Modern Steel Construction / November 2000

By Robert J. Dexter

In the beam-to-column connectionused in welded moment frames,the through-thickness strength ofthe column flange is relied upon totransmit the cyclic forces from thebeam flanges to the column. Testingsponsored by the SAC Joint Venture,TradeARBED, and Nucor-YamatoSteel recently performed at LehighUniversity and the University ofMinnesota has shown that in theconstrained configuration of thebeam-to-column joint, the columnflange through-thickness strength ismuch higher than the demand thatcould be generated by a beam flange.The through-thickness failure of col-umn sections is very unlikely inmoment connections or other typesof T-joints. Therefore, it is recom-mended that the through-thicknessstrength does not need to be explicit-ly checked in the design of weldedbeam-to-column connections.

Many of the moment connectionfractures that occurred in the 1994Northridge Earthquake and fracturesthat occurred in subsequent labora-tory tests on beam-column jointspropagated into the column basemetal, including fractures thatappeared to scoop out some of thecolumn material, referred to asmouse-ears or divots. Fractographyhas shown that the primary cause ofthe fractures was identified as low-toughness weld filler metal combinedwith the existence of backing-barnotches and lack-of-fusion defects at

TThhrroouugghh-tthhiicckknneessssSSttrreennggtthh ooff CCoolluummnn FFllaannggeess

Column flange through-thickness strength is muchhigher than the demand thatcould be generated by abeam flange, makingthrough thickness failure ofcolumn sections highlyunlikely.

the weld root. These fracturesshould not be considered through-thickness failures of the columnflange since the cause of the frac-tures is not related to the columnflange properties. Therefore, the factthat some of these cracks propagatedinto the column material should notbe considered a deficiency of the col-umn material.

There are no documented caseswhere lamellar tearing or inadequatethrough-thickness strength or ductil-ity was responsible for any of thefractures in the NorthridgeEarthquake or in subsequent fullbeam-column joint tests. Ultrasonictesting (UT) has been performed onhundreds of buildings that wereloaded significantly by the earth-quake. There has been only onedocumented case of a lamellar sepa-ration in the column flange materiallarge enough to require repair.Investigators concluded this separa-tion was probably caused by weldshrinkage during construction andthat it did not propagate during theearthquake.

Despite no indications that inade-quate through-thickness strength wasresponsible for any of these fractures,there are lingering questions aboutthe through-thickness strength of thecolumn sections. This lingeringdoubt is due to a number of factors,including the awareness of manyengineers that lamellar tearing is apotential problem with throughthickness loading, as described inmany textbooks. However, steel pro-duced since 1980 has typically notbeen susceptible to lamellar tearingbecause producers have been con-trolling the large nonmetallic inclu-sions that used to cause lamellartearing.

Another reason for the questionsabout through-thickness strength ofcolumn sections may be due to theresults of uniaxial through-thicknesstensile tests performed in accord withASTM A770. These results oftenshow a high degree of scatter includ-

ing some low strength values thatmay even be below the minimumspecified yield strength (MSYS) inthe longitudinal direction. In addi-tion, the reduction in area of thesethrough-thickness tests is often lessthan the reduction of area in the lon-gitudinal tests, indicating the ductili-ty of the material is less in thethrough-thickness direction.

The uniaxial through-thicknesstests are not representative of thehighly constrained conditions underwhich the column flange base metalis loaded in the connection, however.The experiments indicate that underthe constrained conditions typical ofa beam flange-to-column flangewelded tee-joint in a moment con-nection, the column sections do notyield in the through-thickness direc-tion. Because it is not possible to getyielding in the through-thicknessdirection, the measured through-thickness yield strength and reduc-tion in area should not be a concern,i.e. the ductility in that direction isnever needed.

This behavior was observed inmore than 46 tee-joint specimensthat were fabricated with high-strength (690 MPa [100 ksi] yieldstrength) pull plates welded trans-versely to opposite flanges of short610 mm (24) lengths of heavy col-umn sections. Column shapes wereobtained from four steel mills (inte-grated and electric furnace) and con-formed to ASTM A572 Gr. 50, A992,and A913 Gr. 50 and 65. A columnsection taken from a building thathad been damaged in the 1994earthquake was also tested. Thesteel included a wide range of chem-istry, including some specimens thatwere intentionally made with anunusually high sulfur content as highas the specification limits of 0.05%.The tee-joint tests were performed athigh and low strain rates, with vari-ous widths for the simulated beamflange, and with axial and bendingloads.

Modern Steel Construction / November 2000

In every case, the through-thick-ness strength of the column flangestested exceeded the 690 MPa (100ksi) yield strength of the pull plates,well above any possible demand thatcould come from Grade 50 or evenGrade 65 beam flanges.

Robert J. Dexter is an associate pro-fessor of structural engineering at theUniversity of Minnesota. ProfessorDexter is currently conducting tworelated studies on shape materialsand design for AISC.