Modeling logistic chain of lettuce on Panama

Humberto lvarez
Universidad Tecnolgica de Panam, [email protected]

Andrs Orozco
Universidad Martima Internacional de Panam, [email protected]

Despite being recognized as an international logistics center, Panama has no records of deep analysis of
their internal logistics chains that allow you to understand their behavior. This paper presents the preliminary
results of a project funded by the National Secretariat for Science, Technology and Innovation of the Republic
of Panama, which aims to model and analyze the supply chain of agricultural products, specifically lettuce, in
the country. The model presents the behavior of an agricultural product logistics chain, considering aspects
such as costs of shipment and the availability of vehicles for transportation.
Keywords: Agricultural logistic chains, mathematical optimization.

RESUMEN
A pesar de ser reconocida como un centro logstico internacional, Panam no tiene registros de anlisis profundos
de sus cadenas logsticas internas que permitan entender su comportamiento. Este trabajo presenta los resultados
preliminares de un proyecto financiado por la Secretara Nacional de Ciencia, Tecnologa e Innovacin de
la Repblica de Panam, cuyo objetivo es modelar y analizar la cadena de suministros de productos
agrcolas, especficamente lechuga, en el pas. El modelo presenta el comportamiento de la cadena logstica de
un producto agrcola, considerando aspectos tales como costos de envo y la disponibilidad de vehculos para su
transporte.
Palabras claves: Cadenas logsticas agropecuarias, optimizacin matemtica.

1. INTRODUCTION
Panama has been considered a logistics center point for the American Continent. According to the Panamanian
Government Five Year Strategic Plan, it has been considered to create in the country a world class center of
logistics added value services, luxury tourism, high value agriculture and thus enhance its privilege geographical
position and the comparative advantages, decreasing the unemployment rate and increasing the economic growth
between 6% and 9% annual, also to generate approximately 500,000 new jobs.
The objective of this paper is to develop a mathematical model that depicts both the behavior of the agricultural
supply chain and to analyze the transportation equipment used in the logistic chain in order to obtain an optimal
transportation policy of agricultural products through the country. This optimization model is one of the outcomes
of an oingoing project funded by the National Secretariat of Science, Technology and Innovation of Panama
entitled Designing a logistic platform through the optimization of distribution networks for the agricultural
sector. The project is aimed to study the distribution network of agricultural products and propose a decision
making model for optimal locations of modal interchange facilities and logistics platforms. The rest of the
document will present a description of the current situation, a brief literature review of work done in this area, the
methodology used and a first approach to the model and its solution.

11th Latin American and Caribbean Conference for Engineering and Technology
Cancun, Mexico

August 14-16, 2013

2. PROBLEM STATEMENT
According to the 2012 Key Performance Indicators of the Republic of Panama, published by the General
Ccounting Office, although agriculture and other related activities are the main sources of employment in the
rural areas, providing the 19% of total employment nationwide the Agricultural Gross Domestic Product (AGDP 2011) it represents just 3% of total Gross Domestic Product (GDP). Thus, lack attention has been provided to the
analysis, modeling and optimization of the internal agricultural supply chain.
The Province of Chiriqu located at the west side of the country (see map in figure 1) provides almost 80% of the
vegetables consumed in Panama, thus being the main supplier node in the food distribution network of the
country. On the other hand, the Province of Panama at about 500 kilometers east of Chiriqu is the largest market
with almost 50% of the population, hence an optimal distribution networks system is required throughout the
country to serve not only the Province of Panama, but also the rest of the provinces in an efficiently way.
To find the contextual description for the current situation, preliminary data from the different distribution points
was gathered. In addition, an exploratory research was conducted in several locations around the country in order
to know the situation and understand the behavior of the distribution of products at these points. The information
from these sources was collected through interviews and questionnaires applied to a group of stakeholders that
were selected more by convenience than by random selection. Data such as transportation costs, operation costs,
vehicles availability, production capacities, market demand, warehouses and distribution capacities were gathered.
In addition, data from the National Secretary of the Cold Chain and the Agro-Marketing Institute allowed the
researchers to have production data. Furthermore, the data collected from these organizations helped the
researchers to compare this information with the one collected from the suppliers and consumers. At this point,
the information is being carefully analyzed since there are significant differences between the information
collected.
With the information provided, a map of the distribution network was developed. This map is shown in figure 1.
As seen, two production areas were located. As seen, the supply chain has a longitudinal configuration, with two
main production points and several transhipment centers. The main issue of the chain is that over 50% of the
products have to be delivered to Panama City, while the rest of the products are delivered to the rest of the central
provinces.
Products are transported from these areas to different distribution points. These points are David, the largest city
of Chiriqu, which distributes products to the rest of the province. Also, products are sent to Santiago, in the
Province of Veraguas, which distributes products to the province, and also sends products to other areas of in the
middle of the country that serve also as intermediate distribution centers. Finally, they send most of their
production capacity to the Central Supply Market, which is the main distribution point in Panama City. This
center supplies products to the west, east, north and central areas of the Province of Panama, and also to the
Province of Colon, located in the Caribbean area of Panama.
No previous study about the distribution network of agricultural products has been previously conducted in
Panama, hence the objective of the project is to develop an optimization model that helps understanding the
behavior of the agricultural supply chain in Panam, finding not only the minimum cost of satisfying supply and
demand of agricultural products, but also to implement the minimum transportation cost of a vehicle assignment
policy for the minimum allocation of products.
In order to better understand the behavior of the supply chain, several assumptions were made.

