A framework for optimizing process parameters in fused deposition modeling using predictive modeling coupled response surface methodology

Fused deposition modeling (FDM) is a widely used additive manufacturing (AM) process due to its simplicity, cost-effectiveness, and low waste generation. The mechanical properties of FDM products are greatly influenced by the process parameters. Thus, it is important to use optimal process parameters to achieve desired mechanical properties. One way to determine the optimal process parameters is using response surface methodology (RSM) which has been widely implemented for determining optimal manufacturing process parameters. For RSM, experimental investigations are performed in a systematic way using design of experiments (DOE). However, in some cases, experimental data is not collected in a systematic manner using DOE. Additionally, it is expensive and time consuming to conduct experimental investigations of a well-designed DOE to acquire data. This results in generating insufficient experimental data for developing response surface models using RSM. To address these challenges, this study presents a streamlined five-step framework to optimize FDM process parameters for achieving desired mechanical properties of products. In this process, once experimental data is collected, it is analyzed and preprocessed to remove or repair any missing data. Next, this data is used to develop an artificial neural network (ANN) predictive model to predict the systematic DOE data necessary for developing response surface models. Finally, these models are used to optimize the process parameters for maximizing mechanical properties of FDM products. The efficacy of this framework is demonstrated in detail for a real-world FDM application with limited experimental data. This framework determines the four optimal process parameters for simultaneously maximizing three mechanical properties. By utilizing ANN and genetic algorithm-based optimization, this approach minimizes the need for extensive experiments. Thus, unlike any other work in the literature, this framework not only determines optimal FDM process parameters which results in achieving the desired mechanical properties, but also significantly reduces the time and resources required in the process, thereby paving the way for a more efficient manufacturing process. The optimized results obtained using this framework are found to be very close to experimental data, thereby establishing that the framework is effective for determining optimal parameters in case of limited or non-systematic DOE data. In future, the generic nature of the framework can be utilized to include other FDM process parameters, material characteristics affecting the properties, and different predictive modeling and optimization techniques. Finally, such a framework can be modified as necessary for utilizing it for other AM and traditional manufacturing processes.