Professional Documents
Culture Documents
MRR (2019)
MRR (2019)
MRR (2019)
https://doi.org/10.1108/MRR-03-2017-0094
Downloaded on: 30 January 2019, At: 21:49 (PT)
References: this document contains references to 35 other documents.
To copy this document: permissions@emeraldinsight.com
The fulltext of this document has been downloaded 55 times since 2019*
Users who downloaded this article also downloaded:
(2018),"Analysis of a multi-echelon supply chain problem using revised multi-choice goal
programming approach", Kybernetes, Vol. 47 Iss 1 pp. 118-141 <a href="https://doi.org/10.1108/
K-05-2017-0189">https://doi.org/10.1108/K-05-2017-0189</a>
Access to this document was granted through an Emerald subscription provided by emerald-
srm:397875 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald
for Authors service information about how to choose which publication to write for and submission
guidelines are available for all. Please visit www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company
manages a portfolio of more than 290 journals and over 2,350 books and book series volumes, as
well as providing an extensive range of online products and additional customer resources and
services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the
Committee on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for
digital archive preservation.
MRR
42,1 A practical supply-demand hub
in industrial clusters:
a new perspective
68 Vahid Kayvanfar
Department of Industrial Engineering, Amirkabir University of Technology,
Received 30 March 2017 Tehran, Islamic Republic of Iran, and
Revised 4 July 2017
9 October 2017 Singapore Institute of Manufacturing Technology, Singapore, Singapore
8 June 2018
S.M. Moattar Husseini
Downloaded by National University of Singapore At 21:49 30 January 2019 (PT)
Abstract
Purpose – This paper aims to optimize the interactions of businesses located within industrial clusters (ICs)
by using a supply-demand hub in ICs (SDHIC) as a conjoint provider of logistics and depository facilities for
small- and medium-sized enterprises (SMEs) as producers, where all of these interactions are under
supervision of a third-party logistics provider (3PL).
Design/methodology/approach – To evaluate the values of SDHIC, three mathematical models are
proposed, optimally solved via GAMS and then compared. Also, a “linear relaxation-based heuristic”
procedure is proposed to yield a feasible initial solution within a significant shorter computational time. To
illustrate the values of SDHIC, comprehensive calculations over a case study and generated sets of instances
are conducted, including several sensitivity analysis.
Findings – The experimental results demonstrate the efficiency of SDHIC for SMEs via combining batches
and integrating the holding space of inventories, while the outcomes of the case study are aligned with those
obtained from random sample examples, which confirms the trueness of used parameters and reveals the
applicability of using SDHIC in real world. Finally, several interesting managerial implications for
practitioners are extracted and presented.
Practical implications – Some of the managerial and practical implications are optimizing interactions
of businesses involved in a supply chain of an IC containing some customers, suppliers and manufacturers
and rectifying the present noteworthy gaps pertaining to the previously published research via using real
assumptions and merging upstream and downstream of the supply chain through centralizing on storage of
raw materials (supply echelon) and finished products (demand echelon) at the same place simultaneously to
challenge a classic concept in which supply and demand echelons were being separately planned regarding
their inventory management and logistics activities and showing the positive consequences of such challenge,
showing the performance improvement of the proposed model compared to the classic model, by increasing
the storing cost of raw materials and finished products, considering some disadvantages of using SDHIC and
Management Research Review showing the usefulness of SDHIC in total, presenting some applied findings according to the obtained results
Vol. 42 No. 1, 2019
pp. 68-101 of sensitivity analysis.
© Emerald Publishing Limited
2040-8269
Originality/value – The key contributions of this paper to the literature are suggesting a new applied
DOI 10.1108/MRR-03-2017-0094 mathematical methodology to the supply chain (SC) of ICs by means of a conjoint provider of warehousing
activities called SDHIC, comparing the new proposed model with the two classic ones and showing the Supply-
proposed model’s dominancy, showing the helpful outcomes of collaborating 3PL with the SMEs in a cluster,
proposing a “linear relaxation-based heuristic” procedure to yield a feasible initial solution within a demand hub
significant shorter computational time and minimizing total supply chain costs of such IC by optimum
application of facilities, lands and labor.
Keywords 3PL, Supply chain management (SCM), Small and medium enterprises (SMEs),
Industrial clusters (IC), Production and operations management, Supply hub,
Supply-demand hub 69
Paper type Research Paper
1. Introduction
The term industry cluster (IC), the connection of interrelated firms cooperating together
Downloaded by National University of Singapore At 21:49 30 January 2019 (PT)
inside the alike profitable sector and which are close to each other physically and enjoy
uncommon economical achievement in their area, was presented and spread by Michael
Porter (Porter, 1990). According to Porter (1998), the economic plot of the world is at the
present time dominated by industrial clusters (ICs). The actions of companies inside ICs
depend on convinced local specificities such as availability of natural resources and centers
of technological development (Oprime et al., 2011). As the companies are geographically
close to each other inside the IC and their activities are also in the same area or nearly so, the
IC elements enjoy the economic advantages of synergies and location-specific externalities
(Shakya, 2009). Also, the ICs benefit from scale without managing the rigidities of formal
relations or vertical combination. Porter (1998) shortened that ICs have a significant impact
on competition and create competitive advantages in one of the following manners:
by increasing the productivity of firms located inside the cluster;
by motivating the trend and innovation pace, which strengthens the development of
upcoming productivity; and
by stimulating the formation of new firms that strengthens and enlarges the cluster,
forming an optimistic feedback.
In spite of such importance, one can barely discover any quantitative methodology in the
literature which investigates the interactions of entities within an IC. In better words, the
optimization of ICs’ activities is inadequately studied so far, and in this research, an
endeavor is made to remove such a significant gap in the literature. Bastida-Ruiz et al.,
(2013) suggested a sustainability signs outline for industrial parks so that information is
inadequately reliable or not enough. They attempted to remove these gaps and made an
indicator outline. Giannoccaro (2015) investigated the correlation between learning and
adaptation in supply chain (SC) positioned in ICs so as to specify the best adaptive SC. The
consequences revealed which SC type suggests the maximum adaptive performance among
different considered types. They presented some managerial insights as well. Sarach (2015)
analyzed cooperative relationships in ICs and implemented the proposed methodology of
collaboration analysis. This study revealed that the internet and the age of globalization
fundamentally have an effect on the collaboration development in ICs.
Today, few organizations are independent; rather, they collaborate and depend on each
other. The most important methodologies developed to improve coordination and
cooperation of different firms toward a common aim are supply chain management (SCM)
and IC (Tolossa et al., 2013). The mixture of two concepts, i.e. SCM and IC, can proficiently
boost the competitive advantage of industries so as to enhance local economic
competitiveness. DeWitt et al., (2006) showed the linking of supply chain and Porter’s cluster
MRR theory and presented indication of their potential joint positive impact on performance of the
42,1 organization. Huang and Xue (2012) presented several hypothetical propositions and applied
approaches in the area of cluster supply chain (CSC). Qu et al. (2015) studied a
comprehensive analysis for the total operations of CSC. By proposing a wide-ranging
investigation of CSC’s formation policies in typical phases, they presented a common CSCC
framework and then segmented into an augmented Lagrangian coordination-based
70 decentralized decision framework through the classic decision autonomy distribution in ICs.
Banasik et al. (2017) proposed a multi-objective model to investigate the consequences of
closed-loop supply chain in a mushroom cluster. Connecting phases of an SC for effective
movements of information and material is one of the purposes of an IC. The frequent
interactions of entities within IC can increase competition and improve coordination,
efficiency and innovation, along with improvement in trust among supply chain partners.