In order to facilitate the initial understanding of the model, only one product will be selected.
The selected produce is lettuce.
Supplies and demands at different sources, transshipments and destinations will be considered weekly.
11th Latin American and Caribbean Conference for Engineering and Technology

Cancun, Mexico

August 14-16, 2013

No inventories are allowed in intermediate points.

Only three types of vehicles will be considered: pick-ups, trucks and trailers, as seen in figure 2.
All costs, demands, supplies and availability of vehicles are known.
No unloading and downloading times are considered.
The unit load considered is the 40 lbs. crate of lettuce, as seen in figure 3.
No returning of products.
The cost is divided on two: the transportation cost, that considers production and loading costs, and the
vehicle related cost that considers fuel, operation costs and depreciation.
Production cost is constant and does not depend on the final destination. Thus, it is a fixed cost and has no
influence on the model.

Figure 1: Agricultural Distribution Network

Figure 2: Types of vehicles

Figure 3 Unit load

3. LITERATURE REVIEW
Studying food distribution systems becomes an important point in supply chain management for several reasons.
First of all, food scarcity becomes critical nowadays due to climate changes. Thus, cost of food and produces are
increasing. On the other hand, people need food to be accessible and safe, in terms of availability, effects in health
and costs, becoming a strategic issue for governments. For example, Pietro and Timpanaro (2012) affirm that the
issue of agricultural logistics is the subject of great interest because it is considered strategic for the development
of a country especially on the possible transport links between different areas of the country.
Boudahri, et al (2012) define the term agri-food supply chains (ASC) to describe the activities from production to
distribution that bring agricultural or horticultural products from the farm to the table. ASC are formed by the
farmers, distribution, processing, and marketing of agricultural products to the final consumers. In Panama, the
interface between producers and supermarkets or municipal supply centers (final destination) is completely
11th Latin American and Caribbean Conference for Engineering and Technology
Cancun, Mexico

