Experimental Characterization and Numerical Modelling of Urea Water Solution Spray in High-Temperature Crossflow for Selective Catalytic Reduction Applications

Engines have improved a lot and reached a new state of the art in terms of combustion technology, but they alone still fall short in achieving emission limitations without any trade-off on performance. Selective Catalytic Reduction (SCR) is the key technology used to meet the increasingly strict Nitrogen Oxides (NOx) emission regulations. The injection of Urea Water Solution (UWS—32.5% urea solution) upstream the catalyst is currently the leading technique for reducing the emission of NOx from the exhaust (DeNOx). A uniform distribution of the spray droplets is very crucial to achieve a good conversion efficiency. Therefore, the size and velocity distribution of the droplets are of high importance in deciding the fate of the DeNOx process. This article describes an approach of modelling the UWS spray and its validation against experimental data collected under realistic exhaust-like conditions. Droplet size distributions and velocities were recorded using Phase Doppler Anemometry (PDA) in a high-temperature wind tunnel. Simulations of the spray were performed using a commercial Computational Fluid Dynamics (CFD) code, ANSYS Fluent, in Lagrangian solver framework under the same flow conditions.

The results show that initializing the droplets with the correct diameter and velocity distributions is a vital element in determining the fate of droplets and their impingement location. Using the proposed methodology, validation of diameter and velocity distributions were performed and the average deviations in Sauter Mean Diameter (SMD) were found to be less than 8%. Deviations in velocity distributions were recorded with larger differences appearing in planes closer to the nozzle. It was seen that the simulation initiated with the correct momentum had an average difference of around 8% whereas the simulation initiated with a single velocity value had an average difference of around 32%, when compared with the actual measurement data.