Download as pdf or txt
Download as pdf or txt
You are on page 1of 24

CIRP Annals - Manufacturing Technology 68 (2019) 677–700

Contents lists available at ScienceDirect

CIRP Annals - Manufacturing Technology


jou rnal homep age : ht t p: // ees .e lse vi er . com /ci r p/ def a ult . asp

Geometrical metrology for metal additive manufacturing


R.K. Leach (2)a,*, D. Bourell b, S. Carmignato (2)c, A. Donmez (2)d, N. Senin a,e,
W. Dewulf (2)f
a
Manufacturing Metrology Team, Faculty of Engineering, University of Nottingham, UK
b
Department of Mechanical Engineering, The University of Texas at Austin, USA
c
Department of Management and Engineering, University of Padova, Italy
d
National Institute of Standards and Technology, USA
e
Department of Engineering, University of Perugia, Italy
f
Department of Mechanical Engineering, KU Leuven, Belgium

A R T I C L E I N F O A B S T R A C T

Article history: The needs, requirements, and on-going and future research issues in geometrical metrology for metal
Available online 11 June 2019 additive manufacturing are addressed. The infrastructure under development for specification standards
in AM is presented, and the research on geometrical dimensioning and tolerancing for AM is reviewed.
Keywords: Post-process metrology is covered, including the measurement of surface form, texture and internal
Additive manufacturing features. In-process requirements and developments in AM are discussed along with the materials
Metrology metrology that is pertinent to geometrical measurement. Issues of traceability, including benchmarking
Surface measurement artefacts, are presented. The information in the review sections is summarized in a synthesis of current
requirements and future research topics.
© 2019 Published by Elsevier Ltd on behalf of CIRP.

1. Introduction complexity in geometries to be measured; this is why there are so


many developments in geometrical metrology for AM and why
There is little doubt that additive manufacturing (AM) will have there are still unsolved measurement issues. These developments
a large effect across manufacturing, from the home workshop to and issues are the focus of this paper.
the most technologically advanced digital factories [304]. Recent AM is still at an early stage of development. There are many
review papers describe the state of the art and future trends in the examples of consumer products using AM with plastics, for
field of AM, and this will not be repeated in depth here example, vacuum cleaner casings and even fashionable clothing.
[36,240,253,273]. In AM, objects are manufactured, not by But, if AM is to be used in earnest in high-value, advanced
removing material, but by generating the desired shape in a manufacturing, for example, in the aerospace or medical indus-
process of adding material layer by layer. This way of producing a tries, then it will be metals (and ceramics, although not addressed
shape has many benefits over the subtractive techniques, but the here) that will be the game-changers. However, right now, the
biggest benefit is the ability to produce almost any desired shape. integrity of metal (or ceramic) parts essentially made of fused
There are some caveats to this (for example, care around powders, is not equivalent to that achievable using more
overhanging features), but AM allows almost infinite design traditional subtractive manufacturing techniques [163]. AM parts
freedom, without the significant constraints on the types of made from metal powders are commonly characterised by high
geometries that can be produced using subtractive techniques. surface roughness values and can suffer from undesired material
This design freedom also applies to internal features, allowing characteristics. Also, where a part would not be manufactured with
advances in areas such as light-weighting for aerospace, internal subtractive techniques without a dimensional tolerance scheme, it
cooling channels for automotive and designed-in porosity for is still not clear exactly how to apply tolerance principles to AM
medical implants. The ability to design with such freedom is a rich parts. AM does not currently have the benefit of the over one
mathematical discipline in itself and can result in complex and hundred years of research into the production of components that
intricate geometries that could not be produced using subtractive is the hallmark of precision subtractive techniques.
techniques (see for example, Fig. 8) [270,311]. From the metrology The following reasons to measure part geometry remind us why
point of view, the extra level of design freedom offered by AM with so much effort is devoted to measuring manufactured components
respect to subtractive techniques, results in an extra level of – all of these also apply to AM, although in many cases, AM can
avoid assembly processes:
 To know whether a part is fit-for-purpose; for example, to
* Corresponding author. determine whether or not a shaft will fit within a hole, but still
E-mail address: richard.leach@nottingham.ac.uk (R.K. Leach). give enough clearance to allow the flow of lubricating fluids

https://doi.org/10.1016/j.cirp.2019.05.004
0007-8506/© 2019 Published by Elsevier Ltd on behalf of CIRP.
678 R.K. Leach et al. / CIRP Annals - Manufacturing Technology 68 (2019) 677–700

 To allow assembly of complex components; without under- was impeded because computers were not well known or
standing the dimensions of parts and their associated tolerances, widespread in society; and modern AM, denoted by the
it becomes almost impossible to determine whether a part will commercial marketing and sale of AM fabricators.
fit to another – this is an especially relevant point when Two prehistorical approaches for production of metal parts
assembling parts that have been manufactured at different involve sheet lamination and weld deposition. Blanther in
companies or different parts of a company. 1890 filed a US patent describing the creation of a mould by
 To allow control of a manufacturing process; for example, to cutting and stacking sheets of wax following topographical
change the intensity of a laser depending on the surface texture elevations [31]. As shown in Fig. 1, each sheet was cut into at
that it is producing – the texture (or something from which least two pieces, and by stacking all the pieces separately a full
texture can be inferred) needs to be measured during the two-piece mould set could the constructed. The mould was used to
production process. press paper feedstock to create “3D” topography maps.
 To avoid unnecessary scrap material and redundant processing Sheet lamination of non-metallic feedstock developed over the
time; metrology is essential for quality control which allows next sixty years (for example, [87,213,314]). In 1974, DiMatteo
things such as net-shape manufacturing – getting it right first proposed to form contours in sheets of metal using milling,
time. followed by stacking to create complex geometric parts [68]. An
 To improve energy-efficiency; the fewer repeat manufacturing example of a part built up by a “cut and stack” method is shown in
processes that are required, the lower the energy required to Fig. 2. Nakagawa at Tokyo University advanced layered production
produce a product. of tooling to include blanking, press forming, and injection
 To give customers confidence in a product; “customers” in this moulding tools [153,194,195].
context could be another manufacturing concern that needs to The use of a moving weld head to build up objects was initially
use your components – without tolerances and quality control, patented by Baker in 1925 [20]. Fig. 3 illustrates a sketch from the
there will be a lack of confidence in the assembly processes down patent. Starting in the 1960s, a number of approaches arose for
the line. producing tooling using a moving weld head, often to build parts
on a moving axis or rotating mandrel [39,40,84,300]. Fig. 4 shows
From the metrology standpoint, AM is no different to subtractive an embodiment from 1964 for creation of pressure rollers using a
manufacturing. But, a lack of integrated metrology in current AM wire-fed welding head.
machines and processes is hindering the commercialisation of the For precursor processes, the earliest is the 1972 disclosure by
resulting products [28,177,246]. The last bullet point above is Ciraud in France [56]. As shown in Fig. 5, the process contains many
especially relevant in this context. For example, an aerospace elements of directed energy deposition, including metal powder,
manufacturer is not going to “fly” a turbine blade made using AM an energy source (laser, electron beam or plasma beam), a part
without the high degree of confidence that metrology can supply. built up from a base plate and even powder recycling. The
The paper will start with a short history of AM. Section 2 will Housholder patent, issued in 1981, describes laser sintering of
discuss the infrastructure under development for specification powdered material [111], eight years before Deckard’s patent [61].
standards in AM, and Section 3 will review the research and Current AM processes for direct metal processing are divided
developments in the geometrical dimensioning and tolerancing for into three categories according to ISO/ASTM [115]: powder bed
AM. Section 4 addresses post-process metrology, including the
measurement of surface form, surface texture and internal features.
Section 5 concerns the in-process requirements and developments
in AM and Section 6 reviews the materials metrology that is pertinent
to geometrical measurement concerns. Section 7 discusses issues of
traceability and reviews the various benchmarking artefacts that are
under development. Finally, Section 8 pulls together the information
in the technical review sections into a synthesis of current
requirements and future research topics.
Note that, to keep the size of the paper to a manageable level,
there are some areas of metrology that could arguably be bracketed
in with geometrical metrology, but have not been included. This
includes the dimensional requirements of the powders and a Fig. 1. Blanther layered manufacturing, 1890s. The images are a topographical map
comprehensive treatment of the multitude of defects found in AM, showing iso-elevation lines (left) and a wax toolset and moulded paper 3D map
but see Refs. [257] and [97] respectively for recent reviews in these (right) obtained by a cut-and-stack approach [31] (Source: United States Patent and
Trademark Office, www.uspto.gov).
areas. Another important area of AM that will be increasingly
important is the combination of AM with more conventional
processes; so-called “hybrid manufacturing”. Again, for the sake of
brevity, this area has not been covered in detail, but see Ref. [132].

1.1. History

The triggering event marking the birth of AM is debatable. It is


generally associated with the commercial sale of the first
industrial fabricator, the 3D Systems stereolithography SLA-1,
in 1988 [35], although the issuing of the associated patent two
years earlier may be used [112]. The difficulty of the latter is that
there is actually a fairly lengthy patent history of AM inventions
that predates the 3D Systems founding patent for stereolitho-
graphy. According to Bourell [35], the development of AM is
categorised into three time periods: prehistory, dating to the mid-
1860s, in which objects were created in layerwise fashion without
part specific tooling and without the use of a computer;
precursors, from about 1968 to the mid-1980s, in which a
computer was used to assist in fabrication but commercialisation Fig. 2. Early sheet stacking to create a part [68].
R.K. Leach et al. / CIRP Annals - Manufacturing Technology 68 (2019) 677–700 679

fabricator based on Helisys’ Laminated Object Manufacturing


technology was in 1991 [23] (later known as sheet lamination
[162]). Ultrasonic AM was invented by White and described in a
patent that was issued in 2003 [301]. This “stack and cut” process
uses a cylindrical sonotrode to solid-state weld a metallic foil onto
a previous layer. The layer boundary is machined during the build
to produce the part. The technology was commercialised initially
by Solidica and now through Fabrisonic, a joint venture between
Solidica and the Edison Welding Institute (EWI).

2. Specification standards
Fig. 3. Weld overlay to create three-dimensional objects from a 1925 patent [20].
2.1. ISO and ASTM specification standards

Advances in metal AM machines and processes in the mid-


1990s led to direct manufacturing of functional components (as
opposed to prototypes for design validation, tooling for other
manufacturing processes, etc.), impacting major industries such as
aerospace, defence, medicine, energy, and automotive. Direct
manufacturing applications, especially for components with
critical functions, created the need for clear communication
between designer, manufacturer and the user of such compo-
nents about: (1) design intent, (2) material specifications, (3)
component test and inspection requirements, and (4) process and
component qualification and certification issues. In the last
Fig. 4. Early wire fed weld deposition for creation of pressure rollers [300].
decade, several technology roadmaps were developed to identify
the important technological barriers to widespread use of AM
[37,198,202]. A common need identified among these roadmaps is
AM-specific specification standards to introduce uniform and
consistent methods to establish requirements, and test the
resulting components, thereby reducing the risk of adopting this
technology. ASTM International is the first standards developing
organisation (SDO) recognising the need and the potential of AM
standards. The ASTM Committee F42 on Additive Manufacturing
Technologies was established in 2009 with a scope of “The
promotion of knowledge, stimulation of research and implemen-
tation of technology through the development of standards for
additive manufacturing technologies” [18]. The International
Organization for Standardization (ISO) also has a technical
committee (TC 261) with the objectives of standardising “the
process of additive manufacturing, their process chains, test
procedures, quality parameters, supply agreements, fundamen-
tals and vocabularies” [118]. ISO and ASTM signed a Partner
Standards Developing Organization agreement in 2011 to jointly
Fig. 5. Embodiment of additive manufacturing dated 1972. Metal powder (2)
develop standards related to AM technology to avoid the
delivery into an energy source/s (7/7a) to create a part (15) built from a baseplate (1)
[56]. development of multiple competing and potentially inconsistent
standards. In 2016, ASTM F42 and ISO TC 261 developed a new
fusion (PBF), directed energy deposition (DED), and sheet framework for the development of AM standards (see Fig. 6)
lamination. [18]. According to this framework, the standards are developed at
Modern laser PBF (LPBF) is based on Deckard’s invention [61] in three levels:
the late 1980s, leading to the foundation of DTM Corp. in the early
1990s. DTM was acquired by 3D Systems in 2001. The earliest
known metallic parts produced using PBF were copper-tin and
copper-(73 Pb/30Sn) solder elemental powder blends [178]. High-
energy LPBF systems were developed in the mid-1990s in Germany
and Belgium which enabled direct fabrication of parts using metals
of commercial interest [150]. These processes have become known
as “selective laser melting”, a term first used by Meiners at the
Fraunhofer Institute for Laser Technology in 1996. Electron beam-
based PBF was developed in the mid-1990s. Based on initial
development in collaboration with Chalmers University of
Technology in Gothenburg, the company Arcam was founded in
1997 with the first product launched in 2002.
DED was initially developed at Sandia National Laboratories in
the United States in the mid-1990s and was termed “laser
engineered net shaping” [131]. In DED, metal powder is sprayed
into an energy beam, typically a laser.
Modern sheet lamination AM began with Feygin who founded Fig. 6. Standards framework agreed by ASTM and ISO for future development of
Helisys in 1985, one year before 3D Systems. The first shipment of a additive manufacturing standards [18].
680 R.K. Leach et al. / CIRP Annals - Manufacturing Technology 68 (2019) 677–700

 General standards (for example, concepts and requirements  ASME B46 project team on AM – describing surface topography
common to all applications). of AM parts.
 Standards for broad categories of materials, processes/equip-  AWS D20 Inspection Task Group – inspection requirements for
ment, and finished parts. AM components based on component classification.
 Specialised standards for specific material, process or applica-  NASA-MSFC-STD-3716 [196] – accepting X-ray computed to-
tion. mography for internal feature measurements, but requiring a
“part analog reference” to be used to confirm the accuracy and
Standards for geometric specifications and measurement of AM precision of the measurements and calibration of the system and
parts may belong to all three levels of this structure depending on data processing methods.
their scope. However, to date, there are very few completed
standards addressing this area. This deficit is mostly due to the
more urgent needs of specifying materials and processes related to 2.3. New efforts to coordinate standardisation activities
this emerging technology.
The most relevant standards and guidelines developed by ISO/ In view of emerging multiple AM standards from multiple SDOs,
ASTM to date for geometric specifications and metrology are in 2016 the American National Standards Institute (ANSI) initiated
related to design for AM [116] and some general principles about an effort to coordinate standardisation activities in the US through
quality characteristics of AM parts, with corresponding appropri- the Additive Manufacturing Standardization Collaborative (AMSC).
ate test procedures [126]. While this effort is currently ongoing, a standardisation roadmap
In the design guidelines for AM, specific references to “part has been developed identifying areas for which standards are
geometry and characteristics” are made [116]. In particular, needed with associated priority classification [10,11]. According to
accuracy of part dimensions, surface texture, minimum feature this roadmap, the following medium-to-high priority standardisa-
size, maximum aspect ratio, minimum feature spacing, maximum tion gaps related to geometric specifications and metrology were
part size, and maximum unsupported feature characteristics are identified:
defined. In addition, these guidelines draw attention to the
influences of conversion of solid models into triangular mesh  Application specific design rules and guidelines (high).
models on the representation of the part geometry, causing  Design guide for surface finish capabilities (medium).
deviations from design specifications. For example, the effect of the  Technical data package addressing form/fit and qualification
triangle sizing on surface texture and accuracy and the lack of units requirements (high).
in triangle mesh models potentially causing unit-related size  Dimensioning and tolerancing requirements (high).
errors in the parts are mentioned [116]. ISO/ASTM 52,910 also  Measurement of AM features such as lattices (medium).
highlights the challenges of conveying design intent through  Protocols for image accuracy (medium).
triangular mesh models, in terms of feature tolerances, shrink  Dimensional metrology of internal features (medium).
factors, and clearance requirements. A series of warnings
associated with the effects of build orientation on the tolerances A recent roadmap published by the European Technology
and surface texture, the effects of overhang features requiring Platform in Additive Manufacturing [5] also highlighted the needs
support structures, thermal distortions caused by abrupt thickness for speedy development of harmonised standards, especially
transitions, and the unused feedstock material trapped in internal related to AM certification. A working group within this platform,
closed features are described to help design engineers. which consists of many central European industries and academia,
ISO 17296-3 [126] “specifies main quality characteristics of has been established to focus on standardisation efforts [6].
parts” fabricated by AM and “specifies appropriate test proce- In light of these requirements, and the research and develop-
dures” for these characteristics. Surface texture, size, dimensional ment needed to produce related standards, ASTM International
and geometric tolerances are among the part characteristics along with Auburn University, EWI, National Aeronautics and
specified. However, ISO 17296-3 does not provide any AM-specific Space Administration and the Manufacturing Technology Centre
test procedures; it provides relevant generic standards used for all (UK) initiated the Additive Manufacturing Center of Excellence in
manufacturing applications. For example, ISO 1302 [123] and ISO March 2018 [19]. The National Institute for Aviation Research
4288 [121] are referred to for surface texture specification and (NIAR) and Singapore’s National Additive Manufacturing Innova-
measurement, along with ISO 1101 [128] and ISO 2768-2 [119] for tion Cluster (NAIMC) have recently joined this centre as its
geometrical tolerancing. Recently, ISO/ASTM joint group issued a strategic partners. The main functions of this centre will include
draft standard providing general descriptions of benchmarking conducting AM research for efficient development of industrial AM
test piece geometries along with quantitative and qualitative standards as well as supporting education, training, proficiency
measurements to be taken to assess the performance of AM testing, and certification.
systems [117].

