Research paper

Evaluating logistics network configurations for

a global supply chain
A. Creazza and F. Dallari
Logistics Research Centre, Carlo Cattaneo University-LIUC, Castellanza, Italy, and
M. Melacini
Department of Management, Economics and Industrial Engineering, Politecnico di Milano, Milan, Italy

Abstract
Purpose – This paper aims to identify and assess different configurations to design logistics networks in global sourcing contexts. A framework to
support the choice of the most suitable logistics network is developed.
Design/methodology/approach – Starting from an overview of previous research on global sourcing and global logistics network design, followed
by a series of interviews with international freight forwarders and logistics providers, five main logistics network configurations are identified and the
paper proposes a framework to evaluate their cost-effectiveness, deriving the overall logistics cost by means of simulation.
Findings – On the basis of the analysis of different scenarios the paper develops a taxonomy for selecting the most suitable logistics network
configuration, with respect to some key logistics factors and purchasing strategies.
Research limitations/implications – The analysis has been focused on ocean container shipping, the primary transportation mode for world trade.
Practical implications – In a context characterised by the widespread adoption of direct shipment with full container load (FCL) ocean shipping (with
implications on inventory levels), the provided taxonomy can represent a useful tool to support companies in choosing the most suitable combination of
configurations for setting their global logistics network.
Originality/value – The impact of globalisation on logistics network configuration has received little attention from supply chain researchers to date.
The originality of the present paper is twofold. First, a framework to assess the overall logistics costs is developed. Second, the paper presents an
original taxonomy for the selection of the most suitable logistics network.

Keywords World economy, Supply chain management, Transportation, Distribution management

Paper type Research paper

1. Introduction components and subassemblies from international sources

for use in fabrication, assembly or for resale, regardless of
Technical progress, liberalisation of markets and the diffusion whether the import source is internal or external to the
of global production systems are creating new opportunities company” (Kotabe and Omura, 1989). In fact global sourcing
and challenges for industrialised countries as well as for the covers a wider meaning: “integration and coordination of
global market economy. One of the most significant procurement requirements across worldwide business units,
consequence of globalisation is represented by supply chains looking at common items, processes, technologies and
progressively becoming more international and complex. This suppliers” (Monczka and Trent, 2003).
phenomenon is gradually increasing: in year 2000 firms The above definition highlights the ability to schedule,
sourced between 21 and 30 percent of their total annual coordinate and synchronise the variety of the goods exchanges
spending on a worldwide basis. In 2005, the total non- and information flows from source to destination as a key
domestic spending increased up to values between 31 and 40 factor. Transport and logistics, according to this new point-of-
percent, compared to year 2000 (Monczka and Trent, 2005). view, turn from costs to keep under control, into strategic
This procurement strategy, in its more advanced form, is levers, being the basis for the supply chain redesign. Global
known as global sourcing. The term “global sourcing” gives a sourcing structures and processes, including logistics
far different meaning than international purchasing, which networks and activities, become increasingly important to
can be defined as “the acquisition of raw materials, the effectiveness of global sourcing (Peterson et al., 2000).
Moreover, global sourcing creates the need to efficiently plan
The current issue and full text archive of this journal is available at intercontinental transportation (Pollit, 1998; Bhatnagar and
www.emeraldinsight.com/1359-8546.htm Viswanathan, 2000; Mattsson, 2003). Geographical distances
in global contexts increase not only transportation costs, but
also make it more complicated to solve the trade-off between
Supply Chain Management: An International Journal inventory and physical distribution costs (Kruger, 2002).
15/2 (2010) 154– 164
q Emerald Group Publishing Limited [ISSN 1359-8546]
Low-cost off-shore sourcing strategies can end up as high-cost
[DOI 10.1108/13598541011028750] supply chain outcome (Christopher et al., 2006).

154
Evaluating logistics network configurations for a global supply chain Supply Chain Management: An International Journal
A. Creazza, F. Dallari and M. Melacini Volume 15 · Number 2 · 2010 · 154 –164

