1 s2.0 S0736584520302374 Main1

See discussions, stats, and author profiles for this publication at: https://www.researchgate.
net/publication/342943093
A big data-driven framework for sustainable and smart additive

manufacturing
Article in Robotics and Computer-Integrated Manufacturing · February 2021

DOI: 10.1016/j.rcim.2020.102026
CITATIONS READS
11 1,023
7 authors, including:
Arfan Majeed Yingfeng Zhang

Northwestern Polytechnical University 155 PUBLICATIONS 3,301 CITATIONS
25 PUBLICATIONS 165 CITATIONS
SEE PROFILE
SEE PROFILE
Jingxiang Lv Tao Peng

Northwestern Polytechnical University Zhejiang University
44 PUBLICATIONS 1,010 CITATIONS 69 PUBLICATIONS 838 CITATIONS
SEE PROFILE SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Call for Papers: Special Issue on Smart Product Service Systems, IET Collaborative Intelligent Manufacturing View project
Industrial Artificial Intelligence for Digital Servitization View project
All content following this page was uploaded by Saad Waqar on 06 October 2020.
The user has requested enhancement of the downloaded file.

Robotics and Computer Integrated Manufacturing 67 (2021) 102026
Contents lists available at ScienceDirect
Robotics and Computer Integrated Manufacturing

journal homepage: www.elsevier.com/locate/rcim
A big data-driven framework for sustainable and smart additive T

manufacturing
⁎
Arfan Majeeda, Yingfeng Zhanga,g, , Shan Rena,b, Jingxiang Lvc, Tao Pengd, Saad Waqare,
Enhuai Yinf
a
Key Laboratory of Industrial Engineering and Intelligent Manufacturing, Ministry of Industry and Information Technology, Northwestern Polytechnical University, Shaanxi
710072, PR China
b
School of Modern Post, Xi'an University of Posts and Telecommunications, Shaanxi 710061, PR China
c
Key Laboratory of Road Construction Technology and Equipment, Ministry of Education, School of Construction Machinery, Chang'an University, Xi'an 710064, Shaanxi,
China
d
Department Key Laboratory of 3D Printing Process and Equipment of Zhejiang Province, Institute of Industrial Engineering, School of Mechanical Engineering, Zhejiang
University, Hangzhou 310027, China
e
School of Mechanical Engineering, Shandong University, Jinan 250061, China
f
Xi'an Research Institute of Navigation Technology, China Electronics Technology Group Corporation, Xi'an 710068, China
g
School of Mechanical Engineering, Shaanxi University of Technology, Shaanxi 723001, PR China
A R T I C LE I N FO A B S T R A C T
Keywords: From the last decade, additive manufacturing (AM) has been evolving speedily and has revealed the great
Big data potential for energy-saving and cleaner environmental production due to a reduction in material and resource
Additive manufacturing consumption and other tooling requirements. In this modern era, with the advancements in manufacturing
Sustainable manufacturing technologies, academia and industry have been given more interest in smart manufacturing for taking benefits
Smart manufacturing
for making their production more sustainable and effective. In the present study, the significant techniques of
Optimization
smart manufacturing, sustainable manufacturing, and additive manufacturing are combined to make a unified
term of sustainable and smart additive manufacturing (SSAM). The paper aims to develop framework by com-
bining big data analytics, additive manufacturing, and sustainable smart manufacturing technologies which is
beneficial to the additive manufacturing enterprises. So, a framework of big data-driven sustainable and smart
additive manufacturing (BD-SSAM) is proposed which helped AM industry leaders to make better decisions for
the beginning of life (BOL) stage of product life cycle. Finally, an application scenario of the additive manu-
facturing industry was presented to demonstrate the proposed framework. The proposed framework is im-
plemented on the BOL stage of product lifecycle due to limitation of available resources and for fabrication of
AlSi10Mg alloy components by using selective laser melting (SLM) technique of AM. The results indicate that
energy consumption and quality of the product are adequately controlled which is helpful for smart sustainable
manufacturing, emission reduction, and cleaner production.
1. Introduction Artificial Intelligence (AI), Cyber-Physical System (CPS) [3], Big Data
Analytics (BDA) [4], Cloud computing and manufacturing [5], Digital
Nowadays, sustainable manufacturing is a more competitive ap- Twin (DT) [6], 5G [7], etc. [8] have significantly strengthened the
proach for manufacturing enterprises as its execution can support progress of smart manufacturing. Smart manufacturing can make the
producers to accomplish complete development plans, decrease re- industry more sustainable, productive, and profitable [9].
source consumption and pollution along the entire lifecycle [1]. In this Additive manufacturing (AM) is an evolving technology for today's
advanced production era, the industry and academia have been dis- manufacturing enterprises [10], [11]. AM is categorized according to
cussing and focusing on the implementation of smart manufacturing in various material states, which include liquid, powder, wire and fused
their arena of research and manufacturing. The recent improvements in material [12], [13]. When categorized by considering materials, a
the smart enabling technologies, such as Internet of Things (IoT) [2], variety of polymers [14], ceramics [15], metals and alloys [16], airy
⁎
Corresponding author.
E-mail address: zhangyf@nwpu.edu.cn (Y. Zhang).
https://doi.org/10.1016/j.rcim.2020.102026
Received 31 March 2019; Received in revised form 24 February 2020; Accepted 21 June 2020
0736-5845/ © 2020 Elsevier Ltd. All rights reserved.
A. Majeed, et al. Robotics and Computer Integrated Manufacturing 67 (2021) 102026
Abbreviations ERP enterprise resource management

FDM fused Deposition Modeling
3D three dimensional GA genetic Algorithm
5G 5th generation technology IIoT industrial internet of things
3DP 3D printing IoT internet of things
AA artificial aging MES manufacturing execution systems
AB as-built MOL middle of life
AI artificial intelligence OM optical microscope/ microscopic
AM additive manufacturing PDKM product data & knowledge management
ANN artificial neural networks PDM product data management
ASTM American Society of Testing and materials RFID radio frequency identification
BDA big data analytics SAM sustainability of additive manufacturing
CAD computer aided design SCM supply chain management
CAE computer aided engineering SEC specific energy consumption
CAM computer aided manufacturing SEM scanning Electron Microscope
CAPP computer-aided process planning SHT solution heat treatment
CC cloud computing SL/SLA stereolithography
CM conventional manufacturing SLM selective Laser Melting
CP cleaner production SLS selective laser sintering
CPS cyber-physical systems SM smart manufacturing
DM data mining SSAM sustainable and smart additive manufacturing
DMLS direct metal laser sintering SSM sustainable smart manufacturing
DSS decision support systems SVM support vector machine
DT digital Twin UHF ultra-high frequency
EISs enterprise information systems WIP work in process
EOL end of life
structures [17], composites [18], functionally-graded materials [19] fly ratios. AM can be applied anywhere in the product lifecycle [21].
and multi-phase materials [20] are used for various type of AM pro- AM is claimed as green technology which holds a lot of potential in
cesses. With a growing emphasis on sustainability and cleaner pro- enhancing effectiveness of materials, dropping life cycle effects, and
duction, AM proposes a significant worth in relation to decrease buy-to- permitting better engineering functionality correlated to traditional
Fig. 1. A route map of sustainable manufacturing milestones [10].
2
approaches, with a minimum requirement of specialized tooling in the 2. Literature review

