Content uploaded by Miguel Mujica Mota
Author content
All content in this area was uploaded by Miguel Mujica Mota on Aug 08, 2018
Content may be subject to copyright.
MORE THAN JUST CHOCOLATE: SUPPLY CHAIN MODEL OF PRODUCTION OF
COCOA CROPS IN CÔTE D’IVOIRE
Miguel Mujica Mota (a), Abdel El Makhloufi(b), Nico De Bock(c),, Paolo Scala(d)
(a),(c),(d)Aviation Academy, Amsterdam University of Applied Sciences, The Netherlands
(b) Research group Smart Mobility & Logistics, Amsterdam University of applied Sciences, The Netherlands
(a) m.mujica.mota@hva.nl, (b) a.el.makhloufi@hva.nl (c) n.de.bock@hva.nl, (c) p.m.scala@hva.nl
ABSTRACT
Côte d’Ivoire produces about 42 percent of the world’s
total Cocoa but processes only a very few amount of the
production. A big part of the country depends on the
commercial benefits of the Cocoa production and supply
chain of it. For that reason, the World Bank asked the
simulation group of the Amsterdam U. of Applied
Sciences in collaboration with the Port of Amsterdam to
develop a simulation model that allows the politicians
assess the performance of the supply chain of the Cocoa
in that region of the world. The simulation model gave
light to the potential of improvement in the supply chain
by identifying inefficiencies, bottlenecks and blockers
that hinder the efficient transport of Cocoa in the chain
with the consequence of low productivity. The most
important results are presented in the article together with
suggestions for improvement in order to increase the
wellbeing of the farmers in that region of Africa.
Keywords: logistics modelling, Ivory Coast, developing
countries, transport
1. INTRODUCTION
Côte d’Ivoire produces about 42,4 percent of the world’s
total Cocoa but processes only 0.51 million tons of Cocoa
beans in the country (2015). The main importer of Cocoa
from Côte d’Ivoire is the Netherlands. The last years,
Côte d’Ivoire has gained a larger market share, both in
production of Cocoa beans and grinding, respectively,
from 36.7 percent in 2013 to 39.3 percent in 2015, and
from 11.3 percent in 2013 to 12.6 percent in 2015 (Port
of Amsterdam 2016). Seventy percent of the total Cocoa
production in Côte d’Ivoire is obtained from the
following production zones: Soubré, San Pedro, Dalao,
Divo and Gagnoa.
There are many challenges facing the development of
Cocoa sector and performance of the logistics system in
Côte d’Ivoire. The organisation of the Cocoa’s supply
chain suffers from various problems such as limited
traceability, poor quality and congested transport roads,
increasing waste, lack of storage facilities/warehouses
and time-consuming administrative processes.
Furthermore, the market of Cocoa sector is highly
concentrated in the sense that the bulk of trade and
processing of the market of Cocoa is dominated by a
limited number of foreign exporters. Because the
multinational companies are strong in terms of capital
and use of sophisticated technologies, barriers of entry
are higher for local firms to enter the export market as
economies of scale require large investments and
volumes of export, especially in case of shipping Cocoa
in liquid or solid forms.
Others main challenges that the Cocoa sector is facing
are:
• The Cocoa supply chain in Côte d’Ivoire is
dysfunctional and not favourable to the majority of
Cocoa farmers that receive frequently low market prices.
The supply chain is often too long and characterized by
the proliferation of many stakeholders, with most
operators not performing any marketing function that
adds value to Cocoa beans, while taking a share of the
market prices.
• Farmers often do not have access to market
information and technology and their understanding of
the quality requirements of the market is very limited.
This translate into low productivity, low income and
decreasing yield.
• A fragmented and inappropriate functioning of the
market that results in a trading system in which quality is
often compromised.
• The majority of Cocoa farmers sell their Cocoa
beans individually to itinerant buyers (not necessarily
retailers), which often operate in areas where it is
difficult for farmers to transport the Cocoa themselves.
• A widespread practice of mixing good and bad
quality Cocoa beans to meet minimum market quality
standards.
• Limited access of farmers to productivity-enhancing
inputs and resources such as fertilizers, agrochemicals,
seedlings, farm tools and credits, which affect the
productivity and competitiveness of the Cocoa sector.
Côte d’Ivoire has made significant progress in the
development of roads, power and ICT networks during
the 1990s. After 1999 this progress slowed down because
of a lack of investments, and political turmoil. Spending
on infrastructure was less than 5 percent of GDP in the
mid-2000s, which is about half of what many
neighbouring West African countries have been devoting
to infrastructure in this period. Various empirical studies
show that improvement of country’s infrastructure
endowment, such as energy supply, roads networks, rail
infrastructure and terminal capacity of ports and airports
could rise growth at a rate of 2%.
Road network (82.000 km) is relatively well developed
in Côte d’Ivoire and although of a low density, it
provides sufficient connectivity to link the capital cities,
secondary towns and international borders. In
opposition, rail network is not developed. The country
has only one rail link for transporting goods which
connects Abidjan with the capital of Burkina Faso
(Ouagadougou).
