Listening for leaks

In many industrial applications, signal-processing techniques offer the ability to locate noise sources and help determine if the system under test is functioning properly. This diagnostic model helps prevent the degradation of a system via preventive maintenance, and can be used to discover faulty parts in a system. Compressors, engines, roller bearings, and electrical fans are good candidates for signal processing diagnostics.

Noise-source detection is also useful for monitoring and testing pipe systems used for liquid transportation. It might, for instance, prove valuable in Nigeria where Shell pipelines are being sabotaged by crude-oil thieves.

The extent of the losses the company is suffering can be gauged by its declaration of force majeure and admission that it may not be able to meet contractual obligations. Last year, said Shell, 98% of oil spillage from its pipelines was caused by sabotage.

The noise in pipelines commonly results from cracks and leakage points where liquid escapes into the environment. The escaping liquid causes specific noise, which travels in the pipe material, in the surrounding environment and in the liquid that remains in the pipe. These leak signals can be recorded and analysed to locate the position of the noise source.

While many tools already exist for leak detection, such as correlators and loggers, there is a need for further research and development in this field, according to Raul Ionel, University of Timisoara, Faculty of Electronics and Telecommunications Timisoara, Romania

Ionel has led a recent study of working conditions such as pipe material, leak debit, environment temperature, and velocity of propagation.

hese parameters, he said, require more complex methods for acquiring and processing signals emanating from noise sources.
The objects of the research include:

• locating the position of a noise source in a pipe system with minimal errors and without unnecessary digging;
• employing traditional and modern signal-processing algorithms for superior results;
• building a system with little signal-processing knowledge;
• automating system operation for minimal user involvement;
• creating a simple and easy-to-read data display for an effortless understanding of results; and
• developing a compact and portable system with reduced set-up and execution time.

The university team selected LabVIEW graphical system design software as the basis for the application. The system can provide signal-processing algorithms that can be used almost immediately and require no prior knowledge of the logic behind them, which has drastically reduced the university’s programming time.

The application hardware includes two piezoelectric sensors that can be attached to the pipes, two amplifiers, and a filter that limits the frequency band to 6kHz. The system employs the principle of cross-correlation between the noise signals and signals must be simultaneously sampled.

Simultaneous sampling

“We chose the NI USB-9215A data-acquisition module for its ability to simultaneous sample on all four channels,” said Ionel. “In addition, the USB-9215A offers high accuracy and resolution and requires no extra power supply, all at a lower cost than alternative products. The USB communication between the USB-9215A and the laptop helped simplify set-up and required no prior configuration.”

The USB-9215A device samples the signals from the sensors and sends the data to a portable PC. The LabVIEW program analyses the data using several signal-processing methods.

“We implemented various graphical user interface elements and functions available in LabVIEW, such as graphs, spectral measurements, VIs, and MathScript”, noted Ionel. “We also created new subVIs to meet our application needs.”

Using LabVIEW and the USB-9215A, the team developed a signal acquisition and analysis application that delivers clear, accurate results comparable with a commercial leak-detection system.