2.2. Other standardisation activities 3. Geometrical tolerancing

Over the last few years, several SDOs in the US have either 3.1. AM-specific challenges in geometrical dimensioning and
initiated efforts or declared interests and plans to develop AM tolerancing
related standards within their areas of expertise. Among them, the
American Society of Mechanical Engineers (ASME), SAE Interna- Determining geometric tolerances by designers and conveying
tional, and the American Welding Society (AWS) are the most them to manufacturers is essential to ensure proper functioning of
active organisations with working draft standards in progress. The products. Geometric dimensioning and tolerancing (GD&T) speci-
following ongoing efforts are most relevant to AM geometric fication standards enable the conveying of the design intent to
specifications: manufacturing process planners. GD&T is also critical for planning
the final part inspection process. The methods for such communi-
 ASME Y14.46 – standard addressing product definition require- cation are well established for traditional manufacturing processes
ments specific to AM (including geometric dimensioning and [16,120,128]. However, the high degree of influence of AM process
tolerancing). parameters on the final product characteristics and the new
 ASME Y14.41 – standard establishing a schema for organising capabilities introduced by these processes create challenges for
information in a three dimensional (3D) model within a digital conveying design intent using existing standards [8]. In addition,
product definition data set. the new capabilities of AM create further challenges associated
R.K. Leach et al. / CIRP Annals - Manufacturing Technology 68 (2019) 677–700 681

with complex, freeform internal and external features [25,38,90], for mass reduction pushes the application of these standards to
lattices, trusses, and cellular materials for generating custom their limits.
material properties and light-weighting applications The introduction of infill patterns with various cell topologies
[176,181,200], customized surface textures [105], and multi- requires new GD&T symbology, including shape, size and tolerance
material or graded-material parts [9]. of cell patterns, and transitions from infill to solid sections. In the
Since most AM processes are based on adding materials layer by case of lattice structures, definitions of representative surfaces
layer, discrete layer thickness has an effect on the geometry of the created by such structures and associated tolerance zones are
part on sloped or freeform features. This effect is also dependent on needed.
the build direction, as shown in Fig. 7. Although layer thickness Similarly, the transition regions for graded materials within the
selection is mostly dependent on the machine capability, there are AM parts have to be clearly specified with associated tolerance
studies to optimise layers for specified geometry constraints zones. Such tolerance zone specifications should take into account
[54,234]. On the other hand, the differing surface textures of the available measurement methods and related measurement
horizontal and vertical surfaces due to layering may also drive the uncertainties.
selection of build direction. Therefore, layer thickness and build
direction are critical pieces of information that need to be captured 3.2. New developments addressing these challenges
in GD&T.
Recent research efforts have focused on addressing many of the
above-mentioned challenges. Rupal and Qureshi categorized
different strategies for geometric deviation control into pre-
process, in-process, and post-process stages and related each
category to GD&T needs [254]. Moylan developed a new Product
and Manufacturing Information (PMI) for AM combining 3D
annotations of GD&T with a voxel-based volumetric representa-
tion, directly linking part design and verification by a volumetric
measuring system, such as X-ray computed tomography (XCT)
[188]. Other researchers addressed the difficulties associated with
rough AM surfaces in establishing datum and resulting geometry
evaluations, highlighting industrial needs and proposing a criteri-
on to estimate their effects [14,28]. Xiao et al. investigated the
application of the skin model concept to represent part tolerances
and tolerance analysis to AM-specific applications [13,306].
Fig. 7. The effect of layer thickness and build direction on the geometry of freeform
feature. Discretisations 1 and 2 correspond to build directions 1 and 2, ISO, ASTM and ASME standards committees responsible for AM
respectively [8]. have been dealing with the above-mentioned challenges for
several years. The first standard establishing the terminology for
The need for structures to support overhang features and coordinate systems and specifying the location and orientation of
balance the thermal conditions creates additional challenges parts within the build volume of the AM machine was developed
when performing GD&T. Removal of support structures by various by ISO/ASTM [114]. ASME standards committee Y14, Engineering
post-processing steps influences the final geometry and surface Drawing and Related Documentation Practices, has been active for
texture. Therefore, designers should carefully consider and the last few years. ASME published a “draft standard for trial use”
convey the placement of support structures around the critical (ASME Y14.46) describing product definition practices for AM
features, as well as the optimal part orientation in build volume [17]. The new ASME standard introduces: unit vectors identifying
[8,89,302]. gravity and build directions; the concept of “derived surface” from
The new capabilities of AM processes challenging the a functional collection of points, surfaces, or any combination
traditional GD&T methods include fabrication of very complex thereof (see Fig. 9); bounded and transition regions; lattice and
shapes, internal features with infill patterns (see Fig. 8), lattice gradient structures; as well as descriptions of fill patterns and
structures, parts with varying material densities or engineered support structures. Ameta et al. demonstrated the use of new
porosities. Although complex freeform features are currently concepts, theoretical supplemental surface (TSS) for geometry
handled with existing product specification standards, the shape specification and derived supplemental surface (DSS) geometry
complexities resulting from highly flexible topology optimisation verification, in a case study involving a part with a lattice infill
pattern [7].

Fig. 8. Example lattice structures in a range of lightweight part designs. (a) shows
a cross-section through an AM part with an internal repeating cellular structure,
(b) shows a tetrahedral lattice conforming to a complex shape (in this case a
component for a commercial airliner), and (c) shows a lattice structure in a
topology optimised part featuring internal routing channels for structural health Fig. 9. Rectangular lattice cuboid with a derived surface on one end with a profile of
monitoring [180]. surface tolerance of 0.5 [17].
682 R.K. Leach et al. / CIRP Annals - Manufacturing Technology 68 (2019) 677–700

4. Off-line metrology number of points on the measured surface. Principles and


characteristics of tactile probing systems are reviewed elsewhere
4.1. Surface form and coordinate metrology [298], and there has been little development of interest here since
then (mainly probing speed enhancements).
This section focuses on coordinate measuring systems (CMS) for When metal AM parts are produced with rough surface
off-line measurement of the external shape of metal AM products. textures, the highest probing performance offered by the most
Together with surface texture metrology (see Section 4.2) and precise CMMs is typically not needed [227]. However, when post-
internal feature metrology (see Section 4.3), coordinate metrology processing steps and conventional subtractive manufacturing
of the external form is fundamental for quality control of metal AM techniques are used to polish AM surfaces (for example, to achieve
products, as well as for providing feedback for AM process tight dimensional tolerances not achievable by AM alone), higher
optimisation. The eventual form errors of metal AM parts are the accuracy contact probes are more frequently needed.
result of complex interactions between the material and the Tactile CMMs are often used for reference measurements to
energy source (see Section 4.2), as well as of warpage due to study metal AM processes and their error sources (see Section 7.3).
thermal gradients and residual stress within the part [22]. The Bauza et al. [22], for example, investigated the impact of the
specific challenges that metal AM brings to coordinate metrology removal of the built part from the build platform. Their
compared to traditional manufacturing are due to the specific investigation revealed a large change in geometrical characteristics
characteristics of metal AM products, including: (i) complex (for example, flatness, as shown in Fig. 12) due to warpage of the
freeform shapes, (ii) characteristic surface texture with typically part after removal from the build platform, caused by residual
high roughness, (iii) multiple occlusions and difficult-to-access stress within the part.
features, and (iv) wide material range with different optical and Optical methods for dimensional metrology are reviewed
surface properties. All these challenging characteristics may elsewhere [104,241], which can be referred to for in-depth
contribute to measurement errors and deviations between explanation of the different measuring principles. According to
different sensors measuring the same AM surface (see Fig. 10). Se and Pears [244], optical coordinate measuring techniques can be
For example, several studies have demonstrated that AM surfaces grouped into two main types: active and passive, depending on the
with high surface roughness cause inherent systematic errors use or not of their own source of illumination in addition to
between dimensional measurements obtained from contact and ambient light. While passive sensors are typically cheaper, lighter
non-contact measuring systems [3,32,49,170,227]. The above and more compact, active sensors are generally offer better
issues also mean that it is still not clear how to obtain measurement uncertainty and speed [53]. Hence, the most
measurement traceability for CMS measurement (addressed in accurate optical sensors for measuring AM metal parts are
Section 7.3). typically active sensors. Se and Pears [244] classify active optical
To select the appropriate coordinate measuring sensor for a measuring techniques in five families, based on their accuracy and
specific metal AM product, the performance of the sensor has to be operating distance, as reported in Fig. 13.
compared to the requirements of the specific application. The Stavroulakis and Leach [264] have recently reviewed the optical
requirements in terms of profile tolerance are reported in Fig. 11 for measuring systems currently available for post-process form
different industrial applications of freeform shaped parts [236]. metrology of metal AM parts. In their review, they concluded that
Contact probing CMSs, such as tactile coordinate measuring taking into account the optically rough surfaces of industrial metal
machines (CMMs), have been used for decades in many different AM parts, the typical object size, the measurement accuracy required
industrial applications and can measure form and dimensions with
high accuracy – usually better than current non-contact CMSs.
However, they are generally slower and often sample a lower

Fig. 10. Different measurement techniques produce different measurement results:


schematic representation of (a) tactile, (b) optical and (c) X-ray computed Fig. 12. Comparison of the flatness before (left) and after (right) removal from the
tomography scanning of a surface. Red lines represent extracted points on the build platform, measured with tactile CMM on the National Institute of Standards
measured surface [52]. and Technology (NIST) AM test artefact (details on the part can be found in Section
7.2) [22].

Fig. 11. Typical tolerances of freeform parts against part dimension, for various Fig. 13. Classification of active optical measuring techniques based on their accuracy
industrial applications [236]. and operating distance [244].
R.K. Leach et al. / CIRP Annals - Manufacturing Technology 68 (2019) 677–700 683

for top-end industrial 3D measurement applications, and practical


matters such as commercial availability and cost, the most applicable
form measurement principles for metal AM have been found to be
laser triangulation and structured light projection. In the same
review, they go further in the analysis of the optical techniques
available today and suggest that if one technique were to be selected,
sinusoidal fringe projection seems to be the more favourable option
as it operates in a regime where both measurement speed and
accuracy are better suited for industrial AM form metrology. Fig. 15. Dental implant produced by direct metal laser sintering of Ti6Al4V and
Structured light (often called fringe projection) systems rely on measured using XCT: (a) picture of the threaded implant; (b) XCT three-dimensional
diffuse reflectance for their correct operation; hence, they typically reconstruction of the implant; (c) virtual sectioning of the XCT scanned volume
operate best on the optically rough surfaces that are normally [316].
produced on AM metal parts. When the surfaces become smoother
and thus produce specular reflectance, thin powder coatings can be overcome the limitations arising when measuring AM parts with small
used [133], as shown for example in Fig.14. Photogrammetry has also features, large form errors and complex surface texture.
been recently investigated for the measurement of AM components Research efforts are going in the direction of reducing the
and, with the use of suitable lighting methods, can be employed with scanning time. For example, Warnett et al. recently studied fast
both as-built and finished parts [252]. XCT reconstruction of an AM part (see Fig. 16) using a technique
With optical techniques, there can be many issues when called real-time tomography (RTT) originally developed for airport
measuring some of the complex freeform parts produced using baggage inspection [297]. This setup allowed a reduction of
AM. Currently, costly and time-consuming manual methods are scanning time for a single part by a factor of 1500 compared to a
used to determine the scanning path or lighting/sensor positions conventional XCT system. However, the smallest obtainable voxel
around a part to minimise the effects of occlusions and maximise size was 1 mm, which is still too large for conventional use in
object coverage. Recently, artificial intelligence (AI) methods have manufacturing metrology.
been applied to help solve the object coverage issues and allow
sensors to be repositioned without the need for recalibration of the
extrinsic parameters [263,264].
X-ray computed tomography (XCT) is increasingly used for
coordinate measurements of AM products, due to several advantages
with respect to tactile and optical measuring systems [274]. In
addition, XCT can simultaneously provide surface texture measure-
ments (see Section 4.2) and inspection of the geometry of internal
features (see Section 4.3), as well as quality inspection of the material
(for example, detection of undesired porosity common to AM parts). A
number of reviews go through the applications of XCT for AM
[55,214,272]. More details on the state of the art of XCT for AM are given
in Section 4.3, while here in Section 4.1 only specific aspects of XCT
used for form and coordinate metrology of AM parts are covered.
Compared to tactile and optical coordinate measuring techniques, XCT Fig. 16. 3D reconstructed volume of an AM part measured by: (a) reference lab XCT
allows dense coordinate measurements to be obtained from an holistic metrology system and (b) fast method applied to an airport XCT scanner. The
volumetric model of the scanned object. XCTalso offers advantages if a scanning time was 1570s for the first case and only 1 s for the second case. The
part includes complex freeform shapes, surfaces with high roughness, influence on reconstruction accuracy of the fast scan is clearly visible [297].
difficult-to-access features, or parts with different optical and surface
properties [237]. Therefore, for coordinate metrology of AM parts, XCT Uncertainties in XCT coordinate measurements are affected by a
is sometimes preferred to tactile and optical probing methods, even large number of influence factors [51]. In particular, when scanning
when measuring external surfaces only. For example, Fig.15 illustrates metal parts, dimensional and form measurements are strongly
the external thread measurement of a dental implant produced by influenced by the material and its penetrated thickness, which
laser sintering of titanium alloy (Ti6Al4V) [316], where XCT was directly affect the X-ray attenuation [51]. To start to address this
preferred to conventional thread measuring techniques [50], to issue when measuring AM parts, Shah et al. [248] designed a test
part for benchmarking different AM processes, choosing a
cylindrical shape to reduce the influence of X-ray imaging artefacts
(see Fig. 17). In addition, other significant influence factors are for

Fig. 17. XCT measurement results on three parts produced with the same nominal
geometry (test artefact) using three different AM processes. The features of interest
Fig. 14. Aluminium alloy (AlSi12) part produced by laser powder bed fusion and are all accessible by tactile CMM for reference measurements [248]. In this figure,
measured by a structured light system. Due to glossy surface, a thin layer of anti- SLA is stereolithography, SLS is selective laser sintering and FDM is fused deposition
reflection coating with titanium powder was applied for measuring this part [133]. modelling.
684 R.K. Leach et al. / CIRP Annals - Manufacturing Technology 68 (2019) 677–700

example beam hardening and scattering [147]. Also the methods effect refers to the presence of visibly offset layers of the build
used for thresholding and segmentation within the 3D voxel model process that approximate the 3D model which changes with build
(see Section 4.2 and 4.3) have a large influence on the accuracy of orientation, which has been investigated thoroughly in the literature
surface determination and geometrical measurements [147]. Fur- for metal PBF [34,79,100,250,269,284]. These offset layers are
thermore, when using XCT for coordinate metrology of AM parts, it present as the larger scale wave-like components and are most
should be considered that big-data need to be processed, as the visibly offset at lower surface slopes, with the adhered powder
size of typical XCT datasets, including X-ray projections and becoming more dominant at steeper slope angles [34]. The surface
reconstructed volumes, can easily exceed several gigabytes topography of side surfaces is mostly dominated by the sphere-like
[51]. Further details on capabilities and limitations of XCT for protrusions made from adhered powder particles from the
scanning AM parts are given in Section 4.3. surrounding powder bed. These particles are protrusions on the
side surfaces and increase in number as the build angle increases,
4.2. Surface texture metrology with upwards facing surfaces possessing less particle coverage than
downwards facing surfaces, and bottom surfaces consisting almost
Metal AM surfaces are often highly complex and irregular (as completely of particles [79,102,142,143,284]; this is the result of
shown in the example scanning electron and optical microscope multiple melting and re-melting phenomena between the build
images in Fig. 18), and texture measurements can show significant layer and layers underneath, and energy conducting through the part
discrepancies between measurement technologies, as demon- into the surrounding powder bed. In LPBF the excess energy from the
strated by Thompson et al. [274] for the PBF family of metal scanning strategy acts to sinter loose adjacent powder around the
additive processes. In metal PBF, the eventual surface topography build geometry [74], whilst for EBPBF the build process pre-sinters a
of a part is the result of complex interactions between the larger region around the build into a ‘cake’ prior to melting the layer
feedstock material and the energy source scanning over the [97]. Defects that can be present on the surface topography of side
current build layer (typically a laser or electron beam). This surfaces include pores and thermal cracks found at smaller scale
interaction involves the formation of melt pools, the ejection of [280], with large scale defects including large recesses due to de-
particles and significant thermal gradients [138,183]. Surface lamination [97] and remnants of the build supports that leave large
topography is typically dominated by weld tracks, resulting from protrusions [280].
the fusion and subsequent solidification of a melt pool [184], which There have been limited attempts to study the deterministic
impart a strong texture directionality indicative of the laser or production of specific topographic features [1,309,310] and
electron beam path [206,247]. At smaller scales, weld tracks are microstructure [231,271], although such studies are problematic
covered by chevron-shaped ripples, indicating the beam scanning due to the need to effectively gain access to the machine control
direction, and may feature smaller-scale thermal cracks and areas software. There have also been some limited studies to try to link
of local oxidisation [109]. Throughout weld tracks, high aspect- theoretical predictions of topography to experimental trials
ratio singularities are observable, typically consisting of deep [80,285,296], but more research is needed in this area to fully
recesses or sphere-like protrusions. Recesses may result from understand the complex interaction between texture and
incomplete seams between weld tracks, balling phenomena (i.e. processing parameters [59,110,221,222,235]. There has been
discontinuities of the track itself) or, at smaller scales, open micro- some recent research on the development of dedicated ‘home-
porosity [95,251,278]. Sphere-like protrusions are formed either made’ AM systems [30,155], but it remains to be proven that such
from unmelted or partially-melted powder particles (appearing systems are representative of the types of surfaces being
alone or in clusters), from spatter particles, i.e. molten material produced by commercial AM systems, which are being developed
ejected from the melt pool during beam traversal, that impact the at a rapid pace.
nearby surface during solidification [172,251] or from balling, Townsend et al. [282] have produced a comprehensive review
especially due to insufficient heat in overhang areas [95]. As the of the measurement of AM surface topography and the key points
metal PBF surface is the result of multiple melting and re-melting are summarised here. Initial research implied the literature is
phenomena, involving the current layer as well as a variable limited and research is at an early stage; research applied to
number of layers underneath, multiple, larger-scale, wave-like process understanding with no introduction of application-specific
components affect the final appearance of the top surface [283] requirements. Ti6Al4V was the most studied material, while LPBF
(just the weld tracks are visible in Fig. 18 but there may also be was the most used process. Contact stylus was found to be by far
longer wavelength components due the effects of the recoating the most common measurement method, coupled with the use of
and spreading process). This partially random, partially determin- profile parameters.
istic topography usually forms a recognisable pattern, i.e. the The metal PBF topography is a challenge for topography
“fingerprint” of the technology [79]. Such top-layer (topographic) measurement. High slopes, variable aspect-ratios, the alternation
features are often referred to as “defects” in the literature, and of dark (for example, deep recesses) and overly bright (for example,
there are a number of reviews that list and qualify them tops of smoother regions of particles and weld tracks) regions, as
[74,97,261]. well as non-uniform optical properties caused by local oxidisation
For the side surfaces produced by metal PBF, the topography is and/or micro-roughness effects, are the main issues for optical
dominated mostly by the effects of local surface angle, the bonding measurement technologies [77,157,247]. It has been recently
between layers and the consolidated part, and the interaction of the demonstrated [45,100,274] that areal topography instruments
part and surrounding powder bed [102,250,284]. The “stair-case” quantifiably react differently to metal PBF topographies, leading to
reconstructions of measured surfaces where measurement error is
of the same order of magnitude as the size of localised topographic
features captured by the measurement process. There have been a
number of recent studies to pinpoint the optimum conditions and
setup parameters that should be used for optical instruments, for
example, those based on coherence scanning interferometry (CSI)
[94] and focus variation [199]. XCT measurement is subject to an
equally complex series of non-optical challenges [276], which
affect the spatial resolution of the measurement, as well as
Fig. 18. (a) and (c) Optical microscope and (b) scanning electron microscope (SEM)
accurate surface determination [279] (see below for more on XCT
images of topographic features typical of a Ti6Al4V metal laser PBF surface. Arrows
1a and 1b point to the same example surface recess in optical and SEM images surface measurement). Vetterli et al. [290] investigated the use of
respectively, while arrow 2 indicates local cracking in a higher resolution image an elastomer pad with a reflective coating (‘Gelsight’) combined
[247]. with an optical sensor for measuring LPBF surfaces, and obtained
R.K. Leach et al. / CIRP Annals - Manufacturing Technology 68 (2019) 677–700 685

