Академический Документы
Профессиональный Документы
Культура Документы
review
Abstract
This article provides a literature review of finite element simulation studies for metallic powder bed additive manufactur-
ing processes. The various approaches in the numerical modeling of the processes and the selection of materials proper-
ties are presented in detail. Simulation results are categorized according to three major findings’ groups (i.e.
temperature field, residual stresses and melt pool characteristics). Moreover, the means used for the experimental vali-
dation of the simulation findings are described. Looking deeper into the studies reviewed, a number of future directions
are identified in the context of transforming simulation into a powerful tool for the industrial application of additive man-
ufacturing. Smart modeling approaches should be developed, materials and their properties should be further character-
ized and standardized, commercial packages specialized in additive manufacturing simulation have to be developed and
simulation needs to become part of the modern digital production chains. Finally, the reviewed studies are organized in a
table and characterized according to the process and material studied, the modeling methodology and the experimental
validation method used in each of them. The key findings of the reviewed studies are also summarized.
Keywords
Additive manufacturing, finite element analysis, residual stresses, distortion
industrial practice. AM processes involve a high level almost every engineering discipline.14 The FEM has
of complexity and the physical mechanisms acting dur- also proved to be very popular in the study of the physi-
ing the process are not yet well understood.5 Thus, the cal phenomena during AM processes and their para-
parts produced using AM processes are often not ful- metric optimization, as it is a resource efficient
filling the mechanical performance requirements set by methodology,16 which can save a large number of phys-
manufacturers or exhibit difficulty to predict distortion ical experiments from being conducted.17 Thus, FEM
behavior.6 This fact, along with long lead times,7 can has eventually become the most common technique for
be considered as the main reason preventing the total predicting the porosity, the residual stresses and the dis-
transition to AM when it comes to the production of tortion of parts produced by AM processes. Other
metallic parts. Poor mechanical performance can be numerical methods used include the Lattice Boltzmann
attributed to the remaining porosity and the presence method (LBM)5,18–20 and the finite volume method
of thermally induced residual stresses and distortion in (FVM),21,22 in studies that take into account the fluid
the finished parts.8 Porosity is generated by instabilities dynamics of the melt pool. The ultimate goal of studies
during material consolidation.9 That phenomenon, in this field is adjusting the process parameters for
which is commonly referred in the literature as ‘‘bal- achieving the desired mechanical integrity and reducing
ling,’’9 does not allow a full surface contact between the number of faulty parts, toward minimizing manu-
melted material particles, leading to parts with lower facturing costs. The literature studies in the field of
density than the respective solid material and therefore modeling and simulation of other types of manufactur-
decreased mechanical strength.10 On the other hand, ing processes already exist.12,23–25 In the field of simu-
residual stresses and distortion are generated by the lating AM processes, the only existing review article is
thermo-mechanical nature of the process itself and the dealing with thermal analysis methods in SLS and
thermal gradients created when building the part.11 SLM.26 The aim of this article is to provide an overview
Residual stresses and distortion affect significantly the of the finite element numerical modeling and simulation
mechanical properties and cause dimensional inaccura- of the AM processes utilizing a metallic powder bed.
cies to the final part. An overview of the current issues The structure of this article is organized as follows.
in the AM industrial practice is illustrated in the dia- Section ‘‘Numerical modeling’’ describes how modeling
gram of Figure 2. for simulation is performed. Details about mathemati-
Numerical methods can be classified into mesh- cal modeling, meshing, simulation techniques, heat
based methods and mesh-free methods.12 The finite ele- source models and temperature-dependent material
ment method (FEM) belongs to the first class and was properties implementation into the model are provided.
originally introduced by Turner et al.13 In the FEM, a Section ‘‘Simulation results’’ presents the simulation
continuum of matter (domain) is discretized into a finite results of the reviewed studies. The results are categor-
number of elements forming a mesh, thus reducing the ized by subject into three groups: (1) temperature his-
problem to that of a finite number of unknowns.14 tory, (2) residual stresses and distortion and (3) melt
Each finite element possesses simpler geometry and pool characteristics. The means of experimental verifi-
therefore, it is easier to analyze than the actual struc- cation of the simulation results are explained in section
ture.15 The FEM has proven to be an essential tool for ‘‘Experimental validation’’ and the relevant applica-
solving boundary, initial and eigenvalue problems in tions are reported. In section ‘‘Challenges,’’ the current
98 Proc IMechE Part B: J Engineering Manufacture 231(1)
challenges in the use of finite element analysis (FEA) in explicit or implicit methods can be used. If t is the cur-
AM process simulation are listed and already imple- rent time, the state of the system yðt þ DtÞ at a later
mented, or potential solutions are presented. Finally, time would be calculated as a function of the current
this study is summarized and a number of concluding state yðtÞ by an explicit method
remarks are given in section ‘‘Conclusion.’’
yðt þ DtÞ ¼ fðyðtÞÞ ð1Þ
where K is the thermal conductivity, T is the tempera- Figure 4. Physical mechanisms during additive processes.31
ture of the part, q_ is the rate at which heat is supplied to
the system, r is the density of the material, Cp is the spe-
cific heat capacity and t is the interaction time between In the general case, where the heat supplied by the
the beam and the material. beam is modeled as a heat flux and heat is lost from the
In order to capture the solid–liquid transformations, system by convection and radiation, the boundary con-
the above equation can be modified so that the enthalpy dition for the free surfaces of the part is33
change is taken into account. The enthalpy change dH
is given by k ∂T q_ s þ hðT T0 Þ þ seðT4 T40 Þ ¼ 0 ð8Þ
∂n
dH ¼ Cp dT ð5Þ where n is the vector normal to the surface, q_ s is the rate
of the heat input from the beam, h is the heat transfer
and the heat conduction equation is transformed to
coefficient, s is the Stefan–Boltzmann constant and e is
the emissivity.
∂ ∂T ∂ ∂T ∂ ∂T ∂H
k þ k þ k þ q_ ¼ r
∂x ∂x ∂y ∂y ∂z ∂z ∂t
ð6Þ Modeling assumptions
The process simulation can be three-dimensional
The initial conditions can be expressed as32
(3D)34,35 or include only two dimensions,36–39 so that
Tðx; y; z; 0Þ ¼ T0 ð7Þ the computational time is reduced. Two-dimensional
(2D) modeling can be used when melting of a single
supposing that the initial temperature of the powder is layer is simulated, but is inadequate when multiple
equal to the ambient temperature T0 . layers are processed. Radiation losses from the free
100 Proc IMechE Part B: J Engineering Manufacture 231(1)
Figure 7. Gaussian heat source. instead, modeling the laser beam so that it irradiates
five elements in a cross layout. A constant energy input
in the model without adding to the overall stiffness of applied to five elements in a cross layout was also used
the matrix,31,68 as their actual stiffness has been multi- by Matsumoto et al.,36 as shown in Figure 8. As stated
plied by a severe reduction factor.67 After solidification, previously, another way of modeling the heat source is
the elements are activated, obtaining back their actual applying a temperature load to certain elements.48
stiffness. A control loop determines when an element is Fixed temperature loading is used by Ma and Bin52 for
activated according to whether its temperature after a a laser beam size of 2 3 2 mm. The temperature is set
load step has surpassed or not the melting point.47 In at 1740 K, which is slightly higher than the melting
other studies, the elements of every single layer are temperature of the material (nickel).
added simultaneously.62,69 From a programming point
of view, controlling the state of the elements and locally
refining the mesh in each load step increase the level of
Temperature-dependent material properties
difficulty in the preparation of the analysis. Detailed definition of material properties is the basis
for performing accurate simulation of AM processes.
