Content area
Abstract
When an aircraft flies through a cloud of water at temperature below freezing point, ice can occur on the aircraft structure. Several ice protection systems can be used to keep the wing clear of ice. Among them, the hot air anti-icing systems are used frequently on turbo-reacted airplanes. This thesis presents a numerical code for the simulation of a hot air anti-icing device for airplane wings.
A 2D steady mathematical model of a hot air anti-icing system is presented. The mass of impinging water is evaluated from the potential flow solution around the airfoil. From a position upstream of the airfoil, the droplet trajectories are calculated by solving a momentum equation that accounts for the drag and inertia forces. The collection efficiency and the impinging water mass are estimated from droplet impinging points on the airfoil.
The convection and evaporation rates are found by solving the boundary layer equation for momentum, energy and mass diffusion. Surface boundary conditions depend on temperature and on the presence or absence of water. Effect of runback water on the momentum boundary layer is taken into account by considering that the water film creates roughness on the airfoil surface.
The 2D conduction equation for solids gives the temperature distribution inside the metal skin. The external boundary condition depends on the surface state. When either only liquid water or only ice is present in an area, an imposed heat flux found from the boundary layer solution is used. When liquid water and ice are present together, surface temperature is fixed to freezing point temperature. On the internal side of the metal skin, a heat flux or a convection coefficient is used.
The boundary layer equations are solved by a finite difference method. A finite volume method is also used to solve the conduction equation inside the metal skin. The final temperatures inside the metal skin and the amount of water evaporated are obtained by solving successively the boundary layer equation and the conduction equation. Iterations stop when energy balance across the metal skin is reached.
Results obtained with the boundary layer code compare well with experimental results, except in the case of a favorable pressure gradient. The heat transfer predictions for a turbulent boundary layer around a cylinder are not entirely satisfactory. The heat transfer around a rough cylinder increases faster with the Reynolds number in the numerical results than in the experimental results. (Abstract shortened by UMI.)
