Eethavae Engineering, Ten Weis Contarence 1852 Gahoma, tera SBN 9064100608 Cyclic tests on large scale steel moment connections Michael D-Engelhardt The University of Tetas ar Austin, Te., USA ‘Abunnasr S. Husain DeLeuw Cather and Company San Francisco, Cali, USA ABSTRACT: Cyelic loading tess were conducted on eight large scale welded fange-bolted web type steel moment connections. The objective was to investigate the influence of various we connection details on the overall connection performance. The eight specimens showed highly variable performance, and developed plastic rotations that were judged #8 poor to marginal for severe seismic applications. Va lity ia the performance of the beam flange groove welds dominated the response of the specimeas. This paper summarizes the test program and compares results to those of previous US. test programs. Concems afe expressed about the reliability ofthis conection deta for seismic applications. L INTRODUCTION ‘The beam-to-column connection is ertical element of a steel moment resisting frame. The most widely used moment connection detail in curent U.S. practice for seismiccesistant framing is the welded flange-bolied ‘web detail. The beam flanges are welded tothe column using complete penetration siagle bevel groove welds, ‘with the welds typically made in the feld using the selfshielded flux core ate welding (FCAW) proces ‘The beam web is field bolted oa single plate shear tb whieh is shop welded to the column. Recent tests by Tsai and Popov (1988) bave indicat. ced that this detail may not provide satisfactory perfor ‘mance when used for beams in which the web accounts {ora substantial portion ofthe beam’s flexural strength This observation has been attbuted tothe inability of| the bolted web connection to transfer bending moment, resulting in exeessive demands on the beam flange connections. As a result of this concern, recent US. seismic code provisions (Recommended lateral force requirements 1988, Seismic provisions 1990, Unifocm building code 1988) require that in addition to web bolts, supplemental welds be provided between the ‘beam web and the shear tab for beam sections with ZL = 0.70. Z is the beam's plastic section modules, and Z, is the plastic section modulus of the beam anges only. The supplemental wed welds must be Sesigned to develop at least 20 percent of the beam ‘web's flexural strength. These provisions recognize that the web connection must be capable of caring not only shear force, but also portion ofthe beams bend- ing moment. These new provisions are based on rather rather limited data, and according to the Recommended nel freequemens (1960, were con “un aon! de abies ‘a exerinena avetgnton was undeaen to colts dat oo eect tte Bf ta fod the web conection dea Te eae ove ct Sheree spotor wis determi fein of te mplemchal we weldgTequemens Cul Ge Jes based sional txpetinenal da, Tas Pepe snares ult ts exetinental pops, 2 TEST PROGRAM ‘Tests were conducted on cantilever type test specimens in which a beam was connected tothe flange of 3.65 m long pin ended column, Cyclic loads were applied the Beam ata distance of 244 m from the clumma face. ‘Atotal of eight specimens with diferent beam sections and web connection details were used, as summarized in Table 1, All eams were of A36 stel (specified F, + 250 MPa). W12e136 sections of AST2 Gr. 50 steel (pecified F, = 345 MPa) were used forthe columns of All specimens. This section was chosen to provide @ very strong panel zone, so that inelastic deformation of the specimens occured primarily as flexural yielding of the beam rater than shear yielding ofthe column panel ‘Connection details fortwo ofthe test speclmens are iustated in Figures 1 and 2 Similar details were sed for all specimens. Complete penetration single bevel [groove welds were used to connect the beam flanges to the columa flange in all specimens. This weld was etailed with © 10 mm root opening and 2 30 degree bevel. Backup strips were used and remained in place 2385‘Table 1. Tes Specimens Seee_Beam _Z/Z_Web Connection _ TT Wau Ost 6 ADS ols 2 Wass 061 6A bots 3 Wass 061 6455 tn + Sea 4 Mame 07s eo Asets 5 Wie 075 6. As2stots 6 womst 067 $4525 es 7 wast 067 5402 ata St ewan Gt ali wer 9 Ue fe is sxe 2 whee ee | fe ances Figure 1. Coonestion deals for Specimen 4, W28(G 501 Taare onsale | i seu { a 4 ways? Ase pe > mie Figure 2. Connection deals foe Specimen 7. after the welds were completed. Welding was by set. Shielded FCAW proces, using a0 E7OE7, 24 mm diameter electrode (specied Fy = 485 MPa). "The wed connection details (Table 1) included at bolted connestion, bolted eonsectons with supplemen- ta we welds, and an all welded web connection. All bots were 22 mm diameter, grade A225 (specified