A Fuzzy-Taguchi Approach for Improving Dimensional Accuracy of Fused Deposition Modelling (FDM) Built Parts

Abstract

Fused deposition modeling is one of rapid prototyping system that produces prototypes from plastic materials such as ABS (acrylonitrile-butadiene-styrene) by laying the tracks of semi-molten plastic filament onto a platform in a layer-wise manner from bottom to top. The present work attempts experimental investigations to study influence of important process parameters viz., layer thickness, part orientation, raster angle, air gap and raster width along with their interactions on dimensional accuracy of Fused Deposition Modelling (FDM) processed part. The part produced from FDM machine does not match with dimension of CAD model due to presence of shrinkage. However, shrinkage is more prominent in length and width direction but a positive deviation is observed in thickness direction.
It is essential to study the effect of each parameter on responses such as percentage change in length, width, and thickness of specimen. A design of experiment (DOE) is used to study the effect of process parameters on responses. Optimum parameters setting to minimize percentage change in length, width and thickness of standard test specimen have been found out using Taguchi’s parameter design. Experimental results indicate that optimal factor settings for each performance characteristic are different. There are number of techniques available for predicting responses using input parameters e.g. genetic algorithm, artificial neural network, fuzzy inference system (FIS) etc. But present work uses Fuzzy Inference System (Mamdani Fuzzy logic) to predict the dimensional accuracy in part produced by FDM machine. This method is capable of taking into account the uncertainty and impreciseness in measurements which is commonly encountered in shop floor. The model uses all input and output variables in linguistic terms enabling it convenient for practitioners. The inference engine in Mamdani type FIS uses rules which are obtained with help of design of experiment technique (DOE).