A study of selected problems related to accidental process fires

An experimental full scale characterization of a turbulent propane jet flame has been made in terms of temperatures and radiation. Sonic propane gas releases were achieved at steady pressure and near steady flow. Commercial propane was used, consisting of a mixture of propane with very small admixtures of sulphur and methanol. The size of the fire was 13-14MW (average burning rate 0,3kg/s). The pressure drop across the horizontally mounted nozzle was 10.3 barg. The experimental setup was simulated using the CFD-code Kameleon FireEx, and characterizations were made for temperature, radiation and gas velocity. The results from experiments and simulations were compared using interpolation techniques for reducing the errors of measurements, and MatLab for visualization. Both transient and time-averaged values were plotted. The main findings in this work were: • the length of the visible flame was ≈ 5.5m, with a lift-off distance of 0.6m • the highest temperature region of the jet flame was ≈ 70% along the visible flame length (i.e not including lift-off). The maximum temperature in the flame was in the region 1200 − 13000C • up to ≈ 3m, there was a fuel rich region along the centre trajectory of the flame, where the temperature was ≈ 2000C less than in the stoichiometric region, 0.3m away from the centre line • the radiation fraction along the jet trajectory at positions 25%, 50%, 70%, and 95% downstream of the visible flame length was 28%, 57%, 73%, and 63%, respectively • moving outside the flame perpendicular to the jet axis, the radiation fraction gradually increased. At 3m distance from the centerline, it was equal to the total heat flux. This indicated that the convection fraction was close to zero • the radiation heat flux sensors were extremely sensitive to unclean environment. Even when applying nitrogen for purging, it did not keep the soot and other particles away from the inner surface of the gauge’s restrictor • the CFD-code KFX predicted a correct flame length, but estimated a slightly shorter lift-off distance • the end part of the KFX-flame was more influenced by buoyancy and deviated some from that of the experiment • the measurements showed more irregular shaped temperature fields compared to the simulated • the measurements showed larger fluctuations in the temperature fields compared to the simulated • the maximum measured radiative heat flux inside the flame was 185kW/m2. The maximum simulated radiative heat flux was 193kW/m2, representing a deviation of 4.3% • the maximum measured total heat flux was 256kW/m2 A steel cylinder of radius 160mm was placed at various positions in the jet, and the relative heat transfer was assessed by means of thermocouples placed radially inside the cylinder. This work showed that: • convection is the major contributor to the total heat transfer from a turbulent jet flame to a steel cylinder impinged by the flame • the largest rate of heat transfer is at the side facing the flame, i.e no high levels of turbulence induced thermal loading could be detected at the back • the heat transfer coefficient, h, is a function of the velocity of the gas flow relative to the impinged object The stability of ignited propane gas jets, discharged from circular cross section outlets of varying diameters and inclinations were examined. This resulted in: • a model, with an accuracy of 0.89, that predicts the upper and lower blowout limits for propane in gas phase, as well as a critical outlet diameter of 14mm • no observations were made indicating that the outlet inclination has any effect on the blowout limits The heat attenuation in water spray in a full scale offshore flare situation was examined by applying a known model for calculations and comparing with measurements. The result of this work was: • the model predictions slightly under estimated the capacity of the water curtain. There were, however, uncertainties regarding the water curtain properties, and more detailed measurements are necessary in order to present a verified model for engineering use • for the extreme situations of an underestimate of drop sizes, where the actual drop sizes are 50% larger than estimated, or where the actual small drop fraction is doubled, calculation errors will be caused in the range -7% to +18%, in absolute terms • the calculation model is not capable of identifying irregularities within the water curtain. This will have importance relating to maximum allowable radiation limits where people are expose