Vous êtes sur la page 1sur 20

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

net/publication/308390849

Additive manufacturing technology in spare parts supply chain: a


comparative study

Article  in  International Journal of Production Research · September 2016


DOI: 10.1080/00207543.2016.1231433

CITATIONS READS

9 370

4 authors, including:

Yang Cheng
Aalborg University
20 PUBLICATIONS   158 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Yang Cheng on 22 November 2017.

The user has requested enhancement of the downloaded file.


International Journal of Production Research

ISSN: 0020-7543 (Print) 1366-588X (Online) Journal homepage: http://www.tandfonline.com/loi/tprs20

Additive manufacturing technology in spare parts


supply chain: a comparative study

Yao Li, Guozhu Jia, Yang Cheng & Yuchen Hu

To cite this article: Yao Li, Guozhu Jia, Yang Cheng & Yuchen Hu (2016): Additive manufacturing
technology in spare parts supply chain: a comparative study, International Journal of
Production Research, DOI: 10.1080/00207543.2016.1231433

To link to this article: http://dx.doi.org/10.1080/00207543.2016.1231433

Published online: 20 Sep 2016.

Submit your article to this journal

Article views: 30

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at


http://www.tandfonline.com/action/journalInformation?journalCode=tprs20

Download by: [Aalborg University Library] Date: 25 September 2016, At: 06:56
International Journal of Production Research, 2016
http://dx.doi.org/10.1080/00207543.2016.1231433

Additive manufacturing technology in spare parts supply chain: a comparative study


Yao Lia,b, Guozhu Jiaa, Yang Chengc,d* and Yuchen Hud
a
School of Economics & Management, Beihang University, Beijing, P.R. China; bSchool of Economics & Management, Tsinghua
University, Beijing, P.R. China; cCenter for Industrial Production, Aalborg University, Aalborg, Denmark; dSchool of Business
Administration, Jiangxi University of Finance and Economics, Nanchang, P.R. China
(Received 8 December 2015; accepted 26 August 2016)

Additive Manufacturing (AM) technology has the potential to significantly improve supply chain dynamics. The purpose
of this paper is to investigate the impact of AM on spare parts supply chain. Three supply chain scenarios are
investigated in this paper, namely conventional supply chain, centralised AM-based supply chain and distributed
AM-based supply chain. Based on system dynamics simulations, this paper specifically compares three supply chain
scenarios, in terms of total variable cost and carbon emission. The results show the spare part supply chain utilising AM
is indeed superior to the traditional one in sustainable performance. It is also expected that AM can facilitate the spare
parts supply chain to achieve more economic benefits along with its development. To our knowledge, this paper is one
of the early studies that explores the impact of AM on supply chain performance and quantitatively examines the
superiority of utilising AM in spare parts supply chain. Some suggestions are also provided to help managers adopting
AM in their spare parts supply chains.
Keyword: advanced manufacturing technology; supply chain dynamics; additive manufacturing; spare parts supply
chain; system dynamics; comparative study

1. Introduction
Currently, the biggest challenge in supply chain management is the need to efficiently and effectively deliver ever more
valuable products and services to customers (Holmström et al. 2010; Christopher 2011). In the context of a spare parts
supply chain, customer service is related to meeting customers’ demands reliably and consistently to keep their equip-
ment in operational condition. Maintenance, repair and operations are closely tied to the accessibility of appropriate
parts whenever a customer requires them and to reduce downtime costs (Khajavi, Partanen, and Holmström 2014). Thus,
the requirements for a spare parts supply chain differ from those of supply chains for other products in several ways:
service requirements are higher as the effects of stock-outs may be financially significant, the demand for parts may be
extremely sporadic and difficult to forecast, and the prices of individual parts may be very high (Huiskonen 2001). The
ability to supply spare parts with high fulfilment rates at low costs therefore becomes both a strategic opportunity (Wise
and Baumgartner 1999) and a challenge for many original equipment manufacturers (OEMs) (Markillie 2012).
Typically, companies must invest heavily in their spare parts supply chain operations to reach high fulfilment rates
and reliability (Cohen, Agrawal, and Agrawal 2006). The need to hold a relatively large inventory of parts close to the
consumption locations leads to issues such as high warehousing and inventory obsolescence costs, and capital costs
related to slow-moving parts (Khajavi, Partanen, and Holmström 2014). Throughout recent decades, ICT-enabled tech-
nologies such as enterprise resource planning have effectively addressed some of these problems. Now, the emergence
of additive manufacturing (AM) technology further creates an opportunity to manufacture spare parts on demand to
improve supply chain dynamics (Liu et al. 2014).
AM is a kind of rapid prototyping technology (Hague, Mansour, and Saleh 2004; Berman 2012; Birtchnell and Urry
2013). Significantly different from the conventional subtractive manufacturing, AM relies on particular powdered, fila-
mentous or fluid materials, which are then accumulated, cured and bonded layer upon layer to make final products
according to 3D model data (ASTM 2010). It is suitable for producing products that are small in volume, complex in
structure, and with uncertain demand. In addition to reducing the waste of raw materials and saving the costs of produc-
tion line construction and tool development, AM further provides the opportunity of reducing the number of stages in
the conventional supply chain (Reeves 2008b). In a supply chain using AM, products no longer need to be transported

*Corresponding author. E-mail: cy@business.aau.dk

© 2016 Informa UK Limited, trading as Taylor & Francis Group


2 Y. Li et al.

to distributors or retailers after being produced by manufacturers. Instead, they can be manufactured on demand and
near the customers, i.e. distributed manufacturing (Liu et al. 2014). The net effect is a reduction in the need for
warehousing, transportation and packaging. These characteristics of AM indeed imply that it can have tremendous impli-
cations to spare parts supply chains, especially considering the features of a typical spare parts supply chain explained
above.
This study investigates the effects of utilising AM to produce spare parts within the structure of a spare parts supply
chain. The objective of this paper is to compare a spare parts supply chain that utilises AM with a conventional one, in
terms of cost and carbon emission. By doing so, it is possible to explore the impact of AM on supply chain perfor-
mance and quantitatively examine the superiority of utilising AM in a spare parts supply chain.
The remainder of this paper is organised as follows: Section 2 presents a literature review, Section 3 explains the
research methodology and Section 4 presents the findings and results of our analysis. This paper ends with conclusions
summarising the research outcomes, practical implications and future research.

2. Literature review
2.1 AM technology
AM, also known as direct manufacturing or three-dimensional printing, is a digital technology for producing physical
objects layer by layer from a three-dimensional (3D) computer aided design (CAD) file (Khajavi, Partanen, and
Holmström 2014; Gardan 2016), rather than through moulding or subtractive techniques (such as machining). The AM
production process begins with scanning an object to generate a 3D CAD model of it with all its details and dimensions
(Khajavi, Partanen, and Holmström 2014). In addition to 3D CAD, the initial 3D data can be generated by computer
tomography, magnetic resonance imaging, or using 3D digitising systems (Wohlers 2008). Next, the 3D CAD file is
transformed into STL format files, which can be directly processed by the AM printing machine. Each slice of the STL
file represents a two-dimensional (2D) cross section (layer) of the object to be produced. These 2D layers are then sent
to the printing machine one layer at a time. The machine produces the object by building each layer on top of the previ-
ous one, utilising different solidification methods pertinent to the raw material in its production chamber (Kruth, Leu,
and Nakagawa 1998; Gibson, Rosen, and Stucker 2009). The process may take from a few hours to a few days, depend-
ing on the product’s size and required production precision (Khajavi, Partanen, and Holmström 2014).
AM was invented in the US in the 1980s and initially used as a method for producing rough physical prototypes of
products. Since then, it has continued to evolve in different aspects along with the development of information technol-
ogy, processing technology and the emergence of new materials. Major AM techniques have gradually been developed,
including selective laser sintering (SLS), direct metal laser sintering, fused deposition modelling, stereolithography
(SLA), laminated object manufacturing and inkjet bioprinting. In all cases, objects are formed one layer at a time, each
layer on top of the previous, until the final object is complete. After decades’ development, AM can now create objects
from a variety of materials including plastic, metal, ceramics, glass, paper and even living cells. The input materials for
these can be in the form of powders, filaments, liquids or sheets (Manyika et al. 2013). Furthermore, increasingly more
parts produced by AM today possess the required precision and quality to be used as final functional parts for special
applications, for example air-cooling ducts for aircrafts or hearing aid and prosthesis equipment (Levy, Schindel, and
Kruth 2003). Therefore, AM is no longer viewed as a technology merely for prototype production. In fact, proclaimed
to bring ‘the next industrial revolution’ (Berman 2012; Manyika et al. 2013), AM with its swift advancements (e.g.
precision, speed, affordability and materials range) and inherent capabilities, has the potential to fundamentally
revolutionise manufacturing operations and supply chains (Economist 2012; Markillie 2012; Khajavi, Partanen, and
Holmström 2014). Its characteristics enable manufacturers to produce any part at any time in various locations with
customised batch sizes. Therefore, the continuous development of AM offers OEMs the opportunity to change their
manufacturing operations and their supply chain configurations.

