THE RELATIVE RIGIDITY OF WELDED AND RIVETED CONNECTIONS

Abstract

Little quantitative information being hitherto available concerning the degree of restraint developed at the ends of steel beams and girders with welded connections, and the extent of possible related economies and improvements in frame stiffness being correspondingly uncertain, the experimental investigation of two distinct aspects of joint rigidity was undertaken. These were:(a) The relative capacity under gravity load of beam-girder and beam-column connections for developing beam restraint, or continuity, designed with a view to reducing the required weight of the connected beam.(b) The relative values of certain typical connections in resisting the deformation of a frame due to lateral loads, either in one direction or subject to reversal.Two series of specimens were fabricated and tested in relation to plain rating beams. The first series, designed for the purpose of studying beam continuity, consisted of ten 9-in., 20.5-lb. Bethlehem beam double-cantilever specimens, employing five different types of welded connections. The second, designed primarily for the study of wind bracing rigidity, consisted of twelve 18-in., 47-lb. Carnegie beam double-cantilever specimens with three different types of connections. These types included the welded T, the welded gusset and, for comparison, the riveted T. For half the specimens of the second series the connections were to a central transverse plate simulating the web of a column, and for the other half to the flanges of a 12 by 12-in., 110-lb. H-column.Coefficients of restraint, that is, the ratio of actual to 100% restraint, were ascertained by determining end slope angles with the aid of a special type of longitudinal extensometer attached to the beam flanges. Although, for convenience, the specimens were tested as double-cantilevers the results were readily transformed to be applicable to the partially restrained beams simulated by the specimens.Story drift angles, that is, the angle through which the columns of a building subjected to horizontal force would tilt in one story of height due to the deformation of the connections, were ascertained by utilizing the end slope angles mentioned above and the observed deflections of the specimens. The results for the specimens were transformed by calculation to be applicable to a 20-ft. bay of an actual building.Broadly stated, the results obtained in the study of the connections tested were:(1) Welded connections as ordinarily designed for end restraint fall short of the ideal value of 0.75 for the coefficient of restraint by from 10 to 25%.(2) Both welded and riveted connections designed primarily for capacity wind moment develop a coefficient of restraint in excess of the most economical value (0.75) required to resist gravity loads only.(3) A beam welded to a column web may develop less story drift angle than would a plain beam integral with the column. T's riveted to a column flange may, for a 20-ft. bay, give rise to a lineal drift of as much as 0.36 in. per 12-ft. story under a capacity wind in one direction, while welded T's under the same conditions showed 0.09 in. and welded gussets only 0.03 in.(4) In the connections designed for beam continuity only, the compression attachment was in all cases more yielding than the tension one. Transfer of axial compression by longitudinal welds to the edges of a thin flange introduces much yield due to horizontal shear in the flange.(5) In wind connections either to a column web or to a column flange, the top and bottom attachments, under capacity reversing load, gave approximately equal total longitudinal deformations for each of the three types.(6) From a third to a half of the flexural slope in a connection to an unstayed column flange may be charged to the deformation of the column flange.(7) In wind connections to an unstayed column flange the welded connections gave regional flexural slopes from 17 to 63% of the corresponding ones developed in the riveted T-connections, depending on the region, or longitudinal zone. Consequently, the welded connections were definitely stiffer than the riveted ones in all regions.(8) The aggregate longitudinal elastic deformation in tension for capacity reversing load was greater than the aggregate deformation in compression, the maximum excess being 70%. This was for the riveted T-connection to a column web.(9) For the riveted T's a large part of the longitudinal deformation was non-elastic slip. It amounted to 31% for the connection to a column web and 46% for a connection to a column flange.(10) Welding of stiffeners to the column flanges reduced the longitudinal deformation of the welded T and the welded gusset connections to 75 and 85% of their previous values, respectively.(11) For the specimens designed for beam continuity only, the ratios of the test factor of safety to the design factor of safety, based on an estimated ultimate strength of 10,000 lb. per lineal inch of [Formula: see text]-in. fillet weld with uniform distribution of stress, varied from 0.65 to 0.97.(12) For the specimens designed primarily as wind connections, the ratio of test factor of safety to design factor of safety, based on assumed uniform stress distribution, varied from 0.88 to 0.98 for the riveted specimens, from 0.49 to 0.83 for the welded T-specimens and from 1.06 to 1.29 for the welded gusset plate specimens. It is thus evident that non-uniformity of stress, which obviously existed to a marked degree for the welded T-connections, may cause the capacity of certain types of connections to fall considerably below that estimated on the basis of uniform distribution.