Consequently, in order to obtain the best supply chain The second research stream deals with the management of
outcome, logistics patterns (e.g. the choice of transport mode global supply chains, that are more complex than local ones.
and logistics networks configuration) should be changed as Different local cultures along with diverse languages and
well. practices decrease the effectiveness of such processes like
Since previous literature has given little attention to the demand forecasting and material/production planning. Some
effects of globalisation on logistics networks, we present a authors (Monczka and Trent, 2003; 2005; Li et al., 2006)
framework in order to support supply chain managers in the studied the need for integrating facilities and for sharing
design of global logistics networks. We study the shipment of information in global supply chains and the consequent
finished goods from the point of origin (plants, generally benefits. To this aim, in a recent study, Zeng (2003) proposed
suppliers’ facilities) to the point of destination (companies’ a material and information flow model, while other studies
regional warehouses), focusing on the physical logistics (Bozarth et al., 1998; Choy and Lee, 2003) focus on demand
network. In particular, this paper addresses an increasingly planning coordination with suppliers. Dealing with tactical
relevant issue for European companies, i.e. the sourcing from production-distribution allocation problems, various
the Far East countries. Starting from a literature review mathematical programming models have been introduced
concerning global supply chain design (section 2), and by (Goetschalckx et al., 2002). The most advanced ones take
means of interviews with international freight forwarders and into account price/exchange as well as risk effect (Cohen and
logistics service providers, we identify and present five Huchzermeier, 1996).
configurations to design a logistics network on a global scale With respect to the third group of researches, the focus of
(section 3). Subsequently we propose a framework to assess the present study, the configuration of global logistics
the effectiveness of the proposed configurations (section 4). networks implies the choice of the most suitable
Within this framework, after having defined the parameters international transport mode (mainly ocean container
for a base case (section 4.1) and described the implemented shipping or airfreight), the design of infrastructures for
simulation model, we assess the overall logistics cost with freight consolidation and for serving end markets and the
respect to the presented base case, for each configuration definition of the number of echelons composing the logistics
(section 4.2). We then perform a sensitivity analysis to network (Kruger, 2002). Prior literature analyses global
evaluate the influence of some key logistics parameters on the logistics network design from two main perspectives: from
suitability of the proposed configurations (section 4.3). On global carriers’ point-of-view (international forwarders or
the basis of the results of the sensitivity analysis, we propose a logistics service providers), or from manufacturers’ side. In
taxonomy to select the most suitable logistics network (section the first case the studies deal with the use of transport
5). Concluding remarks and further research areas are infrastructures or logistics models like hub and spoke (Zapfel
presented in the final section. and Wasner, 2002; Lee and Yang, 2003; Imai et al., 2006).
The attention is often placed on shipment consolidation
strategies (Tyan et al., 2003), as well as internationalisation
2. Literature review and globalisation policies of freight forwarding companies and
logistics service providers (Lemoine and Dagnaes, 2003).
The growth of globalisation and the consequent challenges for Papers about manufacturers’ point-of-view consider
management have motivated both practitioners and transportation systems as an origin-destination process
academics’ interest in global supply chain management characterised by a transport unit cost but paying little
(Gargeya and Meixell, 2005). Recent studies, which attention to the structure of the transport service and to the
considered the extent of this phenomenon, concentrate on relative implications on the entire supply chain (e.g. Cohen
some relevant aspects/decisions that can be subdivided into and Lee, 1989). Various relevant studies for the redesign of
three main groups: strategic alignment of the supply chain, the logistics network on a global scale have been developed,
coordination of the players operating in the global supply even if they do not allow significant generalisations of the
chain and design of the global logistics network. results (Arntzen et al., 1995; Bhatnagar and Viswanathan,
Studies on global manufacturing strategies (e.g. facility 2000; Chung et al., 2004). Only in the most recent studies, a
location and vertical integration) belong to the first research deeper analysis on the possible configurations for logistics
stream (MacCormack et al., 1994; Chopra and Meindl, 2004; networks is provided. For instance, Zeng (2003) considered
Kotabe and Murray, 1989; Brown et al., 2007; Liu et al., three global transportation service categories (airfreight, full
2008). Depending on the level of vertical integration, there container load shipping, less than container load shipping),
can be different supply chain configurations (Hong and including inventory costs as part of the decision making
Holweg, 2002). Production facility location and the distance problem. Lovell et al. (2005) proposed a taxonomy for global
from the final markets involve a different degree of network design mainly based on product value density and
responsiveness and agility of the supply chain towards the throughput volume. Cheong et al. (2007) evaluate a
fluctuations of the market demand (Miemczyk and Howard, configuration, which implies the adoption of consolidation
2008). Some current researches try to evaluate the benefits of hub facilities in Asian sourcing countries (to collect shipments
a supply chain strategy based on the concept of “lean” coming from several suppliers) and the delivery of the
(Womack and Jones, 1996) and “agile” (Christopher, 2000; consolidated shipments to the respective worldwide
Christopher and Towill, 2000). With respect to the choice manufacturing plants. This study includes shipping
between the “lean” and “agile” approaches, there should be frequency as a key element to be taken into account in
no real contrast between them. On the contrary companies ocean shipping planning.
should integrate the appropriate and coherent aspects of these In addition, the literature review showed a number of
paradigms in their specific supply chain strategy (Christopher specific case studies about logistics networks configuration
et al., 2006). characterised by a strong emphasis on operational research

155
Evaluating logistics network configurations for a global supply chain Supply Chain Management: An International Journal
A. Creazza, F. Dallari and M. Melacini Volume 15 · Number 2 · 2010 · 154 –164

tools and approaches. In the meantime it also highlighted a goods are shipped by means of road haulage to the
lack in the definition of exhaustive frameworks for choosing company’s CH where goods are temporary stored. After
the potential options for designing a global logistics network. consolidation, containers are shipped by means of FCL
from a unique LP to a final destination (RWj).
3. Global logistics network configurations 4 Configuration 4 – One echelon logistics network with
central warehouse (CW) in Europe: all the suppliers ship
In the present study we consider the product sourcing process their goods by means of FCL to a unique UP, the closest
performed by a manufacturing company or by a retailer (from to the CW. Here shipments are deconsolidated and non-
suppliers/plants located in a specific geographical area (e.g. containerised goods are hauled to one or more regional
south eastern Asia)), selling goods all over Europe throughout warehouses (RWj).
a network of regional warehouses (i.e. warehouses 5 Configuration 5 – Two echelons logistics network with
characterised by a delivery area including one or more both CH and CW: non containerised goods are shipped
European countries (Lovell et al., 2005)). We consider ocean by means of road haulage to the company’s CH in Far
container shipping, the primary transportation mode for East. After consolidation, containers are shipped from the
world trade (because of its low freight rates, despite long LP by means of FCL to a unique UP the closest to the
transit time). The sourcing activities are organised according CW. Here shipments are deconsolidated and non-
to Ex Works Incoterm, which allows the manufacturing containerised goods are hauled to one or more regional
company (or retailer) to have full visibility and control of the
warehouses (RWj).
entire transportation process.
After having performed an in-depth analysis of the Besides the above-described main configurations, we did not
literature, a complementary series of interviews with take into account three alternatives entailing the use of LCL
international freight forwarders and logistics providers along with one or two echelon logistics network. This choice
followed (i.e. Maersk, Kuehne & Nagel, DHL), in order to was suggested by freight forwarders, which underline how the
validate the logistics network options and to derive real costs. use of these configurations results in a redundancy of logistics
Ocean container shipping can be performed in two different activities and, as a consequence, in inefficiencies in terms of
ways (Zeng, 2003; Dallari et al., 2006), namely shipment with costs and delivery lead times.
full container load (FCL) and shipment with groupage The five considered configurations differ one from another
container (less than container load (LCL)). Therefore it is in terms of cost structure, supply lead times, risk of delay and
possible to optimise the ocean container shipping through a operational complexity (i.e. the intensity of the effort
consolidation hub, i.e. a node of the logistics network where necessary for logistics network planning and control).
shipments from several suppliers are collected and As suggested by Zeng and Rossetti (2003), the differential
consolidated similar to a cross-docking facility and they are logistics cost categories connected to the 5 configurations are
shipped to the final destinations (Cheong et al., 2007). represented by transportation cost, handling cost, inventory
Furthermore, three main different options can be considered carrying cost and order processing cost (see Table I).
to configure a logistics network: direct shipment, one echelon Transportation’s cost category can be subdivided in two
network and two echelon logistics network (Simchi-Levi et al., different cost items: LCL rate and FCL rate. LCL rate,
2003). In the one echelon the consolidation hub can be connected to configuration 2 only, includes road haulage
located close to the suppliers (i.e. upstream the ocean (initial/final), ocean container shipping and handling costs at
shipping leg) or close to the market (i.e. downstream the forwarders’ facilities. As far as the other configurations are
ocean shipping leg). Thus, four different options to configure regarded, FCL rate is composed by an ocean leg (ocean
a logistics network can be considered: direct shipment, one
container shipping) common to all configurations and a road
echelon with consolidation upstream, one echelon with
leg, function of the distance to be covered in each
consolidation downstream and two echelon logistics
configuration as highlighted in Figure 1.
network. Combining these options with the aforementioned
The configurations differ also in terms of handling and
shipping ways, we consider five logistics network
inventory carrying costs: in configuration 1 and 2 the
configurations (Figure 1):
handling activity is performed in RWs only, while in the
1 Configuration 1 – Direct shipment with full container load
(FCL): a single supplier (Si) based in Far East ships a full other options goods are handled also in the additional logistics
container load to a regional warehouse (RWj) located in nodes (CH and CW, where a certain cycle stock is also
Europe, by means of a first container haulage to a loading present). A key decision in logistics network configuration is
port (LPk), an ocean shipping leg and a final container represented by safety stock allocation. We hypothesise to hold
haulage from the unloading port (UPh) to a final safety stocks in RWs only, thus considering CW or CH as
destination (RWj). transit points, where no safety stock is held (Cheong et al.,
2 Configuration 2 – Direct shipment with groupage 2007) and thus assuming the same service level for all the
container (less than container load (LCL)): non considered configurations. Consequently, the stock-out cost is
containerised goods are shipped by means of road not differential for the various configurations. The different
haulage to freight forwarder’s facility located near the amount of safety stocks in the RWs, related to demand and
loading port (LPk) for container consolidation; after the lead time distributions (Ballou, 2001), depends only on the
ocean shipping leg, the container is deconsolidated at the mean value and variability of the lead time connected to each
unloading port (UPh) and non containerised goods are configuration. In conclusion, it should be remarked that order
hauled to one or more regional warehouses (RWj). processing cost is only a function of each RW’s reorder
3 Configuration 3 – One echelon logistics network with frequency and it is not dependent on the considered
consolidation hub (CH) in Far East: non containerised configuration.

