Discussion of “Empirical Design Rules for Effective Utilization of Orthotropic Decks” by Roman Wolchuk

Abstract

The author’s statement that there is no stress-strain proportionality of welded connections because there are residual stresses that are at the yield point at the weld toes or in the weld, and therefore that linear-elastic analysis is not applicable at such locations, is in error as worldwide research has shown in published experimental fatigue studies of welded steel details provided in detailed National Cooperative Highway Research Program (NCHRP) reports (Fisher et al. 1970, 1974; Keating and Fisher 1986) and several books (Fisher 1984; Barsom and Rolfe 2001; Maddox 1991; Fisher et al. 1998; Gurney 2006). These results are also incorporated into design specifications around the world (every edition of the AASHTO standards from 1974 to the current version (AASHTO 2002); every edition of the AASHTO LRFD bridge design specifications from 1994 to the current version (AASHTO 2012); every edition of the American Railway Engineering and Maintenance-of-Way Association (AREMA) specification from 1978 to the current version [AREMA 2014; British Standards Institution 1980; European Convention for Constructional Steelwork (ECCS) 1985; European Committee for Standardization (CEN) 2006; Japanese Society of Steel Construction 1993]). There is no cumulative plasticity at these welded details as the author implies in Fig. 3 of the paper; plasticity mostly occurs at gross section yielding. At the weld toe section, the residual stresses are in equilibrium on the cross section, and the surrounding material remains elastic under subsequent dead and live loads. The flowwithin the high tensile residual stress region at the weld toe is contained by the surrounding, primarily elastic, material, which prevents cumulative plasticity at these regions. This has been verified in the experimental and analytical studies cited previously. Accordingly, the nominal elastic stress in the gross section is used for fatigue design, without any explicit consideration for the local residual stresses, which are implicitly considered by experimentally obtained fatigue resistance curves. This approach to fatigue design has been accepted universally; the phenomenon of metal fatigue was formally acknowledged in the 1860s (Barsom and Rolfe 2001; Bannantine et al. 1990). Only loads that create yielding on the entire cross section, such as that which occurs in cyclic hysteresis loops (or gross section yielding), as applied in seismic loading, redistribute the residual stress states. This does not occur in the orthotropic deck, nor in other welded bridge elements,which are designed for nominal elastic stresses under service conditions. Measurements on full-scale orthotropic decks and other bridge members have verified this finding in the laboratory and in the field (Fisher 1984; Connor and Fisher 2000; Tsakopoulos and Fisher 2005a, b; Connor et al. 2003). These measurements have also verified the analytical evaluation of orthotropic decks (Fanjiang et al. 2004).