Вы находитесь на странице: 1из 27

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/323609224

A review on quality control in additive manufacturing

Article  in  Rapid Prototyping Journal · March 2018


DOI: 10.1108/RPJ-03-2017-0048

CITATIONS READS

19 4,948

3 authors, including:

Hoejin Kim Bill Tseng


University of Texas at El Paso University of Texas at El Paso
44 PUBLICATIONS   231 CITATIONS    125 PUBLICATIONS   1,302 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Piezoelectric and Dielectric Nanocomposites Devices for Pressure/Temperature/Strain Sensing and Energy Storage Applications using Fused-Deposition Modeling 3D
Printing Technique View project

Photopolymer Resin based Piezoelectric Composites for Pressure Sensing Applications and Other Applications using Stereolithography 3D Printing View project

All content following this page was uploaded by Hoejin Kim on 27 April 2018.

The user has requested enhancement of the downloaded file.


Rapid Prototyping Journal
A review on quality control in additive manufacturing
Hoejin Kim, Yirong Lin, Tzu-Liang Bill Tseng,
Article information:
To cite this document:
Hoejin Kim, Yirong Lin, Tzu-Liang Bill Tseng, (2018) "A review on quality control in additive manufacturing", Rapid Prototyping
Journal, Vol. 24 Issue: 3, pp.645-669, https://doi.org/10.1108/RPJ-03-2017-0048
Permanent link to this document:
https://doi.org/10.1108/RPJ-03-2017-0048
Downloaded on: 27 April 2018, At: 09:51 (PT)
References: this document contains references to 144 other documents.
To copy this document: permissions@emeraldinsight.com
The fulltext of this document has been downloaded 56 times since 2018*
Downloaded by The University of Texas at El Paso At 09:51 27 April 2018 (PT)

Users who downloaded this article also downloaded:


(2018),"Quality control issues in 3D-printing manufacturing: a review", Rapid Prototyping Journal, Vol. 24 Iss 3 pp. 607-614 <a
href="https://doi.org/10.1108/RPJ-02-2017-0031">https://doi.org/10.1108/RPJ-02-2017-0031</a>
(2012),"Additive Manufacturing Technologies – Rapid Prototyping to Direct Digital Manufacturing", Assembly Automation, Vol.
32 Iss 2 pp. - <a href="https://doi.org/10.1108/aa.2012.03332baa.010">https://doi.org/10.1108/aa.2012.03332baa.010</a>

Access to this document was granted through an Emerald subscription provided by emerald-srm:459066 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald for Authors service
information about how to choose which publication to write for and submission guidelines are available for all. Please visit
www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company manages a portfolio of
more than 290 journals and over 2,350 books and book series volumes, as well as providing an extensive range of online
products and additional customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee on Publication Ethics
(COPE) and also works with Portico and the LOCKSS initiative for digital archive preservation.

*Related content and download information correct at time of download.


A review on quality control in
additive manufacturing
Hoejin Kim, Yirong Lin and Tzu-Liang Bill Tseng
Department of Mechanical Engineering, University of Texas at El Paso, El Paso, Texas, USA

Abstract
Purpose – The usage of additive manufacturing (AM) technology in industries has reached up to 50 per cent as prototype or end-product. However,
for AM products to be directly used as final products, AM product should be produced through advanced quality control process, which has a
capability to be able to prove and reach their desire repeatability, reproducibility, reliability and preciseness. Therefore, there is a need to review
quality-related research in terms of AM technology and guide AM industry in the future direction of AM development.
Design/methodology/approach – This paper overviews research progress regarding the QC in AM technology. The focus of the study is on
manufacturing quality issues and needs that are to be developed and optimized, and further suggests ideas and directions toward the quality
improvement for future AM technology. This paper is organized as follows. Section 2 starts by conducting a comprehensive review of the literature
studies on progress of quality control, issues and challenges regarding quality improvement in seven different AM techniques. Next, Section 3
provides classification of the research findings, and lastly, Section 4 discusses the challenges and future trends.
Downloaded by The University of Texas at El Paso At 09:51 27 April 2018 (PT)

Findings – This paper presents a review on quality control in seven different techniques in AM technology and provides detailed discussions in each
quality process stage. Most of the AM techniques have a trend using in-situ sensors and cameras to acquire process data for real-time monitoring
and quality analysis. Procedures such as extrusion-based processes (EBP) have further advanced in data analytics and predictive algorithms-based
research regarding mechanical properties and optimal printing parameters. Moreover, compared to others, the material jetting progresses technique
has advanced in a system integrated with closed-feedback loop, machine vision and image processing to minimize quality issues during printing
process.
Research limitations/implications – This paper is limited to reviewing of only seven techniques of AM technology, which includes photopolymer
vat processes, material jetting processes, binder jetting processes, extrusion-based processes, powder bed fusion processes, directed energy
deposition processes and sheet lamination processes. This paper would impact on the improvement of quality control in AM industries such as
industrial, automotive, medical, aerospace and military production.
Originality/value – Additive manufacturing technology, in terms of quality control has yet to be reviewed.
Keywords Additive manufacturing, Systematic review, Quality control, Data analytic and algorithm, In-situ correction,
Real time closed-feedback loop
Paper type General review

1. Introduction medical, aerospace and military, which require high precision


and reliability, lead 48 per cent in total. The global AM market
Additive manufacturing (AM) technology is a process of making sales reached up to $2.2bn in 2012 and is forecasted for growth of
a three-dimensional (3D) object-based on layer-by-layer or drop- the market size by $10.8bn in 2020 (Wong and Hernandez,
by-drop deposition of materials under a computer controlled 2012). Therefore, final and functional part productions and
system. It is composed of various types of printing methods such relevant researches are growing faster than the general market.
as binder jetting, material extrusion, material jetting and sheet Most of AM applications were purposed for prototyping and
lamination. AM industries have grown rapidly since 2000, and tooling; however, nowadays, industries invest 10 times more on
have shown almost six times the growth during the 2000s as end-part production than on prototyping (Cotteleer and Joyce,
compared with the growth during the 1990s (Wong and 2014).
Hernandez, 2012). AM technology has a high potential to exploit Likewise, this new paradigm of AM technology is being
ideas and applications on various areas. In 2013, Wohlers dramatically developed and brings high feasible applications
Associates reported that AM system sales revenue of each sector into real industry. However, before AM product is directly used
are industrial products (19 per cent), consumer products (18 for real-functional part fabrication, certified quality control
per cent), automotive (17 per cent), medical (14 per cent), (QC) is of vital importance to address specific requirements,
aerospace (12 per cent), military (5 per cent). Automotive, especially quality and reliability. For high-quality products that
are used in aerospace and automobile industries, AM is
The current issue and full text archive of this journal is available on required to certify high precision and mechanical properties to
Emerald Insight at: www.emeraldinsight.com/1355-2546.htm be used and functioned as real part (Wong and Hernandez,
2012). As a variety of materials are being used in the AM

Rapid Prototyping Journal


24/3 (2018) 645–669 Received 19 March 2017
© Emerald Publishing Limited [ISSN 1355-2546] Revised 9 June 2017
[DOI 10.1108/RPJ-03-2017-0048] Accepted 15 June 2017

645
Quality control in additive manufacturing Rapid Prototyping Journal
Hoejin Kim, Yirong Lin and Tzu-Liang Bill Tseng Volume 24 · Number 3 · 2018 · 645–669

technology, such as ceramic, steel, alloy, composites and living part and laser speed. Critical exposure (Ec) and penetration depth
tissue, compared with traditional manufacturing, the current (Dp) are two primary parameters of resin determined by
AM technology around part quality and performance needs to windowpane analysis. In addition, molecular weight of polymers
be constantly improved. Therefore, more the variety of (i.e. viscosity) plays an important role to determine final
advanced materials being used with AM technology, the more mechanical properties (Lee et al., 2007).
quality processes that will be necessary. Specific design tools used in the biomedical engineering for
For quality control improvement, there have been studies patient such as computed-tomography (CT) and magnetic
that focus on quality-approached researches. For example, resonance imaging (MRI) enable the SL technique to remove
Kaveh et al. (2015) studied the effect of printing parameters on elaborate design process and provide accurate custom-design
precision and internal cavity of fabricated parts using fused data. In addition, these design tools allow evaluating the
deposition modeling 3D printer (Kaveh et al., 2015). Another dimensional accuracy by comparing the scanned data with
research, Farzadi et al. (2015) investigated the effects of layer designed model. Cohen et al. (2009) used a commercial 3D
printing delay on physical and mechanical properties (Farzadi workstation (EBW 4.0, Philips medical) which has capabilities
et al., 2015). If these types of quality-related researches on AM of both MRI and CT, and can significantly enhance scanning
technology keep happening and produce improved quality time from 2 hours to 5 minutes for mandibular reconstruction
compared with the traditional manufacturing, AM products (Cohen et al., 2009).
will eventually substitute current technologies and expand to Cooke et al. (2003) incorporated the SL technique in
areas that require super-advanced technology in the near manufacturing biodegradable polymeric scaffold for use in
Downloaded by The University of Texas at El Paso At 09:51 27 April 2018 (PT)

future. ingrowth repairs of human bone defects. In biodegradable resin


This paper provides an overview of the research progress (mixtures of Diethyl fumarate and poly pylene fumarate), it was
regarding the QC in AM technology and the future direction of observed that approximately 10 per cent of supports did not
AM development in the QC aspect, and especially discusses attach to the build table, which caused problems of
quality issues and the needs to be developed and optimized; dimensional/geometric inaccuracy and rough surface finish
further, it suggests ideas and directions toward quality built parts. By customizing build tables and hole patterns that
improvement for future AM technology. This paper is organized control the resin flow rate, the initial layer of cross-linked
as follows. Section 2 starts by conducting a comprehensive review biodegradable resin mixtures can be securely adhered to the
of the literature studies on progress of quality control, issues and surface of the build table so that it can prevent the part from
challenges regarding quality improvement in seven different floating freely in the resin. It was found that the viscosity and
techniques of AM. Next, Section 3 provides classification of the density of vat solution influence the lack of adherence as well as
research findings, and lastly, Section 4 discusses challenges and design of build table and convex-surface effect (Cooke et al.,
future trends. 2003). Many efforts to fabricate complex tissue scaffold has
been attempted to improve part quality with different types of
2. Overview of quality control in additive solutions (Gauvin et al., 2012; Dhariwala et al., 2004). These
attempts are increasing usage area and better quality control as
manufacturing
variety of materials are being used (e.g. epoxy or acrylate or
2.1 Photopolymer vat processes additional additives (PPF/DEF–HA, PDLLA/HA) as based
Photopolymer vat processes (PVP) are based on the selective resins with photo-initiators, support material such mix of
solidification of a liquid photopolymer using light sources (e.g. propylene/polyethylene glycols, glycerin and/or acrylate).
ultraviolet; Vaezi et al., 2013). Stereolithography (SL or SLA) Developing materials to be used in the SL technique must take
is the main technique that is widely used especially in into consideration variety, composition, strength and finishing
biomedical engineering field such as implant and tissue procedures to increase the fabrication versatility and quality
engineering (Gauvin et al., 2012; Cohen et al., 2009). Melchels (Gross et al., 2014; Bose et al., 2013).
et al. (2010) reviewed two different SL techniques: scanning SL Kim et al. (2010) decreased contamination when fabricating
and projection SL. The scanning SL builds a structure from multi-materials using SL technique. When draining and
bottom with scanning laser, while the projection SL builds from cleaning the previous material before switching another vat, it
top to bottom with digital light projection (Melchels et al., was difficult to remove materials entirely owing to high viscosity
2010). The latter technique not only provides better surface of the previous resin. By using low viscosity polymer resin and
finish and dimensional accuracy but also saves materials and improving a process-planning algorithm, contamination was
fabrication time (Vaezi et al., 2013). reduced (Kim et al., 2010). Han et al. (2010) developed an
Some researchers (Huang and Jiang, 2003; Adolf et al., 1998) automatic material switching method by cleaning out the
have documented significant heat release owing to thermal current solution in the vat using solvent purging after UV-
reaction between laser source and resin solution during photo- exposure so that it was able to fabricate high-quality
polymerization, which causes residual stress development heterogeneous 3D hybrid scaffolds (Han et al., 2010). For high
associated with thermal shrinkage. Likewise, temperature profile quality fabrication process, Zhou et al. (2011) proposed to
has an important role on quality. Corcione et al. (2006) found remove the previous material sticking on part surface using a
that temperature increase on vat surface is dependent on energy soft brush for rough cleaning and then ultrasonic removal for
dose during SL process. Lower scan speed leads to higher final cleansing before introducing a different material (Zhou
temperatures and reaction rates, which involve shrinkage and et al., 2011).
curl distortion (Corcione et al., 2006). Lee et al. (2007) studied Orientation in which a design is fabricated can influence the
the effects of laser energy with respect to thickness of building SL performance. Lan et al. (1997) investigated that deposition

646
Quality control in additive manufacturing Rapid Prototyping Journal
Hoejin Kim, Yirong Lin and Tzu-Liang Bill Tseng Volume 24 · Number 3 · 2018 · 645–669

orientation mostly influences on three significant parameters: there are still many quality and process problems that occur
surface quality, build time and support structures. On the basis owing to vat contamination, wasting materials, long process
of the parameter consideration, decision criteria and algorithms time, material limitation and detachment of build table. Living
were established and developed to identify the desirable biological cells can be incorporated using this technique so that
fabrication orientations (Lan et al., 1997). Pham et al. (1999) human artificial organ or skin can be fabricated. In the near
developed feature-based system using an object-oriented future, more materials will be incorporated which will, in effect,
programming language and solid modeling CAD environment demand more quality related studies on the variety of materials
to obtain the best tradeoff by considering factors based on build needed.
time, cost and accuracy while having optimal part orientation
(Pham et al., 1999). 2.2 Material jetting processes
The SL technique has also been used in micro/nano scale Material jetting processes (MJP) are based on the use of inkjet
fabrication (Maruo, 2015). Ha and Yang (2014) studied that printing technique to deposit droplets selectively through a
surface friction factor taken into consideration when fabricating nozzle or orifice to build up a 3D structure. This droplet is a
closed structures does not affect open complex structure. liquid that is jetted out of the nozzle and becomes solid after
Therefore, a micro-scaled open structure was erected securely deposit through cooling or UV curing. There are two methods
on the substrate from the mentioned process, as shown in
for creating appropriate material droplets: drop on demand
Figure 1. This nanostructure fabrication is capable of having
(DoD) and continuous inkjet (CIJ). The latter uses fluids of
high quality structures with higher efficiency than closed
low viscosity and high drop velocity for fast printing speed; in
Downloaded by The University of Texas at El Paso At 09:51 27 April 2018 (PT)

structures. In this study, 10  10  10 m m open structures


contrast, the former uses smaller droplets and velocity for high
were fabricated, CCD camera were used to monitor the
dimensional accuracy; therefore, DoD has a higher potential
nanoscale structure during fabrication (Ha and Yang, 2014).
for microfabrication (Vaezi et al., 2013). In the DoD system,
West et al. (2001) developed a new adaptive slicing algorithm
in the design process for improving build performance. This two main kinds of pressure pulse methods in the manner of
algorithm assists SL operators in developing a process plan by ejecting droplet from ink vat are divided: thermal and
quantifying tradeoffs among surface finish, geometry tolerances piezoelectric actuators. Thermal DoD has a limitation on ink
and build time. By adjusting process variables within algorithm, material used, which is volatile, while the piezoelectric DoD has
it can predict geometric and dimensional properties, and can no such limitations, but is required to have droplet generation
compare the desired and actual values so that SL operators can rates in the range of 3-10 kHz to allow the meniscus to reach
approach the specific design requirements (West et al., 2001). capillary equilibrium (Sachs and Vezzetti, 2005).
In summary, PVP has advanced in providing accurate data Vaezi et al. (2013) reviewed the several significant
acquisition during design applications that use CT and MRI, phenomena affecting the quality of the material jetting
which makes this technique particularly useful in biomedical processes. The shape of deposited droplet influences critically
applications. In addition, these procedures allow part on resolution, precision and accuracy (Vaezi et al., 2013).
evaluation by comparing design and 3D printed model. Beaman et al. (2004) demonstrated through fluid mechanics
Finding optimal build orientation to improve the quality and theory that there is an important relationship between the
build time was mostly studied by incorporating a new type of Reynolds number Re = r gvd/ m , denoted as a ratio between
solid modeling and slicing algorithm. This indicates that build inertial and viscous forces, and Weber number We = r 1v2d/s ,
orientation mostly affects the quality. For printing process, denoted as a ratio between kinetic and surface energy, where r g

