Wind Tunnel Balance Calibration: Are 1,000,000 Data Points Enough?

Measurement systems are typically calibrated based on standard practices established by a metrology standards laboratory, for example the National Institute for Standards and Technology (NIST), or dictated by an organization's metrology manual. Therefore, the calibration is designed and executed according to an established procedure. However, for many aerodynamic research measurement systems a universally accepted standard, traceable approach does not exist. Therefore, a strategy for how to develop a calibration protocol is left to the developer or user to define based on experience and recommended practice in their respective industry. Wind tunnel balances are one such measurement system. Many different calibration systems, load schedules and procedures have been developed for balances with little consensus on a recommended approach. Especially lacking is guidance the number of calibration data points needed. Regrettably, the number of data points tends to be correlated with the perceived quality of the calibration. Often, the number of data points is associated with ones ability to generate the data rather than by a defined need in support of measurement objectives. Hence the title of the paper was conceived to challenge recent observations in the wind tunnel balance community that shows an ever increasing desire for more data points per calibration absent of guidance to determine when there are enough. This paper presents fundamental concepts and theory to aid in the development of calibration procedures for wind tunnel balances and provides a framework that is generally applicable to the characterization and calibration of other measurement systems. Questions that need to be answered are for example: What constitutes an adequate calibration? How much data are needed in the calibration? How good is the calibration? This paper will assist a practitioner in answering these questions by presenting an underlying theory on how to evaluate a calibration based on objective measures. This will enable the developer and user to design calibrations with quantified performance in terms of their capability to meet the user's objectives and a basis for comparing existing calibrations that may have been developed in an ad-hoc manner.