some comparative results with contact and optical methods, albeit can provide surface topography measurements for metal AM
with lateral resolution and surface slope limitations. surfaces with a wide range of surface features, and recommenda-
Most reported studies focus on exploring how changes in tions for optimisation of future measurements on metal AM
reconstructed topography ultimately affect the results of the surfaces.
computation of texture parameters (ISO 4287 [122] profile X-ray computed tomography (XCT) is discussed as a coordinate
parameters and ISO 25,178-2 [125] areal parameters). An approach measuring tool in Section 4.1 and for measuring internal features in
dedicated to computing texture parameters is justified as the Section 4.3. Although the spatial resolutions typically achievable
majority of surface characterisation tolerances in industrial by XCT have not historically been at the level required to capture
applications are currently addressed via texture specifications; the smaller-scale formations of a surface in addition to the overall
however, when the interest is in localised surface formations shape, advanced systems are approaching these resolutions in
(surface features), then the investigation of how local topography their best-case measurement scenarios, and XCT is becoming a
is reconstructed through different measurement technologies is of viable option for measurement of surface topography [272]. When
fundamental importance. Senin et al. [247] discuss the require- considering the fact that AM parts commonly feature complex,
ment for an understanding of the manufacturing process finger- internal geometries, the prospect of using XCT for surface
print in metal AM processes. Using a Ti6Al4V test surface, the topography measurement appeals further, as a method of
authors compare measurements of problematic features on the overcoming the access requirement problems intrinsic to contact
metal AM surface, particularly: a large measured area, approxi- and optical measurements. The use of XCT for this purpose is
mately spherical particles attached to the surface, surface recesses, highlighted in a number of recent studies [135,216,275,279]. Spe-
high spatial frequency weld ripples, and weld tracks (encompass- cifically, Pyka et al. [135,216] performed the first surface
ing several of the other features). To assess these features, the topography measurement using XCT, by extracting profiles from
authors first measure using a number of optical and non-optical slice data obtained from XCT measurement of lattice struts.
methods, and then finely align topographic reconstructions in Townsend et al. [279,280] and Thompson et al. [274] extended this
space, assessing discrepancy between measurements for these work by initiating a more extensive examination of XCT
features. Results show high slopes and high aspect ratio features topography measurement performance in comparison to conven-
playing a dominant role in measurement discrepancy, and they tional optical surface measurements. It has also been shown that
find that quantitative comparison of topography data obtained by surface texture analysis for internal surfaces does not differ
using different measurement technologies shows that disagree- significantly from that of external surfaces, at least for the types of
ment between reconstructions can be significant, and that for surfaces found in PBF processes [275]. As XCT is a relatively
small-scale features (tens of micrometres), local discrepancy complex measurement process, it is important to establish the
between measurements can be of similar magnitude the size of sensitivity of the various measurement parameters and influence
the feature (see Fig. 19). For larger-scale features (on the order of factors to surface determination (especially sampling and resolu-
hundreds of micrometres to millimetres), all measurement results tion), and there have been a number of recent studies to establish
are consistent. Finally, Senin et al. state that no particular these relationships [275,276,318]. The rendered surface topogra-
technology should be considered inferior or superior to another, phy from an XCT measurement is found to be highly dependent on
as results can be heavily dependent on specific instruments and the surface/edge determination algorithms employed
setups. [49,166,167,227,239] and this is expected to be a major factor
There has been relatively little detailed work on the response of when investigating the types of surfaces found in metal AM. A
measurement technologies to surface features produced using number of methods exist for determining surfaces in XCT data (for
EBPBF processes, but Sidambe [250] compared a CSI-based example, user-determined thresholding, ISO-50 thresholding
instrument with a 3D scanning electron microscope and showed [147], gradient-based [69]). Different methods produce different
large differences in the results when comparing a range of ISO results, providing alternative representations of the surface with
25178-2 [125] areal surface texture parameters. Gomez et al. [94] respect to the position of all parts of the surface in 3D space. For
presented an empirical sensitivity analysis using a different CSI example, individual features on the surface as determined using
system for the top and side surfaces made from metal LPBF and one method could be eroded or dilated compared to that
EBPBF processes. In the work of Gomez et al., topographic determined using another method, or whole surfaces offset from
measurements were described through the use of ISO 25178-2 one another in some way. Lateral and vertical resolution are also
[125] areal surface texture parameters Sq and Sdq and were affected by the surface determination method used, with some
analysed for data coverage, measurement time, and area. The methods and implementations of those methods sampling at
results show that modern implementations of the CSI technique higher rates than others.
Artefacts for benchmarking form and dimensional accuracy are
detailed in Section 7, but there have been a small number of
attempts to include surface texture-specific features [192]. An
artefact suite, proposed by the ASTM F42/ISO TC 261 includes user-
selectable application-specific artefacts, which include a series of
patterns to be built over a range of angles to the build plate
[188]. Townsend [281] has developed a range of artefacts that can
be built using any AM technology and which include a significant
range of feature types based on determining the metrological
characteristics under development in ISO TC 213 [158], see
Fig. 20. Such benchmarking artefacts can help the user of an
instrument to elucidate the effect of various AM features and
textures on a measurement result and may in some cases be used
to determine metrological characteristics.
Surface texture characterisation is conventionally carried out
using surface texture parameters. Whilst there are many hundreds
of parameters in use, by far the most common are those from ISO
4287 [122] in the profile case and ISO 25178-2 [125] in the areal
Fig. 19. Reconstruction of large measured areas using four different instruments: (a)
surface case (see Ref. [277] for a recent review of parameter usage).
a confocal microscope, (b) a coherence scanning interferometer, (c) a focus variation However, surface texture parameters (at least field parameters)
microscope, and (d) an XCT system [247]. were originally developed to characterise random or near-random
686 R.K. Leach et al. / CIRP Annals - Manufacturing Technology 68 (2019) 677–700

location). Early studies on how to mathematically represent full-


3D topography and how to define new texture parameters based
on such representations are in progress, for example, by using
parametric surfaces and profiles [207].
Research into the functional behaviour of AM surfaces has been
limited due to the frequent requirement for post-process finishing
– in this case it is not the AM surface that is functional, but the
finished surface. There have been some developments in
Fig. 20. Surface texture artefacts developed by University of Huddersfield [281].
correlating surface topography to mechanical properties; often
simple texture field parameters are not sufficient and feature
machined surfaces. AM surfaces tend to be nothing like this and are parameters are used, for example, linking topography to strength
a mixture of deterministic patterns (for example, weld tracks) and and elastic properties [101], fatigue performance [98,144] or bio-
random features (for example, spattered particles). It is not clear compatibility [299]. Several groups have used topography analysis
whether the more conventional parameters can be applied to to investigate the mechanical properties of lattice structures
typical AM surfaces. However, many groups have investigated the [182,216]. The effects on AM surfaces caused by various finishing
use of texture parameters and this work is summarised in processes have also been investigated, using shot-peening [174],
Townsend et al. [282] (and see Refs. [45,100,250,274,317]). There laser polishing and re-melting [4,154,173,286,312], and abrasive
may also be issues when using default ISO filter settings (for processes [24,169].
example, to set the cut-off length) with AM surfaces, although it
has been shown that multi-scale approaches can be applied to aid 4.3. Internal feature metrology
the selection of filters [171,284].
As may have been expected, Townsend et al. [282] report that As AM progressively enters the application domain of high-
Ra was the most common texture parameter, followed by Rq, Rz, quality functional components, the metrology challenges related
and Rt. Regarding areal parameters (used in 20% of the examined to internal and inaccessible features are also increasing [55]. Exam-
literature), Sa was the most common. The authors comment on the ples of internal features include conformal cooling channels, for
advantages of areal over profile, and mention that NIST recom- which traditional GD&T symbology can already be applied, but this
mend the use of Ra/Sa, Rz/Sz, Rsk/Ssk, and Rku/Sku for AM surface is not the case for the infill patterns and lattice structures for which
characterisation. The authors note that filtering for parameters is new GD&T symbology is still under development (see Section 3.1).
often not reported in the literature and that measurements are Moreover, since the process variability of AM has not reached the
rarely reproducible as a result. Several people, therefore, prefer to levels of subtractive techniques, inspection for internal material
use the less filtered profile parameters Pa instead of Ra. Parameters defects, such as cracks and pores remains essential [272].
can also be applied to differentiate between top and side/angled X-ray computed tomography (XCT) is currently the only
surfaces [100,250,282]. available technique that is able to non-destructively inspect both
Potentially more fruitful methods to define surface texture material and dimensional quality of internal structures [147],
parameters that capture process or function information about AM although there has been some research into the measurement of
processes, may be found from the less conventional approaches, buried but near-surface defects, for example using ultrasonics [73],
and there has been limited work on multi-scale analysis spatially-resolved acoustic spectroscopy [258], and flash thermog-
[41,45,217,218]. Quinsat et al. use a multi-scale discriminative raphy [320]. XCT relies on the fact that X-rays are only partially
method to analyse the relative area of the pores and pore attenuated while traversing a material [48,51] (Fig. 21). Therefore,
distributions. a well-selected set of radiographic images taken from multiple
An alternative set of parameters that can be used for surface perspectives contain the data necessary to reconstruct a band-
texture characterisation is feature parameters. First introduced by width-limited 3D model of a workpiece, comprising both internal
Scott [243] as a complement to texture field parameters, feature and external features, and even material defects. Subsequent
parameters describe a surface topography in terms of the thresholding and segmentation within the 3D greyscale voxel
properties of the peaks and valleys that populate it (number, model allows the distinguishing of different materials and surface
area, height/depth, etc.). A full list of currently recognised feature determination (see Section 4.2), hence geometrical inspection.
parameters is reported in the standard ISO 25178-2 [125]. Current-
ly, a thorough investigation of the potential of using feature
parameters to describe the topography of metal AM surfaces has
not been carried out. A discussion on the subject is reported in the
review on surface texture characterisation for AM surfaces by
Townsend et al. [282]. In recent work by Lou et al. [171] on the
extraction of signature features from a LPBF surface, the use of
feature parameters is not discussed; however, feature identifica-
tion is carried out by using methods prescribed by ISO 25178-2
[125] for the computation of feature parameters, i.e. morphologic Fig. 21. Principle of X-ray computed tomography [48].
segmentation into hills and dales and change tree pruning
[243]. The authors also report on a method to determine the The metrological capabilities of XCT still need to be enhanced,
nesting indices of filters when using areal parameters to since a full traceability chain has not yet been established and the
characterise AM surfaces. influence of a large number of parameters and settings has not yet
Additional research is needed to mathematically capture the re- been fully unravelled [66,288]. Nevertheless, significant progress
entrant nature of many topographic features that are typically has been made over the last decade in this respect.
appearing on AM surfaces. Optical and contact areal topography A series of international interlaboratory comparisons have
measurement technologies have limited possibility of capturing provided useful insights into the overall performance of XCT
undercut topography; however, the emerging use of XCT as a viable systems, yet have confirmed the difficulty of determining adequate
option to capture surface topography has shown that more measurement uncertainties due to the multitude of influence
comprehensive, full-3D measurements of surface topography are factors [12,47,268]. Operator settings, such as X-ray source current
possible [274,279]. In such a scenario, conventional surface texture and acceleration voltage, number of projections, integration time,
parameters may not suffice, as they are limited to the characteri- and workpiece orientation, heavily influence XCT measurement
sation of height maps (only one z value possible at each x, y results [12,140,291]. Recently, methods have been proposed to
R.K. Leach et al. / CIRP Annals - Manufacturing Technology 68 (2019) 677–700 687

autonomously optimise settings for a given workpiece geometry the pore size that can be reliably detected, and is influenced by the
[44] (see also Section 4.2 which refers to optimisation of sampling focal spot size, the magnification (hence workpiece size), noise
and resolution settings for XCT AM surface measurement). (hence also by workpiece material), and many other factors
Workpiece dimensions influence the XCT measurement uncer- [136,315]. XCT is, therefore, reported to yield slightly lower
tainties by determining the allowable magnification as well as the porosity values compared with the Archimedes method
penetration lengths, hence device settings. Moreover, surface [256,262,303]. Moreover, voxel size errors and thresholding both
texture directly influences surface determination, see Section influence the measured size of the pores, hence of the overall
4.2. Research has also been dedicated to understanding the effect porosity value. Different authors have tested thresholding
of deviations in the XCT instrument kinematics [78]. Kinematic techniques on various porous materials in order to achieve optimal
deviations can be detected by conventional reference measuring correspondence with reference techniques [113,134]. Comparison
instruments or by analysing radiographs or reconstructions of with references, however, remains challenging, due to the
dedicated reference objects, followed by physical adjustment or difficulty for traceability (see Section 7.3). Therefore, Hermanek
software compensation [64,140,193]. Over the past few years, a et al. have proposed a fully dismountable reference artefact with
multitude of artefacts have been proposed for XCT system small yet well-controlled internal cavities [108].
verification [51]. They often rely on features that are simple to Although XCT based porosity measurements typically deviate
measure with low uncertainty, such as spheres, calottes, cylinders from other reference methods, a carefully designed XCT measure-
or gauges blocks. It has been argued that standardised artefacts ment protocol can ensure that pore sizes, with a significant
that are manufacturable by AM and optimised for XCT measure- contribution to the overall porosity [210] and which are above the
ments would be expedient (see Section 7.2) and an example can be critical size for fatigue [27,232], exceed the structural resolution,
found elsewhere [211]. In addition, the structural resolution of XCT hence allowing reliable relative measurements.
systems is a key characteristic. The VDI/VDE 2630-1.3 [289] Several authors have employed XCT to develop an increased
guideline proposes verification of the structural resolution by understanding of the influence of AM parameters on the workpiece
identifying the diameter of the smallest sphere that can be quality. Parameters assessed include the employed laser power
measured with a deviation compared to the calibrated value that and scan speed [65,88,233], scanning strategy [65], build direction
remains below a specified error. Carmignato et al. [51] propose [159], powder properties [71] and powder recycling [88].
alternative ways to determine the “metrological structural
resolution”, which is of particular relevance for porosity measure- 5. Materials properties affecting AM part geometry
ments and internal feature metrology.
Despite the challenges, the adequacy of XCT for comparative The main effect of materials on final part shape stems from part
measurements of AM parts can already be confirmed, as shown by distortion during the build, and, for post-processed parts, after the
Bauza et al. [22], who observed that the repeatability of XCT build. In both cases, the distortion arises from residual stress
measurements of AM lattice structures (around 4 mm variability) formation and its relief. Residual stress is directly related to the
was well below the between-parts deviations caused by the AM thermally induced strain, which itself is proportional to a DT,
and post-processing steps (Fig. 22). Next to dimensional inspec- where a is the coefficient of thermal expansion and DT is the
tion, recent research efforts are also targeting the use of XCT for change in temperature. Excepting tungsten and lead, a is relatively
surface topography measurements, as described in Section 4.2. In invariant for metals, changing by only a factor of two for all metals,
addition, XCT can simultaneously provide dimensional and and its effect on distortion in AM is small [293]. There are several
porosity measurements. materials properties that impact the temperature gradient:
thermal diffusivity and specific volume change on melting.
Excepting precious metals, the thermal diffusivity varies among
metals by a factor of twenty, and it was shown not to correlate
strongly to distortion in metallic AM [293]. Finally, several
mechanical properties have an effect on the residual stress. Low
modulus materials generally exhibit low residual stress, although
the nature of the thermal field and the base plate upon which the
part is built also have an impact on the development of residual
stress [185]. If the residual stress exceeds the yield strength, plastic
deformation will take place. In extreme cases, the part or weaker
supports may fracture.
In a study of the effects of nine different metals on residual
stress development in AM [292,293], it was concluded that internal
cracking at weak interfaces and oxidation affected part residual
stress to a greater degree than the intrinsic material properties.
Fig. 22. Octet lattice test artefact with edge length of 14 mm. Between-part Residual stress development in metallic AM is characterised by
variability of the PBF process determined by the direct comparison of XCT large, in-build-plane tensile residual stress at the top of the build
measurements of two parts [22].
and compressive residual stress in the centre of the part
[164,185,219,249,292,293,295], as shown for example in
Industry standards for porosity measurements of AM parts Fig. 23. Residual stress diminishes as the scan area during the
often rely on the Archimedes method, which has been shown to be build increases [308], but it has also been reported that the residual
reliable, and pycnometry provides an alternative, yet assesses only stress increases with build height for certain alloys; true for
one or a few separate cross-cuts. Mercury porosimetry allows stainless steel 316L but not for Ti-6Al-4V [308]. For thin-walled
determination of the pore size distribution, but only for open parts, residual stress tends to be compressive in the centre and
pores. XCT non-destructively provides additional information, tensile on the edges [60,165,287,319]. Distortion is manifested by a
such as pore size distributions for both open and closed pores, as buckling phenomenon in which the centre bulges, Fig. 24. In some
well as pore morphologies and local porosity information cases, the distortion is a unidirectional bowing curvature in the
throughout the entire volume [137,179]. direction of the scan line [60,292].
As with geometric metrology, XCT based porosity measure- Residual stress and distortion effects have been characterised
ments are influenced by a number of factors, including the sample and in some cases modelled for the following metallic systems: Al
attributes (size, geometry, material, etc.), the XCT scan settings, and [43,164,293], stainless steel [2,185,287,293,308], low allow steel
the post processing procedures. The spatial resolution of XCT limits [249], maraging steel [293], tool steel [319], Fe–Ni–Cu–P [165], Ni
688 R.K. Leach et al. / CIRP Annals - Manufacturing Technology 68 (2019) 677–700

[2,60,249,259,292]. Beuth concluded that the process effect on


residual stress was universally defined by the effect on the
temperature gradient in the build direction [2]. Spiral and random
scan patterns result in lower residual stress than raster scanning
[60,185]. Kruth’s group has proposed a materials solution which is
to pursue alloy development for AM resulting in tough, crack-
resistant alloys [292], coupled to high heat input, thin layers, slow
scan speed and high power.
Residual stress has been relieved by internal cracking during
AM [228,229,292,293]. While this is not acceptable for structural
parts, it does result in parts with reduced distortion from residual
stress [292,293]. Cracking has multiple origins in metallic AM. Off-
eutectic metals solidify over a large range of temperature with
slushy zones prone to crack. In AM, this effect may be extended
across multiple layers as new feedstock is deposited, Fig. 25,
resulting in directional cracks parallel to the build direction. Prior
particle boundaries weakened, for example, by oxide formation,
are a source of delamination in the presence of residual stress
[228].
Fig. 23. Simulated and measured residual stress (neutron diffraction) as a function
of distance from the build plate for a 7 mm thick walled build in Inconel 625 using
directed energy deposition [295].

Fig. 25. Cracking in laser powder bed fusion processed Al. The build direction is
vertical. 6061 Al prealloyed feedstock, showing vertical cracks formed during
cooling through a large melt range and extended as additional layers were
Fig. 24. Simulated and measured normalized distortion in a 0.3 mm thin walled deposited [229].
plate built in an Fe-30Ni-10Cu-15 P alloy using powder bed fusion (laser) [165],
showing exaggerated bowing.
The extent of deformation due to process residual stress
[60,219,293–295], refractory metals (Ta and W) [293], and Ti development in AM metals has been assessed in a number of
[160,245,293,294]. The magnitude of the residual stress is also AM investigations [149,148,164,165,185,201,208,215,313]. There is
process dependent. In general, residual stress is highest for LPBF presently no standard method for assessing and reporting
[43,160,308,164,165,185,245,249,292–294], followed in decreasing distortion in AM metal parts. One “worst case” approach to
order by DED [2,60,219,287,295,319], electron beam PBF [294], and assessing distortion is to add one or more layers of metal onto a
sheet lamination. thin support-free substrate while monitoring the in-build distor-
For the most part, distortion in metallic AM is controlled using tion [165]. A single 0.15 mm layer of 63Fe-20Ni-15Cu 2.3 P powder
process modifications. Residual stress in as-built metallic AM parts was LPBF deposited onto a 1 mm thick rectangular sheet of steel
typically ranges from 80% to 100% of the yield strength (22 mm by 45 mm) [165]. The maximum in-build distortion of the
[160,249,287,294,308]. Currently, parts are built on rigid base- layer-plate assemblage was 0.35 mm, which gave a maximum
plates using extensive support structures to minimise distortion. distortion of 0.35/22 = 0.016 mm/mm. A “worst case” approach
The baseplate/support structure/part assemblage is post-pro- incorporating support structure is to build a long cantilever using
cessed by heat treatment to relieve the stresses prior to removal supports. The distortion of the cantilever is measured after
of the part from the support structure [148,292]. Preheating the removing the supports. One example [164] shows a maximum
baseplate [43,83,185,287] or the entire part during the build distortion of 0.038 mm/mm (2.1 mm maximum distortion) for a
[2,43,165] has significant but lesser effects on residual stress 55 mm long cantilever (width 6 mm, thickness 4 mm) of LPBF
reduction compared to post-process heat treatment. Buchbinder processed AlSi10Mg.
reported that heating an AlSi10Mg LPBF part to 250  C during the Distortion is also present in AM parts produced indirectly; that
build eliminated all residual stress [43]. Localised pre- and post- is, using a transient binder which is removed or transformed in a
heating of the melt zone during the build has a limited effect on post-processing step. Stevinson and Bourell [265,266,267] studied
reducing part residual stress [2]. Slow cooling [259] and laser shot support-free builds and post-processing of a silicon carbide
peening [245] also reduce the residual stress. Double scanning material with a phenolic transient binder using LPBF. The binder
each layer without adding feedstock reduces residual stress [249], was reduced to carbon in a post-processing step, followed by part
but it also increases the build time significantly. Energy beam scan infiltration with molten silicon. Green parts immediately after the
parameters have an effect on residual stress formation. Velocity, build showed uniform linear expansion of 2.4% with slightly less
scan speed, hatch spacing and the scan pattern have been assessed expansion; 2.0%, along the build direction. The part shrunk as the
R.K. Leach et al. / CIRP Annals - Manufacturing Technology 68 (2019) 677–700 689

