Experimental and numerical analysis of isothermal turbulent flows in interacting low NOx burners in coal-fired furnaces

Abstract

Coal firing power stations represent the second largest source of global NOx emissions. The current practice of predicting likely exit NOx levels from multi-burner furnaces on the basis of single burner test rig data has been proven inadequate. Therefore, to further improve current NOx reduction technologies and assist in the assessment of NOx levels in new and retrofit plant cases, an improved understanding of the impact of burner interactions is required.

The aim of this research is two-fold: firstly, to experimentally investigate isothermal flow interactions in multi-burner arrays for different swirl directions and burner pitches in order to gain a better understanding of burner interaction effects within multi-burner furnaces. Secondly, to carry out numerical modelling in order to determine turbulence models which give the best agreement to experimental data. Experimental investigations were carried out using flow visualisation for qualitative and 3D laser Doppler anemometry for quantitative measurements. Numerical modelling was performed using the computational fluid dynamics software, Fluent, to compare performance between k-ε, k- ω and RSM turbulence models.

Experimental investigation showed that the recirculation zone of the chequerboard configuration is more sensitive to the change in pitch than that of the columnar configuration. Further, it was found that the smaller pitch is more sensitive to change in configuration than the wider pitch. The analysis of fluctuating components, u’, v’ and w’ showed that the burner flow is highly anisotropic at burner exit.

Numerical investigation showed that the k-ω turbulence model consistently performed below the other two models. The statistical comparison between k-ε and RSM turbulence models revealed that, for prediction of the swirl velocity profiles, the RSM model overall performed better than the k-ε turbulence model. The visual and statistical analyses of turbulent kinetic energy profiles also showed that the RSM turbulence model provides a closer match to the experimental data than the k-ε turbulence model.