ADDITIVE MANUFACTURING SIMULATION SOLUTION AS ENABLER FOR CONFIDENT LIGHTWEIGHT AUTOMOTIVE DESIGN

Transcription

1 ADDITIVE MANUFACTURING SIMULATION SOLUTION AS ENABLER FOR CONFIDENT LIGHTWEIGHT AUTOMOTIVE DESIGN Laurent Adam, Roger Assaker, Philippe Hébert, Olivier Lietaer, Sylvain Mathieu e-xstream engineering Abstract Additive manufacturing of polymers is transitioning from rapid prototyping to a true industrial production technique. While it brings valuable opportunities to the automotive industry, such as drastically facilitating the spare part supply chain or allowing multi-material and multi-functional designs, it also comes with a series of challenges for the engineers. The reliability of the mechanical properties of the final part still has some uncertainty and is not fully supported by standard engineering tools. Dimensional accuracy is not always met and cannot be predicted prior to printing. To support this transition, the engineering workflow which is daily applied for traditional manufacturing processes needs to be replicated and adapted to the additive manufacturing. Printer manufacturers, material suppliers and end-users need predictive simulation tools to bring the additive manufacturing efficiency and performance to the next level required by the industry. This paper presents a holistic simulation approach for additive manufacturing of plastics and composites, covering material engineering, process simulation and structural engineering of both SLS and FDM. Introduction Additive manufacturing (AM), by definition opposed to subtractive methods, regroups a number of manufacturing processes allowing to create parts from 3D numerical CAD model by building up the components in layers by depositing material, hence not requiring specific mold tooling design. Low volume, complex parts can be produced at fixed cost in less time. As lightweighting becomes a top design priority for the automotive and aerospace markets, the capability to reduce the number of parts by manufacturing an assembly directly is a promising source of gain. Lattice structures are also increasingly interesting the industry seeking the optimum mix of mechanical performance and low density. Such design opportunities offered by AM technologies can successively lead to weight reduction of several tens of percent. Other advantages of the AM technology include: More integrated/functional parts; Product lifecycle: reduced costs and time to market; Onsite - On demand production / Customization; o End-of-life products; o Medical & dental; o Aerospace; Reduced material usage (lightweight design or less manufacturing waste); Reduced energy usage (no need to remove material); No need for tooling; Page 1

2 Agile & Lean development. Until recently, the application of 3D printing was limited to rapid visual prototyping. Additive manufacturing of plastics is experiencing a paradigm shift as the industry is now looking into the technology as a full production technique to achieve not only a tailored design but also, new lightweighting solutions which are not viable with other manufacturing methods. While the advantages of AM technology are appealing for the industry, additive manufacturing is still a recent technology and requires multiple trials and errors to reach the right consistency and expected performance. The printed part may show important warpage due to thermally induced distortion, thus dimensional tolerance may not be ensured. Process control and optimization is also a matter of concern: a large number of parameters drive the manufacturing and the part-to-part variability is high. New high performance materials need to be developed specifically for the AM industry. Simulation tools are also needed to connect the structural analysis with the manufacturing process by predicting the part performance as a function of the printing setup. To unveil the full potential offered by AM technology, predictive simulation tools are required by process engineers and end-users. This paper will address the current research and developments of e-xstream engineering with the goal to provide the industry with a holistic simulation approach for additive manufacturing of plastics and composites, covering process simulation, material engineering and structural engineering of both SLS and FDM. Process simulation An increasing number of AM technologies are applicable to plastics and reinforced plastics. The most widely used in the industry today are the Selective Laser Sintering (SLS) process and Fused Deposition Modeling (FDM), also referred to as Fused Filament Fabrication (FFF). In SLS, powdered material is sintered using a laser heat source. In FFF, plastic filaments are melted, extruded, and deposited through a moving head. Additive manufacturing modeling is a true multi-scale challenge. In this section, insights on how the simulation of the 3D-printing process can be solved via multiscale thermo-mechanical models are presented. The numerical simulation follows the real printing workflow, takes into account all process parameters and allows to predict the deformed shape of the part and residual stresses. The first sub-section describes the modeling approach, while specific developments for Fused Filament Fabrication and Selective Laser Sintering are addressed in the two following subsections. Finally, a warpage compensation technique is further considered to minimize part distorsion. Modeling approach The main goals of AM processes simulations include predicting the final distorted shape of a part, the residual stresses distribution and the process-induced microstructure (such as porosity and fiber orientation for reinforced plastics). However, the development of the AM process simulation has to cope with several challenges: 1. The complex thermomechanical loadings that occur during the layer-by-layer deposition of the material and the successive cooling of the part. 2. Additive manufacturing is a true multiscale challenge. Let us consider the FFF process (Figure [1]): at the macroscopic scale, the part is composed of contours and infill. One scale below, the infill is made of the superposition of beads inducing some specific mesostructures. These structures are Page 2