2.2 AM technology and its impact on a (spare parts) supply chain


AM technology can potentially have a significant effect on the conventional supply chain configuration of various indus-
tries (Petrovic et al. 2011), such as the hearing aid industry (Wohlers 2008), dental industry (Holmström et al. 2010)
and aircraft industry (Khajavi, Partanen, and Holmström 2014; Liu et al. 2014). The impact of AM could be complex as
it provides the opportunity to reduce the number of stages in the conventional supply chain (Reeves 2008b). In this
case, it not only influences current methodologies but also the constituent parts of the supply chain (Reeves 2008a; Liu
et al. 2014). Accordingly, several researchers have investigated the introduction of AM in supply chains from various
International Journal of Production Research 3

perspectives, such as Birtchnell and Urry (2013), Tuck and Hague (2006), Tuck, Hague, and Burns (2007), Huang, Liu,
and Mokasdar (2013), Nyman and Sarlin (2014) and Gebler, Uiterkamp, and Visser (2014).
Several studies, such as Walter, Mcmanus, and Yrjölä (2004), Peres and Noyes (2006), Hasan and Rennie (2008),
Holmström et al. (2010), Liu et al. (2014) and Khajavi, Partanen, and Holmström (2014), have specifically paid atten-
tion to the spare parts supply chain and investigated the use of AM technology in such a context. Spare parts supply
chain management attempts to reduce operating costs while keeping the customers’ satisfaction at an acceptable level
(Andersson and Marklund 2000). To accomplish this, the suppliers have to overcome the unpredictability of demand
(Simao and Powell 2009) and make decisions about the trade-off between operating cost, inventory level and delivery
time (Huang, Liu, and Mokasdar 2013; Liu et al. 2014). These challenges make it difficult for the spare parts supply
chain managers to deliver a high level of service with a low cost. However, the development of AM technology offers
the potential to address these challenges (Peres and Noyes 2006). Although noting that a fully functional AM supply
chain was not yet available in the spare parts industry, Hasan and Rennie (2008) proposed a business model to enable
such a supply chain, which provides various services based on an e-business platform.
Holmström et al. (2010) described two different approaches to integrate AM technology in a spare parts supply
chain. The first approach is to use centralised AM capacity to replace inventory holding. AM machines are deployed in
centralised distribution centres to produce slow-moving spare parts on demand. This approach is desirable when a lim-
ited number of parts are needed and the required response time is not critical. The second approach, the distributed
deployment of AM at each service location, is suitable when the demand of AM-producible parts is sufficiently high to
justify the capital investment. The advantages include the elimination of inventory holding and transportation costs, and
a fast response time. Further, Liu et al. (2014) evaluated the impact of AM in the aircraft spare parts supply chain based
on the well-known supply chain operation reference model. They investigated three supply chain scenarios, i.e. conven-
tional (as-is) supply chains, centralised AM supply chains and distributed AM supply chains. They showed that the use
of AM offers various opportunities to reduce the required safety inventory of aircraft spare parts in the supply chain.
Khajavi, Partanen, and Holmström (2014) evaluated the potential impact of AM improvements on the configuration of
spare parts supply chains through scenario modelling of the spare parts supply chain of the F-18 Super Hornet fighter
jet. They found that using current AM technology, centralised production is clearly the preferable supply chain configu-
ration, but distributed spare parts production becomes practical as AM machines become less capital intensive and more
autonomous, and offer shorter production cycles.

2.3 Gaps identified in the existing literature


It is shown above that a number of studies have been conducted investigating the application of AM in the context of
spare parts supply chains. However, because AM technology is constantly evolving, the conclusions should be revisited
periodically and a step-by-step procedure must be used to determine AM’s suitable applications (Huang, Liu, and
Mokasdar 2013; Liu et al. 2014). The existing studies (e.g. Walter, Mcmanus, and Yrjölä 2004; Peres and Noyes 2006;
Hasan and Rennie 2008; Holmström et al. 2010; Khajavi, Partanen, and Holmström 2014 and Liu et al. 2014) have
indicated that AM can have tremendous implications for a spare parts supply chain since AM, due to its unique charac-
teristics, creates the opportunity to manufacture spare parts on demand and to reduce the number of stages in the supply
chain. It thereby facilitates a spare parts supply chain to address the challenges related to the trade-off between operating
cost, inventory level, and delivery time (Huang, Liu, and Mokasdar 2013; Liu et al. 2014). Most of these studies quali-
tatively imply that the spare parts supply chain adopting AM is supposed to be superior to a conventional one, but this
comparison lacks the research to quantitatively verify the positive implications of AM on a spare parts supply chain. To
our knowledge, there is only one article, which is Liu et al. (2014), that makes an informed comparison between con-
ventional and AM-based spare parts supply chains. Nevertheless, this study mainly addresses the comparison of safety
inventory, which is indeed not sufficient.
As AM technology has yet to reach the level of maturity of conventional manufacturing, a comparison between
them is interesting and important; in particular, cost comparison. Cost is usually the key point for decision-making, with
break-even points for different manufacturing technologies being the dominant information for decision-makers (Ruffo,
Tuck, and Hague 2006). Many of the current cost comparison studies tend to examine the production of single parts,
and those that examine assemblies tend not to examine supply chain effects such as inventory and transportation costs
(Thomas 2015). To address this gap, it is important to compare the spare parts supply chain adopting AM with the con-
ventional one, specifically in terms of total cost from a supply chain perspective as discussed in Holmström et al.
(2010).
Furthermore, some of the existing studies indicate that AM technology is more environmentally friendly than
conventional subtractive manufacturing, in terms of waste reduction, energy consumption and carbon emission
4 Y. Li et al.

(Morrow et al. 2007; Campbell et al. 2011; Serres et al. 2011). But, at the same time, some other research shows that
AM technology can be more energy intensive under certain conditions (Reeves 2008a; Baumers et al. 2011). Therefore,
we also need to compare the spare parts supply chain adopting AM with the conventional one in terms of carbon
emission, especially considering there is very limited research about AM on sustainability aspect (Xu et al. 2015).
In summary, this paper aims to explore whether the spare parts supply chain adopting AM technology is really supe-
rior to a conventional one, both in terms of total cost and carbon emission.

3. Methodology
Considering that the adoption of AM technology in spare parts supply chains is still in its infancy, it is relatively diffi-
cult to identify its best practices and obtain the actual data for the comparison with conventional supply chains. There-
fore, this paper chooses to rely on simulation to explore the differences in total cost and carbon emission. The research
process is described in Figure 1. In the following sections, the selected simulation method and the corresponding
methodological considerations will be introduced in detail.

3.1 Simulation method – system dynamics (SD)


Researchers on spare parts management usually focus on designing inventory or ordering policies (Spengler and
Schröter 2003). Thereby, they normally develop optimisation models based on techniques such as stochastic program-
ming that determine the optimal final orders based on estimates of total demand, and optimal strategies for spare parts
supply and recovery. However, considering this current study aims to compare conventional and AM-based spare parts
supply chains in terms of total cost and carbon emission, these techniques might not be suitable because they often
require researchers to make a number of simplifications that disregard the dynamic aspects of the supply chain. More-
over, we in this study try to obtain a holistic understanding of spare parts management so that we can identify which
supply chain could be more beneficial in which circumstances.
For all these reasons, in this current study, we specifically selected SD as the simulation method. SD, originally
developed in the 1950s (Forrester 1971), is a methodology and mathematical modelling technique for framing, under-
standing, and discussing complex issues and problems. Due to its unique features, SD is considered as a suitable
approach to describe and analyse a supply chain system in which a range of different materials, information flows and
complex dynamic problems intersect (Angerhofer and Angelides 2000). SD has been widely used in the research of
(spare parts) supply chains (Spengler and Schröter 2003). Furthermore, we use Vensim PLE 6.3 as the software for SD
simulation. As user-friendly software, it includes some common models such as causal loop, and ordering and inventory
models. To build models in Vensim, one only needs to develop relationship equations between variables according to
SD’s graphical representation. Using the software, it is easy to compare multiple results derived from different simula-
tion tests.

