Modeling and optimization of developed cocoa beans extractor parameters using box behnken design and artificial neural network

Breaking the cocoa pod and extracting the beans is a strenuous and laborious task; hence an attempt was made to design and optimize a mechanized continuous cocoa beans extractor. Cocoa fruit was fed manually into a breaker unit through the hopper. The tangential force of the roller pushed the cocoa pod towards the gap resulted in breakage. The strainer attached with pod breaker separate the cocoa kernels from cocoa pod and placenta. The developed machine was tested and the parameters have been optimized by using box-behnken design (BBD) and artificial neural network (ANN). The input variables of bod breaker were roller speed (260, 360, and 460 rpm), cocoa pod size (140, 160, and 180 mm) and inclination angle of strainer (40, 45, and 50 degrees) and out put variables involving capacity, machine efficiency, energy requirement, shelling efficiency, beans separation efficiency, and bean damage percentage.. The coefficient of determination (R-2) of BBD ranged between 0.95 and 0.99, and the sum of squared error (SSE) ranged between 8.38 and 158.3, Whereas the R-2 of ANN ranged between 0.94 and 0.99, and mean squared error (MSE) ranged between 0.0013 and 0.0074. This indicates ANN design superior performance over BBD. The performance parameters of 360 rpm roller speed, 160 mm pod size and 45 degrees inclination angle were determined as optimum conditions to obtain maximum capacity (626.8 kg/h), machine efficiency (96.78%), shelling efficiency (96.3%), beans separation efficiency (86.42%), and low energy requirement (10.54 kJ) with less beans damage percentage (< 1%).