3 defined by the printing pattern (toolpath) that can be dense or sparse (e.g., lattices) and will drive the macroscopic mechanical behavior typically inducing anisotropy due to the inter-beads porosity. At the microscale, inside the beads, some porosity or fillers are susceptible to be present and will also impact the effective response of the material. 3. The thermal history of the material deposition generates differential shrinkage between adjacent beads or layers that shapes the part. Figure 1: Additive manufacturing is a multiscale challenge (illustrated for the FFF process). Overall, modeling the printing process requires users to account for the material state evolution and model the stress build-up as well as the stress relaxation, which can occur over time. Numerical predictions of warpage need to account for the process parameters, which can be a function of the type of AM technology considered the material characteristics and the printing strategy (part orientation, toolpath, supports ). The general workflow for process simulation is independent of the type of AM process (Figure [2]). The first step is to define the component to print: the initial part geometry (STL file) and the material specifications must be characterized. The AM process is further defined via the tool path and the process parameters, as well as some process-specific inputs to describe the manufacturing setup. This description of the process set up is then translated in an actual thermomechanical simulation, which will model the AM process and accounts for conduction, convection, and radiation inside the printer build. The part geometry is meshed based on a voxel approach to facilitate the layer-by-layer modelling of the AM. Once the FEA simulation is performed, residual stresses can be obtained as well as, the final deformed shape of the part. Page 3

4 Figure 2: Workflow in Digimat-AM to predict warpage and residual stresses for a given application. Fused Filament Fabrication applications Specifically, for FFF process, information about the time deposition and the microstructure are deduced from the toolpath (typically, from GCode file), which is loaded and read in the software (Figure [3]). The beads orientation are associated with the corresponding voxels and consistent microstructures are stacked up. These process-induced microstructures are dramatically influencing both the material properties and the structural performance, and in particular the failure. Figure 3: Toolpath loaded in the software to generate the process-induced microstructure. Green corresponds to support and red to part infill. An application example of FFF process simulation results is shown hereafter. Material Page 4

5 considered is ABS, which requires engineers to model: Isotropic elastic material response with thermal dependency of Young s modulus; Isotropic thermal expansion with thermal dependency; Specific volume with thermal dependency; Specific heat; Thermal conductivity. Part considered for the process simulation is a dumbbell. Support is idealized via clamping of all bottom nodes, which are released at the end of the thermomechanical simulation. Process parameters are summarized in the table below. Table I: Process parameters used for dumbbell application FFF process parameters Material temperature from printing head ( C) 230 Processing chamber temperature ( C) 80 Final temperature ( C) 23 Printing head speed (mm/s) 60 Bead width (mm) 0.4 Layer thickness (mm) 0.1 The process simulation of the part aims at evaluating the effect of the carbon fibers reinforcement on the warpage. It is assumed that the fibers are aligned with the toolpath. A comparison is performed for two toolpath patterns (see Figure [4]): beads aligned with the loading direction; beads transverse to the loading direction. Figure 4: Toolpath loaded on the dumbbells: beads aligned with the loading direction (left, 0 ) or transverse (right, 90 ). The results of the thermomechanical simulations are presented below. It can be observed that adding carbon fibers has a strong influence on the predicted warpage (Z deflection) for the Page 5

