The research summarized by this thesis was comprised of an experimental analysis of beams loaded perpendicular to grain at midspan by a bolted double-shear laterally-loaded connection. Connection specimens were loaded monotonically until capacity was reached. Variables of consideration included the loaded edge distance of the connection main member, the span:depth ratio of the main member, and the main member material. Southern pine machine-stress-rated (MSR) lumber, laminated veneer lumber (LVL), and parallel strand lumber (PSL) were the three material types included in the program. Experimental results were compared with theoretical predictions from three models: the yield theory-based general dowel equations, which are currently the standard for laterally-loaded connection design in the U.S., and two models based upon fracture mechanics. All material property inputs required by the three models, were measured in the experimental program of this research and used to produce theoretical predictions. Comparisons were also made with respect to design values in the form of calculated factors of safety, over-strengths, and design factors of safety. Test results and observed trends are provided for all connection and material property tests. Notable trends included failure by splitting for all connections at low loaded edge distances, and variable span:depth ratios generally having a negligible effect on both connection and model performance. In most cases, the general dowel equations were more accurate than the two fracture models, however it should be noted that all three models over-predicted connection capacity at low loaded edge distances. / Master of Science

Over recent decades, the public tendency toward using the structural composite lumber (SCL), a common composite of wood made of wood strands or veneers glued and compressed together, as structural members (especially the main load bearing members such as beams and columns) has risen considerably. In contrast to the fast-paced market growth of these products, development is slow. The experimental development is gradual and time-consuming and the computational development is even slower. The objective of this project is to introduce appropriate numerical models for limit state analysis of a certain type of SCL material called PSL. Parallel strand lumber (PSL), has mesostructures characterized by the presence of voids that renders the mesostructure highly heterogeneous. In addition to material phase aberrations such as grain angle variations and defects, void heterogeneities play an important role in determining the failure modes and strength of PSL. In this study, virtual void structures were defined to form part of the input to finite element analysis of PSL for the purpose of investigating the sensitivity of strength to the void structure. Assuming the wood phase to be homogeneous and orthotropic, the following 2D and 3D characteristics of voids were investigated: volume fraction, volume, alignment and moments of inertia of voids, as well as second moment properties, lineal path function and chord length functions of the two phase mesostructure. In addition, a method was developed to generate virtual voids in order to simulate PSL and investigate the possible effects of the void distribution on material strength. An experimental program along with a statistical survey was conducted to quantify the mentioned characteristics of the voids in two 133 mm * 133 mm * 610 mm 2.0 E Eastern Species PSL billets. As expected, most of the voids lie on the longitudinal direction of the specimen and have approximately an ellipsoidal shape. Based on this shape data, the characteristics of the ellipsoids which best t the voids were calculated. Using the statistical data of the fitted ellipsoids, a random field of virtual ellipsoid shaped voids to simulate the mesostructure of PSL was generated. In this study, the simulation of PSL material is based on two simplifying assumptions: 1) The wood phase is continuum, homogeneous and orthotropic. While in reality, the wood phase consists of glued wood strands that are heterogeneous due to their mechanical variability and only roughly orthotropic on a macro scale as a result of the varying fiber angle; 2) Voids are the mere source of uncertainty. The linear elastic analysis of carefully defined (in mesostructural aspect) PSL models can be the first step of mechanical study of the material. The effective modulus of elasticity of material in presence of voids and the distribution of conventional, principal and effective stresses considering the effect of volume fraction and shape of the voids are the target of this preliminary study. Linear elastic uniaxial analyses showed good mechanical consistency between the models including actual void shapes and the models including ellipsoidal void representations. Also, they showed that the stress mutliaxiality at the tip of the voids is negligible. The study of mechanics of PSL is incomplete unless the question of material anisotropy is taken into consideration. PSL is brittle in tension and ductile in compression. The material heterogeneity increases the complexity of the problem by affecting the stress distribution in the member. A detailed nonlinear approach has been proposed in order to investigate the mechanical behavior of PSL structural members under different uniaxial loading scenarios. This approach introduces proper constitutive models for the wood phase along with good void generation techniques. In other words, this approach suggests what models should be used for the continuum-assumed wood phase to simulate its brittle behavior in tension and ductile behavior in compression; and moreover, tests the applicability and accuracy of ellipsoidal void representation. The models are calibrated using the results of experiments on PSL material. Because of the brittle behavior, all wood products show significant mechanical dependency to the member's size under tensile loading. Once good constitutive model and mesostructural simulation is found for tensile loading, it is easy to make and analyze PSL models with different sizes and investigate the effect of size on mechanical behavior. The simulation results have been compared to the available results of a previously done experimental study.