F, 2350 MPa) or grade A490 (epecfied F, = 1035 MPs) The AIS bolt were tensioned 10 182 KN and the ‘A490 bots 10-227 RN, using a carefully callie torque control method. For each ofthe beam sections tse ia the lst program, iheee 22 mm diameter A325 bolts would have been sufcient to satisfy connection shear strength requirements ofthe Recommended ata force requirements (1988). It was judged, however, that 2 larger numberof boli would be used for these beam sections in curent California praccs. The numberof bolts was therefore increased ab indicted in Table 1 more closely reece current detain practice. or the specimens with supplemental wed welds (Woe. 3 aad 7), these welds were sized to provide 4 nominal Srength of 20% of Mya Myay Was compat fdas 280 MPa"(J-2:)?*4/4, where d and are he veal beam depth, ange thickness and web thickness Weld seageh wes compued using 17 times te usual allowable sess, per the Recommended lateral fore Fequitements (1988). The supplemental web welds were plhotd atthe top and bottom edges of the shat 2d Gigue.Z) to provide the maximum Hexural wes onzesion capacity with the smallest possible wells. For Specimen 8 the beam web was directly welded to the column fange using a single bevel complete penetration groove wel with thee AS25 bolts provi 410 simulate ecection bolt. Allewelded connections Sue as Specimen 8 have sown xclen efor in past tet programs (Popov and Stephea 1972, Ts tod Popo 185 OR Oe Si ‘A specimens were provided with 13 mm thick continuity plates, as showa in Figures land 2. Accord ing to the Recommended lateral force requirements (0988), continuity plates were not required for any of thetest specimens. These plates were provided, bowev= 10 eliminate the possible inuenes of loeal column ‘Benge beading as an addtional variable affecting test results. No doubler plates were cequced for any ofthe test specimens, and none were provided. “Tne connection design and dealing for Specimens 3, 4,5, 7, and 8 met or exceeded all current US. Seismic code requirements. Specimens 1, 2 and 6 intentionally violated code requirements for supplemen- fal web welds, but satisfied or exceeded all other requirement. "The fabsication sequence was denial fr al speci- sexs. Fist, the comity pats and shear tb were ‘welded to the column. Bolts were then installed and filly tensioned. The top beam Bange was then welded, followed by the botom flange. After the beam fange welds cooled, the supplemental web welds were made on Specimens 3 and andthe complete web weld wasmade on Specimen 8 All groove welds were made with the specimens in an upright position to simulate field ‘welding pastions. The fabrication and welding se- uence were chosen to simulate typical US. field onstrection p “All specimens were constructed by = commercial scuctural steel fabricator. Welders were certified per the Structural welding code (1988). Inspection was provided by a commereal welding inspection frm and Fluded ultrasonic testing of the column Manges for laminations prior to fabrication, 2 pre-weld fitup inspection, and ultzasonic testing of all complete penetration groove welds. The individual conducting the Inspections had all appropriate certifications from the “American Welding Society and the American Society for Nondestructive Testing. The ultrasonic inspetions revealed a defect in a groove weld of Specimen 3, ‘which was then repaired and successfully retested ‘lrasoncaly. 3 TEST RESULTS Each specimen was subjest to cyclically increasing displacements atthe tip ofthe cantlever. When sp smens entered the inelastic cange, yielding was typically Concentrated inthe besm Manges. Specimens wi bolted webs only (Nos. 1, 2, 4, 5, and 6) showed virtually no yielding inthe beam web, reflecting the lack of flexoral participation of the web connection. Specimens with web welding (Nos. 3, 7 and 8) typial- ly showed significant beam web yielding. Some speci mens also showed minor yielding inthe column panel ‘Cyclic loading was continved for each specimen tnlit connection failure. Failure for all specimens fceurted as fracture ata beam flange groove wel. For Specimens 1, 2, and 4, failure occurred unexpectedly teary in the foading program. Fer each of these three specimens, the beam had barely entered the inelastic range when sudden fale occurted at the bottom ‘ange weld. Complete facture occurred nearly instanta- ‘neously at the interface of the weld and the column ‘ange. For each ofthese specimens, afte failure atthe bottom flange occurred, the load was reversed for one final, monotonic half-cycle of loading, which was continved nti failure of the top lange cecurred. The top flange failures occurred by a gradual tearing ofthe beam flange well outside of the groove weld. The