Reliability Assessment Based on Multisource Information Fusion Method for High Pressure Natural Gas Compressors

Abstract

Abstract Compressor units used for long-distance transportation natural gas pipeline pressurization have high-pressure and high-risk characteristics. Hence, the scientific reliability assessment is important for high-pressure compressor units, to evaluate the reliability standard, find the performance deficiency, and provide references for operation and maintenance. The classical reliability assessment method is not suitable for the complex and high-reliability equipment, like high-pressure compressor units and pipelines. The reliability assessment of the high-reliability equipment is faced with the challenge of the multisource information. A reliability assessment method based on the multisource information fusion is proposed in this work. The fusion resources consist of design information, component test information, and trial operation information. The reliability of high-pressure compressor units can be assessed by fusing the characteristic parameters, from component-based assessment, function-based assessment, quality evaluation, and life model, by D–S evidence theory. A case study is conducted to verify the proposed reliability assessment method in a 20 MW-class high-pressure compressor. There are four information resources in the case, i.e., component test data, design information, operation data, and simulation data. The compressor reliability is assessed as 99.32%, validated by the statistical assessment result based on long-term shutdown reports. This application points out the existing weakness in the high-pressure compressor units and indicates the directions for improving the design, analysis, operation, and failure prevention technologies. It reveals that the reliability assessment based on multisource information can provide a guarantee for the operation and maintenance of high-pressure compressor units. Meanwhile, the proposed method has good expansibility, which may be used in more fields.