Downloaded by National University of Singapore At 21:49 30 January 2019 (PT)
Moreover, the 3PL is given larger responsibilities in collaborating with SMEs in this
research. The adopted approach in this research is more comprehensive, which considers
different criteria simultaneously in a real situation.
The supply hub was first introduced by Barnes et al. (2000) as a place physically
close to a manufacturer’s site. All or part of its suppliers are warehoused through the
agreement that the price of materials will be paid only when consumed. Barnes et al.
(2000) also specified the advantages and disadvantages of supply hub’s application, as
well as an overview of its development. Shah and Goh (2006) proposed an organized
hierarchical model for managing the inventories according to the supply hub so as to
aid in achieving the stability between different bodies existing in the supply chain.
Several researchers such as Li et al. (2008), Li et al. (2009) and Shuang-yan et al. (2013)
showed benefits of using the supply hub. Qiu et al. (2014) studied how manufacturers
and SHIP can interact so as to optimize their decision on replenishment, delivery and
storage pricing. Qiu and Huang (2016) investigated the cooperating decision-making
problem between existing enterprises inside the IC and SHIP in transportation service
sharing thorough bi-level attitude.
The rest of sections in this research is as follows. Section 2 explains the problem
description. Section 3 describes the three proposed mathematical models. Section 4 defines
solution approach. Section 5 demonstrates experimental results, including sensitivity
analysis, and Section 6 states conclusions and future works.
2. Problem description
The considered supply chain in this study comprises multiple external suppliers, an
SDHIC and several SMEs/manufacturers located in a proximity of each other inside an IC.
The suppliers may be close to IC. Raw materials are first distributed from suppliers to the
SDHIC, which is managed by 3PL (all inbound and outbound transportation are handled
by 3PL) and then delivered to the SMEs/manufacturers based on their production
planning requirements. The manufacturers/SMEs as decision-makers in this IC
coordinate their decisions regarding production with 3PL. Actually, a given space to
possible extent is assigned to each SME by 3PL to hold both its raw materials and
finished products. According to this allocated storage capacity, each SME should make
MRR decision about two issues: how many products should be manufactured in each period
42,1 and how often. The order is initially sent to the manufacturers by customers where its
duplicate is immediately directed to the 3PL. Then, 3PL requests to correspondent SME
(s) for producing such product(s). There is information sharing between manufacturers
and 3PL to better manage the logistics activities. Thereafter, the manufacturer(s) start to
produce the given product once they receive the raw materials as no inventory is kept at
72 manufacturers’ sites. Each producer has also a limited capacity for manufacturing, and
after production, all finished products are directly shipped to SDHIC. One can store both
raw materials and finished products only in SDHIC, and as the holding capacity is
limited, the materials and products enter a competition for getting more storage space in
each period. Each supplier and manufacturer only provide one type of raw material and
one type of finished product, respectively. Owing to trucks’ capacity limitation besides
Downloaded by National University of Singapore At 21:49 30 January 2019 (PT)
restricted storage space of SDHIC, shortage may occur. Besides minimizing total logistics
cost of the system, to manage the clusters from the supply-demand viewpoint is a key
objective of this paper. The interactions of entities through supply hub, SDHIC and
without any hub are depicted in Figure 1.
The main difference of using supply hub [Figure 1(b)] compared to using SDHIC
[Figure 1(a)] is that only needed raw materials are stored in supply hub and each
manufacturer has its own private warehouse for holding its finished product. All
ordered finished products by customers are shipped from each manufacturer’s private
warehouse to them. The other difference is that freight consolidation occurs only in
“supply echelon” in Figure 1(b), while it happens in both supply and distribution
echelons by using SDHIC [Figure 1(a)]. In the classic model [Figure 1(c)], there is no
central warehouse and the most assumed assumptions are similar to those of the second
model [Figure 1(b)], excluding the following ones:
Each manufacturer has a private warehouse for storing its needed raw materials
and produced finished products.
No freight consolidation occurs.
73
Downloaded by National University of Singapore At 21:49 30 January 2019 (PT)
Figure 1.
Interactions of
entities in an IC with
SDHIC (a) with
supply hub (b) and
with no public hub (c)
MRR The aim of this paper is to minimize the whole imposed logistics costs in both supply and
42,1 distribution sides including:
ordering costs of commodities to producers, including fixed costs of communications,
administration tasks, etc.;
storing costs of inventories comprising both raw materials and finished products at
SDHIC throughout time horizon, which is calculated based on inventory level at the
74 end of period;
shipping costs of raw material from suppliers to central hub, i.e. SDHIC and
transferring commodities from SDHIC to customers, comprising costs of different
used trucks types; and
shortage costs of finished products, which is back ordered, and in the last period, all
Downloaded by National University of Singapore At 21:49 30 January 2019 (PT)
According to the aforementioned elucidations, the key contributions of the current paper
could be presented as:
Propose a practical model to the supply chain of ICs using SDHIC.
Compare the new proposed model (supply-distribution model with SDHIC, in
Section 3.3.1) with the two classic ones (“supply-distribution model with supply
hub” in Section 3.3.2 and “the classic supply chain model” in Section 3.3.3) and
showing our model’s dominance.
Propose a “linear relaxation-based heuristic” procedure to yield a feasible initial
solution within a significant shorter computational time.
Reach the optimal logistics costs level by means of optimum usage of existing
facilities, lands, labor, etc.
3.1 Assumptions
The main assumptions considered in this research are as follows:
In each period, the demand of customers is independent and prearranged. This
assumption is valid especially for general commodities such as sugar.
A cluster of manufacturers/SMEs and SDHIC is managed by a 3PL. As already
mentioned, most SMEs face major challenges and problems comprising profitability,
business planning, low productivity, people, business versus personal, managerial
experience, cash flow, decision-making information and succession planning (Bentley-
Jennison, 2006). To resolve so, SMEs are persuaded to outsource their logistics
undertakings to 3PLs. This assumption is actually one of the properties of such
supply chains.
The manufacturing capacity of each producer is restricted in each period. This is a
practical assumption, as in reality, the production volume of any manufacturer
could not be extreme and is restricted.
SDHIC has a restricted space for storing raw materials and finished Supply-
products. This assumption is aligned with the reality as any warehouse has demand hub
limited capacity for storing inventories, either raw materials or finished
products.
Both materials and commodities are simultaneously stored in SDHIC and both
compete to achieve more holding space. As the SDHIC has restricted space and both
raw material and finished products should be stored in SDHIC and in each period, 75
the volume of held materials and commodities in SDHIC may vary, obviously,
competition happens for achieving more storage space from raw material and
finished products’ sides.
Each supplier can only provide one type of raw material and each producer can only
manufacture one type of commodities/products. This assumption is considered as
Downloaded by National University of Singapore At 21:49 30 January 2019 (PT)
the base assumption in this area (owing to simplicity), and the more complicated
situations which assume “each supplier and manufacture can procure/produce more
than one item” could be considered in future studies based on an extension of this
research.
No inventory is kept at manufacturers’ sites. As the manufacturers have no private
warehouse (owing to existence of SDHIC as the central warehouse in IC), obviously
no inventory could be held in their sites/warehouses.
The shortage is allowable thanks to the capacity limitation of trucks and holding
area of SDHIC and is backlogged. This assumption is one of the realistic
assumptions which frequently happens in real world and is neglected in many
previous research.
