Development of a Framework using Orthogonal Grids and Immersed Boundary Methods for Large Eddy Simulation of Indoor Airflows

Abstract

Energy efficiency is regarded as a key measure to reduce greenhouse gas emissions and minimize the dependency on energy imports. In the European Union, building energy consumption, particularly in heating, represents a significant share of energy usage. Innovative strategies and measures are required to promote energy conservation. Ventilation systems play a crucial role in this scenario, constituting a significant fraction of building energy use. This includes both the electricity consumed by the fans and the heat losses associated with ventilation through the building envelope. Although ventilation consumes energy, its primary purpose is to ensure excellent Indoor Air Quality (IAQ) while directly contributing to the indoor thermal environment and overall comfort within buildings. The significance of IAQ is particularly pronounced in environments such as hospitals or clean rooms, where the effective dispersion of pollutants and prevention of airborne diseases are critical. The advent of the COVID-19 pandemic has increased the importance of understanding airborne diseases, emphasizing the dominant role of indoor spaces in disease transmission. This Ph.D. thesis aims to facilitate high-resolution Large Eddy Simulation (LES) for indoor airflow. A numerical framework is implemented and tested by adapting an existing flow solver initially developed for hydrodynamic simulations, namely REEF3D. REEF3D is an incompressible flow solver based on staggered orthogonal grids. It can simulate the fluid-solid interaction of rigid moving structures within a viscous fluid using Immersed Boundary Method (IBM). To adapt REEF3D, a new low-dispersion central scheme for the convective term is required to perform explicit LES on staggered grids. This scheme is developed, implemented and validated in this thesis. The new finite difference scheme (HCDS6) conserves the discrete mass and momentum with limited production or dissipation of discrete kinetic energy. The performance of the numerical approach is evaluated on viscous and inviscid flow simulations conducted on both uniform and non-uniform grids. First, a set of benchmark test cases without IBM is selected, such as the convection of an isentropic vortex, 3D Taylor-Green vortex flow and simulation of turbulent channel flow. The results indicate that the proposed scheme is more accurate than the standard second-order scheme. Moreover, it has a numerical stencil that is more compact compared to existing fourth-order kinetic energy-conserving schemes, which makes its implementation and the treatment of boundary conditions easier. In the second validation, two additional benchmark test cases are introduced to assess the capabilities of the numerical approach in scenarios where the IBM is employed. These cases include two generic benchmark test cases: the flow past a wall-attached cube and steady non-axisymmetric flow past a sphere. The findings indicate that the IBM can accurately capture the detached flow around an object, whether it has a smooth slope (such as a sphere) or sharp edges (like a wall-mounted cube). Finally, the potential of the framework is investigated for contaminant breach in isolation rooms with a sliding door. This containment failure due to airflows induced by sliding door movement is a critical concern, particularly in healthcare facilities. Traditional hinged doors exacerbate this issue, making sliding doors an attractive alternative. The study employs Computational Fluid Dynamics (CFD) simulations using the LES approach to predict detailed airflow patterns during sliding door operations. An improved version of a continuous, direct forcing Immersed Boundary Method is used for modeling a rigid sliding door. It is based on an implicit representation of the body on a stationary grid using a level set function. This test case demonstrates that IBM can simulate moving objects for airflow inside buildings. The findings can provide practical knowledge for healthcare facility design and overall occupant safety. An inherent challenge of IBM is to accurately represent high Reynolds number flows. Therefore, for further work, wall functions should be implemented in REEF3D, with a thorough investigation into their impact on the solution.