Besides low density and low quality of road network,
there are several problems that have direct effect on
transport and logistics sector in Côte d’Ivoire such as the
increasing transport prices, high operational costs and
unpredictable delays to the transport of goods due to the
extraction of significant bribes from trucks along the
roads by police. As result, transporters tend to overload
their trucks to compensate for the costs of the bribes and
other additional charges (for example, charge load per
axle).
In order to develop and implement a wide logistics
(supply) chains and network that capture various
dimensions of performance at various levels in a
consistent way, there is the need of using adequate and
valuable tools (i.e. set of indicators) covering several
levels; the strategic level, the tactical and the operational
level together with novel techniques and methodologies
that allow more transparency in the expected outcomes
of policy processes.
Globally, the focus on the Cocoa’s logistics chain and
network in Côte d’Ivoire may be approached by looking
at the following indicators dedicated to evaluate
performance and trends in logistics practices:
• Physical state of road infrastructure and transport
intensity (tonnes-km/total output).
• Freight volume through load capacity/factor of
vehicle by mode (ton/vehicle).
• Distance by transport mode (km).
• Vehicle utilization (vehicle-km/ton-km).
• Freight movements and energy and emissions by
supply chain link (energy consumed/vehicle-km).
• Energy consumption/emissions.
• Time (total time for transport and storage and related
procedures, average and maximum number of
hours/days).
• Cost (total costs of transport and storage and related
procedures).
• Variability (total time of document processing
hours/days).
• Complexity (total number of documents per trade
transaction).
• Financial cost of logistics services and hidden costs
(costs of delays and uncertainties). These costs include
financial charges, obsolescence, and loss of damaged or
stolen goods.
In addition, the domestic Cocoa supply chain is formed
by approximately 800.000 farmers, 500 cooperative
companies, 5400 traders (5000 pisteurs and 400
traitants), 50 exporters and about 6 local grinders and
local processing firms (see Figure 1 below). What
characterizes the Cocoa supply chain in Côte d’Ivoire is
that farmers sells their crop to three different actors: to
domestic processing firms through cooperatives, to
traders (pisteurs and traiteurs) or market them through
traders to exporters. The domestic processing firms sells
processed Cocoa products directly on international
market.
Due to the complex nature of the supply chain of Cocoa,
the only approach that enabled to assess all the important
aspects and the variability of the chain was Simulation.
In the current work, we present a simulation model of the
su pply chain of Coco a in Co te D’Iv oire wh ich can be a lso
used as a decision-support tool. The model represents the
different relationships already mentioned, so that it
enables the decision-makers with a wide-angle view for
policy making.
Figure 1: Conceptual Model of Supply Chain
2. STATE OF THE ART
Modelling and simulation is a technique that has been
widely used for supply chain management (SCM), as it
is mentioned in Longo (2011), it is “a powerful tool for
analysis, investigation, examination, observation and
evaluation of real-world industrial and logistics
systems”. In the work of Ingalls (1998), the author makes
a review of the simulation technique applied to SCM
where he outlines the advantages of using simulation. In
his work he points out that simulation fits best for tactical
planning, usually where the time horizon is long, and for
supply chains affected by variance. Supply chain is
usually modelled as a multi-agent and in a dynamic
fashion (Swaminathan et al. 1998; Kahiara 2003) in order
to be able to consider all the interactions between the
different actors of the supply chain. Simulation models
for SCM are usually based on agent-based simulation
(ABS) and discrete-event simulation (DES), and even a
combination of the two as it is proposed by Lee et al.
(2002). Regarding ABS models we can find many of
them in the literature, considering different policies and
objectives. In the work of Gjerdrum et al. (2001) an ABS
model was presented with the objective of simulating
different demand-driven supply chains. They included an
optimization model for the manufacturing component,
with the objective of reducing operative costs and
keeping a high level of customer order fulfilment. Albino
et al. (2007), modelled a supply chain by focusing on
cooperation between supply chain actors in industrial
district. The concept of industrial district consists of an
area where many small and medium enterprises (SMEs)
are located and work together in the supply chain; this
concept has been also translated in the model proposed
in this work by clustering the production sites. In other
works we can find supply chain modelled using ABS
focusing on different strategies such as: analysis of
alternative production-sales policies (Amini et al. 2012)
and different combined contract competition (Meng et al.
2017). DES models have been also widely exploited in
the SCM, as it can be found in Longo and Mirabelli
(2008), where they proposed a decision-making tool for
different supply chain scenarios. Their scenarios were
based on multiple performance measures and user-
defined set of data input parameters. In the work of
Mensah et al. (2017) another DES model was developed
for a resilient supply chain with the aid of ICT
implementation. The model follows a six sigma approach
to improve the overall supply chain resiliency against
disruption.