There are different types of trucks, each of which can ship both materials and
commodities/products between different corporations. This assumption is
expansion of the base state which considers “there is only one type of trucks in the
system.” Moreover, taking different types of trucks into account increases the
flexibility of the system and can reduce the total costs, as one can profit truckload
privilege in the supply chain.
The procurement lead time of raw materials for manufacturers is equal to zero. As
already mentioned, the manufacturers have no private warehouse (owing to existing
the SDHIC) and it is understandable that no inventory is kept in their warehouses/
sites. Also, due to frequent deliveries to the SMEs and negligible distance from
SDHIC to manufactures [Figure 1(a)], Just-in-Time procurement happens and
consequently, the procurement lead time of raw materials for the manufacturers is
equal to zero.
In each period, delivered quantity of raw materials by suppliers to the producers is
equal to their requirements. As the manufacturers cannot hold any amount of
inventories, they should receive at most the required amount of raw materials in
each period.
Freight consolidation is included in shipping of materials both from suppliers
to central hub (SDHIC) and from SDHIC to customers. As in any
transshipment from supplier to SDHIC and from SDHIC to manufactures or
vice versa, different needed types of raw materials or finished products could
be carried in one truck; therefore, freight consolidation happens in the
system.
MRR 3.2 Symbolizations
42,1 3.2.1 Subscripts
m (m = 1, 2,. . .,M) = Finished product/producers;
u (u = 1,2,. . .,H) = Truck;
s (s = 1,2,. . .,S) = Raw material/supplier;
c (c = 1,2,. . .,C ) = Customer; and
76 t (t = 1, 2,. . .,T) = Period.
3.2.2 Parameters
T = Number of cycles;
fsm = The unit amount of needed material type s used to create one unit of product type m;
i = The rate of storage cost for stocks;
Downloaded by National University of Singapore At 21:49 30 January 2019 (PT)
period t; and
Ymt = 1 if producer m in period t makes product type m; 0 otherwise..
X
S1 ¼ Ordering cost ¼ OCmt : Ymt (1-1)
m;t
X
S2 ¼ Transportation cost of all types of materials ¼ aus : Numstu (1-2)
s;t;u
X 0u
S3 ¼ Transportation cost of all types of products ¼ b uc : Num ct (1-3)
c;t;u
X
S4 ¼ Storing costs of different types of materials ¼ i:cs : Invst (1-4)
s;t
X 0
S5 ¼ Storing costs of different types of products ¼ i:cm : Inv mt (1-5)
m;t
MRR S6 ¼ Functioning cost of storing stocks at SDHIC ¼ T : h : CaSDH (1-6)
42,1 X
S7 ¼ Shortage cost ¼ BCmct : Backmct (1-7)
m;c;t
Constraint (2) ensures that in each period total materials received from all suppliers as
well as total finished products received from all producers are fewer than the unoccupied
capacity of SDHIC. In other words, as the volume of SDHIC is restricted and both
products and materials should be stored at SDHIC simultaneously, in each period, any
newly added amount of inventories should only be held in the left space of SDHIC from
the previous period:
Constraint (3) limits production amount of each producer to its manufacture capacity in each
period. Constraint (3) exists in real world as in reality, the production volume of any
manufacturer could not be extreme and is restricted:
X
M
QSsmt # CaSst 8s; t (4)
m¼1
Constraint (4) makes sure that in each period, total distributed material type s needed for
creating all types of products to SDHIC is fewer than the capacity of supplier s. Similar to
Constraint (3), this inequality should be established as a practical constraint, as no supplier
can produce more than its capacity:
X
M X
M
Invst þ QSTsmt ¼ Invst1 þ QSsmt 8s; t (5)
m¼1 m¼1
Constraint (5) in each period balances stock amount (inventory level) of any kind of
materials at SDHIC’s space. In other words, the amount of material type s at SDHIC at the
end of previous period (period t-1) plus amount of material type s necessary for fabricating
product type m sent to SDHIC in present period (period t) are equal to amount of material
type s carried to producer m from SDHIC in current period (period t) plus left amount of Supply-
material type s at SDHIC at the end of current period (period t): demand hub
X
C
0 0
QPTmct ¼ Inv mt1 þ QPmt Inv mt 8m; t (6)
c¼1
Similar to Constraint (5), Constraint (6) in each period balances stock amount (inventory 79
level) of products at SDHIC’s space. In other words, amount of product type m made in
current period (period t) plus left amount of product type m from previous period (period t-1)
are equal to quantity of product type m carried to customer c from SDHIC in current period
(period t) plus the left volume of product type m in current period (period t):
Downloaded by National University of Singapore At 21:49 30 January 2019 (PT)
X
M X
H
g Ts QSsmt # Numust : Cavu 8s; t (7)
m¼1 u ¼1
Constraint (7) guarantees that all needed material type s for making all products is smaller
than what is carried from supplier s to SDHIC in each period. Simultaneously, Constraint (7)
identifies, in each period, how much material type s should be distributed by which truck
type from that supplier to SDHIC. Indeed, all required raw materials ordered by all
manufacturers are consolidated into a single batch by each supplier, separately:
X
M X
H
0u 0
QPTmct : g Tm # Num ct : Cav u 8c; t (8)
m¼1 u ¼1
Similar to Constraint (7), Constraint (8) specifies, in each period, how much of product type
m from SDHIC to each customer should be sent out by which type of trucks. It should be
stated that both upstream and downstream of chain, i.e. supply and distribution echelons,
take benefit of freight consolidation in batches. In Constraint (8), all finished products which
are ordered by each customer are combined into a single batch/truck. Also, in calculating
objective function, owing to different distances between suppliers and central warehouse
(SDHIC) and between SDHIC and customers, the utilization costs of different trucks types
are taken different into account:
Dmct þ Backmct1 ¼ QPTmct þ Backmct 8m; c; t (9)
Constraint (9) in each period establishes equilibrium in inventory level of different types of
products in SDHIC regarding the customers’ demand and sends out volume of products. As
in each period, the whole amount of demand ordered by all customers may not be satisfied,
the unfulfilled quantity is backlogged. In better words, the demand of customer c
from product type m in current period (period t) plus backordered amount of product type m
from previous period (period t-1) are equal to amount of product type m sent to customer c
from SDHIC in current period (period t), and backordered quantity of product type m
belongs to customer c in current period (period t):
QPTmcT ¼ DmcT þ BackmcT1 8m; c (10)
Constraint (10) ensures that in the last period, all backordered quantities should be
fulfilled. That is to say that in the last period (period T), the volume of product type m
sent to customer c from SDHIC is equal to the demand of customer c from product type
MRR m, and backordered volume of product type m goes to customer c from previous
42,1 period:
fsm QPmt ¼ QSTsmt 8s; m; t (11)
Constraint (11) is concerned with the sent amount of all materials to each producer in each
period and the created products by that producer. In fact, Constraint (11) makes sure that in
80 each period, the manufactured quantity of each product at producer m is equal to the
received quantity of material type s by that producer which is required for making this
product:
X
M
0
X
S
Inv mt : g H
m þ Invst : g H
s # CaSDH 8t (12)
Downloaded by National University of Singapore At 21:49 30 January 2019 (PT)
m¼1 s¼1
Constraint (12) assures that the total amount of inventory level of each type of materials and
products in each period does not disrupt the volume of SDHIC. In other words, Constraint
(12) in each period causes a competition between materials and products to be allotted more
holding area:
0
QPmt ; QSsmt ; QSTsmt ; QPTmct ; Invst ; Inv mt ; Backmct 2 Rþ
0u
Numstu ; Num ct 2 Zþ ; Ymt 2 f0; 1g 8s; m; c; t
(13)
Constraint (13) displays the types of used decision variables. As it can be seen, there are one
binary and two integer variables, while the rest of variables are continuous.
