A predominant problem in the water and wastewater industry is leaking below-ground pipes. If these leaks are not located and treated, catastrophic situations and disruption of services may result. At a lower level, uncorrected leaks can represent considerable cost, as well as waste of a valuable natural resource in regions where water is a scarce commodity.
In the past, the common practice was to detect leaks or faults in buried pipelines using tracer gas, or flow and pressure modeling. More recent techniques such as infrared thermography, ultrasonic and electromagnetic scanning, and ground-penetrating radar (GPR) have being used. Although these methods may be accurate, a more reliable, real-time, accurate monitoring and reporting system for buried underground pipelines would be beneficial for drinking water and wastewater utilities by cutting cost of long term repairs and maintenance. Moreover, the attendant reduction of catastrophic failures and their associated health hazards would also benefit the environment and society as a whole.
The use of Acoustic Emission (AE) for inspection and monitoring involves using acoustic waves at frequencies above the audible range to locate flaws in the structural integrity of the pipeline. These acoustic waves are generated by the fluid escaping from the leaking pipes. During the leak detection process, any noise produced by escaping fluid through small openings such as cracks or holes in the pipe will generate acoustic energy. Additional acoustic energy may be produced by the interaction of the escaping fluid with the surrounding soil. This produces localized impacts that mimic acoustic burst type signals rather than a typical continuous leak signal. ?
The New Jersey Institute of Technology (NJIT)/U.S. EPA Edison Pipeline Testing Facility was constructed with $2.5 million in funding from the National Standards Foundation and the Strategic Environmental Research and Development Program (SERDP). The Pipeline Test Facility in Edison, N.J., is composed of five different pipelines, a test-pit and above- and below-ground storage tanks. NJIT investigated AE methods for pipeline leak detection and location in pressurized underground pipelines with no flow. The study was funded by SERDP and delivered very encouraging results, with leaks detected to within an accuracy of 1 ft. The results reported here were part of a follow-up study funded by EPA to investigate the AE methods for pipeline leak detection and location in pressurized underground pipelines with flowing water. ?
The NJIT team used AE leak detection techniques to locate leaks in an underground pipeline under flow conditions. The objective of the study was to determine the effects of the noise produced by water flowing in the pipeline simulating actual operating conditions using AE leak detection. In this study, the tests were carried out on a 1,000-ft long, 12-in. diameter, Schedule 40 welded steel pipeline. The tests were performed by the NJIT team using sensors developed by MISTRAS Group, i.e., two Model PLS-1, 15 kHz peak sensitivity piezoelectric sensors were used to receive the AE signal. These acoustic sensors were located along the length of the test pipeline, so they were positioned to intercept known leak locations. Access to the buried pipeline was attained through vertical access tubes installed at discrete locations along the pipeline using 17 PVC pipes as access points for the sensors. The graphic to the right provides a schematic representation of the pipeline layout.
To receive the data, a Digital Signal Processor (DiSP), also developed by MISTRAS Group, was used. The DiSP workstation measures AE signal features and records complete waveforms on as many as 56 channels. It also records as many as eight parametric inputs (such as pressure and flow rate). In this particular test, only two digital AE channels and two parametric channels were utilized. The DiSP instrument pinpoints leaks by using coincidence detection. This determines the precise time (accurate to 0.25 μs) when a leak signal crosses a specific voltage threshold of a given sensor. ?
MISTRAS Group has made an operational modification for use in detecting leaks, which is called tuned linear location. The tuning automatically adjusts the sensitivity of each channel by altering the voltage trigger threshold to produce hits within a preset range, i.e., a form of floating threshold. Coincidence detection is very robust in the field because it is relatively insensitive to dispersion, whereby a signal increases in duration and decreases in amplitude in a frequency-dependent manner as it propagates.
