HUMANITARIAN CRISIS: WHEN SUPPLY CHAINS
REALLY MATTER
RAFAEL CUERVO, FABIO DIAZ, ISABEL NAMEN, CRISTINA PALACIO &
CATHERINE SIERRA
UNIVERSIDAD DE LOS ANDES
DEPARTAMENTO DE INGENIERIA INDUSTRIAL
Calle 1 No 18210. Bogota, Colombia
Email: fa-diaz@ uniandes.edu.co, rs.cuervo97@ uniandes.edu.co
mi.namen36@ uniandes.edu.co, mc.palacio145@ uniandes.edu.co,
c.sierral 36@ uniandes.edu.co
“Delivering the right supplies to the right people, in the right place, at the right time and in
the right quantities” Davidson
ABSTRACT
The supply chain of humanitarian aid is a complex and interlinked network in which
different actors, processes, decisions and information are mixed to serve the needs of the
victims in a catastrophe. As aid can be vital, also the correct management of the supply
chain in each one of its stages is important not only to support the reconstruction efforts,
but to allow a correct and pertinent assistance of the needy population. In a disaster,
arrivals, transportation, storage and delivery of necessary goods become a difficult task
that requires the mobilization of a great amount of resources. The supply chain is therefore
stretched to their limits; that is why, it is so important to understand its behavior, the
handling of shelter, food, water, and health care (among other necessities). The main
objective of this paper is to model and describe the different phases comprising the supply
chain of food aid in a humanitarian crisis using the system dynamics methodology,
describing the particular case of the National Relief Agency of the Colombian Red Cross.
Keywords: Supply Chain, humanitarian aid, system dynamics.
1. Introduction
A natural disaster can occur anywhere at any time, and all countries in the world must be
prepared in order to minimize the eventual damage of the catastrophe. Therefore, the
effective management of aid determines the amount of lives that can be saved.
The process of delivering supplies to a community that suffered a natural disaster implies
the coordination and execution of multiple processes. Due to the importance of aid, its
relevance in the event of a catastrophe (Gustavsson 2003), and an adequate understanding
of how humanitarian aid is handled by donors, governments, international, local NGO’s,
community based organizations (local partners), support teams and finally beneficiaries, is
vital.
According to the type of disaster (earthquake, flood, hurricane, etc), the specific
requirements of the affected population vary. For example, when a flood occurs, the lack of
food is the major concem; but when an earthquake occurs, shelter is one of the immediate
requirements. The specific conditions of the place where the disaster occurs also determine
the kind of assistance needed. (Wilches-Chaux 2000)
Although all types of catastrophes are different, they share similarities on their
consequences: destruction of roads, health diseases, and destruction of infrastructure.
(Oloruntoba 2006) This is why an effective coordination of the supply chain and a constant
flow of information is the key to improve the delivery and the performance of organizations
of aid. (Cruz Roja Colombiana 2009)
This paper emphasizes on the importance of the supply chain in the effort to provide food
assistance to the needy ones, considering how different actors interact across the supply
chain to face the needs of the population, considering the case of the Colombian Red Cross.
2. Problem Definition
Supply chains are complex. Systems composed by series of stages in which materials and
information flow through different steps to fulfill the requirements of the recipients and the
objectives to those who send help (Davidson 2006). Profit and non-profit organizations
have different types of supply chains; most of them are measured under defined
performance parameters and directed to achieve certain objectives.
2.1 The humanitarian supply chain: stages, processes and inefficiencies.
Food aid implies the recollection and delivery of large amounts of goods from different
sources: people who want to help sending aid to those affected, governments, NGO's and
private enterprises.
The initial flow of material reflects the immediate reaction of actors according to the
information transmitted. (Organizacion Panamericana de la Salud 2001) As time passes,
supply efforts are reduced because donor’s perception of the necessities decreases and with
it, the amount of products delivered. According to this behavior, a humanitarian supply
chain needs to incorporate the recollection, transportation, classification and delivery of the
products from the donors to the affected people, and all the information processes that
coordinate each stage with the others. (Gray 2006)
The humanitarian supply chain for food assistance in the Colombian Red Cross begins once
a disaster takes place. The national entity in charge (National Relief Agency of the
Colombian Red Cross) elaborates a detailed evaluation of the disaster, and suggests the
most suitable action procedure according to the stocks of available products and logistic
teams in the place of the disaster. Then it verifies if the local teams request for assistance
can be received, processed, and delivered according to their logistic resources.
If the local unit is unable to answer to these necessities, the second level of assistance is
“activated”, the regional and strategic units are contacted to provide aid. If stock levels in
these units are not still enough to provide needs of the affected population, the principal
unit of assistance (National Unit) is alerted (involving the request for aid from the Regional
Unit). Depending on the inventory and the amount of food requested by the Regional Unit,
the National Unit determines if it is necessary to buy aid kits, or to ask for donations.
Emergency assistance supply chains share some similarities to “conventional” chains, both
consider as an indicator the lack of coordination (management of information along the
process) and the inefficient procedures that lead to long delays on the process of delivering
aid (Davidson 2006). So, the structure and coordination of the supply chain conditions the
success or failure of the effort. (Paché 2009)
On the literature review we have found that several efforts have been made to standardize
the process of delivering humanitarian aid ( Cruz Roja Colombiana Socorro Nacional
1998). However, some authors have correctly pointed that extreme standardization might
be a mistake, because all disasters are different, not only on their causes, but also on the
particular characteristics of the region in which they occur (geographical or cultural) and
this should be taken into account at the time of delivering aid. (Paché 2009)
A humanitarian supply chain is characterized by the presence of different kind of delays.
