Measurement Technique

Principle of Sensing:
Exploit any property which is significantly different for the two phases

Techniques:
• X ray,  ray, Neutron, RF
Additionally:
• Conductivity
Flow rate
• Capacitance • Ultrasonic
Pressure
• Optical • Magnetic Resonance
Temperature
➢ Photographic • Chemical/Electrochemical
➢ Attenuation

1
Measurement of void fraction
• Preliminary idea of void
fraction can be obtained by
two quick closing valves

• Valves are closed quickly to

enclose two phase mixture in
between\

• Allow gravity separation

using buoyancy

• Associated with error

2
 Digital System
1. Sensors

2. Processing circuit

3. Data acquisition
system

4. Computer
Signal conditioning (multiplexing, amplification,
filtering, compensation etc.)
Signal processing, feature extraction, qualitative
information, quantitative estimation

3
Conductivity probe

Gas- liquid up flow through concentric annulus

• Sensor- Parallel plate conductivity probe

• Signal analysis through PDF

Single sensor, single feature

4
Conductivity probe

Parallel Wire Conductivity Probe Ring Conductivity Probe

5
Details of the processing circuit:

6
Slug
flow

7
Churn
flow

8
Optical probe

Simultaneous flow of two immiscible liquids

• Sensors- Conductivity probe, Optical probe

• Signal processing through
• PDF (mean, skewness, kurtosis)
• Wavelet (Different frequencies, standard
deviation)

Single sensor, multiple feature

9
11
 Bubbly flow (USW=0.05 m/s; USK=0.05 m/s)
0.20
1
0.15 M 0.411
V/Vmax
0.5 P  0.172
0.10 S 0.573
D
Blue
F 0.05
0
0 20 Time (s) 40 0.00
0.0 0.2 0.4 0.6 0.8 1.0

Churn turbulent flow (USW=0.05 m/s; USK=0.18 m/s)

0.20

1
M 0.714
V/Vmax

0.15

P  0.164
0.5
D 0.10 S -0.732
0 F 0.05

0 20 40
Time (s) 0.00
0.0 0.2 0.4 0.6 0.8 1.0

V/Vmax
13
 Core annular flow (USW=0.05 m/s; USK=1.2 m/s)
1.0 1.0

0.8 M 0.045
V/V max  0.0002
0.5
P 0.6
S 1.0
D 0.4
Bluish
0.0 F 0.2
0 20 Time (s) 40
0.0
0.0 0.2 0.4 0.6 0.8 1.0

V/Vmax
Dispersed bubbly flow (USW=1.2 m/s; USK=0.05 m/s)
0.5

1 0.4
M 0.920
σ 0.018
V/Vmax

0.3
P
0.5 S -0.171
D 0.2

F 0.1
0
0.0
0 20 40 0.0 0.2 0.4 0.6 0.8 1.0
Time (s)
V/Vmax
14
Estimation of void fraction

Point based measurement

Single Needle Contact Prob Deformation of bubble at probe tip

15
 Voltage
Estimation of void fraction
time
a. Ideal Signal

 po int =
 t gas

t + t
gas liquid

Voltage
time
b. Actual Signal

16
Estimation of void fraction

17
Multisensor measurement and data
fusion (MSDF)
Data fusion

Identification of core
MSDF is useful in handling annular flow- a simple
complex flow situation
involving multiscale and time implementation of multi-
variation
sensor measurement
Multiple sensor, comparison and data fusion
of sensor outputs

18
 Centerline point and wall mounted
conductivity probe

Central probe

1
Pipe wall
8 2

Pipe
7 3
wall Wall mounted
probe
Wall mounted
6
probe
4
5

19
Conductivity and optical probes

20
 Flow regime Identification through data fusion
❑ In the experiment three
Multiple types of flow regimes (Slug,
sensor, Churn, Annular) have been
multiple observed.
❑ The mean, standard
feature deviation, skewness and
Information along a
Near wall chordal Length (Parallel kurtosis of the raw signals
Information (Ring wire probe))
probe) have been used for
clustering which separates
Statistical Statistical
data of similar properties
Features Features
based on those.
Feature Fusion ❑ The mean(Me) is the
average value of the time
Algorithm for Clustering series voltage data counted
throughout the total
Different Clusters timespan of the signal.
• Standard deviation(Sd) is used to measure the variability of the signal. It shows how much
variation or dispersion the data have from the mean value of the time series voltage data.
• Kurtosis(Ku) is a measure of how outlier- prone a distribution is. Skewness (Sk) is utilized
to measure the asymmetry of the data around the sample mean
21
 Data fusion in tomographic measurement

