Vous êtes sur la page 1sur 14

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

net/publication/257337095

Additive manufacturing and its societal impact: A literature review

Article  in  International Journal of Advanced Manufacturing Technology · July 2012


DOI: 10.1007/s00170-012-4558-5

CITATIONS READS

394 1,133

4 authors, including:

Samuel Huang Peng Liu


University of Cincinnati Jilin University
117 PUBLICATIONS   4,621 CITATIONS    2 PUBLICATIONS   455 CITATIONS   

SEE PROFILE SEE PROFILE

Hou Liang
Xiamen University
47 PUBLICATIONS   542 CITATIONS   

SEE PROFILE

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

CMII, CDS Lab, University of Cincinnati View project

Assessment of Crack in Turbo machinery Blade View project

All content following this page was uploaded by Samuel Huang on 03 December 2017.

The user has requested enhancement of the downloaded file.


Production Planning & Control, 2014
Vol. 25, Nos. 13–14, 1169–1181, http://dx.doi.org/10.1080/09537287.2013.808835

The impact of additive manufacturing in the aircraft spare parts supply chain: supply chain
operation reference (scor) model based analysis
Peng Liua,b, Samuel H. Huangb,c*, Abhiram Mokasdarb, Heng Zhoub and Liang Houc
a
Department of Industrial Engineering, Jilin University, Jilin, P.R. China; bSchool of Dynamic Systems, University of Cincinnati,
Cincinnati, OH 45221, USA; cDepartment of Mechanical and Electrical Engineering, Xiamen University, Xiamen, P.R. China
(Received 4 May 2013; final version received 4 May 2013)

Additive manufacturing (AM) technology has the potential to significantly improve supply chain dynamics, reduce
shipping costs and shorten delivery lead times. Using AM technology, manufacturers can produce parts on demand
Downloaded by [University of Louisville] at 18:13 10 February 2015

and thus reduce the need of maintaining safety inventory. This is especially useful in the aircraft spare parts industry
where currently there is a need to maintain a high level of safety inventory for high-cost long-lead time metallic parts.
Therefore, more and more companies in the aerospace industry are interested in using AM technology. There are dif-
ferent approaches to configure the aircraft spare parts supply chain using AM technology. This paper evaluates the
impact of AM in the aircraft spare parts supply chain based on the well-known supply chain operation reference
model. Three supply chain scenarios are investigated; namely, conventional (as-is) supply chain, centralized AM sup-
ply chain and distributed AM supply chain. A case study is conducted based on data obtained in the literature. The
result shows that the use of AM will bring various opportunities for reducing the required safety inventory of aircraft
spare parts in the supply chain. A sensitivity analysis is performed and some key factors affecting the choice of AM
scenarios are studied.
Keywords: additive manufacturing; supply chain operations reference (SCOR) model; aircraft spare parts; safety
inventory

1. Introduction a design in the form of a computerized 3D solid model


Aircraft spare parts manufacturers are often under great can be directly transformed to a finished product without
pressure of the demand of products with high quality the use of additional fixtures and cutting tools. As a
and high cost materials. They have to spend many result, manufacturing lead time can be reduced signifi-
machine hours at high cost rates on precision machine cantly. In addition, the high material and resource wast-
tools machining away anything up to 98% of the original age factors are also avoided. It has been pointed out that
billet to achieve the desired form (Allen 2006). The AM has the potential to reduce the number of stages in
manufacturing lead time for aircraft parts is very long the traditional supply chain (Reeves 2008). Specifically,
using such conventional manufacturing processes. It is AM technology offers two opportunities: (1) to redesign
not uncommon to have lead time in the range of several products with fewer components and (2) to manufacture
months for these parts. On the other hand, the aircraft products near the customers (i.e. distributed manufactur-
spare parts industry requires a very high cycle service ing). The net effect is the reduction in the need for ware-
level (CSL) to meet customer demands. The long pro- housing, transportation and packaging. Therefore, there
curement lead time and high CSL place a heavy burden is a need to study the impact of AM on supply chain
on the aircraft spare parts industry to maintain a high performance.
level of safety inventory, which significantly increases The objective of this paper is to study the impact of
supply chain cost. AM in the aircraft spare parts industry, with an emphasis
The emergence of additive manufacturing (AM) tech- on the use of distributed manufacturing strategy to
nology creates an opportunity to manufacture parts on reduce inventory cost. Section 2 reviews prior work in
demand to improve supply chain dynamics. AM is the two areas: (1) the use of AM in the aircraft spare parts
“process of joining materials to make objects from 3D supply chain and (2) supply chain analysis using the sup-
model data, usually layer upon layer” (ASTM 2010). By ply chain operations reference (SCOR) model. Section 3
the means of creating the final shape by adding material, presents three different supply chain configurations for

