Title : Artisanal Gelato 4.0: Enhancing Quality and Efficiency with Digitalization and Artificial Intelligence
Abstract:
The study aimed to assess whether digitalizing the batch plant to produce Italian-style artisanal gelato could help assess the technological quality of starting blends and control dynamic process conditions. At present, producers use a trial-and-error approach, which leads to varying quality of the finished product. Relationships between blend composition, technological performance, and physical mechanisms during the water crystallization and air emulsification process were understood using sensing, digitalization, and numerical analysis.
METHOD: IoT sensors were utilized to monitor bulk temperature variation, electrical conductivity, and shear stress acceleration throughout the freezing process. The final quality of the ice cream was evaluated based on texture profile and rheological properties under controlled flow conditions. Numerical analysis of cooling curves was conducted to identify the time-temperature domains of water crystallization and compare the effects of different sweeteners. The temperature profile in the cold zone was calibrated to ensure precise analysis.
RESULTS: The study revealed that ice cream blends exhibit a sigmoidal decrease in electrical conductivity during three cooling phases, indicating cooperative mechanisms involved in water crystallization and air incorporation. These mechanisms contribute to the exponential increase in the viscoelastic properties of the ice cream blend's microstructure. Shear stresses increase at a variable rate depending on the kinetics of internal structure formation. Mechanical vibration sensors provide more detailed information than electrical conductivity and temperature signals and can divide the freezing process into four distinct phases based on internal structure evolution. Emulsifiers and stabilizers interact during the third cooling phase to form a consistent viscoelastic network, reducing the size of air cells and retaining emulsified air. Shorter process times result in smaller ice crystals and lower sensory quality, as the quality of ice cream is closely related to its structural consistency.
CONCLUSION: Digitalizing freezing and whipping processes can effectively evaluate the technological quality of starting premix under real processing conditions. Machine and deep learning analysis can be used to create an artificial intelligence platform able to recognize significant freezing and whipping events, to trigger alarms, and to allow decision making on technological variable modulation and extrusion times to obtain tailor-made final quality of the Italian-style ice cream.
Audience Take Away:
- Digitalizing freezing and whipping processes can effectively enable standardization of method to determine the technological quality of starting premix under real processing conditions.
- Machine and deep learning analysis can be used to support decision making activities about technological variable modulation and extrusion times to obtain tailor-made final quality of the ice cream.