specimens all developed substantial duelity dring this final monotonic half-cyce of loading. tn fact, it was ‘ot possible to fail the top flange of Specimen 4 before running out of stroke on the hyéraule loading ram. “The hysteretic response of Specimen 4is stated in Figure 3. Results are ploted as moment versus plastic rotation. Moment as computed a the load Imultiplied by the distance to the face of the columr. Rotation was computed asthe tip displacement divided by the distance to the face of the column. Plastic was obtained by subtracting out the measured ‘portion ofthe rotation. This method of comput- Figure 3. Hysteretc response of Specimen & (1 kip-in = .113 kN-m). Figure 4. Hysteretic response of Specimen 7 (Ukip-in = 113 EN), ing plastic rotation is consistent with previous experi mental investigations, and therefore provides mean- Inf basis for comparison "The dotted lines in Figure 3 represent the response during the final monotone half-ecle of loading, after failure atthe bottom flange connection had occurte. “The remaining specimens (Nos. 3 and 5-8) showed somewhat beter performance. These specimens all failed by the development of + facture ator near the interface of the groove weld and the beam flange. ‘These fractures generally iniited atthe edge of the ‘beam flange, and gradually spread across the width of the lange as the loading progressed. In many cass, the fracute occured atthe interface of the weld and the beam flange, ronaing along the bevel inthe ange. igure 4 displays the byserti response of Speci= men 1 Feiloce ofthis specimen occurred at the top ange weld in the manner described above. The dotted fine present the specimen’s response during the final ‘monotone balf-cycle of loading after failure ofthe top flange connection had occurred. This final monotonic Toading was continued unt filire also occurred atthe bottom ange connection. ‘Of he eight specimens tested inthis program, Speci ren 7 provided the best performance in ters of plastic rotation, whereas Specimen 4 (Figue 3) provided the worst peeformance. 2887Table 2. Summary of plastic rotations and failure modes. Spee 6, Failure Mechanism (at 1 0.004 sudden fracture at weld-columa Interface at bottom flange 2 02003 sudden fracture at weld-column Interface at bottom flange 3 02009 gradual fracture through bottom beam flange, ouside of weld 4 0.002 sudden fracture at weld-column imerface at bottom flange 5 0013 gradual fracture at weld-beam Interface a bottom flange © 0.013 gradual fracture at weld-beam interface a bottom flange 7 O15 gradual fracture at weld-beam Interface a tp flange 8 0.012 gradual facture at weld-beam Interface a bottom flange ‘Teble 3. Maximum moments. Spec. Man MM," M,, Mead,” 1 612 55 6812088 2 610 S45 641128 3 72 SS 6h 1331.04 4 540 500 62710886 So 500 67134107 6 664 5508 Tt 7 716 sz 59413712 8 TT Sh 584140128 Notes: moments in kN-mt ‘M.., = maximum measured moment in beam at column face, prior to connection flare 1M, = beam plastic moment based on nominal specitied dimensions and yield stess My = beam plasde moment based on measured dimensions and yield stess ‘Table 2 summarizes the plastic rotation, 8,, and xe mechanism for each test specimen. The tabulat- ced value of 6, is the maximum plastic rotation attained before failure, as measured from the beam’s original ‘undeformed position. Most specimens were loaded until fracture occurred at both the top and bottom flange connections. However, the point at which the first ange connection fractured is taken as the actual “fallote™ foreach specimen, since beyond this pont, the ability to resist eyelic loading is exhausted ‘The maximum moment, Mya, developed by the beam atthe column face prior {6 connection flue is listed in Table 3. For comparison, the nominal specified plac moment My and the eximated ata lsc moment, M,., are siso listed, In the final monotonic cycle afier thre, ‘some specimens developed bending ‘moment substantially higher than the value of Moy listed in Table 3. 4 DISCUSSION OF TEST RESULTS 4.1 Weld performance ‘As described abave, the failures of Specimens 1,2, and 4 occurred unexpectedly early in the cyclic loading program. These three specimens failed ina neatly Identical manner: a sudden fracture atthe interface of the column and the groove weld atthe beam’s bottom flange. Examination of the failure surfaces revealed 2 small region of incomplete fusion nes the lower central portion of each of the welds, suggesting poor welding the cause of the failures. Further investigations and concullaion with welding specialists supported this ‘observation. A different welder was used for Specimens 5 to 8, and similar premature failures were not ob- served. ‘As described earlier, failures in Specimens 5 to 