Analysis of the Strength Data of wood Structures on Limit States Design I. The influence of probabilistic and statistic method on specified values

Strength data of structural lumber must be fitted to empirical distributions for limit states design to replace the allowable stress design in Timber Engineering. There are some methods for fitting sample data to empirical distributions. The differences of parameters, goodness-of-fit test and the statistical lower limit due to the methods were investigated. The results were as follows: When we estimate parameters for the strength distribution of structural lumber, (1) We should not use the moment method because the calculation does not converged in some cases. (2) Normal distribution or Log Normal distribution is appropriate for 100% data, but Log Normal distribution is better because there is a probability that the strength is minus in Normal distribution. Also 2P or 3P Weibull distribution is appropriate for the lower 15% data, but 2P Weibull distribution is better because the theoretical and physical meanings of the location parameter in 3P Weibull distribution is not clear. (3) The statistical lower limit, fifth percentile value with 75% confidence, in Log Normal distribution or order statistic should be calculated because the value in Weibull distribution has not been analyzed theoretically. However the value in Log Normal distribution is better because the order of these values change due to grades and the value in order statistically is directly influenced by the data in small populations.