Fluid flow simulation over complex shape objects using image processing to achieve mesh generation

In the domain of flow simulation, avoiding the manual conception and numerisation of the domain can lead to the saving of a certain amount of time. Some processes, using heavy devices like LASER metrology, allow the numerical reconstruction of a real object. The aim of this paper is to propose a more simple tool requiring a commercial digital camera (like a smartphone), to transform a digital picture into a ready to use mesh. Besides simplicity, the tool has to be precise enough to bring accurate simulation results. Then, image processing object detection and reconstruction are used to generate a 2D mesh that can be integrated in a finite volume transient CFD simulation. Cars and airfoils are chosen as objects and the DNS fluid flow Gerris solver performs the simulations. After a validation on a circular shape object, simulations, conducted at different Reynolds number, provide accurate results plotting the Von Karman alley regime.