fabrication of a product, quick tooling manufacture, and material wa-
stage is also reduced [22]. Furthermore, AM has reduced cost and There are five subsections which are categorized as follow:
processing time for customized and low-volume components produc-
tion. • Additive manufacturing and its qualification.
Smart manufacturing (SM) is an innovative, service-oriented, and • Sustainability of additive manufacturing.
networked manufacturing model, which progressed from, but extends • Smart manufacturing and its key enabling technologies.
beyond, conventional manufacturing (CM) and service methods, and • Big data and its architecture in smart manufacturing.
integrates various cutting-edge technologies such as IIoT, AI, BDA, CPS, • Research Gaps.
CC and DM [23]. In the SM environment, there are lots of monitoring
and control points organize along entire manufacturing processes, like
from delivery of raw materials to shop-floor to final delivery and 2.1. Additive manufacturing and its qualification
packaging of the products [1].
Consequently, a vast amount of manufacture data is produced and AM was established about three decades ago, alias as rapid proto-
collected. Producers can investigate these data by using BDA to enhance typing (RP), additive layer manufacturing, layer-based manufacturing
the performance of manufacturing and management of the whole pro- (LBM), stereolithography (SL), 3D printing, etc. [30]. ASTM
duction processes for complex products, such as optimizing process characterized AM into seven categories as vat photopolymerization,
parameters, reducing process flaws, improving product quality and material jetting, binder jetting, powder bed fusion, material extrusion,
productivity, etc. Such managerial improvement and optimization may directed energy deposition and sheet lamination [30], [31].
significantly contribute to reducing energy consumption, material The quality of AM products is characterized by dimensional and
waste, carbon emission, and environmental impact [24–27]. Currently, surface quality, mechanical properties, hardness, densification, and
theories and applications of the innovative ways of manufacturing can residual stress behavior. Numerous research work is done on the qua-
be further studied and analyzed for better resource efficiency, e.g., lification of different materials components by applying different kinds
collaborative manufacturing, AM, green supply chain, etc. of AM technologies as per requirement. Previous researchers’ studies
The product lifecycle management (PLM) comprises three stages, have shown that for metal AM processes, appropriate combinations of
which are beginning of life (BOL), middle of life (MOL), and end of life process parameters such as scan speed, laser power, hatch distance,
(EOL). Majeed, et al. [28] proposed a BDA framework for process layer thickness, etc. influence on above-mentioned quality character-
analysis and optimization of AM and implemented it on the BOL phase istics. Most of the research is focused on the manufacturing of fully
of AM. Brundage, et al. [29] analyzed environmental sustainability dense parts; it means that the relative density is nearly equal to 100%.
calculation approaches to enable more precise decisions earlier in de- But it is difficult to achieve almost 100% or above 99% relative density
sign phase of product life cycle. As previously discussed, AM and big in every material in every AM technique [13]. So, the most commonly
data both dealt separately in the manufacturing enterprises. There is used AM technique in which the relative density is higher than 99% is
minimal work done in the combined form. SM and sustainable manu- selective laser melting (SLM), which belongs to powder bed fusion
facturing are also a niche for AM industries. The sustainability in the (PBF) technology of metal AM.
manufacturing enterprises was started about four decades ago, and the Mumtaz and Hopkinson [32] investigated that the high scan powers
route of sustainability for the subtractive and additive manufacturing is persuaded to reduce surface quality, and comparatively low scanning
shown in Fig. 1 [10]. speed is mostly preferred because of the longtime permits the melt pool
It can be seen that manufacturing enterprises identify the sustain- to restructure and attain a smooth surface. Islam, et al. [33] presented a
ability issue after a century of industrialization. After recognition, many comparative experimental study on the dimensional accuracies of fab-
researchers have worked on sustainable manufacturing. With the de- ricated parts of two broadly used AM techniques and the results ex-
velopment in the AM technology, the efforts on the implementation and posed that in both processes, the primary reason of dimensional dis-
adoption of sustainability of additive manufacturing (SAM) in AM en- similarities was the volumetric changing of the process, and SLA
terprises have also been expedited. So, the AM leaders are interested in accuracy was better than PBP produced parts. Wang, et al. [34] ex-
benefiting their companies by effectively using SAM. plained that scanning track characteristics were very critical for at-
In the present paper, big data analytics, additive manufacturing, taining a successful surface finish on the SLM products. Relvas, et al.
and sustainable smart manufacturing (SSM) have combined to form a
new interdisciplinary research area, namely big data-driven sustainable
and smart additive manufacturing (BD-SSAM), as shown in Fig. 2.
Therefore, a framework of big data-driven sustainable and smart ad-
ditive manufacturing (BD-SSAM) is proposed, which is an initial step for
the development of smart manufacturing environment in the additive
manufacturing enterprises. This framework is applied for the BOL phase
of whole product life cycle of AM process. Also, very supportive of AM
enterprises and produces products with energy-efficient and working on
cleaner production strategies for the environment.
The paper structure is mentioned as follows. In Section 2, a detailed
literature survey is conducted for BDA, AM, and SSM. In Section 3, the
overall framework of BD-SSAM is established. Section 4 describes the
key enabling technologies of SSAM. In Section 5, the benefits of the
framework are briefly mentioned. In Section 6, an application scenario
of the additive manufacturing industry is presented. Section 7 explains
the results and discussions of the case study. The managerial implica-
tions of the framework are presented in Section 8, and in Section 9,
conclusions with future research direction are explained.
Fig. 2. Combinations of research of BDA, AM, and SSM.
3
[35] investigated a comparison study on the 4 AM processes SLA, SLS, main aspects which are, resource consumption, management of waste
FDM, and 3DP for the dimensional and geometrical accuracy, and and controlling pollution which are shown in Fig. 3.
concluded that the worst dimensional performance of all the AM pro- Material consumption and energy consumption are mainly related
cesses was from 3DP, but the geometrical performance of 3DP com- to the resource consumption of the AM industry. In comparison to CM,
pared to other processes was in favorable condition. the AM processes material efficiency is better, but the energy con-
Calignano, et al. [36] investigated the influence of main process sumption is more due to low productivity and other accessories re-
parameters on surface roughness of DMLS, built AlSi10Mg alloy spe- quirements of different AM processes [40]. Peng et al. [10] studied the
cimens, and analyzed that scan speed had the highest impact on surface SAM, and discussed the SAM with the main concentration on energy
roughness, also described that the surface roughness could be mini- and environmental effects, and also predicted that the most crucial
mized by shot peening. Delgado, et al. [37] evaluated DMLS and SLM feature for SAM is resource consumption. By the implementation of AM
built parts mechanical properties and resulted that part mechanical techniques, there is a considerable saving of material by reducing
performance is inversely proportional to layer thickness. waste; sometimes it reaches more than 90%. The pollution is also
controlled by applying AM technologies because there are no cutting
fluids, lubricants, etc. used by AM processes, which make the en-
2.2. Sustainability and additive manufacturing
vironment clean and step to green production. From the life cycle
perspective of AM products, there is very little work done, and there is a
With the progress in sustainable manufacturing, the sustainable
requirement to work on the long-term influence of AM on the en-
development of the AM industry is also vital, and researchers have in-
vironment [41].
fluenced the SAM for implementation and adoption in the AM en-
Energy consumption is a critical approach to calculate the sustain-
terprises [10]. The development route of sustainability for subtractive
ability of AM technology [42], [43]. Huang et al. [44] calculated the
and additive Manufacturing industries is shown in Fig. 1. Previous re-
life-cycle energy and greenhouse gas emissions savings possibility of
searchers have correlated the sustainability of AM or other processes
AM techniques for aerospace components. A new methodology of multi-
with energy consumption and its environmental effects [38]. Jawahir,
flow consumption of materials, fluids, and energy was investigated by
et al. [39] presented the 6R concept (i.e. reduce, redesign, recover,
Le Bourhis et al. [45] on a direct laser solid forming (DLSF) technique of
remanufacture, recycle and reuse) for three main dimensions of sus-
AM, and the flow of energy is based on the energy consumption during
tainability, which are environmental, society and economy as shown in
laser system, cooling system, and motor drives. Ford and Despeisse [46]
Fig. 3. The environmental effect of AM is further divided into three
Fig. 3. The three dimensions of SAM [10].
4
investigated the implementation of AM from the perspective of the life • 5G networking is a new emerging technology for the data-driven
cycle and concluded that sustainability benefited in the redesigning of industries, SM, smart cities and infrastructure management because
product, processing of input material, manufacturing of the product, it will be possible to have many more devices working, reliably,
and product reuse by remanufacturing of the AM industry. Gebler et al. securely and uninterrupted in the similar zone [59]. 5G has several
[47] analyzed the life cycle perspective of AM and resulted that AM has benefits such as higher speeds, less latency, capacity for a larger
potential for reduction in costs, energy, and CO2 emissions, and also number of connected devices, less interference and better efficiency
predicted that cost saving of $113 ~ $370 billion till 2025 due to re- [7]. As discussed above, it is very beneficial for the big data and
duction in material consumption with reduction in the supply chains. internet of things. For example, with 5G technology, more manu-
Mani et al. [48] and Chen et al. [49] defined the potential benefits of facturing resources and devices can be connected and to carry out
sustainability of AM which are reduce wastage during production, op- large scale communication. This is benefit to promote information
timization of products and creation of lightweight structures, reduction and knowledge share among heterogeneous manufacturing re-
in material and energy consumption, reduction in the supply chain, sources and thereby to optimize the manufacturing service abilities
reduction in inventory management, etc. Wits et al. [21] investigated that involved in the proposed framework. Meanwhile, the ad-
the sustainability of AM from end-user perspectives and concluded that vantages of 5G technology on higher reliability and less latency are
AM is a more sustainable substitute from maintenance, repair, overhaul benefit to improve the real-time operation and control as well as
(MRO) or replace point of view. decision-making capabilities of SSAM processes.
2.3. Smart manufacturing and its key enabling technologies 2.4. Big data and its architecture in smart manufacturing
SM is adopted by different industries and manufactures, and es- Laney [60] has defined the big data on the theory of 3Vs (i.e. Vo-
tablished its significant enabling technologies which are utilized to lume, Variety, and Velocity), which means that a great volume of data
report data acquisition, storage, communication, processing, analysis, is generated and collected, the speed of data collection and analysis is
the discovery of knowledge and pattern [1]. The sustainability and high, and various types of semi-structured and unstructured data are
proficiency of the SM systems are enhanced with the provision of va- received for final processing. Big data is classified according to different
luable and reliable data by using these enabling technologies of SM. life cycle stages, which are BOL, MOL, EOL, on the systematic changing
Some of the key enabling SM technologies are described as follows: like static and dynamic, and also categorized on a semi-structured,
structured, and unstructured basis. Recently, Ren et al. [1] conducted a
• Abell et al. [50] has defined that BDA is the method of investigative big data analytics review to support sustainable smart manufacturing
vast and diverse data groups to expose hidden correlations, un- for the whole lifecycle and proposed a framework to deal with the fu-
known patterns, customer likings, market trends and further valu- ture challenges of smart manufacturing. Jiang et al. [61] used big data
able material which can support organizations to enhance sustain- based neural network to predict the printable bridge length in additive
ability, make more-informed business decisions and to initiative the manufacturing for reducing support material waste.
society on the way to the globular economy and also do sound With the developments in the BDA, it has been applied by many
business decisions. Gunasekaran et al. [51] studied the influence of companies to improve their production, maintenance, services, etc.
BDA on sustainable and green SCM for the improvement of orga- Leitão et al. [62] described that challenges of the industrial and man-
nizational performance. Tao et al. [52] designed a fiber channel (FC) ufacturing automation arena would be efficiently handled by applying
switch based on field programmable gate array (FPGA) and applied BDA due to its ability of management of enormous volume of rapidly
due to its high speed, low latency, and high-performance transmis- created data. Siemens [63] implemented big data for remote diag-
sion capacities. The FC switch higher capacity of communicating nostics services to investigate the operational behaviors of their power
and processing big data opens a bright perspective for SM. plants from all over the world by taking 100,000 measurements.
• Data Mining (DM) is extensively used in manufacturing and service
industries for making decisions on the available data, and it is also 2.5. Research gaps
applied in a sustainable manufacturing environment. Köksal et al.
[53] studied and analyzed the applications of DM for improving the From the literature review, there is significant progress in the field
manufacturing and product quality. With the development in DM of AM, SM, Sustainable manufacturing, and BDA, but all the fields were
technologies, it is executed in various phases of the lifecycle, like discussed or investigated separately. Based on this, there are following
designing of product, production, maintenance, service, recycling, research gaps for SSAM which needs to be examined:
etc. [54].
• For the advanced sustainable manufacturing and SM, the IoT sys- • A big data-driven framework in a combined form of additive man-
tems are chiefly employed to monitor energy consumption, emis- ufacturing and smart manufacturing is not available or discuss
sions-reduction, and improving the efficiency of recycling during the previously, because both fields have been studied in parallel. So,
life cycle [55]. Shrouf and Miragliotta [56] established an energy there is a need to develop a framework that jointly works and how
management system on IoT-based and concluded that this system is to establish a framework for BD-SSAM?
beneficial to manufacturing company managers for making deci- • In the SM environment for AM, an enormous amount of process
sions, and also helpful for sustainable manufacturing and energy control and performance of product data is produced, and the most
management. vital is to extract valuable information from the big data, which can
• Lee et al. [57] developed a CPS framework in Industry 4.0 system be possible by applying BDA in Smart AM environment.
which was utilized in the production line of machine tools for in-
tegration of CPS in SM and concluded that CPS could be im- 3. Big data analytics framework for sustainable and smart
plemented for improving product quality and reliability of the additive manufacturing
system, reducing production downtime and optimizing production
planning and inventory management. Song and Moon [58] devel- With the developments in the information and manufacturing
oped an architecture of cyber manufacturing systems (CMS) by technologies, there is a massive amount of data generated during AM
using CPS methodology and concluded that CMS has potential sus- processes, which becomes a substantial challenge for conventional ar-
tainability benefits in comparison to conventional manufacturing chitecture to handle it. With time, the requirement of sustainable and
systems. smart additive manufacturing is developing, and there is a need for
5
particular architecture which will help the AM companies’ engineers 3.1. Perception and acquisition of big data for SSAM
and managers to improve their capability with the growing world
standards and to compete with the market demands. Considering the It is the main layer of the framework, as shown on the top left of
previous literature survey and discussion, the role of BDA is very sig- Fig. 4. Firstly, the IoT devices such as smart devices, RFID readers, RFID
nificant and vital for manufacturing enterprises like additive manu- tags, smart sensors, etc. [66] are configured on the whole product
facturing [64] due to its ability of collection and storage of data, pre- manufacturing cycle (i.e. smart design, smart production, smart main-
paration, cleaning, reduction, integration and transformation of data tenance and services, smart delivery) of the smart AM environment.
and the more vital phase of big data is the data mining. Then, decision Then, the huge heterogeneous and multi-source data of AM such as
making is performed from the extracted knowledge of big data [65]. product planning and design, material and procurement, AM systems,
Based on the presented overview, an overall framework of BD-SSAM AM production control and status, product qualification, energy con-
is designed, which is shown in Fig. 4. By using the framework, real-time sumption, products delivery, and customer feedback, maintenance and
and non-real time data of the BOL stage of product life cycle of AM is services, products recovery, etc. are sensed and captured for further
determinedly monitored, controlled and captured. By the usage of evaluation. The smart sensors and calibration devices are used to con-
communication and network technologies, the gathered data can be trol and monitor the AM system working and manufacturing of quality
transferred and stored in databases. Furthermore, reliable and available products according to customer requirements. The product quality data
data after preprocessing is provided for supporting data mining and is also monitored and collected at each stage of production to the as-
decision making. There are mainly four phases of the developed fra- sembly phase. Then, the standard communication procedures (i.e.
mework which work in a closed-loop and interconnected with each Modbus, intranet, Internet RS- 485/323, 5G, etc.) [66] are utilized to
other. The details are mentioned in the upcoming subsections. transfer the large captured data for further processing in the next
Fig. 4. A BDA framework for SSAM.
6
working layer of the framework. that the SSAM big data could be integrated by pre-defined logics [71].
The established data warehouses are used to store and manage the
3.2. Storage, preprocessing, integration and management of SSAM big data manufacturing cycle data cubes, and to describe the complex logic re-
lationship among massive manufacturing cycle data cubes. Next, the
The real-time and non-real-time data of the entire product manu- generated manufacturing cycle data cubes have a great number of re-
facturing cycle of AM enterprises is collected, which composes of a dundancies, which is reduced by data cleansing operation. The general
great amount of unstructured, structured and semi-structured data models which are applied for data cleaning are information structure
[67]. The conventional data storage and management tools and tech- model [72] and RFID-Cuboid model [73]. During cleaning operation,
nologies are not enough to further process and handle the too huge, set of manufacturing cycle data cube from manufacturing cycle data
complex and complicated data effectively. Consequently, this layer is warehouse is the input which gives the output in an organized set of
supportive for the further proceeding of the storage disordered and manufacturing cycle data cube that transmits significant information
huge datasets or data cubes by using Extensible Markup Language about manufacturing cycle status.
(XML), Not only Structured Query Language (NoSQL) and Distributed Moreover, meta-models are developed to integrate the huge data
Database system (DDBS)[68]. For the non-real-time product manu- which consists of design, market, production, maintenance, dimen-
facturing cycle data processing, the Hadoop computing framework is sional quality, product delivery, logistics, etc. [74]. The integrated data
applied [68]. The real-time computing framework Storm is applied for sets are obviously still massive, which may be infeasible for data ana-
heavy real-time processing of AM product manufacturing cycle data lyses. Consequently, a data reduction process is made to acquire a re-
[69], [70]. In this layer, the flow of data for preprocessing and man- duced illustration of the data set that is much smaller in volume, but it
agement is followed by the sequence mentioned below and also shown must hold the veracity of the original data. Furthermore, the concise
in Fig. 4 (bottom left): manufacturing cycle data is transformed so that the subsequent data
mining method may be more effective, and the patterns found may be
I Data storage method easier to recognize [74]. Finally, the preprocessed SSAM big data are
II Data Cleaning method stored in various above-mentioned data management systems such as
III Data integration method DDBS for further utilization in decision making.
IV Data reduction method
V Data transformation method 3.3. Data mining and decision-making of SSAM big data
Initially, AM product manufacturing cycle data are chosen to form a Throughout the entire product manufacturing cycle of AM en-
data cube, and to construct a manufacturing cycle data warehouse so terprises, a huge amount of data is generated in the form of 4Vs
Fig. 5. Big data perception and acquisition framework of product manufacturing cycle for SSAM.
7
(volume, velocity, variety, and value), which is challenging to in-
To monitor the electrical energy consumption information of AM machines, compressors, pumps, etc.
vestigate it by using the traditional methods of analysis. In this fra-
Monitoring temperature data of machine, processing bed of product, and also the workshop work.
mework, an impressive technology that has potential to determine
hidden knowledge from the large AM product data sets is applied,
To identify and track the critical components of the overall product manufacturing cycle.
which is data mining models (e.g. clustering, classification, association,
neural network, prediction, etc.) [66]. By combining the AM product
To trace and track the real-time data of product, manufacturing process, usage, etc.
To monitor the pressure data of machines, product manufacturing, air supply, etc.
processing analysis approaches and methods of big data mining [66],
valued information and knowledge can be revealed from these AM
To trace and monitor real-time data information of drawings and material.

