Professional Documents
Culture Documents
Additive Manufacturing: Danielle Strong, Michael Kay, Brett Conner, Thomas Wake Field, Guha Manogharan
Additive Manufacturing: Danielle Strong, Michael Kay, Brett Conner, Thomas Wake Field, Guha Manogharan
Additive Manufacturing
journal homepage: www.elsevier.com/locate/addma
A R T I C LE I N FO A B S T R A C T
Keywords: The ever-growing applications of Additive Manufacturing (AM) in the production of low volume- high value
Hybrid manufacturing metal parts can be attributed to improving AM processing capabilities and complex design freedom. However,
Additive manufacturing secondary post-processing using traditional processes such as machining, grinding, heat treatment and hot
Facility location isostatic pressing, i.e., Hybrid Manufacturing, is required to achieve Geometric Dimensioning and Tolerancing
Supply chain
(GD&T), surface finish and desired mechanical properties. It is often challenging for most traditional manu-
Logistics
Manufacturing hubs
facturers to participate in the rapidly evolving supply chain of direct digital manufacturing (DDM) through in-
Direct digital manufacturing house investments in cost prohibitive metal AM. This research investigates a system of strategically-located AM
hubs which can integrate hybrid-AM with the capabilities and excess capacity in multiple traditional manu-
facturing facilities. Using North American Industry Classification System (NAICS) data for machine shops in the
U.S., an uncapacitated facility location model is used to determine the optimal locations for AM hub centers
based on: (1) geographical data, (2) demand and (3) cost of hybrid-AM processing. Results from this study have
identified: (a) candidate US counties to build AM hubs, (b) total cost (fixed, operational and transportation) and
(c) capacity utilization of the AM hubs. It was found that uncapacitated facility location models identified
demand centroid as the optimal location and was affected only by AM utilization rate whereas a constrained p-
median model identified 22 AM hub locations as the initial sites for AM hubs which grows to 44 AM hubs as
demand increases. It was also found that transportation cost was not a significant factor in the hybrid-AM supply
chain. Findings from this study will help both AM companies and traditional manufacturers to determine lo-
cation in the U.S and key factors to advance the metal hybrid-AM supply chain.
1. Introduction uneven and rough surface profile) and inaccurate dimensional toler-
ances in AM metal parts, along with very high investment costs for AM
Additive manufacturing (AM) has emerged as a solution to machines, operating costs and materials are identified as the primary
streamline supply chains through both centralized and decentralized challenges in furthering the adoption of AM [5].
applications [1]. As defined by ASTM [2], AM is the process of adding Traditional manufacturers such as machine shops have reported
material layer by layer based on 3D model data. AM applies to a wide that despite the realization of AM benefits to their industry, access to
range of materials from polymers and ceramics to metals. There is a metal AM capabilities remain the primary barrier and challenge for
growing interest in processing metallic super-alloys for low-volume their adoption of AM [7]. Currently, costs of gaining direct access to AM
aerospace, medical, industrial, and transportation applications using capabilities in-house is cost prohibitive for small and medium en-
AM due to complex design freedom, higher degree of customization and terprises (SME). Due to this challenge, Original Equipment Manu-
better material utilization when compared to traditional manu- facturers (OEMs) surveyed also expressed their interest in supply chain
facturing. Although there are many benefits to AM such as the wide strategies in which AM processes can be accessed externally [6]. The
range of materials [3], significant part weight reduction, freeform need for finishing and non-traditional post processing of metal AM parts
fabrication and lack of fixtures and tooling [4], several challenges in is widely reported [7]. This is the premise behind hybrid-AM, in which
metal AM for direct adoption exists. Inadequate surface finish (i.e., a metal part is first near-net produced via AM and, subsequently, post-
⁎
Corresponding author.
E-mail address: gum53@psu.edu (G. Manogharan).
https://doi.org/10.1016/j.addma.2018.03.010
Received 21 November 2017; Received in revised form 2 February 2018; Accepted 8 March 2018
Available online 09 March 2018
2214-8604/ © 2018 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/BY-NC-ND/4.0/).
D. Strong et al. Additive Manufacturing 21 (2018) 159–173
processed via traditional manufacturing processes such as machining, facilities will result in unnecessarily poor customer service and expensive
grinding, etc., to achieve the desired surface finish, dimensional toler- transportation costs. Optimally located sites can help a company gain
ances and material properties. In this paper, hybrid-AM refers to two or competitive advantages and improve operational performance not only in
more sequential discrete processes (e.g. AM + Machining) employed to the short term but also in the long term [13]. The ability of a firm to
achieve the final part specifications. Several studies have highlighted market and produce its products effectively and/or to deliver high-quality
the economic benefits [8] and detailed the current state of hybrid services is dependent in part on the location of the facilities in relation to
manufacturing [9,10]. Almost 94% of the OEMs surveyed in a recent other facilities and to its customers [14]. Daskin [15] also presented a list
study have reported an interest in offering post-processing of AM parts of comprehensive taxonomy for multiple location model problems in
as a service through hybrid-AM6. More than 53% of OEMs have a contrast to vehicle routing problems which are often limited to fewer lo-
minimum of 20% excess machine capacity to offer post-processing cations (i.e. nodes).