Figure 1 Schematic diagram of nano-stereolithography apparatus (left) and SEM images of fabricated open structure (right)

647
Quality control in additive manufacturing Rapid Prototyping Journal
Hoejin Kim, Yirong Lin and Tzu-Liang Bill Tseng Volume 24 · Number 3 · 2018 · 645–669

and r 1 represent the densities of the process gas and liquid not only are accurate and optimal printing temperatures and
drop, respectively. The variables v, d, m and s are droplet voltages essential factors in nanofluids’ control for micro scaled
velocity, droplet diameter, liquid dynamic viscosity and liquid structures, but also that thermal optimization such as substrate
surface tension, respectively. It was found that Re and We temperature and laser annealing are to be taken into
p
should satisfy 1 < Re= We < 10 to maintain the overall consideration (Kullmann et al., 2012).
printing quality of DoD system (Beaman et al., 2004). This Inkjet printing technique is used widely in biological tissue
author reported that droplets should not exceed more than 10 engineering and electric circuit design due mainly to its multi-
m m in diameter owing to problems with air resistance. Ko et al. material capability (Nakamura and Henmi, 2008; Mei et al.,
(2010) reported that fundamentally considerable and 2005; Xu et al., 2006; Cui and Boland, 2009; Czy_zewski et al.,
interconnected conditions for high desired quality control on 2009). However, Rengier et al. (2010) rated MJP technique on
MJP are the following: medical applications as inferior in accuracy and strength in
 ink properties (viscosity and surface tension); comparison with other techniques (SLA, SLS and FDM;
 jetting parameters (signal width, voltage magnitude and Rengier et al., 2010). Ebert et al. (2009) found the process-
jetting frequency); and related defects as a result of single clogged nozzles owing to
 environment (pressure, environment and substrate either dried up or blocked by agglomerates while printing
temperature and humidity). dental prostheses made of zirconia-based ceramics (Ebert et al.,
Research on metal nanoparticle inkjet printing investigated 2009). To resolve this issue, sintering was adopted as post
processing to enhance the strength and accuracy, and
Downloaded by The University of Texas at El Paso At 09:51 27 April 2018 (PT)

thoroughly that proper substrate temperature and sufficient


drying of nanoparticle inks in each layer guarantee high quality preheating substrate was implemented to avoid internal stress
(Ko et al., 2010). and bending of layers during drying. In addition, aqueous
Jiang et al. (2010) studied on the accurate micro scale- solution of 10 wt. per cent ethanol was found to be sufficient for
droplet creations. As shown in Figure 2, droplets merge with nozzle cleaning to prevent clogging.
each other to form bigger droplets owing to different droplet Wüst et al. (2011) mentioned that protein and biological cells
velocities. These merged droplets degrade the droplet’s are more likely to be denaturant depending on temperature;
diameter accuracy, which significantly affects surface therefore, temperature of ink bubbles in thermal-based printing
roughness and total geometry. It was observed that uniform should not exceed more than 41°C (Wüst et al., 2011). Owing
spherical droplets can be developed at optimal frequency of to this reason, Wilson and Boland (2003) modified commercial
2.842 kHz and at low oxygen concentration environment (Jiang inkjet printers by implementing temperature controller to avoid
et al., 2010). Kullmann et al. (2012) contributed on building denaturation of proteins and cells (Wilson and Boland, 2003).
3D microstructures of gold nanofluids with high aspect ratio Henmi et al. (2007) developed a gelation technique to align
and low-cost basis using DoD technique. It was discussed that printed cell on the substrate without drying and distributed
randomly so that living cells and biomaterials can be easily
Figure 2 Water droplet stream using orifice diameter of 150 m m patterned to create sophisticated 3D printed tissue (Henmi
at a gas pressure of 8 kPa et al., 2007). Cui et al. (2010) noted the frequency of
piezoelectric inkjet printers to be within a frequency range that
causes cell damage and lysis after sonication at 15-25 kHz (Cui
and Boland, 2009). For a thermal inkjet printer, its heating
temperature at the ejected nozzle increases up to 300°C for a
couple of microseconds during the printing. Therefore, there
are significant concerns in printing processes that cause cell
damage and death (Cui et al., 2010). In 2012, authors
comprehensively evaluated cell viability and damage on
thermal inkjet cells and found 4-10°C above ambient for only
2 m s avoids cell damage up to an average cell viability of 90 per
cent. Therefore, thermal inkjet printing technology is becoming
more biocompatible to the living system compared to
piezoelectric printing (Cui et al., 2012).
Derby (2010) and Yeong et al. (2006) discussed the
importance of fluid properties for quality and investigated the
main factors of quality in the MJP technique to be:
 generation of droplet;
 position and interaction of droplet on a substrate; and
 solidification mechanism.
In terms of interaction between droplets and substrates, part
quality is significantly determined by properties such as
gravitational forces, fluid density, surface energy, drop
velocity, droplet size, surface tension, capillarity forces,
contact angle, drop spacing and print head traverse speed.
In addition, during droplet transition from liquid to solid,

648
Quality control in additive manufacturing Rapid Prototyping Journal
Hoejin Kim, Yirong Lin and Tzu-Liang Bill Tseng Volume 24 · Number 3 · 2018 · 645–669

which may occur, volume change is important and can be issues aim to control the droplet mechanism, which
controlled by these considerations such as solvent significantly affects resolution and dimensional accuracy. The
distribution, fluid flow rate and evaporation rate. Proposed main issue during printing process is nozzle clogging occurring
nano/micro-scale patterning and structuring of substrate owing to dry and agglomerates, which are detrimental to
can overcome this resolution issue limited by drop size and medical application. Using thermal inkjet printing technique,
can achieve higher resolution more than drop size (Derby, sophisticated tissue structures such as biomaterials and living
2010; Yeong et al., 2006). cells can be easily fabricated without cell damage caused by
Tourloukis et al. (2015) identified factors that affect quality piezoelectric inkjet frequency and high temperature. Positive
issues and printing performance and studied the potential of early stage research was made on data-driven approach using
computational intelligence algorithms particularly in neural neural networks to predict the quality. A great achievement of
networks to assess and forecast quality of 3D inkjet-printed MJP is a real-time closed-feedback loop system incorporated
electronic products. Results validated that one-step ahead with machine vision. This can correct the design and printing
predicting model using nonlinear autoregressive neural parameters in real-time and eventually open up real-time
network with external input brought up better overall processing during printing process.
prediction behavior compared with multi-step ahead one.
However, more investigation is needed with larger data set to 2.3 Binder jetting processes
be trained with respect to a wider range of values and trends for Binder jetting processes (BJP) have similar principals to
enhancing predictive performance (Tourloukis et al., 2015). material inkjet process, but based on nozzles depositing
Downloaded by The University of Texas at El Paso At 09:51 27 April 2018 (PT)

Recently, Sitthi-Amorn et al. (2015) developed a material droplets of a water-based binder material over surface of a
jetting systems allowing for self-calibration of print heads, 3D powder bed (Wong and Hernandez, 2012; Stucker, 2012). The
scanning and a closed-feedback loop to enable print powder particles are held together by bonding with deposited
corrections. Real-time printing control technique of machine binder material and eventually shape 3D structures. This
vision built in enables to correct overall platform design and process is followed by lowering the powder bed through a
printing parameter in real-time through closed-feedback loop. piston and a fresh layer of powder spread over the previous
The image processing algorithm computes depth difference layer and deposited by binder again over the new layer. The
between droplets and adjusts amount of jet droplet. This possible materials used as powder includes ceramics, metals,
technique opens up the real-time quality control during shape-memory alloys and polymers (Vaezi et al., 2013). Gibson
printing process. In addition, 3D scanning (i.e. optical et al. (2014) found that BJP technique provides better quality
coherence tomography, OCT and scanner) provides high control ability for ceramic and metal fabrications to be
resolution (less than 1 m m) to scan the depth map and overlay produced compared with MJP technique. However, a BJP
the model to be printed over the 3D scanned auxiliary printed part tends to have lower accuracy and surface finish
component on the build platform. Furthermore, users can than the part made by MJP. In addition, infiltration processes
interactively translate and rotate the model over the image to are typically needed to fabricate dense parts or to ensure good
facilitate the alignment (Sitthi-Amorn et al., 2015). (Figure 3) mechanical properties (Gibson et al., 2014).
In summary, MJP have significantly advanced in micro-scale Deposition ways of the binder are the same as the MJP
fabrication. Most of the studies implemented for quality related technique. The CIJ system of BJP technique is designed to

Figure 3 (A) Optical coherence tomography 3D scanner: the optical components are arranged in a Michelson configuration. The light source is a
collimated, red LED. Polarizers are placed in the LED and camera paths to control the light density. The points in the sample that are at the same distance
as the reference mirror cause constructive interference patterns. The platform is moved along the Z axis while the interference patterns are detected by
the camera. The scanner features a circular scanning area with a diameter of 15 mm; (B) the machine vision feedback loop proceeds in the stages and
image processing pipeline for 3D scanner: We compute the 3  3 standard deviation. Next, we choose the depth in which the standard deviation is the
highest. We also keep the maximum standard deviation as the confidence map. We perform the depth correction on the depth map from previous stage.
Finally, we stitch the depth maps from all the location weighted by the confidence. We assume that the area where the confidence is below certain
threshold is a hole, and fill those holes at the end

649
Quality control in additive manufacturing Rapid Prototyping Journal
Hoejin Kim, Yirong Lin and Tzu-Liang Bill Tseng Volume 24 · Number 3 · 2018 · 645–669

print in either uni or bidirectional mode. Bidirectional printing particles during sintering process (Chumnanklang et al., 2007).
mode is faster than unidirectional one, but causes dynamic shift For better powder distribution, Snelling et al. (2013) used a
errors. A part requiring higher quality surface finish should be sieve test to determine particle size distribution, which is of
printed in unidirectional mode. Critical point on quality in this critical importance to mechanical properties and dimensional
system is a catcher, which helps to achieve a selective printing accuracy (Snelling et al., 2013).
stratagem in the presence of a continuous stream as shown in Lu and Reynolds (2008) analyzed printing variables mainly
Figure 4(A). Figure 4(B) display how the catcher is operating. influenced on integrity, dimensional accuracy and minimal
Deflected plates are guiding droplets to catchers when feature sizes that can be printed for less than 500 m m 3D mesh
unwanted droplet and drained for reuse. However, owing to structures (TiNiHf shape memory alloy). This author classified
some disadvantages such as binder accumulation, pump three groups in terms of the variables: 3D printer-, powder- and
clogging and low cross-section of individual catchers, Sachs binder-related variables. Two major 3D printing variables (i.e.
and Vezzetti (2005) proposed a single catcher system that is printing layer thickness and binder saturation) are evaluated. It
shared by all the nozzles, proportional deflection reducing the was found that breaking strength increases with the binder
nozzle gap and 45 degree print head rotation improving saturation level up to 170 per cent at the same printing layer
dynamic shift and enabling the droplets to land on exact thickness, and 35 m m printing layer yields 3D mesh structure
powder bed at certain time to improve surface finish and with the highest integrity and the lowest dimensional deviation
dimensional accuracy, as shown in Figure 4(C) (Sachs and (Lu and Reynolds, 2008).
Vezzetti, 2005). By comparison, the DoD method provides In bone tissue engineering area, Farzadi et al. (2014)
Downloaded by The University of Texas at El Paso At 09:51 27 April 2018 (PT)

better quality (e.g. resolution and minimum feature width) but investigated the effects of layer thickness and printing
is slower in regard to printing process (Cima et al., 2001). orientation (parallel to X, Y and Z) on dimensional accuracy
The BJP is capable of achieving accurate geometries in nano- and mechanical properties of printed porous scaffolds. It was
biomaterial fields (Liu et al., 2013). Factors that determine the found that concurrence between model’s orientation (i.e.
final part dimensions are accumulative accuracy of deposited longitudinal direction) and printing head movement results in
layer thickness, accuracy of droplet placement, reproducibility better dimensional accuracy and mechanical properties, and
of droplet spread and of dimensional change during curing layer thickness as well as printing direction, which has a
process. In addition, the dimensional accuracy is strongly significant effect on the compressive strength of scaffolds
influenced by interaction performance between powders and (Farzadi et al., 2014).
binders such as powder’s surface treatment, size, shape, Some of researchers have used mathematical analyses to
packing density, distribution, binder’s viscosity, surface improve the quality for BJP technique. Asadi-Eydivand et al.
tension, droplet size, velocity and temperature. (2016) found that the delay time of applying the new layer
Chumnanklang et al. (2007) studied the effects of binder and printing orientation significantly influences the
concentration in pre-coated particle on part strength. It was dimensional accuracy, compressive strength and porosity. A
found that part strength can be increased by increasing the full factorial design of experiment (DOE), signal to noise (S/
binder concentration and pre-coated particle size. The N) ratio and analysis of variance (ANOVA) were used to
hydroxyapatite particle used was coated with maltodextrin and investigate optimal process parameters (Asadi-Eydivand et
water. Coated powders have easier-to-flow characteristic than al., 2016). Another research used statistical analysis
raw powders so that it could lead to high flow-ability and (ANOVA) to determine significant factors affecting product
uniform distribution. These higher binder concentration and quality. Parameters of binder saturation, layer thickness,
pre-coated particle enables better densification of intra- building orientation were optimized through statistical

Figure 4 (A) Components of the continuous printing system; (B) flight path of a droplet during printing; (C) binary deflection and proportional one

