Enhancing Liquefied Natural Gas supply chain robustness through digital twin-driven machine learning models: A special case of cryogenic heat exchanger

The Liquefied Natural Gas (LNG) supply chain plays a critical role in the global energy sector but remains vulnerable to technical disruptions that compromise operational stability. Equipment failures pose significant risks, leading to production halts and quality degradation. This paper proposes a proof-of-concept resilience-enhancing framework to mitigate minor disruptions that, if left unaddressed, could escalate and impact LNG production continuity. Focusing on the Cryogenic Heat Exchanger (CHE) as a case study, an essential component of liquefaction, the framework integrates digital twins (DT), machine learning (ML), and predictive model to enable real-time monitoring, early failure detection, and proactive mitigation. First, randomly online simulated data on critical parameters (temperature, pressure, and flow rate) is collected using IoT sensors. Next, this data is processed through Aspen HYSYS-based and ML-driven DT to assess the system performance and predict potential failures, respectively. Finally, a Vector Autoregressive model is employed to forecast future operating conditions based on recent observations, capturing system dynamics and correlations. The resulting forecasts will feed the ML model to predict the next operational state. The framework is validated through an extensive computational study on randomly generated scenarios. The results demonstrate that the proposed system monitoring enhances LNG supply chain robustness, achieving early failure detection averaging 57.21% and significant downtime reduction reaching 31.57% on average compared to corrective maintenance strategies. These findings underscore the framework's potential to improve operational efficiency and stability in LNG production, offering a scalable solution for supply chain robustness.