- Email: [email protected]
An insight to the failure of FDM parts under tensile loading: finite element analysis and experimental study
Author’s Accepted Manuscript An Insight to the Failure of FDM Parts under Tensile Loading: Finite Element Analysis and Experimental Study Ashu Garg, Anirban Bhattacharya www.elsevier.com/locate/ijmecsci
PII: DOI: Reference:
S0020-7403(16)30959-6 http://dx.doi.org/10.1016/j.ijmecsci.2016.11.032 MS3511
To appear in: International Journal of Mechanical Sciences Received date: 2 June 2016 Revised date: 10 September 2016 Accepted date: 30 November 2016 Cite this article as: Ashu Garg and Anirban Bhattacharya, An Insight to the Failure of FDM Parts under Tensile Loading: Finite Element Analysis and Experimental Study, International Journal of Mechanical Sciences, http://dx.doi.org/10.1016/j.ijmecsci.2016.11.032 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
An Insight to the Failure of FDM Parts under Tensile Loading: Finite Element Analysis and Experimental Study Ashu Garg, Anirban Bhattacharya* Measurement and Process Analysis Laboratory, Department of Mechanical Engineering, Indian Institute of Technology Patna, Bihta-801103, Bihar, India * Corresponding author, email: [email protected]
Abstract Fused deposition modeling (FDM) is one of the widely adopted additive manufacturing techniques to fabricate complex three dimensional parts to a near-net shape. FDM parts are now-a-days not limited to prototype building for product realization but used as functional parts in widespread applications. The present work aimed at unveiling the deformation behavior of the FDM samples in general and individual rasters of different thicknesses (layer thickness), in particular, laid at different directions under uniaxial tension. Finite element (FE) analysis is carried for FDM tensile specimens to simulate elasto-plastic behavior and results are validated with the experimental observations. More realistic models for FE analysis are generated that include the layers of different thicknesses (0.178 mm, 0.254 mm, 0.330 mm) and rasters at different angles (0°, 90°, 0°/90°) maintaining the inter-layer and intralayer bonded region (developed in the present work). FE analysis and experimental results indicate that developed stress, strain at yield, elongation and tensile strength first decreases with layer thickness and then increases. Number of layers along the loading direction is more in sample with 0.178 mm layer thickness thus more elongation and load bearing capacity whereas for 0.330 mm layer thickness, less number of air voids and higher intra-layer bonded region are the reasons for higher tensile strength of the specimen. FE results and fractographic analyses reveal that first failure and layer separation in 90º rasters occurs followed by brittle failure of 0º rasters where pulling and necking takes place. Results also show that 0° raster layers fail under pulling and rupture of fibers and numerous micro-hills indicating micro-pulling of each raster fiber within a layer of material are the reasons for the improved tensile strength for 0° raster specimen. Keywords: Fused deposition modeling; Finite element analysis; Tensile testing; Layer thickness; Raster angle.
1. Introduction Rapid prototyping (RP) has become focus of manufacturing in recent years to accommodate part complexity and fabrication of parts within the time and cost constraints. Fused deposition modeling (FDM) is one of the widely adopted RP technique for fabricating three-dimensional (3D) complex components by depositing the material layer by layer through a very fine liquefier nozzle that moves in X and Y direction (in the plane of build platform). After depositing one layer, the build platform (or the worktable) is lowered in Z direction and the next layer is added. In FDM technique, the model and
support materials are deposited through a separate liquefier nozzles mounted on the extrusion head. To fabricate components without much geometric distortion, the parts are built with simultaneous deposition of support materials at desired locations. The deposited support material can be removed easily once the fabrication is complete and the part is taken out of the build chamber. Both model and support structure materials are available in the form of filaments which melt at preselected temperature before deposition. The deposited material solidifies rapidly and adheres to adjoining layers due to thermally driven diffusion bonding [1]. In FDM process components’ mechanical strength (like tensile, flexural strength) and surface roughness are observed to be highly anisotropic [2, 3]. Strength, roughness and geometric accuracy of the final manufactured parts depend on various process parameters such as contour width, raster angle, raster width, layer thickness, part orientation, air gap and machine settings [4]. Selection of optimum process parameters settings can improve the mechanical strength, surface roughness and geometric accuracy to considerable amount. The influence of model material temperature, raster width, air gap, raster angle and material colors were investigated on tensile and compressive strength of ABS P400 FDM samples [2]. Garg et al. [3] investigated the effect of raster angles on tensile strength, flexural strength and surface roughness of FDM parts. Results revealed that 0° raster angle offers more mechanical strength than 90° raster specimen. The influence of air gap, part orientation, raster width, layer thickness and raster angle was investigated on the tensile, flexural and impact strength of FDM parts using statistical approach [5]. Croccolo et al. [6] investigated the effect of part build directions, number of contours on tensile strength and stiffness of FDM parts using both experimental and analytical modeling techniques. They proposed analytical model to predict the failure of FDM samples considering the FDM parts as slender beam and sharing the load between longitudinal and inclined rasters and validated with experimental results with mean error reported as 4%. Espin et al. [7] investigated the effect of part orientations on the tensile strength of polycarbonate parts fabricated by FDM technique and verified the results with finite element method. They concluded that mechanical strength of FDM parts are anisotropic in nature and depend upon part building direction. To increase the mechanical strength, the part should be orientated in such a direction that produces longest contours aligned with tensile stresses. However, the other process parameters such as raster angle, layer thickness etc. also significantly affects the mechanical strength of FDM parts and needs to be modeled for detailed study. Huang and Singamneni [8] carried out analytical modeling considering a hypothesis that mechanical properties of FDM parts are structure sensitive in nature. They studied the influence of raster angles on the tensile strength, shear modulus, elastic modulus and Poisson’s ratio of FDM parts. Results showed that mechanical strength decreases with increase in raster angle from 0° to 90° which are also validated with experimental observations. The influence of layer thickness, scanning speed and road width was investigated on the distortion and residual stresses of FDM samples using finite element based thermo-mechanical model [9]. The results showed that scanning speed has significant effect on the part distortion whereas residual
stresses increases with increase in layer thickness. Bellehumeur et al. [10] studied the influence of envelope (build chamber) temperature, extrusion temperature and width of extruded filament on bond quality and neck formation between adjacent filaments in FDM process and found that extrusion temperature is most significant factor the effects the neck growth and bonding. Ravari et al. [11] investigated the effect of struts diameter on elastic modulus and collapse stress of cellular lattice structure fabricated by FDM process using finite element method. The influence of raster orientation was investigated on tensile, flexural and impact strength of FDM parts [12]. They observed that failure of the specimen mainly occurred along the layer interface. The effect of building direction and model interior on volumetric shrinkage and tensile strength was investigated for FDM parts using multi-objective optimization technique [13]. They further investigated the effect of inter-layer bonding, intra-layer bonding and neck formation between adjacent filaments on tensile strength of FDM parts using both experimental and mathematical modeling [14]. They observed that in a FDM sample with 0° raster layers the failure of a specimen is due to inter-layer fracture whereas in 45° raster layers, specimen fails under both inter-layer and intra-layer fracture. Dawoud et al. [15] investigated the effect of raster angle and raster air gap on mechanical strength of FDM parts and found that negative air gap significantly improved the mechanical strength of FDM parts. The influence of part build direction and raster orientation was investigated on tensile strength, tensile modulus and elongation of FDM samples [16]. The relationship between failure and mechanical properties are derived through fractographic analysis. Aforementioned literature review reveals that several authors have made attempt to determine the optimal process parameter settings to obtain desirable mechanical strength of FDM parts. The FDM parts, may it be the functional parts or prototypes, behave differently under loading condition and depend upon the layer thickness and alignment of the rasters with respect to loading direction [3]. Adaptive layer thicknesses [17], depending upon part shapes and structure, could optimize the surface roughness but may affect the strength of the build parts non-uniformly. Few authors have also studied the effect of process parameters on the performance of FDM samples through finite element methods [7, 9] however, finite element models that accounts for the layer pattern with their orientation and thickness can effectively predict the behavior of the FDM specimens. Thus, the present article is aimed at providing an insight towards the behavior during tensile failure of FDM samples built at different raster angles and layer thicknesses. In order to study the elasto-plastic behavior of the individual raster, their orientation, layer thickness, realistic FE models are developed that include the layers of different thicknesses (0.178 mm, 0.254 mm, 0.330 mm) and rasters at different angles (0°, 90°, 0°/90°) maintaining the inter-layer and intra-layer bonded region and subsequently results are validated with experiments. Later, fractographic analyses are carried out using optical microscope, 3D optical profiler and scanning electron microscopy (SEM) to study the failure modes of FDM samples.
