A flexible framework for supply chain modeling and simulation

flutheronioneyedΛογισμικό & κατασκευή λογ/κού

13 Δεκ 2013 (πριν από 3 χρόνια και 11 μήνες)

84 εμφανίσεις

1

ISIC 2012

A flexible framework for supply chain
modeling

and simulation


George Paes, Nagesh Phaniraj, Anil Kumar

DELMIA Solution Pvt. Ltd
,
J. P. Nagar, Bangalore
-

560 078, India

george_paes@3dplmsoftware.com
,
nagesh_phaniraj@3dplmsoftware.com
,
anil_kumar@@3dplmsoftware.com



Abstract

A supply chain is a network of entities
-

suppliers, manufacturers,
distributors and
retailers
,

that collectively converts raw materials into goods, delivering them to
customers. Due to the complex
ity

of a supply chain, simulation is a powerful tool to
study the
behavior
, assess efficiency

and

evaluate new solutions. This

paper proposes a
framework to model a supply chain
where each

supply chain entity is
modeled

as a
system, hierarchically composed of sub
-
systems to the required level of abstraction.
Each s
ystem
has
operations which determine
its
inputs, outputs, time and

capacity.
A
model

can be simulated to the desired level of detail

depending on decisions that are
needed
.
The framework
is

tested in DELMIA V6 Production System.
A sample of
experiments are described. Performance is evaluated using

system
utilization, through
-
put, order
fulfillment

time, inventory, collected during simulation. This framework
allows incremental
modeling

and easy modification of
system
structure and operations.


Keywords:

Simulation, Discrete
-
event,
Hierarchical
Modeling
,
Supply Chain


1.
Introduction

1.1. Supply chain

-

entities and functions

A supply chain is a network of facilities that perform the functions of sourcing of materials,
transformation of these materials into intermediate and finished products, distribution of
these finished products to customers and the return of defective or ex
cess products.
A
typical supply chain consists of a number of entities
-

starting with
suppliers,
provid
ing

raw
materials; producers
,

manufactur
ing

products
; distributors,
suppl
ying

finished goods
to
retailers and c
ustomers who

receive the good
s

from the r
etailers. There
may

be other
separate
entities
like the transporters, transport
ing

raw material
, intermediate products,
finished goods
and waste material b
etween the
other
entities. Alternatively, transportation
could be a function of suppliers,
distributors or even retail
er
s.

Similarly, warehouses may be
independent or part of another entity.

While

products primarily
flow from supplier
s

to manufactur
es, to
distributors, to retailers
,

to customers
, the
re is
also
a

reverse flow
,

of defective and
excess products.

Besides, there is a flow of waste for recycling and disposal. There is
also flow of orders and information and cash between entities.

According to
Vieira & Júnior

(2005)
,
t
he
increasing competition

and
technological innovation
in
manufacturing and communication
has forced
companies
to reconsider their production structure
.

Expanding

markets, both for suppliers
and
customers, has resulted in

competition among companies
giving
w
ay to competition
among supply chains.

2

ISIC 2012

1.2
Challenges

in Supply chain
design and operations

Cope
et al

(2007)

have
found

that
problems in
supply chain
are unique because they are
subject to u
ncertain
ty

and
h
igh
v
ariability,
d
ynamic
ally changing environment

and
physically d
istributed.

The
uncertainty
of one entity
interacts with another greatly affecting the overall
supply chain activities and performance.

It tends to be amplified at each stage upstream
through
the supply chain

giving

rise to the bullwhip effect
.

Lee

et al

(1997)

have

identified
that
, demand forecast updating, order batching, price fluctuation and shortage gaming
,
together with the rational decision
s

made at each entity
create the bullwhip effect.
It

results
in
cycles of excessive inventory and severe backlogs, poor product forecasts, unbalanced
capacities, poor customer service, uncertain production plans,
and
even lost sales
.

Understanding
its

causes

and coordination among entities
helps managers design and
develop strategies to counter it.

S
upply chain
s are

highly dynamic

due to changes in
its
entities over time
.

E
ntities
are added
,
removed

and
t
he entities change due to changes in
specification
s
,

workflow, or
technology.

Besides, the
social and economic environments
in

which the entities operate
undergo
changes.

S
upply chains
being

physically distributed, the information is
generated

and owned
by different entities.
W
hen mak
ing

decisions
for

the supply

chain as a unit, d
ata
may be

available but the
interpretation or
knowledge

maybe inaccurate.

