- Email: [email protected]
Seismic retrofit of welded steel moment connections with highly composite floor slabs
Journal of Constructional Steel Research 139 (2017) 62–68
Contents lists available at ScienceDirect
Journal of Constructional Steel Research
Seismic retrofit of welded steel moment connections with highly composite floor slabs Kim Sung-Yong, Lee Cheol-Ho ⁎ Department of Architecture and Architectural Engineering, Seoul National University, Seoul 151-742, Republic of Korea
a r t i c l e
i n f o
Article history: Received 2 June 2017 Received in revised form 26 August 2017 Accepted 5 September 2017 Available online xxxx Keywords: Seismic retrofit Connections Steel moment frames Slab effect Haunch Heavy shear tab
a b s t r a c t In the 1994 Northridge earthquake, connection damage initiated from the beam bottom flange was prevalent in welded steel moment frames. The composite action due to the presence of a concrete floor slab was speculated as one of the critical causes of the prevalent bottom flange fracture. Close review of past experimental studies recently conducted by the authors clearly indicated that conventional steel moment connections with highly composite slabs are much more vulnerable to the bottom flange fracture. In this study, three seismic retrofit schemes are presented for welded steel moment connections with highly composite floor slabs typical of existing steel moment frames in Korea. Because top flange modification of existing beams is not feasible due to the presence of a concrete floor slab, beam bottom flange or web modifications by using welded triangular or straight haunch and heavy shear tab were cyclically tested. Test results of this study showed that all the retrofit schemes used are effective in eliminating the detrimental effect caused by high composite action and can ensure excellent connection plastic rotation exceeding 4% rad. Side effects resulting from retrofit as well as design recommendations are also discussed. © 2017 Published by Elsevier Ltd.
1. Introduction Most of steel moment frame buildings have composite floor slabs which usually consist of steel beam, shear studs, metal decks, steel rebars or wire meshes, and topping concrete. As was recently discussed by the authors [1], understanding the slab influence on connection seismic performance is very important for both new construction and retrofit. Under earthquake loading, composite floor slab acts as diaphragm and shear studs are provided to help transfer diaphragm loads to the steel moment frame, to laterally support the beams, and to increase overall structural continuity. For gravity beams with simple connection at both ends, it is usual practice to design them as fully composite. In this case, adequate number of shear studs must be provided to ensure full composite action according to the well-established design procedure. It is the moment frame beams that much attention should be paid to. In the past design and construction practice in Korea, more shear studs than needed to transfer diaphragm forces were often provided for steel moment frame beams. For example, the use of a pair of shear studs of 16–19 mm diameter per deck rib (or about 200 mm on center) for moment frame beams was very common even when they were designed as non-composite beams, or when composite action was not explicitly considered in computing lateral stiffness and strength of the moment frame. Typical slab details in many countries ⁎ Corresponding author. E-mail address: [email protected] (C.-H. Lee).
https://doi.org/10.1016/j.jcsr.2017.09.010 0143-974X/© 2017 Published by Elsevier Ltd.
appear to be empirical and vary somewhat depending on locality. For example, providing 19-mm diameter shear studs spaced 300 mm on center has been a popular US west coast practice to transfer seismic forces from the slab to steel frame. The effects of composite slabs on cyclic seismic behavior of welded steel moment connections received only limited attention before the 1994 Northridge earthquake and most of cyclic seismic connection tests were conducted for bare steel beam-column subassemblies. However, unprecedented and widespread brittle failures were observed in connections of welded steel moment frames in the 1994 Northridge earthquake. Numerous causes were conjectured as potentially contributing to the poor seismic performance of the pre-Northridge steel moment connections [2]. Especially, the majority of reported damages occurred in the beam bottom flange. Other than the flaws recognized in the welding and detailing issues like the use of low toughness welding electrode and notch-like condition present in the bottom flange, the upward shift of the neutral axis due to the unintended composite action between steel beam and concrete floor slab was speculated as one of the critical causes for the prevalent bottom flange fracture. After the earthquake, the effects of a concrete slab on seismic performance were investigated by many researchers. However different observations and conclusions were made among different researchers for slab effect on seismic performance depending upon the details of beam-to-column connections and floor slabs. Recently the authors conducted a close review of extensive relevant experimental
S.-Y. Kim, C.-H. Lee / Journal of Constructional Steel Research 139 (2017) 62–68 Table 1 Test specimens. Specimen
Slab
Retrofit scheme
PN500 PN500C PN500C-HST PN500C-SH PN500C-TH
None Yes Yes Yes Yes
None None Heavy shear tab Straight haunch Triangular haunch
studies conducted after the 1994 Northridge earthquake [1]. Based on the review, the authors were able to draw the following conclusions: i) When connections are composed of conventional welded steel moment connections (similar to pre-Northridge (PN) type) and relatively shallow beams with a high degree of composite action, composite slab effect is quite detrimental to seismic performance because of significant upward shift of the neutral axis under positive moment, often leading to premature fracture of the beam bottom flange [3–5].
63
ii) Composite specimens with deep beams, a low degree of composite action, and connection modification by using RBS (reduced beam section) or haunch showed little or no negative effects due to the presence of a composite slab. The presence of a composite slab did not cause any significant shift of the neutral axis and increased strain demand on the bottom flange. The dominance of the beam bottom flange fracture was not observed. Instead, the composite slab was beneficial and helped improve seismic performance because local and lateral instability of the beams was decreased due to the bracing effect of the slab [6–10]. However, several experiments also showed that composite slab effect could be detrimental to the bottom flange even in RBS connections when composite action is very strong (for example, see [11]). iii) Connection improvement by using modification scheme like RBS or haunch provides a structural fuse at the critical section of the beam. In such connections with a promoted plastic zone, which can help relieve increased strain demand, the slab effect on the bottom flange is expected to be less critical than in PN type connections.
(a) PN500
(b) PN500C Fig. 1. Side view and top view of PN500 and PN500C.
(a) PN500C–HST
(b) PN500C–SH Fig. 2. Details of retrofitted specimens.
