Radiation field modeling of the NETmix milli-photocatalytic reactor: Effect of LEDs position over the reactor window

The radiation field in the meso-scale NETmix photoreactor for gaseous streams purification was evaluated employing a Monte Carlo based algorithm through the use of the software Ansys Speos. The illumination system consists of 18 high power UV LEDs placed over the reactor window. A thin TiO2 -P25 catalyst film was immobilized in the NETmix network or in the reactor window, allowing for two illumination mechanisms: front-side (FSI) and back-side illumination (BSI). The reflection properties of the catalyst were considered to be a mixture of specular and Lambertian reflection with proportions varying as a function of light incident angle. Simulations of both illumination mechanisms were validated with experimental data returning a relative error of 2.3 and 7.0% for BSI and FSI, respectively. The distance between each LED and the distance between LEDs plate to the reactor window were varied in order to determine the impact of LEDs geometrical arrangement in the registered irradiance as well as in the radiation homogeneity over the catalyst surface. Twenty seven illuminations schemes were proposed and simulated for each BSI and FSI. Results shows that light absorption efficiency is dependent of both the distances of LED to LED and from LED to reactor window, allowing to appoint a most favorable illumination set that balances the absorption efficiency - that tends to increase with LEDs proximity - and homogeneity - that tends to decrease with proximity.

Automated summary

The radiation field in a meso-scale NETmix photoreactor for gaseous streams purification was evaluated using a Monte Carlo based algorithm with Ansys Speos software. The reactor has 18 high power UV LEDs and utilizes a thin TiO2 -P25 catalyst film for front-side and back-side illumination mechanisms. The study considered the reflection properties of the catalyst and variations in light incident angles, with simulations validated by experimental data showing a relative error of 2.3% for BSI and 7.0% for FSI.