2. Materials and Methods 2.1 Finite element modeling In the present work, finite element modeling and simulation are carried out for acrylonitrile butadiene styrene (ABS) FDM specimen using finite element (FE) package ABAQUS. The samples are designed at three different layer thicknesses (0.178, 0.254 and 0.330 mm) and three different raster angles (0°, 90° and 0°/90°) using Pro/Engineer 5.0 CAD modeling software. The created models are exported to ABAQUS in STEP file format. The overall dimensions for tensile test specimens are decided as per ASTM D638 standard (standard for tensile testing for plastic components) [18]. The detailed dimensions of the tensile specimen and raster angle arrangements are schematically shown in Fig. 1. Tensile specimen modeled for finite element analysis Loading direction
R14
19 4
0º R25 5 90º All dimensions are in mm Fixed gripper
(b)
(a)
0º/90º (c)
Fig.1. (a) Schematic of gripper arrangement, (b) tensile specimen as per ASTM D638 standard, (c) schematic illustration of different raster angles The geometric model of the tensile specimen is prepared by generating the layers at different raster angles for different layer thicknesses. The layer thickness (
) implies the nozzle diameter through
which the material is deposited. Instead of using the theoretical circular cross section of layers, the layers after depositing owns an elliptical cross section with certain amount of intra-layer necking. The growth of neck is a time dependent diffusion driven phenomenon attaining the final shape after the model is fully solidified and this has been validated by the optical microscopic visualization. Moreover, there is also some overlap region between two adjacent raster layers also known as interlayer necking which occurred due to diffusion of two raster layers at interface. General increase in necking with layer thickness has been observed from the microscopic measurements. The shape of individual rasters and overlap region observed under microscope can clearly be seen in Fig. 2. Using
the optical microscope (STM6, Olympus, Nagano-Ken, Japan, 0.5 µm least count), the height of the filament (h) is measured as 0.160 mm, 0.227 mm and 0.296 mm for layer thickness 0.178 mm, 0.254 mm and 0.330 mm. The other dimensions i.e width of the filament (w) is measured as 0.370 mm, 0.525 mm and 0.684 mm and inter-layer neck (a) region as 0.07 mm, 0.10 mm and 0.13 mm for different layer thickness (0.178 mm, 0.254 mm and 0.330 mm respectively). Thus, the dimensions, raster cross section and necking region are measured using the optical measuring microscope and used to generate a more realistic geometric model of the FDM test parts. To build the entire model for the tensile test specimen, first the cross section is created using the dimensions obtained from the microscopic measurements and then 465 to 9423 numbers of patterns (depending upon layer thickness and raster angle) are generated to prepare a rectangular model. Finally the model is trimmed as per the exact dimension of the test specimen (refer Fig. 1(b)). In order to reduce the number of elements, nodes and save computational time, the regions under the grip (in the tensile testing machine) of the specimen is not modeled (depicted schematically in Fig. 1). Air gap (0 mm) h0 During initial layer deposition
Neck growth and diffusion at interface
Intra-layer neck formation at interface
Inter-layer necking
Air voids
Intra-layer necking
Fig. 2. Necking formation between two adjacent raster layers, where a, w and h stands for necking region length, width of filament and height of filament The modeled FDM samples are imported to ABAQUS platform for the FE simulation of the elastic-plastic deformation under the uniaxial tensile loading. The FE simulation is carried out with implicit scheme and modeling it as load/displacement problem where the tensile load or the observed total displacement till fracture is applied along the longitudinal axis (length of the specimen). The material used for FDM model is assumed to be homogeneous and isotropic in nature. The material properties (stress-strain relationship) of ABS model material are derived from tensile test conducted by Anton et al. [19] for solid ABS samples. The tensile behavior of the FDM sample built with different layer thickness and/or raster orientation are different from the behavior and property of solid ABS material. However, while modeling by FEM, the material property of the ABS material is to be
defined thus the stress-strain data of solid ABS sample are considered from reference [19].The other material properties used for the analysis are density ρ = 1.04 g/cm3, Young’s modulus E = 2.2 GPa and Poison’s ration ν = 0.35. In explicit FE analysis the results may deviate from the actual if numbers of increment are not sufficient and may have continuous stability problems (conditionally stable), whereas implicit analysis often provides better results for elasto-plastic analysis although it requires larger computational time and space. Finer mesh size improves accuracy but demands higher computational resources thus the model is meshed by tetrahedron elements keeping the mesh size fine enough so that each raster is meshed properly for different layer thickness and raster orientations, and contains nodes in thickness directions also (as shown in zoomed view of the meshed model in Fig. 3). Continuum 3 dimensional 4 noded solid linear tetrahedron (geometrically versatile) element C3D4 (each node with 3 degrees of freedom) is used to mesh the specimen. The complete meshing of the model and 3D tetrahedron element is shown in Fig. 3.
N3
Inter-layer bonding
N4
N2 N1
Intra-layer bonding 0º
90º
Fig. 3. Representation of workpiece meshing and C3D4 tetrahedron mesh element The boundary conditions applied in the current FE model are shown in Fig. 4. As carried out during the tensile testing under uniaxial loading, one end of the work piece (or the model) is kept fixed and load and/or displacement is applied from another end (schematically represented in Fig. 1(a)). Fig. 4 shows the boundary condition where one end of the workpiece is encastred by constraining all six degree of freedoms, U1 = U2 = U3 = UR1 = UR2 = UR3 = 0. The displacement along longitudinal axis (z-axis in Fig. 4) is applied from the other end and displacement/ rotation from all
other sides are fixed, U1 = U2 = UR1 = UR2 = UR3 = 0, U3 = amount of displacement noted during failure test.
U1 = U2 = UR1 = UR2 = UR3 = 0, U3 = 0
U1 = U2 = UR1 = UR2 = UR3 = 0, U3 = - 2.4 mm
Fig. 4. Applied boundary conditions for FDM tensile specimen 2.2 Experimentation In the present work, all the parts are fabricated on FDM UPrint SE Plus and Mojo machines (Stratasys, Eden Prairie, USA) using ABS P430TM model material. The CAD models are designed using modeling software and converted into .STL file for slicing, layer path generation and building of the components. All specimens are built by depositing the materials in form of semi molten state extruded through a liquefier nozzle in form of layers with thickness of 0.178 mm, 0.254 mm or 0.330 mm and with constant raster angle 0°/90°. The dimensions for the tensile specimens are decided as per ASTM D638 standard [18] as shown in Fig. 1.
(a) (b) Fig. 5. Gripper arrangement and tensile specimen (a) before testing, and (b) after testing The tensile testing of the specimens are conducted on a Zwick/ Roell Z050 (Zwick Roell, Ulm, Germany) ultimate testing machine (UTM) equipped with 50 kN load cell and testing are conducted at a crosshead speed of 5 mm/min according to ASTM standard. The FDM sample fabricated for tensile testing, gripped in machine jaw for testing as shown in Fig. 5(a) and fractured specimen after testing is shown in Fig. 5(b). The optical measuring microscope and 3D optical profiler (Zegage, Zygo, Middlefield, CT, USA) are used to analyze the fracture surface of FDM samples to study the mode
and nature of raster failure when specimen built at different layer thickness and raster angles. Fractographic analyses are also carried out by studying the fracture surfaces using scanning electron microscopes at different magnifications.
3. Results and Discussions The FE simulations are carried out on Windows© based systems using ABAQUS for different models created with different raster angles and layer thicknesses. From the output database file, the results are extracted by creating path by joining the nodes along a certain direction. The simulated results are subsequently validated and compared with experimental observations. Later fractographic studies are also carried out to analyze the mode of failure of the FDM samples. 3.1 Finite element analysis The stress distribution on the samples with three different layer thicknesses and 0 º raster angles is shown in Fig. 6. The peak stresses developed in the samples are in the order of 39 MPa and small amount of necking region can be observed within the gage length region. Higher peak pressure is observed for 0.178 mm sample (39.2 MPa) which first decreases with increase in layer thickness (38.2 MPa for 0.254 mm layer thickness) and then again increases (38.8 MPa for 0.330 mm layer thickness). However an average stress of 34 MPa is noted in and around the neck region. This behavior shows that in the specimen with 0° raster angle, layers are aligned parallel to the loading direction (along the longitudinal or length of specimen). For specimen with 0.178 mm layer thickness the numbers of fine layer in the loading direction are more and load is borne by each layer with noticeable amount of individual necking. On the contrary, for 0.330 mm layer thickness numbers of layers are less but the amount of intra-layer necking region is high thus more close to behave like solid metal thus produces more strength to the specimen. A noticeable amount of necking both in width and thickness i.e. diffuse and localized necking are verified from the displacement plot of the FE results. The variations of stresses along the length of the specimen for all the samples with different layer thicknesses are shown in Fig. 6. The stress variations are captured along the three different paths (Path 1, 2 and 3) on the top, middle and bottom surface of the specimen joining the nodes in sequence. For the sample with 0.178 or 0.254 mm layer thickness, the maximum stress and the necking is observed close to the end of gage length whereas for sample with 0.330 mm layer thickness the maximum stress is distributed over the almost entire gage length and necking is observed on both sides of the gage length. An increase in maximum stress magnitude is observed with increase in layer thickness from 0.254 mm to 0.330. This increase in stress at 0.330 mm layer thickness specimen could be due to increase in overlap region and subsequent bonding strength between adjacent rasters at higher layer thickness. Moreover, with increase in layer thickness, number of layers required to fabricate the specimen decreases which further decreases the number of air voids formed at interstitial sites thus
load bearing capacity of the specimen increases. This air voids are major source for crack initiation and propagation in the specimen therefore, desirable to have lesser number of air voids. Stress distribution pattern for 0.330 mm layer thickness also shows that stress is more uniformly distributed over entire narrow region (gage length) due to strong bonding between adjacent rasters (observation is found similar as that of analysis for a solid model, discussed later in Fig. 8) whereas in sample with layer thickness 0.178 or 0.254 mm, stress is more concentrated towards the end of the gage length where the area starts decreasing. 45 Gage length
40 35
Path 1 30
Path 2 Path 3
Stress (MPa)
25 20 Path 1
15 Fixed end
10
Top layer
Path 2
Pulling direction Path 3
5 Middle layer
Bottom layer
0
0
10
20
(a)
30 40 50 Distance along path (mm)
60
70
80
45 Gage length
40 35
Path 1
30
Path 2 Path 3
Stress (MPa)
25
20 Path 1 15 Top layer
Fixed end
10
Path 3
Path 2
Pulling direction
5
Bottom layer
Middle layer 0 0
(b)
10
20
30 40 50 Distance along path (mm)
60
70
80
45 Gage length
40 35
Path 1
Stress (MPa)
30
Path 2 Path 3
25
20 Path 1 15 Fixed end
10
Pulling direction
Top layer Path 2
Path 3
5
Middle layer
Bottom layer
0 0
(c)
10
20
30 40 50 Distance along path (mm)
60
Fig. 6. Stress-displacement and stress distribution pattern for specimen built at 0° raster angle with layer thickness (a) 0.178 mm, (b) 0.254 mm, (c) 0.330 mm
70
80
The stress-displacement curve for specimen with 90° raster angle is shown in Fig. 7. For 90° raster angle specimen, layers are laid perpendicular to the loading direction thus failure of the specimen mainly depends upon adhesion between adjacent rasters (intra-layer bonding). During tensile loading/displacement applied perpendicular to raster alignment, the stress is concentrated to narrow region in and around the intra-layer bond region which causes the failure of the specimen due to delamination/ separation of layers from bonded region. The sudden increase in stress, concentrated to small region can also be seen form Fig. 7 where sharp peaks within the gage length region indicates high stress concentration region.
