(28g) Discrete Element Method Simulations of Dry Impregnation Processes in Catalysts
- Conference: AIChE Annual Meeting
- Year: 2008
- Proceeding: 2008 Annual Meeting
- Group: Particle Technology Forum
- Time: Monday, November 17, 2008 - 10:18am-10:36am
Catalyst impregnation is one of the most crucial steps for preparing industrial catalysts. In this process, metal complexes are dissolved in an aqueous solution and contacted with a porous oxide catalyst support such as alumina (Al2O3) or silica (SiO2). Generally, the process of impregnation is performed in rotating vessels with sprinklers that pour the aqueous solution with the metal precursor into the catalyst support. As the liquid penetrates into the dry powder, the particles start adsorbing the fluid, increasing the density of the particles, until they become saturated, after which the excess liquid forms liquid bridges between particles, which causes wet cohesive forces that disrupt flow and affect granular mixing processes. This can adversely affect the homogeneous dispersion of the metal precursor in the solid. Some of the open questions are (i) how mixing and flow are affected when the particles have a certain degree of moisture or are saturated with liquid, (ii) whether the liquid is homogeneously distributed into the powder bed and (iii) the extent of dead zones for a given impregnator configuration.
We performed DEM (Discrete Element Method) simulations of the dry impregnation process with different nozzle configurations that distribute an aqueous solution into the catalyst support. A novel water transfer algorithm was implemented in the simulations to account for the actual distribution of water in the particle bed. We computed several parameters, which include the residence time distribution (time that the particles spend under the spraying zone), the extent of the dead zones (zones that do not receive any water during the impregnation process) and the spatial distribution of the sprayed water throughout the powder bed in different cut planes. Our results show that the residence time distribution of the particles in the spraying zone is not significantly affected when we vary the speed of the blender. For four, eight and ten revolutions the particles remain almost a similar amount of time in the spraying zone. The distribution of water is inhomogeneous across different planes parallel to rotation axis reflecting poor radial mixing typical of double cone blenders. When we increase the number of nozzles the extent of dead zones significantly reduces and 70 to 80 percent of particles show a level of saturation that exceeds 50 percent after ten revolutions. For all impregnator configurations the liquid is never distributed completely homogeneously throughout the powder bed. However, if we correctly manipulate the variables and the configuration of the spraying guns we can obtain a nearly homogeneous distribution.