This article is licensed under CC-BY 4.0 pubs.acs.org/IECR Article Population Balance Modeling of Particle Size and Porosity in Fluidized Bed Spray Agglomeration Published as part of Industrial & Engineering Chemistry Research special issue “Marco Mazzotti Festschrift”. Eric Otto,* Aisel Ajalova, Andreas Bück, Evangelos Tsotsas, and Achim Kienle Downloaded via 180.247.62.68 on March 6, 2025 at 02:35:20 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles. Cite This: Ind. Eng. Chem. Res. 2024, 63, 17545−17556 ACCESS Read Online Metrics & More Article Recommendations ABSTRACT: Fluidized bed spray agglomeration is a unit operation applied for the size enlargement of solid granules by aggregation. End-use agglomerate properties crucially depend on agglomerate size and porosity, which, in turn, depend on the operating conditions of the fluidized bed process. In the context of plant automation in an inherently uncertain process environment, the application of process control algorithms is crucial. To facilitate the application of advanced control algorithms relying on accurate yet computationally efficient models, we present a novel population balance model for the evolution of the agglomerate size and porosity. In contrast to other models presented in the literature, the porosity is incorporated by means of a power law relationship with the agglomerate size, which is parametrized by the agglomerate fractal dimension. In order to test the new model, the fractal dimension of agglomerates, produced in a series of batch and continuous experiments, is investigated and correlated to the fluidized bed process conditions. Additionally, based on the experiments, an empirical model for the aggregation kinetics is proposed and kinetic parameters are estimated and also correlated to the process conditions. The proposed model is validated by comparing measured particle size distributions from experiments with model predictions. The results show good agreement; i.e., the relative error of the particle size distributions is below 7% for all validation experiments. ■ INTRODUCTION Agglomeration (also termed aggregation or granulation) processes with a binding agent are widely used in different industries for the formulation of solid particles, e.g., for the production of food powders, pharmaceuticals, detergents, fertilizers, and more.1 During the process, the liquid binder in the form of a solution or suspension is sprayed on the surface of a particle population. When particles collide at wet surface spots, agglomerates may be formed, initially bound by liquid bridges, which subsequently solidify into solid bridges upon drying (see Figure 1a). When carried out in a fluidized bed, the particles are fluidized by an upward-facing stream of air, resulting in well-mixed behavior with high energy- and mass-transfer rates. The process is realized in the fluidization chamber, either in batch mode or continuously with particle feed and withdrawal as presented schematically in Figure 1b. Although the process is primarily used for size enlargement, various other agglomerate properties are of interest. Besides the particle size distribution, the agglomerate morphology, i.e., the internal structure, is relevant due to its influence on properties such as dispersibility, solubility, bulk density, and more.2−5 In general, the economic value of the product agglomerates is determined by these end-use properties. To produce agglomerates with predefined properties under inherently uncertain process conditions, process control is a crucial tool. Furthermore, from an industrial point of view, it is © 2024 The Authors. Published by American Chemical Society mandatory to optimize the production process with respect to product quality, throughput, and energy considerations such as heat consumption for the heating of the fluidization medium. Both process control6−11 and optimization,12,13 can be enhanced by accurate yet computationally efficient dynamic models, capturing the evolution of the desired properties depending on the operating conditions. In this contribution, we present a population balance model for the fluidized bed spray agglomeration process that considers particle size represented by volume or diameter, respectively, and morphology represented by agglomerate porosity. The literature provides some population balance models for a fluidized bed and more general agglomeration processes that include both properties. Iveson14 presents a comprehensive, four-dimensional population balance equation (PBE) for wet granulation processes with granules consisting of two different solid fractions and a binder. The granule solid phase mass, binder to solid ratio, granule porosity, and solid Received: May 10, 2024 Revised: August 15, 2024 Accepted: September 18, 2024 Published: October 3, 2024 17545 https://doi.org/10.1021/acs.iecr.4c01660 Ind. Eng. Chem. Res. 2024, 63, 17545−17556 Industrial & Engineering Chemistry Research