22 kHz and 44 kHz second harmonics appear in the fiber sensor signal 34 Figure 4.12: Coupling of 27 kHz acoustic vibration in fiber sensor and AE sensor. 36 Figure 4.15: AE sensor peak detection in acoustic vibrations 36 Figure 4.16: Fiber signal peak detection in acoustic vibrations 37 Figure 4.17: Coupling effect of acoustic vibration 57 kHz lower than resonant.
Non-Revenue Water (NRW) 1
On the other hand, the fiber sensor signal and its background resonant peaks are shown in Figure 4.10. In the other sides, the fiber signal is not varied much in Figure 4.16, and the background resonance signal also creates a similar trend line. In Figure 4.22, a Gaussian broadband can be observed at the low frequency range, which is around 0 to 40 kHz.
There are two lower peaks that fit the position observed in Figures 4.22 and 4.26.
Existing pipeline monitoring syste 3
Literature review on acoustic sensor 4
Literature review on optics fibre sensor 5
On the other hand, the optical fiber interferometer method is more useful than the single fiber sensor in terms of stability and accuracy. An optical fiber with an array of gratings embedded inside the core can be considered as an FBG.
Problem statement 7
An OTDR is an optoelectronic instrument by injecting a series of laser pulses into a special fiber and measuring the time dilation of the scattered light pulse. However, the OTDR is expensive and the operating frequency range is limited by measuring over long distances.
Aims and Objective 8
Erbium Doped Fibre Laser 9
In the metastable state, the excited ions will de-excite and fall back to the ground state by releasing a radiative energy such as photons or non-radiative energy such as phonons. The photons released by spontaneous emission will propagate in a random direction that is incoherent in phase and generate a wide spectral width. These emissions will then be amplified by the rest of the excited Erbium ions, and the end product of this process is called amplified spontaneous emission (ASE).
On the other hand, stimulated emission occurs when there is an incoming photon with the same energy as the band gap between the metastable state and the ground state. Then the emitted photons will be coherent in nature with the incoming photon, and this usually helped to amplify the incoming photon source by stimulating more metastable state ions to de-excite and release more photon energy (Bransden & Joachain, 2003). The two external connections on the left side are used to release the ASE to the fiber sensor and inject the signal to the photodetector.
Laser dynamic behaviour- Turn on transient 11
In Figure 2.3, a practical EDFL is shown, and the circuit board in the middle is a laser diode that pumps the laser source into the EDF. Where n is the photon number, D is the electron number between level 1 and level 2, 𝐷 = 𝑁2− 𝑁1, W is the spontaneous emission coefficient and k is the photon decay loss.
EDFL operation near threshold 14
During recovery, the on state will be reset to the “on” state and a clear pattern in terms of amplitude will be seen.
Loss Modulation 15
Lossy modulation can be modeled by considering a two-level system as shown in Figure 2.9. Then, simplify these equations by introducing the difference in the number of electrons between the two levels, D, and get the general formula. For general perspective, the arbitrary wave function can be included and modified by the user.
Where m is the amplitude modulation number and G(t) is the pump modulation function, which can also be related to the flow-in-pipe acoustic wave model here.
Failure investigation 17
The magnitude of the stress can be less than pipeline metal yield strength, but the failure can happen as soon as the process time is long enough for the defect inside the metal to reproduce. On the other hand, material degradation, reversible process, manufacturing process and leakage flow problem also generate some troublesome problem with the pipeline (Wild & Hinckley, 2008). Next, the reversible process such as crystallographic phase transformations, solidification, thermoplastic effect and friction between surfaces can also be the underlying factors that cause a crack.
This research will focus on leakage detection on the pipe and deepening the correlation between multiple point measurements. In the next chapter, a real pipeline is installed in the lab to test the sensor in the simplest case. Then the sensor is verified by introducing a more realistic situation and trying to prove its capabilities.
Implementation on the Pipeline 20
The function of the WDM is to separate and combine the 980 nm pump laser in the EDF. From the fiber free end to the end of the WDM will produce a linear laser oscillating cavity as shown in Figure 1.5(b). The long SMF attached to the WDM will act as a sensing arm in the experiment.
Any acoustic or vibrational wave that couples and causes loss to the fiber cavity will cause the transient effect.
Pipeline construction 22
Vibration modulation 23
The acoustic vibrations produced by the piezoelectric transducer show a strong coupling effect in the fiber signal. The piezoelectric transducer drives the function generator, and its frequency range can vary from 2 kHz to 2 MHz, where the frequency range is the limit from the speaker.