*Corresponding author. Email: sam.huang@uc.edu

Ó 2013 Taylor & Francis


1170 P. Liu et al.

the aircraft spare parts industry. The performance of AM technology can potentially change the traditional
these three different configurations is analyzed in Sec- supply chain configuration of various industries. The
tion 4. Finally, conclusions are drawn in Section 5. impact of AM could be complicated, as it not only
impinges on current methodologies but also on the con-
stituent parts of the supply chain (ATKINS 2007). Tuck,
Hague, and Burns (2007) discussed how AM technology
2. Literature review
could change supply chain management thinking.
2.1. Aircraft spare parts supply chain and AM Because AM only requires 3D data and raw material in
Companies along the aircraft spare parts supply chain order to produce a complex part, it will reduce setup and
face significant challenges in providing fast repair and changeover time, and the number of assemblies. Several
maintenance services while minimizing costs. A large researchers have investigated the introduction of AM in
commercial airplane is made up of several millions of the aircraft spare parts supply chain (Walter, Holmstrom,
parts. Most parts are infrequently needed but they have and Yrjola 2004; Hasan and Rennie 2008; Holmstrom
to be kept in stock in order to ensure fast service time. et al. 2010). However, as AM technology has yet to
The current supply chain for aircraft spare parts func- reach the level of maturity of conventional manufactur-
tions as follows. A safety stock of standard replacement ing processes, only a limited range of parts can be pro-
Downloaded by [University of Louisville] at 18:13 10 February 2015

parts is kept at the airline’s in-house warehouse. These duced economically. Therefore, it is necessary to use a
parts can be ordered at regular intervals according to a systematic step by step procedure to determine suitable
maintenance plan. Other less-frequently-used parts are applications. The decisions should be revisited periodi-
purchased from suppliers and often require a 24 h deliv- cally because AM technology is constantly evolving.
ery time to ensure fast service. Therefore, suppliers must Detailed discussion of this procedure can be found in
keep these parts in stock and use an overnight delivery (Walter, Holmstrom, and Yrjola 2004).
service. To keep the stock of those slow moving (i.e. Holmstrom et al. (2010) proposed two different
infrequently needed) parts as low as possible, the strat- approaches to integrate AM technology in the aircraft
egy of demand aggregation is used which requires cen- spare parts supply chain. The first approach is to use
tral warehousing. The trade-off is increased delivery centralized AM capacity to replace inventory holding.
cost. Another problem is related to the production batch AM machines are deployed in centralized distribution
size of these parts. Current production technology centres to produce slow moving spare parts on demand.
requires parts to be produced in large batch size to take Producing parts in a centralized location has the advan-
advantage of economies of scale. However, for slow tage of aggregating demand from various regional ser-
moving parts, it means a long lead time and a large vice locations to ensure that the investment in AM
amount of capital tied up in the form of inventory. capacity is well utilized. The disadvantage is that the
Aircraft spare parts demand pattern follows a 20/80 produced parts need to be shipped to the service loca-
Pareto curve. This means that 80% of the parts are tions, which results in the increase of response time. For
needed frequently and make up most of the airline certain parts that are needed in the first line maintenance,
demands. However, they only account for 20% of the inventory still needs to be carried in the service loca-
supply chain expenditure in terms of holding inventory tions. This approach is desirable when parts that can be
and moving materials. In other words, the majority of produced using AM are limited and the required
the supply chain cost is due to the small amount of slow response time is not critical. The second approach, dis-
moving parts. The spare parts supplier often subsidize tributed AM deployed at each service location, is suit-
the holding cost of these parts with the profits they make able when the demand of AM producible parts is
from the fast moving spare parts. sufficiently high to justify the capacity investment. The
AM technology can cost-effectively produce a small advantage is the elimination of inventory holding and
batch of parts. It also has the potential to reduce the need transportation costs and a faster response time.
of safety inventory. AM describes a collection of produc- In addition to these two approaches, Holmstrom
tion techniques that manufacture parts layer-by-layer et al. (2010) also contemplated the feasibility of mobile
using digital data and raw material as inputs (Baumers AM but conceded that there are many challenges. The
et al. 2011). The technology has advanced rapidly since authors further discussed the trade-off between batch
its inception in the late 1980s. The AM industry is pro- production and on-demand production, and that between
jected to reach $3.1 billion worldwide by 2016 and specialized and general purpose AM. The key variables
$5.2 billion by 2020 (McCue 2012). AM technology is considered are material and production costs, distribution
capable of producing components using a variety of and inventory obsolescence costs and life-cycle costs for
materials. Many aircraft manufacturers are already using the user. They concluded that on-demand and centralized
AM technology to produce high-cost long-lead metallic production of spare parts is most likely to succeed. How-
components. ever, the implications of a move toward AM extend far
Production Planning & Control 1171

beyond the manufacturing process itself. This agile pro- configuration of process elements (Wall, Jagdev, and
duction strategy enables a truly distributed supply chain, Browne 2007). The main objective behind developing
where manufacturing can take place concurrently at mul- SCOR is to help large corporations compare and assess
tiple locations that are located close to the customer. This their different supply chain activities by providing a stan-
new structure could eliminate many stages of the tradi- dard framework (SCC 2012). This framework incorpo-
tional supply chain, affecting lead times, inventory man- rates a set of evaluation metrics and connects business
agement, and transaction and logistics costs. To better processes, technology and best practices for effective
understand the impact of AM on supply chain dynamics, exchange of materials and information between different
it is necessary to apply a systematic modelling and levels of the supply chain. SCOR identifies five core
analysis methodology. supply chain performance attributes: reliability, respon-
siveness, agility, costs and asset management. The SCOR
model provides a common process-oriented language for
2.2. Supply chain management and the SCOR model communicating among supply-chain partners in the fol-
In the past few decades, supply chain management prac- lowing five distinct management processes: PLAN,
tice has been adopted by more and more companies to SOURCE, MAKE, DELIVER and RETURN. Based on
reduce costs, increase market share and sales and build the above expression methods, SCOR provides three lev-
Downloaded by [University of Louisville] at 18:13 10 February 2015