pubs.acs.org/IECR Article Figure 1. (a) Stages of agglomeration: (1) primary particles, (2) wetted primary particles, (3) wet agglomerate after primary particle collision, and (4) dried agglomerate. (b) Fluidized bed process including spraying, particle feed, and withdrawal. phase mass fraction of the first component are the internal coordinates. Verkoeijen et al.,15 Poon et al.,16 and Chaudhury et al.,17 referring specifically to fluidized bed granulation, restrict themselves to agglomerates with only one solid component and use a PBE with the solid volume (vs), liquid volume (vl), and air volume (va) as independent variables describing an agglomerate. The porosity is then given by the ratio of air plus liquid to total volume. Lastly, Barrera Jiménez et al.18 formulate a PBE with agglomerate volume and porosity (i.e., the ratio between pore and total volume) for a twin-screw wet granulation process. In the models of Verkoeijen et al.,15 Poon et al.,16 and Chaudhury et al.,17 it is assumed that the respective volumes of solid, liquid, and air components are conserved during aggregation. With this assumption, the porosity of newly formed agglomerates is uniquely determined by the compositions of the parent agglomerates. Although using a different formulation of the PBE, the same assumption is made by Iveson.14 Barrera Jiménez et al.18 neglect any liquid components and assume that the porosity of a newly formed agglomerate is the weighted average of the parent agglomerates’ porosities. Speaking in terms of solid and air volumes, this is equivalent to the volume-conserving approaches described above. All of these contributions have in common that, besides agglomeration, a so-called consolidation term is considered in the population balance equation, which accounts for changes in porosity independent of any aggregation (and breakage) events. Independently, from population balance modeling, the porosity of aggregates in a population is often described by means of a power law relationship, e.g., in Rosner and Tandon,19 Logan and Kilps,20 Sorensen,21 and Singh and Tsotsas,22 which is parametrized by the agglomerate fractal dimension and in some cases a so-called prefactor. From this relationship, as will be shown in the upcoming section, it follows that, at least in batch and continuous steady-state processes, the air (or pore) volume increases during a binary agglomeration event; i.e., the aggregates become more porous the larger their solid volume becomes. This nonconservation of pore volume, which is supported by the experimentally investigated agglomerates in this contribution and in Singh and Tsotsas,22 contradicts the assumptions made in the models presented above. In order to incorporate this observed behavior into a model that can be used for process control, we present a novel extension of the PBE, with the goal of more accurately representing the morphogenisis of agglomerates in fluidized beds. In contrast to the previously reported models, we choose a one-dimensional particle size distribution (PSD) with solid volume as the internal coordinate, which is possible since the porosity and solid volume are not independent particle properties due to their power law relationship. Since the solid volume is conserved during aggregation, the traditional birth and death terms are used in the PBE,23,24 where the agglomeration kinetics are described by the so-called agglomeration kernel. The functional relationship between solid volume and porosity, obtained from the above-mentioned power law, is used to transform the PSD with solid volume as an internal coordinate from the population balance equation to a PSD with total volume as an internal coordinate, which is the particle size distribution that can be measured online. For the model application in a process control context, the influence of the process conditions has to be modeled. In the PBE model presented here, the agglomeration kinetics on the one hand and the agglomerate fractal dimension on the other hand depend on the gas inlet temperature and the binder concentration. For both dependencies, empirical correlations are proposed. To validate these correlations and the overall PBE model, three batch and five continuous fluidized bed spray agglomeration (FBSA) experiments have been conducted, and the model predictions of the particle size distributions are compared to measurements. Finally, a modification of the model, accounting for timevarying process conditions, is proposed. Initially, the dependence between porosity and solid volume, parametrized by the fractal dimension, is given for batch and continuous processes at steady state and therefore only for constant process conditions. Thus, for the description of the transient phase in continuous processes as well as generally continuous processes with changing process conditions, we propose a 17546 https://doi.org/10.1021/acs.iecr.4c01660 Ind. Eng. Chem. Res. 2024, 63, 17545−17556