Supply chains are complex, a change in an entity may have unexpected and
dispro
por
tional effect on the supply chain as a whole.

Due to
the

complexity,

an

improvement of
an

entit
y

may

not lead to improvement of the supply chain as a whole.
It
may result in an undesirable effect.
Therefore an effective supply chain management
requires a joint effort of the suppliers, manufacturers and distributors and r
etailers, with a
focus on processes across organizational boundaries.
(Olugu & Wong, 2009)

1.3 Modeling and
Simulation

of Supply Chains

S
imulation is one of the few tools that can capture
the stochastic, dynamic,
distributed
and
complex
nature in a realistic manner. It permits the evaluation of operating performance
prior to the implementation; it enables what
-
if analyses leadin
g to better planning decisions,
without interrupting the real system; it permits time compression so

that timely policy
decisions can be made.

D
esign
and operation
of a supply chain encompasses a set of decisions

that may be
broadly classified as
,
strategic,
tactical
and operational
.
Strategic decisions such as
facilities

location
,

suppli
er
selection
,
have long
-
term significance

and
impacts
most
entities

in the
supply chain.
The
y

change the structure of the supply chain
.

Tactical decisions
, mainly
related to
resource allocation
,
production
, inventory

and logistics
policies,

have a

med
ium
term (
from
a month to
a

year)

impact
.

They
impact

one or more

supply chain
entities.
Operational decisions are short term,
mostly

related to the day
-
to
-
day activities.
Their scope
is restricted to

individual area of the supply chain (e.g. pl
ant and warehouse)
.

S
imula
tion
models, can be used to support decision making both at the strategic level and the
operational level

Collaborative management
strategies to reduce
the bullwhip effect
,
involve
high
investments

and
require
behavior changes across
organizations.
Vieira
& Júnior

(2005)

suggest that s
imulation
can
help managers analy
ze

these

strategies and also
help
in the
administration.

According to Cope
et al

(2007)

simulation modeling provides the flexibility
3

ISIC 2012

to model processes and events to the desired level of complexity, in a risk free,
dyn
amic and stochastic environment
.
Bottani & Montanari

(2005)

observe that it
enables the

examination of the sensitivity of a system.

Jain
(2008)

has

attempt
ed to

identify
the

different approaches for building generic
supply chain simulation capability and provide a comparison among them
. S
upply chain
specific
modeling
language, sub
-
models of supply chain nodes, templates, and different
types of simulators
,
vary in their expertise required for building models the flexibility
to
model

configurations and operation
al policies
.

1.4
Review of

Simulation and Modeling


Real systems are much more
complex
compared to those modeled in
existing
studies,
hence
the contribution of such studies to the optimization of supply chain design is
quite limited

(Bottani & Montanari
, 2005)
.

Pundoor & Herrmann
(2006)

observe that
with increase in computing power
,
though
it has become easier to simulate
large and
complex systems
,
the time needed to develop the
simulation model
remains
quite high.
T
he features and
modules

in

s
tandard simulation
software

are at a very basic level compared to those
required

in supply chain simulation
models.
They suggest that d
eveloping hierarchical models with
l
ibraries of reusable sub
-
models supply chain models
can be built
with less time and effort
. Th
erefore they

approach
the problem by
build
ing

hierarchical modules based on the SCOR model

(Supply Chain
Council, Inc. (SCC), 2012)
.
They propose a three level modeling framework
. The first level
is the
supply chain
. The second level has sub
-
models correspond
ing

to
its
entities
. The third
level has sub
-
models that correspond to the process
es
of

entity
.

Planning
is done in

Excel
VBA

and
Arena is used for execution
.

P
erformance
is measured
using
cycle time, percent
tardiness, inventory, cost performance

(
based on job order costing
)
,

resource utilization
and
orders
.

They use this model to study on the impact of rescheduling frequency on the

performance of supply chains.

Vieira & Júnior
(2005)

find that the

greatest difficulty
in creating

a
supply chain
simulation model is the

required

level of detailing of each part of the
supply
chain
. They
also
have adopted a
hierarchical
model
ing

approach
. The first level is composed
of

the four
entities, suppliers, produce
rs, distributors and
customers
,
integrat
ed

by orders and
material/products flows.
Each entity is modeled at t
he second
level
. Detailed modeling of
entity's

functions is at the third
level.

