Fouling growth modeling of kraft recovery boiler fume ash deposits with dynamic meshes and a mechanistic sticking approach

• Published in: Fuel (Guildford) Vol. 185; pp. 872 - 885
• Main Authors: García Pérez, Manuel, Vakkilainen, Esa, Hyppänen, Timo
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.12.2016
• Subjects: Ash deposition; Computational fluid dynamics; Dynamic mesh; Fouling; Kraft recovery boilers; Fouling; Ash deposition; Computational fluid dynamics; Kraft recovery boilers; Dynamic mesh
• ISSN: 0016-2361; 1873-7153
• DOI: 10.1016/j.fuel.2016.08.045

•A CFD model for prediction of growing ash deposits with dynamic meshes is presented.•A periodical 2D 4-tube row of a kraft recovery boiler bank is modeled.•The model is unsteady, and includes a particle stick/rebound mechanistic approach.•Two different ash particle sizes are considered: 0.7 and 3.62μm (diameter).•The grid resolution and the constant particle sticking assumption are also studied. The buildup of ash deposits represents a major challenge in the operation of industrial boilers since it entails significant losses in heat transfer performance. Therefore, the research of predictive tools is of great value in boiler operation. A CFD model for ash deposition and fouling growth prediction of industrial furnaces is presented. Considerations are taken regarding grid resolution and accuracy requirements available in literature. A 2D transversally-periodic bundle of four in-line tubes of a kraft recovery boiler bank is modeled with the gas laden with discrete solid fume ash particles. A sticking-rebound submodel determines whether to account the mass of a hitting particle for the deposit growth or to make it rebound, re-entraining it back into the flow. A dynamic mesh modifies periodically the model grid, simulating the expansion of deposits. 0.7μm diameter particles showed round and uniform deposits caused by thermophoresis. The deposit thickness grew up to about 3mm after 100min of fouling. 3.62μm particle showed more irregular distributions with some high local peaks, but lower deposition rates in average. For this particular case modeled, the constant particle sticking probability approach showed no major differences with respect to the particle sticking submodel.