These occur during three stages of the aid process. In the evaluation of the disaster and the
measurement of the quantity of people affected, then during the process of delivery of aid
including the logistics needed to transport the aid for the place where donations are
received to the location of the disaster, and the process of relief and distribution of aid to
the recipients. (Cruz Roja Colombiana y Holandesa s.f.)
2.2 Is system dynamics helpful to understand the problem?
System Dynamics can be used to analyze different stages of the Supply Chain in order to
define what could happen under different scenarios, in example if aid were not enough to
supply the “demand”, and therefore understand possible side effects or delayed
consequences of different policies implemented (J. D. Sterman 2000).
Particularly, it is important to understand the so called “bullwhip effect’ were distorted
information from one end of a supply chain to the other can lead to remarkable
inefficiencies (Hau L. Lee 1997).
In the case of the aid supply chain, inventory moves up in the chain and it fluctuates more
with donations recollected and in the distribution centers creating a distortion on the
demand information.
To study this phenomenon, system dynamics presents an endogenous approach, in which
the behavior of the system can be described as a consequence of their underlying structure
(J. Sterman 1991), and also makes a distinction on the variables of the model (among
instantaneous variables and accumulated variables) (J. Forrester 2003) and the modeling of
feedback effects on it related to actors and variables of the system, allowing the modeling
of the structure, the decision processes involved and the flow of information (Richardson
2009).
System dynamics is useful to model this situation because the interactions among the
agents (1) People affected by the disaster 2) Members of the aid organizations 3) Donors of
aid) on the chain are complex, and a model of stocks and flows could elicit how aid (or
information) passes among them and affects the performance and effectiveness of the
chain.
The model is built to represent the flow of aid, and how it is affected by the information
feedbacks, that explain the presence of bullwhips effects on the chain at different periods of
time depending on the initial conditions (initial stocks of aid storage on the different parts
of the chain) and on exogenous variables.
The goal of this paper is not to calculate the precise amount of aid that should be sent at
every period of time or to calculate the optimum quantity of capacity that should be used,
because as it is said before, it depends on the particular characteristics of a disaster and of
the region in which it takes place. This paper wants to assess, through system dynamics, a
preliminary model to favor a better comprehension and planning prior to the problem, to
understand its complexity, in order to provide a tool that supports the strategic planning of
emergency response on the National Relief A gency of the Red Cross of Colombia.
3. Model
After a disaster or an emergency, it is necessary to deliver the required aid, and in our
model, we assess the delivery of food aid kits,! because for The National Red Cross is food
this stock is the most important on the supply chain , it is being modeled specifically the
flow of food represented in kits. First, the Red Cross evaluates the damages and needs of
the population, and then it determines how many kits of food they will deliver to the
population. The main operation center of the National Relief A gency of the Colombian Red
Cross is located in Bogota, where the main warehouse is located. Colombia is divided in 32
departments, and the Red Cross has located one regional warehouse in each of these
departments. In addition, it has local warehouses, located in the towns in which the Red
Cross has presence. They also have “strategic” warehouses, located in some departments of
the country that are used when the aid in regional centers is not enough in the face of an
emergency. There are four main variables in the model:
- The requirements of food by people affected by the emergency or disaster.
- The delivery of kits of food available in regional and local warehouses.
- The delivery of kits of food available in the strategic warehouse and in the main
national warehouse.
- The donations of food that are received by the Red Cross before and after a call for
aid has been made.
The model is divided in two main sectors, the orders and the flow of material.
a) The orders made and received by each of the different warehouses of the Colombian
Red Cross (regional, strategic, local and national). For example, based on the
community requirements, the local warehouses decide if they ask resources to the
regional warehouses or not. (Figure 1)
Figure 1: Orders made and received by the Red Cross
b) The material flow through the supply chain. In this case, the material flows in the
opposite direction of the information. The aid that is stored in the national
warehouse or in the strategic warehouses, including the aid received in donations, is
distributed to the regional centers, and when all the aid required is gathered in this
? Food aid kit: Represents de quantity of food that supplies the needs of a family of five persons
during two weeks.
point (because in aid delivery processes it is better to assist all the population in one
intervention) then is delivered the aid to the local centers. (Figure 2)
Figure 2: Material flow in the supply chain
3.1 Orders
3.1.1. Orders made by local warehouses based on initial requirements
The model assumes that the initial requirements are given at the beginning of the
simulation. This means that only an initial requirement is given, and then the aid chains
answers to that initial requirement, which represents the result obtained through the
evaluation of damages and needs made by the National Relief Agency of the Colombian
Red Cross. Initially, the Red Cross has previously positioned resources in its different
warehouses, because many emergencies in Colombia, such as floods or droughts, for
example, occur in determined seasons and these can be foreseen but not predicted.
The local center compares this requirement, measured in kits of food kits, with the
resources they have stocked in their warehouses. Based on this, they determine how much
food kits they will ask (in the model this is defined every 24 hours).
It is assumed that the distribution of food kits occurs almost immediately, and this is made
according to the food available at the local warehouses. The local warehouses will deliver
aid if there are food requirements. (Figure 3)
food requirement
po bey
food alain reduced
food that is delivered
deficit in local
warehouses
how often is deficit revised?
new food requirements
2; Food available in (
_, local centers initial food requirement
Figure 3: Initial requirements and deficit of local warehouses
3.1.2 Orders made and received by regional warehouses
The deficit that was previously mentioned is the amount that the local centers ask to the
regional centers. This value is adjusted by some percentages: it is possible that the regional
center would not be able to commit to deliver the whole amount of aid because the
resources must also be used to cover other emergencies. They can also send an additional
percentage based on the possibility to have some losses or damages in the food kits during
their delivery, guaranteeing the supply of aid required.
In this case, the orders received by the regional centers are compared with the amount of
food available in the regional warehouse, and a deficit is also calculated. This deficit
represents the amount that is ordered to the next stage, the national center.
