Coherent Anti-Stokes Raman Spectroscopy measurements of temperature fluctuations in turbulent natural gas-fueled piloted jet diffusion flames

Abstract

Temperature fluctuations have been measured in a turbulent, natural gas-fueled, piloted jet diffusion flame with a fuel jet exit Reynolds number of 9700, using broadband Coherent Anti-Stokes Raman Spectroscopy (CARS) nitrogen thermometry with a best spatial resolution of 0.9 mm. Radial profiles of mean temperature and root mean square (rms) temperature fluctuations have been acquired and temperature probability density functions (pdfs) have been constructed for streamwise distances in the range 4.2 ≤ x d ≤ 66.7 . Comparison with thermocouple measurements shows very good agreement (within 50 K) in regions of moderate temperature gradients (jet core and jet tails), whereas in the steeper gradient jet flank areas spatial averaging leads to CARS mean temperatures lower by as much as 60 K for x d ≥ 25 and by 150 K for x d = 8.3 . The largest rms temperature fluctuations (640 K) occur at radial locations corresponding to about 70% of the full jet height, for all streamwise distances. Comparison with numerical predictions that employ a k-ε model for turbulence and a constrained equilibrium model for chemistry along with a presumed pdf shape (β function) for the mixture fraction, shows very good agreement, for x d ≥ 16.7 , between computed and measured peak and centerline mean temperatures and rms temperature fluctuations, and fair agreement for x d = 8.3 . The thermal jet widths are underpredicted for solx d ≥ 25 . Measured pdfs attain a variety of shapes, from nearly symmetric around the centerline and bimodal near the average reaction zone location, to nearly uniform in parts of the jet flanks and, finally, to triangular at the jet tails. In addition, measured centerline pdfs evolve from triangular to nearly Gaussian as x d increases. The agreement between predicted and measured pdf shapes is excellent at the centerline and is very good in other parts of the jet flame, with the possible exception of bimodal shapes around the average reaction zone location, attributed to inherent limitations in the beta function representation.