3.2 Causal analysis of a spare parts supply chain and a basic simulation scenario
The spare parts supply chain is a typical example of a ‘demand pull’ supply chain. The demand for spare parts is nor-
mally due to components that malfunction or after periodic inspection in the process of using equipment. When there is

Developing mathematical Gathering publicly available


formulations of cost for AM data on cost for AM based
Developing basic simulation scenarios based supply chain supply chain
and models for AM based supply chain Developing mathematical Gathering publicly available Carbon
formulations of carbon emission data on carbon emission for emission
for AM based supply chain AM based supply chain comparison
Setting the Choosing the
goals of simulation Causal analysis
research method and tool
Developing mathematical Gathering publicly available
Cost
formulations of cost for data on cost for
Developing basic simulation scenarios comparison
conventional supply chain conventional supply chain
and models for conventional supply
chain Developing mathematical Gathering publicly available
formulations of carbon emission data on carbon emission for
for conventional supply chain conventional supply chain

Figure 1. Research process.


International Journal of Production Research 5

adequate stock in the warehouse, the necessary spare parts are shipped directly from the warehouse, leading to a
decrease in spare parts inventory. Correspondingly, the warehouse has to order from suppliers according to the inventory
level and the difference between supply and demand. Figure 2 shows a causal loop diagram of a spare parts supply
chain, where ‘+’ represents a positive causal chain polarity and ‘−’ represents a causal chain with negative polarity.
This figure is used to guide the development of simulation models based on SD because several assumptions can be
made to establish the basic scenario for simulation. First, the demand rate of spare parts (x) follows normal distribution.
It is accumulated over time, leading to the increase in the demand of spare parts (X). Second, after being used for
replacement, the usage rate (s) of spare parts decreases, resulting in a reduction in the demand (X). Therefore, in the
continuous-time, the demand of spare parts (X) is the integration of the difference between demand rate (x) and usage
rate (s) over time. The initial demand is given as zero as equipment has no failure. Third, the spare parts inventory (K)
increases due to inbound delivery to the warehouse and decreases due to outbound delivery from it. Therefore, the spare
parts inventory (K) can be defined as the integration of the difference between inbound and outbound rates over time.
Fourth, a status variable, i.e. demand in order (D) is defined to represent the number of spare parts that have been
ordered but are not yet put into storage. The reason for formulating this variable is the delays due to raw material trans-
portation and spare parts manufacturing. The demand in order (D) is defined as the integration of the difference between
order rate (d) and inbound rate (r) over time. Fifth, the inventory of spare parts can change instantly. Inbound and out-
bound delivery can happen at the same time. The amount of outbound spare parts from the warehouse is determined
based on the demand. The amount of spare parts put in use is consistent with the outbound amount, but the time delay
due to transportation between the distributor and the service location where the demand of spare parts is generated has
to be taken into consideration. Sixth, the ordered spare parts are directly put into storage as inventory after being pro-
duced. The order amount of spare parts is determined based on the outbound amount of spare parts from warehouse. At
the same time, the difference between spare parts inventory and demand is also considered as grounds for an adjust-
ment. When demand is greater than the existing inventory, the order amount is certainly more than the previous cycle
and vice versa. Therefore, an adjustment coefficient is introduced to represent the difference between supply and
demand, ranging between 0 and 1. Seventh, some accessorial variables and constants are given to facilitate the develop-
ment of the model. Specifically, the aid variables are used to link different status or rate variables, whereas the constants
reflect the basic settings of the simulation model. Definitions of these accessorial variables and constants can be seen in
Table 1. Finally, the delay in information transfer in the supply chain system is not taken into consideration; the spare
parts supply chain considered in this paper is relatively closed, meaning that the material and information exchanges
between supply chain members and external entities are not taken into consideration. All the mentioned variables are
summarised in Table 1 and the statements and equation used in the Vensim simulation are presented in Appendix 1.

3.3 The SD model of a spare parts supply chain adopting AM


In this study, we only consider a simple-structured spare parts supply chain adopting AM, i.e. only including one supplier
and one distributor (manufacturer), and one service location where the demand of spare parts is generated. The reasons for
using such a supply chain structure are threefold. First, although simple, this ‘one supplier-one distributor-one user’ struc-
ture still enables us to fulfil the research objective of this study, i.e. comparing the spare parts supply chain adopting AM

Spare Parts Shipped


Inventory - out of Warehouse
+ +

+
Order Spare Parts Equipemt
+ Replacement Malfunction

+
Diference between - +
Supply and Demand +
Spare Parts
Demand

Figure 2. Spare parts supply chain causal loop diagram.


6 Y. Li et al.

Table 1. Variables classification in the SD model.

Type Variable Unit Variable description Remarks

Status D n Demand in order The demand has been ordered but not yet put into storage
variable K n Spare parts inventory
X n Spare parts demand The demand has been generated but has not yet been
fulfilled
Rate x n/day Spare parts demand rate
variable s n/day Spare parts usage rate
d n/day Spare parts order rate
r n/day Spare parts inbound rate
c n/day Spare parts outbound rate
Accessorial G n Differences between demand and The difference between the demand and inventory
variable inventory
T1 day Spare parts production time Time to produce the ordered spare parts
Y day Inbound delay time The time intervals between demand generation and inbound
delivery of spare parts to the warehouse
Constant T2 day Material transportation time
T3 day Spare parts transportation time
P n/day AM equipment production rate
A day Adjustment coefficient of difference An adjustment coefficient to represent the difference between
between demand and supply demand and inventory, ranging from 0 to 1

with the conventional one in terms of total cost and carbon emission from a supply chain perspective. Second, including
more suppliers, distributors and service locations will require consideration of inventory design and ordering policies
(Spengler and Schröter 2003), and developing an optimisation model, which are essential to determine optimal final orders
based on estimates of total demand and optimal strategies for spare parts supply and recovery. The choice of optimisation
approach can directly influence the reliability of the results of cost and carbon emission comparisons between the spare
parts supply chain adopting AM and the conventional one. Third, such a simple supply chain structure enables us to
develop a deep understanding of the distribution of cost and carbon emissions between different activities in a supply chain,
which would be difficult to achieve if more suppliers, distributors and service locations were included.
Moreover, we assume that the locations of the members in this supply chain are pre-determined and the supplier pro-
vides the raw materials and STL files needed in the manufacturing processes. Nevertheless, according to the existing
research (Holmström et al. 2010; Liu et al. 2014), there are two different approaches to integrate AM technology in the
spare parts supply chain. One approach is to deploy AM machines in centralised distribution centres, in which the dis-
tributor is responsible for producing spare parts with allocated AM equipment as well as holding the inventory of spare
parts. The other approach is to decentralise AM machines to each service location and to have smaller holding invento-
ries that are established near the destinations (Liu et al. 2014). Obviously, the decentralised structure is simpler, as its
material transfer is just between the supplier and service location. In the context of an SD model, the main differences
between these two kinds of structures are the product transportation process and the shifted responsibility of manufactur-
ing from the distributor to the service location, when only one service location is considered. For the decentralised struc-
ture, if assuming that the spare parts can be put into use as soon as they are produced, it can actually be regarded as an
extreme situation of the centralised one. In this case, it is possible to simulate these two types of spare parts supply
chains adopting AM in the same SD model (as illustrated in Figure 3) by adjusting some parameters. In detail, in the
decentralised SD model, the variable of spare parts transportation time, i.e. T3 mentioned above, is eliminated, since it
is assumed that the spare parts can be put into use as soon as they are produced. At the same time, the variables related
to T3 are also adjusted by changing T3’s value to zero. The other variables that do not have direct relationships with T3
remain the same as in the centralised structure. In the decentralised structure, however, the materials need to be trans-
ported to the service location rather than the distributor. Therefore, material transportation time would be longer and is
accordingly assumed to be three days in order to reflect the difference with two-day material transportation time in the
centralised structure.

