Cognitive Digital Process Engineering of Electrified Heat Exchange Networks with Physics Guided AI for Fouling Corrosion Prediction and Carbon Neutral Refinery Energy Systems

The petroleum refining industry confronts an inflection point: the simultaneous imperatives of decarbonization, operational efficiency, and asset integrity management demand approaches that transcend conventional steady-state thermodynamic models. This paper presents a unified framework Cognitive Digital Process Engineering (CDPE) that integrates electrified heat exchange network (e-HEN) optimization with physics-guided artificial intelligence (PGAI) for real-time fouling–corrosion prediction and carbon-neutral energy target setting. A multi-fidelity digital twin architecture couples first-principles conservation equations with graph neural network (GNN) surrogate models, enabling simultaneous optimization of heat recovery efficiency, electric heater deployment, and corrosion inhibitor injection schedules. Fouling resistance trajectories are predicted using a hybrid Kern–Seaton mechanistic core augmented by long short-term memory (LSTM) networks trained on plant historian data from two Middle Eastern crude distillation units (CDU). Physics-informed loss functions enforce thermodynamic consistency and electrochemical Tafel kinetics across all learned representations. Validated against ۱۸ months of operational data, the CDPE framework achieved a ۳۴.۷% reduction in fired heater fuel gas consumption, a ۲۱.۳% decrease in unplanned heat exchanger cleaning interventions, and a projected CO۲-equivalent abatement of ۴۷,۲۰۰ tonnes per annum for a representative ۱۰۰,۰۰۰ BPSD refinery. The results establish CDPE as a credible pathway toward Scope ۱ and Scope ۲ net-zero targets without compromising process throughput or product quality.

کلیدواژه ها:

heat exchanger networks ، electrification ، physics-informed machine learning ، fouling prediction ، corrosion modeling ، digital twin ، carbon neutrality ، Refinery energy systems ، graph neural Networks ، Process Optimization