio equal to 0.2, column wrapping allowed a gain in strength of approximately 30% and compensated for the induced lack of confinement of ACI Structural Journal/September-October 2004

Fig. 14—Column shear versus story drift angle for Specimens: (a) L1; (b) L2; (c) L3; and (d) L4.

Fig. 15—Column shear versus story drift angle for Specimens: (a) H1; (b) H2; (c) H3; and (d) H4. Subassemblage H2U, whose behavior was very similar to H2. A difference of approximately 13% was observed between Connections H3 and H4, whose strength was approximately 83% larger than that shown by Connection H1. For Type 3 specimens, the experimental evidence provided values of ultimate column shear independent of the axial load ratio. Measurements of linear variable displacement transducers (LVDTs) placed on the beams allowed for the calculation of the story drift angle of the connection. Within a performancebased approach to earthquake at-risk buildings (ERBs), the story drift angle is the key acceptance criterion (FEMA 302 1998) as it has been recognized that the deficiency of seismic performance is mainly related to the lack of ductility (Priestley 1997). Increasing story drift angle and energy dissipation capacity is then the main objective in the seismic upgrade of atrisk structures. Because the tested subassemblage represents an extracted portion of the frame, values of story drift angles are ACI Structural Journal/September-October 2004

representative of the behavior of the real structure (Bonacci and Wight 1996). Their analysis is then crucial to understand how different levels of upgrade influence the global seismic performance of strengthened RC frames. For the low axial load ratio, column wrapping alone caused a loss of ductility equal to approximately 11%, while further column strengthening with FRP bars (that is, L3) and joint strengthening (that is, L4) improved the ductility by approximately 6 and 73%, respectively, compared with the control subassemblage. In the case of Type H connections (highest axial load), the column wrapping boosted ductility by approximately 24%, while their strengthening with both FRP laminates and bars (that is, H3) lowered it by approximately 14%. The upgrade of the joint region (that is, H4) increased the story drift angle by approximately 51% compared with the control specimen. In the results evaluation, the differences in concrete strength surely affected Series H more than L, as Table 2 shows. Finally, the ultimate story 705

Fig. 16—Column shear versus story drift angle at cracking of joint (left-hand side) and at failure of subassemblage (righthand side). drift angle was almost the same for Connections L3 and M3, while a reduced value was provided by H3: the lower concrete strength seems to affect the performance of the subassemblage in terms of ductility. Figure 14 and 15 depict trends of column shear versus story drift angle for connections of Series L and H, respectively; a similar diagram characterized Subassemblage M3. Values of the story drift angle corresponding to joint crack initiation were also computed and summarized in Table 2; seismic guidelines (such as ATC40 [1996]) consider such a stage as a first level of degradation of the frame. Because the joint region was not instrumented, crack initiation was detected based on the appearance of first diagonal cracks in the nodal zone. Such visual methodology could not be applied in the case of Connections H4 and L4, where the joint was covered by FRP. For subassemblage Schemes 1 and 2, the story drift when the joint started cracking was very close to 10% of the value of connection failure. Such values increased for Scheme 3 specimens, ranging between 15 to 19% of the ultimate drift angles; this is due to the contribution provided by NSM bars to the tensile strength of the joint (Prota et al. 2001a). COMPARISON BETWEEN UPGRADE SCHEMES The previous analysis of test results underlined that, depending on the axial load ratio and concrete strength, the upgrade of a component (that is, column or joint) has different effects on strength and ductility of the subassemblage. Based on the experimental outcome, the fuses defined by similar specimens in terms of column shear versus story drift angle were analyzed. This was done at both first cracking of the joint and failure of the subassemblage. Because the back of each specimen was hidden during tests, measurements provided by strain gages installed on the side facing the floor were used to verify the symmetry of strain and crack distribution; this was also checked postmortem when the specimen was removed from the floor. For each level of upgrade (that is, 1, 2, 3, and 4), areas having experimental points as vertexes are depicted in Fig. 16. At joint crack initiation, both control connections reach the same story drift for the same cracking shear in the column (that is, point indicated by the arrow in Fig. 16). The column wrapping does not imply a significant change both in terms of cracking force and ductility (Scheme 2), while the combination of FRP laminates and bars allows a gain of approximately 60% in both column shear and story drift at 706

