Milling optimisation of removal rate and accuracy with uncertainty: part 2: parameter variation

Milling models provide tools for estimating stability and surface location error. Existing models are deterministic, though inherent variations in the model inputs propagate to uncertainty in the model outputs. In this paper the experimental procedures used to estimate the model parameters are presented. The effect of correlation between parameters is addressed. The variability of the stability boundary and surface location errors are determined using Latin Hypercube sampling. It is seen that including the correlation between parameters reduced the output variability by as much as 55% with a minimum reduction of 10%. Comparisons between mean model predictions and experimental results are provided.