The Future of Simulation

1
The Future of Simulation

• Jerry Banks
• Atlanta, Georgia 30319
• USA

2
The 80s

• We need to understand the past to predict the
  future
• As of 1980, simulation languages were reasonably
  popular in manufacturing, computer and
  communications modeling, and military and
  government applications

3
The 80s

• The larger organizations were the users of the
  technology
• A large number of universities were teaching
  simulation

4
Software Developments 80s, early 90s

• AutoMod
• Development of a special purpose language for
  material handling
• SIMAN
• Simulation software on PCs
• SEE-WHY
• Animation and graphics on PCs and workstations

5
Software Developments 80s, early 90s

• FACTOR
• Simulation-based scheduling technology
• ProModel and WITNESS
• Tools designed for graphical model building

6
Software Developments 80s, early 90s

• AutoSched AP
• Simulation-based scheduling for day-to-day
  operations in a major industry
• Extend and SLX
• Tools geared toward flexibility

7
Making it Possible

• Making much of this possible was the rapid
  improvement in graphics and processing power of
  the PC
• Another notable feature was the penetration of
  simulation analysis into virtually all capital
  projects of the automotive industry

8
Why simulation?

• One of the top 3 technologies
• Simulation is no longer the technique of last
  resort
• It is an indispensable problem-solving methodology

9
Why simulation?

• Real systems are random
• Demands
• Time to failure
• etc.
• Random systems cannot be easily analyzed with
  static modeling tools

10
Why simulation?

• Modern manufacturing system
• Investment costs are high
• Tolerance for error is low
• Simulation helps to
• Hold costs down
• Decrease errors

11
Spread of Simulation

• Steady progression of simulation users
• But, not growth by leaps-and-bounds

12
A paradox

• Simulation is a widely accepted tool for
  predicting performance of complex systems
• Simulation is used in only a small fraction of
  the cases where it can have a significant impact

13
Competition

• My biggest competitor is the decision not to use
  simulation
• Brian Stanley

14
Spread of Simulation

• Simulation is both an art and a science
• Bob Shannon said it in 1975
• Still true today
• My first PC in 1981
• Wordstar
• Open a book and start typing
• Simulation model building requires a system
  perspective and understanding

15
Spread of Simulation

• Complex problems require complex solutions
• Averill Law
• Reducing the complexity is a great challenge

16
Spread of Simulation

• Reducing complexity
• Not to be achieved at the expense of flexibility
• Simulators, templates, palettes, etc.
• Good, but must be robust

17
Spread of Simulation

• Simulation takes time and money
• No wine will be sold before its time
• Simulation takes time
• Cases where decision makers cant wait for the
  simulation results
• Bad for the simulation business
• Good for the simulation business
• John Carson

18
Spread of Simulation

• Interpreting simulation results can be difficult
• Especially true for those whose statistical
  competence is low
• Use an output analyzer

19
Spread of Simulation

• Simulation is slow
• CPU speed going up
• Still there are users complaining about the time
  to perform a replication
• As models get bigger

20
Spread of Simulation

• Visualization is a great benefit
• Simulation must look as good as Saturday morning
  cartoons
• Enhance visualization
• But, dont increase complexity of use

21
Introductory Tutorial at WSC 1999

• What do you see happening in simulation over the
  next five years?

22
WSC 1997 and 1998

• Seven areas in 1997
• Five areas in 1998
• Internet applications
• ERP applications
• Embedded simulation
• Optimization
• Object-oriented simulation

23
WSC 1997 and 1998

• Only two areas in 1997 were in the 1998 list
• 2/7 0.29
• 0.295 0.002
• That is, believe me with an accuracy of 0.2

24
Further disclaimer

• The future is not just unknown, it is unknowable
• Simulation will be customer-driven, and
  market-based
• Market is demanding easier development, more
  functionality, greater accuracy

25
WSC 1999

• Mixed it up
• Consultant
• Academic
• Two corporate simulationists
• Two software vendors
• Pull rather than push
• What does industry need from simulation vendors
  in Y2k and after?

26
WSC00

• The viewpoint of seven consultants

27
WSC01

• Panel Session
• Emulation
• Data exchange standards
• Optimization

28
Panel Session

• Representatives in seven different areas talked
  about their needs
• Manufacturing
• Services
• Consulting
• Logistics
• Construction
• Military
• Wafer fabrication

29
Emulation

• Two presentations
• Reducing commissioning costs in a high-speed
  bottling line
• Test control logic on a simulated 3D model prior
  to startup
• Debug it in the lab, not on the floor
• Control logic is connected to a simulation model
  that imitates the actual machine/material
  handling device to correct logic problems in the
  office

30
Data Exchange Standards

• From CAD drawings to simulation modeling for
  material handling system eliminating potential
  errors and saving lots of time

31
Simulation Optimization

• Recent Ph.D. graduate from Northwestern gave an
  introduction
• Developers of the following reacted
• OptQuest
• SimRunner
• RiskOptimizer

32
WSC02

• Alexander Verbraeck
• Emulation?
• Chuck McClain
• Standards (for VV?, for output analysis?, for
  certification as a simulation analyst?)
• Jerry Banks
• Opportunities for simulation in SCM
• Averill Law
• Panel session, vendors

33
Whats in?

• Better and tighter integration of simulation
  software and other software for the exchange of
  data (CAD, ERP/MES, WMS, MCS)

34
Data Exchange Standards

• Lowest level
• CAD to simulation software
• Most software can import a CAD file
• Three components
• Movement systems
• Dynamic entities
• Static backgroundthe CAD drawing

35
Data Exchange Standards

• Material handling system
• Standardized format
• Conveyors
• Pathways
• Vehicles
• etc.
• Collaborative modeling would be possible