joint cracking (Scheme 3). First cracking of joints reinforced with FRP (Scheme 4) is not discussed as the visual method (that is, based on observation) used for detecting cracking initiation was not applicable in these cases. In terms of ultimate subassemblage performances, the upgrade Scheme 2 determines a gain in strength ranging between 10 and 26%, while Scheme 3 generates an increase between 43 and 55%. Both schemes do not allow for a significant improvement in terms of ductility, which can be even reduced by the wrapping of columns with low axial load ratio or by a combined application of FRP laminates and bars to columns with low concrete strength. In these cases, the presence of FRP increases the sectional ductility of the column (in terms of curvature of its cross section), but reduces its deformability as a member and also provides a stiffening effect to the entire subassemblage. Because the strengthening of the joint causes a considerable improvement of the performances of the connection (that is, gain between 44 and 75% in strength and between 50 and 75% in story drift angle), it is expected that such an upgrade will perform satisfactorily under seismic loads. CONCLUSIONS The combination of externally bonded FRP laminates and NSM bars was experimentally validated for the seismic upgrade of underdesigned beam-column connections. Experimental outcomes confirmed that varying the external reinforcement amount (number of plies and bars), the location (column or column plus joint), and reinforcement type (laminates, bars or their combination) could allow the engineer to decide the level of the hierarchy of strength and failure mode that the connection should attain. Laboratory findings highlighted that the joint region needs to be strengthened to achieve a significant gain both in strength and ductility. For different axial load levels, shear failure of the joint occurred at very similar values of column shear; previous experimental outcomes evidenced that the axial load ratio does not significantly influence shear failure of the joint (Murakami et al. 2000). Prota et al. (2001a) proved that this occurs due to a particular combination of principal compression and tension stresses for which the joint approaches critical values of its nominal shear stress (Paulay and Priestley 1992). The experiments pointed out the following: • Along with the amount and location of FRP, axial load and material properties can play an important role on the global performances of upgraded subassemblages; • Strengthening the column can improve the behavior of the subassemblage, but, due to the brittle failure of the joint, it does not provide much in terms of ductility. The upgrade of the joint zone increases its deformability and also provides a significant contribution to the ductility of the system; • The termination of the laminate in the direction perpendicular to the beam axis that is required by actual field conditions could determine a shear failure at the column joint interface; and • Performed tests highlighted the influence of axial load level and concrete strength on the global behavior. This suggests that a reliable assessment of the conditions of the original structure, particularly with respect to applied loads, material properties, and actual hierarchy of strength, could represent a crucial step towards a successful upgrade. ACI Structural Journal/September-October 2004

ACKNOWLEDGMENTS The authors wish to acknowledge the support of the NSF Industry/University Cooperative Research Center—Repair of Buildings and Bridges with Composites at University of Missouri-Rolla.