solid customer relations (Ferguson 2000). A supply chain els of process detail. Each level of detail assists a
is a network of facilities that procure raw materials, company in defining scope (Level 1), configuration or
transform them into intermediate goods and then final type of supply chain (Level 2) and process element
products and deliver the products to customers through a details, including performance attributes (Level 3).
distribution system. A supply chain has three key parts Below level 3, companies decompose process elements
(SCW 2012): and start implementing specific supply chain manage-
ment practices. It is at this stage that companies define
• Supply focuses on the raw materials supplied to
practices to achieve a competitive advantage, and adapt
manufacturing, including how, when and from
to changing business conditions.
what location.
SCOR helps companies to examine and measure
• Manufacturing focuses on converting these raw
their supply chain processes, to determine where weak
materials into finished products.
links exist and to identify possible improvements. As a
• Distribution focuses on ensuring these products
cross-industry standard supply chain model, SCOR
reach the consumers through an organized net-
guides the application of analytical tools to the supply
work of distributors, warehouses and retailers.
chain on the basis of process, performance evaluation
Supply chain is very important for all businesses and best practice (Feng, Wua, and Chia 2010). Along
involved in raw materials acquisition, component manu- with today’s higher attention on supply chain perfor-
facturing, assembly, distribution, logistics, retailing, mance, the role of SCOR is more and more significant
information management and customer relation manage- as it can offer a practical approach to determine supply
ment. Designing a reasonable model to describe the ele- chain process decisions and performance metrics. In
ments and performance of a supply chain is the research practice, the SCOR model is typically used by manufac-
priorities of many practitioners and researchers. Gener- turing companies to define their supply chain processes
ally, there are four categories of supply chain models: and activities, and associated performance measures (Li,
deterministic model (Voudouris 1996), in which the vari- Su, and Chen 2011). Building on the terminology and
ables are known and specified; stochastic model (Tzafes- process definitions provided by SCOR, Verdouw et al.
tas and Kapsiotis 1994), where at least one of the (2011) modelled demand-driven supply chain configura-
variables is unknown, and is assumed to follow a partic- tions as specific sets of transformations, control systems
ular probability distribution; economic model (Christy and coordination mechanisms. Naesens, Gelders, and
and Grout 1994); and simulation model (Towill 1991). Pintelon (2007) used the SCOR model to measure the
Among all kinds of supply chain models, the SCOR strategic fit of inventory pooling in a horizontal
model, intended to be an industrial standard, is the most collaborated supply chain. GE used the SCOR model in
widely accepted one. It is a reference model that is used its transportation systems unit, which reported sales of
as a tool to map, benchmark and develop the operations $2.6 billion in year 2001 (Zhou et al. 2011). Huang,
of supply chains. Sheoran, and Keskar (2005) summarized the perfor-
The SCOR model was developed by the Supply mance improvement achieved by AVON, LEGO and
Chain Council in 1996. It provides a balanced horizontal Siemens Medicals using SCOR. The application field of
(cross-process) and vertical (hierarchical) view as SCOR is so wide that over 650 organization members
compared to the classical process decomposition worldwide have taken the initiative in implementing
models, which are developed to address one specific such a model.
1172 P. Liu et al.

3. Configurations and SCOR representations of distributed inventory holding warehouses that are nearer
aircraft spare parts supply chain to the destination. Therefore, three conceptual configura-
The aircraft spare parts supply chain configuration is one tion scenarios of aircraft spare parts supply chain are
of the key success factors in achieving desired supply defined in this work; namely, the conventional as-is sup-
chain performance. According to SCOR, we can ply chain, centralized AM supply chain and distributed
configure the ‘as-is’ thread-diagram or the ‘to-be’ thread- AM supply chain.
diagram of a particular supply chain based on the user’s The conventional as-is aircraft spare parts supply
input. In order to describe and evaluate the performance chain (Figure 1) is described as follows. The original
of the aircraft spare parts supply chain, the SCOR mod- equipment manufacturers (OEM) manufacture the spare
els should be defined depending on the context. Gener- parts and supply them to the regional distribution centres
ally, a supply chain can have either a centralized (RDC) of maintenance, repair and overhaul (MRO) com-
inventory holding capacity that distributes goods to a panies. The RDCs distribute parts to different service
large number of retailers, or more number of smaller locations (SL), where the actual maintenance and
Downloaded by [University of Louisville] at 18:13 10 February 2015

Figure 1. The as-is aircraft spare parts supply chain.