8 were generally observed at the weld-beam interface ‘These failures occurted after the beams had achieved substantially higher moments and plastic rotations as ‘compared to Specimens 1, 2, and 4. Clearly, the welds ‘were of higher quality in these later specimens. None- theless, subsequent studies of these failures indicated that these welds likely did not develop the fll tensile strength of the beam flanges. Thus, the groove welds for Specimens 5 to 8 also do not appear to have per {formed ina completely satisfactory manner. There were fo obvious defects in these welds, and possible cause of theie somewhat unsatisfactory” performance is sti ‘under investigation. Tes important to recall tha all test specimens inthis program were constructed by a commercial structoral sel fabricator. All welders were certified, and all _roove welds were ultrasonically tested by an indepen- dent welding inspection firm. Nonetheless, a number of ‘welds were found to be inadequate when the coanec- jons were tested to destruction. This suggests the possibilty that current industry practices in the USS. for ‘welding and quality control may not be adequate for this moment conneetion detal when subject to large cyclic deformations. 42 Strength and plastic rotation capacity Based on the data in Table 3, all specimens devel- 2008g eam Plas Rotation oie? (©) Specimens with 24/2 « 0.70. Figure 5. Comparison of beam plastic rotations with ast tests ‘oped their nominal specified plastic moments. Further, all specimens, except Nos. 1, 2, and 4, developed theit estimated actual plastic moments. The detrimental effect of the flange weld failures on the strength of Specimens 1, 2, and 4is clear from Table 3 ‘A few comparisons between the performance of specimens withthe same Z/Z ratios can also be made. ‘The most interesting trend in behavior ean be observed by comparing Specimens 6 to 8. These specimens were provided wit increasing degrees of flexural capacity in their web connections. The data in Table 3 show increasing strengths progressing from Specimen 6 to 8 This suggests that when reasonably good quality lange welds are provided, the web connection details may havea significant effect on overall connection strength. criteria for judging the performance of ens was thelr plastic rotations. A value fof 20.015 radian was chosen a a reasonable estimate ‘of beam plastic rotation demand in steel moment resisting frames subject to severe earthquakes. This is based on a review of recent inelastic dynamic frame analyses and appears to be consistent with criteria used in past test programs (Popov et al 1986, Tsai and Popov 1988). At joints in which the colume panel zone ‘an effectively take pat in developing inelastic defor- mations, the plastic rotation demand on the beam can be relaxed (Tsai and Popov 1988). Based on the citerion of 20.015 radian of bear Plastic rotation at the connection measured from the ‘original undeformed postion, the performance of the ight specimens was Unsatisfactory overall. Based of the data in Table 2, the performance of Specimens I, 2, and 4 was particularly poor. The other specimens performed significantly better. However, Specimen 7 was the only one to achieve 20.015 radian of plastic rotation. All others were judged as marginally accept- abl, with plastic rotations varying from 20.009 radian 10 200013 radian. Although the test results showed some effet of web connection details on strength, there is no clear evi- dence from the test data thatthe 2/2 rata or the web connection detils hed a significant influence on uctlty. Specimens § and 6, for example, both had bolted webs, but had different Z/Z ratios (0.75 for Specimen 5 end 0.67 for Specimen 6). Yet, the plastic ‘rotation was nearly identical fo these two specimens. Compare also Specimens 6, 7, and 8. These specimens had the seme Z,/Z ratio, bu had increasing degrees of ‘web participation. Specimen 7 with the supplemental web welds did develop greater plastic rotation than Specimen 6, which hed only s bolted wed. However, Specimen 8, witha fully welded web, developed less plastic rotation than elther Specimens 6 or 7. ‘Overall, variability inthe performance of the beam ‘ange welds appears to heve had a much greater {nglience on plastic rotation capacity than ZZ rato or web connection det. This vaieilty is clearly evident ‘when comparing Specimens 4 and 5. These specimens had nominally identical beam sections and connection details. Yet, Specimen 5 achieved approximately 6 times greater plate rotation, 5 COMPARISON WITH EARLIER TESTS Results ofthis investigation were compared with results of teats conducted by Popov and Stephen (1972), Popov et al (1986), and Tsai and Popov (1988). Although tis is not an exhaustive