6 toolpath aligned with the dumbbell, while it has almost no effect when the toolpath is transverse to the main deflection direction. These simulation results aim at demonstrating that the effects of material reinforcement on warpage can be identified by a simulation approach. Figure 5: Z deflection predicted with unfilled ABS (upper left), 30% carbon-fibers reinforcement aligned with the toolpath (0 upper right) and transverse (90 bottom). Note: the same colorscale was used for all figures. Figure 6: Layer-by-layer build of an aerospace rocker printed part and warpage prediction. Page 6

7 Selective Laser Sintering applications Selective Laser Sintering process consists of the sintering of powdered material using a laser heat source. The laser selectively fuses the powder by scanning cross-sections of the part. Once a layer is sintered, an additional layer of powder is deposited from the powder bed, and a new sintering step starts. The SLS process involves several steps, which all need to be modelled by the simulation: Heating; Layer deposition; Laser movement and local temperature rise; Powder sintering; Heat diffusion; Additional layer deposition, which can be repeated as many times as required; Until global cool down. As for FFF process, the meshing procedure is based on voxels. Neighbouring powder is also modeled for actual part support. Overall, a parallelepiped containing the printed part and the neighbouring powder is generated. The thermomechanical simulation aims at modeling the building chamber environment of the SLS process, and relies on a multiscale thermomechanical material model. The progressive layer-by-layer manufacturing of an engine bracket is finally modeled and global distortion is shown on Figure [8]. Displacement magnitude is represented, such that areas colored in red undergo the most important distortion compared to the initial geometry shape. In order to highlight the effect of the printing direction on the part warpage, the same geometry is printed in exactly the same conditions, except the printing direction (a rotation of the part of 90 is performed along X-axis). The results of the warpage simulation are illustrated on Figure [7] with the same scale for both configurations. As can be observed, different printing directions lead to different final deformed shapes. Figure 7: Different printing directions (left: flat; right: rotated by 90 ) lead to different warpage. Figure [8] illustrates the impact of using reinforced plastics (i.e., ABS reinforced with 30% glass beads) on the part distortion. As expected, the glass beads leads to a decrease of the engine bracket distortion. Page 7

8 Figure 8: Effect of the material reinforcement on the warpage prediction (left: unfilled; right: +30% GB). Same scale is used for both figures. Warpage compensation case study The objective of this section is to illustrate the workflow performed to compensate for the warpage of a part. We will distinguish, on the one hand, shrinkage (contraction of the part due to the progressive cooling between the extrusion/sintering temperature and room temperature) from warpage (differential shrinkage e.g. between successive layers) and, on the other hand, the asdesigned geometry from the as-printed geometry, i.e. the final distorted shape of the part once the printing process is completed. Most printers actually integrate this distinction by preforming the geometry following a user-defined scale factor. However, this functionality is limited to shrinkage compensation. A solution to circumvent warpage consists of compensating the geometry to be printed, by specifying it such that after warpage, the as-printed geometry is sufficiently close to the asdesigned geometry. The proposed workflow for warpage compensation consists in allowing to export a counter-warped shape resulting from a first warpage analysis, and then using it as a compensated geometry (alternatively, a simple anisotropic scale factor can also be applied on the as-designed geometry to compensate for shrinkage). By running a new simulation on the compensated geometry, it can be checked that the as-printed geometry becomes similar to the as-designed geometry. By the end of this workflow, the geometry to send to the physical printer is identified, such that physical printing should yield the right as-printed geometry. Let us consider an engine bracket printed in FFF with a generic ABS. Warpage prediction workflow is illustrated in Figure [9] and the as-designed and as-printed geometries are superimposed in Figure [10], highlighting some clear mismatches. To improve the dimensional accuracy, based on the first warpage simulation, the as-designed geometry is deformed by applying to the nodes of the initial STL geometry the opposite of the resulting displacements. The geometry is reconstructed and can be exported (i.e., STL file of the compensated geometry). Performing a second process simulation using this compensated geometry (and re-computing the toolpath with the slicer software) yields the expected part dimensional accuracy (Figure [11]). The warpage compensation workflow is synthetized on Figure [12]. Page 8