Figure 2. SCOR thread diagram of the as-is aircraft spare parts supply chain.
Production Planning & Control 1173

repairing of the out-of-order aircraft subassemblies take make-to-order, safety inventory is still required in order
place. These out-of-order subassemblies are received to maintain a high CSL. Other spare parts are still manu-
from line replacement units (LRU) near the aircraft oper- factured by OEMs using conventional manufacturing
ators. Inventory of functional subassemblies are kept at facilities. OEMs also supply raw materials to the AM
LRU so they can be brought to the installed bases to facilities. The introduction of AM leads to a number of
replace out-of-order subassemblies of an aircraft quickly. advantages to the aircraft spare parts supply chain. The
The SCOR thread diagram of this as-is supply chain is foremost advantage is parts produced using AM will
shown in Figure 2. have a much shorter manufacturing lead time compared
In the centralized scenario, AM technology is to those produced using conventional manufacturing pro-
introduced into the aircraft spare parts supply chain as cesses. Another advantage is reduced logistics lead time
shown in Figure 3. Specifically, AM facilities are built because the RDCs are closer to the end customers. As a
within the RDCs to manufacture spare parts that can result, the amount of safety inventory that needs to be
be economically produced. Although these parts are held is reduced, which in turn reduces total supply chain
Downloaded by [University of Louisville] at 18:13 10 February 2015

Figure 3. Aircraft spare parts supply chain with centralized AM.

Figure 4. SCOR thread diagram of the centralized AM supply chain.


1174 P. Liu et al.

inventory cost. The SCOR thread diagram of this central- operators remains unchanged. The differences occur in
ized AM supply chain for the aircraft spare parts indus- the configuration of the OEM and the MRO. Therefore,
try is shown in Figure 4. our analysis should focus on these two stages of the sup-
In the distributed scenario (shown in Figure 5), AM ply chain. In the aircraft spare parts supply chain, avail-
facilities are built within the SLs instead of the RDCs. ability of parts to ensure fast service is of the utmost
Compared to the centralized scenario, this arrangement importance. Therefore, supply chain reliability, respon-
has the advantage of producing parts closer to the cus- siveness and asset management are the three performance
tomer. The trade-off is a lower level of inventory aggre- categories that should be emphasized. Under supply
gation. The SCOR thread diagram of this distributed AM chain reliability, perfect order fulfillment (RL.1.1) is the
supply chain for the aircraft spare parts industry is most suitable performance metric. However, due to data
shown in Figure 6. availability issues, we use a more generic reliability mea-
Comparing the three thread diagrams (Figures 2, 4 sure of CSL, which facilitates the calculation of required
and 6), one can see that configuration of the aircraft safety inventory under the assumption of normally
Downloaded by [University of Louisville] at 18:13 10 February 2015

Figure 5. Aircraft spare parts supply chain with distributed AM.

Figure 6. SCOR thread diagram of the distributed AM supply chain.


Production Planning & Control 1175

distributed demand. Under supply chain responsiveness, (SL1), Phoenix, ZA (SL2), Austin, TX (SL3) and Bur-
we select source cycle time (RS.2.1), make cycle time bank, CA (SL4), respectively. RDC2 also supplies to 4
(RS.2.2) and delivery cycle time (RS.2.3) as the perfor- SLs, located at Chicago, IL (SL5), Albuquerque NM
mance metrics. For simplicity sake, source cycle time (SL6), Fort Lauderdale, FL (SL7) and Atlanta GA (SL8),
and delivery cycle time are combined as logistic lead respectively.
time. Under supply chain asset management, we select Six (6) parts are considered for production using AM
inventory (AM.2.8) as the performance metric. Specifi- technology. The basic information (part name, billet
cally, we focus on the required safety inventory because; weight and part weight) of these six parts were taken
it is a critical issue in the aircraft spare parts supply from Allen (2006). Additional information, i.e. make
chain. In fact, the selection of all of these metrics is cycle time using conventional manufacturing (CM) and
driven by the goal of analysing supply chain safety AM were assumed by taking into account part complex-
inventory. ity and weight. Specifically, the EOS M270, a direct
metal laser sintering machine is assumed for AM produc-
tion. The output of this machine is 40H/kg or
4. Performance comparison study 0.055 months/kg (Allen 2011), which is used to calculate
4.1. Hypothetical supply chain the make cycle time for each part based on part weight.
Downloaded by [University of Louisville] at 18:13 10 February 2015

In order to compare the performance of different supply The information is shown in Table 1. The make cycle
chain configurations, a case study is conducted. Figure 7 times in Table 1 are mean cycle times. It is assumed that
shows the hypothetical supply chain designed for analys- each make cycle time follows a normal distribution with
ing the 3 scenarios. The supply chain has an OEM at a standard deviation that is 20% of the mean.
Carson, CA, which uses a conventional manufacturing The logistic lead time for sourcing and delivering a
facility. The OEM supplies spare parts to the MRO, part from one location to another is determined by taking
which has 2 RDCs, located at Seattle, WA (RDC1) and into account travel distance and administrative time
Torrance, CA (RDC2), respectively. Within the MRO, (which is assumed to be the same for each of the 6
RDC1 supplies to 4 SLs, located at Broomfield, CO parts). The mean logistic lead times from the OEM to

Figure 7. A hypothetical aircraft spare parts supply chain.


1176 P. Liu et al.

Table 1. Information of parts considered for AM production (modified from Allen 2006).

