Bikas 2015
Bikas 2015
Bikas 2015
DOI 10.1007/s00170-015-7576-2
ORIGINAL ARTICLE
Abstract Additive manufacturing is a technology rapidly layer upon layer, as opposed to subtractive manufacturing
expanding on a number of industrial sectors. It provides de- methodologies, such as traditional machining” [1]. AM can
sign freedom and environmental/ecological advantages. It deliver parts of very intricate and complex geometries with a
transforms essentially design files to fully functional products. minimum need for post-processing, built from tailored mate-
However, it is still hampered by low productivity, poor quality rials with near-zero material waste, while being applicable to a
and uncertainty of final part mechanical properties. The root variety of materials, including plastics and metals. Therefore,
cause of undesired effects lies in the control aspects of the AM is a tool that offers increased “design freedom” and en-
process. Optimization is difficult due to limited modelling ables designers and engineers to create unique products that
approaches. Physical phenomena associated with additive can be manufactured at low volumes in an economical way.
manufacturing processes are complex, including melting/ An indicative example of the design freedom offered is that
solidification and vaporization, heat and mass transfer etc. conventional assemblies can be redesigned in a single com-
The goal of the current study is to map available additive plex structure that could not be manufactured with the current
manufacturing methods based on their process mechanisms, manufacturing processes. Another driver of the AM technol-
review modelling approaches based on modelling methods ogy is that it is environmentally and ecologically promising.
and identify research gaps. Later sections of the study review Additive manufacturing technologies and methods are in-
implications for closed-loop control of the process. creasing constantly in terms of application and market share,
spreading into various manufacturing divisions, such as auto-
Keywords Additive manufacturing . Process control . motive, medical and aerospace, and are expected that this
Process mechanisms . Modelling methods heavy growth will continue over the next few years. Accord-
ing to the Gartner report [2], sales of sub-$100,000 AM ma-
chines have grown worldwide by 49 % in 2014 and are ex-
pected to have reached a 75 % growth rate by the end of 2014.
1 Introduction In terms of materials processed, plastics are currently lead-
ing the AM market, with around 30,000 machines in produc-
Additive manufacturing (AM) is defined as “the process of tion [3]; however, the metal AM market is also growing. With
joining materials to make objects from 3D model data, usually over 1500 machines sold to date (Fig. 1), it is expected that the
metal AM machines will see double-digit percentage growth
* G. Chryssolouris in their sales, over the next 5 years, despite the global
xrisol@lms.mech.upatras.gr recession.
Indeed, according to [5], the industry’s average annual
1
Laboratory for Manufacturing Systems and Automation, Department
growth (CAGR) over the past 25 years is impressive, namely
of Mechanical Engineering and Aeronautics, University of Patras, 25.4 % (Figs. 2 and 3).
Patras 265 00, Greece In the last few years, there is a significant trend towards
2
Department of Aeronautical Studies, Hellenic Air Force Academy, metal AM for the production of structural components, mainly
Dekelia Air-Force Base, TGA 1010 Athens, Greece in areas, such as aerospace and motorsport applications, that
Int J Adv Manuf Technol
bears high operating and changeover costs due to the system’s reversible metastable state, which is subsequently polymer-
complexity. ized by the radiations of the second laser. The process is as-
sociated with various technical limitations such as insufficient
Liquid thermal polymerization (LTP) LTP is a process sim- absorption of laser radiation at higher depths, shadowing ef-
ilar to SLA in the way that the part is built by solidification of fects of the already solidified material and diffraction of laser
successive layers of liquid polymer. However, the polymers light, leading to difficulties in obtaining the precise intersec-
used in LTP are thermosetting instead of photopolymers and tion of the beams [13].
hence, the solidification is induced by thermal energy rather
than light. The thermal nature of the process makes the control Holographic interference solidification (HIS) In this pro-
of the size of the polymerization zone difficult, due to the cess, a holographic image is projected on a liquid photosensi-
dissipation of heat [13], therefore, the parts produced by this tive polymer contained in a vat so as for the entire surface of
method are less accurate. Nevertheless, the process has a rel- the polymer to be solidified, instead of point-by-point [13]. In
atively high throughput and can be considered in applications that essence, the process is really similar to that of solid
where accuracy is not an issue. ground curing.
a desktop prototyping facility, since it uses cheap, non-toxic, 2.3.2 Inkjet printing (IJP)
odourless materials, in a variety of colours and types, such as
acrylonitrile butadiene styrene (ABS), medical ABS, PLA, IJP is a type of computer printing that creates a digital image
investment casting wax and elastomers [17]. The simplicity by propelling droplets of ink onto paper, plastic or other sub-
of the FDM process, the relatively cheap equipment and the strates. Inkjet printers are the most commonly used types of
raw materials render its use ideal by hobbyists as well as the printers and range from small inexpensive consumer models
production of low-cost plastic parts. However, accuracy and to very large professional machines that can cost tens of thou-
surface quality are relatively poor when compared to those of sands of dollars or more [23].
powder-based plastic AM processes.
2.3.4 Ballistic particle manufacturing (BPM) 2.4.1 Laminated object manufacturing (LOM)
The BPM process involves a stream of molten droplets ejected The material used in LOM is a special kind of paper, having a
from piezoelectric inkjet printing nozzles to be deposited on heat-sensitive adhesive applied to one of its sides. The paper is
the target substrate. The process still uses the 3D data of the supplied from a roll and is bonded to the previous layer with
solid model to position the stream of material on the substrate. the use of a heated roller, which activates the paper’s adhesive.
Since the process is based on the material’s melting, it is par- The contour of the layer is cut by a CO2 laser, carefully mod-
ticularly suited for the materials, namely thermoplastics and ulated to penetrate into a depth of exactly one layer (paper)
metals that easily melt and solidify [13]. thickness. Surplus waste material is trimmed to rectangles to
facilitate its removal but remains in place during build-in order
to be used as support. The sheet of material used is wider than
2.3.5 Thermojet the building area, so that, when the part layer has been cut, the
edges of the sheet remain intact in order to be pulled by a take-
Thermojet is a process similar to multijet modelling. The sys- up roll and thus to continuously provide material to the next
tem generates wax-like plastics models, albeit with less accu- layer [17].
racy than SLA. The machine uses a wide area head with mul-
tiple spray nozzles. These jetting heads spray tiny droplets of
melted liquid material which cool and harden on impact to 2.4.2 Solid foil polymerization (SFP)
form the solid object. This process is commonly used for the
creation of casting patterns in the jewelry industry and other The process is based on complete polymerization of semi-
precision casting applications. polymerized plastic foils on exposure to suitable light source.