T
he model allows

modeling of
safety stock

levels
,
ordering process,
with varying quantities, production

and delivery lead times, different types
of

products and their respective bill
-
of
-
mater
ials,

inventory replenishment policies, random
demand patterns

and demand forecast.

It

does

not consider

minimum purchase

& production

lot sizes; and functions such as for
ecasting, shop
-
floor scheduling and capacity planning.

Cope
(2008)

has
develop
ed

a tool that allows users to define a supply chain
simulation model using SCOR based ontologies. The ontology includes the supply chain
knowledge and the knowledge required to build a simulation mod
el of the supply chain
system. A simulation model is generated automatically from the ontology to provide the
flexibility to model at various levels of details
by
changing the model structure on the fly
.

Umeda & Zhang
(2010)

describe
a
hybrid framework
that combine
s

discrete
-
event

and system dynamics models. The discrete
-
event models represent operational processes
inside
the
supply chain, and the system dynamics models represent supply chain reactions
to
external factor
s, like customers' reaction to varying service levels.


4

ISIC 2012

2.
Modeling

and Simulation
Framework

2.1
Modeling

supply chain structure

In this proposed framework
,

supply chain entities

-

supplier
s
, producer
s
, distributor
s
,
retailer
s

and consumer
s

are

modeled

as
s
ystem
s.

They

are
connected by

directed
product
flow link
s
. The five basic types of systems
-

g
eneral
,
s
ource
,
s
ink
,
b
uffer
and
t
ransfer
,
available in
DELMIA
V6
Production System Simulation

can be used to
model entities
. A
system can
contain other systems, thus allowing a system to be hierarchically composed. A
system contains operations.
An

operation define
s

the required input products, the time, the
output products produced.

It also defines batch size, stock capacity and safety stock

for each
product in the system.
Time

may
be defined as a statistical distribution allowing
modeling
uncertainties
.

Suppliers are
modeled

as
s
ource
s
ystems.

Source operations specify the products
suppl
ied,
the batch size
and lead time
.
If a supplier also transports the products, a transport
system can be added to the supplier.
Customers are
modeled

as
s
ink
s
ystems
. The sink
operations specify the demand in terms of
p
roduct
,

the batch size

and

the

time between two
consecutive
batches
.

Uncertainty in supply and demand can be modeled by defining lead
time and inter
-
demand time

distributions
.

Retailers

are

modeled

as
b
uffer
system
s
. The
b
uffer
operations at the system specify
the products, the capacity and the safety stock.

Distributor
s
,

who transport products from
their warehouses to retailers,

are

modeled

as
g
eneral
system
s

containing
b
uffer
and
t
ransfer
systems.

Generally
, since

there is no product transformation at the retailers and distributors
,
they can be composed of
only
b
uffer
and
t
ransfer
systems
.
In case there is need to model
operations like inspection and re
-
packaging,
g
eneral
systems may be added with suitable
operations.

Producers are
modeled

as
g
eneral
systems. These may be composed of suitable
systems as necessary
-

the
raw material and finished goods inventory as buffer system
s
, the
production facility

that
transforms

raw material into products is
modeled

as a
g
eneral
system.
If need be, other systems for testing, packing or intermediate transformations

can be
added to the
production facility
.
If required,

systems for intermediate storages or
transportation

can also be modeled
. Additionally, operations may be sequenced, if it is
required to model
such
operations detail
s
.


Figure
1
: Higher level model


5

ISIC 2012


Figure
2
: Model

with details at second level

Similarly, other independent entities such as, warehouses and transporters can be
modeled. Recycling and waste disposal can also be easily modeled.

Interruptio
ns
in

normal
operations can be
modeled

by
defining

interrupt

operations
, for example repair or
maintenance
.
For
interrupt
operations
,

both
operation
time

and

time between two
consecutive interruptions

is defined
.


2.2
Supply chain operations

The
s
ink
operation
generates
demand

for products.
To satisfy

the demand

the
it

requests

the

upstream retailer

for the product(s). Products are routed
downstream

,
if available
, otherwise
the request
s

remains pending till a suitable product
s

are

available. If the
system

is connected
to multiple upstream

systems
, the request routing
parameters

a
re used to select the
upstream
system

to

satisfy the request.
Request r
outing
parameters

can be used to specify the routing
strategy
, for each type of product
. It may be
in
terms of

percentage, in order of priority or to
any system where the product is available.

Normally
,

retailer
s do not accept orders

and they order
according to
their stock

levels.