Part Name Billet weight (kg) Part weight (kg) Mean CM cycle time (Month) Mean AM cycle time (Month)
1 Intercase 182.00 30.00 8.000 1.650
2 Simple duct flange 1 67.00 11.14 3.000 0.613
3 Simple Duct Flange 2 67.00 7.66 2.000 0.421
4 Complex duct flange 1 149.00 7.65 1.500 0.421
5 Complex duct flange 2 206.60 10.28 7.000 0.565
6 Large blisk 810.00 97.00 9.000 5.335

Table 2. Logistic lead time (in month) information.

RDC1 RDC2 SL1 SL2 SL3 SL4 SL5 SL6 SL7 SL8
OEM 0.140 0.070 – – – – – – – –
– – – – – –
Downloaded by [University of Louisville] at 18:13 10 February 2015

RDC1 0.162 0.148 0.157 0.136


RDC2 – – – – – – 0.196 0.187 0.239 0.217

Table 3. Demand (in unit of part per month) information.

Part SL1 SL2 SL3 SL4 SL5 SL6 SL7 SL8


1 80 50 50 79 47 61 64 41
2 489 475 477 502 504 491 511 512
3 297 290 319 299 297 318 296 292
4 147 116 141 170 132 144 153 172
5 121 124 116 113 120 123 112 129
6 24 25 24 23 30 23 26 25

the two RDCs and from the two RDCs to the eight SLs μD is the mean demand (in month) and
are shown in Table 2. It is assumed that the logistic lead σD is the standard deviation of demand
time follows a normal distribution with a standard devia-
tion that is 20% of the mean. The mean demands of Under the as-is scenario, demands from SLs are
these six parts at the SLs are shown in Table 3. It is aggregated in RDCs and demands from RDCs are aggre-
assumed that the demand at each SL follows a normal gated in the OEM. The demands are assumed to be inde-
distribution with a standard deviation that is 80% of the pendent. The aggregated demands at the two RDCs and
mean. the OEM are calculated as shown in Table 4. The replen-
ishment lead time at a SL is simply the logistic lead time
from the RDC that supplies the SL; the replenishment
4.2. Safety inventory comparison lead time at a RDC is simply the logistic lead time from
Safety inventory is calculated as follows the OEM to the RDC; whereas the replenishment lead
time at the OEM is the CM make cycle time.
ss ¼ FS1 (CSL)  rL (1) Under the centralized scenario, demands from SLs
are still aggregated in RDCs; but there are no longer
where σL, the standard deviation of demand during demands of AM produced parts at the OEM. In other
replenishment lead time, is calculated as words, the OEM does not need to hold any inventory for
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi such parts. The replenishment lead time at a SL is still
rL ¼ Lr2D þ l2D r2E (2) the logistic lead time from the RDC that supplies the
SL; but the replenishment lead time at a RDC is now the
AM make cycle time. Under the distributed scenario,
where
there is no longer demand aggregation and thus no
L is the mean replenishment lead time (in month),
inventory holding requirement, either at the OEM or at
σE is the standard deviation of replenishment lead
the RDCs. Inventory is held only at the SLs. The
time,
Production Planning & Control 1177

Table 4. Aggregated demand (in unit of part per month) at RCDs and the OEM.

Demand at RDC1 Demand at RDC2 Demand at OEM


Part Mean Standard deviation Mean Standard deviation Mean Standard deviation
1 259 106.26 213 86.56 472 137.05
2 1943 777.39 2018 807.31 3961 1120.75
3 1205 482.31 1203 481.47 2408 681.50
4 574 231.65 601 241.53 1175 334.66
5 474 189.72 484 193.85 958 271.24
6 96 38.42 104 41.80 200 56.77

Table 5. Required safety inventory (in unit of part) under different scenarios.

Scenario Part SL1 SL2 SL3 SL4 SL5 SL6 SL7 SL8 RDC1 RDC2 OEM Total
As-Is 1 60 36 37 54 39 49 59 36 94 54 1975 2493
Downloaded by [University of Louisville] at 18:13 10 February 2015

2 368 342 353 346 418 397 468 447 688 501 7139 11,468
3 224 209 236 206 246 257 271 255 427 299 3170 5800
4 111 83 104 117 109 117 140 150 205 150 1258 2545
5 91 89 86 78 99 100 103 113 168 120 3539 4585
6 18 18 18 16 25 19 24 22 34 26 926 1145

Centralized 1 60 36 37 54 39 49 59 36 375 306 – 1051


2 368 342 353 346 418 397 468 447 1521 1579 – 6239
3 224 209 236 206 246 257 271 255 765 764 – 3434
4 111 83 104 117 109 117 140 150 367 383 – 1683
5 91 89 86 78 99 100 103 113 354 362 – 1475
6 18 18 18 16 25 19 24 22 315 342 – 816

Distributed 1 201 126 126 198 118 153 161 103 – – – 1185
2 726 705 708 745 748 729 759 760 – – – 5881
3 363 355 390 366 363 389 362 357 – – – 2946
4 180 142 172 208 161 176 187 210 – – – 1437
5 172 177 165 161 171 175 159 184 – – – 1364
6 119 124 119 114 149 114 129 124 – – – 993

Figure 8. Supply chain–safety inventory under different scenarios.