45 Gage length
40 35
Path 1
Stress (MPa)
30
Path 2 Path 3
25 20 Path 1
15 Fixed end
10
Top layer
Path 2
Pulling direction Path 3
5 Middle layer
Bottom layer
0
(a)
0
10
20
30 40 50 Distance along path (mm)
60
70
80
45 Gage length
40 35
Path 1
Stress (MPa)
30
Path 2 Path 3
25 20 Path 1
15 Fixed end
10
Pulling direction
Top layer Path 3
Path 2
5 Bottom layer
Middle layer 0 0
(b)
10
20
30 40 50 Distance along path (mm)
60
70
80
45 Gage length
40 35
Path 1
Stress (MPa)
30
Path 2 Path 3
25 20 Path 1
15 Fixed end
10
Top layer
Path 2
Pulling direction Path 3
5 Middle layer
Bottom layer
0
(c)
0
10
20
30 40 50 Distance along path (mm)
60
70
Fig. 7. Stress-displacement and stress distribution pattern for specimen built at 90° raster angle with layer thickness (a) 0.178 mm, (b) 0.254 mm, (c) 0.330 mm
80
It can be seen from Fig. 7 that amount of peak stress and distribution in and around the neck regions is similar for all specimens which predict that for 90° raster specimen failure is not significantly dependent on layer thickness. This is due to the fact that specimen fails at intra-layer bond and number of layers or air voids does not play much significant role and 90° raster specimen fails due to delamination and separation of layers from intra-layer bond region.
45 Gage length
40 35
Stress (MPa)
Path 1 30
Path 2
25
Path 3 Path 4
20 15
Path 1
10
Fixed end
Pulling direction Path 4
Top layer (0 ) Path 2
5
Path 3
Middle layer (90 )
Middle layer (0 )
Bottom layer (90 )
0
0
(a)
10
20
30 40 50 Distance along path (mm)
60
70
80
45 Gage length
40 35
Stress (MPa)
Path 1 30
Path 2
25
Path 3 Path 4
20 15
Path 1
10
Fixed end
Path 2
5
Middle layer (90 )
0
0
(b)
10
Pulling direction
Top layer (0 ) Path 3
20
Path 4
Middle layer (0 )
Bottom layer (90 )
30 40 50 Distance along path (mm)
60
70
80
45 Gage length
40 35
Path 1
Stress (MPa)
30
Path 2
25 20 Pulling direction
15 Fixed end
10
Path 1
Path 2
5 Top surface
(c)
Bottom surface
0
0
10
20
30 40 50 Distance along path (mm)
60
70
Fig. 8. Stress-displacement and stress distribution pattern for specimen built at 0°/90° raster angle with layer thickness (a) 0.254 mm, (b) 0.330 mm, and (c) solid tensile specimen
80
The stress distribution and variations of stresses for sample with 0°/90° raster (alternate layers with raster at 0º and 90º) is shown in Fig. 8. In 0°/90° raster specimen layers are laid in two orientations 0° and 90° alternately and show different behavior for 0° raster angle and 90° raster angle layers. From Fig. 8 it can be noticed that for path 1 and path 3 (layer with raster at 0°), average as well as peak stress is considerably higher as compared to path 2 and path 4 i.e. in the layers where rasters are laid in 90° to the length of the specimen. The rasters at 0º (to the length of the specimen) are aligned parallel to loading direction and under tensile loading these rasters continues to get pulled exhibiting a small amount of necking however, 90° raster layers are aligned perpendicular to the loading direction thus produces weak direction. It can also be seen that the overall stress generation lies between 0° and 90° raster specimen because the premature bond separation of 90° raster layers causes the earlier failure of 0° raster layers. For 0º/90º raster samples the load bearing capacity of the specimen increases with layer thickness due to increase in overlap region between adjacent layers. It may be also noted from Fig. 8 that the average stress variation in 0.330 mm layer thickness specimen is alike the variation as noted for the solid specimen (equivalent to specimen receives from injection molding processes) with only difference that necking is observed for solid sample on both sides of the gage length. Thus the stress variation and distribution reveals that during the failure first the 90º rasters get separated from the intra-layer bond region (fracture eventually starts from these regions) but 0º rasters pulled and experience diffused necking and finally brittle failure of the 0º rasters lead to the fracture of the specimen. Elemental stress development for the three different samples with 0º, 90º, 0º/90º rasters and solid samples are represented in Fig. 9. The element is considered at the highly stress generated region of FE model (neck region) with different layer thickness and raster angles. Elemental stress measurement is important to study the behavior of individual element under uniaxial tensile loading. It may be noted that experimentally it is difficult to measure/capture the elemental stress thus finite element results are extracted to study the development of stress on an element when tensile loading is applied. It can be seen from Fig. 9(a) that for 0° raster, yielding starts earlier for samples of 0.178 mm layer thickness as compared to higher layer thickness. For all the samples with 0º raster, where the individual small rasters are aligned with loading direction, after short duration of yielding diffuse necking starts. Due to the low plastic region of the ABS materials strain hardening no longer able to compensate the weakening due to the reduction in cross sectional area. On the contrary, for 90º rasters an almost similar behavior is observed for all samples with different layer thicknesses. Rather than a clear yielding the deformation starts and irrespective of layer thickness the separation of layers from intra-layer bond region takes place. The variation of elemental stress for 0º/90º rasters of different layer thickness and solid samples are shown in Fig. 9(c). Elements are taken from both the 0° and 90° raster layers and stress developments are shown in the Fig. 9(c). After the yielding region a quick drop in elemental stress is noted for 90º rasters of 0.254 mm layer thickness indicating the initiation of failure at such rasters. However the same is delayed for 0.330 mm layer thickness samples due to the
fact of thicker intra-layer bonding. The loss of homogeneous state of deformation and subsequent necking leading to failure is observed for all the specimens but the failure in 0.330 mm layer thickness
45
40 45
40 45
35 40
35 40
30 35
30 35
25 30
Stress (MPa) Stress (MPa)
45
0.178 mm
20 25 15 20 10 15 5 10 0 5 0
0.254 mm 0.330 mm 0.178 mm
0.254 mm 0.330 mm 0.2
0.4
0.6
0.8
1
Time step
0 0
0.5
1
25 30 20 25 15 20 10 15 5 10 0 5 0
0.178 mm
0.254 mm 0.330 mm 0.178 mm
0.254 mm 0.330 mm 0.2
0.4
0.5
1
2
2.5
0.6
0.8
1
2
2.5
Time step
0
1.5
0
Displacement (mm)
1.5
Displacement (mm)
(b)
(a) 45 40 45 35 40 30 35
Stress (MPa) Stress (MPa)
Stress (MPa) Stress (MPa)
sample occurs later as compared to 0.254 mm thickness.
25 30 20 25 15 20 10 15 5 10 0 5 0
Solid specimen 0.254 mm 0.178 mm 0.330 mm 0.254 mm 0 raster layer 0.330 mm 90 raster layer 0.2
0.4
0.6
0.8
1
2
2.5
Time step
0 0
0.5
1
1.5
Displacement (mm)
(c)
Fig. 9. Elemental stress development during tensile loading of FDM specimen built with (a) 0° raster layers, (b) 90° raster layers, and (c) 0°/90° raster layers 3.2 Experimental validation Tensile testing for samples with three different layer thickness and 0°/90° rasters are conducted experimentally. For each specimen built at different layer thickness (0.178 mm, 0.254 mm and 0.330 mm) and 0°/90° raster angles experimentally obtained stress-strain curves are shown in Fig. 10. From
Fig. 10 it can be noted that with increase in layer thickness tensile strength first decreases and then increases. The similar behavior is also observed from FE analysis and discussed in previous subsection. This phenomenon is mainly because in 0.330 mm layer thickness number of layer decreases thus air voids, distortion and shrinkage is less as compared to other layer thickness. Moreover, for 0.330 mm layer thickness the overlap region between adjacent layers is more than the other layer thickness. Therefore, the loading bearing capacity of the specimen increases and more stresses is observed in the specimen as compared to other layer thickness (0.254 mm). In 0.178 mm layer thickness even though large numbers of layers are required to build the specimen but still it exhibit more tensile strength than 0.330 mm layer thickness. This is because in 0.178 mm layer thickness relatively more numbers of layer are aligned along the loading direction compared to other layer thickness. Thus during tensile testing more number of layers are pulled and breaks under tensile loading. With increase in layer thickness the yielding is shifted (takes place at higher strain, indicates more elastic region) and strain at failure is also increased.
35 30
Stress (MPa)
25 20
15
0.178 mm 0.254 mm
10
0.330 mm 5
0 0
0.5
1
1.5
2 2.5 3 3.5 Strain (%)
4
4.5
5
Fig. 10. Stress-strain curve for specimen built at 0°/90° raster angle and with different layer thickness Photographic view of the failure specimen and the stress distribution as obtained from the FE analysis is shows (Fig. 11) as a sample representation for 0.178 mm, 0.254 mm and 0.330 mm layer thickness. The failure in the sample after tensile testing took place close to the end of the gage length marked on the specimen. Corresponding stress distribution also shows that the maximum stress and necking occurred towards the end of the gage length region. Fig. 11(a) also shows the stretched marks (indicative of stressed regions) on fabricated specimen of 0.178 mm layer thickness under uniaxial tensile loading which is equivalent to stress distribution pattern in FE model. This stretched mark also shows that stresses are more uniformly distributed over entire gage length in 0.178 mm layer
thickness sample. The more number of fine layers in the loading direction and strong bonding between the adjacent rasters increases the loading bearing capacity of each raster under uniaxial tensile loading. The small amount of necking is also observed towards the end of the gage length where the area starts decreasing. On the contrary, for 0.254 mm and 0.330 mm layer thickness numbers of layers are less as compared to specimen with 0.178 mm layer thickness but the amount of intra-layer necking region is high thus stresses are more concentrated to narrow region i.e. towards the end of the gage length (refer Fig. 11(b) and (c)).
Gage length
Gage length
Stretch marks during tensile loading
Necking region
(a) Gage length Gage length
Necking region
(b)
Gage length Gage length Necking region
(c) Fig. 11. Failure specimen after tensile testing and the stress distribution for sample built at 0º/90º raster angle and layer thickness (a) 0.178 mm, (b) 0.254 mm, and (c) 0.330 mm
3.3 Fractographic analysis Cross sectional views of the samples at the maximum stress regions for all the samples at different raster angle and layer thickness along with the solid sample are shown in Fig. 12. Fig. 12 shows that the maximum stress generated in the neck region is distributed almost over the entire cross sectional area when layer thickness is 0.178 mm as well as 0.330 mm as in 0 º raster angles each layer is pulled under tensile loading whereas for 0.254 mm layer thickness the maximum stresses are distributed mostly over the peripheral regions. For 90º raster angles since layers are laid perpendicular to loading direction less stresses are generated and failure of specimen mainly depend upon adhesion between successive layers. Fig. 12 also shows that failure of specimen started from outer layers and moves towards the center of the workpiece for all samples. However, for 0º/90º raster angle specimen less stresses are generated in 90º raster layers and comparatively higher stresses in 0º raster layers can be easily noted. With increase in layer thickness more stresses are generated in each specimen due to increase in overlap region between adjacent layers and layers are more strongly bonded together and require higher force to cause failure of the specimen.
