Assessing variation in package modeling

摘要

Predictions from empirical models are affected by variability in the input parameters for the data set used to build the models.For corrugated boxes， the difference between actual and modeled compression strength creates a real cost associated with box production， often resulting in boxes that may need to be over-designed to compensate for a lack of model precision.No work to date has attempted to assess the limitation in these compression estimates due to input parameter testing variability.In this paper we approach that problem， initially for the McKee equation and then conceptually for other box models.For our industry to do a better job at meeting the needs of our corrugated packaging customers， we need to reduce the variation in the tests we all rely on， particularly for evaluating material strength(edge crush test [ECT])and package compression performance(box compression test [BCT]).Application:Though there are many models for estimating box compression strength， few include an assessment of the limitations on modeling accuracy based on variations in measured inputs.This paper provides an assessment of estimation accuracy， as well as what must be done from a testing perspective to reduce the inherent uncertainty in model output arising from testing variation.