Digital Twin-Driven Optimization in Natural Gas Processing: Methodology and Applications
Abstract
This paper aims to develop a practical methodology for digital twin-driven optimization in natural gas processing plants, addressing key practical challenges of low process efficiency, high energy consumption, and heavy reliance on operator experience. Focusing on two core processes—LPG recovery and gas decarbonization-dehydration—the study demonstrates the framework's adaptability across diverse unit operations. It seeks to validate the methodology's capability to boost product yield and reduce energy use, thereby improving operational stability and sustainability in the natural gas industry. SLB digital twin technology implements an integrated Real-Time Optimization (RTO) and Advanced Process Control (APC) system on an industrial Edge IoT platform at the terminal. A high-fidelity, first-principles model built with Symmetry™ was executed locally on the Edge device, leveraging real-time data. It carried out continuous optimization to minimize total energy consumption while meeting all process and product constraints. The resulting optimized operating conditions were generated and directly applied at the Edge through the APC system. The comprehensive implementation, involving data analysis, rigorous modeling, and validation, ensured the solution's accuracy, robustness, and reliable performance under variable conditions. The methodology delivered significant, validated results in two distinct application cases. In Gaolan natural gas terminal, the digital twin system increased the LPG recovery rate from 93.5% to over 95%, generating estimated annual economic benefits on the order of tens of millions of Chinese Yuan (RMB) while concurrently reducing specific energy consumption. In the decarbonization and dehydration unit of a Yanchang Petroleum CPF facility, the system effectively lowered the unit's energy consumption per volume of gas processed by optimizing amine circulation rates and regenerator operation. Key observations from both deployments include the system's consistent high model convergence rate (exceeding 95%), good robustness, and its notable contribution to enhanced process stability and reduced reliance on operator practical judgement. In conclusion, this work successfully establishes a verifiable, replicable, and economically impactful universal Edge-enabled digital twin methodology for the continuous optimization of core natural gas processing operations. This study presents an industrial application of an Edge-deployed digital twin for real-time optimization in China's natural gas processing. Combining first-principles simulation with real-time data at the Edge enables immediate decision-making and reliable, cloud-independent optimization. The achieved gains in efficiency and stability demonstrate the solution's scalability, establishing a practical reference for improving economic and operational performance across the industry.