Abstract:The digitalization of chemical research and industry generates large amounts of data, offering new opportunities for developing and parameterizing kinetic models. Leveraging this data requires machine learning techniques capable of autonomously extracting kinetics from reactor datasets. Recently, neural ordinary differential equations (neural ODEs) were coupled with reactor models to learn kinetics from ideal reaction systems, such as plug-flow reactors. However, real reactor set-ups commonly feature non-idealities such as heat- and mass transfer limitations, described by partial differential equations (PDEs) or differential-algebraic equations (DAEs). Discretizing such non-ideal PDE or DAE reactor models by finite volumes yields algebraic balance equations, which are solved by numerical schemes. In this work, we propose to learn kinetics from non-ideal reactor data using these balance equations and implicit neural networks, avoiding expensive backpropagation through the numerical solution by utilizing the implicit function theorem. The approach is demonstrated for the example of a mass transfer limited flat plate reactor for the preferential catalytic oxidation of CO. We show that Global Reaction Neural Networks, embedding thermodynamic and stoichiometric prior knowledge, coupled with a discretized two-phase CSTR cascade reactor model extract intrinsic kinetics from 50 integral reactor experiments. These kinetics generalize to new reactor geometries, where kinetics obtained by neural ODEs fail. Further, the approach is robust, recovering accurate kinetics even when training data is perturbed by 10% Gaussian noise. We expect that combining neural network-based kinetics with non-ideal reactor models will broaden the scope of kinetic model discovery and improve access to accurate kinetic models.

Nonlocal modeling of continuously stirred tank reactors with residence time distribution

Study on Residence Time Distribution of Fluid in the Taylor Reactor

A New Nonlinear Optimal Control Design Approach for a Chemical Reactor

A CSTR State Observer Based on Residual Neural Network

CFD Simulation of Shear Flow and Mixing in a Taylor–Couette Reactor with Variable Cross-Section Inner Cylinders

Modeling and Control of a Continuous Stirred Tank Reactor Based on a Mixed Logical Dynamical Model

A method-of-lines formulation for a model of reactive settling in tanks with varying cross-sectional area

Using CFD and Machine Learning to Explore Chaotic Mixing in Laminar Diameter-Transformed Stirred Tanks

Residual neural network-based observer design for continuous stirred tank reactor systems

Continuous Time Random Walks for Non-Local Radial Solute Transport

Reaction kinetics in open reactors and serial transfers between closed reactors

Measurement, Modelling and Analysis of Residence Time Distribution Characteristics in a Continuous Hydrothermal Reactor

Mathematical modeling of the residence time distribution in loop reactors

Learning kinetics from non-ideal reactors by implicitly solved finite volumes and global reaction neural networks

Modeling Chemical Reactors I: Quiescent Reactors

Numerical and experimental studies of mixing characteristics in a T-junction microchannel using residence-time distribution

Nonequilibrium dynamic modeling of a simple batch reactive distillation

Residence time distributions in microchannels with assistant flow inlets and assistant flow outlets

Trans-level multi-scale simulation of porous catalytic systems: Bridging reaction kinetics and reactor performance

Linear Model Predictive Control for a Coupled CSTR and Axial Dispersion Tubular Reactor with Recycle

A new flow-kinematics-based model for time-dependent effective dispersion in mixing-limited reactions