Computers in Industry 65 (2014) 50–63

Contents lists available at ScienceDirect

Computers in Industry
journal homepage: www.elsevier.com/locate/compind

Additive manufacturing in the spare parts supply chain

Siavash H. Khajavi, Jouni Partanen, Jan Holmström *
Aalto University/School of Science, Department of Industrial Engineering and Management, PB 15500, 00076 Aalto, Espoo, Finland

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

Article history: As additive manufacturing (AM) evolves to become a common method of producing final parts, further
Received 2 May 2013 study of this computer integrated technology is necessary. The purpose of this research is to evaluate the
Accepted 31 July 2013 potential impact of additive manufacturing improvements on the configuration of spare parts supply
Available online 19 September 2013
chains. This goal has been accomplished through scenario modeling of a real-life spare parts supply chain
in the aeronautics industry. The spare parts supply chain of the F-18 Super Hornet fighter jet was selected
Keywords: as the case study because the air-cooling ducts of the environmental control system are produced using
Additive manufacturing
AM technology. In total, four scenarios are investigated that vary the supply chain configurations and
Spare parts supply chain
Distributed production
additive manufacturing machine specifications. The reference scenario is based on the spare parts
Aerospace industry supplier’s current practice and the possible future decentralization of production and likely
improvements in AM technology. Total operating cost, including downtime cost, is used to compare
the scenarios. We found that using current AM technology, centralized production is clearly the
preferable supply chain configuration in the case example. However, distributed spare parts production
becomes practical as AM machines become less capital intensive, more autonomous and offer shorter
production cycles. This investigation provides guidance for the development of additive manufacturing
machines and their possible deployment in spare parts supply chains. This study contributes to the
emerging literature on AM deployment in supply chains with a real-world case setting and scenario
model illustrating the cost trade-offs and critical requirements for technology development.
ß 2013 Elsevier B.V. All rights reserved.

1. Introduction costs related to slow-moving parts. Throughout recent decades,

The digitalization of after-sales, particularly spare parts sales, is
both a strategic opportunity [1] and a threat [2] for many original
equipment manufacturers, as the focus of the competition
increasingly shifts away from the price and quality of the offerings
toward the delivery of value to customers [3]. Customer value is
related to keeping the products in operational condition with high
reliability. Maintenance, repair and operations (MRO) are closely
tied to the accessibility of proper parts and skills whenever a
demand occurs to satisfy the needs of customers and reduce
downtime costs. However, the ability to provide the necessary
parts with high fulfillment rates at low costs is a major challenge to
overcome [4] and one that digital manufacturing technologies
promise to resolve [5].
Conventionally, firms must invest heavily in their spare parts
supply chain operations to reach high fulfillment rates and
reliability [6]. 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
* Corresponding author. Tel.: +358 503675973.
E-mail address: jan.holmstrom@aalto.fi (J. Holmström).
ICT-enabled technologies such as enterprise resource planning
(ERP) have effectively addressed some of these problems. Now the
question is if this trend can continue as technologies based on
digital manufacturing mature. This study investigates the effects of
utilizing additive manufacturing (AM) to produce spare parts
within the structure of the spare parts supply chain. The scenario
analysis indicates how developments of AM machines enable novel
configurations of the spare parts supply chain for simultaneously
improved efficiency and increased customer value.
The remainder of this paper is organized as follows. Section 2
presents a literature review. Section 3 explains the research
methodology. Section 4 presents the findings and results of our
analysis. This paper ends with conclusions summarizing the
research outcomes, and suggestions for future investigation are
provided.
2. Literature review
In the following section, our aim is to introduce the reader to
concepts that have been utilized in this research and to construct a
theoretical foundation based on prior published literature. The
challenges to spare parts provision, the additive manufacturing
production method and distributed production are reviewed.

