Corroboration of sparse failure data through Bayesian estimation for natural gas cross country pipelines: A case study with reference to the Indian scenario. (September 2018)

• Record Type: Journal Article
• Authors: Raveendran, Anagha; Renjith, V.R.
• Is Part Of: Journal of loss prevention in the process industries. Volume 55(2018)
• Page Start: 347
• Page End: 358
• Publication Date: 2018-09
• Subjects: Bayesian -- Natural gas transmission -- City gas distribution network -- Pipeline failures -- PNGRB; Chemical industries -- Safety measures -- Periodicals; 660.2804
• DOI: 10.1016/j.jlp.2018.07.011
• Languages: English
• ISSNs: 0950-4230

Abstract

Abstract: The rising demand for Natural Gas (NG) in India, a developing country, necessitates adequate, uninterrupted and safe transmission and distribution of the gas to consumers at various points within the country. Though pipelines are considered as the safest mode of transportation, the failure of a flammable and highly pressurized natural gas pipeline can cause fatalities and extensive damage to properties. A reliable estimation of occurrence probabilities of 'actual' natural gas pipeline incident scenarios helps to evaluate the safety management and control system existing in the country. Unfortunately, India does not maintain any quantitative failure data base of the natural gas pipeline incidents due to lack of operational history. This paper outlines the use of Bayesian logic for corroborating the existing sparse and imprecise occurrence proportion of natural gas pipeline incident scenarios obtained from the incident analysis reports. This is done by incorporating the specific unreported Piped Natural Gas (PNG) incident scenario frequency per year. Posterior occurrence probability estimates of the incident scenarios lie between the sparse prior proportions obtained from the published incident analysis reports and the likelihood proportion of specific PNG incidents happening in India per year. The newly generated data through the Bayesian estimation therefore can be considered more reliable and sensible and are also validated like such data available in the developed countries. The posterior estimates also necessitate more improved safety management and control system in each City Gas Distribution (CGD) entity within the country to reduce specific failure cause-consequences. Highlights: Identification of most prevalent 'actual' incident scenarios obtained in the Indian natural gas pipelines. Reliable estimation of the incident scenario occurrence probabilities to improve the safety concern in future. Bayesian approach to corroborate sparse prior proportion of each incident scenario obtained from the incident reports. Incorporation of likelihood frequency of specific 'unreported' piped natural gas incident scenarios in a year. Distinctive analyses of each incident scenario to prioritise them based on posterior estimates and provide safety measures.