Pairmatching and Prediction Models: Their Use in Predicting Particleboard Properties

Robert J. Tichy

Abstract


It is quite valuable for the research scientist to be able to predict accurately the strength properties of his treatment specimens. This is frequently done through the use of control or untreated test specimens.

This study explored several methods of predicting treatment specimen MOR's for two types of oriented particleboard, using side- and end-matched control specimens. The effect of control specimen positioning was investigated. Side-matching proved to be more accurate than end-matching. With the treatment specimen flanked on either side by control specimens, the average percent error was only 3.4% with a standard deviation of 2.8 when the prediction parameter was obtained by averaging the MOR's of the flanking control specimens. A freehand curve fit method was attempted with similar results.

The introduction of another material property (i.e., modulus of elasticity) into the prediction model was examined. Because of a low correlation of the parameters used (MOR and MOE), this method did nothing to improve prediction accurary.


Keywords


Statistical analysis;experimental design;modulus of rupture;modulus of elasticity;particleboard;prediction equations

Full Text:

PDF

Refbacks

  • There are currently no refbacks.