Hydrogen–air mixer: CFD and genetic optimization

Project · Project 34

Design and optimization of a passive hydrogen-air mixing system aimed at enhancing combustion stability through improved outlet mixture uniformity, leveraging Venturi-induced acceleration, controlled turbulence generation, and hydrogen's high diffusivity.

Geometry concept

The device geometry incorporated a Venturi flume combined with radial hydrogen injectors and a downstream cylinder array, the latter placed deliberately to induce vortex shedding and enhance mixing via turbulent diffusion. The mixing is therefore achieved with a mechanically simple, no-moving-parts architecture.

Method

High-fidelity CFD simulations were developed in Ansys Fluent, solving the compressible, unsteady Reynolds-Averaged Navier–Stokes equations with k-ω SST turbulence modelling and species transport, capturing the coupled diffusion–turbulence effects.

The numerical setup was rigorously validated against canonical benchmark cases, ensuring agreement with experimental Strouhal number and drag coefficient trends across Reynolds numbers before any optimization was attempted.

To enable scalable design exploration, the full CFD workflow was automated using Python (PyFluent), converting GUI-based simulations into scriptable pipelines and enabling parametric studies without manual intervention. This framework was then integrated with a Genetic Algorithm (PyGAD) to iteratively optimize inlet conditions, maximizing a custom-defined outlet uniformity index.

Results

The optimization process demonstrated clear performance improvements over the nominal design, achieving significantly enhanced mixing quality while maintaining the simple architecture.