introduction in simulation dr. christoph laroque summer 2012
TRANSCRIPT
Fakultät Informatik, Institut für Angewandte Informatik, Professur Modellierung und Simulation
Dresden, 04.07.2012
Introduction in Simulation
Dr. Christoph LaroqueSummer 2012
04.07.2012
Einführung in die Simulation Folie Nr. 2
Folie Nr. 2
Next week
04.07.2012
1. Answer open questions to the lecture, either content or organisation. Please reflect and send open questions till Monday, 6 p.m. via email.
2. Practise with FlexSim… 3. Decide, if presentation by google can take place on
Thursday 12th - yes, still an open issue..
• extra-practise – self-management – interest?
Einführung in die Simulation Folie Nr. 3
Folie Nr. 3
VALIDATION & VERIFICATION
Introduction in SimulationSummer 2012
04.07.2012
Einführung in die Simulation Folie Nr. 4
Folie Nr. 4
Objectives of V&V
The objectives of the validation and verification of simulation models and experiments are:
• Create comprehensible principles about the credibility of the model• Recognize errors during the modeling at an early stage• Ensure, that lessons-learned is incorporated in further modeling• Proper use of good models and correct interpretation
Einführung in die Simulation Folie Nr. 5
Folie Nr. 5
Approach in a simulation study - II
DATA MODELSproblemdefinition
Planningthe Data
collection
Datacollection
Dataprocessing
Experiments
System analysis
Model formalization
Implemen‐tation
Evaluation
Validation and Ve
rification of data
Validationand Verification of m
odels
Interim results
Einführung in die Simulation Folie Nr. 6
Folie Nr. 6
Terms
Validation is a continuous review, whether the models reproduce the behavior of the illustrated system with sufficient accuracy.
(Does it the right thing?)
Verification is the review, whether a model has been transformed from a mode of description in a different mode of description correctly.
(Does it the right thing right?)
Einführung in die Simulation Folie Nr. 7
Folie Nr. 7
Terms
Testing occurs regularly from the beginning of the development to completion. We distinguish:
• Factual correctness (sufficiently accurate reproduction)• Functional correctness (all functions available for the target
position?)• Technical correctness (If all conditions met, for example, computer
time, interfaces?)
Einführung in die Simulation Folie Nr. 8
Folie Nr. 8
Criteria
A Model can be only checked against the background of a specific purpose for its validity.
General criteria can be determined as:
• Completeness (correctness of the content and structure)• Consistency (correctness of the content and structure)• Accuracy (reasonableness of the result)• Relevance (adequacy of the result)• Appropriateness (adequacy of the result)• Plausibility (reasonableness of the result)• Intelligibility (adequacy of the result)• Feasibility (feasibility)• Availability (feasibility)
Einführung in die Simulation Folie Nr. 9
Folie Nr. 9
Result documentation
The individual phases of the simulation study can be secured by specified documents:
• target description• task specification• concept model• formal model• executable model• simulation results• raw data• formatted data
Einführung in die Simulation Folie Nr. 10
Folie Nr. 10
Result documentation task specification
(Extract from outline proposal)
1. Description of the target and tasks2. Description of the investigated system3. Necessary information and data4. Planned use of model
Period of use User community and qualification Type of model use
5. Approach and method6. Requirements of model and modeling
Einführung in die Simulation Folie Nr. 11
Folie Nr. 11
Result documentation task specification
(Excerpt of evaluation questions)
• If the task specification being examined with the department heads?
• Is the selection of the chosen solution methods substantiated?• If the purpose of the simulation study clearly defined?• Is the description of the system as a whole coherent and
understandable?• Is the named V&V approach to cost and benefit points of view
adequate?
Einführung in die Simulation Folie Nr. 12
Folie Nr. 12
Techniques of verification and validation
• Animation• Assessment• Dimension test• validity test event• Read-only test• Limit Test• Monitoring• Writing test• Sensitivity analysis• Statistical techniques• Structured runaway• Test of internal validity• Test of partial models
• Trace Analysis• Turing test• Cause-effect graph • Validation dialog• Validation of predictions• Comparison with other models• Compared with recorded data
Einführung in die Simulation Folie Nr. 13
Folie Nr. 13
Techniques: Animation
With animation can be particularly pointed out that the behavior of a model in certain situations is not valid.
advantage:• Verification of the model behavior in selected model sections• Localization errors of detail• Direct observation of the model curve
disadvantage:• Misrepresentation of detail• Rarely occurring error is not detected
Einführung in die Simulation Folie Nr. 14
Folie Nr. 14
Techniques: Writing Test
Careful review of their own work in terms of completeness, accuracy, consistency and clarity
advantage:• Own work will be reconsidered again at rest
disadvantage:• Certain errors are not detected (blindness)
note:• Consult any other person and validation dialog
Einführung in die Simulation Folie Nr. 15
Folie Nr. 15
Techniques: sensitivity analysis
This will select critical input parameters of the model and determines the impact on output parameters
advantage:• identification of particularly sensitive parameters• comparison with real behavior
disadvantage: • dynamic effect relationships and interaction effects of the
parameters are not recognized
Einführung in die Simulation Folie Nr. 16
Folie Nr. 16
Techniques: Structured runaway
Technology in software development. Responsible go together through each statement in the program.
advantage:• 4-eye principle (at least)• unity of all project managers
disadvantage:• very time consuming, especially for complex models and / or long
documents
Einführung in die Simulation Folie Nr. 17
Folie Nr. 17
Process Model - Guidelines
1. V&V are to be performed integrated; all the measures, procedures, etc. for verification for validation are treated together
2. V&V is a process which always accompanies the entire simulation study
3. Basis are documented period results
4. The data must be considered a special way.
Einführung in die Simulation Folie Nr. 18
Folie Nr. 18
Process Model
Hier kommt das Bild
Einführung in die Simulation Folie Nr. 19
Folie Nr. 19
Quality criteria
Quality is the totality of features and feature values are one unit on its ability to meet all specified and assumed requirements.
• Refers not only on results but also on the approach• Quality of a simulation project is depending on project-specific
requirements• Is determined by the care and systematic implementation of the
project preparation• Comparisons deadly sins of Liebl
Einführung in die Simulation Folie Nr. 20
Folie Nr. 20
Quality criteria
Five major quality criteria can be distinguished:
1. Careful project preparation2. Consistent documentation of all project activities3. Collaborative verification and validation4. Continuous integration of the client5. Systematic project implementation
Einführung in die Simulation Folie Nr. 21
Folie Nr. 21
Quality criterion for project preparation
As part of project preparation some early policy decisions are made:
• Check dignity of simulation• Determine the acceptance of the simulation method • Formulate a clear major task for study• Selecting a service provider• Selecting of a simulation tool
• Consistent documentation
Einführung in die Simulation Folie Nr. 22
Folie Nr. 22
QUESTIONS?
Introduction in SimulationSummer 2012
04.07.2012