Industrial operation of particulate processes have historically been reliant on accumulated corporate knowledge that is assimilated by trial-and-error through decades of process operation. However, this accumulated knowledge is often poor at dealing with new issues which may arise due to new product formulations or new process conditions as the underlying science is not well established or misunderstood. Issues such as overfilling of packaging due to poor control of the bulk density and product which does not meet specification are recurring issues which the process engineer has to regularly tackle. As a response, food, pharmaceutical, detergent and, relevant to this thesis, the agrochemical industries have emphasised increased investments into developing understanding of the core scientific fundamentals of their processes. A convenient way to capture the fundamentals is to deploy a model-based approach.
Through this thesis, a coupled one-dimensional population balance model is proposed and implemented. The model framework considers the granule in four parts, the solid composition, the solvent content, gas content and solid within the solvent phase. With these internal co-ordinates, a novel wetting model is developed and validated which builds upon the work of \cite{kariuki2013distribution} and integrates the particle coating number into the continuous form of the population balance framework. Further to this, a novel agglomeration model is implemented which builds upon the wetting model. The model considers agglomeration-likelihood as a function of the surface wetting in addition to the collision behaviour via the Stokes criterion. The model demonstrates that the collision and wetting behaviour both play a key role to growth and integrates them within a single framework for the first time.
The model is extended into a two-compartment framework. One compartment considers spray drying and the other compartment considers the bulk particle bed. The bulk bed compartment integrates the population balance framework, where the spray drying compartment is linked by a mass flow which feeds the population balance model with nuclei generated by spray drying. A sensitivity study is performed to characterise model behaviour and the role of drying in the overall granulation mechanisms.
This model framework is finally validated against experimental data performed on a pilot-scale continuous fluidised bed granulator. The process parameters which were varied are the spray rate, inlet air temperature, inlet fluidisation air flow rate and a combination of these. Lab-scale experiments were performed to feed the model with key formulation properties; the single particle drying curve and the sorption isotherm. Key model parameters such as the agglomeration rate constant, spray bypass ratio, spray droplet diameter and standard deviation were estimated using a single model experiment. The model then predicted successfully changes in all process parameters, which is a powerful validation of the model. This highlights that we can capture granulation phenomena such as the dependence on wetting and drying within particle growth, while not being reliant on computationally intensive methods such as computational fluid dynamics and discrete element modelling. This work provides a framework which may be attractive for industrial integration due to the low computational intensity whilst still capturing the complex physics of granulation.