3.3.2 Supply distribution model with supply hub. The considered supply chain in this
subsection [Figure 1(b)] includes some suppliers, a supply hub and several manufacturers in
the proximity of each other within an IC, as well as some customers. Its objective function,
Constraint (14) is equivalent to Constraint (1) and minimizes the whole imposed logistics
expenses in the considered supply chain comprising seven parts, i.e. Part 1 (S1) to Part 7 (S7):
(14-1) production costs of all types of products by manufactures, (14-2) transportation costs
of different types of materials from suppliers to supply hub, (14-3) transportation costs of
products from manufacturer’s warehouse to customers, (14-4) inventory storing expenses of
all types of materials at supply hub, (14-5) stocks storing expenses of all products at
manufacturer’s warehouse, (14-6) operational fixed cost of storing stocks at supply hub
along with manufacturers’ warehouses and (14-7) shortage costs of finished products. Thus,
the objective function is as follows:
X
S1 ¼ Production cost ¼ FPCmt : Ymt (14-1)
m;t
X
S2 ¼ Transportation cost of all types of materials ¼ asu : Numust (14-2)
s;t;u
X 000 u
S3 ¼ Transportation cost of all types of products ¼ u
h mc : Num mct (14-3) Supply-
m;c;t;u demand hub
X
S4 ¼ Storing costs of different types of materials ¼ i:cs : Invst (14-4)
s;t
X 0 81
S5 ¼ Storing costs of different types of products ¼ i : cm : Inv mt (14-5)
m;t
X
S6 ¼ Functioning cost of storing stocks ¼ h: T : CaH þ CaMmt (14-6)
m;t
Downloaded by National University of Singapore At 21:49 30 January 2019 (PT)
X
S7 ¼ Shortage cost ¼ BCmct : Backmct (14-7)
m;c;t
In this model, Constraints (3)-(7) and Constraints (9)-(11) are the same as the main model
presented in Section 3.3.1, and the rest of constraints are as follows:
X
H
00 u 0
g Tm QPTmct # Num mct : Cav u 8m; c; t (15)
u ¼1
Constraint (15) specifies how much of product type m should be dispatched by which truck
type from each producer’s warehouse to each customer in each period. Against the
equivalent constraint in the main model [Constraint (8)], there is no freight consolidation in
any of demand/customer’s side/echelon in this model. In calculating the objective function,
owing to different distances between private warehouses of manufacturers and customers,
the utilization costs of different trucks types are taken different into account.
X X X
g Hs : Invst1 þ g Hs QSsmt
s s m
X S
M X
QSTsmt g H
s # CaH 8t (16)
m¼1 s¼1
Constraint (16) makes sure that the total inventory level of all types of materials in each
period does not violate the capacity of supply hub. As the capacity of supply hub is limited, in
each period, any newly added amount of raw materials should only be held in the left space of
supply hub from the previous period. To be precise, the left quantity of raw material from the
previous period (period t-1) plus the new amount of raw material transferred to supply hub in
the current period (period t) are less than the capacity of supply hub:
XC
0H 0 0H 0H
g m : Inv mt1 þ g m : QPmt QPTmct g m # CaMmt 8m; t (17)
c¼1
Constraint (17) ensures that the sum of inventory level of finished product at each
manufacturer’s warehouse in each period does not violate its capacity. That is to say, the left
quantity of all types of products from the previous period (period t-1) plus the new amount
MRR of products transferred to each producer’s warehouse in the current period (period t) are less
42,1 than the capacity of each producer’s warehouse:
QPmt ; QSsmt ; QSTsmt ; QPTmct ; Invst ; Inv mt ; Backmct 2 Rþ
0
000 u
Numust ; Num mct 2 Zþ ; Ymt 2 f0; 1g 8s; m; c; t
(18)
82
Constraint (18) displays the types of used decision variables. As it can be seen, there are one
binary and two integer variables, while the rest of variables are continuous.
3.3.3 The classic supply chain model. The considered supply chain in this subsection
[Figure 1(c)] includes some suppliers, several manufacturers in the proximity of each other
within an IC and some customers. Its objective function, Constraint (19), is equivalent to
Downloaded by National University of Singapore At 21:49 30 January 2019 (PT)
Constraint (1) and minimizes the whole imposed logistics expenses in the considered supply
chain comprising seven parts, i.e. from Part 1 (S1) to Part 7 (S7): (19-1) production costs of
finished products by manufactures, (19-2) transportation costs of different types of materials
from suppliers to producer’s warehouse, (19-3) transportation costs of all types of products
from producer’s warehouse to customers, (19-4) stocks storing expenses of materials at
producer’s warehouse, (19-5) stocks storing expenses of products at producer’s warehouse,
(19-6) operational fixed cost of storing stocks at producers’ warehouses and (19-7) shortage
costs of finished products. The objective function could be stated as follows:
min ðTPC þ TTC þ THC þ TSC Þ¼ S1 þ ðS2 þ S3Þ þ ðS4 þ S5 þ S6Þ þ S7
X
S1 ¼ Production cost ¼ FPCmt : Ymt (19-1)
m;t
X 00 u
S2 ¼ Transportation cost of all types of materials ¼ d usm : Num smt (19-2)
s;t;u
X 000 u
S3 ¼ Transportation cost of all types of products ¼ h umc : Num mct (19-3)
m;c;t;u
X 00
S4 ¼ Storing costs of different types of materials ¼ i : cs : Inv smt (19-4)
s;t
X 0
S5 ¼ Storing costs of different types of products ¼ i : cm : Inv mt (19-5)
m;t
X
S6 ¼ Functioning cost of storing stocks ¼ h: CaMmt (19-6)
m;t
X
S7 ¼ Shortage cost ¼ BCmct : Backmct (19-7)
m;c;t
In this model, Constraints (3), (4), (6), (9), (10) and (15) are the same as the previous models Supply-
presented in Sections 3.3.1 and 3.3.2, and the rest of constraints are as follows: demand hub
00 00
Inv smt þ QPmt : fsm ¼ Invsmt1 þ QSsmt 8m; t; s (20)
Constraint (20) in each period establishes equilibrium in inventory quantity of any type of
materials at each producer’s warehouse. In other words, the amount of material type s at
producers’ warehouses at the end of previous period (period t-1) plus quantity of material
83
type s necessary for making product type m sent to producer m in current period (period t)
are equal to consumed amount of material type s for manufacturing product type m in
current period (period t) plus left volume of material type s at producers’ warehouses at the
end of current period (period t):
Downloaded by National University of Singapore At 21:49 30 January 2019 (PT)
X
H
00 u
g Ts QSsmt # Num smt : Cavu 8s; m; t (21)
u ¼1
Constraint (21) makes sure that in each period, the necessary material type s for building
product type m is fewer than what is carried from supplier s to producer m. In other words,
Constraint (21) concurrently specifies in each period how much material type s should be
shipped by which truck type from that supplier to producer m. In such a classic supply
chain, there is no freight consolidation in any supply side/echelon. In calculating objective
function, owing to different distances between suppliers and manufacturers’ warehouses,
the utilization costs of different trucks types are taken different into consideration:
0 1
XS X
S
g m : Inv mt1 þ g m :QPmt þ @ g Hs : QSsmt A
0H 0 0H 00
g Hs : Inv smt1 þ
s¼1 s¼1
XS C
X
0H
fsm QPmt g H
s þ QPTmct g m # CaMmt 8m; t (22)
s¼1 c¼1
Constraint (22) guarantees that the total stock amount of all types of materials besides
products in each period does not disrupt the volume of each producer’s warehouse. In other
words, as the volume of each producer’s warehouse is restricted and all types of materials
and products should be stored simultaneously, in each period, any newly added amount of
inventories should only be held in the left space of each producer’s warehouse from the
previous period:
00 u 000 u (23)
Num smt ; Num mct 2 Zþ ; Ymt 2 f0; 1g 8s; m; c; t
Constraint (23) exhibits the types of used decision variables. As it can be observed, there are
one binary and two integer variables, while the rest of variables are continuous.
