A New Multiple Proof Loads Approach For Estimating Correlations
By employing reliability based design to structures, engineers can make efficient use of building materials. One important input to the design is the correlation between two or more strength properties such as bending and tensile strength. However, with wood products, there is the added complication that it is difficult to create good pairs of specimens. When specimens are tested to failure in one strength mode the specimen is destroyed. Somehow, information on both strength properties must be obtained from a single specimen. Our approach is to load each specimen to a specified load in strength mode one and then load survivors to failure in strength mode 2.
We first review our earlier experimental designs which do allow for the estimation of correlation in a bivariate normal distribution. Then, we present some results for a new design that uses multiple proof loading. Some preliminary results are also presented for estimating correlation under a bivariate Weibull distribution.
Keywords and phrasesDestructive testing estimation of correlation
Unable to display preview. Download preview PDF.
- 3.Galligan, W.L., Johnson, Richard A. and Taylor, J.R. (1981), Examination of the Concomitant Properties of Lumber, Proceedings of the Metal Plate Wood Truss Conference(1979), Madison, WI: Forest Products Laboratory.Google Scholar
- 4.Johnson, Richard A. (1980), Current Statistical Methods for Estimating Lumber Properties by Proofloading, Forest Products Journal, 30, 14–22.Google Scholar
- 5.Johnson, Richard A., Evans, J. W. and Green, D. W. (1999), Some Bi-variate Distributions for Modeling the Strength Properties of Lumber, United States Department of Agriculture—Forest Product Laboratory.Google Scholar
- 9.Wu, K.T. (1993), An Experimental Design for Nonparametric Estimation of Correlation Under Destructive Testing, unpublished Ph.D. dissertation, University of Wisconsin-Madison, Dept. of Statistics.Google Scholar