Layer thickness 0.178 mm
Layer thickness 0.254 mm
Layer thickness 0.330 mm
(a) 0° raster angle
Layer thickness 0.178 mm
Layer thickness 0.254 mm
Layer thickness 0.330 mm
(b) 90° raster angle 0° rasters
90° rasters Solid specimen
Layer thickness 0.254 mm
Layer thickness 0.330 mm
(c) solid specimen and 0°/90° raster angle
Fig. 12. Cross sectional surfaces after FE simulations of samples with different raster angles at different layer thicknesses and solid sample
Fractographic analysis of built specimens at different layer thickness is shown in Fig. 13. It can be easily noted from the Fig. 13 that the 0° raster layers first undergoes pulling followed by small necking and subsequently fails due to brittle fracture with tearing of individual raster layers. Some air voids present between adjacent raster layers decreases the strength of the specimen. For 90° raster layers quick delamination from the bonded region between adjoining layers take place.
Air gap
(a)
Delamination failure for 90° raster layers
(b)
Tearing and brittle fracture of 0° raster layers (c)
Fig. 13. Fractographic analysis of FDM sample built at 0°/90° raster angle with layer thickness (a) 0.178 mm, (b) 0.254 mm, and (c) 0.330 mm
Fig. 13 also reveals that with increase in layer thickness the overlap region between adjoining layers increases (maximum in 0.330 mm layer thickness). The 3D optical profiler image for 0° raster layers indicates that the 0° raster layers fails under pulling and rupture of raster fibers and numerous microhills indicate micro-pulling of each raster fiber within a layer of a material which also indicate good tensile strength for 0° raster specimen as shown in Fig. 13. For 90° raster layers material fails under delamination with small region of rupture/ brittle fracture which indicate relatively low tensile strength as compared to 0° raster layers. For 0º rasters unevenness in failure surface indicates the pulling and necking of the fibers (refer Fig. 14) and a significant amount of stepped failure region in 0º raster for relatively higher layer thickness can also be seen in Fig. 14(b) (inset). However, mostly co-planar separation with micro-hills in 90º rasters indicate the bond separation from the adhered regions (refer Fig. 14).
(a)
(b)
Fig. 14. Failure surface patterns for 0° and 90° raster layers for samples build with layer thickness (a) 0.254 mm, and (b) 0.330 mm
Additionally, few parts are fabricated at different part orientations (X, Y and Z orientation). Building of parts on thickness face (horizontally), width face and thickness face (vertically) indicates X, Y and Z orientations respectively. Buildings of all previous samples used for tensile testing are done about Y orientation for which the results and fractographic analysis are presented before. SEM images of the fracture surfaces of the samples build at X0 (X orientation, 0º raster), Z90, X60 are shown in Fig. 15. The maximum tensile strength was achieved for X and Y orientations whereas for Z orientations strength is inferior due to weak bonding between adjacent rasters [20]. For 0° raster layers in X or Y orientation, layers are laid parallel to loading direction more stresses are generated in each layers as compared to 90° raster layers thus 0° layers fails under brittle fracture i.e. tearing and rupture of individual rasters. Fig. 15(a) shows the fracture surface of FDM sample fabricated at X orientation where it can clearly be seen that 0° raster layers separated by tearing and rupture of individual layers.
Fig. 15. SEM images of tensile failure sample built at (a) X0, (b) Z90, and (c) X60 In Z orientation both 0°/ 90° raster layers are of very small size and aligned perpendicular to the loading direction therefore strength of the part depend upon adhesion between successive layers and specimen fails due to delamination or separation of layers form weak bonding. Fig. 15(b) shows the corresponding fracture surface where layers separation at the weak interlayer bond can be easily seen thus provides the lowest tensile strength (14.8 MPa in comparison of 34.5 or 32.2 MPa for X, Y orientation). For X and Y orientations when specimen is fabricated either with rasters at 30° or 60° grain pullout, necking and tearing takes place (refer Fig. 15(c)) thus highest tensile strength are possible to achieve. Therefore, it is desirable to fabricate the FDM parts either in X orientation or Y orientation rather than the Z orientation. Higher strength of the samples would be achieved for 0° raster layers as compared to 90° raster layers however, layers at 30° or 60° raster layers could also be preferred and parametric optimization would be needed to decide the orientation and raster angle so as to achieve the best possible mechanical strength of the FDM parts.
4. Conclusions In the preset work, FE modeling and simulation are carried out developing realistic models that considers layers of different thicknesses and rasters at different angles maintaining the inter-layer and
intra-layer bonded region. Experimental studies, fractographic analysis are also performed in order to validate the results. From the results the following conclusions are drawn. FE analysis indicates that necking is present in the 0º rasters and strain at failure and elongation increases with increase in layer thickness whereas tensile stress first decreases with increase in layer thickness then increases. More number of layers in loading direction, particularly when layer thickness is small, contributes to the load bearing and more stress generation whereas the higher layer thickness provides more amount of intra-layer bonded region thus provides higher tensile strength. Failure first takes place through intralayer separation in rasters 90º to the loading direction followed by the pulling and tearing of the 0º rasters along with the inter-layer (between 0º-90º rasters) bond separation. Fractographic analysis indicates that 0° raster layers fails under brittle fracture with tearing and rupture of individual layers. In 90° raster layers failure of the specimen occurred due to delamination or separation of layers from the adjacent bonds. Higher strength of the components are possible to achieve when specimen is fabricated either in X or Y orientations whereas inferior strength is obtained for Z orientation parts. In case of raster angles 30° or 60°, shearing of individual layers adds up to the contribution of higher tensile strength in addition to the pulling, rupture and bond failure. Numerous micro-hills are observed on the failure surfaces of the specimens that indicate micro-pulling of each raster fiber within a layer of material and contribute to the improved tensile strength of the specimen.
Acknowledgement Authors thankfully acknowledge the support and help received from Mr. Sandeep Shukla of M/s DesignTech Systems, New Delhi, India.
References 1.
2. 3.
4. 5. 6.
7.
Sood AK, Ohdar RK, Mahapatra SS. Parametric appraisal of fused deposition modelling process using the grey taguchi method. Proc IMechE, Part B: J Engineering Manufacture 2010; 224:135– 45. Ahn SH, Montero M, Odell D, Roundy S, Wright PK. Anisotropic material properties of fused deposition modeling ABS. Rapid Prototyping J 2002; 8:248–57. Garg A, Bhattacharya A, Batish A. Failure investigation of fused deposition modelling parts fabricated at different raster angles under tensile and flexural loading. Proc IMechE, Part B: J Engineering Manufacture 2015; http://dx.doi.org/10.1177/0954405415617447. Garg A, Bhattacharya A, Batish A. On surface finish and dimensional accuracy of FDM parts after cold vapor treatment. Mater Manuf Process 2016; 31:522–29. Sood AK, Ohdar RK, Mahapatra SS. Parametric appraisal of mechanical property of fused deposition modelling processed parts. Mater Des 2010; 31:287–95. Croccolo D, Agostinis MD, Olmi G. Experimental characterization and analytical modelling of the mechanical behaviour of fused deposition processed parts made of ABS-M30. Comput Mater Sci 2013; 79:506-18. Espin MD, Forcada JMP, Granada AAG, Llumà J, Borros S, Reyes G. Mechanical property characterization and simulation of fused deposition modeling polycarbonate parts. Mater Des 2015; 83:670–77.
8. 9.
10. 11.
12.
13. 14. 15. 16.
17. 18. 19.
20.
Huang B, Singamneni S. Raster angle mechanics in fused deposition modelling. J Compos Mater 2015; 49:363–83. Zhang Y, Chou K. A parametric study of part distortions in fused deposition modelling using three-dimensional finite element analysis. Proc IMechE, Part B: J Engineering Manufacture 2008; 222:959–68. Bellehumeur C, Li L, Sun Q, Gu P. Modeling of bond formation between polymer filaments in the fused deposition modeling process, J Manuf Process 2004; 6:170-78. Ravari MRK, Kadkhodaei M, Badrossamay M, Rezaei R. Numerical investigation on mechanical properties of cellular lattice structures fabricated by fused deposition modeling. Int J Mech Sci 2014; 88:154–61. Es-Said OS, Foyos J, Noorani R, Mendelson M, Marloth R, Pregger BA. Effect of layer orientation on mechanical properties of rapid prototyped samples. Mater Manuf Process 2000; 15:107–22. Gurrala PK, Regalla SP. Multi-objective optimization of strength and volumetric shrinkage of FDM parts. Virtual Phys Prototyp 2014; 9:127–38. Gurrala PK, Regalla SP. Part strength evolution with bonding between filaments in fused deposition modelling. Virtual Phys Prototyp 2014; 9:141–49. Dawoud M, Taha I, Ebeid SJ. Mechanical behaviour of ABS: An experimental study using FDM and injection moulding techniques. J Manuf Process 2016; 21:39–45. Riddick JC, Haile MA, Wahlde RV, Cole DP, Bamiduro O, Johnson TE. Fractographic analysis of tensile failure of acrylonitrile-butadiene-styrene fabricated by fused deposition modeling. Addit Manuf 2016; 11:49–59. Pandey PM, Reddy NV, Dhande SG. Slicing procedures in layered manufacturing: a review. Rapid Prototyping J 2003; 9:274–88. ASTM D638-14. Standard test method for tensile properties of plastics. West Conshohocken, PA: ASTM International; 2014. Anton S, Thomas A, Markus G, Martin F, Wagner M, Hierl S. Finite element analysis of thermoplastic probes under tensile load using LS-DYNA compared to ANSYS WB 14 in correlation to experimental investigations. ANSYS Conference & 30th CADFEM Users’ Meeting 2012. Garg A, Bhattacharya A, Batish A. Chemical vapor treatment of ABS parts built by FDM: Analysis of surface finish and mechanical strength. Int J Adv Manuf Technol 2016; http://dx.doi.org/10.1007/s00170-016-9257-1.
Highlights Realistic FE models for different layer thickness and raster angles maintaining inter-layer and intra-layer bonded region The developed stress, strain at yield, elongation and tensile strength first decreases with layer thickness and then increases. The 0° raster layers fail under pulling and rapturing of fibers and presence of micro-hills indicates micro-pulling of each raster fiber within a layer.
