Imaging measurements and LES-CMC modeling of a partially-premixed turbulent dimethyl ether/air jet flame

Abstract

Turbulent dimethyl ether (DME) jet flames provide a canonical flame geometry for studying turbulence-flame interactions in oxygenated fuels and for developing predictive models of these interactions. The development of accurate models for DME/air flames would establish a foundation for studies of more complex oxygenated fuels. We present a joint experimental and computational investigation of the velocity field and OH and CH2O distributions in a piloted, partially-premixed turbulent DME/air jet flame with a jet exit Reynolds number, ReD, of 29,300. The turbulent DME/air flame is analogous to the well-studied, partially-premixed methane/air jet flame, Sandia Flame D, with identical stoichiometric mixture fraction, ξst = 0.35, and bulk jet exit velocity, Vbulk = 45.9 m/s. Measurements include particle image velocimetry (PIV) and simultaneous CH2O and OH laser-induced fluorescence (LIF) imaging. Simulations are performed using a large eddy simulation combined with conditional moment closure (LES-CMC) on an intermediate size grid of 1.3 million cells. Overall, the downstream evolution of the mean and RMS profiles of velocity, OH, and CH2O are well predicted, with the largest discrepancies occurring for CH2O at x/D = 20-25. LES-CMC simulations employing two different chemical reaction mechanisms (Kaiser et al., 2000 [20] and Zhao et al., 2008 [21]) show approximately a factor of two difference in the peak CH2O mole fractions, whereas OH mole fractions are in good agreement between the two mechanisms. The single-shot LIF measurements of OH and CH2O show a wide range of separation distances between the spatial distributions of these intermediate species with gaps on the order of millimeters. The intermittency in the overlap between these species indicates that the consumption rates of formaldehyde by OH in the turbulent DME/air jet flame may be highly intermittent with significant departures from flamelet models.