650
Quality control in additive manufacturing Rapid Prototyping Journal
Hoejin Kim, Yirong Lin and Tzu-Liang Bill Tseng Volume 24 · Number 3 · 2018 · 645–669

analysis (Uma Maheshwaraa et al., 2008). In addition, BJP Figure 5 Microphotograph of the cross-sectional area of a 38  38 
technique has been studied for quality in metal casting area 30 layer part
(Meisel et al., 2012; Snelling et al., 2013), foundry
engineering (Budzik, 2007) and pharmaceutics (Yu et al.,
2009).
In summary, BJP have highly advanced in ceramic and metal
fabrications compared with other techniques because BJP can
fabricate high melting point materials at low temperature using
binder and post-treatment. The impressive achievement in BJP
was a study on DoD to develop a single catcher system on each
nozzle to control individual droplet and pre-coating particle to
enhance part strength by solving drawback of BJP such as low
density and porosity. In addition, mathematical analyses were
conducted to find optimal printing parameters. The main
issues during printing process were the interaction performance
between binder and powder, accumulative accuracy of
deposited layer thickness, droplet placement, the delay time of Boschetto and Bottini (2016) used a design for manufacturing
applying a new layer and dimensional change during post- (DFM) method to overcome inconsistent geometrical
Downloaded by The University of Texas at El Paso At 09:51 27 April 2018 (PT)

treatment. These issues significantly influence dimensional deviations that cause poor surface quality and depend on layer
accuracy and challenges to investigate for better quality. thickness, deposition angle, etc. Proposed DFM permits a
redesign of the components during preprocessing to predict the
2.4 Extrusion-based processes obtainable dimensional deviations. Modifications carried out in
Extrusion-based processes (EBP) deposit material in form of a design stage can compensate for the deviations generated in the
continuous flow layer-by-layer to build a 3D structure. These next manufacturing stage, thus, providing improved part
processes have diverse technique concepts but are classified accuracy, as shown in Figure 6 (Boschetto and Bottini, 2016).
into two main groups: melting-based extrusion and non- Reducing in build time and increasing part surface quality
melting-based extrusion (Vaezi et al., 2013). 3D-bioplotting is contradict each other and there have been a number of
the most representative of non-melting-based extrusion attempts to solve these issues. Pandey et al. (2003a, 2003b)
techniques and fused deposition modeling in melting based proposed a method of real-time adaptive slicing procedures.
extrusion techniques. In here, fused deposition modeling This method predicts surface quality expressed by standard Ra
(FDM) will be the main point of discussion. value using direct slicing and tessellated model (STL) so that in
2.4.1 Researches on general quality control the design stage, build time and surface quality can be
In the FDM technique, there are two kinds of deposition methods maximized based on the predicted Ra value (Pandey et al.,
mainly used: contour and raster. In contour method, extrusion 2003b). Huang et al. (2009) proposed a robust algorithm to
path is following outside contour and offset in until it fills the entire generate the support slice data that enable part’s self-support
domain. This technique’s, main disadvantage is slow printing ability and reduce redundant support volume at maximum
operation owing to multiple paths creation. In a raster method, the extent. This slice data based algorithm has the same efficiency
material is deposited in zig–zag fashion. Raster method is a faster as the STL based algorithm, however, it is more stable and
process but less accurate and gives rise to voids in deposition robust to generate support slicing process. The advantage of
process in comparison with the contour one. This can lead to this algorithm provides necessary support for hanging not only
deterioration in the in-plane properties of the structure. Kulkarni structure surface but also vertexes and edges, which can
and Dutta (1999) proposed spiral geometric path, which can guarantee dimensional accuracy (Huang et al., 2009).
eliminate multiple paths and inaccuracy to provide faster process In the area of casting, Kaveh et al. (2015) used high impact
and reduce voids. In result, in-plane stiffness spiral method was polystyrene (HIPS) materials to fabricate wax patterns for
improved up to 18,000 lb/in2 compared with 10,300 lb/in2 of casting technology that requires accuracy without nay internal
contour paths (Kulkarni and Dutta, 1999). cavity. They designed proper benchmarks regarding printing
Sun et al. (2008) evaluated the effects of processing parameters and statistical equations that can calculate
conditions in the bonding quality of FDM technique. It was calibration factors (precision and internal cavity) to prevent
found that the neck growth between adjacent filaments of the errors between designed and actual dimensions; eventually,
same layer is expected to be larger in the bottom layers than in these can be used to determine optimal printing parameters
the top layers because bottom layer remains longer period than (Kaveh et al., 2015).
upper layers, indicating voids size in the lower region is clearly
smaller than in the upper region. It is the fact that convective 2.4.2 Researches on surface quality and post-processing
heat from adjacent extruded molten layers during process FDM technique presents limitations in surface quality.
allows neck growth between filaments, which occurs above Staircase effect influences surface quality of 3D printed part
glass transition temperature. This neck growth can be increased using thick filament approximately 0.24 mm, which is the most
by proposed lateral geometric path to have slow cooling rate. used thickness and 0.127 mm at the most. This is lower
Therefore, bonding strength and voids were improved and resolution than other techniques such as SL (0.05 mm), SLS
minimized, respectively (Sun et al., 2008). (Figure 5) (0.02 mm), BJP (0.05 mm) and MJP (0.016 mm). Other

651
Quality control in additive manufacturing Rapid Prototyping Journal
Hoejin Kim, Yirong Lin and Tzu-Liang Bill Tseng Volume 24 · Number 3 · 2018 · 645–669

Figure 6 (A) Modeled (left) and fabricated (right) air fan blade; (B) 3D maps of the thickness deviation measurements for original (left) and modified
(right) blade
Downloaded by The University of Texas at El Paso At 09:51 27 April 2018 (PT)

problems are related to thermal distortion and stress resulting Galantucci et al. (2009) proved significant improvements on
from shrinkage (Boschetto and Bottini, 2015). surface finish of FDM 3D printed acrylonitrile butadiene styrene
Armillotta (2006) assessed textured surface quality printed (ABS) part using the chemical post-processing method.
by FDM technique. Critical issues on surface quality are Dimethyl ketone (acetone) was used with 10 per cent of water as
dependent on the rate of material deposition, shrinkage and chemical because ABS has weak interaction with polar solvents
residual stresses during solidification. It was found that such as dimethyl ketone, ester and chloride solvents. Therefore,
sufficient resolution of surface without stair-stepping, as shown chemical reaction is controlled by the soaking and exposure time
in Figure 7(A), can be obtained under the conditions provided to ABS parts (Galantucci et al., 2009). In biomedical micro-
which feature in-plane size  1 mm on surfaces parallel to the devices, McCullough and Yadavalli (2013) also used the same
build direction can be fabricated without excessive stair- chemical as dimethyl ketone-based sealing method to modify and
stepping. Moreover, sinking effects made by contour offsetting, seal the surfaces of ABS to render the surface of 3D printed
as shown in Figure 7(B), can be avoided when adjacent features micro-structured features to be impervious to water and increase
are spaced out at least 1 mm away from one another hydrophilicity and biocompatibility, as shown in Figure 8
(Armillotta, 2006). Campbell et al. (2002) implemented a (McCullough and Yadavalli, 2013). With this chemical method,
visualization algorithm (AutoLisp language within AutoCAD Galantucci et al. (2010) studied how this chemical reaction
14) that represents varying surface roughness of 3D designed affects mechanical properties as well as surface finish. It was
model as color shading within a CAD image. Using this observed that mechanical properties such as ductility, flexural
algorithm, the designer is given clear visual image of the overall and compressive strength were improved as treated chemicals but
surface roughness of a model and any potential problem areas tensile strength was slightly reduced because of a different action
(Campbell et al., 2002). of the solution on different surface pattern or different reaction of

Figure 7 (A) Graphical simulation of the stair-stepping effect on textured surfaces and (B) resolution limits (right): (a) spacing of layer contours; (b),
(c) feature sinking

652
Quality control in additive manufacturing Rapid Prototyping Journal
Hoejin Kim, Yirong Lin and Tzu-Liang Bill Tseng Volume 24 · Number 3 · 2018 · 645–669

Figure 8 (A) Permeability of micro-channels fabricated using FDM before and after treatment; (B and C) FDM ABS feature fidelity before and after
treatment; (B) channel and (C) ridge shape fidelity; (D), percentage change in ridge radius of curvature (E) and channel wall incline angle with varying
acetone solution concentration, treatment length constant at 4h
Downloaded by The University of Texas at El Paso At 09:51 27 April 2018 (PT)

treated filaments (Galantucci et al., 2010). In addition, Percoco through statistical analysis validated in comparison with
et al. (2012) studied compressive strength and surface roughness experimental results (Boschetto and Bottini, 2015).
of FDM 3D printed ABS part using chemicals, 90 per cent Many researchers have found that build orientation affects
dimethyl ketone solution and 10 per cent water. Results showed surface quality on FDM parts. Vijay et al. (2011) investigated
an increase in compressive strength and reducing roughness up to that surface roughness is proportional to the layer thickness for
90 per cent when the immersion time is at 300 s, as compared 20° and 45° build orientation; however, roughness is an inverse
with non-treated part (Percoco et al., 2012). proportion to layer thickness when 70° is used for building
Addanki Sambasiva Rao et al. (2012) used design of orientation (Vijay et al., 2011). Sreeram (1995) developed an
experiment, ANOVA, to find significant factors affecting auto-compute method to decide optimal orientation direction
surface finish by analyzing various parameters in chemical based on variable slicing thickness for a polyhedral part
treatment process, such as different levels of solution (Sreeram PN, 1995). Frank and Fadel (1995) proposed a user-
concentration, exposure time, temperature and initial friendly decision-making tool that recommends the optimal
roughness. Two different chemicals, dimethyl ketone and orientation direction before printing (Frank and Fadel, 1995).
methyl ethyl ketone, were used. In the case of dimethyl ketone, Chakraborty et al. (2008) developed a new method for
solution concentration, temperature and initial roughness were curved layer 3D structure. This curved layer FDM (CLFDM)
significant factors that mostly affect surface roughness. For method has an advantage in fabricating thin and curved part
Methyl ethyl ketone, solution concentration, temperature and with improved surface finish owing to reduction of stair–step
exposure time were the most important factors to be surface effect as well as the reduction of layers. The method also
quality (Addanki Sambasiva Rao et al., 2012). Moreover, increased mechanical strength, reduced material waste and
vapors treatment was introduced by Sratasys. Vapors of build time, as shown in Figure 10(A). In regular FDM, table
tetrahydrofuran created by applied heat interact with surface of and head movement do not deviate from vertical (z-axis) when
3D part, which settles in the side of the closed vessel with a non- x- and y-axes are being controlled, as shown in Figure 10(B).
air tight lid (Stratasys, 2016). However, this chemical However, the proposed method uses x- and y-axes with
treatment can dramatically improve surface finish but has been contouring control and z-axis with linear interpolation control.
only applied to ABS polymer and drawback limits using to Therefore, the table and head movement can have a 3D linear
chemical sensitive materials. In addition, it provides poor interpolator for the deposition of curved layers. In addition,
repeatability, small variations in geometry of the part and they worked on determining adjacent filament paths for proper
unpredictable dimensional accuracy. reproduction of part shape (Chakraborty et al., 2008).
Another method used was hot cutter machining, as shown in Ahn et al. (2009) proposed a theoretical model to express
Figure 9(A). Pandey et al. (2003a, 2003b) addressed a problem surface roughness with respect to surface angle u . As
of surplus stock material, necessary tooling phase and path the surface angle is closer proportionally to fabrication
planning despite improved surface at low time (Pandey et al., direction, surface roughness improves quality. The validity of
2003b). A method mostly used for surface finish is a barrel this proposed model was further proved by comparing
finishing reproducing better surface finish using abrasive or empirical data with computed values (Ahn et al., 2009).
lubricative polishing agents with 3D printed parts put in
rotating closed chamber creating friction against each other, as 2.4.3 Data analytic and algorithm researches
shown in Figures 9(B) and 9(C). This method provides high Many researchers used statistical analyses to optimize both
repeatability and surface finish. Boschetto and Bottini (2015) processes and product design for quality control. Anitha et al.
studied the quality control for predicting surface roughness (2001) performed experiments to assess the effects of the
after barrel finishing by using theoretical models developed parameters on quality characteristics using the Taguchi

653
Quality control in additive manufacturing Rapid Prototyping Journal
Hoejin Kim, Yirong Lin and Tzu-Liang Bill Tseng Volume 24 · Number 3 · 2018 · 645–669

Figure 9 (A) Different directions of machining using hot cutter; (B) scheme of barrel finishing; (C) 3D map of surface before machining and after 960
min barrel finishing working time

Figure 10 (A) Model development of different complex geometries using CLFDM – (a) a hemispherical surface with a straight trim on one side
(b) trimmed surface of a part in the interior, discontinuous filament deposition method (c) trimmed surface of a part in the interior, continuous filament
deposition method (d) surface made of two patches, continuous filament method; (B) (a) sectional view of a part produced by FDM. The overlapped
Downloaded by The University of Texas at El Paso At 09:51 27 April 2018 (PT)

section of two adjacent layers in FDM is shown by shaded lines. (b) Adjacent filaments of CLFDM

method. Analysis of the Taguchi results using methods of signal a staircase structure on 3D printed parts and used ANOVA and
to noise (S/N) ratio, ANOVA, correlation and regression fractional factorial design of experiment to analyze the effects of
analyses indicated that without pooling only one layer thickness three important machining parameters (i.e. cutting speed and
is effective to 49.37 at 95 per cent level of significance. But on direction) and for developing statistical model, respectively
pooling, the layer thickness is mostly effective to 51.57 at 99 per (Pandey et al., 2003a). Sood et al. (2009) adopted the Grey–
cent level of significance. This result was further strengthened Taguchi method to obtain optimum level of printing parameters
by afore mentioned analyses which indicate a strong by finding the significant factors and their interactions and to
relationships with surface roughness. Moreover, based on the minimize the percentage change in length, width and thickness
S/N analysis, the layer thickness was mostly effective at 0.3556 into a single objective known as the Grey relation grade. This
mm, road width at 0.537 mm and deposition speed at 200 mm/s Grey–Taguchi method provided good results on overall
(Anitha et al., 2001). Similarly, Lee et al. (2005) used the improvement in part dimension. In addition, artificial neural
Taguchi method and orthogonal array, main effect, S/N and networks (ANN) was used to predict overall dimensional
ANOVA analyses to investigate the optimal printing parameters accuracy and provide errors between predicted data and
needed to achieve maximum flexibility of ABS material. Through observed value varying between 0 and 3.5 per cent. Smaller error
this statistical analysis, it was concluded that layer thickness, percentages mean better suitability of present model (Sood et al.,
raster angle and air gap has significant impact on the elastic 2009).
performance of ABS printed structure (Lee et al., 2005). Pandey Sood et al. (2010) conducted central composite design for
et al. (2003a, 2003b) studied the improvement of surface finish design of experiment and ANOVA for validity to investigate the
by staircase machining using the FDM technique. They important impact of printing parameters related to mechanical
proposed hot cutter machining method to improve the surface of properties such as tensile, flexural and impact strength. It was

654
Quality control in additive manufacturing Rapid Prototyping Journal
Hoejin Kim, Yirong Lin and Tzu-Liang Bill Tseng Volume 24 · Number 3 · 2018 · 645–669

concluded that the increase in layer thickness, which means such as raster orientation, air gap, bead width, color and model
decreased number of layers required, leads to minimizing temperatures. It was found that the air gap and raster
distortion effects within the structure so that total mechanical orientation significantly influence the tensile strength of printed
strength will be increasing (Sood et al., 2010). Sood et al. part a lot more than bead width, model temperature and color
(2012) studied that compressive strength of 3D structure (Ahn et al., 2002). Thrimurthulu et al. (2004) used real coded
exhibits high dependence on five important printing genetic algorithm to obtain optimum parameters of part
parameters (i.e. layer thickness, part orientation, raster angle/ deposition orientation, which significantly affects the surface
width and air gap). This research not only provides an insight roughness and build time. Also, this optimized parameter
into the complex dependency of compressive stress on process considers minimization of support structures estimated by
parameters but also created a statistically validated predictive genetic algorithm. This genetic algorithm helped predicting
equation to find optimum parameters for maximum surface quality and build time (Thrimurthulu et al., 2004).
compressive stress using a quantum-behaved particle swarm Patel et al. (2012) emphasized that the Taguchi method is a
optimization (QPSO). However, to solve involved large versatile tool for process and product design optimizations and
number of conflicting factors and complex phenomena for part to solving current quality control issues on 3D printing
building to predict output characteristics accurately. ANN with technology (Patel et al., 2012), based on the references (Sood
back propagation algorithm was adapted to predict the final et al., 2009, 2010; Shaji and Radhakrishnan, 2003; Çaydas  and
structure’s compressive stress as shown in Figure 11. In result, Hasçalik, 2008; Gopal and Chakradhar, 2012).
this optimized statistically predictive equations that gave a
Downloaded by The University of Texas at El Paso At 09:51 27 April 2018 (PT)