product manufacturing cycle data sets. From these mined results, smart
and sustainable AM product decision makings for application services
will be provided to AM enterprise managers [66].
Purpose/ Objective
3.4. SSAM big data application services
This application services layer can be seen from the top right of
Fig. 4; big data application services are applied to deliver significant
real-time and non-real-time applications based on mined information
and knowledge for end-users [66], [74]. For the various manufacturing
cycle phases, numerous forms of application services are planned in this
layer of the framework. Specifically, sustainable and smart product
design, smart shop floor scheduling, optimization of AM process, AM
processing parameters optimization, reduction in energy consumption,
logistics optimization, sustainable and smart maintenance, predictive
maintenance, smart delivery, etc. are designed for SSAM, which is fa-
vorable for promoting CP strategy in the sustainable and smart AM
environment. By the implementation of smart decision making and real-
time feedback, the services mentioned above are applied in the AM
enterprises for efficient and sustainable smart production.
4. Key technologies for big data analytics sustainable and smart Product/ tools/AM machines/ pallet/ operators/ AGV/ QC document/etc.
AM
Big data of SSAM plays a vital role in the whole product life cycle of
Information of configuration of IoT devices for the data acquisition of additive manufacturing environment.
AM enterprises. In the BOL, it is an important asset for innovative

products, and it can also reduce waste and emissions with the EOL
decision making which means that the product can be remanufactured
and reused. Though it is problematic to capture overall and real-time
data of product life cycle, particularly products’ data after being dis-
tributed to customers [75].
Material /Product/ maintenance/ etc.
4.1. Big data perception and acquisition for SSAM

Monitoring Resources/Location
Drawings/ Models/Material
An overall framework model for heterogeneous and real-time pro-

AM machines/ workshop
AM machine/ workshop
Machines/ Compressors
duct manufacturing cycle, big data perception, and acquisition of SSAM

are developed which is shown in Fig. 5 [75]. The configurations of
several IoT and smart devices are the fundamentals for gathering het-
erogeneous, real-time and multisource data of fabricated products and
things of SSAM environment (Table 1). During the entire product
manufacturing cycle of SSAM process, IoT and smart devices are posi-
tioned at the manufacturing resources and main areas of the products in
the machine, workshop, and factory level (see in Fig. 5) [66].
Voltage and current sensors
Ultra-high frequency (UHF)
Primarily, the RFID tags are configured adequately on the technical

documents and reports of quality, design, production of products and
Temperature sensor
Type of IoT device
maintenance of machines which play significant roles throughout the

Pressure sensors
product manufacturing cycle. Moreover, the RFID tags are configured

on the parts design and models for accurate tracking. RFID tags are also
deployed to different types of AM materials like powder, wire, and li-
UHF
UHF
quid which are used at different AM systems. RFID readers are installed
at different locations of the entrance of factory, CAD rooms, material
RFID Reader/ Scanner
handling, warehouse, tool stores, laboratories, etc. The AM machines

are also properly tagged.
The energy consumption data capturing is also vital for the SSAM
Smart sensors
Smart Sensor
Smart sensor
IoT devices
and CP strategy. The energy consumption can be measured from the

RFID Tags
RFID Tags
power consumption of the AM systems which is equivalent to the