(c) PN500C–TH
64
S.-Y. Kim, C.-H. Lee / Journal of Constructional Steel Research 139 (2017) 62–68
Table 2 Measured tensile mechanical properties of steels. Specimen
H-500 × 200 × 10 × 16 (SS400) H-400 × 400 × 13 × 21 (SM490) Heavy shear tab (SN490) Straight haunch (SN490) Triangular haunch (SN490)
Flange Web Flange Web Flange Web Flange Web
Yield stress (MPa)
Tensile stress (MPa)
Yield ratio (%)
Elongation (%)
308 306 373 370 412 375 379 363 367
442 442 523 531 565 545 546 546 550
69 69 71 70 73 69 69 66 67
24 26 29 24 24 23 25 24 23
iv) The bottom flange in a composite beam is always more strained than that in a bare steel beam for a given beam rotation demand, and therefore welded seismic steel moment connections and concrete slab should be designed and constructed to minimize the composite action, if possible, especially when shallow beams and conventional connections are involved. In this study, as an attempt to salvage conventional (or PN type) welded steel moment connections with highly composite floor slabs, several connection retrofit schemes involving welded bottom haunches or heavy shear tab were tested and evaluated based on full-scale cyclic test results. 2. Testing program 2.1. Design of specimens When modifying existing connections for improved seismic performance, the presence of a concrete floor slab often dictates modification in the bottom flange only. The bottom flange modification by using welded triangular haunch is well described in the AISC Design Guide 12 [12]. The bottom flange modification by using welded straight haunch was also proposed based on the analytical and experimental studies by Lee and Uang [13] and Lee et al. [14]. Test specimens were designed to simulate one-sided (or exterior) moment connections. Five specimens were tested in this study as shown in Table 1 and Figs. 1 and 2. The steel beam and column sizes of all the specimens were identical. The grade of steel for the beams was SS400 with a specified minimum yield strength of 235 MPa; the steel used for the columns was SM490 with a specified minimum yield strength was 325 MPa. SN490 or seismic steel recently developed in Korea was used for fabricating haunches and other retrofit steel plates because of its improved weldability. The tensile test results based on the average of three coupons taken from the flanges and the webs are reported in Table 2. Each specimen consisted of H–500×200×10×16 (SS400) beam and H–400×400×13×21 (SM490) column. The beam and column sizes typical of past construction practice in Korea were selected. Especially, relatively shallow beams, 400–500 mm deep, were typically used because all the frames were designed as moment resisting; the use of a perimeter moment frame system is not a popular practice yet in Korea. The beam and the column used in specimens satisfy the seismically compact requirements of the AISC seismic provisions [15]. All specimens were fabricated
Fig. 3. Typical test set-up.
by using all-welded beam-to-column connection per the past practice in Korea. A notch-tough welding electrode with a specified minimum Charpy V-Notch (CVN) toughness of 26.7 J at − 28.9 °C (20 ft-lb. at − 20 °F) and a tensile strength of 490 MPa was specified for flux-cored arc welding in all specimens. Nominally identical 175 mm-thick composite slab with rebars (13 mm diameter @200 in both directions) and 100 mm concrete topping on 75 mm-deep corrugated steel deck was provided to all the specimens except PN500. The average 28-day compressive strength of three cylinder specimens taken from the concrete topping was fc′ = 21 MPa. The flutes of the deck were oriented perpendicular to the beam axis. A pair of shear studs of 19 mm diameter per deck rib (or 200 mm on center) was attached to the beam top flange of all the composite slab specimens following the past practice in Korea. The specified minimum tensile strength of shear stud (Fu) was 400 MPa. Two cross beams (with one row of shear studs of 19 mm diameter 200 mm on center) were also provided to simulate the presence of the gravity beams. The shear studs provided corresponded to 92% fully composite action. The degree of composite action was calculated per Eq. (1) as the shear strength of shear studs (∑ Qn) divided by tensile yielding strength of the steel beam (AsFy). The shear strength of a stud was calculated according to Eq. (2) following the AISC Specification [16]. P The degree of composite action ¼
Q n ¼ 0:5Asa
Qn As F y
ð1Þ
qffiffiffiffiffiffiffiffiffi 0 f c Ec ≤ Rg Rp Asa F u
ð2Þ
where ∑ Qn = sum of nominal shear strength of shear studs, Asa = cross-sectional area of shear stud, fc′ = specified compressive strength of concrete, Ec = modulus of elasticity of concrete, Rg = coefficient to account for group effect, Rp = position effect factor for shear studs, and Fu = specified minimum tensile strength of a shear stud. Table 3 summarizes the relative strength of the beam, the panel zone and the column of specimen PN500 based on the measured yield
Table 3 Relative strength of beam, panel zone, and column of PN500. Beam (SS400) Mpb = ZbFyb
Column (SM490) Mpc = ZcFyc
Panel zone (SM490) Rv = 0.60Fycdctw × [1 + (3bcft2cf / dcdbtw)]
SC-WB requirement ∑Mpc/∑Mpb
PZ requirement Rv/(Mpb/db)
671.4 kN-m
1368.9 kN-m
1400.6 kN
2.04
1.04
Note: All the strengths were calculated using the measured yield strength reported in Table 2; the panel zone strength Rv was computed per Eq. (J10-11) in the 2010 AISC Specification.
S.-Y. Kim, C.-H. Lee / Journal of Constructional Steel Research 139 (2017) 62–68
(a)PN500
65
(b)PN500C
(c) PN500C–HST
(d) PN500C–SH
(e) PN500C–TH
Fig. 4. Photographs showing plastic hinge region after completion of testing.
1.236 0.80
0.80 1.147
with a composite slab have not been tested yet, and limited tests were conducted for triangular haunch connections mostly for deep beams with lightly composite slabs. The motivation of including the heavy shear tab specimen PN500C–HST is as follows. As mentioned previously, when modifying existing connections for improved seismic performance, the presence of a concrete floor slab often dictates modification in the bottom flange only. Strengthening existing beam web by using heavy shear tab can be a viable alternative when even the bottom flange modification is not feasible. As was shown by Lee [17] and Lee-Kim [18], the load transfer mechanism close to the beam to column welded joint is completely different from the engineering beam theory; the flange near the welded joint should transfer a significant portion of the beam shear as well as the horizontal tensile force, and is overstressed. Further, the flange near the welded joint becomes brittle due to the welding heat affection and the tri-axial strain restraint condition there. Thus, the purpose of
Normalized moment at column centerline M/Mpb
Total plastic rotation, ×10-2 rad
1.591 0.80
0.80 1.132
Total plastic rotation, ×10-2 rad
(c) PN500C–HST
Normalized moment at column centerline M/Mpb
Normalized moment at column centerline M/Mpb
(a) PN500
1.546 0.80
0.80 1.118
Total plastic rotation, ×10-2 rad
(b) PN500C 2.013 0.80
0.80 1.458 Total plastic rotation, ×10-2 rad
(d) PN500C–SH Fig. 5. Total connection plastic rotation.
