Grant: Discovery Grants Program - Individual

Next-generation Canadian aircraft, jet engines, and electric propulsion systems must be optimized to be quieter and more fuel efficient whilst remaining commercially viable amidst increasing environmental regulation and competition. Their design will rely heavily on high-fidelity computational fluid dynamics solvers, specifically Large Eddy Simulation (LES), and leveraging these for aerodynamic optimization. Unlike current industry-standard tools, specifically the Reynolds Averaged Navier-Stokes approach, LES resolves the large-scale unsteady turbulent structures in a flow. This provides accurate results in flow regimes where RANS often fails, specifically transitional and separated flow. Furthermore, existing aerodynamic optimization frameworks are ill-suited for use with LES, due to the chaotic nature of the resulting solutions. The objective of this research program is to address these limitations, enabling the use of LES for design. It is expected that this will seed a step change in aerodynamic optimization frameworks, yielding reduced environmental impacts and ensuring commercial competitiveness. The first proposed work package focuses on robust and efficient time stepping algorithms. One of the primary distinguishing features of LES is that, unlike RANS, it is an unsteady simulation technique. This means that flow features have to be resolved in both space and time using accurate, efficient, and stable discretizations. The first portion of this work package will focus on internally-stable partitioned time stepping. This allows different time stepping algorithms to be used in different parts of the computational domain, based on local flow physics. It is expected that this will yield a 1-2 order of magnitude reduction in computational cost, facilitating the use of LES for optimization. The second work package focuses on creating new optimization frameworks for chaotic systems, specifically gradient-hybrid and gradient-based approaches. The gradient-hybrid approach will integrate within an existing gradient-free framework, extracting estimates of the sensitivity of objective functions, such as lift and drag, to design parameters, such as airfoil shape. The gradient-based approach will focus on reduced order modelling of LES simulations, and how to extract gradients from these. These gradients can then be used to rapidly towards optimal designs. It is expected that these will facilitation the use of LES for optimization, greatly increasing the accuracy of the design tools available for commercial aircraft design.