Algorithm for soot sheet quantification in a piloted turbulent jet non-premixed natural gas flame

Abstract

A novel method to quantify soot sheets from planar images of soot volume fraction is presented and demonstrated in a well-characterised turbulent, non-premixed flame, known as the ‘Delft-Adelaide Flame’. The image processing algorithm presented is based on the adaption and combination of two existing computational methodologies that were presented in the literature. The algorithm starts by identifying the longest line spanning the object. The line is subsequently segmented repetitively to generate ‘anchor points’ that are forced to lie along the centreline of the object. The length of the soot sheet is obtained by fitting straight lines to the anchor points, whilst the average characteristic width of the sheet is determined from the mean thickness of the ellipses that are fitted to the segmented soot sheet. The algorithm employed in this method, which has an uncertainty of ∼11 %, is found to be well suited to extract the characteristic dimension and position information of soot sheet with bend, irregular shape and random orientation. Statistical assessments of these dimensions in this flame reveal that: (1) the characteristic width and length of the soot sheets range from ∼6 to ∼10 mm, and ∼30 to ∼50 mm, respectively, (2) a strong correlation exists between the soot characteristic width and length, and (3) the soot sheets display high sensitivity to local flow dynamics, with measured normalised interaction lengths ranging from 3.4 to 3.9, for the present flame.