The semi-polymerized foil is first stacked on the previously
solidified part and then illuminated such that bonding is
2.4 Adhesive achieved after complete polymerization. The excess foil that
is not illuminated can be removed by being dissolved into
Adhesive-based processes are of limited use nowadays. The suitable solvent, leaving behind the desired part [13].
operating principle involves (usually a laser) a cutter, which
cuts a thin film of paper or plastic in the desired outlines. The
film is then pressed down onto the previous one by a heated 2.5 Electron beam
compactor, thus activating a heat curing adhesive present on
the downwards face of the film, in order to be bonded to the Electron beam processes are identical to the laser-melting pro-
substrate (Fig. 8). cesses but instead of a laser beam, an electron beam is used as
an energy source in order to melt or sinter the material (Fig. 9).
Fig. 8 Adhesive AM process schematic [9] Fig. 9 Electron beam AM process schematic
Int J Adv Manuf Technol
2.5.1 Electron beam manufacturing (EBM) depending on the development approach. The analytical
models are the output of the process’s mathematical analysis,
EBM is a relatively new but rapidly growing process similar to taking into consideration the physical laws and the relevant
SLS, albeit suitable for building metallic parts only. Powder is physical processes. The main advantage of such models is that
melted by an electron beam powered by a high voltage, typically the derived results can be easily transferred to other pertinent
30–60 KV. The process takes place in a high vacuum chamber to processes. The limits of the analytical modelling are deter-
avoid oxidation issues. EBM can also process a high variety of mined by the underlying assumptions. The empirical models,
pre-alloyed metals [24]. When compared to SLS, EBM can offer on the other hand, are the outcome of a number of experi-
much higher throughput and more uniform thermal field distri- ments, whose results are evaluated; one model type is chosen,
bution; however, accuracy and surface quality are lower. the coefficients are determined, and then the empirical model
can be verified by further tests. The quality of the models’
2.6 Comparative table results is limited in the special conditions of the specific pro-
cess. Their major advantage, compared with that of the ana-
The categorization of AM processes can be summarized in lytical models, is that they require minimum effort. Numerical
Table 1. models are in between, in essence, stem from the physics of
the process, but a numerical step-by-step method is employed
over time in order for useful results to be produced. Following
3 Modelling approaches is a list of authors who have presented modelling attempts in
AM.
The AM processes are hampered mainly due to their low Table 2 presents a categorization of those attempts, based
productivity, relatively poor surface quality and dimensional on the method (analytical, numerical and statistical/empirical)
stability as well as uncertainty regarding the mechanical prop- used to model the desired attribute. Moreover, Table 3, at the
erties of the products. Therefore, those manufacturing attri- end of Section 3, presents modelling approaches categorized
butes should be optimized in order for AM to get established on the basis of the modelled attribute, both for the AM prod-
in production. For the optimization of any manufacturing pro- ucts (accuracy, roughness, integrity etc.) as well as for the
cess, a deep knowledge of the process itself is required. This process (phase change, heat transfer etc.)
knowledge could be gained, either by experimentation or by
analysing the physical mechanisms of the process. A model is 3.1 Laser-based processes
the abstract representation of a process that establishes a rela-
tion between input and output quantities. The real system is In laser-based processes, the most usual modelled aspects are
simulated by the models that aim to predict its performance. the interaction between the laser beam and the material and the
Models met in literature can be divided into three major associated phase changes, either from liquid to solid
categories, namely analytical, numerical and empirical ones, (photopolymerization) or from solid to liquid (melting).
3.1.1 Laser polymerization techniques and process the results with the response surface
methodology and analysis of variance (ANOVA) methods, to
Chryssolouris in [88] has used a semi-empirical approach investigate into the optimal parameters (including layer thick-
through statistical design to extract a model, regarding the ness, hatch spacing, hatch style, hatch overcure, blade gap,
dimensional accuracy of AM parts built via stereolithography; position on the build plane) of the platform of the SLA pro-
thus identifying the most crucial process parameters that affect cess. The issue has also been experimentally modelled, by
accuracy in each direction built. Retraction (the distance of the [36–38, 89] having presented experimental studies on the di-
hatch vector end from the layer border) is not only the dom- mensional accuracy of the SLA parts and the associated pro-
inant parameter in the x-direction but also important to the y- cess parameters. A process planning method was developed
direction accuracy; whereas, in the z-direction, the shrinkage by Lynn et al. [41, 42] and was further improved by West et al.
