Abstract
Since the early 1990s, the international wind tunnel balance community has made a concerted effort to establish internationally accepted and adopted balance calibration, evaluation, and uncertainty estimation practices. However, consensus on an international standard for wind tunnel balances remains to be achieved. Non-standard balance calibration and performance estimation practices, coupled with ambiguous definitions of balance performance metrics, complicate efforts by balance end-users to investigate and validate the load prediction uncertainty (accuracy statements) of a supplied balance independently from the balance vendor.
At the Aeronautic Systems Competency (ASC) of the Council for Scientific and Industrial Research (CSIR), conventional discrete check-loading practices are applied to validate balance performance. This practice involves comparing predicted balance loads against known values of several unrelated discrete check-loads and investigating whether the predicted load and its statement of uncertainty intersect the single discrete data point and their statement of uncertainty. A drawback of this approach to balance performance validation is the limited insight obtained and the difficulty associated with identifying the reason(s) for residuals falling outside the coverage probability of the uncertainty estimates.
ASC of the CSIR provided two balance calibration modelling practices. Different calibration models and balance uncertainty quotes are provided when a balance is subjected to both calibration practices. Without standard balance calibration and performance estimation practices, the seemingly straightforward task of selecting the “best” performing calibration is problematic. Selecting the “best” calibration model requires an objective and calibration-data-independent balance evaluation method. As an alternative to the traditional method of applying several check-loads and the subsequent analysis of the residuals, this study investigated the feasibility of balance performance evaluations using force-functions (FF). It was postulated that balance FF evaluations would produce objective metrics that are calibration data independent that would also serve as a basis for conducting comparative balance studies and the verification of balance uncertainty data.
This thesis investigated the capabilities and limitations of the proposed balance FF evaluation approach by performing FF evaluations on a balance that had been calibrated by both of ASC’s balance calibration offerings. The FF evaluation results revealed that one of the calibration models demonstrated superior load prediction capabilities. Furthermore, it was demonstrated that since the parameters (coefficients) of FFs represent unique function characteristics, each parameter served as individual metrics of balance performance. An assessment of the individual parameters provided a higher level of separation of effects and, thereby, greater insight into balance load prediction performance compared to traditional discrete check loading evaluations. Insight into the load prediction capabilities, load prediction asymmetry, non-orthogonality of the calibration model, and signal drift were obtained for both calibration models from an analysis of the referenced and independent performance metrics.
The successful demonstration of the balance FF evaluation approach’s ability to objectively assess balance performance specifications whilst simultaneously delivering additional insight into different aspects of load prediction performance led to an investigation into its use in different evaluation contexts. The four primary FF evaluation contexts identified were
• balance design analysis using Computer-Aided Design (CAD) and Finite Element Analysis (FEA),
• pre-calibration,
• post-calibration,
• and re-installation.
Performing FF evaluations in each of these contexts provides valuable and distinct insight into different aspects of balance performance.