M
odeled as
buffer system
s
, they
route
available
product
s

when
customer
s

request

and

place orders to upstream
systems
w
hen
the

stock
of
a give
n

product becomes
less than
the safety stock
.

D
istributor
s normally accepts orders and transport products to the retailers. The
system

accepts the orders
from
downstream systems
as
and w
hen
they arrive. At the
beginning of each period, it executes its inventory and delivery planning methods. The
inventory planning method
calculates

the requirements
for

the period based on
orders

received during the earlier periods
and forecasted orders.
The delivery planning method
creates a delivery schedule for the transfer system.

P
roducer
s produce products from raw material

based on orders
. The producer
,

modeled

as a general system
,
composed of buffer
,

general and transfer system
s,

follow the
planning

method

similar to the distributor. It defines the production schedule, in addition to
the inventory and delivery plans.

The supplier

creates parts
with the given

batch sizes
.
It
operate
s

using either the

planning period

or lead time.
With planning period,

it accepts orders and
starts
supplies at
the
beginning of the

period.
With lead time
, it supplies
as an when it receives
orders after
the defined lead time.

6

ISIC 2012

A system
with a defined planning period
execute
s

its planning method
,

a
t the start of
the period
.

It
defines a schedule for each sub
-
system
, that is
in the 'scheduled' mode
.
The
other two operating modes available are, 'push' and 'pull'
A system
in

the push mode,
operates

based on available inputs, whereas in the pull mode it operates based on orders
f
rom the downstream systems.

Since entities like producers and distributors operate, based
on orders from downstream systems, they are in the pull mode. Their sub
-
systems, like
production and transport system
s

operate based on schedules, are in the schedule
d mode.

2.
3

Hierarchical
Simulat
ion

In this framework, the
system
's operations

model its
overall behavior, in terms of input,
outputs, time, batch and stock capacity
, whereas, it's

sub
-
systems model the
details
.


Th
us

the

framework facilitates hierarchical simulation, where the details contained in any system
can

be simulated
, if required
.
Simulation

with details, simulates its sub
-
systems, otherwise
only the operation
s

of the system are simulated
. Therefore
,

with the same model, it is
possible to simulate
at any
required level of
details to support
strategic
tactical or
operational decisions.


3. Experiments with framework

3.1 Model description

The structure of model used to test and experiment

with

the framework is similar to the one
in Figure 1

and 2
, except that each supplier also transports the products to the producer,
hence it
has
a transfer system in addition to a source system
.

The customers
demand two
types of products

at the rate of 720 pr
oducts
/
day
. Six type
s

of raw materials are required to
produce the products
, four of which are supplied by one supplier and one each by the other
two suppliers
.
P
roducts are transported
in
batch sizes of 10

and transportation time is 1 hour
.
The production

facility can produce 10 parts
/
hour.

The total raw material
stock
capacity is
6000 and with a safety stock of 1200.

The planning period is 1 day.

3.2 Experiments and results

The model
was
s
imulated for
25

days
, in each experiment.
Five
experiments and
results are
discussed below.

Aggregate results are shown in Figure 3.

1.

Experiment
: Model as described

above
.

Result
:
The supply chain meets only 25% of the demand. The utilization of the
production facility is above 75%.

2.

Experiment:

Therefore a second
production facility (same as the first) is added in
parallel to the first and connected to the raw material inventory on the upstream side
and to the finished good buffer on the downstream side.


Result
:
As expected
the supply chain now meets about 32% of demand, but
the
overall utilization of the production
facilities now decreases to 45%. It is observed
that the average raw material is low, about 28% of capacity
resulting in shortages at
the production facility for
about 35% of the time.

3.

Experiment
:
Therefore
supplier transport batch size

is double to 20 products per
batch

Result
:
Now the supply chain can meet about 40% of the demand.
But the
raw
material inventory

again increases to the levels it was in the first experiment.

7

ISIC 2012

4.

Experiment
:
Therefore
raw material stock capacity
reduced to
half

the original
capacity

Result
:
The average i
nventory reduces

while the supply chain still meets about 40%
of the demand. It is a
lso noted
producer's transport system

utilization
is high at about
78%.

5.

Experiment
:
Therefore the
producer's transport

batch size is now double to 20 per
batch.

Result
: This results in a
small
increase of 2% demand satisfaction

but the order
average order
lead
time significantly increases.