0166-3615/$ – see front matter ß 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.compind.2013.07.008
 S.H. Khajavi et al. / Computers in Industry 65 (2014) 50–63
2.1. Challenges of managing spare parts supply chains
According to the provided description by the International
Institute of Management, the main characteristics of an excellent
supply chain are the following [7]: delivering high quality
customer responses, efficiently converting inputs into outputs
and improving asset utilization. These characteristics also play a
crucial role in the excellence of spare parts supply chains. Spare
parts supply chain management attempts to reduce operating
costs while keeping the customers’ satisfaction level at an
acceptable level [8]. To accomplish this, the company needs to
overcome a number of challenging issues. A major issue is the
unpredictability of demand, especially for new product launches
for which the data on parts failure rates are unavailable [9]. In
uncertain demand situations, delivering customer satisfaction
(low downtime costs) leads to higher inventory levels in more
locations. Another important challenge is that companies need to
support the previous generation of their product as well as their
new products. This obligation magnifies the number of stock-
keeping units in after-sales inventories. Moreover, the manage-
ment of the after-sales supply chain addresses a combination of
workforce, parts and equipment issues [6]; thus, to be effective and
efficient, a proper combination must be employed. The correct
parts without the right technician are useless and vice versa. These
challenges make it difficult for the supply chain managers to
deliver a high level of service with a low cost with regard to spare
parts.
Based on the data from the United States Logistics and
Materiel Readiness Office, in 2009 the military spent $194bn on
its logistics operations and spare parts supply chain manage-
ment, consisting of $104bn in supply, $70bn in maintenance and
$20bn in transportation. At the end of the same year, the US
military held an astonishing 4.6 million stock-keeping units of
spare parts inventory, valued at $94 billion [10]. These eye-
catching numbers present the opportunity for considerable
gains, even with small improvements in the spare parts and
logistics operations [11].
2.2. Additive manufacturing
Additive manufacturing (AM), also known as direct
manufacturing, is a digital technology for producing physical
objects layer by layer from a three-dimensional computer aided
design (CAD) file. A simple explanation of this production process
is as follows. The process begins with generating a three-
dimensional CAD model of the object with all its details and
dimensions. Next, the three-dimensional CAD file is sliced into very
thin two-dimensional (2D) cross sections (layers) by a computer
program. Then, the 2D layers are sent to the three-dimensional
printing machine one layer at a time. The machine produces the
object by building each layer on top of the previous one, utilizing
different solidification methods of raw material in its production
chamber [12,13]. The process may take from a few hours to a few
days to produce an object, depending on its size and required
production precision.
This technology, originally introduced as rapid prototyping (RP)
and three-dimensional printing, was invented and put into use
throughout the 1980s [14] as a method for producing rough
physical prototypes of final products. Since then, it has continued
to evolve in different aspects. Currently, increasingly more parts
produced with this method are reaching the suitable precision and
quality necessary to be used as final functional parts for special
applications, such as air-cooling ducts for aircrafts or hearing aid
and prosthesis equipment [15]. The technical committee within
ASTM (American Society for Testing and Materials) International
recently adopted the term additive manufacturing for the
51
technology, as it is no longer solely for prototype production.
The advancements that have made AM possible are widening the
material range used by the machines, improving precision and final
quality and reducing machine acquisition cost [13]. During the last
few decades, additive manufacturing equipment has followed the
digital technology progress model, aka ‘‘Moore’s law’’, in that more
capable and cheaper machines have been introduced year after
year.
AM technology, with its swift advancements (e.g., precision,
speed, affordability, and materials range) and inherent capabilities,
has the potential to fundamentally revolutionize manufacturing
operations and supply chains [2,16]. Jeff DeGrange, former
manager of Boeing Phantom Works’ Direct Manufacturing Process,
once said, ‘‘One day, we’ll be building parts on demand in space, on
aircraft carriers and at other points of use’’ [17]. Prior research
[5,18] studied this potential for a spare parts supply chain and
indicated the feasibility of introducing distributed production by
utilizing AM technology. Additionally, Lindemann and his collea-
gues [19] conducted a study comparing additive manufacturing to
conventional manufacturing. Their research focuses on the
additive manufacturing of metal parts and indicates the benefits
of AM on the lifecycle costs of parts.
What makes this production method a potentially disruptive
technology for supply chain management is its characteristics.
Holmström et al. [5] highlight the following benefits of AM
methods over the conventional manufacturing methods:
- No need for tooling (economies of scale does not exist, which
makes customization and design revisions possible).
- Feasibility of producing small production batches economically.
- Possibility for quickly change design.
- Product optimization for function.
- More economical custom product manufacturing (batch of one)
plus the capability to produce complex geometries.
- Potential for simpler supply chains with shorter lead times and
lower inventories.
In addition, there is the possibility of reducing material waste
by as much as 90% according to an Economist special report [2] on
additive manufacturing.
These characteristics may enable the supply chain managers to
manufacture any part (including customized) at any time in
various locations and batch sizes without the need to be concerned
about massive tooling costs. The continued evolution of AM will
offer original equipment manufacturer companies the opportunity
to change their supply chain configurations in the future by
introducing distributed production.
2.3. Distributed production
One of the decisions that may lead to significant improvements
in spare parts supply chain performance is changing the location of
production facilities. There are two basic options to select from.
Concentrating production facilities in a centralized location and
serving the world market from that location is one option. The
other option is decentralizing production in various regional or
national locations close to major markets [20].
There are several cases where companies or public organiza-
tions in different industries have chosen to decentralize their
production activities (goods or services). However, a common
theme in all cases has been an organization’s ability to balance its
customers’ value with supply chain costs. Production supply chain
configuration decisions may bring essential benefits such as lead-
time reduction, which according to De Treville et al. [21] enables
significant improvements to the service level and profitability of
the firm. Thus, such decisions have the potential to increase the
52 S.H. Khajavi et al. / Computers in Industry 65 (2014) 50–63
competitiveness of the company if chosen in a suitable case and if
implemented correctly [22,23].
Different studies have suggested various benefits for the
implementation of distributed strategies in the production of
goods and services [20,24,25,26,27,28]. There are also case
studies related to deficiencies and disadvantages from a supply
chain management perspective [29]. The most common gains
are achieving greater responsiveness and flexibility, improved
decision making based on more accurate information from local
conditions, improved efficiency and higher supply chain
reliability [28]. Moreover, for tangible goods production, lower
capital investment in each facility, lower shipping costs and in
some cases lower inventory-related costs (which might also be
categorized under efficiency improvements) can also be among
the gains from a distributed production strategy [26]. On the
other hand, the most important disadvantages are closely
related to a lack of effective management and information flow
between and from distributed production sites [24,28,29].
Although the abovementioned benefits and shortcomings of
distributed production are generic, there is a necessity for
companies to conduct a cost-benefit analysis (CBA) before
making any decision on the implementation of a distributed
strategy, as different industries and distinct operation settings
affect the trade-offs.
2.4. An identified gap in the literature
There are studies that investigate the utilization of additive
manufacturing to produce functional parts. Lindemann et al. [19]
compared traditional production and additive manufacturing
products based on their lifecycle costs, Hopkinson [30] and Ruffo
[31] investigated the same question, solely focusing on the
production cost of different methods. Other efforts, by Holmström
et al. [5], Pérès and Noyes [18] and Chawla et al. [32], go one step
further and study AM in supply chains. However, the previous
research has not investigated which are the critical improvements
required to enable distributed production in specific supply
chains. The research reported here addresses this gap by
investigating the impact of the evolution of AM technology on
the spare parts supply chain of air-cooling ducts for the F-18 Super
Hornet fighter jet.
3. Methodology
The methodology of research is scenario modeling [33,34,35] in
a deductive case study [36]. The scenario models investigate the
Fig. 1. Design, production, testing and implementation of additive manufactured parts (air-cooling ducts), deployed in the F-18E/F aircraft [38].
use of additive manufacturing to produce functional spare parts for
F-18 Super Hornet fighter jets. The starting premise is that the
evolution of AM technology will enable a switch to the distributed
production of spare parts. To be determined are the critical
requirements on the development of technology to enable the
change.
The case was selected because it is one of the first
implementations of AM in final product manufacturing. Super
Hornet jets carry approximately one hundred additive manufac-
tured parts for their air-cooling ducts. This case selection provided
us with the necessary data to model alternative spare parts
provision scenarios. The sources of information used in this paper
consist of books, articles, e-books and subject-related websites. In
addition, one of the authors has been involved in the provision of
AM technology to the aeronautics industry, providing a basis for
the expert interpretation of public sources of information.
3.1. Case history
In the year 2000, the US Navy selected Boeing Company as its
contractor to develop F/A-18E/F Super Hornet fighter jets. The
project goals were to reduce the production cost of the forward
fuselage and to shorten the planes’ manufacturing cycle from 34 to
18 months. Improving product quality and adding six new avionics
systems were also important objectives. These new systems
required additional cooling and associated parts such as electrical
covers and shrouds. All these parts were subject to limited space
inside the airplane. To manufacture the air-cooling ducts
(environmental control system, or ECS) and meet the functional
requirements and limitations, selective laser sintering (SLS)
additive manufacturing technology was selected (Fig. 1). SLS AM
technology enabled the engineering team to combine different
ducts into single parts, to integrate the attachment mechanisms
into them and to reduce the overall number of parts. This was
made possible by complex geometries utilizing the AM method.
The results were the simplification of the installation process,
shortening the required time for installation and weight reduction
[37].
3.2. Scenario modeling
A total of four scenarios in two dimensions were modeled.
The first dimension was the supply chain configuration and the
second dimension was the development of the AM machine
technology (Fig. 2). The total operating costs of the four
scenarios are compared. The downtime cost reflecting customer
 S.H. Khajavi et al. / Computers in Industry 65 (2014) 50–63
Scenario 3 Scen
nario 4
Future
AM machine tech.
Scenario 1 Scenario 2
Current
Centralize Distribute
D d
Supp ly Chain Config
guration
Fig. 2. Differences among the investigated scenarios.
value is included in the total operating cost. For instance,
Scenarios 1 and 4 are distinct both from the viewpoint of
technological evolution and supply chain configuration. Technolog-
ical evolution refers to either current or future AM machines. Supply
chain configuration refers to centralized production or distributed
production.
Fig. 3(a) displays the centralized production configuration for
spare parts provision. It indicates naval air stations and master jet
bases as the consumption centers being supplied from a single
production facility in the southwest of the U.S. (California). For the
purpose of simplification, in this paper we limited the spare parts
supply chain investigation to the contiguous United States,
yielding twenty consumption locations.
Fig. 3(b) shows the distributed production scenarios. In this
setting, every naval air station and master jet base (consumption
points) has deployed AM technology to meet its spare parts
demand. In this setting, parts transportation for inventory
replenishment is assumed to be null.
Our reference scenario, Scenario 1, is centralized spare parts
production deploying current AM machines. The operating cost of
this scenario was modeled to reflect the current practice of the
spare parts supplier. To accomplish this task, we combined
publicly accessible data and expert views where sufficient data
were unavailable (Table 1). Detailed data related to the reference
scenario are presented in Appendix A.
The operating cost of each scenario consists of the following
cost components: personnel, material, transportation, inventory
carrying, aircraft downtime, inventory obsolescence, initial
investment in AM machine depreciation and annualized cost
of initial inventory production. To calculate the quantity of each
component, we relied on the accessible data and optimization
method, which will be explained here briefly. After setting
the annual expected demand for the air duct spare parts at a
specific level (10% of all installed parts on the fighter jets
according to Lyons [38]), a simulation was run to determine the
stockout probabilities. The utilized method was Monte Carlo
simulation for various inventory levels and replenishment time
intervals.
By applying the probability results from the simulation, the
unique spare parts provision operation settings were identified
that yielded each scenario with the least operating costs. In the
next step, by implementing the optimal settings, the total cost
components of the current and future AM machine specifications
53
were analyzed to understand why and how the shift between the
centralized and distributed supply chain configuration occurs. In
other words, we wanted to know under which circumstances
distributed production becomes more cost effective. Furthermore,
the analysis indicates the cost components with a significant effect
on the total operating cost. The research process is described in
Fig. 4. For a detailed description of the simulation and optimization
processes, refer to Appendix B.
The future additive manufacturing machine scenarios were
designed to assess the impact of AM machine development in the
reduction of the overall cost of spare part provision. The future
machine concept also helped us indicate, among the AM machine
specifications, which specifications have the greatest influence on
the supply chain configuration.
In the model, a future AM machine is realized by altering two
of the machine’s key specifications: the automation level and the
size of the production chamber. In this paper, the automation
level corresponds to the number of AM machines operable by a
single dedicated employee. Therefore, by increasing the auto-
mation level, additive manufacturing machines become less
dependent on the human factor in the pre- and post-production
preparation activities. Additive manufactured parts usually
require pre- and post-production cycles before they are suitable
for utilization. According to Kumar [44], cleaning, surface
polishing and testing are among the most typical postproduction
activities.
In our model, we assumed the future machine would contain
production chambers that are six times smaller, with a consequent
six-fold reduction in acquisition price (Table 2). The future AM
machine has the same production rate as the current machine;
however, it requires less time to produce parts because the future
machine only manufactures a single part in each run in contrast to
six parts for the current AM machine. For further details regarding
the future AM machine and information regarding its feasibility,
refer to Appendix C.
4. Results
This section presents outcomes of our analysis to answer the
research questions. Tables and graphs have been utilized to
accommodate the illustration of findings.
As mentioned above, the total operating cost is used as
the performance measure to compare scenarios along our
defined dimensions. The purpose is to identify the favorable
spare parts supply chain configuration in different conditions. The
annual total cost formula for this case includes the following cost
factors:
 Personnel cost is an important cost component that depends on
the automation level of AM machines and the supply chain
configuration. This cost component is a major differentiator
among current and future scenarios.
 Material cost depends on the level of demand. Therefore, because
we compare different scenarios on the same demand level, it
does not influence the outcome of this study. (To avoid too much
complexity, we fixed the annual demand for spare parts at a
constant level of 5000 and estimated the total costs of scenarios
based on this level.)
 Transportation costs depend on the replenishment policy, which
varies among scenarios.
 The inventory carrying cost is a component of total cost that is
closely related to the inventory level. This cost varies for the
scenarios in each category, as different supply chain configura-
tions have different stock levels.
 Aircraft downtime cost is among the most important cost
components in this case, as we address capital-intensive
54 S.H. Khajavi et al. / Computers in Industry 65 (2014) 50–63