3.4 The SD model of a conventional spare parts supply chain


The spare parts supply chain adopting AM is further compared with the conventional three-tier spare parts supply chain,
which includes four basic functional members, i.e. the raw material supplier, manufacturer, distributor and service
International Journal of Production Research 7

Figure 3. System dynamics model diagram of a spare parts supply chain adopting AM.

Table 2. Differences between a conventional supply chain and one adopting AM.

Parameter Conventional supply chain Supply chain adopting AM

Production methods Conventional manufacturing – Subtractive AM equipment – Additive manufacturing


manufacturing
Supply chain Supplier – Manufacturer – Distributor – Centralised structure: Supplier – Distributor (Manufacturer) –
configuration Service location Service location
Decentralised structure: Supplier – Service location
Members’ Supplier: Provide raw materials Supplier: Provide raw materials and manufacturing STL files
responsibilities Manufacturer: Manufacturing and holding Distributor (Manufacturer)/Service location:
inventory
Distributor: Distributing and holding Manufacturing and distributing based on demand and holding
inventory inventory

location. This supply chain has a similar causal relationship as the ones adopting AM. They merely differ from each
other in terms of supply chain configuration and transportation process. The conventional supply chain has more levels
of transportation, which in turn leads to longer lead-times and higher inventory levels. Table 2 summarises the differ-
ences between these two kinds of supply chains.
Addressing these differences, we further develop the SD simulation model for the conventional spare parts supply
chain (see Figure 4) based on the SD simulation model for the spare parts supply chain adopting AM by adding more
levels of transportation. The development of this simulation model is also guided by the causal analysis illustrated in
Figure 1 and based on the same set of assumptions mentioned in Section 3.2.

4. Results and discussion


4.1 Cost comparison
There have been several studies that compare cost structures of different AM techniques with those of conventional
manufacturing methods (Hopkinson and Dicknes 2003; Allen 2006; Ruffo, Tuck, and Hague 2006; Atzeni and Salmi
2012). The cost categories considered in these studies generally include material cost (Hopkinson and Dicknes 2003;
Ruffo, Tuck, and Hague 2006; Atzeni and Salmi 2012), labour cost (Hopkinson and Dicknes 2003; Atzeni and Salmi
2012), machine cost (Hopkinson and Dicknes 2003) and administrative cost (Ruffo, Tuck, and Hague 2006; Atzeni and
8 Y. Li et al.

Figure 4. System dynamics model diagram of a conventional spare parts supply chain.

Table 3. Cost structure comparison between a conventional supply chain and one adopting AM.

Cost category Conventional supply chain Supply chain adopting AM

T : Transportation Material (from supplier to manufacturer) Material (from supplier to distributor)


cost Product1 (from manufacturer to distributor) Product (from distributor to service location)
Product2 (from distributor to service location)
M: Manufacturing The labour, material and machine cost The labour, material and machine cost
cost
I: Inventory cost The cost to the distributor and manufacturer The cost to the distributor
For a decentralised supply chain, there is no inventory at the
distributor
A: Administrative The cost to the supplier, manufacturer and The administrative cost to the supplier and distributor
cost distributor

Salmi 2012). There are also studies that measure the costs and benefits of using AM from a supply chain perspective
(Holmström et al. 2010; Thomas 2015). In these studies, inventory cost, processing cost, transportation cost and some
trading costs are usually considered as the elements of the cost model for a supply chain adopting AM. Based on these
studies, this paper chose to include four main cost categories, i.e. transportation cost, manufacturing cost, administrative
cost and inventory cost. Note that the cost structures of these four categories are slightly different in the conventional
spare parts supply chain and the one adopting AM technology, as shown in Table 3.
In order to compare the costs of three different supply chains based on our established SD models, several rules
have to be defined. First, this paper mainly considers supply chains for a single spare part, i.e. a lever as studied in
Hopkinson and Dicknes (2003) and Ruffo, Tuck, and Hague (2006), which is produced on demand. Second, the amount
of spare parts produced is calculated based on the spare parts inbound rate. Third, inventory cost is calculated according
to the daily inventory level. In this paper, inventory cost can be viewed as the aggregation of various inventory-related
costs, for example inventory obsolescence costs and capital costs related to slow-moving parts. Fourth, transportation
cost is positively related to the amount transported as a fixed rate, rather than the transportation time. Specifically, trans-
portation cost of raw materials is calculated according to the amount of raw materials transported, which is further con-
verted based on the amount of spare parts produced and material conversion rate. The transportation cost of spare parts
is calculated based on the amount of spare parts that is shipped out. Fifth, the administrative cost is considered related
to the spare part inventory. This is based on the assumption that the distributor in the centralised supply chain adopting
AM is also responsible for manufacturing, thus the administrative cost is believed to be higher than the administrative
International Journal of Production Research 9

cost of a distributor in the conventional supply chain. The latter needs to hold inventories for both manufacturing and
distribution. Sixth, only the variable costs associated with the spare parts supply processes are taken into consideration,
meaning, for example, the equipment purchasing cost and the production equipment depreciation are not taken into
account. Last, this study compares the total variable costs of three supply chains in the same demand status for
1000 days. This specific demand is selected to ensure that all three supply chains achieve the same service level, so that
the costs of different supply chains in a relatively equal condition can be compared.
Based on these rules, the mathematical formulations of the cost models for the three supply chain scenarios were
developed as shown below.

! !
X
1000 X
1000 X
1000 X
1000 X
1000
CAM ¼ rðt Þ  cT 1AM þ cðt Þ  cT 2AM þ rðt Þ  cM 1AM þ rðt Þ  cM 2AM þ rðt Þ  cLAM
t¼0 !t¼0 !t¼0 t¼0 t¼0
X
1000 X
1000
þ K ðt Þ  cAAM þ K ðt Þ  cIAM
t¼0 t¼0

!
X
1000 X
1000 X
1000
CTrad ¼ r2ðt Þ  cT 1Trad þ c2ðt Þ  cT2Trad þ c1ðt Þ  cT 2Trad
t¼0 t¼0 t¼0 ! !
X
1000 X
1000 X
1000 X
1000 X
1000
þ r2ðt Þ  cM 1Trad þ r2ðt Þ  cM 2Trad þ r2ðt Þ  cLTrad þ K1ðt Þ  cATrad þ K2ðt Þ  cATrad
t¼0 t¼0 ! t¼0 t¼0 t¼0
X
1000 X
1000
þ K1ðt Þ  cITrad þ K2ðt Þ  cITrad
t¼0 t¼0

where CAM is the total cost of the supply chain adopting AM; CTrad is the total cost of the conventional supply chain.
As introduced in Table 1, r is the spare parts inbound rate; c is the spare parts outbound rate and K is the spare parts
inventory. Specially, in the conventional supply chain, K1 and c1, are the parameters related to distributor; K2, c2 and
r2 are the parameters related to manufacturer. The other cost parameters are listed in Table 4 in detail.
Considering that the adoption of AM technology in spare parts supply chains is still in its infancy, it is relatively
difficult to identify its best practices and obtain the actual data for the comparison with conventional supply chains.
Furthermore, there are various kinds of AM machines, techniques and materials, each of which is associated with
different manufacturing processes and cost calculations. In this case, we rely on the information of AM manufacturing
and costs presented in the existing studies and surveys; specifically, we refer to the data presented in Hopkinson and
Dicknes (2003), Ruffo, Tuck, and Hague (2006) and Thomas (2015) for the costs of activities involved in a spare parts
supply chain adopting AM, such as transportation, manufacturing, administration and inventory. According to these
studies, laser sintering (LS) is used as the main AM technology to produce levers. The basic data for this current study’s
cost calculations is shown in Table 4. It is verified by the data collected from other sources. For example, we collected
the data about raw material costs from other studies, as well as from Internet trading platforms such as Alibaba.

Table 4. Basic data for cost calculation (unit: euro).

Parameter Definition Conventional supply chain Supply chain adopting AM

Transportation cost
cT1 Material transportation cost per part 2.02
cT2 Product transportation cost per part 2.85
Manufacturing cost
cM1 Material cost per part 0.165 1.0725
cM2 Machine cost per part 1.170 1.2350
cL Labour cost per part 0.065 0.4225
Administrative cost
cA Administrative cost per part 0.52
Inventory cost
cI Inventory cost per part 0.31
10 Y. Li et al.

