Edwin Nagy

Date of Award


Level of Access

Campus-Only Dissertation

Degree Name

Doctor of Philosophy (PhD)


Civil Engineering


Eric N. Landis

Second Committee Member

William G. Davids

Third Committee Member

Roberto Lopez-Anido


There has been considerable effort over several decades to create a useful model to predict failure in wood. One of the chief stumbling blocks that has stood in the way of success is the combination of wood's complex morphology with the extreme dependence of wood strength on its morphology. Our attempt to surmount this block is a 2D/3D morphologically-based lattice-type model. The model has a quasi-regular rectangular array of truss elements braced with diagonals. Element properties are randomly created based on empirically-derived means and variations. Morphological features represented in the model include non-straight grain, growth rings and specimen geometry including notches. A relatively complex 2D model and a much-simplified 3D model have been created. The 2D model is in anticipation of a complex 3D model - even the 2D model representing a reasonably sized specimen may run upwards of several weeks on today's faster desktop computers. The model was tuned using simple stress states - shear and tension parallel and perpendicular to grain. After the element property values were set based on the simple experiments, the results of more complicated numerical simulations and laboratory experiments were compared. Target strength properties were mean and variation of peak strength, deformation at peak strength and initial stiffness in compression parallel and perpendicular to grain, notched tension parallel and perpendicular to grain, and bending. In general, mean peak strengths and deflection and peak strength are well predicted by the model, but several of the coefficients of variation were highly under-predicted by our model. The model also predicts peak strength for notched tension. In addition, the model predicts relative conversion of external work to elastic, plastic and fracture energy. The fracture energy predictions were compared with acoustic emission measurements from laboratory specimens. While far from the engineering tool that is eventually desired, the model represents a large step forward in our ability to predict the mechanical behavior of wood under a variety of stress states based on statistical studies (computer based in-grade testing). In addition, the model allows the investigation of the relative behavior of different connection geometries and grain angle states.

Files over 10MB may be slow to open. For best results, right-click and select "save as..."