Burn Angle and Its Implications on Flame Spread Rate, Mass Burning Rate, and Fuel Temperature for Downward Flame Spread over Thin PMMA. Issue 8 (2nd August 2020)

• Record Type: Journal Article
• Authors: Carmignani, L.; Bhattacharjee, S.
• Is Part Of: Combustion science and technology. Volume 192:Issue 8(2020)
• Page Start: 1617
• Page End: 1632
• Publication Date: 2020-08-02
• Subjects: Downward flame spread -- mass burning rate -- pyrolysis temperature -- thin PMMA -- burn angle; Combustion -- Periodicals; Combustion engineering -- Periodicals; 541.36105
• DOI: 10.1080/00102202.2019.1618286
• Languages: English
• ISSNs: 0010-2202

Abstract

ABSTRACT: A burn angle is defined as the acute angle subtended by the pyrolyzing surface with the virgin fuel surface and is an easily measurable overall characteristic parameter of flame spread, just like flame height and spread rate. Downward flame spread and space-based microgravity experiments are conducted using polymethyl-methacrylate (PMMA) samples with thickness between 50 μm and 3 mm to correlate the burn angle with mass burning rate, an average temperature of the pyrolysis region, and flame spread rate. For very thin fuels, the burn angle is deduced from the ratio of fuel thickness and length of the pyrolysis region, whereas for relatively thicker fuels it is directly measured. Direct and indirect measurements of burn angles, as well as data from literature, show an initial increase of their values with fuel thickness, and after reaching a peak they decrease asymptotically in the limit of semi-infinite fuel beds. A simplified phenomenological model is developed to describe the burn angle behavior as a function of fuel thickness and is used to successfully predict the average burning rate and the length of the pyrolysis zone. The peak reached by the burn angle after the initial rise can only be explained if the average temperature of the pyrolysis region is assumed to vary with fuel thickness; the thinner the fuel, the higher the temperature. Although direct experiments are needed to verify this prediction, numerical solutions obtained with an existing code corroborate this conclusion.