| dc.description.abstract | The key to controlling accidental fires, is early detection by the correct placement of devices such as smoke detectors and sprinkler systems. Smoke is accompanied by a temperature rise, brought about by the ceiling jet resulting from a fire. Accurate prediction of temperatures in the ceiling jet, by either empirical correlations or numerical methods, therefore gives ground for suggesting placement of devices. Numerical methods such as CFD (Computational Fluid Dynamics) codes are continuously improved and need validation. In this work, two CFD models, FLACS and FDS, are evaluated against experimental measurements and empirical correlations presented by Alpert. By varying heat release rate (HRR), ceiling height and burner surface area, both weak and strong plume-driven flows are produced. Maximum temperatures for radial positions along the ceiling are found to be well predicted with both FDS and the correlation, but is over-predicted to a variable degree by FLACS. FDS also shows temperatures closer to experimental values near the fire, but for distances further away, temperatures drop too quickly towards room temperature, resulting in a too narrow ceiling jet. FLACS captures more of the temperature distribution below the ceiling and thereby predict ceiling jet thicknesses more accurately than FDS. The correlation predict thicknesses even closer to experimental values. Effects of changing HRR, ceiling height and burner size are found to be present to a variable degree, while the effect of grid size is found to be critical for accurate numerical results. | en_US |