database, i is believed to represent 4 significant portion of eycic tests performed on large seale specimens in the US, Figure 5 shows beam plastic rotations achieved by specimens in these test programs, divided according 9 Z,/Z ratio. Only bearn to column flange connections with bolted webs, with ot without supplemental web weld, ae included. Speci ‘mens with supplemental web welds are highlighted. The beam plastic rotations as shown in this figure were calculated as half of the maximum plastic rotations measured from the total width ofthe hysteretic loops, fas 8 basis for comparing results ftom many different 2369tests, The tests by Popov et al (1986) included signiti- ‘cant panel zone deformations. However, only the rotation atteibutable to the beam is included in the bar charts in Figure 5. observations can be made from the data in 1. The results of the current test program, in terms of the magnitude and variability of plastic rotations, are similar to previous test programs. 2 There is a large variability in performance of test ‘Specimens in both the current test program as well as in previous test programs. 13.A significant numberof specimens have not achieved plastic rotations of 15 radian. A rather large number hhave not even achieved .005 radian. 4. Specimens with Z/Z > 0.70 have shown beter performance, on average, than those with Z,/Z 0.70. Both groups of specimens, however, show variable performance. '. Specimens with supplemental web welds (all have been on beams with Z/Z = 0.70) have performed ‘somewhat beter on average than their counterparts with ‘no web welding. There is again, however, considerable variability. 5 CONCLUSIONS ‘The results of this test program revealed no clear influence of the 2/2 ratio or web connection detail on the performance of welded ange-bolted web moment connections. Rather, variability in the performance of the beam flange groove welds dominated the response of the specimens. One of the objectives of this invest- ‘ation was to determine if current U.S. seismic code Fequirements for supplemental web welds could be relaxed. Although the influence of web welds was ‘masked by variability in the flange welds, specimens ‘with the supplemental web welds performed somewhat ‘better than their counterparts without web welds. Consequently, no relaxation in current code require- ments is recommended. ‘An important observation from this and previous test programs is that the welded flange-bolted web connec- tion detail has shown highly variable performance. ‘While some specimens have performed satisfactorily, a significant number have demonstrated poor or marginal performance in cyclic test programs. Some of variability can be attributed to the influence of Z/Z ratio and web connection details. However, a great deal ff the variability also appears to be related to the performance of the beam flange groove welds. ‘The final recommendation of this investigation calls for 8 thorough review of current US. industry practice for seismic steel moment connections. The results of this and previous tests programs leads to questions on the reliability of the current welded flange-bolted web 2890 etal for severe seismic applications. A careful review of design and detailing practices, as well as welding and quality contol issues is needed. ACKNOWLEDGEMENTS Financial support of this work was provided by the Steel Commitiee of California and is gratefully ac- knowledged. The assistance of the project's technical advisory committee is appreciated. Contributions of ‘materials from Bethlehem Steel and Nucor-Yamato Steel are also gratefully ecknowledged. The views expressed inthis paper are solely those of the authors. REFERENCES Popov, E.P., Amin, N.R., Louie J, and Stephen, RM. 1986. Cyclic behavior of large beam-columa assem- blies. Engineering Journal. First Quarter. American Institute of Stee! Construction Inc. Popov, E.P. and Stephen, RM. 1972. Cyclic loading of fullsize stel connections. Bulletin No. 21. Ameri= can Iron and Steel Institute. Recommended lateral foree requirements and tentative commentary. 1988. Seismology Committe, Stuctur- al Engineers Assocition of California. Seismic provisions for structural steel buildings, load and resistance facor design. 1990. American Insti- tute of Steel Construction, Inc. Chicago, Iinois. Siructural welding code ~ steel, AWS D1.1-88. 1988. American Welding Society. Miami, Florida. ‘Tsui, KC, and Popov, EP, 1988, Steel beam-column Joins in seismic moment resisting frames. 1988, ‘Report No. UCB-EERC 88/19. Earthquake Engineer ing Research Center. University of California at Berkeley. Berkeley, California, Uniform building code. 1988. Intemational Conference of Bullding Orficiats, Whittier, California.