Fig. 3. Visual illustration of production configurations. (a) Centralized production (single manufacturing site). (b) Distributed production (manufacturing capability at every
service facility).

Table 1 equipment such as aircraft. The downtime cost depends on the

Model figures for the reference scenario, Scenario 1. supply chain configuration, inventory fulfillment policy, inven-
tory level and pace of parts production.
Item Quantity
 Inventory obsolescence cost is a function of inventory level. This
F/A-18E/F Super Hornet flight lifetime (hours) 9000
cost component is not significant due to the special character-
Aircraft life span (years) 30
F/A-18E/F Super Hornet average unit M$66.9
istics of this case, as the stock levels are low. More generally,
cost (year 2012) inventory obsolescence cost is a major problem for slow-moving
Number of F-18 Super Hornets (units) 500 spare parts [6].
Number of different parts (SKUs) per 100  The depreciation costs of initial investment in AM machines are
environmental control system
derived from the AM machine acquisition price and are
Number of aircraft deployment locations 20
inside the US (Naval Air Stations) influenced by the supply chain configurations.
Average material cost for production of $100  The annualized cost of producing the initial inventory is the
each part (assumption) summation of material, personnel and machine depreciation
Procurement price of each appropriate k$350
costs derived from the production of initial inventory divided by
AM machine (sProTM 60 HD)
Average annual salary per employee k$60
the lifespan of the project.
(assumption)
Total number of parts at the installed base 50,000 For more information regarding cost components, refer to
Sources: [37,39,40,41,42,43]. Appendix D.
 S.H. Khajavi et al. / Computers in Industry 65 (2014) 50–63 55

Fig. 4. Research process.

4.1. Current AM machine scenarios: more efficient centralized 4.2. Future AM machine scenarios: distributed production becomes
production feasible

Table 3 presents the results for Scenarios 1 and 2, where the
production uses AM machines with current specifications. The
table clearly demonstrates the favorability of centralized produc-
tion, as it yields a lower total cost compared with distributed
production. The reason for this significant operating cost difference
is the number of AM machines utilized in each scenario. Another
reason is the high acquisition cost of AM machines, which leads to a
high annual depreciation cost. Moreover, the distributed produc-
tion requirement for widespread AM machine deployment leads to
significant personnel costs. Centralized production, on the other
hand, yields a higher cost for spare parts transportation, total
inventory carrying, inventory obsolescence and aircraft downtime.
To clarify the effect of AM machine specifications on the spare
parts supply chain configuration, a similar total cost analysis was
conducted on different specifications for AM machines. These
specification differences are assumed to affect the form factor
(size) and capital intensity (money and labor) of AM machines. The
changes led to the ‘‘future AM machine concept’’ of this study,
which was described in Section 3.
The results for the total cost analysis of Scenarios 3 and 4 are
provided in Table 4. According to the model, there is a slight cost
advantage for the distributed production scenario compared with a
clear disadvantage when the current AM machine is used. This
illustrates a significant change in total cost values of the

Table 2
Comparing the current and future AM machine specifications.

