DiscussionDesign method for bolted connections loaded perpendicular-to-grain: Discussion

Abstract

This paper and recent papers by the same authors (Quenneville and Mohammad 2000; Mohammad and Quenneville 2001) present a tremendous amount of valuable laboratory data on the subject of bolted connection behaviour in timber. The observations of behaviour and failure mechanisms are particularly relevant. The authors present new design equations, based on failure mechanisms, to predict connection strength — an improvement on the current code requirements found in O86.1 (CSA 1994). These new equations were shown to produce improved results in the paper’s figures; however, the equations lack a detailed analytical basis. All bolted connections, regardless of the material, will develop a three-dimensional stress field — the magnitudes of the stresses depend on the connection geometry and material properties. The ductility in the steel connector also plays a role. In the wood member, the stress distribution is complicated by material orthotropy and nonlinear behaviour of wood in compression. The connection behaviour is further complicated by brittle failure of wood in tension and in shear. The authors provide between three and six equations to predict the connection strength for different connection con figurations loaded perpendicular-to-grain, when strength is governed by bearing. They provide only one additional equation to predict connection strength when splitting governs. The authors later conclude that, based on their tests, tension perpendicular-to-grain splitting was found to govern connection strength in most cases. From their conclusion, one would expect that their proposed equation for the prediction of brittle failures would have been developed using a rigorous analysis of the stresses in a bolted connection, rather than curve-fitting to an expression originally devel oped to predict the strength of timber rivets. Foschi and Longworth (1975) used a combination of material nonlinearity and the weakest link theory to predict the strength of timber rivets undergoing brittle failure. This work forms the basis of the current O86.1 rivet design procedure. Their one-dimensional finite element model predicted stresses around each rivet over a series of load steps. At each step, the stresses were analyzed to determine if the tension or shear stresses exceeded the allowable stresses according to the Weibull weakest link theory (Weibull 1939). The model also pinpointed the governing mode of failure (i.e., tension perpendicular-to-grain or shear). This model was specifically developed for timber rivets — the boundary conditions for the governing displacement field equations were chosen specifically for the rivet connection analyzed at that time. Quenneville and Mohammad modified Foschi’s (1973) governing equation but gave no direct indication that the original field equations can be applied to the design of multiple-bolt connections. There are a number of reasons to believe that recalibrating