4. Solution approach
The proposed models are coded in GAMS and solved by the CPLEX, which are intended to
solve large-scale and sophisticated problems rapidly with least user intervention. For those
MRR problems with integer variables, the branch and cut (B&C) algorithm is used by CPLEX
42,1 through solving a series of linear programming (LP) and sub-problems. As a single mixed
integer problem (MIP) produces many sub-problems, and even small MIPs could be so
calculate-intensive and need noteworthy amounts of physical memory. B&C algorithm is a
combinatorial optimization technique used to solve integer linear programs (ILPs) including
LP problems where some or all the unknowns are constrained to integer values. B&C
84 comprises running a branch and bound technique and using cutting planes so as to tighten
the LP relaxations. If cuts are only used to tighten the initial LP relaxation, the algorithm is
called “cut and branch.” The pseudo-code of the general B&C technique is stated as follows:
1. Add the primary ILP to Q (the list of active problems)
2. Set p* = null and r* = 1
3. while the Q is not empty
Downloaded by National University of Singapore At 21:49 30 January 2019 (PT)
Set N^ ¼ ½N ~ þ 1; (N
~ is the set of relaxed values of variables
Numust and Numctu ).
Report W ^ ;N^ as the feasible solution of P0; (P0 is the original 85
problem).
Solve the relaxed problem P3 by fixing the set of variables W ^ and N
^;
Report S = {all continuous variables}, as the complete initial
feasible solution.
Downloaded by National University of Singapore At 21:49 30 January 2019 (PT)
The proposed linear relaxation-based heuristic should be solved in each period, where in the
first step, all binary variables Ymt should be relaxed (named as Y ~ mt ) to continuous variables
range of (0, 1] and then Problem P1 should be solved. It should be pointed out that P0 is the
original/main problem in which neither binary nor integer variables are relaxed, while P1 is
the problem P0 in which the variables Ymt are relaxed. If Y ~ mt 2 ð0; 1, fix Y ^ mt ¼ 1,
otherwise Y ^ mt ¼ 0. Now, to solve Problem P2, as there are two integer variables Numu and
st
Numctu , they should first be relaxed to continuous variables (named as the set N ~ ) and the
values of the set Y^ should also be fixed. P2 is the problem P0 in which the variables Numu
st
and Numuct are relaxed and variables Y ~ (the relaxed values of variables
Ymt) are fixed. Then,
^ ¼ ½N
to obtain the feasible values, one should set N ~ þ 1. The set Y ^;N ^ shows the feasible
solution of the original Problem P0. Finally, to attain a complete feasible initial solution
including the values of the rest of continuous variables, one should solve P3 by fixing the set
of variables Y ^ and N^ . P3 is the problem P0 in which all variables Y ^ and N ^ (the relaxed
values of variablesNumst and Numct ) are fixed. The above-mentioned steps should be
u u
1 2 2 5 5
2 3 5 8 5
3 5 10 15 10 Table I.
4 8 12 20 10 Characteristics of test
5 10 20 30 20 instances
MRR hub (CaH) are 54,000 and 28,000 units, respectively. Also, the capacity of manufacturers’
42,1 private warehouses (CaMmt) are regarded as 2,000 units for all of them.
It should be mentioned that OCmt = FPCmt (Vm,t) should be able to compare all proposed
mathematical models, i.e. order expense of product type m by 3PL in period t is equal to the
fixed production cost of product type m in period t and takes 1,500 into account.
The shortage cost at SDHIC and manufacturers’ sites are regarded the same for the
86 similar reason, i.e. 10. The primary stock levels of all types of materials and products at
SDHIC and producers’ private warehouses are set to zero. In the conducted numerical
experiments, two truck types are considered, each of which is assigned to different capacity
values/levels in implementations, i.e. (50,75) and (80,120). In better words, the capacity of
truck Type 1 is set to 50 and 80, and the capacity of truck Type 2 is set to 75 and 120.
Utilization constants of storage capacity at SDHIC/supply hub/producer’s warehouse to
Downloaded by National University of Singapore At 21:49 30 January 2019 (PT)
Parameter #1 #2 #3 #4 #5
h umc ð8mÞ h 1mc ¼ 150; 240 h 1mc ¼ 150; 240 h 1mc ¼ 150; 240 h 1mc ¼ 150; 240 h 1mc ¼ 150; 240
Table II.
h 2mc ¼ 187; 300 h 2mc ¼ 187; 300 h 2mc ¼ 187; 300 h 2mc ¼ 187; 300 h 2mc ¼ 187; 300
The used values for
parameters
Downloaded by National University of Singapore At 21:49 30 January 2019 (PT)
a1s ¼ d 1sm ¼ 100 a1s ¼ d 1sm ¼ 100 a1s ¼ d 1sm ¼ 160 a1s ¼ d 1sm ¼ 160
a2s ¼ d 2sm ¼ 112 a2s ¼ d 2sm ¼ 180 a2s ¼ d 2sm ¼ 112 a2s ¼ d 2sm ¼ 180
b 1c ¼ h 1mc ¼ 150 b 1c ¼ h 1mc ¼ 240 b 1c ¼ h 1mc ¼ 150 b 1c ¼ h 1mc ¼ 240 b 1c ¼ h 1mc ¼ 150 b 1c ¼ h 1mc ¼ 240 b 1c ¼ h 1mc ¼ 150 b 1c ¼ h 1mc ¼ 240
cs cm CaV
0.5 1.5 (50,75) 18.21 18.42 18.43 19.47 19.62 20.06 20.30 20.76
34.84 36.91 36.90 37.59 38.07 41.03 40.50 41.99
(80,120) 22.54 23.55 22.22 24.27 23.80 23.93 25.51 25.96
37.03 40.51 40.06 41.61 41.33 44.64 44.26 45.60
5 (t50,75) 20.00 20.46 20.19 22.01 21.82 21.38 22.15 22.37
37.62 41.20 40.76 42.30 42.53 45.41 47.06 48.78
(80,120) 23.16 23.89 23.09 24.94 25.34 24.79 26.35 26.51
42.36 47.16 45.63 46.92 47.36 49.92 51.47 53.15
1.2 1.5 (50,75) 21.12 21.84 21.98 24.07 23.22 24.06 24.00 24.66
42.46 46.54 46.97 49.67 48.78 53.04 51.34 54.46
(80,120) 25.88 26.09 25.89 27.54 26.72 27.40 27.47 28.14
48.26 51.13 52.54 54.01 54.62 58.83 58.22 62.03
5 (50,75) 24.59 25.43 25.47 26.97 26.98 26.05 27.11 28.13
48.53 51.82 54.45 56.80 54.56 59.54 57.44 61.16
(80,120) 28.63 28.85 29.02 29.72 29.91 30.30 30.15 30.70
56.01 60.75 60.19 63.12 61.83 64.69 62.85 66.07
Table III.
