- Abstract
Evaluating the heat transfer between the building envelope and the indoor environment is crucial in building simulations. Accurate values for the heat transfer can contribute to the design of more energy-efficient buildings. Typically, heat transfer processes, especially the convective heat transfer, are calculated using empirical models such as heat transfer coefficients, since fluid simulations are normally computationally intensive. These empirical models are generally developed under steady-state conditions. Therefore, the models may not be suitable for dynamic room environments. In such cases, a dynamic heat transfer simulation based on a faster fluid simulation could be a viable solution. The “Fast fluid dynamics” algorithm, originally developed for fast fluid visualization, shows potential. This paper explores and implements optimization possibilities for this fluid simulation algorithm for the use in building simulations with natural convection. Benchmark comparisons demonstrate that the results achieve reasonable accuracy. The computational speed was improved by approximately 1.2 times through the implementation of a conjugate gradient (CG) solver. Additionally, an extra acceleration of 1.65 times was achieved with an optimized multigrid preconditioned CG solver, resulting in a computational time up to 46.5 times faster than conventional computational fluid dynamics simulations for the investigated problem. The results demonstrate significant potential for integrating the algorithm into dynamic building heat transfer simulations.