The aid delivery from the regional centers considers that is not possible to deliver resources
continuously from one place to another, and it is preferred to make one or few deliveries
with great amount of food kits , based on efficiency in logistics, costs, and the difficulties
of making many deliveries. This is why a minimum lot size is determined, and based on
this; the delivery is made when the amount of food available in the warehouses reaches this
amount. In an extreme case, this lot size could be the order itself, making only one delivery.
The same rules apply to the strategic warehouses and national warehouse. This is just an
overview of how the model works; the detailed equations are found in the annex presented
at the end of this document. (Figure 4)
Extra percentage to
orders in regional centers
ask to regional centers
d orders by
minimum lot size rpaeiel centers
at regional center
amount that Fan be sent
from Regionbl Centers
percentage of aid
delivered by the red cross
deficit in regional warehouses
deficit in local
warehouses
regional cfnters
A how often is deficit revised?
©) | Food available in
local centers
Figure 4: Orders made and received by regional warehouses
3.1.3 Orders received by national and strategic warehouses
The national center receives the orders from the regional center, which correspond to the
deficit calculated in the previous stage.
The national center has two ways of attending an order (Figure 5):
- Use the strategic warehouse, authorizing the regional centers to use these resources
that have been previously positioned in these warehouses.
- Send kits from the national warehouse located in the capital. There is priority in
delivering aid first from the strategic warehouses, and then using the national
warehouse, because these ones (strategic warehouses) are usually closer to the place
where the emergency occurred, and therefore costs and transport times are reduced.
orders in national center
extra percentage to
ask to national centers
servg orders by
‘a naffonal centers
orders received by
minimum lot size at
national warehouse
served ordety by
strategic warehd\ses
amount tafe served
by the natjonly centers amount that cane sent
from strategic Caters
(=
minimum lot size at ‘a
2 at strategic centers e
Food on National Warehouse
gap of inventory to be covered
by national warehouses
Figure 5: Orders received by national and strategic warehouses
3.2 Delivery of aid
The second sector of the model describes the flow of aid through the supply chain. The
material flows from national to regional warehouses answering the need to distribute the
aid.
3.2.1 Local delivery of aid
Locally, the food kits are delivered immediately as mentioned before. But the model
assumes that the resources do not arrive immediately from the regional to the local centers,
and there is a time to transport the resources. There is a delay in the delivery of resources.
The material flow depends only of the orders made in each of the stages. No material
passes from one stage to the next one, unless this is done to deliver an order. Additionally,
it is considered the fact that the food in the warehouses can only last for a certain time, so
there is a small and constant loss of resources in the warehouses. (Figure 6)
peseéntage of loss
pergéntage of loss in local centers {cod requirenent
‘regional centers
served ortlers by =
regional cAnters loss jf local centers
loss in regional cenders
Food bvailable j
regioffal centey
food that is delivered
food arrives from
regional to local centers
Figure 6: Local delivery of aid
3.2.2 Regional delivery of aid
Food kits are also delivered from the national and strategic centers, depending on the
amount of orders that must be served, considering the transport times between one
warehouse and the other, and the possible loss of food in the warehouses. Food kits from
strategic warehouses can be sent to regional warehouses or directly to the local warehouses.
(Figures 7 and 8)
Food available in
local centers
e ~ @
food arrives from
regional to local centers
strategical to
regional 1 or lofal centers 0?
food arrives rom the ~
| centers =
strategic to the lo
time from
strategic td regional centers
food gies from the strategic
fo the regional centers
my transport time from
strategical to local centers|
served orders by
strategic warehouses
served orders by
strategic warehouses
loss in strategic centers
(
percentage of loss
in strategic centers
Figure 7: Delivery of aid from strategic to local or regional warehouses
served orders by
national cehters
Food available in
Food on National Warehouse regional centers
2
2
food arrives from the national
to the rgional center
transport time from
national to regional centers
Figure 8: Delivery of aid from national to regional warehouses
3.2.3 Food purchasing and donations
There are certain cases in which the food stored in the national or strategic warehouses is
not enough, and one of these two policies must be implemented.
- Buy food kits, a process that takes a certain period of time, before it is delivered.
The model allows the user to define this amount.
10
- Collect resources through donations. Initially, it is known that the national
warehouse receive a flow of donations constantly, but a call for aid can be made to
all the national population (this is usually made when the disaster requires a big
amount for aid, in example, an earthquake). This option is presented in the model,
and a great amount of food can be obtained with the help of donors (from 5 to 20
tons every day). Classifying and processing donations takes certain time, so they
can be finally distributed. (Figure 9)
(a
is there afcall for aid (w
time of pulehasing
in National Center
orders in national center
food in process to be purchase
new arrivals of food New food available
that is purchased in National Center
Food on Natiffnal Warehouse
Q
food donations ale processed
‘s percentage of loss in
loss in national ware
ieanegua national warehouse
(=) in national center
donations received
aeuee seal donations received
Folld donations in
proces in national center
Figure 9: Food purchasing and donations
The model considers that some donations are not useful, and when donations are classified
these are rejected and eliminated. (Figure 10)
11
Food donations in
process in national center
T =
rejected donations in
the national center
peseeftage of rejection
in the national center
donations to be eliminated
in the national center
( Gonations are eliminated
in the national center
Figure 10: Rejection and elimination of donations
4. Model validation
In order to assess the validity of the model we analyzed the structure and parameters of the
model (J. D. Sterman 2000), focusing primarily on the consistence between the purpose and
the model itself, and on the consistency of the structure of the model to “mimic” reality,
according to the members of the National Relief A gency of the red Cross.
