Scribd
Upload
17 views

Lecture - 1

Uploaded by

omerfaruk200141
Copyright
© © All Rights Reserved
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
17 views

Lecture - 1

Uploaded by

omerfaruk200141
Copyright
© © All Rights Reserved
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 19

Lecture 1: Introduction to Simulation

END 322E System Simulation Mehmet Ali Ergün, Ph.D.


System
❖ A system is any set of related components that together work toward some
purpose.
❖ Manufacturing facility
❖ Bank operations
❖ Airport operations (passengers, security, planes, crews, baggage)
❖ Transportation/logistics/distribution operation
❖ Hospital facilities (emergency room, operating room, admissions)

2
Ways to Study a System

3
Model
❖A model is the conceptualization of a real system
❖ Requires set of assumptions, hence a simplification
❖ Relations between entities of the system can be expressed in
terms of
❖ Mathematical

❖ Logical

❖ Can be physical or mathematical

4
Modeling as an activity
❖ Why do we use models?
❖ Show some aspects of the real-world systems
❖ Ask what-if questions
❖ They help us to understand what is going on so we can make decisions

❖ Famous Quotation:
❖ Essentially, all models are wrong, but some are useful.
George E. P. Box (Professor Emeritus of Statistics at the University of Wisconsin)

❖ Good to remember: A model is an approximation


What is Simulation?
❖ Simulation – very broad term – simply: A
modeling approach to imitate or mimic
complex real systems, usually via computer. (A
computer model of a real system)
Simulation is everywhere
❖ Entertainment
❖Many popular games are based on simulation.

❖ Aviation
❖Wind tunnel simulations help design more efficient planes.

❖Pilots trained on flight simulators mean fewer crashes .

❖ Manufacturing
❖Simulation is used to optimize production lines and schedule productions

❖ Military
❖Planning logistic operations

❖War scenario games


Some Application Areas
❖ Design/improve a processing center for shipping orders.
❖ Design/improve visitor flow at an amusement park.
❖ Design/improve an automobile assembly line.
❖ Design/improve the back office processes for a bank.
❖ Design/improve the baggage handling system at an airport
❖ Reduce waiting times in an emergency department.
❖ Process improvement at an ambulatory surgery center.
❖ Evaluate a disaster plan for emergency services.
When Simulation is Appropriate?
❖ To study and experiment on complex systems
❖ Informational, organizational, and environmental changes can be
simulated to explore better alternatives
❖ Knowledge gained when building simulation model toward suggesting
improvement
❖ To verify analytical solutions
❖ To prepare what-if scenarios
❖ Animations help visualization of the operations
Drawbacks of Simulation
❖ Results are not exact due to randomness, only approximations
❖ May lead to misinterpretations
❖ Statistical design, analysis of simulation experiments requires attention in
detail
❖ Not easy to explore all possible alternatives, no optimality guarantee
❖ Sometimes very time-consuming/costly
❖ One single run tells us little, often with high variance
Don’t simulate when… (Banks &
Gibson)
❖ A “common sense” solution exists
❖ The problem can be solved analytically*
❖ i.e. simple queuing system
❖ If direct experiments are easier and cheaper to perform
❖ Simulation cost exceeds potential savings:
❖ Remember to budget for: planning/problem definition, data collection,
model validation, experimentation and analysis
❖ There are not enough resources (people, money, time) available
1
1
Types of Simulation
Static vs. Dynamic
• Does time have a role in model?

Continuous-change vs. Discrete-change


• States change continuously or at discrete points of
time?

Deterministic vs. Stochastic


• Does randomness play a role?

1
2
Randomness
❖ Simulation models that contain no random variables are called deterministic
simulations
❖ Scheduled arrivals to a dentist
❖ A stochastic simulation model has one or more random variables as inputs
❖ Simulation of a bank involves random arrivals of customers and service
times
❖ Random inputs lead to random outputs
❖ Output measures are averages, variances, and distributions, hence are
statistical estimates
Randomness and System Performance
Patients arrive with The average
an average time treatment time is
between arrivals of 55 minutes.
1 hour

Server

How will this system perform?


Arrival/Service Process Result
Constant/Constant No waiting.
Random*/Constant 5 hour average wait.
* Exponential Distribution Random*/Random* 10 hour average wait.
Randomness in a simulation
❖ Typical random components: processing time, service time, customer
arrival times, transportation times, machine failures or repair times, etc.

❖ For example, if you head to the drive-through window at a local fast-


food restaurant for a late-night snack, you cannot know exactly how
long it will take you to get there, how many other customers may be in
front of you when you arrive, or how long it will take to be served, to
name just a few variables
Organizing a simulation study
❖ Most of the work is not the simulation itself

❖ Start with: what is the study about?


❖ What will it be used for?
❖ Who has to be involved?
❖ What will be the measure of effectiveness?
❖ How long will the study take?
❖ What is the cost?
Conceptual Model
❖ Get the required information and the data
❖ Not the same!

❖ Write an initial assumptions document


❖ What is the model?
❖ What will be ignored or simplified?
❖ Where can you get additional data, if needed?

❖ Discuss the document with key people


Program the model
❖ An issue: code the model in a programming language, or use a simulation package?
❖ Programming language may run faster, and marginal software cost usually low, but
it may be buggy and take longer to develop
❖ Simulation package may produce many fewer errors, lower total cost, but less flexible
❖ Pilot runs and debugging (verify and validate)
❖ Compare results with real system's output, if real system exists
❖ Even if it doesn't, get opinions from Subject Matter Experts (SME) on the model
❖ Important not only to be correct (valid), but also to look correct to influential people
(credible)
Analyze data and report results
❖ Design experiments that will answer the desired questions
❖ Determine how these will run: replications, warm-up length, length of each run
❖ Statistical analysis is essential
❖ Run the model as required within these policies
❖ Document, present, and use the model and its results
❖ People often underestimate importance of credibility
❖ Decision-makers often aren't technically competent
❖ They won't act unless they feel that the model is reasonable
❖ Important to keep them informed

You might also like