Developments in Lattice Boltzmann Flow Solver
This project is focused on building technology to model the flow of single and multiphase fluids. This will rely on simulating the fluid flows with a Lattice-Boltzmann approach, as a complementary capability to other fluid methodologies that are being developed in PREMIER. Lattice Boltzmann approaches are very flexible and allow the simulation of granular flows, multiphase flows and flows through porous media. They also allow the easy incorporation of extra physics where needed, for example around walls and interfaces.
The development of this capability will also allow the modelling of flows in “high-Knudsen” regimes, where the particles in the fluid do not collide enough to be accurately modelled by the traditional Navier-Stokes equation. This can be important in multiphase flows, in flows very close to walls or obstacles and when considering kinetic reactions on surfaces.
Specifically, this work will build upon the numerical technology developed in the AMCG (Applied Modelling & Computation Group) at Imperial College London for the Boltzmann Transport Equation (BTE) in nuclear applications (). Boltzmann approaches for modelling fluid flows, like Lattice Boltzmann and other “kinetic” methods, share many important similarities with the BTE. Our work on the BTE has shown the viability of performing adaptivity in a Boltzmann context, allowing the description of the direction/velocity of fluid particles to become more accurate only where needed, which enables the simulation of difficult flow conditions that would otherwise be inaccessible.
Furthermore, traditional Lattice-Boltzmann methods often use structured grids in space to represent their domains, whereas this work will develop unstructured grid technology allowing flexibility in accurately modelling curved surfaces. This will be enabled by the adaptivity technology described above, but additionally applied to focus on spatial resolution where required. Also, flexibility in time-stepping methods will be supported by the parallel iterative methods we developed previously for the BTE, that give good scaling on modern high-performance computers. The combination of these technologies will significantly enhance the types of fluid flows that can be accurately modelled.
