A New Approach to Stress Grading of Lumber

Forcing low strength certification on many pieces of lumber which should have higher ratings makes lumber vulnerable to substitution by other materials which may be inferior in strength. To make grading rules more precise and efficient, the wide range of actual safety factors could be reduced by grading each wooden member on the basis of its characteristics, rather than on the basis of a species average. Automating the grading process would eliminate the imprecise element of judgment. The Purdue University Wood Research Laboratory tested two hundred 10-foot pieces of nominal 2 by 6 Douglas-fir lumber, using 40 pieces in each of five stress grades (1100f, 1450f, 1700f, 1900f, and 2150f). The correlation between modulus of rupture and various strength-predicting factors was first determined. Each piece was loaded at third points of a 9-foot span by means of a hydraulic apparatus, and the load was recorded for midpoint deflections of 0.200 and 0.700 inch; the loads were labelled P1 and P2 and the load increment to produce a 0.500 inch deflection was defined as P2 – P1, or E. Data from the machine tests was treated statistically using a multiple regression analysis. Relationships between the simple correlation coefficients determined which variables were significant in predicting MOR, and a linear predicting equation was derived by the least squares method: MOR = -3596 11.28E 136.4D, where MOR is modulus of rupture (psi), E is the increment of load (P2 – P1 in pounds), and D is density in lb./ft.3 at ovendry weight. The multiple correlation coefficient for this relationship was 0.702; though a coefficient of 0.681 can be obtained by using E alone to predict MOR, D was retained in this analysis to illustrate the principle of using multiple prediction equations. Performance of the visual and statistical systems was measured by relating safety factors and grades assigned to each piece of lumber. The visual grading safety factor was the actual modulus of rupture divided by the visual grade number, and the statistical grade safety factor was the modulus of rupture divided by the statistical grade number. Comparison of these pairs of safety factors showed the relative efficiency of the two systems. An ideal grading system should provide safety factors for each board as close to 2.424 as possible, given the spread in grade intervals. Though the test sample was small and included only one species, machine grading appears a promising possibility for more study and development.

You must be logged in to download any documents. Please login (login accounts are free) or learn how to Become a Member