Model #1 vs Models
reduction (%) of
The average cost
sets of instances
#2 and #3 for all five
Supply-
87
demand hub
MRR run 2 2 2 24 = 27 = 128 times with different combinations of parameters, and as each
42,1 problem has run on all three models, there are 128 3 = 384 runs for each case. Altogether,
there are 384 5 = 1,920 runs for all five test problems. In Constraint (25), the relative
percentage deviation (RPD) is used to calculate the cost reduction percentage of Model #1
with respect to the other models, as follows:
The more the value of RPD, the better performance of Model #1 with respect to the other
methods. As the distance between manufacturers’ warehouses and SDHIC are negligible,
one can consider a1s ¼ d 1sm (V s, m) and a2s ¼ d 2sm (V s, m) for simplicity. The values of a1s
andd 1sm are set as a1s ¼ d 1sm ¼ 100; 160 and the values of a2s and d 2sm are set as
a2s ¼ d 2sm ¼ 112; 180, respectively. Similarly, the values of b uc and h umc are set to
b 1c ¼ h 1mc ¼ 150 and b 2c ¼ h 2mc ¼ 240 (V c, m), respectively.
70
Tge Average cost Reducon (%)
60
50
40
30
Figure 2.
The average cost 20
reduction (%) of
Model #1 vs Model 10
#2 and Model #1 vs
Model #3 0
(0.5, 1.5) (0.5, 5) (1.2, 1.5) (1.2, 5)
private warehouse for holding both material and products. Also, in Model #2, each Supply-
manufacturer should store its products in its own private warehouse; however, all of the demand hub
needed materials by all manufacturers are held in supply hub. But, in Model #1, all storage
spaces are merged in SDHIC in which both materials and products can be stored
simultaneously. In better words, land utilization is at maximum level in Model #1 compared
to Models #2 and #3, because by increasing the storing expense of materials and storing
expense of products, the total storing cost in Model #1 increases slower than in Models #2 89
and #3. Consequently, the higher the holding cost is, the more logistics costs could be saved.
According to the obtained experimental results and accomplished sensitivity analyses, it
could be concluded that via merging storing spaces, the application of SDHIC could bring
advantages in total logistics cost diminutions to the whole supply chain.
Downloaded by National University of Singapore At 21:49 30 January 2019 (PT)
1 10 15 30 40
2 10 20 35 50
3 15 25 40 50
4 15 30 45 60
5 20 30 55 60
6 25 40 60 75
7 30 45 60 80
8 35 50 70 80 Table IV.
9 40 50 75 100 The structure of test
10 45 50 80 100 instances
MRR 5.5 The industrial case study
42,1 In this study, the moquette and nonwoven fiber industry was recognized with a good
capability to apply the concept of supply-demand hub in a cluster, where in an industrial
group in this sector, a main manufacturing plant, a number of related plants and logistics
facilities are located in an industrial park. Therefore, a well-established and highly reputed
manufacturing group in this sector (referred to as ABCD in this paper) is considered in the
90 current study.
The ABCD produces several commodities, which are as follows:
Different types of moquette such as corduroy, velour and tufted for office and
household consumption.
Polyamide, polyester and polypropylene fibers.
Textile oils, polymer powders, industrial brushes, water-based homo- and co-
Downloaded by National University of Singapore At 21:49 30 January 2019 (PT)
polymer resins.
Different types of carpet comprising tufted, printed, brocaded, needle punched and
carved.
Geo-synthetics including geo-composite, geo-membrane and geo-textile.
In this study, an effort is made to fit the considered assumptions into the chosen case study.
A conjoint warehouse supervised by 3PL offers some SMEs (which are playing the role of
manufacturers) positioned inside an IC. All types of materials and products are warehoused
in the central storeroom. All logistics activities of the enterprises in such a supply chain are
outsourced to a 3PL containing holding inventories, transportation of different types of
materials and products, etc. The schematic of the adopted aerial picture/map of the facilities
in the concerned industry is depicted in Figure 3.
As it could be seen in Figure 5, there is a public hub (SDHIC) surrounded by ten SMEs as
manufacturers which are close to each other physically. The needed raw materials of these
SMEs are provided through eight suppliers, while this system has 15 customers, where the
final products should be distributed to them. The important data are presented in Table VI.
The value of fsm for each producer and their necessary number of materials are as follows (for
each producer): (2,2,1,3,1,3,1,2), (2,1,1,1,2,1,1,2), (1,2,1,2,1,3,3,3), (1,1,1,1,1,1,1,1), (2,2,1,2,2,2,1,3),
(1,2,1,2,2,3,1,4), (1,4,1,2,1,1,1,2), (2,2,1,3,1,3,1,2), (2,2,1,3,3,2,2,1) and (4,1,1,2,2,3,1,2).
As indicated in Table VI, there are three types of trucks for distributing different types of
materials with the volume of 60, 110 and 140 units (Cavu ) and three different types of trucks
91
Downloaded by National University of Singapore At 21:49 30 January 2019 (PT)
Figure 3.
The schematic of
adopted aerial image
of the facilities in the
concerned industry
0
for distributing all types of products with the volume of 70, 140 and 190 units (Cav u ). The
number of raw material s used to make one piece of different types of products (i.e. Types 1-
10) is displayed as fsm. As we have ten types of finished products, for each one, we present
the details of requirements. Also, utilization constants for storing at SDHIC for different
types of materials ( g H H
s ) and products ( g m ), besides utilization constant for storing in
different types of trucks to carry each unit of material ( g Ts ) and product ( g Tm ), are generated
according to the mentioned uniform distributions. aus is the utilization cost of truck Type u
to distribute between supplier s and SDHIC, which are a1s ¼ 50 , a2s ¼ 75 and a3s ¼ 90 for
different types of trucks. Similarly, b uc is the utilization cost of truck Type u to distribute
from SDHIC to customer c, which are b 1c ¼ 65, b 2c ¼ 90 and b 3c ¼ 110 for different types of
trucks. The capacity of SDHIC and also the number of time periods are regarded as 6,5000
and 5, respectively. Moreover, the received orders of ten types of products from 15
customers are shown in Table VI. For instance, Customer #2 needs 11 units of Product 1, 31
units of Product 2, 22 units of Product 3, 10 units of Product 4, 22 units of Product 5, 14 units
of Product 6, 17 units of Product 7, 29 units of Product 8, 33 units of Product 9 and 18 units of
Product 10. The order type of all customers is fixed during five studied time periods.
In better words, a challenge is created here by two issues: growing the constant expense
of using such a central hub by expanding the SDHIC from one side and from the other
side and the cost reduction of supply chain through enlarging the storage space, which
results in stopping of extra opportunity cost. Such a challenge reaches an equilibrium
Total cost vs Capacity Supply-
180,000
demand hub
170,000
160,000
Total cost (TC)
93
150,000
140,000
130,000 Figure 4.
Downloaded by National University of Singapore At 21:49 30 January 2019 (PT)
point when the SDHIC’s capacity is between 55,000 and 75,000, and then the whole
expense of SC is raised, owing to losing the above-mentioned attractiveness in capacity
growth.
When the SDHIC’s capacity is becoming greater than 75,000, the growth of fixed cost
is bigger than its decrease and consequently, the whole expenses of SC raises. The
optimum volume for SDHIC is about 44,000 according to the obtained objective function’s
value in the performed sensitivity analysis. It should also be emphasized that if the
SDHIC’s capacity is considered smaller than 17,000, the problem is infeasible and no
solution could be found. Likewise, if the SDHIC’s capacity violates 120,000, the total cost
of supply chain increases. Therefore, one can say that the optimum size of SDHIC is
44,000.