compensation function was found to be the most influential. [34] in order to develop response surface models for the eval-
Zhou et al. [47] as well as others [39, 40] have used a similar, uation of the SLA parts’ accuracy. An interesting approach
semi-empirical approach by the Taguchi experimental design was presented by Cho et al. [48], modelling the SLA process
Surface Dimensional Dimensional Build Mechanical Heat transfer Material Phase Other
roughness accuracy stability time properties deposition/ change
melt pool
Laser-based SLA [25, 26, [34–42, 48, [27, 43–46, [30–32, [25, 29, 30] [25, 29,
(polymerization) 49] 88, 89] 90] 91] 30]
Laser-based SLS [54, 59] [53, 92, 93] [50, 51, 53, 84, [51, 52,
(melting) 92–94] 62]
SLM [61] [63] [60, 63] [52, 62]
DMD [56] [64] [64]
LPD [64, 65] [64]
SLC [67] [66] [66] [55]
Extrusion FDM [58] [74] [77, 80] [68, 70, 72, 75, [73] [57, 68–71,
(thermal) 76] 79]
Material jetting 3DP [81] [65, 82]
Adhesive LOM [33]
Electron beam EBM [83] [83–85] [86, 87] [83]
Int J Adv Manuf Technol
via a genetic algorithm model in order to determine the opti- volume built by SLA. Flach et al. [27] integrated an analytical
mal process parameters (which include layer thickness, hatch resin shrinkage model into the general SLA process model
spacing and hatch overcure) that would yield the minimum developed in [28], to have a theoretic prediction of the dimen-
part build error. Chryssolouris et al. [25] has estimated the sional stability due to resin shrinkage, concluding that faster
average surface roughness of SLA-produced parts as a func- shrinking resins should result in lower overall shrinkage
tion of the layer thickness and the angle of the inclined surface values. It has been found that the overall linear shrinkage,
(Fig. 10). Modelling was based on simplistic trigonometry due to cure for a line of plastic, was estimated to have been
assumptions, while the surface roughness could be calculated given by the equation:
according to the following equation:
ZL
"rffiffiffiffi #
2 PL sinθ sinθ F C ¼ 1=L f r ðY Þdy ð2Þ
Ra ¼ Dp ⋅ln −OC ð1Þ
π W 0 V S E C 4tanθ 4tanθ 0
where:
where:
fr(y) residual fractional linear shrinkage at position y
Dp depth of penetration
FC overall fractional linear shrinkage due to cure
PL laser power
L length of strand of plastic (cm)
W0 laser beam spot diameter
t time (sec)
VS laser scanning speed
ts time for laser to scan from position y to L (sec)
EC critical exposure time
OC overcure
Chryssolouris [25], Jelley [29] and Jacobs [30] investigated
Reeves and Cobb in [26] and [49] presented an analytical the polymerization process that occurs during SLA
model for SLA surface roughness that took into consideration manufacturing, based on the modelling of the laser source,
the layer profile as well whether the plane was up-facing or the modelling of the photo-initiated free radical polymeriza-
down-facing, which was verified with experimental data. tion and the modelling of the heat transfer involved in the
Podshivalovab et al. [35] has used a 3D model to verify the process. A few have dealt with modelling the build time in
dimensional accuracy of scaffold-like structures used in bone the SLA process. Chen [31], Giannatsis [91] and Kechagias
replacement via CAD and FEA. Part dimensional stability has [32] have calculated the process time analytically. Kechagias
been experimentally studied by a number of researchers. [32] has presented a method where the total distance travelled
Rahmati [43] studied dimensional stability in SLA as a result by the laser beam is directly calculated from the part geometry
of resin shrinkage; Wang et al. [44] studied the effect of the (STL file). The time required for each layer to be produced is
post-curing duration, the laser power and the layer pitch on the then calculated analytically on the basis of the laser velocity,
post-cure shrinkage and empirical relations were established keeping in mind whether the laser is performing border draw-
on the basis of the least squares method. The shrinkage strains ing, hatching or filling. Furthermore, the time required for all
were investigated by Karalekas and Aggelopoulos [45] based the auxiliary steps is estimated. Contouring and hatching ve-
on a simple experimental setup and the elastic lamination the- locities were calculated by:
ory. Narakahara et al. [46, 90] studied the relationship between vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u2 P
the initial linear shrinkage and resin temperature in a minute CvðiÞ ¼ u tπ L . ; ð3Þ
Cd ðiÞ Dp
W 0 EC e
mP
HvðiÞ ¼ L. ð4Þ
Cd ðiÞ Dp
hsE C e
where:
PL laser power
W0 laser beam half width
Cd curing depth
hs hatching space (distance between neighbour scanning
vectors)
m number of times the slice area is hatched
Ec critical energy
Fig. 10 Trigonometry used by Chryssolouris [25] Dp penetration depth
Int J Adv Manuf Technol
It has been found that the hatching time, calculated as the temperature-transforming 2D numerical model. This model
ration between the total hatching length Hl(i) and the hatching converts the enthalpy-based energy equation into a nonlinear
velocity Hv(i) deviates from the actual one, due to delays equation with temperature being the only dependent variable.
occurring during hatching; further experimental study of the Vasinonta et al. [64] have developed two process maps for laser-
hatching procedure has led to the following equation, the ac- based solid freeform fabrication processes, through the investi-
curacy of which was found to be within 0.5 %. gation of the melt pool size and the residual stresses developed
via finite element thermomechanical models. Michaleris [60]
Hl ðiÞ
THatch ¼ þ 0:0005NoVectors investigated two finite element techniques for modelling metal
HvðiÞ deposition and transient conducting heat transfer in SLM. Par-
Hl ðiÞ ticularly, quiet and inactive element activation is probed. In the
þ −0:206: ð5Þ quiet case study, the elements are manipulated through their
54HvðiÞ þ 6114
properties in order to be neutral. In the inactive case study, the
Jacobs [30] assumed that the laser presents a Gaussian dis- elements are not included in the analysis until the related mate-
tribution and the absorption of the laser radiation within the rial has been added. Analytical equations have been used for the
resin follows the Beer-Lambert Law. Based on these assump- formulation of transient conductive heat transfer. The evaluation
tions, he calculated the maximum depth of the cured line. Based has been performed by the use of 1D and 3D FEA models.
on the relation among the maximum exposure, the laser power, Zhiqiang [62] investigates the topic of fusion-based additive
the half beam width and the scanning speed, required for their manufacturing of titanium through the use of mathematical
solidification to a specific cure depth, were calculated: and numerical modelling in order to model the melt pool. The
rffiffiffiffi s numerical model was correlated with experimental measure-
2 PL C ments. Melt pool monitoring was an attribute also investigated
Vs ¼ exp − d ð6Þ
π W 0 EC Dp into by Hu and Kovacevic in [52] using a 3D numerical model.
Matsumoto et al. [63] proposed a simulation method about a
Chen et al. [31], based on Jacobs’ work [30], developed an single layer deposition of the SLM process. The stress and
analytical model for predicting total build time, by incorporat- temperature within the solid layer have been calculated with
ing a correction factor based on experimental work. The mod- the heat conduction and linear finite elements. Kolossov et al.
el has been found to be in good correlation with experimental [94] modelled the heat transfer and thermal field using a non-
data for a number of different parts. linear 3D model, which took into consideration the thermal
characteristics of bulk material, such as thermal conductivity
and heat capacity and the thermal history of the material in each
3.1.2 Laser melting step. Dong et al. [53] created a 3D FEA model to predict statis-
tically the temperature and density distribution in the SLS of
Wang and Kruth [50] modelled the laser beam scanning and polycarbonate. The parameters analysed include the laser beam
the energy absorption, taking place in an SLS machine, by velocity, laser power and laser diameter, and the results have
using the analytical ray-tracing model which allowed the cal- been verified against experimental values found on related lit-
culation of the sintering zone dimensions (width and thick- erature. A similar approach was followed by Liu [92], develop-
ness). The energy necessary to fuse a certain powder particle ing a micro-scale 3D FEM in order to investigate the character-
can be calculated according to the following equation: istics of the temperature distribution within the powder bed.
