Development of a flow-based predictive model for the coalescence of fused deposition modeling filaments

[ACCESS RESTRICTED TO THE UNIVERSITY OF MISSOURI AT AUTHOR'S REQUEST.] As rapid prototyping processes continue to be developed, there is increasing use of such processes for the production of end-use parts. Fused deposition modeling (FDM) is a particularly favorable method for fabricating end-use parts because of the wide selection of materials available for the process such as Ultem 9085, prized by the aerospace industry for its high strength-to-weight ratio. To confidently employ FDM parts in service requires a thorough understanding of their behavior under expected loading conditions and the ability to predict their success for failure in a particular application. The strength of an FDM part is derived from the amount of bonding that occurs between the polymer filaments as they are deposited. Thus, an accurate prediction of this bond length should lead naturally to an accurate prediction of part strength. Models simulating the heat transfer and coalescence experienced by a pair of adjacent filaments are developed and presented. The models are executed across a range of build parameters to help determine flexibility, and provide a value for the predicted bond length. To validate the models, FDM parts are built from Ultem 9085, cross sectioned, and imaged. The images allow measurements of actual bond lengths to be obtained. The measured bond lengths are compared to the predicted bond lengths. Only a select number of bond lengths measurements are obtained because of variations in microstructure corresponding to various build parameters. A predictive accuracy of 95 % is desired, but the model is unable to achieve it due to estimates of critical data that is unavailable and the variability inherent in the FDM process. However, the simulations provide a significant foundation for future modeling efforts aimed at providing a model capable of predicting bond lengths, and therefore strengths, of FDM parts.