Introduction
Two-phase flow measurement is a challenging task for transit
time ultrasonic flowmeters. Due large air bubbles or particles, existing
ultrasonic flowmeters may suffer to the problem of unstable readings, or even
stop working completely when the signal-to-noise ratio (SNR) drops below a
limit. This paper introduced a more stable and robust ultrasonic flowmeter
technology, which can better cope with liquid flow with enriched gas bubbles.
The flowmeter uses broadband ultrasonic transducers and multi-frequency digital
signal processing to enhance the signal to noise ratio under gas bubble
interference. The same flowmeter can thus measure both dirty and clean flows,
widening its application area.
Theory
In a transit time ultrasonic flowmeter shown in Figure 1, two
ultrasonic transducers are placed at an angle θ the pipe axis. Ultrasound wave is
sent first along the flow direction, e.g., from A to B with a transit time of tA,B.
Another ultrasound wave is then sent against the flow direction, from B to A
with a transit time of of tA,B. The wave traveling in the same
direction of the flow has a shorter transit time. By determining the time
difference of the wave transmissions, the average flow velocity v can be
determined as
(1)
where K is is a proportional flow factor determined by the
distance of the two ultrasonic transducers and the angle θ. The flow rate can easily be obtained by multiplying the
cross-sectional area of the pipe.
The multi-frequency ultrasonic flowmeter operates in the following
measurement steps:
(1) Adaptively adjust the number of transmitted pulses
according to SNR;
(2) Transmitting a burst of signal, which includes
multi-frequency pulses in sequence;
(3) The burst of signal is transmitted and received by
broadband ultrasonic transducers; and
(4) The received signal is processed to get ultrasonic
transit time and flow rate.
Refer to Figure 2(a), a multi-frequency sinusoid waves are
transmitted in sequence and received by the broadband ultrasonic transducers.
The sent signal (referred as a burst) includes a sequence of 1.0 MHz, 0.8 MHz,
and 1.2 MHz sinusoid waves (referred as pulses), each with 10, 8, and 12
cycles, respectively. The total length of this burst is 30 μs. During this
short period the flow turbulence (i.e., jitter) does not influence the
measurement much. 
Passing through the sending transducer and the flow path, the
burst is received by the receiving transducer. Due to the broadband nature of
the ultrasonic transducers, the received signal keeps the frequency
information, as shown in Figure 2(b), which is advantageous for the transit
time detection.
In step (1) of “adaptively adjust the number of transmitted
pulses according to SNR”, the signal-to-noise ratio of current measurement is
calculated by
(2)
where Psignal,
Pnoise, Asignal,
and Anoise are the power and
amplitude of signal and noise, respectively. When SNR drops below a limit, the
pulse number in a burst will be increase to compensate for the interference.
To effectively calculate transit times, the received signal
is cross correlated with a reference signal. For the ultrasonic signal x(n) and
a reference signal y(n), the cross correlation is defined as
The maximum point of rxy(l) indicates the
ultrasonic wave’s transit time, and must be calculated to a high precision
(e.g., in nano-second or pico-second) even under severe interference, which can
be achieved either by direct time domain linear correlation or by using an
FFT-based frequency method, together with proper data interpolation technique.
The reference signal is chosen as a received signal when the SNR is good (e.g.,
SNR > 20 dB), which can be stored in the flowmeter during factory
calibration or field set up.
The
effectiveness of the MFUF is explained by comparing with two prior art methods.
The received signals and their corresponding cross-correlation curves are shown
for the three methods, as shown in Figure 3. Precision and robustness of the
method is determined by the amplitude and the sharpness of the
cross-correlation curve (as marked in red circles in the figure).
Compared with transmitting a single-frequency pulse with 6
sinusoid of 1 MHz (Figure 3(a)), MFUF improves the maximum value three-fold
(from 0.44 to 1.7, Figure 3(c)). This is achieved by transmitting four times
more ultrasound energy. On the other hand, by transmitting a single-frequency
pulse with 30 sinusoid of 1 MHz (Figure 3(b)), the maximum value of the cross
correlation curve is larger. However, the maximum is ambiguous to identify.
Especially when SNR is low, cycle-skip might occur and leads to big measurement
error. The MFUF gives a high and sharp peak, so that the transit time can be
uniquely identified with high precision.
EXPERIMENTAL RESULTS
The experimental
setupis shown in Figure 4. The flow system includes a water tank, a centrifugal
pump with capacity of 4m3/h, controlled by an inverter. The liquid
volume flow rate can be adjusted up to a maximum of 54 L/min, corresponding to
a flow speed of 2.9 m/s. Gas bubbles are introduced into the flow by a gas pump,
which can be controlled.
An ultrasonic flowmeter and the MFUF are connected in series
in the flow path. The ultrasonic transducers in MFUF have a centre frequency of
1 MHz and -6dB bandwidth of 57%. A function generator and an oscilloscope are
used to stimulate and receive the ultrasonic signals, respectively. Data
acquisition, digital signal processing, and display are conducted in NI
LabVIEW.
A series of
water flow rate measurement was conducted, as shown in Figure 6. It’s apparent
that when the flow is free of gas bubble interference, both flowmeters work
properly with high precision.
When gas bubbles were introduced into the flow system, both
flowmeters were immediately influenced by the interference (Figure 7). In
Figure 7(a), a small proportion of gas bubbles (~1% in volume) were first
introduced near the time point of 60. Then a big proportion of gas bubbles (~5%
in volume) were introduced near the point of 90. The actual water flow rate was
kept constant at about 20 L/min. It’s apparent that the MFUF technology can
better tackle the problem posed by two-phase flow.
In Figure
7(b),when a big proportion of gas bubbles (~5% in volume) were introduced
between the points 75 and 230, and the liquid flow rate was adjusted by the
pump, the benchmark flowmeter stopped working under the severe interference but
the MFUF with improved SNR still gave acceptable measurements for three flow
rates.
For two
phase flow, the flowmeter showed superiority in results.
CONCLUSIONS
By transmitting multi-frequency ultrasonic pulses using
broadband transducers, improved SNR and stability is achieved with computation
efficiency. More ultrasonic energy(more pulses) is transmitted under severe
noise interference, and the SNR is improved several times. MFUF can measure
both dirty and clean flows, widening its application area and simplifies the
flowmeter product portfolio. When the two-phase flow contains 10% or more gas
distributed evenly in the liquid, and the gas bubbles are small in size,
ultrasound signal is completely blocked from transmission in liquid, the
transit time ultrasonic flow meter cannot work for this situation, and other
measurement technologies need to be developed.
Reference:Authors: Fan Shunjie,
Zhuo Yue
Journal: IEEE 2011
By
Satya Swarup
Roll 89
Branch-Instrumentation and Electronics
No comments:
Post a Comment