Giovanni et al. [61] investigated the surface roughness and
E m ¼ cp ΔT þ cl ρ V ð7Þ
morphology in SLM parts. A mathematical model is introduced
where: for the prediction of surface roughness. Both the stair step effect
and the increasing presence of particles on the top surface are
cp specific heat (KJ/KgK)
considered key contributing factors to the surface morphology.
ΔT temperature rise required for melting (K)
In Khairallah [93], a mesoscopic 3D model has been introduced
cl latent melt energy (KJ/Kg)
to simulate the SLM processes by using the commercial code
ρ density (kg/mm3)
ALE3Dmulti-physics. The simulation includes the substrate,
V volume of the (spherical) particle (mm3)
the random particle melding and re-solidification either into a
A simple comparison of the absorbed energy Ei to Em will continuous or a discontinuous track. The model is taking into
determine whether any particle absorbs enough energy to melt consideration the thermal diffusion to hydrodynamics, the
or not. The sintering zone dimension is evaluated from the temperature-dependent material properties, the surface tension
most side-wise molten particles. and the random particle distribution.
Chen and Zhang [51] investigated the parameters affecting Various researchers have investigated shrinkage in SLS.
the sintering depth and shape of the liquid pool by formulating a Chen and Zhang [54] created a partial shrinkage model of
Int J Adv Manuf Technol
two component metal powder mixture with different melting na apparent viscosity determined using a capillary
points, in order to investigate the effects of volume fraction of rheometer
the gas in the liquid or sintered regions on the shape and size of Q volumetric flow rate
the heat-affected zone. Furthermore, Raghunath and Pandey
It was calculated that the filament material would buckle if:
[59] utilized the Taguchi method for the determination of the
influence of the SLS process parameters to shrinkage, as well as 8QlðL=RÞ2
to define the optimum shrinkage conditions. E=ηα < ð9Þ
π3 r 4 k
LPD, DMD and SLC are variations of the same process,
where a powder is injected into a laser beam and melted simul- where:
taneously with a thin layer of the substrate to form a continuous L/R slenderness ratio of the filament
track of material. Costa et al. [95] developed a thermo-kinetic FE k scaling factor (experimentally determined)
model of multilayer LPD for the investigation of the microstruc-
tural transformations and hardness variations that occur during Yardimici [70] and Agarwala [71] have studied the issue of
the deposition of steel parts for the calculation of the final hard- filament buckling in a liquefier entry by theoretical means. An
ness distribution in the part. The results obtained were in agree- experimental investigation into the same phenomenon was
ment with published experimental observation. Gockel et al. [83] conducted by Venkataraman [79]. Heat transfer to the built
have developed an FE model capable of simulating the material material inside the liquefier was theoretically studied by [68,
solidification of Ti64 in an electron beam wire feed AM process 70, 72], while a number of authors dealt numerically with the
resulting in the prediction of the material microstructure and same issue [68, 72, 75, 76].
properties. A similar model of the LPD of titanium was devel- Ramanath [72] having used Bellini’s previous work [68]
oped by Crespo et al. [66] in order to study the influence of the has mathematically modelled the pressure drop during the
deposition path geometry on the melt pool stability as well as to extrusion of PCL.
estimate the adjustments of the deposition parameters, necessary Using the law of non-Newtonian polymer melt flow
to avoid hot spots. The same thermo-kinetic model has been γ=φτm, the pressure drops for each of the five zones were
used in order for the hardness distribution of the SLC process derived, by considering Bellini’s work [68].
to be predicted [67]. Toyserkani et al. [55] investigated the ef- m1 1
fects of laser pulse shaping on the SLC process with a 3D FE V mþ3 m 1 1
ΔP1 ¼ 2L1 exp α −
model. In Muller’s work [56], an analytical model of the direct ϕ r1mþ1 T −T 0 T α −T 0
LPD process has been developed, for functionally graded mate- ð10Þ
rials (FGM). Particularly, this study focuses on the operation of a m1 m1
powder distribution system. The model scope is to be utilized in V mþ3 1 1
ΔP2 ¼ 2L2 exp α −
order for manufacturing strategies, in the production of parts, to ϕ r1mþ1 T −T 0 T α −T 0
be compared with complex material distribution. ð11Þ
m1 m1
V mþ3 1 1
ΔP3 ¼ 2L3 exp α −
3.2 Extrusion processes ϕ mþ1
r1 T −T 0 T α −T 0
ð12Þ
FDM is a process modelled for various attributes. Zhang and ! 1
Chou [73] developed a 3D FEA model to simulate the FDM 2m 1 1 V m 2 mþ3 1 1 1
ΔP4 ¼ α − r1 2 ðm þ 3Þ m exp α −
3tan
3 3
r2m r1m ϕ T −T 0 T α −T 0
process melt pool. The same model was used and enhanced by 2
Zhang in [74] for the simulation of the residual stresses in order ð13Þ
for part distortion to be evaluated. Prototype parts were built and m1 2 m1
used to validating the simulated results. Bellini et al. [68] and V r 1 ð m þ 3Þ 1 1
ΔP5 ¼ 2L5 exp α −
Venkataraman et al. [69] have analytically modelled the material ϕ r2mþ3 T −T 0 T α −T 0
flow on the extrusion nozzle. Venkataraman et al. [69] investi- ð14Þ
gated the material buckling in the liquefier using Euler’s analysis
for buckling and a capillary rheometer. The pressure drop (ΔP) ΔP ¼ ΔP1 þ ΔP2 þ ΔP3 þ ΔP4 þ ΔP5 ð15Þ
in a capillary rheometer required that a non-Newtonian fluid be
driven through a tube of length l and radius r is: where:
R2 radius of the channel at zone 2 Key flow parameters, namely pressure, velocity and tempera-
r1 radius of the cylindrical area of the melt flow channel ture, have been investigated into.
r2 the exit radius
a nozzle angle
V filament velocity at the entry 3.3 Material jetting processes
u fluidity
m flow exponent 3DP has also been the subject of various researchers’ study.
