0% found this document useful (0 votes)
44 views17 pagesM&S Chapter 2
Modeling is the process of creating a simplified representation of a system to analyze its behavior and the effects of changes. It involves techniques such as abstraction and idealization, and can be classified into various types including dynamic, static, deterministic, and stochastic models. The model development lifecycle includes stages such as problem formulation, conceptual model development, validation, and simulation analysis.
Uploaded by
oluwatomisin oyaniyiCopyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
0% found this document useful (0 votes)
44 views17 pagesM&S Chapter 2
Modeling is the process of creating a simplified representation of a system to analyze its behavior and the effects of changes. It involves techniques such as abstraction and idealization, and can be classified into various types including dynamic, static, deterministic, and stochastic models. The model development lifecycle includes stages such as problem formulation, conceptual model development, validation, and simulation analysis.
Uploaded by
oluwatomisin oyaniyiCopyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
/ 17
Chapte,
e
Z
MODELLING | @
2.0 INTRODUCTION
Modeling is the process of producing a model; a model is the
fon of the construction and working of a particular
One purpose of modeling is to enable the analyst
: effect that certain changes will have on a system. In
going, a model should be a close approximation of
nd consequently incorporate most of its salient
's ely, it should not be too complex to the extent it
mes rather impossible to either understand or experiment with
of fact, a good model is expected to be a careful
eoff between realism and simplicity. Experts recommend that
the complexity of model should be increased iteratively.
‘A model is a set of assumptions about the operation of the
system. These assumptions can be algorithmic (sequence of steps),
snathematical (equations), logical (conditions). This model can be
run (simulated) in order to study the real system. There are two
main techniques for building models:
@ Abstraction: this means leaving out unnecessary details. Here,
only selected attributes of a customer are represented.
e Idealization: this means replacing real things by concepts, Here,
sets of measurements are replaced by a function,
fication of a real system. Modelling is the process
Amodel isa simpli i
ecific tool to study its behavior.
of representing a system with a sp
e@ Amodel can be:
> Analytic: when a
Queuing Model)
mathematical approach is feasible (¢
18
Scanned with CamScanner » Simulation: model used for complex s
» Experimental: when a model of the real
Amodelisa pattern, plan, representation (es speciz
or description designed to show the main object or w
object, system, or concept
» Model may also refer to;
> Abst » and theories
y representations of objects
> human and animal behavior
> occupations
> history and culture
> lighting
> In geography
actions, concep!
In general, modelling is used for systems with
uncertainty:
© Waiting time in a restaurant/airport
Time to walk from the hostel to the classroom
Response time and Throughput of a web server
* The productivity of manufacturing systems
¢ Design of multi-processor machine
* Performance of MAC protocols (e.g. CSMA/CA)
2.1 PRACTICAL SITUATIONS
a. Industrial problem: this includes the design of queuing syste:
communication networks, inventory control and chemical
Processes.
. Business and economic problems: this includes consumer
behaviour price determination, economic forecasting, and total
firm operation
A and social problems: this includes population
dynamics, environmental health effects, epidemiological studies
and group behaviour
Scanned with CamScanner 20 — General Introduction (0 Modelling and Simulation
+ om: this i les fluid balar
id Reliavioutal syesme va eel roliferatic, ety
distribution in human body, bloo P ‘on, andy,
activities.
e. War strategies and tactics
Example: Movement
} Consider a system when a given
object move
This system canbe modelled by the
equation
Where:
> Sis the distance run through
> Vis the speed of the object
> tis the time that has been observed,
e This simplifies the real world phenomenon
¢ Another model can take into account the direction o
movement, or the three dimension Coordinate
e Itis therefore a case of studying the behavi
based on a specific model 'viour of the systen
2.2 CHARACTERISTICS OF MODELS
i, A model is never equal to the rea
simpler than the reality
ii, The accuracy of a model is determi
approach the real system
Is (ii) above a problem?
> Yes, if the model ignore important
system (over simplification)
> No, ifthe model takes into account the important
(ignoring some details is sometimes Not probe, matigy tS
IC)
Scanned with CamScanner
I system because it is always
ined by its tendency to
iti.