For 90° raster layers material fails under delamination and bond separation.
PII: DOI: Reference:
S0020-7403(16)30959-6 http://dx.doi.org/10.1016/j.ijmecsci.2016.11.032 MS3511
To appear in: International Journal of Mechanical Sciences Received date: 2 June 2016 Revised date: 10 September 2016 Accepted date: 30 November 2016 Cite this article as: Ashu Garg and Anirban Bhattacharya, An Insight to the Failure of FDM Parts under Tensile Loading: Finite Element Analysis and Experimental Study, International Journal of Mechanical Sciences, http://dx.doi.org/10.1016/j.ijmecsci.2016.11.032 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
An Insight to the Failure of FDM Parts under Tensile Loading: Finite Element Analysis and Experimental Study Ashu Garg, Anirban Bhattacharya* Measurement and Process Analysis Laboratory, Department of Mechanical Engineering, Indian Institute of Technology Patna, Bihta-801103, Bihar, India * Corresponding author, email: [email protected]
Abstract Fused deposition modeling (FDM) is one of the widely adopted additive manufacturing techniques to fabricate complex three dimensional parts to a near-net shape. FDM parts are now-a-days not limited to prototype building for product realization but used as functional parts in widespread applications. The present work aimed at unveiling the deformation behavior of the FDM samples in general and individual rasters of different thicknesses (layer thickness), in particular, laid at different directions under uniaxial tension. Finite element (FE) analysis is carried for FDM tensile specimens to simulate elasto-plastic behavior and results are validated with the experimental observations. More realistic models for FE analysis are generated that include the layers of different thicknesses (0.178 mm, 0.254 mm, 0.330 mm) and rasters at different angles (0°, 90°, 0°/90°) maintaining the inter-layer and intralayer bonded region (developed in the present work). FE analysis and experimental results indicate that developed stress, strain at yield, elongation and tensile strength first decreases with layer thickness and then increases. Number of layers along the loading direction is more in sample with 0.178 mm layer thickness thus more elongation and load bearing capacity whereas for 0.330 mm layer thickness, less number of air voids and higher intra-layer bonded region are the reasons for higher tensile strength of the specimen. FE results and fractographic analyses reveal that first failure and layer separation in 90º rasters occurs followed by brittle failure of 0º rasters where pulling and necking takes place. Results also show that 0° raster layers fail under pulling and rupture of fibers and numerous micro-hills indicating micro-pulling of each raster fiber within a layer of material are the reasons for the improved tensile strength for 0° raster specimen. Keywords: Fused deposition modeling; Finite element analysis; Tensile testing; Layer thickness; Raster angle.
1. Introduction Rapid prototyping (RP) has become focus of manufacturing in recent years to accommodate part complexity and fabrication of parts within the time and cost constraints. Fused deposition modeling (FDM) is one of the widely adopted RP technique for fabricating three-dimensional (3D) complex components by depositing the material layer by layer through a very fine liquefier nozzle that moves in X and Y direction (in the plane of build platform). After depositing one layer, the build platform (or the worktable) is lowered in Z direction and the next layer is added. In FDM technique, the model and
support materials are deposited through a separate liquefier nozzles mounted on the extrusion head. To fabricate components without much geometric distortion, the parts are built with simultaneous deposition of support materials at desired locations. The deposited support material can be removed easily once the fabrication is complete and the part is taken out of the build chamber. Both model and support structure materials are available in the form of filaments which melt at preselected temperature before deposition. The deposited material solidifies rapidly and adheres to adjoining layers due to thermally driven diffusion bonding [1]. In FDM process components’ mechanical strength (like tensile, flexural strength) and surface roughness are observed to be highly anisotropic [2, 3]. Strength, roughness and geometric accuracy of the final manufactured parts depend on various process parameters such as contour width, raster angle, raster width, layer thickness, part orientation, air gap and machine settings [4]. Selection of optimum process parameters settings can improve the mechanical strength, surface roughness and geometric accuracy to considerable amount. The influence of model material temperature, raster width, air gap, raster angle and material colors were investigated on tensile and compressive strength of ABS P400 FDM samples [2]. Garg et al. [3] investigated the effect of raster angles on tensile strength, flexural strength and surface roughness of FDM parts. Results revealed that 0° raster angle offers more mechanical strength than 90° raster specimen. The influence of air gap, part orientation, raster width, layer thickness and raster angle was investigated on the tensile, flexural and impact strength of FDM parts using statistical approach [5]. Croccolo et al. [6] investigated the effect of part build directions, number of contours on tensile strength and stiffness of FDM parts using both experimental and analytical modeling techniques. They proposed analytical model to predict the failure of FDM samples considering the FDM parts as slender beam and sharing the load between longitudinal and inclined rasters and validated with experimental results with mean error reported as 4%. Espin et al. [7] investigated the effect of part orientations on the tensile strength of polycarbonate parts fabricated by FDM technique and verified the results with finite element method. They concluded that mechanical strength of FDM parts are anisotropic in nature and depend upon part building direction. To increase the mechanical strength, the part should be orientated in such a direction that produces longest contours aligned with tensile stresses. However, the other process parameters such as raster angle, layer thickness etc. also significantly affects the mechanical strength of FDM parts and needs to be modeled for detailed study. Huang and Singamneni [8] carried out analytical modeling considering a hypothesis that mechanical properties of FDM parts are structure sensitive in nature. They studied the influence of raster angles on the tensile strength, shear modulus, elastic modulus and Poisson’s ratio of FDM parts. Results showed that mechanical strength decreases with increase in raster angle from 0° to 90° which are also validated with experimental observations. The influence of layer thickness, scanning speed and road width was investigated on the distortion and residual stresses of FDM samples using finite element based thermo-mechanical model [9]. The results showed that scanning speed has significant effect on the part distortion whereas residual
stresses increases with increase in layer thickness. Bellehumeur et al. [10] studied the influence of envelope (build chamber) temperature, extrusion temperature and width of extruded filament on bond quality and neck formation between adjacent filaments in FDM process and found that extrusion temperature is most significant factor the effects the neck growth and bonding. Ravari et al. [11] investigated the effect of struts diameter on elastic modulus and collapse stress of cellular lattice structure fabricated by FDM process using finite element method. The influence of raster orientation was investigated on tensile, flexural and impact strength of FDM parts [12]. They observed that failure of the specimen mainly occurred along the layer interface. The effect of building direction and model interior on volumetric shrinkage and tensile strength was investigated for FDM parts using multi-objective optimization technique [13]. They further investigated the effect of inter-layer bonding, intra-layer bonding and neck formation between adjacent filaments on tensile strength of FDM parts using both experimental and mathematical modeling [14]. They observed that in a FDM sample with 0° raster layers the failure of a specimen is due to inter-layer fracture whereas in 45° raster layers, specimen fails under both inter-layer and intra-layer fracture. Dawoud et al. [15] investigated the effect of raster angle and raster air gap on mechanical strength of FDM parts and found that negative air gap significantly improved the mechanical strength of FDM parts. The influence of part build direction and raster orientation was investigated on tensile strength, tensile modulus and elongation of FDM samples [16]. The relationship between failure and mechanical properties are derived through fractographic analysis. Aforementioned literature review reveals that several authors have made attempt to determine the optimal process parameter settings to obtain desirable mechanical strength of FDM parts. The FDM parts, may it be the functional parts or prototypes, behave differently under loading condition and depend upon the layer thickness and alignment of the rasters with respect to loading direction [3]. Adaptive layer thicknesses [17], depending upon part shapes and structure, could optimize the surface roughness but may affect the strength of the build parts non-uniformly. Few authors have also studied the effect of process parameters on the performance of FDM samples through finite element methods [7, 9] however, finite element models that accounts for the layer pattern with their orientation and thickness can effectively predict the behavior of the FDM specimens. Thus, the present article is aimed at providing an insight towards the behavior during tensile failure of FDM samples built at different raster angles and layer thicknesses. In order to study the elasto-plastic behavior of the individual raster, their orientation, layer thickness, realistic FE models are developed that include the layers of different thicknesses (0.178 mm, 0.254 mm, 0.330 mm) and rasters at different angles (0°, 90°, 0°/90°) maintaining the inter-layer and intra-layer bonded region and subsequently results are validated with experiments. Later, fractographic analyses are carried out using optical microscope, 3D optical profiler and scanning electron microscopy (SEM) to study the failure modes of FDM samples.