T working temperature Jee and Sachs [81] proposed a visual simulation technique so
Ta the temperature at which m and u are calculated as to facilitate the surface texture designs to be produced by
To the absolute temperature 3DP. This technique simulates 3DP by taking into consider-
ation all the necessary geometric attributes of physical phe-
The model’s results were verified with those having de- nomena and therefore enables the realization of a manufactur-
rived from the use of an FE model in ANSYS. able design with minimum iterations. Sachs and Vezzetti [82]
Crockett has investigated the deposition and liquid-to-solid numerically modelled the deposition process of a new 3DP
transition phase of the FDM process, by developing an ana- head design in order to ensure a reliable and continuous jet
lytical model for bead spreading [57]. Sood et al. [80] have deposition, resulting in an order of magnitude increase in the
developed a semi-empirical methodology using neural net- printing speed. Curodeau [65] modelled the drop merging (the
work algorithms to develop a model that would predict the phenomenon where, in a uniformly spaced train of drops, the
compressive strength of FDM built parts. Martinez et al. [77] leading drop is retarded by air drag and tends to merge with
have used a methodology usually applied to fibrous composite the drop behind it) in order to evaluate the number of merged
materials for the characterization of parts built by FDM using drops for various distances and printing conditions.
FEA. Anitha et al. [58] focused on optimizing the FDM pro-
cess surface quality. Taking into consideration the Taguchi’s
analysis, three variables have been investigated into, those of 3.4 Adhesive processes
the road width, the build layer thickness and the speed of
deposition. In addition, analysis of variance (ANOVA) has Chryssolouris et al. [33] have used a semi-empirical approach,
been performed with the same parameters. Mostafa et al. through a statistical design in order to model the surface
[78] have performed both 2D and 3D numerical analysis on roughness of the LOM process. The specific model can pre-
flow behaviour of ABS-iron composite in FDM, through the dict the part’s surface roughness for any combination of pro-
ANSYS FLOTRAN and the CFX finite element packages. cess variables.
Fig. 11 Sectional view of melt flow channels showing five zones considered by [75]
Int J Adv Manuf Technol
Laser power/ Melt pool Nozzle Jet status Chamber Chamber Platform Head
distribution temperature temperature temperature vacuum position position
radiation intensity emitted from a black body. On brightness or the material phase change. Subsequently, modelling processes
colour temperature determination, the intensity of the thermal in the field of additive manufacturing were presented and cat-
radiation from the surface, in the region of laser action, is re- egorized, based not only on the operating principle but also on
corded by a CCD camera or a pyrometer in one or two spectral the modelled process attribute and the modelling
intervals and is correlated with the ones of the black body methodology.
simulator, located in the same surface region. The degree of The most commonly modelled AM process is that of the
its approximation to the thermodynamic temperature is defined SLA, followed by the SLS/SLM and FDM. Most authors deal
by the accuracy of the material emissivity determination. In with modelling dimensional accuracy/stability, while quite a
creation of the temperature monitoring system, a combination few others deal with predicting the mechanical properties of
of two types of optical sensors—2D sensor—a digital CCD the finished product as well as the total build time.
camera and a single spot sensor pyrometer, based on photodi- The most utilized approach to the issue of dimensional
odes, which integrate thermal radiation emitted by a surface of accuracy is by empirical models, via statistical methods
certain size, are used [97]. This technique could also be used in (ANOVA etc.). Mechanical properties and dimensional stabil-
extrusion, requiring that the nozzle temperature for the mate- ity modelling are usually made by numerical heat transfer
rial’s liquidity control be measured. models, studying mainly the melt pool and the material phase
change, while build time has been investigated both analyti-
4.2 Nozzle monitoring (jetting) cally and numerically.
However, most studies present either a theoretical approach
As it has been referred to, in the previous section, one of the with little to no verification compared to that of real-life re-
dominant elements, in many of the existing additive sults, or semi-empirical approaches that may correlate well
manufacturing machines, is that of the nozzles. To ensure with specific experiments, but their results are not directly
productivity and reliability as manufacturing tools, the nozzle transferrable and expandable to other machines, requiring fur-
status needs to be monitored, while the jet failures should be ther experimentation. The AM processes could significantly
identified. To detect jet failures, the use of piezo self-sensing benefit from accurate, verified models, aided by the use of
signals has been proposed. A piezo inkjet head uses a piezo machine-integrated monitoring systems in order to be able to
actuator to jet ink droplets. In addition, the piezo actuator can back up the models with real, measured data. Given the turn of
be used as a sensor, by sensing the force that results from the industry to metal AM, the models of laser-based AM and
pressure wave of ink inside the inkjet dispenser. The possible EBM metals are of utmost relevance, especially in the thermal
causes of jet failures include the inkjet head temperature con- field, dimensional stability and residual stresses, since these
trol failure, the backpressure control failure, wetting on the factors significantly affect the quality and safety of the final
nozzle surface, the nozzle blockage, etc. For the verification product.
and the detectability of these jet failures, jet images could be
acquired for comparison with self-sensing signals [98].
Open Access This article is distributed under the terms of the Creative
Commons Attribution 4.0 International License (http://
4.3 Mechanical parts: rollers—building creativecommons.org/licenses/by/4.0/), which permits unrestricted use,
platforms—material supplier distribution, and reproduction in any medium, provided you give appro-
priate credit to the original author(s) and the source, provide a link to the
Creative Commons license, and indicate if changes were made.
