DFG SPP 2331

Machine Learning in Chemical Engineering

Project Title: Hybrid Physics-Neural Network Soft Sensors for Dynamic Operation of Liquid-Liquid Separation Processes

Joint project with: Institute for Fluid Process Engineering, RWTH Aachen (AVT.FVT)

Goals:

Development of a physics-informed neural network (PINN) based soft sensor to estimate the dispersion layer height (key performance indicator) for the dynamic operation of a horizontal gravity separator with varying operating conditions.

Contribution from ICE-1:

• Evaluation and development of a physics-informed neural network (PINN) based separator model by infusing partially available physicochemical knowledge with scarce plant data
• Estimation of dispersion layer height for varying operating conditions with the PINN-based model
• Validity range analysis of the developed PINN-based models for effective and safe operation

Website: https://gepris.dfg.de/gepris/projekt/466656378?language=en