Assessment of the distributed and concentrated pipeline corrosion by means of ultrasonic testing and machine learning methods

 

Project no.: P-MIP-21-180

Project description:

Corrosion is one of the most common degradation mechanisms of metallic engineering infrastructure. The pipelines of petroleum, chemical and gas industries carry products that contain corrosive substances leading to progressive degrada????on of the equipment. EU reports at least 137 major accidents in petroleum
refineries since 1984, while three of them happened in Lithuania. Around 20% of the incidents worldwide are corrosion related, leading to annual global losses up to US$2.5 trillion excluding the costs of forced shutdowns and environmental hazards. To avoid severe environmental and economic consequences, the periodic inspecton of corrosion rates is a must to ensure long-term integrity of the assets. Ultrasonic tes????ng stands out among the classic corrosion detection techniques as the direct and non-intrusive method that allow to assess the corrosion. However, currently existng ultrasonic methods suffer from inspection time, accuracy and sensitivity trade-of meaning that either the accurate localized inspection or rough long range screening is available for corrosion assessment. As a result, localized inspec????ons can only detect widely distributed corrosion with a-priori known locations, while the long range systems are dedicated for detection of randomly distributed large flaws, which itself cannot be characterized properly due to poor resolution. In this project, the hybrid corrosion
assessment technique is proposed that combines ultrasonic tomographic reconstruction and machine learning methods to reconstruct and characterize both distributed and localized corrosion at large sections of pipes with high accuracy. This will be achieved through development of sophisticated tomographic reconstruc????on techniques that will be able to identify the corrosive regions of the pipe, while the distribution, type of corrosion and wall thickness loss will be defined with machine learning models and cumulative distribution functions. Throughout the project the proposed technology will reach TRL level 6 and will be patented at Lithuanian patent office.

Project funding:

Projects funded by the Research Council of Lithuania (RCL), Projects carried out by researchers’ teams


Project results:

During implementation of the project, a method for measuring the phase velocity of guided waves was developed and investigated, which allows the identification of modes propagating in pipes and the detection of corrosion defects by identifying modes resulting from defect-induced mode conversion. Different methods of signal phasing and defect position reconstruction have been investigated, including tomography and total focusing methods.

Period of project implementation: 2021-04-01 - 2024-03-31

Project coordinator: Kaunas University of Technology

Head:
Renaldas Raišutis

Duration:
2021 - 2024

Department:
Laboratory of Numerical Simulations, Prof. K. Baršauskas Ultrasound Research Institute