·Table of Contents
·Industrial Plants and Structures
Acoustic Emission Detection of Underground Pipeline Leakage
Shifeng Liu, Luming Li, Jian Cui, Tie Li
Department of Mechanical Engineering, Tsinghua University, Beijing, China
Bang Xian Li, Center of Boiler and Pressure Vessel Inspection and Research, Beijing, China
Aihong Dong, Beijing Organic Chemical Plant, Beijing China
Acoustic emission detection of underground pipeline leakage was studied. Several methods of locating the leakage sources, based on attenuation and waveform analysis, were developed and tested in site. The ability of extracting leakage signal from various background situations was also evaluated and tested. The flowing mediums of the pipeline investigated are air, water, and industrial gases. Different size holes were drilled on the over 100 m length research pipeline for artificial leakage sources. The real sources of industrial pipeline leakage are studied too. The differences between underground pipeline and above pipeline are compared such as acoustic influences by coating of anticorrosion and earth. The waveforms of the leakage signals are studied to get better understanding of the generation of the acoustic emission signal caused by the leakage and the influence of the signal propagation. FFT, Wavelet and neural network pattern recognition are employed to study the leakage signals and background noise. The location precision and the sensitivity of the acoustic emission detection are estimated by both theory and experiment. The diameter, materials, thickness and the coating of the pipeline have different influence on leakage signal propagation. The sensitivity is strongly depended on the pressure of the working medium and the size of the leakage. For some steel pipeline a leakage with 1mm diameter hole could be detected by a sensor placed over 100m away from the leakage source. The error of the location can be better than 10% of the distance of two sensors. The results of theory analysis and experiments show the positive conclusion on the ability of acoustic emission detection of underground pipeline leakage.
Underground pipeline leakage inspection is an open problem. Industrial users such as gas pipeline, oil pipeline and water pipeline, are waiting impatiently for the incoming innovation technology of underground pipeline leakage inspection. Scientific workers and engineers investigate and develop several technologies such as gas detection, detection of sound above ground generated by leakage, for underground pipeline leakage detection. Unfortunately all current methods can not meet the minimum requirement of industrial users. Most of them don't have enough sensitivity to detect at least applicable amount of leakage even sometimes do well for above pipeline leakage detection. Another problem is the poor ability of leakage source location. For example the method based on the gas detection can only detect the leakage after enough accumulative total amount, too late for prediction of leakage. Further more the methods based on the gas detection usually are poor in the location of the leakage source of the underground pipeline since the leakage gas may flow a long distance before escaping to above ground. Another example is the method based on the leakage sound detection in the air. Theoretically the sound is generated together with leakage and is spread in the air so could be detected. But after transmitting through the crust soil the sound become very weak and is difficult to be detected. There are some other methods such as negative pressure detectors, infra-red thermography, flow/pressure change detection and mass/volume balance. Dr. Jun Zhang list some other pipeline leakage detection methods.
Acoustic emission is one of underground pipeline leakage detection methods investigated. The sensitivity and the ability of location for acoustic emission method at moment are not so satisfied by the application users. Miller and Pollock etc. investigated the acoustic emission method for underground pipeline leakage detection and gave a very encourage result. The accuracy of the location is about 1 foot and the sensitivity is about 25 feed sensor spacing in their field trials. In our investigation some signal processing technologies (FFT, wavelet analysis, fractal and AR spectrum) were used for extract weak leakage acoustic emission signals for improving the sensitivity. A model of energy-attenuation was established for better location accuracy. Further a pulse-enlarge method was designed to enlarge the leakage signal and then improve sensitivity.
The underground pipeline over 100m was prepared by Beijing Organic Chemical Plant. Part of the pipeline is above ground for generating leakage. Some holes, diameter 1mm and 2mm, were drilled on the pipeline with different distance. Leakage is generated when a thin plug is pulled out of the hole and stopped when the plug is pushed in the hole. The water or air is the media inside the pipeline. The pressure of the pipeline is generated and controlled by a pump or compressed nitrogen. The acoustic emission instrument is WAE200, a waveform data base instrument from Soundrey Corporation. Wide band and narrow ban sensors were used. After comparing the results of the signal analysis the instrument system (sensors, preamplifiers, main amplifier and soft filters) was set sensitive to frequency window 15KHz-40KHz.
