A reliability prediction methodology based on improved radial basis function neural network (RBFNN): a condensate light crude oil stabilization facility as case study. Issue 3 (3rd July 2021)

• Record Type: Journal Article
• Authors: Riverol, Carmen; Pilipovik, V.
• Is Part Of: Safety & reliability. Volume 40:Issue 3(2021)
• Page Start: 157
• Page End: 164
• Publication Date: 2021-07-03
• Subjects: Gas stabiliser -- RBFNN -- risk analysis -- neural networks; Reliability (Engineering) -- Periodicals; System safety -- Periodicals; Risk assessment -- Periodicals; Industrial safety -- Periodicals; Industrial safety; Reliability (Engineering); Risk assessment; System safety; Electronic journals; Periodicals; 620.00452
• DOI: 10.1080/09617353.2021.1989227
• Languages: English
• ISSNs: 0961-7353

Abstract

Abstract: In this article, a risk analysis was carried out on condensate stabilisation facility. The condensate or light crude oil produced from oil and gas facilities can be highly volatile i.e. the liquid can vaporise very easily, forming a flammable atmosphere, thereby making it difficult or unsafe for storage in atmospheric tanks. Due to its volatility, a high vapour pressure can result in seal damage to floating roof tanks with the potential to result in roof collapse and/or loss of containment of the stored liquid to atmosphere (on top of the roof). The Radial Basis Function Neural Network (RBFNN) proposed method herein is based on clustering of input space vectors and computing weights of Euclidian distances; histogram equalisation within each cluster will determine the centre and width of each receptive field for the estimation of the risk parameters. The RBFNN is designed to estimate the quality of performing consistently well (Reliability) to minimise any risk in the tank. The obtained results demonstrate the high performances of the proposed method where the deviation between Weibull function and the predicted value using RBFNN is less than 0.5%.