Predicting filament material extrusion energy consumption: a comparative study

Life-cycle assessment and other sustainability evaluation methods rely on energy flow data, but such data are not always readily available for products made using material extrusion of polymer filament (fused filament fabrication). While some predictive models for energy consumption have been proposed, they have not yet been rigorously evaluated or compared. In this paper, we combine new experimental data and existing data from literature to form a comprehensive database with 374 unique energy-consumption measurements with data describing printers, print parameters, and material type. We refit several existing models using this expanded dataset and present a new semi-empirical model for energy prediction that is computationally inexpensive. In addition, we evaluate the performance of these energy-prediction models, with the mean absolute error based on the estimated build time falling between 0.15 and 0.34 MJ. The analysis showed the importance of build time in energy-consumption prediction and identified filament material type and printed volume as two variables that are not currently captured well by existing models. This work helps to inform decision making for sustainability during the engineering-design process by improving our ability to predict energy consumption.