3. Signal analysis
As the distance between the sensors increases and as the pressure inside the pipeline decreases it is more and more difficult and even impossible to identify leakage from just looking and comparing the waveforms of the signals with and without leakage since the signals are very weak and the amplitude error is large considering the coulping conditions. Some signal analysis methods are investigated for developing a technique to identify leakage from weak signals. The signal analysis methods selected are those based on frequency analysis, i.e. independent of amplitude. FFT, wavelet analysis, fractal and AR spectrum are investigated. Figure 1 shows results of the signal analysis investigated. Figure 1a is the waveform and FFT of background noise without leakage, Figure 1b is the waveform and FFT of leakage signals, Figure 1c is the results of wavelet and fractal, Figure 1d is the AR results, blue line for background noise and red line for leakage signals, It is obvious that AR spectrum and fractal can give the right classification of background noise and leakage signals. Further the applicable distance of the sensors and the pressure of the pipeline are investigated for fractal analysis method. Table 1 &2 show the results. From the table1 and table 2 the criteria of fractal for leakage identification is fractal number over -1.8. From the results of table2 the narrow frequency window 15-40KHz has better fractal analysis results, over criteria -1.8 more. Maximum distance 86m in our trial the leakage can be identified in Table 1. Figure 2 are the results of AR spectrum for various pressures. It is encourage that both fractal and AR spectrum methods are independent of pressure inside pipeline and distance between sensors. Sensitivity could be increased by applying those signal analysis methods.
c:wavelet(db3) and fractal, left for noise and right for leakage, x:level of coefficients, y:value of the coefficient
above:low frequency part, below:high frequency part,Kis the fractal number
Fig 1: Signal analysis results, a: waveform and FFT of noise, b:waveform and FFT of leakage
C:wavelet and fractal, d:AR spectrum
|Table 1a: results of fractal analysis for dia 1mm leakage|
||-1.6 ~ -3.4
||-2.8 ~ -3.3
|Table 1b: Results of fractal analysis for dia 2mm leakage|
|Table 2: fractal numbers for noise signals(8 groups of noise signals) at different windows of frequncies
4. Leakage Location
The principle of the location is the attenuation. For increasing the accuracy of the location some improvement method were investigated. The first error considered is the error caused by the coupling conditions and system error such as different amplifiers of two channels. The simple improving method is calibration by background noise to get suitable compensate gain to the instrument amplifier. The little complicated compensate way is before doing calibration by background signals filters the signals within the frequency windows 15-40KHz, then do simple improving method. Table 3 shows the results. The accuracy of the location reached is about 6% of sensor spacing.
||8.2/23.6m = 0.3475
|Table 3: Location accuracy improving results.|
By signal analysis the ability of leakage identification and the accuracy of the location could be improved. Fractal analysis based on the spectrum and wavelet pre-processing and AR spectrum methods show the better results than FFT. It is encourage that those methods show some independent of distance and pressure. For location accuracy gain compensate and narrow band signal filtering are useful ways to increase the accuracy. For engineering application the sensitivity and the location accuracy still need to increase.
This project is supported by China national scientific and technology foundation. Pipeline was prepared by Beijing Organic Chemical Plant, Acoustic emission instrument by Beijing Soundrey Corporation.
- Dr Jun Zhang, Designing a cost-effective and reliable pipeline leak-detection system, Pipe & Pipelines International, January-February 1997, pp20-25.
- R. K. Miller, etc., A reference standard for the development of acoustic emission pipeline leak detection techniques, NDT&E International 32(1999)1-8.
- Huo Zheng, Cheng Chuimei and Zhu Renxiang, Acoustic emission inspection of pipeline leakage, Non-destructive Testing Journal( in Chinese), Vol.19, No.4,April 1997,pp105-107.