ir —_—_ ACHIEVING DUCTILE BEHAVIOR OF MOMENT CONNECTIONS The results of recent tests suggest that weld metal toughness holds a key to achieving better moment connection performance in earthquakes ‘Structural Systems (ATLSS) at Lehigh University, are examin- ing, among other issues, the effect of weld metal toughness fand fabrication defects on the Seismic performance of moment fonnections. ‘The programs include: (1) examination of the erack ur faces of tome damaged connec tions removed from buildings in the Los Angeles area as well as af laboratory tested connections (2) dynamie tension tests of sim- ulated beam flange-to-column Flange connections; and (3) dynamic eyele tosts of large-size Figure I: above) Fracture By Erie J.Kaufmann, _beam-to-column connections. turface of column divot type Ming Xue, Le-WaLwand This paper is a brief resort of fracture. Tue brite fracture ‘John W. Fisher some of the findings of these Initiates from a flaw atthe weld Studtes, based on tests ofa limit- ‘root caused by inadequate root ‘uci HAs BEEN whirrEN ed number of weld metals. mnetraton and the Back-up bor BOUT THE DAMAGE 70 tack of fasion ° SONMBCTIONS IM ELE FaacTune ANALYSES Suvi ronnser rans observed Detailed failure analyses of Figure 2: opposite top) later the 1994 Northridge earth- nearly 20 fractures removed ‘Simulated beam flange to quake, Preliminary investiga. from damaged bulldings have ‘column laage test specimen tions of the fractured connee. been performed. These have tions, which were carried out included the most common types Figure &jopposite middle) Weld soon’ after the earthquake, of weld and column fractures ‘metal charzy Vinotch test results revealed in most eases very Tow reported in building surveys, In fracture toughness of the weld all fractures of these typos that Figure 4 opposite bottom) metal used in mnaking the lange Were examined by the aathors, Dynamic load vs.crosshead __welds.as welllas defects and the origin of the fracture has displacement for simulated beam been identified with areas of Flange a calumn tests ‘back-up bar. The mechanical and inadequate weld root penetration chemical properties of the struc: arising from entrapped slag or tural steel, however, did meet porosity at the back-up bar. the ASTM requirements, Several Invariably, these have boon esearch programs, which are observed near the midlength of ongoing in the Center for the weld where the weld passes ‘Advanced Technology for Large Uhrough the web cope in bottom 0 Moor Se! Contraction Sanry 1098flange connections and where weld discontinuities are more prevalent, Figure 1 shows the rack surfice of a fracture which resulted in a column divot type of fracture, The inadequate root penetration flaw from which the brittle fracture initiates is clear ly seen, Similar root flaws were also observed to be the origin of Traccures which propagated close to the fusion line of the weld or propagated across the column ange. A common factor in all of theee fractures, rogardloss of the path of erack propagation, has been the initiation of cleavage fracture from the weld root adja cent to the back-up bar. The ther common factor has been that the welds were deposited ‘with E70T-4 electrodes n com: ‘mon uso prior to the earthquake Charpy Venoteh torts of weld mnetal Yom fractured eonnections have shown this weld metal to have very low fracture resistance (lose than 10 ft-lbs. at 70 degrees F). This combination of factors is consistent with frac tures that hve occurred in other "Muetures in the past where smn ilar geometrie conditions and low toughness weld metal were both present. SiuLarED BEAM FLAN ‘Cotumy Planer Tests ‘To study the effects of materi a property and fabreation vart- fables om connection performance ‘without the complexity and high Cost of large-scale testing, the Small-scale tension specimen Shown in Figure 2 was devel- oped. The intent of this specimen twas to study aspects of eonnoe- ton behavior and the relative fffecte of individual variables, fand identity conditions whieh improved or deteriorated weld joint performance. A pilot study hhas utilized this specimen to study the effects of beam and col- ‘umn material properties, filler metals, and weld procedures on ‘moment connection performance. In addition to material and fabri. cation variables, the effect of Strain rate was also studied by Conducting teats both statically Both sel ste Tack ea ee u 2 1 = eros \ ie | gen ag cone ange ove TENPERATURE-e ce owen. a T T Taal T 2 gL 2m 