Predictive Analytic of Corrosion Build Up Rate by Utilizing Artificial Neural Network

• Main Author: NORSALLEHIN, NOOR AISHAH
• Format: Final Year Project
• Language: English
• Published: IRC 2020
• Subjects: Q Science (General)
• Online Access: http://utpedia.utp.edu.my/21810/1/23314_Noor%20Aishah%20Norsallehin.pdf; http://utpedia.utp.edu.my/21810/

Oil and gas industry for a few years has spent a lot of money in handling in technical operation problems such as inefficiencies and failures of pipeline and instrument that occur when the it is not operated optimally due to external and internal corrosion issue. Corrosion is a metal's deterioration due to chemical reactions between it and the environment around it. Corrosion in pipeline and instrument can result in material loss, thickness reduction, and ultimate failure at times. The part will be completely broken down at one point and will have to be replaced during production is stopped. Corrosion must be detected at an early stage in order to prevent a high cost-effective catastrophe for the business and even the lives of humans. According to Prabha et al. (2014), “corrosion affects all aspect of exploration and production”. In placing more emphasis, Asmara and Kurniawan (2018) stated that corrosion predictions are important for corrosion and materials engineering to pick suitable materials, plan testing, management work and estimate future cost analysis repairs. The contribution of certain operating parameters (temperature, salt, chloride and pH) to the oil and gas condenser corrosion rate was focused in this paper. Condenser is used particularly in refinery operations in the oil and gas industries. In refineries, corrosion occurances is usually caused by inorganic factors such as water, H2S, CO2, sulphuric acid and sodium chloride, and not by the organics (Prabha et al., 2014). Many efforts have been made with advanced technology today to tackle this problem. This creates a demand for the use of machine learning to predict corrosion build-up. Historical field data from RAPID Pengerang Petronas are analyzed using the Artificial Neural Network method to determine the dependence of the corrosion rate on the operating parameters. This paper main purpose is to build a machine learning model on corrosion build-up at the condenser and at the same time developing a predictive corrosion analysis dashboard for monitoring purposes.