9 Figure 9: Warpage prediction with Digimat-AM when printing the as-designed geometry. Figure 10: Comparison of as-designed (green) and as-printed geometry (warped geometry, red). Page 9

10 Figure 11: Comparison of as-designed (green) and as-printed geometry (warped geometry, red) before and after warpage compensation. Figure 12: Complete workflow for optimal printing. Material engineering The additive manufacturing process has a direct influence on the material behavior, whether for SLS or FFF processes. By considering and modeling heterogeneous materials with their true multiscale nature, engineers can understand the link between the as-printed microstructure and the material properties. In this section, multiscale material modeling techniques applied to additive manufacturing of polymers (unfilled and reinforced) will be detailed. Applications include the the build-up of nonlinear material models as a function of the printer toolpath, the computation of the effective mechanical response of lattices and the homogenization of the material behavior of reinforced polymers. Page 10

11 Fused Filament Fabrication Fused Filament Fabrication consists in depositing material filaments through a moving head. The superposition of the filaments or beads induces some specific microstructures based on the printing pattern. Such microstructures are shown in the illustration below, where beads have been simplified into straight cylinders showing a given interpenetration. Figure 13: Example of bead deposition pattern and FE mesh for FFF printed material. The deposition pattern considered is unidirectional 0 (Figure [13])). From this idealization of the bead deposition process, it can be seen that porosity is induced, with a given volume fraction, but also a given orientation. The FFF material, even when considering a pure polymer, therefore shows heterogeneities, ultimately leading to anisotropy in its behavior. As heterogeneous material properties depend on the material microstructure including void volume fraction and orientation, they are adequately modeled from micromechanics. Several multiscale modeling techniques can be applied for AM, whether it is mean-field homogenization (semi-analytical approach [2] [3]) or full-field homogenization (finite element based approach). Each of them follows the same general workflow: the multiphase microstructure of a Representative Volume Element (RVE) is described (for instance each phase shape, amount and orientation), the material properties of each phase are defined, so numerical procedures can be applied to compute the equivalent macroscopic properties. It is proposed to evaluate the non-linear response of ABS FFF material accounting for the printing pattern, first using the full-field homogenization approach. This widely used technique allows to model complex shape inclusions and microstructures with almost any degree of freedom. Once a RVE consistent with the process-induced microstructure is generated (assuming 10% of voids volume fraction, see Figure [13]), it can be further defined with a generic ABS J2- plasticity stress-strain curve fitted for pure material. The homogenization procedure of the ABS FFF microstructure yields the mechanical responses illustrated on Figure [14]. It can be seen that a softer response than the pure polymer is obtained, which is a direct consequence of the presence of porosity. However, porosity also induces anisotropy because of its preferential direction. Page 11

12 Figure 14: Full-field (left) and mean-field (right) homogenization of FFF ABS with 10% porosity. While mean-field homogenization is usually applied on fiber reinforced materials to model the effect of stiff inclusions, it can also be extended to void inclusions, which can be idealized as reinforcements showing no stiffness. The same material microstructure behavior is evaluated in Digimat-MF software which implements the Mori-Tanaka homogenization scheme [1]. Beads are modeled as a homogeneous material described by the same ABS material model, while porosity is modeled by voids. Results of the macroscopic equivalent response are shown on Figure [14] and are very similar to the full-field results. Reinforced material and effect of defects In this study, an example of application of multiscale modeling applied to reinforced materials of additive manufacturing materials is provided. A major concern of AM materials is the evaluation of the effect of defects, among them the intrabead porosity. In this application, the equivalent composite properties of a material compound based on a PEKK matrix with 30% of carbon fibers reinforcement are first deduced (Figure [15]). Then 2% of volume fraction of porosity is added to the material in order to evaluate the effect of defects. Table II gives the characteristics of each phase defining the multiphase material. In particular, the fibers are aligned following two distinct orientations: 7 and -7 with respect to the horizontal plane. Figure 15: Representative Volume Element (RVE) of the material microstructure. Carbon fibers are aligned following 7 (blue) and -7 (red) with respect to the horizontal plane. Green spheres depict the porosity content. The rest of the RVE is filled with PEKK matrix (not illustrated). Page 12

