Page 25 - MSAM-2-1

Materials Science in Additive Manufacturing Fast fiber orientation optimization

computer-aided design model of the part in.stl or.obj manufactured on a Markforged X7 printer, without the need
format. Optimal printing can be achieved by adjusting for time-consuming topology optimization algorithms or
parameters, such as layer height, printing temperature, metamodels.
and bed temperature (if equipped). The optimal values of
these printed parameters depend on the printed material, 2. Principal stress-based method
nozzle diameter, and ambient air temperature. MEX is Since printed composite parts are built layer-by-layer,
attractive due to its ease of use, compatibility with various resulting in a laminate composite, fibers are not oriented in
thermoplastics, low cost (for both printer and materials), a 3D space but in a stacking way. Intuitively, a 2D method
and low energy consumption. When combined with applied to each layer of a part seems to be well adapted.
continuous fiber-reinforced filaments, the MEX process However, this could result in a huge simulation cost, as
leverages the benefits of both additive manufacturing and MEX layers are thin (close to a tenth millimeter). For
long fiber composites, resulting in complex freeform parts example, optimization of a ten-centimeter-high part would
with stiffness comparable to aluminum alloys, low density, result in a thousand optimization processes. Furthermore,
and corrosion resistance . commercial machines only allow one angle per layer for
[12]
However, additive manufacturing of long fiber oriented fibers. To solve these two issues, we will not
composites results in highly anisotropic materials. Hence, consider a 2D layer-based method, but a 3D method based
aligning the fibers with the mechanical strain is crucial on a stack division of the part, as explained later.
to obtain the best stiffness and strength in a printed 2.1. 3D model based on a stack division of the part
part [13-15] . Therefore, non-optimized fiber paths can lead to
easy breakage. While previous works have explored fiber To meet the time computation needs of designers, we
optimization, they have limitations. Zhou et al. proposed introduce the concept of a stack, which is a group of several
[16]
a 2D model that divided the part into areas, each with a layers. By working with stacks instead of individual layers,
specific fiber orientation. Ding et al. designed curved we can decrease the mesh size and significantly reduce
[17]
fiber routes using a 2D model. Li et al. combined the two computation time. For example, a ratio of ten between a
[18]
approaches with areas divided into concentric curved fiber layer-based mesh and a stack-based mesh results in a 3D
routes. Safonov and Nomura et al. developed curved mesh that is 1000 times lighter (Figure 1). Working with
[19]
[20]
fiber routing in both 2D and 3D models, and Jung et al. stacks of ten layers provides a good balance between efficient
[21]
created a complex model that considers fiber orientation computation and accurate results from the finite element
and diameter. All these methods are stress-based and analysis, as values greater than ten result in significant loss
consider fiber to be most effective when its direction aligns of precision. Hence, based on this simplification concept,
with the major principal direction. However, they are not an optimization method is proposed to optimize the fiber
easily applied to commercial printers like Markforged or angles for each stack of layers as an optimal configuration
Anisoprint, and while they can be done with the open- to improve the structure’s stiffness and strength.
source slicing software Aura, they require manual coding of Besides an important reduction of the simulation
fiber routes in a gcode file. In addition, slicers like Eiger or cost, using a stack-based 3D model also allows the user
Aura work layer-by-layer, a constraint of the MEX process to allocate different angles on the layers of a stack while a
that makes it impossible to use 3D optimization models and layer-based model allows only one angle per layer. Thus, if
makes it challenging to find an easy and quick optimization a layer-based model is considered, each layer is reinforced
model for commercial printers. To solve the problem above, with only the dominant stress orientation (Figure 2B left).
the best fiber orientation is determined using a standard However, some mechanical loads may result in a complex
stress flow method based on principal stress analysis and stress flow, with several strained directions (Figure 2A). As
corresponding direction. Keeping ease of use for commercial additively manufactured composites have low stiffness and
printers in mind, the concept of a stack is introduced: strength when the load is not aligned with the fibers, it is
A set of layers with optimized orientation angles for each important to consider all the possible strained orientations,
layer, weighted by the dominant principal stress. A cost- as our model allows (Figure 2B right). With our approach,
optimization approach using a multi-layered finite element the computed percentages represented by each orientation
model and the previously computed best orientation angles (20%, 30%, and 50%) are distributed in ten different layers,
is also proposed. This allowed the identification of a reduced so each area is reinforced on at least one layer.
number of layers, where reinforcement is necessary, while
using nylon in other layers. The methodology, implemented 2.2. Optimization process
in Ansys Parametric Design Language, was efficient and This section presents our optimization process with the
demonstrated by a 18% increase in stiffness of wrenches different substeps included in the workflow (Figure 3). It is

Volume 2 Issue 1 (2023) 2 https://doi.org/10.36922/msam.49