Normalized moment at column centerline M/Mpb
Normalized moment at column centerline M/Mpb
strength of steels. The column was much stronger than the beam, but the panel zone strength including the column flange contribution factor was comparable to the beam strength and thus some limited but varying degrees of panel zone yielding were expected among specimens because of different connection details and different shear demand on the panel zone under cyclic loading. Bare steel specimen PN500 was nominally identical to specimen PN500C except the absence of a composite slab. Both may be thought as benchmark specimens. Triangular and straight haunch specimens were designed according to the procedure in the AISC Design Guide 12 [12] and the procedure proposed by Lee-Uang [13] and Lee et al. [14], respectively. The unique load transfer mechanism of triangular and straight haunch connections and their design procedures for haunch connections are not presented here because of space limitations. Within the authors' knowledge, straight haunch connections
1.840 0.80
0.80 1.254
Total plastic rotation, ×10-2 rad
(e) PN500C–TH
S.-Y. Kim, C.-H. Lee / Journal of Constructional Steel Research 139 (2017) 62–68
0.80 0.80
Total plastic rotation, ×10-2 rad Fig. 6. Comparison of cyclic response of specimens with and without composite floor slab.
strengthening the beam web with heavy shear tab is two-fold; i) to reduce tensile strain demand on the beam flange by increasing the plastic section modulus of the retrofitted beam web, and ii) to push the plastic hinging away from the brittle welding area to the inside “ductile” area. As can be seen Fig. 2(a), the width of the heavy shear tab used in PN500C–HST is about half of the beam depth in order to push the plastic hinging sufficiently away from the column face. The height and the thickness of the heavy shear tab were 370 mm and 9 mm, respectively. As a result of the web strengthening, the plastic section modulus of the web was increased from 25% to 40% of the original beam section (H– 500×200×10×16). The two corners of the heavy shear tab were slightly trimmed to reduce stress concentration there (see Fig. 2(a)). 2.2. Test setup and loading The specimens were mounted to a strong floor and a strong wall. An overall view of the test setup is shown in Fig. 3. The column, which was pinned at the top and bottom, were 3.5 m high. The distance between the column centerline and the actuator loading point was 3.5 m. Thus, the beam tip displacement corresponding to 1.0% story drift was 35 mm. Lateral bracing was provided at a distance of 2.5 m from the column centerline. The specimens were tested statically according to the AISC cyclic seismic loading protocol [15]. The test specimens were instrumented with a combination of displacement transducers and strain gages to measure global and local responses. 3. Test results and discussions 3.1. Overall behavior and plastic rotation capacity
Distance from beam mid-depth, mm
Global behavior of each specimen including plastic rotation capacity and failure mode is described in the below. Photographs of the plastic hinging region at completion of testing are shown in Fig. 4. Fig. 5
0.375% 0.5% 0.75% 1%
Strain, ×10-3 mm/mm
(a) PN500
summarizes the hysteretic responses observed in testing in terms of the total connection plastic rotation. The ordinate of each figure is expressed in terms of the normalized moment at the column centerline. Total connection plastic rotation is calculated with assuming that plastic deformation is lumped at the column centerline. All tested specimens exhibited plastic rotation capacity well in excess of 3% rad that corresponds to the minimum requirement for special moment frame connections. Such an excellent performance may be attributed to a relatively shallow beam depth (500 mm) and all-welded quality welding by using notch-tough welding electrode. Below is brief description of the global behavior of each specimen. 3.1.1. PN500 As can be seen in Fig. 5(a), PN500 was able to develop an excellent connection plastic rotation of 5% rad without fracture. This specimen exhibited the maximum moment of 1.1Mpb at the column centerline for both positive and negative bending. Test was terminated because of severe local and lateral buckling at the plastic hinge (see Fig. 4(a)). 3.1.2. PN500C As mentioned previously, this composite specimen is nominally identical to PN500 except the presence of the composite floor slab. As can be observed in Fig. 5(b), the hysteretic response is asymmetric in contrast to that of PN500. The maximum positive moment observed in PN500C is as high as 1.55Mpb because of the composite action while the maximum negative moment is about 1.05Mpb just comparable to that of PN500. Of course, the large increase of the maximum positive moment due to the composite action can be very detrimental to the bottom flange and is never desirable from seismic perspective. This specimen experienced low cycle fatigue fracture near the heat affected zone at 4% plastic rotation level (see Fig. 4(b)) and was less ductile than PN500. 3.1.3. PN500C–HST This specimen is nominally identical to PN500C except the addition of heavy shear tab to the beam web. As a result of the web strengthening, brittle fracture was delayed and the connection ductility and energy dissipation capacity of PN500C–HST was significantly improved compared to PN500C. The maximum positive moment at the column centerline was about 1.59Mpb. This specimen showed gradual strength degradation after 5% rad plastic story drift, and eventually fractured due to low cycle fatigue at the very high plastic story drift of 7% rad. 3.1.4. PN500C–SH and PN500C–TH Both haunch-retrofitted specimens showed similar and excellent plastic rotation capacity exceeding 5% rad (see Fig. 5(d) and (e)) without fracture. As the photos in Fig. 4(d) and (e) show, the plastic hinging was successfully formed outside the haunch region. However, the maximum positive moments at the column centerline observed in PN500C–SH and PN500C–TH were about 2.01 and 1.84 times Mpb, respectively. Both Distance from beam mid-depth, mm
Normalized moment at column centerline M/Mpb
66
0.375%
1% 0.75%
0.5%
Strain, ×10-3 mm/mm
(b) PN500C
Fig. 7. Strain profiles of PN500 and PN500C under positive moment.
67
Beam tip force, kN
Beam tip force, kN
S.-Y. Kim, C.-H. Lee / Journal of Constructional Steel Research 139 (2017) 62–68
Panel distortion, ×10-2 rad
Panel distortion, ×10-2 rad
Beam tip force, kN
Panel distortion, ×10-2 rad
Beam tip force, kN
(b) PN500C
Beam tip force, kN
(a) PN500
Panel distortion, ×10-2 rad
(c) PN500C–HST
Panel distortion, ×10-2 rad
(d) PN500C–SH
(e) PN500C–TH
Fig. 8. Panel shear distortion of specimens.
haunch-retrofitted specimens showed rather severe pinching under unloading stage toward positive moment at high story drift because of severe flange and web local buckling (see Fig. 4(d) and (e)).
3.2. Comparison of responses and design recommendations Fig. 6 compares the cyclic response of the composite specimen PN500C with that of the bare steel counterpart PN500. PN500C showed seismic performance inferior to the bare steel counterpart PN500 because of low cycle fatigue fracture of the beam bottom flange at 4% plastic drift cycle. In addition, the specimen PN500C exhibited an increased beam flexural strength as high as 1.55Mpb under the positive moment compared to the maximum negative moment of about 1.05Mpb. Such asymmetric topology of the response may lead to the column hinging under positive moment and should be properly considered in design. The flexural strain profiles of PN500 and PN500C under positive bending are depicted in Fig. 7. It is observed that the neutral axis of PN500C was shifted upward by about 10% of the beam depth compared to the bare steel specimen PN500. Fig. 8 compares the panel zone shear distortions measured in the test. In the two specimens PN500C and PN500C-HST, the panel zone shear distortion is increased asymmetrically or further increased under the positive moment compared to the bare specimen PN500. This implies that neglecting the presence of composite floor slabs can lead to the underestimation of panel zone shear distortion. In the haunch specimens, the panel zone shear distortion is significantly reduced, especially in the straight haunch specimen. This reduction is due to the effect of enlarged (or dual) panel zone, a side benefit of haunch retrofit. Please refer to Lee-Uang [13] and Lee-Uang [19] for the behavior and mechanical modeling of dual panel zone of haunch connections.
