Physics-informed neural networks estimate liquid-liquid separation phase heights

Researchers have developed a novel framework utilizing Physics-Informed Neural Networks (PINNs) to estimate the dense-packed zone height in liquid-liquid separation processes. This approach combines a PINN, pre-trained on synthetic data and simplified mechanistic models, with readily available volume flow measurements. The system is then fine-tuned with limited experimental data and integrated into an Extended Kalman Filter for accurate, real-time phase height tracking without direct measurement. AI

Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →

IMPACT This PINN-based approach offers a more cost-effective and accurate method for monitoring critical separation processes in chemical and pharmaceutical industries.

RANK_REASON This is a research paper detailing a novel application of PINNs to a specific engineering problem.

Read on arXiv cs.LG →

• arXiv
• Manuel Dahmen
• Physics-Informed Neural Network
• Extended Kalman Filter

• paper
• other