Wind flow in the atmospheric boundary layer is multi-scale and strongly influenced by complex topographies such as urban environments. Rapidly growing, dense cities in combination with extreme weather events present new challenges for the predictive assessment of local wind fields and infrastructure planning. Advances in computational methods, physical modeling and hardware allow an increased use of computer simulation to address these challenges. The feasibility and reliability of such computational models depend to a large extent on the effort needed to prepare the model, the size and quality of the mesh, and the definition of meteorologically consistent wind boundary conditions for the micro-scale. These problems will be tackled in this project by combining multidisciplinary expertise with recent advances in parallel embedded flow computation, adjoint sensitivity analysis, mesh adaptation and co-simulation: The goal is to extend those methods such that a streamlined CFD simulation approach is developed with a robust treatment of complex “dirty” city geometry descriptions and goal-oriented adaptation of meshes using mesh quality indicators based on filtered adjoint sensitivity information to obtain an optimal model size. This is complemented by a co-simulation strategy to couple meso-scale simulations with the new micro-scale CFD model via stochastic inlet generation procedures. Implementations will be done within the open-source software Kratos to maximize dissemination. Demonstrator city-scale simulations and cooperation with world-leading companies will be conducted.
Warnakulasuriya, Suneth J., et al., Stabilization of a Time-Dependent Discrete Adjoint Solver for Chaotic Incompressible Flows (2018), PAMM (Proceedings in applied mathematics and mechanics) [1617-7061], e201800124, doi: 10.1002/pamm.201800124