Parameters of the real a Modelling 21
2.3 SIMULATION MODEL VERSUS ANALYTICAL, MODEL,
e Simulation is not used when a suitable mathematical model
exists
e Simulations are
@ Simulation only produce
e Simulation can take
© Mathematical models ar
efficient
often complex error:
“Prone pieces of software
Approximate answers
ALONG time to execute
¢ less flexible.
» but they are exact and
The problem is: what model better represents the real world?
2.4 MODEL CLASSIFICATION
Dynamic Models
© Represents a system as it
evolves over time
© Example: Cars arriving to a
parking
Static (Monte Carlo) Models
© Time plays no role
© Represents the system at
a particular point in time
Usually referred to as Monté
Carlo Method/Simulation
Deterministic Models
Stochastic Models
° No probabilistic/random
variable or component in
the system
Example: Worst-Case Analysis
of the system
Example: Queuing systems
{Continuous Models
© The state variables of the
system change continuously
Over time.
Example: chemical processes
~ here, the status of the system
is changing continuously over
time. Such systems are usually
modeled using differential
equations.
© Some components of the system
has a probabilistic behavior
(Random variable, event
probability)
Discrete Models
© The state of the system changes
only at discrete or countable
points in time.
A bank is an example of a
discrete system, since the state
variables change only when a
customer arrives or when a
customer finishes being served
and departs. These changes take
place at discrete points in time.
Scanned with CamScanner groduction to Mo
2 General In
nd continuous simulation models ar
Js are called discrete
istic models. Such model
Js, Diserete-event simulation concerns thy
em as it evolves over time by 4
¢ variables change only at discretg
te-event simulation is:
Discrete 4
ete stocha
ation model
a stochastic
in which st
of discret
tod
event simul
modeling of
representation
points in time. Example
server queuing system. Customers arrive into this
population and either go into service
rver is idle or join the waiting queue jf
Jes of this kind of system are:
terminal
ne checkout counter
e Asingle-
system from some P
immediately if the se
the server is busy. Examp!
o Asingle ticket counter at a bus
o Asmall grocery store with only 01
o Aone-person barber shop.
A more in-depth discussion of the types of simulation
modelswillbe made in chapter four of this book.
2.5 MODEL DEVELOPMENT LIFECYCLE
Develop Conceptual Model
Develop Specification Model
Develop Computational Model
Validation
Fl ;
IGURE 2.1 Model Development Lifecycle - a fundamentally erat? gio!
Scanned with CamScanner Modelling — yi
1, Problem formulation
acstitement of the prof
thos sinulation analyst
mus aire that the problem is clearly
understood, Wa problem statement in prepared by the
simulation analyst, itis important that the client understand
nd agree with the formulation, 44 is suggested tha t of
wnplions be prepared by the simulation and agreed
toby the client. Even with all of these precautions, it posable
that the problem will need to be reformulated sa
study progresse:
very simulation oludy beging with
dem. HW the SLANT is provided by
that have the problem (client), the s
ke extreme cate to eny
ithe simulation
2, Determine Goals and Obje
sntific study, including
explicit definition of the objectiv
= the initial stage of
ion task requires an
this should include
questions to be answered, the hypothe
le
ited, and the
Without a clear picture of the
ance of a successful completion
and implem ion is largely undermined. Here, issues
pertaining to performan iteria are highlighted as well as
model parameters and the identification and definition of the
state variables,
alternatives to be coi
problem to be solved, t}
It is very likely that the initial formulation of the problem will
undergo many modifications as the study proceeds and as we leam
more about the situation being studied. However, a clear initial
statement of the objectives is essential.
In a nutshell, this stage captures the following:
a. What do you want to do with the model?
i, It may be an end in itself
ii, More often, it is a means to an end
b. Goals may not be known when you start the project!