156
Evaluating logistics network configurations for a global supply chain Supply Chain Management: An International Journal
A. Creazza, F. Dallari and M. Melacini Volume 15 · Number 2 · 2010 · 154 –164

Figure 1 Global logistics network configurations between n suppliers and m regional warehouses

Table I Logistics cost items

Logistics cost
category Cost item Brief description
Transportation LCL rate Door-to-door groupage cost, including road haulage (pick-up and delivery), port-to-port ocean freight,
terminal handling charge, agency fee, customs clearance, bunker adjustment fee (BAF), currency
adjustment factor (CAF), materials handling at freight forwarders’ facilities, transportation planning
FCL rate Ocean leg: port-to-port ocean freight, terminal handling charge, agency fee, customs clearance, bunker
adjustment fee (BAF), currency adjustment factor (CAF)
Road leg: road haulage cost (pick up and delivery) for a given distance to be covered
Handling Material-handling cost Cost of workforce and equipment used to handle goods in the warehouses (i.e. central warehouse,
consolidation hub, regional warehouses)
Inventory carrying Safety stock cost Safety stock holding costs (capital cost, risk and damage, storage cost)
Cycle stock cost Cycle stock holding costs (capital cost, risk and damage, storage cost)
In-transit inventory Holding costs (capital cost only) during the transit time
cost
Order processing Administration cost Cost related to order processing, stock replenishment, shipment planning and expediting

157
Evaluating logistics network configurations for a global supply chain Supply Chain Management: An International Journal
A. Creazza, F. Dallari and M. Melacini Volume 15 · Number 2 · 2010 · 154 –164