As can be observed in Fig. 5, asymmetric topology of hysteretic response is very pronounced in all composite slab specimens. The maximum positive moments observed in heavy shear tab and straight or triangular haunch specimens were about 1.59, 2.01 and 1.84 times the bare steel plastic moment, respectively (see Fig. 5(c), (d) and (e)). Such increase of the positive moment as a negative effect of retrofit could violate the strong column-weak beam condition and should be considered in retrofit design and analysis. The positive and negative strain-hardening factors observed in this testing program are summarized in Table 4. Although test data is limited, simple guides for sizing and detailing of heavy shear tabs are illustrated in Fig. 9 based on successful test results of specimen PN500C–HST. Note that the web plastic section modulus of PN500 and PN500C–HST was respectively 25% and 40% of the plastic modulus of the whole beam section. Retrofit design by using triangular and straight haunches can be found in AISC Design Guide 12 [12], Lee-Uang [13], and Lee et al. [14]. 4. Summary and conclusions This study investigated seismic retrofit schemes for PN-type welded steel moment connections with highly composite floor slabs by using the beam bottom flange or beam web modification. The results of this study can be summarized as follows. i) This study again confirmed that high composite action of a floor slab is detrimental to seismic performance of PN-type welded steel moment connections; the almost fully composite PN type specimen showed seismic performance inferior to bare steel counterpart and eventually experienced to the beam bottom flange fracture.
Table 4 Positive and negative strain hardening factors observed. Computed at the column face
Positive strain hardening factor α+ Negative strain hardening factor α−
Computed at the tip of strengthened zone
Bare steel connection
Composite connection
Heavy shear tab
Straight haunch
Triangular haunch
1.165 1.082
1.458 1.054
1.386 0.987
1.610 1.166
1.472 1.003
68
S.-Y. Kim, C.-H. Lee / Journal of Constructional Steel Research 139 (2017) 62–68
CJP s
thst tw
s=thst-2 (mm)
Existing bolted shear tab CJP
db
s
s=thst-2 (mm) db thst tw
0.5db
tw Trim to reduce stress concentration
(a) All-welded
0.5db
tw Trim to reduce stress concentration
(b) Bolted web and welded flange
Fig. 9. Sizing and detailing of heavy shear tab recommended.
ii) All the three retrofitted specimens by using heavy shear tab and triangular or straight haunches effectively pushed plastic hinging outside the strengthened region and exhibited excellent total plastic rotation capacity exceeding 4% rad. Although the test data of heavy shear tab retrofit is very limited, simple retrofit design guides are also recommended. iii) Generally, asymmetric topology of hysteretic response was very pronounced in the highly composite slab specimens of this study. The maximum positive moments observed in heavy shear tab and haunch specimens were about 1.6 and 2.0 times the bare steel plastic moment, respectively. This side effect should be considered in retrofit design when checking the strong column-weak beam condition as well as in seismic performance evaluation of the retrofitted frames. iv) The heavy shear tab retrofit tended to increase the panel zone yielding compared to PN-type connections. However, haunch retrofit significantly reduced the panel zone yielding, less than PN-type counterparts, because of the enlarged (or dual) panel zone effect. Both triangular and straight haunch specimens exhibited comparable seismic performance. They showed increased pinching under unloading stage toward positive moment at high story drift because of severe flange and web local buckling. Acknowledgements This research was supported by a grant (17AUDP-B066083-05) from Architecture & Urban Development Research Program funded by Ministry of Land, Infrastructure and Transport of Korean government. The Institute of Engineering Research at Seoul National University provided research facilities for this work. References [1] C.H. Lee, J.H. Jung, S.Y. Kim, J.J. Kim, Investigation of composite slab effect on seismic performance of steel moment connections, J. Constr. Steel Res. 117 (2016) 91–100 (Elsevier). [2] M. Bruneau, C.M. Uang, A. Whittaker, Ductile Design of Steel Structures, McGrawHill, New York, 2011.
[3] R. Leon, J. Hajjar, M. Gustafson, Seismic response of composite moment-resisting connections. I: performance, J. Struct. Eng. ASCE 124 (1998) 868–876. [4] J. Hajjar, R. Leon, M. Gustafson, Seismic response of composite moment-resisting connections. II: behavior, J. Struct. Eng. ASCE 124 (1998) 877–885. [5] Y.J. Kim, S.H. Oh, T.S. Moon, Seismic behavior and retrofit of steel moment connections considering slab effects, Eng. Struct. 26 (2004) 1993–2005 (Elsevier). [6] C.M. Uang, Q. Yu, S. Noel, J. Gross, Cyclic testing of steel moment connections rehabilitated with RBS or welded haunch, J. Struct. Eng. ASCE 126 (2000) 57–68. [7] S. Civjan, M. Engelhardt, J. Gross, Retrofit of pre-Northridge moment-resisting connections, J. Struct. Eng. ASCE 126 (2000) 445–452. [8] S. Civjan, M. Engelhardt, J. Gross, Slab effects in SMRF retrofit connection tests, J. Struct. Eng. ASCE 127 (2001) 230–237. [9] S. Jones, G. Fry, M. Engelhardt, Experimental evaluation of cyclically loaded reduced beam section moment connections, J. Struct. Eng. ASCE 128 (2002) 441–451. [10] X. Zhang, J. Ricles, Experimental evaluation of reduced beam section connections to deep columns, J. Struct. Eng. ASCE 132 (2006) 346–357. [11] S.J. Chen, Y.C. Chao, Effect of composite action on seismic performance of steel moment connections with reduced beam sections, J. Constr. Steel Res. 57 (2001) 417–434 (Elsevier). [12] J.L. Gross, M. Engelhardt, C.M. Uang, K. Kasai, N.R. Iwankiw, Modification of Existing Welded Steel Moment Frame Connections for Seismic Resistance (Vol. 12), AISC, Chicago, Illinois, 2003. [13] C.H. Lee, C.M. Uang, Analytical modeling and seismic design of steel moment connections with welded straight haunch, J. Struct. Eng. ASCE 127 (2001) 1028–1035. [14] C.H. Lee, J.H. Jung, M.H. Oh, E.S. Koo, Cyclic seismic testing of steel moment connections reinforced with welded straight haunch, Eng. Struct. 25 (2003) 1743–1753 (Elsevier). [15] AISC, Seismic Provisions for Structural Steel Buildings, AISC, Chicago, Illinois, 2010. [16] AISC, Specification for Structural Steel Buildings, American Institute of Steel Construction (AISC), Chicago, Illinois, 2010. [17] C.H. Lee, Review of force transfer mechanism of welded steel moment connections, J. Constr. Steel Res. 62 (2006) 695–705 (Elsevier). [18] C.H. Lee, J.H. Kim, Seismic design of reduced beam section steel moment connections with bolted web attachment, J. Constr. Steel Res. 63 (2007) 522–531. [19] C.H. Lee, C.M. Uang, Analytical modeling of dual panel zone in haunch repaired steel MRFs, J. Struct. Eng. ASCE 123 (1997) 20–29.