To analyze the structure of the model, a question must be made: To what extent is the
structure of the model representing the real system? The model represents the process of
distribution of food aid kits. The orders are made from local to national warehouses, and
material flow follows the opposite direction: from national to local warehouses. The model
considers transport times from one warehouse to another, and incorporates situations in
which a call for aid is made to population when an emergency tums into a disaster.
However, there are certain things that the model differs from real life.
The inventory model structure is vital, therefore when we checked the model we considered
that orders that cannot be fulfilled could not imply an accumulation on the stocks of
inventory, nevertheless the backorders cause a reaction on the chain, such as an increase on
the orders for aid. Additionally the chain comprise several stages that imply accumulations
and delays related to the different steps of the supply chain (and therefore different
processes and delays), on the model there are several simplification on the stages, and the
simplification of the chain is done by dividing it into the stages explained in the model, but
a detailed structure of it can be considered. In the case the distinction mentioned before
12
implies changes in the behavior of the system, the model should be re-structured to asses
this particularities.
éHow are backorders considered?
In this model, if an order cannot be served by a certain warehouse, the “smaller warehouse”
orders the deficit to the bigger warehouse up in the chain,. For example, suppose that the
regional warehouse receives an order of 200 kits, and has only 100 kits of food available in
its stock. Therefore, it must order 100 kits to the national or strategic or national warehouse.
Why does the model work then? Every certain amount of time, the local warehouses
compare their amount of stock with the food requirements of the population. This is a short
time, so constantly orders are generated by the local warehouse. If the food that the local
warehouses asked for were not received, they will consider this in the next order, and the
needs of the population will be finally covered.
How is deficit in local warehouses calculated?
For the model to work property, the local warehouses compare their stock with the
population needs every 24 hours. However, there is a difference between the model and the
real system. The local warehouses make orders only considering their levels of stock. They
do not consider the stocks of other warehouses, and food that was asked before to cover
population needs. The model works differently, because if a certain amount of food is not
received, a new order would be made to ask for the deficit.
Are warehouses correctly represented in the model?
The model has four main stocks: Food in local, regional, strategic, and national
warehouses. However, the model does not considers cases in which more than one of each
kind of these warehouses must be considered in the model. These cases occur when two
different strategic warehouses can be used to serve the population, when an emergency
affects a population but nearby towns can also contribute to deal with the emergency, or
when these neighbors are also affected by the emergency. In this case, what should be
represented by the stocks in the model? One solution is to add the capacities of each kind of
warehouse, to obtain the amount of food represented in the stock. The assumption here
would be that food delivered by stocks of the same kind is delivered exactly at the same
time, and they take the same transport time to reach their destiny. This may not be realistic
because distances between warehouses may differ, and some stocks may be more efficient
than others.
In the case mentioned before, in which two different strategic warehouses can be used to
serve a regional warehouse, the main variables to consider is the time that it takes to serve
13
the warehouse and the amount of food stored in each in warehouse. But this case cannot be
analyzed in our model, unless an extension is made.
At the moment of assessing the model, there is also necessary to discuss the pertinence of
the boundaries of the model, and the way model answers the question of how it changes
when assumptions are relaxed.
The model assumed that the population requirements for aid are a pulse caused by the
natural disaster, and the relief provided by aid.
As can be seen in the document, the model has been done in conjunction with The Red
Cross, We have done a weekly a meeting with their members in order to understand and
validate the process that we wanted to represent and to know important details such as data
and basically reserved information that was necessary to model the supply chain.
Sensitivity Analysis
To understand the behavior of the model a base model is considered. This case supposes
that a disaster has occurred, and 50,000 food kits are required. The warehouses were not
prepared for such a disaster. Therefore, they rely on donations from the population, which
must be distributed immediately.
The exogenous variables in the model are initialized, and their values are shown in an
annex 1 presented at the end of the paper.
The following results were obtained.
@ 1: food requirement 2: deficit in local warehouses
4 5000047-1- _ SS
2; aT
du oe
ay |. al ia ——
4 25000 =
2] 0. 2 La 2
1.00 60.75 120.50 180.25 240.00
Page 1 Hours 15:05 jue, 12 de ago de 2010
Lm?
14
Graph 1: Food requirement and deficit in local warehouses.
There is an initial requirement of 50.000 food kits. The deficit in local warehouses is
calculated every 24 hours, which explains the peaks obtained in the graph each day . It can
be seen also that food requirerments decrease as orders are attended and new supplies
arrive.
@ 1: Food available in local centers
lL 5000:
aH 2500:
i of Lh, 1: 1 43 | Lj
1.00 60.75 120.50 180.25 240.00
Page 1 Hours 15:05 jue, 12 de ago de 2010
a2 ?
Graph 2: Food available in local centers
The peaks in this graph are obtained because the local centers receive food kits from
regional centers and these ones arrive in lots that take a certain time to transit.
15
@ 1: Food available in regional centers
1 90004
a 45004
—_ . PL.
1
1.00 60.75, 120.50 180.25 240.00
Hours 15:05 jue, 12 de ago de 2010
Graph 3: Food available in regional centers
@ 1: Food on National Warehouse
1: 400004 {
|
|
|
|
|
|
|
|
|
ci 200004
| ih
|
|
|
|
|
7 |
|
eal 2
a nee £ T
1.00 60.75, 120.50 180.25 240.00
Page 1 Hours 15:05 jue, 12 de ago de 2010
Graph 4: Food available in National Warehouse
After the disaster has been attended, donations are still being received. Unless the call for
aid is interrupted, the amount of food aid kits stocked on the national warehouse will grow
indefinitely. The resi
could be ignored if this last assumption is made.
ults obtained after reducing food requirements to zero in this graph
What would happen if a call of aid was not made?