replenishment lead time at a SL is now the AM make Figure 8 shows the supply chain–safety inventory for
cycle time. The required safety inventories under these the six parts under different scenarios. We can see that
three scenarios are calculated as shown in Table 5. for all six parts, the as-is scenario has the highest supply
1178 P. Liu et al.

chain safety inventory. In other words, the use of AM 4.3. Sensitivity analysis
technology by the MRO allows us to reduce supply In the case study, the standard deviation of the demand
chain–safety inventory compared to the use of CM tech- is assumed to be 80% of the mean. Here, we conduct a
nology by the OEM. For parts 1 and 6, the safety inven- sensitivity analysis by varying the standard deviation
tory under the centralized AM scenario is lower than from 10 to 100% of the mean. The result is shown in
that under the distributed scenario. These two parts have Figure 9. As expected, when demand standard deviation
low demand and longer make cycle time. On the other increases, the required supply chain–safety inventory
hand, parts 2, 3, 4 and 5 have relatively high demand increases for all three scenarios. However, the rate of
and shorter make cycle time; they require less safety increase is different for different parts under different
inventory under the distributed AM scenario. scenarios. When the demand standard deviation is low

3000 14000
Part 2
As-IS Part 1 As-Is
Centralized 12000 Centralized
2500 Distributed Distributed
10000
Downloaded by [University of Louisville] at 18:13 10 February 2015

Safety Inventory
2000
Safety Inventory

8000

1500 6000

1000 4000

2000
500

0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1
Standard Deviation Standard Deviation

Part 3 Part 4
7000 As-Is 3000 As-Is
Centralized Centralized
6000 Distributed Distributed
2500
Safety Inventory

5000
Safety Inventory

2000
4000
1500
3000

1000
2000

1000 500

0 0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1
Standard Deviation Standard Deviation

1300
As-Is Part 5 As-Is Part 6
5000
Centralized 1200 Centralized
Distributed Distributed
1100
4000
Safety Inventory
Safety Inventory

1000

3000 900

800
2000
700

1000 600

500
0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1
Standard Deviation Standard Deviation

Figure 9. Sensitivity analysis by varying demand standard deviation.


Production Planning & Control 1179

2600 12000
As-Is Part 1 As-Is Part 2
2400 Centralized 11000 Centralized
2200 Distributed Distributed
10000
2000
9000
Safety Inventory

Safety Inventory
1800
8000
1600

1400 7000

1200 6000

1000 5000
800
4000
600
3000
0.89 0.90 0.91 0.92 0.93 0.94 0.95 0.96 0.97 0.98 0.99 1.00 0.89 0.90 0.91 0.92 0.93 0.94 0.95 0.96 0.97 0.98 0.99 1.00
CSL CSL

As-Is Part 3 2600 As-Is Part 4


6000 Centralized Centralized
Downloaded by [University of Louisville] at 18:13 10 February 2015

2400
Distributed Distributed
2200
5000
Safety Inventory

Safety Inventory
2000

1800
4000
1600

1400
3000
1200

1000
2000
800

0.89 0.90 0.91 0.92 0.93 0.94 0.95 0.96 0.97 0.98 0.99 1.00 0.89 0.90 0.91 0.92 0.93 0.94 0.95 0.96 0.97 0.98 0.99 1.00

CSL CSL

5000 1200
As-Is Part 5 As-Is Part 6
4500 Centralized 1100 Centralized
Distributed Distributed
4000
1000
3500
Safety Inventory

Safety Inventory

900
3000
800
2500
700
2000
600
1500

1000 500

500 400
0.89 0.90 0.91 0.92 0.93 0.94 0.95 0.96 0.97 0.98 0.99 1.00 0.89 0.90 0.91 0.92 0.93 0.94 0.95 0.96 0.97 0.98 0.99 1.00
CSL CSL

Figure 10. Sensitivity analysis by varying CSL.

(i.e. 10 or 20% of the mean), distributed AM has the higher level of safety inventory for part 6 than the as-is
lowest required safety inventory for all six parts. How- scenario. On the other hand, centralized AM is still able
ever, when the demand standard deviation reaches 30% to reduce the required safety inventory for part 6 when
of the mean, centralized AM requires less safety inven- compared to the as-is scenario even when the demand
tory for parts 1 and 6 than distributed AM. For part 6, a standard deviation is several times of the mean.
part with very low demand, its required safety inventory We then analyse the influence of CSL by varying its
under the distributed AM scenario increases quickly as value from 0.90 to 0.99 while keeping the demand
the demand standard deviation increases. In fact, when standard deviation at 80% of the mean. The result is
the demand standard deviation increases to 120% of the shown in Figure 10. Again, as expected, the required
mean (not shown in Figure 9), distributed AM requires a supply chain–safety inventory increases for all three
1180 P. Liu et al.

