Computational fluid dynamics simulations of thermal flows in various applications

The present thesis belongs to the field of computational physics. In particular, computational fluid dynamics (CFD) methods are utilized to simulate heat transfer and fluid flow phenomena, with the methodological emphasis on using scale-resolving large-eddy simulations (LES). The overall objective of the thesis is to utilize LES in various societally relevant applications on various scales. The dissertation is among pioneering works on the usage of LES in conjugate heat transfer, recovery boiler, and indoor airflow context. In Publication I, a liquid cooling heat exchanger with embedded channels for high power electronics was designed and numerically analyzed. Experiments carried by co-authors are also reported in the publication to confirm the functionality of the heat exchanger. The numerical analysis was performed with three different turbulence models, and the comparison revealed significant differences in the thermal performance based on the capturing of the relevant flow features. In particular, the overprediction in the level of turbulence was observed to also overpredict the turbulent heat transfer and consequently, underpredict the surface temperature levels. In Publication II, a kraft recovery boiler with two different secondary air jet configurations was analyzed with LES. In addition to the flow field, dispersion of black liquor droplets was simulated as solid particles with Lagrangian particle tracking (LPT). The two jet configurations exhibited differences in the mixing of the hot and cool gases as well as in the dispersion of the particles. A staggered jet configuration provided more uniform temperature levels compared to an in-line configuration. However, significantly more fouling was observed in the staggered configuration. In Publication III, a large room with mechanical ventilation was simulated using LES. The simulation case was motivated by a real life SARS-CoV-2 virus super-spreading event during a choir practice. The emphasis of the simulation was on the buoyancy-driven flows generated by the radiators and the choir attendees, and the effect of buoyancy on the dispersion of the respiratory aerosols. The presence of buoyancy was noted to increase the turbulent dispersion of the respiratory aerosols significantly. Pope's criterion was utilized to verify the numerical solution in Publication II and III. As an overall conclusion, usage of scale-resolving techniques (e.g. LES) appears important in predicting flows with transitional features and strong turbulence generation affecting e.g particle dispersion patterns or wall heat transfer.