Identifier

etd-11182013-144154

Degree

Master of Science (MS)

Department

Renewable Natural Resources

Document Type

Thesis

Abstract

Predictive models were developed for stiffness and bending strength of southern pine 2x6, eight ft. lumber using nondestructive measurements of stresswave velocity, density and visual characteristics such as knots, slope of grain and rate of growth. To account for local areas of weakened material due to knots and slope of grain, a grid system was developed to quantify general knot size and location. Multiple regression models were created using these physical and visual measurements. Two sets of models were developed: one that removed influential samples with abnormal wavespeeds (greater than 18,000 ft./s) indicative of poor wood quality; and, models that included all samples. Static modulus of elasticity (MOEs) model performance was significantly better for those that removed influential samples compared to the all-sample models, with an R2 of 0.892 and 0.720, respectively. Modulus of rupture (MOR) model performance was slightly better with influential samples removed – R2 of 0.714 and 0.690, respectively. The location of knots within a board significantly altered the mechanical properties, especially bending strength. The results indicate potential for greater specification of allowable stresses for different orientations during bending. A simulated grading study was conducted to assess the feasibility of the developed models. Thousands of samples were generated according to estimated variable distributions and graded according to the American Lumber Standards Committee Machine Graded Policy. Results suggest that these models may be feasible in an actual lumber grading scenario.

Date

2013

Document Availability at the Time of Submission

Release the entire work immediately for access worldwide.

Committee Chair

Shupe, Todd F.

DOI

10.31390/gradschool_theses.362

Share

COinS