The variable “Is there a call for aid” represents a binary variable that takes the value of 1 if
a call of aid is made to the population, and 0 if the Red Cross considers that this is not
necessary. The disaster requires the call of aid to be made, because donations are an
16
important source of aid. The following graphs shows what would happpen in the absence of
the extra donations received after a call for aid.
Variable Value Units
Is there a call for aid? 0 Binary
@ 1: food requirement 2: deficit in local warehouses
1 | 50000—7-1- 1 1= ie
a:
2 2500044
2
2: 0. 26 2 2 2 1
1.00 60.75, 120.50 180.25 240.00
Page 1 Hours 15:08 jue, 12 de ago de 2010
aa ?
Graph 1A: Food requirement and deficit in local warehouses
@ 1: Food available in local centers
1 1004-1
1 aml 4
1.00 60.75 120.50 180.25 240.00
Page 1 Hours 15:08 jue, 12 de ago de 2010
Graph 2A: Food available in local centers
17
@ 1: Food available in regional centers
1 2004
1 100-71 ——
1
i
1 0 1.
1.00 60.75 120.50 180.25 240.00
Page 1 Hours 15:08 jue, 12 de ago de 2010
Graph 3A: Food available in regional centers
@® 1: Food on National Warehouse
i 2004
1
it
1 100--1-—~—. 1
i
|
|
1 0 —_|
1.00 60.75 120.50 180.25 240.00
Page 1 Hours 15:08 jue, 12 de ago de 2010
Graph 4A: Food available in National Warehouse
Graphs show that the call of aid was necessary for the relief of the disaster. Otherwise, the
food requirements could not be reduced, as shown on graph 1A. Graphs 2A, 3A, and 4A,.
Food requirements are finally not reduced during the next ten days.
What would happen if transport times are reduced?
In the following case transport times among the different warehouses are reduced:
18
Variable Value Units
Transport time to regional to local centers 6 Hours
Transport time from strategic to local centers 4 Hours
Transport time from regional to local centers 3 Hours
Transport time from national to regional centers | 6 Binary
It is expected that food requirements will decrease faster, because food kits take less time
between warehouses. Buto also we could expct a higher volatility on the intermediate
stocks of the chain.
@ 1: food requirement 2: deficit in local warehouses
| Ce
2:
=o 250004-4-
1
3) 0 2s Lame 2
1.00 60.75, 120.50 180.25 240.00
Page 1 Hours 15:12 jue, 12 de ago de 2010
aa ?
Graph 1B: Food requirement and deficit of local warehouses
19
@ 1: Food available in local centers
1 300004
| -o
a 150004
1.00 60.75 120.50 180.25 240.00
Hours 15:12 jue, 12 de ago de 2010
Graph 2B: Food available in local centers
@ 1: Food available in regional centers
L 5000%
1 25004
1 —— 1 J 1
1.00 60.75 120.50 180.25 240.00
Page 1 Hours 15:12 jue, 12 de ago de 2010
Graph 3B: Food available in regional centers
@ 1: Food on National Warehouse
L 20000 oT
fe 700 —— ————————— le
. be AAAAAAAAL
st
1.00 60.75 120.50 180.25 240.00
Page 1 Hours 15:12 jue, 12 de ago de 2010
Graph 4B: Food available in National Warehouse
The main differences in the results show that the food requirements are covered in less
time. Food is delivered quickly, and with a higher frequency. The results in graph 2B show
that if transport times are short, extra orders could be made because local warehouses are
not considering the aid in transport-order before making a new order. Therefore, in the end,
the food available in the regional centers grow indefinetly bacause aid is stocked and not
distributed anymore (it is not necessary to deliver aid that is not required)
What would happen if deficit is revised less often in local centers?
Variable Value Units
How often is deficit revised? | 120 hours
21
@ 1: food requirement 2: deficit in local warehouses
ch | S00 ap ereteererrernerys eee
2
st 250004 |
2 — TO 7 —
1
2 | 0 2 _- 2 Lee 24
1.00 60.75 120.50 180.25 240.00
Hours 15:18 jue, 12 de ago de 2010
Graph 1C: Food requirement and deficit in local warehouses
@ 1: Food available in local centers
1 300004
1 150004
+ 1-———__—|
1 oF 1s EM 1
1.00 60.75 120.50 180.25 240.00
Hours 15:18 jue, 12 de ago de 2010
Graph 2C: Food available in local centers
22
@ 1: Food available in regional centers
1 300004
|
1 150004 t
|
1.
1 it
120.50 180.25 240.00
Page 1 Hours 15:18 jue, 12 de ago de 2010
asf ?
Graph 3C: Food available in regional centers
@ 1: Food on National Warehouse
1 40000
1 2000044
2 y
a
1 ol
1.00 60.75 120.50 180.25 240.00
Page 1 Hours 15:18 jue, 12 de ago de 2010
Graph 4C: Food in National Warehouse
In this case, orders are made less often- every 120 hours. Therefore, less deliveries are also
made, and this explains why there are less peaks in each of the graphs. Food requirements
are reduced almost at the same time because donations are received at the same rate. Lot
sizes would be bigger: Less orders are made but in larder quantities.
23
What would happen if the amount of donations is reduced?