4M yg ee ite sn 3 5 120 t 180 g 4 g dm foie * q s g qe 8 2 ob a 8 2 ce cleans 2 cso nea fee ay pee a TEWPERATURE,“F aaa asp emt gauew — [wonuren 400 [ 2 500 2 0 H § 0 0 4 ComeeahiOdermeeeON yt" 4 2p Shee DISPLACEMENT Us) © e ‘Matern Sel Canstructian January 1998/81Pleaso circle # 87 ea smnropaiaecar ieee : opr ln myn Be iechii wdintecioe ‘RRA Ash oibal nods onl ecraye 2 soa yt E)-E, INDUSTRIES, INC. Please circle # 46 (a strain rate onthe orderof 10% 10" sec')and at dynamic strain ‘ates (a strain rate on the ordor of 10° sec") similar to earth: quake conditions. A total of 15, tests were conducied inthe pilot Study. Several of the tests were ‘designed to parallel the weld Joints tested In full-size connec: tion tests in order to compare their behavior ‘The test specimen, shown in igure 2, consist of @ section of Wide176 (A572 Gr. 50) with one ange removed. A pull plete was welded to the web of the section for gripping in @ universal test machine, A Gin. x 16in.x Lin, (A36 or A572) plate was then groove welded to the eolumn Fange face to simulate the beam flange connection. A coped web plate was also added to intro- fluce welding access restrictions similar to welding the bottom ange of «connection. This plate was removed hefore testing to permit installation of a strain gauge at the beam mid-wideh ‘lose to the weld Test specimens were febricat- ed using threo differen: fillor metals (B70T-4, EOTG-K2 and ETOLS). A 0,120-in.-diemeter ET0T-4 electrode wats used sim Tar to the electrode commonly tiged in moment connections Welds were deposited with this clectrode using both manufactur fre recommended welding para tmoters and parameters that pro Guced high heat input deposits ‘ilar to those observed in humerous fractured eonneetions In buildings. The ET0TG-K2 electrode (in. diameter), a ‘candidate substtate for ETOT-, hhas a minimum CVN toughness requirement of 20 Rlbs. at -20 degrees F, which is similar to ETLT-S Giller metal now used in many connection repairs. Test Specimens were also prapared Using the shielded-metal-are (SMAW) process with an ETO18 electrode. Figure 3 shows the ‘wide range of Charpy V-notch toughness obtained forthe three Biller metals. The Tow noteh toughness measured for the ETOT-¢ weld metal (approxi-mately 10 ft-lbs. at 70 degrees Fy is consistent with CVN tough ness of weld metal meastred in fractured connections in build ings. Lite difference in tough ees was measured in welds deposited using manufacturers recommended parameters and parameters which produced high Reat inputs, There are no AWS. noteh tough requirements for Br0r- electrode: The BTOTG-K2 lectrode provided improved toughness although substantial Tess than E701 In addition to varying filler metal in the test bpecimen, testa were also per formed with and without the back-up bar, using end dams instead of proper end tabs, and using beam flange plates with Yield strengths ranging from 40, esi to 58 kes, A complete report deseribing the results ofall test willbe issued in the near future Figure 4 shows load vs crosshead displacement traces for three specimens fabricated using (1) “pre-Northridge’ fabr- cation practice (E701-4 filler metal "with back-up bar Fetained), (2) pre-Northridge practice retrofitted hy removing the back-up bar and adding fillet reinforcements to the weld root fnd weld te with a noteh tough tlectrode (67178), and (3) using BIOI8 electrode alzo with the back-up har removed. All three font apacimens wore tated under ‘dynamic strain rate conditions Brittle fracture developed in the "pre-Northridge” specimen igure 4a) shortly after the beam plate yielded. The A36 beam plate in these teste had s yield strength of 40 ksi. The frac ture erin was identified at the Weld root adjacent to the notch introduced by the back-up bar at a loestion with inadequate root penetration (see Figure 5) Figure 4b shows the result of an identically fabricated specimen with the back-up bar removed. Brittle fracture again occurred although after greater plastic deformation of the beam plate ise Figure 6), In this ease the Fracture initiated within the weld metal and not close to the Coxtisven Ow Pact 96 Figure 5a (top) “Pre-Northridge” simulated ‘ar flange fo column test, Fracture Figure 5b: (middle) ‘The fracture surface shows the origin fs a well ro defo (arrow at back-up ba Figure 6: (eft) Retrofitted “pre Northridge test specimen fractreTable 1: Connection Details of eam and Column Yest Assemblies] farw.