As one of the main contributions of this study is proposing supply-demand hub, which is
result of integrating supply hub and demand hub, considering the volume of empty area in
SDHIC with different sizes could be attractive. To do so, the portion of used space by raw
materials and finished products in terms of the ratio of “occupied space of SDHIC by raw
materials (S)” to “total occupied space by both raw materials and finished products (S þ M)”
in SDHIC is depicted in Figure 5 with different capacities. As it can be seen from Figure 5,
through growing the SDHIC’s capacity, the raw materials approximately achieve more
storage space, though this growth is nearly stopped after the materials occupy 15 per cent of
whole used area.
If Figures 4 and 5 is compared, on the whole, one can indicate that the maximum portion
of the materials occurs when the whole expense is almost minimized. Such a behavior
happens due to the following reasons:
Each piece of material occupies smaller holding area compared to each piece of
product, and as the unit storing expense of materials is lesser than products, the
whole expense reduces.
The worth of finished products is higher than raw materials, as the value of products is
equivalent to the worth of all used materials in addition to a given profit margin.
The shortage expense is costly as the backorder burdens high expenses to the entire
supply chain.
MRR So, it is obvious that the products are assigned more holding area compared to materials. In
42,1 this context, by growing the size of SDHIC, the system can allocate more holding area to all
types of materials, though this growth is continued up to definite bound. As it can be
observed from Figure 5, when the SDHIC’s capacity attains 55,000, this portion gets nearly
fixed.
The above-mentioned explanations are consistent with the real-world
94 environment. This issue reveals that the proposed attitude in this study is a practical
approach as the consequences approve the used assumptions and can well fulfill the
restrictions.
5.7 Effect of transaction cost on supply-demand hub in industrial clusters and supply hub
Downloaded by National University of Singapore At 21:49 30 January 2019 (PT)
To consider the disadvantage of using SDHIC and supply hub, additional cost terms
called “transaction cost of 3PL with manufacturers for raw material” and “transaction
cost of 3PL with manufacturers for finished products” are defined. These parts and
their calculations are studied in this section as the sensitivity analysis enables to
compare the models with and without this disadvantage. Also, it should be pointed out
that in Model #1, one should define these two new terms in the objective function as
“transaction cost of 3PL with manufacturers for raw materials (S8)” and “transaction
cost of 3PL with manufacturers for finished products (S9)”, but in Model #2, we need
only define the term “transaction cost of 3PL with manufacturers for raw materials
(S8),” as each manufacturer has its own private warehouse and 3PL only undertakes to
procure their raw materials. For the first model, the newly added sections to the
objective function are as follows:
S/(S + M) vs Capacity
0.1600
0.1400
0.1200
Percentage (%)
0.1000
Figure 5.
The ratio of 0.0800
“occupied space of 0.0600
SDHIC by raw
materials” to “total 0.0400
occupied space by 0.0200
raw material and
finished products” in 0.0000
15,000 35,000 55,000 75,000 95,000 115,000 135,000
the SDHIC of ABCD
Capacity
company
X
S8 ¼ Transaction cost of raw material ¼ j : ONsmt (1-8) Supply-
s;m;t demand hub
X 0 0
S9 ¼ Transaction cost of finished products ¼ j : ON mt (1-9)
m;t
X
In Constraint (1-8), the term ONsmt enumerates the total number of shipping materials 95
s;m;t X 0
from SDHIC to all producers in all periods and ON mt counts the total number of
m;t
collecting products from all producers in all periods. Given this, Constraint (1-8) calculates
Downloaded by National University of Singapore At 21:49 30 January 2019 (PT)
total transaction cost of shipping materials from SDHIC to all manufacturers, and Constraint
(1-9) computes total transaction cost of gathering products from all manufacturers (to
SDHIC). Also, parameters j and j ’ denote transaction cost coefficients of materials and
products for each time, respectively. 0
The new two variables ONsmt and ON mt are defined as follows:
ONsmt
0
1 if QSTsmt is positive; 0 otherwise.
ON mt 1 if QPmt is positive; 0 otherwise.
The following Constraint (26) should be added to Model #1:
For the second model, the new added section to the objective function is as follows:
X
S8 ¼ Transaction cost of raw material ¼ j : ONsmt (14-8)
s;m;t
Constraint (14–8) computes the total transaction cost of gathering finished products from all
manufacturers (to SDHIC). Also, the following constraint should be appended to the body of
Model #2:
0
QPmt # ON smt :M (27)
instance, the value 8.51 in column QPmt and second row signifies that the quantity of all
manufacturers in all periods are increased to 8.51 per0 cent in average. As it can be observed
from Table VII, the values of QPmt, QSTsmt and Inv mt are increased in average, while the
quantities of Invst and Ymt are decreased.
Some managerial insights could be driven here; the above-mentioned treatment
happens as the new added part of objective function in this subsection tries to minimize
60
Cost saving (%)
50
20
Figure 6. 40
10
The average cost
30
reduction of Model 5
#1 “without 20 0
transaction cost” 10
compared to “with 0 5 10 20
30
Cost saving (%)
25
Figure 7. 20
The average cost
reduction of Model 15
#2 “without
transaction cost” 10
0 5 10 20
compared to “with
Transacon cost coefficients of raw material
transaction cost”
Cost coefficient Decision variable
Supply-
j j’ QPmt QSTsmt Invst 0
Inv mt Ymt demand hub
0 0 0.00 0.00 0.00 0.00 0.00
5 8.51 3.22 4.83 7.06 4
10 13.34 4.37 4.37 9.76 6
20 13.34 5.75 5.52 9.76 6
5 0 4.37 4.14 6.21 5.75 4 97
5 5.75 5.75 5.06 8.53 6
10 7.36 8.51 6.67 14.29 8
20 12.04 8.51 6.67 14.29 8
10 0 4.83 9.89 8.05 9.37 6
5 6.82 10.58 6.67 15.83 10
6.67 12 Table VII.
Downloaded by National University of Singapore At 21:49 30 January 2019 (PT)
the transportation frequencies between SDHIC and manufacturers, and vice versa. The
customers’ demand should be satisfied finally and the system intrinsically tries to be
exposed with the least possible amount of shortage. Also, the reduction of
transportation frequencies between SDHIC and manufacturers causes an increment in
the amount of raw materials shipped to SMEs (QSTsmt) in each movement. On the other
hand, when the transportation frequencies between manufacturers and SDHIC
decrease, the amount of produced finished products each time are increased (QPmt),
owing to the same reason (satisfying the customers’ demand). Also, as the products are
allocated extra area of SDHIC than materials (owing to reasons mentioned in Section 5.6
and expensive cost of shortage because of failure to meet demand), more fraction of
SDHIC’ capacity is busied by finished products, and therefore, the inventory level of
0
finished products is increased (Inv mt ). The inverse reason happens about the inventory
level of raw material (Invst), which totally is decreased. Thus, the SMEs produce
products in less periods (Ymt), where in each production period, the produced amount of
products is increased.
SDHIC to customers comprise costs of different used trucks types and is brought as
S2 and S3 in the model.
Shortage costs of finished products as backordered is calculated as S7 in the model.