2. Materials and Methods 2.1 Finite element modeling In the present work, finite element modeling and simulation are carried out for acrylonitrile butadiene styrene (ABS) FDM specimen using finite element (FE) package ABAQUS. The samples are designed at three different layer thicknesses (0.178, 0.254 and 0.330 mm) and three different raster angles (0°, 90° and 0°/90°) using Pro/Engineer 5.0 CAD modeling software. The created models are exported to ABAQUS in STEP file format. The overall dimensions for tensile test specimens are decided as per ASTM D638 standard (standard for tensile testing for plastic components) [18]. The detailed dimensions of the tensile specimen and raster angle arrangements are schematically shown in Fig. 1. Tensile specimen modeled for finite element analysis Loading direction
R14
19 4
0º R25 5 90º All dimensions are in mm Fixed gripper
(b)
(a)
0º/90º (c)
Fig.1. (a) Schematic of gripper arrangement, (b) tensile specimen as per ASTM D638 standard, (c) schematic illustration of different raster angles The geometric model of the tensile specimen is prepared by generating the layers at different raster angles for different layer thicknesses. The layer thickness (
) implies the nozzle diameter through
which the material is deposited. Instead of using the theoretical circular cross section of layers, the layers after depositing owns an elliptical cross section with certain amount of intra-layer necking. The growth of neck is a time dependent diffusion driven phenomenon attaining the final shape after the model is fully solidified and this has been validated by the optical microscopic visualization. Moreover, there is also some overlap region between two adjacent raster layers also known as interlayer necking which occurred due to diffusion of two raster layers at interface. General increase in necking with layer thickness has been observed from the microscopic measurements. The shape of individual rasters and overlap region observed under microscope can clearly be seen in Fig. 2. Using
the optical microscope (STM6, Olympus, Nagano-Ken, Japan, 0.5 µm least count), the height of the filament (h) is measured as 0.160 mm, 0.227 mm and 0.296 mm for layer thickness 0.178 mm, 0.254 mm and 0.330 mm. The other dimensions i.e width of the filament (w) is measured as 0.370 mm, 0.525 mm and 0.684 mm and inter-layer neck (a) region as 0.07 mm, 0.10 mm and 0.13 mm for different layer thickness (0.178 mm, 0.254 mm and 0.330 mm respectively). Thus, the dimensions, raster cross section and necking region are measured using the optical measuring microscope and used to generate a more realistic geometric model of the FDM test parts. To build the entire model for the tensile test specimen, first the cross section is created using the dimensions obtained from the microscopic measurements and then 465 to 9423 numbers of patterns (depending upon layer thickness and raster angle) are generated to prepare a rectangular model. Finally the model is trimmed as per the exact dimension of the test specimen (refer Fig. 1(b)). In order to reduce the number of elements, nodes and save computational time, the regions under the grip (in the tensile testing machine) of the specimen is not modeled (depicted schematically in Fig. 1). Air gap (0 mm) h0 During initial layer deposition
Neck growth and diffusion at interface
Intra-layer neck formation at interface
Inter-layer necking
Air voids
Intra-layer necking
Fig. 2. Necking formation between two adjacent raster layers, where a, w and h stands for necking region length, width of filament and height of filament The modeled FDM samples are imported to ABAQUS platform for the FE simulation of the elastic-plastic deformation under the uniaxial tensile loading. The FE simulation is carried out with implicit scheme and modeling it as load/displacement problem where the tensile load or the observed total displacement till fracture is applied along the longitudinal axis (length of the specimen). The material used for FDM model is assumed to be homogeneous and isotropic in nature. The material properties (stress-strain relationship) of ABS model material are derived from tensile test conducted by Anton et al. [19] for solid ABS samples. The tensile behavior of the FDM sample built with different layer thickness and/or raster orientation are different from the behavior and property of solid ABS material. However, while modeling by FEM, the material property of the ABS material is to be
defined thus the stress-strain data of solid ABS sample are considered from reference [19].The other material properties used for the analysis are density ρ = 1.04 g/cm3, Young’s modulus E = 2.2 GPa and Poison’s ration ν = 0.35. In explicit FE analysis the results may deviate from the actual if numbers of increment are not sufficient and may have continuous stability problems (conditionally stable), whereas implicit analysis often provides better results for elasto-plastic analysis although it requires larger computational time and space. Finer mesh size improves accuracy but demands higher computational resources thus the model is meshed by tetrahedron elements keeping the mesh size fine enough so that each raster is meshed properly for different layer thickness and raster orientations, and contains nodes in thickness directions also (as shown in zoomed view of the meshed model in Fig. 3). Continuum 3 dimensional 4 noded solid linear tetrahedron (geometrically versatile) element C3D4 (each node with 3 degrees of freedom) is used to mesh the specimen. The complete meshing of the model and 3D tetrahedron element is shown in Fig. 3.
N3
Inter-layer bonding
N4
N2 N1
Intra-layer bonding 0º
90º
Fig. 3. Representation of workpiece meshing and C3D4 tetrahedron mesh element The boundary conditions applied in the current FE model are shown in Fig. 4. As carried out during the tensile testing under uniaxial loading, one end of the work piece (or the model) is kept fixed and load and/or displacement is applied from another end (schematically represented in Fig. 1(a)). Fig. 4 shows the boundary condition where one end of the workpiece is encastred by constraining all six degree of freedoms, U1 = U2 = U3 = UR1 = UR2 = UR3 = 0. The displacement along longitudinal axis (z-axis in Fig. 4) is applied from the other end and displacement/ rotation from all
other sides are fixed, U1 = U2 = UR1 = UR2 = UR3 = 0, U3 = amount of displacement noted during failure test.
U1 = U2 = UR1 = UR2 = UR3 = 0, U3 = 0
U1 = U2 = UR1 = UR2 = UR3 = 0, U3 = - 2.4 mm
Fig. 4. Applied boundary conditions for FDM tensile specimen 2.2 Experimentation In the present work, all the parts are fabricated on FDM UPrint SE Plus and Mojo machines (Stratasys, Eden Prairie, USA) using ABS P430TM model material. The CAD models are designed using modeling software and converted into .STL file for slicing, layer path generation and building of the components. All specimens are built by depositing the materials in form of semi molten state extruded through a liquefier nozzle in form of layers with thickness of 0.178 mm, 0.254 mm or 0.330 mm and with constant raster angle 0°/90°. The dimensions for the tensile specimens are decided as per ASTM D638 standard [18] as shown in Fig. 1.
(a) (b) Fig. 5. Gripper arrangement and tensile specimen (a) before testing, and (b) after testing The tensile testing of the specimens are conducted on a Zwick/ Roell Z050 (Zwick Roell, Ulm, Germany) ultimate testing machine (UTM) equipped with 50 kN load cell and testing are conducted at a crosshead speed of 5 mm/min according to ASTM standard. The FDM sample fabricated for tensile testing, gripped in machine jaw for testing as shown in Fig. 5(a) and fractured specimen after testing is shown in Fig. 5(b). The optical measuring microscope and 3D optical profiler (Zegage, Zygo, Middlefield, CT, USA) are used to analyze the fracture surface of FDM samples to study the mode
and nature of raster failure when specimen built at different layer thickness and raster angles. Fractographic analyses are also carried out by studying the fracture surfaces using scanning electron microscopes at different magnifications.
3. Results and Discussions The FE simulations are carried out on Windows© based systems using ABAQUS for different models created with different raster angles and layer thicknesses. From the output database file, the results are extracted by creating path by joining the nodes along a certain direction. The simulated results are subsequently validated and compared with experimental observations. Later fractographic studies are also carried out to analyze the mode of failure of the FDM samples. 3.1 Finite element analysis The stress distribution on the samples with three different layer thicknesses and 0 º raster angles is shown in Fig. 6. The peak stresses developed in the samples are in the order of 39 MPa and small amount of necking region can be observed within the gage length region. Higher peak pressure is observed for 0.178 mm sample (39.2 MPa) which first decreases with increase in layer thickness (38.2 MPa for 0.254 mm layer thickness) and then again increases (38.8 MPa for 0.330 mm layer thickness). However an average stress of 34 MPa is noted in and around the neck region. This behavior shows that in the specimen with 0° raster angle, layers are aligned parallel to the loading direction (along the longitudinal or length of specimen). For specimen with 0.178 mm layer thickness the numbers of fine layer in the loading direction are more and load is borne by each layer with noticeable amount of individual necking. On the contrary, for 0.330 mm layer thickness numbers of layers are less but the amount of intra-layer necking region is high thus more close to behave like solid metal thus produces more strength to the specimen. A noticeable amount of necking both in width and thickness i.e. diffuse and localized necking are verified from the displacement plot of the FE results. The variations of stresses along the length of the specimen for all the samples with different layer thicknesses are shown in Fig. 6. The stress variations are captured along the three different paths (Path 1, 2 and 3) on the top, middle and bottom surface of the specimen joining the nodes in sequence. For the sample with 0.178 or 0.254 mm layer thickness, the maximum stress and the necking is observed close to the end of gage length whereas for sample with 0.330 mm layer thickness the maximum stress is distributed over the almost entire gage length and necking is observed on both sides of the gage length. An increase in maximum stress magnitude is observed with increase in layer thickness from 0.254 mm to 0.330. This increase in stress at 0.330 mm layer thickness specimen could be due to increase in overlap region and subsequent bonding strength between adjacent rasters at higher layer thickness. Moreover, with increase in layer thickness, number of layers required to fabricate the specimen decreases which further decreases the number of air voids formed at interstitial sites thus
load bearing capacity of the specimen increases. This air voids are major source for crack initiation and propagation in the specimen therefore, desirable to have lesser number of air voids. Stress distribution pattern for 0.330 mm layer thickness also shows that stress is more uniformly distributed over entire narrow region (gage length) due to strong bonding between adjacent rasters (observation is found similar as that of analysis for a solid model, discussed later in Fig. 8) whereas in sample with layer thickness 0.178 or 0.254 mm, stress is more concentrated towards the end of the gage length where the area starts decreasing. 45 Gage length
40 35
Path 1 30
Path 2 Path 3
Stress (MPa)
25 20 Path 1
15 Fixed end
10
Top layer
Path 2
Pulling direction Path 3
5 Middle layer
Bottom layer
0
0
10
20
(a)
30 40 50 Distance along path (mm)
60
70
80
45 Gage length
40 35
Path 1
30
Path 2 Path 3
Stress (MPa)
25
20 Path 1 15 Top layer
Fixed end
10
Path 3
Path 2
Pulling direction
5
Bottom layer
Middle layer 0 0
(b)
10
20
30 40 50 Distance along path (mm)
60
70
80
45 Gage length
40 35
Path 1
Stress (MPa)
30
Path 2 Path 3
25
20 Path 1 15 Fixed end
10
Pulling direction
Top layer Path 2
Path 3
5
Middle layer
Bottom layer
0 0
(c)
10
20
30 40 50 Distance along path (mm)
60
Fig. 6. Stress-displacement and stress distribution pattern for specimen built at 0° raster angle with layer thickness (a) 0.178 mm, (b) 0.254 mm, (c) 0.330 mm
70
80
The stress-displacement curve for specimen with 90° raster angle is shown in Fig. 7. For 90° raster angle specimen, layers are laid perpendicular to the loading direction thus failure of the specimen mainly depends upon adhesion between adjacent rasters (intra-layer bonding). During tensile loading/displacement applied perpendicular to raster alignment, the stress is concentrated to narrow region in and around the intra-layer bond region which causes the failure of the specimen due to delamination/ separation of layers from bonded region. The sudden increase in stress, concentrated to small region can also be seen form Fig. 7 where sharp peaks within the gage length region indicates high stress concentration region.