36
EAI and SDX

• Engineering Animation Incorporated
• Simulation interface
• SDX
• Simulation Data Exchange
• Several simulation firms can import SDX files
• Works best with Material Handling System

37
XML

• Extensible Markup Language
• World Wide Web Consortium (W3C)
• Rules, guidelines, conventions for designing text
  formats for data
• Such that files are
• Easy to generate and read
• Unambiguous
• Extensible

38
XML

• Large and growing community
• Many tools do not yet exist
• Coordinated by NIST
• Eventually, some group will want to develop XML
  standards for Mfg/MH

39
Whats in?

• Tighter implementation of simulation software
  with control systems and controls emulators to
  test controls at the MES and PLC levels before
  implementation to drastically reduce testing and
  de-bugging

40
Whats in?

• Simulation tools will continue to be integrated
  with other tools to form suites

41
Whats in?

• Component-based simulation environment.
• Includes library of pre-built icons that allow
  simulation models to be built much faster
• Object-orientation via C is too hard for most
  simulation practitioners

42
Object-orientation possibility

• Object-Oriented Simulators that Require no Hard
  Coding
• Completely menu-driven

43
Six possibilities using the web

• Online documentation
• Run the model on client computer
• Remote access
• PADS on the web
• Rent software on the web
• Collaboration

44
Online Documentation

• The documentation can be about the modeling
  components, or the results of output analysis for
  a set of inputs, or even animations that display
  the simulation modeling results
• Hypertext is the medium for documentation

45
Run the Model on Client Computer

• Input parameters are provided through the web

46
Run the Model on Client Computer

• The simulation model and user interfaces run on
  the client as applets
• Applets are programs written in Java that are
  embedded on web home pages
• The simulation model components and user
  interfaces are downloaded seamlessly on the
  browser and are executed at the users desktop

47
Remote Access

• The simulation model is accessed remotely from a
  web browser and executed on a server platform
• The results can then be viewed on a browser
• Requires knowledge of common gateway interface
  (CGI) scripts, Java servelets, or middleware
  technologies such as CORBA
• The different means of enabling server-side
  connectivity have an impact on execution speed

48
PADS on the Web

• Multiple users can interconnect through web
  browsers from different locations
• The simulation model executes on the server side
• The user interfaces could be tailored to
  different users depending on their level of
  abstraction
• A possible application could be for training

49
PADS on the Web

• http//ms.ie.org/websim/survey/survey.html

50
PADS on the Web

• PADS requires that each model be taken apart so
  that each CPU can run a part or module
  simulation software vendors dont want to bother
  with such a solution

51
PADS on the Web

• Instead of screen saver coming on, computer could
  begin processing a simulation
• Over the internet

52
Distributed Simulation

• Replications made on many computers within a
  network
• This is done now

53
Distributed Simulation

• Currently, it is possible to distribute
  replications over a local network
• If 10 replications are needed, each requiring 4
  hours, we can distribute that task to five
  computers, each performing two replications over
  night
• The host computer controls the others and
  collects the output

54
Rent Software?

• Clients could pay for use of software via the web
  on a time basis

55
Collaboration

• Larger models require joint effort in model
  building and maintenance
• Modelers dispersed in time and space can work
  together via the internet

56
Supply Chains

• MHRC at Georgia Tech from 1983 to 1992
• Then logistics became the buzz word
• logistics supply chains
• The Logistics Institute
• Many companies going through big changes

57
Supply Chains

• Simulation is the friend of change
• Dennis Pegden
• Change is the friend of simulation
• Simulation models can be used to understand these
  new processes

58
Supply Chains

• Simulation models are used to study the entire
  supply chain from suppliers to final customer
  delivery
• Complex systems with many interacting and random
  elements

59
Embedded Simulation

• Simulation software needs to become less visible!
• Already true in simulation-based scheduling
• Strength of current simulation software is not in
  the interface
• It is in the ability to model randomness

60
Optimization

• Five years ago optimization was in its infancy
• Now, most of the software vendors have some
  optimization
• But, there are still problems

61
Optimization

• Major problem is the time that it takes
• Say that a simulation takes 10 minutes per
  replication and that we are doing five
  replications
• Thats 50 minutes

62
Optimization

• Genetic algorithms
• Work with function evaluations
• 1000 would not be unreasonable
• 10 x 5 x 1000 50,000 minutes
• More than one month
• Unreasonable

63
Optimization

• Smart user can take short cuts
• Eliminate experiments
• Little impact on outcome
• Use factorial experiments in a clever way
• Much research left to be done
• Lots of opportunities for Ph.D. dissertations

64
HLA

• Much has been said recently about High Level
  Architecture
• To facilitate interoperability
• To promote reuse of simulations

65
HLA

• Started in the simulation of combat
• Exported to the world of manufacturing and
  material handling
• IMS MISSION
• Lots of acronyms
• First five pages had 19 acronyms

66
HLA

• In manufacturing and material handling
• There has to be a data exchange format
• This is an evolutionary process
• Time flow and synchronization have to be worked
  out
• Simulation software is inconsistent in its
  internals
• Schriber and Brunner
• Tendency of vendors
• Stick to your knitting!

67
HLA

• Will go nowhere in industrial use unless it can
  be greatly simplified and revised so that models
  can be developed transparently

68
Next great application area?

• Parcel and letter handling
• Healthcare
• Penetrated but not permeated
• Pharmaceutical and chemicals
• Mining and mineral processing
• Printing and publishing
• Mid-sized manufacturing

69
Next great application area?

• Customized templates reflecting the customers
  niche are needed

70
Summary

• Letting the mind wander allows for many future
  possibilities
• But, my predictive abilities have not been that
  good in the past!

71
End