Physics-Informed Neural Network Digital Twin for Dynamic Tray-Wise Modeling of Distillation Columns under Transient Operating Conditions

arXiv cs.LG / 3/27/2026

💬 OpinionSignals & Early TrendsIdeas & Deep AnalysisModels & Research

共有:

Key Points

The paper proposes a Physics-Informed Neural Network (PINN) digital twin framework to model binary distillation columns in a tray-wise, dynamic way under transient operating conditions.
It embeds key thermodynamic and process laws—modified Raoult’s law for vapor-liquid equilibrium, tray-level mass/energy balances, and McCabe-Thiele methodology—directly into the neural network training loss via physics residual terms.
The model is trained and tested on a synthetic high-fidelity dataset (961 time-stamped measurements over 8 hours) generated in Aspen HYSYS for a 16-sensor-stream binary HX/TX distillation system.
Using adaptive loss weighting, the PINN outperforms several data-driven baselines (LSTM, MLP, GRU, Transformer, DeepONet), achieving an HX mole fraction RMSE of 0.00143 (R^2=0.9887), a 44.6% improvement over the best data-only baseline.
The predicted tray temperature and composition trajectories under transient perturbations closely match expected column dynamics, supporting downstream uses like real-time soft sensing, model-predictive control, and anomaly detection.

Abstract

Digital twin technology, when combined with physics-informed machine learning with simulation results of Aspen, offers transformative capabilities for industrial process monitoring, control, and optimization. In this work, the proposed model presents a Physics-Informed Neural Network (PINN) digital twin framework for the dynamic, tray-wise modeling of binary distillation columns operating under transient conditions. The architecture of the proposed model embeds fundamental thermodynamic constraints, including vapor-liquid equilibrium (VLE) described by modified Raoult's law, tray-level mass and energy balances, and the McCabe-Thiele graphical methodology directly into the neural network loss function via physics residual terms. The model is trained and evaluated on a high-fidelity synthetic dataset of 961 timestamped measurements spanning 8 hours of transient operation, generated in Aspen HYSYS for a binary HX/TX distillation system comprising 16 sensor streams. An adaptive loss-weighting scheme balances the data fidelity and physics consistency objectives during training. Compared to five data-driven baselines (LSTM, vanilla MLP, GRU, Transformer, DeepONet), the proposed PINN achieves an RMSE of 0.00143 for HX mole fraction prediction (R^2 = 0.9887), representing a 44.6% reduction over the best data-only baseline, while strictly satisfying thermodynamic constraints. Tray-wise temperature and composition profiles predicted under transient perturbations demonstrate that the digital twin accurately captures column dynamics including feed tray responses, reflux ratio variations, and pressure transients. These results establish the proposed PINN digital twin as a robust foundation for real-time soft sensing, model-predictive control, and anomaly detection in industrial distillation processes.