Cleanliness Correlation of Mono Ethylene Glycol (MEG) as Thermodynamic Hydrate Inhibitor to forecast Fresh Injection Period using Supervised Machine Learning

Abstract

Production in the context of oil and gas can be defined as the total flow rate of hydrocarbon components with several other components such as water, CO2, H2S, and soil from production wells, then flows through a pipeline to the processing facilities. Oil and gas production can be limited by several factors, one of which is the issue of flow assurance such as gas hydrate, which can result in a loss of production opportunity (LPO) due to gas hydrate blockage. The most common hydrate formation prevention method is injecting hydrocarbon fluid with antifreeze chemicals called thermodynamic inhibitors such as mono ethylene glycol (MEG). However, dissolved solids contained in produced water may precipitate and tend to deposit in surface facilities including the Mono Ethylene Glycol Regeneration Unit (MRU). These can plug the MEG injection system and result in potential hydrate formation. This paper deals with how actual problems of plugging due to scaling or fouling on the MEG injection system can be minimized by analyzing process parameters and laboratory analysis results using supervised machine learning. The study suggests that machine learning can be used to predict the problem occurrence by observing the cleanliness level of lean MEG that correlates with some process parameters such as hydrocarbon flow rate, CO2 content, and wellhead flowing pressure. If the cleanliness level is above specification, the MEG injection system is assumed to be possible plugging, otherwise not plugging. Some supervised learning algorithms are compared to evaluate the performance of plugging possibility prediction. This result can be used to determine and optimize MRU operation, monitoring, and maintenance strategy.