SAFETY INTEGRITY LEVEL EVALUATION BASED ON UPDATING FAILURE RATE USING BAYESIAN INFERENCE AND MAHALANOBIS-MINIMUM COVARIANCE DETERMINANT ALGORITHM ON METHANATOR IN AMMONIA PLANT

• Main Author: Bilhaq Mas Avicena, Ramiza
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/85740
• Institution: Institut Teknologi Bandung

Safety and reliability are crucial parameters in the operation and maintenance of process industries. One form of safety in these industries is the Safety Instrumented System (SIS). However, failures can still occur. SIL Verification evaluations, conducted according to IEC 61508/61511 standards, ensure that the Safety Instrumented Function (SIF) meets the initial requirements. In industrial practice, these evaluations are often static because failure rate data used come from generic and manufacturer data, making it difficult to determine the actual SIL value based on plant operating conditions. This study employs Bayesian Inference to update failure rate data, allowing for dynamic SIL evaluations that account for operational failures. This approach uses operational failure data to influence the failure rate model while maintaining static data from manufacturer and generic sources as the initial knowledge. The study updates failure rates for SIF subsystems, including sensors and final elements. Anomaly detection analysis using the Mahalanobis Minimum Covariance Determinant algorithm identifies failures in SIF sensor subsystems. The study demonstrates the application and impact of a dynamic approach to SIL evaluation through a case study on a methanator in an ammonia plant. With the ???????????????????????? value from the dynamic SIL verification results being 0.1337, which has not yet met the SIL target requiring a ???????????????????????? of at least 0.05, the primary recommendation is to achieve the SIL 1 target with a ????????????????????????value of 0.03088 by changing the configuration of the final element to a 1oo2 configuration. Additionally, an extra recommendation is provided, which involves installing a relief valve to a safe location at the inlet of Methanator 106-D. Keywords: safety integrity level (SIL), safety instrumented system (SIS), Mahalanobis Minimum Covariance Determinant, updating failure rate, Bayesian Inference, methanator, anomaly detection