Search

In this paper a new method for the derivation of settings for grading machines is proposed based on the fulfillment of basic statistic principles of regression, as well as the assumption that the distribution of the non-destructive indicating property of the machine and the model residuals are known. A system is introduced to quantify the uncertainty of the observations with the actual population they should describe. To verify the method the theoretical distribution of assigned strength grades is constructed together with a 90% confidence interval over which the destructive observations are expected. The strength grading system in Europe is based on strength, stiffness and density. Although in some cases stiffness or density may be the governing property this paper will focus on the case that the bending strength is the governing grade determining property. In this paper the principle of the proposed method is explained on simulated data and afterwards verified on real data for strength only. This shows good application for uses in practice. The verification showed that an incorrect combining of samples gives inaccurate settings for the intended grading area. With the proposed method the variation between samples can easily be taken into account. [ABSTRACT FROM PUBLISHER]