Analyzing the Distribution of Moisture Content in Kiln-dried Lumber Using Goodness of Fit Techniques – with Procedures

Abstract

Knowledge of the distribution of moisture content in kiln-dried lumber is essential for quality control work. This paper describes formal methods and procedures for determining the distribution of moisture content in kiln-dried lumber using goodness-of-fit analysis, for three known distributions: Normal, Lognormal, and Weibull. Two graphical methods are employed: investigation of symmetry, and probability plotting. Formal mathematical tests are carried out for each distribution using test statistics based on the empirical distribution function (ECDF). The methods are tested on Douglas-Fir (Pseudotsuga menziesii) lumber collected from a production facility in British Columbia, Canada. Test results show that moisture content of kilndried lumber is well modeled by the threeparameter Lognormal distribution. In order to properly test the distribution of moisture content, all three parameters of the Lognormal distribution need to be considered: 'scale', 'shape' and 'threshold'. For kiln-dried lumber, the left bound (or threshold) of the distribution is greater than zero percent, and slightly smaller than the minimum lumber MC.