2.4.4 Researches on composites


maximum compressive stress of 17.4751 MPa at optimum
Mostafa et al. (2009) investigated the behavior of melt flow of
values of layer thickness, orientation, raster angle, raster width
ABS–iron based composite materials deposited by FDM
and air gap as 0.254 mm, 0.036°, 59.44°, 0.422 mm and
technique. 2D and 3D numerical analysis (i.e. CFD, FEA and
0.00026 mm, respectively (Sood et al., 2012). Li et al. (2010)
ANSYS FLOTRAN/CFX software program) were used to
also used particle swarm optimization algorithm to obtain
analyze melt flow parameters: mainly temperature, velocity and
appropriate fabrication orientation by optimizing support area,
pressure drop at extrusion. These analyses help prevent
fabrication time and surface roughness so that it was proved to
clogging and allow smooth flow out of the extrusion and
enhance processing efficiency and reduce the costs significantly
enhance quality of composite based 3D structure by predicting
(Li et al., 2010). Rao and Rai (2016) used the teaching–
the flow behavior (Mostafa et al., 2009). Bellini and Güçeri
learning-based optimization algorithm (TLBO) and non-
(2003) developed a novel FDM technique mounted a mini-
dominated sorting TLBO (NSTLBO) to solve optimization
extruder on deposition of FDM machine for agile fabrication. It
problems on FDM. When it comes to objective function value
streamlined filament preparation for continuous process of
with a higher convergence rate, it was proved that TLBO shows
fabrication ECG9/PZT ceramic composites. This high-
better performance as compared to QPSO and generic
precision-based extruder was designed for a wide range of
algorithm (GA) in terms of solving single-objective
materials as well as ceramic based composites. The author
optimization aspect and that NSTLBO demonstrated better
studied flow behaviors influencing quality in terms of granules
performance as compared with non-dominated sorting genetic
size, nozzle temperature and nozzle size during extrusion and
algorithm (NSGA-II) in terms of solving multi-objective
deposition. Two thermocouples were used at both the entrance
optimization aspect (Rao and Rai, 2016).
and exit of the extruder qualifier to determine the influence of
Ahn et al. (2002) used design of experiments to study the
temperature in upper and lower parts of the qualifier (Bellini
relationship among the printing parameters of ABS material
and Güçeri, 2003).
In summary, EBP has been studied in various research
Figure 11 The ANN architecture approaches as they are the most popular processes
commercialized in public. The general quality issues currently
identified in FDM technique are the surface quality arose from
staircase effect, limited layer thickness, thermal distortion
shrinkage causing low resolution and dimensional inaccuracy.
Many kinds of post-processing and build orientation methods
have been investigated to alleviate the surface and mechanical
property issues. This EBP has advanced especially in data
analytics and algorithms, which signify that other fabrication
techniques can possibly adopt to predict their optimal design
and printing parameters. Research in these fields can be the
next step to understand the effects of various parameters on the
quality of fabricated parts.

2.5 Powder bed fusion processes


Most of the processes in powder bed fusion (PBF) system use a
high source of thermal energy to melt powder-based materials
(i.e. polymer, metal and ceramic) into desired structures and
shapes (Stucker, 2012). In the PBF processes, three different

655
Quality control in additive manufacturing Rapid Prototyping Journal
Hoejin Kim, Yirong Lin and Tzu-Liang Bill Tseng Volume 24 · Number 3 · 2018 · 645–669

techniques have been developed: selective laser sintering (SLS) dominantly more than thermocouples for temperature
technique, which uses partial melting to fuse powders together, monitoring in PBF processes. Shishkovsky et al. (2008) used
selective laser melting (SLM), which uses full melting on thermocouples to monitor the build of six different
powders transferring into homogenous parts, as well as fusing intermetallic powders and measure the temperature on the
and electron beam melting (EBM), which uses an electron powder bed (Shishkovsky et al., 2008). In addition, Van Belle
beam to design to fuse only metal powders in vacuum et al. (2013) used thermocouple attached to the bottom of the
environment. For medical applications, SLS technique is base plate with a strain gauge to record residual stresses (Van
widely used for human skeleton or organ reconstruction owing Belle et al., 2013), and Taylor and Childs (2001) also used
to its high accuracy and repeatability when compared to FDM, thermocouples positioned under the bed’s surface to monitor
LOM and inkjet printing techniques (Silva et al., 2008; Rengier energy absorption and powder effective conductivity for better
et al., 2010). understanding of heat transfer in metal powder during laser
Edwards et al. (2013) studied mechanical property processing of the powder bed (Taylor and Childs, 2001). Low
improvement of Ti-6Al-4V alloy part printed by EBM. By and Ake (2004) invented a thermocouple control system to
elevating certain bed temperature, residual stresses were improve uniform distribution of temperature on powder bed
minimized; therefore, post heat treatment could be eliminated. during part build. The thermocouple was attached inside of the
In addition, fatigue performance is highly dependent on rough powder bed along with IR sensor and communicates with
surface and porosity, which means that the technique highly temperature transmitter through circuitry in real-time (Low
requires post-process (e.g. peening). However, this approach is and Ake, 2004).
Downloaded by The University of Texas at El Paso At 09:51 27 April 2018 (PT)

not agile and smart enough to enhance part quality. Ideally, the Besides, research was focused on the IR thermography-
printing parameters need to be quickly and dynamically tuned based monitoring and control system incorporated in the
up by optimized input variables. Accordingly, research EBM system. Price et al. (2012) demonstrated the feasibility
attempts have been put into practice to monitor process to of using a near IR thermal camera for temperature
control process input variables (Edwards et al., 2013). Melvin measurement in hatch melting, preheating and contour
et al. (1994) used a video microscopy system to observe melting events during the EBS process (Price et al., 2012).
sintering and flow behavior in real-time, which can evaluate Yadroitsev et al. (2014) used a galvanometric scanner
different sintering characteristics of the materials (Melvin et al., system for temperature distribution monitoring of melt pool
1994). Benda (1994) developed an infrared light thermal of Ti6Al4V alloy in the SLM system (Yadroitsev et al.,
sensor to control laser power for better uniform sintering 2014). Dinwiddie et al. (2013) developed continuous data
performance (Benda, 1994). Zeng et al. (2012) reviewed capturing method using the IR camera to demonstrate
literatures on the thermal modeling method in SLS and SLM feasible work to detect porosities inside materials and
techniques. It was summarized that uniform temperature understand thermal phenomena such as it happens when
distribution of fields during printing processes leads to better beam and powder interact with each other (Dinwiddie et al.,
quality; there is a need to provide information from monitoring 2013). Kruth et al. (2007) developed temperature feedback
temperature of the melt pool to be able to control process control system by positioning photodiodes and CMOS
parameters for part quality. As a temperature monitoring digital camera on laser beam to stabilize melt pool
system, pyrometers and thermocouples were used for temperature distribution in SLM system (Kruth et al.,
monitoring its temperature (Zeng et al., 2012). 2007). In 2013, Mireles developed automatic feedback
Pyrometer is defined as the noncontact measurement of body control system using IR camera for defect detection, and
temperature based on emitted thermal radiation. There are two notification tool to stop printing process when porosity level
types of pyrometer used depending on the application: reaches to certain level. In addition, achieved data can help
photodiodes and digital cameras [e.g. charged-coupled device process parameters to be optimized and stabilize
(CCD), complementary metal oxide semiconductor (CMOS)]. temperature automatically through image processing and
Both detects radiation and light, respectively, and convert them auto-decision-making (Mireles et al., 2013). In 2015,
into electric signal. Bi et al. (2007) developed a novel laser Mireles developed in-situ defect monitoring methods using
cladding head with photodiodes and digital camera (CCD) to IR thermography inside an EBM machine. This captured IR
enable real time monitoring to monitor the status of machine images provided a good indication of measured defects
parts, thus, improve part quality (Bi et al., 2007). Kleszczynski et geometry presence, which can be used to re-melt defect (i.e.
al. (2012) used a high resolution CCD camera for surface error porosity) for the purpose of in-situ correction, as shown in
detection through image processing, as shown in Figure 12 Figure 13 (Mireles et al., 2015). This method saved total
(Kleszczynski et al., 2012). Lott et al. (2011) analyzed images of manufacturing time and decreased altering microstructure
melt pool size to adjust laser output power by using a CMOS compared with HIPing (hot isostatic pressing) process, and
camera with an additional illumination source required for high eventually increased the mechanical properties throughout
scanning velocities, resolution, as well as photodiodes (Lott et the fabricated part.
al., 2011). Kolossov et al. (2004) developed a thermal model In summary, the PBF processes, as compared with other
that the temperature evolution and sintering formation can be processes, have mostly advanced in temperature monitoring
simulated by a 3D FEA to predict thermal properties (i.e. system using sensors. Currently, the main quality issues
thermal conductivity and specific heat; Kolossov et al., 2004). identified was the heterogeneous distribution of powder bed
Thermocouples, in contrast with pyrometers, are defined as temperature and laser output power. Non-contact method,
contact measurement of temperature which can reduce the which is more adaptable owing to the freedom of process, was
freedom of process. Therefore, pyrometer has been used used to solve these quality issues. In addition, thermal

656
Quality control in additive manufacturing Rapid Prototyping Journal
Hoejin Kim, Yirong Lin and Tzu-Liang Bill Tseng Volume 24 · Number 3 · 2018 · 645–669

Figure 12 Setup of the CCD camera system in front of machine window (left) and setup of CMOS camera system with illumination source and
photodiode for high scanning velocities and resolution (right)
Downloaded by The University of Texas at El Paso At 09:51 27 April 2018 (PT)

structures, while in DED metal powders are fed from the


Figure 13 Correction of un-melted powder through layer re-melt deposition head and melt onto a substrate by energy source
shown by the build file with the re-melt area shown by the red arrow coming out from middle of the head. There are different types
(top), showing the defect present in the IR image by the black arrow of techniques such as laser engineering net shape (LENS), laser
(middle) and showing the defect not shown on the left column owing to cladding (LC) and direct metal deposition (DMD). Major
the re-melt (bottom) considerable processing parameters in DED are melt pool
temperature, material delivery rate and nozzle tip-substrate
distance.
The process monitoring and control focused on aforementioned
major printing parameters have been of increasing interests to
improve quality and reproducibility. Bi et al. (2013) developed
closed-loop controller that sends monitored temperature feedback
and adjust laser power density in real-time to improve dimensional
accuracy by using the IR pyrometer that measures melt pool
temperature. This closed-loop controller can eliminate the heat
accumulation and achieve oxidation-free clad surface with feeding
argon gas shielding to avoid disturbance from IR temperature
signals, as shown in Figure 14 (Bi et al., 2013). Fathi et al. (2008)
used similar closed-loop laser cladding process that can reduce
stair–step effects as well as production time (Fathi et al., 2008).
In addition, Bi et al. (2006) investigated printing parameters
relationships by looking at microstructure, hardness and
dimensional accuracy. It was proved that variations of melt
pool temperature and size influencing cooling rate and
solidification conditions result in heterogeneous dimensional
modeling techniques such as 3D FEA were implemented to accuracy and microstructure (Bi et al., 2006). In a similar
predict thermal properties. Post processes (e.g. heat treatment, manner, Hu and Kovacevic (2003) implemented a system to
peening, etc.) are necessary after the 3D printing process to monitor and control the powder delivery rate through a real-
alleviate residual stress, rough surface and porosity. Research time sensing and control using IR high-speed camera
efforts were made to develop feedback control system for positioned with laser beam. Captured images of melt pool by
temperature distribution, defect detection and in-situ this camera were also used for analysis to adjust laser power for
correction. uniform temperature distribution. The reflected data from the
sensor helped improve the 3D printed part quality (Hu and
2.6 Directed energy deposition processes Kovacevic, 2003). Similarly, Hofmeister et al. (1999) used a
Directed energy deposition (DED) processes use a laser beam CCD camera mounted with laser beams to capture melt pool
to fuse materials (e.g. metal powder and wire feedstock) images and analyze the melt pool cooling rate. Some
delivered from a material deposition head and substrate where researchers studied process monitoring with DOE method
materials are deposited (Tapia and Elwany, 2014; Vaezi et al., (Hofmeister et al., 1999). Ermurat et al. (2013) and Lee (2008)
2013). PBF process feeds a powder layer using a blade or roller used a high-speed camera and CCD camera to monitor nozzle
onto the powder bed and selectively melts sections to build 3D and powder delivery rate. With its extracted data, DOE is used

657
Quality control in additive manufacturing Rapid Prototyping Journal
Hoejin Kim, Yirong Lin and Tzu-Liang Bill Tseng Volume 24 · Number 3 · 2018 · 645–669

Figure 14 Front view of the deposited sample, the laser power and IR-temperature signals after (left) and before (right) stabilizing IR–temperature
signals

to analyze the behavior of the nozzle dimension and powder rate (Hu et al., 2011). In addition, Smurov et al. (2012) used
Downloaded by The University of Texas at El Paso At 09:51 27 April 2018 (PT)

size along with changes of standoff distance and gas flow high-speed cameras for calculating the powder speed and flux
(Ermurat et al., 2013; Lee, 2008). by positioning the cameras next to the nozzle head (Smurov
Some researchers adopted analytical models to analyze the et al., 2012).
thermal behavior of the process. Ye et al. (2006) and Wang et al. Some of the researchers conducted studies on the monitoring
(2009) reported that DED fabrication requires an in-depth layer height through the use of CCD cameras and neural
understanding of the entire thermal behavior of the process. In network algorithm. Iravani-Tabrizipour and Toyserkani (2007)
these papers, the finite element method (FEM) was used to used three CCD cameras-based optical detectors positioned at
obtain the thermal distribution and numerical simulation, 120 degree of each other around the nozzle head used in LC
known as Thermaviz. Two-wavelength imaging pyrometer was technique. A novel algorithm composed of an image-based
used to provide temperature data such as temperature tracking protocol and a recurrent neural network were
mapping, cooling and heating rate, modeling and simulation, developed to measure the clad height in real-time. These
etc. The numerical simulation and FEM results provide the features tracking and trained network showed a promising
user further predicting models and parameters (Ye et al., 2006; results in detecting the clad height in various trajectory angles
Wang et al., 2009). Alimardani et al. (2007) also used a 3D in real-time, which measured the clad height with 10 Hz and
numerical method for predicting transient geometrical and 60.15 mm average precision independent from clad path with
thermal characteristics of multilayer laser solid freeform about 12 per cent maximum error (Iravani-Tabrizipour and
fabrication. Thermal and geometrical properties can be Toyserkani, 2007). In a different approach, Toyserkani and
numerically obtained and predicted even with variation of Khajepour (2006) used a CCD camera-based optical detector
parameters such as the powder feed rate, elapsed time, melt in close-loop control to be able to effectively measure not only
pool temperature, substrate roughness, etc. so that the quality clad height but also angle of liquid/solid interface in real-time.
can be more precisely optimized (Alimardani et al., 2007). Using the recorded different images of each layer, it estimates
(Figure 15). its roughness and product quality (Toyserkani and Khajepour,
For the studies on material delivery rate, Hu et al. (2011) 2006).
developed a monitoring and control system using IR diodes and In summary, DED processes have dealt with advanced
low-power laser. This system is designed for sensing the researches such as process monitoring, real-time closed loop
powder flow rate out of nozzle to ensure consistent powder feed feedback system, thermal behavior and data analytic algorithm.