Regarding supply lead time (defined as the time elapsing from Europe, demanding the same product range and characterised
the issue of a replenishment order by a RW and the receiving by identical sales volume and service level (the probability of
of the goods ordered from a supplier) configuration 1 ensures not incurring in a stockout is assumed equal to 95 per cent).
the lowest mean lead time. When the number of transit nodes The weekly demand for each item is represented by a normal
increases, mean lead times become longer for configurations 3 distribution with a mean value (D) equal for all the items of
and 4, up to configurations 2 and 5, characterised by the the same class. Demand variability is described by the
longest mean lead times. Similarly, the risk of delay increases coefficient of variation, i.e. the ratio between the standard
coherently with the number of transit nodes, while the deviation of the weekly demand and the relative mean value
operational complexity for the company progressively (s/D). The items are characterised by an average unit value
increases from configuration 1 to configuration 5. equal to 0.03 m3 and a product value density (PVD) equal to
Each company should analyse and understand which 450 e/m3. The overall annual demand, resulting from the sum
configuration could be the most suitable for its own of the average demand of each item, is equal to approximately
business environment, as well ensuring the required service 200,000 m3.
level. Nowadays most companies adopt configuration 1, Freight rates and lead time parameters differ between the
mainly due to its low operational complexity and five configurations. We considered that ocean container
transportation cost although it does not always represent the shipping is performed according to the hub and spoke model
most cost-effective configuration (Zeng, 2003)., In the next (i.e. feeder services connecting regional ports with
section we present a framework for assessing the overall transhipment hubs). The container shipping rates for 20
logistics costs in order to fully understand the suitability of the and 40 feet containers and supply lead times are reported in
proposed configurations. Tables III and IV. Table V reports the average distances for
the different road transport legs, and Table VI summarises the
4. A framework to evaluate the proposed road transport costs with respect to different distance ranges.
It must be underlined that volatility has to be considered in
configurations the long run, due to the market complexity and to the
The cost-effectiveness of the proposed configurations has variability in oil price and exchange rates (Cohen and
been evaluated according to a framework derived from Zeng Huchzermeier, 1996).
and Rossetti (2003) and comprising three steps: As far as the other cost items are concerned, the average
1 Defining the input parameters required to calculate the handling cost at the European warehouses (RWs and CW) is
overall logistics cost (OLC). supposed to be equal to 0.2 e/unit. The inventory holding cost
2 Simulating the physical logistics flows for a significant (including cost of capital, space, insurance and depreciation)
time horizon (e.g. one year) and calculating the related is assumed equal to 25 per cent of inventory value on a yearly
OLC. basis. However, the costs are different if considered in
3 Performing a sensitivity analysis to assess the impact of sourcing countries: the handling cost in Far East countries is
key parameters on OLC. assumed 40 per cent lower than European handling cost
(because of the lower cost of the local workforce), whereas the
The analysis has been performed with respect to a real life storage cost (increasing for areas nearby the ports) is
case (defined as “base case”), derived from a leading toys considered only 10 per cent lower than the European one.
manufacturing company mainly operating in Italy and
Order processing cost has been set equal to 100 e/order, while
Europe, whose identity has not been reported here for
LCL rate is equal to 75 e/m3 for road legs within 300 km from
confidentially reasons. The analysis of the company’s business
the ports.
environment and industry sector allowed obtaining the input
parameters connected to the demand characteristics and to 4.2 Simulation of the logistics flows and calculation of
the logistics network details. Moreover, interviews with global
the related logistics cost
freight forwarders permitted to obtain freight rates and supply
In the second step of the framework, the flow of goods
lead times.
between suppliers and RWs has been simulated. Considering
a time frame of one year, we generated the weekly demand
4.1 Definition of the input parameters
according to Monte Carlo method. Each RW operates with an
The main parameters common for all the five configurations
(R, S) inventory control policy where R indicates the review
are:
interval and S indicates the order-up-to level (Nahmias,
1 Demand characteristics: number of items, average item
1997). Order-up-to level is updated at the beginning of each
volume, product value density (ratio between item value
replenishment to reflect the changes in demand patterns. The
and unit volume), and weekly demand.
quantity ordered by each RW depends on the number of items
2 Suppliers’ characteristics: number, location, number of
sourced from a supplier, the demand of each item and the
supplied items.
inventory review interval, which corresponds to the shipping
3 Regional Warehouses’ characteristics: number, product
frequency.
range and service level.
The logistics complexity (i.e. the difficulty in planning and
In the base case we consider 20 suppliers located in Far East, organising full container loads on the considered linkage)
globally providing 200 items grouped in three main classes: increases with a progressively higher number of linkages in the
fast moving (A class), medium moving (B class) and slow transportation process (Craighead et al., 2007). For instance
moving (C class) (see Table II). The suppliers are subdivided in configuration 1, with 20 suppliers and 12 RWs, we obtain
into two groups: big suppliers (5 out of 20), characterised by 240 linkages. Considering ceteris paribus configuration 4,
high volumes of shipped goods and medium-small suppliers with the central warehouse located in Europe, we will obtain
(15 out of 20). Then, we considered 12 RWs, located in only 20 linkages to connect the suppliers with the CW in

158
Evaluating logistics network configurations for a global supply chain Supply Chain Management: An International Journal
A. Creazza, F. Dallari and M. Melacini Volume 15 · Number 2 · 2010 · 154 –164

Table II Demand characteristics

Class A items Class B items Class C items
Number of items sourced from big suppliers 30 15 5
Number of items sourced from medium-small suppliers 15 45 90
Number of items 45 60 95
Mean value of weekly demand in each RW (units/week) 600 300 150
Coefficient of variation of weekly demand 0.25 0.4 0.8

Table III FCL rate (ocean leg) the OLC. The resulting OLC depends, in each simulation
run, on the values of the items’ weekly demand. In order to
Type of container Rate [e/container] reduce the impact of random variations, the same random
20’ 1,500 numbers have been used to simulate all the five configurations
40’ 2,500 (thus generating the same item demand for all of the
configurations). In addition to this variance reduction
technique, we performed 50 simulation runs to reach the
stability of the system (Law and Kelton, 1997).
Table IV Supply lead times related to the logistics network Differently from other studies (e.g. Zeng and Rossetti,
configurations 2003), we have not assumed the shipping frequency as an
Logistics network configurations input data. On the contrary, it has been considered as a
1 2 3 4 5 variable because of the high dependency of the overall results
on this element (Cheong et al., 2007). For example,
Mean value (weeks) 4.9 5.9 5.6 5.4 6.2 considering configuration 1, ceteris paribus, we expect
Standard deviation (weeks) 0.50 0.80 0.54 0.54 0.76 the transport cost to progressively decrease with the
contemporaneous reduction of the shipping frequency,
becoming almost constant. On the other hand, inventory
Table V Distances for the different road transport legs (average values) holding cost is expected to rise when shipping frequency
decreases, whereas order processing cost falls and handling
Road transport legs cost does not significantly vary. Therefore, for each single
d1 d2 d’1 d’’1 d *1 d *2 d’2 d’’2 configuration the optimal shipping frequency (or review
Distance (km) 200 200 400 50 300 300 200 500 interval), able to minimise the OLC, has been evaluated. For
instance, with respect to configuration 1, as reported in Table
VII, results highlight that the efforts to achieve higher
Europe through a unique UP and 12 linkages from the CW to economies of scale for transportation inevitably lead to less
the RWs. In this case there will be a significant reduction in frequent shipments, which result in higher inventory levels. In
the number of linkages (from 240 to 32), with a potential the base case, the optimal solution can be obtained by setting
a 6 week shipping frequency.
higher utilisation rate of container capacity, even when
Iterating the computational procedure for all the other
considering an identical shipping frequency.
configurations, it is possible to obtain the optimal shipping
Based on the amount to be shipped from source to
frequency with the respective minimum OLC. Table VIII
destination, is obtained the number of containers shipped per
shows the optimal resulting OLC for all the configurations
year. This value is a function of the container loading capacity highlighting configuration 3 as the most suitable option for
(volume or weight if necessary) and of the logistics flow per the base case.
shipment. The choice of the container type mix (20 or 40
feet) for each shipment is driven by the minimisation of ocean 4.3 Sensitivity analysis
shipping costs. Taking into account the resulting logistics The variation of some input parameters can significantly
flows (aggregate and for single item) the average stock present affect the OLC and consequently the ultimate decision
in the RWs, CH and CW can be obtained, along with the in- regarding the selection of the logistics network configuration.
transit inventory. In the third step of the framework we try to generalise the
On the basis of the assumed unit costs (section 4.1) and of obtained results, performing a sensitivity analysis aimed at
the results of the simulation (in terms of shipped volumes to studying the impact on the OLC of the variation of five key
the different logistics nodes, the number and type of parameters, identified in literature as relevant elements for
shipments and the consequent inventory levels), we derived designing global logistics networks (Kruger, 2002; Vereecke