services for AM parts. In the context of this paper, hybrid-AM is defined Different models employed for facility location include the p-hub
as the integration of dissimilar metal manufacturing processes, i.e., AM median problem, p-hub center problem, uncapacitated hub problem, ca-
linked to traditional manufacturing processes, which are planned to- pacitated hub problem, and hub covering. Methodologies employed include
gether so that the required engineering specifications can be met. mixed integer [16], branch and bound [17], genetic algorithms [18],
This study explores a hybrid-AM based DDM supply chain in which heuristics [19], and Lagrangian relaxation [20,11]. Snyder and Daskin [21]
AM hubs would act as suppliers to traditional manufacturers with studied classical facility location models like the p-median problem and the
capabilities and demand for hybrid AM production. In this system, uncapacitated fixed-charge location (UFL) problem that implicitly assume
OEMs would send demand that qualifies for hybrid AM to a regional that once constructed, the facilities chosen will always be fixed in location
AM hub (e.g. service bureau) which would control all aspects of AM and capacity. The UFL is a classical facility location problem that chooses
fabrication and operations. Subsequently, the AM hub would ship the facility locations and assignments of customers to facilities to minimize the
printed metal part to the OEMs for post-processing, i.e., hybrid manu- sum of fixed and transportation costs. In Melkote and Daskin [22], plant and
facturing using traditional processes to achieve the desired surface warehouse location problems are referred to as UFL with consideration to
finish and dimensional tolerances (e.g. machining, grinding, polishing). routing and geography. The distinguishing feature of the UFL is the decision
The overall goal of this work is to determine the optimal locations maker’s ability to determine the size of each facility without any budgetary,
for metal AM hubs in the US based on estimated demand and produc- technological, or physical restrictions and is closely related to the p-median
tion metrics such as total cost of establishing and operating AM hubs problem. In many cases, it is more realistic to incorporate the capacity
and their allocation as suppliers to existing machine shops based on limitations on the facilities to be established called the capacitated facility
2013 county-level NAICS data. This work employs both: (1) un- location (CFL) problem [23]. The p-median problem is framed to locate p
capacitated facility location and (2) p-medial models based on facility facilities to minimize the demand-weighted average distance between de-
and transportation costs, location and density of current traditional mand nodes and the nearest of the selected facilities [24]. Using sensitivity
manufactures, and hybrid-AM demand estimated by current annual analysis on p-median based number of hubs, it was shown that branch-and-
sales. This paper intends to present a decision-making model for both: bound approach is optimal for large amount demand and supplier locations
(1) AM service companies who can use these models to locate their next (nodes in the order of 1000s). In a supply chain that comprises suppliers,
facility to operate as an AM hub and (2) OEMs who can evaluate par- plants, distribution centers, warehouses and customers, these basic for-
ticipating in the DDM supply chain through hybrid post processing mulations are relevant for making location decisions [23]. Hence, in this
services to fully harness the benefits of AM. Finally, a detailed sensi- study both UFL and p-median models are employed to determine AM hub
tivity analysis is conducted for scenarios with growing AM demand, locations for hybrid-AM supply chain.
varying number of AM machines at each hub and AM machine utili- Prior studies on hub locations have been applied to industries such
zation (i.e. metal AM parts that might not require post-processing). as airlines [25], postal service [26], warehouse and supply chain lo-
In Section 2, background on facility location, hybrid AM and supply gistics [27], emergency services, delivery services, logistics services,
chain implications are presented. The methodology applied for the fa- and transportation [28]. In the case of tiered supply chain, hierarchical
cility location models is detailed in Section 3, followed by results on AM hub network has been shown as an effective method to locate hub fa-
hub location and supply chain costs in Section 4. Discussions and key cilities [29,30]. Figueiredo et al. [31] also used p-median models in two
insights obtained from this study are presented in Section 5. Finally, stages to locate regional hubs for commercial aviation in regional
Section 6 summarizes the conclusions, limitations and future direction market sectors. However, there is a knowledge gap in the literature on
for the hybrid-AM based DDM supply chain. AM hub locations that can be integrated with existing facilities in tra-
ditional manufacturers for hybrid-AM. Also, it is important to note that
2. Literature review lack of real data for facility location models often limits its applications
to real world problems [28]. The uncertainty of a facility location
2.1. Facility location model is typically in the input conditions including the costs and de-
mands which require sensitivity analysis to capture the effects of the
The AM hub problem in this study will be treated as a single allocation inherent variability [12].
