Real-Time Measurements of Turbulence Quantities With a Triple Hot-Wire System

Abstract

This paper describes a technique for measuring turbulence quantities in highly turbulent flows whose mean-flow direction is not well known. Three orthogonal hot-wire anemometers are used, with individual linearizers. Their signals are simultaneously processed, in real time, by a network of high-speed analog devices. Instantaneous values of u, v, and w (velocity components in laboratory coordinates) are calculated, using the directional characteristics of each wire and its orientation within the test tunnel. The instantaneous velocity signals can be either processed by a digital computer or introduced into a second analog network which separates the signals into mean and fluctuating components and forms any desired double or triple correlation, again in real time.

Real-time processing avoids some of the ambiguities which arise in time-averaging methods: those which are a consequence of the non-zero values of the time-averaged product of the fluctuating terms. Those ambiguities remain which arise as a consequence of non-zero wire length or lack of perfect knowledge of the directional characteristics of the wires. The present work does not address either of those problems, and they remain as the principal hindrances to further improvements in hot-wire anemometry.

Electrical qualification tests of the analog system have shown the output amplitude to be independent of frequency to within 0.1 % up to 20 kHz and the phase shift between components to be less than 1° up to 7 kHz.

Fluid dynamic qualification tests were conducted in a two-dimensional channel flow whose mean and turbulence quantities are believed to be well known. A standard DISA triple-wire probe was used and traverses made with various combinations of pitch and roll on the probe.

Both in the central region of the channel and near the wall, the measured streamwise component of mean velocity was independent of roll and independent of pitch up to 20°. Small spurious transverse components of mean velocity were indicated. In the central region, turbulence quantities were generally within ±10 %, so long as the velocity vector was within 20° of the probe axis. Near the wall, in the region of high velocity gradient (on the order of 100 m/sec/m), the acceptance cone for turbulence quantities became smaller, and roll angle became a more important factor. This latter behavior is attributed to the effect of probe size.

Progress is being made in the development of a compact triple-wire probe, which is expected to improve the near-wall behavior.