Related Experiment Video
Updated: Apr 11, 2026
Planar and Three-Dimensional Printing of Conductive Inks
Published on: December 9, 2011
Multi-objective optimization of printer control parameters for 3D printing of millet dough
Sanket Balasaheb Kokane1, Vinkel Kumar Arora1, Senthilkumar Thangalakshmi1
1Department of Food Engineering, National Institute of Food Technology Entrepreneurship and Management-Kundli, Sonipat, India.
Background:
Three-dimensional (3D) food printing enables precise customization and intricate shapes of food materials. The influence of printer control parameters on the printing performance of millet-based dough is still underexplored.
Objective:
This study investigates the effect of different printer control parameters on the millet dough printing performance, which is evaluated using height ratio, mass flow rate, and bending angle to enhance printing precision.
Methodology:
The already optimized best printable formulation, of 40 g composite flour, 30 g shortening, 22 g jaggery, and 25 g water. The printer control parameters included nozzle diameter (ND) at 1.2, 1.6, and 2 mm; printing speed (PS) at 20, 25, and 30 mm/s; layer height (LH) at 35, 50, and 65% of ND; infill density (ID) at 40%, 60%, and 80%. Response surface methodology (RSM) and Artificial neural networks (ANN) were used for predictive modeling and comparing its statistical measures. Multi-objective optimization was performed through response surface methodology with desirability function (RSMDF) and Artificial neural networks with genetic algorithm (ANNGA). The best-performing printer control parameters were determined by validating the optimized conditions.
Results:
The ID and ND strongly influenced the height ratio. LH and ND significantly affect the mass flow rate. ID and LH were the significant parameters affecting the bending angle. While comparing the statistical measures for predictive modeling, the ANN exhibited lower root-mean-square error (RMSE) values (0.0013 for height ratio, 0.0336 for mass flow rate, and 0.202 for bending angle) and higher coefficient of determination (R2) values (0.97, 0.99, and 0.98, respectively) as compared to RSM. These results indicate that ANN has slightly better prediction capabilities than RSM. Based on the prediction capability performance of multi-objective optimization techniques, the ANNGA performs marginally better in predicting height ratio and mass flow rate with lower prediction errors (0.006 and 0.063, respectively) and higher accuracy (99.993 and 99.936, respectively) than the RSMDF model.
Conclusion:
The optimal condition predicted by ANNGA was as follows: 2 mm of ND, 27.75 mm/s PS, 64.98% LH, and 67.80% ID were obtained for maximum height ratio (5.633), mass flow rate (5.633 g/min), and minimum bending angle (1°). © 2025 Society of Chemical Industry.