problem, i.e. all the machine shops in a county are assigned to the same
AM hub. Optimal demand allocations are affected by hub locations and 2.2. Additive – hybrid manufacturing
optimal hub locations are affected by allocation decisions; hence, location
and allocation problems must be simultaneously modeled to develop cost According to ISO/ASTM 52900:2015, additive manufacturing (AM)
effective AM hub networks. Alumur and Kara [11] presented a review of describes manufacturing processes where material is deposited or fused
prior studies on hub locations and allocation of demand nodes to hubs for together layer-by-layer until a net-shape or near net-space is achieved
traffic routing between origin–destination pairs using both single alloca- [2]. This is contrasted with subtractive manufacturing (i.e. milling,
tion and multiple allocations. Daskin and Dean [12] highlighted the im- grinding, cutting, drilling, etc.) where material is removed to obtain the
portance of facility location models. In industry, poorly located facilities or final part shape. Additive manufacturing is often called 3D printing in
the use of too many or too few facilities will result in increased expenses the popular media. A variety of materials can be processed additively
and/or degraded customer service. If too many facilities are deployed, including polymers, metals, ceramics, electronic materials, and biolo-
capital costs and inventory carrying costs are likely to exceed the desirable gical materials [32]. Of interest to this research is metal additive
value. If too few facilities are used, customer service can be severely de- manufacturing. Metals produced using AM include but are not limited
graded. Even if the optimal number of facilities is established, poorly sited to titanium, stainless steel, tool steels, nickel based alloys, aluminum,
160
D. Strong et al. Additive Manufacturing 21 (2018) 159–173
and cobalt chrome [33,34]. ISO/ASTM 52900:2015 also standardizes [53] who noted that the shift of manufacturing facilities closer to the
terminologies for AM equipment technologies into seven categories. customer as a result of AM adoption would lead to fewer opportunities
Four of these are currently used for metals production: powder bed for logistics suppliers to be involved in companies’ upstream supply
fusion, binder jetting, directed energy deposition, and sheet lamination. chains. They further posited that a major new sector of the logistics
This research will focus on the most widely used technology for metals industry would emerge to facilitate the storage and movement of raw
AM production: powder bed fusion [5]. In this AM technology, feed- materials which ‘feed’ the AM equipment and that the home delivery
stock in the form of metal powder is spread on a build plate layer-by- market of these materials would increase.
layer. Based on the 3D part information from a CAD (Computer Aided Petrick and Simpson [7] addressed how the trend of localization is
Drafting) file, an energy source in the form of either a laser or an empowered by direct interactions between consumers and producers.
electron beam selectively melts the powder in each layer. This process The projected transformation is also discussed in Frazier [33] who
is repeated until the final part is produced. The most common form of developed a business case for AM and noted that the reduction in lo-
powder bed fusion is a laser based system invented by Deckard and gistical footprint, cost, and energy associated with packaging, trans-
Beaman [35]. Currently, all metal AM methods require some form of portation and storage of spare parts could be significant for large or-
post-processing to obtain the final part geometric dimensions, surface ganizations. Walter et al. [1] contended that AM production through
finish, and material properties. Additive and subtractive machining can both centralized and decentralized applications will become the basis
occur in the same machine envelope with the directed energy deposi- for new solutions in supply chain management and developed a deci-
tion [36]. However, the vast majority of AM metal processing occurs in sion-support model to capture emergent business opportunities arising
a dedicated additive manufacturing build envelope [5]. Recent studies from AM technology. Reeves [54] argued that the traditional produc-
have presented details on near-net AM and post-processing through tion-distribution-retail model would shift toward DDM based model
traditional machining [9], non-traditional cryogenic machining [37], where electronic retail will initiate AM manufacturing and distribution
laser polishing [38] and thermal processes such as Hot Isostatic activities for the end customer [55].
Pressing – HIP [39]. Despite AM’s promising future and endless potential to simplify supply
The 3D digital model is often created using CAD although reverse chain, rapid adoption of AM into the existing supply chain remains a big
engineering methods like 3D laser scanning or MRI/CT techniques can challenge [56]. In addition, the literature review reveals a critical gap
be used to digitally produce an existing part geometry. The solid model regarding hybrid AM supply chain, particularly in the lack of studies that
will be converted into a standard format suitable for AM processing. use real data to predict demand and location of AM hubs. To the best of
Standard formats include .STL, .AMF and .3MF. The processing soft- our knowledge, this paper is one of the early studies that would in-
ware will slice the CAD model based on the thickness of the powder corporate data on the current state of manufacturing cost, transportation,
layers to create tool path for the laser/e-beam to selectively fuse ma- and logistics in the U.S. to develop a quantitative model that identifies
terial in the AM machine envelope. The digital thread associated with potential locations of AM hubs for a DDM supply chain via hybrid-AM.
AM and corresponding hybrid AM steps is critical in connecting AM
processing (CAD) through traditional CNC (Computer Numeric Control) 3. Methodology
machining and quality control using CMM (Coordinate Measuring
Machine). In today’s world of digital manufacturing [40], it has been This investigation initially aims to develop an uncapacitated facility
acknowledged that AM is well positioned to drastically impact con- location model based on demand, location, fixed cost, production cost and
ventional operations [41] and time-sensitive production scenarios such transportation cost to identify candidate counties in the U.S. that could
as defense [42] and part replenishment in the automotive service sec- serve as AM hubs for existing machine shops. A schematic representation
tors [43]. Hence, there is a critical need to expand the existing DDM of the hybrid-AM based DDM supply chain is shown in Fig. 1. These AM
supply chain to incorporate hybrid AM processing which would include hubs would offer additive manufacturing services for OEMs, which would
traditional manufacturers. provide the post-processing services for the printed parts.
