This multi-year project explores how CFD and Machine Learning can be combined to design capillary microfluidic circuits for lab-on-a-chip devices with far less trial-and-error.
Microfluidics enables precise handling of tiny fluid volumes in diagnostic and biomedical devices, but designing microfluidic circuits is often complex and slow:
- A gap exists between microfluidics engineers and end users such as physicians.
- Design cycles rely on repeated physical prototyping and trial-and-error.
- There is a lack of user-friendly design tools tailored to non-experts.
The project uses Computational Fluid Dynamics (CFD) to simulate fluid flow in microchannels under various conditions, generating a large dataset describing how geometry and operating parameters influence performance.
Machine Learning models are then trained on this dataset to:
- Predict flow behavior and key performance metrics for new geometries.
- Propose design candidates that satisfy functional constraints.
- Reduce the need for repeated experimental iterations.
Build AI-assisted tools that let clinicians and researchers design microfluidic circuits without needing deep CFD expertise.
The project is conducted in collaboration with Lakehead University and supported by the Mitacs Accelerate program. The award provides multi-year funding for internships and research expenses, with a total value exceeding CAD $135,000.
Design of lab-on-a-chip devices for rapid point-of-care testing.
Microfluidic circuits for controlled dosing and formulation studies.
Cell manipulation, organ-on-a-chip, and high-throughput experiments.