Characterizing the Strength of Wood Truss Joints

Abstract

Probability density functions (normal, lognormal, and three-parameter Weibull) were used to characterize strength data for three different types of metal-plate-connected wood truss joints (web at the bottom chord, tension splice, and heel). Modulus of elasticity (MOE) of the lumber used to fabricate the joints was also characterized. A probability-plot technique, in conjunction with Kolmogorov-Smirnov and chi-square statistics, was used to determine which distribution best fit the data. Lumber MOE was best described by a lognormal distribution. No single distributional form fit the strength data for all three joint types with equal accuracy. Lumber MOE and joint strength were unrelated. Strength data for the web at the bottom chord and heel joints were best described by normal distributions; however, none of the distributions considered fit the data for the tension splice joints. The probability-plot technique provided a better visual inspection of fit than did a density function superimposed over a histogram. Fitted distributions are easy to work with and can be used in reliability analyses to simulate strength values of joints. The results presented here are for particular joint types and plates and should not be extrapolated to other truss joints.