45 Gage length
40 35
Path 1
Stress (MPa)
30
Path 2 Path 3
25 20 Path 1
15 Fixed end
10
Top layer
Path 2
Pulling direction Path 3
5 Middle layer
Bottom layer
0
(a)
0
10
20
30 40 50 Distance along path (mm)
60
70
80
45 Gage length
40 35
Path 1
Stress (MPa)
30
Path 2 Path 3
25 20 Path 1
15 Fixed end
10
Pulling direction
Top layer Path 3
Path 2
5 Bottom layer
Middle layer 0 0
(b)
10
20
30 40 50 Distance along path (mm)
60
70
80
45 Gage length
40 35
Path 1
Stress (MPa)
30
Path 2 Path 3
25 20 Path 1
15 Fixed end
10
Top layer
Path 2
Pulling direction Path 3
5 Middle layer
Bottom layer
0
(c)
0
10
20
30 40 50 Distance along path (mm)
60
70
Fig. 7. Stress-displacement and stress distribution pattern for specimen built at 90° raster angle with layer thickness (a) 0.178 mm, (b) 0.254 mm, (c) 0.330 mm
80
It can be seen from Fig. 7 that amount of peak stress and distribution in and around the neck regions is similar for all specimens which predict that for 90° raster specimen failure is not significantly dependent on layer thickness. This is due to the fact that specimen fails at intra-layer bond and number of layers or air voids does not play much significant role and 90° raster specimen fails due to delamination and separation of layers from intra-layer bond region.
45 Gage length
40 35
Stress (MPa)
Path 1 30
Path 2
25
Path 3 Path 4
20 15
Path 1
10
Fixed end
Pulling direction Path 4
Top layer (0 ) Path 2
5
Path 3
Middle layer (90 )
Middle layer (0 )
Bottom layer (90 )
0
0
(a)
10
20
30 40 50 Distance along path (mm)
60
70
80
45 Gage length
40 35
Stress (MPa)
Path 1 30
Path 2
25
Path 3 Path 4
20 15
Path 1
10
Fixed end
Path 2
5
Middle layer (90 )
0
0
(b)
10
Pulling direction
Top layer (0 ) Path 3
20
Path 4
Middle layer (0 )
Bottom layer (90 )
30 40 50 Distance along path (mm)
60
70
80
45 Gage length
40 35
Path 1
Stress (MPa)
30
Path 2
25 20 Pulling direction
15 Fixed end
10
Path 1
Path 2
5 Top surface
(c)
Bottom surface
0
0
10
20
30 40 50 Distance along path (mm)
60
70
Fig. 8. Stress-displacement and stress distribution pattern for specimen built at 0°/90° raster angle with layer thickness (a) 0.254 mm, (b) 0.330 mm, and (c) solid tensile specimen
80
The stress distribution and variations of stresses for sample with 0°/90° raster (alternate layers with raster at 0º and 90º) is shown in Fig. 8. In 0°/90° raster specimen layers are laid in two orientations 0° and 90° alternately and show different behavior for 0° raster angle and 90° raster angle layers. From Fig. 8 it can be noticed that for path 1 and path 3 (layer with raster at 0°), average as well as peak stress is considerably higher as compared to path 2 and path 4 i.e. in the layers where rasters are laid in 90° to the length of the specimen. The rasters at 0º (to the length of the specimen) are aligned parallel to loading direction and under tensile loading these rasters continues to get pulled exhibiting a small amount of necking however, 90° raster layers are aligned perpendicular to the loading direction thus produces weak direction. It can also be seen that the overall stress generation lies between 0° and 90° raster specimen because the premature bond separation of 90° raster layers causes the earlier failure of 0° raster layers. For 0º/90º raster samples the load bearing capacity of the specimen increases with layer thickness due to increase in overlap region between adjacent layers. It may be also noted from Fig. 8 that the average stress variation in 0.330 mm layer thickness specimen is alike the variation as noted for the solid specimen (equivalent to specimen receives from injection molding processes) with only difference that necking is observed for solid sample on both sides of the gage length. Thus the stress variation and distribution reveals that during the failure first the 90º rasters get separated from the intra-layer bond region (fracture eventually starts from these regions) but 0º rasters pulled and experience diffused necking and finally brittle failure of the 0º rasters lead to the fracture of the specimen. Elemental stress development for the three different samples with 0º, 90º, 0º/90º rasters and solid samples are represented in Fig. 9. The element is considered at the highly stress generated region of FE model (neck region) with different layer thickness and raster angles. Elemental stress measurement is important to study the behavior of individual element under uniaxial tensile loading. It may be noted that experimentally it is difficult to measure/capture the elemental stress thus finite element results are extracted to study the development of stress on an element when tensile loading is applied. It can be seen from Fig. 9(a) that for 0° raster, yielding starts earlier for samples of 0.178 mm layer thickness as compared to higher layer thickness. For all the samples with 0º raster, where the individual small rasters are aligned with loading direction, after short duration of yielding diffuse necking starts. Due to the low plastic region of the ABS materials strain hardening no longer able to compensate the weakening due to the reduction in cross sectional area. On the contrary, for 90º rasters an almost similar behavior is observed for all samples with different layer thicknesses. Rather than a clear yielding the deformation starts and irrespective of layer thickness the separation of layers from intra-layer bond region takes place. The variation of elemental stress for 0º/90º rasters of different layer thickness and solid samples are shown in Fig. 9(c). Elements are taken from both the 0° and 90° raster layers and stress developments are shown in the Fig. 9(c). After the yielding region a quick drop in elemental stress is noted for 90º rasters of 0.254 mm layer thickness indicating the initiation of failure at such rasters. However the same is delayed for 0.330 mm layer thickness samples due to the
fact of thicker intra-layer bonding. The loss of homogeneous state of deformation and subsequent necking leading to failure is observed for all the specimens but the failure in 0.330 mm layer thickness
45
40 45
40 45
35 40
35 40
30 35
30 35
25 30
Stress (MPa) Stress (MPa)
45
0.178 mm
20 25 15 20 10 15 5 10 0 5 0
0.254 mm 0.330 mm 0.178 mm
0.254 mm 0.330 mm 0.2
0.4
0.6
0.8
1
Time step
0 0
0.5
1
25 30 20 25 15 20 10 15 5 10 0 5 0
0.178 mm
0.254 mm 0.330 mm 0.178 mm
0.254 mm 0.330 mm 0.2
0.4
0.5
1
2
2.5
0.6
0.8
1
2
2.5
Time step
0
1.5
0
Displacement (mm)
1.5
Displacement (mm)
(b)
(a) 45 40 45 35 40 30 35
Stress (MPa) Stress (MPa)
Stress (MPa) Stress (MPa)
sample occurs later as compared to 0.254 mm thickness.
25 30 20 25 15 20 10 15 5 10 0 5 0
Solid specimen 0.254 mm 0.178 mm 0.330 mm 0.254 mm 0 raster layer 0.330 mm 90 raster layer 0.2
0.4
0.6
0.8
1
2
2.5
Time step
0 0
0.5
1
1.5
Displacement (mm)
(c)
Fig. 9. Elemental stress development during tensile loading of FDM specimen built with (a) 0° raster layers, (b) 90° raster layers, and (c) 0°/90° raster layers 3.2 Experimental validation Tensile testing for samples with three different layer thickness and 0°/90° rasters are conducted experimentally. For each specimen built at different layer thickness (0.178 mm, 0.254 mm and 0.330 mm) and 0°/90° raster angles experimentally obtained stress-strain curves are shown in Fig. 10. From
Fig. 10 it can be noted that with increase in layer thickness tensile strength first decreases and then increases. The similar behavior is also observed from FE analysis and discussed in previous subsection. This phenomenon is mainly because in 0.330 mm layer thickness number of layer decreases thus air voids, distortion and shrinkage is less as compared to other layer thickness. Moreover, for 0.330 mm layer thickness the overlap region between adjacent layers is more than the other layer thickness. Therefore, the loading bearing capacity of the specimen increases and more stresses is observed in the specimen as compared to other layer thickness (0.254 mm). In 0.178 mm layer thickness even though large numbers of layers are required to build the specimen but still it exhibit more tensile strength than 0.330 mm layer thickness. This is because in 0.178 mm layer thickness relatively more numbers of layer are aligned along the loading direction compared to other layer thickness. Thus during tensile testing more number of layers are pulled and breaks under tensile loading. With increase in layer thickness the yielding is shifted (takes place at higher strain, indicates more elastic region) and strain at failure is also increased.
35 30
Stress (MPa)
25 20
15
0.178 mm 0.254 mm
10
0.330 mm 5
0 0
0.5
1
1.5
2 2.5 3 3.5 Strain (%)
4
4.5
5
Fig. 10. Stress-strain curve for specimen built at 0°/90° raster angle and with different layer thickness Photographic view of the failure specimen and the stress distribution as obtained from the FE analysis is shows (Fig. 11) as a sample representation for 0.178 mm, 0.254 mm and 0.330 mm layer thickness. The failure in the sample after tensile testing took place close to the end of the gage length marked on the specimen. Corresponding stress distribution also shows that the maximum stress and necking occurred towards the end of the gage length region. Fig. 11(a) also shows the stretched marks (indicative of stressed regions) on fabricated specimen of 0.178 mm layer thickness under uniaxial tensile loading which is equivalent to stress distribution pattern in FE model. This stretched mark also shows that stresses are more uniformly distributed over entire gage length in 0.178 mm layer
thickness sample. The more number of fine layers in the loading direction and strong bonding between the adjacent rasters increases the loading bearing capacity of each raster under uniaxial tensile loading. The small amount of necking is also observed towards the end of the gage length where the area starts decreasing. On the contrary, for 0.254 mm and 0.330 mm layer thickness numbers of layers are less as compared to specimen with 0.178 mm layer thickness but the amount of intra-layer necking region is high thus stresses are more concentrated to narrow region i.e. towards the end of the gage length (refer Fig. 11(b) and (c)).
Gage length
Gage length
Stretch marks during tensile loading
Necking region
(a) Gage length Gage length
Necking region
(b)
Gage length Gage length Necking region
(c) Fig. 11. Failure specimen after tensile testing and the stress distribution for sample built at 0º/90º raster angle and layer thickness (a) 0.178 mm, (b) 0.254 mm, and (c) 0.330 mm
3.3 Fractographic analysis Cross sectional views of the samples at the maximum stress regions for all the samples at different raster angle and layer thickness along with the solid sample are shown in Fig. 12. Fig. 12 shows that the maximum stress generated in the neck region is distributed almost over the entire cross sectional area when layer thickness is 0.178 mm as well as 0.330 mm as in 0 º raster angles each layer is pulled under tensile loading whereas for 0.254 mm layer thickness the maximum stresses are distributed mostly over the peripheral regions. For 90º raster angles since layers are laid perpendicular to loading direction less stresses are generated and failure of specimen mainly depend upon adhesion between successive layers. Fig. 12 also shows that failure of specimen started from outer layers and moves towards the center of the workpiece for all samples. However, for 0º/90º raster angle specimen less stresses are generated in 90º raster layers and comparatively higher stresses in 0º raster layers can be easily noted. With increase in layer thickness more stresses are generated in each specimen due to increase in overlap region between adjacent layers and layers are more strongly bonded together and require higher force to cause failure of the specimen.