Contents lists available at ScienceDirect
Journal of Constructional Steel Research
Seismic retrofit of welded steel moment connections with highly composite floor slabs Kim Sung-Yong, Lee Cheol-Ho ⁎ Department of Architecture and Architectural Engineering, Seoul National University, Seoul 151-742, Republic of Korea
a r t i c l e
i n f o
Article history: Received 2 June 2017 Received in revised form 26 August 2017 Accepted 5 September 2017 Available online xxxx Keywords: Seismic retrofit Connections Steel moment frames Slab effect Haunch Heavy shear tab
a b s t r a c t In the 1994 Northridge earthquake, connection damage initiated from the beam bottom flange was prevalent in welded steel moment frames. The composite action due to the presence of a concrete floor slab was speculated as one of the critical causes of the prevalent bottom flange fracture. Close review of past experimental studies recently conducted by the authors clearly indicated that conventional steel moment connections with highly composite slabs are much more vulnerable to the bottom flange fracture. In this study, three seismic retrofit schemes are presented for welded steel moment connections with highly composite floor slabs typical of existing steel moment frames in Korea. Because top flange modification of existing beams is not feasible due to the presence of a concrete floor slab, beam bottom flange or web modifications by using welded triangular or straight haunch and heavy shear tab were cyclically tested. Test results of this study showed that all the retrofit schemes used are effective in eliminating the detrimental effect caused by high composite action and can ensure excellent connection plastic rotation exceeding 4% rad. Side effects resulting from retrofit as well as design recommendations are also discussed. © 2017 Published by Elsevier Ltd.
1. Introduction Most of steel moment frame buildings have composite floor slabs which usually consist of steel beam, shear studs, metal decks, steel rebars or wire meshes, and topping concrete. As was recently discussed by the authors [1], understanding the slab influence on connection seismic performance is very important for both new construction and retrofit. Under earthquake loading, composite floor slab acts as diaphragm and shear studs are provided to help transfer diaphragm loads to the steel moment frame, to laterally support the beams, and to increase overall structural continuity. For gravity beams with simple connection at both ends, it is usual practice to design them as fully composite. In this case, adequate number of shear studs must be provided to ensure full composite action according to the well-established design procedure. It is the moment frame beams that much attention should be paid to. In the past design and construction practice in Korea, more shear studs than needed to transfer diaphragm forces were often provided for steel moment frame beams. For example, the use of a pair of shear studs of 16–19 mm diameter per deck rib (or about 200 mm on center) for moment frame beams was very common even when they were designed as non-composite beams, or when composite action was not explicitly considered in computing lateral stiffness and strength of the moment frame. Typical slab details in many countries ⁎ Corresponding author. E-mail address: [email protected] (C.-H. Lee).
https://doi.org/10.1016/j.jcsr.2017.09.010 0143-974X/© 2017 Published by Elsevier Ltd.
appear to be empirical and vary somewhat depending on locality. For example, providing 19-mm diameter shear studs spaced 300 mm on center has been a popular US west coast practice to transfer seismic forces from the slab to steel frame. The effects of composite slabs on cyclic seismic behavior of welded steel moment connections received only limited attention before the 1994 Northridge earthquake and most of cyclic seismic connection tests were conducted for bare steel beam-column subassemblies. However, unprecedented and widespread brittle failures were observed in connections of welded steel moment frames in the 1994 Northridge earthquake. Numerous causes were conjectured as potentially contributing to the poor seismic performance of the pre-Northridge steel moment connections [2]. Especially, the majority of reported damages occurred in the beam bottom flange. Other than the flaws recognized in the welding and detailing issues like the use of low toughness welding electrode and notch-like condition present in the bottom flange, the upward shift of the neutral axis due to the unintended composite action between steel beam and concrete floor slab was speculated as one of the critical causes for the prevalent bottom flange fracture. After the earthquake, the effects of a concrete slab on seismic performance were investigated by many researchers. However different observations and conclusions were made among different researchers for slab effect on seismic performance depending upon the details of beam-to-column connections and floor slabs. Recently the authors conducted a close review of extensive relevant experimental
S.-Y. Kim, C.-H. Lee / Journal of Constructional Steel Research 139 (2017) 62–68 Table 1 Test specimens. Specimen
Slab
Retrofit scheme
PN500 PN500C PN500C-HST PN500C-SH PN500C-TH
None Yes Yes Yes Yes
None None Heavy shear tab Straight haunch Triangular haunch
studies conducted after the 1994 Northridge earthquake [1]. Based on the review, the authors were able to draw the following conclusions: i) When connections are composed of conventional welded steel moment connections (similar to pre-Northridge (PN) type) and relatively shallow beams with a high degree of composite action, composite slab effect is quite detrimental to seismic performance because of significant upward shift of the neutral axis under positive moment, often leading to premature fracture of the beam bottom flange [3–5].
63
ii) Composite specimens with deep beams, a low degree of composite action, and connection modification by using RBS (reduced beam section) or haunch showed little or no negative effects due to the presence of a composite slab. The presence of a composite slab did not cause any significant shift of the neutral axis and increased strain demand on the bottom flange. The dominance of the beam bottom flange fracture was not observed. Instead, the composite slab was beneficial and helped improve seismic performance because local and lateral instability of the beams was decreased due to the bracing effect of the slab [6–10]. However, several experiments also showed that composite slab effect could be detrimental to the bottom flange even in RBS connections when composite action is very strong (for example, see [11]). iii) Connection improvement by using modification scheme like RBS or haunch provides a structural fuse at the critical section of the beam. In such connections with a promoted plastic zone, which can help relieve increased strain demand, the slab effect on the bottom flange is expected to be less critical than in PN type connections.
(a) PN500
(b) PN500C Fig. 1. Side view and top view of PN500 and PN500C.
(a) PN500C–HST
(b) PN500C–SH Fig. 2. Details of retrofitted specimens.
