Thermal modelling of extrusion based additive manufacturing of composite materials

One of the hottest topics regarding manufacturing these years is additive manufacturing (AM). AM is a young branch of manufacturing techniques, which by nature is disruptive due to its completely different manufacturing approach, wherein material is added instead of removed. By adding material layer by layer, mould and customised tooling requirements from the conventional manufacturing are reduced or removed, which leads to increased customisation options and enables new part complexities without increasing the manufacturing cost.

AM hence enables customised small volume productions of composite parts not feasible by conventional manufacturing techniques. This sets up new requirements to the part verification and validation, while conventional destructive tests become too expensive. This initial study aims to investigate alternative options to this destructive testing by increasing process knowledge, and validating the generated toolpaths before the real manufacturing process takes place: Hence removing time consuming and expensive trial-and-error processes for new products.

This study applies a 2D restricted finite volume model aimed to describe thermoplastic Acrylonitrille-butadiene-styrene (ABS) and thermosetting polyurethane (PU) material extrusion processes. During the experimental evaluation of the produced models it is found that some critical material properties needs to be further investigated to increase the precision of the model.

It is however also found that even with only sparse material property information, the simulations show quite accurate temperature simulations when compared to the experimental results. Additionally it is during the thermoplastic experiments seen that the temperature characteristic of the simulations is in in good agreement with the ones obtained from the experiments.

Moreover it is found that the thermosetting experiments show increased reaction rate at higher catalyst concentrations which is in good agreement with the conducted simulation.