Work

Predictive modeling for the optimization of specialty fats in margarine using computational tools

Year: 2026
Type: dissertation
Abstract: The present work proposed the application of computational tools and statistical modeling for the prediction of fat parameters with a focus on margarine formulation. For this purpose, a second-degree Simplex-Lattice design was used in the formulation of 14 lipid blends (F1–F14) from soybean oil, palm oil, palm stearin, and palm kernel oil, ranging from pure raw materials to binary and ternary mixtures. The correlation heatmap revealed a strong and positive association between the MPT and solid contents at 35°C and 40°C (r close to 1.0), and principal component analysis (PCA) showed that the first two components explained 50.8% and 35.1% of the total variance, respectively (approximately 86% cumulative), reducing the system to two main axes related to hardness/saturated long-chain fatty acids and lauric/polyunsaturated content. Based on the desirability function, formulation F11 (33% soybean oil, 33% palm oil, and 33% palm stearin) was identified as the optimal mixture, with a desirability of approximately 83%. This formulation was used in the production of a pilot-scale margarine with 80z% lipid phase, which presented a fat content of 81.87%, moisture of 16.20%, MPT of 47.20°C, and SFC of 41.70%, 9.60%, and 7.00% at 10°C, 35°C, and 40°C, respectively — values close to those of the pure blend (MPT of 46.30°C; SFC of 40.60%, 8.40%, and 6.70%), attributed to the margarine plasticization process. The fatty acid profile of the margarine remained predominantly composed of palmitic acid (37.03%), oleic acid (29.21%), and linoleic acid (22.22%), with a residual trans fatty acid content of only 0.62%. Experimental results consolidate the database as an application of statistical modeling based on chromatographic profile, solid curve, and melting point. The integration of experimental design with statistical models shows potential to significantly reduce the reliance on extensive experimental testing, contributing to innovation and sustainability in the food industry and to the development of stable and functional margarines, aligned with market demands for zero trans and nutritionally adequate alternatives.
Language: Portuguese
Subfield: Food Science
Open Access status: green
OpenAlex ID: W7202015642
Cites: 60
Cited by:
Related to: 10