Agglomeration Mechanisms during Fluidized Bed Combustion of Biomass

Abstract

Biomass, as a carbon neutral and renewable fuel, has a potential to partially replace coal for heat and power production. Fluidized bed combustion is a promising technology for utilization of biomass due to its fuel flexibility and high efficiency. However, bed agglomeration is one of the main operating problems in fluidized bed combustion. Formation of agglomerates may change the hydrodynamics and influence the stability of fluidized bed boilers. The severe agglomeration may cause defluidization, leading to an unscheduled shut down of a power plant. Therefore, understanding the mechanisms of agglomeration and developing the effective countermeasures are of essential importance. This PhD project aims to achieve and improved understanding of the agglomeration mechanisms, as well as to evaluate effective measures for reducing agglomeration problems. The agglomeration phenomena are investigated by: 1) evaluating the interactions between model potassium compounds and silica sand in a lab-scale fluidized bed reactor; 2) examining the reaction mechanisms between K2CO3 and silica sand; 3) performing combustion experiments of biomass and K-compounds doped biomass under different temperature and air staging conditions in the lab-scale fluidized bed reactor. For reducing the agglomeration problems, various solid additives and different dosing methods are evaluated. In addition, the segregation phenomena of bed material caused by formation of agglomerates are simulated by Computational Fluid Dynamics (CFD) modelling. The studies of agglomeration mechanisms by using potassium model compounds, such as KCl, K2SO4, K2CO3 and K2Si4O9, and their mixtures indicate two interaction mechanisms between model K-compounds and silica sand particles: without reaction and with reaction occurring. When KCl and a eutectic mixture of KCl/K2SO4 are applied, defluidization occurs near KCl melting temperature and KCl/K2SO4 eutectic temperature, respectively. Such defluidization is caused by the formation of liquid phase in form of low viscous melting salts, with limited reaction between K-species and silica sand. The minimal amount of salt required for defluidization occurring depends on the hydrodynamics, particularly the Ug/Umf ratio. However, the defluidization temperature appears to be independent of the amount of K-salts in the bed for this type of agglomeration. On contrary, the addition of K2CO3 results in defluidization well below its melting point, and it is induced by reaction between K2CO3 and silica sand. The defluidization temperature decreases with increasing amount of K2CO3 loaded, by using coarser sand particles, and by reduced gas velocity. Finally, unique behavior is observed when K2Si4O9 is added, since defluidization is not observed up to 850 °C, even for a very high amount of K2Si4O9 added to bed. However, large number of agglomerates is observed because high-viscous K2Si4O9 particles act as nuclei for binding several sand particles, and the interaction is localized at contact point between particles. The reaction mechanism between K2CO3 and silica sand is further studied by thermogravimetric analysis (TGA). It is showed that the reaction occurs in a solid-solid phase already at temperatures around 700 °C. The reaction rate increases with increasing temperature, but decreases in presence of CO2. The reaction is enhanced by using smaller particle size and by mixing of solid reactants. It is observed that the reaction is initiated at the contact points between K2CO3 and silica sand, forming a thin product layer. The layer acts as a reactive media further reacting with K2CO3 and silica sand. For different type of biomass, the appearance of potassium may vary. In order to investigate the transformation of K-compounds and their interaction with bed material during biomass combustion, KCl, K2CO3, KCl/K2CO3 and KOH doped pine wood and washed straw are continuously burned in lab-scale fluidized bed reactor. It is observed that KCl and KOH might be transferred to K2CO3 during combustion process, which further reacts with silica sand and causes defluidization. The transformation of KCl to K2CO3 depends on fuel properties, for example, the transformation is more pronounced for pine wood compared to washed straw. Continuous combustion experiments of wheat straw and sunflower husk are also conducted in order to investigate the influence of fuel type with an emphasis of the impact of stage combustion (widely used for NOx emission reduction) on agglomeration tendencies. It is observed that firing the biomass with high content of both, K and Si, results in a short combustion time before defluidization. On the other hand, firing the biomass containing Ca, Mg, S, and Cl, appears to reduce agglomeration and defluidization tendencies. The agglomeration tendency is increasing for lower ratio between primary air and total air (λ1/λ), and the trend is particularly pronounced for sunflower husk. However, no defluidization is observed when pyrolysis condition is applied in the dense bed, i.e. λ1/λ ratio equal to 0, probably due to the fact that the presence of char hinders the interaction between the biomass ash and the silica sand particles. A fast defluidization occurs during char combustion at very low oxygen level. Different additives, as well as dosing methods are examined as a countermeasure to reduce the agglomeration tendency during combustion of wheat straw and sunflower husk. The additives can either react with potassium present in biomass ash and reduce the amount of potassium reacting with silica sand, or they can increase the viscosity of formed melts, thus reducing their stickiness and agglomeration tendency. The results show that kaolin is the most efficient additive, followed by magnesium carbonate, lime, ammonium sulphate, coal fly ash, and clay. It is also showed that mixing of the additives with fuels is more effective than direct feeding of additives to the bed. Formed agglomerates may segregate from the bed material particles, which accelerates further agglomeration. Computational Fluid Dynamics (CFD) modeling is applied to investigate the segregation process due to the agglomerates formation. Segregation data of the binary particle system in literature are used for the CFD simulations. It was found that the EMMS drag model coupled with Ma-Ahmadi radial distribution and solid pressure models predicted more reasonable axial distribution of solid phases compared to commonly used Syamlal O´Brian drag model coupled with Lun et al. radial distribution and solid pressure models. An increase in the solid-solid drag further improved the simulation results. Developed and optimized simulation scheme can be further used for simulation of mixing/segregation behaviour of sand/agglomerates mixtures in fluidized bed reactors.