Layer thickness 0.178 mm
Layer thickness 0.254 mm
Layer thickness 0.330 mm
(a) 0° raster angle
Layer thickness 0.178 mm
Layer thickness 0.254 mm
Layer thickness 0.330 mm
(b) 90° raster angle 0° rasters
90° rasters Solid specimen
Layer thickness 0.254 mm
Layer thickness 0.330 mm
(c) solid specimen and 0°/90° raster angle
Fig. 12. Cross sectional surfaces after FE simulations of samples with different raster angles at different layer thicknesses and solid sample
Fractographic analysis of built specimens at different layer thickness is shown in Fig. 13. It can be easily noted from the Fig. 13 that the 0° raster layers first undergoes pulling followed by small necking and subsequently fails due to brittle fracture with tearing of individual raster layers. Some air voids present between adjacent raster layers decreases the strength of the specimen. For 90° raster layers quick delamination from the bonded region between adjoining layers take place.
Air gap
(a)
Delamination failure for 90° raster layers
(b)
Tearing and brittle fracture of 0° raster layers (c)
Fig. 13. Fractographic analysis of FDM sample built at 0°/90° raster angle with layer thickness (a) 0.178 mm, (b) 0.254 mm, and (c) 0.330 mm
Fig. 13 also reveals that with increase in layer thickness the overlap region between adjoining layers increases (maximum in 0.330 mm layer thickness). The 3D optical profiler image for 0° raster layers indicates that the 0° raster layers fails under pulling and rupture of raster fibers and numerous microhills indicate micro-pulling of each raster fiber within a layer of a material which also indicate good tensile strength for 0° raster specimen as shown in Fig. 13. For 90° raster layers material fails under delamination with small region of rupture/ brittle fracture which indicate relatively low tensile strength as compared to 0° raster layers. For 0º rasters unevenness in failure surface indicates the pulling and necking of the fibers (refer Fig. 14) and a significant amount of stepped failure region in 0º raster for relatively higher layer thickness can also be seen in Fig. 14(b) (inset). However, mostly co-planar separation with micro-hills in 90º rasters indicate the bond separation from the adhered regions (refer Fig. 14).
(a)
(b)
Fig. 14. Failure surface patterns for 0° and 90° raster layers for samples build with layer thickness (a) 0.254 mm, and (b) 0.330 mm
Additionally, few parts are fabricated at different part orientations (X, Y and Z orientation). Building of parts on thickness face (horizontally), width face and thickness face (vertically) indicates X, Y and Z orientations respectively. Buildings of all previous samples used for tensile testing are done about Y orientation for which the results and fractographic analysis are presented before. SEM images of the fracture surfaces of the samples build at X0 (X orientation, 0º raster), Z90, X60 are shown in Fig. 15. The maximum tensile strength was achieved for X and Y orientations whereas for Z orientations strength is inferior due to weak bonding between adjacent rasters [20]. For 0° raster layers in X or Y orientation, layers are laid parallel to loading direction more stresses are generated in each layers as compared to 90° raster layers thus 0° layers fails under brittle fracture i.e. tearing and rupture of individual rasters. Fig. 15(a) shows the fracture surface of FDM sample fabricated at X orientation where it can clearly be seen that 0° raster layers separated by tearing and rupture of individual layers.
Fig. 15. SEM images of tensile failure sample built at (a) X0, (b) Z90, and (c) X60 In Z orientation both 0°/ 90° raster layers are of very small size and aligned perpendicular to the loading direction therefore strength of the part depend upon adhesion between successive layers and specimen fails due to delamination or separation of layers form weak bonding. Fig. 15(b) shows the corresponding fracture surface where layers separation at the weak interlayer bond can be easily seen thus provides the lowest tensile strength (14.8 MPa in comparison of 34.5 or 32.2 MPa for X, Y orientation). For X and Y orientations when specimen is fabricated either with rasters at 30° or 60° grain pullout, necking and tearing takes place (refer Fig. 15(c)) thus highest tensile strength are possible to achieve. Therefore, it is desirable to fabricate the FDM parts either in X orientation or Y orientation rather than the Z orientation. Higher strength of the samples would be achieved for 0° raster layers as compared to 90° raster layers however, layers at 30° or 60° raster layers could also be preferred and parametric optimization would be needed to decide the orientation and raster angle so as to achieve the best possible mechanical strength of the FDM parts.
4. Conclusions In the preset work, FE modeling and simulation are carried out developing realistic models that considers layers of different thicknesses and rasters at different angles maintaining the inter-layer and
intra-layer bonded region. Experimental studies, fractographic analysis are also performed in order to validate the results. From the results the following conclusions are drawn. FE analysis indicates that necking is present in the 0º rasters and strain at failure and elongation increases with increase in layer thickness whereas tensile stress first decreases with increase in layer thickness then increases. More number of layers in loading direction, particularly when layer thickness is small, contributes to the load bearing and more stress generation whereas the higher layer thickness provides more amount of intra-layer bonded region thus provides higher tensile strength. Failure first takes place through intralayer separation in rasters 90º to the loading direction followed by the pulling and tearing of the 0º rasters along with the inter-layer (between 0º-90º rasters) bond separation. Fractographic analysis indicates that 0° raster layers fails under brittle fracture with tearing and rupture of individual layers. In 90° raster layers failure of the specimen occurred due to delamination or separation of layers from the adjacent bonds. Higher strength of the components are possible to achieve when specimen is fabricated either in X or Y orientations whereas inferior strength is obtained for Z orientation parts. In case of raster angles 30° or 60°, shearing of individual layers adds up to the contribution of higher tensile strength in addition to the pulling, rupture and bond failure. Numerous micro-hills are observed on the failure surfaces of the specimens that indicate micro-pulling of each raster fiber within a layer of material and contribute to the improved tensile strength of the specimen.
Acknowledgement Authors thankfully acknowledge the support and help received from Mr. Sandeep Shukla of M/s DesignTech Systems, New Delhi, India.
References 1.
2. 3.
4. 5. 6.
7.
Sood AK, Ohdar RK, Mahapatra SS. Parametric appraisal of fused deposition modelling process using the grey taguchi method. Proc IMechE, Part B: J Engineering Manufacture 2010; 224:135– 45. Ahn SH, Montero M, Odell D, Roundy S, Wright PK. Anisotropic material properties of fused deposition modeling ABS. Rapid Prototyping J 2002; 8:248–57. Garg A, Bhattacharya A, Batish A. Failure investigation of fused deposition modelling parts fabricated at different raster angles under tensile and flexural loading. Proc IMechE, Part B: J Engineering Manufacture 2015; http://dx.doi.org/10.1177/0954405415617447. Garg A, Bhattacharya A, Batish A. On surface finish and dimensional accuracy of FDM parts after cold vapor treatment. Mater Manuf Process 2016; 31:522–29. Sood AK, Ohdar RK, Mahapatra SS. Parametric appraisal of mechanical property of fused deposition modelling processed parts. Mater Des 2010; 31:287–95. Croccolo D, Agostinis MD, Olmi G. Experimental characterization and analytical modelling of the mechanical behaviour of fused deposition processed parts made of ABS-M30. Comput Mater Sci 2013; 79:506-18. Espin MD, Forcada JMP, Granada AAG, Llumà J, Borros S, Reyes G. Mechanical property characterization and simulation of fused deposition modeling polycarbonate parts. Mater Des 2015; 83:670–77.
8. 9.
10. 11.
12.
13. 14. 15. 16.
17. 18. 19.
20.
Huang B, Singamneni S. Raster angle mechanics in fused deposition modelling. J Compos Mater 2015; 49:363–83. Zhang Y, Chou K. A parametric study of part distortions in fused deposition modelling using three-dimensional finite element analysis. Proc IMechE, Part B: J Engineering Manufacture 2008; 222:959–68. Bellehumeur C, Li L, Sun Q, Gu P. Modeling of bond formation between polymer filaments in the fused deposition modeling process, J Manuf Process 2004; 6:170-78. Ravari MRK, Kadkhodaei M, Badrossamay M, Rezaei R. Numerical investigation on mechanical properties of cellular lattice structures fabricated by fused deposition modeling. Int J Mech Sci 2014; 88:154–61. Es-Said OS, Foyos J, Noorani R, Mendelson M, Marloth R, Pregger BA. Effect of layer orientation on mechanical properties of rapid prototyped samples. Mater Manuf Process 2000; 15:107–22. Gurrala PK, Regalla SP. Multi-objective optimization of strength and volumetric shrinkage of FDM parts. Virtual Phys Prototyp 2014; 9:127–38. Gurrala PK, Regalla SP. Part strength evolution with bonding between filaments in fused deposition modelling. Virtual Phys Prototyp 2014; 9:141–49. Dawoud M, Taha I, Ebeid SJ. Mechanical behaviour of ABS: An experimental study using FDM and injection moulding techniques. J Manuf Process 2016; 21:39–45. Riddick JC, Haile MA, Wahlde RV, Cole DP, Bamiduro O, Johnson TE. Fractographic analysis of tensile failure of acrylonitrile-butadiene-styrene fabricated by fused deposition modeling. Addit Manuf 2016; 11:49–59. Pandey PM, Reddy NV, Dhande SG. Slicing procedures in layered manufacturing: a review. Rapid Prototyping J 2003; 9:274–88. ASTM D638-14. Standard test method for tensile properties of plastics. West Conshohocken, PA: ASTM International; 2014. Anton S, Thomas A, Markus G, Martin F, Wagner M, Hierl S. Finite element analysis of thermoplastic probes under tensile load using LS-DYNA compared to ANSYS WB 14 in correlation to experimental investigations. ANSYS Conference & 30th CADFEM Users’ Meeting 2012. Garg A, Bhattacharya A, Batish A. Chemical vapor treatment of ABS parts built by FDM: Analysis of surface finish and mechanical strength. Int J Adv Manuf Technol 2016; http://dx.doi.org/10.1007/s00170-016-9257-1.
Highlights Realistic FE models for different layer thickness and raster angles maintaining inter-layer and intra-layer bonded region The developed stress, strain at yield, elongation and tensile strength first decreases with layer thickness and then increases. The 0° raster layers fail under pulling and rapturing of fibers and presence of micro-hills indicates micro-pulling of each raster fiber within a layer.
For 90° raster layers material fails under delamination and bond separation.