High-speed real-time processing of cross-wire data

A method is described for processing cross-wire data at high speed in real time. The basis for the high speed is the assumption of simple cosine cooling for the heat transfer because the expression for the effective cooling velocity is reduced to a linear function of the applied velocity components. The calibration inversion is greatly simplified because the measurements are linearly dependent on the effective cooling velocity of each wire, which can be expressed uniquely in terms of a single variable such as output voltage. This makes it practical to use a pair of one-dimensional look-up table arrays such that the number of elements corresponds exactly to the measurement resolution over the full range. This is the fastest possible method because the word returned by the analog-to-digital converter can be used directly as the index of the array and linear interpolation is avoided. It is shown that it is unnecessary to calculate U and V for each pair of cross-wire samples when Reynolds stress and triple products are required, and this further reduces the number of floating-point calculations. Finally, the method has been implemented in a true doublebuffered scheme in which there is a high degree of parallelism between the data acquisition and data processing calculations. The parallelism means that the extra experimental run time introduced by the data processing calculations is only minimal. Although the method assumes simple cosine cooling, the difference (typically 2° or 3°) between the effective wire angle obtained by neglecting the longitudinal cooling component and the actual wire angle provides compensation. The errors are less than 2% for typical wires and flow angles less than 25. The results of a Monte Carlo simulation indicate that the errors are only slightly larger than the errors observed when the longitudinal cooling component is fully accounted for. The method is suitable for flows in which the turbulence intensity is moderately high.