Loading Subspace Selection for Multidimensional Characterization Tests via Computational Experiments
To pursue characterization of composite materials, contemporary automated material testing machines are programmed to follow loading paths in multidimensional spaces. A computational methodology for selecting the best loading subspace among all those possible is formulated and presented in this paper. The criterion for subspace selection employed is based on the assessment of which among the possible subspaces generates the richest set of strain-states as compared to those of the union of all possible 4D loading spaces. A systematic program of simulation sequences of virtual experiments is presented and the concept of strain state cloud (SSC) is introduced as a high dimensional volumetric histogram describing the frequency of appearance of each strain state within the corresponding strain space. Comparison of the SSCs for each of the fifteen 4D subspaces relative to the full 6D space allows a ranking classification of each subspace. Based on this ranking we select the three top cases as being those considered for actual testing.