Due to the thermal character of the process, the mate-
Heat source models rial properties should be stated with relation to the
The impact of the heat source in the powder bed can be temperature and also the transition of the material
modeled as a heat flux load or as a temperature load. from powder to liquid and finally to solid. The proper-
In the vast majority of studies where a heat flux load is ties that affect significantly the process evolution and
used, the heat source follows a Gaussian intensity dis- their dependence on temperature is of high interest are
tribution. An example of a Gaussian intensity distribu- the density, the thermal conductivity and the specific
tion is presented in Figure 7. The intensity IðrÞ of a heat capacity. Other material properties, such as the
Gaussian heat flux at a radial distance from the beam Young modulus,37,44,52 the tangent modulus,52 the yield
center equal to r is given by the following equation45 strength44,52,59 and the thermal expansion coeffi-
cient,37,45,52 can also be considered as temperature
2
IðrÞ ¼ 2AP2 exp 2r2 ð9Þ dependent. Those properties are used when a mechani-
pv v cal analysis is performed, either coupled or uncoupled
with the thermal analysis. Thus, their temperature
where A is the absorptivity of the powder, P is the beam dependency affects the residual stresses and distortion
power and v is the radius where the intensity is reduced results.
from the intensity at the center of the beam by a factor
of e2 .
Heat flux loading that does not follow a Gaussian Density. Density as a physical quantity describes the
distribution has been implemented in a number of stud- mass per unit volume of a substance. As a temperature-
ies. Van Belle et al.37 applied the heat flux created by dependent property, material density changes during
the laser beam on the surface of a single element, which the process and at most times the density of the final
is of equal size with the beam diameter. They are claim- part is higher than the powder bed density.34 However,
ing that due to the very small size of the finite elements, in a number of studies, the density has been considered
this will have minor effect to the simulation results. to be constant, in order to reduce the complexity of the
Contuzzi et al.35 did not consider a Gaussian power analysis.34,41,49,51,52,59–62,69 Other studies consider two
density distribution but a constant power density different densities for the powder and the solid state.35–
102 Proc IMechE Part B: J Engineering Manufacture 231(1)
37,42,44,70
In those studies, the powder density can vary
from 40% to 60% of the solid density, depending on
which assumption for the powder packing is followed.
For example, Antony et al.50 and Patil and Yadava46
calculated the powder density on the bed, rbed , using
the following formula, where rs is the density of the
solid material, assuming that all solid particles in the
powder bed have a spherical shape, are equal in size
and are densely arranged in a cubic array
prs
rbed ¼ ð10Þ
6
For a more detailed analysis, the density is changing
with temperature.31,71,72 The relation of density with Figure 9. Temperature profiles of consecutive spots.31
temperature can be linear, as in the study of Yin et
al.,42 where a linear increase of density with tempera- Specific heat. The amount of the energy needed for a
ture has been assumed. In a few SLS studies, along certain temperature rise can be found using the specific
with temperature, the effects of sintering are accounted, heat capacity of the material, which is a temperature-
through applying sintering laws, for determining the dependent property as well. As in the case of the ther-
density.39 mal conductivity, several studies use pre-defined values
for the heat capacity and the value for any given tem-
perature point is calculated by linear interpolation or
Thermal conductivity. The thermal conductivity of a mate- extrapolation.34,41–47,49,51,52,59–62,64,69,73,78 The specific
rial determines the rate at which heat transfers through heat capacity of powder beds can be found taking into
the material. Thermal conductivity is a temperature- account the powder bulk density or for powder mix-
dependent material property. In addition, the thermal tures applying the law of Kopp–Neumann.63 Other
conductivity of powder is different than the solid mate- studies consider constant specific heat. In some cases,
rial, as the contact area among adjacent particles is the material while in powder form has different specific
smaller.73 It has been proved that the effective thermal heat capacity than the material in solid form,37,71 while
conductivity of powders is a function of multiple fac- in other studies the same value is assumed for both
tors, such as particle size and shape, solid volume frac- states.19,35,36,70 A different approach is calculating the
tion, thermal conductivity of the solid and the thermal evolution of specific heat capacity during the process,
conductivity of the gaseous media.74 Several studies according to the current temperature.39,40,50 In studies
consider constant thermal conductivities, both for the where a melt pool stability analysis is performed, the
powder and the solid state, in order to reduce the com- specific heat capacity can be defined for the solid and
plexity of the model.35–37,50,54,75 However, numerous the liquid phase as well.50,75
studies consider a temperature-dependent thermal con-
ductivity, in order to reach more realistic results. The
thermal conductivity of the solid material has been Simulation results
defined experimentally at certain temperature points.
The conductivity for the intermediate temperature
Temperature field
points is calculated with linear interpolation.41 The The temperature field history of the part during the
powder’s effective thermal conductivity is calculated in process can provide useful input for determining the
relation to the solid thermal conductivity according to distribution of the thermal stresses and predicting the
theoretical models, which take into account the poros- residual stresses and distortion.31 The temperature at
ity and the thermal conductivity of the surrounding every scanned region of the part undergoes similar ther-
gas.22,31,32,39,40,51,63,69,71,72,76 Some models take into mal cycles,26,31,44,45,66 except the edges.31,44 This phe-
account only the porosity without considering the sur- nomenon is presented in the diagram of Figure 9, where
rounding gas.45 In SLS, the models can also include the the temperature profiles of a number of consecutive
sintering potential,17,34 the effect of the sintering spots have been recorded. Van Belle et al.,37 when
necks47 and the initial sintering temperature42 as para- explaining the cyclic temperature history, state that
meters. In some cases, a constant conductivity is each element reaches peak temperature values when it
assumed for the powder and a temperature-dependent is scanned itself and also when neighboring elements
conductivity for the solidified material.44,46 Finally, a are scanned. The laser scanning of additional layers also
zero thermal conductivity for powders can be assumed leads to secondary peaks in the temperature of the
as well. In a study published by Wang et al.,77 this underlying layers.31 Roberts et al.31simulated the tem-
approach did not have a noticeable effect on the simula- perature field during SLM. Their findings indicate that
tion results, as the inter-particle heat transfer was so the upper layers reach higher temperatures than the
weak at high scanning speeds that could be disregarded. bottom layer which is in contact with the base plate.
