Dr Gerald Pereira1, Dr David Howard1, Dr Paulus Lahur1, Dr Mike Horne1
1Csiro, Melbourne/Clayton, Australia
Static mixers are used in a variety of industries from petrochemicals, minerals processing to chemicals and bio-technology to name but a few. Small volumes of different fluids need to be mixed together in small-scale devices. Nowadays, many industries require to do this in a continuous process and so the fluid is passed through static-mixers which have no moving parts, but need to have sufficiently complex geometries to maximize mixing. Since the fluids pass through narrow channels, the flow is laminar meaning the streamlines do not intersect. Mixing then will be very slow – only due to molecular diffusion. One needs to devise clever geometries of the static-mixer to induce chaotic flows which enhance the fluid mixing. Up to now, design of smart mixing devices has been based on experience and best guesses.
A significant recent technological advance for this field is the capacity to print solid motifs at a relatively fine-grained scale using 3D printing methods. This advance allows one to build solid structures of immeasurably more complexity than up to now. This opens up the parameter space for possible smart mixer designs but one needs to identify and then print these designs. We outline a scientific workflow which aims to quantitatively predict optimal mixer designs in different mixing regimes via a combination of numerical analysis (lattice Boltzmann method) for the fluid flow modelling and artificial intelligence using generative design algorithms. The workflow is implemented over many computer nodes with multi-cores to rapidly converge to an optimal solution.
Dr Pereira completed his PhD in Applied Mathematics where he implemented random walk models to describe polymer behaviour and fluid mechanical models for flow in confined random environments. He gained experience in a variety of numerical techniques. Subsequently he took up a research fellowship at the Cavendish Laboratory (Cambridge, UK), followed by an ARC QEII fellowship before obtaining a tenured academic position in the School of Physics at Victoria University Wellington (NZ). He returned to Australia, due to family commitments, where he took up a continuing scientist position in the CFD group in the (then) Division of Mathematics and Statistics at CSIRO.
Currently he develops numerical methodologies to solve fluid dynamical problems. These methods include the lattice Boltzmann method (LBM), Smoothed Particle Hydrodynamics (SPH) and Molecular Dynamics (MD). He also applies these methods to solve real-world problems in various industries including minerals, energy and manufacturing.