binder was dissociated during post-processing, resulting in an targets and metrological requirements is deeply tied to the current
overall slight expansion in the lateral directions, 0.8%, while the understanding of what variables are relevant in determining the
height direction shrunk, -0.2%. All distortions were measured insurgence of relevant drifts that ultimately make the fabricated
relative to the computer solid model dimensions. The final part part non-conforming to the specification requirements [74,97]. Al-
after silicon infiltration exhibited an overall volume expansion of though such knowledge is intimately tied to process understand-
less than 2% compared to the computer solid model. ing, a series of process-related variables is already commonly
accepted as relevant, and is consequently targeted in in-situ
measurement for monitoring. Typical targets considered in all the
6. On-machine metrology major metal AM processes (LPBF, EBPBF, powder-DED and wire-
DED) are:
6.1. Types of metrology solutions and their role in metal AM
 the melt pool: properties such as emitted energy (intensity
On-machine metrology investigates measurement solutions or frequency spectrum), size and shape
dedicated to observing the manufacturing process or the part itself [15,29,33,62,139,141,145,186,260,305];
executed within an AM machine (see Ref. [85] for recent definitions  the topography of the layer (or individual track) after processing,
of the terminology associated with on-machine metrology). or any other information pertaining the state of the layer (for
Measurement may be used to acquire information while the example, temperature) [21,33,67,82,92,230,307,321].
process is being executed (in-process metrology) either by
inspecting the part itself as it is being fabricated, or by observing For LPBF and EBPBF, additional targets pertain to:
other process-related variables of interest within the machine. In-
process measurement solutions may be developed with direct  the powder bed before processing [21,58,129,168];
access to where the actual physical phenomena involved in the  spatter particles during processing: size, number, trajectories,
transformation process are taking place. For metal AM, except for temperature, and distance of travel from the ejection point [224].
those processes operating in an open environment (usually wire-
DED), this type of metrology typically considers events taking An increasing number of authors is acknowledging the
place in the build chamber (LPBF, EBPBF, and powder-DED). importance of acquiring information from multiple targets at
Comprehensive reviews on recent research pertaining to on- the same time [62,139,307]; several are also considering the sensed
machine and in-process metrology for metal AM can be found in information in combination with additional process parameters
the works by Everton et al. [74] and Mani et al. [177]. recorded from the AM machines, for example, related to the energy
In metal AM, in-process measurement is currently implemen- source or atmosphere in the build chamber [307].
ted with two primary purposes: to understand the physics of the
process, and to implement process monitoring. In-process 6.3. In-process measurement technologies and strategies
measurement for process understanding typically requires cus-
tom-built manufacturing stations, designed to accommodate the In-process measurement for process understanding often
sophisticated measurement solutions needed to observe the implies the development of prototype AM rigs explicitly designed
physical phenomena that take place during transformation. For to be compatible with measurement solutions [161]. Open and
example, high-speed synchrotron X-ray imaging combined to a custom-built AM machine architectures also favour more control
custom-built rig for LPBF was used to capture the melt pool shape of the manufacturing process in combination with the adopted
and evolution during processing [161,322]. Again, custom-built sensing solutions [15,62,96]. Using existing, commercial AM
additive systems, designed to accommodate high-speed optical machines on the contrary provides the advantage of more
imaging, were used to observe melt pool formation and to capture consolidated and robust manufacturing processes, but has the
the trajectories of ejected spatter [30,155,172]. Similarly, Schlieren drawback that the in-situ monitoring solution must be designed
imaging can be used to reconstruct the gaseous flows taking place around a stricter set of existing constraints [21,92,224,305]. For
during laser processing [30]. example, AM processes based on a closed chamber (LPBF, EBPBF
Different goals and thus different design criteria typically and powder-DED) offer limited possibilities of installing in-process
characterise in-process measurement solutions dedicated to sensing equipment within the chamber, because of paucity of
process and part quality monitoring, as the primary objective in available space, necessity of not colliding or interfering with
this case is to detect out-of-control conditions which may lead to elements of the process, and prohibitive operating conditions
defective parts being produced [74,97]. Early detection of issues (temperature, gas flows, floating particles, etc.) [74].
undermining part quality during fabrication is particularly The most frequently adopted technologies for in-process
important in the manufacture of high value-added AM parts, measurement include optical and infra-red imaging for the layer
where failure to meet specifications at the end of the production before or after processing, and for the melt pool [21,29,33,62,82
cycle may be unacceptable, because of the costs associated with ,92,129,141,186,224,230,242,260] (see also Figs. 26–28), and the
discarding or reworking the part [81,91,162]. Fortunately, AM use of photodiodes or pyrometry sensors for the melt pool
processes are inherently sequential, because they proceed layer by [29,62,139,212]. The above measurement technologies are suffi-
layer, and thus are intrinsically slower than other bulk manufactur- ciently fast to allow for sampling rates compatible with the melt
ing processes, such as casting or forging. This makes AM processes pool dynamics and thus also with the layer fabrication cycle. The
more suitable to in-process monitoring, akin to material removal region of the melt pool and its immediate surroundings in both
processes, the latter sharing many similarities with AM, being also DED and PBF processes are characterised by a wide range of
sequential and layer-based [91].

6.2. Targets for in-process measurement

The choice of what events to observe in-process varies


depending on application requirements. For process understand-
ing, the largest possible coverage is often sought for in terms of
measurement targets and spatial-temporal scales, because of the
relatively recent emergence of many metal AM processes makes
the full breadth of involved physical phenomena still partially Fig. 26. Layer imaging with in-situ infra-red camera as proposed by Rodriguez
unknown [30,161,172,322]. For process monitoring, the choice of et al. [230].
690 R.K. Leach et al. / CIRP Annals - Manufacturing Technology 68 (2019) 677–700

In many cases, measurement systems (for example, based on


optical or infra-red imaging) may be placed outside the build
chamber, accessing the chamber through pre-existing or custom-
designed transparent windows [224,321]. However, in such cases,
all the optical elements must be designed to achieve acceptable
compromises between measurement performance (mainly field of
view and resolution) and bi-directional protection to and from the
process [33], i.e., the process should not damage the measurement
system, and the measurement system should not significantly alter
the process. Observation of the process from outside the chamber
also typically results in a viewpoint not orthogonal to the layer
surface, and not coaxial with the laser/electron beam (for example,
see Fig. 27 [145]). In such cases, data correction to compensate for
the viewpoint may be implemented [260]. Examples of imaging
sensors mounted internally to the chamber can also be found
[82,92], requiring appropriate protection from the harsh environ-
mental conditions. Finally, measurement solutions for the melt
Fig. 27. Monitoring temperature patterns inside the build chamber in PBF by means
pool, fully co-axial to the energy beam usually require dedicated,
of an external sensor; infra-red camera configuration example (adapted from
Krauss et al. [145]). fully integrated architectures (for example, see Fig. 28) [29,62,186].
Additional in-process monitoring solutions are being explored
based on non-destructive testing technologies. For example, Du
et al. [70] have explored the in-process measurement of Eddy
currents for detecting subsurface defects in DED processes. Rieder
et al. [226] have proposed the use of an ultrasonic transducer
mounted under the build platform of a LBPF machine: the
ultrasonic wave generated by the transducer is reflected by the
walls of the part being built and by internal discontinuities, thus
reflection patterns can be analysed to detect defects. As opposed to
generating acoustic signals (active acoustics), existing acoustic
emissions can be captured from the process or workpiece
themselves (passive acoustics). For example, Eschner et al. [72]
Fig. 28. Melt pool monitoring solution proposed by Berumen et al. [29]; (a) the have placed a piezoelectric sensor under the base plate of a LPBF
reflected emissions from the melt pool are captured by a photodiode (intensity machine to sample structure-borne acoustic signals as the part is
monitoring) and a camera system (melt pool shape and size); (b) melt pool image being built. The signals are used to detect pores. Patel et al. [209]
captured by the camera.
have investigated the use of spatially resolved acoustic spectros-
copy to monitor the state of the current layer in LPBF.
Experimentation so far has focused on off-line measurements,
emissions at different wavelengths, thus useful information can be but investigations are in progress for in-process integration.
extracted from both the visible and the infra-red range. For Finally, stress-strain information can be collected in-process. Van
example, Griffith et al. have used both ranges of wavelengths to Belle et al. [26] have used a strain gauge mounted under the build
image the shape of the melt pool in powder-DED [99]. plate to monitor the workpiece in LPBF.
Electron imaging solutions have also been recently investigated
for monitoring the current layer in EBPBF, given the availability of 6.4. Sensor data analysis and processing
both back scattered and secondary electrons. Arnold et al. [15] have
used a detector to capture back scattered electrons and generate The importance of in-process measurement, in particular for
electron optical images of the layer in a way comparable to monitoring the process, is now widely recognised, and even
scanning electron microscopy. The solution is used to detect pores. manufacturers of AM machines have begun equipping their latest
Wong et al. [305] have developed a custom electron imaging sensor offerings with a wide array of sensors. Since sensor data streams
for capturing both back scattered electrons and secondary can be stored, interesting scenarios are beginning to arise. When
electrons, and have used the solution to detect deviations of the any process variable (for example, temperature of the melt pool) is
part contour from nominal via image analysis on the current layer. recorded through time, together with encoder information from
The possibility of acquiring areal topography information the AM machine axes, then a complete history of that variable in
(height maps) from the layers is being explored as well: as time and space is available for the complete duration of process
opposed to conventional 2D imaging of the layer, a height map execution. This type of data logging can be found in recent
contains quantitative information about the local elevations literature [15,33], but also in commercial AM machines (for
(heights) of a surface, providing a 3D model of its reliefs. example, see [29] or the InfiniAM Spectral software from Renishaw
However, not many options for acquiring areal topography data [223]). An increasingly popular term in recent literature is also
are currently available and capable of achieving sufficiently high “optical tomography”, used in the context of in-process monitoring
spatial resolution within limited scan times. The most frequent of AM to refer to digital images collected from the layers of an AM
solutions include the use of fringe projection (which requires one part and aggregated into a vertical stack. An optical tomography
or two cameras and a pattern projector) [67,168,321], or the use of dataset may offer hints at the spatial localisation of defects and
laser line scanners [21,307] combined with some type of other singularities detected on individual images of the stack, as
mechanical traversal mechanism to implement raster scanning long as local differences in pixel colours or intensities may be
(otherwise a line scanner would only acquire a single line profile). deemed as a reliable indicators of the defect/singularity in the first
The traversal approach has been so far implemented by mounting place. Camera images may also be used to extract the contour of
the line scanner on the recoater blade [21], or on the (moving) the part on the current layer, thus allowing tracking of the vertical
processing head (for example, the welder head of a wire-DED evolution of the contour as new layers are created. Similar results
process [307]). Areal topography data from LPBF layers has also may be obtained by looking at slices of XCT data (both for contour
been acquired by means of high-speed spectral-domain optical tracking and for internal defect identification), however, XCT
coherence tomography [63]. measurement can only be done on the final part for the time being,
R.K. Leach et al. / CIRP Annals - Manufacturing Technology 68 (2019) 677–700 691

whilst optical imaging can be implemented in-process. Research volumetric datasets (XCT data and optical tomography data)
on how optical tomography data correlates with other types of [92]. Instead of trying to detect singularities/regions on layer
volumetric datasets, such as that from XCT, is in progress (see for images that are representative of defects, an alternative approach
example [93]). The idea is to see whether a defect visible in the XCT consists of using the entire contents of a series of images
stack (for example, a pore) is also visible in the optical tomography representing "good" layers to create a statistical model to be used
stack. in conjunction with methods from statistical process control.
As the amount of stored data increases, new challenges are When new images (new layers) are analysed, they can either be
beginning to arise, related to how to manage and handle found consistent with the statistical model representative of the
increasingly larger amounts of information. One of the main "in-control" state (i.e. the process is doing well), or they can be
issues at the moment is understanding what pieces of information found inconsistent with the model, meaning that an anomaly is
are really important to retain within a metal AM process, because being registered (out of control state). Multiple options are
of some proven correlation with final part quality and the available to turn the complex content of an image into forms
insurgence of defects. Thus on one hand, efforts are being made compatible with statistical modelling and statistical process
to reach a satisfactory classification of defects in metal AM parts control; in the work by Colosimo et al. [57] for example,
(for example, see Ref. [176] for LPBF). On the other hand, statistically weighted T-mode principal component analysis is
correlations between sensed variables and part quality are sought used to encode image information.
for, for example, by comparing porosity or other internal defects
identified at specific locations in volumetric datasets obtained by 6.5. In-process quality control for metal additive manufacturing
XCT on the final part, with sensor data streams corresponding to
those same locations [92]. An example comparison of volumetric A final mention should be reserved to in-process control. As
datasets involving multiple sensors as well as XCT data is shown in opposed to monitoring, whose goal is only to detect or predict the
Fig. 29. insurgence of conditions which will ultimately affect the quality of
the build, control implies the implementation of corrective actions.
A well-known problem that can be approached by implementing
closed-loop feedback architectures is the stabilisation of the melt
pool (for example, see the patent and products by Concept Laser
[151]). The sequential nature of layer-based manufacturing is well
suited to handle approaches where defects detected within a layer
can be compensated by means of modifications of the following
layers. For example, in Ref. [106] the current layer generated by
wire-DED is scanned in 3D, and deviations from intended height
are compensated by controlling the wire feed rate on the next
deposition layer. Nassar [197] changed the scan path of a powder-
DED process in real-time based on the temperature of the previous
layer as measured by a pyrometer. The implementation of in-
Fig. 29. Volumetric datasets from different sensor types and comparison to XCT data
process control systems for the continuous quality control and
[255]; STL: original geometric model in standard tessellation language (STL) format;
XCT: computed tomography of the final build (grey levels correspond to material real-time optimisation of metal AM processes is one of the most
density values); quality management (QM) PHD: melt pool data from the important avenues of investigation for the development of
photodiode (grey levels correspond to melt pool intensity values); QM CAM: increasingly advanced AM processes.
melt pool shape from the camera (grey levels correspond to melt pool area values).
The values stored in the voxels of the CT, QM PHD and QM CAM datasets can be used
to search for correlations amongst the measured variables. Note that grey level
variations are scarcely visible by the naked eye in the figure. 7. Traceability and measurement uncertainty

7.1. National measurement institute work


Machine learning solutions are being investigated by several
research groups and commercial machine manufacturers, to Most of the National Measurement Institutes (NMIs) have
support the development of classifiers/predictors capable of programs of research in AM, focussing on metrology and
spotting relevant patterns in sensed data streams; i.e., patterns standardisation, but the largest program is The Measurement
most likely leading to defects in the final parts. Examples include Science for Additive Manufacturing program at NIST. The NIST
support vector machines [92,139], decision trees, k-nearest program currently focuses on material characterisation, real-time
neighbour, discriminant analysis [139], and convolutional neural monitoring and control of AM processes, qualification methodolo-
networks [242]. Especially for image-like sensor data (for example, gies, and systems integration for the complete AM process chain
optical or infra-red images), research efforts are also often [204]. Critical geometric measurements include metal powder
dedicated to the algorithmic extraction of relevant features for particle size and shape for powder characterisation and determi-
use as inputs for predictors/classifiers. For example, the melt pool nation of particle size distribution, size and shape of heat affected
presence and position must be automatically identified within the zone (melt pool) for process monitoring and control, location, size
image field of view, and its boundaries must be accurately isolated and shape of internal features of AM parts, and surface texture
as a means to assess the shape and size of the pool itself. To be characterisation of AM surfaces to be used for part qualification
compatible with in-process measurement, algorithmic image used as the process signature for process qualification. Dynamic
processing must be fast, thus simple solutions are often privileged, imaging techniques [124] are used for particle size and shape
for example, based on thresholding/blob detection or consolidated measurements. Establishing uncertainty of these measurements is
edge detection methods [33,46,62,224,305]. A more complex currently under investigation. The size and shape measurement of
method of edge detection for melt pool images, based on log-Gabor the heat affected zone relies on infra-red (IR) thermography and
filters and phase congruency, can be found in the work by Song image processing. The uncertainty of temperature measurements
et al. [260]. Areal topography data from fringe projection or line by IR thermography has been established [156]. Size measurement
scanners is similarly processed with simple and fast methods, such uncertainty based on image processing is also currently under
as height thresholding [321]. Interestingly, some authors are also investigation. Another ongoing effort is related to internal feature
beginning to apply 3D image processing methods directly on size and shape measurements using XCT. In order to evaluate the
692 R.K. Leach et al. / CIRP Annals - Manufacturing Technology 68 (2019) 677–700

performance of XCT instruments and verify the accuracy of the other specific features, including overhangs, ramps, cones, hemi-
images obtained from such instruments, NIST scientists are spheres, etc. An example is shown in Fig. 30 right [175].
developing well characterised reference artefacts [203]. A literature review of existing AM test artefacts is given
The European Metrology Program for Innovation and Research elsewhere [189]. In this review, the test artefacts are classified in to
(EMPIR) has the Metrology for Additively Manufactured Medical four categories: (i) artefacts for comparing systems or technologies
Implants project running to carry out cross-European research to for decision making, (ii) artefacts for evaluating individual systems
enable the production, on demand, of customised implants which for optimisation, (iii) artefacts for evaluating metal-based pro-
match a patient’s anatomy, with grafts that promote bone growth, cesses, and (iv) artefacts for other uses. The same review points out
as well as surgical guides that help the surgeons. The objective of the need for a standardised test artefact and builds upon the
this project is to provide a comprehensive basis to enable the safe findings of previous work by Richter and Jacobs [225], which
use of medical AM products. In-vivo bone characterisation has identified the following “rules” for test artefacts:
been performed after installing lattice structured implants
fabricated by AM [205] and customised dental and spinal implants  should be large enough to allow testing of the whole build area,
that match the patient’s anatomy have been manufactured by AM  should include many small, medium, and large features,
and are undergoing geometrical and mechanical testing.  should include in- and out-wards pointing features,
 should allow a short build time,
7.2. Transfer artefacts  should allow low material consumption,
 should be easy to measure,
In recent years, several test artefacts have been developed for  should include “real part” features.
metal AM, with the main purpose of determining a system’s
manufacturing performance. This section primarily discusses Based on these rules and on rigorous design criteria, NIST
artefacts developed for quantitative evaluation of the performance developed a test artefact intended for standardisation (see Fig. 31)
of AM systems and technologies. In addition, the last part of the
section briefly introduces artefacts used for performance verifica-
tion of measuring systems and traceability establishment for
specific dimensional measurements of metal AM products.
Manufacturing and measuring a test artefact is one of the two
main approaches for evaluating the performance of manufacturing
systems. The other approach consists of direct measurement of
system components or characteristics, but it is generally more
difficult for end users, as it requires access to positioning and
control of single system’s components, to measure them using
specific instruments mounted in the system’s working volume. The
approach of producing and measuring test artefacts is normally Fig. 31. Geometrical model of the “NIST part” [191].
simpler for end users, as manufacturing parts is the usual function
of AM systems and, in addition, specific measuring instruments are [191]. The “NIST part” incorporates many of the necessary
typically not required, as common manufacturing measuring structures and features for the evaluation of the process and
equipment for inspection of products can be used. Measuring a test machine capabilities, allows the assessment of systematic errors at
artefact enables the verification of multiple errors, as most errors the same time, and contains datum features for alignment and
of the manufacturing system combine and contribute to the errors measurement. Despite this and other standardisation efforts, a
of the manufactured artefact. The main drawback is that the large number of different test artefacts exist and still continue to be
correlation of the specific error source to a specific error in the developed by individual users to address very specific require-
artefact may be complex [238]. The approach of using test artefacts ments.
is not unique to AM, as for example, it is well known and For example, the “NIST part” was used by Bauza et. al. to
established for many years to characterise machining centres evaluate a specific AM system, but in addition they produced a
[127]. However, test artefacts for AM require specific character- lattice structure test part (see Fig. 22) to evaluate the system’s
istics [152]. capabilities in producing lattice structures and to show common
One of the first test artefacts developed to quantitatively metrology problems when inspecting AM parts with lattice
evaluate the performance of AM systems is the so-called “user structures [22]. Moreover, Bauza et. al. showed that a typical
part” (see Fig. 30 left). The part focuses on determining the problem with test artefacts is that their form and dimensions can
accuracy of the AM machine in the x-y plane [86]. Many change after removal from the machine build plate or after post
researchers developed test artefacts, often called benchmarking treatments, and thus the actual system errors can be masked and
parts, for comparing the performance of systems and technologies. changed by these post-built variations (see Fig. 12) [22].
A recent review on benchmarking parts for evaluating the Fahad and Hopkinson focused on the simplicity of the part and
geometrical performance of AM processes was produced by the possibility to easily measure it using tactile CMM (see Fig. 32a)
Rebaioli and Fassi [220]. One of the first authors proposing a test [75]. They also showed that the amount of surrounding powder has
artefact for comparing AM machines was Kruth [146]. Other a significant influence on the final properties of parts and has to be
researchers implemented different benchmarking parts, adding accounted for when testing AM machines [76]. A more complex

Fig. 30. Examples of geometries of early test artefacts. Left: “user part” focusing on
evaluating the machine accuracy in the x-y plane [86]. Right: artefact including Fig. 32. (a) Artefact developed by Fahad et. al. [76]. (b) Micro- and meso-scale
specific features: overhangs, ramps, cones, hemi-spheres, etc. [175]. artefact developed by Hao et al. [102].
R.K. Leach et al. / CIRP Annals - Manufacturing Technology 68 (2019) 677–700 693

and much smaller artefact (see Fig. 32b) was proposed by Hao et al.
which was especially designed for the evaluation of micro- and
meso-scale production capabilities, as well as process stability and
repeatability [102].
Möhring et al. developed a micro-scale test artefact including
simple geometrical features as well as freeform structures (Fig. 33)
to compare the capabilities of milling and different AM processes
in a CIRP round robin experiment involving different micro-
manufacturing processes. The comparison revealed large form
errors especially for thin walled features, caused by insufficient
resolution of AM processes, but also showed the advantages of
using XCT systems for measuring some of the key features [187].
Following the trend towards XCT inspection, Shah et al.
presented their test artefact especially tailored for inspection
with XCT (see Fig. 17). The part and the features are arranged on a
cylinder to reduce X-ray imaging artefacts and minimise additional
error sources. For reference measurements, the features are also
accessible by tactile CMM probes [248]. Other XCT inspection
artefacts are presented elsewhere [147].