(%GMl dimensions, but i Backupfur Bearish Coninuiy | Beam tothe column. Spocimen Soccimen WoidAttal Stat Concio —Pe Al was detailed to represent al (OT —NotRemoved Bolle No typical pre-Northridge practice. Be ee ee te Specimen A? was the sume at 2 Soh famed Wee ys Atv hue with the backup bar i ee aes Wa is and weld tabs removed and a in reinforcing filet wold ( column fusion ine as cecurred in lies have been tested dymami- 8) added to its weld rook. EI0T-4 the previous test. In contrast, cally, utilizing the facilities in was used as the filler metal in when an 27018 filler metal was the ATLSS Center. Figure 8 both specimens. The rraove Used to febrieate the joint, no shows a schematic test assembly welds were ultrasonielly ested weld cracking occurred and the laid on tho laboratory floor and Specimen A'S was a fully welded Ultimate strength of the beam the test setup used, The column connection, using EVOI8 elec Plate was developed (see Figure was a WI4xd11, 12.7-t. long, trode. The beam web was welded fteand Figare 7) and made of A572 Gr. 60 sicol. 1€ to the eolumn flange through ‘These (rts indicate that the was supported by a pin at ite groove weld. The dffereees in notch toughness of the weld hottom and a roller at its top, detailing of the three specimens total useé in the fabrication has The A36 steel WaGx150 beam are summarized in Table 1 strong effect on the perfor: (flange yield strength equal to 38 "Fully reversed dynamic dis mance of te connections ks), approximately 10-f. long, placement cycles were applied at was connected to tho column at predetermined rates, Twe issues the mid-height, A dynamic eyelic Were considered in determining load was applied at the free end the rates. One was the response Lanok Size Beant & Couuney Parallel to the simulated ton- of the beam by wo eervo-con- frequency ofa typieal muli-tary sion connection studies, fall. trolled actuators a shown, seel structure andthe other was seale bearr-and-eolumn assem- ‘The specimens tested have the the appropriate kinetic delorme Sas tMN DI DAE) LON(OR gaeeensncctliy ec Epes eer paear toa center Tae graphic interface yo always wanted, wit the advanced analysi power yourelly nee v Dynamic Analysis ¥ Interactive 3D Modeling ¥ Rendering & Animation ¥ Code Checking, ¥ True Mac Interface Tet (408) 464 1949. Fax 408) 964 07 th ADDITION TO ANGLE, RAL, PPE, CHANNEL, AND BEAM ROLLING pence nen ter Please circle # 55 Please circle #92tion rate of steel in the plastic range, which ean be represented fas strain rate. At small displace rents a rate of 2 eles per see- tnd was used. Detailed deserip tions ofthe rates adopted are not prosent here but will be included Ina fortheaming report. Spocimen A-I fractured at the bottom flange connection in a brittle manner. The maximum bending moment achieved dur- ing teat was only 87% of the yield moment of the beam. The Fracture occurred when both the beam and column were still in the elastic range, because no vis- ible sign of boam yielding was observed, except some flaking of the whitewash in small areas around the access holes. The track started at the root of the ‘weld, extended into the column ange (eausing divot type frac- ‘wire and eventually eased the Separation of the bottom beam flange from the eolumn. These characteristics are similar to those nbserved in the simulated tonneetions and in some of the damaged connections after the Northridge earthquake, ‘The removal of the back-up bars in Specimen A. resulted in fan improvement of its perfor ‘ance, but the brittle Feature of the lange welds again led to fal ure, Other than the limited Iywhen the welds ofthe top and ottom Minges fractured almost simultaneously during a reversed loading eyele. The frac tre surface of the welds showed the same characteristics as the fracture surfaces of the simulat fed tests, The maximum bending ‘moment resisted by the connec: tion was approximately the yield moment of the beam. Figure 9 Shows the fractured bottom ‘ange and Figure 10 i a plot of the load vs. displacement (of the load point) relationship of the specimen. “Much improved performance in torms of both strength and uctility” was observed in Specimen "A-3," A. typical Smoment gradient” plastic hinge Pater otene sick! ves | oasan éy | Figure 7 (op) Ductile beam plate fracture in ETOI8 welded test ‘specimen Figure 8: (bottom) Fullsize assembly test specimen and setupPlease circle # 54 40TH EDITION INCLUDES METRIC [AND STANDARD UNITS. ‘nutes spss nea able cote ine KES sear fer coat ads sot nin + Cort Kad Se