Theoretically, coincidence detection works best with pulses, and does not work well when the leak signal is continuous and has no discrete characteristics. However, the random impact of soil with the surface of the pipe at the leak location produces a pseudo-pulsed signal, which allows for successful testing. During one successful test, a leak location test performed with a sensor spacing of 215 ft, and with a leak imposed at 113 ft from sensor 1. During this particular test, the flow rate of the water in the pipe was 180 gallons per minute, the pressure in the pipe was 39 psi, and the leak rate was 15.4 gallons per hour. The vertical spike illustrates that the leak was located at approximately at 112 ft from the Channel 1 sensor, i.e., an error of minus 1-ft. The overall accuracy of the computed locations of various detected leaks with discernible spikes for comparable sized leak rates ranged from minus 1-ft m to plus 2.6 ft, or approximately plus or minus 2 percent.
In another series of experiments conducted in 2012 at the same pipeline test facility, MISTRAS Group assessed the leak detection performance of several AE location techniques and several sensors and frequency bands. Different leak types such as jets and sprays were described. In efforts to make more realistic simulations of actual leaks, sand and small stones were used in various test runs to baffle the water exiting the leak. A high-capacity pump used to circulate water in the loop generated considerable background noise which reduced the leak-detecting capability.
In a test with lower-frequency sensors (Model AE accelerometers, resonant at 30 kHz instead of the previous 60 kHz AE sensors), a lower-pressure leak was well detected using a greater sensor spacing. The distance from this leak to the more distant sensor was 136 ft. As in many of the tests carried out, the physical mechanism creating the emission was not the actual egress of liquid from the pipe. In this case, the emission was caused by small droplets thrown 5 to 6 ft into the air by the pressure behind the leak, falling back and impacting the pipe.
The above discussion shows that many variables are relevant to the AE generated by leaks in pipes and its detectability. These include sensor type and frequency, source-sensor distance, pressure, leak path and leakage rate, background noise sources and the acoustic propagation and attenuation properties of the pipeline. With so many variables, the results of experimental measurements tend to take on an anecdotal flavor. It would be desirable to make the above more systematic and its capabilities more predictable. In the MISTRAS Group project discussed above, which was sponsored by the Electric Power Research Institute, an effort was made to develop a comprehensive model which would take all of these variables into account and allow predictions to be made regarding the detectability of leaks in the many situations that might arise. While this model was not completed, it was laid down in concept, and it is hoped to be able to complete it at some future time. It is already known that the greatest source of uncertainty will be the difficulty of dealing with the wide variety of source mechanisms that become apparent as soon as one starts working with this technology under actual field conditions.
The measurements discussed in this article were conducted at higher frequencies than those used in the correlation-based leak location instruments that have been used for decades in the water industry, for example to locate leaks in municipal water mains. The main effect of operating at a lower frequency is that the sensitivity may be higher, and it may be possible to work at a greater sensor spacing, but on the other hand, background noise will present greater challenges. The lower-frequency correlation technique has been established for much longer, but this may be because a long time ago, lower-frequencies were more easily handled by the electronic engineers who were designing the instruments. Now that AE equipment instruments operating at much higher frequencies are readily available, it is a good time to find out by practical experience the benefits that they may offer.
As documented above AE, leak detection technology can locate leaks and hence municipal water and wastewater utility managers will be able to benefit from having their crews use this technology in the field. A proactive leak detection program can make substantial savings by avoiding catastrophic situations and disruption of services. With wide use of this technology and scientific advancements, it is expected that AE leak detection technology will also become a quite inexpensive.
EDITOR?S NOTE: This paper was developed by the authors for publication in UIM and is based on a paper by Dr. Meegoda and Dr. Juliano on Acoustic Emission leak detection published by the American Society of Civil Engineers in 2013. ?
Dr. Jay Meegoda is a professor of civil and environmental engineering at the New Jersey Institute of Technology.
Dr. Thomas M. Juliano works for the New Jersey Institute of Technology and has worked in industrial and academic research and consulting.
Dr. Adrian Pollock is principal scientist at MISTRAS Group.
Dr. Obdulia Ley is manager of the Research Contracts and Application Group of MISTRAS Group.?