| |
Curve Fitting
A number of functions for curve fitting the hotwire calibration data have been proposed
in the literature. These functions are individually applied to the two wires of the  probe. The most commonly used functions are polynomials, power laws (King's
law) and look-up tables. Low order polynomials, say second or third, do not show an
oscillatory trend between calibration points. Jorgensen (2002) has shown that for a wide
range of velocities  power law fits are less accurate than polynomial ts
and show a systematic error distribution. The fourth- and fifth-order polynomials are
preferable and gives less error, almost of equal accuracy. The reported maximum errors
for these polynomials are 0.74 and 0.53% respectively.
For velocity measurements in the lower range, say 0.5 to 4 m/s, a number of special
calibration procedures is available. In present investigation, two different approaches
were examined, but the difference between them was found to be negligible. These
approaches are based on the linearizer function (recommended by DANTEC) and a high
order polynomial. In linearizer approach the mathematical relationship used is
Here A,B,C,D and E are curve fitting parameters. They are selected to give a fluid
velocity equal to 10 m/sec for a wire output of 10 V. The quantity y is the output
quantity equal to normalized velocity of fluid, defined as
|
