APL390 Expt 4

Hot Wire Anemometry (HWA)

Group- 4

Ashutosh Sharma 2020AM10643

Bhavesh Bhati 2020AM10645

Hridey Gupta 2020AM10650

Jai Shankar 2020AM10651

Parveen 2020AM10659

Raunak Choudhary 2020AM10664

Shubham Varshneya
Aim: To obtain the King’s Law constants (A, B and n) and analyze
submerged-jet profile variation

Description of Experiment

Technique (HWA)-
Hot-wire anemometry is a flow velocity measurement technique that relates the cooling
characteristics of a flow to the velocity at that location.

There are two configurations possible for an HWA setup-

The latter is affected by thermal lag and prone to damage at low fluid velocities. Therefore, CTA
is mostly used. In CTA, the voltage required to run a current to maintain a constant wire
temperature is related to the local flow velocity through the King’s Law.
2 𝑛
𝐸 = 𝐴 + 𝐵𝑢

Apparatus-
The apparatus consists of-
1. A setup under a flat table top to generate a submerged jet in air.
2. A hot wire connected across a pair of prongs.
3. A boundary layer probe
4. A setup to mount the probe body and move it by desired amount in two directions.
5. BNC cables
6. Anemometer box
7. A manometer
8. A motor controller to control blower speed
Experimental Procedure
Calibration data was obtained by positioning the probe close to the nozzle outlet and running
the blower at different speeds (corresponding to different jet exit velocities).
The voltage across the hot wire leg of wheatstone was measured and displayed in the
anemometer box.
The pressure difference between the inlet (equal to atmospheric pressure) and outlet of the
nozzle was measured using a manometer which was later related to the flow velocity using the
Bernoulli’s equation.

The calibration was again repeated at the end of the measurement to ensure consistency.

Measurement involved taking voltage measurements at 13 radial locations at 7 difference axial

Thus, different x values will yield different radial extents, increasing with increasing x.

The first 12 radial readings at a given axial location were equally spaced, the last one was a
little farther away to ensure that the velocity has fully died down (by ensuring variation of voltage
signal was negligible between the last two readings).

The constants obtained from calibration and results from the measurement are summarized
below.
Results

Calibration

2
The results of the biquadratic fit (trendline, equation, and 𝑅 value) for calibration data before
and after the experiment are shown below.
Though the data doesn’t visibly change much (note the axes of both plots are the same so that
the plots can be visually compared), the trendline equations are vastly different.

The results of King’s law fit using the “Shooting” method is shown below. Again, though the data
points for calibration before and after the experiment are visually identical, the regression leads
to vastly different A, B and n in the King’s law relation, suggesting that the coefficients are highly
sensitive to the data points.

A third calibration (titled “together”) was done taking both the sets of data points (before and
after the experiment) to get a better estimate of “average” calibration over the measurement.
This was used for finding the velocity field from measurement data.
NOTE: Though the regression results for calibration before and after are very different, the
coefficients from any calibration will give velocity values very close to those from others.

Therefore, the fitted King’s Law is given by-

2 5 6 0.21
𝐸 =− (7. 25 × 10 ) + (4. 18 × 10 ) 𝑈
Measurement-

The velocity profiles at different axial distances are plotted against the radial coordinate. The
measurement was done starting from origin and going along one direction, the velocity values
were mirrored and plotted to show the full sectional profile.

The initial few velocity profiles (starting from x = 1.1D upto 3D) can be identified as the top-hat
velocity profile, it is only after x = 9D that it sufficiently spreads out and starts to take the usual
Gaussian-like form.
Error analysis-
Though the data on the error in Voltage signal (which was being noted manually) was not
collected, the percentage (or fractional) error in velocity due to the same can be formulated as
follows-

2 𝑛
𝐸 = 𝐴 + 𝐵𝑈

Taking differential of both sides-

𝑛−1
2𝐸 δ𝐸 = 𝐵𝑛𝑈 δ𝑈

2 𝑛
(2𝐸 ) δ𝐸/𝐸 = (𝑛𝐵𝑈 ) δ𝑈/𝑈

2(δ𝐸/𝐸)
δ𝑈/𝑈 = 2
𝑛(1−𝐴/𝐸 )

For a more accurate analysis, the errors in the (effectively) linear regression coefficients A and
B can also be included using the expressions for standard error in Linear regression
coefficients-

2
σ
𝑆𝐸(β1 ) = (𝑛 − 2) · 𝑆𝑥𝑥

2 2
𝑆𝐸(β0) = σ · (1/𝑛 + 𝑥̄ /𝑆𝑥𝑥)

Although, the process of calibration seemed far more accurate than the measurement.
Discussion
The calibration resulted in an equation with n = 0.21, which was significantly lower than the
typical range of n in King’s Law (0.4 to 0.5).

A strong sensitivity of King’s law coefficients (A and B) on the data was observed. Though the
calibration data (before and after the experiment) was visibly very similar, the calibration
relations were very different.

The Velocity profile obtained from the measurements well matched the theory. Close to the jet
outlet, it started off as a top-hat velocity profile and became smooth and spread out for higher
axial distances.

Also, it was the profile of magnitude of velocity that was measured and not individual
components, as opposed to PIV.

The Voltage signal fluctuated a lot during measurement, making manual measurement very
difficult.
This also means that the electronic drift was the main contributor to experimental error as
calibration looked comparatively much more precise.

Logging the data as several time-series signals (one at each x and r location) would be much
more practical. Since the flow is statistically stationary, the signal can be time-averaged (which
should converge to the ensemble average) to obtain a much better estimate of flow velocity by
reducing the random errors significantly.