Professor Peter Willemsen

This work involves developing advanced software and algorithms, particularly leveraging GPU computing to create large-scale, dynamic simulations. The code related to this project is called QES. QES (Quick Environmental Simulation) code is a low-computational-cost framework designed to compute high-resolution wind and concentration fields in complex atmospheric-boundary-layer environments. QES is written in C++ and NVIDIA's CUDA for Graphics Processing Unit (GPU) acceleration. The code uses NVIDIA's dynamic parallelism API to substantially accelerate simulations.  The code includes QES-Winds, QES-Turb and QES-Plume.

• QES-Winds is a fast-response 3D diagnostic urban wind model using a mass-conserving wind-field solver. QES-Winds uses a variational analysis technique to ensure the conservation of mass rather than slower yet more physics-based solvers that include the conservation of momentum. QES-Winds minimizes the difference between an initial wind field that is specified using empirical parameterizations and the final wind field. This method requires the solution of a Poisson equation for Lagrange multipliers. The Poisson equation is solved using the Successive Over-Relaxation (SOR) method (an iterative solver), which is a variant of the Gauss-Seidel method with more rapid convergence.
• QES-Turb is a turbulence model based on Prandtl’s mixing-length and Boussinesq eddy-viscosity hypotheses. QES-Turb computes the stress tensor using local velocity gradients and some emprical non-local parameterizations.
• QES-Plume is a stochastic Lagrangian dispersion model using QES-Winds mean wind field and QES-Turb turbulence fields. QES-Plume solves the generalized Langevin equations to compute the fluctuations of the particle in the turbulent flow fluid. A time-implicit integration scheme is used to solve the Langevin equation, eliminating 'rogue' trajectories. The particles are advanced using a forward Euler scheme. QES-Plume is also a stand-alone dispersion model that can run using fields from diverses sources such as RANS or LES models.