4. Conclusion and future work

This paper
proposes a
framework to model and simulate a supply chain

using
hierarchical

systems
.
It enables the
supply
to
be modeled and simulated
incrementally.
Starting from a
high level of abstraction
,
defining

the concept

and
structure
,

details can be added
as and
when
they

are available

to decide the operations

of the supply chain
.
A model

can also be
built
-
up by composing it of systems that have been modeled and tested

separately
.
The
framework,

facilitates
the
simulation
the model
to

the required
level of details.

Thus the
same model can be used for structural and operational decisions.
I
t captures
most common
behaviors of
product and order flow,
in
both stock and schedule based modes. It provides
the flexibility to model the
various supply chain
entities
as
close to their real

configuration.

4000
5000
6000
7000
8000
1
2
3
4
5
Throughput

350
375
400
425
450
1
2
3
4
5
Order Lead Time

500
600
700
800
900
1000
1
2
3
4
5
Inventory Level

40
50
60
70
80
1
2
3
4
5
Utilization

Figure
3
: Results of Experiments

8

ISIC 2012

Uncertainties and interruptions in normal opera
tions can also be modeled. The framework
allows customized planning and forecasting methods.

The frameworks needs to be extended to allow flow of cash
.

Modeling of social and
environmental factors which influence supply chain operations also need to be captured.
Given the size and complexity of real supply chains, know methods of simulation analysis
and optimization
limits the
model to

only a sub
-
set of
the decision variables
. Therefore a
similar hierarchical framework for analysis and optimization need
s to

explored

so that
simulation modeling

may be useful in supply chains decision making
.


References

Bottani, E., &

Montanari, R. (2005). Supply Chain Design: Guidelines from a Simulation
Approach. In A. G. (Ed.),
Discrete Event Simulations

(p. 63). InTech, August 2010.

Chang, Y., & Makatsoris, H. (2001, Vol.2, No.1). Supply Chain Modeling using Simulation.
Internation
al Journal of Simulation: Systems, Science and Technology

, pp. 24
-
30.

Cope, D. (2008).
Automatic generation of supply chain simulation models from SCOR based
ontologies.

Orlando, Florida.: Ph.D. Dissertation, University of Central Florida.

Cope, D., Fayez
, M. S., Mollaghasemi, M., & Kaylani, A. (2007). Sypply chain simulation
modeling made easy: An innovative approach.
Proceedings of the 2007 Winter Simulation
Conference

(pp. 1887
-

1896). S. G. Henderson, B. Biller, M.
-
H. Hsieh, J. Shortle, J. D.
Tew, and

R. R. Barton, eds.

Jain, S. (2008). Tradeoffs in building a generic Supply Chain simulation capability.
Proceedings of the 2008 Winter Simulation Conference

(pp. 1873
-

1881). S. J. Mason, R.
R. Hill, L. Mönch, O. Rose, T. Jefferson, J. W. Fowler.

Lee, H.

L., Padmanabhan, V., & Whang, S. (1997, Spring, Volume 38, Issue 3). The
Bullwhip Effect In Supply Chains.
Sloan Management Review

, pp. 93
-
102.

Olugu, E. U., & Wong, K. Y. (2009, Vol 2 1). Supply Chain Performance Evaluation:
Trends and Challenges.
Ameri
can Journal of Engineering and Applied Sciences

, pp. 202
-

211.

Pundoor, G., & Herrmann, J. W. (2006, Volume: 2 Issue: 3/4). A hierarchical approach to
supply chain simulation modelling using the Supply Chain Operations Reference model.
International Jour
nal of Simulation and Process Modelling

, p. 124.

Supply Chain Council, Inc. (SCC). (2012, January 5).
About SCOR
. Retrieved January 9,
2012, from Supply Chain Council: http://supply
-
chain.org/about/scor

Umeda, S., & Zhang, F. (2010). A simulation modeling

framework for supply chain system
analysis.
Proceedings of the 2010 Winter Simulation Conference

(pp. 2011
-

2022). B.
Johansson, S. Jain, J. Montoya
-
Torres, J. Hugan, and E. Yücesan, eds.

Vieira, G. E., & Júnior, O. C. (2005). A Conceptual Model for the
Creation of Supply Chain
Models.
Proceedings of the 2005 Winter Simulation Conference

(pp. 2619
-

2627). M. E.
Kuhl, N. M. Steiger, F. B. Armstrong, and J. A. Joines, eds.