(c) PN500C–TH
64
S.-Y. Kim, C.-H. Lee / Journal of Constructional Steel Research 139 (2017) 62–68
Table 2 Measured tensile mechanical properties of steels. Specimen
H-500 × 200 × 10 × 16 (SS400) H-400 × 400 × 13 × 21 (SM490) Heavy shear tab (SN490) Straight haunch (SN490) Triangular haunch (SN490)
Flange Web Flange Web Flange Web Flange Web
Yield stress (MPa)
Tensile stress (MPa)
Yield ratio (%)
Elongation (%)
308 306 373 370 412 375 379 363 367
442 442 523 531 565 545 546 546 550
69 69 71 70 73 69 69 66 67
24 26 29 24 24 23 25 24 23
iv) The bottom flange in a composite beam is always more strained than that in a bare steel beam for a given beam rotation demand, and therefore welded seismic steel moment connections and concrete slab should be designed and constructed to minimize the composite action, if possible, especially when shallow beams and conventional connections are involved. In this study, as an attempt to salvage conventional (or PN type) welded steel moment connections with highly composite floor slabs, several connection retrofit schemes involving welded bottom haunches or heavy shear tab were tested and evaluated based on full-scale cyclic test results. 2. Testing program 2.1. Design of specimens When modifying existing connections for improved seismic performance, the presence of a concrete floor slab often dictates modification in the bottom flange only. The bottom flange modification by using welded triangular haunch is well described in the AISC Design Guide 12 [12]. The bottom flange modification by using welded straight haunch was also proposed based on the analytical and experimental studies by Lee and Uang [13] and Lee et al. [14]. Test specimens were designed to simulate one-sided (or exterior) moment connections. Five specimens were tested in this study as shown in Table 1 and Figs. 1 and 2. The steel beam and column sizes of all the specimens were identical. The grade of steel for the beams was SS400 with a specified minimum yield strength of 235 MPa; the steel used for the columns was SM490 with a specified minimum yield strength was 325 MPa. SN490 or seismic steel recently developed in Korea was used for fabricating haunches and other retrofit steel plates because of its improved weldability. The tensile test results based on the average of three coupons taken from the flanges and the webs are reported in Table 2. Each specimen consisted of H–500×200×10×16 (SS400) beam and H–400×400×13×21 (SM490) column. The beam and column sizes typical of past construction practice in Korea were selected. Especially, relatively shallow beams, 400–500 mm deep, were typically used because all the frames were designed as moment resisting; the use of a perimeter moment frame system is not a popular practice yet in Korea. The beam and the column used in specimens satisfy the seismically compact requirements of the AISC seismic provisions [15]. All specimens were fabricated
Fig. 3. Typical test set-up.
by using all-welded beam-to-column connection per the past practice in Korea. A notch-tough welding electrode with a specified minimum Charpy V-Notch (CVN) toughness of 26.7 J at − 28.9 °C (20 ft-lb. at − 20 °F) and a tensile strength of 490 MPa was specified for flux-cored arc welding in all specimens. Nominally identical 175 mm-thick composite slab with rebars (13 mm diameter @200 in both directions) and 100 mm concrete topping on 75 mm-deep corrugated steel deck was provided to all the specimens except PN500. The average 28-day compressive strength of three cylinder specimens taken from the concrete topping was fc′ = 21 MPa. The flutes of the deck were oriented perpendicular to the beam axis. A pair of shear studs of 19 mm diameter per deck rib (or 200 mm on center) was attached to the beam top flange of all the composite slab specimens following the past practice in Korea. The specified minimum tensile strength of shear stud (Fu) was 400 MPa. Two cross beams (with one row of shear studs of 19 mm diameter 200 mm on center) were also provided to simulate the presence of the gravity beams. The shear studs provided corresponded to 92% fully composite action. The degree of composite action was calculated per Eq. (1) as the shear strength of shear studs (∑ Qn) divided by tensile yielding strength of the steel beam (AsFy). The shear strength of a stud was calculated according to Eq. (2) following the AISC Specification [16]. P The degree of composite action ¼
Q n ¼ 0:5Asa
Qn As F y
ð1Þ
qffiffiffiffiffiffiffiffiffi 0 f c Ec ≤ Rg Rp Asa F u
ð2Þ
where ∑ Qn = sum of nominal shear strength of shear studs, Asa = cross-sectional area of shear stud, fc′ = specified compressive strength of concrete, Ec = modulus of elasticity of concrete, Rg = coefficient to account for group effect, Rp = position effect factor for shear studs, and Fu = specified minimum tensile strength of a shear stud. Table 3 summarizes the relative strength of the beam, the panel zone and the column of specimen PN500 based on the measured yield
Table 3 Relative strength of beam, panel zone, and column of PN500. Beam (SS400) Mpb = ZbFyb
Column (SM490) Mpc = ZcFyc
Panel zone (SM490) Rv = 0.60Fycdctw × [1 + (3bcft2cf / dcdbtw)]
SC-WB requirement ∑Mpc/∑Mpb
PZ requirement Rv/(Mpb/db)
671.4 kN-m
1368.9 kN-m
1400.6 kN
2.04
1.04
Note: All the strengths were calculated using the measured yield strength reported in Table 2; the panel zone strength Rv was computed per Eq. (J10-11) in the 2010 AISC Specification.
S.-Y. Kim, C.-H. Lee / Journal of Constructional Steel Research 139 (2017) 62–68
(a)PN500
65
(b)PN500C
(c) PN500C–HST
(d) PN500C–SH
(e) PN500C–TH
Fig. 4. Photographs showing plastic hinge region after completion of testing.
1.236 0.80
0.80 1.147
with a composite slab have not been tested yet, and limited tests were conducted for triangular haunch connections mostly for deep beams with lightly composite slabs. The motivation of including the heavy shear tab specimen PN500C–HST is as follows. As mentioned previously, when modifying existing connections for improved seismic performance, the presence of a concrete floor slab often dictates modification in the bottom flange only. Strengthening existing beam web by using heavy shear tab can be a viable alternative when even the bottom flange modification is not feasible. As was shown by Lee [17] and Lee-Kim [18], the load transfer mechanism close to the beam to column welded joint is completely different from the engineering beam theory; the flange near the welded joint should transfer a significant portion of the beam shear as well as the horizontal tensile force, and is overstressed. Further, the flange near the welded joint becomes brittle due to the welding heat affection and the tri-axial strain restraint condition there. Thus, the purpose of
Normalized moment at column centerline M/Mpb
Total plastic rotation, ×10-2 rad
1.591 0.80
0.80 1.132
Total plastic rotation, ×10-2 rad
(c) PN500C–HST
Normalized moment at column centerline M/Mpb
Normalized moment at column centerline M/Mpb
(a) PN500
1.546 0.80
0.80 1.118
Total plastic rotation, ×10-2 rad
(b) PN500C 2.013 0.80
0.80 1.458 Total plastic rotation, ×10-2 rad
(d) PN500C–SH Fig. 5. Total connection plastic rotation.
