Leveraging Digital Twin for operational resilience in the oil and gas industry

Abstract

Motivation

The oil and gas industry is a highly complex and interconnected environments, where system failures or cyber threats can lead to severe operational and safety risks. Digital Twin enables real-time monitoring and predictive analysis to enhance resilience and decision-making, and existing studies often lack a comprehensive integration of system components and interdependencies, limiting the effectiveness of DT applications in critical scenarios.

Methodology

This paper presents the development of a DT for an experimental oil and gas transportation system at Università Politecnica delle Marche. The DT integrates the same digital PID controllers used in the real plant to replicate system dynamics and predict anomalous behaviours. Developed within the EU- and MUR-funded RESIST project, the DT is part of a broader digital environment that includes both the plant and operator models, aimed at enhancing system resilience.

A Gradient Boosted Tree algorithm, selected through comparative analysis, proved the most reliable technique for predictive modelling. The model was validated using real data, focusing on two key components: the ejector-pump system and the vertical tank.

Results

The findings demonstrate the potential of DTs as decision-support tools for plant operators, enabling proactive risk management and enhanced operational resilience despite, the presence of air phase variability introduces fluctuations in some critical conditions, requiring further data acquisition and refinement.

Future work

Future improvements will focus on refining the air flow modelling and expanding the dataset to improve predictive performance in extreme conditions providing an efficient anomaly detection system based on the DT prediction.