In the present work, a DES model has been implemented
to describe and analyse the supply chain of Cocoa
products from producers, distribution centre and final
centre of collection and shipment. Because of the nature
of the network and operations modelled and also the
tactical nature of the analysis, DES was considered as a
best approach by the authors. DES allowed us to make a
scenario-based analysis based on different policies, and
it enabled us to measure the performance of the supply
chain based on different aspects (productivity, economic,
and environment).
3. METHODOLOGICAL APPROACH
In this work, we simulated the different links of the
supply chain, the main boundary is the Port of San Pedro
where the exporting function is performed. The model
was based on the supply chain mapping made by the Port
of Amsterdam (Port of Amsterdam 2017). Figure 1
presents the description of the supply chain relationships
that will be considered for the developed model.
The farmers produce the Cocoa beans which in turn are
transported by a merchant or the production is
concentrated by a cooperative of producers (Farmers) in
warehouses. In the next link, the product is transported
as raw material directly to the Port and some percentage
is transported to the Grinders (30% of the production). In
the grinder or refinery, the raw Cocoa is transformed into
Cocoa butter and then transported as a higher-value
product to the Port of San Pedro. The next link in the
supply chain is the transport of either the raw material
(beans) or the refined product (butter or oil) by sea to the
destination Port, in our study the destination Port in
Europe corresponds to the Port of Amsterdam. However,
as mentioned before, the transport to the Port of
Amsterdam will be out of the scope of the developed
model.
3.1. Conceptual Modelling
The first modelling phase corresponds to the conceptual
development in which the relationships between the main
elements are identified based on public information and
the discussion with subject-matter experts. The following
table presents the elements that are included in the
simulation model.
Table 1: Main elements of the conceptual model
Elemen
t
Description
Production
quantities
The quantities of Cocoa beans
produced in the specific region
under stud
y
Transport routes The routes of transport that are
used b
y
the different transports
Grinder
Facilities
Cocoa that is transported in the
grinder facilities as well as the
added value to the product for
evaluating the impact of
decisions
Transport trucks The trucks that transport the
products for being export at the
Port. The speeds and capacities
will be considered as well
Warehouses The warehouses dwell time will
b
e considere
d
Checkpoints
along the
transport routes.
These points will be considered
since they hinder the smooth
flow of truc
k
s towards the Port.
State of roads
and pavement
The quality of the road will be
considered, since it has a direct
impact in the transport time from
the different locations in the
re
g
ion to the Por
t
Market value of
the products
The market value of the modelled
p
roducts will be considere
d
Emissions In the model the green-house
pollutants are considered, mainly
CO2 and NO2
3.2. Production modelling
Some functionality is key for the correctness of the
developed mode. Production is one of them, as it is
appreciated in Figure 2, the production varies with the
region of the country. Due to the objectives pursued, the
production at an atomic level was not considered,
instead, a high-level model was developed. The
complexity of the production network was coped by
developing clusters within the region under study (San
Pedro Region), taking into consideration the production
zones and the political boundaries.
The production of Cocoa in the model is generated in
batches of 6 tons within the cluster and located randomly
within the cluster. The model developed has stochastic,
dynamic and flexible characteristics; the amount of
entities is generated in such a way that they match the
amount of production of the region within a year.
00
0
1
)(
x
xe
xf
x
Figure 2: Production of Cocoa in Cote D’Ivoire
The production is modelled assuming a Poisson process
where the main probability distribution is an exponential
one with the average inter-arrival time of 3.33 minutes
(1).
(1)
Assuming this inter-arrival time, the average production
of a year will be 946 000 tons. It is important to mention
that this production is stochastic meaning that every time
the model is run, the production might vary, due to the
interlinks and causal relationships present in the supply
chain model. However, the average mean will be around
946 000 tons/yr.
3.3. Road Network Infrastructure
According with Openstreets (OpenStreets.org 2017), the
road infrastructure of Côte d’Ivoire consists of 5
classifications of roads, however, for reducing the
complexity, the model accounts with three types of
roads:
Primary Road (Paved). These roads are paved
roads, with good maintenance, the achievable
speeds of the trucks could get up to 100-120
km/hr. However, the maintenance of these
roads is very scarce, thus, the roads are filled
with potholes and stones making the vehicles
reduce their speed and sometimes break their
tires, decreasing drastically the average speed in
the road. For the paved motorways we specified
a stochastic speed following a Triangular
distribution of T(40,50,60) km/hr.
Secondary Road(s). The secondary roads are
roads that are unpaved. In these roads, the
average expected speed is also very uncertain.
In this roads we assumed a stochastic speed
following a triangular distribution of T
(20,30,40) km/hr.
Tertiary Road(s). These roads are also unpaved,
and these types of roads are the ones used by the
primary producers (individual and local
families), in these roads, the farmers can use
small vehicles or even bicycles for the transport
of Cocoa beans. For the modelling of these
tertiary roads, we specified also a stochastic
speed of T (10, 15, 20) km/hr.
