Proceedings of the Eighth International Conference on Probabilistic Safety Assessment & Management (PSAM)
285 Condition Assessment of Production Separators Using Neuro-Fuzzy Tools for Signal Validation and Diagnosis (PSAM-0185)
Download citation file:
- Ris (Zotero)
- Reference Manager
Among the most crucial equipment to be found on offshore oil installations are the production separators, the purpose of which is to separate the mixture of crude oil, water, gas, foam, emulsion, sand and grit from the oil wells through gravity-induced coalescence. The very hostile operating environment, with high pressure and high temperatures, places heavy strain on internal equipment installed to facilitate the coalescence process, and makes the condition monitoring of said equipment extremely difficult.
Currently no standard condition monitoring method of separator internals exists, and only corrective maintenance is carried out during planned routine inspections, typically every 3 years (time based maintenance). As the maintenance shutdowns are associated with very high expenses (which can run at more than $10,000,000 per day), it is considered a high priority to both shorten the downtime and to lengthen the interval between shutdowns.
Condition based maintenance facilitates achieving both these goals by allowing maintenance planners to have the necessary spare part and repair equipment at hand, and by avoiding unnecessary maintenance shutdowns where the internals turn out to be in acceptable working condition.
This paper describes a sub-activity within the CORD Production Separator project focusing on applying neural network-based signal validation and diagnosis tools for abnormality detection and condition assessment of the internal equipment of production separators in the petroleum industry. The tools have been successfully applied for the monitoring of various processes within the nuclear industry, and perform their functions by learning how various process signals are correlated. PEANO will validate and calibrate a sensor by determining any deviations from the expected correlations to other sensors, while ALADDIN recognizes a fault by its associated time-based “signature” in the combined data matrix.
The focus of the CORD Production Separators project is to investigate several technologies, including acoustics, microwave and gamma measurements, to perform condition monitoring of internal equipment. In the activity described in this paper it is proposed to combine the data from these technologies with standard process measurements, such as flow, pressure, temperature and composition at separator inlet and outlets, and apply them to the neural network-based tools to achieve reliable information regarding the condition of the internal equipment. This information can reveal possible sources of deficiencies, and thus help to assess the risks associated with different levels of the subsequent maintenance activities.