Schoinochoritis et al. 103
Figure 10. (a) Powder stacking strategy, (b) laser irradiation region and (c) meshing of a simulation model considering powder
arrangements.49
The lower conductivity of the bottom layer, if com- temperature difference between the top and bottom of
pared with the steel base plate, shields to some extent the part is so high (i.e. 2513 °C) that a 3D model is nec-
the upper layers. They also observed that as the number essary for obtaining accurate results. In their 3D model,
of layers increased, the temperature of the bottom layer a higher maximum temperature was observed as the
experienced a slight but steady rise, which can cause the laser beam moved from the center of the first layer to
generation of residual stresses after completion of the the center of the second layer. In SLS, the temperature
process and during cooling. Thus, determining the gradient under the beam spot is greater than that in the
number of layers should be done in accordance with the already sintered areas, according to Ren et al.63 In a
rest of the process parameters and not only considering semi-scanned layer, the residual heat in the powder-half
dimensional accuracy requirements. Despite the time is higher than that of the solid-half.65 Liu et al.49 devel-
for cooling that powder recoating offers, a small but oped a 3D finite element model, which takes into
incremental rise of temperature was also observed in account powder arrangements as shown in Figure 10,
the rest of the part and the base plate. A great tempera- in order to study SLS. They came to the conclusion that
ture gradient is observed from the surface of the powder the temperature field of laser sintering is intermittent in
bed to the substrate.41 Li and Gu51 pointed out that the the micro scale, because of the discretely distributed
temperature gradient decreases as the distance from the particles, and that the temperature distribution is
top of the melt pool in the Z-axis increases. The reason remarkably inhomogeneous with the maximum tem-
for that is the higher conductivity that the solidified perature to be observed in the top layer. This approach
material in the previous layers has, if compared with offers enhanced accuracy. However, it would not be
the molten material. According to Kolossov et al.,34 the recommended for real parts, as modeling in the micro-
104 Proc IMechE Part B: J Engineering Manufacture 231(1)
Figure 11. Temperature versus scanning speed.51 Figure 12. Temperature versus laser power.51
scale, would mean an enormous number of elements been studied. Zhang et al.40 observed that a preheating
and a huge computation cost. A number of studies temperature of 200 °C, when melting composite W-Ni-
point out that the cooling rate of overhang structures is Fe powder, can lead to an increased maximum tem-
relatively smaller than normal geometries, as they exhi- perature by 6 °C, allowing for less energy input by the
bit inferior heat removal due to the insulating behavior laser beam. The temperature field is remarkably sensi-
of the unconsolidated powder below them.48,78 tive to changes in the scanning pattern. Ma and Bin52
Overhang structures should be given extra attention, as developed a 3D finite element model to investigate the
they may also require a support in order to avoid col- effect of different scanning patterns in the evolution of
lapsing. Papadakis et al.79 reported that when using lat- temperature, distortion and thermal stresses during
tice structures to support parts during fabrication, a SLS. More specifically, they compared the ‘‘S’’ and the
densely meshed support leads to higher temperature fractal scanning pattern, as shown in Figure 13(a) and
gradients between the scanned region and the substrate. (b), respectively. Their findings indicate that the fractal
On the other hand, wide-meshed supports can lead to pattern leads to much more symmetrical temperature
increased building platform temperatures and therefore field if compared to the ‘‘S’’ pattern. Increasing scan
smaller gradients. interval (i.e. the distance the laser beam moves after
The effect of the process parameters to the tempera- each time step) also causes the part temperature to
ture field characteristics has drawn the attention of a decrease, as reported by Patil and Yadava46 However,
significant number of researchers. In several studies, it a very high scan interval can lead to insufficient energy
has been reported that the maximum temperature of density input and inadequate melting of the material
the part is increasing with the laser power32,40,42,50 and between the two scan lines.32
decreasing with the scanning speed.32,40,42,64 The depen-
dence of temperature and of the time that material at a
certain point remains liquid, for an SLM process, with Residual stresses and distortion
the scan speed and the laser power is shown in Figures Thermal stresses are created when part or all of the vol-
11 and 12 respectively. As reported by Li and Gu51 the ume of a body is not free to expand or contract in
cooling rate of the molten pool increased slightly when response to temperature changes.81 In order to numeri-
the power of the laser in the SLM process doubled cally predict the residual stresses and distortion in parts
from 150 to 300 W. A significant rise of the cooling manufactured by AM processes, a mechanical simula-
rate was observed when the scanning speed increased tion, is performed using as thermal loading the tem-
from 100 to 400 mm/s. In addition, the maximum tem- perature distribution history recorded in the thermal
perature gradient increased significantly when doubling analysis. Warping and loss of edge tolerance are the
the laser power while it slightly decreased when increas- effects of distortion from a part quality perspective.44
ing the scanning speed. Higher spot radius causes lower The largest residual stresses are observed at the top
maximum temperature,42,47 due to lower intensity if the layer and the interface between the part and the sub-
power remains constant. According to Shen and strate.37,53,54 Filleting the transition from the substrate
Chou,47 an increased beam diameter in EBM can lead to the actual part can prevent stress concentration in
to a lower cooling rate due to the smaller temperature this region.68 While high tensile stresses were reported
gradients. The cooling rate of a layer was found to be by Hodge et al.78 at the top layer, the bottom layer was
higher when the layer is scanned over a solid substrate dominated by high compressive stresses. In the x-direc-
instead of being scanned on loose powder bed.45 The tion, the part initially encounters compressive stresses
effect of preheating to the temperature history has also which are gradually turning to tensile stresses after
Schoinochoritis et al. 105
Figure 13. (a) The ‘‘S’’ scanning pattern and (b) the fractal scanning pattern.80
beam scanning and when cooling occurs.44 In SLS, temperature distribution history and therefore on the
according to Jiang et al.,44 distortion in the vertical generation of residual stresses and distortion. The effect
direction is mainly caused by shrinkage due to the sin- of two SLS scanning patterns in residual stresses and
tering of porous powders. However, in the horizontal distortion was investigated by Ma and Bin52 The two
direction, the distortion is mostly caused by the thermal scanning patterns used were the ‘‘S’’ pattern and the
loading. Generally, in SLS, major distortion of the part fractal pattern as depicted in Figure 13. The results
appears during the laser scanning process, with the showed that the fractal pattern can offer smaller distor-
effect of the residual thermal stresses on distortion tion due to more symmetrical temperature field com-
being rather small if compared with that of the transi- pared to the ‘‘S’’ pattern. Matsumoto et al.36 observed
ent thermal stresses.52 Dai and Shaw59 state that in AM that when the neighboring track begins to solidify high
processes, the transient thermal stresses are those caus- tensile stresses appear between the solidified tracks at
ing the distortion, while the residual thermal stresses the side end of the solid part. This may lead to cracking
have little and basically opposite effect to that of the of the layer. They also report that the amount of deflec-
transient ones. The effect of residual stresses appears tion caused by SLM increases as the laser track length
when, as described by Papadakis et al.,79 detaching the increases and thus shortening the path may be used to
part from the building platform. That causes a relaxa- prevent distortion. Cantilever beams are experiencing
tion of the residual stresses and a subsequent develop- larger distortion when being built in the lengthwise
ment of deformations. According to Hussein et al.,45 direction than in the orthogonal direction, according to
high temperatures lead to a decreasing elastic modulus Neugebauer et al.66
which causes stress reduction. Therefore, high thermal
stresses appearing in the already solidified regions tend
to reduce as moving closer to the melt pool. In addi- Melt pool characteristics
tion, when scanning neighboring tracks, the previously Melt pool dimensions are important for ensuring that a
scanned tracks are reheated, due to track overlap, and strong bonding between powder particles will be
the residual stresses are thus released. Thin-walled achieved. Bonding of the particles significantly affects
structures exhibit high residual stresses after SLM pro- the porosity and therefore the part quality. In addition,
cessing, according to Neugebauer et al.54 It is indicative melt pool width and length should be thoroughly con-
that in the blades of an impeller, which are thin-walled sidered in the way they affect the already fabricated
structures, the stresses were found to be the highest areas. Remelting of such areas may cause dimensional
over the whole part. inaccuracies and loss of surface quality. The point with
The effect of process parameters on residual stresses the highest temperature in the melt pool is named as
and distortion has been given great attention. ‘‘center of the melt pool.’’ It has been observed by vari-
Preheating has been found to lead to larger longitudi- ous researchers that this point is not located at the cen-
nal compressive stresses divisions at the lower side of ter of the beam spot, as it is normally expected. It has
the part, which is connected to the supports, and also been reported that the center of the melt pool is slightly
to larger deformations of the part in the negative z- displaced backward from the laser scanning direc-
direction.79 Preheating was found to lead to higher ten- tion.42,46,75 This displacement can increase as the beam
sile stresses by Krol et al.16 High scanning speeds are power increases.51 According to Yin et al.,42 the melt
reported to lead to distortion due to sudden drop of pool displacement can be attributed to the heat accu-
temperature versus time.50 In SLS, according to Wang mulation in the previously scanned area. This phenom-
et al.,77 lower distortion in the final part can be accom- enon is graphically presented in Figures 14 and 15. The
plished with lower laser power. As described previ- melt pool with the surrounding area is eventually, as
ously, scanning patterns have a significant effect on the the beam scanning progresses, taking the shape of a
106 Proc IMechE Part B: J Engineering Manufacture 231(1)
Figure 14. Displacement of the melt pool center with respect to the laser beam.75
Experimental validation
Some level of abstraction is used when developing pro-
cess simulation models in order the computational time
required to be realistic and feasible.83 Thus, validation
of the simulation model, through comparison with
experimental results, is recommended.16
The temperature history of a part during fabrication
can be experimentally recorded using conventional
measurement instruments such as thermographic cam- Figure 17. Melt pool dimensions change with beam power:82
eras46,48,66 or thermocouples.71 While thermocouples (a) 20 W (b) 40 W (c) 60 W and (d) 80 W.