The road infrastructure is modelled as a set of nodes and
edges with two directions at a certain scale in which the
edges correspond to the direction, sense and length of the
actual roads. The links (edges of the network) in turn are
placed over a GIS map from OpenStreet.org, then the
entities modelling the production and the trucks
transporting the product are placed over the GIS layer.
Thus, two layers are used, one which is a GIS layer of the
region under study and the second layer which is
composed by the entities, nodes, edges and other objects
like servers that model the performance of the
production once it enters to a warehouse, location or to
the Port (simulation layer). Figure 3 illustrates the layers
used for constructing the scaled model of the regions of
San Pedro.
Figure 3: The GIS from OpenStreet.org for Côte
d’Ivoire
The simulation layer has been developed using a
discrete-event systems approach, using a commercial
simulator called SIMIO (SIMIO 2018).We took the
advantages of the functionalities of the simulator for
making a more detailed model of the supply chain.
3.4. Grinders, warehouses and Ports
These elements are represented by functional nodes that
are connected via the edges that model the roads, in
addition, these nodes will have some functionalities.
These functionalities, model characteristics such as
capacity, delays of all the internal processes that the
product undergo when they get to the node (e.g. grinding,
loading, unloading, packing, unpacking, sorting, etc.).
Figure 4 illustrates the network approach model of the
supply chain under study.
The links or segments are geographically aligned with
the GIS layer so that they have the right proportion and
length that the vehicles need to traverse in order to go
from one location to the other. Some of the nodes also
have functionalities that are used to model operations
performed in the locations of the GIS Map such as check
points, warehouses, grinding operations or check points
along the road.
Figure 4: model of the road network
The warehouses and location nodes will have as main
functions, the storage, transformation and transhipment
of the product in the network layer. The processes related
to those activities, consist of loading, unloading,
processing, and storing. Table 2 illustrates the
characteristics of these elements in the system.
Table 2: Characteristics of the Warehouses, Port and
Grinders
Facilit
y
Processin
g
Time Capacit
y
Warehouse Triangular(5,6,7)
da
y
s
Unlimited
Grinde
r
Uniform (12, 24) hrs. Unlimited
Por
t
N
ULL Unlimited
3.5. Vehicles
The production and transport of Cocoa is modelled using
different entities, mainly two entities. One entity will
represent the amount of 6 tons of Cocoa; the other vehicle
will model the heavy trucks or trailers whose capacity is
maximum of 60 tons. The entities will have other
characteristics besides the capacity, such as speed, CO2,
NO2 emissions and they will move through the edges of
the network layer. Table 3 illustrates the characteristics
of the different vehicles used in the model.
Table 3: Vehicle characteristics
Vehicle Capacit
y
Speeds
Small Truck 6 ton
(Fixed)
60 km/hr *
Trailer [0..60] tons Triangular (30, 45,
60) km/hr *
* Maximum speed, the model will be restricted by the
roads limitations
The different types of vehicles are parameterized with the
characteristics of the entities they represent. These
parameters, will be used as variables that can be modified
during the experimental design in order to compare the
impact of different policies in the system.
In the warehouses, the entities are batched up to 60 ton
(10 entities) and transported by another entity with
similar characteristics but with a different representation.
Figure 5 presents the entities used for modelling the
transport and production of Cocoa with their
characteristics.
.
Figure 5: Vehicle characteristics
Inter-arrival time per Region
The amount of production per region was estimated
based on the yearly production rate. In the case of San
Pedro region, the production per year is approximately
946,000 tons/yr. In order to generate a stochastic
approach for this production, it was assumed that the
production follows a Poisson process where the
production is modelled by an exponential distribution
with an inter-arrival time with the correspondent mean.
The matching value for the region under study is 3.33
min between entities for batches of 6 tons. The model
assumes that during the year the number of tons is
produced relatively evenly.
3.6. Check Point Implementations
The checkpoints at the roads are important elements to
be considered in the model. Based on public information
(World Bank, 2008), check points in the region of San
Pedro have been located in the model, with a
correspondent processing time and probability of being
checked. The following figure illustrates the details of
the check points in the road network.
Figure 6: Check Point Implementation
For the implementation of the checkpoints, the network
has a detour based on a probability for the vehicles to be
checked. The checkpoint is composed by three segments
with functionality where the trucks are deviated from
their original route to the Port. They have a specific
probability of being checked, initially is 10%. In the
experiments, the variation of this probability is used to
evaluate the impact of the checkpoints in the lead time of
the products. In this way, it also enables the decision
makers to evaluate policies that reduce the hassle or
inefficiencies that the checkpoints add to the supply
chain. The following table presents the characteristics of
the checkpoints in the supply chain model.
Table 4: check point characteristics
Element Processing
Time
Capacity Initial
Probabili
ty of
Inspectio
n
Checkpoint
enter
segment
Heavy
Truck:
Triangular
(1,2,3) hrs.