An overview was done of the multiple relationships between the prices of raw materials used for AM and those for
conventional manufacturing. The cost data obtained from Hopkinson and Dicknes (2003), Ruffo, Tuck, and Hague
(2006) and Thomas (2015) are further adapted to reflect these relationships and then used for simulation. The simulation
results are reported in Table 5.
According to Table 5, it is evident that the total variable cost of the conventional supply chain is higher than that of
both centralised and decentralised supply chains adopting AM. Upon further investigation of the cost structures of dif-
ferent supply chains, it is found that transportation cost is the dominant factor for the conventional supply chain,
amounting to 57% of its total cost. Similarly, for the centralised supply chain adopting AM, the dominant cost category
is also transportation cost, accounting for 49% of its total cost. However, the situation is different for the decentralised
supply chain with AM where the manufacturing cost is the biggest factor, amounting to 44% of its total cost.
Comparing the cost structures of different supply chains also indicates that those adopting AM are more economical
than the conventional supply chain in terms of transportation, inventory and administration. In fact, this verifies the
qualitative analysis in some of the existing studies such as Holmström et al. (2010). As mentioned in these studies, the
conventional spare parts supply chain normally has more tiers compared to the supply chain adopting AM, which in
turn results in more product transportation and thus higher transportation costs. Also, the more tiers in the supply chain,
the higher the possibility of it being out of stock of parts. In this case, the manufacturer and distributor in the conven-
tional supply chain are forced to hold more inventory in order to avoid stock-outs, which inevitably leads to higher
inventory and administrative costs. In contrast, the supply chains adopting AM hold inventory only at the manufacturing
sites. A shorter delay in transportation leads to a relatively lower inventory level, which accordingly leads to the reduc-
tion in inventory and administrative costs. However, it is also found that the AM-based supply chains have much higher
material and labour costs than the conventional supply chains, which accordingly leads to their higher manufacturing
cost. This finding is also consistent with some of the existing studies indicating that material costs constitute a signifi-
cant portion of an additive manufactured product (Thomas 2015).
Comparing two supply chains adopting AM, the results show that the total cost of the decentralised structure is
lower but the cost compositions of these two structures are actually similar. The main difference between them is related
to the costs of product transportation, which is due to the reduction of one or more tiers in the decentralised structure.
However, it is still necessary to note that the SD model in this paper only contains one service location. If more service
locations are considered, the costs may increase rapidly due to the large amount of labour costs, as well as the initial
investment in AM machines (Thomas 2015) that is not considered in this study. It thus seems that the decentralised
structure could be preferable only when the number of service locations is limited.
In summary, the analysis shows that the AM-based supply chain is superior to the conventional one in terms of total
variable costs, since the characteristics of AM enable it to reduce the number of tiers in a supply chain. The fewer tiers
in a supply chain, the fewer costs are associated with it. The analysis also indicates that both manufacturers and distrib-
utors have an opportunity to take advantage of AM to avoid the scale of economies with having more cost effective
spare parts supply chains while enjoying more manufacturing flexibility. Nevertheless, it should be noted that this study
merely addresses the variable costs without considering some other fixed costs such as the purchasing cost of AM
equipment. If these fixed costs are taken into consideration, the AM-based supply chain might not be more cost effec-
tive than the conventional one in terms of the total cost. However, as AM is being further developed, it is expected that
the price of AM equipment will gradually decrease. In fact, the average selling price of a professional-grade industrial
AM system has decreased by 42% between 2001 and 2013 (Wohlers 2014).

Table 5. Simulation results of supply chain costs (unit: euro).

Conventional supply Centralised supply Decentralised supply


Type of cost Category chain chain adopting AM chain adopting AM

Transportation cost Material 19,234.44 15% 19,234.44 20% 19,214.24 32%


Product 54,261.15 42% 27,109.20 29% 0.00 0%
Manufacturing cost Machine 16,282.62 13% 11,759.67 12% 11,747.32 20%
Material 1571.13 1% 10,212.35 11% 10,201.62 17%
Labour 618.93 0% 4023.05 4% 4018.82 7%
Inventory cost Manufacturer 11,367.70 9% 0.00 0% 5445.15 9%
Distributor 2523.71 2% 8373.41 9% 0.00 0%
Administrative cost Manufacturer 19,068.40 15% 0.00 0% 9133.80 0%
Distributor 4233.32 3% 14,045.72 15% 0.00 15%
Total Total variable cost 129,161.40 100% 94,757.83 100% 59,760.95 100%
International Journal of Production Research 11

4.2 Carbon emission comparison


A carbon footprint is generally defined as ‘a measure of the exclusive total amount of carbon dioxide emissions that is
directly and indirectly caused by an activity or is accumulated over the life stages of a product’ (Wiedmann and Minx
2008). A spare parts supply chain indeed involves many activities that result in carbon emissions. Figure 5 illustrates
the carbon footprint of the spare parts supply chain modelled in this paper. It is developed based on PAS 2050, which
was published by the British Standards Institution in 2008 to evaluate and measure the greenhouse gas emissions of
goods and services (Sinden 2009; Whittaker, Mcmanus, and Hammond 2011). However, in order to simplify the calcula-
tion, this paper mainly considers carbon emissions within the processes of raw material processing, manufacturing and
transportation, while omitting waste and recycling in the processing of raw materials processing and production from
the discussions. This is because the carbon emissions related to waste and recycling are more associated with the design
of the products and the specific recycling mechanisms, which are not the main concerns of this study.
Since the real carbon emission data of a spare parts supply chain is difficult to measure and account for, this study
chooses to use the carbon emission data derived from the Atkins Project as reference (Reeves 2012). The Atkins Project
was conducted from 2009 to 2012 to explore the approaches to integrate AM technology into product design, manufac-
turing and distribution. The project aimed to seek the environmental benefits of AM and facilitate the sustainable devel-
opment of AM in a practical application, while ensuring the cost efficiency (Reeves 2012). The project specifically
compared the energy usage and carbon emission generated in the entire lifecycle, covering the raw materials, manufac-
turing, distribution, usage and recycling, when using three different manufacturing approaches (i.e. conventional manu-
facturing and two AM-based manufacturing approaches) to produce the same aviation product. The Atkins Project’s
data about energy usage and carbon emission is presented in Appendix 2. It is used as reference data in this paper but
with certain simplifications (as shown in Table 6) and further used for the simulation. Additionally, a number of specific
rules are developed to facilitate the carbon emission calculation applied in the simulation model as stated below:
• This paper mainly considers supply chains that produce a single spare part on demand.
• The products produced by both conventional manufacturing and AM technology (i.e. SLS) are assumed to be the
same in terms of specification and quality.
• The transportation distance is fixed and the carbon emissions generated during transportation are only related to
the number of products transported.
• The three supply chains face the same demand conditions and the simulation cycle is set as 1000 days. The given
demand ensures all three supply chains achieve the same service level, which further allows for comparison
between the carbon emissions of different supply chains in a relatively equal condition.
Based on these rules, the mathematical formulations of the carbon emission models for the three supply chain scenarios
are developed and shown below.

Figure 5. The carbon footprint of a spare parts supply chain.


12 Y. Li et al.

Table 6. Basic data for carbon emission calculation (unit: kg).

Parameter Definition Conventional supply chain Supply chain adopting AM

Product weight 0.80 0.37


Total weight of material required in supply chain 3.20 0.40 (10% support)
Primary material production carbon emission 100.50 17.46
Powder production carbon emission – 1.34
eM1 Cumulative raw material carbon emission per part 100.50 18.80
eM2 Manufacturing carbon emission per part 1.55 6.72
eT1 Material transportation carbon emission per part 6.40 3.00
eT2 Product transportation carbon emission per part 2.00 2.00

!
X
1000 X
1000 X
1000 X
1000
EAM ¼ rðt Þ  eT1AM þ cðt Þ  eT 2AM þ rðt Þ  eM 1AM þ rðt Þ  eM 2AM
t¼0 t¼0 t¼0 t¼0