Normalized moment at column centerline M/Mpb
Normalized moment at column centerline M/Mpb
strength of steels. The column was much stronger than the beam, but the panel zone strength including the column flange contribution factor was comparable to the beam strength and thus some limited but varying degrees of panel zone yielding were expected among specimens because of different connection details and different shear demand on the panel zone under cyclic loading. Bare steel specimen PN500 was nominally identical to specimen PN500C except the absence of a composite slab. Both may be thought as benchmark specimens. Triangular and straight haunch specimens were designed according to the procedure in the AISC Design Guide 12 [12] and the procedure proposed by Lee-Uang [13] and Lee et al. [14], respectively. The unique load transfer mechanism of triangular and straight haunch connections and their design procedures for haunch connections are not presented here because of space limitations. Within the authors' knowledge, straight haunch connections
1.840 0.80
0.80 1.254
Total plastic rotation, ×10-2 rad
(e) PN500C–TH
S.-Y. Kim, C.-H. Lee / Journal of Constructional Steel Research 139 (2017) 62–68
0.80 0.80
Total plastic rotation, ×10-2 rad Fig. 6. Comparison of cyclic response of specimens with and without composite floor slab.
strengthening the beam web with heavy shear tab is two-fold; i) to reduce tensile strain demand on the beam flange by increasing the plastic section modulus of the retrofitted beam web, and ii) to push the plastic hinging away from the brittle welding area to the inside “ductile” area. As can be seen Fig. 2(a), the width of the heavy shear tab used in PN500C–HST is about half of the beam depth in order to push the plastic hinging sufficiently away from the column face. The height and the thickness of the heavy shear tab were 370 mm and 9 mm, respectively. As a result of the web strengthening, the plastic section modulus of the web was increased from 25% to 40% of the original beam section (H– 500×200×10×16). The two corners of the heavy shear tab were slightly trimmed to reduce stress concentration there (see Fig. 2(a)). 2.2. Test setup and loading The specimens were mounted to a strong floor and a strong wall. An overall view of the test setup is shown in Fig. 3. The column, which was pinned at the top and bottom, were 3.5 m high. The distance between the column centerline and the actuator loading point was 3.5 m. Thus, the beam tip displacement corresponding to 1.0% story drift was 35 mm. Lateral bracing was provided at a distance of 2.5 m from the column centerline. The specimens were tested statically according to the AISC cyclic seismic loading protocol [15]. The test specimens were instrumented with a combination of displacement transducers and strain gages to measure global and local responses. 3. Test results and discussions 3.1. Overall behavior and plastic rotation capacity
Distance from beam mid-depth, mm
Global behavior of each specimen including plastic rotation capacity and failure mode is described in the below. Photographs of the plastic hinging region at completion of testing are shown in Fig. 4. Fig. 5
0.375% 0.5% 0.75% 1%
Strain, ×10-3 mm/mm
(a) PN500
summarizes the hysteretic responses observed in testing in terms of the total connection plastic rotation. The ordinate of each figure is expressed in terms of the normalized moment at the column centerline. Total connection plastic rotation is calculated with assuming that plastic deformation is lumped at the column centerline. All tested specimens exhibited plastic rotation capacity well in excess of 3% rad that corresponds to the minimum requirement for special moment frame connections. Such an excellent performance may be attributed to a relatively shallow beam depth (500 mm) and all-welded quality welding by using notch-tough welding electrode. Below is brief description of the global behavior of each specimen. 3.1.1. PN500 As can be seen in Fig. 5(a), PN500 was able to develop an excellent connection plastic rotation of 5% rad without fracture. This specimen exhibited the maximum moment of 1.1Mpb at the column centerline for both positive and negative bending. Test was terminated because of severe local and lateral buckling at the plastic hinge (see Fig. 4(a)). 3.1.2. PN500C As mentioned previously, this composite specimen is nominally identical to PN500 except the presence of the composite floor slab. As can be observed in Fig. 5(b), the hysteretic response is asymmetric in contrast to that of PN500. The maximum positive moment observed in PN500C is as high as 1.55Mpb because of the composite action while the maximum negative moment is about 1.05Mpb just comparable to that of PN500. Of course, the large increase of the maximum positive moment due to the composite action can be very detrimental to the bottom flange and is never desirable from seismic perspective. This specimen experienced low cycle fatigue fracture near the heat affected zone at 4% plastic rotation level (see Fig. 4(b)) and was less ductile than PN500. 3.1.3. PN500C–HST This specimen is nominally identical to PN500C except the addition of heavy shear tab to the beam web. As a result of the web strengthening, brittle fracture was delayed and the connection ductility and energy dissipation capacity of PN500C–HST was significantly improved compared to PN500C. The maximum positive moment at the column centerline was about 1.59Mpb. This specimen showed gradual strength degradation after 5% rad plastic story drift, and eventually fractured due to low cycle fatigue at the very high plastic story drift of 7% rad. 3.1.4. PN500C–SH and PN500C–TH Both haunch-retrofitted specimens showed similar and excellent plastic rotation capacity exceeding 5% rad (see Fig. 5(d) and (e)) without fracture. As the photos in Fig. 4(d) and (e) show, the plastic hinging was successfully formed outside the haunch region. However, the maximum positive moments at the column centerline observed in PN500C–SH and PN500C–TH were about 2.01 and 1.84 times Mpb, respectively. Both Distance from beam mid-depth, mm
Normalized moment at column centerline M/Mpb
66
0.375%
1% 0.75%
0.5%
Strain, ×10-3 mm/mm
(b) PN500C
Fig. 7. Strain profiles of PN500 and PN500C under positive moment.
67
Beam tip force, kN
Beam tip force, kN
S.-Y. Kim, C.-H. Lee / Journal of Constructional Steel Research 139 (2017) 62–68
Panel distortion, ×10-2 rad
Panel distortion, ×10-2 rad
Beam tip force, kN
Panel distortion, ×10-2 rad
Beam tip force, kN
(b) PN500C
Beam tip force, kN
(a) PN500
Panel distortion, ×10-2 rad
(c) PN500C–HST
Panel distortion, ×10-2 rad
(d) PN500C–SH
(e) PN500C–TH
Fig. 8. Panel shear distortion of specimens.
haunch-retrofitted specimens showed rather severe pinching under unloading stage toward positive moment at high story drift because of severe flange and web local buckling (see Fig. 4(d) and (e)).
3.2. Comparison of responses and design recommendations Fig. 6 compares the cyclic response of the composite specimen PN500C with that of the bare steel counterpart PN500. PN500C showed seismic performance inferior to the bare steel counterpart PN500 because of low cycle fatigue fracture of the beam bottom flange at 4% plastic drift cycle. In addition, the specimen PN500C exhibited an increased beam flexural strength as high as 1.55Mpb under the positive moment compared to the maximum negative moment of about 1.05Mpb. Such asymmetric topology of the response may lead to the column hinging under positive moment and should be properly considered in design. The flexural strain profiles of PN500 and PN500C under positive bending are depicted in Fig. 7. It is observed that the neutral axis of PN500C was shifted upward by about 10% of the beam depth compared to the bare steel specimen PN500. Fig. 8 compares the panel zone shear distortions measured in the test. In the two specimens PN500C and PN500C-HST, the panel zone shear distortion is increased asymmetrically or further increased under the positive moment compared to the bare specimen PN500. This implies that neglecting the presence of composite floor slabs can lead to the underestimation of panel zone shear distortion. In the haunch specimens, the panel zone shear distortion is significantly reduced, especially in the straight haunch specimen. This reduction is due to the effect of enlarged (or dual) panel zone, a side benefit of haunch retrofit. Please refer to Lee-Uang [13] and Lee-Uang [19] for the behavior and mechanical modeling of dual panel zone of haunch connections.