13 Figure 16: FE results of a uniaxial tensile loading (strain). The analysis is conducted using a full-field homogenization approach. Results of the equivalent compound are detailed in Table III and illustrated in Figure [16]. The presence of 2% volume fraction porosity generates a drop of 10-15% of most of the equivalent homogeneous material properties. Table II: Characteristics of the multiphase material PEKK Carbon fibers Porosity Young Modulus (MPa) 3, ,000 0 Poisson ratio 0,38 0,2 0 Thermal Expansion (K 1 ) 2,1E-5 1,8E-6 0 Thermal Conductivity (W/mK) 0, ,0001 Electric Conductivity (S/mm) 2,04E E-20 Volume Fraction 30% 2% Aspect Ratio 10 1 Orientation +7 /-7 Table III: Predicted equivalent material properties Material properties Material properties No Porosity 2% volume fraction porosity Young Modulus (MPa) Poisson ratio Max Matrix Strain 18% 15% Page 13

14 Thermal Expansion (K 1 ) 3.60E E-06 Thermal Conductivity (W/mK) Electric Conductivity (S/mm) Lattice structures The lattice microstructure illustrated on Figure [17] is further considered. It consists in sparse filaments of FFF material superposed with alternated angles (-45/+45 ). We will focus on the effective stiffness response of the lattice structure. Considering that the initial bulk tensile modulus is 2 GPa, the full-field homogenization procedure of the microstructure yields an equivalent modulus in the X direction of 80.5 MPa. Following the designer technical requirements, the lattice can then be modified to further lightweight the part or stiffen it. Figure 17: Predicting lattice structure effective response. Selective Laser Sintering Material characterization of Sinterline grade from Solvay (polyamide 6 reinforced with a 40% loading of glass beads) targets effect of printing direction. Plates have been printed using the SLS process, as shown on the figure below. From the plate, specimen have been cut out every 15 angle, with 0 being the in-plane reference, and 90 the build direction. Figure 18: Specimen cutting from SLS printed plate. Page 14

15 Specimens have then been tested both in tension and compression in order to obtain the effect of printing direction on the material response. When analyzing the observed material response, very little anisotropy is shown from a stiffness point of view, whether in tension or compression. However, still from a stiffness and plasticity point of view, a pressure sensitivity is highlighted, as compressive response shows a distinctly stiffer response. For a given type of loading (tension or compression), while little anisotropy is observed for stiffness, strain at failure is exhibiting strong anisotropy, with 0 strength almost twice the 90 strength. In other words, maximal strength is reached when material is loaded in the directions of the layer, while minimum strength is reached when material is loaded in the direction of powder bed progression. Figure 19: Tensile response of Sinterline at several loading angles compared to printing direction. Figure 20: Compressive response of Sinterline at several loading angles compared to printing direction (refer to Figure [19] for color legend). Page 15