Table 1
product of real time electrical current and voltage. Different smart

sensors and smart meters (i.e. temperature sensors, pressure sensors,
8
current sensors, voltage sensors, vibration sensors, etc.) are applied to to enterprises databases.
monitor and collect big data of temperature, pressure, energy con-
sumption, etc., during the whole production process. The log of the data 4.2. Big data mining and knowledge sharing for SSAM
is maintained in the AM systems for the each second during the AM
processing for building of a product and it is also displayed on the Fig. 6 is the closed-loop big data mining and knowledge sharing
control panel on machine. By the IoT devices and sensors, the tem- structure of SSAM, which comprises of the data and knowledge layer,
perature, gas, pressure, etc., data is continuously monitored and col- model layer, and objective layer.
lected.
Furthermore, the maintenance data of AM machines during the 4.2.1. Data and knowledge layer
product manufacturing process is also recorded by suitable smart sen- Data and knowledge layer contain heterogeneous, multi-source,
sors and tags (see Fig. 4). Due to the implementation of the SSAM en- real-time, non-real-time, and knowledge data of product manufacturing
vironment, some maintenance and malfunctioning issues of AM systems cycle of SSAM which has four dimensions which are as follows:
are solved by the online internet or other sources like cloud computing.
The RFID tags and readers are also configured on the product inventory, • The dimension of data type contains the data of whole product
delivery and logistics sections for making the smart delivery. Moreover, manufacturing cycle for the SSAM, such as AM design data, material
through the standard communication protocols [66], such as RFID, RS- data, AM process data, market analysis data, customers’ demand
232/485, Internet, Intranet, WLAN, Modbus, 5G, etc. technologies, the data, maintenance data, quality data, inventory data, delivery data,
collected multi-source, non-real-time, and real-time data is transferred energy data, etc.
Fig. 6. Big data mining and knowledge sharing structure for SSAM.
9
• The dimension of data storage consists of various database en- 5. Benefits of SSAM
terprises which are used for storage of the heterogeneous data,
structured data, semi-structured data, and unstructured data, such The leading-edge technologies applications are growing con-
as DDBS, RDBMS, XML, NoSQL, etc. tinuously in the current AM environment to make it SSAM, which may
• The third dimension of this layer is data processing. Different data be produced a massive amount of data during AM product designing,
processing computing software is applied for the cleaning, reduc- AM fabrication processes, and maintenance of the product, etc., and it
tion, integration, and transformation of real-time and non-real-time can be composed during the whole product manufacturing cycle. The
captured data [66]. The mainly applied computing systems are following benefits can be extracted by the customers and manufacturers
Hadoop, Storm, MapReduce, etc.
• The final dimension of this layer is the knowledge data, which is 5.1. Market demands perception and prediction
very helpful for deciding on the available data from the previous
knowledge sharing. This dimension contains data of design optimi- With the revolution in manufacturing style from mass to customized
zation, optimized processing parameters, quality improvement data, production, determining customer demands and preferences have
energy optimization (reduction in energy consumption and carbon grown gradually significant for the manufacturers [1]. Mourtzis [76]
emission), maintenance data, data relevant to the processing of focused on the aspects that affect manufacturing network performance
different materials in the AM, and other various types of data. for mass customization and also presented the work relevant to the
design, planning, and control of manufacturing networks in the field of
mass customization and personalization.
4.2.2. Model layer AM products demand increased due to its ability to build unique and
The model layer is also identified as a method layer which princi- complex structure parts. Precise perception and prediction of customers'
pally discusses the data mining models. Data mining models can gen- demands and preferences are productive resources for producers to
erally be characterized as descriptive and predictive [53]. Descriptive make their products better fit the requirements of customers, and to
data mining models include summarization, clustering, and association make sophisticated faithfulness and revenue. The huge capacities of
general model. Summarization models used correlation analysis and data associated with customer demands (e.g., customer behaviors and
scatter plots for determining the relationship between the input and evaluations, user feedback, online reviews, and sentiments, etc.) can be
output variables [53]. Association general model can be used to assess gathered and integrated from numerous sources for mining actionable
the AM product design scheme, and to investigate the quality factor of visions by the use of big data analytics. These visions can be applied to
AM product, etc. Predictive DM modeling can be characterized as sta- forecast market demands appropriately, and the probable market
tistical-based (S-based) methods, DT-based algorithms, ANN-based al- magnitude, margin, the number of competitors and the level of dis-
gorithms, classification methods, etc. [53] The prediction general tinction among AM products can also be predicted [1].
model can be used to estimate the maintenance cycle in AM systems.
Furthermore, the other DM models include genetic algorithms (GA), 5.2. Improvement in product design
Bayesian estimation, Apriori algorithms, support vector machine
(SVM), regression analysis, etc. Then an appropriate model of data In the SSAM framework, the remote product manufacturing cycle
mining will be nominated to extract data from the data layer and dis- data that affect product design can be combined and investigated to
covering knowledge from them. create imperative intuitions about enhancements and revolutions in the
AM product [1]. Topology optimized products have also been designed
with minimum material consumption and higher strength. Producers
4.2.3. Objective layer have recognized that BDA is an effective tool for categorizing the
The objective layer is also known as an application or demand layer. hidden requirements and cultivating the efficiency of selection about
On the basis of review of sustainable smart manufacturing [1] and several design substitutes [1]. So, smart products are redesigned ac-
additive manufacturing [10], in present investigation, the manu- cording to the feedback of the customer.
facturing application of the objective layer principally comprises of
improving product design, optimization of processing parameters, im- 5.3. AM product quality improvement
proving productivity, product quality improvement, optimizing AM
process, reducing energy consumption, and improvement in material The real-time data of AM resources (e.g., materials, operators, WIP,
management. This layer also includes improving delivery and shop- etc.) can be monitored by the configuration of smart devices. There are
floor logistics, providing predictive maintenance service, reduction in lots of quality control points organized along the manufacturing line,
carbon emissions, reducing environment effects, etc. and a huge amount of data are generated during the raw materials (i.e.,
Conferring to diverse demands of the objective layer, appropriate powder, wire, etc.) provision to production workshop to the packing of
data mining model, and real-time data of the product manufacturing AM products for final delivery [1]. BDA is applied by AM manufacturers
cycle are carefully chosen to carry out the knowledge discovery [66]. to discover supplementary methods to lessen faults of process and to
For example, the processing parameters can be optimized and predicted rising production, which can be obtained by applying several data
by determining the mining algorithm like a regression, genetic algo- analysis algorithms and models to the manufacturing processes.
rithm or artificial neural network. Moreover, BDA can be utilized to connect the process and equipment
The big data mining and knowledge sharing analysis of the SSAM level data to metrology data to make more precise predictions about
environment illustrate that closed-loop structure begins from applica- production failures [1].
tion and knowledge sharing objectives, and in conclusion, meets ap-
plication objectives [66]. Initially, application objectives are predicted 5.4. Energy consumption control and reduction
in the data and knowledge layer. Then, by different goals or demands,
the adaptive data mining models are designated and developed. The In the current manufacturing scenarios, management of energy and
data mining models have been selected on the main rule which means lessening of emissions are two essential responsibilities for any manu-
the data mining models have abilities of store-memory, evolution, and facturing enterprise (i.e., subtractive or additive manufacturing). With
self-learning. Moreover, appropriate real-time data is extracted to im- the continuous implication of smart devices, meters, and sensors
plement data mining. Lastly, information and knowledge are attained to throughout the entire manufacturing cycle of product and process, huge
meet the application objectives of AM enterprises [28], [66]. size of real-time energy consumption data from manufacturing and
10
operation process can be gathered [28], [66]. The energy consumption and smart AM was done by applying advanced information and man-
data offer a huge potential to reduce energy consumption and to en- ufacturing systems. The performance of the AM processes can be en-
hance the decisions of energy proficiency management. Big data inputs hanced by the execution of proposed BD-SSAM framework, such as
can also help managers to recognize and calculate the wastage points design innovation, reduction in energy consumption, reduction in
and to lessen or disregard them in real-time [1]. processing time with enhancement in productivity, improvement in
product quality with optimizing processing parameters, etc. A full ex-
5.5. Intelligent predictive maintenance and services planation of application study is stated in the forthcoming sections.
Through IIoT, real-time data of the product manufacturing cycle can 6.1. Case description
be collected and evaluated, which would help to improve maintenance
and service decisions [1]. The AM product operation status data is The framework is implemented in a company, which is specialized
collected and communicated to the producer who is a vital asset for in the arena of AM technology and working on the different AM tech-
maintenance decisions, and AM producers can investigate data to es- niques. The case company is Xi'an Ruite 3D Technology Co. Ltd.
timate indicators to control whether system performance is declining. (http://www.xaruite.com/), which has a wide range of AM systems for
These investigations can support producers to precisely forecast when manufacturing of products of metals, polymers, ceramics, etc. The
the products may fail [1]. This may lead to intelligent predictive company is providing facilities of AM to their respected customers from
maintenance and services. various industrial fields like automobiles, aerostructures, biomedicals,
electronics, customer goods, etc.
6. Case study scenario The objective of the case study is to demonstrate an association
between the implementation of SSAM big data analytic practices and
For the implementation of the proposed framework of BDA for economic benefits for the AM company. So, we have considered the
SSAM, a case study of the AM industry is demonstrated for proof of BOL stage (product manufacturing cycle). The company has different
concept. The principal objective of the case study was to show how the AM systems like SLM systems, SLS systems, FDM systems, etc. The
proposed framework benefited the AM enterprises and how sustainable present case study was performed on the SLM 280HL systems, which is
Fig. 7. The overall framework of BD-SSAM for the application scenario.
11
a product of SLM solutions, Germany. mentioned, initially the parameters and their levels were defined by the
The company has many new tasks to accomplished on time to meet preliminary experimentation on the single track, single layer and then
customer requirements. Previously, the company has produced pro- the multi-layer fabrication from the SLM process. Then, the full ex-
ducts of titanium alloys, nickel alloys, steel alloys, etc. The company perimental study flowchart is shown in Fig. 10. The processing condi-
has new tasks of fabricating parts of the pump from AlSi10Mg alloy. So, tions and parameter details are presented in Table 3.
it was also a pleasant task for us to implement our BD-SSAM framework
for manufacturing of AlSi10Mg alloy products in the company which
would be beneficial for it. Previously, the company provides different 6.2.2. Acquisition and sensing of AM big data
products to their customers in a traditional business model like other The AM production setup of the company is shown at the top of
Chinese manufacturing companies. So, the company decided to trans- Fig. 7. It can be observed that manufacturing is done in a closed-loop
form its conventional business model into a sustainable smart manu- structure, and all departments of the company are interconnected with
facturing model, which would be a system integration and service- each other. All departments' data is continuously monitored and con-
driven [1], [77]. trolled by the execution of different IoT devices at different locations
and positions according to their requirements (already discussed in
Section 4.1), and data is also gathered from other sources, which are
6.2. BDA framework of SSAM for the application scenario
mentioned in Table 4. The closed-loop starts with the customers’ de-
mands. Then, the demands come to the production planning and con-
Firstly, the framework of BD-SSAM for the product manufacturing
trol (PPC) department. The PPC department sees the requirement of the
cycle (BOL stage) is developed which is shown in Fig. 7.
customer if the product already manufactured from previous experience
and knowledge available, then the demand is forward to R & D de-
6.2.1. Case study experimental methodology partment for drafting and issue of proper drawings and model, other-
Before going to the framework discussion, it is better to discuss the wise the R & D of the new product required. Then, the models and
experimental methodology of the case study. An SLM 280 HL system is drawings are issued to production department for adequate manu-
used for the manufacturing of products, which is shown in Fig. 8, and facturing of the product.
also an energy monitoring system is shown on the right side of Fig. 8. During the production, the maintenance and services department is
The SLM system is equipped with 02 fiber laser beams of 0.40 kW laser interconnected with the production and other departments for proper
power. The powder material is AlSi10Mg, which is broadly used in tackling of any malfunctioning of AM systems or any other related
automobiles and other industries. The powder morphology of AlSi10Mg equipment like pumps, compressors, electric supply, etc. After the
alloy is shown in Fig. 9, and its chemical composition is mentioned in fabrication of SLM products, it is appropriately cut off from the sub-
Table 2. strate by applying the wire EDM. Then, the different other processes
There are numerous types of processing parameters used in SLM, were performed on the product as per requirement such as sand-
but the most suitable and relevant parameters are laser power, scan blasting, stress-relieving, heat treatment, machining, shot peening, etc.
speed, hatch distance, and layer thickness. The combined effect of these Furthermore, the surface and dimensional quality, strength, etc. of
four processing parameters is called the Energy density, which is de- the product is measured. Then, the qualified product goes to the fin-
noted as ED and described in Eq. (1). ished goods area and finally it would be delivered to the customers.
PL Then, the customer's feedback is taken for further improvement of the
ED = product or service of the company. For the all above discussion, a
(Vs × hd × tl ) (1)
considerable amount of multi-source, real-time, and non-real-time data
where PL is laser power, Vs is scanning speed, hd is hatch distance, and tl is collected, such as design, drawings, production, various processes,
is layer thickness. quality, maintenance, energy consumption, etc.
The suitable combination of these processing parameters is very For the collection of the big data of the whole product manu-
significant for fully dense and good quality products with better facturing cycle, various types of IoT devices like UHF RFID tags, RFID
strength. For this case study, we were started from a single-track, then a reader, smart meters, smart sensors, the voltage sensor, the current
single layer, and finally fabricated bulk samples for parameter optimi- sensor, etc. are configured at various locations (Table 4). A massive
zation. Additionally, the thin-walled specimens were produced for amount of data is also gathered from different external types of
further testing, and finally, the final product is manufactured. As above- equipment such as surface roughness meter, digital vernier caliper,
Fig. 8. Experimental system for fabrication of product and data acquisition.
12
Fig. 9. AlSi10Mg powder; (a) morphology, (b) EDS analysis.
Table 2 non-time data is collected during all samples and products’ manu-
Chemical Composition of AlSi10Mg alloy powder (EDS analysis). facturing. During manufacturing, all malfunctioning data of the SLM
%Al %Si %Mg %Fe %Sn %Mn
system is also gathered. There were also performed different heat
treatment cycles like stress relieving, solution heat treatment, artificial
Bal 11.28 0.49 0.11 0.32 0.12 aging on the test specimens and the final product [78]. All data of the
heat treatments is also stored for future processing. The qualification
data of the test specimens is also stored.
hardness testing machine, tensile testing, SEM, OM, etc. which are It would be observed that key additive manufacturing responses
applied for the qualification of the SLM product (presented in Table 4). such as the densification, porosity, surface roughness, dimensional
Using the devices mentioned above, processing parameters and SLM quality, tensile strength, processing time, energy consumption, etc. are
products' real-time data can be sensed and collected. influenced by processing parameters of laser power, scanning speed,
During the SLM process, the data from different IoT devices and hatch distance and layer thickness. The overall big data is collected for
sensors is monitored on the control panel, and stored in the SLM ma- the a/m responses and processing parameters, which is further in-
chine database system. The data is gathered for every second of SLM tegrated and analyzed for sustainable and smart manufacturing.
processing and continuously monitored on the control panel screen. The
sampling rate of data display on the control panel screen is 5 s. Any
6.2.4. AM big data mining and knowledge sharing
malfunctioning or variations in the sensor data is observed con-
By using the BDA theories, the data mining models are developed,
tinuously, and tackled efficiently. The processing time, and energy
which are applied for attaining hidden knowledge from the enormous
consumption data were also taken and monitored by the smart devices,
real-time and overall manufacturing phase big data. The optimization
which are shown in Fig. 8.
of manufacturing phase of different responses and parameters are also
Also, any malfunctioning in the SLM system parts, e.g. problem in
achieved. By uniting the knowledge with product data & knowledge
powder recoater which not properly placed powder during printing is
management (PDKM) or decision support system (DSS), the objectives
monitored and solved online by smart cloud computing network. It
of the SSAM enterprise can be accomplished, and it can also be applied
means smart maintenance is performed during processing of SLM pro-
to select more environmentally beneficial raw material and cleaner
cess within a very short time.
energy [75].
After the accomplishment of the SLM process, the heat treatment
It can be observed from the bottom right of Fig. 7 that the data
processes have been performed on the fabricated product according to
mining universal and specific models are developed according to sev-
the final requirement and usage. Then, the dimensional and surface
eral requirements of AM enterprises. There are four kinds of general
quality of the SLM manufactured product has performed by applying
models formed in this framework, which are GA, artificial neural net-
various dimensional measurement equipment. All the data of the pro-
work (ANN), Pareto front, regression analysis. The GA model can be
cesses and devices as mentioned earlier have been gathered and sensed.
applied to optimize the processing parameters and also investigate the
improvement in the AM product quality. The ANN general model can be
6.2.3. AM big data storage, management, and processing applied to predict the optimized energy consumption conditions and
A considerable amount of multi-source, real-time, and non-real-time even for the reduction or prediction of maintenance cycle.
data of different data interfaces of design, material, SLM production, Furthermore, different forms of particular data mining models are
quality, maintenance, etc. is communicated and transmitted through also developed, like product design improvement, AM processing
the various data transmitted technologies like Intranet, Wireless, parameters optimization, product quality improvement, energy con-
Internet, etc. Distributed methods are applied for storage and man- sumption optimization, prediction of maintenance, improvement in
agement of the whole AM process data, but here the data was stored in productivity, etc. These data mining models have been constructed for
the SLM system and company PCs for the handling of unstructured, achieving specifically targeted objectives so that the associations
semi-structured, and structured data. among BOL of AM big data, the unique models, and the data mining
For the present study, the company PCs have stored and managed results are ‘one-to-one’ [74]. For example, if the customer projects the
his previous data to manufactured products for their customers. For this improvement of product quality, then the unique model of product
new task of AlSi10Mg products, the company needs R&D, which is quality improvement is selected. For achieving this objective, the con-
thoroughly described in the flowchart of Fig. 10. All the real-time and trol processing parameters of the model are optimized, which have been
13
Table 3
Control processing parameters with technical description of SLM system.
Experimental conditions Processing Description / Values
parameters/
Conditions
General parameters (used in AM System SLM 280 HL (SLM