Figure 15 Finite element mesh and geometry to simulate the LENS process for ten layer wall (left) and molten pool size distribution for each layer at
P = 600 W and V = 2.5 mm/s (right)

658
Quality control in additive manufacturing Rapid Prototyping Journal
Hoejin Kim, Yirong Lin and Tzu-Liang Bill Tseng Volume 24 · Number 3 · 2018 · 645–669

The main quality issues identified were related to part and waste is determined for the burn-out rule with
heterogeneous melt pool temperature, laser power density and calculated optimal laser energy levels, velocity and burn-out
material delivery rate. These heterogeneities can be solved by a distance to predict better de-cubing process. It is suggested that
closed-loop controller with data acquisition equipment such as optimal laser paths can be obtained by ordinary laser power
CCD and IR high speed cameras, which can sense printing with the burn-out rule and low cutting velocity (Chiu and Liao,
process data in real time and adjust them to optimal values. 2003).
Researches such as FEM simulation and data analytic The most important quality characteristic of LOM part is the
algorithm were also executed to predict the quality. vertical surface roughness (Ra). Kechagias (2007a, 2007b)
worked on experimental investigation of the surface roughness
2.7 Sheet lamination processes on LOM part. Improving vertical surface quality leads to
Sheet lamination processes (SLP) are one of the additive reducing post-processing time, easier disengagement, easier de-
manufacturing techniques to build a 3D structure by cubing, process optimization and less finishing (Kechagias,
sequentially laminating, bonding and cutting 2D cross-section 2007a). Using statistical models, Chryssolouris et al. (1999)
sheets of such polymer, ceramic coated paper or composite found that vertical surface roughness (Ra) is mainly influenced
paper by either a laser or a knife, and layers bonded by glue or by process parameters such as heater temperature, layer
adhesive coating. This technique is also called laminated object thickness and laser speed when measured on ZX plane,
manufacturing (LOM). The bonding process is accomplished whereas pressure and heater speed have more effects on surface
by heat or pressure applied from a heated cylinder rolling along roughness when on ZY plane according to ANOVA analysis
Downloaded by The University of Texas at El Paso At 09:51 27 April 2018 (PT)

the sheet. While there is another technique called ultrasonic (Chryssolouris et al., 1999).
consolidation (UC), which has the capability of ultrasonic Another critical issue is a delamination between bonded
welding of metal sheets, it will be not covered in this paper as layers owing to weak bonding or process malfunctioning (Vaezi
this technique is a hybrid fabrication method combining the et al., 2013; Kechagias and Iakovakis, 2009). Some researchers
additive manufacturing with subtractive process (Vaezi et al., proposed mathematical models to enhance the bonding of
2013; Kechagias, 2007a). laminates. Pak (1994) developed a mathematical equation
Although LOM process provides faster processes with sufficient containing a relation between temperature at arbitrary depth of
quality characteristics on large physical prototypes than other the part (Z-axis) and critical parameters in terms of bonding of
techniques (Kruth, 1991), this technique has many difficulties and lamination such as temperature (Th), speed (), and pressure
drawbacks in manufacturing process, which are poor surface finish (d ) of heater roller. Reeves and Cobb (1995) presented an
before post-processing, difficult disengagement and only good analytical model to predict optimal heater temperature needed
tensile strength properties in laminates direction compared with for good laminate bonding. Sonmez and Hahn (1998) analyzed
other techniques (Chryssolouris et al., 2003; Upcraft and Fletcher, heat transfer and stress using FEA simulation and found the
2003). following:
The LOM process is separated into two methods when it  surface temperature of heater should not be higher than
comes to the way of laminating a sheet of material: cut-then- glass transition temperature of the used paper;
bond and bond-then-cut. The cut-then-bond method, which  surface stresses of laminate increase with the decrease in
cuts first and then re-position on workpiece to bond, is effective layer thickness or roller diameter; and
for de-cubing, however, causes a loss in positioning precision  higher rolling speed is more likely to cause delamination or
and accumulated error. The bond-then-cut, which bonds a malfunctioning on sheet bonding (Sonmez and Hahn,
sheet on workpiece and then cuts it by laser, is not able to 1998).
perform complicated parts (e.g. hollow and vase) owing to
Flach et al. (1998) developed a thermal analytical model with
difficulty in de-cubing inner waste material so that it consumes
various process parameters for fabrication of functional ceramic
high laser power for de-cubing and crosshatching. In addition,
and composite parts and founded the following:
it is time-consuming and labor-dependent.  chamber air temperature affects temperature of part being
Liao et al. (2003) proposed an online de-cubing process in
built;
the bond-then-cut method and with a developed bridge-  roller speed significantly affects surface layer in direct
supporting algorithm. Shielding paper laminated on a sheet was
contact; and
torn for laser cutting a sheet material and then used to remove  delay time of the process affects the temperature of the
waste materials online by using adhesion force between
part being built (Flach et al., 1998).
shielding paper and waste material, as shown in Figure 16 (Liao
et al., 2003). For easy de-cubing process, an algorithm for Some researchers studied on predicting surface roughness.
building bridges from 2D drawing was developed to support Reeves and Cobb (1995) developed a general model for
and position islands part while de-cubing process (Chiu et al., predicting roughness on sloped surface. The surface angle (u )
2003). This proposed process enabled the fabricating process and the layer thickness (Lt) are critical parameters in predicting
to perform various geometric shapes with significantly reduced sloped surface roughness. Paul (2001) reported that the
processing time up to 20-50 per cent. Chiu and Liao (2003) orientation angle and sheet thickness are statistically significant
proposed a novel laser path planning strategy to improve de- effects on surface roughness and proposed a mechanistic model
cubing process problem. Burn-out rule proposed in this study for sloped surfaces considering the two critical parameters
was applied in laser path planning to reduce damaging part and (Reeves and Cobb, 1995). Chryssolouris et al. (1999)
bonding strength between the part and waste, eventually saving developed a semi-empirical model in consideration with
68 per cent of de-cubing time. Overlap zone between fabricated parameters such as changes of layer thickness, heater speed,

659
Quality control in additive manufacturing Rapid Prototyping Journal
Hoejin Kim, Yirong Lin and Tzu-Liang Bill Tseng Volume 24 · Number 3 · 2018 · 645–669

Figure 16 (A) The mechanism of the proposed online de-cubing process; (B) the original sliced 2D contour drawing; (C) the altered 2D drawing
developed by bridge-supporting algorithm
Downloaded by The University of Texas at El Paso At 09:51 27 April 2018 (PT)

heater temperature and platform retraction, based on matrix In summary, there have been issues on de-cubing process,
experiments with different surface angles and cross of laminates vertical surface roughness and delamination. Development of
direction (Chryssolouris et al., 1999). Ahn et al. (2012) online de-cubing process and bridge-supporting algorithm has
proposed theoretical expressions to quantify average surface solved several issues such as saving energy, post-process time
roughness according to surface angle variation. In case of and complicated structure arose by de-cubing. Research in
upward or downward facing surface (0° < u < 90°, 90° < u < mathematical models and neural network were implemented to
180°), theoretical model was expressed by RðaÞu ¼ 1l A, enhance bonding of laminates and predict optimal process
where R(a) denoted the surface roughness, l the base length of parameters. Vertical surface roughness and delamination are
unit profile, which is distance between edge point of sheets and the future challenges to be solved to improve quality. Seven
A the total surface area; therefore, R(a) is affected by the base techniques in AM technology was summarized and compared
length. On the other hand, in case of vertical surface, roughness in quality point of view, as shown in Table I.
layer thickness (t) affected R(a) expressed by RðaÞ90 ¼ 1t A.
These expressions were verified by comparing with empirical 3. Classification of research findings
data (Ahn et al., 2012).
Researches of QC in AM are being rapidly investigated recently
Kechagias (2007a, 2007b) used a design of experiment to
owing to high feasibility, reliability and growth of mechanical
investigate dimensional accuracy of the LOM part. It was and material aspects applied to various areas. The first
found that parameters such as heater, speed, temperature and consideration on AM technology when it comes to QC is that
nominal layer thickness mainly affect dimensional accuracy in several steps are classified to achieve design requirement, as
X and Y directions (Kechagias, 2007b). In 2008, he proposed a described by Hollister et al. (2015) who studied a splint used in
simple model to predict and optimize LOM process medical area. This paper designed five steps of the design
performance using a feed-forward back propagation neural control process, as shown in Figure 17. The first step is a design
network (FFBP–NN). It was concluded that this model is a input considering all parameters on designing and modeling a
quick prediction of each performance measurements and can part to meet mechanical and specific requirements such as laser
be used to select optimal process parameters values (Kechagias power, speed and particle size. Design output as the second
and Iakovakis, 2009). Vaezi et al. (2013) reported that quality step comprises tests performed to measure the design input
of the LOM also requires well-prepared material sheets to based on the standard operating procedures by determining the
reduce variations in layer thickness and materials content. compressive, bending and opening stiffness of the splint by
Depth of cuts varied with variations of layer thicknesses and using the method of destructive evaluation. The first two steps
material content so that these variations lead to more damage to are to test sample parts to generate the design variables for
a bonded layer below. In addition, the LOM part can be post- optimized final part being considered in design implementation
processed to reaction bonding process during a pyrolysis cycle as third step. Design verification as fourth step is to ensure the
to enhance surface quality and laminates bonding (Vaezi et al., final manufactured part meets design input via geometric and
2013). Chryssolouris et al. (2003) investigated the influence of mechanical evaluation (e.g. calipers, Micro-CT, etc.). The last
different process parameters of LOM on the tensile strength of design validation is to test pre-clinical model and human
LOM part. It was found that the most significant process clinical implantations of final part.
parameter affecting the tensile strength of part was layer This outline suggested as a good framework for providing
thickness. Therefore, it can be concluded that layer thickness can steps of QC in medical-related research. However, there are
provide a prediction on tensile strength of part (Chryssolouris drawbacks when figuring out the design variables. It is time-
et al., 2003). consuming and wastes materials in making samples for

660
Downloaded by The University of Texas at El Paso At 09:51 27 April 2018 (PT)

Table I Summary and comparison of seven AM techniques in quality point of view


Category Important quality parameters Advantages Disadvantages Advanced skills Challenges

Photopolymer vat Resin viscosity, density and temperature Excellent dimensional accuracy Only UV curable polymer resin Projection SL technique Development of variety vat solution
processes (PVP) Build orientation Good surface finish Material contamination and wastes Accurate custom-design and evaluation Expensive scanning equipment such as
Laser scan speed Biocompatible process Mechanical property limited by resin- processes through MRI and CT MRI and CT
Exposure energy Microscale fabrication based polymer Data analytic and algorithm Resolution
Penetration depth (10 m m) Thermal shrinkage Biomaterial fabrication
Nano/micro scale fabrication
Material jetting Droplet shape, velocity, diameter, viscosity Excellent resolution Nozzle clogging DoD technique Exposure energy
processes (MJP) and surface tension Fast process Cell damage in certain temperature Micro-scale fabrication Nozzle clogging
Jetting frequency, signal width and Excellent dimensional accuracy or frequency Biomaterial fabrication More fabrication freedom on biomaterials
voltage magnitude Multiple material printing capability Nano particles agglomerates causes Real-time close-feedback loop for printing
Substrate temperature and evaporation Microscale fabrication clogging correction through image processing on
rate (100 m m) Low viscosity droplet and depth map
Humidity Self-calibration of print head
Quality control in additive manufacturing

Quality prediction
Hoejin Kim, Yirong Lin and Tzu-Liang Bill Tseng

Binder jetting Powder surface treatment, size, shape, Excellent quality control ability for ceramic Poorer accuracy and surface finish DoD technique Interaction performance between binder
processes (BJP) packing density and distribution and metal fabrication compared to MJP compared with MJP technique Ceramic and metal fabrication and powder
Binder viscosity, surface tension, droplet technique Infiltration process needed for post- Accumulative accuracy of deposited layer
size, velocity and temperature DoD treatment thickness
Printing layer thickness, orientation, Droplet placement
binder saturation and delay time Delay time of applying new layer
Dimensional change during post-process
Extrusion-based Layer thickness, build orientation, raster Fast process Fair resolution FDM technique Resolution

661
processes (EBP) width and angle and extrusion Good material properties Staircase effect Data analytic and algorithm Dimensional accuracy
temperature Thermal distortion Post-processing Shape change
Post-processing (exposure time and Thermal shrinkage Optimal build orientation Clogging issues on composites materials
chemical temperature) Low dimensional accuracy
Delamination
Powder bed fusion Powder bed temperature Wide range of materials Fair resolution limited by particle size SLS technique Resolution
processes (PBF) Laser/beam output powder Excellent dimensional accuracy Residual stress Temperature monitoring system Laser output feedback control system in
Powder size Excellent repeatability Porosity Thermal modeling and image processing real time
Atmosphere Good material property Temperature feedback control system Atmosphere control
Real-time defect detection and in-situ
correction
Directed energy Melt pool temperature Wide range of materials Fair resolution Closed-loop controller system for real-time Resolution
deposition Material delivery rate Good material property Thermal stress feedback of temperature, laser output, Surface post-process
processes (DED) Distance between nozzle tip and substrate clad height and delivery rate Atmosphere control
Laser power density
Sheet lamination Heater temperature Fast process Poor resolution LOM technique Resolution
processes (SLP) Layer thickness Sufficient quality on large prototype De-cubing or crosshatching process Burn-out rule Consistent sheet thickness
Laser speed/power Good tensile strength in laminated Delamination Data analytic and algorithm Repositioning precision and accumulation
Rolling speed/pressure direction Shrinkage Online de-cubing process error (cut-then-bond process)
Chamber air temperature Poor tensile strength in Z direction Vertical surface roughness
Rapid Prototyping Journal

Delay time Waste Material damage by laser cut


Orientation Tensile strength in Z direction
Volume 24 · Number 3 · 2018 · 645–669

Delamination
Quality control in additive manufacturing Rapid Prototyping Journal
Hoejin Kim, Yirong Lin and Tzu-Liang Bill Tseng Volume 24 · Number 3 · 2018 · 645–669

Figure 17 Schematic outline of the splint design control process by reference


Downloaded by The University of Texas at El Paso At 09:51 27 April 2018 (PT)

mechanical test and generating optimal design variables. If the variables and printing parameters generated from mechanical
design variables are adjustable during the design and printing tests. This approach finally brings the optimized splint part
process, the entire process of QC can be more agile and individually customized for patients who needs perfect clinical
reliable. Figure 18 shows proposed classification of four main recovery (Hollister et al., 2015). However, it requires a long
processes to take account of QC during AM cycle and describes time-consuming procedure to fabricate and iterations of
detail works recently studied and also included in Section 2. mechanical testing depending on patients to generate optimal
On the basis of the four steps, modeling, printing, post- design variables. Recently, one research (Sitthi-Amorn et al.,
treatment, and part evaluation during the entire 3D printing 2015) investigated on developing real-time printing control
process, the following four categories are described briefly and technique to correct design and printing factors in real-time
issues are discussed. through closed-feedback loop. This technique analyzes each
printed layer in real-time through the machine vision,
3.1 Design process computing depth difference between the layers, and sends
This is so-called pre-QC considering from raw material’s feedback to central computer to control amount of jet droplet
quality to model design. Material selection is important injection. This technique opens up the real-time design process
because it may be contaminated or may have air porosities, during printing process, but is limited to MJP. Some
which affects directly to the part property. Good design ability researchers have studied optimizing design parameters by
by setting up optimized values of design parameters (e.g. layer developing analytical modeling and data-mining based on
thickness, infill pattern, printing speed and delay, mesh, etc.) is quality prediction algorithms as mentioned in Section 2.
directly correlated to part’s good quality. Recently, Hollister However, this research is limited to BJP, FDM and SLP
et al. (2015) studied on designing splint structure meeting techniques. There are needs to launch this research in other
quantitative clinical requirement by finding optimized design printing techniques.