The recent emergence of AM emerged as one of the most important The primary data set used for this analysis was obtained from the
disruptive technologies can be attributed to the continuous stream of AM North American Industry Classification System (NAICS) Association for
supply chain research. This includes potential benefits of AM to supply NAICS code 332710 (Machine Shops) in 3109 U.S. counties. It included
chain and logistics management as well as how manufacturer’s long-run the following information for each county: (1) region of U.S., (2) state,
strategy (e.g. location decisions) may be impacted in the future. For in- (3) number of machine shops, (4) annual sales volume ($M/yr) and (5)
stance, Thomas [44] used operational cost data for a representative me- number of employees per county. To assist in estimating demand, five
chanical product to identify the avenues when AM can potentially improve sample CAD drawings of AM metal parts for mechanical/aerospace
overall efficiency and reduce total cost. Scott and Harrison [45] developed applications of various complexities and metal materials were sent to
a stochastic cost model that determines when AM is an optimal choice multiple AM service bureaus in the U.S. to obtain current market pri-
(over traditional manufacturing) to the supply chain as a whole and cing. The parts and their specifications are presented in Table 1. The
identified raw material cost and demand as the key decision criteria. In the part design included a part overgrowth of 0.255 mm (0.01in) for ma-
domain of spare parts supply chain, the characterization of cost factors chining during hybrid processing which resulted in geometric mean
that favor or hinder the adoption of AM are presented in Knofius et al. unit price of $1303 per metal AM part for varying batch size (1–10
[46], Li et al. [47], Sirichakwal and Conner [48], Savastano et al. [43] and parts/batch), weight of 2–5 lbs. per part and volume of 15–30 in [3].
Khajavi et al. [49], among others. Table 2 presents the input and output parameters employed in this
Nyman and Sarlin [50] summarized the four key principles of AM in study. It is evident that metal hybrid-metal AM is suitable only for low
the context of supply chain strategies with emphasis on improving the volume, highly complex and customized production runs and is a subset of
sustainability aspect of an operation through efficient material utiliza- the current annual demand for metal production [57]. While 82% of metal
tion in AM. Other studies have outlined scenarios where relatively manufacturers in the U.S estimate that 50% of their products are low vo-
lower fixed costs and small batch sizes production capability will move lume production6, this study conservatively assumes two scenarios of 5%
AM production towards the point of wider adoption [51,52]. This and 10% of current metal parts as candidates for hybrid-AM. Annual de-
transformation to localization of production and sourcing activities is a mand for hybrid-AM parts for each county is estimated based on their re-
notion shared among many researchers, including Manners-Bell et al. spective annual sales volume and average AM unit price.
161
D. Strong et al. Additive Manufacturing 21 (2018) 159–173
Annual operating costs for the hubs were also estimated using data hybrid post-processing. The transportation costs for shipping metal AM
from literature. The fixed cost for a new hub facility with one metal AM parts to traditional manufacturers are based on FedEx ground shipping
machine at 57% utilization or 5000 h/yr [58] is presented in Table 3. rates, detailed in Section 3.3.
Fixed costs for facilities with 1, 2, 5 and 10 metal AM machines at 90% As illustrated in Table 2, two major categories of results are de-
utilization or 7884 h a year for both 57% and 90% utilization are pre- ducted from this study: Cost (production, shipping) and logistics for AM
sented in Tables 3 and 4 [58]. The number of metal AM machines and hubs. Based on longitude and latitude of AM hub locations from the
utilization rates at each hub is indicative of AM production that require facility location models, the city within the county with the highest
Table 1
Hybrid metal AM parts surveyed among existing AM service bureaus in the U.S.
Part Name Figure Materials Quoted Bound Volume (in) Weight (lbs) Batch Size Avg.
Quote
Alcoa Bearing Ti64, Inconel 718 Inconel 625, Stainless Steel (316), 4.99 × 1.3 × 3.25 3.42 1, 5, 20 $1103
Bracket Direct Metal Titanium (6Al–4 V ELI), and Aluminum
(AlSi10Mg)
162
D. Strong et al. Additive Manufacturing 21 (2018) 159–173
Table 2
Model parameters for AM hub location problem.
Outputs Cost Component Total Fixed Cost $M/year Literature (Baumers [58]), see Table 3
Production Cost Metal AM production per year at each hub
Transportation cost Cost of shipping AM parts from hub to assigned regional manufacturers
No. of Hubs – Required AM hubs in the U.S based on demand and machine capacity-utilization
AM Hubs Location Cities-County-State Largest city within the county and nearest metropolitan for AM hub locations in the U.S (U.S.
Census Bureau [59])
Allocations – Allocations of AM hubs to serve regional manufacturers
Table 3
Annual fixed cost per AM hub: 1 AM machine at 57% utilization (5000 h/yr).
Machine Depreciation $97,702 Baumers [58]; Lindemann et al. [67]; Production Overhead
Rent $34,170 Thomas and Gilbert [68]
Utilities $12,562 Baumers [58]
Technician Salary $26,732 Labor Overhead
Indirect Cost/Machine hr $223,104 Administrative Overhead
Indirect Consumables $1540 Thomas and Gilbert [68]
Indirect Software Cost $462 Thomas and Gilbert [68]; Baumers [58]
Indirect Hardware Cost $462 Thomas and Gilbert [68]; Baumers [58]
Machine Software Cost $3081 Machine Costs
Machine Hardware Cost $924
Machine Maintenance $23,104 Thomas and Gilbert [68]
Direct Machine Consumables $2700 Baumers [58]
Total Fixed Cost $340,335
Table 5
Assumptions for hybrid metal AM supply chain in the U.S.
Justification
Locations Existing traditional manufacturers will remain fixed in their current locations and capacities
Demand A small percentage of recent sales volume is an indicator of the demand for AM metal parts for each county (e.g. 5 and 10%)
Contagious U.S. Since FedEx ground shipping rate is applied, District of Columbia (D.C) is included in the study and states of Hawaii and Alaska are not considered
along with the U.S. territories
Supply Chain Integration Hybrid-AM operates in sequence with in-built costs for potential part failure/scrap: Traditional manufacturers receive orders for low production run
from customers → CAD models sent to AM hub → AM hub produces ‘near net’ metal parts → AM hub ships metal parts to traditional facilities who
perform hybrid post-processing the part → fulfill orders to customers using existing delivery methods. See Fig. 1.