Data from
i) Change of flow
multiple regime
measuring
station ii) Void fraction wave
iii) Phase velocity / flow
rate
Tomography is a very versatile technique which can provide the information of
a non-homogeneous medium at a minute level of space and time using
multiple sensors and the principle of data fusion

23
 Wire Mesh Tomography

Wire mesh sensor Provides detailed two

phase distributions at a
cross section

Sensors based on resistance (conductivity) and capacitance

principle have been used very extensively. A special case of
conductivity based tomography is wire mesh tomography.
24
25
Area average void fraction vs time plot for Taylor Bubble
1
0.9
0.8
0.7
Note: Shows passage of bubble
C.S Avg. V.F

0.6
0.5
0.4
0.3
0.2 X axis- Time
0.1
0
0 10 20 30 40
Tim e fram e
Slow initial rise indicates Nose
region.
Flat top indicates cylindrical body
Sharp fall indicates blunt tail portion
Sectional view shown below as
reference.

27

27
Capacitance based tomography

28
Image analysis

Step 1 Step 2 Step 3

Read Original image Read background image Subtract original image from background

Step 4 Step 5 Step 6

Convert subtracted image to binary Find contour Original image with contour

29
Signal analysis
Resolution: 1280920
1000 FPS

Pressure signal is recorded at 1000 sample per second

30
 Normalization of pressure
Step 2 Divide pressure with Pmax
signals

Step 3 Plot PExp/Pmax

Step 1
Find maximum of temporal pressure series

31
 Finding Probability Density Function (PDF) of Normalized
pressure signals

Step 1 Copy temporal Step 2 Go to Data tab in

pressure series in separate menu bar of excel.
sheet.

Step 3 Go to Data
analysis in right of Data.
Step 4 Select Histogram in data analysis
tools as shown below.

32
 Finding Probability Density Function (PDF) of Normalized
pressure signals

Step 6
In new sheet, you will get bin and
frequency data.
The bin data is your normalized
pressure and PDF is calculated by
dividing frequency with total no of
data points.
Step 5
In Histogram dialogue box, input
and output appears as shown below.
Select normalized pressure in input
range, and output will be in new
sheet

33
 Finding Fast Fourier Transformation (FFT) of Normalized
pressure signals

Step 1 Step 2
Open Origin pro>copy paste time and In Open dialog, go to plot and select only
normalized data in A and B column > select both column amplitude as shown below. After clicking OK,
> Go to analysis tool > Signal Processing > FFT > Open you will get frequency and amplitude data
dialog shown in next step.

Step 3
Delete zero frequency data> copy all data in
Step 4
excel and plot amplitude vs frequency data.
Plot Amplitude vs frequency

34
Circuit for analog
signal processing:

38
 Analog System

Morse and Ballou, 1951 Recording

Fluidized bed 3. Magnetic Oscillograph with inking brush
1. Sensor- Strain gauge pressure transducer Characterization
2. Processing circuit 4. Polar planimeter+ Manual counting

39
Propagation of Rewetting front- Data recording and typical
analog processing of signal

Temperatur

Time
e

40
 Techniques for image Techniques for signal processing
processing • Visual inspection
• Edge detection • Probability density function (PDF)
• Pattern recognition • Power spectrum density function
(PSDF)
• Cross correlation
• Fourier transform- FFT, STFT
• Auto correlation
• Wavelet analysis
• Fuzzy logic based techniques
• Cross correlation • Chaos analysis
• Auto correlation • Neural network
• Fractal analysis • Complex network

41