AM machine improvement aspects Current AM machines Future AM machines (assumptions)

Automation level (operator: machine) 2:5 1:15

Production chamber size 381 mm  330 mm  457 mm 190 mm  165 mm  305 mm
Producible parts per run in this case 6 1
Production speed (hours) 50 8.33
Procurement price k$350 k$58.33
Production rate (parts per year) 1050 1051
Source: [45].

Table 3
Total cost breakdown of scenarios while ‘‘current AM machines’’ are deployed.

Item description Scenario 1 Scenario 2

Personnel cost $120,000 $480,000

Material cost $500,000 $500,000
Spare parts transportation costs $58,700 $0
Inventory carrying cost $91,500 $60,000
Aircraft downtime cost $17,107 $1909
Inventory obsolescence cost $47,638 $31,238
Initial investment in AM machines, $175,000 $700,000
depreciation costa
(Total initial investment in AM machines $1,750,000 $7,000,000)
Annualized cost of initial inventory productionb $31,759 $20,825
(Total production cost of initial inventory $952,762 $624,762)
Expected total cost of scenario per annum $1,041,704 $1,793,972
a
10% of total initial investment in AM machines
b
Total production cost of initial inventory divided by the lifespan of the aircrafts (30 years)
56 S.H. Khajavi et al. / Computers in Industry 65 (2014) 50–63

Table 4
Total cost breakdown of scenarios while ‘‘future AM machines’’ are utilized.

Item description Scenario 3 Scenario 4

Personnel cost $20,000 $80,000

Material cost $500,000 $500,000
Spare parts transportation costs $58,700 $0
Inventory carrying cost $91,500 $30,000
Aircraft downtime cost $17,107 $12,723
Inventory obsolescence cost $33,354 $10,936
Initial investment in AM machines, depreciation costa $29,167 $116,667
(Total initial investment in AM machines $291,667 $1,166,667)
Annualized cost of initial inventory productionb $22,236 $7,290
(Total production cost of initial inventory $667,073 $218,712)
Expected Total Cost of Scenario per Annum $772,064 $757,616
a
10% of total initial investment in AM machines.
b
Total production cost of initial inventory divided by the life span of the aircrafts (30 years).

distributed production scenario compared with the current AM
machines, which means changing the specification of AM
machines potentially enables distributed production, providing
customers with simultaneously faster after-sales services (lower
downtime) to lower total costs. However, three major cost
components of this scenario remain: material, AM machine
depreciation and personnel costs. These costs are similar (but in
Scenario 3 √ Scenario 4
Future
AM machine tech.
a different order) for Scenario 2 which relies on current AM
technology.
Fig. 5 briefly presents the outcomes of the previous sections.
First, considering the current situation, where additive
manufacturing machines are capital- and labor-intensive and
large (sizable production chamber), the analysis suggests that the
centralized spare part production (Scenario 1) is more efficient
than distributed production by a wide margin. In contrast, when
we investigated a future situation in which AM machines are

√ Scenario 1 Scenario 2
Current

Centralized Distributed

Supply Chain Configuration

Fig. 7. Impact of AM machine acquisition price on total cost of future machine
Fig. 5. Optimal spare parts supply chain scenarios. scenarios.

Fig. 6. Impact of annual personnel salary on the total cost of future machine
scenarios. Fig. 8. Impact of future AM machine automation level on total cost.
 S.H. Khajavi et al. / Computers in Industry 65 (2014) 50–63 57

Fig. 9. Total operating cost breakdown for all investigated scenarios.

smaller, cheaper and more automated, the distribution of
production (Scenario 4) becomes the supply chain of choice. This
result is significant because it suggests the possibility for a radical
change in spare parts supply chains.
In this case, the influential cost components that affect the shift
between cost efficient scenarios are personnel, transportation,
inventory carrying and aircraft downtime costs in addition to
depreciation cost of investment in additive manufacturing machines.
4.3. Analysis of outcomes
To identify the critical values (related to shifting points) for
important model variables, the impacts of annual personnel
salary, AM machine procurement price and AM machine
automation level on the total cost of the future scenarios were
investigated (Figs. 6–8).
It is observed in our analysis that the personnel salary level has
a positive linear relationship with the total cost of both Scenarios 3
and 4. However, the same increase in the salary of AM machine
operators in the two scenarios brings about a higher total cost for
the distributed production configuration scenario (Fig. 6). For
every $10,000 personnel salary increase, the total cost of Scenario 3
grows by $3650, while the increase is $13,400 for Scenario 4. The
reason for this phenomenon has its roots in the higher number of
deployed AM machines in Scenario 4. In this study, the threshold
for annual salary per AM machine operators, which entails the
switch between efficient supply chain configurations, is approxi-
mately $70,000.
58 S.H. Khajavi et al. / Computers in Industry 65 (2014) 50–63

Table 5
Total cost difference sensitivity analysis for future AM machine supply chain configurations (positive values indicate lower cost for distributed production).

Acquisition price of AM machine (k$) Automation level of AM machine

4 5 7 10 12 15a

5 7.88% 3.04% 3.45% 9.18% 11.64% 14.26%

10 8.62% 3.90% 2.43% 8.00% 10.39% 12.93%
15 9.35% 4.73% 1.44% 6.85% 9.18% 11.64%
20 10.06% 5.55% 0.47% 5.74% 8.00% 10.39%
30 11.43% 7.12% 1.40% 3.60% 5.74% 8.00%
40 12.74% 8.62% 3.17% 1.58% 3.60% 5.74%
50 14.00% 10.06% 4.85% 0.34% 1.58% 3.60%
58.33a 15.01% 11.20% 6.19% 1.87% 0.03% 1.91%
60 15.21% 11.43% 6.46% 2.17% 0.34% 1.58%
65 15.79% 12.09% 7.23% 3.04% 1.27% 0.60%
70 16.36% 12.74% 7.99% 3.90% 2.17% 0.34%
100 19.57% 16.36% 12.19% 8.62% 7.12% 5.55%
a
Assumed AM machine acquisition price and automation level for the future AM machine in this study.