are measuring temperature only at the spots where
installed, thermal cameras can provide a temperature
distribution imaging, giving therefore more insight in 3 2.8 mm2, having a resolution of 256 3 256 pixels.
the temperature field evolution. Raytheonä infrared The frame rate was set at 10 frames per second (fps).
cameras have been used by Kolossov et al.34 and Zäh and Lutzmann71 utilized a thermocouple to mea-
Roberts et al.31 to measure the temperature during sure the temperature evolution during EBM. In order
SLM. In the case of Kolossov et al.,34 the camera was not to interrupt the process sequence, the measuring
equipped with a spectral band pass filter with transmis- device is attached to the build chamber from the reverse
sion between 3 and 5 mm and was calibrated using a side of the platform. In fact, the temperature is mea-
pyrometer. The size of the image was approximately 2.8 sured slightly below, about 300 mm, the surface layer. A
108 Proc IMechE Part B: J Engineering Manufacture 231(1)
number of time steps to one. This approach was fol- stage. However, it must be mentioned that efforts being
lowed by Ding et al.,87,88 who achieved the steady-state done for the integration of design and FEA modules
transformation by attaching an Eulerian reference provide a promising vision of the future.97
frame at the heat source, when simulating the wire and In a more and more computer-enabled world, AM is
arc additive layer manufacturing (WAALM) process. the last step in a digital production chain. This is proba-
According to the researchers, additional time can be bly the main driver for its rising popularity. An holistic
saved, as using an Eulerian frame does not require as approach to all the stages from design to manufacturing
dense mesh as the typical transient approach does. The is the current trend. CAD, computer-aided manufactur-
total thermo-mechanical analysis time saved was ing (CAM) and computer-aided process planning
almost 80% of the transient analysis time. (CAPP) are becoming integrated98–100 with special
Simulation, in general, has an inductive character, applications in AM.101,102 In this direction, it is essen-
meaning that it should allow researchers to draw con- tial that simulation of the process becomes part of this
clusions beyond the nature of the specific problem stud- chain. Adjusting the process parameters based on
ied. At this point, AM simulation lacks transferability design features and requirements or considering design
of results. This lack can be attributed to the great diver- modifications due to machinery limitations would be
sity in the modeling of the material properties and the therefore possible, saving significant amounts of faulty
assumptions followed and also to the limited under- parts. Taking into account the high computational
standing of the physics of the process.89,90 Regarding demands and the extended geographical span of the
materials characterization and standardization, most of modern manufacturing networks, deploying such an
the advancements are coming from official sources. approach into a cloud-based architecture, as the one
ASTM International91 has created a committee devoted presented by Ari and Muhtaroglu,103 using advanced
to AM, namely the F42, who have published a number algorithms for parallel processing as those proposed by
of material specifications for AM. NIST, the National Bangerth et al.97 and Paszyński et al.104 and exploiting
Institute of Standards and Technology,92 has also the capabilities of knowledge management105 can be the
began an initiative since 2013 for the characterization key factors for a successful implementation.
of AM materials. As far as it concerns the understand-
ing of the physical mechanisms acting during the pro-
cess, the studies utilizing the FEM, as those studied in Conclusion
this article, can only give a macroscopic insight and are
The latest studies in the finite element simulation of
mostly aimed at drawing quick but safe conclusions.
metallic powder bed additive processes are reviewed in
Deeper understanding can be gained by studies that are
this article. Those studies are summarized in Table 1
modeling the material at the powder particle level and
and are characterized according to the process and
consider the effects of fluid flow at the melt pool and
material studied, the modeling methodology and the
the gas–liquid–solid interactions. The LBM19,93 and the
experimental validation method used. In addition, the
FVM82 are methods commonly used in this type of
scope of each study is reported.
studies, while other numerical methods such as the
The following remarks can be concluded from the
Monte Carlo-based ray tracing method94 are more
rarely used. table mentioned above:
Commercial FEA tools, specialized to AM modeling
and simulation, exist only at a limited extend regarding The majority of the simulation studies are focused
their capabilities. This weakens the position of FEA as on the SLM process, with the EBM given the least
a tool for the global industry. The reason is that highly attention.
qualified personnel is required in order to address the Steel is the material used in most of the studies.
challenges of modeling an AM process and incorporate Regarding the scope of the studies, the temperature
techniques such as ‘‘element birth and death’’ and field is what most studies target at. Melt pool char-
‘‘moving heat source’’ into the existing general-purpose acteristics seem to be given the least attention by
FEA packages. The industrial needs for specialized researchers. However, other methods, such as the
software can be observed if looking at the welding FVM, are more favorable in the study of the melt
simulation current situation. Welding, as a process with pool than FEM.
long tradition in the manufacturing domain and a vast There are not many new experiments reported. A
number of different applications, can be simulated by great number of researchers use previously pub-
dedicated welding simulation tools, such as the NX lished experiments in order to validate their simula-
Weld Assistant95 developed by Siemens PLM Software. tion models.
The area of design for AM, including topology optimi-
zation, layer generation and STL file conversion has Based on the above, strengthening the role of simula-
been covered by commercial solutions developed in the tion in AM process optimization would require greater
last years by well-established companies, such as attention to be given in the development of models for
Altair.96 On the other hand, commercialization of processes other than SLM and models that consider
simulation for AM processes is still at an immature materials for special applications, such as Inconel
Table 1. Summary of reviewed publications.
110
Authors Year Process Material Number of Model Element FEA code used Experimental Scope
dimensions size (mm) size (mm) validation
Krol et al.16 2013 N/A AlSi12 3 23233 N/A N/A Neutron Residual
diffraction stresses—
verification of
simulation
results
Boillat et al.17 2004 SLS Titanium 3 Hollow cylinder: N/A N/A N/A Process
0.3 inner radius, optimization
1.2 outer radius, combining FEA
0.6 depth and neural
networks
Zeng et al.26 2012 SLM Ti6Al4V 3 Powder bed: 3 3 3 Powder bed: Free ANSYS N/A Temperature
3 3 Substrate: 1 3 meshing Substrate: field—melt pool
1 3 0.3 0.025 3 0.025 3 characteristics
0.03
Roberts et al.31 2009 SLM Ti6Al4V 3 Powder bed: Scanned region: ANSYS N/A Temperature
13130.015 0.025 3 0.025 3 field
Substrate: 33333 0.025 Coarser
mesh away from
the beam
Li et al.32 2009 SLM 316L Stainless Steel 3 Powder bed: 2.2 3 Powder bed: 0.1 3 ANSYS Microscopy Temperature
1.2 3 0.1 0.1 3 0.1 field—melt pool
Substrate: 3 3 1.8 Substrate: free grid characteristics
3 0.4 at the contact
surface. Coarser
mesh at the rest
Kolossov et al.34 2004 SLS Titanium 3 53532 Scanned region: N/A Infrared camera Temperature
0.01 3 0.01 3 field
0.01Away from the
beam: 0.1 3 0.1 3
0.1
Contuzzi et al.35 2011 SLM 316L Stainless Steel 3 6 3 9 3 1.3 Scanned region: ANSYS Optical Temperature
0.056 3 0.056 3 microscope field
0.056 Coarser
mesh away from
the beam
Matsumoto et al.36 2002 SLM Nickel alloy 2 N/A N/A N/A N/A Residual
stresses—
distortion
Van Belle et al.37 2012 SLM 15-5PH Steel 2/3 0.6 3 0.5 3 0.5 N/A ABAQUS N/A Temperature
field—residual
stresses
(continued)
Proc IMechE Part B: J Engineering Manufacture 231(1)
Table 1. Continued
Authors Year Process Material Number of Model Element FEA code used Experimental Scope
dimensions size (mm) size (mm) validation
Bai et al.38 2006 SLS Polymer-coated 2 N/A 10 3 0.2 Code written Infrared thermometer— Temperature
molybdenum powder in FORTRAN Thermocouple field
Schoinochoritis et al.