August 14-16, 2013

monopolized by intermediaries (carriers) which increases the final product cost. All of these intermediates make
up almost all the distribution networks for agricultural products in the internal food supply. This carriers transport
the products of more than 60% of total producers and the remaining 40% is transported by private companies. In
addition, there are losses of 40% of transported products. Therefore, these wastes in produce are transferred to the
final customer (Orozco and Tun, 2012).
Moving agricultural products between different points in the country implies handling issues regarding
perishability of products, long and tortuous supply chains marked by the presence of several operators and no
alternate roads, the need to maintain a cold chain to guarantee the quality of the final product, consumption
behavior and habits, and the role that health aspects and organoleptic quality play in purchasing decisions of
consumers, among others (Estrada, 2007). According to Pietro and Timpanaro (2012), the cost the agricultural
logistics varies between 20-30% of the cost of the product. This can be even higher depending on the type of
chain involved, e.g the distance from origin and the type of transportation considered. Thus, it is important to
view the transportation and logistics system as a whole since the production, exchange, distribution and
consumption of agricultural products constitute the organic chain of agriculture reproduction. Any deficiency of
them will affect the development of agriculture (Tan, 2012, p. 106).
Moreover, it is important the study of distribution networks in order to address the different issues existing
between the diverse parties involved in the transportation and distribution systems of products. Daganzo (1992),
for example, establishes the principle of distribution network application with the goal of uniting one origin with
one destination, one origin to many destinations and many to many systems using transshipment centers and
providing methods to solve it. On the other hand, Agra (2008) demonstrated that the costs associated with the
transport of goods represent a large part of the final cost.
Several papers have been found in the literature concerning the modeling of the agri-food supply chain. Boudahri,
et al (2012), for example, presented a document concerned with the planning of a real agri-food supply chain for
chicken meat for the city of Tlemcen in Algeria. The agri-food supply chain network design is a critical planning
problem for reducing the cost of the chain. More precisely the problem is to redesign the existing supply chain
and to optimize the distribution planning. The authors applied the Allocation Problem Model in order to define
points in the network with the objective of minimizing the total distance between customers and these sites, or to
minimize the maximum distance.
Jones, et al (2001), on the other hand, consider a production-scheduling problem arising when there are random
yields and demands as well as two sequential production periods before demand occurs. The paper presented a
two-period model with random yield and random demand in which production can occur in either or both periods.
The model is solved optimally as a sequential decision problem and it demonstrated that the two-period
production strategy has substantial economic payoff for the seed industry.
Shu-quan and Ling (2010) focused the research in the multi-dimension and uncertainty of logistics performance
evaluation for agricultural products distribution centers and the lack of evaluation methods, The authors proposed
a hierarchy model of evaluation factors that combines fuzzy analytical hierarchy process (FAHP) with fuzzy
comprehensive evaluation to generate quantitative comprehensive evaluation of logistics performance for
agricultural products. They proved the rationality and application of this method through a practical case. Finally,
Jang and Klein (2011), developed models for agricultural supply chains based on the stochastic aspects of risk and
return on investment that small enterprises face, and suggest their uses and future considerations.
Related with the purpose of this paper, Mejia and Castro (2007) worked in the logistics optimization in a
Colombian frozen and refrigerated food company. The authors developed a decision model based on linear
programming to determine packing and distribution policies of frozen products. Zhang, et al (2011), on the other
hand, focused on the research of a distribution model and vehicle routing optimization of fresh agricultural
products. On the basic of detailed researching of agricultural products logistic characters, the paper establishes a
vehicle optimization model suitable for transferring kinds of perishable agricultural products, to solve the severe
11th Latin American and Caribbean Conference for Engineering and Technology
Cancun, Mexico

August 14-16, 2013

losing of fresh produce logistics with transportation distance. The model is solved by a genetic algorithm and the
algorithms effectiveness is verified using different examples.

4. MODEL DESCRIPTION AND SOLUTION

To develop the mathematical model, the Minimal Cost Network Flows (Bazaraa et al, 2005) approach was used,
considering the different elements involved in the network. Thus, points as production centers, distribution points
and final markets will be introduced, and a transshipment approach will be structured, and the optimal amount of
lettuce through the network will be determined. In addition, a minimal flow problem consisting on modeling the
optimal amount and types of vehicles used to deliver the lettuce will be included in the general model.
4.1 MODEL DESCRITPION
Consider a general network G = (V, A) where V is a vertex set representing either production centers, distribution
centers of final markets, and A a set of directed arcs connecting different points in the set V. Every arch A is
defined by the pair of indexes i, j indicating the origin and destination of such arch.
Let xi,j be the amount of products sent from point i to point j. In addition, consider y(k)i, j the type of vehicle k used
to transport products from point i to point j. Let ci,j, the cost of moving one unit of product and and b(k)i,j the cost
of moving vehicle type k from point i to point j. Moreover, consider distribution or transshipment points l that
define the transportation policy of the logistic system. The objective of the problem is to optimize the amount of
products sent from the origins to destinations and the optimum amount and type of vehicles used to move the
products, at a minimum cost.
Consider the following parameters:
Z
N(k)i
A(k)
Si
Dj
Wl
m
n
L
K

:
:
:
:
:
:
:
:
:
:

Total weekly cost of the transportation policy.

Amount of vehicles type k available at point i.
Capacity of vehicle type k in terms of unit loads.
Weekly supply of point i.
Weekly demand at point j.
Weekly capacity of the distribution or transshipment points.
Number of origins.
Number of destinations.
Number of distribution centers.
Vehicle types, in this case pick-ups, trucks and trailers.

The model is expressed bellow:

(1)
Subject to:

Weekly capacity of the sources:

(2)

The first constraint requires that the different supply points send no more than the production available at each of
them, thus the upper limit of the distribution policy is the maximum supply available at the different production
points.

Weekly demand of the destination points:

(3)

11th Latin American and Caribbean Conference for Engineering and Technology
Cancun, Mexico

August 14-16, 2013

For every destination point, the amount sent by the sources must be at least the demand required by each
destination point.

No inventory in the transshipment points:

(4)

Due to the perishability of the lettuce, no inventory will be allowed at the different origin, transshipment and
destination points. Thus, any amount sent from the origins to the transshipment points has to be sent to the
destination points.