scenarios when CSL increases. The rate of increase the MRO have to develop a strategic partnership in order
seems to be quite stable for different parts under differ- to truly realise the potential of AM technology in the air-
ent scenarios. The best performing supply chain scenario craft spare parts industry. The best solution may be for
does not change for any of the six parts. Therefore, CSL the OEM to be responsible for AM production and
is not deemed a sensitive factor in the analysis of the inventory management at the MRO’s facility. In this
required safety inventory under different supply chain way, the OEM maintains its manufacturing business and
scenarios. reduces the level of its safety inventory; whereas the
MRO can focus on its core competency.
5. Conclusion
Notes on contributors
This paper outlined the effects of AM in the aircraft Peng Liu is an Associate Professor at the
spare parts supply chain. Three supply chain scenarios Department of Industrial Engineering, Jilin
were presented and their performance analysed in terms University, Changchun, P.R. China. He
of the required safety inventory. The factors taken into received the B.S., M.S., and PhD. degrees
consideration include demand characteristics, manufac- in Mechanical Engineering from Jilin
turing lead time, logistics lead time and CSL. It is University in 2002, 2006. and 2010,
respectively. Dr Liu’s research focuses on
Downloaded by [University of Louisville] at 18:13 10 February 2015

clearly demonstrated that AM has the potential to change


the conventional configuration of the aircraft spare parts manufacturing processes and systems, and
engineering education. He has received
supply chain to achieve safety inventory reduction; thus
several research grants from the National High Technology
cutting inventory holding cost across the entire supply Research and Development Program of China and other
chain. agencies.
AM technology can be introduced into different
stages of the aircraft spare parts supply chain. The cen- Samuel H. Huang is a Professor at the
tralized AM supply chain is more suitable for parts with School of Dynamic Systems, the University
low average demand, relatively high demand fluctuation of Cincinnati. He was previously Assistant
and longer manufacturing lead time. The distributed AM Professor of Industrial Engineering at the
supply chain is suitable for parts with high average University of Toledo and Systems Engineer
at EDS/Unigraphics (now Siemens PLM
demand. It is also suitable for parts with very stable
Software). He received the B.S. degree in
demand, because in this case the benefit of demand Instrument Engineering from Zhejiang
aggregation in a centralized distribution centre is greatly University, Hangzhou, P. R. China, in 1991
diminished. It can also be used for parts with very short and the M.S. and PhD. degrees in Industrial Engineering from
manufacturing lead time even if their demand is low and Texas Tech University, Lubbock, Texas, in 1992 and 1995,
unpredictable; because such parts can be produced respectively. Dr Huang has been awarded five grants from the
on-demand. National Science Foundation for his research in healthcare
This study only focuses on total safety inventory in analytics and manufacturing research and education.
the aircraft spare parts supply chain. Note that when AM
Abhiram Mokasdar is a Master of Science
technology is used by the MRO, the OEM does not need
Graduate in Mechanical Engineering, from
to hold any safety inventory. However, the MRO needs
the School of Dynamic Systems, University
to hold a higher level of safety inventory when com- of Cincinnati, in December 2012. He
pared to the conventional supply chain scenario. Its received the Bachelor of Technology degree
inventory holding cost will increase. This may be offset in Mechanical Engineering in 2010
by the decrease in part cost because of the benefit of from Visvesvaraya National Institute of
AM technology. However, deploying AM technology Technology, a prestigious university in
requires substantial initial investment so a full return on India. His research interest includes additive
investment analysis is necessary. manufacturing and supply chain management.
In summary, this paper explores the idea of introduc-
Heng Zhou is an Master of Science student
ing AM technology into the aircraft spare parts supply
in Mechanical Engineering in the School of
chain. The impact of AM technology on the reduction of Dynamic Systems at the University of
supply chain safety inventory is analysed quantitatively. Cincinnati. He received the B.S. degree in
The analysis clearly demonstrates the potential of AM Mechanical Engineering from Shanghai
technology to make the aircraft spare parts supply chain Maritime University, Shanghai, P.R. China,
more efficient. Due to the lack of cost data, the analysis in 2011. His reserch interests are in
is limited and at an aggregated level. It is unclear development and use of mathematical
whether the OEM or the MRO or both can benefit from optimization tools to solve complex supply
the use of the AM technology. Most likely the OEM and chain management problem.
Production Planning & Control 1181

Liang Hou is a Professor and Director of Huang, S. H., S. K. Sheoran, and H. Keskar. 2005. “Computer-
Mechanical and Electrical Engineering, assisted Supply Chain Configuration Based on Supply
Xiamen University, China. He received his Chain Operations Reference (SCOR) Model.” Computers
PhD. from Tianjin University, China, in &Industrial Engineering 48: 377–394. doi: 10.1016/j.
2002, and then worked as a postdoctoral in cie.2005.01.001.
Li, L., Q. Su, and X. Chen. 2011. “Ensuring Supply Chain
Zhejiang University. His research interests
Quality Performance Through Applying the SCOR Model.”
include design methodologies and new International Journal of Production Research 49 (1):
product R&D management for mechanical 33–57. doi: 10.1080/00207543.2010.508934.
equipment, including product family and McCue, T. J. 2012. “3D Printing Industry will Reach $3.1 Bil-
platform design and optimization. His researches have been lion Worldwide by 2016.” http://www.forbes.com/sites/
applied in the R&D activities in the enterprises of machine tool, tjmccue/2012/03/27/3d-printing-industry-will-reach-3-1-bil-
engineering machine, and electrical product. He supervised 5 lion-worldwide-by-2016/
PhD. and more than 30 MS students. He has published more Naesens, K., L. Gelders, and L. Pintelon. 2007. “A Swift
than 80 papers and has been awarded 15 patents. Response Tool for Measuring the Strategic Fit for Resource
Pooling: A Case Study.” Management Decision 45 (3):
434–449. doi: 10.1108/00251740710745061.
Reference Reeves, P. 2008. “How the Socioeconomic Benefits of Rapid
Manufacturing can Offset Technological Limitations”.
Downloaded by [University of Louisville] at 18:13 10 February 2015