Table VI FCL rate (road leg) [e/container]

Distance ranges (km)
Up to 50 51-250 251-499 500-750 751-999 1,000-1,500
Transport cost in Far East 250 420 600 840 1,200 1,500
Transport cost in Europe 200 350 500 700 1,000 1,300

159
Evaluating logistics network configurations for a global supply chain Supply Chain Management: An International Journal
A. Creazza, F. Dallari and M. Melacini Volume 15 · Number 2 · 2010 · 154 –164

Table VII Logistics cost as a function of the annual shipping frequency (direct shipment with FCL configuration, with reference to the base case)
(average values)
Annual shipping frequency (weeks)
2 4 6 8
Transportation cost (thousands e/year) 20,187 17,303 14,324 13,860
Order processing cost (thousands e/year) 624 312 208 156
Inventory holding cost (thousands e/year) 2,683 3,480 4,275 5,070
Handling cost (thousands e/year) 2,462 2,462 2,462 2,462
Overall logistics cost (thousands e/year) 25,956 23,557 21,269 21,548
Cost-to-value ratio * (%) 20.17 18.3 16.5 16.8
Note: *Defined as the ratio between overall logistics costs and the value of annual demand

Table VIII Optimal logistics costs for the considered logistics configurations in the base case (average values)
Logistics configurations
1 2 3 4 5
Overall logistics cost (thousands e/year) 21,269 20,685 20,115 20,423 23,682
Cost-to-value ratio (%) 16.5 16.1 15.8 15.9 18.4

and Dierdonck, 2002; Zeng and Rossetti, 2003; Lovell et al., Consequently configuration 3 becomes even less cost-
2005): effective, proving the suitability of configuration 4.
1 Overall annual demand. The fourth test examined the effect of an increase in the
2 Annual demand between each supplier and each RW. average Product Value Density (PVD) on the choice of the
3 Suppliers’ dispersion, defined as the average distance best logistics configuration (Table IX – part D). When PVD
between CH and suppliers. becomes lower, configuration 1 can result as the most cost-
4 Product value density. effective option, due to the low impact on inventory levels
5 Differential labour cost between sourcing and destination corresponding to low shipping frequencies. The higher the
countries. average PVD, the more suitable are the solutions that are able
We ran five test problems for each parameter, mounting up to to keep transport and storage costs low with high shipping
a total of 25 cases. In the first test, we varied the annual frequencies. In the examined case configuration 3 is the most
suitable solution. Potentially, in the considered context,
demand (the total amount of goods sourced from suppliers,
configuration 2 could also represent a significantly cost-
expressed in m3) while keeping constant the other inputs
effective configuration when inventory holding cost becomes
(Table IX – part A). The analysis showed that, with
relevant.
increasing overall demand, companies can find it more cost-
Table IX (part E) shows the effect of variations in labour
effective to manage flow consolidation activities in house by
cost expressed as percentage difference between handling
means of a CH. If overall demand further increase, direct
costs in sourcing and destination countries. A reduction of the
shipment with FCL becomes the most cost-effective option.
percentage difference makes configuration 4 virtually more
At the same time, configurations 3, 4 and 5 become
cost-effective than configuration 3.
progressively less cost-effective, because of the consequent
In conclusion, it is worth to underline that configuration 5
additional handling costs.
is dominated by all the other options. In fact, it does not imply
As shown in Table IX – part B, keeping constant the overall
a different inventory re-allocation within the logistics network
annual demand, configuration 1 returns to be cost-effective
and consequently the additional handling cost due to the
when the demand between a supplier and a RW in each
presence of more transit nodes is not justified.
linkage increases (e.g. by decreasing the number of suppliers)
due to the opportunity to achieve higher transportation
economies of scale by means of frequent shipments without 5. A taxonomy for the selection of the global
increasing inventory levels. Vice versa, when the demand
between a supplier and a RW in each linkage decreases, the
logistics network configurations
logistics complexity grows and configuration 3 potentially A number of classification schemes have been proposed in the
becomes suitable again. If the demand between a supplier and literature to guide the choice of supply chain strategy (Fisher,
a RW in each linkage significantly reduces (e.g. the number of 1997; Christopher et al., 2006) or network design (Lovell
suppliers significantly increases), configuration 2 turns out to et al., 2005). Moving from the classification proposed by
be the most suitable option. Lovell et al. (2005) and from the outcomes of the sensitivity
Our third test, shown in Table IX (part C), demonstrated analysis, we can draw some general considerations on the
the effect of suppliers’ dispersion on OLC without varying the choice of the most cost-effective configuration in global
other input parameters. A growth in the average distance sourcing contexts. With respect to standard goods (i.e. items
between CH and suppliers will increase transportation costs. stable in demand with no, or limited customisation) we

160
Evaluating logistics network configurations for a global supply chain Supply Chain Management: An International Journal
A. Creazza, F. Dallari and M. Melacini Volume 15 · Number 2 · 2010 · 154 –164