Variable Value
Units
Donations received after call of aid 6000/24
food kits
@ 1: food requirement 2: deficit in local warehouses
4) 500004-1- — '
2 “KI |
+1 im |
| |
|
1 ——
2: 250004 |
re
1: N [\
3] 0 2. 2 —2. a) es ee
1.00 60.75 120.50 180.25 240.00
Page 1 Hours 15:23 jue, 12 de ago de 2010
Graph 1D: Food requirements and deficit in local warehouses
@ 1: Food available in local centers
1 60004
1 30004
' L, i me ios ll t |
1.00 60.75 120.50 180.25 240.00
Page 1 Hours 15:23 jue, 12 de ago de 2010
Graph 2D: Food available in local centers
24
@ 1: Food available in regional centers
1 60004
|
1 30004 t
|
1 oie 1 1 1
Page 1
aer
1.00 60.75 120.50 180.25 240.00
Hours 15:23 jue, 12 de ago de 2010
Graph 3D: Food available in regional centers
@ 1: Food on National Warehouse
1 40004
a 200044
a
Page 1
aa
Saal
1.00 60.75 120.50 180.25 240.00
Hours 15:23 jue, 12 de ago de 2010
Graph 4D: Food available in National Warehouse
The graphs show that when the amount of donations given by the population is reduced,
food requirements persist for a longer time (60 hours in the following example).Food in the
warehouses follow the same behavior. The different peaks are given by the time between
orders made by the local warehouses.
The structure analysis was made with the Colombian Red Cross, institution that supported
our work and validated the obtained results with real life data. Data is treated as
deterministic in the model, and therefore, the model does not take into account the
variability of most data. Deecisions that are taken in the chain depend on population needs,
25
and therefore the vaues that are taken as exogenous, can change constanty. Nevertheless,
the behavior of inventory through the different stages of the chain is correctly represented,
and the understanding of the humanitarian supply chain in the Colombian case was
achieved, which was one of the main goals of this work.
Conclusions
The correct management of the supply and the efficiency in each one of its stages is
important not only to support the reconstruction efforts, but to allow a correct and
pertinent assistance of the needy population.
Although all types of catastrophes are different and the conditions of the place where
they occur determine the kind of assistance needed, they share some similarities related
to the immediate effects that are originated. The particular case of chain that is analyzed
in this paper represents the general behavior of a supply chain.
System dynamics was useful to model this situation because it shows the interactions
that represent a complex system that was modeled using differential equations (stocks,
flows and their respective delays).
The model results correspond to the behavior that was expected according to the
relation of the actors, and the mobilization of the resources (i.e. arrivals, transportation,
storage and delivery of necessary goods). However, some break points and
inefficiencies that arise in each of the stages were identified (i.e. the distortion of the
demand because of the information delay), as in the bullwhip effect.
Related concepts such as inventory, and its relation with infrastructure capacity were
incorporated to approach the supply chain. Also, the behavior of the system changes
when the assumptions presented were implemented (i.e. given the fact that the supply
chain is directly related with an emergency, not necessarily a cost efficient perspective
defines the assignment of resources).
According to the data and the information recollected, the model structure reflects the
paper purpose; water and food are satisfied locally for a certain period of time, then
donations are asked, and the humanitarian chain tries to satisfy water and food needs.
Population needs are satisfied until there is no need to ask for donations, and local
production is enough to supply the population.
26
For institutions like the Colombian Red Cross, the implementation of the supply chain
in a model, and its respective results (variables, graphs, and break points) is really
useful for the planning, of policies and actions towards the prevention and the correct
performance of the supply chain in all kinds of disasters. A dditionally it is important to
remark the necessity to built the model together with the actor involved, in order to
make the model more operational and similar to reality; on the other hand this
involvement will help to built a common language towards the model so actors, as well
as modelers can understand how are they actually working, and what would happen if
they improve some of the critical variables of the process.
27
Annex 1: Initial values of the model
Variable Value Units
Initial food requirement 500 food kits
How often is deficit revised? 24 hours
Food available in local centers 100 food kits
Food available in regional centers 100 food kits
Food available in national centers 100 food kits
Food in national warehouse 100 food kits
% of aid delivered by the Red Cross 100% adimensional
Extra % to ask to regional centers 10% adimensional
Extra % to ask to national centers 10% adimensional
Minimum lot size at regional center 5 food kits
Minimum lot size at strategic center 5 food kits
Minimum lot size at national warehouse 5 food kits
Percentage of loss in local centers 0,1% Adimensional
Percentage of loss in regional centers 0,1% Adimensional
Percentage of loss in strategic centers 0,1% Adimensional
Percentage of loss in national warehouse 0,1% Adimensional
Transport time from regional to local centers 12 Hours
Transport time from strategic to local centers 8 Hours
Transport time from regional to local centers 6 Hours
Transport time from national to regional centers 12 Hours
28
Is there a call for aid? 1 Binary
How much is purchased? 10 Hours
Donations received 1.6/24 food kits
Donations received after call of aid 10000/24 | food kits
Time of purchasing in national center 24 Hours