Childs et al.39 2000 SLS Stainless Steel 2/3 N/A N/A N/A N/A Melt pool
characteristics
Zhang et al.40 2010 SLM W-Ni-Fe composite 3 Powder bed: 1 3 2 Powder bed: 0.05 ANSYS SEM Temperature
powder 3 0.05 Substrate: 2 3 0.05 3 0.05 field—melt pool
3 3 3 1.5 Substrate: free characteristics
mesh
Song et al.41 2012 SLM Ti6Al4V 3 Powder bed: 2 3 1 0.005 3 0.005 3 ANSYS Optical microscope Temperature
3 0.05 Substrate: 0.005 field
25 3 10 3 5
Yin et al.42 2012 SLS Iron 3 0.9 3 0.9 3 0.03 0.0075 3 0.0075 3 ANSYS N/A Temperature
0.0075 field—melt pool
characteristics
Zhang et al.43 2011 SLS W-Cu composite powder 3 3.4 3 1.6 3 0.3 Powder bed: 0.1 3 ANSYS N/A Melt pool
0.1 3 0 Substrate: characteristics
coarser mesh
Jiang et al.44 2002 SLS 316L Stainless Steel 3 N/A Scanned region: ABAQUS N/A Temperature
0.183 3 0.183 3 field—residual
0.083 Coarser stresses—
mesh away from distortion
the beam
Hussein et al.45 2013 SLM 316L Stainless Steel 3 12.2 3 3.2 3 1.0 Scanned region: ANSYS N/A Temperature
0.75 3 0.75 3 0.75 field—residual
Coarser mesh stresses—melt
away from the pool
beam characteristics
Patil and Yadava46 2007 SLS Titanium 3 N/A N/A Code written Infrared camera Temperature
in MATLAB field—melt pool
characteristics
Shen and Chou47 2012 EBM Ti6Al4V 3 N/A N/A ABAQUS N/A Temperature
field—melt pool
Schilp et al.48 2014 SLS Inconel 718 3 N/A 0.5 3 0.5 3 0.5 ANSYS Infrared camera Temperature
field
Liu et al.49 2012 SLS Pure Fe 3 N/A N/A ANSYS Optical microscopy/SEM Temperature
field—melt pool
characteristics
Antony et al.50 2009 SLM 316L Stainless Steel 3 N/A 0.05 3 0.05 3 0.05 N/A SEM Temperature
field—melt
pool—
microstructure
(continued)
111
Table 1. Continued
112
Authors Year Process Material Number of Model Element FEA code used Experimental Scope
dimensions size (mm) size (mm) validation
Li and Gu51 2014 SLM AlSi10Mg 3 Powder bed: 1.54 Powder bed: ANSYS OM/SEM Temperature
3 0.7 3 0.1 0.0175 3 0.0175 3 field—melt pool
Substrate: 2 3 1 3 0.025 Substrate: characteristics
0.4 coarser mesh
Ma and Bin52 2007 SLS Nickel 3 60 3 60 3 2 23232 N/A N/A Temperature
field—residual
stresses—
distortion
Zaeh and Branner53 2010 SLM Tool Steel 1.2709 3 N/A N/A ANSYS Neutron diffraction Residual
stresses—
distortion
Neugebauer et al.54 2014 SLM Inconel 718 3 100 3 100 3 100 N/A MSC Marc N/A Residual
stresses
Ren et al.63 2011 SLS W-Ni composite powder 3 Powder bed: 3.4 3 Powder bed: 0.1 3 ANSYS N/A Temperature
1.6 3 0.2 0.1 3 0.1 field
Substrate: 4.4 3 3 Substrate: coarser
3 0.6 mesh
Zhang et al.64 2010 SLS Copper 3 Single layer powder Powder bed: 0.1 3 ANSYS N/A Temperature
bed: 3.4 3 1.6 3 0.1 3 0.1 field—melt pool
0.1 Multi-layer Substrate: coarser characteristics
powder bed: mesh
3.4 3 1.6 3 0.3
Substrate: 4.8 3 2
3 0.5
Shen and Chou65 2012 EBM Ti6Al4V 3 300 3 300 3 150 N/A N/A N/A Temperature
field
Neugebauer et al.66 2014 SLM X4CrNiCuNb 3 N/A 0.5 3 0.5 3 0.4 MSC Marc/Mentat Thermal camera Temperature
field—
distortion
Zäh and Lutzmann71 2010 EBM 316L Stainless Steel 3 N/A N/A N/A Thermocouple Melt pool
characteristics
Childs et al.72 2005 SLM/SLS M2/H13 Tool Steel 314- 3 N/A N/A N/A Photography Melt pool
HC Stainless Steel characteristics
Chen and Zhang76 2007 SLS AISI 1018—Bni-2 3 N/A N/A N/A N/A Melt pool
composite powder characteristics
Hodge et al.78 2013 SLM 316L Stainless Steel 3 0.6 3 0.6 3 0.2 0.0025 3 0.0025 3 Diablo N/A Temperature
0.0025 field—residual
stresses—
distortion
Papadakis et al.79 2012 SLM Tool Steel 1.2709 3 Powder bed: 70 3 1 3 1 3 0.5 ANSYS Neutron diffraction Temperature
15 3 12 Supports: field—residual
30 3 15 3 8 stresses—
distortion
Proc IMechE Part B: J Engineering Manufacture 231(1)
FEA: finite element analysis; SLS: selective laser sintering; SLM: selective laser melting; EBM: electron beam melting; SEM: scanning electron microscopy; OM: optical microscopy.
Schoinochoritis et al. 113
alloys. Furthermore, a solid experimental base should in terms of residual stresses, distortion and porosity.
be created, in order to be used as a reference point for However, determining the process parameters can also
future simulation studies. be influenced by other factors such as productivity,
Regarding the simulation results of the studies energy consumption and manufacturing costs. Thus,
reviewed in this article, conclusions can be drawn as the selection of the optimal process parameters should
follows. The fabricated part undergoes similar tempera- be dealt as a multi-dimensional problem.
ture cycles at every scanned region, except its edges. The FEM has proven valuable in the simulation of
However, the upper layers tend to reach higher tem- AM for metallic parts. By understanding the nature of
peratures than the bottom layer which is in contact the process and predicting the effect of process para-
with the base plate. The bottom layer itself experiences meters into the mechanical characteristics of the final
a slight temperature increase as more layers are added, part, the process can be adjusted and optimized in
with this fact implying the need for cautious consider- order to produce right first time parts at greater rates.
ation of the slicing strategy and the total number of Scrap rates and faulty parts can thus be reduced lead-
layers. The major temperature gradients, developed ing to more eco-friendly and cost-effective produc-
during the process, are those in the z-axis between the tion. However, there are still issues that need to be
top layer and the substrate. As a result, 3D models addressed. Long computational times do not allow
should be preferred to 2D ones for obtaining accurate for on-line process optimization based on FEA. In
results. Process parameters have a significant impact in addition, transferability of simulation results to differ-
the temperature history of the part. The maximum tem- ent case studies is limited. A common ground in mate-
perature in the part has been observed to increase with rial properties modeling is hard to be found between
the laser power and decrease with the scanning speed. different researchers of the field. Furthermore, specia-
The same observations were also made for the maxi- lized software for AM simulation is limited only to
mum temperature gradient. The maximum temperature product design solutions, with the area of process
was also found to increase when lowering the spot dia- simulation being at its very first stages of develop-
meter of the heat source and when preheating the pro- ment. Finally, the total shift of manufacturing to a
cess chamber. Residual stresses and distortion play a web and cloud-enabled paradigm is fueling the inte-
significant role in the structural integrity and the qual- gration of FEA tools with other digital manufacturing
ity of the fabricated part. High tensile stresses are tools in order to serve the needs of today’s global
observed at the top layer and high compressive stresses manufacturing networks.
at the interface between the part and the substrate.
