Open Collections

Abstract

Fused Filament Fabrication (FFF) has emerged as a viable manufacturing process for structural components, but part performance is frequently limited by mechanical anisotropy arising from the discrete inter-layer interfaces formed during deposition. Strength at these interfaces is governed by a transient fusion bonding process driven by polymer reptation, wherein chain segments diffuse across the weld plane at temperatures above the glass transition temperature. Predicting this interfacial strength requires calibration of the Arrhenius parameters defining the temperature-dependent welding time in the Degree of Bonding (DoB) model. Existing approaches carry significant limitations: isothermal mechanical methods neglect the transient thermal history of the process, while rheological methods face inherent scaling challenges in translating molecular relaxation times to macroscopic weld performance. This thesis presents an integrated experimental and numerical framework to calibrate a fusion bonding model for amorphous thermoplastics, using Acrylonitrile Butadiene Styrene (ABS) as the model material. Inter-layer shear specimens tested in a Boeing-modified ASTM D695 compression shear fixture isolated the weld response across a DoB range of 0.39 to 0.86, spanning limited interfacial wetting to near cohesive failure, with coefficients of variation below 5%. Shear strengths were normalised against injection-moulded bulk ABS to anchor the DoB to a physically grounded maximum. These data were coupled with interfacial temperature histories from a G-code-driven transient finite element heat transfer model in Abaqus/Standard, and a hybrid Genetic Algorithm and Sequential Quadratic Programming optimisation routine regressed the Arrhenius parameters directly from the macroscopic weld strength data. The regression was evaluated under two cutoff strategies corresponding to the floor and ceiling of the ABS glass transition window: 95°C and 125°C. The 95°C floor model yielded an activation energy of 61.74 kJ/mol and a mean absolute predictive error of 3.94% against independent validation specimens, competitive with the rheological state of the art and substantially outperforming the melt-restricted 125°C ceiling model, which exceeded 10% error. These results confirm that meaningful fusion bonding continues throughout the cooling phase as the interface descends through the glass transition window, a physical finding with direct implications for the certification of rapidly cooled FFF components.