Some observations towards improved predictive models for box compression strength

Abstract

A McKee-Urbanik type equation for box compression testing (BCT) strength prediction was evaluated using the box data of McKee, Gander, and Wachuta and the tube strength data of Batelka and Smith. Instead of separating BCT into separate panel loads, an effective width was deemed adequate for the data set. Although not much can be done to improve the fit of the McKee data over the McKee formula, the fit of the tube data was slightly better than previously reported in the literature. Unlike previous literature, the fitting parameters used in the model directly target the uncertainty we have in predicting a critical buckling load for a panel (K = edge restraint, κ = twisting resistance, α = relative scaling). The lack of ability to improve the McKee data set is attributed to the variety of boxes tested and the dependence of BCT on unmeasured properties. The century-old take-home message reiterated in this paper is that focused research is needed to replace estimated or fitted parameters with independently measured quantities, which should result in a decrease in the number of boxes with unsatisfactory predictions of BCT.