Thin-walled structures are exhibiting high residual
Declaration of conflicting interests
stresses. In SLS, residual stresses have a little effect on
distortion, which is mainly caused by transient thermal The authors declare that there is no conflict of interest.
stresses during processing. Distortion can be prevented
by applying lower scanning speeds, lower power at the Funding
heat source and suitable scanning strategies that utilize
shorter paths. The melt pool dimensions and shape are This research received no specific grant from any fund-
strongly influencing the bonding between the scanned ing agency in the public, commercial or not-for-profit
region and the already solidified regions. The center of sectors.
the melt pool is displaced backward the scanning direc-
tion, while its shape is taking the form of a comet tail References
as the process evolves. Steeper thermal gradients are 1. Mellor S, Hao L and Zhang D. Additive manufacturing:
observed at the front of the melt pool than in the back. a framework for implementation. Int J Prod Econ 2014;
The melt pool length increases with the scanning speed 149: 194–201.
and the beam power and decreases with the beam dia- 2. Levy GN, Schindel R and Kruth JP. Rapid manufactur-
meter and the initial powder porosity. The melt pool ing and rapid tooling with Layer Manufacturing (LM)
width and depth have been found to decrease with the technologies, state of the art and future perspectives.
scanning speed and the beam diameter, while they CIRP Ann: Manuf Techn 2003; 52(2): 589–609.
increase with the beam power. Generally, the melt pool 3. Santos EC, Shiomi M, Osakada K, et al. Rapid manu-
facturing of metal components by laser forming. Int J
dimensions depend on the interaction time with the
Mach Tool Manu 2006; 46(12–13): 1459–1468.
beam and the rate of energy input. The length-to-width
4. Gu DD, Meiners W, Wissenbach K, et al. Laser additive
ratio, which is critical for the development of Plateau– manufacturing of metallic components: materials, pro-
Rayleigh instabilities and the subsequent propagation cesses and mechanisms. Int Mater Rev 2012; 57: 133–164.
of the ‘‘balling’’ effect, has been observed to increase 5. Bauereiß A, Scharowsky T and Körner C. Defect genera-
with the scanning speed and decrease with the beam tion and propagation mechanism during additive manu-
power. As described in this study, process parameters facturing by selective beam melting. J Mater Process
have a significant impact on the quality of the final part Tech 2014; 214(11): 2522–2528.
114 Proc IMechE Part B: J Engineering Manufacture 231(1)
6. Chantzis D, Van der Veen S, Zettler J, et al. An indus- 23. Dandekar CR and Shin YC. Modeling of machining of
trial workflow to minimise part distortion for machining composite materials: a review. Int J Mach Tool Manu
of large monolithic components in aerospace industry. 2012; 57: 102–121.
Procedia CIRP 2013; 8: 281–286. 24. Shi B and Attia H. Current status and future direction in
7. Williams CB, Cochran JK, Rosen DW. Additive manu- the numerical modeling and simulation of machining pro-
facturing of metallic cellular materials via three-dimen- cesses: a critical literature review. Mach Sci Technol 2010;
sional printing. Int J Adv Manuf Technol 2011; 53(1-4): 14(2): 149–188.
231–239. 25. Arrazola PJ, Özel T, Umbrello D, et al. Recent advances
8. Leuders S, Thöne M, Riemer A, et al. On the mechanical in modelling of metal machining processes. CIRP Ann:
behaviour of titanium alloy TiAl6V4 manufactured by Manuf Techn 2013; 62(2): 695–718.
selective laser melting: fatigue resistance and crack 26. Zeng K, Pal D and Stucker B. A review of thermal analy-
growth performance. Int J Fatigue 2013; 448: 300–307. sis methods in Laser Sintering and Selective Laser Melt-
9. Kruth JP, Levy G, Klocke F, et al. Consolidation phe- ing. In: National science foundation solid freeform
nomena in laser and powder-bed based layered manufac- fabrication symposium, Austin, TX, 6–8 August 2012.
turing. CIRP Ann: Manuf Techn 2007; 56(2): 730–759. Austin, TX: Laboratory for Freeform Fabrication.
10. Gu D, Wang H and Zhang G. Selective laser melting 27. Yang DY, Jung DW, Song IS, et al. Comparative investi-
additive manufacturing of Ti-based nanocomposites: the gation into implicit, explicit, and iterative implicit/explicit
role of nanopowder. Metall Mater Trans A 2014; 45(1): schemes for the simulation of sheet-metal forming pro-
464–476. cesses. J Mater Process Tech 1995; 50(1–4): 39–53.
11. Kruth JP, Froyen L, Van Vaerenbergh J, et al. Selective 28. Sun JS, Lee KH and Lee HP. Comparison of implicit and
laser melting of iron-based powder. J Mater Process Tech explicit finite element methods for dynamic problems. J
2004; 149(1–3): 616–622. Mater Process Tech 2000; 105(1–2): 110–118.
12. Parandoush P and Hossain A. A review of modeling and 29. Harewood FJ and McHugh PE. Comparison of the
simulation of laser beam machining. Int J Mach Tool implicit and explicit finite element methods using crystal
Manu 2014; 85: 135–145. plasticity. Comp Mater Sci 2007; 39(2): 481–494.
13. Turner MJ, Clough RW, Martin HC, et al. Stiffness and 30. Prior AM. Applications of implicit and explicit finite ele-
deflection analysis of complex structures. J Aeronaut Sci ment techniques to metal forming. J Mater Process Tech
1956; 23(9): 805–823. 1994; 45(1–4): 649–656.
14. Madenci E and Guven I. The finite element method and 31. Roberts IA, Wang CJ, Esterlein R, et al. A three-
applications in engineering using ANSYS. 1st ed. New dimensional finite element analysis of the temperature
York: Springer, 2006, p.1. field during laser melting of metal powders in additive
15. Markopoulos AP. Finite element method in machining pro- layer manufacturing. Int J Mach Tool Manu 2009; 49:
cesses. 1st ed.New York: Springer, 2013, p.30. 916–923.
16. Krol TA, Seidel C, Schilp J, et al. Verification of struc- 32. Li R, Shi Y, Liu J, et al. Effects of processing parameters
tural simulation results of metal-based additive manufac- on the temperature field of selective laser melting metal
turing by means of neutron diffraction. Phys Procedia powder. Powder Metall Met C + 2009; 48(3–4): 186–195.
2013; 41: 849–857. 33. Labudovic M, Hu D and Kovacevic R. A three dimen-
17. Boillat E, Kolossov S, Glardon R, et al. Finite element sional model for direct laser metal powder deposition and
and neural network models for process optimization in rapid prototyping. J Mater Sci 2003; 38(1): 35–49.
selective laser sintering. Proc IMechE, Part B: J Engineer- 34. Kolossov S, Boillat E, Glardon R, et al. 3D FE simula-
ing Manufacture 2004; 218(6): 607–614. tion for temperature evolution in the selective laser sinter-
18. Körner C, Bauereiß A and Attar E. Fundamental conso- ing process. Int J Mach Tool Manu 2004; 44: 117–123.
lidation mechanisms during selective beam melting of 35. Contuzzi N, Campanelli SL and Ludovico AD. 3D finite
powders. Model Simul Mater Sc 2013; 21(8): 085011. element analysis in the Selective Laser Melting process.