Fig. 34. Artefact geometries proposed for standardization by ASTM F42/ISO TC 261:
(a) linear accuracy, (b) circular accuracy, (c) resolution pins, (d) resolution holes, (e)
resolution slots, (f) resolution ribs, (g) surface texture [188]. Note that the scale is
different for each geometry.

8. Synthesis and future research

In the main body of the paper, the current state of the art,
progress, and requirements in industry for geometrical metrology
for AM have been reviewed. This section will pull out the main
take-home messages and future research topics from each
section.
Fig. 33. Micro test artefact for interdisciplinary analyses in micro manufacturing There are significant gaps between existing AM specification
[187]. standards and the industrial needs for widespread use of AM
technology. The most important areas of future AM standards are
summarised in Section 2.3. There are many globally- and
To solve the need for standardisation, the ASTM F42/ISO TC regionally-coordinated research and development efforts to
261 Joint Group for standard test artefacts (Joint Group 52) is expedite the standardisation in these areas such as AMSC, ASTM
proposing a new approach to avoid the problem that one single Centers of Excellence, and AM-Platform. It is expected that due to
artefact might not address the many different needs of different such coordinated efforts, resulting regional and international
users. The Joint Group 52 decided that the specification standard standards will be consistent and harmonised among each other,
under development will describe a suite of test geometries, each which is a key requirement for smooth transfer of AM parts and
testing a different characteristic, and users will be allowed to technologies within multi-tiered supply networks. On the other
configure the individual geometries to whatever best fits the specific hand, urgent need for industrial standards and relatively low
needs [188]. Fig. 34 shows the proposed artefact geometries which maturity level of AM technologies create challenges for standards
are still being debated and subject to change in future reviews. developers. The boundaries between pre-normative research and
Examples of artefacts developed to include surface texture specific standardisation efforts are blurred. Therefore, it is foreseen that
features are shown in Fig. 20 (see Section 4.2). the existing AM standards and the ones currently being drafted
Besides artefacts for testing the performance of AM systems or may require more frequent revisions than the standards
XCT measuring systems, in recent years a number of artefacts have associated with more mature technologies, as the research
been developed for testing measuring systems and establishing community increases its understanding and improvement of
traceability in specific dimensional measurements of metal AM the various AM technologies.
products [42,130]. An example of such artefacts, recently proposed Coordinate measurements of surface form and dimensions
by Hermanek et al. [107], was designed to evaluate the accuracy of are complicated by specific characteristics of metal AM
porosity measurements by XCT. The artefact consists of a products, including complex shape and surface texture, with
dismountable assembly, which includes hemispherical features typically high roughness. For example, the roughness of AM
of different sizes serving as artificial calibrated pores. The surfaces causes significant deviations between dimensional
advantage of this design is the possibility to calibrate all features measurements performed using different contact or non-
using reference measuring systems such as tactile CMMs [107]. contact coordinate measuring systems. Besides tactile CMMs,
There has been recent work to exploit the design freedom optical methods – such as laser triangulation, photogrammetry,
inherent in AM to build specific features into a manufactured and fringe projection – as well as X-ray computed tomography
object that allow the object to be measured at a later stage. Rivas are increasingly used for coordinate measurements of metal AM
Santos et al. present a benchmarking artefact for a high-speed parts; however, the use of such non-contact measuring systems
sintering system that has specific features that facilitate its brings new challenges for traceability and comparability of
measurement with tactile CMM, optical photogrammetry and measurements. Current research efforts are focused on improv-
XCT [227]. ing the measurement results in terms of both accuracy and
694 R.K. Leach et al. / CIRP Annals - Manufacturing Technology 68 (2019) 677–700

measuring time. Research is under way to both maximise object Probably the biggest challenge for in-process metrology lies
coverage whilst at the same time minimising both measuring deeper, in the choice of what should be measured. The metal is
time and the amount of manual intervention in the measure- undergoing continuous and rapid evolution, yet the knowledge
ment planning. Integrating methods from artificial intelligence of process performance and behaviour is lagging behind. As a
into the measurement process is showing promise but requires result, it is often unclear what process variables should be
much more research into measurement automation, versatility, preferentially monitored. Because of such a lack of knowledge,
and traceability. the current tendency is to capture as much information as
AM surfaces challenge existing surface texture measurement possible, by equipping AM machines with an increasing number
and characterisation techniques. The mix of deterministic and of heterogeneous sensors. The amount of information that can
random surface geometry, high slope angles (including undercut be generated, even during the fabrication of a single part, is
features), and mixed optical characteristics mean that existing rapidly growing. The identification of what variables should be
measurement methods struggle and there needs to be an monitored and how involves significant research efforts going
international effort to try to develop good practice in how to ahead, both in terms of process and material understanding, and
optimise a measurement. Also, in many cases, existing surface data mining.
texture field parameters do not capture enough information to The principal impact of materials on geometry in metal AM is
allow process or functional improvements, so more advanced, distortion of parts produced. Current approaches to control
multi-scale and feature-based approaches need to be developed residual stress include part construction by attachment to a
and standardised. There probably is not a single “parameter for massive base plate to maintain rigidity coupled with an extensive
AM”, but there will be a suite of potential methods and good support structure connecting the part to the plate. After the build,
practice guidance is needed to navigate this suite for a given the part/base plate assembly is removed and given a stress relief
application. XCT methods are showing promising results for heat treatment prior to removing the part from the base plate.
surface measurement, especially for internal surfaces, but there is Since residual stresses are typically compressive in AM part
still much research to do to understand the sensitivity of the interiors and tensile on the part surface [219], parts may distort
various influence factors. There also needs to be a more concerted sufficiently to tear away from the support structure or base plate
effort to try to correlate surface topography (and sub-surface prior to the stress relief heat treatment. Generation of supports
features) with processing parameters. This information can be and their subsequent removal is the cause of significant cost
used to inform control strategies and later in-process metrology increase for production, loss of material, and energy waste.
developments. Increasing the build temperature has been partially effective.
With both coordinate metrology and surface metrology, there Indirect metal AM approaches with use of transient binders
is still an elephant in the room for AM surfaces: traceability. largely eliminates distortion during the AM build. Post-processing
Whilst there have been significant developments to allow heat treatments to remove the binder and densify the metal may,
comparison of various instruments, the “ground truth” is still however, also result in minor distortions. Effective process and
missing. The complexity and rough nature of AM surfaces means manufacturing developments are needed to eliminate the need
that the use of contact systems (stylus instruments and contact for supports.
CMMs) is problematic. It is still difficult to provide traceability for Since the dawn of metal AM, several researchers have been
relatively smooth surfaces with optical instruments and the developing test artefacts for studying and comparing systems or
standards are still to be developed [158]. Once again, whilst an technologies. Today, a large number of different artefacts exist
AM surface is “just another surface”; the added complexity brings and still continue to be developed by individual users to address
extra challenges when establishing the traceability route. This specific requirements. Current standardisation activities are
traceability issue needs urgent attention as AM processes carried out by the ASTM F42/ISO TC 261 Joint Group for standard
establish quality systems and industry compliance with estab- test artefacts, which is proposing a new approach to avoid the
lished standards. problem that one single artefact might not address the many
On-machine metrology for AM, and in particular in-process different needs of different users. The specification standard
metrology, i.e., measurement while the part is being built, is one of under development will describe a suite of test geometries, each
the hottest topics currently being tackled by the manufacturing testing a different characteristic, and users would configure the
community. The appeal of in-process measurement for AM resides individual geometries to whatever best fits the specific needs.
in being able to detect problems early during fabrication, offering Close attention needs to be paid to the measurability of such
the potential opportunity to salvage defective parts via the artefacts. In the same vein, there is an opportunity offered by the
implementation of corrective actions, or at least offering the design freedom of AM to incorporate features that make the
chance to discard parts early, avoiding the waste of time and measurement of, not only verification artefacts, but also actual
resources, which could be prohibitively high, especially for low products, simpler, faster and more effective (“design for
volume, high value-added parts. metrology”).
The challenge of in-process metrology for AM pertains to both
making sure the measurement is not disturbed by the process
(taking into consideration temperature, random disturbances due Acknowledgements
to volatiles, etc.) and making sure the process itself is not
disturbed by the measurement (least interaction possible, and The authors would like to thank the following for contribu-
least modifications possible to the process in order to accommo- tions to, and in-depth reviews of, the manuscript: Adam
date the measurement systems). In-process measurement may Thompson, Lewis Newton, Ian Maskery, Xiaobing Feng (Universi-
often be subjected to severe time constraints, depending on the ty of Nottingham); Markus Baier, Filippo Zanini (University of
physical phenomena that must be monitored. Some measure- Padova); Peter de Groot (Zygo); Jean-Pierre Kruth (KU Leuven);
ments may be performed once for each fabricated layer or every Stephen Newman (University of Bath); and Christopher Evans
few layers; others multiple times within each layer; others may (UNCC).
be required to operate at much higher sampling frequencies, for Certain commercial entities, equipment, or materials may be
example, to capture the dynamics of the melt pool (Raleigh identified in this document in order to describe an experimental
instabilities, occurrence of balling). There are evidently signifi- procedure or concept adequately. Such identification is not
cant challenges ahead, but also room for growth, for the intended to imply recommendation or endorsement by the
measurement technologies and systems dedicated to in-process authors, nor is it intended to imply that the entities, materials,
metrology for AM. or equipment are necessarily the best available for the purpose.
R.K. Leach et al. / CIRP Annals - Manufacturing Technology 68 (2019) 677–700 695

References [37] Bourell DL, Leu MC, Rosen DW (2009) Roadmap for Additive Manufacturing –
Identifying the Future of Freeform Processing, The University of Texas at Austin.
[1] Aboulkhair NT, Maskery I, Tuck C, Ashcroft I, Everitt NM (2016) On The [38] Brackett D, Ashcroft I, Hague R (2011) Topology optimization for additive
Formation of AlSi10Mg Single Tracks And Layers In Selective Laser Melting: manufacturing. Proceeding of SFF Symposium.
Microstructure And Nano-Mechanical Properties. Journal of Materials Proces- [39] Brandi HT, Luckow H (1976) Method of Making Large Structural One-Piece
sing Technology 230:88–98. Parts of Metal, Particularly One-Piece Shafts. US Patent 3,985,995.
[2] Aggarangsi P, Beuth JL (2006) Localized Preheating Approaches For Reducing [40] Brown CO, Breinan EM, Kear BH (1982) Method for Fabricating Articles by
Residual Stress in Additive Manufacturing. Proceeding of SFF Symposium 709–720. Sequential Layer Deposition. US Patent 4,323,756.
[3] Aloisi V, Carmignato S (2016) Influence of Surface Roughness on X-Ray [41] Brown CA, Hansen HN, Jiang X, Blateyron F, Berglund J, Senin N, Bartkowiak T,
Computed Tomography Dimensional Measurements Of Additive Manufac- Dixon B, Goïc GL, Quinsat Y, Stemp WJ, Thompson MK, Ungar PS, Zahouani H
tured Parts. Case Studies in Nondestructive Testing and Evaluation 104–110. 6/B. (2018) Multiscale Analyses and Characterizations of Surface Topographies.
[4] Alrbaey K, Wimpenny D, Tosi R, Manning W, Moroz A (2014) On Optimization of CIRP Annals 67:839–862.
Surface Roughness of Selective Laser Melted Stainless Steel Parts: A Statistical [42] Brown S, Sun W, McCarthy MB, Woolliams P (2016) Assessment of Dimen-
Study. Journal of Materials Engineering and Performance 23:2139–2148. sional Measurement Tools to Measure Additively Manufactured Metallic
[5] AM-Motion (2018) A Strategic Approach to Increasing Europe’s Value Proposi- Parts. Proceeding of ASPE Summer Topical 200–205. June.
tion for Additive Manufacturing Technologies and Capabilities – First Draft AM [43] Buchbinder D, Meiners W, Pirch N, Wissenbach K, Schrage J (2014) Investi-
Roadmapwww.rm-platform.com/images/D5.3_rev_July2018-rev9.pdf. gation on Reducing Distortion by Preheating During Manufacture of Alumi-
[6] AM Platform, www.rm-platform.com/. num Components Using Selective Laser Melting. Journal of Laser Applications
[7] Ameta G, Fox J, Witherell P (2018) Tolerancing and Verification of Additive 26:1–10.
Manufactured Lattice With Supplemental Surfaces. Proc CIRP 75:69–74. [44] Kraemer A, Lanza G (2016) Methodology For The Evaluation of Ct Image
[8] Ameta G, Lipman R, Moylan S, Witherell P (2015) Investigating the Role of Quality in Dimensional Metrology. Proceeding of WCNDT. June.
Geometric Dimensioning and Tolerancing in Additive Manufacturing. The [45] Cabanettes F, Joubert A, Chardon G, Dumas V, Rech J, Grosjean C, Dimkovski Z
Journal of Mechanical Design 137:111401–111410. (2018) Topography of as Built Surfaces Generated in Metal Additive
[9] Ameta G, Witherell P (2019) Representation of Graded Materials and Struc- Manufacturing: A Multi Scale Analysis From Form to Roughness. Precision
tures to Support Tolerance Specification for Additive Manufacturing Appli- Engineering 52:249–265.
cation. Journal of Information Science and Engineering 19. 21008-1–21008-9. [46] Caltanissetta F, Grasso M, Petrò S, Colosimo BM (2018) Characterization of in-
[10] AMSC (2017) Standardization Roadmap for Additive ManufacturingFebruary situ Measurements Based on Layerwise Imaging in Laser Powder Bed Fusion.
2017 . Additive Manufacturing 24:183–199.
[11] AMSC (2018) Standardization Roadmap for Additive Manufacturing v2June 2018 . [47] Carmignato S (2012) Accuracy of Industrial Computed Tomography Measure-
[12] Angel J, De Chiffre L (2014) Comparison on Computed Tomography Using ments: Experimental Results From an International Comparison. CIRP Annals
Industrial Items. CIRP Annals 63:473–476. 61:491–494.
[13] Anwer N, Ballu A, Mathieu L (2013) The Skin Model, A Comprehensive [48] Carmignato S (2016) Computed Tomography. in Laperrière L, Reinhart G,
Geometric Model for Engineering Design. CIRP Annals 62:143–146. (Eds.) CIRP Encyclopedia of Production Engineering, Springer, Berlin, Heidel-
[14] Armillotta A (2016) Tolerance Analysis Considering Form Errors in Planar berg.
Datum Features. Procedia CIRP 43:64–69. [49] Carmignato S, Aloisi V, Medeossi F, Zanini F, Savio E (2017) Influence of
[15] Arnold C, Pobel C, Osmanlic F, Körner C (2018) Layerwise Monitoring of Surface Roughness on Computed Tomography Dimensional Measurements.
Electron Beam Melting via Backscatter Electron Detection. Rapid Prototyping CIRP Annals 66:499–502.
Journal. [50] Carmignato S, De Chiffre L (2003) A New Method for Thread Calibration on
[16] ASME Y14.5 (2009) Dimensioning and Tolerancing, ASME, New York, NY. Coordinate Measuring Machines. CIRP Annals 52:447–450.
[17] ASME Y14.46 (2017) Product Definition Practices for Additive Manufacturing, [51] Carmignato S, Dewulf W, Leach RK (2017) Industrial X-Ray Computed Tomog-
ASME, New York, NY. raphy, Springer.
[18] ASTM (2017) www.astm.org/COMMIT/F42_AMStandardsStructureAndPrimer. [52] Carmignato S, Savio E (2011) Traceable Volume Measurements Using Coor-
pdf. dinate Measuring Systems. CIRP Annals 60:519–522.
[19] ASTM (2018) amcoe.org/about/. [53] Carmignato S, Voltan A, Savio E (2010) Metrological Performance Of Optical
[20] Baker R (1925) Method of Making Decorative Articles. US Patent 1,533,300. Coordinate Measuring Machines Under Industrial Conditions. CIRP Annals
[21] Barrett C, MacDonald E, Conner B, Persi F (2018) Micron-Level Layer-Wise 59:497–500.
Surface Profilometry to Detect Porosity Defects In Powder Bed Fusion Of [54] Chen JSS, Feng HY (2011) Optimal Layer Setup Generation in Layered
Inconel 718. JOM 1–9. Manufacturing With a Given Eror Constraint. International Journal of Indus-
[22] Bauza MB, Moylan SP, Panas RM, Burke SC, Martz HE, Taylor JS, Smokovitz JD trial and Manufacturing Systems Engineering 30:165–174.
(2014) Study of accuracy of parts produced using additive manufacturing. [55] De Chiffre L, Carmignato S, Kruth JP, Schmitt R, Weckenmann A (2014)
ASPE Spring Topical 86–91. Industrial Applications Of Computed Tomography. CIRP Annals 63:655–677.
[23] Beaman JJ, Barlow JW, Bourell DL, Crawford RH, Marcus HL, McAlea KP (1997) [56] Ciraud PA (1972) Process and Device for the Manufacture of any Objects Desired
Solid Freeform Fabrication: A New Direction in Manufacturing, Kluwer Academ- from any Meltable Material, FRG Disclosure Publication: 2663777.
ic Publishers, Boston35–38. [57] Colosimo BM, Grasso M (2018) Spatially Weighted Pca for Monitoring Video
[24] Beaucamp AT, Namba Y, Charlton P, Jain S, Graziano AA (2015) Finishing of Image Data With Application to Additive Manufacturing. International Journal
Additively Manufactured Titanium Alloy By Shape Adaptive Grinding (SAG). of Quality Assurance in Engineering and Technology Education 50:391–417.
Surface Topography: Metrology and Properties 3024001. [58] Craeghs T, Clijsters S, Yasa E, Kruth JP (2011) Online Quality Control Of
[25] Becker R, Grzesiak A (2009) Rapid Manufacturing in Automation Applica- Selective Laser. Melting Proceeding of SFF Symposium 212–226.
tions. Innovative Developments in Design and Manufacturing. 4th Interna- [59] Criales LE, Arisoy YM, Lane B, Moylan S, Donmez A, Özel T (2017) Laser
tional Conference on Advanced Research in Virtual and Rapid Prototyping, Leiria, Powder Bed Fusion Of Nickel Alloy 625: Experimental Investigations Of
Portugal, . Effects of Process Parameters on Melt Pool Size and Shape With Spatter
[26] van Belle L, Vansteenkiste G, Boyer JC (2013) Investigation of Residual Stresses Analysis. International Journal of Machine Tools and Manufacture 121:22–36.
Induced During The Selective Laser Melting Process. Key Engineering Materials [60] Dai K, Shaw L (2002) Distortion Minimization Of Laser-Processed Compo-
554–557:1828–1834. nents Through Control Of Laser Scanning Patterns. Rapid Prototyping Journal
[27] Benedetti M, Fontanari V, Bandini M, Zanini F, Carmignato S (2018) Low- and 8:270–276.
High-cycle Fatigue Resistance Of Ti-6al-4v Eli Additively Manufactured Via [61] Deckard C (1989) Method and Apparatus for Producing Parts by Selective
Selective Laser Melting: Mean Stress And Defect Sensitivity. International Sintering. US Patent 4,863,538.
Journal of Fatigue 107:96–109. [62] Demir AG, De Giorgi C, Previtali B (2018) Design And Implementation of a
[28] Berglund J, Söderberg R, Wärmefjord K (2018) Industrial Needs and Available Multisensor Coaxial Monitoring System With Correction Strategies For Se-
Techniques for Geometry Assurance for Metal AM Parts with Small Scale lective Laser Melting of a Maraging Steel. Journal of Manufacturing Science and
Features and Rough Surfaces. Procedia CIRP 75:131–136. Engineering 140041003.
[29] Berumen S, Bechmann F, Lindner S, Kruth JP, Craeghs T (2010) Quality Control [63] DePond PJ, Guss G, Ly S, Calta NP, Deane D, Khairallah S, Matthews MJ (2018)
Of Laser- and Powder Bed-based Additive Manufacturing (AM) technologies. In situ Measurements of Layer Roughness During Laser Powder Bed Fusion
Physics Procedia 5:617–622. Additive Manufacturing Using Low Coherence Scanning Interferometry.
[30] Bidare P, Maier RRJ, Beck RJ, Shaphard JD, Moore AJ (2017) An open-architec- Materials & Design 154:347–359.
ture metal powder bed fusion system for in-situ process measurements. [64] Dewulf W, Ferrucci M, Ametova E, Hermánek P, Probst G, Boeckmans B,
Additive Manufacturing 16:177–185. Craeghs T, Carmignato S (2018) Enhanced Dimensional Measurement by Fast
[31] Blanther JE (1892) Manufacture of Contour Relief Maps, US Patent 473,901. Determination and Compensation of Geometrical Misalignments of X-Ray
[32] Boeckmans B, Tan Y, Welkenhuyzen Y, Guo F, Dewulf W, Kruth JP (2015) Computed Tomography Instruments. CIRP Annals 67:523–526.
Roughness offset differences between contact and non-contact measure- [65] Dewulf W, Pavan M, Craeghs T, Kruth JP (2016) Using X-ray Computed
ments. Proceedings - euspen 189–190. June. Tomography to Improve the Porosity Level of Polyamide-12 Laser Sintered
[33] Boone N, Zhu C, Smith C, Todd I, Willmott JR (2018) Thermal Near Infrared Parts. CIRP Annals 65:205–208.
Monitoring System for Electron Beam Melting With Emissivity Tracking. [66] Dewulf W, Tan Y, Kiekens K (2012) Sense and Non-Sense of Beam Hardening
Additive Manufacturing 22:601–605. Correction in Ct Metrology. CIRP Annals 61:495–498.
[34] Boschetto A, Bottini L, Veniali F (2018) Roughness Modeling of AlSi10Mg Parts [67] Dickins A, Widjanarko T, Lawes S, Leach RK (2018) Design of a multi-sensor in-
Fabricated by Selective Laser Melting. Journal of Materials Processing Technol- situ inspection system for additive manufacturing. Proceeding of ASPE/euspen
ogy 241:154–163. Summer Topical. 248–242.
[35] Bourell D (2016) Perspectives in Additive Manufacturing. Annual Review of [68] DiMatteo PL (1976) Method of Generating and Constructing Three-Dimen-
Materials Research 46:1–18. sional Bodies. US Patent 3,932,923.
[36] Bourell D, Kruth JP, Leu M, Levy G, Rosen D, Beese AM, Clare A (2017) Materials [69] Ding L, Goshtasby A (2001) On the Canny Edge Detector. Pattern Recognition
for Additive Manufacturing. CIRP Annals 66:659–681. 34:721–725.
696 R.K. Leach et al. / CIRP Annals - Manufacturing Technology 68 (2019) 677–700

