A study on the effect of inlet turbulence on gas mixing for single point aerosol sampling

Abstract

The efficiency of certain mixing elements in achieving conditions suited for single point sampling is evaluated. Experimental measurements of velocity and tracer gas concentration are taken to determine the same. Readings are taken under conditions of statistically steady developing flow in a straight duct. Mixing is evaluated for inlet intensities of 1.5%, 10% and 20%, achieved by introducing various bi-plane grids and for a commercial static gas mixer. Reynolds number is varied between 5000 and 16000 and has negligible effect on mixing. The obtained data highlighted the importance of inlet turbulence intensity over Reynolds number in the process of turbulent dispersion of a dilute gas. All mixing data are obtained for Reynolds number around 15000. A semi-empirical correlation to predict the extent of mixing, as characterized by the Coefficient of Variation (COV) over a U.S. EPA sampling grid, with a given turbulent intensity profile is proposed and its results match favorably with the data. The correlation incorporates the idea of a history of intensity influencing the mixing at any downstream point and is much better than an earlier correlation which failed to incorporate the history of fluctuations. It could be included as a sub-model in software like DEPOSITION. Experiments with the commercial static gas mixer show that, unlike the bi-plane grids, the turbulence downstream of the mixer is not homogenous. The results showed enhanced mixing that attained the specified ANSI N13.1 1999 criteria rapidly and selection of the release point for tracer gas plays a significant role in determining the extent of mixing. The gas mixer does not introduce significant pressure losses. It is also seen that while flow straighteners reduce flow swirl, they may not be effective for achieving a uniform velocity profile. Numerical computations are performed with commercially available computational fluid dynamics (CFD) software (FLUENT[], Version 5.4), and the performance of the turbulence and particle tracking models is evaluated. Flow field predictions match favorably with experimental data. Results from the particle-tracking model show good qualitative trends, but they cannot be used to determine compliance with the requirements of the ANSI standard.