Table IX Sensitivity analysis

Logistics network configurations
Parameter Test problem 1 2 3 4 5
3
(A) Overall annual demand 50,000 [m ] £
100,000 [m3] £
200,000 [m3] £
300,000 [m3] £
400,000 [m3] £
(B) Annual average demand between a 200 [m3]
supplier and a RW £
400 [m3] £
800 [m3] £
1,600 [m3] £
3,200 [m3] £
(C) Suppliers’ dispersion (average distance 200 [km]
between CH and suppliers) £
400 [km] £
800 [km] £
1,000 [km] £
1,500 [km] £
(D) Product value density 300 [e/m3] £
450 [e/m3] £
600 [e/m3] £
1,200 [e/m3] £
2,400 [e/m3] £
(E) Differential labour cost (percentage 10%
difference between handling costs in
sourcing and destination countries) £
20% £
30% £
40% £
50% £
Notes: “ £ ” represents the most suitable logistics configuration in each scenario

suggest a two-dimensional classification, whose dimensions end market (configuration 4). If the sourcing volumes in each
and binary gradations are: linkage are so relevant direct shipment from each supplier
1 Overall annual demand (low or high), defined as the with FCL could represent the most suitable option
sourced volume by RWs from a specific geographical area (configuration 1).
(e.g. south eastern Asia). After checking the most suitable logistics network
2 Demand between suppliers and regional warehouses in configuration according to the proposed taxonomy, a
each linkage (low or high), defined as the part of each company can deepen the analysis by examining the different
RW’s yearly demand sourced from a single supplier. options for the chosen configuration (e.g. the number of hubs,
Figure 2 shows the resulting taxonomy with the proposed with respect to the number and the location of suppliers) by
configurations. In a business environment characterised by a means of tools such as mathematical programming models
limited relevance of purchases from international suppliers in (see for more details Goetschalckx et al., 2002). The
terms of sourced volume, it is suitable to rely on LCL services taxonomy can be customised in compliance with the specific
(configuration 2). On the contrary with an increase of the needs of each company, characterising the different solutions
overall annual demand and of the sourcing volumes, according to the various supplier and RW sizes. For instance,
companies can optimise their international transport flows a company characterised by high overall annual demand and
by means of an owned consolidation facility (configuration 3 by relevant demand on a subset of the linkages between
or 4). The choice concerning the adoption of the CH rather suppliers and RWs could adopt a hybrid solution (e.g.
than the CW as a consolidation facility is mainly driven by implementing configuration 1 for high demand linkages and
suppliers’ dispersion and by the percentage difference configuration 3 for the remainders).
between handling cost in sourcing and in destination With respect to the proposed taxonomy, the definition of
countries. For low suppliers’ dispersion values and high the threshold for separating “low” and “high” values
percentage handling cost difference, it is appropriate to adopt represents a critical issue specific to the business contexts.
a CH near suppliers’ locations (configuration 3). Otherwise, On the basis of the sensitivity analysis results, considering a
the most suitable configuration is to adopt a CW close to the PVD lower than 600 e/m3, the resulting threshold value for

161
Evaluating logistics network configurations for a global supply chain Supply Chain Management: An International Journal
A. Creazza, F. Dallari and M. Melacini Volume 15 · Number 2 · 2010 · 154 –164

Figure 2 A taxonomy for selecting the logistics network configurations

the overall annual demand can be set approximately equal to (direct shipment with FCL), a widespread solution, does not
75,000 m3 (i.e. 2,500 TEUs per year), while for the other always represent the most cost-effective one.
dimension it can be considered about 1,600 m3 (i.e. On the basis of the analysis of different scenarios we
approximately 50 TEUs per year). proposed a taxonomy to support the choice of logistics
It should be noticed that the proposed classification network configurations considering some key elements such
dimensions are tightly related to the specific company’s as: the overall demand, the demand between suppliers and
purchasing strategy in terms of management of the global RWs, the suppliers’ geographical dispersion, the product
supply base (Gelderman and Semeijn, 2006). In fact, the value density and the differential labour cost. The proposed
greater the relevance of a certain geographical area in a taxonomy, even if it has been developed considering a specific
particular business environment, the bigger the annual context (i.e. international shipments from Far East to
demand arising from it. In addition, the volume of goods Europe), can be profitably applied to different countries of
sourced from a specific supplier depends on the company’s origin and destination, reflecting their diversity. In fact, the
purchasing policy: the search for the cheapest supplier for specific features of each potential country to be taken into
every single product type will result in a consequent higher account (e.g. number and location of ports, transportation
fragmentation of the overall sourcing volumes and the search network development stage) mainly impact on the definition
for more stable relationships with suppliers will lead to a of the threshold values considered for the choice among the
resulting decrease in the number of suppliers involved in the different logistics network configuration options. Moreover,
sourcing process. With a procurement strategy focused on the the proposed taxonomy could be integrated with some
minimisation of purchasing costs through multiple sourcing, considerations about purchasing strategies (e.g. multiple
the demand on each supplier-RW linkage will be lower and as sourcing vs a limited number of suppliers) and business
a result the logistics complexity will increase. As far as the relationships with foreign countries, established not only for
level of customisation is concerned, the higher this level, the sourcing activities (as supposed in the present paper) but also
lower the opportunity to derive general results. In this case a
for sales purposes. In fact, in this paper the CH is assumed as
higher level of variability of demand usually occurs, thus
a node/facility for optimising the sourced flows. The overall
requiring to exploit also risk pooling strategies for inventory
assessment could result in different outcomes if the CH had
allocation (Simchi-Levi et al., 2003; Lovell et al., 2005) and
to fulfil the orders for the local market as well.
postponement strategies for customising products (Van Hoek,
The proposed framework can be useful also for those
2001).
logistics service providers operating as fourth party logistics
(4PLs), involved in supply chain planning and design on
6. Conclusions behalf of global clients rather than merely providing
As global sourcing dramatically alters the landscape of transportation and materials handling services.
business activities, there needs to be a similar change in the With regards to its practical implications, the presented
way in which logistics networks are designed. However, the framework provides supply chain managers and decision
performed literature review showed some relevant research makers with a useful tool for supporting the design of global
gaps in the design of logistics networks in global sourcing logistics networks. By implementing this framework, decision
contexts, from both a theoretical and a practical point-of-view. makers will be able to study different scenarios and, therefore,
With reference to the theoretical implications of the present to determine the optimal logistics network configuration with
research, this paper contributes to the development of the respect to the existing and prospective features of their supply
current debate on the considered issues, thus expanding the chain.
existing knowledge on the configuration of international The obtained results open a number of other issues calling
logistics networks. In particular, we identified five logistics for in-depth studies and additional investigations. For
network configurations, implying the adoption of instance, the number of CHs to be included in the network,
international container shipping and we developed a with respect to the number and location of suppliers, is a key
framework to support the choice of the most suitable (but hard to generalise) issue to further evaluate the
logistics network option from a manufacturing company’s or effectiveness of “one echelon logistics network with
a retailer’s point-of-view. We showed that configuration 1 consolidation hub” configuration. Then, in a context