Regular
Truck :
Triangular
(5,10,15)
mins
5 Entities 0.1
Checkpoint
parking
se
g
men
t
NULL Unlimited n/a
3.7. Emissions
In the simulation model, the CO2, NO2 and CH4
emissions were considered. A lineal equation was used
according to EPA factors and following the approach of
different authors (Kenney et al. 2014). The emissions are
dependent on the type, age and distance travelled by the
vehicles. The formulas used for estimating the emissions
of the trucks are the following ones:
E_tot=(RL x NV x EF ) (2)
Where
E_tot : Total Gas Emissions
RL: route length
NV: Kilometres traversed
EF: Emission factor
Regarding the characteristics of the vehicles, Table 5
presents the values used for the vehicles in the model
(EPA 2018).
Table 5: Vehicle emission characteristics
Type of
Vehicle
Year
Assum
ed
Emissio
n Factor
CO2
Emissio
n Factor
NO2
Emissi
on
Factor
CH4
Regular
Truck
(Gasoline)
1973 –
1974
1.67
[kg/km]
0.31
[gr/km]
0.28
[gr/Km
]
Heavy
Truck
(Gasoline)
1981 1.67
[kg/km]
0.31
[gr/km]
0.28
[gr/Km
]
4. EXPERIMENTAL DESIGN
This section presents the scenarios that have been
evaluated using the supply chain model. The evaluation
started with a base-case scenario which simulates the
status quo of the system under study. Then with the use
of the other scenario, the impact of different policies are
compared. The following table illustrates the different
scenarios that can be possibly analysed by modifying the
factors and levels. As it can be inferred, a full factorial
design was not feasible since the number of
combinations would be 216 scenarios, however, based on
the authors experience, we selected the scenarios that
were considered most relevant for illustrating the
capabilities of the tool and that allowed to identify
potential areas of improvement of the whole system to
bring more productivity to the region.
Table 6: Design of Experiments for the Supply Chain
Model
FACTOR LEVELS
Speed
secondary
roa
d
Low
speed
Medium speed High
speed
Speed tertiary
roa
d
Low
spee
d
Medium speed High
spee
d
Probability of
checkpoin
t
10% 5 % 0%
Check point
times regular
trucks
Long
waiting
times
Short waiting
times
Check Point
times heavy
trucks
Long
waiting
times
Short waiting
times
Grinder
Percenta
g
e
30 % 50%
Regarding the Performance indicators (PI) considered,
the ones used for the study are the following:
Cocoa Productivity (Ton/day): This PI
measures the amount of Cocoa that is
transported to the Port of San Pedro every day.
By using this PI, we evaluate the impact of
different infrastructure policies. The
expectation is that if one policy has good impact
in the system, the amount of Tons in the Port
will increase, while if it has a negative effect,
the amount in the Port will be reduced.
Cocoa Butter Productivity (ton-day): This PI is
similar to the one from Cocoa, and follows the
same reasoning. The difference is that butter is
a more valuable product, and with the
simulation model it is possible to investigate
when is more economically attractive to invest
in the butter than just transporting the beans.
Cocoa Market Value (USD/Day): With this
indicator, we are able to evaluate and simulate
the value of the produced Cocoa in the market.
The value will be correlated with the
productivity of Cocoa. For the simulation model
it was assumed a constant value of 3.2
USD/KG.
Butter Market Value (USD/Day): with this
indicator, we are able to compare the value of
butter versus the value of Cocoa and how the
policies and decisions impact this indicator. For
the simulation model the value was considered
constant with a value of 7.78 USD/KG
CO2 emissions (Kg/day): This indicator is
directly correlated with the distance travelled,
age and type of trucks used to transport the
Cocoa and the butter in the model. By using this
indicator, it will be possible to measure the
impact of some policies in the emissions of this
pollutant.
NO2 emissions (kg/day): This pollutant is also
measured in the model. The reasoning is the
same as the CO2.
The following section presents the description of the
different scenarios evaluated and the explanation of why
they were chosen.
4.1. Scenario I. Base Case Scenario
Every simulated scenario had 30 replications for a period
of 14 weeks (3 months) in order to evaluate the
performance of the considered PIs. The following table
presents the main parameters and Table 8 the results
obtained.
Table 7: Parameters of the base-case scenario
Parameter Value
Speed Primary Road Triangular(40,50,60)
km/h
r
Speed Secondary Road Triangular(20,30,40)
km/h
r
Speed Tertiary Road Triangular(10,15,20)
km/h
r
Checkpoint probabilit
y
0.1
Checkpoint times
re
g
ular truc
k
Triangular( 1 , 2 , 3 ) hrs
Checkpoint times heavy
truc
k
Triangular(5 , 10 , 15 )
mins
Percenta
g
e of Grindin
g
30%
The following results are the average values out of all the
30 replications measured every day. The results mean
that, for instance, the average productivity values
represent the measurement of 84 days (3 months) times
30 (number of replications) therefore it is an average of
2520 measurements of the productivity.