!
X
1000 X
1000 X
1000 X
1000 X
1000
ETrad ¼ r2ðt Þ  eT 1Trad þ c2ðt Þ  eT 2Trad þ c1ðt Þ  eT 2Trad þ r2ðt Þ  eM 1Trad þ r2ðt Þ  eM 2Trad
t¼0 t¼0 t¼0 t¼0 t¼0

where EAM is the carbon emission of the supply chain adopting AM; ETrad is the carbon emission of the conventional
supply chain. Besides, as introduced in Table 1, r is spare parts inbound rate; c is spare parts outbound rate. As for the
conventional supply chain, c1 is the outbound rate of distributor; c2 and r2 are the outbound rate and inbound rate of
the manufacturer, respectively. The other carbon emission parameters are listed in Table 6.
The simulation results relating to the carbon emission comparison are reported in Table 7. It is evident that the car-
bon emissions of the spare parts supply chains adopting AM are much lower than those of the conventional spare parts
supply chain. Although it is reasonable to examine the carbon footprint comparison based on the whole supply chain, it
is also worth probing further into whether the difference is largely caused by higher/lower direct emissions from produc-
tion, higher/lower indirect emissions from energy use, or higher/lower emissions from up- and/or down-stream supply
chain activities, especially considering that conventional manufacturing and AM differ in, for example, how materials
are produced, the production approaches and equipment used, the sources of energy used for production, and transporta-
tion. Therefore, the results in Table 7 are further classified into three categories as shown in Table 8.
It is evident from Table 8 that for both conventional and AM-based spare parts supply chains, most of the carbon
emissions are derived from raw materials. It is generally believed that the processes of preparing raw materials for AM
manufacturing demand more energy (314 MJ/kg compared to 209 MJ/kg for conventional manufacturing methods) and
generate more carbon emissions, especially when AM technology is still in its infant stage. Nevertheless, Table 8 shows
that the carbon emissions associated with the raw materials for a conventional supply chain are still much higher than
those for the supply chain adopting AM. This is partially due to the fact that in the conventional supply chain, each pro-
duct consumes much more raw materials (3.2 kg compared to 0.4 kg raw material consumed in the AM-based supply
chain). In fact, the difference in the total carbon emissions between the conventional and AM-based supply chains is lar-
gely caused by the carbon emissions related to raw materials. This is somehow determined by the characteristics of the
conventional manufacturing methods. Similarly, the conventional supply chain generates more carbon emissions in its

Table 7. Supply chain simulation results of carbon emissions (unit: kg).

Centralised supply chain Decentralised supply chain


Sources of carbon emission Conventional supply chain with the adoption of AM with the adoption of AM

Raw materials production 956,961.00 87.82% 166,254.12 59.17% 166,079.50 63.46%


Raw materials powder production – – 12,711.87 4.52% 12,698.52 4.85%
Manufacturing 14,711.49 1.35% 63,920.64 22.75% 63,920.64 24.42%
Material transportation 60,940.80 5.59% 19,044.00 6.78% 19,024.00 7.27%
Products transportation 57,117.00 5.24% 19,024.00 6.77% – –
Total 1,089,730.29 100% 280,954.63 100% 261,722.68 100%
International Journal of Production Research 13

Table 8. Three scopes of simulation results of carbon emissions (unit: kg).

Centralised supply chain Decentralised supply chain


Category Conventional supply chain with the adoption of AM with the adoption of AM

Raw materials 956,961.00 87.82% 178,965.99 63.70% 178,778.04 68.31%


Manufacturing 14,711.49 1.35% 63,920.64 22.75% 63,920.64 24.42%
Transportation 118,057.80 10.83% 38,068.00 13.55% 19,024.00 7.27%

transportation. This result is also consistent with our expectation. On the one hand, the conventional subtractive manu-
facturing demands a large amount of raw materials, which correspondingly leads to a significant increase in carbon
emissions during materials transportation. On the other hand, the conventional spare parts supply chain normally has
more tiers compared to the supply chains adopting AM, which in turn results in more transportation and thus higher car-
bon emissions during transportation. In contrast, AM enables a shortened operating channel, in turn leading to less car-
bon emissions in transportation. However, it is can also be observed in Table 8 that AM consumes more energy and
generates more carbon emissions in its manufacturing processes, as it generally consumes 44.8 MJ to produce one part,
compared with the 10.3 MJ used by conventional manufacturing methods. Therefore, the AM-based supply chains are
superior to the conventional one in terms of indirect carbon emissions from energy use in raw material preparation and
transportation, but not in terms of direct emissions from manufacturing.
The two supply chain structures that adopt AM have similar carbon emission patterns, with higher carbon emissions
from raw materials and manufacturing processes. There is no doubt, however, that the decentralised structure is more
environmentally friendly with lower carbon emissions. The advantage of the decentralised structure is its fewer tiers in
the supply chain, thus leading to a smaller carbon footprint.
Based on the comparison results, we conclude that the decentralised structure is a better choice to reduce the total
carbon emissions from a supply chain perspective. Besides, our analysis also provides some insights to different role
players within a spare parts supply chain. Raw material suppliers might be able to reduce the carbon emissions derived
from transporting raw materials when adopting AM, as less material is needed and most of the required materials are in
powder form. Nevertheless, they might have to consider how to reduce the carbon emissions derived from producing
raw material powders for AM. For manufacturers, adopting AM might not enable them to directly reduce the carbon
emissions from manufacturing processes. Instead, the reduction of the carbon emissions might be achieved due to the
fact that fewer raw materials are needed for manufacturing and the pre-treatment of raw materials might not be neces-
sary. Adopting AM might help distributors to significantly reduce the carbon emissions generated in transporting spare
parts. The distribution of spare parts is increasingly disappearing in decentralised supply chains that implement AM.
Whether AM is an environmentally friendly manufacturing method might be still debatable, but our analysis shows
that in some conditions, the adoption of AM can significantly reduce carbon emissions in a spare parts supply chain. It
is expected that along with the development of AM technology, the carbon emissions related to its materials and manu-
facturing processes will be further reduced and AM can play a more important role in the sustainable development and
more environmental benefits of the spare parts supply chain.

4.3 Sensitivity analysis


In the above analysis, the standard deviation of the demand is assumed to be 30% of the mean, but it is generally
believed that the demand of spare parts can fluctuate dramatically. Therefore, we conducted a sensitivity analysis by
varying the standard deviation from 10 to 100% of the mean. The results are shown in Figures 6 and 7.
Figure 6 shows that, when the demand standard deviation ranges between 10% and 60% of the mean, the cost
increases too for all three types of supply chains, as expected. However, the cost starts to decrease when the demand
standard deviation reaches 70% of the mean. This is especially so for the supply chains adopting AM where the cost
drops dramatically when the demand standard deviation is 80% of the mean. Figure 6 also shows that one type of sup-
ply chain performs the best in all the scenarios – the decentralised supply chain adopting AM has the lowest cost, while
the conventional supply chain has the highest.
Similarly, as shown in Figure 7, the decentralised supply chain adopting AM has the lowest carbon emission in all
the scenarios, followed closely by the centralised supply chain adopting AM. Again, the conventional supply chain has
the highest carbon emission in all the scenarios.
14 Y. Li et al.

Figure 6. Sensitivity analysis by varying demand standard deviation for cost.

Figure 7. Sensitivity analysis by varying demand standard deviation for carbon emission.

It is also surprising to see that the carbon emission decreases dramatically for all three types of supply chains when
the demand standard deviation reaches 80% of the mean. This pattern can also be observed in the sensitivity analysis of
total variable costs for the supply chains adopting AM as illustrated in Figure 6, but the mechanisms behind it still need
to be investigated further.
International Journal of Production Research 15

