Development of mathematical models for quantitative risk and safety assessment in oil and gas refineries

• Main Author: Rachid, Ouache
• Format: Thesis
• Language: English
• Published: 2016
• Subjects: T Technology (General); TA Engineering (General). Civil engineering (General)
• Online Access: http://umpir.ump.edu.my/id/eprint/15843/13/Development%20of%20mathematical%20models%20for%20quantitative%20risk%20and%20safety%20assessment%20in%20oil%20and%20gas%20refineries.pdf; http://umpir.ump.edu.my/id/eprint/15843/

Risk assessment is a systematic process to identify hazards, analyze and evaluate the risks associated with hazards that can harm the workers, people, environment or properties using qualitative, semi-quantitative or quantitative approach to determine the appropriate ways to eliminate or control the hazards. Quantitative risk assessment (QRA) is an effective approach used in petrochemical industries to estimate the likelihood of an accident and the severity of its consequence. However, uncertainty is still the main problem faces quantitative risk assessment in spite of its significant progress. Therefore, this thesis proposes mathematical models to address the uncertainties of quantitative risk assessment as follows: i) reliability guide analysis (REGUIA) is developed to identify the main components of accident scenarios and to determine the factors which can affect the failure probabilities, ii) human reliability model based on five matrices with mathematical equations is developed to determine the level of human reliability and to precise probability of human error, iii) characteristics of hazard analysis based on logic factors intermediates (KHALFI), linear and nonlinear models are three models developed to determine the failure probability of the events at any geographical location, considering the factors: temperature, humidity and wind speed where root mean square error (RMSE) of the three developed models are 2.38E-5, 2.10 and 1.94, respectively, iv) risk and safety models to analyze the accident scenarios based on Bowtie method and Bayesian network with new classification of safety integrity level with mathematical equation are developed, v) probability binary state is employed to define the range of failure probability, vi) probability determination (PRODET) is a mathematical model developed in this study to determine the exact probability of the equipment at the specific operation time, vii) occurrence time (OT) is also developed to find the required time for the event to occur, viii) risk matrix model with mathematical equations are developed to compute the level of risk. Finally, Simulink model is developed to implement the developed models to automate the calculation and to facilitate the analysis of the results with graphical representation of the inputs and the outputs. The results show plausible and reliability of the models and demonstrate that the developed models are more reliable and precise than the classical models. The results of risk and safety analyses revealed that 86% of the basic events on average gained 180% increased reliability.