A lot of research has been done in the past about the importance of good indoor climate for health and well-being. Indoor air quality as well as thermal comfort are impacted by mechanical ventilation. Several different mechanisms for mechanical ventilation are known, such as forced convection, natural convection, or Coanda effect. Often, ventilation strategies are investigated using computationally expensive Computational Fluid Dynamics (CFD) modeling. This limits the scope of parametric studies. Recently, researchers started utilizing machine learning (ML) approaches for faster prediction of the flow field. The goal is to conduct a sensitivity analysis of three different mechanical ventilation systems using a coupled approach of CFD and ML modeling. Specifically, the possibility of transfer learning within the ML models of the three ventilation systems shall be investigated to reduce the required amount of training data. The sensitivity analysis shall include factors of thermal comfort and indoor air quality. Tasks: - Literature research - Setup of the numerical models in ANSYS Fluent including geometry and mesh generation - Validation of the numerical model - Parametric simulation of the ventilation systems - Generation of the ML models for data prediction based on CFD data - Analysis of the possibility of transfer learning - Sensitivity analysis utilizing CFD and ML data of the ventilation systems |