The mechanical parts’ motion is usually accompanied by
monitoring the applied torque or force, with the use of strain
gauges or by monitoring the back EMF of the actuator. Fur-
References
thermore, for the machine’s general function, the optical data
could be harvested with the use of photo and video recording
1. ASTM Standard. Standard terminology for additive manufacturing
devices. technologies, vol. 10.04
2. Gartner AM report. Information on http://www.gartner.com/
document/2598122 Accessed 16 June 2014
5 Conclusions 3. AM Platform: Additive Manufacturing Strategic Research Agenda:
Release 2014. Available online http://www.rm-platform.com/
linkdoc/AM%20SRA%20-%20February%202014.pdf Access 5
In the present paper, a review of all AM techniques has been Mar 2015
conducted, followed by a review and assessment of modelling 4. Wohlers TT (2014) Wohlers report 2014: additive manufacturing
approaches. The AM techniques have been categorized based and 3D printing state of the industry: annual worldwide progress
report. Fort Collins, Wohlers Associates
on their operating principle rather than on the materials used, 5. Wohlers TT (2013) Wohlers report 2013: additive manufacturing
albeit this has also been kept in mind. The categorization was and 3D printing state of the industry: annual worldwide progress
made in a way that it respected the underlying physics behind report. Fort Collins, Wohlers Associates
Int J Adv Manuf Technol
6. Wohlers TT (2012) Wohlers report 2012: additive manufacturing 29. Jelley C, Thompson CP The development and application of a
and 3D printing state of the industry: annual worldwide progress Stereolithography build simulator. Proceedings of the First
report. Fort Collins, Wohlers Associates National Conference on RP & Tooling Research
7. Kruth JP, Levy G, Klocke F, Childs THC (2007) Consolidation 30. Jacobs PF (1992) Fundamentals of stereolithography. Proceedings
phenomena in laser and powder-bed based layered manufacturing. of the Solid Freeform Fabrication Symposium, pp 196–211
CIRP Ann Manuf Technol 56(2):730–759 31. Chen C, Sullivan P (1996) Predicting total build-time and the re-
8. Pham DT, Dimov SS (2001) Rapid manufacturing, Springer- sultant cure depth of the 3D stereolithography process. Rapid
Verlag, p 6 Prototyp J 2(4):27–40
9. Woesz A (2010) Rapid prototyping to produce porous scaffolds 32. Kechagias J, Anagnostopoulos V, Zervos S, Chryssolouris G
with controlled architecture for possible use in bone tissue engineer- (1997) Estimation of build times in rapid prototyping processes.
ing. Virtual Prototyp Bio Manuf Med Appl 171–206 6th European Conference on Rapid Prototyping &
10. 3DSystems SLA production series brochure, Information on http:// Manufacturing, Nottingham, pp 137–148
www.3dsystems.com/sites/www.3dsystems.com/files/sla-series- 33. Chryssolouris G, Kechagias JD, Kotselis JL, Mourtzis DA, Zannis
0514-usen-web.pdf Accessed 3 May 2015 SG Surface roughness modelling of the Helisys laminated object
11. Salonitis K, Tsoukantas G, Stavropoulos P, Stournaras A (2003) A manufacturing (LOM) Process. 8th European Conference on Rapid
critical review of stereolithography process modeling, (VRAP 03), Prototyping and Manufacturing, Nottingham, pp 141–152
3rd International Conference on Advanced Research in Virtual and 34. West AP, Sambu SP, Rosen DW (2001) A process planning method
Rapid Prototyping, Leiria, pp 377–384 for improving build performance in stereolithography. J Comput-
12. Gebhardt IA (2003) Rapid prototyping: industrial rapid prototyping Aided Des 33:65–79
system: prototyper: solid ground curing, Cubital, pp. 105–109 35. Podshivalovab L, Gomesc CM, Zoccac A, Guensterc J, Yosephb
13. Dahotre NB, Harimkar S (2008) Laser fabrication and machining of PB, Fischerb A (2013) Design, analysis and additive manufacturing
materials, Springer of porous structures for biocompatible micro-scale scaffolds.
14. Kruth JP (1991) Material incress manufacturing by rapid Procedia CIRP 5:247–252
prototyping techniques. Keynote Paper, CIRP Ann - Manuf 36. Schaub DA, Chu K, Montgomery DC (1997) Optimizing
Technol 40(2):603–614 stereolithography throughput. J Manuf Syst 16(4):290–303
15. Castoro M (2013) Impact of laser power and build orientation on 37. Lan PT, Chou SY, Chen LL, Gemmill D (1997) Determining fab-
the mechanical properties of selectively laser sintered parts. rication orientations for rapid prototyping with stereolithography
Proceedings of The National Conference on Undergraduate apparatus. J Comput-Aided Des 29(1):53–62
Research (NCUR). University of Wisconsin La Crosse, WI 38. Pang TH (1995) Accuracy of stereolithography parts: mechanism
16. 3DSystems SLS production series brochure, Information on http:// and models of distortion for a letter-H diagnostic part. Proceedings
www.3dsystems.com/sites/www.3dsystems.com/files/sls-series- of the Solid Free Form Fabrication Symposium, pp 170–180
0214-usen-web_1.pdf Accessed 23 Apr 2015 39. Onuh SO, Hon KKB (1998) Optimizing build parameters for im-
17. Chryssolouris G (2005) Manufacturing systems: theory and prac- proved surface finish in stereolithography. J Mach Tools Manuf
tice, 2nd edn. Springer, New York 38(4):329–392
18. Khaing MW, Fuh JYH, Lu L (2001) Direct metal laser sintering for 40. CarosiA, Pocci D, Luluiano L, Settimeri L (1996) Investigation on
rapid tooling: processing and characterization of EOS parts. J Mater stereolithography accuracy on both solid and QuickCast parts.
Process Technol 113:269–272 Proceedings of the 5th European Conference on Rapid
19. Vilar R (2014) Laser powder deposition. comprehensive materials Prototyping and Manufacturing
processing 2014. Adv Addit Manuf Tooling 10:163–216 41. Lynn CM, West A, Rosen DW (1998) A process planning method
20. Ready for Printing - 3D Printing at Siemens video. Information on and data format for achieving tolerances in stereolithography.
