Abstract :
[en] The design of a bridge deck with respect to wind action usually relies on a simple principle: as the average and standard deviation of aerodynamic coefficients are responsible for the buffeting excitation, minimizing them is supposed to be beneficial for extreme wind loads. Traditional bridge deck shapes are optimized in this view.
Seminal developments related to the establishment of peak factors for the estimation of the extreme loads are generally attributed to Cartwright and Higgins (1956). They have been widely applied, but fail to offer an accurate model in some circumstances as they are developed within the framework of Gaussian processes. Some less restrictive formulations are discussed and compared by Floris and Iseppi (1998). The non-Gaussianity of the wind loading is indeed an important issue as it affects the extreme wind forces, i.e. those that precisely have to be taken into account for the structural design. Such discussions indicate that a skewed random process might evince, even for a skewness coefficient as small as γ_3=0.5, peak factors that are 20% to 30% higher than those obtained with Gaussian developments (e.g. based on a model proposed in (Gurley et al., 1997)).
With that in mind, and being restricted to buffeting analysis, this paper proposes to supplement the traditional considerations about deck shape optimization with a discussion about the skewness of aerodynamic loading.
We consider a nonlinear wind loading model in which the skewness results, in part, from the nonlinearity of aerodynamic coefficients. Actually, in a quasi-steady context, these coefficients are measured for various angles of attack on the bridge deck. This typically results in a nonlinear coefficient .vs. incidence relation, which in turn is partly responsible for the non Gaussian distribution of the aerodynamic loading.
As a final objective, we endeavor at providing an estimation of the skewness of the aerodynamic loading that could be typically expected for a given bridge deck typology. In doing so, we believe the general trends that are observed in this study will help in assessing the effect of the choice of a bridge deck typology on the extreme wind force. This objective is reached in a two-step procedure. First a database analysis of several bridge decks is performed in order to provide, for three families of deck typology –box, girder, streamlined- ranges of variation of drag and lift coefficients. Then, based on a quasi-steady nonlinear aerodynamic model, these ranges are translated into skewness coefficient of aerodynamic forces.