Table 8: Simulated Results for Base Case Scenario
Parameter Average Value Standard
Distribution
Value Beans
Production
(USD/da
y
)
2,278,031 171,884
Value Butter
Production
(USD/da
y
)
826,915 84,008
Productivity
Beans
(Ton/da
y
)
989 474
Productivity
Butter
(Ton/da
y
)
146 69
Total CO2
Emission
[KG](3
months)
802 9.8
Total NO2
Emission [KG]
(3 months)
14 0.18
It was considered only intensive measures
(measurement/day) in order to make the results not time-
dependent. The only absolute values were the emissions
which measure the total emissions after the 84 days for
the 30 replications. Thus, the results can also be
interpreted as intervals, for example in the case of the
expected market value of the Beans: assuming a normal
distribution one can expect that every day, in
approximately 95% of the time, the market value
productivity would be within and interval of [1 934 263,
2 621 799] USD/day, using a 2 standard deviation
interval.
4.2. Scenario II. Impact of reducing the checkpoints
on the roads
This scenario evaluates the impact of reducing the check
points along the road. As it has been mentioned, the
checkpoints produce inefficiencies in the supply chain,
since it increases the lead time of the Cocoa. However, it
is not known exactly how much is the impact in the
potential of productivity of the system. The following
results help to give light on this matter and on what
impact would it have in the system if the government
reduces the checkpoint frequency. This situation was
modelled by reducing the checkpoint probability on 50%
(from 10% to 5%). The other parameters of the model
were left the same as the base-case scenario.
The results of the simulation are presented in the
following table. In this scenario, the other parameters
than checkpoint probability were left intact.
Table 9: Results of Scenario II
Parameter Average
Value
Standard
Distribution
Value Beans
Production (USD/da
y
)
2,355,652 43,117
Value Butter
Production (USD/da
y
)
871,415 25,969
Productivity Beans
(Ton/da
y
)
1,289 231
Productivity Butter
(Ton/da
y
)
190 34
Total CO2 Emission
[KG](3 months)
1057 13
Total NO2 Emission
[KG] (3 months)
19.64 0.25
In comparison with the base case, the productivity of
transported beans and butter increases around 30%, the
value increases approximately 3% and 5% for Beans and
Butter respectively. On the other hand, the pollution is
increased with the increase of transport activity.
4.3. Scenario III. Impact of Improvement in Road
infrastructure
This scenario evaluates the impact that the improvement
in maintenance of the secondary and tertiary road might
have. As it has been mentioned and illustrated, the road
infrastructure is in a very bad shape in the country. This
situation impacts directly the supply chain efficiency.
This policy has been modelled by matching the speeds of
the secondary and tertiary roads to the speed of the
primary roads; Triangular(40,50,60) km/hr. Table 10
presents the results of this policies.
Table 10: Results of Scenario III
Parameter Average
Value
Standard
Distribution
Value Beans
Production (USD/da
y
)
2,302,323 159,967
Value Butter
Production (USD/da
y
)
848,185 66,620
Productivity Beans
(Ton/da
y
)
1,083 424
Productivity Butter
(Ton/da
y
)
159 62
Total CO2 Emission
[KG](3 months)
873 10
Total NO2 Emission
[KG] (3 months)
16.21 0.20
It can be appreciated that the productivity of Beans and
Butter is increased with 9.5% and 9% respectively, while
the increase in value is only 1%. It is noticeable as well,
that in this case the investment in secondary and tertiary
roads does not have a big impact as with the previous
scenario.
4.4. Scenario IV. Impact of Improvement in road
infrastructure and check points
This scenario evaluates the impact of the combined effect
of improving the road infrastructure and reducing the
checkpoints at the same time. By having a combined
policy of maintenance and more efficient flow of goods
the expected impact is important as the following results
illustrate.
Table11: Results of Scenario IV
Parameter Average
Value
Standard
Distribution
Value Beans
Production (USD/da
y
)
2,353,241 54,509
Value Butter
Production (USD/da
y
)
869,877 21,440
Productivity Beans
(Ton/da
y
)
1,262 251
Productivity Butter
(Ton/da
y
)
186 36
Total CO2 Emission
[KG](3 months)
1,036 13
Total NO2 Emission
[KG] (3 months)
19.25 0.25
As it was expected, with this scenario, the values
increased as well. However, it is noticeable that on the
contrary as to what was expected, the combination of
reducing the checkpoint values and investing in
improving the secondary and tertiary roads does not
produce a higher value than the one from only reducing
the checkpoint probability. The standard distributions are
also similar, thus, it is an interesting result that needs to
be further investigated.
4.5. Scenario V. Impact of investment in producing
more butter than Cocoa
The Cocoa butter has more value in the market than the
Cocoa beans. For that reason, this scenario investigates
what the impact would be if the butter percentage is
increased by diverting more production to the grinders.
