A parameter for the selection of an optimum balance calibration model by Monte Carlo simulation

Abstract

The current trend in balance calibration-matrix generation is to use non-linear regression and statistical methods. Methods typically include Modified-Design-of-Experiment (MDOE), Response-Surface-Models (RSMs) and Analysis of Variance (ANOVA). These methods are used to reduce the number of calibration loads, to identify relevant calibration matrix coefficients, and to avoid or eliminate invalid or detrimental coefficients. The methods are currently focused on the balance loading scheme and the curve-fitting processes. They do not take into account the load dependent uncertainty inherent in the calibration loads, nor do they estimate the uncertainty of the balance when installed in a wind tunnel. This paper proposes a simulation approach which extends the matrix generation process to take into account both the uncertainties of the applied calibration loads as well as the uncertainty of the electrical response of the balance to these loads. Further, calibration uncertainties are propagated by simulation from calibration through to balance installation to produce uncertainties which may be expected in a wind tunnel. A Performance Weighted Efficiency (PWE) parameter is defined and used to select the best calibration matrix with respect to optimum in-tunnel balance performance. It is proposed that the PWE parameter, either in this form, or with modification, may be an effective balance comparison parameter.