Spouted bed drying of milk–blackberry mixture: Material thermal properties and process energy aspects

Abstract

This work presents the determination of the thermal properties of a milk–blackberry pulp mixture (25–75%, vol/vol), and of the powder obtained by spouted bed drying. Based on the experimental data, polynomial and artificial neural networks (ANNs) correlations were proposed to predict the specific heat as a function of temperature. The effects of operating variables (air temperature, atomization pressure, and feed flow rate), on the energy efficiency (EE) and on the specific consumption of energy (ε) were evaluated using a 23‐factorial design. Afterward, the effects of feeding mode (atomization or dripping), on energy aspects were analyzed. The fusion of powder fat crystals occurred at 1°C and at 16°C, glass transition between 27°C and 34°C, and proteins denaturation starting at 87°C. The ANNs presented higher accuracy in the specific heat prediction, with mean deviation as low as 0.5%. The inlet air temperature and the feed flow rate have affected the energy parameters, but the atomization pressure and the interaction among these variables did not. The EE ranged from 7.3 to 23.2% and the ε from 10.6 to 33.8 MJ/kg. Besides the reuse of the outlet air, the drip‐feeding mode could be an alternative to decrease the net energy consumption of the drying process. Practical Applications This work presents the determination of melting and glass transition temperatures, melting enthalpy, and specific heat of the paste milk–blackberry pulp (25%:75%, vol/vol) and/or powder, which are important characteristics for equipment dimensioning, process control, product quality, and to determine storage conditions. We also proposed ANN‐based correlations to relate the specific heat of the paste and powder as a function of temperature. The ANNs equations showed high accuracy and were used in the analysis of the energy aspects of the spouted bed drying process. Statistical analysis of energy aspects as a function of operating variables was performed and two linear response models were obtained, allowing to predict the specific consumption of energy and energy efficiency of the process. Herein, we suggest the potential use of DSC analysis and ANN techniques as a powerful combination to obtain information about food products and process design.