dariadtom ove to yrs of sya al esearch ye pstd and rod ferLRFDasogn ‘ORDER TODAY! $20.0 gr yn US sd is possi pyateusee fit cls stage anna ae rust accmosny oe $2000 (U. earn) Number Cope, ovo ws srAcoueNT i Figure 9: (op) Fracture of bottom flange in Specimen A:2 (eee arr.) Figure 10: (above) Load ve. displacement relationship of Specinen A'2 dovolopod in the beam near the column face with extensive veld lng aeeurring in both the bes flanges as weil asthe web. There ‘waa algo yielding of the panel zone. During the application of the large amplitude displace ‘ment eyeles, some out-of plane movement of the beam was bsarved, but the major eause af failure wae inelastic local buck- ling of the beam flanges and web. The test stopped when the futatplane movement continaed {t inerease ‘and local buckling deformation became excessive. igure 11 shows yielding. and local buckling in the eritical region after the specimen was completely unloaded. The load vs. displacement reiat.onship recorded during the testis pre sented in Figure 12. The maxi ‘mum displacement applied was plus/minus 3.84 inches and the forresponding maximum bend: ing momenta the face ofthe eo. tama Was 1.22 times of te plas He moment of the beam. ‘The plastic hinge rotation reached ‘was about 0.027 radian (aleulat bd by assuming that the ninge is Tocated ata distance 2 frm the ‘where dis the ‘The issue of strength androtation capacity of plastic hinges in WE beame was studied in detall at Lehigh University as part of the researeh program on Plastic design of tho steal frame tructares, It jx known that the Dbehavior is affected by euch fae tors as: moment gradient, strain hardening properties of mater als width-to-thickneas ratios of flange and web, and the precence (or absence) of stiffeners. The Itoral bracing system was later modified and an attempt was made ta perform additional test ing with larger displacement eyeles. But ductile fracture of base metal of the tap flange soon developed duo to accumulated local strains. The crack, initiat data lange tip atthe weld toe find extended in a stable fashion in the base metal until fracture, ‘caused the flange to separate fom the column, Coxer.usioss 1) Both small-scale simulated beam flange-to-column Mange teats and fullsize dynamic eyelie tests of pre-Northridge connee tions show failure modes similar to those observed in the field Brittle fractare developed in the clastic range of response from flaws in the low toughness BIOTA weld metal and geomet: ie conditions 2) Only marginal improve an be obtained by retrofitting pre- Northridge connections by Temoving back-up bars and ‘adding higher toughness weld Teinforeement. Fracture instabil: ity is still likely to develop at internal weld flaws within the Tow toughness weld. '3) Acceptable connection per formance appears to be obtain able by using @ higher toughnoss weld metal in conjunction with Femoving back-up bars and a fully welded web. The simulated beam flange test with B71T-8 tnd BUOTG-K2 weld metal pr ‘ided the same performance 83 the B7018 test, A connection fssembly with E70TG-K2 welds twas tested in December, after the press deadline for this are igure 11: (tp) Yielding and lcal buckling in Specimen A-3 igure 12: above) Load vs. displacement relationship of Specimen A'S ee “D) The simulated tests show thet the behavior of connections uunder static and dynamic load: ings significantly diffrent. Further research is needod to examine more fully the major variables of welding and weld metal selection, the connection dlsign details, higher beam yield ‘strengths, and dynamie loading. The results presented were obtained from three research projects investigating fractare of Steel moment connections from the 1994 Northridge earthquake, ‘The following agencies provided fands for these projects: National Science Foundation, National Institute of Standards. and Technology and SAC Joint Venture, Additional funds were received from the Institute of Tronworking Industry and California Ironworkers, The Structural Shape Producers Council arranged donation of ‘material for the test speciments by Bethlehem Steel Corp. snd Lincoln Elestrie Company” Bric J. Kaufinann and Ming Xue are research engineers with the “Center for Advanced ‘Technology for Large Structural Syetoms (ATESS) at Lehigh University, Le-Wu Lu ta profes sor and’ chairman of the Department of Civil Enaincering fat Lehigh University, and ohn W. Fisher is a professor of cloil engineering and director of ATES Center