http://youtu.be/VyEgbyNg0Q8?t=3m46s Accessed 16 June 2014 Proceedings of Solid Freeform Fabrication Symposium
21. Liu J, Li L (2004) In-time motion adjustment in laser cladding 42. Lynn CC, Rosen DW (2000) Usage of accuracy models in
manufacturing process for improving dimensional accuracy and stereolithography process planning. Rapid Prototyp J 6(2):77–86
surface finish of the formed part. Optics Laser Technol 36:477–483 43. Rahmati S, Dickens PM (1995) Stereolithography process improve-
22. Cesarano J, King BH, Denham HB (1998) Recent developments in ment. Proceedings of the First National Conference on Rapid
robocasting of ceramics and multimaterial deposition, Proceedings Prototyping and Tooling Research, pp 111–126
of the 9th Solid Feeeform Fabricatuion Symposium, pp 697–704 44. Wang WL, Cheah CM, Fuh JYH, Lu L (1996) Influence of process
23. Pandremenos J, Paralikas J, Chryssolouris G, Dybala B, Gunnink parameters on stereolithography part shrinkage. J Mater Des 17:
JW (2008) RM product development: design principles, simulation 205–213
and tool. International Conference on Additive Technologies, Ptuj 45. Karalekas D, Aggelopoulos A (2003) Study of shrinkage strains in
24. Wong KV, Hernandez A (2012) A review of additive manufactur- a stereolithography cured acrylic photopolymer resin. J Mater
ing. Int Sch Res Netw Mech Eng 2012:1–10 Process Technol 6590:1–5
25. Chryssolouris G et al Deliverable D.2.2.2—development of an an- 46. Narahara H, Tamaka F, Kishimani T, Igarashi S, Saito K (1999)
alytical model. Research project “Flexible Assembly and Reaction heat effects on initial linear shrinkage and deformation
Manufacturing Engineering (FLAME)” funded by Greek in stereolithography. Rapid Prototyp J 5(3):120–128
Secretariat for Research and Technology (GSRT) 47. Zhou JG, Herscovici D, Chen CC (2000) Parametric process opti-
26. Reeves PE, Cobb RC (1997) Surface roughness investigation of mization to improve the accuracy of rapid prototyped
Stereolithography ACES components. Proceedings of the Second stereolithography parts. J Mach Tools Manuf 40:363–379
National Conference on Rapid Prototyping and Tooling Research, 48. Cho HS, Park WS, Choi BW, Leu MC (2000) Determining optimal
pp 17–26 parameters for stereolithography process via genetic algorithm. J
27. Flach L, Chartoff RP (1994) A simple polymer shrinkage model Manuf Syst 19(1):18–27
applied to stereolithography. Proceedings of the Solid Freeform 49. Reeves PE, Cobb RC (1997) Reducing the surface deviation of
Fabrication Symposium, pp 225–233 stereolithography using in-process techniques. Rapid Prototyp J
28. Flach L, Chartoff RP (1992) Stereolithography process modelling – 3(1):20–31
a step towards intelligent process control. Proceedings of the Third 50. Wang XC, Kruth JP (2000) A simulation model for direct selective
International Conference on Rapid Prototyping, pp 141–147 laser sintering of metal powders. Computational Techniques for
Int J Adv Manuf Technol
Materials, Composites and Composite Structures, Civil-Comp, 70. Yardimci MA, Hattori T, Guceri SI, Danforth SC (1997) Thermal
Edinburgh, pp 57–71 analysis of fused deposition. In: Solid freeform fabrication proceed-
51. Chen T, Zhang Y (2004) Numerical simulation of two-dimensional ings. University of Texas at Austin, Austin
melting and resolidification of a two-component metal powder lay- 71. Agarwala MK, Jamalabad VR, Langrana NA, Safari A, Whalen PJ,
er in selective laser sintering process. Numer Heat Tran Part A 46: Danforth SC (1996) Structural quality of parts processed by fused
633–649 deposition. Rapid Prototyp J 2:4–19
52. Hu D, Kovacevic R (2003) Sensing, modeling and control for laser- 72. Ramanath HS, Chua CK, Leong KF, Shah KD (2008) Melt flow
based additive manufacturing. Int J Mach Tools Manuf 43(1):51–60 behaviour of poly-epsilon-caprolactone in fused deposition model-
53. Dong L, Makradi A, Ahzi S, Remond Y (2009) Three-dimensional ing. J Mater Sci-Mater Med 19:2541–2550
transient finite element analysis of the selective laser sintering pro- 73. Zhang Y, Chou YK (2006) Three-dimensional finite element anal-
cess. J Mater Process Technol 209(2):700–706 ysis simulations of the fused deposition modelling process. J Eng
54. Chen T, Zhang Y (2006) A partial shrinkage model for selective Manuf 220(10):1663–1671
laser sintering of a two-component metal powder layer. Int J Heat 74. Zhang Y, Chou YK (2008) A parameter study of part distortions in
Mass Transf 49:1489–1492 FDM using 3D FEA. J Eng Manuf 222:959–967
55. Toyserkani E, Khajepour A, Corbin S (2004) 3-D finite element 75. Mostafa N, Syed HM, Igor S, Andrew G (2009) A study of melt
modeling of laser cladding by powder injection: effects of laser flow analysis of an ABS-iron composite in fused deposition model-
pulse shaping on the process. Opt Lasers Eng 41:849–867 ling process. Tsinghua Sci Technol 14:29–37
56. Muller P, Mognol P, Hascoet JY (2013) Modeling and control of a 76. Ji LB, Zhou TR (2010) Finite element simulation of temperature field
direct laser powder deposition process for functionally graded ma- in fused deposition modeling. Manuf Sci Eng 97/101:2585–2588
terials (FGM) parts manufacturing. J Mater Process Technol 213(5): 77. Martínez J, Diéguez JL, Ares E, Pereira A, Hernández P, Pérez JA
685–692 (2013) Comparative between FEM models for FDM parts and their
57. Crockett RS, Calvert PD (1996) The liquid-to-solid transition in approach to a real mechanical behaviour. Procedia Eng 63:878–884
stereodeposition techniques. In: Solid freeform fabrication proceed-
78. Nikzad M, Hasan Masood S, Sbarski I, Groth A (2009) Thermo-
ings. University of Texas at Austin, Austin, pp 257–264 mechanical properties of a highly filled polymeric composites for
58. Anitha R, Arunachalam S, Radhakrishan P (2001) Critical param- fused deposition modeling. Tsinghua Sci Technol 14:29–37
eters influencing the quality if prototypes in fused deposition
79. Venkataraman N, Rangarajan S, Matthewson MJ, Safari A,
modeling. J Mater Process Technol 118(1–3):385–388