Processing time in national center 24 Hours
Percentage of rejection in national center 10% Adimensional
Food goes from strategical to regional or local centers? 1 Binary
Annex 2 : Equations of the model
29
ORDERS
CO food_requirement(t) = food_requirementt(t - dt) + (new_food_requirements -
food_requirements_are_reduced) * dt
INIT food_requirement = 0
INFLOWS:
= new_food_requirements = pulse(initial_food_requirement,5,10000)
OUTFLOWS:
= food_requirements_are_reduced = food_that_is_delivered
(© orders_in_national_center(t) = orders_in_national_center(t- dt) +
(orders_received_by_National_Centers - served_orders_by_strategic_warehouses -
served_orders_by_national_centers) * dt
INIT orders_in_national_center = 0
INFLOWS:
=» orders_received_by_National_Centers =
deficit_in_regional_warehouses*(1+extra_percentage_to_ask_to_national_centers)
OUTFLOWS:
< served_orders_by_strategic_warehouses =
amount_that_can_be_sent_from_strategic_Centers-mod(amount_that_can_be_sent_from_str
ategic_Centers,minimum_lot_size_at_at_strategic_centers)
<p served_orders_by_national_centers
amount_to_be_served_by_the_national_centers-mod(amount_to_be_served_by_the_national_
centers,minimum_lot_size_at_national_warehouse)
(2) orders_in_regional_centers(t) = orders_in_regional_centers(t- dt) +
(orders_received_by_regional_centers - served_orders_by_regional_centers) * dt
INIT orders_in_regional_centers = 0
INFLOWS:
= orders_received_by_regional_centers =
deficit_in_local__warehouses*(1+extra_percentage_to_ask_to_regional_centers)*percentage_
of_aid_delivered_by_the_red_cross
OUTFLOWS:
=p served_orders_by_regional_centers =
amount_that_can_be_sent_from_Regional_Centers-mod(amount_that_can_be_sent_from_Re
gional_Centers,minimum_lot_size_at_regional_center)
30
Oo. 0
000000 000 0 0
amount_that_can_be_sent_from_Regional_Centers =
min(Food_available_in_regional_centers,orders_in_regional_centers)
min(Food_available_in_strategic_centers,orders_in_national_center)
amount_to_be_served_by_the_national_centers =
if(gap_of_inventory_to_be_covered_by_national_warehouses=0)then(min(Food_on_National_Wareho
use,gap_of_inventory_to_be_covered_by_national_warehouses))else(0)
deficit_in_local__warehouses = if(mod(time,how_often_is_deficit_revised?)=0) then
max(food_requirement-Food_available_in__local_centers,0) else (0)
deficit_in_regional_warehouses =
max(orders_received_by_regional_centers-Food_available_in_regional_centers,0)
extra_percentage_to_ask_to_national_centers = 0.1
extra_percentage_to_ask_to_regional_centers = 0.1
gap_of_inventory_to_be_covered_by_national_warehouses =
max(orders_in_national_center-Food_available_in_strategic_centers,0)
how_often_is_deficit_revised? = 24
initial_food_requirement = 50000
minimum_lot_size_at_at_strategic_centers = 5
minimum_lot_size_at_national_warehouse = 5
minimum_lot_size_at_regional_center = 5
percentage_of_aid_delivered_by_the_red_cross = 100/100
31
SUPPLY CHAIN
(J) donations_to_be_eliminated_in_the_national_center(t) =
donations_to_be_eliminated_in_the_national_center(t - dt) +
(rejected_donations_in_the_national_center - donations_are_eliminated_in_the_national_center)* dt
INIT donations_to_be_eliminated_in_the_national_center =
INFLOWS:
=e rejected_donations_in_the_national_center =
Food_donations_in_process_in_national_center*percentage_of_rejection_in_the_national_cen
ter
OUTFLOWS:
=<» donations_are_eliminated_in_the_national_center =
donations_to_be_eliminated_in_the_national_centeritime_of_elimination_in_the_national_cent
er
(2 Food_available_in_regional_centers(t) = Food_available_in_regional_centersi(t - dt) +
(food_arrives_from_the_national_to_the_regional_center +
food_arrives_from_the_strategic_to_the_regional_centers -
food_arrives_from_regional_to_local_centers - loss_in_regional_centers) * dt
INIT Food_available_in_regional_centers = 100
INFLOWS:
=~ food_arrives_from_the_national_to_the_regional_center =
delay(served_orders_by_national_centers transport_time_from_national_to_regional_centers)
= food_arrives_from_the_strategic_to_the_regional_centers =
delay(served_orders_by_strategic_warehouses .transport_time_from_strategic_to_regional_ce
nters,0)*food_goes_from_strategical_to_regional_1_or_local_centers_0?
OUTFLOWS:
= food_arrives_from_regional_to_local_centers =
delay(served_orders_by_regional_centers,transport_time_from_regional_to_local_centers)
=<» loss_in_regional_centers =
percentage_of_loss_in_regional_centers*Food_available_in_regional_centers
(2 Food_available_in_strategic_centers(t) = Food_available_in_strategic_centers(t- dt) + (-
food_arrives_from_the_strategic_to_the_regional_centers - loss_in_strategic_centers -
food_arrives_from_the_strategic_to_the_local_centers)* dt
INIT Food_available_in_strategic_centers = 100
OUTFLOWS:
=» food_arrives_from_the_strategic_to_the_regional_centers =
delay(served_orders_by_strategic_warehouses.,transport_time_from_strategic_to_regional_ce
nters,0)*food_goes_from_strategical_to_regional_1_or_local_centers_0?
=» loss_in_strategic_centers =
Food_available_in_strategic_centers*percentage_of_loss_in_strategic_centers
=<» food_arrives_from_the_strategic_to_the_local_centers =
delay(served_orders_by_strategic_warehouses.transport_time_from_strategical_to_local_cent
ers,0)*food_goes_from_strategical_to_regional_1_or_local_centers_0?
32
( Food_available_in__local_centers(t) = Food_available_in__local_centers(t- dt) +
(food_arrives_from_regional_to_local_centers +
food_arrives_from_the_strategic_to_the_local_centers - food_that_is_delivered -
loss_in_local_centers) * dt
INIT Food_available_in__local_centers = 100
INFLOWS:
=» food_arrives_from_regional_to_local_centers =
delay(served_orders_by_regional_centers transport_time_from_regional_to_local_centers)
=~ food_arrives_from_the_strategic_to_the_local_centers =
delay(served_orders_by_strategic_warehouses , transport_time_from_strategical_to_local_cent
ers,0)*food_goes_from_strategical_to_regional_1_or_local_centers_0?