To illustrate the values of SDHIC, comprehensive calculations over a case study and generated
sets of instances are conducted including some sensitivity analysis. According to the obtained
results and extensive sensitivity analyses, it could be determined that via combining
consignments, use of SDHIC can yield advantages in whole logistics cost diminutions to the
whole supply chain. Comparing Models #1 and #2 and Models #1 and #3, it was shown that
by increasing the storing expenses of materials and storing expenses of products, the cost
saving percentage increases and Model #1 outperforms the other two models. However, the
average cost reduction compared to model #3 is more brilliant than the average cost reduction
of model #1 compared to model #2.
Also, the optimum capacity of SDHIC in the case study was determined through
conducting another sensitivity analysis. The results showed that the whole expenses are
diminished to a given extent at first and then increase because through increasing the
SDHIC’s capacity, the system avoids more backorders and allocates extra space to both
materials and products. The cost saving is stopped after a given interval as the
attractiveness in increasing the SDHIC’s capacity is missing and making the storage space
larger merely increases the fixed cost. Moreover, the maximum portion of occupied space of
SDHIC by the raw material is determined, and it happens when the whole expense is nearly
minimized. As each piece of material requires smaller holding area compared to each piece
of product and, on the other hand, the unit storing expense of materials is lesser than
products, the whole expenses reduce.
Generally speaking, according to the consequences, one can say that existing SDHIC in
ICs is beneficial (even by considering disadvantages of using SDHIC) for SMEs via
combining batches and integration of the holding space of stocks. Also, it should be
mentioned that the results of the case study are aligned with those obtained from random
sample examples, which confirm the trueness of used parameters and reveal the
applicability of using SDHIC in the real world.
Based on the obtained calculations and sensitivity analysis, several managerial
implications for the practitioners could be extracted as follows:
(1) merging upstream and downstream of the supply chain through centralizing on
storage of materials (supply echelon) and products (demand echelon) at the same
place simultaneously (the usage of SDHIC can yield advantages in whole logistics
cost reduction);
(2) showing the positive outcomes of collaborating 3PL with the SMEs as producers in Supply-
the considered supply chain; demand hub
(3) optimizing communications of industries involved in supply chain of an IC
containing some customers, suppliers and manufacturers;
(4) challenging a classic concept in which supply and demand echelons were being
separately planned regarding their inventory management and logistics activities
(in this research, we consolidate and manage them in a centralized way which 99
brings lots of benefits to the whole supply chain);
(5) rectifying the present important gaps regarding the previously published
research in this field by applying real-world expectations, such as capacity
constraint;
Downloaded by National University of Singapore At 21:49 30 January 2019 (PT)
(6) improving the performance of the proposed model compared to the classic model
by increasing the storing expenses of materials and products;
(7) presenting some practical findings based on the obtained results of sensitivity
analysis; as an instance, in the case study:
By increasing the SDHIC’s capacity, the total cost is diminished to a given
extend, and after that, the costs rise, as the attractiveness in increasing the
SDHIC’s capacity is missing and getting larger the storage space merely
increases the fixed cost
The optimum size of SDHIC is 44,000 for the studied case study based on
minimizing total costs of the supply chain
diminishing the attractiveness of using SDHIC to some extent by considering
“transaction cost of 3PL and manufacturers” as the disadvantage of SDHIC’s
application (however, SDHIC is still beneficial to the supply chain).
(8) A trend for future studies is to extend the proposed models to a multi-objective
area to consider some other important objectives. Another opportunity may be
taking the vehicle routing methods into consideration to optimize the sequence
of pickups and deliveries of materials and products. Also, a study on the
considered applied problem in a stochastic condition which is closer to a real-
world situation could be another stimulating stream of research in future. In a
stochastic condition, one practical assumption could be assuming shortage cost
as “service level” rather than “shortage cost.”
References
Banasik, A., Kanellopoulos, A., Claassen, G.D.H., Bloemhof-Ruwaard, J.M. and van der Vorst, J.G.A.J.
(2017), “Closing loops in agricultural supply chains using multi-objective optimization: a case
study of an industrial mushroom supply chain”, International Journal of Production Economics,
Vol. 183 (Part B), pp. 409-420, doi: http://dx.doi.org/10.1016/j.ijpe.2016.08.012
Barnes, E., Dai, J., Deng, S., Down, D., Goh, M., Lau, H.C. and Moosa, S. (2000), “On the strategy of
supply hubs for cost reduction and responsiveness”, White Paper on Electronics Supply Chain,
Georgia Institute of Technology and National University of Singapore.
Bastida-Ruiz, E., Franco-García, M.-L. and Kreiner, I. (2013), “Analysis of indicators to evaluate the
industrial parks contribution to sustainable development: Mexican case”, Management Research
Review, Vol. 36 No. 12, pp. 1272-1290, doi: 10.1108/MRR-06-2013-0145.
Bentley-Jennison, (2006), “Major Challenges Facing SMEs”.
MRR Capasso, M. and Morrison, A. (2013), “Innovation in industrial districts: evidence from Italy”,
Management Decision, Vol. 51 No. 6, pp. 1225-1249.
42,1
Chen, C. (2006), “The investigation for the establishment of science parks: the case of Taiwan”, Journal
of American Academy of Business, Vol. 8 No. 1, pp. 62-66.
DeWitt, T., Giunipero, L.C. and Melton, H.L. (2006), “Clusters and supply chain management: the
Amish experience”, International Journal of Physical Distribution & Logistics Management,
100 Vol. 36 No. 4, pp. 289-308.
Fattahi, M., Mahootchi, M., Govindan, K. and Husseini, S.M.M. (2015), “Dynamic supply chain network
design with capacity planning and multi-period pricing”, Transportation Research Part E:
Logistics and Transportation Review, Vol. 81, pp. 169-202, available at: www.sciencedirect.com/
science/article/pii/S1366554515001295
Giannoccaro, I. (2015), “Adaptive supply chains in industrial districts: a complexity science
Downloaded by National University of Singapore At 21:49 30 January 2019 (PT)
Shuang-yan, L., De-zhi, Z. and Fang-ping, J. (2013), “Base inventory cooperation strategy of multi-parts
with supply-Hub”, International Journal of Business and Management, Vol. 8 No. 20, pp. 96-104.
Tolossa, N.J., Beshah, B., Kitaw, D., Mangano, G. and De Marco, A. (2013), “A review on the integration
of supply chain management and industrial cluster”, International Journal of Marketing Studies,
Vol. 5 No. 6, pp. 164-174.
Yan, B., Cai, C.-Y. and Shi, P. (2017), “Multi-level two-way cross-chain inventory coordination
model that combines the same level with different levels in multi-supply chains”,
Simulation, Vol. 93 No. 11, pp. 973-983, doi: 10.1177/0037549717706169.
Zain, Z.M., Khalili, J. and Mokhtar, M. (2008), “Export problems among small and medium scale
industries in Klang Valley: a preliminary finding”, Gading Business and Management Journal,
Vol. 12 No. 1, pp. 23-39.
Further reading
Lenny Koh, S.C., Demirbag, M., Bayraktar, E., Tatoglu, E. and Zaim, S. (2007), “The impact of supply
chain management practices on performance of SMEs”, Industrial Management & Data
Systems, Vol. 107 No. 1, pp. 103-124.
Xiaoqiang, H. (2009), “Research of relevance of supply chain and industry cluster”, International
Journal of Marketing Studies, Vol. 1 No. 2, pp. 127-130.
Corresponding author
Vahid Kayvanfar can be contacted at: vkayvanf@gmail.com
For instructions on how to order reprints of this article, please visit our website:
www.emeraldgrouppublishing.com/licensing/reprints.htm
Or contact us for further details: permissions@emeraldinsight.com