all testing) Solutions, Germany)
Powder Material AlSi10Mg
Atmosphere Argon inert gas
Oxygen content (%) <0.1%
Beam focus diameter 0.08
(mm)
Building direction Vertical
Scanning strategy 67 °Checkboard
Layer Thickness 0.03
(mm)
Single track printing (1st Exp) Laser Power (kW) 0.06 - 0.36 (6 levels)
Scan Speed (m/s) 0.10 - 1.10 (6 levels)
Single track printing (2nd Laser Power (kW) 0.30 - 0.40 (6 levels)
Exp) Scan Speed (m/s) 0.30 - 1.05 (6 levels)
Single layer samples Laser Power (kW) 0.32 - 0.40 (3 levels)
Scan Speed (m/s) 0.60 - 0.90 (3 levels)
Overlap rate (%) 0.20 - 0.35 (4 levels)
Multiple layers samples (Bulk Laser Power (kW) 0.32 - 0.40 (3 levels)
and tensile samples) Scan Speed (m/s) 0.60 - 0.90 (3 levels)
Overlap rate (%) 0.25 - 0.35 (3 levels)
Hatch distance (mm) 0.071 - 0.116
Thin-walled samples (Wall Laser Power (kW) 0.32
thickness study) Scan Speed (m/s) 0.90
Hatch distance (mm) 0.080
Wall Thickness 0.50 - 5.0 (12 levels)
(mm)
Thin-walled samples Laser Power (kW) 0.32 – 0.38 (3 levels)
(parameters Scan Speed (m/s) 0.73 – 1.07 (3 levels)
optimization) Hatch distance (mm) 0.08 - 0.13 (3 levels)
Wall Thickness 1.0 – 3.0 (3 levels)
(mm)
step by following the flowchart of Fig. 10. It starts from a single layer
and have been completed to the final product of the customer.
7.1. Results of the single-track scanning
SSAM framework work starts with defining the initial processing

parameters, which are best suited to AlSi10Mg material parts. From the
previous knowledge and literature survey, the two processing para-
meters are prime important for SLM process which are laser power and
scanning speed. We have initially defined the laser power from 0.06 kW
to 0.36 kW and scanning speed from 0.10 m/s to 1.10 m/s for printing
of single-track samples.
The results of the surface were observed on the optical microscope
(OM), which is shown in Fig. 11. In the SLM process, a moving heat
source is applied for the melting of material, which causes the flow of
material in the assigned path of the laser beam. Consequently, single
tracks during SLM solidify with a distinctive chevron pattern, which can
be seen on top of each track (see Fig. 11) [79]. The distinctive chevron
pattern represents the direction of motion of the heat source (in SLM is
laser beam), which is shown in optical micrographs of Fig. 11.
Fig. 11 shows the processing maps of a single-track for the SLM
processed AlSi10Mg powder. The laser power and scan speed have a
Fig. 10. Overall flowchart for the experimental methodology. significant influence on the surface stability of the single track. The
analysis of single track could provide a solid basis for determining the
process window. Four types of zones existing in the single track were
given the results of excellent product quality; the obtained results are
identified as partial making, good consolidation, excessive balling, and
stored in PDKM, whose knowledge is further applied in the future. The
over-burning which are also shown in Fig. 11. The partial making and
other unique models also have similar establishing principles.
excessive balling are presented by discontinuous tracks and the drops
formation, and the good consolidation is characterized by the con-
7. Results and discussions tinuous tracks [80].
The sets of processing parameters for the working powder express
In this section, the experimental results will be discussed step by geometrical characteristics and continuity of SLM tracks. The molten
14
material builds a cylinder-like track due to surface tension. For con-
Monitoring and controlling temperature data of the SLM system, processing bed of product, and
To trace and monitor real-time data information of drawings, material, product, SLM process,
To measure the mechanical properties of SLM products such as UTS, yield strength, breakage
tinuous tracks formation, penetration into the substrate or the pre-
To monitor the pressure data of the SLM machine, Sandblasting machine, compressor, air
To monitor the electrical energy consumption information of AM machines, compressors,
To measure the morphology of powder and products, EDS analysis, Microscopic analysis,
viously sintered layer has the additional stabilizing effect [81]. But
deep penetration causes keyhole, which is unacceptable in SLM because
it can stimulate porosities in the final SLM product due to the collapse
of molten pool and gas bubbles locked in the material. Balling is de-
pendent on scan speed and is built due to instability promoted by
higher scanning speeds and surface tension of material when using
large layer thickness powder. Surface tension plays an important role in
the crystal growth and porosity formation of metal during solidification
[82]. The mechanism of discontinuity, irregularities, distortion, and
balling-effect may be connected with granulomorphometric character-
Measuring the dimensional quality of built product
istics of the powder and inhomogeneity in powder layer thickness,

Measuring the surface quality of the built product
To measure the hardness of the built components

thermo-physical properties of materials, energy input parameters in-
cluding laser power, scan speed, and spot size; melt hydrodynamics,
etc. [79].
elongations, stress-strain curves, etc.
It can be observed from Fig. 11 that the single tracks are non-con-
tinuous at the laser power of 0.06 kW and 0.12 kW, which would be
caused due to low laser energy which could not melt the powder
fracture surfaces analysis.
properly. Specifically, it can be seen at # 6 single tracks at a laser power

of 0.06 kW and a scan speed of 1.10 m/s (see Fig. 11). Stable zones of
single tracks are formed at a scan speed from 0.30 to 0.90 m/s and at
the laser power of 0.18 kW, 0.24 kW, 0.30 kW, and 0.36 kW (see
pumps, etc.
supply, etc.
Fig. 11). At higher scan speed (1.10 m/s), the laser energy cannot melt
usage, etc.
workshop.
the particles completely and initiate balling effect, which leads to un-
stable and irregular tracks. At low scan speed (0.10 m/s), over burning
of the tracks occurs due to excessive energy exposed on the powder (see
Purpose/ Objective
Drawings/ Models/Material/Product/ tools/AM machines/operators/
Fig. 11). Consequently, stability of the single tracks is achieved by

optimizing the laser power and scanning speed, and proper layer
thickness deposition.
A single track is built not only from powder positioned directly
under the laser spot during the SLM process. The adjacent powder
particles are also involved in this process due to conduction through the
substrate and neighboring particles, scattering of radiation, capillary
phenomena, etc. [79]. The denudation zone, i.e., area without powder
after scanning of laser can be twice as large as the width of the track
[83], which describes the geometrical features of the tracks and finally
the morphology of layers. For stable single tracks, the widths of the
Monitoring resources/Location
single line were measured, as shown in Fig. 12. It was found that width
Machines/ Compressors/etc.
SLM built tensile specimens

SLM powder and products
of the single track varied from 37 to 288 μm. The track width decreased
AM machines/ workshop
AM machine/ workshop
with the reduction of laser power and the increase of scanning speed.
QC document/etc.
From the above analysis, it was found that the single tracks manu-
SLM components
Information of configuration of IoT and other devices for the data acquisition.
factured with laser power of 0.18–0.36 kW and scan speed of

