Sooting behaviour of turbulent non-premixed jet flames.

Soot measurements using planar laser-induced incandescence (LII) have been performed in three commercial propane-fired flames produced by simple jet (SJ), precessing jet (PJ-P) and bluff-body jet (BB) burners. The flames span a wide range of different global mixing rates for the same nozzle diameter and fuel flow rate. Measurements of their radiation, NOₓ emissions and residence times are available. LII has also been performed in a natural gas-fired precessing jet (PJ-NG) flame with the same nozzle diameter and throat Reynolds number as the PJ-P and, in the well characterised “Delft III Flame” from the TNF workshop firing simulated Dutch natural gas. The PJ-P and the SJ have a similar time-averaged soot volume fraction, FV [F in italics & ¯ accented; V in italics & subscript], which is about an order of magnitude higher than in the BB and about 20 times higher than in the PJ-NG. The integrated total volume of soot, however, in the PJ-P is about 2.5 time higher than in the SJ, 25 times higher than in the PJ-NG and 220 times higher than in the BB. The axial location of the peak, time-averaged soot volume fraction, FV,max [F in italics & ¯ accented; V,max in italics & subscript], in the two propane flames issued from a long pipe, the SJ and the BB, are very similar when normalised to the flame length (x/Lfl≈0.6) [x/Lfl in italics; fl subscript] and consistent with other measurements in simple jet flames. The PJ-P and PJ-NG flames show much closer peaks at x/Lfl≈0.37 [x/Lfl in italics; fl subscript] and 0.53, respectively, also consistent with the peak heat flux in a precessing jet burner. When normalised to the length of the sooting part, the Delft flame shows a similar location of FV,max [F in italics & ¯ accented; V,max in italics & subscript] as the SJ and the BB. The soot in the Delft flame is highly intermittent with FV [F in italics & ¯ accented; V in italics & subscript] about 70 times less than in the SJ. The burnout of soot in the two propane flames issued from a long pipe, the SJ and the BB, and in the natural gas Delft flame, proceeds mainly by increasing intermittency, I [in italics], evident by the high instantaneous soot volume fraction, FV [F in italics; V in italics & subscript], and the high intermittency in the burnout region. In addition, the soot sheet dimensions do not vary significantly in the burnout region of these flames indicating that burnout proceeds by less number of soot sheets rather than a reduction in their dimensions. The burnout in the PJ-P proceeds by a different mechanism than the other turbulent flames, the asymptotic behaviour of I [in italics] and the low FV [F in italics; V in italics & subscript] reveal that the burnout in this flame is mainly due to low soot concentration. The reduction in the size of the soot sheets also indicate that both the dimensions of the sheets and the soot concentration within them drop in the burnout region. The distribution of the PDF of FV [F in italics; V in italics & subscript] in the five investigated flames is well characterised by an exponential function, with some departure at low FV [F in italics; V in italics & subscript]. The axial and radial distributions of the curve fitting parameters KP,F [K in italics; P,K in italics & subscript] and KP,O [K in italics; P,O in italics & subscript] of the exponential function for the Delft flame are presented. They were found mostly to vary smoothly with the axial height above the burner. The data provided on the well-defined Delft flame should be of great usefulness for model validation.Thesis (Ph.D.) -- University of Adelaide, School of Chemical Engineering, 201