Process Optimization Under Uncertainty for Improving the Bond Quality of Polymer Filaments in Fused Filament Fabrication

This paper develops a computational framework to optimize the process parameters such that the bond quality between extruded acrylonitrile butadiene styrene (ABS) filaments is maximized in fused filament fabrication. Application of additive manufacturing has been limited due to the variability in the product quality. Thus, improving the product quality, and reducing the variability in the manufactured parts are crucial. Porosity is one of the key metrics of additively manufactured parts, and poor bond quality between adjacent filaments will cause increased porosity. The common practice to optimize the quantity of interest in additive manufacturing is to use the traditional trial and error approach to achieve the desired microstructure and properties of the manufactured parts. However, this approach of optimizing process parameters such as printer nozzle temperature and printer nozzle speed based on physical experiments is expensive and time-consuming. To address these issues, the trial and error approach is replaced in this paper with a model-based approach for predicting the variability of the additively manufactured product; thus reducing the number of trial and error experiments. An analytical solution for the transient heat transfer analysis during the printing process in fused filament fabrication is coupled with a sintering neck growth model to assess the bond quality that occurs at the interfaces between adjacent filaments. Predicting the variability in the fused filament fabrication process is essential for achieving the desired quality of the manufactured part; however, the models used to predict the variability are affected by assumptions and approximations. Therefore, this work systematically quantifies the uncertainty in the bond quality model prediction due to various sources of uncertainty, both aleatory (natural variability) and epistemic (lack of knowledge), and includes the uncertainty in the process parameter optimization. The model parameters of the neck growth model, the temperature-dependent material viscosity and surface tension, and the material properties are considered epistemic uncertainty sources. The natural variability in the printer nozzle temperature is identified. Variance-based sensitivity analysis based on Sobol’ indices is used to quantify the relative contributions of the different uncertainty sources to the uncertainty in the bond quality between filaments. A Gaussian process surrogate model is constructed to quantify and include the model error within the optimization. The surrogate model is used to estimate the model discrepancy for given process parameter values and correct the physics model predictions. Then, the corrected prediction model is used in the optimization framework to select the optimal process parameters to maximize the bond quality at each layer. Physical experiments are conducted to verify that the proposed formulation for process parameter optimization under uncertainty results in high bond quality between adjoining ABS filaments.