Exergetic sustainability assessment of a semi-industrial convective dryer employing waste heat recovery for drying wood chips: A BPANN-based approach

Abstract

This study investigates the exergy-based performance and sustainability of a semi-industrial convective dryer equipped with a waste heat recovery unit for drying poplar wood chips. Four key exergy-related indicators, namely exergetic improvement potential (EIP), exergetic sustainability index (ESI), universal exergetic efficiency (UEE), and overall exergetic efficiency (OEE) were examined and then predicted using a feedforward backpropagation multilayer perceptron neural network (FFBP-MLPNN) with the Levenberg-Marquardt (LM) learning algorithm and a single hidden layer. The network evaluated different numbers of neurons in the hidden layer and utilized tansig and purelin activation functions in the hidden and output layers, respectively. Experimental trials demonstrated that increasing the air recirculation ratio enhances the ESI due to improved heat recovery and reduced exergy losses, while it reduces the EIP, indicating lower thermodynamic inefficiencies and less potential for further improvement. In contrast, lower recirculation ratios yielded lower ESI values and higher EIP, highlighting greater exergy destruction and larger optimization potential. Additionally, increased air temperatures and flow rates improved both indices. The results indicated that the neural network can predict all four outcomes with R2 > 0.97. Additionally, 0.013601 (using a 4–23–1 topology), 6.4137 × 10−6 (4–15–1 topology), 3.186 × 10−6 (4–30–1 topology), and 0.036108 (4–32–1 topology) were the mean squared error (MSE) values for predicting the EIP, ESI, UEE, and OEE, respectively. Hence, this study suggests that the ANNs approach could be an effective tool for analyzing thermal sustainability indicators in industrial convective drying processes.