Weekly capacity of the distribution points:

(5)
Each distribution or transshipment point has an specific capacity of storage that must be satisfied with every
shipment of products from the production point.

Weekly availability of vehicles:

(6)

The amount of every type of vehicle used to transport lettuce at any supply point (considering also the
transshipment points) must be less or equal to the available amount of vehicles at each of these points.

Weekly transportation capacity of the vehicles at every distribution point:

(7)

At every distribution point, the capacity of all the available vehicles must be at least the amount ready to be sent
to every destination point, thus vehicles are to be used only with this product.

All variables are integer and bounded by their upper limits.

(8)

4.2 MODEL SOLUTION METHOLOGY AND RESULTS

A feasible initial solution using Excel Solver was found. This approach was used to see the solvability of the
model, and if the preliminary solutions had a plausible answer.
The solution shown in the following tables depict the optimal solution of the problem divided in two sections:

The first section shows the amount of crates moved from the supply to demand points.
The second section shows the amount of vehicles used to comply with the distribution program.

Table 1 shows the distribution policy for the lettuce. As seen, the total demand is satisfied using all the supplies
from the different distribution points. This distribution policy takes into account the amount delivered to the
intermediate points, Santiago and Mercado that are then delivered to the final consumption points.

11th Latin American and Caribbean Conference for Engineering and Technology
Cancun, Mexico

August 14-16, 2013

Table 1: Crates Moved (xij)

David

Santiago

Chitr

Las
Tablas

770

2,000

1,980

Mercado

Santiago

750

200

2,000

2,750

750

2,000

2,750

750

From/To
Boquete
Cerro
Punta

Total
Demand
(Dj)

San
Miguelito

Panam
Este

Coln

5,500

6,270

6,500

3,200

7,180

8,000

1,200

1,400

600

1,000

4,200

5,500

250

350

1,550

3,200

200

250

350

1,200

8,700

1,400

600

1,000

19,200

23,300

200

250

350

1,200

8,700

1,400

600

1,000

Aguadulce

Penonom Chorrera Mercado

Supply
(Si)

Total

Tables 2 to 4 show the optimum amount of vehicles recommended to accomplish the distribution policy shown in
table 1. As seen, the amount of vehicles used satisfies the availability of the corresponding vehicles: pick-ups,
trucks and trailers. Further, the model recommends the use of large transport for longer routes rather than small
vehicles, taking advantage of the low unitary cost of transportation in large vehicles.
It is important to recall that, since the model is limited to one product, it considers that the vehicles are dedicated
only to transport one product, not to share space with other products.
Table 2: Pick-ups used
Available
Pickups
(Nki)

Chorrera

Mercado

San
Miguelito

Panam
Este

12

15

13

45

20

23

Chorrera

Mercado

San
Miguelito

Panam
Este

Santiago Chitr

Las
Aguadulce Penonom
Tablas

From/To

David

Boquete

Cerro
Punta

Mercado

Santiago

Total

Coln Total

Table 3: Trucks used

Santiago Chitr

Las
Aguadulce Penonom
Tablas

Coln Total

Available
Trucks
(Nki)

From/To

David

Boquete

Cerro
Punta

Mercado

12

12

Santiago

10

Total

15

36

Table 4: Trailers used

Available
Trailers
(Nki)

Chorrera

Mercado

San
Miguelito

Panam
Este

10

Santiago Chitr

Las
Aguadulce Penonom
Tablas

From/To

David

Boquete

Cerro
Punta

Mercado

Santiago

Total

Coln Total

11th Latin American and Caribbean Conference for Engineering and Technology
Cancun, Mexico

August 14-16, 2013

Finally, table 5 shows a summary of the optimal solution for the problem. Minimum cost for delivering 19,200
crates per week is $80,790. The distribution program uses 23 pick-ups, 36 trucks and 10 trailers at a total weekly
transportation cost of $6,190.
Table 5: Optimal Solution Summary
Objective Function $
80,790
For delivering 19,200 Lettuce Crates
per week
At a total weekly
Using
cost of:
23 Pick-ups
$ 415.00
36 Trucks
$ 1,825.00
10 Trailers
$ 3,950.00
*Solved using Solver with Simplex LP