16 Based on these experimental observations, a first modeling approach for the SLS PA6-GB40 material is chosen as following: Polyamide constituent behavior is described by a pressure sensitive material model: Drucker- Prager model has been selected [1]. Composite behavior is obtained by mean-field homogenization of the elastoplastic matrix with elastic glass beads. Failure anisotropy as a function of the loading direction is described by a Tsai-Wu Transversely Isotropic failure indicator: The calibration of the material model parameters is done to ensure the best fit with the experimental data. Failure model calibration results are shown below. While a very good fit between model and experiment is achieved for 0, 45 and 90, other directions show a less good fit. However, since the observed experimental data standard deviation is at maximum 5.7 MPa, the current fit is considered as satisfying and the model is judged sufficient to represent the evolution of material strength as a function of the loading direction. (1) Figure 21: Comparison of material failure model with experimental data for various loading angles. The final material model is thus capable of representing two keys aspects of the sintered polyamide material which were identified during the experimental characterization campaign: Pressure sensitivity of the plasticity via the Drucker-Prager model; Anisotropy of failure via a Tsai-Wu 3D Transversely Isotropic failure indicator. Page 16

17 By taking into account those key aspects, the material model can predict non-linear material response and failure for any generalized 3D loadings such as those which can be encountered in FEA analysis of structural parts. A structural engineering application and validation of the material model is proposed in the following section of the paper. Structural engineering Finally, to bridge the gap between process and as-printed part performance, a strongly coupled process-structure methodology will be shown in this section, as key enabler for predictive simulation of new lightweight high performance designs. The aim of structural modeling of AM parts is to predict the as-manufactured mechanical response and optimize part performance as a function of material and printing setup. This requires a strong coupling between process and structure, which is typically ensured by the usage of a material model which is itself dependent on the process parameters, such as the printing direction. SLS part Polimotor case study This section proposes an application and validation of the material model on a structural simulation of the air intake plenum part shared last year. However, in this first modeling and experimental testing approach, the loading conditions that the air chamber must withstand (a given inside pressure and specific temperature environment conditions) were limited to 3 bars and ambient temperature. The major outcome of the previous study was the evaluation of the effects of the printing direction on the plenum chamber ultimate strength. The numerical results have highlighted that the initial manufacturing direction was not optimal (see Table IV) and that the part strength could be improved. Table IV: Ultimate load of the plenum (internal pressure of 3 bars) in function of the printing direction. Ultimate load Printing direction X Y Z Internal pressure (bars) In this paper, the full technical requirements of the Polimotor project (6 bars at 100 C) are taken into account with a new material model built upon mechanical data obtained at 110 C and the previous results are updated accordingly. The procedure to reverse-engineer the new material model is similar to the one developed for SLS in Section 3 and typically provides the required dependency of material behavior on printing direction. To predict the ultimate strength of the part, a non-linear finite element analysis coupled with the Digimat multiscale material model is performed to simulate the progressive pressure increase. As soon as the failure indicator has reached a value of 1 in any element of the FEA, we assume the ultimate strength of the part is met (Figure [22]), which occurs once internal pressure reaches a value of 6.2 bars. Page 17

18 Figure 22: Failure prediction of the plenum chamber. Yellow field indicates critical failure spots. Courtesy of Solvay Engineering Plastics. To validate the simulation trends, an experimental set up has been designed to test the plenum chamber at higher temperature (about 100 C): Pressure increase by steps up to 6 bars positive air pressure inside the plenum; Pressure release to ambient pressure after about 60 minutes at 100 o C. Figure 23: Left: Experimental set up for internal pressure test campaigns; right: experimental step-by-step pressure increase until 6 bars. Courtesy of Solvay Engineering Plastics. The results of the pressure test campaign show that no burst of the part is happening, thus validating the part strength for the working load. As a conclusion, part can thus be used in the race car. Page 18