163
D. Strong et al. Additive Manufacturing 21 (2018) 159–173
Table 6
Using UFL results to identify p-Median parameters.
of the counties in the U.S. is imported into the model from Matlog. The
imported data includes latitude and longitude, population and land
area of each county. Population and land area are used to adjust the
latitudes and longitudes based on population centroids and land bar-
riers. Based on model parameters noted in Section 3.1, we assume that
the standard FedEx large box of dimensions 8.75” × 7.75” × 11.3125”
will be used for shipping. The transportation costs are then determined
Fig. 2. Linear Regression analysis for Fixed Cost. using distances based on population centroids and land area, 2016
FedEx ground rates, order weight, order size and order cost. Un-
in figure 7.6 of Daskin [14], which uses an additive heuristic to con- discounted 2016 FedEx Ground rates [61,62] were used to estimate
struct an initial solution and then an exchange procedure to improve transportation costs. The ground rates cover shipments between 1 and
the solution, followed by the addition or dropping until there is no 150 lbs. within the continental United States. The rate is determined by
change in the number of facilities. the chargeable weight and distance of the shipment, with the distance
While it is usually straightforward to estimate the transportation falling into one of seven different zones. The chargeable weight is the
costs from each new facility to each existing facility (cij), it is difficult to maximum of a shipment’s actual and dimensional weight. Dimensional
estimate the fixed cost of each new facility in UFL since it is in- weight is the product of a package’s cubic dimensions in inches divided
dependent of the quantity of product produced at the facility. One by 166 and is meant to account for the fact that the actual weight of a
means of estimating the fixed cost is to fit a straight line to the pro- low-density package would underestimate its utilization of the cubic
duction and procurement costs (TPC) of a representative set of existing capacity of a transportation vehicle. The rate includes additional
facilities (Nʹ), and then the y-intercept of the line (k) can be used as the charges for excessive linear dimensions, weight beyond 70 pounds, and
fixed cost for the UFL problem (the variable production cost (cp) is not declared value in excess of $300 (as in this study). One notable feature
used in UFL). Thus, of the rate with respect to its use in location procedures is that distance-
based zones result in transport charges that remain flat for extensive
TPC = ∑ieN ' TPCi = ∑ieN ' (k + cp fi ) changes in shipment distance; for example, the first zone (Zone 2)
TLC = TDC + nk covers all distances up to 150 miles and the other zones have distance
ranges from 300 to 400 miles. An uncapacitated facility location
= ∑ ∑ cij + nk
heuristic [14] is then used to choose the hub locations based on the
ieN jeMi (1)
fixed facility costs and transportation costs. The results include the
where TPC = total production and procurement cost variable y which represents the hub locations, the variable x which
N’ = set of representative existing plants represents the hub allocations, and the variable TC which represents
k = fixed portion of TPCi total costs of the hubs.
cp = variable portion of TPCi per unit of f Similar to earlier studies [15,29–31] multiple location stages are
f = annual plant production applied to this problem. Initially, the model will employ uncapacitated
TLC = total logistics cost of UFL AM hubs to determine required capacity and corresponding annual
TDC = total outbound distribution costs from each NF fixed costs for AM hubs. Subsequently, a p-median heuristic [14] will be
n = number of NFs applied to account for capacity needs of the AM hubs based on re-
cij = distribution cost from NFi to EFj spective regional demand and transportation cost (TPC+TLC). Using
N = set of NFs metal AM powder bed fusion machine specifications and average metal
Mi = EFs allocated to NFi part volume (Table 1), it is found that one AM machine can print up to
Since the AM hub being located is a new facility, county data for all about 400 parts per year at 57% utilization and about 630 parts per
Fig. 3. UFL results for Optimal AM hub locations at: (a) Washington County, IL (b) Preble County, OH and Kern County, CA and (c) Schuylkill County, PA, Perry County, IL and Kern
County, CA.
164
D. Strong et al. Additive Manufacturing 21 (2018) 159–173
year at 90% utilization. Since AM hubs under UFL scenario will be Table 7
dedicated systems, only 90% utilization will be considered. These ca- Locations for 22 AM hubs for 5% Demand at 90% utilization.
pacities are used to determine the number of hubs needed from the UFL
results based on an average upper bound of $40 M to establish a new
AM hub. The rationale for this assumption is based on the total in-
vestment cost reported for a dedicated AM hub by General Electric –
Center for Additive Technology Advancement (CATA) which has been
actively pursuing metal AM in 2016 [63].
Fig. 4. p-Median results for AM hubs based on: (a) 5% demand and (b) 10% demand.
165
D. Strong et al. Additive Manufacturing 21 (2018) 159–173
90% utilization rates and AM demands of 5% and 10% which would Table 8 (continued)
require 2245 and 4490 AM machines at enormous fixed costs of $0.88B
and $1.76B respectively. The city within the county with highest po-
pulation is Nashville, Illinois and is near St. Louis, Missouri. It was
interesting to observe that this county was about 200 miles northwest of
Plato, Missouri which is the current population centroid of the U.S (U.S.
Census Bureau, 2010).