©. One often learns things along the way
»
Develop Conceptual Model :
a. An abstract representation of the system. It is recommended
that modeling begin simply and that the model grow until
Scanned with CamScanner auction to Modelling and Simulation
a model of appropriate complexity has been developed
consider the model of a manufacturing ang
tem. The basic model with the arrivals,
Then, add the failure:
erial-handling
For example,
material handling svs
quenes and servers is constructed.
and shift schedul ext, add the m
capabilities, Finally, add the special features. Constructiny
an unduly complex model will add to the cost of the study
and the time for its completion without increasing th,
quality of the output. -
i, What should be included in model? What can be lef
out?
ii, What abstractions should be used? What is the level
details? :
b. Appropriate choice depends on the purpose of the mode
Develop Specification Model
a. Amore detailed specification of the model including mor.
specifics
b. Collect data to populate model
i, Example:
o Traffic: Road geometry, signal timing, expected traf,
demand, driver behavior
o Communication: network topology, message type
inter-arrival time, data rates
ii. Empirical data or probability distributions often used,
5. Develop a Computational Model - this usually involvs
developing a computer program for the model. One of the ke;
decisions here is the choice of the language. Note that special
purpose languages require less programming than the general
purpose languages but are less flexible and tend to requir
long computer running times. In either case, the.programminy
part of the study is likely to be a time-consuming proces
because simulation programs tend to be long and complex.
summary, this stage captures:
a. Executable simulation model
Scanned with CamScanner — Modelling 9.
Tt
p. Software approach
i, General purpose programming language
ii, Special purpose simulation language
¢. Other (non-functional) requirements
i. Performance
ii, Interoperability with other models/tools/data
6, Verification - solving the problem right. This is the
process of assuring that the implemented model accurately
represents the conceptual model. This concerns the operational
model. Is it performing properly? It is highly advisable that
verification take place as a continuing process. It is not wise
for the simulation analyst to wait until the entire model is
complete to begin the verification process.
a. Did I Build the Model Right?
b. Does the computational model match the specification
model?
c. Debugging: checking if the program contains any
programming errors.
d. Once the program has been developed and debugged, we
determine whether the program is working properly.
Inotherwords, is the program doing what it is supposed to
do?
e. This verification step is usually difficult because for most
simulations, there won't be ready-made results to compare
with the computer output.
£ Verification is different from Validation: (see model
validation)!
7. Validation - solving the right problem. This is the process
of assuring that the conceptual model accurately represents
the behaviour of the real system. Can the model be substituted
for the purposes of experimentation? If there is an existing
system (base system) then an ideal way to validate the model
is to compare its output with that of the base system.
Unfortunately, there isn't always a base system.
Scanned with CamScanner —————
P
simulation
26 General Introduction to Modelling and oo
Right Model? this i critical part of th.
Tere, we UY
the syste
mine if the mode
pid 1 Build the
nalyzed and
steps.
presents
is are reliable,
to det
m being
simulation
realistically 1h
whether the resul
p. Ifat this stage we are Sa
model, we can then use t
yswer the questions at
iments Must be
hr the performance of th,
conduct experiment
| The data generate
collected, pro
stisfied wit
he model to
hance
so as 10 aT
py the simulation expe
and analyzed
_ Does the computational model match the actual (9;
envisioned) em?
a. Typically.the vé
by comparing
j. Measurements of actual system
ji, An analytic (mathematical) model of the system
iii, Another simulation model
e. Results are analyzed not only as the solutions to the mos:
put also in terms of statistical reliability and vali
‘Afterwards, decision must be made whether to perform ax
additional experiments.
f, Asamatter of necessity, model validation techniques includ:
simulating the model under known input conditions a:
comparing model output with system output.