19. Körner C, Attar E and Heinl P. Mesoscopic simulation Int J Simulat Model 2011; 10(3): 113–121.
of selective beam melting processes. J Mater Process Tech 36. Matsumoto M, Shiomi M, Osakada K, et al. Finite ele-
2011; 211(6): 978–987. ment analysis of single layer forming on metallic powder
20. Scharowsky T, Bauereiß A, Singer RF, et al. Observation bed in rapid prototyping by selective laser processing. Int
and numerical simulation of melt pool dynamic and beam J Mach Tool Manu 2002; 42(1): 61–67.
powder interaction during selective electron beam melt- 37. Van Belle L, Vansteenkiste G and Boyer JC. Compari-
sons of numerical modelling of the Selective Laser Melt-
ing. In: National science foundation solid freeform fabrica-
ing. Key Eng Mat 2012; 504–506: 1067–1072.
tion symposium, Austin, TX, 6–8 August 2012. Austin,
38. Bai PK, Cheng J, Liu B, et al. Numerical simulation of
TX: Laboratory for Freeform Fabrication.
temperature field during selective laser sintering of
21. Chen T and Zhang Y. Numerical simulation of two
polymer-coated molybdenum powder. T Nonferr Metal
dimensional melting and resolidification of a two-
Soc 2006; 16: 603–607.
component metal powder layer in selective laser sintering
39. Childs THC, Hauser C, Taylor CM, et al. Simulation and
process. Numer Heat Tr A: Appl 2004; 46(7): 633–649.
experimental verification of crystalline polymer and direct
22. Dai D and Gu D. Thermal behavior and densification
metal Selective Laser Sintering. In: National science foun-
mechanism during selective laser melting of copper matrix
dation solid freeform fabrication symposium, Austin, TX,
composites: simulation and experiments. Mater Design
7–9 August 2000. Austin, TX: Laboratory for Freeform
2014; 55: 482–491.
Fabrication.
Schoinochoritis et al. 115
40. Zhang DQ, Cai QZ, Liu JH, et al. Select laser melting of 56. Sypkens Smit M and Bronsvoort WF. Efficient tetrahe-
W–Ni–Fe powders: simulation and experimental study. dral remeshing of feature models for finite element analy-
Int J Adv Manuf Tech 2010; 51: 649–658. sis. Eng Comput 2009; 25(4): 327–344.
41. Song B, Dong S, Liao H, et al. Process parameter selec- 57. Vartziotis D, Wipper J and Papadrakakis M. Improving
tion for selective laser melting of Ti6Al4V based on tem- mesh quality and finite element solution accuracy by
perature distribution simulation and experimental GETMe smoothing in solving the Poisson equation.
sintering. Int J Adv Manuf Tech 2012; 61(9–12): 967–974. Finite Elem Anal Des 2013; 66: 36–52.
42. Yin J, Zhu H, Ke L, et al. Simulation of temperature dis- 58. Riedlbauer D, Steinmann P and Mergheim J. Thermome-
tribution in single metallic powder layer for laser micro- chanical finite element simulations of selective electron
sintering. Comp Mater Sci 2012; 53(1): 333–339. beam melting processes: performance considerations.
43. Zhang J, Li D, Li J, et al. Numerical simulation of tem- Comput Mech 2014; 54(1): 109–122.
perature field in selective laser sintering. In: Li D, Liu Y 59. Dai K and Shaw L. Distortion minimization of laser-
and Chen Y (eds) Computer and computing technologies in processed components through control of laser scanning
agriculture IV. 1st ed.New York: Springer, 2011, pp.474– patterns. Rapid Prototyping J 2002; 8(5): 270–276.
479. 60. Dai K and Shaw L. Finite-element analysis of effects of
44. Jiang W, Dalgarno KW and Childs THC. Finite element the laser-processed bimaterial component size on stres-
analysis of residual stresses and deformations in direct ses and distortion. Metall Mater Trans A 2003; 34(5):
metal SLS process. In: National science foundation solid 1133–1145.
freeform fabrication symposium, Austin, TX, 5–7 August 61. Dai K, Klemens P and Shaw L. Numerical simulation of
2002. Austin, TX: Laboratory for Freeform Fabrication. bi-materials laser densification. In: National science foun-
45. Hussein A, Hao L, Yan C, et al. Finite element simula- dation solid freeform fabrication symposium, Austin, TX,
tion of the temperature and stress fields in single layers 7–9 August 2000. Austin, TX: Laboratory for Freeform
built without-support in selective laser melting. Mater Fabrication.
Design 2013; 52: 638–647. 62. Dai K and Shaw L. Thermal and stress modeling of
46. Patil RB and Yadava V. Finite element analysis of tem- multi-material laser processing. Acta Mater 2001; 49(20):
perature distribution in single metallic powder layer dur- 4171–4181.
ing metal laser sintering. Int J Mach Tool Manu 2007; 47: 63. Ren J, Liu J and Yin J. Simulation of transient tempera-
1069–1080. ture field in the selective laser sintering process of W/Ni
47. Shen N and Chou K. Thermal modeling of electron beam powder mixture. In: Li D, Liu Y and Chen Y (eds) Com-
additive manufacturing process: powder sintering effects. puter and computing technologies in agriculture IV. 1st
In: ASME international manufacturing science and engi- ed.New York: Springer, 2011, pp.494–503.
neering conference, Notre Dame, IN, 4–8 June 2012, 64. Jian Z, Deying L, Longzhi Z, et al. Simulation of tem-
pp.287–295. New York: ASME. perature field in selective laser sintering of copper pow-
48. Schilp J, Seidel C, Krauss H, et al. Investigations on tem- der. In: IEEE international conference on mechanic
perature fields during laser beam melting by means of automation and control engineering, Wuhan, China, 26–28
process monitoring and multiscale process modelling. June 2010, pp.3282–3285. New York: IEEE.
Adv Mech Eng 2014; 2014: 217584 (7 pp.). 65. Shen N and Chou K.Numerical thermal analysis in elec-
49. Liu FR, Zhang Q, Zhou WP, et al. Micro scale 3D FEM tron beam additive manufacturing with preheating
simulation on thermal evolution within the porous struc- effects. In: National science foundation solid freeform fab-
ture in selective laser sintering. J Mater Process Tech rication symposium, Austin, TX, 6–8 August 2012. Aus-
2012; 212(10): 2058–2065. tin, TX: Laboratory for Freeform Production.
50. Antony K, Arivazhagan N and Senthilkumaran K. 66. Neugebauer F, Keller N, Ploshikhin V, et al. Multi scale
Numerical and experimental investigations on laser melt- FEM simulation for distortion calculation in additive
ing of stainless steel 316L metal powders. J Manuf Pro- manufacturing of hardening stainless steel. In: BIAS 4th
cess 2014; 16(3): 345–355. international workshop on thermal forming and welding dis-
51. Li Y and Gu D. Parametric analysis of thermal behavior tortion, Bremen, 9–10 April 2014. Bremen: BIAS Verlag.
during selective laser melting additive manufacturing of 67. ANSYS Inc. Help manual (ANSYS 15.0). Canonsburg,
aluminum alloy powder. Mater Design 2014; 63: 856–867. PA: ANSYS Inc., 2013.
52. Ma L and Bin H. Temperature and stress analysis and 68. Denlinger ER, Heigel JC and Michaleris P. Residual
simulation in fractal scanning-based laser sintering. Int J stress and distortion modeling of electron beam direct
Adv Manuf Tech 2007; 34(9–10): 898–903. manufacturing Ti-6Al-4V. Proc IMechE, Part B: J Engi-
53. Zaeh MF and Branner G. Investigations on residual neering Manufacture. Epub ahead of print 16 July 2014.
stresses and deformations in selective laser melting. Prod DOI: 10.1177/0954405414539494.
