This paper presents the first-ever industrial application of the digital design of a complex, large-scale industrial continuous crystallization network. The aim of the work is to optimize a large-scale continuous crystallization network used for the purification of dextrose monohydrate, by introducing a novel robust model-based digital design framework with the objective to minimize the fine index of the product size distribution, by re-adjusting the key operating parameters such as feed concentration, cooling profile, and recirculation strategy, for the current process as well as for the technically feasible configurations. The system consists of a complex network of over 40 large-scale continuous crystallizers operated in parallel as well as in cascade of multiple stages. The product quality attribute is the fine index, which represents the mass fraction of crystals under 75 mu m. The considered crystallization mechanisms are secondary nucleation and crystal growth, also accounting for growth inhibition by higher polymerization degree sugars. The nucleation and growth rate parameters were re-estimated based on laboratory-scale crystallization experiments. Uncertainty analysis and robust optimization were performed to provide design and operating condition recommendations, and the key outcomes were validated using scaled-down laboratory experiments. The optimizations and the experimental results revealed that, under the investigated conditions, the crystal fine index can be improved more significantly by applying recirculation from the last stage to the first-level crystallizers of the cascade, than by adding more active volume or increase residence time. The results aligned well with the company's process experience, including observations in another plant, and gave enough technical support to initiate the change in current operating conditions of the industrial system.