As can be observed in Fig. 5, asymmetric topology of hysteretic response is very pronounced in all composite slab specimens. The maximum positive moments observed in heavy shear tab and straight or triangular haunch specimens were about 1.59, 2.01 and 1.84 times the bare steel plastic moment, respectively (see Fig. 5(c), (d) and (e)). Such increase of the positive moment as a negative effect of retrofit could violate the strong column-weak beam condition and should be considered in retrofit design and analysis. The positive and negative strain-hardening factors observed in this testing program are summarized in Table 4. Although test data is limited, simple guides for sizing and detailing of heavy shear tabs are illustrated in Fig. 9 based on successful test results of specimen PN500C–HST. Note that the web plastic section modulus of PN500 and PN500C–HST was respectively 25% and 40% of the plastic modulus of the whole beam section. Retrofit design by using triangular and straight haunches can be found in AISC Design Guide 12 [12], Lee-Uang [13], and Lee et al. [14]. 4. Summary and conclusions This study investigated seismic retrofit schemes for PN-type welded steel moment connections with highly composite floor slabs by using the beam bottom flange or beam web modification. The results of this study can be summarized as follows. i) This study again confirmed that high composite action of a floor slab is detrimental to seismic performance of PN-type welded steel moment connections; the almost fully composite PN type specimen showed seismic performance inferior to bare steel counterpart and eventually experienced to the beam bottom flange fracture.
Table 4 Positive and negative strain hardening factors observed. Computed at the column face
Positive strain hardening factor α+ Negative strain hardening factor α−
Computed at the tip of strengthened zone
Bare steel connection
Composite connection
Heavy shear tab
Straight haunch
Triangular haunch
1.165 1.082
1.458 1.054
1.386 0.987
1.610 1.166
1.472 1.003
68
S.-Y. Kim, C.-H. Lee / Journal of Constructional Steel Research 139 (2017) 62–68
CJP s
thst tw
s=thst-2 (mm)
Existing bolted shear tab CJP
db
s
s=thst-2 (mm) db thst tw
0.5db
tw Trim to reduce stress concentration
(a) All-welded
0.5db
tw Trim to reduce stress concentration
(b) Bolted web and welded flange
Fig. 9. Sizing and detailing of heavy shear tab recommended.
ii) All the three retrofitted specimens by using heavy shear tab and triangular or straight haunches effectively pushed plastic hinging outside the strengthened region and exhibited excellent total plastic rotation capacity exceeding 4% rad. Although the test data of heavy shear tab retrofit is very limited, simple retrofit design guides are also recommended. iii) Generally, asymmetric topology of hysteretic response was very pronounced in the highly composite slab specimens of this study. The maximum positive moments observed in heavy shear tab and haunch specimens were about 1.6 and 2.0 times the bare steel plastic moment, respectively. This side effect should be considered in retrofit design when checking the strong column-weak beam condition as well as in seismic performance evaluation of the retrofitted frames. iv) The heavy shear tab retrofit tended to increase the panel zone yielding compared to PN-type connections. However, haunch retrofit significantly reduced the panel zone yielding, less than PN-type counterparts, because of the enlarged (or dual) panel zone effect. Both triangular and straight haunch specimens exhibited comparable seismic performance. They showed increased pinching under unloading stage toward positive moment at high story drift because of severe flange and web local buckling. Acknowledgements This research was supported by a grant (17AUDP-B066083-05) from Architecture & Urban Development Research Program funded by Ministry of Land, Infrastructure and Transport of Korean government. The Institute of Engineering Research at Seoul National University provided research facilities for this work. References [1] C.H. Lee, J.H. Jung, S.Y. Kim, J.J. Kim, Investigation of composite slab effect on seismic performance of steel moment connections, J. Constr. Steel Res. 117 (2016) 91–100 (Elsevier). [2] M. Bruneau, C.M. Uang, A. Whittaker, Ductile Design of Steel Structures, McGrawHill, New York, 2011.
[3] R. Leon, J. Hajjar, M. Gustafson, Seismic response of composite moment-resisting connections. I: performance, J. Struct. Eng. ASCE 124 (1998) 868–876. [4] J. Hajjar, R. Leon, M. Gustafson, Seismic response of composite moment-resisting connections. II: behavior, J. Struct. Eng. ASCE 124 (1998) 877–885. [5] Y.J. Kim, S.H. Oh, T.S. Moon, Seismic behavior and retrofit of steel moment connections considering slab effects, Eng. Struct. 26 (2004) 1993–2005 (Elsevier). [6] C.M. Uang, Q. Yu, S. Noel, J. Gross, Cyclic testing of steel moment connections rehabilitated with RBS or welded haunch, J. Struct. Eng. ASCE 126 (2000) 57–68. [7] S. Civjan, M. Engelhardt, J. Gross, Retrofit of pre-Northridge moment-resisting connections, J. Struct. Eng. ASCE 126 (2000) 445–452. [8] S. Civjan, M. Engelhardt, J. Gross, Slab effects in SMRF retrofit connection tests, J. Struct. Eng. ASCE 127 (2001) 230–237. [9] S. Jones, G. Fry, M. Engelhardt, Experimental evaluation of cyclically loaded reduced beam section moment connections, J. Struct. Eng. ASCE 128 (2002) 441–451. [10] X. Zhang, J. Ricles, Experimental evaluation of reduced beam section connections to deep columns, J. Struct. Eng. ASCE 132 (2006) 346–357. [11] S.J. Chen, Y.C. Chao, Effect of composite action on seismic performance of steel moment connections with reduced beam sections, J. Constr. Steel Res. 57 (2001) 417–434 (Elsevier). [12] J.L. Gross, M. Engelhardt, C.M. Uang, K. Kasai, N.R. Iwankiw, Modification of Existing Welded Steel Moment Frame Connections for Seismic Resistance (Vol. 12), AISC, Chicago, Illinois, 2003. [13] C.H. Lee, C.M. Uang, Analytical modeling and seismic design of steel moment connections with welded straight haunch, J. Struct. Eng. ASCE 127 (2001) 1028–1035. [14] C.H. Lee, J.H. Jung, M.H. Oh, E.S. Koo, Cyclic seismic testing of steel moment connections reinforced with welded straight haunch, Eng. Struct. 25 (2003) 1743–1753 (Elsevier). [15] AISC, Seismic Provisions for Structural Steel Buildings, AISC, Chicago, Illinois, 2010. [16] AISC, Specification for Structural Steel Buildings, American Institute of Steel Construction (AISC), Chicago, Illinois, 2010. [17] C.H. Lee, Review of force transfer mechanism of welded steel moment connections, J. Constr. Steel Res. 62 (2006) 695–705 (Elsevier). [18] C.H. Lee, J.H. Kim, Seismic design of reduced beam section steel moment connections with bolted web attachment, J. Constr. Steel Res. 63 (2007) 522–531. [19] C.H. Lee, C.M. Uang, Analytical modeling of dual panel zone in haunch repaired steel MRFs, J. Struct. Eng. ASCE 123 (1997) 20–29.