Wet granulation is a key process in industries such as pharmaceuticals, where it improves powder flowability, compressibility, and blend homogeneity. With the transition from batch production to continuous manufacturing, twin-screw wet granulation (TSWG) has emerged as a promising technology due to its flexibility and suitability for continuous processing [1]. However, process design and optimisation still rely heavily on empirical approaches and costly experimental trial-and-error studies [2]. This work presents a multiscale framework for the virtual prototyping of wet granulation processes, combining theory, experiments, discrete element method (DEM) simulations, population balance (PB) modelling, and machine learning approaches. The objective is to enable predictive modelling and optimisation of TSWG processes directly on the computer, reducing the need for extensive experimental prototyping. First, a systematic experimental investigation on a lab-scale twin-screw granulator is presented. Using a full-factorial design of experiments, the influence of specific feed load, liquid-to-solid ratio, and screw configuration on granule size distribution, particle shape, and residence time is analysed. These experiments provide insights into granulation kinetics and establish a validation basis for further modelling efforts. Second, we investigate DEM and PB methods individually to capture particle-scale dynamics and the evolution of particle size distributions. Particular emphasis is placed on applying quadrature-based moment methods. In addition, different calibration strategies for DEM simulations are compared, and machine-learning-based surrogate models are introduced to accelerate parameter calibration using MercuryDPM [3]. Finally, the talk introduces MercuryPBM, an open-source software framework for population balance modelling of granular systems with distributed properties. Future coupling with DEM will enable the prediction of granule properties under varying screw geometries and material formulations, enhancing modelling accuracy for TSWG and beyond. By combination of experiments and advanced modelling techniques, MercuryPBM will potentially enable virtual prototyping and rapid optimisation of twin-screw wet granulation processes and thereby reduce reliance on costly trial-and-error studies.