Leaks, whether internal to the process equipment or external to the environment, can be major threat to production and to reputation. Often they start small when they are difficult to detect. A method for detecting leaks while the process is online can prevent serious future issues.
When a large pipeline burst the client came to Sygnology requesting an investigation into methods to detect the presence of both large volume and minor leaks using statistical methods to compare expected pressures throughout the flow system. Due to slugging the client had trouble developing a system that did not generate a large amount of false alerts. Sygnology agreed to an analytical investigation and proposed employing its signal extraction technology as well.
Similarly to how water utility companies use microphones to locate leaks in buried pipelines, the Feature Extraction System developed by Sygnology provided the client with the ability to treat pressure transmitters as a type of pipeline listening device, effective for detecting the presence of moderate and high amplitude – or volume – repeating sound waves caused by leaks. Feature identification and tracking discovered signatures in acoustically connected piping systems with confirmed leaks, but not elsewhere. Therefore, the signal processing and statistical methods employed by acoustic leak detection allowed Sygnology to recognize telltale leak signatures and eliminates the false positives generally produced by traditional methods. Acoustic resonances created by leaking fluid can be traced downstream through the pressure transmitters of a piping system to their origin. The acoustic sensor with the highest amplitude acoustic feature is most likely the origin of the leak.
Because sound is a mechanical wave, in compressible fluids, it takes the form of a compression wave and the form of a bending wave in solids. In piping systems, aerodynamic sounds are transmitted downstream by the fluid, but also tend to couple with and excite the structural-borne sounds present in a system. Leaks, in particular, are high velocity jets of fluid across small orifices which produce a “static” type of noise – hissing with distinct tonal frequencies. The excitation of mechanical and acoustic resonances cause one frequency in particular, along with its harmonic equivalents, to occur more prominently and exact a tonal pitch to the leakage noise. The frequency and amplitude of this resonance are functions of leak geometry and differential pressure across the leak location. They nearly always couple with the structural borne sounds, traveling the lengths of the piping system at resonant frequencies.
The historical data analysis performed by Sygnology leveraged our Systems Analysis techniques. Coupled with associated verified failure reports, including ROV inspection results, coincided with acoustic signatures consistent with small leaks to create a repertoire of confirmed leaks detected by this method. These included multiple wellhead, well completion, piping system, and valve leaks. Expanded application of the method yields less than one false positive per year, a significant improvement over current, previous, and alternatively proposed systems based on time domain data.