EXPERIMENTAL INSIGHTS AND MACHINE LEARNING PREDICTIONS ON FLEXURAL STRENGTH OF 3D-PRINTED POLYLACTIC ACID

The 3D-printed parts are anisotropic in nature and require a lot of analysis to achieve mechanical excellence. It is achieved with an in-depth knowledge of the choice of major printing process parameters such as nozzle size, layer height, infill density and the infill pattern. In this study, the mechanical behavior of 3D-printed polylactic acid (PLA) was analyzed in regard to flexural strength and an attempt has been made to predict future values with machine learning (ML) algorithm. The flexural analysis has demonstrated that the greater nozzle size and layer height have manifested better strength. However, there are a few exceptions in infill shapes like wiggle, which have performed poorly at higher layer heights because of frequent changes in rasters. The investigation of infill pattern has demonstrated better flexural strength with wiggle at 80% infill, rectilinear at 50% infill, and triangle at 20% infill. Furthermore, the results were critically analyzed with the help of analysis of variance (ANOVA) and morphological studies. According to the ANOVA results, the choice of infill density has the greatest impact on the outcome of the flexural analysis. The nozzle size and layer height also affect the final outcome significantly, as is evident from the F-values of 186.76 and 101.53, respectively. The infill pattern only has an F-value of 6.44, suggesting that flexural strength is unaffected by its changes. However, retrospection in combination with infill density and layer height has increased F-value to 79.54 and 50.24, respectively. Morphological analysis reported layer delamination, noncircularity in filaments, cavities, inter-layer gap, and coalescence. Additionally, an extreme gradient boosting model (XGBM) was developed to predict flexural strength, showing promising performance with a mean absolute error (MAE) of 6.78, a mean absolute percentage error (MAPE) of 14.39%, and an accuracy of 85.6%.