[70] Du W, Bail Q, Wang Y, Zhang B (2018) Eddy Current Detection of Subsurface [100] Grimm T, Wiora G, Witt G (2015) Characterization of Typical Surface Effects in
Defects for Additive/Subtractive Hybrid Manufacturing. The International Additive Manufacturing With Confocal Microscopy. Surface Topography: Me-
Journal of Advanced Manufacturing Technology 95:3185–3195. trology and Properties 3014001.
[71] Dupin S, Lame O, Barrès C, Charmeau JY (2012) Microstructural Origin of [101] Grimm T, Witt G, Wiora G (2016) Three-Dimensional Surface Measurement
Physical and Mechanical Properties of Polyamide 12 Processed by Laser for the Quantification Of Mechanical Properties Of Laser-Sintered Parts.
Sintering. European Polymer Journal 48:1611–1621. Materials Testing 58:293–301.
[72] Eschner N, Weiser L, Häfner B, Lanza G (2018) Development of an acoustic [102] Hao B, Korkmaz E, Bediz B, Ozdoganlar OB (2014) A Novel Test Artifact for
process monitoring system for selective laser melting (SLM). Proceeding of SFF Performance Evaluation of Additive Manufacturing Processes. Using Additive
Symposium. Manufacturing, ASPE Spring Topical, Berkeley, USA.
[73] Everton S, Dickens P, Tuck C, Dutton B (2018) Using Laser Ultrasound to Detect [104] Harding K (2013) Handbook of Optical Dimensional Metrology, CRC Press.
Subsurface Defects in Metal Laser Powder Bed Fusion Components. JOM 1–6. [105] Harrison N, McHugh PE, Curtin W, McDonnell P (2013) Micromotion and
[74] Everton SK, Hirsch M, Stavroulakis P, Leach RK, Clare AT (2016) Review of In- Friction Evaluation of a Novel Surface Architecture for Improved Primary
situ Process Monitoring and In-situ Metrology for Metal Additive Fixation of Cementless Orthopaedic Implants. Journal of the Mechanical
Manufacturing. Materials & Design 95:431–445. Behavior of Biomedical Materials 21:37–46.
[75] Fahad M, Hopkinson N (2012) A New Benchmarking Part for Evaluating the [106] Heralic A, Christiansson AK, Lennartson B (2012) Height Control of Laser
Accuracy and Repeatability of Additive Manufacturing (AM) Processes. Pro- Metal-wire Deposition Based on Iterative Learning Control and 3D Scanning.
ceeding of 2nd International Conference on Mechanical, Production and Auto- Optics and Lasers in Engineering 50:1230–1241.
mobile Engineering, 234–238. [107] Hermanek P, Carmignato S (2016) Reference Object for Evaluating the Accu-
[76] Fahad M, Hopkinson N (2017) Evaluation and Comparison of Geometrical racy of Porosity Measurements by X-Ray Computed Tomography. Case Studies
Accuracy of Parts Produced by Sintering-Based Additive Manufacturing Pro- in Nondestructive Testing and Evaluation 6:122–127.
cesses. The International Journal of Advanced Manufacturing Technology [108] Hermanek P, Carmignato S (2017) Porosity Measurements by X-Ray Comput-
88:3389–3394. ed Tomography: Accuracy Evaluation Using a Calibrated Object. Precision
[77] Fay MF, Colonna de Lega X, Schmidt M (2014) Measuring high-slope parts Engineering 49:377–387.
using coherence scanning interferometry. Proc ASPE. [109] Hirsch M, Catchpole-Smith S, Patel R, Marrow P, Li W, Sharples SD, Clare AT
[78] Ferrucci M, Leach RK, Giusca CL, Carmignato S, Dewulf W (2015) Towards (2017) Meso-Scale Defect Evaluation of Selective Laser Melting Using Spa-
Geometrical Calibration of X-Ray Computed Tomography Systems — A Re- tially Resolved Acoustic Spectroscopy. Proceedings of the Royal Society A
view. Measurement Science & Technology 26092003. 47320170194.
[79] Fox JC, Moylan SP, Lane BM (2016) Preliminary Study Toward Surface Texture [110] Hong M-H, Min BK, Kwon T-Y (2016) The Influence of Process Parameters on
as a Process Signature in Laser Powder Bed Fusion Additive Manufacturing. the Surface Roughness Of A 3d-Printed co-cr Dental Alloy Produced via
Proceeding of ASPE Summer Topical 168–173. Selective Laser Melting. Applied Sciences (Basel Switzerland) 6:401.
[80] Francois MM, Sun A, King WE, Henson NJ, Touret D, Bronkhorst CA, Carloson [111] Housholder R (1981) Molding Process. US Patent 4,247,508.
NN, Newman CK, Haut T, Gibbs JW, Livescu V, Vanfer Weil SA, Schraad MW, [112] Hull C (1986) Apparatus for Production of Three-dimensional Objects by
Blacker T, Lim H, Rodgers T, Walton O (2017) Modeling of Additive Stereolithography. US Patent 4,575,330.
Manufacturing Processes for Metals: Challenges and Opportunities. Current [113] Iassanov P, Gebrenegus T, Tuller M (2009) Segmentation of X-Ray Computed
Opinion in Solid State & Materials Science 21:198–206. Tomography Images of Porous Materials - a Crucial Step For Characterization
[81] Frazier WE (2014) Metal Additive Manufacturing: A Review. Journal of Mate- and Quantitative Analysis of Pore Structures. Water Resources Research 45:9.
rials Engineering and Performance 23:1917–1928. [114] ISO/ASTM 52921 (2013) Standard Terminology for Additive Manufacturing –
[82] Furumoto T, Alkahari MR, Ueda T, Aziz MSA, Hosokawa A (2012) Monitoring of Coordinate Systems and Test Methodologies, ASTM, International, West Con-
Laser Consolidation Process of Metal Powder With High Speed Video Camera. shohocken, USA.
Physics Procedia 39:760–766. [115] ISO/ASTM 52900 (2015) Standard Terminology for Additive Manufacturing –
[83] Furumoto T, Ueda T, Abdul Aziz MS, Hosokawa A, Tanaka R (2010) Study on Generaĺ Principles – Part 1: Terminology, ASTM, West Conshohocken, USA.
Reduction of Residual Stress Induced During Rapid Tooling Process: Influence [116] ISO/ASTM 52910 (2017) Standard Guide for Design for Additive Manufacturing,
of Heating Conditions on Residual Stress. Key Engineering Materials 447– ASTM, West Conshohocken, USA.
448:785–789. [117] ISO/ASTM 52902 (2018) Additive Manufacturing – Test artefacts – Standard
[84] Gale PL, Fair JE (1978) Method of Making Aluminum Piston with Reinforced guideline for geometric capability assessment of additive manufacturing systems,
Piston Ring Groove. US Patent 4,125,926. International Organization for Standardization, Geneva, Switzerland.
[85] Gao W, Haitjema H, Chenc YL, Fang FZ, Leach RK, Cheung CF, Savio E, Linares [118] ISO/TC261 (2015) Additive Manufacturing, Business Plan, International Orga-
JM (2019) On-machine and in-process Surface Metrology for Precision nization for Standardization, Geneva, Switzerland.
Manufacturing. CIRP Annals. in press. [119] ISO 2768-2 (1989) General Tolerances – Part 2: Geometrical Tolerances for
[86] Gargiulo EP (1992) Stereolithography Process Accuracy: User Experience. Features Without Individual Tolerance Indications, International Organization
Proceeding of 1st European Conference on Rapid Prototyping 187–207. for Standardization, Geneva, Switzerland.
[87] Gaskin TA (1973) Earth Science Teaching Device, US Patent 3,751,827. [120] ISO 10303-1 (1994) Industrial Automation Systems and Integration – Product
[88] Ghita O, James E, Trimble R, Evans K (2014) Physico-chemical Behaviour of Data Representation and Exchange – Part 1: Overview and Fundamental Prin-
Poly (Ether ketone) (PEK) in High Temperature Laser Sintering (HT-LS). ciples, International Organization for Standardization, Geneva, Switzerland.
Journal of Materials Processing Technology 214:969–978. [121] ISO 4288 (1996) Geometric Product Specifications (GPS) – Surface Texture:
[89] Ghorpade A, Karunakaran KP, Tiwari MK (2007) Selection of Optimal Part Profile Method – Rules and Procedures for the Assessment of Surface Texture,
Orientation in Fused Deposition Modelling Using Swarm Intelligence. Pro- International Organization for Standardization, Geneva, Switzerland.
ceedings of the Institution of Mechanical Engineers Part B 221:1209–1219. [122] ISO 4287 (1997) Geometrical Product Specification (GPS) – Surface Texture:
[90] Gibbons GJ, Hansell RG (2005) Direct Tool Steel Injection Mould Inserts Profile Method – Terms, Definitions and Surface Texture Parameters, Interna-
Through the Arcam Ebm Freeform Fabrication Process. Assembly Automation tional Organization of Standardization, Geneva, Switzerland.
25(300):305. [123] ISO 1302 (2002) Geometric Product Specifications (GPS) – Indication of Surface
[91] Gibson I, Rosen DW, Stucker B (2010) Additive Manufacturing Technologies: Texture in Technical Product Documentation, International Organization for
Rapid Prototyping to Direct Digital Manufacturing, Springer. Standardization, Geneva, Switzerland.
[92] Gobert C, Reutzel EW, Petrich J, Nassar AR, Phoha S (2018) Application of [124] ISO 13322-2 (2006) Particle Size Analysis – Image Analysis Methods – Part 2:
supervised machine learning for defect detection during metallic powder bed Dynamic Image Analysis Methods, International Organization for Standardi-
fusion additive manufacturing using high resolution imaging. Additive zation, Geneva, Switzerland.
Manufacturing 21:517–528. [125] ISO 25178 part 2 (2012) Geometrical Product Specification (GPS) – Surface
[93] Gögelein A, Ladewig A, Zenzinger G, Bamberg J (2018) Process Monitoring of Texture: Areal – Part 2: Terms, Definitions and Surface Texture Parameters,
Additive Manufacturing by Using Optical Tomography. Proceeding of 14th International Organization for Standardization, Geneva, Switzerland.
Quantitative InfraRed Thermography Conference, 266–272. [126] ISO 17296-3 (2014) Additive Manufacturing – General Principles – Part 3: Main
[94] Gomez C, Thompson A, Su R, DiSciacca J, Lawes S, Leach RK (2017) Optimisa- Characteristics and Corresponding Test Methods, International Organization for
tion of Surface Measurement for Metal Additive Manufacturing Using Coher- Standardization, Geneva, Switzerland.
ence Scanning Interferometry. Optical Engineering (Redondo Beach Calif) [127] ISO IS 10791-7 (2014) Test Conditions for Machining Centers - Part 7. Accuracy
56111714. of a Finished Test Piece International Organization for Standardization, Geneva,
[95] Gong H, Gu H, Zeng K, Dilip JJS, Pal D, Stucker B, Christiansen D, Beuth J, Switzerland, .
Lewandowski JJ (2014) Melt Pool Characterization For Selective Laser Melting [128] ISO 1101 (2017) Geometric Product Specifications (GPS) – Geometrical Toleran-
of Ti-6Al-4V Pre-alloyed Powder. Proceeding of SFF Symposium 256–267. cing — Tolerances of Form, Orientation, Location and Run-out, International
[96] Grantham S, Lane B, Neira J, Mekhontsev S, Vlasea M, Hanssen L (2016) Organization for Standardization, Geneva, Switzerland.
Optical Design and Initial Results from NIST’s AMMT/TEMPS facility. Proceed- [129] zur Jacobsmühlen J, Kleszczynski S. Witt G, Merhof D (2015) ) Elevated region
ing of International Society for Optics and Photonics, . 97380S. area measurement for quantitative analysis of laser beam melting process
[97] Grasso M, Colosimo BM (2017) Process Defects and in-situ Monitoring stability. Proceeding of SFF Symposium.
Methods in Metal Powder Bed Fusion: A Review. Measurement Science & [130] Jansson A, Zekavat AR, Pejryd L (2015) Measurement of internal features in
Technology 28044005. additive manufactured components by the use of computed tomography.
[98] Greitemeier D, Dalle Donne C, Syassen F, Eufinger J, Melz T (2016) Effect of Proceeding of Digital Industrial Radiology and Computed Tomography (DIR
Surface Roughness on Fatigue Performance of Additive Manufactured Ti-6Al- 2015).
4V. Materials Science and Technology 32:629–634. [131] Jantette FP, Keicher DM, Romero JA, Schanwald LP (2000) Method and System
[99] Griffith ML, Schlienger ME, Harwell LD, Oliver MS, Baldwin MD, Ensz MT, for Producing Complex-Shape Objects. US Patent 6,046,426.
Essien M, Brooks J, Robino CV, Smugeresky JE, Hofmeister WH, Wert MJ, [132] Joshi PC, Dehoff R, Duty CE, Peter WH, Ott RD, Love LJ, Blue CA (2012) Direct
Nelson DV (1999) Understanding Thermal Behavior in the LENS Process. digital additive manufacturing technologies: path towards hybrid integra-
Materials & Design 20:107–113. tion. Proceeding of Future of Instrumentation International Workshop, 1–4.
R.K. Leach et al. / CIRP Annals - Manufacturing Technology 68 (2019) 677–700 697

