Data-driven rheological analysis and process optimization of food inks for additive manufacturing

Purpose This study aims to enhance the printability and structural quality of food inks for 3D Food Printing (3DFP), also known as Food Layered Manufacturing (FLM), by optimizing key process parameters and integrating machine learning (ML) for predictive modeling of rheological behavior. Design/methodology/approach Four food inks, chocolate syrup, mashed potato, corn starch and beetroot were systematically investigated. Mashed potato and corn starch were printable in their natural form whereas, chocolate syrup and beetroot juice required the addition of xanthan gum (XG) to become printable. Rheological characterization was performed alongside proximate and color analyses to assess material suitability. Experimental optimization of nozzle diameter, printing speed and extrusion rate was conducted through line and cylinder tests. Printed samples underwent microwave drying, followed by texture analysis to evaluate mechanical properties, and sensory evaluation for consumer acceptance. In addition, rheological data sets were extracted from 50 test graphs per material to construct an ML data set, where extrusion suitability was labeled as “Desired” or “Anomaly.” Seven ML models (Logistic Regression, Decision Tree, Random Forest, Gradient Boosting, Support Vector Machine SVM, K-Nearest Neighbors and Naive Bayes) were trained and evaluated using accuracy, precision, recall, F1 score and AUC metrics. Findings Experimental optimization improved structural fidelity and printability across all food inks. Post-processing enhanced textural strength, while sensory evaluation confirmed consumer acceptance. Among ML models, SVM achieved the highest classification accuracy of up to 98.87%, demonstrating its strong potential for data-driven prediction of printability. The integrated experimental-ML framework presented in this work reduces trial-and-error, supports rational process optimization and demonstrates the potential of ML to automate decision-making in FLM. Originality/value The study establishes a dual-framework methodology combining experimental optimization with ML-driven rheological analysis, reducing manual effort and improving reliability in FLM. This approach advances intelligent automation for the personalized fabrication of nutritious and sustainable food structures.