Leakage creation 25
Multiple points measurement 25
Signal Analysing 26
The photodetector signal will be transmitted by the digital output post and a coaxial cable to a digital oscilloscope manufactured by Textronic TDS 1012B series (Tektronix, 2016). The result obtained is in the time domain, which will be converted to the frequency domain using the Fast Fourier Transform (FFT). This can improve the data and clearly show the influence of a specific frequency when the external acoustic wave is disturbed by the laser signal from the laser pump.
Next, a simple smoothing and filtering method can be introduced for comparison between the single point results. One of the most common signal analysis methods used here is the robust local regression and it can be manipulated using MatLab 2013 (Mathworks, 2000). The reason for choosing this filtering method is to make use of the robust weight function, which makes the calculation resistant to the outlier.
Sensor characterization 28
These two peaks form the benchmark for the following experiment signal and are used to justify the difference between the commercial sensor and the fiber optic sensor. The background peak is caused by vibrations in the data cable, which are irrelevant with external acoustic vibrations. This indicates that a background resonant can act as a measure of fiber sensor data and then be used to observe the external modulation effect.
Airborne sound vibration 30
If we compare the two graphs, both peaks are moving uniformly, but the FFT plot of the optical sensor has a second peak and multiple peaks are also moving. If the higher order frequency has started to dominate due to the coupling effect of external vibrations, more than one peak can be observed in the fiber sensor signal but not in the AE sensor signal. In Figure 4.9, the resonant background signal varies by about 0.5 dB, which is more stable compared to the peak of the signal, which has a moving range of 1.2 dB.
The dynamic range of the fiber sensor signal is about 20 dB and 10 dB for the resonant background signal. An important trend can be observed, which is that the signal does not weaken after 10 kHz, but is stable within the 5 dB range. This made the fiber sensor more reliable in practical signal measurement of small and extreme vibrations.
Acoustic vibration 34
The reason is that the vibration of the speaker comes to an unstable and weak vibration when it meets the frequency limit and causes signal degradation. After observing all the figures, each peak of the fiber signal is moving along the signal peak of the AE sensor, which also occurs exactly the same phenomena in section 4.2. This experiment provides the best evidence for extending the measurement limit of the fiber sensor up to 50 kHz.
In Figure 4.15, an increasing trend of the AE sensor peak is observed, which means that the signal peak continues to increase as the external modulation frequency increases. From the overall performance, the background AE signal is not increased much compared to the AE signal.
Ultrasonic frequency region 37
When the external vibration increases to 67 kHz, the signal peak superimposed with original background frequency becomes -59 dBm. In other words, the limitation of the fiber sensor is determined by the background resonant frequency. Fortunately, the background resonant frequency can be manipulated and shifted by adjusting pump laser power and the length of laser cavity.
Frequency calibration of pipe leakage 39
The water leakage directly hit the fiber and caused additional vibrations in the sensor area. Now the part of the fiber that was previously at the top of the leak is moved slightly towards the edge of the hole to obtain an "off-center signal" where there is no direct contact between the water and the fiber. Then the fiber is now wound twice through the front of the hole defect with equal water pressure.
Note that the fiber parts passing through the hole are not closely spaced to investigate the effect on the fiber signal. The collected signal replaces the dominant resonant frequency in Figure 4.22 and moves to about 55 kHz in Figure 4.26. Finally, the fiber is wound through the hole three times, creating three rings on top of the leak.
Multiple point measurement 44
Now, the selected threads are shown in Figure 4.31 with a circle and the split is about 1.62 meters long. The purpose of not selecting fiber 2 and fiber 3 as a pair is to make the distance between them almost equal and make the time delay in both signals the same. Finally, each data set will be applied the same method to identify its time delay and will be plotted in Figure 4.32.
The following setup is similar to figure 4.31, but the target fiber pair changes to fiber 2 and fiber 4. The time delay measurement of fiber 2 and fiber 4 in time domain signal is exactly the same with the method used in fiber 1 and fiber 3 Finally, all 9 set data analyzed and combined in figure 4.36, follow the same method in setup 1.
Conclusion 49
Future work and recommendation 50
Available at: https://www.kistler.com/is/en/products/components/accelerometer-sensors/#k__beam__triaxial__m_e_m_s__capacitive__d_c__accelerometer_2_200g_. Non-Revenue Water, Service Impact, Environment and Finance, Putrajaya: Kementerian Tenaga, Teknologi, Hijiau dan Air. Available at: http://www.span.gov.my/index.php/en/statistic/water-statistic/non-revenue-water.