The purchasing price of AM machines is similar to personnel
salary level and affects the total cost of both scenarios in a positive
linear way. However, the growth rate is higher for Scenario 4. For
every $10,000 AM machine price increase, the total costs of
Scenarios 3 and 4 grow by nearly $5500 and $20,150, respectively.
This is a wider impact than the personnel salary increase. In this
case, distributed production loses its total cost advantage when the
AM machine acquisition price approaches $65,000 per AM
machine (Fig. 7).
We introduced the AM automation level as one of the future
additive manufacturing machines’ features, which corresponds to
the number of AM machines operable by one operator. Our analysis
indicates this feature’s importance as a facilitator for a shift toward
the distribution of production (Fig. 8). This factor has a negative
non-linear relationship with the total cost of the scenarios;
however, it affects Scenario 4 more than Scenario 3. In this case,
Scenario 4 becomes the most cost efficient when the AM machine
automation level increases above 13. This scenario refers to
additive manufacturing machines with the capability to work
autonomously for most of the pre- and postproduction processes
with low human interference.
A sensitivity analysis for Scenarios 3 and 4 was performed to
indicate the effect of the AM machine’s acquisition price and
automation level on the scenarios’ total cost. The positive values in
Table 5 suggest the cost advantage of the distributed over the
centralized production configuration.
It is important to mention that the capacity utilization of
machines in centralized production is nearly 95% compared with
approximately 25% for the distributed production configuration.
The lower capacity utilization gives the distributed production
scenarios a wide flexibility potential. The extra capacity might be
utilized to adapt to possible demand fluctuations or to manufac-
ture other parts.
A cost factor we ignored in our model for the sake of simplicity
is the role of an ICT-based inventory control system. In centralized
production there are extra costs related to necessary personnel and
equipment. However, this cost does not exist in distributed
production, as each location is capable of producing its own parts
independent of other sites.
A visual comparison of all the investigated scenarios is
presented in Fig. 9. This figure facilitates a comparison of cost
components among our four scenarios.
5. Conclusion
This study is a direct response to Holmström et al.’s [5] call for
further research. The paper examined the readiness of additive
manufacturing (AM) technology in its current state and with the
future evolution of technology to affect change in spare parts
supply chains. We investigated the provision of spare parts for the
F-18 Super Hornet environmental control system, which is one of
the earliest and most well-publicized deployments of AM
technology for functional product manufacturing. The results
suggest the AM machine acquisition price and personnel
intensiveness are major obstacles to a distributed deployment of
this technology in spare parts supply chains. The slow production
rate compared with spare part shipments from a central
warehouse is a further issue to be solved.
Analysis of the modeling results led us to consider what the
critical requirements on a future AM machine to resolve the
current shortcomings of this technology for distributed deploy-
ment in spare parts supply chains are. By utilizing the exact same
model and only changing the specification of AM machines (future
machine), distributed production scenarios could be identified
where the total cost is lower than of the centralized production.
This suggests that the higher automation, lower acquisition price
and shorter production time of AM machines are the factors that
the producers of these machines should aim to develop to enable
radical change in spare parts supply chain operations. Another
team of researchers who have identified similar factors for
choosing between AM and conventional production technologies
are Lindemann et al. [19]. Their research indicated that the main
cost drivers of AM-produced parts are AM machine investment,
raw material costs and labor costs of pre- and post-production
activities.
In addition to the potential for economic benefits (sharp
decline in transportation costs, inventory carrying cost, part
obsolescence cost and less necessity for sophisticated ERP
systems), a shift toward distributed production also provides
the organization with a production capacity buffer that may be
utilized, as spare parts are increasingly produced using AM. The
capacity buffer also increases firm flexibility during unexpected
surges in demand or enables capacity being sold to other vendors.
Moreover, a distributed logistics operation may decrease the need
for 3PLs (third-party logistics providers) and freight forwarders,
as there will not be as great need to transport parts around the
globe.
The following potential advantages of using AM machines for
the distributed production of spare parts can be summarized as
follows: lower overall operation costs, lower down time, higher
potential for customer satisfaction, lower capacity utilization,
higher flexibility, higher robustness to supply chain disruptions,
reduced need for inventory management and logistics information
systems, and potential for sustainability improvements as AM
machines become smaller and more energy efficient. This is in line
with the findings of Sutherland et al. [46].
 S.H. Khajavi et al. / Computers in Industry 65 (2014) 50–63
The field of digital manufacturing is developing even now, and
there are many possibilities for future studies in various aspects of
digital manufacturing, which may affect business operations.
Important areas we consider worth exploring in future research are
the following: extending the model presented in this paper to
study stochastic demand levels and various production config-
urations, the utilization of AM technology in non-military
operations and industries with lower downtime costs, such as in
the auto industry, a comparison among conventional production
methods and additive manufacturing alternatives when produc-
tion volumes are low [30,31]. In addition, the development of
generic models to compare different AM production configurations
is needed.
Appendix A. Case study facts and figures
Case summary
 The number of F-18E/F Super Hornet aircrafts operated by the
United States navy is approximately 500 units [40]. The Navy
deploys these fighter jets in 25 squadrons that are based at three
Master Jet Bases (Oceana and Lemoore inside the United States
and Atsugi in Japan). These squadrons can be deployed all over
the world in different air bases or on aircraft carriers.
 Moreover, we assume there are 20 Naval Air Stations [42] inside
the United States, two of which are Master Jet Bases. The two
Master Jet Bases are located at Lemoore in the western part of the
United States and Oceana in the eastern part of the country. The
bases are indicated by stars in Fig. 3(a) and (b). In the centralized
spare parts supply chain configurations, we assumed these bases
to be our spare parts distribution centers.
The aircraft
 The downtime cost of a fighter jet depends on its mission and can
be very high [6], although for the sake of simplicity it assumed to
be a constant value. This value is $255 per hour for this case and
is the result of dividing the aircraft’s acquisition price by its
lifetime, which is approximately 30 years.
 Each airplane integrates approximately 100 additive manufac-
tured parts (stock-keeping units) in its onboard environmental
control system.
The additive manufacturing machines and parts
 The acquisition price and specifications of the AM machines
deployed to manufacture parts for the F-18 Super Hornet
(originally VanguardTM SLS si2TM) were replaced by a similar
model from the same provider. An sProTM 60 High Definition
(HD) Selective Laser Sintering (SLS) AM machine from the 3D
Systems Company, with an acquisition price of k$350 per unit, is
the machine utilized in our case study. The lifetime of this SLS
machine is assumed to be 10 years with linear depreciation
(Table 7).
 The average cost of the required material for the production of
each part is assumed to be 100$, which also includes the raw
material transportation costs.
 We decided to conduct our analysis on an annual demand of
5000 parts as the fixed demand rate. This number is related to the
environmental control system parts having a lifespan of 10 years
[38].
 The average cost of the carrying inventory is assumed to be 15$
per part per year.
 The transportation costs were retrieved from the UPS Company
website according to the size and destination of the deliveries.
59
 A fixed 5% of inventory parts were assumed to become obsolete
annually in each scenario.
 The annual number of parts each sProTM 60 HD AM machine can
produce is calculated using the production capacity per machine
run (six parts) and a production time of 50 h (the 50 h includes
pre- and postproduction activities as well as the production time
span).
 The specifications of the sPro60 SLS HD (High Definition) AM
machine are presented in Table 6.
Appendix B. Detailed description of our probability simulation
and optimization method
In this research, we simulated the occurrence of expected demand
for spare parts (which was assumed to be 5000 parts annually)
utilizing the Monte Carlo simulation method [47,48]. The process was
performed for different replenishment periods (e.g., every week,
every two weeks) and various inventory levels (e.g., one set of parts at
every location, two sets of parts at every location). To model the
probability of stockouts, spreadsheet software was employed. Table 8
briefly illustrates the actual simulation table.
To study the lumpy demand of spare parts, the period, location and
stock-keeping unit values were modeled using the ‘‘randbetween’’
function. For instance, while inventory replenishment is performed
every two weeks, the function would be ‘‘randbetween (1;26)’’ which
corresponds to the possibility of demand occurrence during any
period between 26 planned inventory refills. It is the same for
location, which is the 20 assumed Naval Air Stations, and for the
demand that may happen for any of the airplanes’ 100 air cooling duct
parts. After modeling the components of the lumpy demand, we
combined them into a single value.
Subsequently, in the next step, combinations with equal values
were counted using the ‘‘count-if’’ function throughout 5000
iterations of the table and were recorded in the last column as 1
or 2 and so on. For instance, a 1 in this column indicates that one
unique demand combination happened twice. Consequently, a
stockout of one specific part at one location occurs if only one set
of parts is being kept at each location. In such cases, there will be a
need for emergency part replenishment.
After determining the number of stockout occurrences for parts,
we calculated the probability of that event by dividing the sum of all
stockout events by 5000, which is the total annual demand. By
counting the separate stockouts for a specific part in a specific
location and period, we can also determine the probability of
stockouts for different levels of inventory.
To complete the modeling, the optimum replenishment interval
and inventory level, which yields the lowest cost for each scenario,
has to be identified (Table 10). By deploying the total cost formula and
input in different inventory levels, the number and type of shipments
(related to inventory replenishment policy) and the simulated
downtime probabilities, the lowest total cost for each supply chain
Table 6
Specifications of the sPro60 HD AM machine [45].
Build envelope capacity (XYZ) 381 mm  330 mm  457 mm
Powder layout Precision counter rotating roller
Layer thickness range (typical) Min 0.08 mm; Max 0.15 mm
Imaging system ProScanTM DX Dual Mode,
High Speed (digital)
Scanning speed (actual) 6 m/s and 12 m/s
Laser power/type 70 W/CO2
Volume build rate 1.8 L/h
60 S.H. Khajavi et al. / Computers in Industry 65 (2014) 50–63

