Abstract
Accurate point cloud representations of complex geometries are today relatively easy to obtain at a low cost with a 3D-scanner. The majority of present CFD simulation tools require a body-fitted mesh in order to impose the correct boundary conditions of the fluid flow equations on geometrical objects. Further, the generation of the mesh is generally based on some form of surface representation of the geometrical object itself. This means that a high-quality surface representation must first be generated from the point cloud which is a challenging and time-consuming task
In this study the fluid flow solver IBOFlow, developed at the Fraunhofer-Chalmers Research Centre for Industrial Mathematics, is used for CFD simulation directly on a point cloud. In IBOFlow an adaptive octree mesh is combined with novel immersed boundary conditions to handle interior objects in the computational domain. By avoiding a body-fitted mesh the grid generation can be made fully automatic and the manual preprocessing reduced to a minimum. The immersed boundary method normally uses a surface triangulation to represent the geometrical objects. In this work the method is extended to also handle point cloud data directly with a voxelization-technique, and the flow may therefore instead be simulated directly on the point cloud. It is further also possible to include CAD-objects or even additional scanned point clouds in the same simulation, so that the effect of the placement of an object in a scanned environment may be analyzed. A similar analysis in software using body fitted meshes would require the complex manual preprocessing to be executed again for each new geometrical setup. With the immersed boundary method, on the other hand, arbitrary combinations may be simulated with minimal extra effort from the user.
A case study is performed by simulating the air flow in part of the campus of Chalmers University of Technology in Gothenburg, Sweden. In the numerical model a scanned point cloud is used as direct representation of the campus’ geometry. The obtained results demonstrate that the method is both fast and stable. For planning and design of buildings and urban environment, CFD simulations can therefore be used to, for example, predict wind loads, temperatures or spreading of pollution.