Discrete-Event System

Simulation

An Introduction to the
Basic Principles of
Simulation
1
 Modeling
 Modeling involves observing a system,
noting the various components, then
developing a representation of the
system that will allow for further study
of or experimentation on the system
 Focus – computer model
 Data Structures & Implementation
 Interaction of the components
2
 Simulation
The process of running a (computer) model
of a real system to study or conduct
experiments
 For understanding the model or its behavior
 To evaluate strategies for operation of the
system
 Involves generation of an artificial history, used
to draw conclusions about the real system

3
 Modeling & Simulation
 Often described as one process
 Should distinguish between the two

4
 Simulation as the
Appropriate Tool
 Enables study and experimentation
 Changes simulated & results observed
 Gain knowledge of system
 Determining importance of variables and
how variables interact
 Experiment before implementation
 Verify analytic solutions
5
 Simulation as the
Appropriate Tool (cont’d.)
 Try different capabilities (of a machine)
 Training
 Animation (graphics)
 Complexity of modern systems almost
require simulation

6
 When Simulation is
Not Appropriate
 If can be solved by
 Common sense or simple calculations

 Analytical methods

 Direct experiments

 If simulation costs exceed savings

 If resources & time are not available

7
 When Simulation is
Not Appropriate (cont’d.)
 If Data is not available
 If verification & validation are not practical
due to limited resources
 If users have unreasonable expectations
 If system behavior is too complex

8
Advantages of Simulation
1. Control
2. Time compression
3. Sensitivity Analysis
4. Training tool
5. Doesn’t disturb real system

9
 Advantages

 New policies, operating procedures, decision rules,

information flows, organizational procedures, etc.
can be explored w/o disrupting ongoing
operations
 New hardware designs, physical layouts,
transportation systems, etc. can be tested w/o
committing resources for their acquisition
 Hypotheses about how or why certain phenomena
occur can be tested for feasibility
10
 Advantages #2
 Time can be compressed or expanded allowing for
speedup or slowdown of the phenomena under
consideration
 Insight about the interaction of variables or the
importance of variables on performance of the
system
 Bottleneck analysis can be performed indicating
where processes are being delayed
 “What if?” questions can be answered –
particularly for a new system
11
 Disadvantages of Simulation

1. Expensive
2. Extensive time needed
3. Lack of experienced
personnel
12
 Disadvantages
(Pegden et al. 1995)

 Model building requires special training and

experience
 Results may be difficult to interpret
 Time consuming and expensive
 Use of simulation when analytical models are
available and preferable, particularly for
closed-form models

13
Why Simulate?

 To save money
 To do things you could not

physically or morally do
within the actual system

14
 Areas of Application
 Manufacturing, Semiconductor Mfg.
 Construction & Project Management
 Military
 Logistics, Supply Chain, Distribution
 Transportation & Traffic
 Business Processes
 Health Care
15
 Current General Trends
 Risk Analysis
 Insurance, options pricing, portfolio analysis
 Call Center Analysis
 Large Scale Systems
 Internet backbones, wireless networks, supply
chains
 Automated Materials Handling (AMHS)
 Control system sw - emulator
16
 System
 A set of inputs which pass through certain
processes to produce outputs
 A set of related components which work
together toward a given goal
 A group of objects joined in regular
interactions or interdependence for the
accomplishment of some purpose
 Helpful if a system is observable,

measurable, systematic
17
 System Environment
 “World” in which the system exists
 System is affected by elements outside the
system – the system environment
 Boundary – “line” between the system & its
environment
 Decision on boundary is dependent upon
simulation purpose

18
 System Components
 Consists of objects called ENTITIES
 Entities have a set of properties called
ATTRIBUTES that describe them
 There exist interactions called ACTIVITIES and
or EVENTS that occur between the entities that
cause them to change
 The STATE OF A SYSTEM is a snapshot of the
system at a given time
i.e. variables necessary to describe system
 The model starts in its INITIAL STATE
19
Activities & Events
 Cause changes in the attributes of
the entities, and, therefore, the
state of the system
 Event: instantaneous
 Activity: has a length of time

20
 System Component Examples
 Bank
 Computer Network
 Hospital Emergency Room

(Homework)

21
 Activities & Events
 2 types of Events or Activities
 Endogenous: variables affecting the

system which are (can be)

manipulated within the system
 Exogenous: variable which affect the

system but cannot be manipulated by

the system because they are outside
the system.

22
 Classifications of Systems
1. Static (Monte Carlo) vs. Dynamic
2. Deterministic vs. Stochastic
3. Continuous vs. Discrete
 D: state vars. change at discrete points in time
 C: state vars. change continuously over time
 Simulate
 Stochastic - Dynamic - Discrete or Continuous

23
 Model
 The representation of an object in some form
other than the form of the object itself, usually
for the purpose of study or experimentation

 Why Model???

24
 Classification of Models
 1. Physical: an actual representation
 2. Schematic: a pictorial representation
 3. Descriptive: a verbal description
 4. Mathematical: components are
described mathematically, in the form of
equations
 5. Heuristics: descriptive model based on
rules; algorithmic; - computer based
25
Characteristics of a Good Model
 Simple to understand
 Goal directed

 Robust

 Easy to control

 Complete on important issues

 Adaptive and easy to update

 Evolutionary

26
27
 Steps in a Simulation Study
(Figure 1.3)
1. Problem Formulation
1. Statement of the problem
2. Set Objectives & Project Plan
1. Questions to be answered
2. Is simulation appropriate?
3. Methods, alternatives
4. Allocation of resources
1. People, cost, time, etc.

28
 Steps in a Simulation Study
(cont’d.)
3. Model Conceptualization
1. Requires experience
2. Begin simple and add complexity
3. Capture essence of system
4. Involve the user
4. Data Collection
1. Time consuming, begin early
2. Determine what is to be collected
29
 Steps in a Simulation Study
(cont’d.)
5. Model translation
1. Computer form
2. general purpose vs. special purpose
6. Verification
1. Does the program represent model and
run properly?
7. Validated?
1. Compare model to actual system
2. Does model replicate system?
30
 Steps in a Simulation Study
(cont’d.)
8. Experimental Design
1. Determine alternatives to simulate
2. Time, initializations, etc.
9. Production & Analysis
1. Actual runs + Analysis of results
2. Determine performance measures
10. More Runs?

31
 Steps in a Simulation Study
(cont’d.)
11. Documentation & Reporting
1. Program & Progress Documents
2. Thoroughly document program – will
likely be used over time
3. Progress reports are important as project
continues – history, chronology –
changes, etc.
12. Implementation
32
 Ten Reasons for Failure (notes)

1.Failure to define an achievable goal

2.Incomplete mix of essential skills
 Project leadership

 Modeling

 Programming

 Knowledge of modeled system

3.Inadequate level of user participation

4. Inappropriate level of detail
5.Poor communication
33
 Failure (cont.)
6. Using the wrong computer language
7. Obsolete or Nonexistent Documentation
8. Using an unverified model
9. Failure to use modern tools and
techniques to manage the development of
a large complex computer program
10. Using Mysterious Results
34
Stochastic Behavior
 Monte Carlo
 Random, but not over time
 E.G. Darts on a dart board

 Pseudorandom
 Time dependent, Reproducible
 E.G. Customer arrivals

35