The fatigue behaviour of welded web attachments

Abstract

Several superstructures located in the United States contain welded web details possessing low fatigue resistance. The cracking of several steel pier bents on the Dan Ryan Rapid Transit Line in Chicago and small cracks observed in other bridges with similar details has demonstrated that this is true. The objective of this paper is to examine the fatigue resistance of two commonly used web attachments. The first attachment considers a flange plate passing through the web of a girder and attached by welds on one or both sides. The second detail considers flange plates welded on both sides of the girder web. The stress distribution along the eventual crack path is used to predict the fatigue resistance of these details. The predicted fatigue resistance under cyclic load for the detail without web penetration and for details which penetrate relatively thin webs and were attached by welds on each side were in good agreement with other test results on long attachments. The predictions and test results fall within the fatigue limits of cover-plated beam data used to derive design provisions for the AASHTO Bridge Specification ( Category ). Tests with web penetrations in girders with thick webs typically gave results much more severe. This can mainly be attributed to lack of fusion in the web behind the weld and its adverse affect on the stresses in the weld. The applicability of Category E to such details is dependent upon the thickness of the web to which it is attached and the size of the welds used to connect the flange to the girder web that it passes through.