. By necessity, validation is always an incomplete activity!
h. Often can only validate portions of the model
i, Ifyoucan validate the simulation with 100% certainty
build the simulation? :
wht
ae DESIGNING A SIMULATION EXPERIMENT
simulation experiment connotes a test or a series of tests in wh
Bani :
eaningful changes are made to the input variables ofa simula!
int
i canes
: Bes as ro observe and identify the reasons for changes
iis ae The number of experiments in a simulat®
er than or equal to the number of questions Le
asked
ed about the model (e.g., Is there a significant aiffere™
il
Scanned with CamScanner “ Yr
petween the mean Alelay MM communication networks 1 and 2?
which network has the least delay: 1, 2, or 32 How will a new
routing algorithm affect the performance of network 32). The
design o mulation experiment involves answering the question:
what data ts needed, in what form, and how much? The following
steps illustrate the processes associated with designing a simulation
experiment.
step 1 - Select appropriate experimental design. Select a
performance measure, a few input variables that are likely to
influence it, and the levels of each input variable.
Step 2 - Establish experimental conditions for runs.
Address the question of obtaining accurate information and the
most information from each run. Determine if the system is
stationary (performance measure does not change over time) or
non-stationary (performance measure changes over time).
Generally, in stationary systems, steady-state behavior of the
response variable is of interest. Determine whether a terminating
ora nonterminating simulation run is appropriate. Select the run
length. Select appropriate starting conditions (e.g., empty and idle,
six customers in queue at time 0). Select the length of the warm-
up period, if required. Decide the number of independent runs -
each run uses a different random number stream and the same
starting conditions - by considering output data sample size. Sample
size must be large enough (at least 3-5 runs for each configuration)
to provide the required confidence in the performance measure
estimates. Alternately, use common random numbers to compare
alternative configurations by using a separate random number
stream for each sampling process in a configuration. Identify output
data most likely to be correlated.
Scanned with CamScanner 28 General ntroauetion to Modeling ard Simulation
£8 __ General Introduction to Mole ~
Jerform runs acco
Step 3 - Perform simulation runs: perform rity
above.
steps
Problem formulation
fan OCS
als a
bxtine
Tava Conve Mee
on Medel
‘Simulation Runs and Analysis
ec
[amas
FIGURE 2.2 Model Development Lifecycle (Expanded) - a fundamer!:
iterative process
2.7 HOWTO PERFORM SIMULATION ANALYSIS ‘
stan
Most simulation packages provide run statistics (mean,
deviation, minimum value, maximum value) on the per!
measures, e.g., Wait time (non-time persistent statistic), inve™
on hand (time persistent statistic). Let the mean wait time it
M/M/1 queue observed from n runs be W1, W2, - Wa +
important to understand that the mean wait time W is 4 ‘al
erformet
—_
Scanned with CamScanner Modelling — 29
ple and the objective of output an
mean of W and to quantify its variabilit
‘alysis is to estimate the true
¥. Notwithstanding the facts
shat there are no data collection errors in simulation, the underlying
model is fully known, and replications and configurations are user
controlled, simulation results are difficult to interpret. An
observation may be due to system characteristics or just a random
occurrence. Normally, statistical inference can assess the
nificance of an observed phenomenon, but most statistical
nference techniques assume independent, identically distributed
(iid) data. Most types of simulation data are auto-correlated, and
pence, do not satisfy this assumption. Analysis of simulation output
data consists of the following steps.
varial
Step 1. Interpret and present results. Compute numerical
estimates (e.g., mean, confidence intervals) of the desired
performance measure for each configuration of interest. To obtain
confidence intervals for the mean of auto-correlated data, the
technique of batch means can be used. In batch means, original
contiguous data set from a run is replaced with a smaller data set
containing the means of contiguous batches of original
observations. The assumption that batch means are independent
may not always be true; increasing total sample size and increasing
the batch length may help. Test hypotheses about system
performance. Construct graphical displays (e.g., pie charts,
histograms) of the output data. Document results and conclusions.
Step 2. Recommend further course of action. This may include
further experiments to increase the precision and reduce the bias
of estimators, to perform sensitivity analyses, etc.