[133] Keller P, Mendricky R (2015) Parameters Influencing The Precision of slm [163] Lewandowski JJ, Seifi M (2016) Metal Additive Manufacturing: A Review Of
Production. MM Science Journal 10. Mechanical Properties. Annual Review of Materials Research 46:151–186.
[134] Kerckhofs G (2009) Morphological and Mechanical Quantification of Porous [164] Li C, Liu JF, Guo YB (2016) Efficient multiscale prediction of cantilever
Structures by Means of Micro-CT. PhD thesis, KU Leuven. distortion by selective laser melting. Proceeding of SFF Symposium 236–246.
[135] Kerckhofs G, Pyka G, Moesen M, Van Bael S, Schrooten J, Wevers M (2013) [165] Li C, Liu JF, Guo YB, Li ZY (2015) A Temperature-Thread Multiscale Modeling
High-resolution Microfocus X-ray Computed Tomography for 3D Surface Approach for Efficient Prediction of Part Distortion by Selective Laser Melting.
Roughness Measurements of Additive Manufactured Porous Materials. Ad- Proceeding of SFF Symposium 1166–1181.
vanced Engineering Materials 15:153–158. [166] Lifton JJ, Carmignato S (2017) Simulating the Influence of Scatter and Beam
[136] Khademzadeh S, Carmignato S, Parvin N, Zanini F, Bariani PF (2016) Micro Hardening in Dimensional Computed Tomography. Measurement Science &
Porosity Analysis in Additive Manufactured Niti Parts Using Micro Computed Technology 28104001.
Tomography and Electron Microscopy. Materials & Design 90:745–752. [167] Lifton JJ, Malcolm AA, McBride JW (2015) On the Uncertainty of Surface
[137] Khademzadeh S, Zanini F, Bariani PF, Carmignato S (2018) Precision Additive Determination in X-Ray Computed Tomography For Dimensional Metrology.
Manufacturing of Niti Parts Using Micro Direct Metal Deposition. The Inter- Measurement Science & Technology 26035003.
national Journal of Advanced Manufacturing Technology 96:3729–3736. [168] Liu Y, Blunt LA, Gao F, Jiang X, Zhang Z, Saunby G, Dawes J, Blackham B,
[138] Khairallah SA, Anderson AT, Rubenchik King WE (2016) Laser Powder-bed Rahman HA, Smith C (2018) In-Situ Areal Inspection of Powder Bed for
Fusion Additive Manufacturing: Physics of Complex Melt Flow and Formation Electron Beam Fusion AM System Based on Fringe Projection. Proceeding of
Mechanisms of Pores, Spatter, and Denudation Zones. Acta Materialia 108:36– ASPE/euspen Summer Topical Meeting: Advancing Precision in Additive
45. Manufacturing.
[139] Khanzadeh M, Chowdhury S, Marufuzzaman M, Tschopp MA, Bian L (2018) [169] Löber L, Flache C, Petters R, Kühn U, Eckert J (2013) Comparison of Different
Porosity Prediction: Supervised-Learning of Thermal History for Direct Laser Post Processing Technologies for SLM Generated 316l Steel Parts. Rapid
Deposition. International Journal of Industrial and Manufacturing Systems Prototyping Journal 19:173–179.
Engineering 47:69–82. [170] Lou S, Brown SB, Sun W, Zeng W, Abdul-Rahman HS, Jiang X, Scott PJ (2017)
[140] Kiekens K, Welkenhuyzen F, Tan Y, Bleys P, Voet A, Kruth J-P, Dewulf W (2011) Use of Morphological Method to Investigate the Influence of Surface Texture
A Test Object With Parallel Grooves for Calibration and Accuracy Assessment on Dimensional Measurement of Additively Manufactured Parts. Proceeding
of Industrial Computed Tomography (Ct) Metrology. Measurement Science & of euspen/ASPE Special Interest Group Meeting: Additive Manufacturing.
Technology 22115502. [171] Lou S, Jiang X, Sun W, Zeng W, Pagani L, Scott PJ (2018) Characterisation
[141] Kleszczynski S, zur Jacobsmühlen J, Sehrt JT, Witt G (2012) Error Detection in Methods for Powder Bed Fusion Processed Surface Topography. Precision
Laser Beam Melting Systems by High Resolution Imaging. Proceeding of SFF Engineering. in press.
Symposium. [172] Ly S, Rubenchik AM, Khairallah SA, Guss G, Matthews MJ (2017) Metal Vapor
[142] Körner C, Bauerei A, Attar E (2013) Fundamental Consolidation Mechanisms Micro-Jet Controls Material Redistribution in Laser Powder Bed Fusion Addi-
During Selective Beam Melting of Powders. Modelling and Simulation in tive Manufacturing. Scientific Reports 7:4085.
Materials Science and Engineering 21085011. [173] Ma CP, Guan YC, Zhou W (2017) Laser Polishing of Additive Manufactured Ti
[143] Körner C (2016) Additive Manufacturing of Metallic Components by Selective alloys. Optics and Lasers in Engineering 93:171–177.
Electron Beam Melting — A Review. International Materials Reviews 61:361– [174] Maamoun AH, Elbestawi MA, Veldhuis SC (2018) Influence of Shot Peening on
377. AiSi10Mg parts fabricated by additive manufacturing. Journal of Manufactur-
[144] Koutiri I, Pessard E, Peyre P, Amlou O, Terris TD (2018) Influence of Slm ing and Materials Processing 2:40.
Process Parameters on The Surface Finish, Porosity Rate and Fatigue Behavior [175] Mahesh M, Wong Y, Fuh JYH, Loh HT (2004) Benchmarking for Comparative
of As-built inconel 625 Parts. Journal of Materials Processing Technology Evaluation of Rp Systems and Processes. Rapid Prototyping Journal 10:123–
255:536–546. 135.
[145] Krauss H, Eschey C, Zaeh MF (2012) Thermography for monitoring the [176] Malekipour E, El-Mounayri H (2018) Common Defects and Contributing
selective laser melting process. Proceeding of SFF Symposium 999–1014. Parameters in Powder Bed Fusion Am Process and Their Classification for
[146] Kruth JP (1991) Material Incress Manufacturing by Rapid Prototyping Tech- Online Monitoring and Control: A Review. The International Journal of Ad-
niques. CIRP Annals 40:603–614. vanced Manufacturing Technology 95:527–550.
[147] Kruth JP, Bartscher M, Carmignato S, Schmitt R, De Chiffre L, Weckenmann A [177] Mani M, Lane BM, Donmez MA, Feng SC, Moylan SP (2017) A Review on
(2011) Computed Tomography for Dimensional Metrology. CIRP Annals Measurement Science Needs For Real-Time Control of Additive Manufactur-
60:821–842. ing Metal Powder Bed Fusion Processes. International Journal of Production
[148] Kruth J-P, Deckers J, Yasa E, Wauthle R (2012) Assessing and Comparing Research 55:1400–1418.
Influencing Factors of Residual Stresses in Selective Laser Melting Using a [178] Manriquez Frayre JA, Bourell DL (1990) Selective Laser Sintering of Binary
Novel Analysis Method. Proceedings of the Institution of Mechanical Engineers Metallic Powder. Proceeding of SFF Symposium 99–106.
Part B 226:980–991. [179] Maskery I, Aboulkhair NT, Corfeld MR, Tuck C, Clare AT, Leach RK, Wildman
[149] Kruth J-P, Froyen L, Van Vaerenbergh J, Mercelis P, Rombouts M, Lauwers B RD, Ashcroft IA, Hague RJM (2016) Quantification and Characterisation of
(2004) Selective Laser Melting of Iron-Based Powder. Journal of Materials Porosity in Selectively Laser Melted Al-Si10-Mg using X-ray Computed To-
Processing Technology 149:616–622. mography. Journal of Materials Processing Technology 111:192–204.
[150] Kruth JP, Leu M, Nakagawa T (1998) Progress in Additive Manufacturing and [180] Maskery I, Aremu AO, Tuck C, Wildman RD, Ashcroft IA, Hague RJM (2015)
Rapid Prototyping. CIRP Annals 47:525–540. Mechanical Properties of Ti-6Al-4V Selectively Laser Melted Parts With Body-
[151] Kruth JP, Mercelis P (2009) Procedure and Apparatus For in-situ Monitoring centred-Cubic Lattices of Varying Cell Size. Experimental Mechanics 55:1261–
and Feedback Control of Selective Laser Powder Processing. U.S. Patent 1272.
Application 12/308,032. [181] Maheswaraa U, Burell D, Seeparsad CC (2007) Design and Freeform Fabrica-
[152] Kruth JP, Vandenbroucke B, Vaerenbergh JV, Mercelis P (2005) Benchmarking tion of Deployable Structures With Lattice Skins. Rapid Prototyping Journal
of different SLS/SLM processes as rapid manufacturing techniques. Interna- 13:213–225.
tional Conference on Polymers and Moulds Innovations (PMI) ;(April)20–23. [182] Maszybrocka J, Stwora A, Gapinski B, Skrabalak G, Karolus M (2017) Mor-
[153] Kunieda M, Nakagawa T (1984) Development of Laminated Drawing Dies by phology and Surface Topography of Ti6Al4V Lattice Structures Fabricated by
Laser Vutting. Bulletin of the Japan Society for Precision Engineering 184:353– Selective Laser Sintering. Bulletin of the Polish Academy of Sciences Technical
354. Sciences 65:85–92.
[154] Lamikiz A, Sanchez JA, Lopez de Lacalle L, Arana JL (2007) Laser Polishing of [183] Matthews MJ, Guss G, Khairallah SA, Rubenchik AM, Depond PJ, King WE
Parts Built up by Selective Laser Sintering. International Journal of Machine (2016) Denudation of Metal Powder Layers in Laser Powder Bed Fusion
Tools and Manufacture 47:2040–2050. Processes. Acta Materialia 114:33–42.
[155] Lane B, Mekhontsev S, Grantham S, Vlasea ML, Whiting J, Yeung H, Fox J, [184] Mazumder J (1991) Overview of Melt Dynamics in Laser Processing. Optical
Zarobila C, Neira J, McGlauflin M, Hanssen L (2016) Design, developments, Engineering (Redondo Beach Calif) 30:1208–1219.
and results from the NIST additive manufacturing metrology testbed [185] Mercelis P, Kruth JP (2006) Residual Stresses in Selective Laser Sintering and
(AMMT). Proceeding of SFF Symposium 1145–1160. Selective Laser Melting. Rapid Prototyping Journal 12:254–265.
[156] Lane BM, Whitenton EP, Madhavan V, Donmez MA (2013) Uncertainty of [186] Mercelis P, Kruth JP, Van Vaerenbergh J (2007) Feedback Control of Selective
Temperature Measurements by Infrared Thermography for Metal Cutting Laser Melting. Proceeding of 15th International Symposium on Electromachining
Applications. Metrologia 50:637–653. (ISEM-15), 421–426.
[157] Leach RK (2011) Optical Measurement of Surface Topography, Springer, Berlin. [187] Möhring HC, Kersting P, Carmignato S, Yagüe-Fabra JA, Maestro M, Jiménez R,
[158] Leach RK, Giusca CL, Haitjema H, Evans C, Jiang X (2015) Calibration and Ferrarisf E, Tuncg T, Bleicher F, Witsi WW, Walczak K, Hedlind M (2015) A
Verification of Areal Surface Texture Measuring Instruments. CIRP Annals Testpart for Interdisciplinary Analyses in Micro Production Engineering. CIRP
64:797–813. Annals 28:106–112.
[159] Léonard F, Tammas-Williams S, Pragnell PB, Todd I, Withers PJ (2012) Assess- [188] Moylan S (2015) Progress toward standardized additive manufacturing test
ment by X-ray CT of the Effects of Geometry and Build Direction on Defects in artefacts. Proceeding of ASPE Topical Meeting.
Titanium ALM Parts, ICT, Wels, Austria. [189] Moylan S, Cooke A, Jurrens K, Slotwinski J, Donmez MA (2012) A review of test
[160] Leuders S, Thöne M, Riemer A, Niendorf T, Tröster T, Richard HA, Maier HJ artifacts for additive manufacturing. NISTIR 7858.
(2013) On the Mechanical Behaviour of Titanium Alloy Tial6v4 Manufactured [191] Moylan S, Slotwinski J, Cooke A, Jurrens K, Donmez MA (2014) An Additive
by Selective Laser Melting: Fatigue Resistance and Crack Growth Perfor- Manufacturing Test Artefact. Journal of Research of NIST 119:429–459.
mance. International Journal of Fatigue 48:300–307. [192] Moylan S, Slotwinski J, Cooke A, Jurrens K, Donmez MA (2012) Proposal for a
[161] Leung CLA, Marussi S, Atwood RC, Towrie M, Withers PJ, Lee PD (2018) In Situ standardized test artefact for additive manufacturing machines and process-
X-ray Imaging of Defect and Molten Pool Dynamics in Laser Additive es. Proceeding of SFF Symposium.
Manufacturing. Nature Communications 9:1355. [193] Müller P (2013) Coordinate Metrology by Traceable Computed Tomography. PhD
[162] Levy GN, Schindel R, Kruth JP (2003) Rapid Manufacturing and Rapid Tooling Thesis, Technical University of Denmark.
With Layer Manufacturing (lm) Technologies, State Of The Art And Future [194] Nakagawa T (1979) Blanking Tool by Stacked Bainite Steel Plates. Press
Perspectives. CIRP Annals 52:589–609. Technique 93–101.
698 R.K. Leach et al. / CIRP Annals - Manufacturing Technology 68 (2019) 677–700

[195] Nakagawa T, Kunieda M, Liu S-D (1985) Laser Cut Sheet Laminated Forming [225] Richter J, Jacobs P (1992) Accuracy. in Jacobs P, (Ed.) Rapid Prototyping &
Dies by Diffusion Bonding. Proceedings of the Twenty-Fifth International Manufacturing, Society of Manufacturing Engineers, pp. 287–315.
Machine Tool Design and Research Conference 505–510. [226] Rieder H, Spies M, Bamberg J, Henkel B (2016) On- and Offline Ultrasonic
[196] NASA-MSFC-STD 3716 (2017) Standard for Additively Manufactured Spaceflight Inspection of Additively Manufactured Components. 19th World Conference
Hardware by Laser Powder Bed Fusion in Metals, Marshall Space Flight Center, on Non-Destructive Testing, 13–17.
Alabama, US. [227] Rivas Santos VM, Thompson A, Sims-Waterhouse D, Maskery I, Leach RK,
[197] Nassar AR, Keist JS, Reutzel EW, Spurgeon TJ (2015) Intra-Layer Closed-Loop Woolliams P (2018) Design and Characterisation of an Additive Manufactur-
Control of Build Plan During Directed Energy Additive Manufacturing of Ti- ing Benchmarking Artefact Following a Metrology Approach. Additive
6Al-4V. Additive Manufacturing 6:39–52. Manufacturing. in press.
[198] NCMS Report 0199RE98 (1998) 1998 Industrial Roadmap for the Rapid Proto- [228] Roberts CE (2018) Doctoral dissertation, University of Texas at Austin, USA.
typing Industry, National Center for Manufacturing Sciences, Michigan, US. [229] Roberts CE, Bourell D, Watt T, Cohen J (2016) A Novel Approach for Additive
[199] Newton L, Senin N, Gomez C, Blunt LA, Helmli F, Danzl R, Leach RK (2018) Manufacturing of Commercial Aluminum Alloys. Physics Procedia 83:909–
Focus Variation Areal Topography Measurement of Metal Additive Surfaces. 917.
Additive Manufacturing 25:365–389. [230] Rodriguez E, Medina F, Espalin D, Terraraz C, Muse D, Henry C, MacDonald E,
[200] Nguyen J, Park SI, Rosen D (2013) Heuristic Optimization Method for Cellular Wicker RB (2012) Integration of a thermal imaging feedback control system
Structure Design of Light Weight Components. International Journal of Preci- in Electron beam melting. Proceeding of SFF Symposium.
sion Engineering and Manufacturing 14:1071–1078. [231] Roehling TT, Wu SS, Khairallah SA, Roehling JD, Soezeri SS, Crumb MF,
[201] Nickel A, Barnett D, Prinz F (2001) Thermal Stresses and Deposition Patterns Matthews MJ (2017) Modulating Laser Intensity Profile Ellipticity for Micro-
in Layered Manufacturing. Materials Science and Engineering A 317:59–64. structural Control During Metal Additive Manufacturing. Acta Materialia
[202] NIST (2013) Measurement Science Roadmap for Metal-Based Additive Manu- 128:197–206.
facturingwww.nist.gov/document-3511. [232] Romano S, Brückner-Foit A, Brandão A, Gumpinger J, Ghidini T, Beretta S
[203] NIST News (2017) Getting the Inside Story on Products with Computed Tomo- (2017) Fatigue Properties of Alsi10mg Obtained by Additive Manufacturing:
graphywww.nist.gov/news-events/news/2017/02/ Defect-Based Modelling and Prediction of Fatigue Strength. Engineering
getting-inside-story-products-computed-tomography. Fracture Mechanics 187:165–189.
[204] NIST Programs (2013) Measurement Science for Additive Manufacturing Pro- [233] Rouholamin D, Hopkinson N (2014) An Investigation on The Suitability of
gramwww.nist.gov/programs-projects/ Micro-Computed Tomography as a Non-Destructive Technique to Assess the
measurement-science-additive-manufacturing-program. Morphology of Laser Sintered Nylon Parts. International Journal of Precision
[205] Obaton AF, Fain J, Djemaï M, Meinel D, Léonard F, Mahé E, Lécuelle B, Fouchet Engineering and Manufacturing-Green Technology 228:1529–1542.
JJ, Bruno G (2017) In vivo XCT bone characterization of lattice structured [234] Sabourin E, Houser SA, Bohn JH (1996) Adaptive Slicing Using Stepwise
implants fabricated by additive manufacturing. Heliyon 3. Uniform Refinement. Rapid Prototyping Journal 2:20–26.
[206] Özel T, Altay A, Donmez MA, Leach RK (2017) Surface Topography Investiga- [235] Safdar A, He HZ, Wei L-Y (2011) Effect Of Process Parameters Settings And
tions on Nickel Alloy 625 Fabricated Via Laser Powder Bed. Fusion The Thickness On Surface Roughness Of EBM Produced Ti-6Al-4V. Rapid Proto-
International Journal of Advanced Manufacturing Technology 1–8. typing Journal 18:401–408.
[207] Pagani L, Qi Q, Jiang X, Scott PJ (2017) Towards a New Definition of Areal [236] Savio E, De Chiffre L, Schmitt R (2007) Metrology of Freeform Shaped Parts.
Surface Texture Parameters On Freeform Surfaces. Measurement 109:281– CIRP Annals 56:810–835.
291. [237] Sbettega E, Zanini F, Benedetti M, Savio E, Carmignato S (2018) X-ray
[208] Parry LA, Ashcroft IA, Wildman RD (2018) Geometrical Effects on Residual Computed Tomography Dimensional Measurements of Powder Bed Fusion
Stress in Selective Laser Melting. Additive Manufacturing 25:166–175. Cellular Structures. Proceeding of euspen 467–468.
[209] Patel R, Hirsch M, Dryburgh P, Pieris D, Achamfuo-Yeboah S, Smith R, Light R, [238] Scaravetti D, Dubois P, Duchamp R (2008) Qualification of Rapid Prototyping
Sharples S, Clare A, Clark M (2018) Imaging Material Texture of As-Deposited Tools: Proposition of a Procedure and a Test Part. The International Journal of
Selective Laser Melted Parts Using Spatially Resolved Acoustic Spectroscopy. Advanced Manufacturing Technology 38:683–690.
Applied Sciences (Basel Switzerland) 8:1991. [239] Schild L, Kraemer A, Reiling D, Wu H, Lanza G (2018) Influence of Surface
[210] Pavan M, Craeghs T, Kruth JP, Dewulf W (2018) Investigating the Influence of Roughness on Measurement Uncertainty in Computed Tomography.iCT2018
X-ray CT Parameters on Porosity Measurement Of Laser Sintered PA12 Parts Wels.
Using a Design-of-Experiment Approach. Polymer Testing 66:203–212. [240] Schmidt M, Merklein M, Bourell D, Dimitrov D, Hausotte T, Wegener K,
[211] Pavan M, Sinnaeve M, Leyssens S, Craeghs T, Kruth J-P, Dewulf W (2017) Overmeyer L, Vollertsen F, Levy GN (2017) Laser Based Additive Manufactur-
Leuven, BelgiumDetermining the Dimensional Accuracy Limits of Laser Sin- ing in Industry and Academia. CIRP Annals 66:561–583.
tered PA12-parts: From Artefact Design to Dimensional Characterization by [241] Schwenke H, Neuschaefer-Rube U, Pfeifer T, Kunzmann H (2002) Optical
X-Ray Computed Tomography Proc. euspen/ASPE Special Interest Group Methods for Dimensional Metrology in Production Engineering. CIRP Annals
Meeting2017. Determining the Dimensional Accuracy Limits of Laser Sintered 51:685–699.
PA12-parts: From Artefact Design to Dimensional Characterization by X-Ray [242] Scime L, Beuth J (2018) A Multi-Scale Convolutional Neural Network for
Computed Tomography Proc euspen/ASPE Special Interest Group Meeting. Autonomous Anomaly Detection and Classification in a Laser Powder Bed
[212] Pavlov M, Doubenskaia M, Smurov I (2010) Pyrometric analysis of thermal Fusion Additive Manufacturing Process. Additive Manufacturing 24:273–286.
processes in SLM technology. Physics Procedia 5:523–531. [243] Scott PJ (2004) Pattern Analysis and Metrology: The Extraction of Stable
[213] Perera BV (1940) Process of Making Relief Maps. US Patent #2,189,592. Features From Observable Measurements. Proceedings of the Royal Society
[214] du Plessis A, Yadroitsev I, Yadroitsava I, Le Roux SG (2018) X-ray Microcom- 460:2845–2864.
puted Tomography in Additive Manufacturing: A Review of the Current [244] Se S, Pears N (2012) Passive 3D Imaging. in Pears N, Liu K, Bunting P, (Eds.) 3D
Technology And Applications. 3D Printing and Additive Manufacturing Imaging, Analysis and Applications, Springer.
5:227–247. [245] Sealy MP, Madireddy G, Li C, Guo YB (2016) Finite element modeling of hybrid
[215] Prabhakar P, Sames WJ, Dehoff R, Babu SS (2015) Computational Modeling Of additive manufacturing by laser shock peening. Proceeding of SFF Symposium
Residual Stress Formation During The Electron Beam Melting Process For 306–316.
Inconel 718. Additive Manufacturing 7:83–91. [246] Seifi M, Salem A, Beuth J, Harrysson O, Lewandowski JJ (2016) Overview of
[216] Pyka G, Burakowski A, Kerckhofs G, Moesen M, Van Bael S, Schrooten J, Materials Qualification Needs for Metal Additive Manufacturing. JOM
Wevers M (2012) Surface Modification of Ti6Al4V Open Porous Structures 68:747–764.
Produced by Additive Manufacturing. Advanced Engineering Materials [247] Senin N, Thompson A, Leach RK (2017) Characterisation of the Topography of
14:363–370. Metal Additive Surface Features With Different Measurement Technology.
[217] Quinsat Y, Lartigue C, Brown C, Hattali L (2016) Multi-scale Surface Charac- Measurement Science & Technology 28095003.
terization In Additive Manufacturing Using CT. International Joint Conference [248] Shah P, Racasan R, Bills P (2016) Comparison of Different Additive
on Mechanics Design Engineering and Advanced Manufacturin. Manufacturing Methods Using Optimized Computed Tomography. Case Stud-
[218] Quinsat Y, Lartigue C, Brown C, Hattali L (2017) Characterization of Surface ies in Nondestructive Testing and Evaluation 6:69–78.
Topography of 3D Printed Parts By Multi-scale Analysis. International Journal [249] Shiomi M, Osakada K, Nakamura K, Yamashita T, Abe F (2004) Residual Stress
on Interactive Design and Manufacturing 1–8. Within Metallic Model Made by Selective Laser Melting Process. CIRP Annals
[219] Rangaswamy P, Griffith ML, Prime MB, Holden TM, Rogge RB, Edwards JM, 53:195–198.
Sebring RJ (2005) Residual Stresses in LENS1 Components Using Neutron [250] Sidambe AT (2017) Three Dimensional Surface Topography Characterisation
Diffraction and Contour Method. Materials Science and Engineering 399:72– of the Electron Beam Melted Ti6Al4V. Metal Powder Report 72:200–205.
83. [251] Simonelli M, Tuck C, Aboulkhair NT, Maskery I, Ashcroft I, Wildman RD,
[220] Rebaioli L, Fassi I (2017) A Review on Benchmark Artifacts for Evaluating the Hague R (2015) A Study on the Laser Spatter and the Oxidation Reactions
Geometrical Performance of Additive Manufacturing Processes. The Interna- During Selective Laser Melting of 316L stainless steel, Al-Si10-Mg, and Ti-6Al-
tional Journal of Advanced Manufacturing Technology 93:2571–2598. 4V. Metallurgical and Materials Transactions A 46:3842–3851.
[221] Reese ZC, Fox JC, Kim FH, Taylor J, Evans C (2018) Effect of subsurface defects [252] Sims-Waterhouse D, Piano S, Leach RK (2017) Experimental Comparison of
on the surface topography of additive manufactured components. Proceeding Photogrammetry for Additive Manufactured Parts With and Without Laser
of ASPE/euspen Summer Topical Meeting. Speckle Projection. Proceedings of SPIE—the International Society for Optical
[222] Reese ZC, Fox JC, Taylor J, Evans C (2018) Evolution of cooling length in parts Engineering 10329. 103290L1-7.
created through laser powder bed fusion additive manufacturing. ASPE/ [253] Singh S, Ramakrishna S, Singh R (2017) Material Issues in Additive
euspen Summer Topical Meeting. Manufacturing: A Review. Journal of Manufacturing Process 25:185–200.
[223] Renishaw InfiniAM Spectral (2018) Energy Input and Melt-pool Emissions [254] Singh Rupal B, Jawad Qureshi A (2018) Geometric Deviation Modeling and
Monitoring for AM Systemswww.renishaw.com/en/infiniam-spectral–42310. Tolerancing in Additive Manufacturing: A GD&T Perspective. 1st Conference of
[224] Repossini G, Laguzza V, Grasso M, Colosimo BM (2017) On the Use of Spatter NSERC Network for Holistic Innovation in Additive Manufacturing (HI-AM).
Signature for In-Situ Monitoring of Laser Powder Bed Fusion. Additive [255] Sintavia (2018) Quality Assurance - Is Data From In-situ Monitoring Similar to a
Manufacturing 16:35–48. CT Scan?sintavia.com/ct-vs-qm-meltpool-data-case-study/.
R.K. Leach et al. / CIRP Annals - Manufacturing Technology 68 (2019) 677–700 699