Figure 18 The proposed classification of the four main processes during AM cycle

662
Quality control in additive manufacturing Rapid Prototyping Journal
Hoejin Kim, Yirong Lin and Tzu-Liang Bill Tseng Volume 24 · Number 3 · 2018 · 645–669

3.2 Printing process 3.4 Evaluation process


This is the so-called in process-QC considering from printing This process is to evaluate part’s reliability. This includes all
setup to process finish of part. Agarwala et al. (1996) reported activities measuring external and internal characteristics of
that internal and surface defects of fabricated wax pattern could printed parts whether it meets specific design and mechanical
be eliminated by the optimization of FDM parameters requirements. The external characteristics are measured by
(Agarwala et al., 1996). Likewise, optimized set-up of printing such a calipers, machine vision, roughness tester, etc., while the
parameters (e.g. extruder temperature, calibration, pattern, internal characteristics are measured by destructive or non-
feed and flow rate, etc.) is critical for the quality. In recently, destructive evaluations. Some studies have been attempted on
researchers have studied the optimization of printing optimizing evaluation process. Muller and De Jean (2015)
parameters through real-time monitoring and printing control developed a novel stereo microscope adapter, called SweptVue,
as mentioned in Section 2. To date, more research has been for cost-effective microfabrication quality control, as shown in
conducted in terms of the printing process compared to other Figure 19. Using this optical technique, surface elevation and
processes. Farzadi et al. (2015) studied the effects of the layer accuracy were examined over multiple regions on plateaus and
printing delay on 3D printed part’s physical and mechanical hemispherical surfaces, and building errors such as decreased
properties for better selections of printing parameters and the heights were discovered when approaching the edges of
best printing conditions (Farzadi et al., 2015). In addition, plateaus, inaccurate height steps and poor tolerances on
channel width (Muller and De Jean, 2015). In addition to
Kaveh et al. (2015) designed proper benchmarks (e.g. extruded
external evaluation, Hollister et al. (2015) used a Micro-CT
Downloaded by The University of Texas at El Paso At 09:51 27 April 2018 (PT)

temperature, etc.) and statistical equations calculating


method, which non-destructively builds an image of internal
calibration factors (precision and internal cavity) used to
structures allowing quantitative and qualitative measures
determine or to optimize the value of printing parameters
(Hollister et al., 2015). These two approaches used traditional
(Kaveh et al., 2015). Similarly, Hollister et al. (2015) created
evaluation methods (e.g. calipers and SweptVue) and high tech
standard test coupons with consistent geometry and known
evaluation methods (e.g. Micro-CT) for the evaluation of
mechanical properties included in every build providing a geometry accuracy and internal structure characteristics.
standard with which to compare builds at different performing However, the latter evaluation method requires high cost and
times because printing parameter should be changed according big volumetric facility.
to custom devices for different patients (Hollister et al., 2015).
These studies aim to optimize printing process for quality along
4. Challenges and future trends
with different printing parameters in different 3D printing
techniques such as PBF, EBP, BJP and MJP techniques. As As a consequence of what has been discussed above, it is
mentioned above, a research by Sitthi-Amorn et al. (2015) and evident that vast researches and variety approaches have been
Mireles et al. (2015) studied real-time optimization of printing studied to improve the quality of AM processes. However, it
parameters by closed-feedback loop control in MJP (Sitthi- has been seen that there are still many issues that occur from
Amorn et al., 2015) and in-situ defects correction in EBM research reviews. To achieve agile and streamlined AM
technique (Mireles et al., 2015). These techniques might not be processes, from design to part evaluation and enhanced quality
limited to other 3D printing techniques when it comes to its AM product, novel AM processes for QC could have the
application. However, PVP and BJP techniques could not be potential to address these issues:
 Prediction of optimal printing parameter: Iterations works by
applied owing to the inaccessibility of the machine vision
external and internal evaluation to find optimal printing
system and IR thermography. Therefore, there is a need for
parameters; these evaluations are tedious and not agile. If
alternatives for real-time monitoring, closed-feedback loop
it can predict printing parameter based on data analytic
printing control, and in-situ defect correction processes to be
analyses (e.g. data mining) at design process, final part can
applied in these techniques.
be expected with better dimensional accuracy and surface
quality.
 Prediction of mechanical property: As mentioned above,
3.3 Post process iterative mechanical testing to meet the requirements
This is so-called post-QC considering finished part’s quality eventually will waste materials. If a designing tool
after printing process and before part evaluation. Some of the (software) before printing process has simulation
techniques such as PBF heavily rely on this process for capability or mechanical prediction module, better design
improving better surface and mechanical property of printed process can be performed by users based on the predicted
part through sintering and peening processes. Edwards et al. data.
(2013) attempted to remove post process on EBM printed alloy  Robust real-time monitoring and process control in build
parts by using the elevating bad temperature to minimize process: In MJP, PBF and DED, real-time monitoring and
residual stresses (Edwards et al., 2013). In the EBP and SLP close-feedback loop control and defect correction during
techniques, post processing was used in chemical, thermal and build process has been recently developed. These
hot cutter processes for surface quality improvement. However, techniques are advanced technology that integrates
this research is limited to certain materials such as ABS machine vision, data collection, image processing and
polymer. Therefore, there is a need to launch research on other closed-feedback loop (Sitthi-Amorn et al., 2015). In PVP,
materials. In the MJP technique, sintering was adopted as post real-time monitoring systems have been used to monitor
process for 3D printed Zr-based ceramic dental prostheses. temperature, layer height, etc. for data collection but

663
Quality control in additive manufacturing Rapid Prototyping Journal
Hoejin Kim, Yirong Lin and Tzu-Liang Bill Tseng Volume 24 · Number 3 · 2018 · 645–669

Figure 19 (A) A pair of hard-baked photoresist pads used for a microfluidic device, seen in monochrome when imaged through a traditional
microscope (left). The SweptVue adapter measures the relative depth of the features in the image and colorizes them in software using a jet color map,
with feature heights increasing blue to red (right); (B) depth maps can be exported to MATLAB software for further image processing and 3D plotting;
(C) the SweptVue stereo microscope adapter, shown mounted to an Olympus SZX7 stereo microscope

image processing (computation) and closed-feedback loop and printing process are displayable and controllable
Downloaded by The University of Texas at El Paso At 09:51 27 April 2018 (PT)

control. If these can be achieved in other processes, they under remote control, massive and customized AM
can provide better quality and agile AM process. productions would be feasible in less time and cheaper
 Feedback interaction between design or printing process and cost of production in the near future.
part evaluation: Closed-feedback loop recently developed
for correcting design and printing factors in real-time is a
good example of intercommunication between design and 5. Conclusion
printing processes. There is another research that AM technology has reached a crossroad right before being used
compares CT scanned images with initial design model to as end-part production. Many techniques have been developed
calculate dimensional discrepancy. If there is a module or up to now to respond to the demand of high-quality 3D printed
software that enables feedback of part evaluation (e.g. parts but more studies are still required to improve their
inner defect, dimensional inaccuracy and surface crack) to systems and quality. This paper presented a review on quality
be considered when redesigning to remove these control in seven different techniques in the AM technology,
discrepancies, high quality of product can be expected in and provided detail discussions in each quality process stages.
shorter time. For example, on an unresolvable part defect Most of the AM techniques follow the trend of using in-situ
in spite of several printing efforts, design software can sensors and cameras to acquire processing data for real-time
adjust design or printing parameter (e.g. printing speed, monitoring and quality analysis. This effort opens up the in-situ
extrusion rate, etc.) using the feedback data generated correction by the closed-feedback loop and image processing.
from part evaluation; therefore, defects can be prevented However, owing to the closed geometry of the 3D printing
before printing process.
system, some techniques such as BJP are not eligible to adopt
 Agile part evaluation: Part evaluation is also time-consuming
this technique. More sophisticated process monitoring and
like the printing process and waste lots of materials owing to
agile feedback control are required to enable end-product to be
iterative destructive evaluation. Non-destructive evaluation
better quality and faster process. Processes such as EBP are
(NDE; offline inspection) is needed to enable validation of
more advanced regarding data analytics and predictive
process performance. However, it is expensive and requires
algorithms-based research regarding mechanical properties and
skills to operate the machine. If machine vision or NDE
optimal printing parameters. This data analytics research is
technique can be imbedded into the 3D printer for external
critical for optimization and prediction of design and printing
and internal quality evaluation as online inspection, part
process. This research effort would lead to artificial control
evaluation process can be agile and finished in 3D printers
without separated evaluation process. For example, 3D system that enables to produce perfect quality of AM product.
scanners that are commercially available are imbedded into MJP technique has advanced with systems that are integrated
3D printers can be used for surface quality evaluation with closed-feedback loop, machine vision and image
through image processing right after printing process. processing to minimize quality issues during printing process. It
 High speed fabrication and scale: Future technology will is expected that future challenges would use these predictive
require more rapid and bigger scale customized production algorithms based research and integrated system in the near
skills. However, in the current technology, the faster printing future for proved QC-based agile and accurate printing
process is, the worse its quality will be. Fabrication speed is process.
significantly correlated to part quality in all techniques and
dependent on part price. Increasing this speed with an
References
increase in quality will be the challenge in the future of AM.
 Cyber quality control: AM technology can be incorporated Addanki Sambasiva Rao, M.A.D., Venkatesh, J.V.L. and Ojha,
with remote connection for quality monitoring and D. (2012), “Investigation post processing techniques to
control. If all the processed data attained from the sensor reduce the surface roughness of fused deposition modeled

664
Quality control in additive manufacturing Rapid Prototyping Journal
Hoejin Kim, Yirong Lin and Tzu-Liang Bill Tseng Volume 24 · Number 3 · 2018 · 645–669

parts”, International Journal of Mechanical Engineering & International Journal of Machine Tools and Manufacture,
Technology (IJMET), Vol. 3, pp. 531-544. Vol. 47, pp. 555-561.
Adolf, D.B., Martin, J.E., Chambers, R.S., Burchett, S.N. and Boschetto, A. and Bottini, L. (2015), “Roughness prediction in
Guess, T.R. (1998), “Stresses during thermoset cure”, coupled operations of fused deposition modeling and barrel
Journal of Materials Research, Vol. 13, pp. 530-550. finishing”, Journal of Materials Processing Technology,
Agarwala, M.K., Jamalabad, V.R., Langrana, N.A., Safari, A., Vol. 219, pp. 181-192.
Whalen, P.J. and Danforth, S.C. (1996), “Structural quality Boschetto, A. and Bottini, L. (2016), “Design for
of parts processed by fused deposition”, Rapid Prototyping manufacturing of surfaces to improve accuracy in fused
Journal, Vol. 2, pp. 4-19. deposition modeling”, Robotics and Computer-Integrated
Ahn, D., Kweon, J.-H., Choi, J. and Lee, S. (2012), Manufacturing, Vol. 37, pp. 103-114.
“Quantification of surface roughness of parts processed by Bose, S., Vahabzadeh, S. and Bandyopadhyay, A. (2013),
laminated object manufacturing”, Journal of Materials “Bone tissue engineering using 3D printing”, Materials
Processing Technology, Vol. 212, pp. 339-346. Today, Vol. 16, pp. 496-504.
Ahn, D., Kweon, J.-H., Kwon, S., Song, J. and Lee, S. (2009), Budzik, G. (2007), “Possibilities of utilizing 3DP technology
“Representation of surface roughness in fused deposition for foundry mould making”, Archives of Foundry Engineering,
modeling”, Journal of Materials Processing Technology, Vol. 7, pp. 65-68.
Vol. 209, pp. 5593-5600. Campbell, R.I., Martorelli, M. and Lee, H.S. (2002), “Surface
Ahn, S.-H., Montero, M., Odell, D., Roundy, S. and Wright, roughness visualisation for rapid prototyping models”,
Downloaded by The University of Texas at El Paso At 09:51 27 April 2018 (PT)