In the case of reduced AM machine costs (see Table A2), the UFL
model showed that Washington County, IL was the optimal location for
AM hubs at both 5 and 10% demand. This indicates that reduction in
AM machine cost and increased demand does not affect optimal hub
locations when the fixed cost is unconstrained. In addition, when the
model is forced to pick more than one hub shown in Fig. 3(b), the
Table 8
Locations for 44 AM hubs for 5% Demand at 90% utilization.
optimal locations were identified as: (1) Kern County, CA and Preble
County, IL (See Table A3). Similar to single UFL facility (Tables A1 and
A2), the fixed cost required was $0.44B and $1.32B respectively. It was
also found that in the unlikely event of 57% utilization rate of AM hubs
(see Table A4), 3 AM hubs across CA, IL and PA were found as optimal
locations with a minimum fixed cost of $0.36B, which is still 9 times the
upper limit for fixed costs (Fig. 3(c)). It is evident that fixed cost is the
main driver behind the location of hubs.
Table 9
Sensitivity analysis for adding AM hubs.
166
D. Strong et al. Additive Manufacturing 21 (2018) 159–173
Fig. 5. AM hub locations at (a) 5% demand – 22 hubs and (b) 10% demand – 44 hubs.
167
D. Strong et al. Additive Manufacturing 21 (2018) 159–173
168
D. Strong et al. Additive Manufacturing 21 (2018) 159–173
Appendix A
See Table A5 .
Table A1
AM hub in Washington County, IL based on UFL model and existing AM machine costs.
169
D. Strong et al. Additive Manufacturing 21 (2018) 159–173
Table A2
AM hubs based on UFL model and lowered AM machine costs at 90% utilization.
Table A3
AM hub in Kern County CA and Preble County OH based on UFL model and existing AM machine costs.
10% Demand
90% Utilization
Kern, CA Preble, OH
Table A4
AM hubs in Kern County CA, Perry County IL and Schuylkill County PA based on UFL model and existing AM machine costs.
10% Demand
57% Utilization
Table A5
UFL results hubs and related cities for 5% and 10% demand.
170
D. Strong et al. Additive Manufacturing 21 (2018) 159–173
Table A6
P-median results for 5% demand, 22 hubs.
171
D. Strong et al. Additive Manufacturing 21 (2018) 159–173
Table A7
P-median results for 10% demand, 44 Hubs.
172
D. Strong et al. Additive Manufacturing 21 (2018) 159–173
References 23rd Annual International Solid Freeform Fabrication Symposium, 2012 Austin, TX,
USA: University of Texas, 2012, pp. 821–827.
[37] A. Bordin, S. Sartori, S. Bruschi, A. Ghiotti, Experimental investigation on the fea-
[1] M. Walter, J. Holmström, H. Tuomi, H. Yrjölä, Rapid manufacturing and its impact sibility of dry and cryogenic machining as sustainable strategies when turning
on supply chain management, Proceedings of the Logistics Research Network Ti6Al4V produced by Additive Manufacturing, J. Clean. Prod. 142 (2017)
Annual Conference (2004) 9–10 (September). 4142–4151.
[2] A. ASTM, F2792-12 Standard Terminology for Additive Manufacturing [38] B. Rosa, B. Rosa, P. Mognol, P. Mognol, J.Y. Hascoët, J.Y. Hascoët, Modelling and
Technologies, ASTM International, 2012. optimization of laser polishing of additive laser manufacturing surfaces, Rapid
[3] N. Guo, M.C. Leu, Additive manufacturing: technology, applications and research Prototyp. J. 22 (6) (2016) 956–964.
needs, Front. Mech. Eng. 8 (3) (2013) 215–243. [39] M. Qian, W. Xu, M. Brandt, H.P. Tang, Additive manufacturing and postprocessing
[4] J. Seppälä, A. Hupfer, Topology optimization in structural design of a LP turbine of Ti-6Al-4V for superior mechanical properties, MRS Bull. 41 (2016) 775–783.
guide vane: potential of additive manufacturing for weight reduction, ASME Turbo [40] D. Wu, D.W. Rosen, L. Wang, D. Schaefer, Cloud-based design and manufacturing: a
Expo 2014: Turbine Technical Conference and Exposition, American Society of new paradigm in digital manufacturing and design innovation, Comput.—Aided
Mechanical Engineers, 2014 (pp. V07AT28A004-V07AT28A004) June. Des. 59 (2015) 1–14.
[5] T. Wohlers, Wohler’s Report 2016, Wohlers Associates, Inc., 2016. [41] T. Birtchnell, J. Urry, A New Industrial Future? 3D Printing and the Reconfiguring
[6] D. Strong, I. Sirichakwal, G. Manogharan, T. Wakefield, Current state and potential of Production, Distribution, and Consumption, Routledge, 2016.
of additive-hybrid manufacturing for metal parts, Rapid Prototyp. J. 23 (3) (2017) [42] C.E. Scheck, J.N. Wolk, W.E. Frazier, B.T. Mahoney, K. Morris, R. Kestler, A. Bagchi,
577–588. Naval additive manufacturing: improving rapid response to the warfighter, Nav.
[7] I.J. Petrick, T.W. Simpson, 3D printing disrupts manufacturing: how economies of Eng. J. 128 (1) (2016) 71–75.
one create new rules of competition, Res.—Technol. Manag. 56 (6) (2013) 12–16. [43] M. Savastano, C. Amendola, D. Fabrizio, E. Massaroni, 3-D printing in the spare
[8] G. Manogharan, R.A. Wysk, O.L. Harrysson, Additive manufacturing–integrated parts supply chain: an explorative study in the automotive industry, Digitally
hybrid manufacturing and subtractive processes: economic model and analysis, Int. Supported Innovation, Springer International Publishing, 2016, pp. 153–170.