REFERENCES ACI Committee 318, 1963, “Building Code Requirements for Reinforced Concrete (ACI 318-63),” American Concrete Institute, Farmington Hills, Mich., 144 pp. ACI Committee 440, 2002, “Guide for the Design and Construction of Externally Bonded FRP Systems for Strengthening Concrete Structures (ACI 440-2R-02),” American Concrete Institute, Farmington Hills, Mich., 45 pp. ATC40, 1996, “Seismic Evaluation and Retrofit of Concrete Buildings,” Report SSC 96-0, California Seismic Safety Commission. Beres, A.; Pessiki, S. P.; White, R. N.; and Gergely, P., 1996, “Implications of Experiments on the Seismic Behavior of Gravity Load Designed RC Beamto-Column Connections,” Earthquake Spectra, V. 12, No. 2, pp. 185-198. Bonacci, J. F., and Wight, J. K., 1996, “Displacement-Based Assessment of Reinforced Concrete Frames in Earthquakes,” Mete A. Sozen Symposium: A Tribute from His Students, SP-162, J. K. Wight and M. E. Kreger, eds., American Concrete Institute, Farmington Hills, Mich., pp. 117-138. Bracci, J. M.; Reinhorn, A. M.; and Mander, J. B., 1992, “Evaluation of Seismic Retrofit of Reinforced Concrete Frame Structures,” Technical Report NCEER-92-0031, State University of New York at Buffalo, Buffalo, N.Y., pp. 3-7. Calvi, G. M.; Magenes, G.; and Pampanin, S., 2002, “Relevance of Beam-Column Joint Damage and Collapse in RC Frame Assessment,” Journal of Earthquake Engineering, V. 6, No. 1, pp. 75-100. FEMA 273, 1997, “NEHRP Guidelines for the Seismic Rehabilitation of Buildings,” Prepared for FEMA by the Building Seismic Safety Council (BSSC), Washington, D.C., 257 pp. FEMA 302, 1998. “NEHRP Recommended Provisions for Seismic Regulations for New Buildings and Other Structures,” Prepared for FEMA by the Building Seismic Safety Council (BSSC), Washington, D.C., 337 pp.

ACI Structural Journal/September-October 2004

Micelli, F., and Nanni, A., 2001, “Issues Related to Durability of FRP Reinforcement for RC structures Exposed to Accelerated Aging,” Paper No. 056, Proceedings of the 6th American Society for Composites Technical Conference, M. W. Hyer and A. C. Loos, eds., Virginia Tech University, Blacksburg, Va., 12 pp. Murakami, H.; Fujii, S.; Ishiwata, Y.; and Morita, S., 2000, “Shear Strength and Deformation Capacity of Interior R/C Beam-Column Joint Subassemblage,” Proceedings, Twelfth World Conference on Earthquake Engineering, Auckland, New Zealand, Paper No. 679. Nanni. A., and Faza, S., 2002, “Design and Construction of Concrete Reinforced with FRP Bars: An Emerging Technology,” Concrete International, V. 24, No. 11, Nov., pp. 53-58. Paulay, T., and Priestley, M. J. N., 1992, Seismic Design of Reinforced Concrete and Masonry Buildings, John Wiley & Sons, Inc., New York, 768 pp. Priestley, M. J. N., 1997, “Displacement-Based Seismic Assessment of Reinforced Concrete Buildings,” Journal of Earthquake Engineering, V. 1, No. 1, pp. 157-192. Prota, A.; Nanni, A.; Manfredi, G.; and Cosenza, E., 2000, “Seismic Upgrade of Beam-Column Joints with FRP Reinforcement,” L’Industria Italiana del Cemento, V. LXX, No. 11, pp. 868-877. (in English and Italian) Prota, A.; Manfredi, G.; Nanni, A.; and Cosenza, E., 2001a, “Capacity Assessment of GLD RC Frames Strengthened with FRP,” Proceedings, Twelfth European Conference on Earthquake Engineering, London, Paper No. 241 (CD-ROM). Prota, A.; Nanni, A.; Manfredi, G.; and Cosenza, E., 2001b, “Selective Upgrade of Beam-Column Joints with Composites,” Proceedings, International Conference on FRP Composites in Civil Engineering, V. I, Hong Kong, J.-G. Teng, ed., Elsevier Science Ltd., Dec., pp. 919-926. Prota, A.; Parretti, R.; and Nanni, A., 2003, “Upgrade of RC Silos Using Near Surface Mounted FRP Composites,” L’Industria Italiana del Cemento, V. LXXIII, No. 2, pp. 170-183. (in English and Italian) Realfonzo, R.; Prota A.; Manfredi, G.; and Peece, M., 2002, “Flexural Strength of FRP-Confined RC Columns,” Proceedings, First fib Congress 2002, Osaka, Japan, pp. 41-50 (CD-ROM B). Yang, X.; Nanni, A.; Haug, S.; and Sun, C. L., 2002, “Strength and Modulus Degradation CFRP Laminates from Fiber Misalignment,” Journal of Materials in Civil Engineering, V. 14, No. 4, pp. 320-326.

707