For this scenario the percentage of beans that go to
grinder is 50%, the other parameters are left the same as
with the base-case scenario. This scenario also assumes
that the grinders have enough capacity to process the
beans and that 20% of the mass of the beans is
transformed into butter. The characteristics and results
are presented in the following tables.
The results show that since half of the beans now are
converted to butter, the productivity of butter is
increased. It is noticeable that beans productivity also
increase a bit in comparison with the base case scenario.
This is another interesting result that requires further
investigation. The most remarkable result is that despite
the butter value production is increased, the combined
average productivity value (beans and butter) is
decreased in comparison with the base case. This might
be due to the market value of both products and also
because in the simulation model other side products such
as Cocoa oil or secondary products were not considered.
Table 12: Results of Scenario V
Parameter Average
Value
Standard
Distribution
Value Beans
Production (USD/da
y
)
2,053,387 67,685
Value Butter
Production (USD/da
y
)
1,002,133 32,503
Productivity Beans
(Ton/da
y
)
1,021 290
Productivity Butter
(Ton/da
y
)
199 57
Total CO2 Emission
[KG](3 months)
956 11
Total NO2 Emission
[KG] (3 months)
17.75 0.21
4.6. Scenario VI. Impact of Investment in road
infrastructure and Butter production
This scenario explores the situation of investing in
improving the road infrastructure and increasing the
butter production at the same time. This is simulated by
modifying the speeds of the secondary and tertiary roads
and increasing to 50% the percentage of beans that go to
the grinders. This scenario again provides better results
than the base-case scenario, but as the results show, it is
not the best configuration for improving the PIs of the
system.
Table 13: Results of Scenario VI
Parameter Average
Value
Standard
Distribution
Value Beans
Production (USD/da
y
)
2,048,255 47562
Value Butter
Production (USD/da
y
)
999873 34182
Productivity Beans
(Ton/da
y
)
959 317
Productivity Butter
(Ton/da
y
)
189 62
Total CO2 Emission
[KG](3 months)
903 10
Total NO2 Emission
[KG] (3 months)
16.76 0.20
This final scenario illustrates that the change to
producing more butter increases the productivity of
butter but reduces the one of the beans, and the total sum
of the average values of the combined production is less
than the value for the base-case scenario. Furthermore, it
illustrates that even though the productivity of butter is
increased, the sum of the final market values do not,
which means that for the improvement of the wellbeing
of the population, probably the first policy impacts more
than this one.
5. CONCLUSIONS
The current paper presents a novel approach of the
analysis of the supply chain of the Cocoa in Côte d’Ivoire
. The model presents an approach for modelling and
simulating the evolution of production of the current
system taking into consideration public information. The
developed model considers the most important
stakeholders of the system and it enables the evaluation
of different policies for the current system. Several
scenarios from the total combination are proposed in
which different policies are analysed. They explore
different policies like improving the road infrastructure,
increasing the amount of production of butter or reducing
the check in points in the road network. These policies
are evaluated using different performance indicators. The
following table summarizes the main findings of the
study.
Table 14: Summary of results
Scenario Comparison with
current situation
Sc. II: Reducing check
points on the roads
•Productivity increased
by 30%
•Value increased 3%
and 5% of beans and
Butte
r
Sc. III: Investing in
improving secondary
and tertiar
y
roads
•Productivity increased
by 9%
•Value increased 1%
Sc. IV: Impact of
Improvement in road
infrastructure and
checkpoints
•Productivity increased
by 27%
•Value increased by 3%
Sc. V: Increasing butter
production
•Increase in productivity
Beans by 3.2% and
Butter by 36%
•Value of Beans reduced
by 10%, value of butter
increase b
y
21%
Sc. VI: Investment in
road infrastructure and
butter production
•Reduction of
productivity beans by
3%, increase by
productivity in butter
29%
•Value of Beans
increased by 11%, value
of butter increased by
21%
Pollution in all the cases increased, and the smallest
contribution was achieved in scenario III followed
by
scenario VI
Some results from the study were expected and other
were counter-intuitive, for instance, the results suggest
that by only decreasing the frequency of checkpoints by
50%, the productivity is increased by 30% and the
monetary value at the Port by 5% in the case of the beans.
This increase in value is translated into approximately
200,000 USD more per day of operation. The results also
suggest that by investing in the least developed road
infrastructure like secondary and tertiary roads will not
have a big impact. This might be because there might be
a bottleneck in the warehouses. On the other hand, by
making the last link of the chain more efficient, which in
this case is the transport to the Port, will have a big
impact. Another interesting result is that from the
environmental point of view, the results suggest that the
investment in secondary and tertiary roads will reduce
the pollution, however, the increase in market value will
not be high. This might be due to the amount of trucks
that are required to take the product from the farms to the
warehouses, which will be reduced once the investment
takes place. Last but not least, with the current market
values of butter and Cocoa, the investment in butter
apparently does not increase the total value of the supply
chain. However, in the model it was assumed that only
20% of the beans mass is transformed into butter taking
out other products that might make the transformation to
butter and other products more economically attractive.