J. Comput. Integr. Manuf. 29 (5) (2016) 473–488. [44] D. Thomas, Costs, benefits, and adoption of additive manufacturing: a supply chain
[9] Z. Zhu, V.G. Dhokia, A. Nassehi, S.T. Newman, A review of hybrid manufacturing perspective, Int. J. Adv. Manuf. Technol. (2015) 1–20.
processes–state of the art and future perspectives, Int. J. Comput. Integr. Manuf. 26 [45] A. Scott, T.P. Harrison, Additive manufacturing in an end-to-end supply chain
(7) (2013) 596–615. setting, 3D Print. Addit. Manuf. 2 (2) (2015) 65–77.
[10] G. Manogharan, R. Wysk, O. Harrysson, R. Aman, AIMS–a metal additive-hybrid [46] N. Knofius, N. Knofius, M.C. van der Heijden, M.C. van der Heijden, W.H.M. Zijm,
manufacturing system: system architecture and attributes, Procedia Manuf. 1 W.H.M. Zijm, Selecting parts for additive manufacturing in service logistics, J.
(2015) 273–286. Manuf. Technol. Manage. 27 (7) (2016) 915–931.
[11] S. Alumur, B.Y. Kara, Network hub location problems: the state of the art, Eur. J. [47] Y. Li, G. Jia, Y. Cheng, Y. Hu, Additive manufacturing technology in spare parts
Oper. Res. 190 (1) (2008) 1–21. supply chain: a comparative study, Int. J. Prod. Res. (2016) 1–18.
[12] M.S. Daskin, L.K. Dean, Location of health care facilities, Operations Research and [48] I. Sirichakwal, B. Conner, Implications of additive manufacturing for spare parts
Health Care, Springer US, 2005, pp. 43–76. inventory, 3D Print. Addit. Manuf. 1 (2016) 56–63.
[13] L. Chen, J. Olhager, O. Tang, Manufacturing facility location and sustainability: a [49] S.H. Khajavi, J. Partanen, J. Holmström, Additive manufacturing in the spare parts
literature review and research agenda, Int. J. Prod. Econ. 149 (2014) 154–163. supply chain, Comput. Ind. 65 (1) (2014) 50–63.
[14] M.S. Daskin, Network and Discrete Location: Models, Algorithms, and Applications, [50] H.J. Nyman, P. Sarlin, From bits to atoms: 3D printing in the context of supply chain
2nd edition, John Wiley & Sons, 2011. strategies, System Sciences (HICSS), 2014 47th Hawaii International Conference on
[15] M.S. Daskin, What you should know about location modeling, Naval Res. Logist. IEEE (2014) 4190–4199 (January).
(NRL) 55 (4) (2008) 283–294. [51] C. Weller, R. Kleer, F.T. Piller, Economic implications of 3D printing: market
[16] J. Ebery, Solving large single allocation p-hub problems with two or three hubs, structure models in light of additive manufacturing revisited, Int. J. Prod. Econ. 164
Eur. J. Oper. Res. 128 (2) (2001) 447–458. (2015) 43–56.
[17] G. Mayer, B. Wagner, HubLocater: an exact solution method for the multiple allo- [52] Robin Kleer, Frank T. Piller, Modeling benefits of local production by users: welfare
cation hub location problem, Comput. OR 29 (2002) 715–739. effects of radical innovation in flexible manufacturing utilizing additive manu-
[18] H. Topcuoglu, F. Corut, M. Ermis, G. Yilmaz, Solving the uncapacitated hub location facturing and 3D printing, Presented at the 73rd Annual Meeting of the Academy of
problem using genetic algorithms, Comput. OR 32 (4) (2005) 967–984. Management 2013, Orlando, FL, 2013.
[19] J.F. Chen, A hybrid heuristic for the uncapacitated single allocation hub location [53] J. Manners-Bell, K. Lyon, The implications of 3D printing for the global logistics
problem, Omega 35 (2007) 211–220. industry, Transp. Intell. (2012) 1–5.
[20] S. Elhedhli, F.X. Hu, Hub-and-spoke network design with congestion, Comput. [54] P. Reeves, How rapid manufacturing could transform supply chains, Supply Chain
Oper. Res. 32 (2005) 1615–1632. Q. 2 (04) (2008) 32–336.
[21] L.V. Snyder, M.S. Daskin, Reliability models for facility location: the expected [55] C. Achillas, D. Aidonis, E. Iakovou, M. Thymianidis, D. Tzetzis, A methodological
failure cost case, Transp. Sci. 39 (3) (2005) 400–416. framework for the inclusion of modern additive manufacturing into the production
[22] S. Melkote, M.S. Daskin, An integrated model of facility location and transportation portfolio of a focused factory, J. Manuf. Syst. 37 (2015) 328–339.
network design, Transp. Res. Part A: Policy Pract. 35 (6) (2001) 515–538. [56] A. Kieviet, S.M. Alexander, Is your supply chain ready for additive manufacturing?