19 FFF parts Influence of the toolpath on the FEA results The influence of the printing direction on the part performance is a very well-known fact, as highlighted in the previous sub-section. When dealing with FFF-printed parts, another manufacturing data has to be connected to the structural analysis: the toolpath information. The toolpath is the path followed by the printer nozzle while depositing molten material to create the part to print, layer by layer. Thus, the toolpath consists of a set of disjoint continuous polylines located in horizontal planes, usually following the contour of the part to print or filling its interior. Two main reasons to take the toolpath information into account: the position of bead deposition creates specific microstructures based on the printing toolpath pattern, which drives the macroscopic mechanical behavior typically inducing anisotropy; the thermal history of the material deposition generates differential shrinkage between adjacent beads or layers that affects the material bonding, thus impacting the end performance of the part. The objective from the structural analysis point of view is thus to read the toolpath, extract the required information from the deposition sequence and transfer (through a mapping process) that information onto the target FEA mesh under a usable & efficient format for the FEA: Usable: must be readable by the FEA; Efficient: with the right compromise of accuracy/integrity (e.g., keep contour vs infill effects, but do not store each single printer nozzle displacement). The final output of this procedure is the creation of a local orientation tensor describing the direction of the path at each element. In this section, we will illustrate the workflow with the objective to predict the non-linear structural behavior with influence of the process parameters (toolpath & printing direction) and the material anisotropic mechanical behavior. For that purpose, a numerical tensile test is performed on a dumbbell on the non-linear anisotropic elasto-plastic material model of FFF ABS developed in Section 3. Figure 24: Mapping procedure of 0 unidirectional toolpath onto the FEA mesh. The first component of the orientation tensor gives the alignment of the filaments with the X direction (1 value is fully aligned, 0 is transverse). Page 19

20 Figure 25: Mapping procedure of 45 unidirectional toolpath onto the FEA mesh. The first component of the orientation tensor gives the alignment of the filaments with the X direction (1 value is fully aligned, 0 is transverse). Figures [24] and [25] illustrate the toolpath transfer procedure on the FE mesh of the dumbbell. In both cases, the orientation tensor highlights the contour (transverse to the X direction at the dumbbell extremities) and the infill (aligned for 0 and with a consistent 0.5 value for 45 ). The results of the structural analysis with a 90 toolpath are shown in Figures [26] and [27]. In particular, the influence of the toolpath on the Von Mises stress is shown as well as the different mechanical response of the contour with respect to the infill. Finally, the macroscopic failure indicator pattern demonstrates that the ultimate strength of the dumbbell will be first reached in the infill. Figure 26: Von Mises stress [MPa] of the dumbbell tensile test. Page 20

21 Figure 27: Macroscopic failure indicator (value of 1 means the ultimate strength has been reached). Finally, the global response of the dumbbell can be compared to the material mechanical behavior in the contours (Figure [28]). Contours follow the 0 mechanical behavior while the global response follows essentially the infill that is predominant. The FFF structural analysis taking into account the process setup (printing direction and toolpath) and the nonlinear anisotropic elastoplastic material behavior helps thus understand the mechanical behavior of the dumbbell. Page 21

22 Figure 28: Stress-strain curves of the FEA global response and in the contours. Summary e-xstream Engineering offers a holistic simulation chain for the additive manufacturing of plastics and composites with solutions for material engineering, process simulation, and part performance. This integrative approach is needed to accelerate the adoption of AM by the industry and promote new innovative structural designs needed to save energy and weight. This paper has covered the AM process simulation based on a multiscale simulation approach. Methodology and applications for AM process simulations of both SLS and FFF technologies were covered with a particular focus on warpage prediction and compensation. Additive manufacturing material engineering applications based on mean-field and full-field homogenization schemes were presented. Finally, structural engineering of SLS and FFF parts was conducted, highlighting the impact of the manufacturing setup, as the printing direction and toolpath. Acknowledgements The authors would like to acknowledge Solvay Engineering Plastics for the sponsoring of the Polimotor project. Bibliography 1. e-xstream engineering (2017), Digimat Users' Manual Release Eshelby, J. (1957), The determination of the elastic field of an ellipsoidal inclusion and related problems: Proceedings of the Royal Society of London, Doghri, I., Ouaar, A. (2003), Homogenization of two-phase elasto-plastic composite materials and structures: study of tangent operators, cyclic plasticity and numerical algorithms: Int. J. Solids Struct. 40, Page 22