Geometry-free prediction of inertial lift forces in microfluidic devices using deep learning
Authors: Jesse Ward-Bond, Ali Mashadian, Timothy C. Y. Chan et al.
Summary
The authors built a deep learning model that predicts how particles move through microfluidic channels (tiny fluid-handling devices) without needing to separately train a model for each channel shape. Previous machine learning approaches required training individual models for rectangular channels, triangular channels, etc., which was tedious. This new approach uses a geometry-free parameter set that generalizes across unseen channel shapes, making it much faster to simulate particle behavior in new device designs.
Main takeaways:
- Predicts particle lift forces (the push/pull particles experience in flowing fluid) without explicit geometric parameters like channel width or angle
- A single trained model works across multiple channel cross-section types (rectangular, triangular, etc.) instead of requiring separate training per geometry
- Generalizes well to channel shapes it has never seen during training
- Produces particle migration patterns consistent with published experimental results when plugged into simulation software
- Shifts the computational burden away from both simulation and per-geometry training
Relevance
Not directly related to my persona/midtraining work — this is a computational fluid dynamics paper about physical particle manipulation in lab-on-a-chip devices, not language models or behavioral installation.
Abstract
arXiv:2605.08109v1 Announce Type: new Abstract: Inertial microfluidic devices (IMDs) offer low-cost, high-throughput alternative techniques for many traditional particle- (or cell-) manipulation tasks, but simulating them requires being able to predict particle migration, and thus particle lift forces, under a variety of possible channel geometries. Recent work has demonstrated that machine learning models can be used to drastically speed up these numerical simulations, but doing so required training individual models for every unique channel cross-section type (e.g., rectangular, triangular) -- shifting the burden from the simulation step to the training step. In this paper, we develop a novel approach for predicting particle lift forces that contains no explicit geometric parameters. We train a neural network model using a new parameter set and show that while it performs comparably to existing models on channel geometries in the training set, it is able to generalize to unseen channel geometries far more effectively. We show that the lift force model developed herein can be easily transferred to particle tracing simulation software, where it is capable of predicting particle migration patterns consistent with the literature across a variety of channel designs.