Original title: Differentiable Hybrid Modelling for Learning and Optimising Chemical Transport Processes from Experimental Data
Arthur Jessop, Mohammed Alsubeihi, Ben Moseley, Ashwin Kumar Rajagopalan · September 2026
Chemical engineers rely on transport models to track how matter and energy move. Standard models use hand-picked rules and guess starting conditions, which causes systematic errors. Black-box neural networks can fit data better, but they cannot be checked for physical consistency or used beyond their training task. To fix this, researchers built a hybrid framework. It combines a physics-based equation solver with neural networks. The system works by letting the neural network discover the missing physical rules and learn the correct initial conditions directly from real experimental data. Because the entire framework is differentiable, meaning math can trace how every part affects the outcome, it can directly calculate the best experimental settings for chosen goals. This lets engineers accurately model and optimize real transport systems.
Imagine cooking a new dish without knowing the exact recipe or how much of each ingredient you started with. If you guess the ingredients and amounts, the dish might turn out poorly. Here, researchers use data from past cooking attempts to figure out the actual ingredients and starting amounts, so they can improve the recipe and even adjust cooking methods for better results.
Chemical engineers often use models with guessed rules, which can lead to wrong predictions and control issues. This approach lets models learn from actual data, making processes more reliable and efficient.
The abstract specifies this framework is for population balance equations, so it may not directly apply to all chemical transport processes. It also does not detail the extent of experimental validation or generalizability beyond this scope.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
Differentiable Hybrid Modelling for Learning and Optimising Chemical Transport Processes from Experimental Data — Arthur Jessop, Mohammed Alsubeihi, Ben Moseley et al.
Abstract licence: CC0
Explanation generated by Synthesis from the abstract — it is not the original paper. This is a preprint, so it hasn't necessarily been peer reviewed.