Characteristics of the churning process in continuous butter manufacture and modelling using an artificial neural network

The characteristics of the churning process in continuous butter manufacture were modelled on the basis of industrial-scale test data sets using an artificial neural network (ANN). The three-layered ANN model accurately predicted the water content of the butter, using as input variables the fat content of the cream, the physical ripening time of the cream, the cream feed temperature, the cream flow rate, and the shear rate on the beater. Further analysis of the churning process indicated that the water content of the butter decreased with an increasing shear rate on the beater, and, after reaching a minimum level, increased linearly. The characteristics of the churning process were affected by the fat content of the cream, the cream flow rate, and in particular the cream feed temperature, which influenced the water content of the butter.