OUTFLOWS:
=» food_that_is_delivered = if(Food_available_in__local_centers>=food_requirement)then
food_requirement else Food_available_in__local_centers
= loss_in_local_centers =
Food_available_in__local_centers*percentage_of_loss_in_local_centers
(2 Food_donations_in_process_in_national_center(t) =
Food_donations_in_process_in_national_center(t - dt) + (donations_are_received_in_national_center -
food_donations_are_processed - rejected_donations_in_the_national_center) * dt
INIT Food_donations_in_process_in_national_center = 0
INFLOWS:
=p donations_are_received_in_national_center = iffis_there_a_call_for_aid>0) then
donations_received_after_call_of_aid else 0+donations_received
OUTFLOWS:
=e food_donations_are_processed =
(delay(donations_are_received_in_national_center,processing_time_in_national_center,0))
=< rejected_donations_in_the_national_center =
Food_donations_in_process_in_national_center*percentage_of_rejection_in_the_national_cen
ter
(J food_in_process_to_be_purchased(t) = food_in_process_to_be_purchased(t - dt) +
(new_arrivals_of_food_that_is_purchased - New_food_available_in_National_Center) * dt
INIT food_in_process_to_be_purchased = 0
INFLOWS:
=p new_arrivals_of_food_that_is_purchased = if(orders_in_national_center=0) then
how_much__is_purchased? else 0
OUTFLOWS:
= New_food_available_in_National_Center =
delay(new_arrivals_of_food_that_is_purchased,time_of_purchasing_in_National_Center,0)
33
| Food_on_National_Warehouse(t) =F Food_on_National_Warehouse(t -dt)+
lofelerololelerelelelelelelelerere)
(food_donations_are_processed + New_food_available_in_National_Center -
food_arrives_from_the_national_to_the_regional_center - loss_in_national_warehpuse) * dt
INIT Food_on_National_Warehouse = 100
INFLOWS:
<p food_donations_are_processed =
(delay(donations_are_received_in_national_center,processing_time_in_national_center,0))
=e New_food_available_in_National_Center =
delay(new_arrivals_of_food_that_is_purchased,time_of_purchasing_in_National_Center,0)
OUTFLOWS:
=~ food_arrives_from_the_national_to_the_regional_center =
delay(served_orders_by_national_centers,transport_time_from_national_to_regional_centers)
=e loss_in_national_warehpuse =
Food_on_National_Warehouse*percentage_of_loss_in_national_warehouse
donations_received = (40/24/25
donations_received_after_call_of_aid = 10000/24
food_goes_from_strategical_to_regional_1_or_local_centers_0? = 1
how_much__is_purchased? = 10
is_there_a_call_for_aid = 1
percentage_of_loss_in_local_centers = 0.001
percentage_of_loss_in_national_warehouse = 0.001
percentage_of_loss_in_regional_centers = 0.001
percentage_of_loss_in_strategic_centers = 0.001
percentage_of_rejection_in_the_national_center = 0.1
processing_time_in_national_center = 24
time_of_elimination_in_the_national_center = 48
time_of_purchasing_in_National_Center = 24
transport_time_from_national_to_regional_centers = 12
transport_time_from_regional_to_local_centers = 6
transport_time_from_strategical_to_local_centers = 8
transport_time_from_strategic_to_regional_centers = 12
34
REFERENCES
Anderson, Mary. Humanitarian NGOs in Conflict Intervention. Washington: United States
Institute of Peace Press, 1996.
Colombiana, Cruz Roja. Logistica en La CRC Unidad de Servicios Logisticos para la
Asistencia Humanitaria. 2009.
Davidson, Anne Leslie. Key Performance Indicators in Humanitarian Logistics.
Massachusetts: Massachussets Institute of Technology, 2006.
Forrester, Jay. « Dynamic models of economic systems and industrial organizations.»
System Dynamics Review, 19, 4 , 2003: 329-345.
Forrester, Jay W. Counterintuitive Behavior of Social Systems. Springer Netherlands, 2004.
Gray, Richard. «Humanitarian aid: an agile supply chain?» Supply Chain Management: An
International J ournal, 2006.
Gustavsson, Lars. «Humanitarian Logistics: Context and challeges.» FMR No.18, 2003: 6-
8.
Hau L. Lee, V. Padmanabhan, Seungjin Whang. The Bulwhip Effect in Supply Chains.
Massachussetts: MITSloan Management Review, 1997.
Holandesa, Cruz Roja Colombiana y. Taller ENI Relief: Trabajo coordinado, Relief y
Logistica.
Nacional, Cruz Roja Colombiana Socorro. Maual de Logistica en Operaciones. Modulo
3800, 1998.
Oloruntoba, Richard. «Humanitarian Aid: an agile supply chain.» Supply chain
Mangement: An International Journal, 2006.
Organizacion Panamericana de la Salud, Organizacion Mundial de la. Logistica y gestion
de suministros en el Sector Salud. Washington D.C., 2001.
Organizacion panamericana de la Salud, Organizacion mundial de la Salud. Logistica y
gestion de suministros en el Sector Salud. Washington D.C., 2001.
35
Paché, Jerome Chandes y Gilles. «To Ponder on the Collective Actions in the Context of
Humanitarian Logistics: Lessons from the Earthquake in Pisco.» Journal of Economics,
Finance & Administrative Science, 2009.
Richardson, George. «System Dynamics, The Basic Elements of.» Encyclopedia of
Complexity and Systems Science, 2009: 8967-8974.
Sterman, John. «A Skeptic's Guide to Computer Models.» In Barney, G. O. et al. (eds.),
Managing a Nation: The Microcomputer Software Catalog. Boulder, CO: Westview Press,
1991: 209-229.
Sterman, John D. Business Dynamics: Systems Thinking and Modeling for a Complex
World. Boston, MA: Mc. Graw Hill., 2000.
Ulrich, Werner. «A Primer to Critical Systems Heuristics for Action Researchers.» Hull:
Centre for Systems Studies, 1996.
Wilches-Chaux, Gustavo. El Terremoto de Armenia: Los primeros meses vistos un afo
después. CRID Centro Regional de Informacion sobre Desastres para América Latina y el
Caribe, 2000.
36