P.K. (2002), “Anisotropic material properties of fused Computer-Aided Design, Vol. 34, pp. 717-725.
deposition modeling ABS”, Rapid Prototyping Journal, Vol. 8, Çaydas, U. and Hasçalik, A. (2008), “A study on surface
pp. 248-257. roughness in abrasive waterjet machining process using
Alimardani, M., Toyserkani, E. and Huissoon, J.P. (2007), “A artificial neural networks and regression analysis
3D dynamic numerical approach for temperature and method”, Journal of Materials Processing Technology,
thermal stress distributions in multilayer laser solid freeform Vol. 202, pp. 574-582.
fabrication process”, Optics and Lasers in Engineering, Vol. 45, Chakraborty, D., Reddy, B.A. and Choudhury, A.R. (2008),
pp. 1115-1130. “Extruder path generation for curved layer fused deposition
Anitha, R., Arunachalam, S. and Radhakrishnan, P. (2001), modeling”, Computer-Aided Design, Vol. 40, pp. 235-243.
“Critical parameters influencing the quality of prototypes in Chiu, Y. and Liao, Y. (2003), “Laser path planning of burn-out
fused deposition modelling”, Journal of Materials Processing rule for LOM process”, Rapid Prototyping Journal, Vol. 9,
Technology, Vol. 118, pp. 385-388. pp. 201-211.
Armillotta, A. (2006), “Assessment of surface quality on Chiu, Y., Liao, Y. and Hou, C. (2003), “Automatic fabrication
textured FDM prototypes”, Rapid Prototyping Journal, for bridged laminated object manufacturing (LOM)
Vol. 12, pp. 35-41. process”, Journal of Materials Processing Technology, Vol. 140,
Asadi-Eydivand, M., Solati-Hashjin, M., Farzad, A. and pp. 179-184.
Osman, N.A.A. (2016), “Effect of technical parameters on Chryssolouris, G., Kechagias, J., Moustakas, P. and
porous structure and strength of 3D printed calcium sulfate Koutras, E. (2003), “An experimental investigation of the
prototypes”, Robotics and Computer-Integrated Manufacturing, tensile strength of parts produced by laminated object
Vol. 37, pp. 57-67. manufacturing (LOM) process”, CIRP Journal of
Beaman, J., Atwood, C., Bergman, T., Bourell, D., Hollister, Manufacturing Systems, Vol. 32, pp. 319-322.
S. and Rosen, D. (2004), “Assessment of European research Chryssolouris, G., Kechagias, J.D., Kotselis, J.L., Mourtzis, D.
and development in additive”, Subtractive Manufacturing, A. and Zannis, S.G. (1999), “Surface roughness modelling
Final Report from WTEC Panel, available at: http://wtec.org/ of the Helisys laminated object manufacturing (LOM)
additive/report/welcome.htm process”, 8th European Conference on Rapid Prototyping and
Bellini, A. and Güçeri, S. (2003), “Mechanical characterization Manufacturing, Nottingham, pp. 141-152.
of parts fabricated using fused deposition modeling”, Rapid Chumnanklang, R., Panyathanmaporn, T., Sitthiseripratip, K.
Prototyping Journal, Vol. 9, pp. 252-264. and Suwanprateeb, J. (2007), “3D printing of hydroxyapatite:
Benda, J. (1994), “Temperature controlled selective laser effect of binder concentration in pre-coated particle on part
sintering”, Proceedings of the Solid Freeform Fabrication strength”, Materials Science and Engineering: C, Vol. 27,
Symposium, DTIC Document, pp. 277-284. pp. 914-921.
Bi, G., Sun, C. and Gasser, A. (2013), “Study on influential Cima, M., Oliveira, M., Wang, H., Sachs, E. and Holman, R.
factors for process monitoring and control in laser aided (2001), “Slurry-based 3DP and fine ceramic components”,
additive manufacturing”, Journal of Materials Processing Proceedings of Solid Freeform Fabrication Symposium, Austin,
Technology, Vol. 213, pp. 463-468. TX, pp. 216-223.
Bi, G., Gasser, A., Wissenbach, K., Drenker, A. and Poprawe, Cohen, A., Laviv, A., Berman, P., Nashef, R. and Abu-Tair, J.
R. (2006), “Characterization of the process control for the (2009), “Mandibular reconstruction using stereolithographic
direct laser metallic powder deposition”, Surface and Coatings 3-dimensional printing modeling technology”, Oral Surgery,
Technology, Vol. 201, pp. 2676-2683. Oral Medicine, Oral Pathology, Oral Radiology, and
Bi, G., Schürmann, B., Gasser, A., Wissenbach, K. and Endodontology, Vol. 108, pp. 661-666.
Poprawe, R. (2007), “Development and qualification of a Cooke, M.N., Fisher, J.P., Dean, D., Rimnac, C. and Mikos,
novel laser-cladding head with integrated sensors”, A.G. (2003), “Use of stereolithography to manufacture

665
Quality control in additive manufacturing Rapid Prototyping Journal
Hoejin Kim, Yirong Lin and Tzu-Liang Bill Tseng Volume 24 · Number 3 · 2018 · 645–669

critical-sized 3D biodegradable scaffolds for bone ingrowth”, accuracy of 3D printed porous prototypes in bone tissue
Journal of Biomedical materials Research Part B: Applied engineering”, Ceramics International, Vol. 41 No. 7,
Biomaterials, Vol. 64, pp. 65-69. pp. 8320-8330.
Corcione, C.E., Greco, A. and Maffezzoli, A. (2006), Fathi, A., Khajepour, A., Durali, M. and Toyserkani, E.
“Temperature evolution during stereolithography building (2008), “Geometry control of the deposited layer in a
with a commercial epoxy resin”, Polymer Engineering & nonplanar laser cladding process using a variable structure
Science, Vol. 46, pp. 493-502. controller”, Journal of Manufacturing Science and Engineering,
Cotteleer, M. and Joyce, J. (2014), “3D opportunity: additive Vol. 130, p. 031003.
manufacturing paths to performance, innovation, and Flach, L., Jacobs, M.A., Klosterman, D.A. and Chartoff, R.
growth”, Deloitte Review, Vol. 14. (1998), “Simulation of laminated object manufacturing
Cui, X. and Boland, T. (2009), “Human microvasculature (LOM) with variation of process parameters”, Proceedings of
fabrication using thermal inkjet printing technology”, Solid Freeform Fabrication Symposium, pp. 407-416.
Biomaterials, Vol. 30, pp. 6221-6227. Frank, D. and Fadel, G. (1995), “Expert system-based
Cui, X., Boland, T., D’lima, D. and Lotz, M.K. (2012), selection of the preferred direction of build for rapid
“Thermal inkjet printing in tissue engineering and prototyping processes”, Journal of Intelligent Manufacturing,
regenerative medicine”, Recent Patents on Drug Delivery & Vol. 6, pp. 339-345.
Formulation, Vol. 6, pp. 149-155. Galantucci, L., Lavecchia, F. and Percoco, G. (2009),
Cui, X., Dean, D., Ruggeri, Z.M. and Boland, T. (2010), “Cell “Experimental study aiming to enhance the surface finish
Downloaded by The University of Texas at El Paso At 09:51 27 April 2018 (PT)

damage evaluation of thermal inkjet printed Chinese hamster of fused deposition modeled parts”, CIRP Annals-
ovary cells”, Biotechnology and Bioengineering, Vol. 106, Manufacturing Technology, Vol. 58, pp. 189-192.
pp. 963-969. Galantucci, L., Lavecchia, F. and Percoco, G. (2010),
Czy_zewski, J., Burzynski, P., Gaweł, K. and Meisner, J. (2009), “Quantitative analysis of a chemical treatment to reduce
“Rapid prototyping of electrically conductive components roughness of parts fabricated using fused deposition
using 3D printing technology”, Journal of Materials Processing modeling”, CIRP Annals-Manufacturing Technology, Vol. 59,
Technology, Vol. 209, pp. 5281-5285.
pp. 247-250.
Derby, B. (2010), “Inkjet printing of functional and structural
Gauvin, R., Chen, Y.-C., Lee, J.W., Soman, P., Zorlutuna, P.,
materials: fluid property requirements, feature stability, and
Nichol, J.W., Bae, H., Chen, S. and Khademhosseini, A.
resolution”, Annual Review of Materials Research, Vol. 40,
(2012), “Microfabrication of complex porous tissue engineering
pp. 395-414.
scaffolds using 3D projection stereolithography”, Biomaterials,
Dhariwala, B., Hunt, E. and Boland, T. (2004), “Rapid
Vol. 33, pp. 3824-3834.
prototyping of tissue-engineering constructs, using
Gibson, I., Rosen, D. and Stucker, B. (2014), Additive
photopolymerizable hydrogels and stereolithography”, Tissue
Manufacturing Technologies: 3D Printing, Rapid Prototyping,
Engineering, Vol. 10, pp. 1316-1322.
and Direct Digital Manufacturing, Springer, Berlin.
Dinwiddie, R.B., Dehoff, R.R., Lloyd, P.D., Lowe, L.E.
Gopal, A.V. and Chakradhar, D. (2012), “Parametric
and Ulrich, J.B. (2013), “Thermographic in-situ process
optimization in electrochemical machining of EN-31 steel
monitoring of the electron-beam melting technology used
in additive manufacturing”, SPIE Defense, Security, and based on grey relation approach”, Applied Mechanics and
Sensing, International Society for Optics and Photonics, Materials, Vols 110/116, pp. 1649-1656.
pp. 87050K-87050K-9. Gross, B.C., Erkal, J.L., Lockwood, S.Y., Chen, C. and
Ebert, J., Özkol, E., Zeichner, A., Uibel, K., Weiss, Ö., Koops, Spence, D.M. (2014), “Evaluation of 3D printing and its
U., Telle, R. and Fischer, H. (2009), “Direct inkjet printing potential impact on biotechnology and the chemical
of dental prostheses made of zirconia”, Journal of Dental sciences”, Analytical Chemistry, Vol. 86, pp. 3240-3253.
Research, Vol. 88, pp. 673-676. Ha, C.W. and Yang, D.-Y. (2014), “Fabrication of micro open
Edwards, P., O’conner, A. and Ramulu, M. (2013), “Electron structure using 3D laser scanning method in nano-
beam additive manufacturing of titanium components: stereolithography”, International Conference on Manipulation,
properties and performance”, Journal of Manufacturing Manufacturing and Measurement on the Nanoscale (3M-NANO),
Science and Engineering, Vol. 135, p. 061016. IEEE, pp. 299-303.
Ermurat, M., Ali Arslan, M., Erzincanli, F. and Uzman, I. Han, L.-H., Suri, S., Schmidt, C.E. and Chen, S. (2010),
(2013), “Process parameters investigation of a laser- “Fabrication of three-dimensional scaffolds for heterogeneous
generated single clad for minimum size using design of tissue engineering”, Biomedical Microdevices, Vol. 12,
experiments”, Rapid Prototyping Journal, Vol. 19, pp. 721-725.
pp. 452-462. Henmi, C., Nakamura, M., Nishiyama, Y., Yamaguchi, K.,
Farzadi, A., Solati-Hashjin, M., Asadi-Eydivand, M. and Mochizuki, S., Takiura, K. and Nakagawa, H. (2007),
Osman, N.A.A. (2014), “Effect of layer thickness and printing “Development of an effective three dimensional fabrication
orientation on mechanical properties and dimensional technique using inkjet technology for tissue model samples”,
accuracy of 3D printed porous samples for bone tissue AATEX, Vol. 14, pp. 689-692.
engineering”, PloS One, Vol. 9 No. 9, p. e108252. Hofmeister, W., Knorovsky, G.A. and Maccallum, D.O.
Farzadi, A., Waran, V., Solati-Hashjin, M., Rahman, Z.A.A., (1999), Video Monitoring and Control of the LENS Process,
Asadi, M. and Osman, N.A.A. (2015), “Effect of layer Sandia National Labs, Albuquerque, NM, and Livermore,
printing delay on mechanical properties and dimensional CA.

666
Quality control in additive manufacturing Rapid Prototyping Journal
Hoejin Kim, Yirong Lin and Tzu-Liang Bill Tseng Volume 24 · Number 3 · 2018 · 645–669

Hollister, S.J., Flanagan, C.L., Zopf, D.A., Morrison, R.J., Kruth, J.-P. (1991), “Material incress manufacturing by rapid
Nasser, H., Patel, J.J., Ebramzadeh, E., Sangiorgio, S.N., prototyping techniques”, CIRP Annals-Manufacturing
Wheeler, M.B. and Green, G.E. (2015), “Design control for Technology, Vol. 40, pp. 603-614.
clinical translation of 3D printed modular scaffolds”, Annals Kruth, J.-P., Duflou, J., Mercelis, P., van Vaerenbergh, J.,
of Biomedical Engineering, Vol. 43, pp. 774-786. Craeghs, T. and De Keuster, J. (2007), “On-line monitoring
Hu, D. and Kovacevic, R. (2003), “Sensing, modeling and control and process control in selective laser melting and laser
for laser-based additive manufacturing”, International Journal of cutting”, Proceedings of the 5th Lane Conference, Laser Assisted
Machine Tools and Manufacture, Vol. 43, pp. 51-60. Net Shape Engineering, pp. 23-37.
Hu, X.D., Kong, F.Z. and Yao, J.H. (2011), “Development of Kulkarni, P. and Dutta, D. (1999), “Deposition strategies and
monitoring and control system for laser remanufacturing”, resulting part stiffnesses in fused deposition modeling”,
Applied Mechanics and Materials, Vols. 44/47, pp. 81-85. Journal of Manufacturing Science and Engineering, Vol. 121,
Huang, X., Ye, C., Mo, J. and Liu, H. (2009), “Slice data pp. 93-103.
based support generation algorithm for fused deposition Kullmann, C., Schirmer, N.C., Lee, M.-T., Ko, S.H., Hotz, N.,
modeling”, Tsinghua Science & Technology, Vol. 14, Grigoropoulos, C.P. and Poulikakos, D. (2012), “3D micro-
pp. 223-228. structures by piezoelectric inkjet printing of gold nanofluids”,
Huang, Y.-M. and Jiang, C.-P. (2003), “Numerical analysis of a Journal of Micromechanics and Microengineering, Vol. 22,
mask type stereolithography, pprocess using a dynamic finite- p. 055022.
element method”, The International Journal of Advanced Lan, P.-T., Chou, S.-Y., Chen, L.-L. and Gemmill, D. (1997),
Downloaded by The University of Texas at El Paso At 09:51 27 April 2018 (PT)

Manufacturing Technology, Vol. 21, pp. 649-655. “Determining fabrication orientations for rapid prototyping
Iravani-Tabrizipour, M. and Toyserkani, E. (2007), “An with stereolithography apparatus”, Computer-Aided Design,
image-based feature tracking algorithm for real-time Vol. 29, pp. 53-62.
measurement of clad height”, Machine Vision and Lee, B., Abdullah, J. and Khan, Z. (2005), “Optimization of
Applications, Vol. 18, pp. 343-354. rapid prototyping parameters for production of flexible ABS
Jiang, X.-S., Qi, L.-H., Luo, J., Huang, H. and Zhou, J.-M. object”, Journal of Materials Processing Technology, Vol. 169,
(2010), “Research on accurate droplet generation for micro- pp. 54-61.
droplet deposition manufacture”, The International Journal of Lee, H.-K. (2008), “Effects of the cladding parameters on the
Advanced Manufacturing Technology, Vol. 49, pp. 535-541. deposition efficiency in pulsed Nd: YAG laser cladding”,
Kaveh, M., Badrossamay, M., Foroozmehr, E. and Etefagh, Journal of Materials Processing Technology, Vol. 202, pp. 321-
A.H. (2015), “Optimization of the printing parameters 327.
affecting dimensional accuracy and internal cavity for Lee, K.-W., Wang, S., Fox, B.C., Ritman, E.L., Yaszemski, M.
HIPS material used in fused deposition modeling J. and Lu, L. (2007), “Poly (propylene fumarate) bone tissue
processes”, Journal of Materials Processing Technology, engineering scaffold fabrication using stereolithography:
Vol. 226, pp. 280-286. effects of resin formulations and laser parameters”,
Kechagias, J. and Iakovakis, V. (2009), “A neural network Biomacromolecules, Vol. 8 No. 4, pp. 1077-1084.
solution for LOM process performance”, The International Li, A., Zhang, Z., Wang, D. and Yang, J. (2010),
Journal of Advanced Manufacturing Technology, Vol. 43, “Optimization method to fabrication orientation of parts in
pp. 1214-1222. fused deposition modeling rapid prototyping”, 2010
Kechagias, J. (2007a), “An experimental investigation of the International Conference on Mechanic Automation and Control
surface roughness of parts produced by LOM process”, Engineering (MACE), IEEE, pp. 416-419.
Rapid Prototyping Journal, Vol. 13, pp. 17-22. Liao, Y., Chiu, L. and Chiu, Y. (2003), “A new approach of
Kechagias, J. (2007b), “Investigation of LOM process quality online waste removal process for laminated object
using design of experiments approach”, Rapid Prototyping manufacturing (LOM)”, Journal of Materials Processing
Journal, Vol. 13, pp. 316-323. Technology, Vol. 140, pp. 136-140.
Kim, H., Choi, J.-W. and Wicker, R. (2010), “Scheduling and Liu, W., Li, Y., Liu, J., Niu, X., Wang, Y. and Li, D. (2013),
process planning for multiple material stereolithography”, “Application and performance of 3D printing in
Rapid Prototyping Journal, Vol. 16, pp. 232-240. nanobiomaterials”, Journal of Nanomaterials, Vol. 2013
Kleszczynski, S., Zur Jacobsmühlen, J., Sehrt, J. and Witt, G. No. 2013, p. 13.
(2012), “Error detection in laser beam melting systems by Lott, P., Schleifenbaum, H., Meiners, W., Wissenbach, K.,
high resolution imaging”, Proceedings of the Solid Freeform Hinke, C. and Bültmann, J. (2011), “Design of an optical
Fabrication Symposium. system for the in situ process monitoring of selective laser
Ko, S.H., Chung, J., Hotz, N., Nam, K.H. and Grigoropoulos, melting (SLM)”, Physics Procedia, Vol. 12, pp. 683-690.
C.P. (2010), “Metal nanoparticle direct inkjet printing for Low, S.C. and Ake, B.E. (2004), “Thermocouple control
low-temperature 3D micro metal structure fabrication”, system for selective laser sintering part bed temperature
Journal of Micromechanics and Microengineering, Vol. 20, control”, Google Patents.
p. 125010. Lu, K. and Reynolds, W.T. (2008), “3DP process for fine
Kolossov, S., Boillat, E., Glardon, R., Fischer, P. and Locher, mesh structure printing”, Powder Technology, Vol. 187,
M. (2004), “3D FE simulation for temperature evolution in pp. 11-18.
the selective laser sintering process”, International Journal of McCullough, E.J. and Yadavalli, V.K. (2013), “Surface
Machine Tools and Manufacture, Vol. 44, pp. 117-123. modification of fused deposition modeling ABS to enable