162
Evaluating logistics network configurations for a global supply chain Supply Chain Management: An International Journal
A. Creazza, F. Dallari and M. Melacini Volume 15 · Number 2 · 2010 · 154 –164

characterised by high relevance of inventory holding cost, it disruptions: design characteristics and mitigation
could be interesting to evaluate the benefits deriving from capabilities”, Decision Sciences, Vol. 38 No. 1, pp. 131-56.
different safety stock allocation strategies in the logistics Dallari, F., Marchet, G. and Melacini, M. (2006),
network, considering further elements such as currency “Transportation strategies in the global supply chain”,
exchange rate risks. Proceedings of the 4th International Logistics and Supply Chain
Congress, Izmir, Turkey, 29 November-1 December 2006,
References pp. 527-34.
Fisher, M. (1997), “What is the right supply chain for your
Arntzen, B.C., Brown, G.G., Harrison, T.P. and Trafton, L.L. product?”, Harvard Business Review, March/April,
(1995), “Global supply chain management at digital pp. 105-16.
equipment corporation”, Interfaces, Vol. 25 No. 1, Gargeya, V. and Meixell, M. (2005), “Global supply chain
pp. 69-93. design: a literature review and critique”, Transportation
Ballou, R.H. (2001), Business Logistics Management, Prentice- Research part E, Vol. 41 No. 6, pp. 531-50.
Hall, New York, NY. Gelderman, C.J. and Semeijn, J. (2006), “Managing the
Bhatnagar, R. and Viswanathan, S. (2000), “Re-engineering global supply base through purchasing portfolio
global supply chains: alliances between manufacturing firms management”, Journal of Purchasing and Supply
and global logistics services providers”, International Journal Management, Vol. 12 No. 4, pp. 209-17.
of Physical Distribution & Logistics Management, Vol. 30 Goetschalckx, M., Vidal, C.J. and Dogan, K. (2002),
No. 1, pp. 13-34. “Modelling and design of global logistics systems: a review
Bozarth, C., Handfield, R. and Das, A. (1998), “Stages of of integrated strategic and tactical models and design
global sourcing strategy evolution: an exploratory study”, algorithms”, European Journal of Operations Research,
Journal Operations Management, Vol. 16 No. 2, pp. 241-55. Vol. 143 No. 1, pp. 1-18.
Brown, S., Squire, B. and Blackmon, K. (2007), Hong, E. and Holweg, M. (2002), “Evaluating the
“The contribution of manufacturing strategy involvement effectiveness and efficiency of global sourcing strategies:
and alignment to world-class manufacturing performance”, a conceptual note research”, working paper, Institute of
International Journal Operations & Production Management,
Management, University of Cambridge, Cambridge.
Vol. 27 No. 3, pp. 282-302.
Imai, A., Nishimura, E., Papadimitriou, S. and Liu, M.
Cheong, M., Bhatnagar, R. and Graves, S. (2007), “Logistics
(2006), “The economic viability of container mega-ships”,
network design with supplier consolidation hubs and
Transportation Research Part-E, Vol. 42, pp. 21-41.
multiple shipment options”, Journal of Industrial and
Kotabe, M. and Murray, J.Y. (1989), “Global sourcing
Management Optimization, Vol. 3 No. 1, pp. 51-69.
strategies and sustainable competitive advantage”, Industrial
Chopra, S. and Meindl, P. (2004), Supply Chain Management:
Marketing Management, Vol. 33 No. 1, pp. 7-14.
Strategy, Planning and Operations, Prentice-Hall, New York,
Kotabe, M. and Omura, G. (1989), “Sourcing strategies of
NY.
European and Japanese multinationals: a comparison”,
Choy, K.L. and Lee, W.B. (2003), “A generic supplier
Journal of International Business Studies, Vol. 20 No. 1,
management tool for outsourcing manufacturing”, Supply
Chain Management: An International Journal, Vol. 8 No. 2, pp. 113-30.
pp. 140-54. Kruger, R. (2002), “Global supply chain management:
Christopher, M. (2000), “The agile supply chain: competing extending logistics’ total cost perspective to configure
in volatile markets”, Industrial Marketing Management, supply chains”, in Goldbach, M. and Seuring, S. (Eds),
Vol. 29 No. 1, pp. 37-44. Cost Management in Supply Chains, Springer, Berlin,
Christopher, M. and Towill, D. (2000), “Supply chain pp. 309-24.
migration from lean and functional to agile and Law, A.M. and Kelton, W.D. (1997), Simulation Modeling and
customised”, Supply Chain Management: An International Analysis, McGraw-Hill, Singapore.
Journal, Vol. 5 No. 4, pp. 206-13. Lee, H. and Yang, H.M. (2003), “Strategies for a global
Christopher, M., Peck, H. and Towill, D. (2006), logistics and economic hub: Incheon International
“A taxonomy for selecting global supply chain strategies”, Airport”, Air Transport Management, Vol. 9 No. 2,
The International Journal of Logistics Management, Vol. 17 pp. 113-21.
No. 2, pp. 277-87. Lemoine, W. and Dagnaes, L. (2003), “Globalisation
Chung, W., Yam, A. and Chan, M. (2004), “Network strategies and business organization of a network of
enterprise: a new business model for global sourcing”, logistics service providers”, International Journal of Physical
International Journal of Production Economics, Vol. 87 No. 3, Distribution & Logistics Management, Vol. 33 No. 3,
pp. 267-80. pp. 209-28.
Cohen, M.A. and Huchzermeier, A. (1996), “Global supply Li, G., Yi, L. and Wang, S. (2006), “Enhancing agility by
chain network management under price/exchange rate risk timely sharing of supply information”, Supply Chain
and demand uncertainty”, Operation Research, Vol. 44 No. 1, Management: An International Journal, Vol. 11 No. 5,
pp. 100-13. pp. 425-35.
Cohen, M.A. and Lee, H.L. (1989), “Resource deployment Liu, B.L., Berger, P., Zeng, A. and Gerstenfeld, A. (2008),
analysis of global manufacturing and distribution “Applying the analytic hierarchy process to the offshore
networks”, Journal of Manufacturing and Operations outsourcing location decision”, Supply Chain Management:
Management, Vol. 2, pp. 81-104. An International Journal, Vol. 13 No. 6, pp. 435-49.
Craighead, C.W., Blackhurst, J., Rungtusanatham, M.J. and Lovell, A., Saw, R. and Stimson, J. (2005), “Product value-
Handfield, R.B. (2007), “The severity of supply chain density: managing diversity through supply chain

