### CFD prediction of thermal radiation of large, sooty, hydrocarbon pool fires

Duisburg, Essen (2009), 183 S.

Dissertation / Fach: Chemie

Fakultät für Chemie » Technische Chemie

Schönbucher, Axel (Doktorvater, Betreuerin)

Redeker, Tammo (GutachterIn)

Volltext in DuEPublico: Dokument 20400

Duisburg, Essen, Univ., Diss., 2009

Signatur des Buches: ZZX663086

##### Abstract:

Computational Fluid Dynamics (CFD) simulations of large-scale JP-4 pool fires with pool diameters of d = 2 m, 8 m, 16 m, 20 m and 25 m in a calm condition, as well as with pool diameters of d = 2 m, 20 m and 25 m under cross-wind conditions with wind velocities in a range of 0.7 m/s < uw < 16 m/s are performed. CFD prediction of emission temperatures T, surface emissive power (SEP) and irradiances E(Δy/d) at relative distances Δy/d in horizontal direction from the pool rim is carried out. Also, for the theoretical understanding of large pool fires, the time dependent flame temperatures are of great interest. CFD predicted vertical temperature profiles for different relative radial distances y/d = 0, y/d = 0.05 and y/d = 0.1 show that the absolute maximum flame temperatures are away (y/d = 0.05) from the flame axis and depend on d: 1300 K (d = 2 m), 1250 K (d = 8 m), 1230 K (d = 16 m), 1200 K (d = 25 m) which agree well with the measured temperatures. CFD predicted radial temperature profiles dependent on x/d are in agreement with measurements. For pool fire with d = 25 m, at x/d = 0.125 bimodal profiles are found, while for x/d = 0.25 unimodal temperature profiles exist. The CFD simulation of the "derived" quantity SEP requires a definition of the flame surface. The present work presents three different ways to predict SEPCFD. The first way is the determination of isosurfaces of constant temperature which is defined as the flame surface. The second way considers that the flame surface results from the integration of many parallel two-dimensional distributions of incident radiation G(x, y) along the z-axis perpendicular to the xy-plane. In the third way a virtual wide-angle radiometer is defined at the pool rim and the irradiance E(Δy/d) as a function of Δy/d is simulated. To simulate the SEP, more exactly, the temperature dependent effective absorption coefficient of the dissipative structures (reaction zones, hot spots and soot parcels) and air as a four-step discontinuity function is developed. CFD predicted values of JP-4 pool fires, obtained by the third way, are: 105 kW/m² (d = 2 m), 65 kW/m² (d = 8 m), 45 kW/m² (d = 16 m) and 35 kW/m² (d = 25 m). The SEPCFD value for d = 2 m under predicts the SEPexp by a factor of 0.8 whereas a good agreement is found between SEPCFD(d) and SEPexp(d) for d = 8 m, 16 m and 25 m. Based on the first way the SEPCFD values agree well with the measured SEPexp values if the flame surface temperature of 1100 K is used for d = 2 m, 500 K for d = 8 m and 400 K for d = 16 m and 25 m. Instantaneous h(T), h(SEP) and time averaged histograms, lead to probability density functions of the emission surface temperatures (flame temperatures) pdf(T)and the surface emissive power pdf(SEP), determined by the second way. For example, from the predicted pdf(TCFD) and pdf(SEPCFD) for d = 16 m temperature and SEP are in the intervals of 648 K < TCFD < 1100 K and 10 kW/m² < SEPCFD < 80 kW/m². The measured values are in the intervals 633 K < Texp < 1200 K and 9 kW/m² < SEPexp < 114 kW/m². The CFD predicted functions pdf(T), pdf(SEP) are consistent with the measured pdfs. CFD predicted time averaged irradiances E(Δy/d, d) under predicts the measured Eexp(Δy/d) at the pool rim Δy/d = 0 for d = 2 m by a factor of 0.8 and over predicts Eexp(Δy/d) up to the factor of 1.6 at Δy/d = 0.5 whereas for d = 8 m, 16 m and 25 m the irradiances ECFD(Δy/d) agree well with the measured Eexp(Δy/d). For example, Eexp(Δy/d) as a function of d at Δy/d = 0.5 the following values are found: 28 kW/m² (d = 2 m), 18 kW/m² (d = 8 m) and 5 kW/m² (d = 25 m). The wind influence on large pool fire is a complex phenomenon. CFD simulation shows that the wind influences the flame length, flame tilt, flame drag, the flame temperatures T, the SEP and the irradiances E. With increasing wind velocity uw from 4.5 m/s to 10 m/s SEPCFD and ECFD(Δy/d) at the pool rim increase downwind by a factor of about 2 – 6 for d = 2 m and by a factor of about 2 – 7 for d = 20 m. In both cases ECFD(Δy/d) do not increase if uw > 10 m/s as it is found in experiments. In the upper section of the flames, depending on the flame tilt and drag, a decrease of flame temperature of several hundreds K is found. With increasing wind velocity uw > 2.3 m/s the predicted flame tilt from the vertical becomes significant. The CFD results show that two counter rotating vortices at the leeward side of the fire are formed at the minimum uw = 1.4 m/s as observed in experiments. The predicted flame tilt and drag for d = 20 m begins from 20° and 1.1 at uw = 1.4 m/s and ends with 80° and 2.5 at uw = 16 m/s which agree with the experimental data. In a case of d = 2 m the flame tilt and drag reach values of 60° and 2.5 for uw = 4.5 m/s and 80° and 2.8 for uw = 16 m/s as in the experiments. Flame temperatures T, surface emissive power SEP, irradiances E and the wind influence on large pool fires were at the first time predicted with CFD simulations. The CFD predictions are generally in good agreement with the measured values. The CFD simulations allow (for future), the estimation of wind effects and also the important influence of multiple fires on the hazard potential. The present work has, also shown that the hazard potential of large pool fires with CFD simulations of the thermal radiation can be estimated much better than before.