5. Conclusions
In order to explore the impact of AM on supply chain performance and quantitatively examine the superiority of utilis-
ing AM in a spare parts supply chain, this paper compared the spare parts supply chain utilising AM with a conven-
tional one. The literature on spare parts supply chains and AM was reviewed and the research objective was established
based on the theoretical gaps that were identified –exploring whether the spare parts supply chain adopting AM technol-
ogy is really superior to the conventional one in terms of both total cost and carbon emission. To realise the research
objective, two system dynamic (SD) models were designed for the spare parts supply chain adopting AM and the con-
ventional one, respectively. Using these models, the researchers were able to compare these different types of spare parts
supply chains in terms of cost and carbon emission.
The simulation results showed that the AM-based supply chains are indeed superior to the conventional one in terms
of total variable costs. The dominant cost category for the centralised supply chain adopting AM is transportation cost,
whereas for the decentralised one manufacturing is the biggest cost source. This study also indicated that if the fixed
costs such as the purchasing cost of AM equipment are also taken into consideration, the AM-based supply chain might
not be more cost effective than the conventional one in AM’s current developmental stage. Fortunately, it is believed
that the fixed costs will be reduced along with the development of AM technology.
With regard to the carbon emissions, AM is usually considered as a green manufacturing technology. According to
this study’s simulation results, the total carbon emissions of the spare parts supply chain adopting AM are indeed lower.
The dominant source of the carbon emissions is related to raw materials used for AM manufacturing. Thus, AM can be
a method to control the carbon emission of a spare parts supply chain by shortening its channel length.
In conclusion, this paper bridges the theoretical gaps identified in the existing studies based on SD simulation. It is
a new attempt to explore the influence of AM on spare parts supply chain performance in terms of cost and carbon
emission. By identifying the distribution of variable costs and carbon emissions at various steps in the spare parts supply
chain, this paper deepens our understandings about when and where costs and carbon emissions are being generated for
both conventional and AM-based spare parts supply chains, which can further be used to explore the reduction potential
on costs and carbon emissions. Moreover, it also shows AM, along with its development, can play a more important
role in the sustainable development of AM, resulting in more environmental benefits.
Although AM displays some advantages over the conventional manufacturing technology, company managers still
have to pay attention to their supply chain circumstances when they would like to adopt AM. First, they have to face
the trade-off of environmental benefits and economic benefits. According to the study’s simulation results, managers
might still hesitate to adopt AM for their spare parts supply chains today, if they emphasise economic benefits. Never-
theless, they could be convinced to adopt AM in their supply chains due to the emphasis on energy conservation, such
as the carbon tax and quota. Whether AM is environmentally friendly is still uncertain but it has been shown that the
adoption of AM can shorten the length of supply chains, resulting in smaller carbon footprints and higher energy con-
sumption. In other words, AM indeed provides the possibility for companies to improve the performance of their spare
parts supply chains in terms of carbon emissions by reducing the tiers of their supply chains in practice. In addition,
after deciding to adopt AM in their supply chains, managers still need to make decisions on which supply chain struc-
tures to implement. When the number of service locations is small, the decentralised structure is more preferable due to
lower costs and carbon emissions. In contrast, when there are many service locations in the spare parts supply chain,
the centralised structure might be more suitable because of its cost reduction related to equipment purchasing.
The limitations of this paper are threefold. First, the SD simulation models cannot completely reflect the actual situa-
tions. Many constraints were experienced and assumptions were made to simplify the SD models, which also made the
models unable to reflect the complexity and flexibility of the real situations. When calculating cost and carbon emission,
far too many rules have not been taken into consideration although some have been presented in this paper. Therefore,
the SD models as well as the calculation rules need to be improved in the future research. Second, due to the lack of
real, practical numerical examples, this paper mainly made the comparison based on the data from an existing survey,
which in turn influences the reliability of the simulation results. In future, more empirical studies are needed to obtain
more first-hand data as the input for simulation. Finally, mainly focusing on logistics and transportation, this study has
not considered other aspects of supply chain management such as information, people and organisation. Future research
is thereby needed to explore the influence of these aspects.

Disclosure statement
No potential conflict of interest was reported by the authors.
16 Y. Li et al.

Funding
This paper was supported by the National Natural Science Foundation of China under [grant number 71372007], [grant number
71272164], [grant number 71462009]; the National Social Science Foundation of China under [grant number 15BGL033], [grant
number 14CJY053], [grant number 15BGL132], and Sino-Danish Center for Education and Research.

References

Allen, J. 2006. “An Investigation into the Comparative Costs of Additive Manufacture Vs. Machine from Solid for Aero Engine
Parts.” Accessed July 15, 2016. http://www.dtic.mil/dtic/tr/fulltext/u2/a521730.pdf.
Andersson, J., and J. Marklund. 2000. “Decentralized Inventory Control in a Two-level Distribution System.” European Journal of
Operational Research 127 (3): 483–506.
Angerhofer, B. J., and M. C. Angelides. 2000. “System Dynamics Modelling in Supply Chain Management: Research Review.”
Proceedings of the 2000 Winter Simulation Conference 1: 342–351.
ASTM. 2010. “Standard Terminology for Additive Manufacturing Technologies.” Accessed April 3, 2015. http://enterprise.astm.org/fil
trexx40.cgi?+REDLINE_PAGES/F2792.htm.
Atzeni, E., and A. Salmi. 2012. “Economics of Additive Manufacturing for End-usable Metal Parts.” International Journal of
Advanced Manufacturing Technology 62 (9): 1147–1155.
Baumers, M., C. Tuck, D. L. Bourell, R. Sreenivasan, and R. Hague. 2011. “Sustainability of Additive Manufacturing: Measuring the
Energy Consumption of the Laser Sintering Process.” Proceedings of the Institution of Mechanical Engineers Part B Journal
of Engineering Manufacture 225 (12): 2228–2239.
Berman, B. 2012. “3-D Printing: The New Industrial Revolution.” Business Horizons 55 (2): 155–162.
Birtchnell, T., and J. Urry. 2013. “3D, SF and the Future.” Futures 50: 25–34.
Campbell, T., C. Williams, O. Ivanova, B. Garrett. 2011. “Could 3D Printing Change the World: Technologies, Potential, and Implica-
tions of Additive Manufacturing.” Washington, DC: Atlantic Council. Accessed July 15, 2016. http://www.atlanticcouncil.org/
images/files/publication_pdfs/403/101711_ACUS_3DPrinting.PDF.
Christopher, M. 2011. Logistics and Supply Chain Management. Harlow: Pearson.
Cohen, M. A., N. Agrawal, and V. Agrawal. 2006. “Winning in the Aftermarket.” Harvard Business Review 84 (5): 129.
Economist. 2012. “Additive Manufacturing: Print Me a Jet Engine.” The Economist. Accessed December 12, 2012. http://www.econo
mist.com/blogs/schumpeter/2012/11/additive-manufacturing.
Forrester, J. 1971. “Counterintuitive Behavior of Social Systems.” Technology Review 73 (3): 52–68.
Gardan, J. 2016. “Additive Manufacturing Technologies: State of the Art and Trends.” International Journal of Production Research
54 (10): 3118–3132.
Gebler, M., A. J. S. Uiterkamp, and C. Visser. 2014. “A Global Sustainability Perspective on 3D Printing Technologies.” Energy Pol-
icy 74: 158–167.
Gibson, I., D. W. Rosen, and B. Stucker. 2009. Additive Manufacturing Technologies: Rapid Prototyping to Direct Digital Manufac-
turing. New York: Springer.
Hague, R., S. Mansour, and N. Saleh. 2004. “Material and Design Considerations for Rapid Manufacturing.” International Journal of
Production Research 42 (22): 4691–4708.
Hasan, S., and A. Rennie. 2008. “The Application of Rapid Manufacturing Technologies in the Spare Parts Industry.” Nineteenth
Annual International Solid Freedom Fabrication Symposium, University of Texas at Austin, Austin, August 4–8.
Holmström, J., J. Partanen, J. Tuomi, and M. Water. 2010. “Rapid Manufacturing in the Spare Parts Supply Chain.” Journal of Manu-
facturing Technology Management 21 (6): 687–697.
Hopkinson, N., and P. Dicknes. 2003. “Analysis of Rapid Manufacturing – Using Layer Manufacturing Processes for Production.”
Proceedings of the Institution of Mechanical Engineers, Part C: Journal of Mechanical Engineering Science 217 (1): 31–39.
Huang, S. H., P. Liu, and A. Mokasdar. 2013. “Additive Manufacturing and Its Societal Impact: A Literature Review.” The Interna-
tional Journal of Advanced Manufacturing Technology 67 (5–8): 1191–1203.
Huiskonen, J. 2001. “Maintenance Spare Parts Logistics: Special Characteristics and Strategic Choices.” International Journal of Pro-
duction Economics 71 (1): 125–133.
Khajavi, S. H., J. Partanen, and J. Holmström. 2014. “Additive Manufacturing in the Spare Parts Supply Chain.” Computers in Indus-
try 65 (1): 50–63.
Kruth, J. P., M. C. Leu, and T. Nakagawa. 1998. “Progress in Additive Manufacturing and Rapid Prototyping.” CIRP Annals-
Manufacturing Technology 47 (2): 525–540.
Levy, G. N., R. Schindel, and J. P. Kruth. 2003. “Rapid Manufacturing and Rapid Tooling with Layer Manufacturing (LM) Technolo-
gies, State of the Art and Future Perspectives.” CIRP Annals-Manufacturing Technology 52 (2): 589–609.
Liu, P., S. H. Huang, A. Mokasdar, H. Zhou, and L. Hou. 2014. “The Impact of Additive Manufacturing in the Aircraft Spare Parts
Supply Chain: Supply Chain Operation Reference (Scor) Model Based Analysis.” Production Planning & Control 25 (13–14):
1169–1181.
International Journal of Production Research 17