5. CONCLUSIONS AND FUTURE WORK

The model provided a solution with a distribution policy consisting on both, amounts to be moved from origins to
destinations and vehicles to be used. All these variables are tied to costs, such that the result provides, in addition,
the minimum cost of the policy.
At his stage of the project several important conclusions can be drawn from this paper. First of all, the lack of
information on costs, routing, demands and supplies makes really difficult to gather valid information to
formulate and evaluate de model. Further, there are no congruence between data from the producers and official
institutions. Thus, it is very difficult to validate the results of the model. Finally, it is necessary for the different
organizations involved in the agro-food supply chain, to work in a more united manner since it is important to
maintain the supply chain efficient and effective for all, producers, suppliers, and final consumers.
Since this is the first stage of a more ambitious project, work is still pending. For example, the model has to be
feed with more current data, and more variables and constrains need to be tested in the model. For instance, to see
what might happen if time and more frequencies are added.
Furthermore, it is necessary to add more products, thus the problem becomes a multicommodity flows problem
(Bazaraa, et al, 2005) which increases the complexity of the problem adding a number of variables and constraints
proportional to the amount of products. In addition, it is necessary to include an additional objective since it is
important to maximize the value of the load in each transport, because the load in each transport has to be the
optimal combination of products. Henceforth, the problem becomes a multicriteria, multicommodity minimum
flow problem with equipment assignment.

6. AUTHORIZATION AND DISCLAIMER

Authors authorize LACCEI to publish the papers in the conference proceedings. Neither LACCEI nor the
editors are responsible either for the content or for the implications of what is expressed in the paper.

AKNOWLEDGEMENT
The project is supported by a grant from the National Secretary of Science, Technology and Innovation of
Panama. In addition it is supported with facilities and infrastructure by the Technological University of Panama
and The International Maritime University of Panama. Finally, it is important to acknowledge the support of Dr.
Mario Serrato from the Tec de Monterrey in reviewing of the model.
11th Latin American and Caribbean Conference for Engineering and Technology
Cancun, Mexico

August 14-16, 2013

REFERENCES
Boudahri, et al. (2012). Application of a Capacitated Centered Clustering Problem for Design of Agri-food
Supply Chain Network. IJCSI International Journal of Computer Science Issues, Vol. 9, No. 4, pp.16941814.
Orozco, A., and Tuon, G. (2012). Designing a Logistic Platform through the Optimization of the Distribution
Networks for the Agricultural Sector. Proceedings of the 5th. International Conference on Maritime
Transportation, Martinez de Oses, F. X., and Castells, M., editors. Universitat Politectnica de Catalunya, pp.
95-109,
Pietro, P., and Timpanaro, G. (2012). Ethics, Sustainability and Logistics in Agricultural and Agri-food
Economics Research. Italian Journal of Agronomy, Vol.7, No. 33, pp. 237 246.
Tan, D. (2012). Developing Agricultural Products Logistics in China from the Perspective of Green Supply
Chain. International Journal of Business and Management, Vol.7, No. 21, 106-112.
Daganzo, C. (1992). Logistics Systems Analysis. 1st. Edition, Springer-Verlag, USA.
Agra D., (2008), Localizacin de centros de intercambio modal y plataformas logsticas. Master's thesis,
Universitat Politecnica de Catalunya.
Estrada, M. (2007). Anlisis de estrategias eficientes en la logstica de distribucin de paquetera. Doctoral
Dissertation, Universitat Politecnica de Catalunya.
Jones, P., et al. (2001). Matching Supply and Demand: The Value of a Second Chance in Producing Hybrid Seed
Corn. Manufacturing and Service Operations Management, Vol. 3, No. 2, pp. 122-137.
Shu-quan, L., and Ling, L. (2010). Comprehensive Evaluation of Logistics Performance for Agricultural
Products Distribution Center. Proceedings of the 2010 2nd International Conference on E-Business and
Information Systems Security, May 4-7, Wuhan, China.
Jan, W., and Klein, K. (2011). Supply Chain Models for Small Agricultural Enterprises. Annals of Operations
Research, Vol. 190, No. 1, pp. 359-374.
Meja, G., and Castro, E. (2007) Logistics Optimization in a Colombian Frozen and Refrigerated Foods
Company. Revista de Ingeniera. Universidad de los Andes. Bogot, Colombia, No, 26, pp. 47-54.
Zhang, S., et al. (2011). Research of Fresh Agricultural Products Logistics Vehicle Optimization. International
Journal of Intelligent Information Processing, Vol.2, No. 2, 45-58..
Bazaraa, M., et al. (2005). Linear Programming and Network Flows. 3rd Edition, Wiley Interscience, New Jersey,
USA.

AUTHORIZATION AND DISCLAIMER

Authors authorize LACCEI to publish the paper in the conference proceedings. Neither LACCEI nor the editors
are responsible either for the content or for the implications of what is expressed in the paper.

11th Latin American and Caribbean Conference for Engineering and Technology
Cancun, Mexico

August 14-16, 2013