Allen, J. 2006. “An Investigation into the Comparative Costs Paper presented at RAPID 2008 Conference and Exposi-
of Additive Manufacture vs. Machine from Solid for Aero tion, Lake Buena Vista, FL, May 20–22.
Engine Parts.” http://www.dtic.mil/cgi-bin/GetTRDoc? SCC. 2012. “What is SCOR?” http://supply-chain.org/scor
AD=ADA521730 SCW. 2012. “What is a Supply Chain?” http://www.wisegeek.
Allen, J. 2011. “The Potential for Aero Engine Component Man- com/what-is-a-supply-chain.htm
ufacture using Additive Layer Manufacturing.” http://www. Towill, D. R. 1991. “Supply Chain Dynamics.” International
cdti.es/recursos/doc/eventosCDTI/Aerodays2011/1D2.pdf Journal of Computer Integrated Manufacturing 4 (4):
ASTM. 2010. “Standard Terminology for Additive Manufactur- 197–208. doi: 10.1080/09511929108944496.
ing Technologies.” http://enterprise.astm.org/filtrexx40.cgi? Tuck, C., R. Hague, and N. Burns. 2007. “Rapid
+REDLINE_PAGES/F2792.htm Manufacturing: Impact on Supply Chain Methodologies
ATKINS. 2007. “ATKINS: Manufacturing a Low Carbon Foot- and Practice.” International Journal of Services and
print.” http://www.atkins-project.com/pdf/ATKINSfeasibili- Operations Management 3 (1): 1–22. doi: 10.1504/
tystudy.pdf IJSOM.2007.011459.
Baumers, M., C. Tuck, D. L. Bourell, R. Sreenivasan, and Tzafestas, S., and G. Kapsiotis. 1994. “Coordinated Control of
R. Hague. 2011. “Sustainability of Additive Manufacturing: Manufacturing/Supply Chains Using Multi-Level Tech-
Measuring the Energy Consumption of the Laser Sintering niques.” Computer Integrated Manufacturing System 7 (3):
Process.” Proceedings of the Institution of Mechanical Engi- 206–212. doi: 10.1016/0951-5240(94)90039-6.
neers, Part B: Journal of Engineering Manufacture 225: Verdouw, C. N., A. J. M. Beulens, J. H. Trienekens, and J. G.
2228–2239. doi: 10.1177/0954405411406044. A. J. van der Vorst. 2011. “A Framework for Modelling
Christy, D. P., and J. R. Grout. 1994. “Safeguarding Supply Business Processes in Demand-driven Supply Chains.”
Chain Relationships.” International Journal of Production Production Planning & Control: The Management of
Economics 36: 233–242. doi: 10.1016/0925527394000247. Operations 22 (4): 365–388. doi: 10(1080/09537287).
Feng, C. M., P. J. Wua, and K. C. Chia. 2010. “A Hybrid 2010.486384.
Fuzzy Integral Decision-making Model for Locating Voudouris, V. T. 1996. “Mathematical Programming Techniques
Manufacturing Centers in China: A Case Study.” European to Debottleneck the Supply Chain of Fine Chemical Indus-
Journal of Operational Research 200 (1): 63–73. doi: tries.” Computer and Chemical Engineering 20 (1):
10.1016/j.ejor.2008.12.004. S1269–S1274. doi: 10.1016/0098-1354(96)00219-0.
Ferguson, B. R. 2000. “Implementing Supply Chain Manage- Wall, B., H. Jagdev, and J. Browne. 2007. “A Review of
ment.” Production and Inventory Management Journal 41 eBusiness and Digital Business – Applications, Models and
(2): 64–67. Trends.” Production Planning & Control: The Management
Hasan, S., and A. E. W. Rennie. 2008. “The Application of of Operations 18 (3): 239–260. doi: 10.1080/
Rapid Manufacturing Technologies in the Spare 09537280601127245.
Parts Industry.” http://edge.rit.edu/edge/P10551/public/SFF/ Walter M., J. Holmstrom, and H. Yrjola. 2004. “Rapid Manu-
SFF% 202008%20Proceedings/Manuscripts/2008-50-Hasan. facturing and its Impact on Supply Chain Management”.
pdf Paper Presented at Logistics Research Network Annual
Holmstrom, J., J. Partanen, J. Tuomi, and M. Walter. 2010. Conference, Dublin, Ireland, September 9–10.
“Rapid Manufacturing in the Spare Parts Supply Chain: Zhou, H. G., W. C. Benton, D. A. Schilling, and G. W. Milli-
Alternative Approaches to Capacity Deployment.” Journal gan. 2011. “Supply Chain Integration and the SCOR
of Manufacturing Technology Management 21 (6): Model.” Journal of Business Logistics 32 (4): 332–344.
687–697. doi: 10.1108/17410381011063996. doi: 10.1111/j.0000-0000.2011.01029.x.

View publication stats

Vous aimerez peut-être aussi