This should be further investigated.
As it can be perceived, the simulation model of the
supply chain of Cocoa is a valuable tool that allows
decision makers and analysts having more insight into
the operation of the supply chain. The tool is flexible
enough to incorporate new elements in addition to the
variability of the system which plays a key role in the
performance of it. With the inclusion of the variability
factor in the different operations and links of the supply
chain, it is possible to evaluate and reduce the risk of
wrong, small-value or even dangerous decisions that
might put at stake the wellbeing of the population that
depends on the production of Cocoa. Moreover, the
developed model of supply chain of Cocoa can be easily
translated into the evaluation of another type of agro-
logistic product. This approach can be used as a tool that
allows to have more informed decisions with the
consequence of reducing the risk of making wrong
decisions on the management of public or private money.
ACKNOWLEDGMENT
The authors would like to thank the World Bank and the
Amsterdam University of Applied Sciences for the
support in this research, as well as the Dutch Benelux
Simulation Society (www.DutchBSS.org) and
EUROSIM for disseminating the results of this work.
REFERENCES
Albino, V., Carbonara, N., Giannoccaro, I., 2007. Supply
chain cooperation in industrial districts: A
simulation analysis. European Journal or
Operational Research, 177 (2007), pp. 261-280.
Amini, M., Wakolbinger, T., Racer, M., Nejad, M. G.,
2011. European Journal of Operational Research,
216 (2), pp.301-311.
Banks, J., Carson, J.S., Nelson, B., Nicol, D.M., 2010.
Discrete-even System Simulation, fifth ed. Pearson.
Bogetic, Z., Espina, C, and Noer, J., 2007. Cote d’Ivoire:
Competitiveness, Cocoa, and The Real Exchange
Rate. Policy Research Working paper Nr. 4416
(WPS4416). The World Bank, November 2007.
Washington DC.
Essoh, N.P.S., 2014. Cote d’Ivoire’s Commodities
Export and Shipping: Challenges for Port Traffic
and Regional Market Size. American Journal of
Industrial and Business Management, 4, 234-245.
http://dx.doi.org/10.4236/ajibm.2014.45031
Gjerdrum, J., Shah, N., Papageorgiou, L. G., 2001. A
combined optimization and agent-based approach to
supply chain modelling and performance
assessment. Production Planing & control, 12 (1),
pp. 81-88.
Ingalls, R.,G., 1998. The value of simulation in
modelling supply chains. Proceedings of the 1998
Winter Simulation Conference, pp. 1371-1375.
December 13-16, Washington (DC, USA).
Kaihara, T., 2003. Multi-agent based supply chain
modelling with dynamic environment. Int J.
Production Economics 85 (2003), pp. 263-269.
Laven, A., Buunk, E and Ammerlaan, T., 2016. Market
Concentration and Price Formation in the Global
Cocoa Value Chain. SEO-report nr. 2016-79A. SEO
Amsterdam Economics, Amsterdam.
Law A., Kelton W., 1991. Simulation Modelling and
Analysis, McGraw-Hill, 2nd Ed.
Lee, Y. H., Cho, M. K., Kim, S. J., Kim, Y. B., 2002.
Supply chain simulation with discrete-continuous
combined modelling. Computers & Industrial
Engineering, 43 (20002), pp. 375-392.
Longo, F., Mirabelli, G., 2008. An advanced supply
chain management tool based on modelling and
simulation. Computers & Industrial Engineering, 54
(2008), pp. 570-588.
Longo, F., (2011). Advances of modelling and
simulation in supply chain and industry. Simulation:
Transactions of the Society for Modeling and
Simulation International, 87(8), 561-656.
10.1177/0037549711418033
Meng, Q., Li, Z., Liu, H., Chen, J., 2017. Agent-based
simulation of competitive performance for supply
chains based on combined contracts. International
Journal of Production Economics, 193 (2017), pp.
663-676.
Mensah, P., Merkuryev, Y., Longo, F., 2015. Using ICT
in developing a resilient supply chain strategy.
Procedia Computer Science, 43 (2015), pp.101-108.
OpenStreets.org 2017 <www.openstreetmap.org>
Port of Amsterdam, 2017. Business Opportunities in
d’Ivoire: Analysis of the Cocoa Sector. PoA
International, Amsterdam.
SIMIO, 2018. Available from:
https://www.simio.com/index.php [accessed 18
April 2018]
Swaminathan, J. M., Smith, S. F., Sadeh, N. M., 1998.
Modeling supply chain dynamics: A multiagent
approach. Decision Sciences, 29 (3) 607-632.
Weisfeld, M., 2009. The Object-Oriented Thought
Process, 3rd ed., USA: Pearson Education.
World Bank, 2008. Etude du Racket sur les Routes en
Cote D’ivoire, Technical report, May, 2018.