Table 7
Information regarding all scenarios.

Item Scenario 1 Scenario 2 Scenario 3 Scenario 4

Expected spare parts demand (part) 5000 5000 5000 5000

Total number of locations with AM machines 1 20 1 20
throughout the spare parts supply chain
Number of AM machines at each production location 5 1 5 1
Total number of AM machines 5 20 5 20
AM machine automation level (the number of 2.5 2.5 15 15
machines one person can operate)
AM machine lifetime (years) 10 10 10 10
AM machine depreciation rate 10% 10% 10% 10%
Spare parts inventory level (part) 6100 4000 6100 2000
Average inventory carrying cost per part $15 $15 $15 $15
Annual inventory obsolescence rate 5% 5% 5% 5%
Spare parts transportation costs $1000 $200 $100 $100 $1000 $200 $100 $100
Number of shipments per annum 21 188 3 0 21 188 3 0
Assumed transportation time in hours 24 24 24 24
(Source: UPS company website)
Required time to produce one part using AM 50 50 8.33 8.33
(also includes pre- and post-production time span)
Production capability of each AM machine 1050 1050 1051 1051
(parts per year)
Average downtime cost of F-18 Super Hornet fighter $255 $255 $255 $255
jets per hour
Estimated average length of jet downtime due to lack 0.013 0.002 0.013 0.01
of parts per airplane per year (hours)
Average downtime for each breakdown in the optimal 48 5 48 36
scenario (seconds)

configuration can be determined. Table 9, presents the outcome of
this modeling for the reference scenario, Scenario 1.
Appendix C. Details regarding the feasibility of future AM
machine specification
When determining the future additive manufacturing machine’s
specification, we aimed to be realistic. Thus, reference to already-
existing AM machines was made in the formulation of the future AM
machine scenarios. Table 11, illustrates the feasibility of our future
AM machine scenarios by listing smaller machines with significantly
lower prices compared with the currently used large AM machines. To
be precise, an existing AM machine from Formlabs is already very
close to the machines assumed in our future machine scenarios.
The criteria, which we took into account while specifying the
future AM machine, were based on studying the current machine’s
major cost components (personnel and depreciation costs of initial
investment in AM machines). Therefore, the ‘‘future AM machine’’
scenario was designed, utilizing the following assumptions.
First, we proposed that alteration to the current AM machine is
related to the increase of automation; thus, single personnel will be
able to perform the necessary pre-production and post-production
tasks, not just for one machine but for numerous machines. In other
words, each machine only requires a portion of the employee’s time.
The second major change in our future AM machine specification
is related to its size, which influences production volume, production
pace and acquisition price. The future machine is assumed to be six
times smaller and six times cheaper than the current option. Although
the specifications for the future AM machine keep the annual
production rate in par with current rate, this change enables the
future machine to produce in a shorter time span because it
manufactures one part at a time.

Table 8
Implementation of the simulation model.

Period (26) Location (20) Part (100)

The formula in this cell compares the demand

Inventory fulfill- Location of 100 different Demand combi-
Demand count ment period demand parts (SKUs) nation
combination value with other values below it
to identify the number of identical demands

1 15 15 44 15 15 44 0
2 8 45 1
Identical demands

2 2 8 45
3 21 1 62 21 1 62 0
… … … … … …
1346 2 8 45 2 8 45 0
… … … … … …
5000 24 5 57 24 5 57 0

Total stockouts 106

 S.H. Khajavi et al. / Computers in Industry 65 (2014) 50–63 61

Table 9
Costa of various inventory levels and replenishment methods, Scenario 1.

T (Number of weeks between replenishments) N (Number of spare part inventory sets at each consumption location)

1 2 3 4

0.5 $ 1,039,587 $ 748,088 $ 820,781 $ 900,241

1 $ 1,111,338 $ 465,271 $ 529,581 $ 609,041
2 $ 1,676,736 $ 355,066 $ 383,981 $ 463,441
3 $ 2,364,907 $ 369,496 $ 343,146 $ 414,908
4 $ 2,967,809 $ 410,430 $ 324,468 $ 390,641
5 $ 3,726,343 $ 510,748 $ 317,981 (Lowest) $ 377,013
6 $ 4,523,883 $ 647,587 $ 321,935 $ 368,237
7 $ 5,073,576 $ 768,198 $ 336,362 $ 366,519
8 $ 5,792,414 $ 942,766 $ 355,379 $ 364,113
9 $ 6,786,871 $ 1,218,463 $ 394,181 $ 364,105
a
Does not include material, personnel and machine depreciation costs because these cost components are independent of inventory level or replenishment policy.

Table 10
Optimal replenishment policies and inventory levels.

Item Scenario 1 Scenario 2 Scenario 3 Scenario 4

Inventory at production location (Set of parts) 1 0 1 0

Inventory at each location (Set of parts) 3 2 3 1
Inventory refill interval (Week) 5 0.3 5 0.05
Total cost ($) $1,041,704 $1,793,972 $772,064 $757,616

Table 11
Comparison of different existing Stereolithography AM machines.

AM machine ProJet HD 6000 Form 1 Comparison

Manufacturer 3D systems Formlabs –

Year of introduction 2011 2012 –
Technology Stereolithography Stereolithography Same technology
Net build volume (xyz) 250 mm  250 mm  250 mm 125 mm  125 mm  165 mm 6 times
Min layer thickness 50 microns 25 microns Twice accuracy
Acquirement Price k$150 $3299 45 times
Diversity of printable material Wide range of VisiJet print material Only one Very limited choice
Sources: [49,50,51].

