Home Team Programs Communications Products
Research Themes
The proposed research takes a holistic view of the process of manufacturing lumber products ranging from the changing resource R (the forests) to the changing supply S of forest products.
The approach is to view the process as a complex dynamic stochastic system, within an engineering context where design standards must be maintained. Eleven specific projects are outlined. They are organized within four main research themes.
Theme I: Monitoring the Resource
The projects in Theme I are about monitoring R through the sampling of S, something seen as prudent even though the grading system for lumber assures its design standards. Modeling R as a slowly varying stochastic process will help development of good ways of the (more practical) sampling of S along with its components of variation. Our holistic approach and hence monitoring plans will model and incorporate spatial relationships (correlations) amongst the sampling units and the components of variance within the system. That correlation structure will be part of the composite models of lumber strength we develop for indexing the aggregate strength of a "lot" of lumber, for example.
At the same time, we will address the abrupt changes in R due to catastrophic regional events such as a severe forest fire season that call for an (analytic) survey of S. We will use Bayesian approaches to computing the sample sizes for such a survey when no historical data is available. A byproduct of this exploration will be a new framework that can be used for modeling the stochastic process as a sequence of random compositional data vectors moving according to a generalized state space vector.
Theme II: Marketing the Supply
Theme II concerns the marketing of lumber, specifically how lumber from different species, regions or countries can be combined within grades and the resulting design values calculated. The aim is an alternative to the current methods, providing greater stability of the design under changes in those combinations, thereby avoiding the great expense of publishing new standards.
Theme III: Modeling the Properties
Theme III concerns the relationships amongst lumber properties: modulus of elasticity (MOE), modulus of rupture (MOR), ultimate tensile strength (UTS), as well as visual characteristics used for grading. Predictive strength models based on those characteristics will be determined so that estimates of the frequencies with which those characteristics occur based on a survey of S can be combined with those predictive models to estimate the population distributions of strength properties. These models may also assist in determining the joint population of strength characteristics, which cannot be simultaneously measured on a piece of lumber. We will also explore use of engineering models in this context to supplement statistical ones. The result will be a theory whose application will enable one design value to be estimated from another, thereby reducing the need for and cost of testing.
Theme IV: Managing the Load
Finally, Theme IV is about new directions. One project concerns the non-destructive testing of lumber when the sample is loaded to a predetermined stress level. The method, "proof loading", breaks only the weakest pieces that fall below the "proof load" stress level. The statistical challenges include finding the critical proof load to use for the destructive test so as to eliminate the low-strength pieces but minimize removal of acceptable pieces. A second project will build predictive models of strength properties for composite lumber products put under a sustained load over a long period of time. Of particular interest here is the ability of the models to accurately predict the long term behaviour of wood-based products with minimal (short-term) testing.