SAE 599 Modeling and Simulation for Systems Architecting and Engineering

1
SAE 599 - Modeling and Simulation for Systems
Architecting and Engineering

• Dr. Raymond Madachy
• September 19, 2007

2
Outline

• Course Project Status
• Simulation Tools Revisited
• Demonstrations of Reference Applications with
  Continuous Systems Modeling
• Homework

3
Course Project Status

• Already chosen
• Jared Fortune EZSIM II architecting and
  acceptance testing (EZSIM)
• Harry Johnson simulation of sustaining
  engineering for the International Space Station
  integration processes (Simulink)
• Jared Smith - Resource adaptability model for
  disasters and terrorist attacks
• Possible
• Ben Haas - Carbon emissions model for China to LA
  cargo transport
• Mentored projects still available
• Motor sports simulation (Settles)
• Limits to change on large SOS projects (Boehm,
  Lane)
• Front-end GUI for web-based simulation tool
  (Madachy)
• Major exercises from Software Process Dynamics
  (Madachy)

4
Simulation Tool Reviews

• Over 250 tools reviewed against more than 50
  attributes
• Recommendations
• EZSIM for discrete event modeling
• Continuous systems modeling
• Ithink/Stella (100 from http//iseesystems.com)
• isee player allows running and viewing of
  Ithink/Stella models (http//www.iseesystems.com/s
  oftwares/player/iseeplayer.aspx)
• Can use for class demonstration models
• Vensim Personal Learning Edition (free from
  http//www.vensim.com/venple.html)
• Agent-based modeling
• NetLogo (http//ccl.northwestern.edu/netlogo/docs/
  )
• Others per agreement for course projects (Extend,
  Simula)
• Question Access other students tool reviews?
  Objections?

5
Simulation Tools Redux

• General purpose, event-oriented languages
• User constructs event logic and changes to system
  state
• Includes standard discrete-event modules (event
  calendar, statistics, output, etc.)
• Low level languages, flexible, but hard to learn
• Examples
• GASP
• SIMSCRIPT
• MODSIM
• General purpose, process-oriented languages
• Contain modules for common processes, with
  varying degrees of implementation in different
  tools
• Entity flow diagram often used to construct
  model, and manually or automatically translated
  into source statements
• Examples
• EZSIM
• SLAM
• GPSS
• SIMAN
• SIMNET

6
Simulation Tools Redux (cont.)

• Continuous system simulation tools implement
  rate/level concepts
• User manually constructs models or uses a GUI
• Examples
• Ithink
• Vensim
• Powersim
• Forio
• Special purpose simulation environments
• Domain specific process-oriented tools (e.g.
  manufacturing, communications, traffic)
• High level model construction with icons and
  menus
• Examples
• SIMFACTORY, WITNESS for factory simulation
• COMNET for communications networks

7
Simulation Tools Redux (cont.)

• Tool selection
• Advantages of traditional computer programming
  languages
• Flexibility and control, existing compilers, no
  training for simulation-specific language,
  efficiency, portability
• Advantages of general purpose simulation
  languages vs. traditional languages
• Leverage of high-level commands and modeling
  framework, faster development, more
  understandable due to modularity
• Advantages of special purpose simulation tools
  vs. general purpose simulation languages
• Simpler to learn for narrower domain, use of
  icons and menus, use of domain-specific
  terminology
• Other considerations
• Price, documentation, training, support and
  maintenance, computer platform, user familiarity

8
Demonstration Problems

• The arrival process of cars on a freeway is
  exponentially distributed with a mean of 3
  minutes. The time to departure from the freeway
  is normally distributed with a mean of 15 minutes
  and standard deviation of 2 minutes. Simulate the
  process and observe the accumulation of cars on
  the freeway.
• Customers arrive uniformly between 10 and 20
  minutes, and the service time is distributed
  uniformly between 8 and 15 minutes. Simulate the
  system for 500 arriving entities or 10,000
  minutes of operation, whichever happens first.
• A device has two components that operate in
  parallel. It fails if both components fail, and
  the probability of each component failing after
  one hour is 0.5. Simulate the system operation
  to find the time of device failure.

9
Freeway with Constant Rates

• Rates 1 / interarrival times
• Minutes converted to hours

10
Freeway with Stochastic Rates

• Departure rate is not per specification. It
  mirrors the arrival rate with a fixed delay time.
  

11
Device Failure

• The simulation time step DT must be set to one
  hour for random failure variables
• Random functions and probability distributions
  sampled each DT in simulation

12
Demonstrations
13
Homework

• Compare and contrast the freeway application
  models using discrete event and continuous
  systems simulation
• Advantages and disadvantages for this type of
  application
• Does either method seem better suited?
• Extra credit specify and/or implement a
  stochastic departure time in the continuous model
• Use EZSIM or another discrete event modeling
  tool
• There are two roads that join into one highway.
  Due to the risk of collision, there is a traffic
  light that controls the highway entry. The
  interrarrival time on one of the roads is
  exponentially distributed with a mean of 4
  seconds while the other is 6 seconds. The light
  for the road with heavier traffic remains green
  for 40 seconds and 30 seconds for the other road.
  Observe the queue buildup on both roads.

14
Homework (cont.)

• Use EZSIM or another discrete event modeling
  tool
• At an air terminal, cargo planes with capacities
  of either 50 or 70 cargo units arrive at the rate
  of one every 60 minutes. Each plane waits until
  it is loaded to capacity, then it takes off.
  Fifty-five percent of the arriving planes have a
  50-unit capacity. Cargo arrives in batches of
  10-unit loads with an exponential interrarrival
  time with a mean of 9 minutes. Modeling the unit
  loads as resources, run the simulation model of
  this system for 100 arriving planes to determine
  the average waiting time of each plane before
  takeoff.