Healthy nasal respiration is essential for the human well-being. Functional deficiencies caused by anatomic deflective positions may lead to an increase of the respiratory resistance and an insufficient moisturization and heating of the inhaled air for the lungs. From a physics point of view, fluid mechanical properties of respiration, such as the pressure loss, the wall-shear stress distribution and the heating capability, allow to evaluate nasal cavities. Their analysis offers new opportunities to extend available diagnostic methods frequently used in surgical decision processes. Since in vivo measurements of the fluid mechanics are not always possible and often error-prone, Computational Fluid Dynamics methods represent a valuable alternative. A software to simulate nasal cavity flows and to conduct virtual surgeries at run time is developed to support physicians in decision making. From computer tomography images the geometry of the nasal cavity is extracted. For the massively parallel simulation, a thermal lattice-Boltzmann method (TLBM) is used. To in-situ analyze the outcome of a virtual surgery, the TLBM is coupled to a level-set solver. The latter together with an immersed boundary approach enables to efficiently implement anatomical changes resulting from a virtual surgery by geometry adaptation at simulation run time using a level-set.