Manyika, J., M. Chui, J. Bughin, R. Dobbs, P. Bisson, and A. Marrs. 2013. Disruptive Technologies: Advances That Will Transform
Life, Business, and the Global Economy. Vol. 180. San Francisco, CA: McKinsey Global Institute.
Markillie, P. 2012. “A Third Industrial Revolution.” The Economist. Accessed December 5, 2012. http://www.economist.com/node/
21552901.
Morrow, W. R., H. Qi, I. Kim, J. Mazumder, and S. J. Skerlos. 2007. “Environmental Aspects of Laser-based and Conventional Tool
and Die Manufacturing.” Journal of Cleaner Production 15 (10): 932–943.
Nyman, H.J., and P. Sarlin 2014. “From Bits to Atoms: 3D Printing in the Context of Supply Chain Strategies.” 2014 47th Hawaii
International Conference on System Sciences (HICSS), Waikoloa, HI, January 6–9, 4190–4199. IEEE.
Peres, F., and D. Noyes. 2006. “Envisioning E-logistics Developments: Making Spare Parts in Situ and on Demand – State of the Art
and Guidelines for Future Developments.” Computers in Industry 57 (6): 490–503.
Petrovic, V., J. Vicente Haro Gonzalez, O. Jorda Ferrando, J. Delgado Gordillo, J. Ramon Blasco Puchades, and L. Portoles Grinan.
2011. “Additive Layered Manufacturing: Sectors of Industrial Application Shown through Case Studies.” International Journal
of Production Research 49 (4): 1061–1079.
Reeves, P. 2008a. “ATKINS: Manufacturing a Low Carbon Footprint: Zero Emission Enterprise Feasibility Study.” 2nd International
Conference on Additive Technologies, Pruj, Slovenia.
Reeves, P. 2008b. “How the Socioeconomic Benefits of Rapid Manufacturing Can Offset Technological Limitations.” RAPID 2008
Conference and Exposition, Lake Buena Vista, FL, May 20–22.
Reeves, P. 2012. Examples of Econolyst Research-understanding the Benefits of AM on CO2. Derbyshire: Econolyst.
Ruffo, M., C. Tuck, and R. Hague. 2006. “Cost Estimation for Rapid Manufacturing-laser Sintering Production for Low to Medium
Volumes.” Proceedings of the Institution of Mechanical Engineers, Part B: Journal of Engineering Manufacture 220 (9):
1417–1427.
Serres, N., D. Tidu, S. Sankare, and F. Hlawka. 2011. “Environmental Comparison of MESO-CLAD 03 Process and Conventional
Machining Implementing Life Cycle Assessment.” Journal of Cleaner Production 19 (9): 1117–1124.
Simao, H., and W. Powell. 2009. “Approximate Dynamic Programming for Management of High-value Spare Parts.” Journal of Man-
ufacturing Technology Management 20 (2): 147–160.
Sinden, G. 2009. “The Contribution of PAS 2050 to the Evolution of International Greenhouse Gas Emission Standards.” Interna-
tional Journal of Life Cycle Assessment 14 (3): 195–203.
Spengler, T., and M. Schröter. 2003. “Strategic Management of Spare Parts in Closed-loop Supply Chains – A System Dynamics
Approach.” Interfaces 33 (6): 7–17.
Thomas, D. 2015. “Costs, Benefits, and Adoption of Additive Manufacturing: A Supply Chain Perspective.” International Journal of
Advanced Manufacturing Technology 85 (5): 1–20.
Tuck, C., and R. Hague. 2006. “The Pivotal Role of Rapid Manufacturing in the Production of Cost-effective Customised Products.”
International Journal of Mass Customisation 1 (2): 360–373.
Tuck, C., R. Hague, and N. Burns. 2007. “Rapid Manufacturing: Impact on Supply Chain Methodologies and Practice.” International
Journal of Services and Operations Management 3 (1): 1–22.
Walter, M., J. Mcmanus, and H. Yrjölä. 2004. “Rapid Manufacturing and its Impact on Supply Chain Management.” Logistics
Research Network Annual Conference, Dublin, Ireland, September 9–10.
Whittaker, C., M. C. Mcmanus, and G. P. Hammond. 2011. “Greenhouse Gas Reporting for Biofuels: A Comparison between the
RED, RTFO and PAS2050 Methodologies.” Energy Policy 39 (10): 5950–5960.
Wiedmann, T., and J. Minx. 2008. “A Definition of ‘Carbon Footprint’.” Ecological Economics Research Trends 1: 1–11.
Wise, R., and P. Baumgartner. 1999. “Go Downstream: The New Profit Imperative in Manufacturing.” Harvard Business Review
77 (5): 133–141.
Wohlers, T. 2008. Wohlers Report 2008: State of the Industry. Fort Collins, CO: Wohlers Associates.
Wohlers, T. 2014. Wohlers Report 2014: Additive Manufacturing and 3D Printing State of the Industry. Fort Collins, CO: Wohlers
Associates.
Xu, X., S. Meteyer, N. Perry, and Y. F. Zhao. 2015. “Energy Consumption Model of Binder-jetting Additive Manufacturing
Processes.” International Journal of Production Research 53 (23): 7005–7015.
18 Y. Li et al.

Appendix 1.
(1) Adjustment A = 0.25; Units: Day
(2) Delay Y = Materials T2 + Production T1; Units: Day
(3) Demand in order D = INTEG (order rate d-input rate r, 0); Units: n
(4) Demand rate x = INTEGER (RANDOM NORMAL (010,203, 0)); Units: n/Day [0, ?]
(5) Demand X = INTEG (Demand rate x-Usage rate s, 0); Units: n [0, ?]
(6) Differences G = Demand X-Inventory K; Units: n
(7) FINAL TIME = 1000; Units: Day; The final time for the simulation.
(8) INITIAL TIME = 0; Units: Day; The initial time for the simulation.
(9) Input rate r = DELAY FIXED (Order rate d, Delay Y, 0); Units: n/Day
(10) Inventory K = INTEG (Input rate r-Output rate c,20); Units: n
(11) Materials T2 = 2; Units: Day
(12) Order rate d = INTEGER (MAX (Output rate c + Differences G*Adjustment A, 0)); Units: n/Day
(13) Output rate c = DELAY FIXED (Demand rate x, Delay Y, 0); Units: n/Day
(14) Production rate p = 25; Units: n/Day
(15) Production T1 = Demand in order D/Production rate P; Units: Day
(16) Products T3 = 1; Units: Day
(17) SAVEPER = TIME STEP; Units: Day The frequency with which output is stored.
(18) TIME STEP = 1; Units: Day [0, ?]; The time step for the simulation
(19) Usage rate s = DELAY FIXED (Output rate c, Products T3, 0); Units: n/Day

Appendix 2.
The data of energy usage and carbon emission used in the comparison study:
Table B1. Carbon emission data reference 1.

Product Primary material Powder Cumulative


Manufacturing weight Total weight of material production energy production material energy
technology (kg) required in supply chain (kg) (MJ) energy (MJ) (MJ)

CNC manufacturing 0.80 3.20 670.0 – 670


SLM lattice 0.31 0.33 97.6 7.5 105
SLM optimal 0.37 0.40 116.4 8.9 125

Table B2. Carbon emission data reference 2.

Energy per kg of Total weight of Embodies process


Process material processed (MJ) material to process (kg) energy (MJ)

CNC manufacturing 4.29 (kg removed) 2.4 (to be removed) 10.3


SLM lattice 112 MJ (kg deposited) 0.3 (to be melted) 33.6

Table B3. Carbon emission data reference 3 (unit: kg).

Process Raw materials Manufacture Distribution Usage Life cycle

CNC manufacturing 100 2 5 43,779 43,886


SLM lattice 16 5 1 16,238 16,260
SLM optimal 18 7 2 20,339 20,366

CO2 based on energy mix of USA (2007) – 1 kW h = 3.6 MJ = 0.54 kg CO2 / kW h


Example based on 90 M km (long haul) application.

View publication stats

Vous aimerez peut-être aussi