[23] H.A. Eiselt, V. Marianov (Eds.), Foundations of Location Analysis, vol. 155, Springer Supply Chain Manage. Rev., 2015, http://www.scmr.com/article/is_your_supply_
Science & Business Media, 2011. chain_ready_for_additive_manufacturing.
[24] M.S. Daskin, K.L. Maass, The p-median problem, Location Science, Springer [57] NAICS, (2016) NAICS Identification Tools, June 28, 2016 (https://www.naics.com/
International Publishing, 2015, pp. 21–45. search/, Accessed 28.6.2016).
[25] C.C. Lin, J.Y. Lin, Y.C. Chen, The capacitated p-hub median problem with integral [58] M. Baumers, Economic Aspects of Additive Manufacturing: Benefits, Costs and
constraints: an application to a Chinese air cargo network, Appl. Math. Model. 36 Energy Consumption (Doctoral Dissertation), Martin Baumers, 2012.
(6) (2012) 2777–2787. [59] U.S. Census Bureau, State and County QuickFacts, (2010) (https://www.census.
[26] S. Çetiner, C. Sepil, H. Süral, Hubbing and routing in postal delivery systems, Ann. gov/quickfacts/table/POP010210/17189 , Accessed 4.3.2017).
Oper. Res. 181 (1) (2010) 109–124. [60] M.G. Kay, Matlog: Logistics Engineering Matlab Toolbox, (2017) (http://www4.
[27] J.J. Wang, M.C. Cheng, From a hub port city to a global supply chain management ncsu.edu/∼kay/matlog/ , Accessed 20.1.2017).
center: a case study of Hong Kong, J. Transp. Geogr. 18 (1) (2010) 104–115. [61] FedEx, FedEx Standard List Rates, (2015) Effective January 4, 2016 (https://www.
[28] R.Z. Farahani, M. Hekmatfar, A.B. Arabani, E. Nikbakhsh, Hub location problems: a fedex.com/us/services/pdf/FedEx_StandardListRates_2016.pdf , Accessed
review of models, classification, solution techniques, and applications, Comput. Ind. 8.7.2016).
Eng. 64 (4) (2013) 1096–1109. [62] FedEx, 2016 Service Guide, (2016) Updated July 1, 2016 (https://www.fedex.com/
[29] H. Yaman, The hierarchical hub median problem with single assignment, Transp. us/services/pdf/Service_Guide_2016.pdf , Accessed 8.7.2016).
Res. B: Methodol. 43 (6) (2009) 643–658. [63] GE Reports, All The 3D Print That’s Fit to Pitt: New Additive Technology Center
[30] C.C. Lin, The integrated secondary route network design model in the hierarchical Opens Near Steel Town, (2016) September 6, 2016 (http://www.gereports.com/all-
hub-and-spoke network for dual express services, Int. J. Prod. Econ. 123 (1) (2010) the-print-thats-fit-to-pitt-new-additive-technology-center-opens-near-steel-town/ ,
20–30. Accessed 4.3.2017).
[31] R. Figueiredo, M.E. O'Kelly, N.D. Pizzolato, A two-stage hub location method for air [64] L. Cooper, Location-allocation problems, Oper. Res. 11 (1963) 331–343.
transportation in Brazil, Int. Trans. Oper. Res. 21 (2) (2014) 275–289. [65] IsBisWorld, Innovation in Creation: Demand Rises While Prices Drop for 3D Printing
[32] I. Gibson, D. Rosen, B. Stucker, Additive Manufacturing Technologies: 3D Printing, Machines, (2017) February 16, 2016 (https://www.ibisworld.com/media/2016/
Rapid Prototyping, and Direct Digital Manufacturing, Springer, 2014. 02/16/innovation-in-creation-demand-rises-while-prices-drop-for-3d-printing-
[33] W.E. Frazier, Metal additive manufacturing: a review, J. Mater. Eng. Perform. 23 machines/ , Accessed 28.2.2016).
(6) (2014) 1917–1928. [66] B. Audretsch, Agglomeration and the location of innovative activity, Oxf. Rev. Econ.
[34] L.E. Murr, S.M. Gaytan, D.A. Ramirez, E. Martinez, J. Hernandez, K.N. Amato, Policy 14 (2) (1998) 18–29.
P.W. Shindo, F.R. Medina, R.B. Wicker, Metal fabrication by additive manu- [67] C. Lindemann, U. Jahnke, M. Moi, R. Koch, Analyzing product lifecycle costs for a
facturing using laser and electron beam melting technologies, J. Mater. Sci. better understanding of cost drivers in additive manufacturing, 23th Annual
Technol. 28 (1) (2012) 1–14. International Solid Freeform Fabrication Symposium–An Additive Manufacturing
[35] Beaman J.J., Deckard C.R., (1990), U.S. Patent No. 4,938,816. Washington, DC: U.S Conference, Austin, Texas, USA 6th–8th August, 2012.
. Patent and Trademark Office. Berman, B., 2012. 3-D printing: The new industrial [68] D.S. Thomas, S.W. Gilbert, Costs and Cost Effectiveness of Additive Manufacturing,
revolution. Business Horizons, 55 (2), pp. 155–162. US Department of Commerce, 2014 (Consulted at: http://nvlpubs.nist.gov/nist-
[36] J.B. Jones, P. Mcnutt, R. Tosi, C. Perry, D.I. Wimpenny, Remanufacture of turbine pubs/Special Publications/NIST. SP, 11, p.76.).
blades by laser cladding, machining and in-process scanning in a single machine,
173