2.8 SCENERIO
Simulate a busy restaurant's seating policy. Assume that the
Testaurant does not take reservations (although adding reservations
to the simulation might be interesting). When customers arrive,
they check in at the main podium and give their name, number of
People in the party, and smoking preference. Some will go to the
bar and may or may not choose to eat there instead of waiting for
atable. Others will simply sit in benches, waiting for their party to
Scanned with CamScanner duction to Mot
30 General Intr
ated, they should have some ranq,
(or you might want to simulate the eating proces,
1- ordering salad, entree, dessert etc.), before leayi,
Tables and seats in the bar would ‘be resources here. What j
hows up? Can some tables (resources) be push
party of 20 sl :
together to accommodate large parties?
be called. When @ party is s
eating tim!
more detai
2.8.1 Goals and Objectives a
The main goal and objective is to Minimize Customers Wai
time through optimization of the seating policy using the follow
performance metrics:
© Maximum number of customers: No of customers ¢
can occupy all tables and chairs.
© Average queuing time: Time spent waiting for a par
be called.
© Average Service time (Eating and Ordering): Tin
takes for a customer to be answered, time a customer take
order and eat (Considered to be random).
© Maximum number of customers the queue cant;
vac :
is row
4
aeseuron
FIGURE 2.3 Work Flow Diagram
oe Conceptual Model
: 5
ae nea model is the result of the data gathering'
aie ormulation in one's mind. It is also an oo
presentation of the system i.e. What should be included"
Scanned with CamScanner Modelling 31
@ Customers
Entities utilizing the resources (tables) to make orders and
eat.
© Queue
» Buffers holding customers waitin:
benches)
e Server (tables)
Resources that are serially reused
customers at a time
ig for tables(bar and
5 Serves one group of
e The System corresponds to M/M/C/N Queuing Model.
e@ The M/M/C/N queue is a multi-server queue model. It is a
generalization of the M/M/1/N queue. It indicates a system
where:
Arrivals are a Poisson process (M)
Service time is exponentially distributed (M)
There are c servers
The length of queue in which arriving users wait before
being served is finite (N)
o The population of users (i.e. the pool of users), or requests,
available to join the system is infinite.
CUSTOMERS:
HOSTESSES
OHOHOE B
exIT
OD OM was
FIGURE 2.4 Schematic showing restaurant plan and structure
Scanned with CamScanner e and Simulation
&
‘al Introduction to
2.8.3 SPECIFICATION MODEL ;
This is a more detailed specification of the model.
° Customers
© What is the arrival process?
«© Schedule of customer arrivals, ¢.g., table reservation recor4,
+ Often, probability distribution defines time betwee,
successive customer arrivals (inter-arrival time)
© Customer attributes?
« Does the customer have a previous reservation or did he
just "walk-in"
e Queue of choice
» Where would the customer prefer to wait?
« He/They could choose to wait at the bar or just move
outside to the benches
« They could also choose to eat at the bar, in that case,
they would be part of the queue
© Servers
« How much service (ordering + eating) time is needed for
each customer?
« We Can use probability distribution to specify customer
service time
e How many servers?
e 25 tables (which can be joined/arranged in any order to
accommodate a party size between 2 to 20)
In both examples below, the restaurant comprises eleven 2-tops,
four 4-tops, and two 6-tops, combinable tables are linked by lines
Scanned with CamScanner Modelling 33
OOO
e Queue fe
« Service discipline - who gets called to the table next?
. First-in-first-out (FIFO), Customers are called to the tables
based on who registered at the Podium first
« Preemption: if there is a custo
reservation, then he will be the next
there is a vacancy,
Queue capacity? What if the queue overflows?
+ Queue capacity is finite i.e. 30 in total (Bar = 10, Benches
= 20)
mer with a previous
t to be called whenever
2.8.4 The M/M/C/N Queue
There are c 21 servers and the waiting room has finite capacity.
If more than one server is available when a new customer arrives
(which necessarily implies that the waiting room is empty) then
the incoming customer may enter any of the free servers. Let }
and u be the rate of the Poisson process for the arrivals and the
parameter of the exponential distribution for the service times,
respectively. This model with exponential inter arrival times with
mean 1/A exponential service times with mean 1/u and c parallel
identical servers. Customers are served in order of arrival. We
suppose that the occupation rate per server, p = Meu, is smaller
than one. The model is considered stable only if u