SLM products
SLM products
0.30–0.90 m/s are stable. But, more stable results of single tracks were
achieved between 0.30–0.40 kW of laser power.
In the 2nd experimental design of the single-track, the laser power
was taken from 0.30 kW to 0.40 kW with an interval of 0.02 kW, and
Voltage and current sensors
scanning speed was chosen from 0.30 m/s to 1.05 m/s with an interval
of 0.15 m/s. The experimental design is based on full factorial DoE
Temperature sensor
design with six levels and two factors. By applying 62 full factorial
Sensor/ UHF type
Pressure sensors
design, total single tracks with 36 combinations of laser power and scan
speed were printed on a substrate. Then, the fabricated single tracks
were observed with the OM of Olympus GX51, and the processing
parameters were more refined for further processing of making single
_
_
_
_
_
_
and multi-layer.
From the analysis of optical microscopic (OM) images, the width of
Scanning electron microscope (SEM)
a single track is measured, which is from 95.42 to 210.56 µm, and the
results are presented in Fig. 13. From OM investigations, it was ob-
served that the smooth, stable, and continuous single tracks and their
Surface roughness tester
Tensile testing machine
width within good range are for the laser power from 0.32 kW to
LECO hardness tester
IoT/ other devices
0.40 kW and scan speed of 0.60 m/s to 0.90 m/s. For 0.30 kW laser
Vernier calliper
power, at some points the single tracks are not stable or not continuous,
Smart sensors
Smart sensor
Smart sensor
and the same behavior is observed for low scan speed below 0.60 m/s
and high scan speed of 1.05 m/s (see Fig. 13). It can also be observed in
Table 4
Tags
Fig. 13 that higher width of track is detected for 0.38 kW and 0.30 m/s.
Furthermore, on the basis of initial measurements, the process
15
Fig. 11. Optical images of the surface morphologies of SLM-processed AlSi10Mg single tracks at laser power from 0.06 kW to 0.36 kW (interval of 0.06 kW) and
scanning speed from 0.10 m/s to 1.10 m/s (interval of 0.20 m/s).
Fig. 12. Width of the weld bead of single tracks versus scanning speed at dif-
ferent power of the laser beam. The thickness of the deposited powder layer is Fig. 13. Width of the single tracks versus scanning speed at different power of
0.03 mm. Scanning speed = 0.10 m/s-0.9 m/s for laser power = 0.18 kW, the laser beam at layer thickness of 0.03 mm. Scanning speed = 0.30 m/s
0.24 kW, 0.30 kW and 0.36 kW. −1.05 m/s and laser power = 0.30 kW – 0.40 kW.
window is selected for parameters, which are mentioned in Table 3. So, 7.2. Results of the single-layer printing
we have taken more measurements for single tracks widths for the 09
parameter combinations (mentioned in Table 5), and their average The single-layer samples were built on the processing parameters of
value is taken for further processing. laser power from 0.32 kW to 0.40 kW (interval of 0.04 kW), scanning
speed of 0.60 to 0.90 m/s (interval of 0.15 m/s) and overlap rate from
20 to 35% (with an interval of 5%). A total of 36 Nos samples were built
by applying full factorial design (6k).
16
Table 5
Optimized and stable width of single tracks for different combinations of laser powers and scan speed.
Levels 1 2 3 4 5 6 7 8 9
Laser Power [kW] 0.32 0.32 0.32 0.36 0.36 0.36 0.40 0.40 0.40
Scan speed [m/s] 0.60 0.75 0.90 0.60 0.75 0.90 0.60 0.75 0.90
Width of single track [μm] 136.5 124.1 109.0 148.0 130.6 118.6 155.2 139.5 125.4
Single-tracks geometric characteristics knowledge is helpful in the mentioned in Table 3 [78]. There is a total of 81 Nos (03 sets) bulk
determination of hatch distance of laser beam scanning, which, to- samples (see Fig. 14(a)) were fabricated to study the effect of processing
gether with selected scanning strategy, governs the quality of a single parameters on the density, porosity, hardness and surface quality [78],
SLM layer [79]. Analysis of the morphology of a single layer, in turn, is [84]. A total of 27 Nos cuboid samples were produced to make tensile
critical and important in the selection of the optimal strategy for the specimens and study tensile behavior. The energy consumption during
fabrication of fully-dense and porosity free products. If the hatch dis- the whole SLM process was also measured for each sample with dif-
tance is more than width of single-track, powder is melted in order of ferent combinations of processing parameters.
different tracks. Non-optimal hatch distance can affect the information The heat treatment processes of solution heat treatment (SHT) at
of gaps between tracks in a single layer, which origins chains of pores in 530 °C and 540 °C for 02 h was performed on 02 sets of bulk samples,
the ultimate SLM product [79]. and one set with SHT at 530 °C was further processed for artificial aging
The hatch distance is calculated for each test trails based on single- (AA) for 12 h [78]. The surface quality, densification, strength, and
track width and the changing overlap rate [by using Eq. (2)]. Previously microstructural behavior were studied in the heat treatment conditions
discussed that hatch distance is very important for stable tracks. The and compared with samples of as-built (AB) condition specimens [78],
single-layer area of 8 mm × 8 mm is printed with 0.03 mm layer [84]. It concluded that the densification was enhanced to 99.94% by
thickness. 36 groups single layer samples were printed by following applying SHT at 530 °C and 99.87% in AA in comparison to as-built
processing parameters shown in Table 3 and Table 5. 98.17%. The porosities were also decreased by the heat treatments due
to fine grain structure and strong bonding among the powder particles.
hd = W (1 − OR) (2)
Heat treatment has an excellent effect on the reduction of porosities of
where hd is the hatch distance [μm], W is the width of the single track the AB samples. It was also confirmed from OM and SEM images that
[μm], OR is the overlap rate. The single layers were examined for sur- highly dense parts had lower porosity [78].
face morphology with the Olympus GX51 optical microscope. The For the sustainable and smart additive manufacturing, the sustain-
measurements were repeated three times for each processing condition. ability factor of energy consumption has been considered for para-
It was observed that morphology of single layer was stable for the meters optimization regarding meeting quality requirements. The re-
overlap rate of 25%, 30%, and 35% because of the low hatch distance, sults of densification, porosity, tensile strength, hardness and surface
which is better for continuity of tracks and pore-free structure. The quality have been investigated and analyzed on different combinations
results of the surface morphology on the OM was revealed that the of processing parameters. We have defined a criterion for parts quali-
processing parameters and their levels were suitable for further man- fication such as densification ≥ 98%, porosity ≤ 2%, tensile strength
ufacturing of bulk samples. But, for further processing, the three levels ≥ 350 MPa, hardness ≥ 110 HV and surface roughness ≤ 5 µm. We
of overlap rate were used as 25%, 30%, and 35%. have optimized process parameters which meeting above-mentioned
quality criteria by applying pareto front and statistical regression ana-
7.3. Results of the bulk samples fabrication lysis.
The best-optimized parameters obtained are the laser power of
The bulk samples were made by using processing parameters of 0.32 kW, scanning speed of 0.90 m/s, overlap rate of 25%, and hatch
laser power, scanning speed, overlap rate and layer thickness as distance of 0.08 mm which have consumed low specific energy
Fig. 14. Fabricated SLM specimens for the case study: (a) Bulk samples; (b)Thin-walled specimens.
17
consumption (SEC) of 369.54 MJ/kg. In comparison to other processing and mechanical characteristics.
parameters combination, these optimized parameters have consumed
27.80% less SEC, which is very beneficial for sustainability and cleaner 7.4.2. Study the influence of processing parameters
production. The thin-walled samples were manufactured by using processing
parameters of wall thickness, laser power, scanning speed, and hatch
7.4. Results of the thin-walled specimens distance as mentioned in Table 3. The specimens were fabricated to
study the effect of processing parameters on the density, porosity,
By the implementation of our SSAM framework in the company, we hardness and tensile behavior. The energy consumption during the
have initially achieved the best processing parameters from the man- whole SLM process is also measured.
ufacturing of thin-walled specimens, which will be further utilized in For the sustainable and smart additive manufacturing, the sustain-
the production of functional products of the aerospace and automotive ability factor of energy consumption has been considered for para-
industry. meters optimization regarding meeting quality requirements. The re-
sults of densification, tensile strength, and hardness have been
7.4.1. Study the influence of wall thickness investigated and analyzed on different combinations of processing
In this study, the thin-walled specimens of various wall thickness parameters. We have defined a criterion for parts qualification such as
(i.e. from 0.5 mm to 5.0 mm, as mentioned in Table 3) were fabricated densification ≥ 98%, porosity ≤ 2%, tensile strength ≥ 320 MPa,
for investigation of densification, porosity, mechanical properties, hardness ≥ 100 HV and breakage elongation ≥ 10%. We have opti-
hardness, surface and dimensional quality, and microstructural char- mized process parameters which meeting above-mentioned quality
acteristics in the AB, SHT, and AA conditions (see Fig. 14(b)) [85–88]. criteria by applying pareto front and statistical regression analysis.
The best relative density was attained for 0.50-mm-wall thickness The best-optimized parameters obtained are the laser power of
specimens in the AB, SHT and AA conditions with a maximum relative 0.35 kW, scanning speed of 1.07 m/s, and hatch distance of 0.13 mm
density of 99.92% in the AA condition, and the minimum relative which have consumed low specific energy consumption (SEC) of
density of 93.42% was achieved for 1.50-mm-wall-thickness specimen 205.4 MJ/kg. On the above -mentioned optimized parameters, the re-
in SHT condition. The AA is beneficial for the low-wall-thickness spe- sponses have achieved values of 98.95% relative density, 1.05% por-
cimens for improving their densification. The AA heat treatment has a osity, 334.1 MPa tensile strength, 101.7 HV hardness and breakage
comparable effect on the porosity of lower-wall-thickness specimens elongation of 10.01%. In comparison to other processing parameters
(i.e., from 0.50 to 2.0 mm) which means that the porosities of thin- combination, these optimized parameters have consumed 57.3% less
walled specimens have been reduced by applying AA in comparison SEC, which is very beneficial for sustainability, carbon emission and
with AB condition. The minimum porosity of 0.08% is observed in the cleaner production.
0.50-mm-wallthickness specimen. It is also observed from OM and SEM We have also optimized processing parameters with better quality
images that the porosities were increased in sizes and quantity till characteristics and minimum energy consumption. These optimized
1.50 mm wall thickness (except 0.50-mm-wall-thickness specimens). parameters are 0.35 kW of laser power, 1.07 m/s of scanning speed, and
Moreover, the porosities were reduced gradually with an increase in 0.105 mm of hatch distance and have achieved densification of 98.75%,
wall thickness to 5.0 mm for the AB, SHT and AA conditions [86]. porosity of 1.25%, tensile strength of 358.1 MPa, breakage elongation
In the as-built (AB) condition, the average minimum hardness of of 14.67% and hardness of 106.2 HV with SEC of 249.2 MJ/Kg. When
102.4 HV was achieved for a 1.0 mm wall thickness specimen, and the comparing to other processing parameters combination, these opti-
average maximum hardness of 137.3 HV was attained for 5.0 mm wall mized parameters have consumed 48.3% less SEC which is very bene-
thickness specimen. It is also observed that the hardness was reduced ficial for SSAM.
from 0.50 mm to 1.0 mm wall thickness and then improved till 5.0 mm By the efficient implementation of SSAM framework, the above-
wall thickness specimens [85]. mentioned parameter collection and their results will be used by the
The SLM processed thin-walled specimens had a tensile strength of manufacturers and designers according to the demands of our valued
188.4–364.0 MPa, 154.2–215.1 MPa and 197.9–274.2 MPa in the AB, customers. Actually, in practical applications, there are different kinds
SHT and AA conditions respectively. The breakage elongation of of needs according to the qualification of a product, and the customer
3.76–12.04%, 10.82–20.66% and 5.13–11.22% was examined in the has a different demand for a product. So, our optimized parameters will
AB, SHT and AA conditions respectively. It has observed that the me- be used for the manufacturing of efficient and effective products ac-
chanical properties of AlSi10Mg thin-walled specimens are comparable cording to the requirements of the customer.
to the casted AlSi10Mg material A360.0 [89]. The strength and ductility
were higher because the thin-walled specimens were fabricated under 7.5. Results of the impeller pump manufacturing
the optimal process parameters, and the relative density is greater than
93% and nearly equal to 100% for some specimens. The above-mentioned optimized parameters have been utilized by
The purpose of above-mentioned studies was to investigate the ef- the company for manufacturing of pump components and other com-
fect of wall thickness on the densification, porosity and mechanical ponents of AlSi10Mg alloy (see Fig. 15). For the SSAM, the sustain-
characteristics of the thin-walled specimens. The heat treatment was ability perspective of AM is also considered that the product quality
also applied on the thin-walled specimens to study its influence on the meeting customer's requirement, use less energy which beneficial to
quality characteristics of AlSi10Mg alloy by SLM process. The heat environment, and product cost and productivity is also better. By
treatment has positively influence on the enhancement of densification considering these SAM perspectives, we have used the optimized
and mechanical properties, and reduction in the porosities of the thin- parameters which gives better product quality with minimum energy
walled specimens. consumption, and also good productivity with minimum processing
It is concluded from this study that the densification and mechanical time and cost. Finally, it can be extracted from our case study that our
characteristics were varied with varying the wall thickness. Moreover, SSAM framework will be effectively implemented by the company by
the study is further extended to study the different process parameters manufacturing different components of AlSi10Mg.
influence on various wall thicknesses for manufacturing of quality
product with minimum energy consumption for the SSAM. So, we have 8. Managerial implications
finalized the three wall thicknesses for our further study which are
1 mm., 2 mm and 3 mm. These wall thicknesses are our optimal wall Managerial significances could be generated from key findings and
thickness which have done significant influence on the densification concealed knowledge of BDA, which are beneficial when several
18
These are the advanced manufacturing technologies and lies in the

Industry 4.0. AM is an emerging technology that is widely applied in
manufacturing enterprises for the production of unique and complex
shapes products. The authors of this paper made the following im-
portant contributions:
• Firstly, by combining the key technologies of sustainable manu-

facturing, smart manufacturing, and additive manufacturing, the
term SSAM was created, which applied throughout the article. The
concept of SSAM did not exist in a precise form before, but it is
critical to advance knowledge in this field.
• Secondly, a conceptual framework of BDA in SSAM (BD-SSAM) was
proposed for the product manufacturing cycle of AM products. The
proposed framework can be utilized as a guideline to select the re-
lated product manufacturing cycle stages that influence on the
sustainable production of a specified AM enterprise.
• Thirdly, the big data acquisition and integration procedures estab-
lished, which were applied for gathering real-time, multisource and
heterogeneous data for AM and then further processing for useful
output.
• Fourthly, the data mining approaches could be applied to expose the
association between sustainable production performance and AM
Fig. 15. The impeller of pump fabricated by AlSi10Mg alloy by SLM.
processing parameters. Furthermore, the processing parameters
have been optimized to production performance by improving
departments’ managers are making sustainable and smart additive product quality, productivity, and reducing energy consumption and
manufacturing decisions subsequently [28]. Targeting at the product emission which is beneficial for smart manufacturing and CP.
manufacturing cycle of additive manufacturing products, four man-
agerial implications of the developed framework are included, parti- The proposed BD-SSAM framework has been evidenced in an ap-
cularly for the marketing department, R&D department, production plication scenario in a company. The new emerging company has the
department, and service department [75]. task of manufacturing of components of pumps of AlSi10Mg alloy by
using the SLM system. A flowchart and framework were developed for
• The marketing department is responsible for classifying the un- the company. The real-time data can be attained and communicated to
spoken needs of the forecast customers' and promising customers. the AM enterprise database. With the help of BDA and by the use of
When a more accurate aiming is established and forwarded, then it algorithms, the process parameters of SLM have been optimized for
is worthwhile to match several products with numerous customers, improving product quality, reduction in energy consumption, and im-
respectively. BDA makes it conceivable to pick the most appropriate prove productivity.
customers from the excessive number of customers' data. The limitation of the present study is that the proposed BD-SSAM
• The R & D department inputs for AM will be increased in the future, framework is applied only on the BOL stage of product life cycle due to
which may be due to the development and usage of innovative the available resources and configuration of IoT devices in the com-
technologies for the manufacturing of cleaner production. There is a pany. The proposed framework is only applied to 1 AM system (i.e. SLM
massive amount of data generated for the making of suitable solu- system), and the algorithm for data analysis, such as association, clas-
tions in the conceptual design stage, for making decisions in the sification, and clustering are not studied in this paper.
detailed design stage, and realization of AM product innovative Future research works will be carried out on the application of this
design. The BDA is applied to categorize the most correlative ex- framework on the whole lifecycle stages, and development of the al-
amples as detailed as possible to give direction for the innovative gorithms to optimize and prediction of the processing parameters of the
product development [75]. AM for various AM techniques and multiple materials.
• In the production department, BDA and SSAM framework should be
applied to manage the production of AM systems, improve the en- Declaration of Competing Interest
ergy efficiency of AM systems, and monitor the quality of AM pro-
duct. With the provision of IoT devices and smart meters, an en-
None.
ormous amount of real-time, heterogeneous, and multi-source data
of production is collected. Moreover, with the assistance of BDA, the
optimization of processing parameters relevant to product quality, Acknowledgments
energy consumption, workshop scheduling decisions, etc. could be
comparatively easy to attain. The energy proficiency of AM systems The authors would like to acknowledge the financial support of the
could also be enhanced. National Science Foundation of China (No. 51675441, 51505423,
• In the services department, the customers' satisfaction is increased 51705428). Natural Science Foundation of Zhejiang Province (No.
LY19E050019) and the International Clean Energy Talent Program
by the state-of-the-art service strategies, such as, real-time mon-
itoring service of product quality and predictive maintenance ser- (iCET) of China Scholarship Council (Liujinfa [2017] 5047 and Liujinfa
vice which should be attained by continuous monitoring of product's [2018] 5023). The author would like to thank all the anonymous re-
status. BDA would be applied to analyze the huge gathered data. viewers for their helpful suggestions on the quality enhancement of our
paper.
9. Conclusions
Supplementary materials
In the present paper, BDA, AM, and SSM have been studied to-
gether, which were investigated in academia and industry separately. Supplementary material associated with this article can be found, in
19
the online version, at doi:10.1016/j.rcim.2020.102026. Review, J. Manuf. Mater. Process. 2 (2018).