Engineer 2010; 4(1): 35–45. 69. Dai K and Shaw L. Thermal and mechanical finite ele-
54. Neugebauer F, Keller N, Hongxiao X, et al. Simulation ment modeling of laser forming from metal and ceramic
of selective laser melting process using specific layer based powders. Acta Mater 2004; 52: 69–80.
meshing. In: Fraunhofer direct digital manufacturing con- 70. Yadroitsev I, Gusarov A, Yadroitsava I, et al. Single
ference, Berlin, 12–13 March 2014. Stuttgart: Fraunhofer track formation in selective laser melting of metal pow-
Verlag. ders. J Mater Process Tech 2010; 210(10): 1624–1631.
55. Parthasarathy VN and Kodiyalam S. A constrained opti- 71. Zäh MF and Lutzmann S. Modelling and simulation
mization approach to finite element mesh smoothing. of electron beam melting. Prod Engineer 2010; 4(1):
Finite Elem Anal Des 1991; 9(4): 309–320. 15–23.
116 Proc IMechE Part B: J Engineering Manufacture 231(1)
72. Childs THC, Hauser C and Badrossamay M.Selective Manufacturing process on large multi-layer parts. Comp
laser sintering (melting) of stainless and tool steel pow- Mater Sci 2011; 50(12): 3315–3322.
ders: experiments and modelling. Proc IMechE, Part B: J 89. Bourell DL, Beaman JJ, Leu MC, et al. A brief history of
Engineering Manufacture 2005; 219: 339–357. additive manufacturing and the 2009 roadmap for addi-
73. Roberts IA. Investigation of residual stresses in the laser tive manufacturing: looking back and looking ahead. In:
melting of metal powders in additive layer manufacturing. National science foundation US—TURKEY workshop on
PhD Thesis, University of Wolverhampton, Wolver- rapid technologies, Istanbul, Turkey, 24–25 September
hampton, 2012. 2009. Istanbul: Istanbul Technical University.
74. Shapiro M, Dudko V, Royzen V, et al. Characterization 90. Hague R, Mansour S and Saleh N. Material and design
of powder beds by thermal conductivity: effect of gas considerations for rapid manufacturing. Int J Prod Res
pressure on the thermal resistance of particle contact 2004; 42(22): 4691–4708.
points. Part Part Syst Char 2004; 21(4): 268–275. 91. ASTM International. Subcommittee F42.05 on materials
75. Gusarov AV, Yadroitsev I, Bertrand P, et al. Heat trans- and processes, http://www.astm.org/COMMIT/SUB
fer modelling and stability analysis of selective laser melt- COMMIT/F4205.htm (2014, accessed 5 September 2014).
ing. Appl Surf Sci 2007; 254(4): 975–979. 92. National Institute of Standards and Technology. Charac-
76. Chen T and Zhang Y. Thermal modeling of laser sinter- terization of additive manufacturing materials, http://
ing of two-component metal powder on top of sintered www.nist.gov/el/isd/sbm/camm.cfm (2014, accessed 5
layers via multi-line scanning. Appl Phys A: Mater 2007; September 2014).
86(2): 213–220. 93. Ammer R, Markl M, Ljungblad U, et al. Simulating fast
77. Wang XC, Laoui T, Bonse J, et al. Direct selective laser electron beam melting with a parallel thermal free surface
sintering of hard metal powders: experimental study and lattice Boltzmann method. Comput Math Appl 2014;
simulation. Int J Adv Manuf Tech 2002; 19(5): 351–357. 67(2): 318–330.
78. Hodge N, Ferencz RM and Solberg JM. Implementation 94. Zhou J and Zhang Y. Numerical simulation of laser irra-
of a thermomechanical model in Diablo for the simula- diation to a randomly packed bimodal powder bed. Int J
tion of selective laser melting. Report, Lawrence Liver- Heat Mass Tran 2009; 52(13–14): 3137–3146.
more National Laboratory, Livermore, CA, October 95. Siemens Product Lifecycle Management Software Inc.
2013. NX Welding Assistant webpage, http://www.plm.auto
79. Papadakis L, Branner G, Schober A, et al. Numerical mation.siemens.com/en_us/products/nx/for-design/process-
modeling of heat effects during thermal manufacturing of specific/welding-design.shtml (2014, accessed 5 September
aero engine components. In: IAENG world congress on 2014).
engineering, London, 4–6 July 2012. Hong Kong: 96. Altair Engineering Inc. Altair Hyperworks home page,
IAENG. http://www.altairhyperworks.com/ (2014, accessed 5
80. Yang J, Bin H, Zhang X, et al. Fractal scanning path gen- September 2014).
eration and control system for selective laser sintering 97. Bangerth W, Burstedde C, Heister T, et al. Algorithms
(SLS). Int J Mach Tool Manu 2003; 43(2): 293–300. and data structures for massively parallel generic adap-
81. Kruth JP, Deckers J, Yasa E, et al. Assessing and com- tive finite element codes. ACM T Math Software 2011;
paring influencing factors of residual stresses in selective 38(2): 14.
laser melting using a novel analysis method. Proc 98. Salonitis K and Stavropoulos P. On the integration of the
IMechE, Part B: J Engineering Manufacture 2012; 226: CAx systems towards sustainable production. Procedia
980–991. CIRP 2013; 9: 115–120.
82. Verhaeghe F, Craeghs T, Heulens J, et al. A pragmatic 99. Sivakumar S and Dhanalakshmi V. An approach towards
model for selective laser melting with evaporation. Acta the integration of CAD/CAM/CAI through STEP file
Mater 2009; 57(20): 6006–6012. using feature extraction for cylindrical parts. Int J Comp
83. Krol TA, Westhaeuser S, Zaeh MF, et al. Development Integ M 2012; 26(6): 561–570.
of a simulation-based process chain—strategy for differ- 100. Houshmand M and Valilai OF. A layered and modular
ent levels of detail for the preprocessing definitions. In: platform to enable distributed CAx collaboration and
ASIM 21. Symposium Simulationstechnik der Arbeitsge- support product data integration based on STEP stan-
meinschaft Simulation, Winterthur, 7–9 September 2011. dard. Int J Comp Integ M 2013; 26(8): 731–750.
Bonn: ASIM. 101. De Amorim Almeida H and Da Silva Bártolo PJ. Vir-
84. Köhler H, Jayaraman V, Brosch D, et al. A novel thermal tual topological optimisation of scaffolds for rapid pro-
sensor applied for laser materials processing. Phys Proce- totyping. Med Eng Phys 2010; 32(7): 775–782.
dia 2013; 41: 502–508. 102. Kerbrat O, Mognol P and Hascoët JY. A new DFM
85. Hutchings MT, Withers PJ, Holden TM, et al. Introduc- approach to combine machining and additive manufac-
tion to the characterization of residual stress by neutron dif- turing. Comput Ind 2011; 62(7): 684–692.
fraction. 1st ed.Boca Raton, FL: CRC Press, 2005, p.2. 103. Ari I and Muhtaroglu N. Design and implementation of
86. Hofmann M, Schneider R, Seidl GA, et al. The new a cloud computing service for finite element analysis.
materials science diffractometer STRESS-SPEC at FRM- Adv Eng Softw 2013; 60–61: 122–135.
II. Physica B 2006; 385–386(2): 1035–1037. 104. Paszyński M, Pardo D, Torres-Verdı́n C, et al. A paral-
87. Ding J, Colegrove P, Mehnen J, et al. A computationally lel direct solver for the self-adaptive hp Finite Element
efficient finite element model of wire and arc additive Method. J Parallel Distr Com 2010; 70(3): 270–281.
manufacture. Int J Adv Manuf Tech 2014; 70(1–4): 105. Gardan N. Knowledge management for topological
227–236. optimization integration in additive manufacturing. Int
88. Ding J, Colegrove P, Mehnen J, et al. Thermo-mechani- J Manuf Eng 2014; 2014: 356256 (9 pp.).
cal analysis of Wire and Arc Additive Layer
Schoinochoritis et al. 117