163
Evaluating logistics network configurations for a global supply chain Supply Chain Management: An International Journal
A. Creazza, F. Dallari and M. Melacini Volume 15 · Number 2 · 2010 · 154 –164

segmentation”, The International Journal of Logistics Zeng, Z. and Rossetti, C. (2003), “Developing a framework
Management, Vol. 16 No. 1, pp. 142-58. for evaluating the logistics costs in global sourcing
MacCormack, A.D., Newmann, L.J.I. and Rosenfield, D.B. processes: an implementation and insights”, International
(1994), “The new dynamics of global manufacturing site Journal of Physical Distribution & Logistics Management,
locations”, Sloan Management Review, Vol. 35 No. 4, Vol. 33 No. 9, pp. 785-803.
pp. 69-84.
Mattsson, L. (2003), “Reorganization of distribution in
globalization of markets: the dynamic context of supply About the authors
chain management”, Supply Chain Management:
An International Journal, Vol. 8 No. 5, pp. 416-26. Marco Melacini received his PhD in Production System and
Miemczyk, J. and Howard, M. (2008), “Supply strategies for Industrial Plant Design in 2002 from Parma University. He is
build-to-order: managing global auto operations”, Supply a Researcher at Politecnico di Milano, Department of
Chain Management: An International Journal, Vol. 13 No. 1, Management, Economics and Industrial Engineering and
pp. 3-8. currently lectures on Modelling the Supply Chain at
Monczka, R.M. and Trent, R.J. (2003), “Understanding Politecnico di Milano. He is Co-director of the Physical
integrated global sourcing”, International Journal of Physical Distribution course for executives at the MIP, the Politecnico
Distribution & Logistics Management, Vol. 33 No. 7, di Milano Business School. His current research interests
pp. 607-29. include global logistics networks, supply chain risk
Monczka, R.M. and Trent, J. (2005), “Achieving excellence management and order-picking systems. He is author and/or
in global sourcing”, MIT Sloan Management Review, Vol. 47 co-author of technical and scientific publications. Marco
No. 1, pp. 24-32. Melacini is the corresponding author and can be contacted at:
Nahmias, S. (1997), Production and Operations Analysis, Irwin/ [email protected]
McGraw-Hill, Homewood, IL. Alessandro Creazza is currently a research assistant at
Peterson, K.J., Frayer, D.J. and Scannel, T.V. (2000), C-log, Logistics Research Centre – Carlo Cattaneo LIUC
“An empirical investigation of global sourcing strategy University, and has participated in a research exchange
effectiveness”, The Journal of Supply Chain Management, program with Cranfield University School of Management at
Vol. 36 No. 2, pp. 29-38. the Centre for Logistics and Supply Chain Management. His
Pollit, D. (1998), “Transport and the community”, research interests are mainly focused on logistics and
International Journal of Physical Distribution & Logistics distribution network design and modelling, global supply
Management, Vol. 28 No. 3, pp. 221-3. chain management, logistics real estate and warehousing.
Simchi-Levi, D., Kaminski, P. and Simchi-Levi, E. (2003), He is an adjunct lecturer in Logistics and Supply Chain
Designing and Managing the Supply Chain: Concepts, Management at Carlo Cattaneo LIUC University and at
Strategies, and Case Studies, Irwin McGraw-Hill, New York, Politecnico di Milano, and in his working background he has
NY.
been developing a wide experience in logistics network design
Tyan, J.C., Wang, F. and Du, T.C. (2003), “An evaluation of
and transport management.
freight consolidation policies in global third party logistics”,
Fabrizio Dallari is currently Associate Professor of Business
Omega, Vol. 31 No. 1, pp. 55-62.
Logistics and Supply Chain Management at Carlo Cattaneo
Van Hoek, R.I. (2001), “The rediscovery of postponement:
University (LIUC) and he is also head of the Logistics
a literature review and directions for research”, Journal of
Research Center (C-log). He is contract professor at
Operations Management, Vol. 19 No. 2, pp. 161-84.
Vereecke, A. and Dierdonck, R.V. (2002), “The strategic role Politecnico di Milano, where he formerly served as research
of the plant: testing Ferdows’ model”, International Journal assistant, after studying for a PhD at the Università degli
of Operations & Production Management, Vol. 22 No. 5, Studi di Firenze. His current research interests include
pp. 492-514. Business Logistics and Production Management with a
Womack, J. and Jones, D. (1996), Lean Thinking: Banish special emphasis on physical distribution network design,
Waste and Create Wealth in Your Corporation, Simon transportation systems, materials handling, forecasting and
& Schuster, New York, NY. demand planning. He has published several papers in leading
Zapfel, G. and Wasner, M. (2002), “Planning and academic journals, conferences and has written three books.
optimisation of hub-and-spoke transportation networks of Since 1994 he has been teaching Logistics and Supply Chain
cooperative third-party providers”, International Journal of Management in several Executive Education Programs and
Production Economics, Vol. 78 No. 2, pp. 207-20. Master’s-degree courses. Fabrizio Dallari has cooperated with
Zeng, Z. (2003), “Global sourcing: process and design for many companies ranging from international manufacturers
efficient management”, Supply Chain Management: such as Nestlé, Tenaris-Dalmine, Pirelli and many national
An International Journal, Vol. 8 No. 4, pp. 367-79. companies, leaders in their respective fields.

To purchase reprints of this article please e-mail: [email protected]

Or visit our web site for further details: www.emeraldinsight.com/reprints

164