Abstract
The purpose of the paper is to develop the model by using the predicted value of creep response based on the regression methodologies that can be effectively used to reduce the test time required to put the stiffness test for the plastic pallets. In order to develop the best model, we examine a wide variety of regression techniques including the OLS (ordinary least squares), the WLS (weighted least squares), the Newey-West robust estimator, and the SUR (seemingly unrelated regression). The new protocol is assessed to remarkably enhance the test productivity by decreasing the standard bending test time by 77% from 26 to 6 h. It is shown that the deflection process under load is best described as a linear model regressing deflections on the natural logarithm of test duration. The model is statistically validated against the various estimation techniques and sample. We demonstrate how the forecasted deflections at a specific time point are statistically tested for a certain performance standard limit. From a standpoint of the parsimony and practical significance, we claim the OLS estimation of a simple regression model is a reasonable choice for the shortened stiffness tests of pallets.
Acknowledgements
This work was supported by the year 2012 grant from the Ministry of Knowledge Economy, the Republic of Korea [grant number B0012352]. The parts of the work have been discussed at the meetings for the ISO/TC 51/WG 2 held in UK for the year 2011 and Seoul for the year 2012. The authors gratefully acknowledge the valuable comments from the committee members of the TC 51, particularly Hideo Takamori, Hiroji Fukumoto, John Mead, David Michael, Gunilla Beyer, John Harvey, Edgar Deomano, Sanjiv Takyar, and Ralph Wunderlich, among others.