[32] K. Mumtaz, N. Hopkinson, Top surface and side roughness of Inconel 625 parts
processed using selective laser melting, Rapid Prototyp J 15 (2009) 96–103.
References [33] M.N. Islam, H. Gomer, S. Sacks, Comparison of dimensional accuracies of stereo-
lithography and powder binder printing, Int. J. Adv. Manuf. Technol. 88 (2017)
[1] S. Ren, Y. Zhang, Y. Liu, T. Sakao, D. Huisingh, C.M.V.B. Almeida, A comprehensive 3077–3087.
review of big data analytics throughout product lifecycle to support sustainable [34] D. Wang, L. Yang, Y. Yongqiang, X. Dongming, Theoretical and experimental study
smart manufacturing: a framework, challenges and future research directions, J. on surface roughness of 316L stainless steel metal parts obtained through selective
Clean. Prod. 210 (2019) 1343–1365. laser melting, Rapid Prototyp J. 22 (2016) 706–716.
[2] S. Liu, Y. Zhang, Y. Liu, L. Wang, X.V. Wang, An ‘Internet of Things’ enabled dy- [35] C. Relvas, A. Ramos, A. Completo, J.A. Simões, A systematic approach for an ac-
namic optimization method for smart vehicles and logistics tasks, J. Clean. Prod. curacy level using rapid prototyping technologies, Proceedings of the Institution of
215 (2019) 806–820. Mechanical Engineers, Part B: J. Eng. Manuf. 226 (2012) 2023–2034.
[3] W. Wang, Y. Zhang, R.Y. Zhong, A proactive material handling method for CPS [36] F. Calignano, D. Manfredi, E.P. Ambrosio, L. Iuliano, P. Fino, Influence of process
enabled shop-floor, Robot Comput. Integr. Manuf. 61 (2020) 101849. parameters on surface roughness of aluminum parts produced by DMLS, Int. J. Adv.
[4] Y. Cui, S. Kara, K.C. Chan, Manufacturing big data ecosystem: a systematic litera- Manuf. Technol. 67 (2013) 2743–2751.
ture review, Robot Comput. Integr. Manuf. 62 (2020) 101861. [37] J. Delgado, J. Ciurana, C.A. Rodríguez, Influence of process parameters on part
[5] Y. Liu, L. Wang, X.V. Wang, X. Xu, L. Zhang, Scheduling in cloud manufacturing: quality and mechanical properties for DMLS and SLM with iron-based materials, Int.
state-of-the-art and research challenges, Int. J. Prod. Res. 57 (2019) 4854–4879. J. Adv. Manuf. Technol. 60 (2012) 601–610.
[6] Y. Lu, C. Liu, K.I.K. Wang, H. Huang, X. Xu, Digital Twin-driven smart manu- [38] M. Akbar, T. Irohara, Scheduling for sustainable manufacturing: a review, J. Clean.
facturing: connotation, reference model, applications and research issues, Robot Prod. 205 (2018) 866–883.
Comput. Integr. Manuf. 61 (2020) 101837. [39] I.S. Jawahir, K.E. Rouch, O.W. Dillon Jr, L. Holloway, A. Hall, J. Knuf, Design for
[7] S. Li, L.D. Xu, S. Zhao, Internet of Things: a survey, J. Ind. Inf. Integr. 10 (2018) 1–9. sustainability (DFS): new challenges in developing and implementing a curriculum
[8] B. Liu, Y. Zhang, G. Zhang, P. Zheng, Edge-cloud orchestration driven industrial for next generation design and manufacturing engineers, Proceedings of the 3rd
smart product-service systems solution design based on CPS and IIoT, Adv. Eng. Inf. SME International Conference on Manufacturing Education CIMEC (CIRP), June 22,
42 (2019) 100984. 2005 - June 25, 2005, Society of Manufacturing Engineers, San Luis Obispo, CA,
[9] A. Kusiak, Smart manufacturing must embrace big data, Nature 544 (2017) 23–25. United states, 2005.
[10] T. Peng, K. Kellens, R. Tang, C. Chen, G. Chen, Sustainability of additive manu- [40] C. Achillas, D. Aidonis, E. Iakovou, M. Thymianidis, D. Tzetzis, A methodological
facturing: an overview on its energy demand and environmental impact, Addit. framework for the inclusion of modern additive manufacturing into the production
Manuf. 21 (2018) 694–704. portfolio of a focused factory, J. Manuf. Syst. 37 (2015) 328–339.
[11] P. Urhal, A. Weightman, C. Diver, P. Bartolo, Robot assisted additive manu- [41] A. Drizo, J. Pegna, Environmental impacts of rapid prototyping: an overview of
facturing: a review, Robot Comput. Integr. Manuf. 59 (2019) 335–345. research to date, Rapid Prototyp J. 12 (2006) 64–71.
[12] F. Michel, H. Lockett, J. Ding, F. Martina, G. Marinelli, S. Williams, A modular path [42] D. Dunaway, J.D. Harstvedt, J. Ma, A Preliminary Experimental Study of Additive
planning solution for Wire + Arc Additive manufacturing, Robot Comput. Integr. Manufacturing Energy Consumption, (2017) V004T005A013.
Manuf. 60 (2019) 1–11. [43] G. Zhang, Y. Zhang, X. Xu, R.Y. Zhong, An augmented Lagrangian coordination
[13] B. Dutta, S. Babu, B. Jared, Chapter 2 - Additive manufacturing technology, in: method for optimal allocation of cloud manufacturing services, J. Manuf. Syst. 48
B. Dutta, S. Babu, B. Jared (Eds.), Science, Technology and Applications of Metals in (2018) 122–133.
Additive Manufacturing, Elsevier, 2019, pp. 11–53. [44] R. Huang, M. Riddle, D. Graziano, J. Warren, S. Das, S. Nimbalkar, J. Cresko,
[14] Z. Li, Z. Zhang, J. Shi, D. Wu, Prediction of surface roughness in extrusion-based E. Masanet, Energy and emissions saving potential of additive manufacturing: the
additive manufacturing with machine learning, Robot Comput. Integr. Manuf. 57 case of lightweight aircraft components, J. Clean. Prod. 135 (2016) 1559–1570.
(2019) 488–495. [45] F. Le Bourhis, O. Kerbrat, J.-.Y. Hascoet, P. Mognol, Sustainable manufacturing:
[15] A.-.N. Chen, J.-.M. Wu, K. Liu, J.-.Y. Chen, H. Xiao, P. Chen, C.-.H. Li, Y.-.S. Shi, evaluation and modeling of environmental impacts in additive manufacturing, Int.
High-performance ceramic parts with complex shape prepared by selective laser J. Adv. Manuf. Technol. 69 (2013) 1927–1939.
sintering: a review, Adv. Appl. Ceram. 117 (2018) 100–117. [46] S. Ford, M. Despeisse, Additive manufacturing and sustainability: an exploratory
[16] J. Xiong, G. Liu, Y. Pi, Increasing stability in robotic GTA-based additive manu- study of the advantages and challenges, J. Clean. Prod. 137 (2016) 1573–1587.
facturing through optical measurement and feedback control, Robot Comput. [47] M. Gebler, A.J.M. Schoot Uiterkamp, C. Visser, A global sustainability perspective
Integr. Manuf. 59 (2019) 385–393. on 3D printing technologies, Energy Policy 74 (2014) 158–167.
[17] J.S. Lee, D. Cha Hwang, J.H. Shim, W. Jung Jin, Y. Kim Jong, D.W. Cho, Effect of [48] M. Mani, K.W. Lyons, S.K. Gupta, Sustainability characterization for additive
pore architecture and stacking direction on mechanical properties of solid freeform manufacturing, J. Res. Natl. Inst. Stand. Technol. 119 (2014) 419–428.
fabrication-based scaffold for bone tissue engineering, J. Biomed. Mater. Res. Part A [49] D. Chen, S. Heyer, S. Ibbotson, K. Salonitis, J.G. Steingrímsson, S. Thiede, Direct
100A (2012) 1846–1853. digital manufacturing: definition, evolution, and sustainability implications, J.
[18] T.D. Ngo, A. Kashani, G. Imbalzano, K.T.Q. Nguyen, D. Hui, Additive manufacturing Clean. Prod. 107 (2015) 615–625.
(3D printing): a review of materials, methods, applications and challenges, Compos. [50] J.A. Abell, D. Chakraborty, C.A. Escobar, K.H. Im, D.M. Wegner, M.A. Wincek, Big
Part B: Eng. 143 (2018) 172–196. data-driven manufacturing—process-monitoring-for-quality philosophy, J. Manuf.
[19] P. Jaiswal, J. Patel, R. Rai, Build orientation optimization for additive manu- Sci. Eng. 139 (2017) 101009-101009-101012.
facturing of functionally graded material objects, Int. J. Adv. Manuf. Technol. 96 [51] A. Gunasekaran, T. Papadopoulos, R. Dubey, S.F. Wamba, S.J. Childe, B. Hazen,
(2018) 223–235. S. Akter, Big data and predictive analytics for supply chain and organizational
[20] M. Sugavaneswaran, G. Arumaikkannu, Modelling for randomly oriented multi performance, J. Bus. Res. 70 (2017) 308–317.
material additive manufacturing component and its fabrication, Mater. Des. (1980- [52] F. Tao, Y. Tang, X. Zou, Q. Qi, A field programmable gate array implemented fibre
2015) 54 (2014) 779–785. channel switch for big data communication towards smart manufacturing, Robot
[21] W.W. Wits, J.R.R. García, J.M.J. Becker, How Additive Manufacturing Enables more Comput. Integr. Manuf. 57 (2019) 166–181.
Sustainable End-user Maintenance, Repair and Overhaul (MRO) Strategies, [53] G. Köksal, İ. Batmaz, M.C. Testik, A review of data mining applications for quality
Procedia CIRP 40 (2016) 693–698. improvement in manufacturing industry, Expert Syst. Appl. 38 (2011)
[22] F. Bechmann, Changing the future of additive manufacturing, Metal Powder Rep. 13448–13467.
69 (2014) 37–40. [54] Y. Cheng, K. Chen, H. Sun, Y. Zhang, F. Tao, Data and knowledge mining with big
[23] C. Qian, Y. Zhang, C. Jiang, S. Pan, Y. Rong, A real-time data-driven collaborative data towards smart production, J. Ind. Inf. Integr. 9 (2018) 1–13.
mechanism in fixed-position assembly systems for smart manufacturing, Robot [55] F. Tao, Y. Zuo, L.D. Xu, L. Lv, L. Zhang, Internet of Things and BOM-Based Life Cycle
Comput. Integr. Manuf. 61 (2020) 101841. Assessment of Energy-Saving and Emission-Reduction of Products, IEEE Trans. Ind.
[24] J. Jiang, X. Xu, J. Stringer, Optimisation of multi-part production in additive Inf. 10 (2014) 1252–1261.
manufacturing for reducing support waste, Virtual Phys. Prototyp. 14 (2019) [56] F. Shrouf, G. Miragliotta, Energy management based on Internet of Things: practices
219–228. and framework for adoption in production management, J. Clean. Prod. 100 (2015)
[25] Z. Guo, Y. Zhang, X. Zhao, X. Song, CPS-Based Self-Adaptive Collaborative Control 235–246.
for Smart Production-Logistics Systems, IEEE Trans. Cybern. (2020) 1–11. [57] J. Lee, B. Bagheri, H.-.A. Kao, A cyber-physical systems architecture for industry
[26] J. Jiang, X. Xu, J. Stringer, Optimization of process planning for reducing material 4.0-based manufacturing systems, Manuf. Lett. 3 (2015) 18–23.
waste in extrusion based additive manufacturing, Robot Comput. Integr. Manuf. 59 [58] Z. Song, Y. Moon, Assessing sustainability benefits of cybermanufacturing systems,
(2019) 317–325. Int. J. Adv. Manuf. Technol. 90 (2017) 1365–1382.
[27] J. Jiang, J. Stringer, X. Xu, R.Y. Zhong, Investigation of printable threshold over- [59] I. Chih-Lin, S. Han, Z. Xu, Q. Sun, Z. Pan, 5G: rethink mobile communications for
hang angle in extrusion-based additive manufacturing for reducing support waste, 2020+, Philos. Trans. Ser. A, Math. Phys. Eng. Sci. 374 (2016).
Int. J. Comput. Integrated Manuf. 31 (2018) 961–969. [60] D. Laney, 3D data management: controlling data volume, velocity, and variety, in,
[28] A. Majeed, J. Lv, T. Peng, A framework for big data driven process analysis and META Group, 2001.
optimization for additive manufacturing, Rapid Prototyp J. 25 (2019) 308–321. [61] J. Jiang, G. Hu, X. Li, X. Xu, P. Zheng, J. Stringer, Analysis and prediction of
[29] M.P. Brundage, W.Z. Bernstein, S. Hoffenson, Q. Chang, H. Nishi, T. Kliks, printable bridge length in fused deposition modelling based on back propagation
K.C. Morris, Analyzing environmental sustainability methods for use earlier in the neural network, Virtual Phys. Prototyp. 14 (2019) 253–266.
product lifecycle, J. Clean. Prod. 187 (2018) 877–892. [62] P. Leitão, A.W. Colombo, S. Karnouskos, Industrial automation based on cyber-
[30] A. F2792-12a, Standard Terminology for Additive Manufacturing Technologies, physical systems technologies: prototype implementations and challenges, Comput.
(Withdrawn 2015), ASTM International, West Conshohocken, PA, 2012, p. 3. Ind. 81 (2016) 11–25.
[31] J. Jiang, X. Xu, J. Stringer, Support Structures for Additive Manufacturing: a [63] Siemens, Big Data Pays Big in Power Plant Operation, in, 2014.
20
[64] N. Elgendy, A. Elragal, Big Data Analytics: a Literature Review Paper, in: P. Perner review on integrated products and services, J. Clean. Prod. 47 (2013) 222–231.
(Ed.), Advances in Data Mining. Applications and Theoretical Aspects, Springer [78] A. Majeed, Y. Zhang, J. Lv, T. Peng, Z. Atta, A. Ahmed, Investigation of T4 and T6
International Publishing, Cham, 2014, pp. 214–227. heat treatment influences on relative density and porosity of AlSi10Mg alloy
[65] K. Zhou, S. Yang, Understanding household energy consumption behavior: the components manufactured by SLM, Compute. Ind. Eng. (2019) 106194.
contribution of energy big data analytics, Renew. Sustain. Energy Rev. 56 (2016) [79] I. Yadroitsev, P. Krakhmalev, I. Yadroitsava, Hierarchical design principles of se-
810–819. lective laser melting for high quality metallic objects, Addit. Manuf. 7 (2015)
[66] Y. Zhang, S. Ma, H. Yang, J. Lv, Y. Liu, A big data driven analytical framework for 45–56.
energy-intensive manufacturing industries, J. Clean. Prod. 197 (2018) 57–72. [80] N.K. Tolochko, S.E. Mozzharov, I.A. Yadroitsev, T. Laoui, L. Froyen, V.I. Titov,
[67] A. Majeed, M. Muzamil, M.Z. Awan, A. Siddiqui, Developing of a manufacturing M.B. Ignatiev, Balling processes during selective laser treatment of powders, Rapid
cycle architecture for Fused deposition modeling technique, Int. J. Lightweight Prototyp. J. 10 (2004) 78–87.
Mater. Manuf. (2019). [81] I. Yadroitsev, A. Gusarov, I. Yadroitsava, I. Smurov, Single track formation in se-
[68] K. Shvachko, H. Kuang, S. Radia, R. Chansler, The Hadoop distributed file system, lective laser melting of metal powders, J. Mater. Process. Technol. 210 (2010)
Proceedings of the IEEE 26th Symposium on Mass Storage Systems and 1624–1631.
Technologies, MSST 2010, May 6, 2010 - May 7, 2010, IEEE Computer Society, Lake [82] D. Shi, D. Li, G. Gao, Relation between surface tension and microstructural mod-
Tahoe, NV, United states, 2010 IEEE. ification in Al–Si alloys, Mater. Charact. 59 (2008) 1541–1545.
[69] W. Yang, X. Liu, L. Zhang, L.T. Yang, Big data real-time processing based on storm, [83] I. Yadroitsev, Selective Laser melting: Direct manufacturing of 3D-objects by
Proceedings of the 12th IEEE International Conference on Trust, Security and Selective Laser Melting of Metal Powders, LAP LAMBERT Academic Publishing,
Privacy in Computing and Communications, TrustCom 2013, July 16, 2013 - July 2009.
18, 2013, IEEE Computer Society, Melbourne, VIC, Australia, 2013, pp. 1784–1787. [84] A. Majeed, A. Ahmed, A. Salam, M.Z. Sheikh, Surface quality improvement by
[70] M.H. Iqbal, T.R. Soomro, Big data analysis: apache storm perspective, Int. J. parameters analysis, optimization and heat treatment of AlSi10Mg parts manu-
Comput. Trends Technol. 19 (2015) 9–14. factured by SLM additive manufacturing, Int. J. Lightweight Mater. Manuf. 2 (2019)
[71] R.Y. Zhong, C. Xu, C. Chen, G.Q. Huang, Big data analytics for physical internet- 288–295.
based intelligent manufacturing shop floors, Int. J. Prod. Res. 55 (2017) [85] A. Majeed, A. Ahmed, B. Liu, S. Ren, J. Yang, Influence of wall thickness on the
2610–2621. hardness of AlSi10Mg alloy parts manufactured by selective laser melting, Procedia
[72] D.F. Xu, Q. Li, H.B. Jun, J. Browne, Y.L. Chen, D. Kiritsis, Modelling for product CIRP 81 (2019) 459–463.
information tracking and feedback via wireless technology in closed-loop supply [86] A. Majeed, M. Muzamil, J. Lv, B. Liu, F. Ahmad, Heat treatment influences densi-
chains, Int. J. Comput. Integr. Manuf. 22 (2009) 648–670. fication and porosity of AlSi10Mg alloy thin-walled parts manufactured by selective
[73] R.Y. Zhong, G.Q. Huang, S. Lan, Q.Y. Dai, X. Chen, T. Zhang, A big data approach laser melting technique, J. Braz. Soc. Mech. Sci. Eng. 41 (2019) 267.
for logistics trajectory discovery from RFID-enabled production data, Int. J. Prod. [87] A. Ahmed, A. Majeed, Z. Atta, G. Jia, Dimensional quality and distortion analysis of
Econo. 165 (2015) 260–272. thin-walled alloy parts of AlSi10Mg manufactured by selective laser melting, J.
[74] Y. Zhang, S. Ren, Y. Liu, T. Sakao, D. Huisingh, A framework for Big Data driven Manuf. Mater. Process. 3 (2019).
product lifecycle management, J. Clean. Prod. 159 (2017) 229–240. [88] Z. Atta, A. Majeed, Y.F. Zhang, M. Abid, Surface quality investigation of alsi10mg
[75] Y. Zhang, S. Ren, Y. Liu, S. Si, A big data analytics architecture for cleaner manu- thin-walled parts built by selective laser melting, Proceedings of the 49th
facturing and maintenance processes of complex products, J. Clean. Prod. 142 International Conference on Computers and Industrial Engineering, CIE 2019,
(2017) 626–641. October 18, 2019 - October 21, 2019, Beijing, China, 2019.
[76] D. Mourtzis, Challenges and future perspectives for the life cycle of manufacturing [89] ASTM-B85-03, Standard Specification for Aluminum-Alloy Die Castings, in: ASTM
networks in the mass customisation era, Logist. Res. 9 (2016) 2. International West Conshohocken, PA, 2003, pp. 1–8.
[77] F.H. Beuren, M.G.G. Ferreira, P.A.C. Miguel, Product-service systems: a literature
21
View publication stats