Table 12
Our model cost component calculation method and examples.

Cost component Formula Example

Personnel cost ðNumber of AM machines utilized in each specific scenarioÞ Scenario 2:

 ðaverage annual salary per employeeÞ 20  60; 000$
¼ ¼
Automation level of machine 1
Material cost = Level of expected demand  Average material cost for production of each part Scenario 2:
X
n = 5000  $100
Spare parts transportation costsa ¼ Number of type n transportations  Cost of type n transportation Scenario 1:
1 = 20.8  $1000 + 188 
$200 + 3  $100
Inventory carrying cost = Level of inventory in hand  Average annual cost of carrying each part of inventory Scenario 2:
= 4000  $15
Aircraft downtime cost = Number of airplane failures (equal to the number of Scenario 2:
expected demand for the spare parts)  Average downtime cost of = 5000  $255  0.005
an airplane per hour (calculated by dividing the cost of each aircraft by
the length of its life span)  Average number of hours downtime for every
maintenance operation
Inventory obsolescence cost = Annual part obsolescence rate (assumed to be 5%)  Total production cost Scenario 2:
of initial inventory for each scenario = 5%  $761, 905
Initial investment in AM = Number of utilized AM machines  Price of acquiring each AM Scenario 2:
machines, depreciation cost machine  Depreciation rate of AM machines (which is assumed to be 10%) = 20  $350, 000  10%
Material cost þ Personnel salary þ AM machines deprecation cost
Annualized cost of initial ¼ Scenario 2:
Project life span $761; 905
inventory production ¼
30
a
Because we assumed no transportation of finished goods in the distributed production supply chain scenarios, this cost is only taken into account for Scenarios 1 and 3.

