Shrinkage Measurement of Nylon as a Build Material in 3-D Printing By Machine Learning

Abstract

Rapid Prototyping (RP) is a very efficient manufacturing technique which is widely used for improving the design quality of products manufactured. The surface quality and dimensional accuracy should be good for success of a RP process and these two important factors mainly depends upon the selected input variables or process parameters. In this study, the input variables are extrusion speed, layer thickness, extrusion temperature, part bed temperature and the nozzle diameter and an attempt has been made to improve the dimensional accuracy of nylon parts fabricated by 3D printing process. Experiments have been performed with respect to central composite rotatable design (CCRD). Empirical statistical model has been developed for predicting the dimensional accuracy of the fabricated parts in x direction laying. Analysis of variance (ANOVA) has been performed to test the significance of process variables on dimensional accuracy. It has been observed that nozzle diameter, layer thickness, part bed temperature and extrusion speed are most significant factors which affect shrinkage. It has been discovered that with increase in nozzle diameter, shrinkage increases, whereas increase in layer thickness, part bed temperature and extrusion speed decreases shrinkage. Verification of developed model was done by doing experiments at different settings which confirm that forecast of model is precise. Further a predictive model has been developed for forecasting shrinkage using Nylon using machine learning. The competence of machine learning based model is checked and the results show that quadric model generated for shrinkage is significant.