1 s2.0 S0736584520302374 Main1

Uploaded by

Copyright:

Available Formats

1 s2.0 S0736584520302374 Main1

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

1 s2.0 S0736584520302374 Main1

Uploaded by

Copyright:

Available Formats

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

A big data-driven framework for sustainable and smart additive

Article in Robotics and Computer-Integrated Manufacturing · February 2021

Arfan Majeed Yingfeng Zhang

Jingxiang Lv Tao Peng

SEE PROFILE SEE PROFILE

Industrial Artificial Intelligence for Digital Servitization View project

The user has requested enhancement of the downloaded file.

Contents lists available at ScienceDirect

Robotics and Computer Integrated Manufacturing

A big data-driven framework for sustainable and smart additive T

Abbreviations ERP enterprise resource management

Fig. 1. A route map of sustainable manufacturing milestones [10].

approaches, with a minimum requirement of specialized tooling in the 2. Literature review

Fig. 2. Combinations of research of BDA, AM, and SSM.

Fig. 3. The three dimensions of SAM [10].

Fig. 4. A BDA framework for SSAM.

(volume, velocity, variety, and value), which is challenging to in-

To trace and monitor real-time data information of drawings and material.

AM enterprises. In the BOL, it is an important asset for innovative

4.1. Big data perception and acquisition for SSAM

An overall framework model for heterogeneous and real-time pro-

duct manufacturing cycle, big data perception, and acquisition of SSAM

Primarily, the RFID tags are conﬁgured adequately on the technical

maintenance of machines which play signiﬁcant roles throughout the

product manufacturing cycle. Moreover, the RFID tags are conﬁgured

handling, warehouse, tool stores, laboratories, etc. The AM machines

and CP strategy. The energy consumption can be measured from the

power consumption of the AM systems which is equivalent to the

product of real time electrical current and voltage. Diﬀerent smart

Fig. 7. The overall framework of BD-SSAM for the application scenario.

Fig. 8. Experimental system for fabrication of product and data acquisition.

Fig. 9. AlSi10Mg powder; (a) morphology, (b) EDS analysis.

General parameters (used in AM System SLM 280 HL (SLM

7.1. Results of the single-track scanning

SSAM framework work starts with deﬁning the initial processing

material builds a cylinder-like track due to surface tension. For con-

To monitor the electrical energy consumption information of AM machines, compressors,

istics of the powder and inhomogeneity in powder layer thickness,

To measure the hardness of the built components

properly. Speciﬁcally, it can be seen at # 6 single tracks at a laser power

Drawings/ Models/Material/Product/ tools/AM machines/operators/

Fig. 11). Consequently, stability of the single tracks is achieved by

SLM built tensile specimens

factured with laser power of 0.18–0.36 kW and scan speed of

Tensile testing machine

These are the advanced manufacturing technologies and lies in the

• Firstly, by combining the key technologies of sustainable manu-

the online version, at doi:10.1016/j.rcim.2020.102026. Review, J. Manuf. Mater. Process. 2 (2018).

View publication stats

You might also like