Appendix D. Detailed description of cost component Appendix E. Detailed description of inventory replenishment
calculation method method for each scenario
We designed different inventory replenishment methods for
Table 12 the centralized and distributed production configurations. In the
62 S.H. Khajavi et al. / Computers in Industry 65 (2014) 50–63
centralized production scenarios, the AM machines are concentrated
in a single location (Camarillo, California). Periodically, large
shipments of parts are delivered to distribution hubs that are master
jet bases (Oceana and Lemoore in the east and west of the United
States, respectively) by the centralized production facility. Afterward,
the master jet bases distribute the spare parts in smaller shipments to
every naval air station and replenish their used inventory. In the case
of a stockout before the periodic replenishment date, an emergency
delivery from the central production facility takes place.
In the distributed production configuration, we assumed that each
consumption location (Naval air stations and master jet base) deploys
the AM machine to produce necessary spare parts to replenish the
inventory. In this setting, there will not be any part transportation. In
the case of a stockout, the part will be produced on the spot.
References
[1] R. Wise, P. Baumgartner, Go downstream, Harvard Business Review 77 (5) (1999)
133–141.
[2] P. Markillie, A third industrial revolution, The Economist, 2012, Available at:
http://www.economist.com/node/21552901 (accessed 05.12.12).
[3] S.L. Vargo, R.F. Lusch, Evolving to a new dominant logic for marketing, Journal of
Marketing (2004) 1–17.
[4] S. Zanoni, I. Ferretti, L. Zavanella, Multi echelon spare parts inventory optimisa-
tion: a simulative study, in: Proceedings 19th European Conference on Modelling
and Simulation, 2005.
[5] J. Holmström, J. Partanen, J. Tuomi, M. Walter, Rapid manufacturing in the spare
parts supply chain: alternative approaches to capacity deployment, Journal of
Manufacturing Technology Management 21 (6) (2010) 687–697.
[6] M.A. Cohen, N. Agrawal, V. Agrawal, Winning in the aftermarket, Harvard Business
Review 84 (5) (2006) 129.
[7] H. Perumal, Improving Supply Chain in Your Business, International Institute of
Management, 2006.
[8] J. Andersson, J. Marklund, Decentralized inventory control in a two-level distri-
bution system, European Journal of Operational Research 127 (3) (2000) 483–506.
[9] H. Simao, W. Powell, Approximate dynamic programming for management of
high-value spare parts, Journal of Manufacturing Technology Management 20 (2)
(2009) 147–160.
[10] A. Estevez, AIA Product Support Conference, DoD Logistics Focus, Florida, Clear-
water, May 4, 2010.
[11] R. Calantone, C. Dröge, Supply chain flexibility: an empirical study, Journal of
Supply Chain Management 35 (3) (2006) 16–24.
[12] I. Gibson, D.W. Rosen, B. Stucker, Additive manufacturing technologies: rapid
prototyping to direct digital manufacturing, Springer, New York, USA, 2009.
[13] J.P. Kruth, M.C. Leu, T. Nakagawa, Progress in additive manufacturing and rapid
prototyping, CIRP Annals-Manufacturing Technology 47 (2) (1998) 525–540.
[14] D.L. Bourell, J.B. Beaman, M.C. Leu, D.W. Rosen, A brief history of additive
manufacturing and the 2009 roadmap for additive manufacturing: looking back
and looking ahead, in: US-Turkey Workshop on Rapid Technologies, 2009.
[15] G.N. Levy, R. Schindel, J.P. Kruth, Rapid manufacturing and rapid tooling with
layer manufacturing (LM) technologies, state of the art and future perspectives,
CIRP Annals-Manufacturing Technology 52 (2) (2003) 589–609.
[16] Schumpeter, Additive manufacturing: Print me a jet engine, The Economist, 2012,
Available at: http://www.economist.com/blogs/schumpeter/2012/11/additive-
manufacturing (accessed 05.12.12).
[17] W. Cole, Breaking the Mold: Boeing engineers and technologists are constantly
developing better ways to design and make products, Frontiers, 2004, Available
at: http://www.boeing.com/news/frontiers/archive/2004/december/ts_sf03.html
(accessed 10.03.13).
[18] F. Pérès, D. Noyes, 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) (2006) 490–503.
[19] C. Lindemann, U. Jahnke, M. Moi, R. Koch, Analyzing product lifecycle costs for a
better understanding of cost drivers in additive manufacturing, in: Paper pre-
sented at the 23rd Annual International Solid Freeform Fabrication Symposium –
An Addictive Manufacturing Conference, Austin/TX/USA, 6th–8th August, 2012,
https://mb.uni-paderborn.de/fileadmin/dmrc/Research/Publications/dmrc_re-
port_2012.pdf (viewed 18 February 2013).
[20] C.W. Hill, International Business: Competing in the Global Marketplace, 6th ed.,
McGraw-Hill/Irwin, New York, 2007, pp. 286–312.
[21] S. De Treville, R.D. Shapiro, A.P. Hameri, From supply chain to demand chain: the
role of lead time reduction in improving demand chain performance, Journal of
Operations Management 21 (6) (2004) 613–627.
[22] C. Anderson, J.J. Bartholdi, Centralized versus decentralized control in
manufacturing: lessons from social insects, Complexity and Complex Systems
in Industry (2000) 92–105.
[23] S. Li, B. Ragu-Nathan, T.S. Ragu-Nathan, S. Subba Rao, The impact of supply chain
management practices on competitive advantage and organizational perfor-
mance, Omega 34 (2) (2006) 107–124.
[24] T.J. Bossert, D.M. Bowser, J.K. Amenyah, Is decentralization good for logistics
systems? Evidence on essential medicine logistics in Ghana and Guatemala,
Health Policy and Planning 22 (2) (2007) 73–82.
[25] J. Fernquest, Hard-disk drive industry hit, 2011, Available at: http://www.bang-
kokpost.com/learning/learning-from-news/261779/hard-disk-drive-industry-
hit (accessed 01.10.12).
[26] P. Hawken, A.B. Lovins, L.H. Lovins, Natural Capitalism: the Next Industrial
Revolution, Earthscan Publications, London, UK, 2010.
[27] T.W. Malone, Making the decision to decentralize, in: Harvard Business School –
Working Knowledge for Business Leaders, 2004.
[28] T.R. Parry, Achieving balance in decentralization: a case study of education
decentralization in Chile, World Development 25 (2) (1997) 211–225.
[29] P. Nersesian, Y. Chandani, J. Durgavich, S. Rao, K. Crowley, Strategic decentraliza-
tion: centralizing logistics, in: The 129th Annual Meeting of APHA, 2001.
[30] N. Hopkinson, P. Dicknes, Analysis of rapid manufacturing—using layer manufactur-
ing processes for production, Proceedings of the Institution of Mechanical Engineers,
Part C: Journal of Mechanical Engineering Science 217 (1) (2003) 31–39.
[31] M. Ruffo, C. Tuck, R. Hague, 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) (2006) 1417–1427.
[32] Chawla, O.S. Sukaran, J.P. Bihade, K.S. Modak, S. Zakiuddin, A.V. Sathishkumar, M.
Suresh, Krishna, et al., Additive manufacturing and its impact on supply chains,
International Journal of Industrial Engineering and Technology 4 (2012) 79–82.
[33] T.J. Chermack, S.A. Lynham, W.E. Ruona, A review of scenario planning literature,
Futures Research Quarterly 17 (2) (2001) 7–32.
[34] A.P. De Geus, Planning as learning, Harvard Business Review (1988) 70–74.
[35] P. Wack, Scenarios: uncharted waters ahead, Harvard Business Review 63 (5)
(1985) 73–89.
[36] M. Barratt, T. Choi, Mandated RFID and institutional responses: cases of decen-
tralized business units, Production and Operations Management 16 (5) (2009)
569–585.
[37] N. Hopkinson, R.J.M. Hague, P.M. Dickens (Eds.), Rapid Manufacturing, Wiley,
Chichester, England, 2006.
[38] B. Lyons, Additive manufacturing in Aerospace; examples and research outlook,
frontiers of engineering, 2011, Available at: http://www.naefrontiers.org/
File.aspx?id=31590 (accessed 15.08.12).
[39] America’s Navy, United states fact file: F/A-18 hornet strike fighter, 2009, Available
at: http://www.navy.mil/navydata/fact_display.asp?cid=1100&tid=1200&ct=1
(accessed 25.07.12).
[40] P. Carder, Boeing, US navy mark delivery of 500th super hornet/Growler, Boeing,
2011, Available at: http://boeing.mediaroom.com/index.php?s=43&item=1715
(accessed 25.07.12).
[41] J. Gertler, Navy F/A-18E/F and EA-18G Aircraft Procurement and Strike Fighter
Shortfall: Background and Issues for Congress, 2009, Available at: http://www.fa-
s.org/sgp/crs/weapons/RL30624.pdf (accessed 25.07.12).
[42] Navy, Navy facilities within the U.S., 2009, Available at: http://www.navy.mil/
navydata/bases/navbases.html (accessed 25.07.12).
[43] Navy, Department of defense, fiscal year 2013 president’s budget submission,
2012, Available at: http://www.finance.hq.navy.mil/FMB/13pres/APN_BA1-
4_BOOK.pdf (accessed 25.07.12).
[44] S. Kumar, Selective laser sintering: a qualitative and objective approach, JOM:
Journal of the Minerals, Metals and Materials Society 55 (10) (2003) 43–47.
[45] 3D Systems Company High Definition SLS Quality and Versatility, 3D Systems Inc.,
2010, Available at: http://production3dprinters.com/sites/production3dprin-
ters.com/files/downloads/sPro-Family-USEN.pdf (accessed 30.01.13).
[46] W.J. Sutherland, S. Bardsley, M. Clout, M.H. Depledge, L.V. Dicks, L. Fellman, E.
Fleishman, et al., A horizon scan of global conservation issues for 2013, Trends in
Ecology & Evolution 28 (1) (2013) 16–22.
[47] A.F. Seila, Spreadsheet simulation, in: Simulation Conference, 2002, Proceedings
of the Winter, IEEE, vol. 1, (2002), pp. 17–22.
[48] J. Zabawa, B. Mielczarek, Tools of Monte Carlo simulation in inventory manage-
ment problems, in: Paper Presented at the 21st European Conference on Modeling
and Simulation, Prague/Czech Republic, 4th–6th June, 2007, http://www.extend-
sim.com/downloads/papers/sols_papers_wroclaw.pdf (viewed 10 March 2013).
[49] 3D Systems Company, 3D Systems Brings Next Generation 3D Printer to RAPID
(online) 3D Systems Inc., 2011, Available at: http://www.3dsystems.com/sites/
www.3dsystems.com/files/051911-3d-systems-brings-next-generation-printer-
to-rapid.pdf (accessed 20.02.13).
[50] 3D Systems Company, Printer utility, Project 6000 & 7000 professional 3D printers
(online) 3D Systems Inc., 2012, Available at: http://printin3d.com/sites/printin3d.
com/files/downloads/ProJet-6000-7000-USEN.pdf (accessed 02.02.13).
[51] Formlabs, Form 1 Tech. Spec. 2012, Available at: http://cdn.shopify.com/s/files/1/
0183/2285/files/Form-1-Tech-Specs.pdf?2383 (accessed 02.02.13).
Siavash H. Khajavi currently is a doctoral student in the
department of Industrial Engineering and Management
at Aalto University, Finland. He also received his
master’s degrees in Service Management and Engineer-
ing from Aalto University, while holding a Bachelor’s in
Industrial Engineering. His research interests are
Logistics and Supply Chain Management and imple-
mentation of new technologies such as Additive
manufacturing to improve operations management.
 S.H. Khajavi et al. / Computers in Industry 65 (2014) 50–63
Dr. Jouni Partanen is Director of the BIT Research
Centre. He is currently transferring to the position of
Professor of Future Production Technologies. He has
industrial and academic background mostly in United
States, but also in Finland and United Kingdom. For his
academic degree, Dr Partanen studied laser physics, but
his later career for almost two decades has evolved
around new production technique called Additive
Manufacturing.
63
Jan Holmström is professor of Industrial Service and
Maintenance and a team member of Logistics Research
Group at Aalto University, Helsinki, Finland. He is an
expert in supply chain management. His research is
practice-oriented and focuses on problem solving and
solution spotting. He has published extensively on the
improvement of industrial, project and retail supply
chains and asset management.