DanforthSC, Yardimci A (1999) Mechanical and rheological prop-
59. Raghunath N, Pandey PM (2007) Improving accuracy through
erties of feedstock material for fused deposition of ceramics and
shrinkage modelling by using Taguchi method in selective laser
metals (FDC and FDMet) and their relationship to process perfor-
sintering. Int J Mach Tools Manuf 47:985–995
mance. In Solid Freeform Fabrication Proceedings, University of
60. Michaleris P (2014) Modeling metal deposition in heat transfer
Texas at Austin, Austin, pp 351–360
analyses of additive manufacturing processes. Finite Elem Anal
80. Sood AK, Ohdar RK, Mahapatra SS (2012) Experimental investi-
Des 86:51–60
gation and empirical modelling of FDM process for compressive
61. Strano G, Hao L, Everson RM, Evans KE (2013) Surface roughness
strength improvement. J Adv Res 3(1):81–90
analysis, modelling and prediction in selective laser melting. J
81. Jee HJ, Sachs E (2000) A visual simulation technique for 3D print-
Mater Process Technol 213(4):589–597
62. Fan Z, Liou F (2012) Numerical modeling of the additive ing. Adv Eng Softw 31:97–106
manufacturing (AM) processes of titanium alloy, from “Titanium 82. Sachs E, Vezzetti E (2005) Numerical simulation of deposition
Alloys—Towards Achieving Enhanced Properties for Diversified process for a new 3DP printhead design. J Mater Process Technol
Applications”. InTech, Missouri University of Science and 161:509–515
Technology 83. Gockel J, Beuth J, Taminger K (2014) Integrated control of solid-
63. Matsumoto M, Shiomi M, Osakada K, Abe F (2002) Finite element ification microstructure and melt pool dimensionsin electron beam
analysis of single layer forming on metallic powder bed in rapid wire feed additive manufacturing of Ti-6Al-4V. Addit Manuf 1–4:
prototyping by selective laser processing. Int J Mach Tools Manuf 119–126
42(1):61–67 84. Shen N, Chou K (2012) Numerical thermal analysis in electron
64. Vasinonta A, Beuth J, Griffith M (2000) Process maps for control- beam additive manufacturing with preheating effects. Proceedings
ling residual stress and melt pool size in laser-based SFF processes. of the 23rd annual international solid freeform fabrication sympo-
In Solid Freeform Fabrication Proceedings, University of Texas at sium, pp 774–784
Austin, Austin, pp 200–208 85. Shen N, Chou YK (2012) Thermal modeling of electron beam
65. Curodeau A (1995) Three dimensional printing of ceramic molds additive manufacturing process—powder sintering effects. Proc.
with accurate surface macro-textures for investment casting of or- the 7th ASME 2012 International Manufacturing Science and
thopaedic implants. Thesis (Ph. D.)-Massachusetts Institute of Engineering Conference, pp 287–295
Technology, Dept. of Mechanical Engineering 86. Markl M, Ammer R, Ljungblad U, Ruede U, Koerner C (2013)
66. Crespo A, Deus AM, Vilar R (2006) Finite element analysis of laser Electron beam absorption algorithms for electron beam melting
powder deposition of titanium. Proceedings of ICALEO 2006, processes simulated by a three–dimensional thermal free surface
Scottsdale, Arizona, Laser Institute of America, Orlando, Florida, lattice Boltzmann method in a distributed and parallel environment.
pp 1016–1021 Procedia Comput Sci 18:2127–2136
67. Costa L, Deus AM, Reti T, Vilar R (2002) Simulation of layer 87. Ammer R, Markl M, Ljungblad U, Koerner C, Rόde U (2014)
overlap tempering in steel parts produced by laser cladding. Simulating fast electron beam melting with a parallel thermal free
Proceedings of the RPD 2002, Marinha Grande surface lattice Boltzmann method. Comput Math Appl 67(2):318–330
68. Bellini A, Güçeri S, Bertoldi M (2004) Liquefier dynamics in fused 88. Chryssolouris G, Kotselis J, Koutzampoikidis P, Zannis S, Mourtzis
deposition. J Manuf Sci Eng Trans ASME 126:237–246 D (1999) Dimensional accuracy modeling of stereolithography
69. Venkataraman N, Rangarajan S, Matthewson MJ, Harper B, Safari parts, 32nd CIRP International Seminar on Manufacturing
A, Danforth SC, Wu G, Langrana N, Guceri S, Yardimci A (2002) Systems, Leuven, pp 213–218
Feedstock material property-process relationships in fused deposi- 89. Wiedemann B, Dusel KH, Eschl J (1995) Influence of the polymer-
tion of ceramics (FDC). Rapid Prototyp J 6:244–252 ization dynamics of stereolithography resins on accuracy.
Int J Adv Manuf Technol
Proceeding of the 6th International Conference on RP University of 94. Kolossov S, Boillat E, Glardon R, Fischer P, Locher M (2004)
Dayton, Ohio 3D FE simulation for temperature evolution in the selective
90. NaraharaH, Tamaka F, Kishimani T, Igarashi S, Saito K (1997) laser sintering process. Int J Mach Tools Manuf 44(2–3):
Reaction heat effect on initial linear shrinkage of stereolithography 117–123
resins. Proceedings of Solid Freeform Fabrication Symposium, pp 95. Costa L, Vilar R, Reti T, Deus AM (2005) Rapid tooling by laser
733–740 powder deposition: process simulation using finite element analy-
91. Giannatsis J, Deboussis V, Laios L (2001) A study of the build-time sis. Acta Mater 53:3987–3999
estimation problem for stereolithography systems. Robot Comput 96. Stavropoulos P, Chantzis D, Doukas C, Papacharalampopoulos A,
Integr Manuf 17:295–304 Chryssolouris G (2013) Monitoring and control of manufacturing
92. Liu FR, Zhang Q, Zhou WP, Zhao JJ, Chen JM (2012) Micro scale processes: a review. Procedia CIRP, 14th CIRP Conference on
3D FEM simulation on thermal evolution within the porous struc- Modelling of Machining Operations, Turin
ture in selective laser sintering. J Mater Process Technol 212(10): 97. Yu C, Smurov I (2010) On-line temperature monitoring in selective
2058–2065 laser sintering/melting. Phys Procedia Part B 5:515–521
93. Khairallah SA, Anderson A (2014) Mesoscopic simulation model 98. Kwon KS, Choi YS, Lee DY, Kim JS, Kim DS (2012) Low-cost
of selective laser melting of stainless steel powder. J Mater Process and high speed monitoring system for a multi-nozzle piezo inkjet
Technol 214(11):2627–2636 head. Sensors Actuators A 180:154–165