Modelli Semi-Empirici e Numerici per l’Analisi, la Progettazione e l’Ottimizzazione dei Sistemi e Componenti Aeraulici nei Processi di Riciclo dei Rottami Metallici

The steel recycling process relies on aeraulic components, including cyclone separators, Venturi wet scrubbers, and Z classifiers, which use airflow to classify scrap streams and collect dust produced during the shredding procedure. Despite their industrial relevance, the design and operation of these components remain largely guided by empirical knowledge. Most of the available analytical and numerical models in the literature to predict the separation performance and energy losses have been developed under laboratory conditions, which differ substantially from industrial environments. As a result, there is a lack of systematic scientific approaches addressing these components under real industrial conditions. The present thesis addresses this research gap by developing a comprehensive methodology for the analysis, modelling, and optimization of aeraulic components used in metal recycling. Extensive in-plant measurement campaigns were carried out in industrial steel recycling plants, covering both dedusting systems and cascade separation systems. Airflow parameters and particle concentrations were measured, enabling the evaluation of pressure drops and separation efficiency under different operating conditions and treated scrap types. The chemical composition of the collected dust from two types of scrap was further analyzed using various techniques to determine its metallic, polymeric, and oxide content, while optical image analysis was employed to quantify particle sphericity. On the modelling side, a generalized semi-empirical model of the cyclone (ML-B) to predict the separation efficiency curve was developed, overcoming limitations of existing approaches by incorporating a new equation for describing particle turbulent diffusion and accounting for particles’ chemical and shape heterogeneity in real operating conditions of the industrial setups. In parallel, a back-propagation neural network (BPNN) was developed and validated for predicting the separation efficiency curve. Building on these models, a multi-objective optimization of cyclone geometry was performed using a Non-dominated Sorting Genetic Algorithm (NSGA-II) with the objective of minimizing the pressure drop while maintaining a sufficiently high separation efficiency. The obtained optimized geometries have been tested with the developed CFD model of the industrial cyclone, which was used as virtual test rig. For Venturi wet scrubbers and Z classifier, the experimental data of particle concentrations and mass flow rates of scrap were used to test and validate the prediction capabilities of analytical models from the literature. In particular, a CFD model of the Z classifier was developed, which allows to predict the separation efficiency of the component considering the industrial operating conditions. Finally, a general model of the dedusting system was developed, integrating the validated meta-models of the components. The conducted experimental analyses enabled to determine the pressure drop and separation ef- ficiency of the aeraulic components under different operating conditions. From the chemical composition analysis, the analyzed scrap types were found with very different material constitution. The ML-B and BPNN models of the cyclone achieved significantly higher predictive accuracy compared to classic semi-empirical approaches from the literature, improving their forecasting capacity by an average value of 87%. The optimization process of industrial cyclone geometry achieved a decrease in the associated pressure drop, which resulted in a reduction in the energy consumption of industrial fans and potential energy savings of up to 2.7% of the total installed power in the recycling plant. The general model of the dedusting system was validated with the prediction of the particle concentration at the system outlet, achieving errors below 15.2%.Overall, the thesis provides a systematic framework for the analisys of scrap recycling process.