667
Quality control in additive manufacturing Rapid Prototyping Journal
Hoejin Kim, Yirong Lin and Tzu-Liang Bill Tseng Volume 24 · Number 3 · 2018 · 645–669

rapid prototyping of biomedical microdevices”, Journal of Price, S., Cooper, K. and Chou, K. (2012), “Evaluations of
Materials Processing Technology, Vol. 213, pp. 947-954. temperature measurements by near-infrared thermography
Maruo, S. (2015), “Development of functional devices using in powder-based electron-beam additive manufacturing”,
micro/nano stereolithography and moulding techniques”, Proceedings of the Solid Freeform Fabrication Symposium,
International Polymer Science and Technology, Vol. 42, p. T7. University of Texas, Austin, TX, pp. 761-773.
Mei, J., Lovell, M.R. and Mickle, M.H. (2005), “Formulation Rao, R.V. and Rai, D.P. (2016), “Optimization of fused
and processing of novel conductive solution inks in deposition modeling process using teaching-learning-based
continuous inkjet printing of 3-D electric circuits”, IEEE optimization algorithm”, Engineering Science and Technology,
Transactions on Electronics Packaging Manufacturing, Vol. 28, an International Journal, Vol. 19, pp. 587-603.
pp. 265-273. Reeves, P. and Cobb, R. (1995), “Surface deviation modeling
Meisel, N.A., Williams, C.B. and Druschitz, A. (2012), of LMT processes–a comparative analysis”, Proceedings of the
“Lightweight metal cellular structures via indirect 3D Fifth European Conference on Rapid Prototyping and
printing and casting”, Proceedings of the International Solid Manufacturing, Helsinki, June, pp. 59-77.
Freeform Fabrication Symposium, pp. 162-176. Rengier, F., Mehndiratta, A., Von Tengg-Kobligk, H.,
Melchels, F.P., Feijen, J. and Grijpma, D.W. (2010), “A review Zechmann, C.M., Unterhinninghofen, R., Kauczor, H.-U.
on stereolithography and its applications in biomedical and Giesel, F.L. (2010), “3D printing based on imaging
engineering”, Biomaterials, Vol. 31, pp. 6121-6130. data: review of medical applications”, International Journal of
Melvin, L.S., III, Das, S. and Beaman, S. Jr (1994), “Video Computer Assisted Radiology and Surgery, Vol. 5, pp. 335-341.
Downloaded by The University of Texas at El Paso At 09:51 27 April 2018 (PT)

microscopy of selective laser sintering”, Proceedings of the Solid Sachs, E. and Vezzetti, E. (2005), “Numerical simulation of
Freeform Fabrication Symposium, DTIC Document, pp. 34-41. deposition process for a new 3DP printhead design”, Journal
Mireles, J., Ridwan, S., Morton, P.A., Hinojos, A. and Wicker, of Materials Processing Technology, Vol. 161, pp. 509-515.
R.B. (2015), “Analysis and correction of defects within parts Shaji, S. and Radhakrishnan, V. (2003), “Analysis of process
fabricated using powder bed fusion technology”, Surface parameters in surface grinding with graphite as lubricant
Topography: Metrology and Properties, Vol. 3, p. 034002. based on the Taguchi method”, Journal of Materials
Mireles, J., Terrazas, C., Medina, F., Wicker, R. and Paso, E. Processing Technology, Vol. 141, pp. 51-59.
Shishkovsky, I., Scherbakov, V., Morozov, Y., Kuznetsov, M.
(2013), “Automatic feedback control in electron beam
and Parkin, I. (2008), “Surface laser sintering of exothermic
melting using infrared thermography”, Proceedings of the Solid
powder compositions”, Journal of Thermal Analysis and
Freeform Fabrication Symposium.
Calorimetry, Vol. 91, pp. 427-436.
Mostafa, N., Syed, H.M., Igor, S. and Andrew, G. (2009), “A
Silva, D.N., De Oliveira, M.G., Meurer, E., Meurer, M.I., Da
study of melt flow analysis of an ABS-iron composite in fused
Silva, J.V.L. and Santa-Bárbara, A. (2008), “Dimensional
deposition modelling process”, Tsinghua Science &
error in selective laser sintering and 3D-printing of models
Technology, Vol. 14, pp. 29-37.
for craniomaxillary anatomy reconstruction”, Journal of
Muller, M.S. & De Jean, P.D. (2015), “3D microscopy for
Cranio-Maxillofacial Surgery, Vol. 36, pp. 443-449.
microfabrication quality control”, SPIE OPTO, International
Sitthi-Amorn, P., Ramos, J.E., Wangy, Y., Kwan, J., Lan, J.,
Society for Optics and Photonics, pp. 937607-937607-11.
Wang, W. and Matusik, W. (2015), “MultiFab: a machine
Nakamura, M. and Henmi, C. (2008), “3D Micro-fabrication
vision assisted platform for multi-material 3D printing”,
by inkjet 3D biofabrication for 3D tissue engineering”,
ACM Transactions on Graphics ( Graphics), Vol. 34 No. 4,
International Symposium on Micro-NanoMechatronics and p. 129.
Human Science, MHS 2008, IEEE, pp. 451-456. Smurov, I., Doubenskaia, M. and Zaitsev, A. (2012),
Pandey, P., Reddy, N.V. and Dhande, S. (2003a), “Real time “Complex analysis of laser cladding based on comprehensive
adaptive slicing for fused deposition modelling”, International optical diagnostics and numerical simulation”, Physics
Journal of Machine Tools and Manufacture, Vol. 43, pp. 61-71. Procedia, Vol. 39, pp. 743-752.
Pandey, P.M., Reddy, N.V. and Dhande, S.G. (2003b), Snelling, D., Blount, H., Forman, C., Ramsburg, K., Wentzel,
“Improvement of surface finish by staircase machining in A., Williams, C. and Druschitz, A. (2013), “The effects of
fused deposition modeling”, Journal of Materials Processing 3D printed molds on metal castings”, Proceedings of the Solid
Technology, Vol. 132, pp. 323-331. Freeform Fabrication Symposium, pp. 827-845.
Patel, J., Patel, C. and Patel, U. (2012), “A review on various Sonmez, F.O. and Hahn, H.T. (1998), “Thermomechanical
approach for process parameter optimization of fused analysis of the laminated object manufacturing (LOM)
deposition modeling (FDM) process and Taguchi approach process”, Rapid Prototyping Journal, Vol. 4, pp. 26-36.
for optimization”, International Journal of Engineering Sood, A.K., Ohdar, R. and Mahapatra, S. (2009), “Improving
Research and Applications, Vol. 2, pp. 361-365. dimensional accuracy of fused deposition modelling
Percoco, G., Lavecchia, F. and Galantucci, L.M. (2012), processed part using grey Taguchi method”, Materials &
“Compressive properties of FDM rapid prototypes treated Design, Vol. 30, pp. 4243-4252.
with a low cost chemical finishing”, Research Journal of Sood, A.K., Ohdar, R. and Mahapatra, S. (2010), “Parametric
Applied Sciences, Engineering and Technology, Vol. 4, appraisal of mechanical property of fused deposition
pp. 3838-3842. modelling processed parts”, Materials & Design, Vol. 31,
Pham, D., Dimov, S. and Gault, R. (1999), “Part orientation in pp. 287-295.
stereolithography”, The International Journal of Advanced Sood, A.K., Ohdar, R.K. and Mahapatra, S.S. (2012),
Manufacturing Technology, Vol. 15, pp. 674-682. “Experimental investigation and empirical modelling of

668
Quality control in additive manufacturing Rapid Prototyping Journal
Hoejin Kim, Yirong Lin and Tzu-Liang Bill Tseng Volume 24 · Number 3 · 2018 · 645–669

FDM process for compressive strength improvement”, Wang, L., Felicelli, S.D. and Craig, J.E. (2009), “Experimental
Journal of Advanced Research, Vol. 3, pp. 81-90. and numerical study of the LENS rapid fabrication process”,
Sreeram, P.D.D. (1995), “Determination of optimal Journal of Manufacturing Science and Engineering, Vol. 131,
orientation based on variable slicing thickness in layered p. 041019.
manufacturing”, ASME Winter Annual Meeting, ASME, San West, A.P., Sambu, S.P. and Rosen, D.W. (2001), “A
Francisco. process planning method for improving build
Stratasys. (2016), “Finishing processes”, available at: www. performance in stereolithography”, Computer-Aided
stratasys.com/solutions/finishing-processes/smoothing-fdm- Design, Vol. 33, pp. 65-79.
parts (accessed 16 July 2016). Wilson, W.C. and Boland, T. (2003), “Cell and organ printing
Stucker, B. (2012), “Additive manufacturing technologies: 1: protein and cell printers”, The Anatomical Record Part A:
technology introduction and business implications”, Discoveries in Molecular, Cellular, and Evolutionary Biology,
Frontiers of Engineering: Reports on Leading-Edge Engineering Vol. 272, pp. 491-496.
From the 2011 Symposium, National Academies Press, Wong, K.V. and Hernandez, A. (2012), “A review of additive
Washington, DC, pp. 19-21. manufacturing”, ISRN Mechanical Engineering, Vol. 2012
Sun, Q., Rizvi, G., Bellehumeur, C. and Gu, P. (2008), “Effect of No. 2012.
processing conditions on the bonding quality of FDM polymer Wüst, S., Müller, R. and Hofmann, S. (2011), “Controlled
filaments”, Rapid Prototyping Journal, Vol. 14, pp. 72-80. positioning of cells in biomaterials–approaches towards 3D
Tapia, G. and Elwany, A. (2014), “A review on process tissue printing”, Journal of Functional Biomaterials, Vol. 2,
Downloaded by The University of Texas at El Paso At 09:51 27 April 2018 (PT)

monitoring and control in metal-based additive pp. 119-154.


manufacturing”, Journal of Manufacturing Science and Xu, T., Gregory, C.A., Molnar, P., Cui, X., Jalota, S., Bhaduri,
Engineering, Vol. 136, p. 060801.
S.B. and Boland, T. (2006), “Viability and electrophysiology
Taylor, C. and Childs, T. (2001), “Thermal experiments in
of neural cell structures generated by the inkjet printing
direct metal laser sintering”, Proceedings of Euro RP.
method”, biomaterials, Vol. 27, pp. 3580-3588.
Thrimurthulu, K., Pandey, P.M. and Reddy, N.V. (2004),
Yadroitsev, I., Krakhmalev, P. and Yadroitsava, I. (2014),
“Optimum part deposition orientation in fused deposition
“Selective laser melting of Ti6Al4V alloy for biomedical
modeling”, International Journal of Machine Tools and
applications: temperature monitoring and microstructural
Manufacture, Vol. 44, pp. 585-594.
evolution”, Journal of Alloys and Compounds, Vol. 583,
Tourloukis, G., Stoyanov, S., Tilford, T. and Bailey, C. (2015),
“Data driven approach to quality assessment of 3D printed pp. 404-409.
electronic products”, 38th International Spring Seminar on Ye, R., Smugeresky, J.E., Zheng, B., Zhou, Y. and Lavernia, E.
Electronics Technology (ISSE), IEEE, pp. 300-305. J. (2006), “Numerical modeling of the thermal behavior
Toyserkani, E. and Khajepour, A. (2006), “A mechatronics during the LENS® process”, Materials Science and
approach to laser powder deposition process”, Mechatronics, Engineering: A, Vol. 428, pp. 47-53.
Vol. 16, pp. 631-641. Yeong, W.-Y., Chua, C.-K., Leong, K.-F., Chandrasekaran,
Uma Maheshwaraa, N., Arumaikkannu, G. and Gowri, S. (2008), M. and Lee, M.-W. (2006), “Indirect fabrication of collagen
“Three-dimensional reconstruction and rapid prototyping of scaffold based on inkjet printing technique”, Rapid
femur bone using multiple digital X-rays”, Journal of Medical Prototyping Journal, Vol. 12, pp. 229-237.
Engineering & Technology, Vol. 32, pp. 30-39. Yu, D.-G., Branford-White, C., Ma, Z.-H., Zhu, L.-M., Li,
Upcraft, S. and Fletcher, R. (2003), “The rapid prototyping X.-Y. and Yang, X.-L. (2009), “Novel drug delivery
technologies”, Assembly Automation, Vol. 23, pp. 318-330. devices for providing linear release profiles fabricated by
Vaezi, M., Chianrabutra, S., Mellor, B. and Yang, S. (2013), 3DP”, International Journal of Pharmaceutics, Vol. 370,
“Multiple material additive manufacturing–part 1: a review: pp. 160-166.
this review paper covers a decade of research on multiple Zeng, K., Pal, D. and Stucker, B. (2012), “A review of thermal
material additive manufacturing technologies which can analysis methods in laser sintering and selective laser
produce complex geometry parts with different materials”, melting”, Proceedings of Solid Freeform Fabrication Symposium,
Virtual and Physical Prototyping, Vol. 8 No. 1, pp. 19-50. Austin, TX.
Van Belle, L., Vansteenkiste, G. and Boyer, J.C. (2013), Zhou, C., Chen, Y., Yang, Z. and Khoshnevis, B. (2011),
“Investigation of residual stresses induced during the “Development of multi-material mask-image-projection-
selective laser melting process”, Key Engineering Materials, based stereolithography for the fabrication of digital
Vol. 554/557, pp. 1828-1834. materials”, Annual Solid Freeform Fabrication Symposium,
Vijay, P., Danaiah, P. and Rajesh, K. (2011), “Critical Austin, TX.
parameters effecting the rapid prototyping surface finish”,
Journal of Mechanical Engineering and Automation, Vol. 1, Corresponding author
pp. 17-20. Hoejin Kim can be contacted at: hkim4@miners.utep.edu

For instructions on how to order reprints of this article, please visit our website:
www.emeraldgrouppublishing.com/licensing/reprints.htm
Or contact us for further details: permissions@emeraldinsight.com

669

View publication stats