[256] Slotwinski JA, Garboczi EJ, Hebenstreit KM (2014) Porosity Measurements [286] Vaithilingam J, Goodridge RD, Hague RJM, Christie SDR, Edmondson S (2016)
and Analysis for Metal Additive Manufacturing. Journal of Research of NIST The Effect of Laser Remelting on the Surface Chemistry of Ti6al4V Compo-
119:429–459. nents Fabricated by Selective Laser Melting. Journal of Materials Processing
[257] Slotwinski JA, Garboczi EJ, Stutzman PE, Ferraris CF, Watson SS, Peltz MA Technology 232:1–8.
(2014) Characterization of Metal Powders Used for Additive Manufacturing. [287] Vasinonta A, Beuth J, Griffith M (2000) Process Maps for Controlling Residual
Journal of Research of NIST 119:460–493. Stress and Melt Pool Size in Laser-Based SFF Processes. Proceeding of SFF
[258] Smith RJ, Hirsch M, Patel R, Li W, Clare AT, Sharples SD (2016) Spatially Symposium 200–208.
Resolved Acoustic Spectroscopy for Selective Laser Melting. Journal of Mate- [288] VDI/VDE 2630-1.2 (2010) Computed Tomography in Dimensional Metrology:
rials Processing Technology 236:93–102. Influencing Variables on Measurement Results and Recommendations for Com-
[259] Sochalski-Kolbus LM, Payzant EA, Cornwell PA, Watkins TR, Babu SS, Dehoff puted-Tomography Dimensional Measurements.
RR, Lorenz M, Ovchinnikova O, Duty C (2015) Comparison Of Residual [289] VDI/VDE 2630-1.3 (2011) Accuracy of Coordinate Measuring Machines: Com-
Stresses in Inconel 718 Simple Parts Made by Electron Beam Melting and puted Tomography in Dimensional Measurement.
Direct Laser Metal Sintering. Metallurgical and Materials Transactions A [290] Vetterli M, Schmid M, Knapp W, Wegener K (2017) New Horizons in Selective
46:1419–1432. Laser Sintering Surface Roughness Characterization. Surface Topography:
[260] Song L, Wang F, Li S, Han X (2017) Phase Congruency Melt Pool Edge Metrology and Properties 3045007.
Extraction for Laser Additive Manufacturing. Journal of Materials Processing [291] Villarraga-Gómez H, Clark D, Smith ST (2016) Effect of the Number of
Technology 250:261–269. Radiographs Taken in CT for Dimensional Metrology. Proceeding of euspen.
[261] Spears TG, Gold SA (2016) In-process Sensing In Selective Laser Melting (SLM) [292] Vrancken B (2016) Study of Residual Stresses in Selective Laser Melting. Doctoral
Additive Manufacturing Integrating Mater. Manufacturing Innovation 5:2. thesis, KU Leuven.
[262] Spierings A, Schneider M (2010) Comparison of Density Measurement Tech- [293] Vrancken B, Wauthle R, Kruth J-P, Van Humbeeck J (2013) Study of The
niques for Additive Manufactured Metallic Parts. Rapid Prototyping Journal Influence of Material Properties on Residual Stress in Selective Laser Melting.
17:380–386. Proceeding of SFF Symposium 393–407.
[263] Stavroulakis P, Chen S, Derlome C, Bointon P, Tzimiropoulos G, Leach RK [294] Wang X, Chou K (2015) Residual Stress in Metal Parts Produced by Powder-
(2018) Rapid Calibration Tracking of Extrinsic Projector Parameters in Bed Additive Manufacturing Processes. Proceeding of SFF Symposium 1463–
Fringe Projection Using Machine Learning. Optics and Lasers in Engineering 1474.
114:7–14. [295] Wang Z, Denlinger E, Michaleris P, Stoica AD, Ma D, Beese AM (2017) Residual
[264] Stavroulakis PI, Leach RK (2016) Review of Post-Process Optical Form Me- Stress Mapping in Inconel 625 Fabricated Through Additive Manufacturing:
trology for Industrial-Grade Metal Additive Manufactured Parts. The Review Method for Neutron Diffraction Measurements to Validate Thermomecha-
of Scientific Instruments 87041101. nical Model Predictions. Materials & Design 117:169–177.
[265] Stevinson B, Bourell DL, Beaman JJ (2006) Dimensional Stability During Post- [296] Wang D, Liu Y, Yang Y (2015) Theoretical and Experimental Study on Surface
Processing of Selective Laser Sintered Ceramic Preforms. Virtual and Rapid Roughness of 316L Stainless Steel Metal Parts Obtained Through Selective
Prototyping 1:209–216. Laser Melting. Rapid Prototyping Journal 22:706–716.
[266] Stevinson BY, Bourell DL, Beaman JJ (2007) Freeform Fabrication of Non- [297] Warnett JM, Titarenko V, Kiraci E, Attridge A, Lionheart WRB, Withers PJ,
Metallic Objects by Selective Laser Sintering And Infiltration. Materials Williams MA (2016) Towards In-process X-ray CT for Dimensional Metrology.
Science Forum 561–565:817–820. Measurement Science & Technology 27:1–14.
[267] Stevinson B, Bourell DL, Beaman JJ (2008) Over-Infiltration Mechanisms [298] Weckenmann A, Estler T, Peggs G, McMurtry D (2004) Probing Systems in
in Selective Laser Sintered Si/SiC preforms. Rapid Prototyping Journal Dimensional Metrology. CIRP Annals 53:657–684.
14:149–154. [299] Weimann V, Drescher P, Seitz H, Hansmann H, Bader R, Seyfarth A, Klinder A,
[268] Stolfi A, De Chiffre L (2018) Interlaboratory Comparison of a Physical and a Jonitz-Heincke A (2018) Effects of Build Orientation on Surface Morphology
Virtual Assembly Measured by CT. Precision Engineering 51:263–270. and Bone Cell Activity of Additively Manufactured Ti6Al4V Specimens.
[269] Strano G, Hao L, Everson RM, Evans KE (2013) Surface Roughness Analysis, Materials 11:915.
Modelling and Prediction in Selective Laser Melting. Journal of Materials [300] White WD (1964) Pressure Roller and Method of Manufacture, US Patent
Processing Technology 213:589–597. 3,156,968.
[270] Tang Y, Zhao YF (2016) A Survey of the Design Methods for Additive [301] White D (2003) Ultrasonic Object Consolidation. US Patent 6,519,500.
Manufacturing to Improve Functional Performance. Rapid Prototyping Journal [302] Witherell P, Herron J, Ameta G (2016) Towards Annotations and Product
22:569–590. Definitions for Additive Manufacturing. CIRP Annals 43:339–344.
[271] Thijs L, Verhaeghe F, Craeghs T, Humbeeck JV, Kruth J-P (2010) A Study of the [303] Wits WW, Carmignato S, Zanini F, Veneker THJ (2016) Porosity Testing
Microstructural Evolution During Selective Laser Melting of Ti-6Al-4V. Acta Methods for the Quality Assessment of Selective Laser Melted Parts. CIRP
Materialia 58:3303–3312. Annals 65:201–204.
[272] Thompson A, Maskery I, Leach RK (2016) X-ray Computed Tomography for [304] Wohlers T (2018) Wohlers Report 2018, Fort Collins: Wohlers Associates.
Additive Manufacturing: A Review. Measurement Science & Technology [305] Wong H, Neary D, Jones E, Fox P, Sutcliffe C (2018) Pilot Capability Evaluation
27:72001. of a Feedback Electronic Imaging System Prototype for In-process Monitoring
[273] Thompson MK, Moroni G, Vaneker T, Fadel G, Campbell RI, Gibson I, Bernard in Electron Beam Additive Manufacturing. The International Journal of Ad-
A, Schulz J, Graf P, Ahuja B, Martina F (2016) Design for Additive Manufactur- vanced Manufacturing Technology 1–14.
ing: Trends, Opportunities, Considerations, and Constraints. CIRP Annals [306] Xiao J, Anwer N, Durupt A, Le Duigou J, Eynard B (2018) Information Exchange
65:737–760. Standards for Design, Tolerancing and Additive manufacturing: A research
[274] Thompson A, Senin N, Giusca CL, Leach RK (2017) Topography of Selectively Review. International Journal on Interactive Design and Manufacturing 12:495–
Laser Melted Surfaces: a Comparison of Different Measurement Methods. 504.
CIRP Annals 66:543–546. [307] Xu F, Dhokia V, Colegrove P, McAndrew A, Williams S, Henstridge A, Newman
[275] Thompson A, Senin N, Maskery I, Körner L, Lawes SDA, Leach RK (2018) ST (2018) Realisation of a Multi-sensor Framework for Process Monitoring of
Internal Surface Measurement of Metal Powder Bed Fusion Parts. Additive the Wire Arc Additive Manufacturing in Producing Ti-6Al-4V Parts. Interna-
Manufacturing 20:126–133. tional Journal of Computer Integrated Manufacturing 31:785–798.
[276] Thompson A, Senin N, Maskery I, Leach RK (2018) Effects of Magnification and [308] Yadroitsava I, Yadroitsev I (2015) Residual Stress in Metal Specimens Pro-
Sampling Resolution in X-Ray Computed Tomography for the Measurement duced by Direct Metal Laser Sintering. Proceeding of SFF Symposium 614–625.
of Additively Manufactured Metal Surfaces. Precision Engineering 53:54–64. [309] Yadroitsev I, Gusarov A, Yadroitsava I, Smurov I (2010) Single Track Formation
[277] Todhunter LD, Leach RK, Lawes SDA, Blateyron F (2017) Industrial Survey of in Selective Laser Melting of Metal Powders. Journal of Materials Processing
ISO Surface Texture Parameters. CIRP Journal of Manufacturing Science and Technology 201:1624–1631.
Technology 19:84–92. [310] Yadroitsev I, Surov I (2010) Selective Laser Melting Technology: From the
[278] Tolochko NK, Mozzharov SE, Yadroitsev IA, Laoui T, Froyen L, Titov VI, Ignatiev Single Laser Melted Track Stability to 3D Parts of Complex Shape. Physics
MB (2004) Balling Processes During Selective Laser Treatment of Powders. Procedia 5:551–560.
Rapid Prototyping Journal 10:78–87. [311] Yang S, Zhao YF (2015) Additive Manufacturing-enabled Design Theory and
[279] Townsend A, Pagani L, Blunt LA, Scott PJ, Jiang X (2017) Factors Affecting the Methodology: A Critical Review. The International Journal of Advanced
Accuracy of Areal Surface Texture Data Extraction from X-ray CT. CIRP Annals Manufacturing Technology 80:327–342.
66:547–550. [312] Yasa E, Kruth J-P, Deckers J (2011) Manufacturing by Combining Selective
[280] Townsend A, Pagani L, Scott P, Blunt LA (2017) Areal Surface Texture Data Laser Melting and Selective Laser Erosion/Laser Re-melting. CIRP Annals
Extraction from X-Ray Computed Tomography Reconstructions of Metal 60:263–266.
Additively Manufactured Parts. Precision Engineering 48:254–264. [313] Zaeh MF, Branner G (2010) Investigations on Residual Stresses and Deforma-
[281] Townsend A, Racasan R, Blunt L (2018) Surface-Specific Additive Manufactur- tions in Selective Laser Melting. Production Engineering 4:35–45.
ing Test Artefacts. Surface Topography: Metrology and Properties 6024007. [314] Zang EE (1940) Vitavue Relief Model Technique US Patent 3,137,080.
[282] Townsend A, Senin N, Blunt L, Leach RK, Taylor JS (2016) Surface Texture [315] Zanini F, Carmignato S (2017) Two-spheres Method for Evaluating the Met-
Metrology for Metal Additive Manufacturing: A Review. Precision Engineering rological Structural Resolution in Dimensional Computed Tomography. Mea-
46:34–47. surement Science & Technology 28114002.
[283] Trapp J, Rubenchik AM, Guss G, Matthews MJ (2017) In Situ Absorptivity [316] Zanini F, Carmignato S (2018) X-ray Computed Tomography for Measurement
Measurements of Metallic Powders Dusing Laser Powder-Bed Fusion Addi- of Additively Manufactured Metal Threaded Parts. Proceedinf of ASPE/euspen
tive Manufacturing. Applied Materials Today 9:341–349. Summer Topical Meeting.
[284] Triantaphyllou A, Giusca CL, MaCaulay GD, Leach RK, Tomita B, Milne KA [317] Zanini F, Pagani L, Savio E, Carmignato S (2019) Characterisation of Additively
(2015) Surface Texture Measurement For Additive Manufacturing. Surface Manufactured Metal Surfaces by Means of X-Ray Computed Tomography And
Topography: Metrology and Properties 3024002. Generalised Surface Texture Parameters. CIRP Annals. in press.
[285] Turner BN, Gold SA (2013) A Review of Melt Extrusion Additive Manufactur- [318] Zekavat AR, Jansson A, Gundlack C, Pejryd L (2018) Effect of X-Ray Computed
ing Processes: II. Materials, Dimensional Accuracy, and Surface Roughness. Tomography Magnification on Surface Morphology Investigation of Additive
Rapid Prototyping Journal 21:250–261. Manufacturing Surfaces. Proceeding of 8th iCT.
700 R.K. Leach et al. / CIRP Annals - Manufacturing Technology 68 (2019) 677–700

[319] Zekovic S, Dwivedi R, Kovacevic R (2005) Thermo-structural Finite Element [321] Zhang B, Ziegert J, Farahi F, Davies A (2016) In Situ Surface Topography of
Analysis of Direct Laser Metal Deposited Thin-Walled Structures. Proceeding Laser Powder Bed Fusion Using Fringe Projection. Additive Manufacturing
of SFF Symposium 338–355. 12:100–107.
[320] Zhang B, Duemmler M, Sripragash L, Adewumi O, Davies A, Zeigart J, Waters [322] Zhao C, Fezzaa K, Cunningham RW, Wen H, De Carlo F, Chen L, Rollett AD, Sun
C, Evans C (2018) Detection of subsurface defects for metal additive T (2017) Real-time Monitoring of Laser Powder Bed Fusion Process Using
manufacturing using flash thermography. Proceeding of ASPE/euspen);(July). High-speed X-ray Imaging and Diffraction. Scientific Reports 7:3602.

You might also like