Digital Twin-Driven Optimization of Pilot-Scale Polyurethane Aerogel Production Using SVR Modelling

The growing demand for sustainable and energy-efficient materials has positioned aerogels as promising candidates for advanced insulation applications. Among them, polyurethane (PU) aerogels are attracting increasing interest due to their thermal insulation properties and mechanical versatility. However, their development commonly relies on trial-and-error experimentation, which is time-consuming and resource-intensive. This study presents a Digital Twin (DT) framework to support PU aerogel design and reduce the experimental workload. A pilot-scale DT was developed using data from 21 synthesis experiments, including process configuration, parameter mapping, model development, and process analysis. Two predictive models were evaluated, with the Support Vector Regression (SVR) model showing good agreement with the experimental data (R² = 0.964) and being selected to estimate aerogel density within the parameter range studied. The DT framework enabled the identification of synthesis conditions associated with lower density, which may contribute to improved thermal insulation performance. These results illustrate the potential of DT-assisted modelling to support material development, improve process understanding, and guide more efficient experimentation in PU aerogel synthesis. Overall, this work highlights a data-driven approach for advancing sustainable and scalable aerogel manufacturing.