Evaluating convective heat transfer coefficients in fused deposition process using infrared imaging and neural networks

During the fused deposition modeling (FDM) process, heat is exchanged by convection with the surrounding air and by conduction with the platform. Therefore, modeling the thermal behavior of the FDM process requires accurate convective heat transfer coefficient (CHTC) and interfacial conduct resistance (ICR) data. The traditional approach to solving this problem is iterative in nature and requires considerable computation time. In this work, an artificial neural network model was designed and trained using a total of 100 sets of data from a finite element model that reproduces the geometry of the manufactured part. A backpropagation neural network was developed, in which thermal profile characteristics of two specific points were used as input variables while the corresponding CHTC and ICR were selected as the output variables. A correlation coefficient of 0.999 was achieved during training. After training the model, the response of the thermally excited surface was monitored and recorded using an infrared camera and characteristics of the experimental condition were used as inputs to evaluate CHTC and ICR. It was estimated that HTC and ICR equal 61.7 and 2894 W/(m²K) in our experimental model, respectively. The developed models could offer significant advantages when heat transfer coefficients need to be estimated repeatedly.